Yolk granule tethering: a role in cell resealing and identification of several protein components

Cell survival in vivo often depends on the capacity for rapid repair or resealing of a plasma membrane disruption. Skeletal (McNeil and Khakee, 1992) and cardiac muscle cells (Clarke et al., 1995) for example, commonly suffer disruptions during normal contractile, force-generating activities. Failure to reseal a plasma membrane disruption results in rapid (seconds to minutes) cell death, probably as a result of excessive entry of Ca²⁺ ions.

The mechanism of resealing large disruptions (>1 μm diameter) has been investigated extensively in the echinoderm egg. Resealing has an absolute requirement for external Ca²⁺ at near physiological concentration (Steinhardt et al., 1994). Ca²⁺ entering through the disruption locally triggers both homotypic (Terasaki et al., 1997) and heterotypic membrane fusion events (Steinhardt et al., 1994). The combined result of these fusion events is the application of a reparative `patch', derived from internal membrane, across the disruption site. The homotypic fusion events occur between yolk granules, and lead to the formation near to the disruption site of abnormally large intracellular vesicles, termed patch vesicles, that correspond in size to the disruption defect (Terasaki et al., 1997). Homotypic fusion of yolk granules can be reconstituted in vitro (Chestkov et al., 1998; McNeil et al., 2000), where it closely mimics the resealing-based process envisioned to occur in vivo: fusion is triggered by Ca²⁺, and strikingly, can produce extremely large vesicular products (>10 μm diameter) very rapidly. The heterotypic fusion events induced by a disruption occur between the patch vesicle and the plasma membrane, and are exocytotic in nature. These exocytotic fusion events join the product of homotypic fusion, the patch vesicle, to the native plasma membrane surrounding the disruption site, completing defect repair. A similar mechanism, involving homotypic and heterotypic fusion events, is thought to mediate resealing in mammalian cells (Bansal et al., 2003), but this process has not yet been reconstituted in vitro.

Two of the paradigmatic protein components in the fusion field have been implicated in resealing-based fusion. Botulinum or tetanus toxins, injected into fibroblasts or sea urchin eggs, inhibit resealing, suggesting a requirement for SNAREs (Steinhardt et al., 1994). Injection of function-neutralizing antibodies or competitive peptide fragments of synaptotagmin VII into fibroblasts (Reddy et al., 2001) or squid giant axons (Detrait et al., 2000a; Detrait et al., 2000b) partially inhibit resealing, suggesting a role for this possible Ca²⁺ sensor. However, a recent study casts doubt on the role of synaptotagmin VII (Jaiswal et al., 2004): it showed that, in fibroblasts at least, synaptotagmin VII actually acts as a brake on the resealing-based exocytotic event it was supposed to mediate. Moreover, another recent study (Shen et al., 2004) failed to confirm that the C2A domain of Syt VII inhibits resealing. The C2B domain, however, which does not block lysosomal exocytosis (Martinez et al., 2000), was effective in blocking resealing.

Does resealing-based fusion use any novel protein components, not so far identified in other systems? An answer to this question requires that one go beyond previously used approaches based on reasoning by analogy from other systems. A biochemical approach is one possible route to the discovery of resealing-specific components.

We previously reconstituted resealing-based yolk granule fusion in vitro, allowing us to identify biochemically protein fusion components and analyze their potential role in resealing. We here show that yolk granules isolated in a non-chaotropic buffer exhibit a potent tethering apparatus and that this tethering apparatus is required for the formation of large patch vesicles in vitro. This suggests a further biological role for tethering: the promotion of large products of fusion suitable for patching large disruptions. Moreover, we describe here methods that allowed us to dissociate tethering activity from isolated yolk granules and to biochemically enrich it. This enriched activity acts as a high molecular weight complex comprised of at least seven protein subunits.

