Three distinct kinetic groupings of the synaptotagmin family: Candidate sensors for rapid and delayed exocytosis

Materials and Methods

Recombinant Proteins. cDNA encoding rat syts I (19), III (20), and IV (21) was provided by T.C. Südhof (University of Texas Southwestern Medical Center, Dallas), S. Seino (Chiba University, Chiba, Japan), and H. Herschman (University of California, Los Angeles), respectively. cDNA encoding mouse syts II and V–XI (22) and squid syt I was provided by M. Fukuda (Institute of Physical and Chemical Research, Saitama, Japan). cDNA encoding syt XII (23) was provided by C. Thompson (The Johns Hopkins University, Baltimore). The cytoplasmic domain, containing both C2 domains (C2A and C2B), of each isoform was subcloned into pGEX-2T or 4T-1 (Amersham Biosciences) as described in ref. 24. The amino acid residues encoding these cytoplasmic domains are as follows: I, 96–421; II, 139–423; III, 290–569; IV, 152–425; V, 218–491; VI, 143–426; VII, 134–403; VIII, 97–395; IX, 104–386; X, 223–501; XI, 150–430; and XII, 114–421. For liposome-binding assays and Ca²⁺ dose responses, constructs were expressed in Escherichia coli as GST fusion proteins. For stopped-flow experiments, cytoplasmic domains were subcloned into pTrcHisA vector (Invitrogen) generating an N-terminal His-6 tag. GST-fusion and His-6-tagged proteins were purified on glutathione-Sepharose beads (Amersham Biosciences) and Ni-nitrilotriacetic acid-agarose (Qiagen, Valencia, CA), respectively, as described in ref. 25. Recombinant syts harbor tightly bound contaminants that may affect their properties (26). These bacterial contaminants were removed by using DNase/RNase and high-salt washes as described in ref. 27. His-6-tagged proteins were eluted from the beads by using imidazole (28).

Lipids. Synthetic 1,2-dioleoyl-sn-glycero-3-phospho-l-serine (phosphatidylserine, PS), 1,2-dioleoyl-sn-glycero-3-phosphocholine (phosphatidylcholine, PC), 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine-N-(5-dimethylamino-1-naphthalenesulfonyl) (dansyl-PE), and N-(lissamine rhodamine B sulfonyl)-1,2-dipalmitoyl-sn-glycero-3-phosphoethanolamine were purchased from Avanti Polar Lipids. l-1,2-dipalmitoyl-sn-glycero-3-phospho[N-methyl-³H]choline was purchased from Amersham Biosciences.

Liposomes. Lipids were dried under a stream of nitrogen and suspended in Hepes buffer (50 mM Hepes/150 mM NaCl, pH 7.4). For fluorescence studies and rhodamine-labeled liposomebinding assays, large (100-nm) unilamellar liposomes were prepared by extrusion as described in ref. 29. For ³H-labeled liposome-binding assays (Fig. 1C), liposomes were prepared by sonication with a Microson ultrasonic cell disruptor (Misonix, Farmingdale, NY), and binding assays were carried out as described in ref. 29 in 150 μl of Hepes buffer by using 6 μg of immobilized protein and 22 nM liposomes per data point. Rhodamine-labeled liposome-binding assays (Fig. 1B) were carried out in the same manner as the ³H-labeled liposome-binding assays. However, bound liposomes were eluted in Hepes buffer by using 1% Triton X-100. The solubilized lipids were collected, and binding was quantified by measuring the fluorescence intensity of rhodamine at 580 nm. Data were normalized, plotted, and fitted with sigmoidal dose-response curves (variable slope) by using prism 2.0 software (GraphPad, San Diego). In all experiments, error bars represent the SDs from triplicate determinations.

Stopped-Flow Rapid-Mixing Experiments. FRET was used to monitor the time course of Ca²⁺-dependent syt·PS/PC liposome interactions by using an SX.18MV stopped-flow spectrometer (Applied Photophysics, Surrey, U.K.) as described in ref. 30. Native Trp residue(s) in syt and dansyl groups attached to phospholipids in the liposomes served as energy donors and acceptors, respectively. The Trps in the cytoplasmic domain of each isoform of syt were excited at 285 nm, and the emission from the dansyl-PE acceptor was collected by using a 470-nm cutoff filter. The dead time was ≈1 ms.

Results

The cytoplasmic domain of syt is largely composed of tandem C2 domains that are connected by means of a flexible linker (19) (Fig. 1 A). In some isoforms, each C2 domain functions as a Ca²⁺-sensing module that interacts with a number of effectors in response to rapid changes in intracellular Ca²⁺ concentration (31). Current evidence indicates that syt operates by binding, in response to Ca²⁺, anionic lipids (e.g., PS) and the t-SNAREs syntaxin and SNAP-25 (15, 18, 32). To measure the kinetics of syt·t-SNARE interactions, we must use micromolar concentrations of protein; these conditions result in complete binding within 1 msec, which falls within the dead time of our stopped-flow instrument (32). Thus, we were unable to compare syt kinetics by using t-SNAREs as the effector. However, the kinetics of syt·liposome interactions, using membranes that contain PS, can be readily monitored in real time in the stopped-flow device (27, 29, 33) because low concentrations (i.e., nanomolar range) of liposomes are used. Therefore, we compared the kinetics of different isoforms of syt by using PS-harboring liposomes as the effector.

