Without the millions of thriving synapses that link them, neurons in the brain are useless. The crumbling of these tiny connective structures is widely viewed as the proximal step to cognitive impairment in Alzheimer’s disease. Yet, researchers struggle to define exactly how and when synapses start to malfunction along the trajectory of the disease in the human brain. At this year’s virtual Alzheimer’s Association International Conference (AAIC), researchers described new techniques, such as mass synaptometry and digital spatial profiling, that reveal how synaptic changes relate to the clinical manifestations of the disease.
- Mass synaptometry, digital spatial profiling parse synaptic signatures in AD versus resilient human brain.
- Synaptic p-tau and neuroinflammation mark the difference between resilience versus vulnerability.
- Imaging and fluid biomarkers of AD pathology, synaptic integrity, and neuroinflammation correlate with cognitive status.
They reported that some synapses in the Alzheimer’s brain are riddled with phospho-tau and markers of cellular stress, while neurons in the brains of resilient fellow citizens have managed to exclude pathological tau from the synaptic sanctum, fending off damaging neuroinflammation. Other researchers brought imaging and fluid biomarkers of synaptic integrity into the growing AD biomarker toolkit, with the hopes of more accurately predicting cognitive decline.
In his talk, Thomas Montine of Stanford University spelled out the problem: Even though synaptic degeneration is a central event in AD, scientists lack powerful methods to study large numbers of individual synapses in the human brain. They can use high-resolution techniques such as electron microscopy to see small numbers of synapses, or they can analyze the contents of millions of synapses in bulk via synaptosome preps. Montine and others use fluorescence-based flow cytometry to investigate what’s inside individual synapses. However, fluorescence-based techniques can only analyze a handful of proteins at once.
To overcome these limitations, Montine and colleagues developed “mass synaptometry” (Gajera et al., 2019). Adapted from CyTOF—a technique that simultaneously detects many proteins at the single-cell level—mass synaptometry merges time of flight (TOF) mass spectrometry with flow cytometry. Essentially, instead of tagging antibodies with fluorescent molecules, researchers label them with heavy metals, each with a distinct mass that can be detected by mass spec. In so doing, they can find dozens of proteins of interest within individual synaptosomes.
Montine showed results from an analysis of 39 synaptic proteins, including those specific for certain cell and synapse types, as well as AD- and PD-related ones. The researchers isolated synaptosomes from postmortem sections of the caudate putamens, hippocampi, and prefrontal cortices of seven cognitively normal controls, nine people who had died with AD, and seven with dementia with Lewy bodies (DLB). An analysis of about 500,000 synaptosomes per sample yielded some striking trends.
Compared with hippocampal synapses from controls or DLB, synapses from AD brain contained an overabundance of p-tau. In contrast, Aβ levels in synapses of the prefrontal cortex—a region littered with Aβ plaques in AD—were no different between the groups.
Using a statistical cluster analysis called CITRUS to identify AD-specific patterns of protein expression in hippocampal synapses, Montine defined three major AD-related synapse types in AD brains: One type had a reduction of VGLUT receptors; another had an increase in p-tau; and a third was chock-full of proteins that denote cell stress and injury, including APP, BIN1, activated Caspase 3, 3NT, and K48. Notably, these three clusters of synapses were distinct from each other, and were absent in people with LBD.
The findings demonstrate a diversity of synaptic changes that take place in the AD brain, which bulk methods would not have detected, Montine said, adding that his group is now deploying machine learning to understand this diversity in more samples.
Data by Jamie Walker of the University of Texas Health Science Center, San Antonio, also placed the synapse at the center of AD. Walker analyzed proteins in individual cells with a multiplexing approach. Rather than focusing on synapses, she used digital spatial profiling to measure 84 proteins within selected cells in fixed hippocampal tissue. Using antibodies conjugated to UV-photocleavable oligonucleotides, DSP allows researchers to measure proteins in precise spatial locations within a tissue. DSP is most widely used to type cancers.
Walker measured neuronal proteins in hippocampal sections from six people who had died with AD and eight people with significant AD pathology who had been cognitively normal at death. This latter group, considered “cognitively resilient,” represents more than a third of samples from postmortem studies, and researchers urgently want to know the mechanisms that drive their resilience.