Gravid starfish Asterias forbesi were obtained from the Marine Biological Laboratory (Woods Hole, MA). Excised ovaries were minced using scissors and frequently swirled, in Ca²⁺-free, ice-cold artificial seawater (Marine Biological Laboratory) and the eggs thus liberated descanted through a meshwork filter. One gravid starfish typically yields between 5-20 ml of wet-packed eggs. Eggs (10-20 wet packed ml) were washed three times (2 minutes, 500 g) in Ca²⁺-free seawater, and twice in glycine-based IM buffer (pH 6.7, 220 mM potassium glutamate, 500 mM glycine, 10 mM NaCl, 5 mM MgCl₂, 2 mM EGTA, protease inhibitor cocktail, EDTA-free; Roche), homogenized by 15-25 passes through a 40-ml Dounce-type glass tissue grinder (Kontes, NJ). Yolk granules were then purified either on a Percoll gradient, as previously described (McNeil et al., 2000), or by differential centrifugation (Vater and Jackson, 1989). No difference was observed between the behavior of granules isolated on the Percoll gradient and those isolated by differential centrifugation, though the latter procedure is a cheaper and faster method for isolation of large quantities of granules. Cytosol was obtained by taking the supernatant of the first high-speed spin used in the differential isolation approach, followed by a centrifugation step (60 minutes, 50,000 g) for removal of residual organelles and membrane fragments. Fusion was initiated in these isolated granules by the addition of calibrated Ca²⁺ buffers (McNeil et al., 2000), and either observed directly in the light microscope as the formation of vesicles greater in diameter than that of the native granule (∼1 μm), or assessed by a semi-quantitative, light scatter-based assay, as previously described (McNeil et al., 2000).

Yolk granules were centrifuged (10 minutes, 3200 g) to remove IM buffer and suspended for 30 minutes at room temperature in a KI-based buffer (pH 6.85; 470 mM KI, 5 mM MgCl₂, 2 mM EGTA, 10 mM PIPES, protease inhibitor cocktail, EDTA-free) to strip the granules of peripheral protein candidate tethering factors. These `KI-washed' yolk granules were pelleted by centrifugation and resuspended in IM buffer and saved for use in assays of tethering. The KI buffer harvested as the supernatant in this centrifugation step was saved as the starting point for further purification of protein candidates of the granule tethering apparatus.

KI-washed, tether-incompetent granules were vortexed for 10-15 seconds, and then a 5-μl volume (suspension OD₄₀₅ ∼0.5) was dispensed onto a glass slide. This was diluted with 20 μl IM buffer alone (negative control) or with 20 μl IM buffer containing test substances. Slides were immediately examined by transmitted light microscopy for the presence of tethered clusters.

KI-washed yolk granules (100 μl) with an OD₄₀₅ of ∼0.350 were mixed with 100 μl IM containing test substances, or with IM alone (negative control), in 96-well, round-bottomed plates. After a 10-minute incubation at room temperature, the plate was centrifuged (3 minutes, 100 g) in order to separate tethered granule aggregates (pellet) from non-tethered granules (supernatant). Finally, 75 μl of the top layer of each well, containing non-tethered granules, was transferred into a new well, and the absorbance of this transferred volume read on a plate reader (Cambridge Technology).

Proteins of interest were salted out of KI wash buffer by the addition of ammonium sulfate (2.4 M final concentration), and the precipitate redissolved in 10 mM sodium phosphate buffer, pH 6.8, preparatory to further separation. Column chromatography (FPLC, Pharmacia) was performed with a strong anion exchanger (HQ, BioRad) and a size exclusion gel (Sephacryl S-300, Amersham). Protein was eluted from the anion exchanger with a linear 1 M NaCl gradient; size exclusion chromatography utilized 10 mM sodium phosphate buffer (pH 6.8). Fractions were analyzed by conventional SDS-PAGE followed by staining with Coomassie Blue; protein amounts were quantified using a modified Lowry assay for all relevant experiments (DC protein assay, BioRad). Prior to all bioassays of tethering activity, protein fractions were desalted in columns (BioRad) into IM.

