Selective interaction of syntaxin 1A with KCNQ2: possible implications for specific modulation of presynaptic activity

Abstract

KCNQ2/KCNQ3 channels are the molecular correlates of the neuronal M-channels, which play a major role in the control of neuronal excitability. Notably, they differ from homomeric KCNQ2 channels in their distribution pattern within neurons, with unique expression of KCNQ2 in axons and nerve terminals. Here, combined reciprocal coimmunoprecipitation and two-electrode voltage clamp analyses in Xenopus oocytes revealed a strong association of syntaxin 1A, a major component of the exocytotic SNARE complex, with KCNQ2 homomeric channels resulting in a approximately 2-fold reduction in macroscopic conductance and approximately 2-fold slower activation kinetics. Remarkably, the interaction of KCNQ2/Q3 heteromeric channels with syntaxin 1A was significantly weaker and KCNQ3 homomeric channels were practically resistant to syntaxin 1A. Analysis of different KCNQ2 and KCNQ3 chimeras and deletion mutants combined with in-vitro binding analysis pinpointed a crucial C-terminal syntaxin 1A-association domain in KCNQ2. Pull-down and coimmunoprecipitation analyses in hippocampal and cortical synaptosomes demonstrated a physical interaction of brain KCNQ2 with syntaxin 1A, and confocal immunofluorescence microscopy showed high colocalization of KCNQ2 and syntaxin 1A at presynaptic varicosities. The selective interaction of syntaxin 1A with KCNQ2, combined with a numerical simulation of syntaxin 1A's impact in a firing-neuron model, suggest that syntaxin 1A's interaction is targeted at regulating KCNQ2 channels to fine-tune presynaptic transmitter release, without interfering with the function of KCNQ2/3 channels in neuronal firing frequency adaptation.

Figure 1. Syntaxin 1A interacts physically with KCNQ2, KCNQ2\3 and KCNQ1, but hardly interacts with KCNQ3 in oocytes.

A, Digitized Phosphorimager scan of SDS-PAGE analysis of [³⁵S] Met/Cys-labeled channels (KCNQ2; KCNQ2/3 and KCNQ3) and syntaxin 1A (syx) proteins coprecipitated by the corresponding antibodies from 1% Triton X-100 homogenates of whole oocytes, that were injected with the channels mRNA alone or with syntaxin 1A mRNA alone or coinjected with the channels and syntaxin 1A mRNAs (as indicated above the lanes). The protein samples were analyzed on an 8% gel. Arrows indicate the relevant proteins. B, KCNQ3 and KCNQ1 channels do not interact with syntaxin 1A as strongly as KCNQ2 in oocytes. Left panel: The channels and syntaxin proteins coprecipitated by the corresponding antibodies (as indicated below the lanes). Right panel: Reciprocal coimmunoprecipitation in oocytes from the same experiment, carried out using a monoclonal syntaxin 1A antibody (IP syx). C, Interaction of KCNQ2, KCNQ2\3, KCNQ3 and KCNQ1 with syntaxin 1A. Bars depict ratios (quantified by ImageQuant) of syntaxin to the channels, precipitated by the corresponding channel antibodies. Numbers in parentheses refer to number of oocyte batches. *p<0.05.

Figure 2. Syntaxin 1A modulates primarily the KCNQ2 currents.

A, Syntaxin 1A (syx) reduces current amplitudes of KCNQ2 (Aa) and KCNQ2\3 (Ba) channels, expressed in Xenopus oocytes, but not those of KCNQ3* (Ca). Representative current traces from single oocytes of the same batch injected with the channels mRNA alone or with syntaxin 1A mRNA (0.75 ng/oocyte; +syx). Inset: the voltage protocol used to elicit currents. B, Syntaxin 1A reduces the maximal conductances of KCNQ2 (Ab), and KCNQ2\3 (Bb), but not of KCNQ3* (Cb). Conductance-Voltage (G–V) relationships for the channels in the presence and absence of syntaxin 1A, normalized to the maximal conductance in the absence of syntaxin 1A or each normalized to itself (inset). G values were obtained from peak currents, assuming a reversal potential of −98 mV for K+ ions. Two component Boltzmann equation G/G_max = 1/(1+exp(−(V_1/2−V)/a), ,was fitted to the data. C, Syntaxin 1A slows down only the rate of activation of KCNQ2 (Ac) but not of KCNQ2\3 (Bc) or KCNQ3* (Cc). Inset: overlay of representative traces elicited at +5 mV showing the activation of KCNQ2 currents in the presence and absence of syntaxin 1A. The rising phase of the currents elicited at all denoted potentials was fitted by a bi-exponential function, deriving fast and slow time constants (τ fast and τ slow). Data in B and C were averaged from three oocyte batches with at least 5 oocytes per batch. *p<0.05, **p<0.01.

