| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Department of Pharmacology, Yale University School of Medicine, New Haven, Connecticut 06510
 |
Abstract |
|---|
 Top Abstract I. Introduction: Concentrating...II. A Well Characterized...III. Sorting of Soluble...IV. Two Models for...V. Models for Formation...VI. Hormone Aggregates in...VII. Hormone Aggregates in...VIII. Does Aggregation Cause...IX. What Keeps Secretory...X. Sorting in the...XI. Sorting Signals and...XII. All Secretory Granule...XIII. SummaryReferences |
|---|
|
I. Introduction: Concentrating Protein Hormones in Secretory Granules |
|---|
|
TopAbstractI. Introduction: Concentrating...II. A Well Characterized...III. Sorting of Soluble...IV. Two Models for...V. Models for Formation...VI. Hormone Aggregates in...VII. Hormone Aggregates in...VIII. Does Aggregation Cause...IX. What Keeps Secretory...X. Sorting in the...XI. Sorting Signals and...XII. All Secretory Granule...XIII. SummaryReferences |
|---|
The secretory pathway in neuroendocrine cells, exocrine cells, peptidergic neurons, mast cells, and other secretory cells that store proteins has an additional process that concentrates specific proteins into secretory granules. Protein hormones, made in large amounts in neuroendocrine cells, are stored in large dense-core vesicles packed with protein, commonly known as secretory granules. Cells retain the granules until stimulated, when they release the contents through exocytosis, so that large amounts of hormones may be rapidly available when needed. Concentration of hormones in granules is extensive; PRL, for example, is 200 times more concentrated in the dense cores of secretory granules than in the lumen of the endoplasmic reticulum (1). How cells concentrate hormones is a major unanswered question in endocrinology.
Although the question is unanswered, models have been proposed based on the possibility of a sorting signal to direct or retain proteins in secretory granules and on the possibility that aggregation of hormones serves both a concentration and a sorting function. Mechanisms are beginning to be understood for sorting of some transmembrane and soluble proteins in the secretory pathway, but in the case of proteins that are concentrated, it is not even clear whether there is any sorting process other than aggregation. Many reviews have been written about sorting of hormones into granules (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20). This review focuses on mechanisms by which soluble protein hormones are concentrated into secretory granules. It differs from previous reviews in that two points are emphasized: 1) new findings and new interpretations of transport through the secretory pathway affect conclusions about sorting into secretory granules; and 2) there is unexpected specificity in handling of secretory granule cargo proteins.
|
II. A Well Characterized Sorting Mechanism: Lysosomal Hydrolases |
|---|
|
TopAbstractI. Introduction: Concentrating...II. A Well Characterized...III. Sorting of Soluble...IV. Two Models for...V. Models for Formation...VI. Hormone Aggregates in...VII. Hormone Aggregates in...VIII. Does Aggregation Cause...IX. What Keeps Secretory...X. Sorting in the...XI. Sorting Signals and...XII. All Secretory Granule...XIII. SummaryReferences |
|---|
) (21, 22). The Golgi complex is a
broad, multilayered ribbon of enclosed membrane stacks (23). Proteins
in the secretory pathway proceed from the cis to the
trans side of this complex, and, in the outermost layer
called the trans-Golgi network, proteins are sorted and
carried to different locations in the cell (15).
|
B. Specific transport is associated with specific vesicles
This example of sorting is also an example of the specificity of
vesicles involved in transport through the secretory pathway, because
clathrin-coated vesicles are only involved in certain phases of
transport, and additional vesicle classes have been characterized.
Clathrin-coated/AP-1 vesicles transport lysosomal hydrolases from the
trans-Golgi network to the endosomes; using a similar
protein, adaptor protein-2 (AP-2), clathrin-coated/AP-2 vesicles
transport proteins from plasma membrane to endosomes (22). Vesicles
with completely different coats designated COPII (coat protein complex
II) transport proteins from the endoplasmic reticulum to the region
where the cis-Golgi layer forms, and vesicles with coats
designated COPI transport vesicles back from the Golgi complex to the
endoplasmic reticulum as well as through the Golgi cisternae (26, 27, 28, 29, 30).
In addition to those identified above, other kinds of vesicles have
been described emerging from the trans-Golgi network (15),
and sorting occurs for which coat proteins have not been identified;
therefore, this area continues to develop.
In neuroendocrine cells, clathrin-coated patches occur on parts of the trans-Golgi membrane where dense cores of secretory granules form and on parts of the membrane of immature secretory granules (31, 32). Immature granules contain lysosomal enzymes that are removed as granules mature (33, 34), and it seemed likely, therefore, that clathrin-coated vesicles remove lysosomal enzymes from immature secretory granules as well as from the trans-Golgi network. Arvan and co-workers have recently demonstrated that sorting of these enzymes from the immature granules uses the same system involving the mannose-6-phosphate receptor as elsewhere in the trans-Golgi network (34, 35).
The elucidation of the role of mannose-6-phosphate in the sorting of soluble proteins in the secretory pathway occurred early on (22) and was one of the factors that led to models in which sorting signals were predicted to be present on other proteins, including those destined for secretory granules. Sorting of proteins clearly occurs in the trans-Golgi network, which led to the assumption that all important sorting occurs there. Recent work has shown, however, that sorting linked to transport also occurs in the endoplasmic reticulum.
|
III. Sorting of Soluble Transported Proteins in the Endoplasmic Reticulum |
|---|
|
TopAbstractI. Introduction: Concentrating...II. A Well Characterized...III. Sorting of Soluble...IV. Two Models for...V. Models for Formation...VI. Hormone Aggregates in...VII. Hormone Aggregates in...VIII. Does Aggregation Cause...IX. What Keeps Secretory...X. Sorting in the...XI. Sorting Signals and...XII. All Secretory Granule...XIII. SummaryReferences |
|---|
The first evidence that sorting occurs came from Mizuno and Singer (39), who showed that the soluble secretory protein serum albumin is concentrated in discrete areas of the endoplasmic reticulum before transport to the cis face of the Golgi complex. Vesicular stomatitis virus glycoprotein, a membrane protein, is also concentrated in the course of transport from the endoplasmic reticulum (40).