Granules were lysed by a 1:10 dilution in distilled water, and a supernatant harvested after centrifugation (60 minutes, 50,000 g). This lysate was then desalted into IM buffer prior to assay for tethering activity.

Purified (anion exchange fraction) tethering complex was used as an antigen for the production of monoclonal antibodies. Clones were selected for expansion based on the production of antibodies that stained intact (not detergent-permeabilized) IM-washed granules more strongly than KI-washed granules. For this purpose, flow cytofluorometry was used to evaluate relative fluorescence levels. Western blotting was done using standard laboratory procedures. All monoclonal antibodies were used at 1:1000 dilution and detected using goat anti-mouse antibody conjugated to HRP (Jackson ImmunoResearch Laboratories). Visualization of HRP was accomplished using an ECL kit (Amersham).

Eggs were fixed in 3.7% formalin/Ca²⁺-free seawater and frozen sections (10 μm) cut for immunostaining. Incubations in primary and secondary antibodies were for 60 minutes at room temperature at dilutions of 1:1000 and 1:200 respectively. Confocal images of the immunostained eggs were acquired on a confocal microscope (Zeiss, Germany).

In a previous study we observed that echinoderm egg organelles extruded from a microneedle into seawater without Ca²⁺, adhere to one another as they leave the tip, forming a strand-like array. In contrast, if Ca²⁺ was present, the organelles fused forming an elongate vesicular structure (Terasaki et al., 1997). A similar phenomenon has been observed in broken echinoderm eggs dating back to the classical experiments of Heilbrunn: breakage in Ca²⁺ elicits a rapid fusion reaction among organelles at the breakage site (termed the surface precipitation reaction), whereas breakage in the absence of Ca²⁺ results in an outflow of organelles that rapidly adhere to one another to form a large extracellular mass (Heilbrunn, 1956). This raised the possibility that aggregation or tethering of egg organelles, in particular the yolk granule, was an important step in the formation of membrane patches for use in the repair of plasma membrane disruptions.

To investigate this possibility further, we developed conditions for isolating yolk granules that retained this tethering capacity. We found that yolk granules isolated in a glycine-based, non-chaotropic buffer (IM) were strongly tethered to one another: they formed clusters consisting of many hundreds of granules (Fig. 1A). Upon Ca²⁺ buffer addition, fusion products of a size proportionate to the tethered sub-population were observed to form rapidly, in less than 2 seconds (the limit of resolution of the video detection method) (Fig. 1B-D). In contrast to previous studies in which the granules were isolated in a KCl-based buffer (Chestkov et al., 1998; McNeil et al., 2000), fusion to form large products did not require that the granules be forced into contact with one another by a centrifugation step prior to Ca²⁺ buffer addition.

As described in previous studies (Chestkov et al., 1998; McNeil et al., 2000), Ca²⁺-triggered granule fusion is a reaction that is initiated rapidly (second to sub-second time scale) and fusion, measured as a decline in light scatter in those studies, was observed to cease within 1 minute of Ca²⁺ buffer addition. To confirm that in our system, in which granules become spontaneously tethered to one another, is a similar temporal limit to fusion, we continued incubations of granules for over an hour. After the initial rapid decline in light scatter (completed before the plate could be introduced into the plate reader, <3 minutes) induced by Ca²⁺ addition, no further change in light scatter was observed for the next 65 minutes (Fig. 2). Neither addition of further Ca²⁺, even that sufficient to produce fusion in granules previously exposed only to IM buffer, nor forcing of vesicles into contact with one another by centrifugation, resulted in additional fusion (data not shown), confirming that neither sequestration of added Ca²⁺, nor fusion induced decreases in vesicle density and therefore contacts, explain the fusion block. Thus, fusion of yolk granules triggered by Ca²⁺ is a self-limiting process, severely constrained temporally: multiple rounds of fusion occurring over a minute-to-hour-long time scale, such as are often recorded in other systems (Merz and Wickner, 2004), are not permitted in the yolk granule system.

