WNT7A-positive dendritic cytonemes control synaptogenesis in cortical neurons

Abstract

Synaptogenesis involves the transformation of dendritic filopodial contacts into stable connections with the exact apposition of synaptic components. Signalling triggered by Wnt/β-catenin and calcium has been postulated to aid this process. However, it is unclear how such a signalling process orchestrates synapse formation to organise the spatial arrangement of synapses along dendrites. We show that WNT7A is loaded on dynamic dendritic filopodia during spine formation in human cortical neurons. WNT7A is present at the tips of the filopodia and the contact sites with dendrites of neighbouring neurons, triggering spatially restricted localisation of the Wnt co-receptor LRP6. Here, we demonstrate that WNT7A at filopodia tips leads to the induction of calcium transients, the clustering of pre- and postsynaptic proteins, and the subsequent transformation into mature spines. Although soluble WNT7A protein can also support synaptogenesis, it fails to provide this degree of spatial information for spine formation and calcium transients, and synaptic markers are induced ectopically along the dendrites. Our data suggest that dendritic filopodia are WNT7A-bearing cytonemes required for focal calcium signalling and initiation of synapse formation, and provide an elegant mechanism for orchestrating the positioning of synapses along dendrites.

Keywords: IPSC-derived human cortical neurons; Lattice structured illumination microscopy; Super-resolution imaging; Synapse formation; WNT7A cytonemes.

Fig. 1.

Characterisation of dendritic filopodia as WNT7A-carrying signalling filopodia. (A) Schematic of iPSC-derived cortical neuron cultures at 60 days in vitro (DIV 60), transfected with membrane marker and co-stained with antibodies to observe localisation of WNT7A, LRP6, and classical filopodia/cytoneme markers. Blue circles denote areas of opposing localisation of proteins across membranes, yellow circles denote proper protein colocalisation, and orange circles no colocalisation. (B) Super-resolution imaging of iPSC-derived cortical neurons stained with anti-WNT7A antibody shows localisation of the protein to dendritic protrusions. Boxed areas are shown at higher magnification on the right. (C,D) Quantification (D) of dendritic protrusion types (filopodia, stubby, mushroom; C) found that nearly 75% of total protrusions were WNT7A positive with no difference in the distribution of WNT7A protein across the class of protrusion. (E) Super-resolution microscopy of neurons co-stained with antibodies against WNT7A and LRP6 shows localisation of WNT7A on dendritic protrusions, whereas LRP6 generally localises at opposing membranes (E, blue circles). WNT7A-positive protrusions also harbour proteins associated with Wnt-signalling filopodia, such as flotillin 2 (FLOT2; F, yellow circles) and Wntless/evenness interrupted (WLS; G, yellow circles), with over 50% of WNT7A-positive protrusions colocalising with these cytoneme markers. Conversely, the Wnt-signalling filopodia marker Myo-10 localises to a WNT7A-negative subset of protrusions (I, orange circles). FLOT2 and WLS colocalise on the same protrusions (H, yellow circles). (J) Quantification of LRP6 and WNT7A co-clustering at apposing membranes identified a significant increase in co-clustering of WNT7A-positive mushroom-shaped protrusions and LRP6 compared to co-clustering at filopodia contacts. (K) Quantification of protrusion number based on cytoneme markers found significantly more WNT7A/FLOT2-positive and WNT7A/WLS-positive protrusions compared to WNT7A-only positive protrusions, and 74% of all protrusion analysed were WNT7A positive. (L,M) Quantification of protrusion density (L) and length (M) found no difference based on the colocalisation of WNT7A with cytoneme protein markers. Statistical significance was addressed using one-way ANOVA with Dunnett's multiple comparison test to compare relevant controls within groups. *P<0.05. ns, not significant.

Fig. 2.

Lipidation of Wnt and calcium are required for colocalisation of WNT7A and synaptic markers. (A-C) Antibody staining for WNT7A and LRP6, WNT7A and PSD95, or WNT7A and bassoon (BSN) in iPSC-derived cortical neurons in the absence (A) or presence of the porcupine inhibitor IWP-2 (B), or the calcium chelator EGTA (C). Phn, phalloidin. (D) Quantification of the co-clustering or colocalisation of WNT7A/LRP6, WNT7A/PSD95, and WNT7A/BSN shows dependence on intracellular Wnt trafficking and calcium signalling. Statistical significance was addressed using one-way ANOVA with Dunnett's multiple comparison test to compare relevant controls within groups. ***P<0.005; ****P<0.001.

Fig. 3.