We first screened the membrane-binding activity of syts I–XII by using liposomes that contained 25% PS, 74% PC, and 1% rhodamine-labeled phosphatidylethanolamine as the tracer. With the exception of syts IV, VIII, XI, and XII, most members of the syt family bound to the PS-harboring liposomes in a Ca²⁺-dependent manner (Fig. 1B). We note that a recent study also addressed the ability of each isoform of syt to bind PS/PC liposomes (34). However, binding was assayed by the cosedimentation of syt with liposomes because syts also oligomerize and precipitate in the presence of Ca²⁺ and membranes. The cosedimentation assay is not an accurate method to measure syt·membrane interaction (35).

We then picked three representative isoforms of syt (from the kinetic analysis presented below), I, III, and IX, to resolve the question of whether they have the same or different Ca²⁺ requirements for binding to membranes. For example, the C2A domains of syt I and III have been reported to have virtually identical [Ca²⁺]_1/2 values in some studies (36–38) and distinct [Ca²⁺]_1/2 values in another study (39). The [Ca²⁺]_1/2 values and Hill slopes for syt I were 18.8 ± 2.2 μM and 2.0, respectively; for syt III the values were 7.0 ± 0.8 μM and 3.6, respectively; for syt IX the values were 13.6 ± 1.3 μM and 2.3, respectively. Among these isoforms, syt III has the lowest [Ca²⁺]_1/2 value, followed by syt IX and syt. In these experiments, the [Ca²⁺] was measured by using a Ca²⁺-sensitive electrode. The data concerning syt I are consistent with previous reports (29, 33).

To investigate further the properties of syts, we determined the affinities of the same three syt isoforms (I, III, and IX) for PS-harboring liposomes (5% dansyl-PE/25% PS/70% PC) by measuring the assembly kinetics of the Ca²⁺·syt·liposome complex, using a stopped-flow rapid-mixing approach. FRET was used to monitor the time course of complex formation (30). The native Trp residues in each syt isoform were used as the energy donors, and a dansyl group that was incorporated into the liposomes (as dansyl-PE) served as the energy acceptor. Rapid mixing of liposome/Ca²⁺ mixtures with syt resulted in a rapid increase in the emission of the acceptor (Fig. 1E). The kinetics traces were well fitted with exponential functions to determine the observed rate constant, k_obs. We then measured k_obs as a function of [liposome]; these data are plotted in Fig. 1F. The slope yields k_on, and the y intercept yields k_off. These values are listed in Table 1. The dissociation constants (K_d) were then calculated by k_off/k_on, and the values for syts I, III, and IX were 1.0, 1.1, and 9.8 nM, respectively (Table 1). The Ca²⁺-dependent liposome-binding affinity for syts I and III are similar; that for syt IX is significantly lower. Together, these experiments indicate that, at least in these three syt isoforms, the affinity of the Ca²⁺·protein complex for membranes is not necessarily related to the affinity of the membrane·protein complex for Ca²⁺.

We extended our study to the disassembly kinetics of Ca²⁺·syt·liposome complexes, again using stopped-flow rapid mixing and FRET. We expected that in these kinetics experiments, the isoforms with higher apparent affinity for Ca²⁺ would retain bound Ca²⁺ longer upon rapid mixing with excess Ca²⁺ chelator, thus yielding slower rates of disassembly. For example, based on the measurements above, we expected that syt III would release Ca²⁺, and hence liposomes, more slowly than would syts I or IX. Moreover, from the measurements above, we anticipated that syts I and IX would exhibit similar disassembly kinetics. Surprisingly, these predications were not borne out by the experimental data (detailed below).

To carry out these experiments, liposomes (5% dansyl-PE/25% PS/70% PC) were premixed with syt in the presence of Ca²⁺ and then rapidly mixed with Hepes buffer containing excess EGTA. The chelation of Ca²⁺ by EGTA resulted in a rapid loss of the fluorescent signal (Fig. 2B). All of the kinetic traces could be well fitted by exponential functions to estimate the rate constants for disassembly (Table 2). According to the overall rate of disassembly, we divided syts into three kinetic groups: fast, medium, and slow. syts I, II, and III fell into the fast group, which disassembled within 50 msecs; syts V, VI, IX, and X belonged to the medium group, which disassembled within 250 msec. For syt VII, the only member in the slow group, disassembly required almost 500 msec. Surprisingly, these groupings do not appear to be correlated with the Ca²⁺ sensitivities as measured under steady-state conditions. As noted above, syt III has the highest apparent affinity for Ca²⁺ (among syts I, III, and IX), yet it dissociated from liposomes upon chelation of Ca²⁺ almost as rapidly as syt I. However, syt IX, which has a lower apparent affinity for Ca²⁺ than does syt III (Fig. 1C), disassembled from liposomes upon mixing with chelator much more slowly than did syt III. These results suggest that Ca²⁺ sensitivities for liposome binding in the steady state might not provide information regarding the ability of the protein·membrane complex to “hold onto” Ca²⁺ upon removal of free Ca²⁺ with chelator. However, it remains possible that some isoforms of syt release Ca²⁺ quickly but dissociate from membrane more slowly. This latter model can be viewed as the persistent activation of a syt·effector complex, as alluded to in the introduction.

