J Mol Cell Cardio122,

771-786

Arrested

(1990)

of Atrial

Exocytosis

Secretory

Granules

T. M. Newman and N. J. Severs* Department of Cardiac Medicine,

The National Heart and Lung Institute, Dovehouse Street, London S W3 6LT, UK

(Received 8 December 1989, accepted in revised form 2 February 1990) N. J. SEVERS. Arrested Exocytosis of Atria1 Secretory Granules. journal of Moleculat and (1990) 22, 771-786. Release of atria1 natriuretic peptide (ANP) from atria1 muscle cells is thought to occur by exocytosis of secretory granules, as in other secretory systems. However, in the atria1 myocyte, exocytosis has previously proved difficult to detect ultrastructurally. In order to study the mechanism of ANP release and related events, we have applied two separate techniques-tannic acid perfusion and ultrarapid freezing-specificially designed to arrest exocytosis. An increased number of fusion sites was found after ultrathin sectioning and freeze-fracture of tannic acid-treated atria enabling a hypothetical release sequence to be constructed. Detail of granule substructure was also obtained, suggesting the presence of a coat to the secretory granule core. Ultrarapid freezing, followed by freeze-fracture and freeze substitution, has confirmed aspects of the proposed exocytotic sequence, suggesting that this release is not due to the application of tannic acid, and these techniques also produced further evidence for the existence of the granule-core coat. Coated pits and vesicles were also found in large numbers in tannic acid-treated atria and interactions between coated vesicles and secretory granules were visualized. The possible role of coated vesicles in an exocytotic/endocytotic membrane retrieval pathway is discussed.

T. M. NEWMAN Cellular

KEY

AND

Cardiology

WORDS:

Freeze-fracture;

Atria1 secretory granule; Atria1 Freeze substitution; Rat.

natriuretic

Introduction Apart from their contractile function, atria1 muscle cells are specialized in the manufacture and secretion of a peptide hormone, atria1 natriuretic peptide. This hormone helps regulate blood pressure and volume and the excretion of sodium, potassium and water. It does this by direct and indirect effects on the kidneys and adrenal glands, control centers in the brain and on the walls of the blood vessels themselves. Although the physiological and molecular properties of atria1 natriuretic peptide have been extensively studied (for reviews see Cantin and Genest, 1985; Maack et al., 1985; Ackerman, 1986), the mechanisms that regulate its release from the cell remain to be clarified (Ferrari and Agnoletti, 1989). Two principal stimuli that initiate peptide release have been found. These are distension of the atria1 wall (Dietz, 1984; Lang et al., 1984; Agnoletti et al., 1987) and the activation of /I-receptors (see Ferrari and Agnoletti, 1989). The sequence of events after stimulation is unclear, but, in common with other * To whom 0022-2828/90/070771

correspondence

should

+ 16 $03.00/O

peptide;

Exocytosis;

Tannic

acid; Ultrarapid

freezing;

peptide-secreting cells, is believed to result in exocytosis of peptide-containing secretory granules. However, until recently (Page et al., 1986; Skepper, 1989), ultrastructural evidence for exocytosis in myocytes eluded detection, and this led to the suggestion of an alternative mechanism for the release of granule contents (Theron et al., 1978). Even though exocytosis has now been demonstrated, the numbers of fusion sites reported remains low and little is known about the dynamics of the event. Reasons that may account for the difficulty in capturing exocytotic events in myocytes include their infrequent occurrence, the transient nature of the event, and the difficulty in preserving membrane components by chemical fixation when, at the instant of fusion, they adopt highly unstable molecular arrangements. Techniques that arrest exocytosis at, or shortly after, fusion of the secretory granule membrane with the plasma membrane would greatly facilitate analysis of the cellular events involved in atria1 natriuretic peptide release.

be addressed. 0 1990 Academic

Press Limited

772

T. M. Newman

Two techniqueshave been usedsuccessfullyto achieve this in other secretory cells. Exposure of cells to tannic acid has been reported to arrest exocytosis of granules whilst allowing continued cell function and the possibility of further granule fusion (Buma et al., 1984)) thus resulting in an increase of ultrastructurallyidentifiable exocytotic sites (see Morris and Pow, 1988 for review). A second technique, ultrarapid freezing, provides a method of physically fixing even very rapid events, such asexocytosis (seePlattner and Knoll, 1987). It has been used in studies of exocytosis in a variety of secretory cells, e.g. oocytes (Chandler and Heuser, 1979), mast cells (Chandler and Heuser, 1980), chroma5n cells (Schmidt et al., 1983) and thrombocytes (Morgenstern et al., 1987). We report here the first application of these two techniques to the study of atria1 peptide secretion, with preparation for both ultrathin section and freeze-fracture electron microscopy, and illustrate how their complementary usage can add to our understanding of the mechanismof atria1 peptide release. Materials

and Methods

Adult Sprague-Dawley rats (225 to 300 g) were used. The animals were maintained on standard pelleted diets and tap water, available ad libitum. Tannic acid perfusion

Anaesthetized rats were perfused with oxygenated Tyrodes Ringer (pH 7.4) containing heparin via the posterior vena cava. The animals were heparinized after cannulation, perfused for 15 min and the blood drained from the severed end of the vena cava. The buffer was then switched over to Tyrodes containing 1% tannic acid (BDH; mol wt, 1701kDa; pH 7.4) for a further 25 min, before 10 min perfusion with 2.5% glutaraldehyde in Tyrodes (pH 7.4). Solutions were gravity fed from 500 ml flasks, held at between 2 and 11 cm above the animal, at a temperature of 36°C. After perfusion fixation for 10 min, the hearts were removed and sampleswere immersed for a further 110 min in the same glutaraldehyde fixative solution at room tem-

and N. J. Severs

perature. Controls were prepared with 40 min Tyrodes perfusion prior to fixation. The dissectedatria were then prepared for electron microscopy as follows: Freeze-fracture

