Gabapentin receptor alpha2delta-1 is a neuronal thrombospondin receptor responsible for excitatory CNS synaptogenesis

Abstract

Synapses are asymmetric cellular adhesions that are critical for nervous system development and function, but the mechanisms that induce their formation are not well understood. We have previously identified thrombospondin as an astrocyte-secreted protein that promotes central nervous system (CNS) synaptogenesis. Here, we identify the neuronal thrombospondin receptor involved in CNS synapse formation as alpha2delta-1, the receptor for the anti-epileptic and analgesic drug gabapentin. We show that the VWF-A domain of alpha2delta-1 interacts with the epidermal growth factor-like repeats common to all thrombospondins. alpha2delta-1 overexpression increases synaptogenesis in vitro and in vivo and is required postsynaptically for thrombospondin- and astrocyte-induced synapse formation in vitro. Gabapentin antagonizes thrombospondin binding to alpha2delta-1 and powerfully inhibits excitatory synapse formation in vitro and in vivo. These findings identify alpha2delta-1 as a receptor involved in excitatory synapse formation and suggest that gabapentin may function therapeutically by blocking new synapse formation.

Figure 1. All thrombospondin isoforms are synaptogenic

(A) TSPs are divided into two subgroups. N-terminal domain (black), the procollagen (red) and properdin-like repeats (orange), EGF-like repeats (blue), calcium binding repeats (grey) and C-terminal L-lectin like globular domain (green). (B) Immunostaining of RGCs for synaptotagmin (red) and PSD-95 (green). White arrows point to co-localized synaptic puncta. Scale bar =30µm. (C) Quantification of the effects of astrocytes, purified TSP1,4 and 5 (8nM each) and (D) conditioned media from COS7 cells overexpressing either TSP3 or empty vector, on synapse number. In all graphs n=20 cells, Error bars mean ± SEM, * p<0.05.

Figure 2. EGF like repeats of TSPs are synaptogenic

(A) The domain structure of TSP1 and 2. N-terminal domain (black), oligomerization domain and a procollagen repeat (red, PC), three properdin-like (TSP type 1, orange, P1-3), three EGF-like (TSP type 2, blue, E1-3), and 13 calcium binding (TSP type 3, grey) repeats (Ca(wire)) and a C-terminal L-type lectin like globular domain (green, C). (B) Quantification of the effect of TSP1 and (C) TSP2 fragments on synapse number. RGCs were treated with astrocytes, full-length TSP1 or a panel of TSP1 or TSP2 fragments (8nM each). (D) Location of epitopes targeted by TSP blocking antibodies (modified from (Carlson et al., 2008)). Inset shows magnified structure of EGF-like repeats and the Ca binding wire region and the C-terminal L-lectin like domain. Highlighted domains indicate putative synaptogenic domain of TSP. (E) Quantification of the effect of monoclonal anti-TSP antibodies on TSP’s synaptogenic activity. In all graphs n=20 cells, Error bars mean ± SEM, * p<0.05.

Figure 3. Thrombospondins interact with α2δ-1

(A) Array tomography analysis of synaptic localization of α2δ–1 in cerebral cortex. RGCs were immunostained for Synapsin I (blue) and for MAGUK (green). α2δ-1 puncta (red) associate both with synapses (white circles) or with isolated presynaptic (diamonds) or post-synaptic (squares) puncta. Scale bar =2 µm. (B) Western-blot analysis of α2δ-1 on the immunoprecipitation (IP) fractions performed using antibodies specific to TSPs1, 2 or 4 as well as calcium channel α1C (Cav1.2), or Agrin (positive and negative controls for IP, respectively). (C) Western blot analysis of α2δ-1 interaction with the synaptogenic domain of TSP2 (SD2). Left panel, HEK293 cell lysates from non-transfected (1), α2δ-1-FLAG alone (2), SD2 alone (3), α2δ-1-FLAG and SD2 (4) and α1δ-1FLAG and Control-myc-His construct (5) transfected cells. SD2 and Control-his-myc protein are marked with red ●. Anti-his antibody cross-reacts with several histidine rich proteins in HEK293 cell lysates (marked with blue *). Anti-α2δ-1 antibody also weakly recognizes the human α2δ-1 expressed endogeneously in HEK293 cells at low levels (blue ♦) Right panel; anti-FLAG IP fractions from α2δ-1-FLAG alone (6), SD2 alone (7), α2δ-1-FLAG and SD2 (8) and α1δ-1FLAG and Control-myc-His construct (9) transfections. (D) Domain structure of α2δ-1 protein and scheme of α2δ-1 protein C (PC) tagged constructs. SP=signal peptide, vWA_N and VGCC_a2=putative domains of unknown structure. Yellow boxes indicate prutative helical regions where no domain has yet been predicted. The red box shows the transmembrane (TM) region. Orange hexagons indicate predicted N-glycosylation sites, and purple bars indicate positions of cysteines. (E) SD2 interacts with the VWF-A domain of α2δ-1. Lane 1 is non-transfected HEK293 cell lysate. SD2 was co-expressed with PC tagged full-length α2δ-1, α2 only or VWF-A only constructs (lanes 3, 4 and 5) as well as CXCR4. SD2 co-immunopurified with the α2 only (8) and VWF-A (9) only constructs of α2δ-1 as well as the full-length protein (7) using anti-PC beads, (red arrows). SD2 did not co-purify with CXCR4 (6).