First, Walker compared, within a given brain, the protein profiles of tangle-bearing neurons to those without tangles in them. Besides detecting multiple species of phospho-tau in the tangle-bearing neurons, Walker also found elevated levels of proteins involved in APP processing and Aβ degradation, including PSEN1, BACE1, ADAM10, neprilysin, and insulin-degrading enzyme (IDE). This would suggest a link between Aβ and tau deposition in the same cell, Walker said.
Curiously, tangle-bearing neurons from resilient brains had a smaller uptick in these proteins, including phospho-tau species, than did tangle-bearing neurons from AD brains. In fact, only one protein was more abundant in tangle-bearing neurons from resilient brains: the presynaptic marker synaptophysin. Elevation of this synaptic vesicle component in tangle-bearing neurons suggests that synapses in resilient brains may have functioned better in the face of tau pathology. Walker also surveyed proteins in the vicinity of tangle-bearing neurons. She found that more proteins involved in oxidative stress and neuroinflammation surrounded tangle-ridden neurons in AD brains than in resilient ones, whereas synaptic and axonal proteins predominated in resilient brains.
What’s inside synapses also emerged as a major determinant of cognitive resilience in findings presented by Teresa Gómez-Isla of Massachusetts General Hospital, Boston. Gómez-Isla scrutinized a large cohort of postmortem brain samples collected at five Alzheimer’s Disease Research Centers. Twenty-eight had died with dementia and a significant burden of AD neuropathology; 61 had had normal cognition, including 27 with little to no AD neuropathology (controls), 20 with a moderate burden of AD pathology, and 27 with a high burden. The latter two groups were considered intermediate or highly resilient, respectively.
Quantifying Aβ plaques and neurofibrillary tau tangles in the entorhinal cortex (EC), superior temporal sulcus (STS), and prefrontal cortex, the researchers found that people in the highly resilient group had a similar burden of both types of pathology as did people with dementia. Those in the intermediate resilient group had a lower pathological burden. In cell-culture assays, tau proteins extracted from the brains of highly resilient people seeded tau aggregation as efficiently as did tau extracted from AD dementia brains.
This ostensibly equal pathological burden had dramatically different consequences in the brains of resilient people versus those with dementia. While resilient brains had as many neurons in the EC and STS as controls, AD brains had lost roughly half of their neurons in these regions. What’s more, their remaining neurons were sorely lacking in both pre- and postsynaptic markers, suggesting profound synaptic loss. Neuronal projections in the hippocampi of AD brains were tortuous and curvy—a sign of dystrophic neurites—while those in highly resilient and control brains remained straight.
Next, the researchers isolated synaptosomes from the brain samples, and analyzed specific proteins via western blot. While synapses from resilient and AD dementia brains had similar amounts of Aβ peptides, those from the latter had at least twice as much phospho-tau, corroborating Montine’s mass-synaptometry data. This synaptic p-tau was accompanied by drastic astrocytic and microglial activation, as well as higher concentrations of proinflammatory cytokines in the entorhinal cortices of people with AD. Resilient brains had no signs of glial activation or inflammation.
The researchers repeated their main findings in the Religious Orders Study and Memory and Aging Project (ROSMAP) cohort, using brains from 25 controls, 25 resilient people, 25 with AD, and 25 who were considered “frail.” This last group had suffered from cognitive impairment before death, despite having no signs of Alzheimer’s or other neuropathology. Interestingly, frail people had levels of microglial activation on par with people who had had AD dementia, casting inflammation as a unifying factor across different types of cognitive impairment. Finally, the researchers reported that synaptic p-tau and glial activation strongly correlated with cognitive decline in those same brain donors before they passed away. Aβ plaque burden did not.
Displaying a tau-PET scan lighting up brain, Gómez-Isla cautioned trial designers that PET or CSF markers of AD pathology may be insufficient to foretell who will develop cognitive decline. Adding biomarkers of synaptic erosion and neuroinflammation might better predict who might develop symptoms, and when, she said.
