| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
J. Biol. Chem., Vol. 278, Issue 33, 31058-31066, August 15, 2003
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
-Latrotoxin (LTXN4C) Does Not Form Pores and Causes Secretion by Receptor Stimulation
|| **
From the
Department of Biological Sciences,
Imperial College London, London SW7 2AY, United Kingdom, the
Medical Research Council, Anatomical
Neuropharmacology Unit, Oxford OX1 3TH, United Kingdom, and the
¶Department of Neuroscience, University of
Edinburgh, Edinburgh EH8 9JZ, United Kingdom
Received for publication, October 10, 2002 , and in revised form, May 19, 2003.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
-Latrotoxin (LTX) causes massive release of neurotransmitters via a
complex mechanism involving (i) activation of receptor(s) and (ii) toxin
insertion into the plasma membrane with (iii) subsequent pore formation. Using
cryo-electron microscopy, electrophysiological and biochemical methods, we
demonstrate here that the recently described toxin mutant (LTXN4C)
is unable to insert into membranes and form pores due to its inability to
assemble into tetramers. However, this mutant still binds to major LTX
receptors (latrophilin and neurexin) and causes strong transmitter exocytosis
in synaptosomes, hippocampal slice cultures, neuromuscular junctions, and
chromaffin cells. In the absence of mutant incorporation into the membrane,
receptor activation must be the only mechanism by which LTXN4C
triggers exocytosis. An interesting feature of this receptor-mediated
transmitter release is its dependence on extracellular
Ca2+. Because Ca2+ is also
strictly required for LTX interaction with neurexin, the latter might be the
only receptor mediating the LTXN4C action. To test this hypothesis,
we used conditions (substitution of Ca2+ in the medium
with Sr2+) under which LTXN4C does not bind
to any member of the neurexin family but still interacts with latrophilin. We
show that, in all the systems tested, Sr2+ fully
replaces Ca2+ in supporting the stimulatory effect of
LTXN4C. These results indicate that LTXN4C can cause
neurotransmitter release just by stimulating a receptor and that neurexins are
not critical for this receptor-mediated action. |
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
-Latrotoxin
(LTX)1 stimulates
exhaustive release of neurotransmitters in vertebrates. This toxin has been
extensively used to probe molecular mechanisms that control exocytosis of both
synaptic vesicles (SVs) and large dense-core vesicles (LDCVs) in such diverse
models as brain, neuromuscular junctions, and endocrine cells (for reviews see
Refs.
1–3).
LTX acts only after binding to presynaptic receptors (4). Once receptor-bound, the toxin can trigger exocytosis by several mechanisms: (i) activation of the receptors (5–8), (ii) formation of non-selective pores in the membrane (9–11), and (iii) hypothetical intracellular interaction with the exocytotic machinery (12).
Because toxin pores damage cell membranes and produce strong cytotoxic effects (e.g. 13–15), only the receptor-transduced LTX action is likely to reveal intact, physiologically important exocytotic mechanisms. Unfortunately, this action is difficult to study using the wild-type LTX (LTXWT) because it easily inserts into membranes and forms ionic pores (9–11, 16). This problem could be overcome by designing LTX mutants that would lack the propensity of membrane insertion, and a promising toxin variant has been described recently (LTXN4C) (17). This mutant had the same affinity for the receptors as LTXWT (17) but failed to form pores in synaptosomes or receptor-transfected BHK cells (8). Lacking the major (ionophore) activity, LTXN4C was originally thought to be altogether inactive (12, 17). However, we have discovered that, in fact, LTXN4C strongly stimulates receptor-mediated transmitter release (8), providing an ideal tool for investigating exocytotic signals transduced by LTX receptors.
If LTXN4C acts via receptor stimulation, which LTX receptor is
crucial for this signaling? Two major receptors have been isolated that bind
LTX specifically and with high affinities: neurexin I (NRX I)
(18) and latrophilin 1 (LPH,
also called CIRL) (19,
20). NRXs are a family of
single-pass transmembrane proteins that resemble cell-contact molecules
(18). There are three long
() and three short (
) forms of NRXs transcribed from three
homologous genes (18,
21). While NRX I binds
LTX strongly, other homologs have low affinities for the toxin
(22); however, all NRXs
strictly require Ca2+ for this interaction
(22,
23). LPH is a G
protein-coupled receptor that binds LTX avidly under all ionic conditions
(5). LPH also has two homologs
(LPH 2 and 3), but these do not interact with LTX appreciably
(22,
24,
25).
