Disease and brain region specific immune response profiles in neurodegenerative diseases with pure and mixed protein pathologies

Alzheimer's disease (AD) is the most common form of dementia worldwide [2, 50]. It affects over 55 million people and impacts not only patients, but also their families and caregivers [91]. Pathologically, the hallmarks of AD are extracellular deposits of amyloid beta (Aβ) in form of plaques, intracellular aggregates of tau in form of neurofibrillary tangles (NFT) and neuronal loss. More recently a prominent response of the local immune system including astrogliosis and microgliosis has been recognized as a key component of AD pathophysiology [7, 13, 27, 29, 46, 48, 68, 82, 86]. Activation of microglia, particularly around Aβ deposits, is observed in AD and has been studied extensively in animal models [63, 64]. These studies have established gene expression signatures of activated microglia in response to Aβ deposition [21, 23, 35, 44, 72, 89] and have demonstrated the impact of genetic AD risk variants in microglia-associated genes on microglia activation [38, 65, 92]. While walling off and phagocytosing of Aβ plaques [21, 29, 59] by microglia is considered beneficial, activation-associated cytokine secretion and overactivation-associated phagocytosis of synapses have been linked to exacerbating disease progression [1, 3]. Thus, it is still a matter of hot debate whether stimulating microglia or inhibiting microglia is the more promising therapeutic approach to combat Aβ deposits.

The immune response to prominent intracellular protein pathologies, such as tau or α-Synuclein (α-Syn) pathology has also been studied and has revealed a more subtle, yet prominent activation of microglia in animal models of tauopathy or α-Synucleinopathy [6, 14, 52, 60]. Translating these groundbreaking studies to the human disease condition has been successful, especially with respect to the activation signatures reported in response to Aβ [51, 88, 93]. Cross-species comparisons however have also documented key differences between model systems and the human disease condition, which are important to consider when moving towards therapeutic interventions. One key difference is prominent age-related changes observed in humans, which can only be modeled in a limited way in animal models [15, 55, 69, 74]. Certain age-related phenotypes in microglia are only observed in humans, such as microglia dystrophy, characterized by beaded microglia processes, changes in nuclear morphology, and an arrested functional state [42, 45, 76‐78]. Furthermore, ND rarely present with “pure” pathology, restricted to one or two protein pathologies. In fact, the majority of patients who come to autopsy exhibit multiple protein pathologies at the time of death [66, 67]. Previous studies have investigated the heterogeneity of microglia in different brain regions. In these studies, the assessment of microglial activation phenotypes was limited to single disease backgrounds in the mouse or human brain [8, 10, 20, 56]. However, the interaction of occurrence of co-pathologies and brain region-specific microglial activation has not been investigated. It is therefore of utmost importance to study the innate immune response in the complex environment of the diseased human brain in order to successfully translate studies in animal models to therapeutic interventions in humans.

We have previously studied the innate immune response in cases with pure (only Aβ and tau) AD pathology, showing that microglia activation follows the progression of Aβ plaques and NFT, while Aβ pathology or NFT pathology alone were not sufficient to drive a strong microglia response [65]. Expanding this study to the complex landscape of ND, we set out to compare and contrast the microglia response in pure AD cases (AD) with the response in cases with intracellular Lewy body disease pathology (LBD) and cases with mixed AD and LBD pathology (MIX).

Next, we studied CD68, a lysosomal marker for macrophage linage cells primarily seen in phagocytically active microglia [24]. Morphologically, CD68 positivity was mainly detected in amoeboid microglia with round cell bodies and shortened, less complex cellular processes. While scattered CD68 pos. cells were evenly distributed throughout the tissue in Control cases, we recognized clustering of CD68 pos. cells in cases with ND-related protein pathology (Fig. 3a). In line with this observation, numbers of CD68 pos. cells were significantly increased in all disease groups (AD, MIX, LBD) compared to Control cases, across all brain regions studied. Cases exhibiting ADNC (AD and MIX) displayed the highest numbers of CD68 pos. cells, and interestingly, among those, MIX cases had slightly higher CD68 counts in the hippocampus, frontal cortex, and midbrain than pure AD cases (Fig. 3b–e).