Along those lines, researchers presented data at AAIC on both PET imaging and CSF biomarkers of synaptic loss. Adam Mecca of Yale University reported that compared with 19 controls, synaptic loss was readily detectable in 34 people with AD using 11C-UCB-J, a PET tracer that binds to the synaptic protein SV2A (Mecca et al., 2020). In 10 cognitively normal people and 10 with AD, Mecca reported that synaptic loss correlated markedly with tau deposition as measured by SV2A and tau tracers. Synaptic loss in the entorhinal cortex correlated with volume loss in the hippocampus in AD, suggesting tau pathology may have gummed up circuitry from the ERC to the hippocampus, leading to degeneration.
Emma Commans of Vrije University, Netherlands, correlated tau- and SV2A-PET in seven people with AD. She saw an odd dichotomy based on a person’s overall tau burden. In those with high tau, brain regions that had more tau had fewer synapses. This might seem to make sense, but consider this: In patients whose total tau burden was low, regions with the most tau had more synapses. Using magnetoencephalography to gauge synaptic activity, Coomans found that in people with high tau pathology and synaptic loss, brain-wave oscillations slowed from alpha into the delta frequency. The bulk of Cooman and Mecca’s findings were presented earlier this year at the Human Amyloid Imaging meeting (Feb 2020 conference news).
Researchers also explored relationships between CSF markers of synaptic loss and neuroinflammation along the AD spectrum. Andréa Benedet of McGill University in Montreal measured four CSF synaptic markers among 115 cognitively normal participants, 41 with MCI, and 31 with AD in that center’s TRIAD cohort. Neurogranin, GAP-43, and SNAP-25 levels correlated with clinical diagnosis, meaning levels were highest in the AD group. Synaptotagmin 1 was highest in MCI. SNAP-25 tracked most closely with clinical symptoms, and also with amyloid- and tau-PET in AD-related brain regions. In people with AD, CSF SNAP-25 correlated tightly with CSF p-tau-181,with the CSF markers of inflammation sTREM2 and YKL-40, as well as with [11C]PBR28 (TSPO-PET), a measure of microglial activation.
Joanna Pereira of the Karolinska Institute in Stockholm showed data on how CSF markers of synaptic and axonal degeneration changed with the stages of AD in the Swedish BIOFINDER cohort. Among 35 participants with evidence of Aβ, but not tau pathology, the SNAP-25, GAP-43, and neurogranin markers of synaptic damage were higher than in 37 controls. However, among 53 people with both Aβ and tau pathology, only neurofilament light (NfL), a marker of axonal damage, had increased further. Pereira believes this could mean that early on, amyloid damages synapses, leading to further increases in Aβ and worsening of synaptic damage. This synaptic damage instigates tau hyperphosphorylation and neurofibrillary tangles, leading to axonal degeneration and ultimately, cognitive impairment.
This sequence may seem at odds with Montine and Gómez-Isla’s findings, which pegged synaptic p-tau, not synaptic Aβ, as the marker of synaptic destruction. It remains unclear how the timeline of changes in CSF and PET markers of AD pathology and synaptic damage fits with the timeline of a shifting protein milieu inside synapses. One thing is clear: Synaptic damage is the defining prelude to cognitive decline.—Jessica Shugart
- Gajera CR, Fernandez R, Postupna N, Montine KS, Fox EJ, Tebaykin D, Angelo M, Bendall SC, Keene CD, Montine TJ. Mass synaptometry: High-dimensional multi parametric assay for single synapses. J Neurosci Methods. 2019 Jan 15;312:73-83. Epub 2018 Nov 20 PubMed.
- Mecca AP, Chen MK, O'Dell RS, Naganawa M, Toyonaga T, Godek TA, Harris JE, Bartlett HH, Zhao W, Nabulsi NB, Wyk BC, Varma P, Arnsten AF, Huang Y, Carson RE, van Dyck CH. In vivo measurement of widespread synaptic loss in Alzheimer's disease with SV2A PET. Alzheimers Dement. 2020 Jul;16(7):974-982. Epub 2020 May 13 PubMed.
No Available Further Reading