Interestingly, the receptor-mediated effect of LTXN4C in central
synapses is Ca2+-dependent
(8), reasonably suggesting that
the Ca2+-dependent receptors (i.e. NRXs) may be
responsible for mediating this LTX action. Although this hypothesis
contradicted our previous results
(5,
26,
27), we decided to test it
using the fact that Sr2+ does not support the
interaction of LTX with NRX I but substitutes
Ca2+ in LTX-evoked exocytosis
(5).
The objectives of the current study were (i) to reveal the molecular basis of the mutant's inability to form ionic pores and (ii) to assess the importance of NRX(s) for the receptor-mediated LTX action. We show that LTXN4C is unable to form pores because it fails to assemble into tetramers. In several model systems that endogenously express both LPH and NRXs (rat synaptosomes; rat hippocampal cultures; mouse neuromuscular junctions, NMJs; and bovine chromaffin cells), we demonstrate that, in the presence of Sr2+, LTXN4C does not interact with any NRX, but still strongly stimulates receptor-mediated neurotransmitter release. We conclude that NRXs are not strictly required for the receptor-transduced LTX signaling and suggest that LPH is the major player in this process.
|
EXPERIMENTAL PROCEDURES |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
Production and Purification of Recombinant Toxins—The construction of baculovirus for LTXWT expression was outlined previously (28). The cDNA encoding LTXN4C (17) was combined from two fragments obtained by polymerase chain reactions on the LTXWT template between the following primer pairs: 5'-TTGGGATCCGAAGGAGAAGATTTAACT/5'-TGTCGCCTCGAGGTACTAGGTATGGTGTGATTTC and 5'-AAATCGTACCTCGAGGATCAAATTTCATCGACATT/5'-GAACTCTCGAGTTACCTCCGAAATTTTCCGCT. These fragments were consecutively ligated into the baculovirus transfer vector pMelBacA (Invitrogen). The final LTXN4C construct was verified by complete sequencing and co-transfected with the viral DNA (Bac-N-Blue kit, Invitrogen) into Sf9 cells cultured in complete TNM-FH medium (Invitrogen). Positive plaque selection, production of high-titer baculovirus stocks and large-scale expression of recombinant toxins in Hi5 cells was carried out as published (28).
Recombinant toxins were purified by affinity chromatography on the A15 mAb
attached to CNBr-activated Sepharose (Amersham Biosciences). For this purpose,
expression medium containing a secreted recombinant LTX was filtered through a
0.22 µm filter (Whatman) and then passed through the antibody column. The
column was washed with 150 mM NaCl, 50 mM Tris-HCl, pH
8.2, and eluted with three column volumes of 1 M MgCl2,
50 mM Tris-HCl, pH 8.2. Chromatographic fractions were analyzed by
SDS-electrophoresis/Western blotting, and those containing the recombinant
LTXs were pooled, dialyzed against TBSM (in mM: 150 NaCl, 2.0
MgCl2, 50 Tris-HCl; pH 8.2) and concentrated in a Vivaspin6 unit
(Sartorious) to 
50 nM.
Channel/Pore Formation by Recombinant
Toxins—Constructs for expression of LPH and NRX have been described
previously (9). To measure
45Ca2+ influx, COS7 cells were transfected
with LPH or NRX I (9)
and plated onto 12-well culture plates at 40% confluence. One day after
transfection, the cells were incubated with different concentrations of
recombinant toxins in buffer A (in mM: 145 NaCl, 5.6 KCl, 5.6
glucose, 0.5 MgCl2, 15 HEPES, and 0.5 mg/ml bovine serum albumin;
pH 7.4) supplemented with 2 mM CaCl2 and 10 µCi/ml of
45Ca2+. After 5 min incubation, the medium
was removed, and the cells washed twice. The amount of radioactivity
accumulated by cells was determined by liquid scintillation counting of cells
solubilized in 1% Triton X-100.
For electrophysiological detection of toxin channels, CHO cells, maintained
in Dulbecco's modified Eagle's medium supplemented with 10% heat-inactivated
fetal bovine serum and antibiotics (Invitrogen), were transiently
co-transfected with receptor constructs mixed with pEGFP-N2 (Clontech)
encoding green fluorescent protein (to identify transfected cells); FuGENE 6
reagent (Roche Applied Science) was used to aid transfection. Recordings were
performed 29–41 h after transfection, on isolated cells expressing green
fluorescent protein. The extracellular solution contained (in mM)
130 NaCl, 3.5 KCl, 1.5 MgCl2, 3 CaCl2, 48
NaHCO3, 1.25 NaH2PO4, 10 glucose, and 1 mg/ml
bovine serum albumin; pH 7.4. The 2–5 M
patch pipettes were
filled with an intracellular solution consisting of (in mM): 135
CsCl, 2 MgCl2, 10 HEPES, 2 Na2-ATP, 0.2
GTP-Na2; pH 7.2 with CsOH. Pulse and Pulsefit (HEKA) software was
used for experimental protocols and analysis. Whole-cell currents were
filtered at 2.9 kHz and digitized at 5 kHz. Inward currents induced by LTXs
were measured and a voltage stimulation protocol (10 mV x 40 ms steps to
–90/+30 mV from holding potential of –60 mV) was applied every 1
min to check membrane conductance.