Dissected tissue pieces were infiltrated with 30% glycerol in Tyrodes Ringer (pH 7.4) for 3 h at room temperature, mounted on Balzers gold stubs and then frozen by immersion in freezing Freon. Freeze-fracture was carried out in a Balzers BAF 400T Unit at a temperature of - 105°C and a vacuum of better than 4 x IO-’ mbar. Standard platinum-carbon replicas were prepared and cleaned in chromic acid. Thin sectioning

Atria1 samples were post-fixed in buffered osmium tetroxide for 2 h at room temperature, dehydrated to 50% ethanol and en bloc stained with uranyl acetate for 30 min. Ethanol dehydration was then completed and the samplesembedded in Emix epoxy resin. Silver sections were cut and counter stained with uranyl acetate and lead citrate. Ultrarapidfreezing

Young adult rats (maximum weight 260 g) were used because their hearts have thin epicardial layers. This helps maximize the amount of underlying muscle that can be frozen without ice crystal damage. The rats were killed by dislocation of the neck, and dissected to remove the left atria. Each left atrium was mounted whole on a portion of fresh lung attached, with polyvinyl alcohol, to a piece of filter paper superglued to a thin mica strip. This was attached via double sided adhesive tape to a piece of adhesive draught excluder on a freezing stage of a Reichert Cryoblock helium-cooled copper block freezer. The assemblywas then transferred to the Cryoblock and the atrium frozen. The time taken from removal of the atrium to freezing was < 150 s. Right atria were also frozen after having been dissected and incubated in oxygenated Tyrodes during the period necessary for freezing the left atrium and resetting the Cryoblock (c. 25 min). After freezing, the

Atrial

Grmtulc

mica strips with the frozen atria were removed from the assemblyunder liquid nitrogen. The atria were then prepared by either freezefracture or freeze substitution. Freeze-fracture

Individual strips were clamped onto a modified Balzers specimen table, with the wellfrozen surface of the atrium uppermost. The atria were fractured by a superficial sweepof the microtome blade through the shallow well-frozen zone. Shadowing and replication were carried out as above. Freeze substitution

Freeze substitution was carried out in the cold table of a Balzers Spray Freezing Unit, which has temperature control over the range -50 to - 150°C. The substitution fluid (2% osmium tetroxide/0.5°/o uranyl acetate in methanol) wascooled to - 100°C in a 3-ml plugged glass vial, inside a protective polypropylene jacket containing methanol. The mica strips and frozen atria were transferred to the vial in a fume cupboard, replaced in the cold table which was then set for -90°C. Substituion times were; -90°C (1 h), warmed to -70°C ( 1 h), warmed to - 60°C (2 h), warmed to

773

Esecytosis

- 50°C (1 h), warmed to - 20°C (1 h). The samples were then transferred to 100% ethanol at - 20°C (two washes),and allowed to warm to room temperature. After two more washesin lOOo/oethanol they were embedded in Emix epoxy resin. Replicas and sections from all experiments were examined on a Philips 301 microscope equipped with a tilt stage.

Results In ultrathin sectionsof control atria, convincing examplesof secretory granules undergoing exocytosis are seldom observed. After exposure of atria1 muscle cells to tannic acid, however, exocytotic profiles, representing granules in the act of discharging secretory product, are easily found (Fig, 1). The granule’s limiting membrane is clearly resolved as a typical trilaminar “unit membrane” (Fig. 2), continuous, in some profiles, with the myocyte plasma membrane [Fig. 2(b)-(d)]. Such “omega” profiles demonstrate that fusion between the granule membrane and the plasma membrane has occurred, and the presenceof intact granule cores demonstrates that it is atria1 peptide release that is being visualized, rather than any other process.

FIGURE 1. Survey thin section of tannic acid-treated atria1 cell. Atrial secretory granules fuse with the plasma membrane. E, endothelium; F, fibroblast; G, granules; M, myocyte.

x

have been arrested 53000.

as they

774

T.M.N

ewman

and N. J. Severs

FIGURE 2. Exocytotic profiles of atria1 secretory granules in tannic acid-perfused atrium. The plate is arranged to show the possible sequence of exocytosis and ANP release. (a) A granule lying at the myocyte surface. The granule and plasma membranes are in such close apposition that they are difficult to distinguish from one another. (b) Fusion has occurred and distinct trilaminar continuity is found between granule and plasma membranes. (c) The pore formed upon fusion reaches a size appropriate for the release of the whole core. (d) The granule membrane flattens as it is incorporated into the plasma membrane. The intact granule appears to deform against the fixed glycocalyx. Note also that there is distinct electrondense material (arrowed) apparently linking the granule coreand theinternal surface ofits membrane. (a) x 12OooO; (b) x 108000; (c) x 15OooO; (d) x 150000.

At&l

Granule

Granules appear to be arrested after different degrees of incorporation of their membrane into the plasma membrane, so that a range ofwidths (30 to 350 nm) to the resulting opening to the exterior is observed (Fig. 2). This enables a hypothetical sequence to be constructed that may reflect the progressive stagesof fusion as they occur in the releaseof an individual granule’s contents [Fig.

2(aH41.

