Noradrenergic axon loss drives olfactory dysfunction in Alzheimer’s disease

doi:10.21203/rs.3.rs-4887136/v1

Download PDF

Biological Sciences - Article

Noradrenergic axon loss drives olfactory dysfunction in Alzheimer’s disease

https://doi.org/10.21203/rs.3.rs-4887136/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Alzheimer’s disease (AD) is often accompanied by early non-cognitive symptoms, including olfactory deficits, such as hyposmia and anosmia¹. These have emerged as solid predictors of cognitive decline, but the underlying mechanisms of hyposmia in early AD remain elusive². Pathologically, one of the brain regions affected earliest is the brainstem locus coeruleus (LC), the main source of the neurotransmitter noradrenalin (NA) and, a well-known neuromodulator of olfactory information processing³. Here we show that early and distinct loss of noradrenergic input to the olfactory bulb (OB) coincides with impaired olfaction in a mouse model of AD, even before pronounced appearance of extracellular amyloid plaques. Mechanistically, OB microglia detect externalized phosphatidylserine and MFG-E8 on hyperactive LC axons and subsequently initiate their clearance. Translocator protein 18 kDa (TSPO) knockout reduces phagocytosis, preserving LC axons and olfaction. Importantly, patients with prodromal AD display elevated TSPO-PET signals in the OB, similarly to APP^NL-G-F mice. We further confirm early LC axon degeneration in post-mortem OBs in patients with early AD. Collectively, we uncover an underlying mechanism linking early LC system damage and hyposmia in AD. Our work may help to improve early diagnosis of AD by olfactory testing and neurocircuit analysis and consequently enable early intervention.

Biological sciences/Neuroscience/Diseases of the nervous system/Alzheimer's disease

Biological sciences/Neuroscience/Olfactory system/Olfactory bulb

Biological sciences/Neuroscience/Neural circuits

Alzheimer’s disease is currently the most prevalent and devastating form of dementia, affecting millions of people worldwide⁴. Extracellular deposition of β-amyloid (Aβ), the formation of Aβ-plaques and the aggregation of microtubule-associated protein tau forming neurofibrillary tangles are the pathological hallmarks of AD⁵. While causal therapies are still not available, recent Aβ targeting antibody-therapies moderately improve cognitive decline in patients at early AD stages^6,7. However, therapeutic success critically depends on the earliest possible diagnosis, warranting a detailed understanding of the mechanisms prior to the onset of first cognitive symptoms. The locus coeruleus (LC) noradrenergic (NA) system is affected particularly early in AD. It is the first site where aberrant tau hyperphosphorylation (pTau) is detected, putatively kickstarting the spread of tau throughout the CNS⁸. Consequently, past research has focused intensely on the effects of pTau on LC physiology, while the role of Aβ in LC dysfunction has attracted only scant attention. Forebrain NA is almost solely derived from the LC and, as a function of its widespread axonal projections, regulates a variety of physiological processes including arousal and attention, sleep-wake-cycles, memory, energy homeostasis, cerebral blood flow and sensory processing, all of which are impaired in the progression of AD, though with differences in temporal progression³. Symptomatically, early olfactory dysfunction frequently marks the early onset of AD with prospective patients remaining cognitively normal and otherwise healthy^9,10. Although decreased olfactory sensitivity is apparent in ~85% of AD cases the underlying mechanisms remain a conundrum^1,11. Here, we ventured for a multifaceted approach to study the neural correlate of olfactory dysfunction in a mouse model of amyloidosis using a plethora of steady-state systems neuroscience techniques both ex vivo and in vivo and studied human post-mortem brain tissue to validate our mechanistic findings.