Native Electrophoresis—The recombinant toxins were subjected
to native alkaline electrophoresis in 6% polyacrylamide gels as described
previously (29). Prior to
loading onto the gel, some samples were incubated for 5 min with 0.2% SDS
(without boiling) or with 1% SDS and 5% -mercaptoethanol (with boiling).
After separation, the proteins were analyzed by immunoblotting using a
polyclonal rabbit anti-LTX Ab.
Cryomicroscopy and Image Processing—Specimens for cryo-electron microscopy (cryo-EM) were prepared by applying 4-µl aliquots of 100 µg/ml LTXN4C in TBSM onto grids containing a porous carbon foil, followed by immediate vitrification; for details and references, see Ref. 29. Images of the vitrified samples kept at –184 °C were taken in a CM200 cryomicroscope equipped with a field-emission gun, at x50,000 magnification and 3.1 ± 0.1 µm defocus. The micrographs were digitized at a step size of 14 µm (2.8 Å on the specimen scale). Molecular images were selected interactively and processed using the IMAGIC-5 software. The images were subjected to multireference alignment and multivariate statistical classification in order to find characteristic molecular views in the sample. Symmetry of the oligomers was determined using free reference alignment followed by multistatistical analysis. Characteristic views of LTXN4C were compared with re-projections of the three-dimensional maps of the LTX dimer and tetramer created using the exact filter back projection algorithm (29).
Binding Experiments—Recombinant toxins were labeled with 125I as outlined previously (30). To study toxin dissociation from membranes, 0.5 nM iodinated LTXWT or LTXN4C were incubated in TBSM (plus 2 mM Ca2+ and 0.5 mg/ml bovine serum albumin) for 10 min, either on ice or at 37 °C, with P2 membranes (0.5 mg protein/ml) isolated from rat brain (31). The samples were then supplemented with 500 nM LTXNAT (to displace the specifically bound iodinated LTXs) and transferred on ice (to prevent any further toxin incorporation into the membranes). Aliquots were taken from these reaction mixtures at specified times and filtered through GF/F filters (Whatman) to determine the remaining bound toxin. The amount of non-specifically bound LTXs was established by including 500 nM LTXNAT during the initial incubation.
For binding studies, COS7 cells were transfected with LPH or NRX I
as described (9), dislodged
from plates and incubated (2.5 x 105 cells/reaction) for 10
min with 1 nM [125I]LTXN4C in buffer A
supplemented with EGTA, CaCl2 or SrCl2 as indicated in
Fig. legends. Throughout this work, Sr2+ buffers also
contained 0.2 mM EGTA to chelate any contaminating
Ca2+, while the [SrCl2] was adjusted to give
the stated free [Sr2+]. Reaction mixtures were filtered
through GF/F filters whose radioactivity was then measured.
The affinities of recombinant toxins for receptors in chromaffin cells were
determined in situ, similar to catecholamine release studies (see
below). After the incubation with 0–10 nM radiolabeled
LTXWT or LTXN4C and washes, the amounts of bound toxins
were established by 
-spectrometry of solubilized cells. The bound and
free toxin concentrations were calculated using quantitative assessments of
oligomeric states of the two toxins (determined by cryo-EM). In these
experiments, the specific binding was calculated by subtracting the
nonspecific binding (determined in the presence of 100-fold excess of
unlabeled LTX) from the total binding.
Immunological Procedures—For immunoprecipitation experiments, the NRX Ab was purified by affinity chromatography on SH-Sepharose conjugated with the cognate peptide (NH2-CSANKNKKNKDKEYYVCOOH). P2 membranes were prepared from rat brain and liver as outlined (31). The membranes were solubilized with 1% Thesit in buffer A (1 mg of protein/ml) and divided into 200-µl aliquots, which were incubated for 10 min with 1 nM125I-LTXN4C in the presence of EGTA, CaCl2, or SrCl2. Then, 5 µg of the affinity-purified NRX Ab and 20 µl of protein G-agarose (Sigma) were added to each tube and incubated for another 3 h. The radioactivity of agarose pellets was determined after two washes with buffer A containing respective additives. The immunoprecipitated proteins were eluted with the SDS-electrophoresis loading buffer, separated in 6% SDS-polyacrylamide gels, transferred onto polyvinylidene fluoride membrane (Immobilon) and visualized using the anti-NRX Ab, a peroxidase-conjugated secondary Ab and a chemiluminescent substrate (Pierce).
