Differential aberrant structural synaptic plasticity in axons and dendrites ahead of their degeneration in tauopathy
Curation statements for this article:-
Curated by eLife
Summary: This paper describes studies of a mouse model of tauopathy with relevance to Alzheimer's Disease. A powerful approach of longitudinal imaging of single synaptic structures over time allows insights into the time course of progressive neurogenerative responses. The strengths of the report are the relevance of the question to human disease, the powerful imaging approach, and the indication that there may be a programmed sequence of structural changes that mediate tauopathy. On the other hand, there were multiple issues with the transgenic mouse model used, which would seriously limit interpretation of results without suitable controls. Further, the data set appeared to be quite noisy, and variable between animals, which may result in part from the nonspecific methods of expressing fluorescent markers, thus leading to uncertainty regarding the specific identity of pre- and post-synaptic elements.
This article has been Reviewed by the following groups
Discuss this preprint
Start a discussion What are Sciety discussions?Listed in
- Evaluated articles (eLife)
Abstract
Neurodegeneration driven by aberrant tau is a key feature of many dementias. Pathological stages of tauopathy are characterised by reduced synapse density and altered synapse function. Furthermore, changes in synaptic plasticity have been documented in the early stages of tauopathy suggesting that they may be a driver of later pathology. However, it remains unclear if synapse plasticity is specifically linked to the degeneration of neurons. This is partly because, in progressive dementias, pathology can vary widely from cell-to-cell along the prolonged disease time-course. To overcome this variability, we have taken a longitudinal experimental approach to track individual neurons through the progression of neurodegenerative tauopathy. Using repeated in vivo 2-photon imaging in rTg4510 transgenic mice, we have measured structural plasticity of presynaptic terminaux boutons and postsynaptic spines on individual axons and dendrites over long periods of time. By following individual neurons, we have measured synapse density across the neuronal population and tracked changes in synapse turnover in each neuron. We found that tauopathy drives a reduction in density of both presynaptic and postsynaptic structures and that this is partially driven by degeneration of individual axons and dendrites that are spread widely across the disease time-course. Both synaptic loss and neuronal degeneration was ameliorated by reduction in expression of the aberrant P301L transgene, but only if that reduction was initiated early in disease progression. Notably, neurite degeneration was preceded by alterations in synapse turnover that contrasted in axons and dendrites. In dendrites destined to die, there was a dramatic loss of spines in the week immediately before degeneration. In contrast, axonal degeneration was preceded by a progressive attenuation of presynaptic turnover that started many weeks before axon disappearance. Therefore, changes in synapse plasticity are harbingers of degeneration of individual neurites that occur at differing stages of tau-driven neurodegenerative disease, suggesting a cell or neurite autonomous process. Furthermore, the links between synapse plasticity and degeneration are distinct in axonal and dendritic compartments.
Key findings
-
Tauopathy driven by tau P301L in rTg4510 mice causes a progressive decrease in density of presynaptic terminaux boutons and postsynaptic dendritic spines in cortical excitatory neurons.
-
Longitudinal imaging of individual axons and dendrites shows that there is a huge diversity of effects at varying times in different cells.
-
Decreases in overall synapse density are driven partly, but not exclusively, by degeneration of dendrites and axons that are distributed widely across the time-course of disease.
-
Suppression of pathological P301L tau expression can ameliorate accumulation of tau pathology, synapse loss and neurodegeneration, but only if administered early in disease progression.
-
Neurite degeneration is preceded by aberrant structural synaptic plasticity in a cell-specific way that is markedly different in dendrites and axons.
-
Degeneration of dendrites is immediately preceded by dramatic loss of dendritic spines.
-
Axonal loss is characterised by a progressive attenuation of presynaptic bouton plasticity that starts months before degeneration.