A. Sorting in yeast
Further evidence that sorting occurs to select proteins for
transport out of the endoplasmic reticulum has been found in yeast, and
the mechanisms are beginning to be identified (28). So far, the
principle that emerged from sorting of lysosomal enzymes, that sorting
of soluble cargo proteins is linked to proteins involved in vesicle
formation, is emerging in this sorting as well, shown by biochemical
and genetic investigations. COPII vesicles carry proteins from the
endoplasmic reticulum, and Sar1, Sec23, and Sec24, three components of
yeast COPII, form a complex with proteins that are transported from the
endoplasmic reticulum, including amino acid permeases, which are
integral membrane proteins, and glycosylated pro-
-factor, which is a
soluble secreted protein (41). Proteins that remain in the endoplasmic
reticulum are not included in the complex, demonstrating specificity.
Binding of the soluble pro--factor inside the lumen to the coat
proteins on the cytoplasmic side of the membrane must be mediated by a
transmembrane adaptor protein not yet identified. There is likely to be
more than one adaptor component, since a consensus sequence among
transported proteins has not been found, and therefore more than one
sorting signal for leaving the endoplasmic reticulum is predicted.
Genetic evidence for more than one sorting system to carry proteins out
of the endoplasmic reticulum comes from yeast with
temperature-sensitive mutants of a COPI component, Sec21. Mutations in
Sec21 allow normal secretion of two proteins, invertase and HSP159, but
two others, carboxypeptidase Y and pro--factor, remain in the
endoplasmic reticulum. Carboxypeptidase Y-invertase hybrids, however,
leave the endoplasmic reticulum, suggesting there is a signal for
forward transport in the invertase molecule (42). COPI vesicles
transport proteins from the Golgi complex back to the endoplasmic
reticulum. The simplest explanation for why mutations in a retrograde
pathway affect transport forward through the secretory pathway is that
mutations in COPI components trap adaptors in the Golgi cisternae that
are unable to return to fulfill their function of causing forward
transport, because COPI vesicles recycle adaptor proteins from the
Golgi complex back to the endoplasmic reticulum. A candidate for these
adaptor proteins in yeast is the p24 family, because deletion of
one member of this family, Emp24, affects the transport of invertase
and not carboxypeptidase Y (43).
B. Sorting in mammalian cells
These investigations in yeast have shown there is active sorting
by more than one system for transport from the endoplasmic reticulum.
Recent evidence indicates such sorting also occurs in mammalian cells,
at least for some proteins. Transfer from the endoplasmic reticulum
occurs through irregularly shaped vesicles that emerge from the
endoplasmic reticulum and fuse in the cis-Golgi region (44).
This area has been called vesicular-tubular clusters or the endoplasmic
reticulum-Golgi intermediate compartment, usually abbreviated ERGIC
(45). Recent genetic studies indicate that ERGIC-53, a protein
found primarily in this location, is involved in selective transport of
certain proteins in people. Combined deficiency of coagulation factors
V and VII is an autosomal recessive bleeding disorder resulting from
alterations to a single gene, ERGIC-53 (46). In addition to its
location in the cell, ERGIC-53 has other characteristics that might be
expected of a protein involved in sorting: it binds mannose and
associates with the COPII component, Sec23 (47). Since COPII vesicles
carry proteins forward from the endoplasmic reticulum, the association
of COPII components with ERGIC-53 is further evidence for a link
between vesicle formation and sorting.
Although ERGIC-53 is necessary for normal secretion of coagulation factors V and VII, what is perhaps surprising is that there are no obvious deficiencies in other secreted proteins in people in whom ERGIC-53 is mutated, suggesting that ERGIC-53 is necessary for a limited set of proteins. The deficiency of the two coagulation factors, however, is not complete, for they are present in the serum at reduced levels (46). Nichols and co-workers (46) suggest that partial deficiencies in the transport of other proteins may exist but may not be as clinically apparent.
C. Implications for sorting into granules
What difference does it make to investigations of sorting into
secretory granules if there is some sorting that occurs in the
endoplasmic reticulum? One immediate difference is that interpretation
of experiments to detect sorting signals becomes more complex. It has
been known for some time that proteins must be folded correctly to
leave the endoplasmic reticulum and proceed further along the secretory
pathway (36, 48), and it has seemed reasonable that mutations in
secreted proteins that resulted in their retention in the endoplasmic
reticulum did so by disrupting folding. There is now an alternate
interpretation: that a sorting signal necessary for transport out of
the endoplasmic reticulum may have been affected. Prosequences of
secretory proteins have been shown to facilitate transport from the
endoplasmic reticulum (49). Such facilitation could be caused because
the sequence increases the rate of correct folding, allows the
formation of oligomers, or reacts with adaptors to increase active
transport from the endoplasmic reticulum.
