Selective vulnerability of tripartite synapses in amyotrophic lateral sclerosis

Abstract

Amyotrophic Lateral Sclerosis (ALS) is a fatal neurodegenerative disorder. Separate lines of evidence suggest that synapses and astrocytes play a role in the pathological mechanisms underlying ALS. Given that astrocytes make specialised contacts with some synapses, called tripartite synapses, we hypothesise that tripartite synapses could act as the fulcrum of disease in ALS. To test this hypothesis, we have performed an extensive microscopy-based investigation of synapses and tripartite synapses in the spinal cord of ALS model mice and post-mortem human tissue from ALS cases. We reveal widescale synaptic changes at the early symptomatic stages of the SOD1^G93a mouse model. Super-resolution microscopy reveals that large complex postsynaptic structures are lost in ALS mice. Most surprisingly, tripartite synapses are selectively lost, while non-tripartite synapses remain in equal number to healthy controls. Finally, we also observe a similar selective loss of tripartite synapses in human post-mortem ALS spinal cords. From these data we conclude that tripartite synaptopathy is a key hallmark of ALS.

Keywords: ALS/MND; Astrocyte; Neurodegeneration; Synapse.

Fig. 1

Mapping excitatory synapses in the spinal cord revealed early symptomatic stage synaptic changes in SOD1^G93a mice. a High-resolution mapping of PSD95-eGFP was performed to analyse synaptic changes in SOD1^G93a mice at different time points of the disease. b Example images of postsynaptic puncta in control and SOD1^G93a mice at different time points of disease. c Heat maps of PSD density in male and female SOD1^G93a mice compared to their respective controls at different ages. d Heat maps of PSD size in male and female SOD1^G93a mice compared to their respective controls at different ages. e Chart plotting PSD density in all 8 laminae of 16-week-old controls, WTSOD1 controls and SOD1^G93a males. Sample sizes denoted as n = number of WTSOD1’s/number of controls/number of SOD1^G93a’s. f Chart plotting PSD size in all 8 laminae of 16-week-old controls, WTSOD1 controls and SOD1^G93a males. g Chart plotting PSD density in all 8 laminae of 16-week-old controls, WTSOD1 controls and SOD1^G93a females. h Chart plotting PSD size in all 8 laminae of 16-week-old controls, WTSOD1 controls and SOD1^G93a females

Fig. 2

Nanoscopic Changes in Substructural Organisation of PSD95 in ALS Synapses. a Diagram of synapse subtypes, defined by the number of PSD95 nanoclusters (NCs) per PSD. PSDs can be visualised using confocal microscopy while NCs are resolved from correlative g-STED images. b Spinal cord tissue from control and SOD1^G93a mice were imaged using confocal and g-STED microscopy to visualise PSDs and NCs, respectively. c Graph showing the size of the PSD95 NCs was not different between controls and SOD1^G93a mice. d Graph showing the average number of NCs per PSD was significantly reduced in SOD1^G93a mice. e Graph showing the number of large 3 + NC-PSDs are specifically reduced in SOD1^G93a mice compared to controls. f High-resolution imaging of synapses (PSD95 and VGLUT2) and the palmitoylation enzyme, zDHHC2, was performed to investigate whether changes in zDHHC2 expression might explain the reduced number of PSD95 NCs at synapses. g Graph showing no difference in zDHHC2 puncta density between control and SOD1^G93a mice. h Graph showing reduced number of PSDs expressing zDHHC2 in SOD1^G93a mice

Fig. 3

Mapping Tripartite Synapses in the SOD1^G93a Mouse Spinal Cord. a Example of high-resolution spinal cord mapping of tripartite synapses from 16-week-old male control and SOD1^G93a mice using PSD95-eGFP, VGLUT2 and p-Ezrin to label the PSDs, presynaptic terminals and PAPs, respectively. Cropped images display individual synaptic and tripartite synaptic structures (denoted by arrows). b Example of high-resolution spinal cord mapping of tripartite synapses from 16-week-old male control and SOD1^G93a mice using PSD95-eGFP, VGLUT2 and EAAT2 to label the PSDs, presynaptic terminals and PAPs, respectively. Cropped images display individual synaptic and tripartite synaptic structures (denoted by arrows). c Colour coded heat maps denote the difference in synapse or tripartite synapse density observed in SOD1^G93a mice as a percentage of the respective controls. Bar charts display the mean synapse density from mice, averaged across all anatomical subregions for simplicity. From p-Ezrin labelling, no changes in non-tripartite or tripartite synapses were observed in 8 week SOD1^G93a mice. d From EAAT2 labelling, no changes in non-tripartite synapses were observed, but a significant increase in tripartite synapses was observed in 8 week SOD1^G93a mice compared to control. e From p-Ezrin labelling, no changes in non-tripartite or tripartite synapses were observed in 12 week SOD1^G93a mice. f From EAAT2 labelling, no changes in non-tripartite synapses were observed, but a significant decrease in tripartite synapses was observed in 12 week SOD1^G93a mice compared to control. g From p-Ezrin labelling, no changes in non-tripartite synapses were observed, but a significant decrease in tripartite synapses was observed in 16 week SOD1^G93a mice compared to control. h From EAAT2 labelling, no changes in non-tripartite synapses were observed, but a significant decrease in tripartite synapses was observed in 16 week SOD1^G93a mice compared to control

Fig. 4

Tripartite Synapses are selectively lost in the cervical spinal cord of human ALS cases. a Diagram of human cervical spinal cord and location of image sampling. b Example images of synapses (PSD95—green) and astrocytic processes (p-Ezrin—magenta) in human spinal cord tissue. c Graph showing the total PSD density (PSD95 puncta) in controls, all ALS cases, and then SOD1 and C9orf72 cases separated. d Graph showing synapse density in patient groups when synapses are separated based on whether they are non-tripartite synapses or tripartite synapses, as determined by the association of PSD95 with p-Ezrin