For detection of LTX receptors in chromaffin cells, equal samples of Triton X-100-solubilized chromaffin cells (105/lane) were separated on an 8% SDS-polyacrylamide gel, transferred onto Immobilon and stained with the anti-NRX and anti-LPH Abs as above.
Biochemical Measurements of Release—Release of [14C]glutamate ([14C]Glu) was done as in Ref. 8. Briefly, synaptosomes were loaded for 5 min with [14C]Glu, washed with physiological buffer (in mM: 145 NaCl, 5 KCl, 1 MgCl2, 10 glucose, 20 HEPES, pH 7.4), and used for release experiments 1 h later (to facilitate the labeling of both the readily releasable and depot pools of vesicles). The release was measured over 5 min after application of 0–2 nM recombinant toxins in the presence of EGTA, CaCl2, or SrCl2. In some experiments, LaCl3 was also included in the buffer to block any toxin pores.
Bovine chromaffin cells were prepared and maintained at a density of 1 x 106 cells/well in 24-well culture plates (32). Five days later, the cells were stimulated for 15 min with 0–5 nM recombinant LTXs in buffer A containing EGTA, CaCl2, or SrCl2. Pore formation was determined simultaneously with release measurements in samples containing 2 mM CaCl2 and 10 µCi/ml of 45Ca2+. After the 15-min stimulation with the toxins, the media were removed; the cells were washed twice with the incubation buffer (without the radioactive tracer) and solubilized in 1% Triton X-100. The uptake of 45Ca2+ was determined by scintillation counting of the solubilized cells. Catecholamine contents of the solubilized cells and media were measured fluorimetrically (32), using an automated microplate fluorimeter, Fluoroscan Ascent FL (Labsystems). Both in synaptosomes and in chromaffin cells, stimulated release was calculated by subtracting basal release in the absence of the toxins from release in their presence and expressed as a percentage of total transmitter contents.
Electrophysiological Recordings of Synaptic Transmission—Organotypic hippocampal slice cultures were prepared as described (33). Whole-cell patch-clamp recordings from CA3 pyramidal cells were performed with pipettes filled with intracellular solution (in mM, 126 K-gluconate, 10 HEPES, 10 Na2-phosphocreatine, 4 KCl, 4 ATP-Mg, and 0.3 GTP-Na2, pH 7.3). Extracellular solution (the same as in toxin channel recordings in CHO cells above) was supplemented with 1 µM tetrodotoxin, 30 µM bicuculline, 3 mM CaCl2, or SrCl2 and oxygenated. The recorded neurons were voltage-clamped at –70 mV; membrane currents were amplified (10 mV/pA), low-pass filtered at 2.9 kHz, and digitized at 5 kHz. Toxin applications were done as described (34). The currents were acquired on-line using the Pulse software.
Intracellular recordings in NMJs were made using standard techniques on isolated preparations of a mouse flexor digitorum brevis muscle. Spontaneous miniature end-plate potentials (mepps) were recorded before and after the addition of 1 nM LTXN4C. The continuously oxygenated extracellular perfusion solution (in mM: 137 NaCl, 5 KCl, 1 MgCl2, 5 HEPES, 5.6 glucose, 0.03 tetrodotoxin, pH 7.4) was supplemented with EGTA, Ca2+, or Sr2+. Recordings were made using an Axoclamp 2B amplifier (Axon Instruments); voltage signals were amplified and low-pass filtered at 3 kHz, digitized at 10 kHz using the 1401+ signal converter and recorded into Spike 4 software (Cambridge Electronic Design). The toxin was added directly to the recording chamber and thoroughly mixed with the perfusion buffer. All electrophysiological data were analyzed off-line with the MiniAnalysis software (Synaptosoft).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
|
We previously showed that LTXN4C fails to form ionic pores in the membrane of synaptosomes and LPH-transfected BHK cells (8). Here, we confirmed and extended these observations by comparing the abilities of the recombinant toxins to induce 45Ca2+ influx into COS7 cells expressing exogenous LPH or NRX. As expected, LTXWT caused massive accumulation of the radioactive tracer in the cells (Fig. 2A). In contrast, LTXN4C stimulated no detectable 45Ca2+ influx (Fig. 2A). The lack of pore formation by even high concentrations of the mutant toxin was not due to its poor interaction with the expressed receptors because, in the same experiment, LTXWT and LTXN4C exhibited similar binding to both NRX- and LPH-transfected cells (Fig. 2B).
|
We then compared the channel-forming properties of LTXWT and
LTXN4C by a more sensitive method, whole-cell patch-clamp
recordings in CHO cells transfected with NRX or LPH. No changes in membrane
conductance and no inward currents were detected after a 20 min
application of LTXN4C (0.5, 1, and 5 nM) to cells
expressing NRX or LPH (Fig. 2, C
and D). In contrast, when 1 nM
LTXWT was subsequently applied to the same cells, a large (up to
400 pA) inward current always developed over several minutes, accompanied by
an increase in membrane conductance (Fig.