In this study, we used a large cohort of human postmortem brains to investigate the local immune response around extra- and intracellular protein pathologies in AD and LBD. We identified a strong, local immune activation pattern in cases with ADNC, regardless of the presence or absence of Lewy body co-pathology (AD and MIX cases), while cases with pure LB pathology showed a more attenuated, global, disease-related immune activation signature. Specifically, activated and dystrophic microglia phenotypes were strongly associated with the presence of extracellular Aβ pathology in cases with ADNC.

In the first part of this study, we focused on the immunohistochemical quantification of different states of microglia (homeostatic, activated, and dystrophic) in brain regions, that are important for the progression of ND (AD and LBD). Previous studies, which mostly focus on individual brain regions, have demonstrated that in the early stages of ND, such as AD or PD, microglia exhibit a specific activation pattern [14, 26, 62, 64, 75]. While there is a large body of literature on microglia activation in AD, there is a lack of studies examining immune activity in LBD. In our study, we found that regardless of the disease or the brain region examined, there were no significant differences in the numbers of Iba1 pos. microglia. Only a slight reduction of Iba1 pos. cells was observed in cases with pure ADNC. While previous studies have reported that the number of Iba1 pos. cells is reduced in high AD cases [42, 53, 65], we did not observe a decrease in cell numbers in cases with LBD pathology (MIX and LBD). Since P2RY12 is often co-localized with Iba1 [87], it is less surprising that we observed a similar expression pattern of these two microglia markers, with a slight reduction of positive numbers in AD cases in HPC and FC. While Iba1 is considered a pan microglia marker, which can be found across all functional microglia states (activated, homeostatic, dystrophic, etc.), P2RY12 is often used as a marker for homeostatic microglia. P2RY12 is important for chemotaxis of microglia (ATP detection), neuron maintenance, and synaptic pruning [19], however, the exact function of this purinergic receptor remains unclear. Interestingly, we noted increased numbers of P2RY12 pos. cells in OCC of cases with LB pathology (MIX and LBD). The visual cortex is considered a brain region with high P2RY12 activity, since visual impacts cause an increase of neuron-microglia interaction [90]. Homeostatic microglia interact with dendritic spines of visual V1 neurons and help to maintain synaptic connections in that area [85]. ND with Lewy bodies are reported to experience well-formed recurrent complex visual hallucinations and visuoperceptual deficits which come with hyperactivation of neurons and neurotransmitter imbalance [36, 81]. The increased neuronal activity in the OCC is thought to cause increased microglia recruitment to maintain neuron health and synapses [16, 90]. In strong contrast, when studying markers associated with microglia activation, we observed a marked increase in activated/phagocytic and dystrophic cells in all tissue samples with ND, especially in the pure AD and MIX cases. This is also consistent with the current literature, which reports increased activation of microglia in the affected brain areas of individuals with AD [3, 65, 84, 92]. Interestingly, our study shows that the largest increase in activated microglia can be observed in HPC and FC. An increase in CD68- and Ferritin pos. cells was noted especially in cases with ADNC (AD and MIX). It seems that these areas, which are affected early in AD, show the greatest extent of chronic microglia activation and dystrophy, while in OCC and MB, the differences between disease groups (AD, MIX and LBD) are less pronounced compared to controls. Ferritin is a microglial marker for dystrophic cells [45] and is also described as a marker for chronically activated microglial cells [33]. Since it can also be associated as a marker for senescence, it was to be expected that Ferritin levels would be elevated in aging brains [43, 47, 49]. However, the results of this study are more in line with the current literature, which cites Ferritin as a marker of microglial cell death in ND [32, 45, 57, 76]. The reduction in Iba1 pos. and homeostatic cells in ADNC cases and the increase in activated and dystrophic microglial cells may be related to the fact that microglial cells undergo a much greater phenotypic shift in response to ADNC than in response to normal aging-related changes or in cases with pure intracellular pathology (e.g. LBs). While this change is less pronounced in control subjects or LBD, microglia tend to become stuck in an activated state around extracellular pathology (Aβ plaques) and enter a "burned-out" state, which leads to increased death and dystrophy later in the course of the disease [45, 79].

Since our IHC data suggested a direct link between microglia activation and microglia dystrophy with extracellular pathology, we next thought to get a more comprehensive profile of microglia activation using whole tissue transcriptomic analysis.