Early LC axon loss exclusive to the OB

LC axon loss has been reported at late disease stages in the APP^NL-G-F mouse model¹². By systematic comparison of multiple brain areas, we set out to analyse if LC axon loss might already be detected earlier in these animals (Fig. 1a). Surprisingly, we discovered an early LC axon degeneration exclusive to the OB starting between 1 and 2 months in APP^NL-G-F mice (Fig. 1a-d). While in 1-month-old animals, the LC axon density was unaltered compared to WT animals, we observed a 14% fibre loss at 2 months of age. This loss further progressed to 27% at 3 months, and 33% at 6 months. Notably, LC axons started to degenerate in other regions such as the hippocampus, piriform cortex and medial prefrontal cortex between 6 and 12 months at the earliest (Extended Data Fig. 1a,b). Similar to the cortex, the OB is composed of different layers, which are disparately innervated by the LC-NA system. We thus analysed layer-specific axon loss and identified the most densely innervated region, the internal plexiform layer to be the site of most prominent axon loss, followed by the external plexiform layer (Fig. 1d). OB microglia increased between 2 and 3 months of age without significant Aβ plaque deposition (Fig. 1f,g). We excluded NA cell loss in the LC to underlie axonal demise as we did not observe differences in LC neuron number in APP^NL-G-F mice when compared to WT animals at 12 months (Fig. 1h,i). We next asked whether early deposition of extracellular Aβ correlates with LC axonal damage. Intriguingly, we found LC fibre loss to be independent of the amount of extracellular Aβ (Extended Data Fig. 2a).

LC axon loss drives hyposmia

Early sensory manifestations such as hyposmia have been well described in prodromal AD (pAD), as have the contributions of NA to olfaction¹³. Thus, we set out to analyse whether LC axon loss results in impaired olfaction. We employed the buried food test, a well-established olfactory task to measure the ability of an animal to detect volatile odours¹⁴ (Fig. 1j). Food deprived WT animals rapidly started exploring the arena and usually uncovered the hidden food pellet within ~40 s. In contrast, 3-month-old APP^NL-G-F mice needed 60% more time to find the buried food pellet. The same phenotype was reproduced in 6 months old animals (Fig. 1k). We did not observe any differences when testing animals at 1 month of age which is consistent with the lack of LC axon degeneration at that time point (Fig. 1b,c,k). To rule out task-specific confounders we aimed to recapitulate our findings in a second olfactory task. To this end, we subjected 3-month-old WT and APP^NL-G-F mice to an odour sensitivity test (Fig. 1l-n). We exposed the animals to ascending concentrations of vanilla, a pleasant odour, and measured the time the animals spent interacting with the odour delivery stick (Fig. 1m,n). WT animals were readily attracted by a low odour concentration (dilution 1:1000) and repeatedly interacted with the odour stick, while APP^NL-G-F mice visited the interaction zone considerably later and less often. The same behaviour was observed when testing a high vanilla concentration (dilution 1:1; Fig. 1m,n). Collectively, these data reveal a consistent olfactory phenotype in APP^NL-G-F mice, starting at 3 months of age, which is hitherto the earliest behavioural manifestation described in this mouse model.

Impaired NA release links to hyposmia

Neurocircuit-homeostasis is able to partially balance molecular and structural changes or loss in case of neuropathological insults¹⁵. We thus aimed to understand whether LC axon loss translates into decreased NA release in the OB. In order to investigate potential changes in the concentration of NA in the OB of APP^NL-G-Fanimals, we performed NA ELISA. Interestingly, we did not observe a significantly different concentration of baseline NA in these animals compared to WT mice (Extended Data Fig. 3a). We thus hypothesized that a change in LC-NA would be more pronounced in stimulus-related NA release. We transduced the OB of 2‑month‑old WT and APP^NL-G-Fanimals with the NA sensitive biosensor GRAB_NE (G-protein-coupled receptor-activation-based sensor for noradrenaline) and implanted a chronic cranial window over the olfactory bulb¹⁶ (Fig. 2a). At 3 months of age, we performed in vivo acousto-optical 2-photon (AO-2P) microscopy in awake animals paired with olfactory stimulation by 10s long vanilla puffs (Fig. 2b-g). WT animals reliably and repeatedly responded to the odour delivery with a strong and long-lasting increase of fluorescence. In contrast, delivering odour to APP^NL-G-Fanimals revealed a drastically decreased response (Fig. 2b-g). As a control, neither an air puff only nor NA measurements in the cortex coupled to odour delivery elicited coherent changes in fluorescence (Fig. 2d, Extended Data Fig. 3b). Immunohistochemical validation revealed a solid transduction of the tissue in the OB of all animals and NA fibre loss in
APP^NL-G-Fmice (Fig. 2h,i). To exclude the possibility of dysfunctional mitral cells, the first order projection neurons of the OB, driving impaired olfaction, we performed perforated patch-clamp recordings of mitral cells in acute OB slices. In line with previous studies, we found mitral cells to be spontaneously active, but we did not detect alterations of intrinsic properties between genotypes at 6 months of age at which hyposmia is well manifested in these animals (Extended Data Fig. 4a-f)¹⁷. The structure-to-function relationship of the LC-NA system and olfaction led us to further probe whether persistent activation of remaining LC axons by chemogenetics would be sufficient to reinstate olfaction (Extended Data Fig. 5a-c). We bilaterally injected an AAV transducing LC neurons of APP^NL-G-F x Dbh-Cre animals with an excitatory ligand-gated G-protein-coupled receptor (h3MDGs, designer-receptor exclusively activated by designed drugs, DREADD). In patch-clamp recordings, we confirmed that the application of Clozapine-N-Oxide (CNO) readily activates LC neurons (Extended Data Fig. 5a,b), however systemic CNO-injection to activate excitatory DREADDs in vivo failed to accelerate the time to find the buried food pellet in these animals (Extended Data Fig. 5c). This strongly suggests a structure-to-function-relationship of LC axons in the OB in the context of olfaction.

