Periodic Reporting for period 1 - HUMANE (Window to the brain: a game changer in the discovery of human neuronal circuitry, cellular heterogenicity and biomarker profile indicative of early Alzheimer's disease -related pathology)
Reporting period: 2022-09-01 to 2025-02-28
Alzheimer’s disease (AD) is the main cause of dementia and possesses a major socio-economic burden. Most AD cases are late onset and sporadic. Multiple animal studies have suggested that microglial dysfunctions contribute to AD progression and even underlie disease onset and that modulation of microglial activation may be beneficial in models of AD. However, although acquisition or loss of function and alteration of the homeostatic roles of microglia has been described in animal models of AD, whether microglia reaction could be beneficial, catastrophic or both to AD progression, is still a subject of intense debate. Single cell RNA sequencing datasets on human post-mortem AD brains and rodent models of AD have revealed that microglia exist in disease-specific cellular states that are fundamentally different between mouse and human. Post-mortem human tissue suffers from post-mortem artifacts, represents only the final stages of the disease and do not allow any functional studies since the tissue cannot be preserved alive. Whilst rodent studies suggest that microglial malfunctions contribute to early AD development, investigation of these events has not been possible in human tissue since fresh human AD patient brains are not available for research. Thus, there is a critical need for analysis of freshly isolated, living human brain tissue that is devoid of post-mortem artifacts and suitable for functional studies of cellular changes related to early AD pathology to enhance the translation of preclinical findings into clinical treatments for AD.
This study overcomes these limitations by using idiopathic normal pressure hydrocephalus (iNPH) as a model disease. iNPH is an neurodegenerative disease characterized by an impaired cerebrospinal fluid (CSF) clearance. iNPH is treated by a shunt surger to drain the excess CSF into the abdominal cavity. During the surgery, 10-20mm3 Broadman area 8 biopsy can be excised by minimally invasive methods for preclinical studies. Due to the early AD-related pathology present in a subpopulation of iNPH patients, the brains of these patients offer a unique window to evaluate cellular events taking place during the course of AD pathology progression.
This study overcomes these limitations by using idiopathic normal pressure hydrocephalus (iNPH) as a model disease. iNPH is an neurodegenerative disease characterized by an impaired cerebrospinal fluid (CSF) clearance. iNPH is treated by a shunt surger to drain the excess CSF into the abdominal cavity. During the surgery, 10-20mm3 Broadman area 8 biopsy can be excised by minimally invasive methods for preclinical studies. Due to the early AD-related pathology present in a subpopulation of iNPH patients, the brains of these patients offer a unique window to evaluate cellular events taking place during the course of AD pathology progression.
During the past years, we have set up a pipeline to evaluate, in a layer and cell-dependent manner, the intrinsic neuronal operational properties in the iNPH biopsies. Our electrophysiological interrogations have shown that the biopsies are viable and retain the required microcircuit to study network and synaptic function in the human cortex. From all incoming iNPH shunt surgeries, we are re-slicing the biopsy for multxielectorde recordings, neuronal patch clamp, spatial transcriptomics, single cell sequencing, immunohistochemistry and electron microscopy. Moreover, any remaining piece of the tissue is stored for later proteomic/lipidomic analysis.
Since our multilectrode array (MEA) setup is equipped with perfusion system, we are able to carry out pharmacological treatments of the iNPH biopsy slices. Following the baseline recordings, we expose the slices to NMDA to elicit glutamatergic neuronal activity or carbachol to elicit cholinergic signalling. Our analysis of the MEA recordings have revealed a hyperexcitable phenotype in response to NMDA stimulation in L2/L3 neurons in biopsies with AD-related pathology. This hyper excitable phenotype was specific to layers 2/3 and was absent in the deeper layers of the cortex. This functional hyperexcitable phenotype was associated with transcriptional initiation of immediate early gene expression in the excitatory neurone reciding in layers 2/3 of the cortex of the patients that had beta-amyloid pathology in their resected biopsies. These data have been published Gazestani et al., Cell 2023. Since we have continued our collection of the biopsies, we have now generated even a larger dataset of samples. Since within this dataset the number of Tau-positive biopsies is substantial, we are re-analyzed the data to evaluate whether co-occuring tau pathology further contributes to the observed hyperexcitable phenotype upon NDMA exposure. Our pipeline was featured in Alzforum (https://www.alzforum.org/news/research-news/fresh-brain-every-friday-biopsies-transform-alzheimers-science(opens in new window)). We demonstrated that the iNPH patient brain tissue with early AD-related pathology manifest hyperexcitability in neural circuit that is relevant to AD.
