Final Report Summary - SYSPHARMAD (A systems pharmacology approach to the discovery of novel therapeutics in Alzheimer´s disease)
During the SysPharmAD project, we described the molecular changes associated with the time course of the AD in the hippocampus of three mouse models, namely the 3xTg and the NL-F and NL-G-F knock-ins (KI). The three AD models represent various pathological processes of the human disease, and therefore they complement each other. For instance, while the 3xTg is the only model that develops Tau tangles, the NL-F enables the study of the initial changes that cause the misprocessing of App and the Aβ accumulation. Complementarily, the most aggressive KI (NL-G-F) is the best to study the AD progression and to test new potential AD drugs, as it develops the cognitive impairment earlier and the accumulation of extracellular Aβ can be observed already at three months. Overall, the combination of the three models facilitates the distinction between changes associated with healthy aging and AD progression.
The full molecular profiling (transcriptomics and proteomics) of mice hippocampi at different AD stages allowed a comprehensive comparison of the information provided by each technique. While the direct overlap between the proteomics and transcriptomics results for the three models is low, due to post-translational modulators and the half-lives difference between mRNA and proteins, allowed us to find interesting cases of protein accumulation in the plaques, mostly due to a deficient clearance processes. In addition, our data indicates that the lysosome system is affected at the initial stages of the disease, which influence App processing. Other onset changes detected in our results are related to osmotic stress and the spliceosome. Later, the accumulation of Aβ in the mouse hippocampus trigger the activation of some processes such us the immune response (innate response, complement, and inflammation), the increment of microglial and astrocytes markers or apoptosis. On the other hand, genes involved in neurotransmitters-dependent synapses, e.g. dopaminergic, glutamatergic or cholinergic synapses, are downregulated along with AD progression. Other neuronal processes like myelination or axon guidance and development, are also downregulated. Finally, the last changes detected are associated with metabolism, apoptosis and glia. We clustered the observed responses and processed the differential proteomics and transcriptomics signatures for the different models and AD stages to disentangle the effects of normal (healthy) aging from those caused by the disease. We then transformed the stage-specific AD signatures to a format that is suitable for the Chemical Checker (i.e. our resource to link biological traits and drug-like molecule bioactivities), and we looked, within the space of 1M bioactive compounds, for molecules able to revert the AD signature modules detected. Overall, we identified 29 compounds that showed good potential, and tested them in our engineered AD SH-SY5Y cells, finding that three of them were indeed able to change the Ab42/Ab40 ratio and revert the transcriptional signature of AD cell to a healthy state. We then implemented a deep-learning strategy to impute bioactivity signatures to any available small molecule, expanding thus our universe of chemical modulators. We finally selected six compounds that we tested in vivo, for a period of 4-8 weeks on mice showing initial cognition impairment. Of these, three compounds showed a good potential on cognitive tests (object recognition) and the capacity to revert some key AD signatures.
Overall, we obtained a comprehensive molecular profiling of three AD mouse models that reflects clinically relevant event found in AD patients and also provides an excellent data set to understand the mechanisms involved in the disorder. Moreover, we identified potential biomarker signatures that might enable an early diagnosis and a collection of initial compounds that show good potential to stop or slow down the progression of the disease.
The full molecular profiling (transcriptomics and proteomics) of mice hippocampi at different AD stages allowed a comprehensive comparison of the information provided by each technique. While the direct overlap between the proteomics and transcriptomics results for the three models is low, due to post-translational modulators and the half-lives difference between mRNA and proteins, allowed us to find interesting cases of protein accumulation in the plaques, mostly due to a deficient clearance processes. In addition, our data indicates that the lysosome system is affected at the initial stages of the disease, which influence App processing. Other onset changes detected in our results are related to osmotic stress and the spliceosome. Later, the accumulation of Aβ in the mouse hippocampus trigger the activation of some processes such us the immune response (innate response, complement, and inflammation), the increment of microglial and astrocytes markers or apoptosis. On the other hand, genes involved in neurotransmitters-dependent synapses, e.g. dopaminergic, glutamatergic or cholinergic synapses, are downregulated along with AD progression. Other neuronal processes like myelination or axon guidance and development, are also downregulated. Finally, the last changes detected are associated with metabolism, apoptosis and glia. We clustered the observed responses and processed the differential proteomics and transcriptomics signatures for the different models and AD stages to disentangle the effects of normal (healthy) aging from those caused by the disease. We then transformed the stage-specific AD signatures to a format that is suitable for the Chemical Checker (i.e. our resource to link biological traits and drug-like molecule bioactivities), and we looked, within the space of 1M bioactive compounds, for molecules able to revert the AD signature modules detected. Overall, we identified 29 compounds that showed good potential, and tested them in our engineered AD SH-SY5Y cells, finding that three of them were indeed able to change the Ab42/Ab40 ratio and revert the transcriptional signature of AD cell to a healthy state. We then implemented a deep-learning strategy to impute bioactivity signatures to any available small molecule, expanding thus our universe of chemical modulators. We finally selected six compounds that we tested in vivo, for a period of 4-8 weeks on mice showing initial cognition impairment. Of these, three compounds showed a good potential on cognitive tests (object recognition) and the capacity to revert some key AD signatures.
Overall, we obtained a comprehensive molecular profiling of three AD mouse models that reflects clinically relevant event found in AD patients and also provides an excellent data set to understand the mechanisms involved in the disorder. Moreover, we identified potential biomarker signatures that might enable an early diagnosis and a collection of initial compounds that show good potential to stop or slow down the progression of the disease.