Figure 4. α2δ-1 is the TSP receptor involved in synaptogenesis

(A) RGCs were transfected with empty vector (pcDNA3, Invitrogen) or pcDNA3 constructs that express full length α2δ-1, δ-1 only or α2δ-1-Adh. The synapses received by transfected cells (marked by GFP co-expression) were then quantified. n=20 cells, Error bars mean ± SEM, * p<0.05. (B) Quantification of the effects of monoclonal antibodies 5A5 and 3B4 (mouse monoclonals raised against the VWF-A domain of α2δ-1, Mazorx Inc.) and anti-Thy1 antibody OX7 in synapse formation. 5A5 and 3B7 mimic TSP’s synaptogenic function. n=30 cells, Error bars mean ± SEM, *p<0.05. (C) Western blot analysis of cell lysates from HEK293 cells, which were co-transfected with an expression vector for rat α2δ-1 and siControl or siα2δ-1 pools, with a monoclonal antibody against α2δ-1 or against β-actin. (D) Immunostaining of siRNA transfected RGCs (marked blue by GFP co-expression) for co-localization of synaptotagmin (red) and PSD-95 (green). RGCs that were transfected with siα2δ-1 did not form many synapses even in the presence of astrocytes (See inserts i versus ii). Scale bars =30µm. (E) Quantification of the effects of siRNA pools on astrocyte and TSP-induced synapse formation in RGCs. n=20 cells, Error bars mean ± SEM, * p<0.05. (F) Overexpression of human α2δ-1, which is resistant to siα2δ-1(9), rescues the inhibition of SD2-induced synapse formation by siα2δ-1(9). n=20 cells, Error bars mean ± SEM, * p<0.05.

Figure 5. α2δ-1 overexpression in vivo increases excitatory synapse number

(A) Immunolabeling of cortices from littermate wildtype (WT) and α2δ-1 overexpressing transgenic (TG) P21 mice for VGlut2 and PSD95. Number of co-localized VGlut2/PSD95 puncta (white arrows in inlays i and ii) was higher in the TGs then the WTs. Scale bars = 20µm. (B) Quantification of VGlut2/PSD95 co-localization in brain sections from WT and TG mice. (*p<0.05). (C) Representative raw data traces of mEPSCs from layer IV cortical pyramidal neurons recorded from a WT and α2δ-1 TG mouse. Top, condensed trace. Bottom, expanded trace. (D) Summary of the frequency of mEPSCs in layer IV cortical pyramidal neurons of α2δ-1 TG and WT. TG = 3.5±0.3Hz (n=11 cells), WT = 2.1±0.2Hz (n=12 cells), p=0.002. (E) Summary of the amplitude of mEPSCs in layer IV cortical pyramidal neurons of α2δ-1 TG and WT. TG = 11.9±0.1pA, WT = 11.6±0.1pA, p=0.1.

Figure 6. Gabapentin inhibits TSP/astrocyte induced synapse formation

(A) Immunostaining for synaptotagmin (red) and PSD-95 (green) in RGCs treated with SD2 in the presence or absence of GBP Scale bars = 30µm. (B) Quantification of the effect of GBP on astrocyte or TSP-induced synapse formation and (C) on SD2-induced synapse formation. GBP blocks SD2’s synaptogenic effect when added simultaneously with SD2 but not when added after synapses have formed. n=20 cells, Error bars mean ± SEM, * p<0.05. (D) Western Blot analysis of effect of GBP on the SD2- α2δ-1 interaction. Red arrows point to SD2 protein co-immunoprecipitated with α2δ-1FLAG. Anti-α2δ-1 antibody also recognizes weakly expressed endogenous human α2δ-1 expressed by HEK293 cells (lanes 2 and 5, top blots). (E) Quantification of co-localization of VGlut2 and PSD95 in brain sections from saline and GBP injected P7 mice (*p<0.05). (F) Immunolabeling of saline and GBP injected P7 cortices for VGlut2 (green) and PSD95 (red). Half of GBP injected mice had a very strong reduction in the number, size and co-localization of synaptic puncta (white arrows, inlays i versus ii). Scale bars = 20µm.

Figure 7. TSP induced synapse formation is involved in barrel cortex plasticity

(A) Schematic presentation of the experimental paradigm: ablation of the C-row of whiskers at P1 causes corresponding reorganization of barrel representations at P7 in contralateral hemisphere. (B Immunolabeling of thalamocortical afferents to the barrel cortex with an antibody against 5HT transporter. Left images show control barrel cortex. Right images are representative examples of lesion-induced plasticity following whisker follicle ablation in mice that were injected with saline (top), with GBP (middle). Bottom row are control (left) and lesioned (right) barrel cortices from a TSP1/2KO mouse. Arrows flank the C-row of barrels corresponding to lesioned whiskers. Brackets and dashed lines show the expansion of D-row barrels. Asterisks denote regions of abnormal lesion-induced plasticity. (C) Hematoxylin staining of the whisker pads from mice whose barrels are shown in (B) showing selective ablation of C row follicles.