Article activity feed
-
Reviewer #3:
The study by Jackson et al. characterizes the progression of the degeneration of axons and dendrites, including metrics on density and dynamics of dendritic spines and terminaux boutons (TBs), in the rTg4510 transgenic mouse model. The authors describe a decrease in the density of both structures, spines and TBs, as well as degeneration of neurites. Repression of the expression of the mutated version of tau was able to partially mitigate some of the negative effects observed in the non-repressed condition. When degeneration of the neuronal process was observed, the loss of a dendritic branch was preceded by a sharp increase in the loss of dendritic spines, while axonal loss was preceded by a long-lasting and progressive loss of TBs. While the findings are interesting, there are several concerns that dampened the enthusiasm …
Reviewer #3:
The study by Jackson et al. characterizes the progression of the degeneration of axons and dendrites, including metrics on density and dynamics of dendritic spines and terminaux boutons (TBs), in the rTg4510 transgenic mouse model. The authors describe a decrease in the density of both structures, spines and TBs, as well as degeneration of neurites. Repression of the expression of the mutated version of tau was able to partially mitigate some of the negative effects observed in the non-repressed condition. When degeneration of the neuronal process was observed, the loss of a dendritic branch was preceded by a sharp increase in the loss of dendritic spines, while axonal loss was preceded by a long-lasting and progressive loss of TBs. While the findings are interesting, there are several concerns that dampened the enthusiasm on the study:
The data obtained with the rTg4510 mouse model must be very carefully interpreted given that the disruption of the endogenous gene Fgf14 that occurs in this mouse model contributes significantly to the neurodegenerative phenotype (Gamache et al., 2019). While the authors acknowledge the possibility that genetic factors other than tau hyperphosphorylation may contribute to the rTg4510 pathology, the results must be put into the perspective of the mouse model rather than into the perspective of the tauopathy exclusively. In this sense, it would be recommended that the caveats of the mouse model be included in the introduction.
The authors do not either mention the sex of the animals used in the study or how many mice from each sex were included in each experimental group. This is an important matter because it has been described that the rTg4510 mouse model presents with sex differences in the degree of accumulation of tau (Yue et al., 2009; Song et al., 2015).
A big concern is the identity of the neurons labeled. The strategy to label cells is very unspecific and no details are given on their identity. Different subtypes of pyramidal neurons with different densities of dendritic spines and axon boutons may be mixed up in different proportions in each group and batch. In fact, the resilience of different neuron subtypes to the pathology may be different too. If the authors cannot pinpoint the identity of the neuron imaged, an elaboration on this issue must be included in the manuscript. In addition, the manuscript must include representative images of the cortex of both genotypes showing the labeling pattern obtained with their approach. It is recommended to the authors to add more information about the vector.
How did the authors estimate the point of divergence between genotypes? The authors mentioned the 30-35 wk and 50 wk as points of divergence - which should be interpreted as the first time points where the differences between groups are significantly different - in lines 180-183. While the Wald test and the Akaike information criterion indicate that genotype is the factor with the most influence on the model estimates, it does not compute statistical differences between phenotypes at a given time point. Regarding the GAMMs, some fits suggest that data at earlier points may be very different between groups (i.e., Fig 2E, 5C, 6C). Is the decrease in density of TBs over time in WT mice significant? How do the authors interpret those fits?
Looking at the data in Figures 1E and 2E, one would expect more negative growth values in figs 5E and 6E, indicating a larger decrease in density. They are flat. Are these analyses well powered? Are the data in Figures 5E and 6E not representative?
-
Reviewer #2:
This manuscript asked the question of how axons vs dendrites are lost by the live-imaging cortex of rTg4510 tau transgenic mice. Overall, this manuscript is well-done and well-written, and confirms previous findings. However, there are a number of key controls missing from the experimental data (please see below). Statistical analyses are satisfactory (with some caveats, please see below).