OB microglia clear LC axons

Microglia have been attracting considerable attention in the pathogenesis of AD¹⁸. Their remarkable heterogeneity has been revealed recently, highlighting the complex nature of microglia and their influence on brain functions¹⁹. Since early LC axon loss coincides with an increased number of microglia, we set out to investigate whether microglia could account for LC axon loss. Thus, we performed bulk RNA sequencing (RNA-seq) of microglia isolated from OBs of WT and APP^NL-G-F mice at the age of 2 months, the very onset of LC axon loss (Fig. 3a). In line with our immunohistological data, we observed an increased number of microglia cells isolated from bulbi of APP^NL-G-F animals (Fig. 3b). After appropriate quality control (Extended Data Fig. 6), we performed differential expression testing using negative binomial models while controlling for sex. This revealed that 2.344 genes (of a total of 17.840) were differentially expressed, with a slight majority of them (1.283) being upregulated in APP^NL-G-F animals (Fig. 3c, Extended Data Table 1). Previous work has demonstrated a so-called “disease-associated” microglia response (DAM) in AD mouse models and humans alike^20,21. To test whether this phenotype was visible in our data, we directly compared our microglia OB RNA-seq data to a publicly available cortical microglia RNA-seq dataset taken from 8‑month‑old APP^NL-G-F mice²². Linear regression of log-fold changes in fact revealed a significant negative relationship (R = -0.44, p < 2e-16), suggesting that no such DAM response is seen in 2-month-old OBs and that these microglia did not yet acquire a similar response to pathological stressors (Fig. 3d). A crucial function of microglia is the removal of debris or apoptotic cells from the parenchyma as well as synaptic remodelling²³. Interestingly, gene ontology (GO) term analysis revealed the 20 most enriched terms relate to neuronal function and synaptic or neuronal plasticity. We thus hypothesized that microglia phagocytosis, a component of synaptic pruning, might be responsible for the selective clearance of LC axons in the olfactory bulb. We compared all identified transcripts annotated to the GO term “phagocytosis”. Here, we identified 121 transcripts, of which only 2 were differentially expressed in our data set (Extended Data Fig. 7a). However, when analysing gene modules related to the GO-term “synapse”, we observed an overarching upregulation of 73 genes, suggesting an increased plastic environment, potentially indicating increased synaptic pruning (Fig. 3e). Thus, we conducted an automated phagocytosis assay from primary OB microglia of WT and APP^NL-G-F mice, aged 2 months (Fig. 3g). Microglia were incubated with pHrodo-labelled synaptosomes to measure their phagocytic uptake over the course of 24 hours (Fig. 3f‑i). Our data revealed an increased efficiency of APP^NL-G-F microglia to phagocytose fluorescently labelled synaptosomes, with OB microglia of APP^NL-G-F mice showing a 33% higher phagocytic capacity already after 12 hours. As expected, Cytochalasin-D application completely abolished phagocytosis in both genotypes (Fig. 3h,i). Based on their increased phagocytic activity, we hypothesized that microglia might indeed be phagocytosing LC axons in OBs from APP^NL-G-Fmice. To test this directly, we performed high-resolution imaging of NET fibres together with microglia and the lysosomal marker CD68 and subsequently performed 3D-reconstructions of these images (Fig. 3j). We found a higher volume of NET⁺ immunosignal in single microglia cells from APP^NL-G-F mice compared to WT animals, as well as increased volumes of lysosomal CD68 (Fig. 3k), corroborating the increase in phagocytic activity observed in vitro. Notably, we did not see significant differences in the cellular volumes of single microglia between groups. Collectively, our data show no overt disease-associated activation of microglia, but a strikingly increased phagocytic activity compared to WT animals of the same age. Consequently, we hypothesized that an inhibition of phagocytosis could prevent the loss of NA axons in the OB. Translocator protein 18 kDa (TSPO) has recently been identified as a key-protein in fuelling synaptic pruning and microglial phagocytosis^24,25. We sought to investigate if TSPO elimination would be sufficient to halt or decelerate the loss of LC axons. To this end, we bred mice with a global knockout of TSPO²⁶ to APP^NL-G-F. We again harvested OBs from these animals at 2-6 months of age and stained for NET⁺LCaxons. Indeed, lack of TSPO in APP^NL-G-F mice abrogated the loss of NA axons in these animals up to an age of 6 months (Fig. 4a,b). This correlated with a decreased uptake of NET⁺ axons in microglia of TSPO-KO x APP^NL-G-Fmice (Fig. 4d,e). We then exposed the TSPO-KO x APP^NL‑G-Fanimals to the buried food task. Importantly, the preservation of LC axons in the OB resulted in a retained ability to find the buried food pellet indistinguishable from WT animals (Fig. 4c).