The earliest detectable stage in exocytosis appears to be represented by a granule that lies in such close contact with the plasma membrane that, over a distance of c. 30 nm, the two membranes apparently merge [Fig. 2(a)]. At this stage, there is no ultrastructurally apparent communication between the extracellular space and granule interior. After this initial event, lipid bilayer continuity between the plasma membrane and granule membrane is establishedaround the periphery of the initial interaction site, creating an opening [Fig. 2(b)]. The smallestopening found so far is of 30 nm; the absenceof smaller pores may indicate that the initial lipid reorganization along the region of apposition may be very rapid and/or involve very labile intermediates. The opening progressively widens [Fig. 2(c)], and as it does so, the profile, initially omega-shaped,is transformed into a depression in the plasma membrane [Fig. 2(d)]. At this final stage, the size of the opening is actually larger than the diameter of the

Jhcytosis

775

original granule [Fig. 2(d)]. The granule contents, though now present on the external surface of the plasma membrane, are still readily identified as an intact, undispersed electron-dense mass beneath the glycocalyx. The granule contents appear compressedin comparison to the spherical appearance they have in the cytoplasm [Fig. 2(d)]. In addition to the hypothetical sequenceas outlined above, other structures can be found that may be relevant to the later stages of releaseand the passageof the peptide through the extracellular space and across the capillary endothelium. These structures are dense aggregates of material, which appear ultrastructurally identical to granule cores, seenin the extracellular spaces of tannic acidperfused atria (Figs 3 and 4). In Figure 3, a presumptive granule core is surrounded by a diffuse band of electron-dense material resembling the glycocalyx, suggestingthat a portion of the glycocalyx might have become associated with the granule during release.The presence of a glycocalyx-like covering is not, however, an invariable feature of such externally located “granules”. In Figure 4, which depicts a presumptive granule lying at the adluminal surface of a capillary endothelial cell, no glycocalyx material is present. An intriguing feature of this granule is that its core is bounded by a thin dense line, giving the appearance of a unit membrane in close apposition to the core. That such a

FIGURE 3. A granule-like structure (arrowed) in the extracellular space in a tannic acid-perfused sample. This structure is associated with a halo of glycocalyx-like material. Comparison of this granule-like structure with the secretory granules in the myocyte above demonstrates the similarity between them. M, myocyte; G, granule. x 77 000.

776

FIGURE 4. A granule-like cell, in a tannic acid-treated (arrowheads). E, endothelium;

T. hi. Newman

and N. J. Severs

structure, attached to the plasma membrane sample. A thin line surrounds the granule large arrow, endothelial plasma membrane.

of the adluminal surface of an endothelial producing a “membrane-like” appearance x 106000.

“membrane-like” image should enclosea discharged granule (whose limiting membrane has already been lost by incorporation into the plasma membrane) is an unexpected observation, yet this feature is consistently found in a proportion of granules seenin all extracellular locations, including those that have only just undergone exocytosis (Fig. 5). This “membrane-like” structure appears to be thinner than both the granule membrane and the plasma membrane. Cytoplastic granules, however, characteristically have one membrane, not two (Fig. 6), although in some sectioning planes it is difficult to distinguish

even this membrane. Occasionally, gaps (Fig. 6), are seen in the limiting membrane of cytoplasmic granules, but no evidence of dispersion of core material through such gaps is apparent (cf. Theron et al., 1978). Freeze-fracture electron microscopy of tannic acid-perfused atria extends the picture of the dynamic events involved in secretory granule discharge suggestedby ultrathin sections. In traversing the frozen tissue,the plane of fracture produces a mixture of crossfractured cytoplasm and large sheetsof fractured plasma membrane, and it is thus possible to obtain a variety of views of the fusing

FIGURE 5. The membrane-like granule coat visualized in an exocytosing granule (arrowheads). The plasma membrane and outer granule membrane are continuous at this stage. Note also the material between the inner coat and outer membrane (arrowed). Tannic acid-treated sampie. x 177000.

FIGURE 6. Cytoplasmic secretory granule from a tannic acid treated sample illustrating the limiting membrane and electron-lucent halo. This halo contains connections between the membrane and the core (arrowheads), but no internal membrane-like coat is apparent. x 148 000.

Atrid

Ganulc

granules (Fig. 7). If the fracture plane deviates from the plasma membrane to enter the cytoplasm at a region of granule fusion, then it can follow the granule membrane, thereby demonstrating the membranous continuity between the two. In such a fracture [Fig. 7(a)] a distinct neck region is evident. The combination of fractures producing such imagesis rare. The most frequent plasma membrane fractures give uninterrupted planar views. In P-face views, protrusions can be found situated within dips in the membrane [Fig. 7(b)]. These protrusions have smooth surfaces in comparison with the surrounding P-face and they appear to represent the core material of granules that have fused asin Figures 2 (c) and (d). The fracture plane, instead of following the sharply-angled fused membrane, has travelled acrossthe opening, producing a view of the core as trapped by the glycocalyx. Figure 7(b) also illustrates a very rare cross-fractured granule, revealing detail of the granule contents. In favourable E-face views of the plasma membrane, the fusing granules can be viewed from underneath, appearing as bulb-like membrane-bound structures attached to the plasma membrane [Fig. 7(c)]. Portions of the granule membrane are sometimes removed by the fracture plane, revealing an underlying smooth layer [Fig. 7(c)] with a similar surface to the protrusions found in P-face fractures of the plasma membrane. Figure 7(d) illustrates a particularly extensive view of this smooth internal layer. The granule limiting membrane, seento be in direct continuity with the plasma membrane via the “fusion neck”, lies exterior to a curved “membrane-like” expanseenclosing, on either side, the cross-fractured granule contents. The smooth inner “membrane-like” layers are characterized by a lack of particles on both fracture faces; the granule-limiting membrane, by contrast, has numerous intramembrane particles of varying size which show the usual polarity on membrane splitting (i.e. appearing more abundant on the P-face than on the E-face). When the fracture plane leaves the smooth layer a distinct step is seensimilar to that shown by membranes [arrowheads Fig. 7(d)]. The smooth layer appears to correspond to the “membrane-like” boundary visible in somethin-sectioned granules.