Figures 1+2 replicate previous findings also in rTg4510 (Crimins et al., 2012; Jackson et al., 2017; Kopeikina et al., 2013); Figures 3+4 (Ramsden et al., 2005; SantaCruz et al., 2005; Spires et al., 2006; Crimins et al., 2012; Kopeikina et al., 2013; Helboe et al., 2017; Jackson et al., 2017). The novelty here are the differing patterns of bouton and spine turnover shortly before axons and dendrites, respectively, are lost, which is a …
Reviewer #2:
This manuscript asked the question of how axons vs dendrites are lost by the live-imaging cortex of rTg4510 tau transgenic mice. Overall, this manuscript is well-done and well-written, and confirms previous findings. However, there are a number of key controls missing from the experimental data (please see below). Statistical analyses are satisfactory (with some caveats, please see below).
Figures 1+2 replicate previous findings also in rTg4510 (Crimins et al., 2012; Jackson et al., 2017; Kopeikina et al., 2013); Figures 3+4 (Ramsden et al., 2005; SantaCruz et al., 2005; Spires et al., 2006; Crimins et al., 2012; Kopeikina et al., 2013; Helboe et al., 2017; Jackson et al., 2017). The novelty here are the differing patterns of bouton and spine turnover shortly before axons and dendrites, respectively, are lost, which is a finding uniquely enabled by 2-photon. Thus, findings in Fig. 5/6 should be highlighted and solidified. Further, the manuscript lacks mechanistic insight.
It is not clear how the authors ensure that the perceived loss of spines/boutons/dendrites/axons is not due to bleaching or loss of the GFP signal. Please validate loss of spines/boutons and actual synapses using fixed tissue imaging or electron microscopy on a separate cohort of mice.
Did the authors control for gliosis after the repeated imaging (very short after viral injection and cranial window implant on the same site)? Could it be that the repeated imaging itself on a damaged tissue induces blebbing on the already more vulnerable spines in the tau mice? Please show Iba1 and GFAP with and without doxycycline administration should be included in supplemental along with area staining quantification. Transgenic mice without manipulation (viral injection/cranial window/2P imaging) should also act as a control to ensure no gliosis is observed.
rTg4510 transgene insertion: Gamache et al. recently showed that the integration sites of both the CaMKIIα-tTA and MAPT-P301L transgenes impact the expression of endogenous mouse genes. The disruption of the Fgf14 gene in particular contributes to the pathological phenotype of these mice, making it difficult to directly ascribe the phenotypes seen in the manuscript to MAPT-P301L transgene overexpression. Although this limitation is acknowledged in the discussion, the T2 mice employed in this paper (Gamache et al., 2019) would be suitable controls to better evaluate the contribution of tauP301L alone on the neuropathology and disease progression observed in the authors' experiments, at least in fixed synapse imaging.
-
Reviewer #1:
Studies in mouse models and humans show synapse loss and dysfunction that precede neurodegeneration, raising questions about timing and mechanisms. Using longitudinal in vivo 2-photon imaging, Jackson et al., investigate pre- and post-synaptic changes in rTg4510 mice, a widely used mouse model of tauopathy. Consistent with cross sectional studies, the authors observed a reduction in density of presynaptic axons and dendritic spines in layer 1 cortex that relate to degeneration of neurites and dendrites over time. Taking advantage of an inducible model to overexpress tau p301L, they show that reducing expression of tau by DOX early in disease progression, resulted in amelioration of synapse loss, also consistent with other studies. Interestingly, the authors observed a significant reduction of dendritic spines less than a …
Reviewer #1:
Studies in mouse models and humans show synapse loss and dysfunction that precede neurodegeneration, raising questions about timing and mechanisms. Using longitudinal in vivo 2-photon imaging, Jackson et al., investigate pre- and post-synaptic changes in rTg4510 mice, a widely used mouse model of tauopathy. Consistent with cross sectional studies, the authors observed a reduction in density of presynaptic axons and dendritic spines in layer 1 cortex that relate to degeneration of neurites and dendrites over time. Taking advantage of an inducible model to overexpress tau p301L, they show that reducing expression of tau by DOX early in disease progression, resulted in amelioration of synapse loss, also consistent with other studies. Interestingly, the authors observed a significant reduction of dendritic spines less than a week before dendrite degeneration. In contrast, they observed plasticity and turnover of presynaptic structures weeks before axonal degeneration, suggesting different mechanisms.