Membrane-tethered WNT7A-GFP can cluster Wnt and synaptic components. (A) Schematic of the tethering experiment using double transfection of WNT7A-GFP and the morphotrap nanobody construct, variable heavy domain of the heavy chain (Vhh)-CD8-mCherry, which binds GFP-tagged proteins. (B,C) AGS cells transfected with soluble-GFP (secGFP) and mem-mCherry (B) and AGS cells transfected with secGFP and morphotrap (C). Orange arrow indicates membrane localisation of secGFP. (D) Quantification identifies a significant colocalisation between the normally secreted secGFP and the morphotrap at the plasma membrane. An unpaired, one-tailed Student's t-test (****P<0.001). (E) The colocalisation of mem-mCherry and WNT7A-GFP on protrusions is shown in white. (F,G) Significantly higher WNT7A-GFP colocalisation on filopodia is observed when co-transfected with morphotrap (F), quantified by Pearson's correlation coefficient (PCC; G). Unpaired, one-tailed Student's t-test (*P<0.05). (H,H′) Super-resolution imaging of iPSC-derived cortical neurons transfected with mem-GFP and mem-mCherry, followed by post-staining for LRP6 and the actin cytoskeleton (phalloidin, Phn). (I,I′) Transfection with WNT7A-GFP and mem-mCherry, followed by post-staining for LRP6 and the actin cytoskeleton identified WNT7A-GFP-positive protrusions that cluster LRP6 in apposed cells (I′, yellow circle). (J,J′) Transfection with WNT7A-GFP and morphotrap, followed by post-staining for LRP6 and cytoskeleton, identified protrusions harbouring membrane-tethered WNT7A-GFP that cluster LRP6 in apposed cells (J′, yellow circles). (K) Quantification of co-clustering of LRP6 with WNT7A-GFP identified a significant increase in both the morphotrap and mem-mCherry conditions compared to mem-Ch alone. Unpaired, one-tailed Student's t-test (*P<0.05, **P<0.01). (L,L′) Transfection with mem-GFP and mem-mCherry, followed by post-staining for PSD95 and the actin cytoskeleton (Phn). (M,M′) Transfection with WNT7A-GFP and mem-mCherry, followed by post-staining for PSD95 and actin, revealed colocalisation on dendritic protrusions (M′, yellow circle). (N,N′) WNT7A-GFP co-transfected with morphotrap and colocalised with PSD95 on dendritic protrusions (N′, yellow circle). (O) Quantification of the colocalisation of PSD95 with WNT7A-GFP also identified a significant increase in both the morphotrap and mem-mCherry conditions compared to mem-Ch alone. Experiments were performed on three independent biological replicates. A minimum of three transfected neurons were quantified per condition. Statistical significance was addressed using one-way ANOVA with Dunnett's post-hoc test for multiple comparisons, comparing groups to the mem-mCherry/mem-GFP control group. *P<0.05. In H-J,L-N, boxed areas indicate the area shown at higher magnification in H′-J′,L′-N′.

Fig. 4.

Calcium transients on dendritic protrusions depend on membrane-associated Wnts. (A) Schematic of the generated mCherry-tagged membrane-tethered NOTUM construct (memNOTUM). mCh, mCherry. (B) Co-transfection of the membrane marker mem-mCherry and WNT7A-GFP shows localisation of WNT7A-GFP to cytoneme tips (yellow circles), whereas co-transfection of memNotum and WNT7A-GFP reduces WNT7A-GFP-positive filopodia. (C) Quantification of mCherry nuclear expression in the receiving cell population identified the ability of memNotum to significantly reduce Wnt-mediated paracrine signalling to a similar level as secreted NOTUM (transfection of full-length untethered human NOTUM). Statistical significance was addressed using one-way ANOVA with Dunnett's multiple comparison test to compare relevant controls within groups and an unpaired, one-tailed Student's t-test to compare specific combinations. *P<0.05; **P<0.01; ***P<0.005. ns, not significant. Experiments were performed in biological triplicate, with three fields per group analysed. (D,E) SH-SY5Y differentiated neurons transfected with WNT7A-GFP and mem-mCherry followed by post-staining with anti-LRP6 show strong colocalisation (D), which is reduced when neurons are transfected with memNotum (E). (F-H) SH-SY5Y neurons transfected with FingR-PSD95 and memNotum and post-stained with anti-WNT7A also show reduced colocalisation of the PSD95/WNT7A proteins compared to the mem-mCherry control (F,G) signal. Quantification is shown in H. signal. Two-way ANOVA with Bonferroni's post-hoc test for multiple comparisons was performed for statistical comparisons between each group. **P<0.01.

Fig. 5.