To confirm the importance of tethering in the yolk granule fusion response, we washed granules in a mildly chaotropic buffer (KI-, NaCl-, KCl-based, in order of decreasing effectiveness). This was found to be capable of completely abolishing tethering: compared to the highly clustered IM-washed granules (Fig. 3A), KI-washed granules were observed to be an approximately monodisperse population (Fig. 3B). This tethering failure in KI-washed granules was further demonstrable in a semi-quantitative assay for tethering (Fig. 3C). Importantly, this KI wash-induced failure to tether had a dramatic effect on the yolk granule fusion response: the size of the fusion products elicited by Ca²⁺ addition was dramatically reduced (Fig. 4A,C). Thus, tethering is required for the production of large fusion products.

To further demonstrate this tethering requirement, we attempted to reconstitute tethering in KI-washed granules. We found that cytosol (Fig. 4B,D) and the KI buffer used for vesicle stripping of tethering activity (Fig. 5) could both restore tethering and Ca²⁺-triggered production of enlarged fusion products. To demonstrate that the tethering activity present in the KI wash buffer was not simply a product of vesicle lysis, we compared the tethering activity of lysed granules with anion-exchange-purified tethering factor (see below). The lysis product had no demonstrable tethering activity (Fig. 6). These results not only confirm the importance of tethering for the production of large fusion products, they suggest that protein components with a tethering function are present in the cytosol and are associated with the granule membrane.

As a starting point for purification of protein `tethering factors' we used KI wash buffer, as it contains a far simpler mixture of proteins than cytosol. Tethering activity was precipitable with ammonium sulfate, allowing us to efficiently concentrate and desalt tether candidate proteins prior to anion exchange chromatography. Tethering activity was found to elute in a linear salt gradient as a single peak from the anion exchanger resin (Fig. 7A). Enrichment, based on the semi-quantitative assay for tethering, was about fivefold over the simplified starting material (Fig. 7C). The active peak consisted of seven major polypeptide bands, of ∼100, 105, 110, 155, 170, 215 and 225 kDa (Fig. 7B).

When the enriched fraction from the anion exchange column was subject to size exclusion chromatography, tethering activity emerged in a single peak running at the column void volume (∼670 kDa) (Fig. 8A). Gel electrophoresis of the material in this peak confirmed that the previously described complex of proteins was present in this peak (Fig. 8A). When subjected to non-denaturing electrophoresis, it ran as single band of ∼ 670 kDa (Fig. 8B). We conclude that the yolk granule tethering factor is a high molecular weight, multimeric complex of proteins.

Monoclonal antibodies were raised against the tethering factor and selected based on higher affinity staining of IM-compared to KI-washed (i.e. tether-competent compared to tether-incompetent) granules. All clones thus obtained were against the 170-kDa subunit (Fig. 9A). Many of these clones were capable of depleting tethering activity from KI wash buffer, confirming that one subunit of the complex at least, is required for granule tethering (Fig. 9B).

Monoclonal antibodies against the tethering complex labeled a ubiquitous and abundant organelle compartment of ∼1 μm diameter in the egg (Fig. 10A), strongly suggesting labeling of yolk granules. That the yolk granules and not other major egg organelles were labeled was confirmed by centrifuging eggs prior to staining, which stratifies the organelle compartments: only the heavy end of the stratified egg, in which the predominant organelle is the yolk granule, were stained with these monoclonal antibodies (data not shown). When the yolk granules were imaged at high resolution, apparent peripheral staining of granules was observed (Fig. 10B). Moreover, staining was observed in the absence of detergent permeabilization of eggs sheared to expose cytoplasmic constituents, and flow cytofluorometry analysis revealed no difference in the intensity of staining between detergent-permeabilized and intact yolk granules (data not shown). These results are all consistent with the antigen having the location expected of a cytosol-facing, peripheral membrane protein. The egg surface, possibly the plasma membrane, was also strongly stained with these monoclonal antibodies (Fig. 10C).