2, C and D); these currents were of the same
order of magnitude as those observed previously
(10,
11,
35,
44), despite any differences
in cell types used, transfection efficiencies, etc.
What could be the reason for such distinct behaviors of the two recombinant toxins? LTXN4C differs from LTXWT only by a 4-amino acid insert (Val-Pro-Arg-Gly) that has been introduced between the predicted N-terminal domain and the ankyrin repeats of LTX (Fig. 3A, top) in order to create a cleavage site for thrombin (17). The borderline between the domains was deduced from the toxin's sequence. However, cryo-EM demonstrates that the domain structure of LTX is actually different (29), and that the mutation has fallen inside a tightly packed "body" domain (Fig. 3A, middle and bottom); consistently, this peptide is inaccessible to thrombin (17). Small structural changes in this region would not be expected to affect the N-terminal wing domain, which is presumed to participate in receptor binding (17, 29), but could perturb the overall structure of LTX and its oligomerization, which is crucial for pore formation (16, 29). To reveal any alteration of the three-dimensional structure of LTXN4C induced by the insert, we used native electrophoresis, a method that had allowed us to demonstrate LTX oligomerization (29). Fig. 3B shows that, in the absence of divalent cations, both LTXWT and LTXN4C normally existed as dimers, although LTXN4C had a slightly lower mobility, which was probably due to its partial unfolding caused by the insert. The addition of 0.2% SDS, known to induce oligomerization of LTX during native electrophoresis (29), clearly produced several bands in the LTXWT sample that corresponded to monomers, dimers, and tetramers. However, the tetramers were absent from the equally treated LTXN4C (Fig. 3B), suggesting that it was easily denatured by 0.2% SDS and lost an ability to assemble into higher oligomers. Upon boiling in 1% SDS, both toxins became monomeric and, as expected, displayed the same electrophoretic mobility.
|
The nature of LTXN4C oligomers was then studied by cryo-EM, which allows the observation of molecules that have been instantaneously frozen in their physiological states. The characteristic molecular images of LTXN4C were similar in size and shape to re-projections of the three-dimensional model of the LTXNAT dimer (Fig. 3C) (29), indicating that the mutant was indeed dimeric. However, the LTXN4C dimers were slightly more flexible and displayed some deviations from the structure of similarly oriented wild-type dimers (e.g. classes III and V in Fig. 3C). Much more striking was the fact that the tetramers, which under the conditions of this experiment constitute 96 ± 3% of identifiable wild-type toxin particles (Fig. 3C, class VI) (16, 29) have never been observed in LTXN4C samples.
It has been postulated that LTX must assemble into cyclical tetramers in
order to insert itself into membranes and form pores
(16,
29), and our results implied
that because the mutant toxin did not make tetramers, it should be incapable
of membrane insertion. To support this idea, we demonstrated that the
dissociation of LTXN4C from receptor-containing membranes did not
depend on the incubation temperature during binding and always followed the
same first order kinetics (Fig.
3D). This experiment could only be explained by the
inability of the mutant to insert into membranes at any temperature. In total
contrast, the dissociation of the wild-type LTX was greatly affected by the
temperature during binding (Fig.
3D): if the binding was carried out at 37 °C, only
25% of LTXWT could dissociate afterward, whereas about 75%
never dissociated due to permanent incorporation into the membrane. As
expected, at 0 °C, the behavior of LTXWT was intermediate,
consistent with restricted incorporation into the membrane at low temperatures
(Fig. 3D).
Thus, our combined data not only ascertain the lack of pore formation by the mutant toxin, but also explain this feature at the molecular level. These findings further indicate that LTXN4C can cause exocytosis only by a reversible interaction with its receptors. Experiments below were designed to answer the question of the relative importance of the two LTX receptors, LPH and NRX, for the receptor-mediated action of the mutant toxin.
NRXs Do Not Bind LTXN4C in the Presence of
Sr2+—We showed previously that
Sr2+ could fully replace Ca2+ in
supporting LTX-evoked release of [3H]norepinephrine from rat
synaptosomes (5). At the same
time, no interaction between NRX I and LTX could be detected in the
presence of Sr2+. However, in the current work we used a
recombinant, mutant toxin, so it was necessary to ascertain that it had the
same divalent cation requirement for binding as native LTX. To test this, we
transiently expressed LPH or NRX I in COS7 cells, which do not contain
endogenous LPH or NRXs. As expected, LPH-expressing cells bound
125I-LTXN4C under all conditions tested, while
NRX-expressing cells were able to bind the mutant toxin only in the presence
of Ca2+ (Fig.