Effect of Wnt ‘scissor’ on calcium transient induction at filopodia contact sites. (A) Schematic of calcium transient analysis from live-cell imaging studies performed on the Elyra7 L-SIM2 system, with (1) showing the analysis of filopodia contacts to near neighbouring neurites leading to (2) calcium transients measured in a 2.5 μm radius of the contact point. (B,C) Representative frames from live imaging of iPSC-derived cortical neurons transfected with either mem-mCherry and mem-gCaMP7s performing continuous calcium imaging (1-min bursts) in HILO mode to obtain 10 μm optical sections (min:s:ms) (B) shows classic calcium transient signals. Quantification is shown in C. (D,E) When cells were transfected with membrane-tethered NOTUM (memNotum) and mem-gCaMP7S (D) calcium transients were lost (E). (F-H) Representative traces from calcium transients occurring specifically on filopodia identified increases in transient amplitudes when either overexpressing WNT7A (G) or when exogenously applying WNT7A (H). (I-K) A reduction in calcium transient amplitude and frequency were observed when cells were co-transfected with membrane-bound NOTUM (memNotum) at basal levels, but also when either transfected with WNT7A (O.E. WNT7A; J) or after exogenous application of recombinant WNT7A (K). Live calcium imaging experiments were performed in three independent batches, with at least two independent transfections per condition. Filopodial calcium transients were analysed on at least eight filopodia/condition/batch. The dendritic calcium transients and protrusion spacing were analysed on a subset of this dataset (nine fields from three independent experiments).

Fig. 6.

Comparison of overexpressed and exogenously applied WNT7A in the induction of local calcium signalling and spine spacing. (A) Quantification of calcium transient amplitude identified significant increases compared with control if neurons overexpressed WNT7A (O.E. WNT7A) or if the soluble protein was exogenously applied to the culture (recWNT7A). These increases were significantly attenuated when cells were co-transfected with memNotum. (B) A significant calcium amplitude peak prominence increase was observed in cells overexpressing WNT7A (O.E. WNT7A). (C) Calcium transient width was significantly altered in cultures expressing memNotum or after exogenous application of WNT7A (recWNT7A), compared to control or O.E. WNT7A groups. (D) Quantification of the dendritic calcium transient intensity identified a significant increase when comparing the overexpression of WNT7A to the exogenous application of recombinant WNT7A. (E) Representative images from iPSC-derived cortical neuron cultures either transfected with mem-mCherry and WNT7A or with mem-mCherry plus exogenous application of recombinant (rec.) WNT7A, followed by analysis of protrusion spacing along dendrites. (F) Quantification of the distance between dendritic protrusions identified a significant reduction in spacing when exogenously applying recombinant WNT7A, compared to overexpression of the protein. For statistical analysis of the calcium transient data, we performed direct comparisons within between groups using an unpaired, two-sided Student's t-test (**P<0.01, ****P<0.001), and used Tukey's Honestly Significant Difference (HSD) test with Holm–Bonferroni adjustments for multiple comparisons. We employed letter-based groupings for a compact representation of the significant differences amongst treatment groups. Groups not statistically different from one another share the same letter, while significantly different groups are designated distinct letters.

Fig. 7.

Membrane-tethered Wnts are required for synaptogenesis. (A-C) Cortical neurons were transfected with FingR-PSD95-GFP to tag PSD95 endogenously. Control neurons were co-transfected with mem-mCherry and showed strong PSD95 puncta localising to protrusions (A), which was lost when neurons were instead co-transfected with memNotum (B) or when control cultures were incubated with soluble, recombinant NOTUM (C). (D) Puncta expression could be rescued in memNotum transfected neurons by pre-treatment with the NOTUM inhibitor LP-922056. (E,F) Overexpression of WNT7A (O.E. WNT7A) did not alter PSD95 localisation to protrusions (E), whereas recombinant WNT7A (Rec. WNT7A) increased protrusion number (F). (G) Quantification of protrusion number in each group. (H) Quantification of PSD95-positive protrusions in each group. (I-O) Quantification (I) of ectopic PSD95 puncta on dendritic arborisation. Neurons were co-stained for PSD95 and bassoon (BSN) after transfection with either mem-mCherry (J), memNotum (K), mem-mCherry+soluble Notum (L), memNotum+LP-922056 (M), mem-mCherry+O.E. WNT7A (N), or mem-mCherry+Rec. WNT7A (O). (P) Quantification of PSD95/BSN puncta on transfected neurons. Statistical significance was addressed using one-way ANOVA with Dunnett's multiple comparison tests to compare relevant controls within groups and an unpaired, two-tailed Student's t-test to compare specific combinations. *P<0.05; **P<0.01; ***P<0.005. Experiments were performed in biological triplicate, with at least five fields per group analysed.