If all proteins that are transported from the endoplasmic reticulum remain mixed in common vesicles as they proceed forward through the Golgi complex, even if more than one set of adaptor proteins is involved, then what happens before the Golgi complex may not be important in so far as secretory granule sorting is concerned. Much sorting occurs on the trans side of the Golgi complex in the vesiculating region known as the trans-Golgi network; if all proteins are in common vesicles, they will still be mixed when they reach this point. There is, however, one piece of evidence that suggests some sorting may occur earlier and that all proteins are not in the same trans-Golgi network. Ladinsky and co-workers (50) developed three-dimensional reconstructions of trans-Golgi regions of rat kidney NRK cells, using high-voltage electron microscopy. They found that individual regions of the trans-Golgi network with no tubular connections to each other each produced vesicles of only one type, either clathrin-coated or a type not previously described with a lacey coat (50). From what has been learned so far about vesicular transport, different coats imply different cargos. The authors predicted from this observation that some sorting of proteins occurs before the trans-Golgi network. What gets sorted, and whether there is any relation to the sorting that occurs in the endoplasmic reticulum is unknown at present.
|
IV. Two Models for the Functioning of the Golgi Complex |
|---|
|
TopAbstractI. Introduction: Concentrating...II. A Well Characterized...III. Sorting of Soluble...IV. Two Models for...V. Models for Formation...VI. Hormone Aggregates in...VII. Hormone Aggregates in...VIII. Does Aggregation Cause...IX. What Keeps Secretory...X. Sorting in the...XI. Sorting Signals and...XII. All Secretory Granule...XIII. SummaryReferences |
|---|
). Each layer of the Golgi complex is a
fixed compartment containing different sets of resident enzymes;
secretory proteins pass through each distinct layer. A second model,
less frequently described, has been called the directed maturation
model (45). In this model, the Golgi stacks are made of progressively
maturing, discontinuous compartments with no forward transport by
vesicles (Fig. 2). Instead, a layer of the Golgi complex forms at the
cis side by fusion of vesicles from the endoplasmic
reticulum, and enzymes that process the proteins are brought by COPI
vesicles to the most recently formed layer back from layers established
earlier. The layer itself progresses from the cis to the
trans side of the complex as new layers form behind it, and
as the ones ahead vesiculate and disappear in turn in the final sorting
process. Implicit in the model is the consumption of each layer in the
final sorting. Bannykh and Balch (45) discuss how this model fits in
well with current experimental findings. Among the evidence for this
model are the results of investigators who examined the Golgi complex
by three-dimensional electron microscopy and concluded that the
trans-Golgi layer is completely consumed by the final
budding of vesicles or by fusing to the plasma membrane in certain cell
types (53, 54).
|
|
V. Models for Formation of Secretory Granules |
|---|
|
TopAbstractI. Introduction: Concentrating...II. A Well Characterized...III. Sorting of Soluble...IV. Two Models for...V. Models for Formation...VI. Hormone Aggregates in...VII. Hormone Aggregates in...VIII. Does Aggregation Cause...IX. What Keeps Secretory...X. Sorting in the...XI. Sorting Signals and...XII. All Secretory Granule...XIII. SummaryReferences |
|---|
A different vision of the formation of secretory granules comes from
the work of Rambourg and co-workers (53), who examined PRL-producing
cells of lactating rats by three-dimensional electron microscopy. PRL
is found in aggregates in the lumen of the trans-Golgi
cisternae; these lumps of PRL are in the plane of the layer of
trans-Golgi complex, scattered throughout (Fig. 3) and not in a sequestered region
projecting above the layer. The morphology is consistent with this
trans-layer beginning to peel away from the Golgi stack as
small vesicles form from it, some clathrin-coated and some not,
removing membrane that is not directly around the aggregated PRL. As
these vesicles are removed, the remaining membrane of the layer, no
longer an uninterrupted surface, takes on the tubular-vesicular
appearance of the trans-Golgi network. What has been termed
immature granules are, in this view, those aggregates in which removal
of membrane by clathrin-coated and other vesicles is not complete (Fig. 3). What has been interpreted as immature granules that have fused to
each other may be two or more protein aggregates that were drawn near
each other as the trans-Golgi layer is consumed (Fig. 3),
although this does not rule out fusion of separate vesicles as well.
Further removal of membrane eventually leads to mature granules.
Membrane removal has been known to occur from secretory granules for
some time (59); the difference in models is whether secretory granules
have budded first. In the more traditional view, secretory granules bud
from the trans-Golgi network; in this view, everything else
buds off, leaving the secretory granules, and the difference between
the trans-Golgi network and immature secretory granules is a
quantitative one, depending on the amount of membrane removal, rather
than a qualitative one.
|
B. Implications for sorting into granules
If everything buds off, leaving membrane containing aggregated
protein inside, what are the consequences for models of secretory
granule formation and sorting into secretory granules? One consequence
is that there is no need for a separate budding mechanism for secretory
granules from the trans-Golgi network. Coat proteins play a
role in formation of vesicles, but not in delivery of vesicles to their
destinations; other proteins in and associated with the membranes of
vesicles direct fusion to the correct sites, and vesicles must lose
their coats to fuse (26, 60). Secretory granules will still need to
obtain the proteins necessary for regulated delivery to the plasma
membrane, but will not need to obtain separate proteins for budding.
The same mechanisms that other cells use for constitutive transport to
lysosomes and plasma membrane and for recovery of proteins that belong
in the Golgi complex will be all that is needed as long as the
components of these pathways exist in sufficient quantity to remove
excess membrane from secretory granules. Several groups have attempted
to determine the requirements necessary for secretory granule formation
in cell-free systems (61, 62, 63, 64, 65). What they may actually be detecting in
these reconstituted systems are requirements for budding of other
vesicles, and what remains is the granules.
A second consequence of a model in which aggregation before vesicle
budding occurs is the potential for aggregation to be an excellent
mechanism for sorting in and of itself. Some vesicle types, such as
those coated with clathrin, are small and, when forming, could enclose
soluble proteins or aggregates of small size, but not the large
aggregates of PRL that have formed in the trans-Golgi lumen
before the vesiculation of that Golgi layer begins (Fig. 3
). The effect
of preexisting large aggregates may be that proteins that are not
aggregated will be carried off by other vesicles as they form, leaving
the proteins that have aggregated in the resulting secretory granules.