4A).
|
This result was based on the use of NRX I and could not be
automatically extended to neurons, which additionally contain other members of
the NRX family that have slightly different ionic requirements for LTX binding
(22). An experiment was then
designed to detect LTXN4C binding to any NRX variant. For this
purpose, solubilized rat brain membranes were mixed with
125I-LTXN4C in the presence of EGTA,
Ca2+, or Sr2+ and then
immunoprecipitated using an antibody that equally recognizes the C termini of
all NRXs. As a control of specificity, we used rat liver membranes that lack
NRXs (18).
Fig. 4B (upper
panel) demonstrates that independently of the ionic conditions used, this
procedure brought down both the - and -forms of NRXs from brain.
However, 125I-LTXN4C co-immunoprecipitated with NRXs
specifically only in the medium containing Ca2+
(Fig. 4B, lower
panel), while neither EGTA nor Sr2+ supported any
interaction between the mutant toxin and NRXs.
These results demonstrate unequivocally that, similar to native toxin, the mutant LTXN4C binds to all NRXs in a strictly Ca2+-dependent manner and that Sr2+ cannot sustain this interaction. Using this approach, we then asked whether the involvement of NRXs was critical for LTXN4C-stimulated release of neurotransmitters in several model systems based on either biochemical or electrophysiological detection of exocytosis.
LTXN4C-evoked Release of [14C]Glu from Rat Synaptosomes—As mentioned above, LTXN4C stimulates receptor-mediated exocytosis of amino acid neurotransmitters from synaptosomes only in the presence of Ca2+ (8). In the current work, we substituted Ca2+ with Sr2+. Synaptosomes were loaded with [14C]Glu and stimulated by different concentrations of LTXN4C in a buffer containing EGTA, Ca2+ or Sr2+. As expected, the mutant toxin was inactive in the presence of EGTA (i.e. when the only divalent cation present was Mg2+), but capably stimulated [14C]Glu release in the Ca2+-containing buffer (Fig. 5A). Interestingly, in the presence of Sr2+, LTXN4C was also able to trigger exocytosis, and the amount of this release was not statistically different from that in Ca2+-containing medium (Fig. 5A).
|
Did LTXN4C require Ca2+ or Sr2+ simply because it produced a small number of pores that allowed these cations to enter nerve terminals and evoke secretion? Although we have demonstrated above that the mutant toxin cannot form pores and also showed previously that LTXN4C did not allow influx of 45Ca2+ into synaptosomes (8), it was still necessary to prove that in our release experiments the mutant toxin acted only via receptor stimulation. To distinguish between the receptor- and pore-mediated actions, we used La3+ known to efficiently block LTX pores (8, 35). The idea was that if the mutant's effect had been even partially based on pore formation, it should be inhibited by La3+. Indeed, this cation strongly attenuated Glu release (Fig. 5B) caused by LTXWT, which acts via both receptor and pores. In particular, the Ca2+-independent release was abolished, while the Ca2+- and Sr2+-dependent secretion was brought to the levels normally evoked by LTXN4C (see also "Discussion"). In contrast, transmitter secretion stimulated by LTXN4C in the presence of either Ca2+ or Sr2+ was not at all affected by La3+ (Fig. 5B). This ability of the pore-blocking cation to partially inhibit the action of the wild type but not the mutant LTX did not depend on the concentration of toxins used (data not shown).
Thus, pore formation does not play any detectable role in the Ca2+/Sr2+-dependent Glu release triggered by LTXN4C in synaptosomes, and this secretion can occur without the participation of NRXs.
Secretion of Catecholamines from Bovine Chromaffin Cells— LTX is known to efficiently stimulate exocytosis of LDCV containing catecholamines and hormones in chromaffin and PC12 cells; this action is strictly Ca2+-dependent (6, 36).
First, we ascertained that bovine chromaffin cells expressed endogenous LPH and NRX. Toxin binding to these cells was higher in Ca2+-containing buffers than in EGTA (Fig. 6A, upper panel), indicating the presence of Ca2+-dependent receptors, which corresponded to NRXs, as demonstrated by Western blotting (Fig. 6A, lower panel). Immunostaining also showed that the Ca2+-independent receptors were represented by LPH.
|
We then compared LTXN4C with wild type toxin in terms of binding to chromaffin cells in situ. As demonstrated in Fig. 6B, the dissociation constants and the numbers of binding sites were equal for both toxins (0.26 nM and 5.2 fmol/106 cells, respectively).