In contrast to earlier reports based on steady-state measurements (37), the kinetics data reported here reveal that syt III is unlikely to function as the “high-affinity” Ca²⁺ sensor that mediates slow asynchronous release; this isoform is too fast. The best candidates for the high-affinity Ca²⁺ sensor that drives asynchronous transmission are members of the slow or medium-speed group: syt V, VI, VII, IX, or X.

We note that for syt VII, the kinetics traces of the disassembly reactions can only be well fitted with double, rather than single, exponential functions, and the first exponential component has a negative amplitude (Table 2). These data indicate that the disassembly reaction is more complex for this isoform than for the others, perhaps owing to the propensity of this isoform to avidly oligomerize (40), which, in turn, could retard the rate of disassembly.

In the final series of experiments, we explored the effect of temperature on the kinetics of assembly and disassembly of Ca²⁺·syt·liposome complexes. For these studies, we compared syt I from two species with distinct internal body temperatures: rat and squid. The squid protein was of particular interest, because this animal's body temperature changes with the environment. Thus, the squid protein might be expected to be less temperature-dependent. We carried out assembly and disassembly assays as described above. For Ca²⁺·syt·liposome assembly reactions, the kinetics traces of rat syt I were single exponentials, whereas those for squid syt I were double exponentials (Fig. 3A Left). The observed rate constants (k_obs) for the assembly of rat syt I and squid syt I with Ca²⁺ and liposomes were very similar over the range of temperatures tested. These data are presented as Arrhenius plots in Fig. 3A Right.

Results from the Ca²⁺·syt·liposome disassembly reactions, carried out at different temperatures, yielded a different conclusion. The kinetics traces for both rat syt I and squid syt I were best fitted with single exponential functions, but, at the same temperature, disassembly of squid Ca²⁺·syt·liposome complexes is much slower than for the rat protein (Fig. 3B Left). Moreover, k_diss (the rate of disassembly) of rat syt I became too fast to observe at temperatures > 25°C. These differences in the disassembly rates between squid and rat syt point to differences in the manner in which these orthologs release Ca²⁺. The temperature dependencies for disassembly of both the rat and squid protein are also presented as Arrhenius plots in Fig. 3B Right. The overall temperature dependencies were similar, although the squid protein had a somewhat higher energy of activation; E_a for squid syt I was 42 kJ/mol, and for rat syt I this value was 30 kJ/mol (Fig. 3B Right).

Discussion

In this study, we observed that Ca²⁺·syt·liposome complexes disassemble in response to rapid mixing with excess Ca²⁺ chelator with distinct kinetics, depending on the syt isoform under study. These experiments were intended to mimic the collapse of local Ca²⁺ microdomains (4, 41) and determine whether complexes with distinct isoforms of syt bound to liposomes respond to drops in [Ca²⁺] with the same or distinct kinetics. The observation that some syts release Ca²⁺, and hence membranes, quickly, whereas other isoforms of syt remain in complexes for longer lengths of time, has ramifications for presynaptic function. For example, the slow isoforms of syt (syts V, VI, VII, X, and IX) are good candidates to serve as Ca²⁺ sensors during asynchronous or delayed release, which is a component of exocytosis that persists tens to hundreds of milliseconds after Ca²⁺ domains have collapsed, presumably because Ca²⁺ stays bound to a Ca²⁺ sensor (e.g., a syt isoform) that is capable of driving exocytosis. In contrast, syts I, II, and III released Ca²⁺ and membranes with very rapid kinetics, and thus are good candidates for mediating the rapid phasic component of transmission. Indeed, genetic studies demonstrated that rat and fly syt I is essential for rapid transmission (14, 42–44). In the absence of syt I, delayed release was enhanced, perhaps owing to the action of a syt isoform with slow kinetics. In this model, expression of syt I would, in principle, titrate out the slow sensor at release sites to give rise to rapid transmission. Thus, an essential issue is to establish the subcellular localization of each isoform of syt. At present, syt I has been unambiguously localized to synaptic and dense-core vesicles, but the subcellular distribution of the other isoforms is the subject of much debate (reviewed in ref. 31).

The distinct kinetics at which syt isoforms release Ca²⁺ and membranes might also underlie aspects of synaptic plasticity, such as paired-pulse facilitation, in which bound Ca²⁺ is thought to enhance the exocytosis in response to another rise in local [Ca²⁺] (2). A test of these hypotheses will involve experimental manipulation of syt isoforms in neurons. For example, the ratio of slow to fast syts could be increased or decreased. Detailed electrophysiological experiments will reveal whether these changes impact the relative levels of fast versus delayed release and/or whether these manipulations impact aspects of short-term synaptic plasticity.