PS labels LC axons for phagocytosis

A plethora of “find-me”- and “eat-me”-signals attracting microglia to their phagocytic targets have been revealed within the last years²⁷. The complement cascade has emerged as one key player of synaptic removal in AD²⁸. We thus aimed to analyse whether LC axons from APP^NL‑G‑F mice would be decorated by Complement component 1q (C1q) as a possible underlying cause of axonal clearance. As expected, staining for C1q resulted in a dense punctate pattern. However, we did not observe any significant changes of C1q colocalisation to NET⁺ axons in the OBs of APP^NL-G-F mice compared to WT mice (Extended Data Fig. 8a,b). In both healthy and diseased brains, the highly coordinated local externalization of phosphatidylserine (PS) leads to the targeted engulfment of neuronal material by microglia and has similarly been described to contribute to synapse loss in AD mouse models^29,30. A variety of microglial receptors are known to recognize exposed PS, such as triggering receptor expressed in myeloid cells 2 (TREM2) and milk fat globule-EGF factor 8 protein (MFG-E8), which in turn binds to microglial vitronectin receptors (the α_vβ3/5 integrins), both of which play major roles in the aetiology of AD^29,31,32. While PS recognized by TREM2 was shown to contribute to synapse loss in APP^NL-Fmice, PS and MFG-E8 are important physiological mediators of microglia-dependent synaptic pruning during adult neurogenesis in the OB of mice²⁹. Considering the increase of mRNAs associated with synaptic plasticity (Fig. 3e), we hypothesized that increased PS externalization might be the underlying cause of LC axon phagocytosis by microglia. To test this, we performed in vivo PS labelling by injecting PSVue550 in the OBs of WT and APP^NL-G-F mice at the age of 5 months. Importantly, as shown previously and in line with its physiological function, we could visualize externalized PS in the OB, both in WT and APP^NL-G-F mice. In order to assess whether PS externalization can be detected on NET⁺ axons, we conducted a colocalisation analysis using 3D reconstruction. When adjusting for the fibre density, we found an elevated colocalisation of PS on NET⁺ axons in APP^NL-G-F mice (Fig. 4f,g). Intriguingly, flipped PS was often accompanied by Iba1⁺ microglia directly contacting LC axons. However, when analysing the contact points between microglia and LC axons, no difference in colocalised volume was found between the genotypes (Fig. 4h,i). Further investigating the possible link, we could show that PS is capped with MFG-E8, serving as the adaptor protein between PS and the microglial integrin receptor (Extended Fig. 9a). Using 3D reconstruction, we found more MFG-E8 colocalised to LC axons of APP^NL-G-F mice than on LC axons from WT animals (Fig. 4j,k). Given the TSPO-KO mediated rescue of LC axons and olfaction, we hypothesized that MFG-E8 decoration should similarly be increased in APP^NL-G-F x TSPO-KO mice. We stained OB tissue from these animals for LC axons and MFG-E8 and again reconstructed both signals. Intriguingly, MFG-E8 decoration of LC axons was clearly increased compared to WT animals and even showed a trend towards an increase compared to APP^NL-G-Fmice (Fig. 4j,k). Overall, we conclude that local PS externalization in