LTXWT and LTXN4C were then used to stimulate release of endogenous catecholamines and influx of Ca2+ into chromaffin cells. Despite similar binding of the two toxins to these cells, LTXWT caused a stronger transmitter secretion than did LTXN4C by acting via both receptor- and pore-dependent mechanisms (Fig. 6C). Consistently, release stimulated by the wild-type toxin was associated with massive influx of 45Ca2+ (Fig. 6D). In agreement with the previously published data (6, 22), the bell-shaped secretion curve indicated that catecholamine release was mainly stimulated by Ca2+ entry via toxin pores. Indeed, while exocytosis correlated with intracellular [Ca2+] at lower [LTXWT], excessive Ca2+ influx induced by 3–5 nM LTXWT inhibited release as in Ref. 6. Note that the apparent decrease in 45Ca2+ uptake at the highest LTXWT concentration was not caused by inhibition of LTX pores but rather by disproportionately easy escape of Ca2+ through multiple toxin pores during extensive washing; this effect was less evident at lower toxin doses. In contrast to LTXWT, the mutant did not cause any detectable uptake of 45Ca2+ (Fig. 6D), confirming that it does not form pores but acts via the receptor-mediated mechanism. Finally, we tested the effect of mutant toxin on transmitter exocytosis in the presence of either Ca2+ or Sr2+. Again, LTXN4C was as efficient in Sr2+ as it was in Ca2+ (Fig. 6E).
LTXN4C-induced Spontaneous Release of Glu in Rat Hippocampal Cultures—Recently, we showed that LTXN4C dramatically increases the frequency of Glu-mediated miniature excitatory post-synaptic currents (mEPSCs) in hippocampal slices (8). In this system too, the mutant toxin is active only in the presence of calcium (45). In the current study, we compared the effects of LTXN4C in the presence of Ca2+ and Sr2+. The results shown in Fig. 7 demonstrate that LTXN4C increased the mEPSCs frequency in Sr2+-containing buffer approximately to the same extent as in the presence of Ca2+ (25-fold on average). At the same time, LTXN4C had no effect on the mean mEPSCs amplitude under any condition, indicating that the mutant acted presynaptically in both media (Fig. 7B).
|
LTXN4C-evoked Spontaneous Exocytosis of Acetylcholine at Mouse NMJs—Although LTX effect of at the NMJ has been extensively studied (37–40) it remains largely unexplained due to the multiple actions of native toxin (1, 40). Therefore, the use of LTXN4C, which fails to form pores or penetrate into the cytosol, should greatly simplify the interpretation of results.
When tested on mouse NMJs, the mutant toxin consistently exhibited the same
effects as in the other systems: it was inactive in the presence of EGTA over
the period of hours2
but caused a profound increase in mepps frequency when the medium was
supplemented with Ca2+ or Sr2+
(Fig. 8A). Normalized
data from 11 muscle fibers (Fig.
8B) indicate that the mutant toxin caused a similar (70-fold)
increase in the mepps frequency in the presence of these divalent cations. As
in hippocampal cultures, the amplitude of the miniature events remained the
same under all conditions, demonstrating the presynaptic site of
LTXN4C action (Fig.
8B).
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
and LPH (2). Although
these structurally unrelated receptors are unlikely to couple to one signaling
system, they both can efficiently support the pore-forming activity of
LTXWT by recruiting toxin tetramers to the cell surface and
facilitating their incorporation into the membrane
(9,
29). This explains why NRXs,
LPH, or even their signal-deficient mutants, over-expressed in chromaffin and
PC12 cells, sensitize these cells to native/wild type LTX
(22,
41): all receptors increase
the number of toxin pores in the cell membrane
(9–11,
41). Receptor gene knockout
has also given unclear results
(42,
43) because the pore-forming
LTXWT remained active even in mice lacking either of these
receptors; however, the simultaneous knockout of both receptors dramatically
reduced toxin activity (43).
Possible interaction of LPH and NRX I
(2) and the ability of LTX to
bind many NRX variants (22)
hinder this approach even further and necessitate a simultaneous knockout of
all NRXs. Therefore, the system had to be simplified. First, it was necessary to prevent all the activities of LTXWT associated with its incorporation into the membrane and resulting in pore formation and/or interaction with intracellular proteins. For this purpose, we used the mutant LTXN4C, which, as we have demonstrated (Figs. 2 and 3) (8), is unable to insert into the membrane and form pores but can only act via receptor stimulation. Second, we needed to exclude the toxin's interaction with one receptor type. NRXs were chosen for two reasons: (i) they were apparent candidates to mediate the receptor-transduced LTX action, which critically required Ca2+ (5, 8), and (ii) toxin binding to all members of the NRX family could be totally abolished by a simple replacement of Ca2+ with Sr2+ (5) (Fig. 4). This eliminated the need for multiple gene knockouts. Importantly, the use of Sr2+ did not alter the interaction of LTXN4C with LPH (Fig. 4A).