Figure 3. Reduction of currents by syntaxin is not associated with either total channel expression or plasma membrane (PM) content.

A, The amount of KCNQ2 (Q2) channel in PM is not affected by coexpression of syntaxin (syx). Data were obtained by measurements of confocal images in whole oocytes expressing KCNQ2 channel with external HA tag (obtained with an anti-HA antibody) or YFP tag, as indicated, alone or together with syntaxin. B, Summary of KCNQ2 PM expression and comparison of the two imaging methods. Gray bars show PM amount of KCNQ2-HA/YFP channels expressed alone. Black bars show the amount of KCNQ2-HA/YFP coexpressed with syntaxin. In both methods the PM expression level in the presence of syntaxin was normalized to the control group of oocytes expressing the channel alone. Numbers above lanes indicate the numbers of oocytes. C, The effect of coexpression of syntaxin on currents (I), corrected to the corresponding PM expression of KCNQ2-HA or KCNQ2-YFP, was measured from the same oocytes as in A. Currents were evoked by a voltage step from a holding potential of −95 mV to +5 mV and normalized to the control group of oocytes expressing the channel alone. Numbers above lanes indicate the numbers of oocytes. *p<0.05, **p<0.01. D, Syntaxin (syx) affects neither total protein expression nor PM content of KCNQ2. Digitized Phosphorimager scan of SDS-PAGE analysis of [³⁵S] Met/Cys-labeled KCNQ2 and syntaxin proteins, immunopurified from 110 plasma membranes (right panel; PM) or 10 internal fractions (left panel; I) of oocytes precipitated by KCNQ2 antibody. KCNQ2 was expressed alone (KCNQ2) or together with syntaxin (+syx) and protein samples were analyzed on an 8% gel. E, Histogram showing normalized amount of KCNQ2 (Q2; quantifies by ImageQuant) expressed with or without syntaxin, precipitated with KCNQ2 antibody from internal fractions (I) of oocytes. F, Histogram showing ratios (quantified by ImageQuant) of KCNQ2 amounts in plasma membranes versus internal fractions in oocytes expressing KCNQ2 alone or together with syntaxin, in the same experiments. Numbers above lanes indicate the numbers of experiments.

Figure 4. Syntaxin 1A binds preferentially to helix A in the C-terminus and is pivotal for the binding.

A, Schematic representation of the C-terminus of KCNQ2, in which the syntaxin 1A (syx) binding domain is indicated. B, Top: Interaction of syntaxin 1A with GST fusion proteins corresponding to different parts of the C-termini of KCNQ2 and KCNQ3. In vitro synthesized ³⁵S labeled syntaxin 1A was reacted with the indicated GST fusion proteins. Bottom: Coomassie blue (CB) staining of the protein gel. Numbers denote molecular weight markers. The bar diagram shows the normalized syntaxin 1A binding values. The intensity of the immunoreactive band of syntaxin 1A (syx) in different groups was normalized to the corresponding intensity of the coomassie blue (CB) staining of the peptide. The data were averaged from several independent experiments and includes the binding of syntaxin 1A to helices A, A+B and A+B+C. C, Top: Stochiometry of the binding of syntaxin 1A to helix A of KCNQ2, derived from binding curves that show saturation. Recombinant hexahistidine-tagged (His₆) cytoplasmic part of syntaxin 1A at the indicated concentrations was bound to immobilized GST-helix A (150 pmol) in a 1 ml reaction volume. Bound syntaxin 1A was determined by SDS-PAGE and immunoblotting with syntaxin 1A antibody (inset). ECL signal intensities were quantified with TINA software and converted to picomoles by the use of standard curves for the corresponding proteins. Bottom: Calibration gel which demonstrates the amount of syntaxin 1A coprecipitated in the experiment. Unbound recombinant hexahistidine-tagged (His₆) cytoplasmic part of syntaxin 1A at the indicated concentrations was loaded on an 8% gel and immunoblotted with syntaxin 1A antibody. D, helix A (aa 339–360) is required for syntaxin 1A's binding to KCNQ2. Oocytes were injected with syntaxin 1A mRNA alone or co-injected with syntaxin 1A and KCNQ2/Δ339–360 deletion mutant/Δ372–493 deletion mutant. The binding assay was performed as described.