Exocytosis

777

Resultsobtained from ultrarapid frozen untreated atria (Fig. 8) complement those obtained by tannic acid perfusion. That the freeze-fracture observations presented in Figure 7(b) are not a product of the glutaraldehyde and glycerol treatments is shown by the presenceof protrusions in P-face sarcolemmal fractures from ultrarapidly frozen atria [Fig. 8(a)]. Ultrarapid freezing followed by freezefracture confirms the neck-like opening and membrane continuity of the fused granule with the plasma membrane [Fig. 8(b)]. In ultrarapid frozen specimens,but not in chemically pre-treated atria, an abrupt transition at the edge of the fusion neck is evident. Intramembrane particles are scarce in this region [Fig. 8(b)]. Ultrarapid frozen atria, prepared by freeze substitution for ultrathin sectioning, illustrate similar granule structures to those of the tannic acid-treated cells [Fig. 8(c) and (d)]. The continuity between the membrane of the fused granule and plasma membrane [Fig. 8(c)] suggeststhat tannic acid is not creating exocytosis artifactually. Additionally, some cytoplasmic granules in freeze-substituted sampleshave a second“membrane-like” layer inside the limiting membrane [Fig. 8(d)]. Figure 9 illustrates an extracellular “granule-like structure” observed at the adluminal endothelial surface in a freeze substituted atrium (Fig. 9). This is bounded by a dark line which may correspond to that of the extracellular “granule-like structure” from the tannic acid-perfused sample in Figure 4. The range of granule structures found after ultrarapid freezing is similar to that found after tannic acid perfusion, The events, however, appear less frequent in the ultrarapid frozen material and it has not, so far, been possible to construct a hypothetical release sequencefrom ultrarapid frozen atria alone. In addition to spherical granules, “teardrop” shaped granules occur. Though in the minority, granules of this “tear drop” shape are consistently observed in thin section in both tannic acid-treated and in freeze substituted myocytes. The rounded tip of the “teardrop” differs from the remainder of the granule-limiting membrane by having a covering of electron dense spikes on its cytoplasmic surface (Figs 10 and 11). Such spiky coated membrane segmentsare conven-

FIGURE 7. Freeze-fractured secretory granules from atria1 samples that were glutaraldehyde-fixed and glycerol cryoprotected after tannic acid perfusion. (a) P-face of the membrane of a secretory granule that has fused with the plasma membrane. Large numbers of membrane particles are found on this face. Note that the area of the fusion neck has few particles. (h) Cross-fractured cytoplasm and P-face view of the plasma membrane showing a smooth-faced protrusion (*) representing the partially expelled core of a secretory granule. A rare cross-fractured secretory granule is also present near the plasma membrane. It has internal structure (arrowed) that may represent the material found in the halo in thin section (cJ Figs Z(d) and 61. (c) E-f ace view of the membrane of a fused secretory granule. The granule is caught at the transition of a fracture from the plasma membrane (E-face) of one cell to an underlying plasma membrane (P-face) of a second cell. The granule is fractured to show a smooth surface (*) which may represent the “E-face” of an inner limiting membrane as seen in thin section. (d) Cross-fractured fused granule with smooth inner surface coat (*) clearly seen to be enclosed by the particulate limiting membrane. Note that the area of the fusion neck of the latter (arrowed) has few particles. For further explanation see text. C, cytoplasm; G, cross-fractured granule core; coat. (a) x 97000; (b) x 88000; (c) arrowheads, transition between granule core and inner “membrane-like” :: 119000; (d) x 119000.

Atrial

Granule

Exocytosis

F ‘IGURE 8. Ultrarapid frozen (non-chemically treated) secretory granules. (a) Protrusion from the pla sma me* nbrane P-face (*) representing a partially expelled secretory core. (b) P-face of a fused secretory granule. Note the clea lrly delineated transition between granule and plasma membrane (arrowed). (c) Fused granule in thin section aifter free se substitution. The granule contents have not been expelled through the developing opening. (d) Cytoplas ,mic gra nules after freeze-substitution. Note that an inner “membrane-like” coat (arrowed) to some cores is found. (4 xl 15000; (b) x83000; (c) x 150000; (d) 75000.

780

T. M. Newman

FIGURE 9. Granule-like structure attached to the plasma membrane of the adluminal surface of the endothelium in a freeze substituted sample. E, endothelium; arrow, endothelial plasma membrane. x 190000.

tionally identified as clathrin-coated vesicles and pits. These images therefore appear to represent coated vesiclesthat are undergoing fusion with, or fission from, the secretory granule membrane. Such coated vesiclesecretory granule “hybrids” are found throughout the cell, from the Golgi apparatus (Fig. 10) to the plasma membrane (Fig. 11). The usual form of coated pit in the plasma membrane is frequently encountered in tannic acid-perfused atria (Fig. 12), and free-floating vesiclesin the cytoplasm are alsoseen. Di8cu6sion The present study demonstratesthat exposure of atria1 muscle cells to tannic acid greatly facilitates the dection of exocytosing secretory granules. This transient event in atria1 secretion has previously proved difficult to detect by electron microscopy (Page et al., 1986; Severs, 1989a; Agnoletti et al., 1989; Skepper, 1989). We show here that tannic acid treatment greatly amplifies the number of granules that can be captured in the act of exocytosis, and that furthermore, a variety of configurations, apparently reflecting different stagesin