Resealing may employ several novel fusion protein components, specialized for this emergency-based cellular reaction (McNeil and Kirchhausen, 2005). Two recently described candidates in mammalian cells, dysferlin (Bansal et al., 2003) and annexin (Lennon et al., 2003), are both Ca²⁺-activated phospholipid-binding proteins. However, the exact roles of dysferlin and/or annexin in resealing remain unknown. Do these proteins participate in early events, such as tethering or docking of membranes, and/or as the Ca²⁺ sensor, and/or in the actual fusion or bilayer merger step? An answer to these questions requires that the various steps of a fusion process can be staged in vitro and the effects of manipulation of a protein at each stage observed. This paper reports the results of just such an approach using reconstituted yolk granule fusion as a model system to study an initial step – tethering – in a resealing-based fusion process.

Tethering is proposed to be an early, reversible event in numerous membrane fusion reactions (Whyte and Munro, 2002). Single, extremely large coiled-coil proteins, or multisubunit tethering complexes, are envisioned to bind to and physically span two membranes destined for fusion, restricting their ability to move apart. This forced physical interaction is hypothesized to promote, in turn, activity of downstream components of the fusion machinery, such as the SNAREs. The proximal nature of the tethering step in fusion suggests that tethers may confer specificity. Such a role is consistent with the known sequence diversity of tethers, and with their precise spatial localization to specific organelles (Waters and Hughson, 2000) and strikingly, to localized domains of the plasma membrane (Guo et al., 1999). Here we propose an additional role for membrane tethers: they are required for the production of large membrane fusion products, such as those utilized in a successful resealing response.

We have shown, by reconstituting yolk granule fusion in vitro and analyzing the staging of this reaction, that a Ca²⁺-independent tethering step precedes Ca²⁺-triggered granule fusion. Fusion clearly occurs preferentially between the members of a tethered granule population: large tethered clusters from large fusion products. We suggest that tethering in the egg is an essential step in the repair of large plasma membrane disruptions, which, as much work has demonstrated, are `patched' by proportionately large products of homotypic fusion.

It is conceivable that large fusion products could be produced in the absence of tethering by multiple rounds of fusion occurring sequentially over time. The behavior of yolk granules, as documented here and elsewhere, rules out this possibility. Granule fusion events triggered by Ca²⁺ are constrained to a narrow, second-scale time window. We propose that tethering of one yolk granule to another ensures that a fusion partner is nearby during this narrow time window: tethering thus acts kinetically to promote fusion in this system. The self-limiting nature of the Ca²⁺-induced fusion response observed in vitro would, in vivo, be of benefit as a means for limiting fusion events to those productive of repair.

This study is the first to identify protein components of resealing-based yolk granule homotypic fusion. From a highly simplified starting material, the approximately 15 major proteins stripped from the granule by a mildly chaotropic salt wash, a fraction enriched in tethering activity was produced by anion exchange chromatography. This fraction, which SDS electrophoresis resolves into seven major protein bands, behaves, under native conditions, as a high molecular weight complex based on analysis by size exclusion chromatography and non-denaturing gel electrophoresis. Immunodepletion of the complex inhibits tethering, confirming its role in this event in vitro. Immunostaining with a monoclonal antibody raised against one of the components of the tethering complex confirm that it associates strongly with the yolk granule, and not other major egg organelles. Yolk granules, and not other known organelles, are required for egg resealing of large disruptions (Terasaki et al., 1997). The yolk granule tether may, in addition to promoting large fusion product formation, also determine the organelle utilized in its formation.

We thank Mark Terasaki and Steve Vogel for many helpful discussions. This work was supported by a grant from National Aeronautics and Space Administration (to P.L.M.) and much of the work was carried out at the Marine Biological Laboratory.