If the directed maturation model of Golgi transport is correct, the
aggregate need not form all at once in the trans-Golgi
layer, but may begin to form as the protein hormone enters the Golgi
complex, developing in size until it reaches the large size detected in
the trans-Golgi lumen.
Finally, it is possible that aggregation is not just a sorting mechanism, but may be, in addition, an important factor driving formation of secretory granules. Such a function will exist if the membrane proteins that do not aggregate, but that must localize in granules for correct function, recognize some aspect of the aggregate to remain in the membrane that becomes part of the granules. This possibility is discussed further in later sections. An important point is that aggregates alone are not sufficient to form secretory granules. Fibronectin, which aggregates easily, forms membrane-enclosed aggregates in AtT20 cells, but these vesicles do not behave as secretory granules, and regulated release of fibronectin can not be induced (66). There must be some form of recognition between the contents of the granule and the membrane proteins necessary for functioning of the granule.
|
VI. Hormone Aggregates in Cells |
|---|
|
TopAbstractI. Introduction: Concentrating...II. A Well Characterized...III. Sorting of Soluble...IV. Two Models for...V. Models for Formation...VI. Hormone Aggregates in...VII. Hormone Aggregates in...VIII. Does Aggregation Cause...IX. What Keeps Secretory...X. Sorting in the...XI. Sorting Signals and...XII. All Secretory Granule...XIII. SummaryReferences |
|---|
The location where the aggregates first start to form is important, because it is the location of PRL aggregates relatively early in the sorting process in the trans-Golgi lumen that raises the possibilities that aggregation is an effective sorting mechanism and is a factor in the formation of granules. Rat PRL is not a complex molecule; the signal sequence that is necessary for its import into the endoplasmic reticulum is rapidly removed, and after that, no further processing or addition is necessary for storage (68).
Other hormones, however, do require processing. Insulin traverses the secretory pathway as proinsulin until it is cleaved into insulin and C-peptide, a process that occurs in immature granules (31, 55). Rat insulin does become insoluble in granules in a crystalline form that sediments rapidly when the contents of the granules are isolated, and, in these conditions, C-peptide is soluble (69). The crystalline form, however, does not occur in all species [e.g., guinea pigs package insulin into secretory granules whose contents remain homogeneous, rather than separating into a compact insulin core with a lucent halo of C-peptide, as rat and mouse insulins do (70, 71)]. Mice that have deficient or mutated processing enzymes do not process proinsulin to insulin well, but package proinsulin into granules that resemble morphologically those of the guinea pig (72, 73). Therefore, the crystalline form of insulin is not necessary for concentrated storage in secretory granules.
The crystalline form of insulin occurs relatively late in the secretory pathway in immature granules and is not analogous to the aggregates of PRL found before extensive formation of vesicles occurs from the trans-Golgi cisternae. Concentrated masses of proinsulin are detected by electron microscopy in the trans-Golgi network in forms that look similar to aggregates of PRL (31, 55). It has not been shown by means independent of electron microscopy that these masses of proinsulin are some kind of aggregate. If aggregation of proinsulin functions as a sorting mechanism, the aggregate must be porous enough or reverse easily enough to allow processing enzymes to function rapidly. Although proteases eventually chew up crystals of proteins, the processing of insulin occurs relatively rapidly, and, if there is an aggregate, it is simplest to imagine cleavage to insulin occurring if the aggregate is loose, as a kind of colloid or gel.
ACTH is produced by processing the prohormone, POMC. Dense masses of POMC are also detected in the trans-Golgi network by electron microscopy (57, 74), but, again, there is no separate evidence that an aggregate of POMC forms. Because processing of POMC, proinsulin, and other precursors is relatively rapid once it begins, it will be difficult to try to isolate aggregates of prohormones from intact cells. Use of mice or cell lines deficient in processing enzymes may allow a better determination of what the physical states of prohormones are in cells.
B. Properties of hormone aggregates in cells
Apart from the demonstrated existence of some hormone aggregates,
the properties of hormone aggregates in the secretory pathway are
largely unknown. One property that may be deduced from electron
microscopy is that aggregation is a specific process. Somatomammotrophs
of the bovine pituitary gland produce both PRL and GH, two proteins of
similar size with structural homology to each other. In these bovine
cells, the two proteins are frequently found in separate granules or in
separate aggregates in the same granules (75, 76). The separation
occurs because these proteins form distinct aggregates even when they
are present in the same trans-Golgi lumen. Another example
of separate aggregation occurs in bag cells of Aplysia
californica. The precursor form of egg-laying hormone is cleaved
in the trans-Golgi lumen into a C-terminal portion and an
N-terminal portion. These are packaged into separate granules, and the
separation appears to occur by aggregation, since separate dense masses
of the N-terminal part and C-terminal part are found in the same
trans-Golgi lumen (77).
In addition to specificity, a second property may be deduced: hormone
in aggregates in granules must remain in a conformation so that it can
dissociate reversibly and relatively rapidly after exocytosis. Changes
in conformations of proteins can trigger aggregation (78). In an
extreme example, denaturing proteins by boiling exposes hydrophobic
surfaces that encourage essentially irreversible intermolecular binding
in an aqueous environment. Changes in conformation in proteins in
vivo are also responsible for some types of aggregation; for
example, the normal cellular form of the prion protein is mostly
-helix with no ß-sheets, but in infected cells the protein has
more than 30% ß-sheet structure and forms aggregates (79). Formation
of ß-sheets is involved in aggregation of other proteins as well (78, 80), but these aggregates do not dissociate easily and are unlikely to
be models for concentrating proteins into secretory granules. It
remains possible that a less drastic change in conformation of the
tertiary structure of the hormone occurs during aggregation and the
change may facilitate the process.
