Alpha subunit-dependent glycine receptor clustering and regulation of synaptic receptor numbers

Accumulation of glycine receptors at synapses requires the interaction between the beta subunit of the receptor and the scaffold protein gephyrin. Here, we questioned whether different alpha subunits could modulate the receptors’ diffusion and propensity to cluster at spinal cord synapses. Using quantitative photoactivated localisation microscopy we found that alpha-1 and alpha-3 containing glycine receptors display the same α3:β2 stoichiometry and gephyrin binding. Despite these similarities, alpha-3 containing receptors are less mobile and cluster at higher density compared to alpha-1, with 1500 versus 1100 complexes µm−2, respectively. Furthermore, we identified a subunit-specific regulation of glycine receptor copy numbers at synapses: when challenged with interleukin 1β, the synaptic occupancy of alpha-1 but not alpha-3 receptors was reduced. This mechanism may play a role in the cell-type dependent regulation of glycinergic currents in response to interleukin 1β and highlights the capacity of the alpha subunits to affect receptor-gephyrin binding at synapses.

(Sigma) was applied in culture medium for 15 minutes at a final concentration of 10 ng/ml prior to fixation and immunolabelling, or added to the external solution for electrophysiological recordings.

Single-particle tracking
The diffusion of Dendra2-tagged GlyRs was measured by single-particle tracking (SPT) using quantum dots (QDs) as described previously 4 . Transfected COS-7 cells or infected spinal cord neurons on glass coverslips were labelled sequentially with Dendra2 antibody (Antibodies-online ABIN361314, 1:10000), biotinylated anti-rabbit F(ab') 2 fragments (Jackson, 1:400) and streptavidin coated QDs emitting at 655 nm (Invitrogen Q10121MP, 1:1000), and imaged in MEM imaging medium (phenol red-free minimal essential medium, 20 mM HEPES, 2 mM   Table S1). This is in line with the absence of endogenous GlyRb subunits in these neurons 5 . Overexpression of recombinant GlyRs in these neurons decreased endogenous gephyrin levels at synapses compared to non-infected controls (MW p = 0.0002 for a1 and p < 0.0001 for a3).  Table S1. Single fluorophore analysis of GlyR stoichiometry.

COS-7 cells, rat spinal cord and hippocampal neurons expressing Dendra2-tagged GlyR subunits
(generally from ≥ 6 cells and 3 independent experiments) were used for quantitative PALM.
Bursts of detections (raw counts) were counted in the time traces of isolated clusters of detections as described in the Methods section (excluding synaptic clusters). The known homopentameric stoichiometry of GlyRa subunits was used to fit the raw counts with a binomial distribution of n = 5, yielding the probability of detection p det of the Dendra2 fluorophore. The goodness of fit was assessed using a chi-square test. The average p det = 0.44 was then applied to the normalised burst frequencies in order to identify the subunit stoichiometries (n = 2 to 5) associated with the smallest residuals (shown in bold). Table S2. Single fluorophore counting of gephyrin complexes.
Using PALM imaging, we counted the number of bursts (raw counts) in the time traces of isolated clusters of detections in COS-7 cells and in the non-synaptic regions of mouse spinal cord neurons expressing mEos2-gephyrin (from ≥ 6 cells and 3 independent experiments). Counts were fitted with a binomial distribution of n = 3 or 6. The trimeric stoichiometry of gephyrin was validated using a chi-square test and by taking into account the associated probability of detection p det of mEos2.