Overall the results are interesting and largely consistent with previous findings. The new findings shown in Figures 5 and 6 address the timing of pre and postsynaptic loss and structural plasticity and reveal interesting differences; however, the data are highly variable and there are several issues that diminish enthusiasm as outlined below. Moreover, this study does not include new biological or mechanistic insight into the differences in pre- and post-synaptic changes from previous work in the field.
The main weakness relates to the significance and relevance beyond this specific mouse model and brain region. I appreciate the strengths but also technical challenges of in vivo longitudinal imaging, including a small field of view. Thus, the rationale and choice of model and brain region, and validation of key findings is critical to support conclusions. In this case, the tau model, although used by others, has several caveats relevant to the investigation of synapse loss (see point 2 below) that weaken this study and its impact.
- Most of the work in the model related to synapse loss and dysfunction have been carried out in hippocampus and other regions of cortex in this model and tau and amyloid models. Here the authors focused on layer 1 of (somatosensory) cortex and followed neurites of pyramidal cells labeled with AAV:GFP, an approach that does not enable one image and track axons and dendrites from large numbers of neurons. They observed divergent dynamics in spine and presynaptic TBS of individual dendrites and axons. Given the small number of neurons sampled, significant noise in their imaging data, these findings need more validation using other approaches. This is particularly important for the data and conclusion drawn from Figures 5 and 6 (see point 3).
To estimate the overall effect of genotype the authors fitted Generalized Additive Mixed Models (GAMMS) to their data given the variability in the data within animals and genotype. It would be helpful to those less familiar to provide more comparisons of data using additional statistical tests and analyses along with power analyses calculations.
Major caveat with inducible Tau mode Tg4510. While this inducible model has the advantage of controlling timing of tau overexpression in neurons, a recent study by Gamache et al (PMID: 31685653) demonstrated that there are issues with the transgene insertion site and factors other than tau expression are actually what is driving the phenotype. Thus, differences in synaptic and behavioral phenotypes are based on the mouse line used and this needs to be carefully controlled. This was not addressed or discussed. See https://pubmed.ncbi.nlm.nih.gov/31171783/ and https://pubmed.ncbi.nlm.nih.gov/30659012/
The interesting new findings presented in Figures 5 and 6 that address timing and differences in axonal and dendritic/spine plasticity and loss need to be validated with more neurons and animals. The sample size is small ( i.e. n= 18 axons from 7 animals and not clear how many neurons. Given the significant variability of the data even within animals, these experiments and data are considered preliminary.
How does anesthesia influence these changes in structural plasticity observed? This was not addressed or discussed.
-
Summary: This paper describes studies of a mouse model of tauopathy with relevance to Alzheimer's Disease. A powerful approach of longitudinal imaging of single synaptic structures over time allows insights into the time course of progressive neurogenerative responses. The strengths of the report are the relevance of the question to human disease, the powerful imaging approach, and the indication that there may be a programmed sequence of structural changes that mediate tauopathy. On the other hand, there were multiple issues with the transgenic mouse model used, which would seriously limit interpretation of results without suitable controls. Further, the data set appeared to be quite noisy, and variable between animals, which may result in part from the nonspecific methods of expressing fluorescent markers, thus leading to uncertainty …
Summary: This paper describes studies of a mouse model of tauopathy with relevance to Alzheimer's Disease. A powerful approach of longitudinal imaging of single synaptic structures over time allows insights into the time course of progressive neurogenerative responses. The strengths of the report are the relevance of the question to human disease, the powerful imaging approach, and the indication that there may be a programmed sequence of structural changes that mediate tauopathy. On the other hand, there were multiple issues with the transgenic mouse model used, which would seriously limit interpretation of results without suitable controls. Further, the data set appeared to be quite noisy, and variable between animals, which may result in part from the nonspecific methods of expressing fluorescent markers, thus leading to uncertainty regarding the specific identity of pre- and post-synaptic elements.
-