conjunction with MFG-E8 decoration constitutes a major “eat-me” signal for microglia interaction with LC axons and subsequent phagocytosis. We finally ventured to elucidate mechanistically as to why PS is externalized on LC axons. In neurons, the protein TMEM16F constitutes a Ca²⁺-dependent scramblase responsible for PS externalization. Earlier work has put much emphasis on the firing properties of LC neurons and the Ca²⁺-dependence of their intrinsic pacemaker, especially in the context of neurodegeneration³³. During pacemaking activity of LC neurons, each action potential (AP) is accompanied by a Ca²⁺-driven supra-threshold oscillation, which leads to the activation of voltage-gated sodium channels underlying the super-threshold AP. We thus hypothesized that increased firing in LC neurons may underlie Ca²⁺-triggered scramblase to flip PS to the outside of the plasma membrane. We performed perforated patch-clamp recordings of LC neurons from WT and APP^NL-G-F mice at the age of 6 months (Fig. 4l-q). Indeed, we found an overall increase in spontaneous AP frequency in acute brain slices from APP^NL-G-F mice (Fig. 4m,n). We did not observe a change in input resistance during hyperpolarization but a slightly decreased intrinsic excitability in response to depolarizing stimuli, likely reflecting an increased activation of Ca²⁺-dependent potassium channels (Fig. 4o-q). We thus conclude that spontaneous hyperactivity in LC neurons and consequently elevated Ca²⁺-signalling instigates Ca²⁺-dependent scramblase/flippase, leading to the externalization of PS and a microglia-mediated removal of hyperactive LC originating axons. In summary, we clearly pinpoint microglial phagocytosis of NA axons in the OB to be the underlying cause of the progressive early axon loss in APP^NL-G-F mice.

LC-APP^NL-G-Fexpression induces hyposmia

In APP^NL-G-F mice, every APP expressing cell harbours three mutations, limiting conclusion about the relative effect of LC axon loss³⁴. Thus, we asked whether APP^NL-G-Fexpression restricted to the LC would be sufficient to recapitulate the neuroanatomical and behavioural findings. We engineered a custom-built Cre-dependent AAV to specifically transduce LC neurons of Dbh-Cre mice with the human APP^NL-G-F (Dbh-hAPP^NL-G-F) or a control virus leading to the expression of a fluorophore only (Dbh-EYPF; Fig. 5a). Three-months post injection, we performed a buried food test. Of note, Dbh-hAPP^NL-G-F mice needed more time to find the buried food compared to the control injected Dbh-EYPFmice (Fig. 5d,e). Immunohistochemical validation revealed an LC axon degeneration of 15% in the OB of Dbh‑hAPP^NL-G-Fmice compared to Dbh-EYPFmice (Fig. 5b,c), without LC neuron loss (Extended Data Fig. 10a,b). We thus asked next, whether again microglia in the OB would phagocytose LC axons and performed the same set of immunohistological staining to assess NET protein within CD68⁺ lysosomes of microglia. Indeed, we observed an increase in the volume of NET⁺ signal inside the lysosomes of microglia (Fig. 5f,g). Collectively, our approach to induce Dbh-hAPP^NL-G-F expression specifically in LC neurons illustrates that this is sufficient to recapitulate both early behavioural and neuropathological phenotypes observed in the APP^NL-G-F mouse line.