There have been suggestions that intermolecular disulfide bonds are involved in storage of PRL in secretory granules (81, 82), although such bonds would not allow rapid dissociation of PRL. Intermolecular disulfide bond exchange may occur between proteins in secretory granules retained in cells for extended periods and may occur during preparation of granules, especially if stored frozen, but are not essential for storage in granules because there are few intermolecular disulfide bonds in rat PRL from freshly prepared granules (83, 84).
Clearly, the environmental conditions in the part of the secretory pathway in which aggregates form will influence that process. In all cells examined, acidification of the lumen occurs in the trans-Golgi network, so that the pH decreases to about 6 there (85, 86). Ca2+ concentrations in the endoplasmic reticulum are high in all cells, since intracellular stores of Ca2+ for signaling are kept there (87). Other than these generalizations, there are only specific pieces of information about a few cell types. In PC12 cells, total Ca2+ in the endoplasmic reticulum, in most of the Golgi complex except for the trans-Golgi network, and in the granule dense cores is at least 10 mM (88). Insulin-containing granules have been isolated and have very high cation concentrations: in addition to 42 mM insulin, they contain 23 mM Zn2+, 120 mM Ca2+, and 42 mM Mg2+ (89). Other granules have not been characterized for divalent cations as quantitatively. Qualitatively, secretory granules and the Golgi complex in the rat anterior pituitary gland stain for Zn2+, especially in somatotrophs, corticotrophs, and thyrotrophs (90). All granules in any one cell do not stain; whether this heterogeneity is an artifact or whether granules are heterogeneous is not resolved. More complete knowledge of divalent cation concentrations and other conditions all along the secretory pathway will be useful in understanding how and when aggregation occurs.
|
VII. Hormone Aggregates in Solution |
|---|
|
TopAbstractI. Introduction: Concentrating...II. A Well Characterized...III. Sorting of Soluble...IV. Two Models for...V. Models for Formation...VI. Hormone Aggregates in...VII. Hormone Aggregates in...VIII. Does Aggregation Cause...IX. What Keeps Secretory...X. Sorting in the...XI. Sorting Signals and...XII. All Secretory Granule...XIII. SummaryReferences |
|---|
The investigations using turbidity or precipitation as measures of aggregation have demonstrated that secretory granule proteins aggregate in the presence of Ca2+, mildly acidic pH, or both, conditions that are likely to exist in the trans-Golgi cisternae and in immature secretory granules (91, 92, 93, 94, 95, 96). There is some specificity, in that proteins not found in secretory granules do not aggregate under these conditions (92, 93, 94). The limitations of these techniques mean that there could be no characterization of the aggregates formed, and so the physical basis for the aggregation has not been elucidated.
Investigations of aggregation of secretory granule proteins under conditions in which they remain in solution have demonstrated these proteins form dimers or oligomers under certain conditions. Proinsulin and insulin form dimers and hexamers that include Zn2+ and Ca2+ (97); human GH will form dimers (98) and human PRL will form oligomers (99) in the presence of Zn2+; and chromogranins A and B will form dimers and, at lower pH, tetramers, in the absence of divalent cations (100, 101). Formation of these oligomers appears to be building blocks for higher molecular weight forms in cells, but the demonstration that they exist in solution does not prove the same forms exist in cells.
One clear-cut finding relative to sorting has emerged from investigations of aggregation in solution, i.e., a single specific sequence in proteins may cause formation of large aggregates. Bovine GH will aggregate so much that it precipitates under conditions in which human GH will not, and a discrete, transferable sequence causes this aggregation, because interchanging 8 amino acids in a 24-amino acid stretch transfers the ability to aggregate to the human hormone (102, 103, 104). Although this aggregation occurs under conditions unlikely to have physiological relevance, the investigations demonstrate that a protein sequence that causes large aggregates may be as small and specific as sequences associated with sorting signals. Therefore, the modification of sequences that affect sorting do not in themselves discriminate between a sorting signal and a sequence that affects aggregation.
|
VIII. Does Aggregation Cause Sorting in Cells? |
|---|
|
TopAbstractI. Introduction: Concentrating...II. A Well Characterized...III. Sorting of Soluble...IV. Two Models for...V. Models for Formation...VI. Hormone Aggregates in...VII. Hormone Aggregates in...VIII. Does Aggregation Cause...IX. What Keeps Secretory...X. Sorting in the...XI. Sorting Signals and...XII. All Secretory Granule...XIII. SummaryReferences |
|---|
Constitutive secretion is caused by release from vesicles not subject to regulated exocytosis and includes 1) release of soluble proteins carried in vesicles from the Golgi complex directly to the plasma membrane, and 2) release of soluble proteins carried first to the endosomes and then to the plasma membrane (106). Hormone secretion through the pathway through the endosomes, described for proinsulin, is called constitutive-like secretion (107).
Basal release of hormone in the absence of stimulation has sometimes been assumed to be caused only by constitutive secretion from vesicles that are not secretory granules, but it need not be. Spontaneous PRL release from lactotrophs, for example, can be almost completely inhibited by addition of dopamine (108), indicating that this basal release is primarily through the regulated pathway of secretory granules, not through a constitutive one, in these cells. There is also little hormone release from insulin-producing cells except through granules. Insulin is present only in granules, so insulin release is a marker for release from granules (69). Perifused pancreatic rat islets release about 5–6% of the total insulin immunoreactivity as proinsulin (109, 110). The amount of proinsulin in rat secretory granules is 2% of the total in the crystalline cores and 8% in the soluble fraction of the granule (69). Proinsulin in the crystalline cores is a minimal estimate of the amount present in granules, because proinsulin is more soluble than insulin; therefore, more may be in the soluble fraction of the granules. Proinsulin in the soluble fractions is an upper estimate because immature granules may have contaminated the granule preparation. At most, however, only a few percent of release of insulin immunoreactivity occurs through nongranular pathways, because most of the proinsulin release is accounted for by what is in the granules. Unlike tumor cells, the primary source of basal release from normal cells or secretory tissue is release of secretory granules.