and N. J. Severs

the processof membrane fusion and hormone extrusion, can be observed. The ability of tannic acid to reveal theseevents representsan important advance in analysing the stepsinvolved in atria1 secretory mechanism and in determining the pathway by which the peptide subsequently reachesthe plasma. The difficulty in capturing exocytosis in atria1 muscle cells by standard ultrastructural techniques may be attributed to a combination of factors. First, the quantity of peptide released,even under experimental conditions designedto stimulate atria1 secretion, requires fusion of relatively few granules (Agnoletti et al., 1989) compared with the massiveexocytosisthat can be triggered in someother secretory systems(Ornberg and Reese,1981). Second, the exocytotic event is itself very rapid, and may, moreover, be difficult to stabilize reliably with standard aldehyde fixation, The tannic acid incubation technique, recently developed and applied to the study of secretion in a number of invertebrate and vertebrate non-cardiac systems(Buma et al., 1984; Buma and Roubos, 1985; Buma and Nieuwenhuys, 1987; Golding and Pow, 1987; Pow and Golding, 1987), is believed to act by specifically arresting the exocytosis processat a stagejust after granule fusion, while at the same time preserving normal cell function to a sufficient degreethat granules can continue to approach and fuse with the plasma membrane (Buma et al., 1984). The net result viewed is thus interpreted as an accumulation of all the exocytosesoccurring during the tannic acid incubation period, held in a state of arrested fusion. The present study also demonstrates that secretory granule exocytoses can be captured by ultrarapid freezing of untreated atria. This provides ultrastructural information unobtainable by other techniques, as well asconfirmatory evidence for results obtained using tannic acid perfusion. The number of release sitesfound after ultrarapid freezing appears to be greater than in aldehyde fixed controls; this is however difficult to quantify, as only a small fraction of the myocardium is well preserved after freezing. It is possiblethat the mounting of atria for freezing is actually causing increasedreleasedue to unintentional stretching of the atria1 well. Freeze fixation may, however, simply be retaining more effectively the samenumber of fusion sitesas occurs in con-

Atrial

Granule

Exocytosis

781

FIGURES 10 and 11. Tear-drop secretory granules. These granules (arrowed) similar to coated vesicles. The secretory granule in 10 is located near the plasma membrane. Figure 10, x 90 000; Figure 11, x 119 000.

have protrusions with a clathrin-like coat near the Golgi apparatus; that in Figure 11

trol tissue,by circumventing two of the problems associated with chemical fixation; the arresting of rapid events and the retention of labile intermediaries. The number of fusion sites obtained does not, however, appear to equal thosefound after tannic acid treatment, followed by conventional chemical fixation. Enhancement of fusion sites by tannin acid

It is important to consider how tannic acid may be causing the increase in ultrastructurally identifiable fusion sites.Tannic acid is the general name given to a group of naturally occurring polymers of galhc acid and glucose, extracted from a variety of plant products. They are reported to have the ability to fix proteins (Mizuhura and Futaesaku, 1972; La Fountain et al., 1977; Burton et al., 1975; Mizuhira et al., 1981), lipids (Kalina and Pease, 1977; Mizuhira et al., 1981), and carbohydrates and glycosaminoglycans (Sannes et al., 1978; Singley and Solursch, 1980). HOWever, the large molecular mass(1701 kDa) of most commercially available tannic acids means that they cannot enter undamaged cells, and so, at least potentially, cytoplasmic

functions may continue with little impairment. Around the surface of the cell, the acid is free to interact with a variety of components. Important amongst these will be glycosaminoglycans, principally hyaluronic acid (Singley and Solursch, 1980), and the carbohydrate portions of integral membrane glycoproteins and glycolipids, that together constitute major components of the glycocalyx. The glycocalyx must allow the passageof the secretory product and the incorporation of new granule membrane into the plasma membrane and thus any alteration to it could have important consequenceson the secretory process.It is not known how exactly the peptide crossesthe glycocalyx and whether the process involves reorganization of glycocalyx components. Using freeze substitution in Limulus amoebocytes, it has been shown that there is, upon fusion, a rapid dispersionof the secretory product through the opening, rather than expulsion of the product as an intact granule core (Ornberg and Reese, 1981), but it was not possible to evaluate what occurs at the glycocalyx. Where tannic acid treatment has been used in the atria, and other secretory systems, the glycocalyx appears to be well

782

T. M. Newman

FIGURE 12. Coated pits (ax-rowed) are common and well defined in the plasma membranes of myocytes in tannic acid-incubated atria1 material. x 119 000.

fixed, and is presumably inflexible. The solid core of secretory product is thus trapped below it, and the plasma membrane may be prevented by the rigidity imposed by the fixed glycocalyx and the presence of the intact core, from readopting its usual post-fusion profile. Tannic acid is theoretically capable of fixing the peptide secretory product, as well as the glycocalyx, and it is thus difficult to decide the degree to which these two factors interplay to arrest exocytosis, i.e. whether the retention of the granule core is a secondary effect due to the prevention of granule membrane incorporation by fixation of the glycocalyx, or whether fixation of the core is preventing secretory product expulsion through the glycocalyx, thereby maintaining the membrane kink. It is probable that both factors are important. A further property of tannic acid-its ability to interact with cholesterol and the choline residues of phosphatidylcholine and sphingomyelin (Kalina and Pease, 1977)-suggests the possibility that direct stabilization of plasma membrane lipids, in addition to the action on the glycocalyx, may also be contributing to the arrest phenomenon. Any such

ad N. J. Severs stabilization would appear to be limited, however, as the capacity of the plasma membrane to undergo fusion with secretory granule membrane evidently remains unimpaired. The use of tannic acid in physiological media to arrest exocytosis but otherwise encourage retention of normal cellular functions is quite distinct in design from the more widespread and established practice of incorporating the acid into aldehyde fixatives (Mizuhira and Futaesaku, 1972). There is, nevertheless, an overlap in these applications. A major aim of the latter procedure-an improvement in the visibility of membranes-is also achieved in the absence of glutaraldehyde, and this undoubtedIy aids detection of exocytotic profiles in the former application. Indeed, arrest” approach was the “physiological developed after the recognition that in aldehydeltannic acid fixed specimens, exocytotic profiles were noticeably more common than with conventional aldehyde fixation; an effect attributable to the combined effects of improved resolution of membranes together with tannic acid-mediated exocytotic arrest prior to complete stabilization by aldehyde fixation (Buma et al., 1984). Ultimately, the chemical properties by which tannic acid intensifiesthe plasma membrane and glycocalyx for improved ultrastructural visualization involve the same or closely related reactions to those that, in the absence of aldehyde fixatives, mediate the arrest phenomenon. In view of the presiding role played by the plasma membrane and glycocalyx in controlling intracellular activities in the myocyte (Frank et al., 1977), this makesit unrealistic to hope that tannic acid exposure will leave all cytoplasmic functions entirely undisburbed for an indefinite period. The degree to which plasma membrane signaldetection and transduction mechanismsresponsible for exocytosis in the atria1 myocyte are affected by tannic acid remains an important question for the future. The exocytotic release sequence