LC axon loss and hyposmia in human pAD

Early impairment of the LC-NA system in humans has recently been in the spotlight of several multimodal imaging studies³⁵. While at the level of the brainstem, LC volume decreases over time and levels of LC integrity predict cognitive outcome in elderly subjects, it is not yet clear whether axon loss also precedes late-phase occurring cell loss in the LC of humans³⁶. Interestingly, both hyposmia and LC integrity are predictors of cognitive decline in humans^9,10. We thus ventured to decipher whether LC axon degeneration is evident in post-mortem tissue from OBs of early AD cases, staged by Aβ and tau immunostainings (Thal-phase 1-2, Braak stage 1-2) and healthy controls. Strikingly, in the OB tissue from early AD cases, we revealed a pronounced degeneration of NET⁺ fibres compared to healthy, age-matched controls, which did not further decline in progressive AD cases (Fig. 6a-c). Moreover, we hypothesized that LC axon loss in humans, similar to mice, may correlate with an increased number of microglia. To this end, we performed TSPO-PET imaging in 16 patients with subjective cognitive decline (SCD)/ mild cognitive impairment (MCI), 16 AD patients and 14 healthy controls, staged by Aβ and tau cerebrospinal fluid (CSF) levels, and investigated their TSPO signal in the respective OBs. We identified increased TSPO signals in the OBs of patients with prodromal AD, indicative of increased numbers or activation of microglia. Interestingly, even transitioning into AD diagnosis did not further elevate OB TSPO signals significantly (Fig. 6d,e). A number of independent longitudinal studies have highlighted olfactory deficits as a predictor of cognitive decline^2,37–40. Thus, we analysed the data of our cohort for signs of hyposmia. While the prodromal AD group showed a trend towards olfactory deficits, patients transitioned into AD indeed revealed a significant decrease in the ability to identify common odours (Fig. 6f). Consequently, we asked whether these findings could be back-translated to APP^NL-G-F mice. Indeed, TSPO-PET imaging in these animals revealed an early elevated signal in the OB compared to WT mice at 2-3 months of age, while the signal in the cortex of the same animals at that age remained unaltered (Fig. 6g-j). Thus, these translational data highlight and assign TSPO-PET imaging of the OB and hyposmia as a potential early bio-marker of AD and LC-NA system dysfunction.

We reveal LC-NA system degeneration as an impaired neuronal network to account for olfactory deficits in AD¹. In humans, ~85% of AD patients exhibit early sensory deficits including hyposmia and anosmia, predicting cognitive decline^{1,2,11,37–40}. Similarly, LC integrity is established as an early biomarker predicting cognitive decline in ageing and neurodegenerative diseases^35,36. Interestingly, hyposmia is well documented in Parkinson’s disease (PD) and LC dysfunction has been implicated to drive prodromal symptoms in PD. In contrast to the LC in AD, the OB and the dorsal motor nucleus of the vagus are the first sites to display α-synuclein pathology, likely suggesting an impairment of first-order olfactory neurons⁴¹. The well-established modulation of olfaction by LC-derived NA, especially in olfactory memory, underscores a possible link from LC vulnerability to hyposmia⁴². In our study, we detected LC axon loss in post-mortem OB tissue from prodromal AD patients. Notably, this pronounced early degeneration of LC axons did not progress further at later stages. Similarly, microgliosis detected by an elevated TSPO-PET signal in the OBs of SCD/MCI patients did not continue to increase in diagnosed AD patients. The same AD patients showed a strong olfactory deficit, while we could only assign a slight trend to the prodromal AD group. Based on the substantial evidence of several independent studies that highlight hyposmia as a common early symptom in AD, we believe that this is likely due to our small cohort size^{2,37–40,43–45}. It is reasonable to hypothesize that hyposmia and LC integrity as independent predictors of cognitive decline may not only be correlating but may be causally linked. Indeed, early sophisticated work suggested that pharmacotoxic lesion of the LC exaggerates olfactory problems in APPPS1 mice, however, the experiments were conducted after nine months of consecutive toxin administration in 12-month-old animals⁴⁶. We here provide the first causal link between the LC and olfactory deficits in mice. While we clearly provide translational data, more research is needed to further confirm this in human patients. The fast progress in MRI resolution and the sophisticated identification of the LC will enable a more detailed examination of the causal link between these two phenomena. Functional connectivity in live patients together with resting-state activity may then be able to delineate putative interconnections between these two widely separated anatomical regions.