Cell lines that grow continuously in culture produce less hormone than primary tissue, usually by at least 1 order of magnitude. They seem likely to have more membrane traffic and therefore more chance to remove hormone to constitutive pathways. For these reasons, perhaps, it has been suggested that pituitary cell lines may not be appropriate for measuring concentration into secretory granules (111). Basal secretion from a PRL and GH-producing cell line, GH3, may be inhibited up to 80% by somatostatin (112), indicating most secretion comes from a regulated pathway. The ratio of stimulated PRL release to basal PRL release and the patterns of stimulated release are quite similar when PRL release from primary cultures of lactotrophs and from GH4C1 cells, a subclone of GH3 cells, were compared, indicating that the cell line behaves as normal cells in culture (113). Pituitary, adrenal, and some insulin-producing cell lines used as models store hormones concentrated in dense core granules (57, 58, 114, 115). Therefore, although cell lines do not usually make as much hormone as primary cultures do, they treat the hormone they do make in the same way normal cells do, indicating that cell lines are appropriate models for investigating how proteins are concentrated into secretory granules.
B. Does sorting in cells correlate with aggregation in
solution?
Experiments to address whether aggregation is important for
sorting are based on expression of mutant proteins in cells. To make
conclusions about the importance of aggregation, a comparison of
results in cells aggregation in solution should be made. Since
at present there is not definitive evidence that aggregates in solution
reflect what occurs in cells, it is not surprising results from these
investigations are somewhat mixed.
1. Insulin. One type of familial hyperproinsulinemia is associated with the mutation, HB10D (histidine at the 10th position in the B chain of insulin mutated to aspartic acid) (116, 117). The histidine that is mutated is a ligand for Zn2+ in insulin and proinsulin hexamers, and the mutation reduces the ability to form hexamers (116). The presence of high levels of proinsulin in the blood is consistent with enhanced secretion through constitutive pathways, since processing to insulin occurs in immature secretory granules (85). This mutant was investigated further in transgenic mice expressing the human HB10D mutant and in AtT20 cells transfected with proinsulin. After islets from the mice received a pulse of radioactive amino acids, newly synthesized human proinsulin was released more rapidly than mouse insulin and proinsulin; the human hormone that remained in the cells was processed to insulin (116). These data are consistent with a mutant proinsulin that aggregates less well being sorted into secretory granules less well. In AtT20 cells, the mutant proinsulin was also rapidly released, and its release could not be stimulated (117).
Quinn et al. (118) also examined packaging of proinsulin mutants into AtT20 cells by immunogold electron microscopy, selecting two mutants, HB10D and SB9D, that both affect aggregation in solution. HB10D allows formation of dimers, but not hexamers, and SB9D prevents formation of either (119). Each mutant was concentrated into granules, although HB10D was less effectively concentrated than SB9D or wild-type insulin, which is consistent with the rapid release found in the studies mentioned above. All forms, however, were still effectively concentrated compared with the density of staining in the lumen of the Golgi cisternae (118). One interpretation is that aggregation is not a sorting mechanism or necessary for concentration, and the authors proposed that concentration occurs because proinsulin is brought to granules by a sorting signal. An alternate interpretation is that the conditions used to measure self-association of insulin in solution do not mimic the conditions that cause aggregation in cells.
Support for the second interpretation comes from the work of Grodsky and colleagues (120). They used a double mutant of insulin, SB9D and TB27E, which is monomeric in the assay conditions usually used, but, as they pointed out, does aggregate in high Ca2+ concentrations (120). These workers made transgenic mice expressing either human wild-type insulin or SB9D,TB27E-insulin; the mice made about half human insulin and half endogenous mouse insulin. They examined basal and stimulated release of insulin immunoreactivity from perfused pancreas preparations of these mice. Release of insulin immunoreactivity was normal in preparations from mice expressing wild-type human insulin. They found no basal release in the absence of Ca2+, and glucose stimulated release and somatostatin inhibited release in the presence of Ca2+, characteristics of insulin release from secretory granules. Release of insulin immunoreactivity showed similar patterns in preparations from mice expressing the mutant hormone in the presence of Ca2+, but differences appeared in the absence of Ca2+. Basal release of insulin immunoreactivity was elevated in preparations with the mutant, and there was no further stimulation of release by glucose and no inhibition of release with somatostatin. These experiments did not distinguish whether it is aggregation of insulin or proinsulin that may be important. The mutations will be expected to interfere with self-association of both forms, and the authors did not measure whether the insulin immunoreactivity was in the form of proinsulin or insulin. The simplest interpretation of these results is that lack of aggregation in the absence of Ca2+ causes insulin immunoreactivity to be diverted to and released from a constitutive pathway, rather than being packaged into secretory granules. The implication of this conclusion is that aggregation properties in certain, but not all, conditions in solution may be relevant to what occurs in the cell, and the studies of insulin in high Ca2+ concentrations are the ones that support aggregation as having a role in cells.