Whatever the precise nature of tannic acid’s action, the ability to visualize exocytosis in action offers immediate opportunities for extending the understanding of the secretory processin atria1 myocytes. From the seriesof micrographs in Figure 2, the essential se-

Atrid

Granule

quence of events appears to involve (i) close approach of the granule to the plasma membrane, (ii) punctate apposition with formation of an incipient fusion site, (iii) fusion and pore formation, followed by (iv) progressive pore widening and finally, (v) emptying of the granule’s contents. In principal detail, this sequencematches that deduced in other secretory cells (e.g. Ornberg and Reese, 1981), although becauseof the action of tannic acid it is difficult to say at exactly what stage of pore formation the granule contents would normally be expelled. That tannic acid permits the detection of all these different stages may indicate that a slowing down rather than total arrest in the exocytotic/discharge process is induced. Alternatively, complete arrest may be effected at a more-or-lessuniform point in the process, the later stages observed representing granules that had progressedbeyond this point before tannic acid reached the cells. There is no indication that caveolae are involved in the fusion of the granules, as has been previously suggested (Page et al., 1986). All granule fusions observed have been directly with the plasma membrane. The events involved in the transport of atria1 peptide to the plasma after exocytosis are lessclear. Our observations on the presence of seemingly intact granule cores in extracellular locations, after freeze substitution, as well as after tannic acid treatment, raises the possibility that dispersal of the granule contents need not always occur. In some cases,extruded cores become wrapped in glycocalyx material, which may be buddedoff as the core crossesthe glycocalyx. How far these features represent the in vivo transport processremains unclear, but even if they reflect a modification brought about by the tannic acid perfusion, the ability of this treatment to disclose extracellular granule cores could neverthelessprove useful in tracing further details of the transport pathway. We have been able to confirm details of the sequenceof release,asobtained by tannic acid perfusion, using ultrarapid freezing of untreated tissue. Pore formation and the continuity of fusing granule and plasma membranes have been shown not to be features causedby tannic acid perfusion. Similarly, the presence of extracellularly located “granulelike” structures in freeze-substituted atria

783

Faocytosis

means that the presenceof these structures is not an artifactual effect of tannic acid. The abrupt boundary between the membrane of the fusing granule and plasma membrane, found after ultrarapid freezing and freezefracture, could indicate that the two membranes remain distinct during pore formation, perhaps not fully mixing until the core has been released. The earliest stages of fusion have been difficult to identity on the myocyte plasma membrane because of the numerous small caveolar openings present. Membrane particle clearance has been suggestedas being characteristic of the initial fusion stages in other systems (e.g. Poste and Papahadjopoulos, 1978), though this is generally thought to be a feature of chemically-fixed secretory systems, not observed when ultrarapid freezing is used (seeKnoll et al., 1988). In the present study, we were unable to characterize these very early stages of granule fusion with certainty, but it appears that an area poor in membrane particles does appear to form asfusion proceeds. Secretory granule structure

Although tannic acid would be expected to stabilize the core of the fusing granule as an intact mass, there is evidence in the atrium that the granule contents may naturally remain compact on expulsion, and that they do not disperse through the plasma membrane opening immediately it forms. By freezefracturing ultrarapidly frozen material, protrusions representing partially expelled core material, similar to those found after tannic acid treatment, are observed; undispersed cores do not, therefore, appear solely to be a tannic acid-induced effect. Moreover, freeze substitution fails to discloseevidence for dispersion of granule content through the fusion opening. An explanation for these observations may be provided by the membrane-like image seen enveloping discharged granules after both freeze substitution and tannic acid treatment. The presence of a membrane-like image bounding extruded cores after tannic acid treatment is puzzling. A trilaminar unit membrane-type image in thin section together with a fracture face-type image in freezefracture, would normally be considered firm

784

T. M. Newman

evidence for a truly membraneous identity. The smooth fracture face would suggest an absence of proteins, making this layer little more than a lipid bilayer. Although a second inner membrane is not found in conventionally prepared ultrathin-sectioned cytoplasmic granules, it might plausibly be supposed that the fixatives are unable to penetrate efficiently through the “outer” membrane to fix such an inner protein-free lipid bilayer. Freeze substitution, where the labile membrane components are kept frozen as chemical fixation proceeds, thus making fixation more effective, does reveal this membrane-like structure. Standard freeze-fracture has previously revealed “smooth membrane vesicles” in intracellular locations (Severs, 1989b) and these were also observed in ultrarapidly frozen atria during the present study. These structures could simply represent the occasional secretory grannule that is fractured along its “inner” rather than its “outer” membrane. Such a coating could function to keep the secretory product compact as it is transported to the endothelium, or it may, under normal circumstances, simply rupture as the concentrated product is exposed to the extracellular fluid. However, this line of interpretation poses a major conceptual obstacle in that it cannot easily be reconciled with the established mechanics of intracellular membrane interaction in secretory protein manufacture (Palade, 1975). Other possible explanations must therefore be considered. Tannic acid could, for example, produce a thin coat to the core by reacting with the surface and then mordanting osmium during post-fixation. Freeze substitution would have to be producing a similar effect by a different mechanism. Smooth faces within exocytosing atria1 granules, similar to those observed in the present study, are apparent in the published micrographs of Page et al. (1986). Where freeze-fracture studies in other secretory systemshave shown such structures they have been assumedto represent the surface of the secretory core (pancreatic /I cells, Orci et al., 1977; neurohypophysis, Theodosis et al., 1978). The latter authors also reported that the secretory core maintains its compact appearance in chemically-fixed tissuesafter it is liberated into the extracellular space. It is interesting to note that the “smooth faces”

and N. J. Severs and compact, extruded cores can be found occurring together in the granules of different secretory systems. Coated vesicle and secretory granule interactions