We also have established protocols to induce synaptic strengthening in these biopsies enabling us to carry out mechanistic studies on how synaptic plasticity is altered during the course of early AD pathology progression. We established a method induce synaptic strengthening using Theta Burst Stimulation (TBS) protocols on multielectrode array (MEA). Our 60-electrode array (Multichannel Systems) enables us to evaluate the capacity of the slices to exhibit potentiation spatially, which can then be correlated with the immunohistochemical stainings for the pathological protein aggregates. The TBS induced MEA-waveforms underwent preprocessing, unsupervised machine learning for quality control and advanced time series analysis. We employed feature selection and dimensionality reduction to prevent data oversimplification. Time series features, as well as potentiation features were extracted from all the waveforms. To investigate whether the presence of early AD-related pathology present in the subpopulation of the biopsies affects the waveform features, we immunostained the biopsies using antibodies against Aβ (WO2) and tau (AT180). The time-series and potentiation features of each pair of individual waveforms (pre- and post-TBS) were used to train a supervised RF classifier. We used SHapley Additive exPlanations (SHAP) values to measure which features contributed the most to differentiation between the patient groups. Interestingly, our analysis algorithm was capable of classifying correctly samples into their respective pathology groups with very high accuracy. t-SNE atlases representing the different waveforms demonstrated significant differences in the waveforms between the patients with no pathology, Aβ pathology, Aβ and tau or tau pathology only.
Since our multilectrode array (MEA) setup is equipped with perfusion system, we are able to carry out pharmacological treatments of the iNPH biopsy slices. Following the baseline recordings, we expose the slices to NMDA to elicit glutamatergic neuronal activity or carbachol to elicit cholinergic signalling. Our analysis of the MEA recordings have revealed a hyperexcitable phenotype in response to NMDA stimulation in L2/L3 neurons in biopsies with AD-related pathology. This hyper excitable phenotype was specific to layers 2/3 and was absent in the deeper layers of the cortex. This functional hyperexcitable phenotype was associated with transcriptional initiation of immediate early gene expression in the excitatory neurone reciding in layers 2/3 of the cortex of the patients that had beta-amyloid pathology in their resected biopsies. These data have been published Gazestani et al., Cell 2023. Since we have continued our collection of the biopsies, we have now generated even a larger dataset of samples. Since within this dataset the number of Tau-positive biopsies is substantial, we are re-analyzed the data to evaluate whether co-occuring tau pathology further contributes to the observed hyperexcitable phenotype upon NDMA exposure. Our pipeline was featured in Alzforum (https://www.alzforum.org/news/research-news/fresh-brain-every-friday-biopsies-transform-alzheimers-science(opens in new window)). We demonstrated that the iNPH patient brain tissue with early AD-related pathology manifest hyperexcitability in neural circuit that is relevant to AD.
We also have established protocols to induce synaptic strengthening in these biopsies enabling us to carry out mechanistic studies on how synaptic plasticity is altered during the course of early AD pathology progression. We established a method induce synaptic strengthening using Theta Burst Stimulation (TBS) protocols on multielectrode array (MEA). Our 60-electrode array (Multichannel Systems) enables us to evaluate the capacity of the slices to exhibit potentiation spatially, which can then be correlated with the immunohistochemical stainings for the pathological protein aggregates. The TBS induced MEA-waveforms underwent preprocessing, unsupervised machine learning for quality control and advanced time series analysis. We employed feature selection and dimensionality reduction to prevent data oversimplification. Time series features, as well as potentiation features were extracted from all the waveforms. To investigate whether the presence of early AD-related pathology present in the subpopulation of the biopsies affects the waveform features, we immunostained the biopsies using antibodies against Aβ (WO2) and tau (AT180). The time-series and potentiation features of each pair of individual waveforms (pre- and post-TBS) were used to train a supervised RF classifier. We used SHapley Additive exPlanations (SHAP) values to measure which features contributed the most to differentiation between the patient groups. Interestingly, our analysis algorithm was capable of classifying correctly samples into their respective pathology groups with very high accuracy. t-SNE atlases representing the different waveforms demonstrated significant differences in the waveforms between the patients with no pathology, Aβ pathology, Aβ and tau or tau pathology only.
We have established electrophysiological recordings in human cortical brain biopsies that show early AD-related pathology. These unique samples are the only sample material there is available that are suitable for investigation on how early AD-related pathology alters cellular states and functions. Our data so far demonstrates that early AD-pathology causes tissue hyperexcitability, impairs single neuron operational properties and synaptic plasticity in human brain. As all prior data is either from animal models or represents late stages of Alzheimer’s disease, our data takes science in this field beyond the state-of-the-art by providing the first insights into alterations of neurocircuit properties in tissues that represent very early stages of AD, prior to the clinical diagnosis.