The above experiments showed that Ca2+ had to be removed to see SB9D,TB27E-insulin immunoreactivity diverted to a constitutive pathway but did not have to be removed to see a diversion of HB10D-proinsulin. The explanation of Carroll and co-workers (116) for why HB10D-proinsulin is diverted to a constitutive pathway is an interesting explanation for the different behavior of the mutants. HB10D-insulin binds with higher affinity to the insulin receptor than wild-type insulin, and the authors propose it is actively diverted to other pathways from secretory granules by binding to the insulin receptor. Further evidence for this proposal was obtained by using an additional mutation to make VA3L,HB10D-insulin, which has reduced affinity for the insulin receptor by about 2 orders of magnitude (121). This mutant was no longer diverted into alternate pathways. These results by themselves suggest that the default pathway for protein hormones may be secretory granules (121). Combined with the evidence for a role of aggregation in the results of Ma et al. (120), they raise the possibility that a loose aggregate of proinsulin forms at high Ca2+ concentrations to get proper localization in secretory granules, but the interactions are sufficiently weak so that higher affinity interactions such as binding to the receptor may divert some proinsulin to constitutive pathways.
2. Chromogranins. Chromogranins are made and stored in secretory granules of many neuroendocrine cell types (122, 123, 124). Chromogranin B and secretogranin II are synthesized by the adrenal cell line PC12, and this line has been used to investigate sorting and storage of these proteins. Chromogranin B, but not secretogranin II, has an intermolecular disulfide bond. Adding dithiothreitol to PC12 cells reduces the disulfide bond in chromogranin B and results in rapid secretion of chromogranin B from the cells (125). Heparin sulfate proteoglycan, a protein of the extracellular matrix, is constitutively secreted by the cells from small vesicles, and the kinetics of secretion of the reduced form of chromogranin B as well as its distribution in subcellular fractions from sucrose gradients resemble that of heparin sulfate proteoglycan, rather than that of proteins in dense core granules, all consistent with the diversion of the reduced form of chromogranin B to a constitutive pathway. The effect of dithiothreitol is not a nonspecific effect on the formation of secretory granules, because the ability of secretogranin II, which does not have a disulfide bond, to be stored in dense granule fractions of cells does not change when cells are treated with dithiothreitol (125).
More recent results have used PC12 cells transfected with a virus expressing chromogranin B, or an N-terminal deletion form of secretogranin B that lacks the sequence with the disulfide bond, or both. The N-terminal deletion mutant is not stored in these cells, when expressed alone, but is secreted via a constitutive pathway. When coexpressed with the full-length protein, the N-terminal deletion form is stored. The authors suggest that the full-length form aggregates and the N-terminal deletion form does not, and when the two are coexpressed, the full-length form causes the deletion form to aggregate (126). An alternate explanation is that secretory granules only form properly when there is a good aggregate around which to form, but when they do form, they become the default pathway.
The aggregation properties of chromogranins have been extensively examined to determine whether there is correlation with what occurs in cells. Chromogranin B forms precipitable aggregates in cells with permeabilized membranes in buffer with high Ca2+ concentrations and acidic pH (95). This aggregation property of chromogranin B does not change when it is reduced with dithiothreitol, an action that drastically affects its ability to localize in secretory granules (127). There is better correspondence with properties in the cell and in solution when formation of oligomers is measured by gel filtration; reduction of the disulfide bond in chromogranin B reduces the ability to make soluble dimers and tetramers (101). These gel filtration assays were performed in the presence of EDTA with no divalent cations, conditions unlikely to resemble what occurs in the trans-Golgi lumen; thus far, it is only investigations in solution in the absence of Ca2+ that support a role for aggregation for chromogranins. The authors suggest the rate of aggregation may be a crucial factor to explain this discrepancy (101).
3. PRL. A series of experiments designed to investigate the role of Zn2+ in human PRL storage also investigated the role of aggregation. Human PRL and human GH bind Zn2+ with similar affinity (98, 99). Zn2+ binding causes human GH to form dimers, which were predicted to be a storage form for GH (98). Human H27A-PRL, a mutant PRL with greatly reduced Zn2+ binding, was not stable and was secreted inefficiently when expressed in clones of transfected GH4C1 cells, unlike wild-type human PRL, suggesting that Zn2+ binding at the site involving H27 does play a role in PRL packaging and storage. Unexpectedly, the aggregation properties in solution of wild-type and H27A-PRL were similar: both aggregated in the presence of Zn2+, assessed by dynamic light scatter and sedimentation equilibrium analysis (99), which indicated that lower affinity Zn2+ binding sites than the one involving the H27 site cause aggregation in solution. It is possible that there are differences in the kinds of aggregates of wild-type and H27A-PRL formed in the presence of Zn2+ that were not resolved by these assays for self-association, or that the conditions in which aggregation was assessed were too far removed from conditions in cells to reflect differences that occur there. As in the first two examples, however, this work also shows a discrepancy between some assays of aggregation in solution and biological effects in cells.
The discrepancies do not rule out a role for aggregation because, as indicated above, most methods of self-association analysis require that proteins be in solution, and conditions to keep proteins soluble may avoid the conditions that cause aggregation in cells. Although it is not satisfying that there is incomplete correlation between biological effects on storage and self-association of proteins in solution, these technical difficulties may preclude obtaining a complete correlation even if aggregation is the chief sorting method.