The occurrence of clathrin-coated pits and vesicles in cardiac myocytes is well known (Severs, 1989a), but interactions between these structures and atria1 secretory granule membranes have not previously been widely reported (Theron et al., 1978). In many cell types, coated pits form the sites at which receptors for specific ligands are displayed at the cell surface (Goldstein et al., 1985). Ligands such as protein hormones or nutrients bind to their receptors and are internalized by the formation of discrete clathrin-coated vesicles. The coated vesicles then transport the internalized proteins to various membranous organelles within the cell. Apart from this function in directed transport of proteins between membrane compartments, the coated pit/vesicle system can also operate as a mechanism for recycling membrane back from the cell surface to the interior following exocytosis. This function has been characterized in nerve cells, using ultrarapid freezing (Miller and Heuser, 1984), but in most secretory cells membrane retrieval after exocytosis is considered to be mediated by smooth-surfaced pinocytotic vesicles (Steinman et al., 1983). However, it has been shown that adrenal chromaffin granule membrane proteins are recycled after exocytosis (Suchard et al., 1981) and that there is a concurrent increase in coated pit and vesicle formation (Geisow et al., 1985). Using antibodies to proteins of the chromaffin granule membrane and immunogold labelling, Patzak and Winkler ( 1986) have shown direct evidence that after exocytosisof thesegranules there is selective retrieval of granule proteins via coated pits and coated vesicles, and that some of the proteins are incorporated into new granules. From these established functions of the coated vesicle, it is not possibleto deduce the precisesignificance of coated vesicle-secretory granule interactions in atria1 muscle cells, but some fascinating possibilities are raised. For example, the coated vesiclesmight be engaged in uptake of proteins, either from the exterior or from within another intracellular compart-

Atrid

Granule

ment, destined for deposition inside the lumen of the secretory granule. Alternatively, transport could be taking place in the reverse direction. Coated vesicles pinching off from the secretory granule membrane could operate in the selective removal of proteins or peptides; unwanted portions of peptide generated during post-Golgi processing of the hormone could be disposedof in this way, for example. On the other hand, it is entirely possible that what is being witnessed is a membrane recycling system similar to that suggested for chromaffin granules, whereby membrane components crucial to the exocytotic process are recovered and recycled via coated vesiclesfor direct reinsertion into the nascent secretory granule. Direct fusion between secretory granules and coated vesicles

Exocytosis

785

was not reported in the chromaffin cell; the recycling processhere is considered to involve the transreticular portion of the Golgi region (Patzak and Winkler, 1986). Further studies,combining tannic acid perfusion and ultrarapid freezing, may help clarify the exact role of coated vesiclesin the atria1 myocyte and provide further insights into the mechanism of granule exocytosis.

Acknowledgements

This work was supported by the British Heart Foundation (grant number 88/120). We greatly appreciate the help of Stephen Rothery in the preparation of the photographic plates.

References ACKERMAN U (1986) Structure and function of atria1 natriuretic peptides. Clin Chem 32: 241-247. ACNOLETTI G, FERRARI R, SLADE AM, SEVERS NJ, HARRIS P (1989) Stretch induced centrifugal movement of atria1 ~ specific granules-a preparatory step in atria1 natriuretic peptide secretion. J Mol Cell Cardiol21: 235239. ACNOLETTI G, RODELLA A, FERRARI R, HARRIS P (1987) Release of atria1 natriuretic peptide during stretching of the rat atrium in vitro. J Mol Cell Cardiol 19: 217-220. BUMA P, NIEUWENHUYS R (1987) Ultrastructural demonstration of oxytocin and vasopressin release sites in the neural lobe and median eminence of the rat by tannic acid and immunogold methods. Neurosci Letts fl: 151-157. BUMA P, ROUBOS EW, BUIJS RM (1984) Ultrastructural demonstration of exocytosis of neural, neuroendocrine, and endocrine secretions with an in uitro tannic acid (TARI) method. Histochemistry 80: 247-256. BUMA P, ROUBOS EW (1985) Ultrastructural demonstration of non-synaptic release sites in the central nervous system of the snail Lymnaea staph, the insect Periplanef americana and the rat. Neuroscience 17: 867-878. microtubules with 12, 13 and 15 protofilaments. J BURTON PR, HINKLEY RE, PIERSON GB (1975) T annic acid-stained Cell Bio165: 227-233. CANTIN M, &NEST J (1985) The heart and the atria1 natriuretic factor. Endocrine Rev 6: 107-127. CHANDLER DE, HEUSER J (1979) Membrane fusion during secretion. Cortical granule exocytosis in sea urchin eggs as studied by quick-freezing and freeze fracture. J Cell Biol83: 91-108. CHANDLER DE, HEUSER J (1980) Arrest of membrane fusion events in mast cells by quick-freezing. J Cell Biol 86: 666674. DXETZ JR (1984) Release of natriuretic factor from rat heart-lung preparations by atria1 distension. Am J Physiol247: Rl093-R1096. FERRARI R, AGNOLETTI G (1989) Atria1 natriuretic peptide: its mechanism of release from the atrium. Int J Cardio124: 137-149. FRANK JS, LANCER GA, NUDD LM, SERAYDARIAN K (1977) The myocardial cell surface, its histochemistry, and the effect sialic acid and calcium removal on its structure and cellular ionic exchange. Circ Res 41: 702-714. GEISOW MJ, CHILDS, J, BURGOYNE RD (1985) Cholinergic stimulation of chromaffin cells induces rapid coating of the plasma membrane. Eur J Cell Bio138: 51-56. GOLDING DW, Pow DV (1987) Neurosecretion by a classic cholinergic innervation apparatus. A comparative study in four vertebrate species (teleosts, anurans, mammals). Cell Tissue Res 219: 421-425. GOLDSTEIN JL, BROWN MS, ANDERSON RB, RUSSEL DW, SCHEIDER WJ (1985) Receptor-mediated endocytosis: concepts emerging from the LDL receptor system. Ann Rev Cell Biol 1: l-39. KALINA M, PEASE DC (1977) The preservation of ultrastructure in saturated phosphatidyl cholines by tannic acid in model systems and type two pneumocytes. J Cell Bio174: 726741. KNOLL G, BURGER KNJ, BRON R, VAN MEER G, VERKLEIJ AJ (1988) Fusion of liposomes with the plasma membrane of epithelial cells: fate of incorporated lipids as followed by freeze-fracture and autoradiography of plastic sections. J Cell Biol107: 25 1 l-252 1. LA FOUNTAIN JR, ZOBEL CR, THOMAS HR, GALBREATH C (1977) F ixation and staining of F-actin and microfilaments using tannic acid. J Ultrastruct Res 58: 78-86. LANG RE, THOLKEN H, GANTEN D, LUFT FG, RUSKOAHO H, UNGER T (1985) Atria1 natriuretic factor. A circulating hormone stimulated by volume loading. Nature 314: 264-266.