Figure 5. Colocalization of KCNQ2 and syntaxin 1A at synaptic sites marked by VAMP-2 immunoreactivity in hippocampal neurons.

A, Immunocytochemistry experiments show colocalization (overlay, yellow) of KCNQ2 (red) and syntaxin 1A (syx; green) in rat hippocampal neurons. High colocalization areas of KCNQ2 and syntaxin 1A are indicated by arrows. B, Colocalization of KCNQ2, syntaxin 1A and VAMP-2 in rat hippocampal neurons as detected by triple immunocytochemistry and illustrated by the merge images. KCNQ2 (red), syntaxin 1A (green) and VAMP-2 (blue) are indicated in the top images from left to right. The bottom images from left to right show the colocalization of KCNQ2 and VAMP-2 (merge, pink), syntaxin 1A and VAMP-2 (merge, light blue) and KCNQ2 and syntaxin 1A (merge, yellow). A varicosity colocalized with VAMP-2, syntaxin 1A and KCNQ2 is indicated by arrow. C, The same image as in B showing all three markers; KCNQ2 (red), syntaxin 1A (green) and VAMP-2 (blue). A linescan was placed through the varicosity indicated by arrow in B. The varicosity was shown to colocalize all three signals and thus, is indeed a synaptic one.

Figure 6. Syntaxin 1A associates with KCNQ2 in cortical and hippocampal synaptosomes.

A, Syntaxin 1A (syx) coprecipitates with KCNQ2 from 2% Chaps synaptosomal lysate by anti-KCNQ2 antibody. Lysates were incubated with KCNQ2 antibody in the absence or presence (+peptide) of antigen peptide-HC (heavy chain). Numbers indicate molecular weight markers. B, GST-syntaxin 1A fusion protein “pulls down” KCNQ2 from synaptosomal lysates. Syx-GST (corresponding to the cytosolic part of syntaxin 1A) or GST immobilized on GSH-agarose beads (each at 150 pmoles) were incubated with 2% CHAPS lysate (200 µg) for 12 h at 4°C. Precipitated proteins were separated by SDS-gel (8% polyacrylamide) and immunoblotted with anti-KCNQ2 antibody (upper panel). The lower panel shows a Ponceau S staining of the blot, which demonstrates the equal protein loading of syntaxin 1A-GST and GST proteins. C, The binding of syntaxin 1A to helices A+B+C of KCNQ2 is stronger than its binding to the same helices in KCNQ3. Syntaxin 1A coprecipitated with KCNQ2 and KCNQ3 from 2% Chaps synaptosomal lysate by anti-KCNQ2/KCNQ3 antibody. Precipitated proteins were separated by SDS-gel (8% polyacrylamide) and immunoblotted with anti-syntaxin 1A antibody.

Figure 7. Simulating the effect of the interaction of syntaxin 1A, KCNQ2 and KCNQ2/3 on neuronal physiology.

A, A regular firing train of action potentials generated by a 2 nA current injection into a spherical neuron containing the Hodgkin-Huxley model (current step is shown below panel c and is similar for A, B, and C). B, Similar simulation to a containing, in addition to the Hodgkin-Huxley model, also a model of the KCNQ2 as described in the Materials and Methods and Table 1 at a density of 5 pS/µm². The scale bar in B applies also to A, C, E, F, and G. C, Similar simulation to a containing, in addition to the Hodgkin-Huxley model, also a model of the KCNQ2+syntaxin 1A (syx) as described in the Materials and Methods and Table 1 at a density of 2.5 pS/µm². D, Simulation of the effect of KCNQ2 on the action potential ADP in a single compartment model containing the Hodgkin-Huxley model and a T-type voltage-gated Ca²⁺ conductance (smooth line). Adding 5 pS/µm² KCNQ2 to the ADP model reduced the amplitude of the ADP (dotted line). Simulating the effect of the interaction of syntaxin 1A with KCNQ2 by using the relevant time constants from Table 1 and halving the maximal conductance generated a larger ADP (dashed lines). The scale bar in D applies also to H. E, Same as in B only with KCNQ2/3. F, Same as is c only with KCNQ2/3.