|
IX. What Keeps Secretory Proteins from Aggregating? |
|---|
|
TopAbstractI. Introduction: Concentrating...II. A Well Characterized...III. Sorting of Soluble...IV. Two Models for...V. Models for Formation...VI. Hormone Aggregates in...VII. Hormone Aggregates in...VIII. Does Aggregation Cause...IX. What Keeps Secretory...X. Sorting in the...XI. Sorting Signals and...XII. All Secretory Granule...XIII. SummaryReferences |
|---|
PRL will precipitate at low pH, even in the absence of cations (94). The pH is lower in the trans-Golgi region than in the earlier part of the secretory pathway in neuroendocrine cells, but a drop in pH in this region is not unique to cells that form secretory granules; the pH decreases in other cell types as well (86, 129, 130). PRL is made by the human decidua, but the cells do not make secretory granules, nor is there any evidence that PRL forms insoluble aggregates in these cells (131, 132). Decidual tissue contains 20 to 100 times less PRL per mg protein than pituitary tissue; the intracellular amounts reflect both synthesis and storage. Estimates of synthesis based on the amount secreted compared with intracellular amounts suggest decidual tissue probably synthesizes 10- to 100-fold less PRL per mg tissue than pituitary tissue (131). The GH4C1 pituitary cell line synthesizes about 100-fold less PRL than normal lactotrophs in culture (113) but still concentrates PRL in aggregates in dense-core secretory granules (114). At this time, we do not know why PRL aggregates only in specific locations in lactotrophs and not in decidual cells, but it seems likely the answer is more than a question of the difference in amounts of PRL made or differences in pH in the trans-Golgi network.
|
X. Sorting in the trans-Golgi Network |
|---|
|
TopAbstractI. Introduction: Concentrating...II. A Well Characterized...III. Sorting of Soluble...IV. Two Models for...V. Models for Formation...VI. Hormone Aggregates in...VII. Hormone Aggregates in...VIII. Does Aggregation Cause...IX. What Keeps Secretory...X. Sorting in the...XI. Sorting Signals and...XII. All Secretory Granule...XIII. SummaryReferences |
|---|
The principles for sorting transmembrane proteins are being elucidated. One mechanism involves the length of the transmembrane segment (136). Cholesterol concentrations are low in the endoplasmic reticulum and increase across the Golgi complex, resulting in a thicker membrane in the trans-Golgi region. Proteins destined for the plasma membrane have longer transmembrane segments than those that remain in the Golgi complex, suggesting that plasma membrane proteins can segregate into thicker patches of membrane when Golgi proteins can not. Since cholesterol-rich and cholesterol-poor domains exist in the same membrane bilayer (136), these domains provide an efficient way of sorting some proteins.
A second mechanism depends on amino acid motifs in cytoplasmic sequences of the transmembrane proteins (15, 111). For example, a tyrosine or double leucine motif in this region may allow transport of the protein that contains it to the basolateral membrane (15, 111). Analyzing signals of membrane proteins is complex because the effect of different sorting signals for apical and basolateral membranes may vary with different cell types (137). In addition, many proteins have more than one sorting motif on the cytoplasmic side; for example, a plasma membrane protein with a motif to allow endocytosis to endosomes may also have a motif for basolateral sorting to allow it to return to the plasma membrane. To add to the complexity, there is a third mechanism: recognition of luminal signals. Destroying the cytoplasmic basolateral sorting motifs on transmembrane proteins causes them to be transported to the apical surface if they have N-linked carbohydrates; those without carbohydrates remain in the Golgi complex (137). O-linked glycosylation sites may also be a signal for apical sorting (138).
In addition to all these sorting mechanisms, there must be some sorting of transmembrane proteins to secretory granule membranes in cells with secretory granules. Some membrane proteins are necessary for regulated exocytosis, such as synaptotagmin and synaptobrevin, which are found both in secretory granule membranes and in the membranes of small vesicles found in neuroendocrine cells, termed synaptic-like microvesicles (105). More interesting, perhaps, are those that localize only in secretory granules. A class of transmembrane proteins that resemble protein tyrosine phosphatase receptors, ICA512 (or IA2) and phogrin, have a location primarily on secretory granules and not in the synaptic-like microvesicles (139, 140, 141, 142, 143). GH4C1 cells, rat pituitary cells, can be induced to contain more secretory granules; such induction stabilizes ICA512 (144), indicating that the stability of this protein is linked to the kind of membrane traffic that occurs.
Other proteins that must go to secretory granules are the processing
enzymes that cleave precursors into hormones. Most of these are not
transmembrane and could be imagined to end up in secretory granules
because of binding to their substrates. Peptidylglycine -amidating
monooxygenase (PAM) amidates the C-terminal end of processed peptides
and can be found in transmembrane and soluble forms, resulting from
either cleavage of the transmembrane product or from alternate mRNA
splicing (145, 146). AtT20 cells, pituitary tumor cells, normally have
very low levels of PAM. Stable clones expressing more of the soluble
form store it in a regulated pathway; clones expressing the
transmembrane form accumulate PAM in the trans-Golgi region
and store less POMC and ACTH (147, 148, 149). These results suggest that PAM
plays more of a role in the cells than merely processing peptides. PAM
appears to have a strong signal on the cytoplasmic side for
localization to the trans-Golgi area, and expressing PAM at
what is a relatively high level for AtT20 cells must interfere with
normal maturation of secretory granules to reduce ACTH storage, perhaps
by binding factors that would normally be used for such a function.
Other proteins to which the cytoplasmic side of PAM binds are currently
being identified (150, 151). Ultimately, understanding how
transmembrane proteins localize to secretory granules and, more
unexpectedly, affect hormone storage should give important information
about the formation and subsequent transport of these granules.
B. Soluble proteins
Soluble proteins are subject to at least some of the same sorting
that affects transmembrane proteins in the lumen. GH is secreted from
both sides of Madin-Darby canine kidney cells unless it is
glycosylated, in which case it is only secreted from the apical
side (152), consistent with apical localization of
transmembrane-glycosylated proteins. Sorting of soluble proteins will
also be affected by proteins to which they bind, as work with the
insulin mutants and the insulin receptor suggests (121). Where will
soluble proteins go if they have no sorting signal?
Investigations of nonglycosylated GH secretion in the kidney cells suggest that soluble proteins m