Using the NanoString nCounter® technology, we analyzed glial cell related signaling pathways in a subset of cases from our histological analysis. Overall, this analysis revealed that in ND cases the entire brain is in a state of general immune activation, regardless of the underlying specific disease process. We noted that the signaling pathway profiles of AD and MIX cases were more similar to each other than those of LBD cases or controls. Previous studies have shown that AD cases exhibit impairment in glucose metabolism and reduced glucose uptake [18, 28, 39]. Our data support this notion but also show that dysregulation of glucose metabolism is not only observed in pure AD cases but also in cases with mixed pathology (MIX). Interestingly, the upregulation of microglia signaling pathways we observed by whole tissue transcriptomics, was less pronounced than expected from the data of our histological analysis. The specific pathways appeared to be dependent on the brain region as well as the underlying disease condition. In HPC, FC and MB cases with ADNC (AD and MIX groups) were more similar than LBD or controls. The differences between experimental groups were smallest in the hippocampus region (comparison of Stage 1 and 2 DAM, MGnD or Primed microglia) than in FC or MB, where we observed a much stronger activation of microglia in AD and MIX than in LBD. The most profound upregulation of immune response pathways however was noted in the OCC. This could reflect more early disease changes, since the OCC is affected late in AD [65]. MIX cases showed the most pronounced upregulation of synaptic signaling pathways, glucose metabolism, or neuronal markers. In comparison, AD cases showed a more prominent and specific upregulation of microglial markers such as Stage 1 and 2 DAM, Primed microglia, and Complement system.

The whole tissue transcriptomic results overall showed a less pronounced activation of microglia pathways in cases with extracellular protein pathologies (AD and MIX) as expected from our immunohistochemical analysis of microglial activation markers. This suggest that there is not a general, tissue-wide activation of microglia, but a focal, pathology-driven event in response to extracellular Aβ plaques.

To confirm this hypothesis and to highlight the focal events around different intracellular and extracellular protein aggregates, we decided to study the local changes with selected protein markers using the NanoString Digital Spatial Profiling platform.

On the NanoString GeoMx® Digital Spatial Profiling platform, we investigated the tissue microenvironment around specific protein pathologies in the last part of this study. By targeting specific ROIs, we were able to perform limited proteomic analyses around disease-specific protein pathologies, such as Lewy bodies or Aβ plaques and investigate specific proteins associated with different microglia activation states in this local environment. Interestingly, when comparing Lewy body-bearing neurons between disease groups (LBD and MIX), we observed no significant increase in microglia or inflammatory markers, albeit a significant increase of phospho-α-Syn (S129) in LB-bearing neurons demonstrated successful selection of the respective ROI. When comparing noLB tissue from disease tissue with noLB tissue from control cases we observed a robust upregulation of microglia activation-associated markers in the disease group. This suggests that the immune activation in cases with intracellular LB pathology is more global, disease-specific than locally concentrated around LB-bearing neurons. This contrasts with the data from the comparison of AD and MIX cases. In the local microenvironment around extracellular Aβ plaques, we identified a significant upregulation of markers for activated and DAM microglia such as MERTK, CLEC7A or GPNMB. The spatial proteomic dataset confirms our histological data, where we noted that activated and dystrophic microglia phenotypes were strongly associated with the presence of extracellular Aβ pathology in cases with ADNC, while cases with pure LB pathology showed a more attenuated, global disease-related immune activation signature.

Our study has several limitations. Although we have made our best efforts to create a cohort of brains of the same age that are balanced in terms of sex, pathology, and APOE genotype, we are dependent on the availability of postmortem tissue. This was most apparent in the sex bias of LBD cases, which are known to be more commonly seen in men, whereas AD is more common in women [17, 58, 61, 71]. Another limitation of this study is that we are studying postmortem human tissue. This represents only the endpoint of the disease and reflects only a short period of disease progression. In particular, the interpretation of our data with respect to a potential change in microglia phenotype during disease progression can only be inferred from cross-sectional analysis and comparison of different brain regions. Longitudinal assessment of microglia activation during disease progression in one person would be desirable, but is not achievable with currently available in vivo microglia markers.

Overall, we conclude that there is a significant correlation between microglial activation and the presence of extracellular protein pathologies, while intracellular pathologies drive a more subtle and more global immune response. Furthermore, there are important brain region differences in the observed microglia and immune activation phenotypes. Our study underscores the complexity of the immune response towards ND-related protein pathologies in human brains and highlights the importance of analysis of post-mortem brains for successful translation of in vitro and in vivo experimental studies.