With this approach, we have found that Sr2+ could fully substitute Ca2+ in LTXN4C-evoked exocytosis of Glu-containing SVs in rat synaptosomes (Fig. 5A) and hippocampal cultures (Fig. 7), acetylcholine-containing SVs at mouse NMJs (Fig. 8), and catecholamine-containing LDCVs in bovine chromaffin cells (Fig. 6E). In all the models, the mutant toxin was inactive when only Mg2+ was present, confirming our previous observation that receptor-mediated LTX-evoked secretion requires extracellular Ca2+ (5, 8).
The dependence of the LTXN4C effect on the presence of Ca2+ or Sr2+, both of which can stimulate exocytosis upon entering the cytosol, could potentially mean that the mutant toxin simply acted as an ionophore, if only less potent than LTXWT. However, we show that LTXN4C does not form cation-permeable pores in synaptosomes (8), hippocampal slice cultures2, chromaffin cells (Fig. 6D), and different types of receptor-transfected cells (8) (Fig. 2, A and C). Our present results further indicate that the pore-forming ability is the main distinction between LTXN4C and LTXWT. Indeed, although LTXWT stimulates a stronger secretion than does LTXN4C, this difference (i) is abolished by La3+, which blocks LTX pores (in synaptosomes); and (ii) correlates with Ca2+ influx (in chromaffin cells). In contrast, the effect of LTXN4C is insensitive to La3+ (Fig. 5B) and does not associate with Ca2+ entry (8) (Fig. 6D).
Our experiments also provide a molecular explanation of the inability of the mutant LTX to form pores. LTXWT induces pores by assembling into cyclical tetramers that possess a central channel and a hydrophobic base, which allows the toxin to incorporate into the membrane (29). We demonstrate here that LTXN4C cannot make tetramers; as a result, it lacks the exposed hydrophobic domain and must be unable to insert into membranes. Indeed, the mutant toxin readily dissociates from brain membranes, whereas LTXWT becomes permanently incorporated into the membrane in a temperature-dependent manner (Fig. 3D). These findings are consistent with the data on protease protection (12), which also suggested that the mutant was incapable of membrane insertion.
Our results, thus, indicate that LTXN4C-evoked release is
mediated by receptors. In addition, the amount of secretion induced by
LTXN4C in all the systems is similar regardless of NRX
participation in toxin binding. Thus, although we cannot totally rule out some
involvement of NRX I in LTX signaling, NRX is not critical for the
receptor-mediated release.
Taken together, these findings strongly implicate LPH as the main LTX receptor that transduces an exocytotic signal. This signaling is masked by massive Ca2+ influx leading to exocytosis when LTXWT is used and can only be clearly demonstrated with LTXN4C. It is currently unclear why the mutant toxin binds LPH under all conditions but requires extracellular Ca2+ or Sr2+ to induce exocytosis. It is possible that these divalent cations (i) serve as extracellular co-factors for LPH and its signaling partners, (ii) participate in the activation of intracellular Ca2+ stores required for LTXN4C-induced release (45) or (iii) directly induce secretion after entering the cell via some LPH-linked cation channels. The latter hypothesis is interesting but at present seems implausible because these hypothetical channels are undetectable electrophysiologically or biochemically and must be very unusual, being insensitive to La3+, Cd2+, SKF 96365 and other drugs (45). Future research will concentrate on the mechanisms of the LPH-transduced LTX action and the roles of Ca2+ and Sr2+ in this process.
|
FOOTNOTES |
|---|
|| Current address: Dept. of Crystallography, Birkbeck College, London WC1E
7HX, UK.
** To whom correspondence should be addressed. Tel.: 44-(0)20-7594-5237; Fax: 44-(0)20-7594-5207; E-mail: y.ushkaryov{at}imperial.ac.uk.
1 The abbreviations used are: LTX, -latrotoxin; LDCV, large densecored
vesicles; LPH, latrophilin; LTXNAT, native LTX; LTXN4C,
mutant LTX; LTXWT, wild-type LTX; mepps, miniature end-plate
potentials; mEPSCs, miniature excitatory postsynaptic currents; NMJ,
neuromuscular junction; NRX, neurexin; Ab, antibody; SVs, synaptic
vesicles.
2 A detailed study of the role of Ca2+ in the
LTXN4C action at the NMJ will be published elsewhere (K. E.
Volynski, D. Thomson, R. R. Ribchester, and Y. A. Ushkaryov, paper in
preparation).
|
ACKNOWLEDGMENTS |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