786

Atrial Gram&

Exttytosis

T, CAMARGO MJF, KLEINERT HD, LARAGH JH, ATLAS SA (1985) Atria1 natriuretic factor: structure and properties. Kidney Int 27: 607615. MILLER TM, HEUSER JE (1984) Endocytosis of synaptic vesicle membrane at the frog neuromuscular junction. J Cell Biol98: 685696. MIZUHIRA V, FUTAESAKU Y (1972) New fixation for biological membranes using tannic acid. Acta Histochem Cytochem 5: 233-235. MIZUHIRA AV, SHIHASI M, FUTAESAKU Y (1981) High speed electron microscope autoradiographic studies of diffusible compounds. J Histochem Cytochem 29: 143-160. MORGENSTERN E, NEUMANN K, S~WEKE H (1987) The exocytosis of human blood platelets. A fast freezing and freezesubstitution analysis. Eur J Cell Biol43: 273-282. MORRIS JR, Pow DV (1988) Capturing and quantifying the exocytotic event. J Exp Biol139z 81-103. ORCI L, PERRELET A, FRIEND DS ( 1977) F reeze-fracture of membrane fusions during exocytosis in pancreatic beta cells. J Cell Biol75: 23-30. ORNBERG RL, REESE TS (1981) Beginning of exocytosis captured by rapid-freezing of Limulus amoebocytes. J Cell Biol 90: 654. PAGE E, GOINGS GE, POWER B, UPSHAW-EARLEY J (1986) Ultrastructural features of atria1 peptide secretion. Am J Physio1251 (Heart Circ Physiol 20): H340-H348. PALADE GE (1975) Intracellular aspects of the process of protein synthesis. Science 189: 347-358. PATZAK A, WINKLER H (1986) Exocytotic exposure and recycling of membrane antigens of chromaffi granules: ultrastructural evaluation after immunolabeling. J Cell Biol 102: 51fk515. PLATTNER H, KNOLL G (1987) Ultrastructural analysis of cynamic cellular processes: A survey of current problems, pitfalls and perspectives. Scanning Micros 1: 1199-l 2 16. POSTE G, PAPAHADJOPOULOS D (1978) The influence of vesicle membrane properties on the interaction of lipid vesicles with cultured cells. Ann NY Acad Sci 308: 164184. Pow DV, GOLDING DW (1987) Neurosecretion by aminergic synaptic terminals in viuo. A study of secretory granule exocytosis in the corpus cardiacum of the flying locust. Neuroscience 22: 1145-I 149. SANNES LP, Knnu~m T, SPICER SS (1978) T annic acid-metal salt sequences for light and electron microscopic localization of complex carbohydrates. J Histochem Cytochem 26: 55-61. SCHMIDT W, PATZAK A, LINGG G, WINKLER H, PLATTNER H (1983) Membrane events in adrenal chromaffin cells during exocytosis: a freeze-fracture analysis after rapid cryofixation. Eur J Cell Biol32: 31-37. SEVERS NJ (1989a) Constituent Cells of the Heart and Isolated Cell Models in Cardiovascular Research. In: Zsolut‘ed Adult Cardiomyo~tcs, Structure and Function, Vol 1, HM Piper, G Isenberg (Eds) Boca Raton, Florida, CRC Press, pp 3-42. SEVERS NJ (1989b) Membrane cytochemistry of the atria1 granule secretory pathway. Am J Physiol 257 (Heart Circ Physio126): Hl587-H1598. SINGLEY CT, SOLUR~CH M (1980) The use of tannic acid for the ultrastructural visualization of hyaluronic acid. Histochemistry 65: 93-98. SKEPPER J (1989) A comparison of myocardial ultrastructure in the hamster (Mesocricetus auratus) with that of a typical non-hibernating mammal. PhD Thesis, Council of National Academic Awards, UK. STEXNMAN RM, MELLMAN IS, MULLER WA, COHN ZA (1983) Endocytosis and the recycling of plasma membrane. J Cell Biol %z l-27. SUCHARD SJ, CORCORAN JJ, PRESSMAN BC, RUBIN RW (198 1) Evidence for secretory granule membrane recycling in cultured adrenal chromaffin cells. Cell Biol Int Rep 5: 953-962. THEODOLUS DT, DREIFU~~ JJ, ORCI L (1978) A freeze-fracture study of membrane events during neurohypophysial secretion. J Cell Biol78: 542-553. THERONJ, BIAGIO R, MEYER AC, BOEKKOOI S (1978) Ultrastructural observations on the maturation and granules in atria1 myocardium. J Mol Cell Cardiol 10: 567-572. MAACK

functional