LC dysfunction has classically been viewed as a consequence of tau-pathology. It is considered to be the first region positive for hyperphosphorylated tau⁴⁷. Due to this tau-centric view of LC dysfunction, the role of APP and Aβ pathology in the LC in the aetiology of AD has only attracted little attention, although Aβ increases as a function of LC connectivity in rats⁴⁸. In line, we provide evidence for an APP mutation-dependent axon loss underlying early olfactory deficits^47,49, marking the earliest described phenotype in this widely used AD mouse model to date. Functionally, the pronounced reduction of NA-release in APP^NL-G-F mice upon odour stimulation can be considered a strong driver of the olfactory phenotype. With our cell-type specific expression of APP^NL-G-F in LC neurons, we were able to demonstrate a coherent relationship between LC axon loss and olfactory deficits. Mechanistically, we present clear evidence that the expression of mutant human APP^NL-G-F instigates the externalization of PS on LC axons. The Ca²⁺-dependence of this externalization is in line with the hyperactivity observed in our study. Moreover, similar AP frequency elevations have been recorded in APPPS1 animals^29,50. In the olfactory bulb, PS-dependent microglial phagocytosis plays a crucial role in both physiology and pathology. During development and adult neurogenesis, microglia mediate synaptic pruning via PS detection which serves as a key mechanism to integrate newborn neurons into functional neuronal networks. Thus, PS located on hyperactive LC axons may be detected with a higher probability and fidelity compared to other regions, providing a rationale for the early axon loss preceding all other highly LC-innervated regions. This is additionally reflected in the lack of an amyloid-driven DAM response in microglia extracted from the OB and the lack of changes in microglia contacts to NET⁺ axons. PS has recently been recognized as an opsonin in AD that marks neuronal structures for removal²⁸. A variety of different receptors or effector-proteins subsequently trigger microglia-dependent clearance, including TREM-2 and MFG-E8. In line with the physiological role of PS-dependent microglia-driven synaptic remodelling, we reveal MFG-E8 as a mediator of microglia-dependent phagocytosis of LC axons. Our data supports the hypothesis that the OB is an anatomical region prone to detection of PS-MFG-E8 complexes by microglia and thus axons of hyperactive LC neurons are cleared with a higher fidelity compared to other regions involving PS-MFG-E8 driven synaptic remodelling. In summary, we provide the first underlying mechanism for hyposmia, a so far underappreciated sensory deficit in AD. Coordinated assessment of structural and functional connectivity, olfactory testing, together with CSF and blood biomarkers could facilitate earlier AD diagnosis and be employed as solid predictors of disease progression and outcome. Ultimately, this may open the window for the earliest treatment to halt or decelerate disease progression.

The use of human post-mortem tissue was approved by the Ethics Committee of the Ludwig-Maximilians University Munich. The human TSPO-PET / Sniffing stick study was approved by the Ethics Committee of LMU University Hospital Munich.

Acknowledgements

We thank T. Saito and T. C. Saido for providing the APP^NL-G-Fmice. Special thanks goes to Fang Zhang, Michael Schmidt, Anke Jürgensonn, Marcel Matt and Brigitte Haslbeck for outstanding technical and administrative assistance; the whole staff, especially Ekrem Göcmenoglu, of the animal facility at the Center for Stroke and Dementia under the lead of Dr. Anne van Thaden, Dr. Carolin Ildiko Konrad and Dr. Manuela Schneider for their continuous support on all animal related efforts. We thank Prof. Dr. Neville Vassallo and Svenja Rumpf for vivid discussions and manuscript revision. This work was partly supported by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) under DFG Research Unit FOR 2858 (project number 403161218; applies to L.P., M.B., S.T., J.H.), DFG Priority Programme SPP2395 (TA 551/2-1; applies to S.T) and under Germany’s Excellence Strategy within the framework of the Munich Cluster for Systems Neurology (EXC 2145 SyNergy– ID 390857198; applies to R.P., M.B., J.J.N and J.H.). R.P. is supported by the German Center for Neurodegenerative Diseases (Deutsches Zentrum für Neurodegenerative Erkrankungen, DZNE), the Davos Alzheimer’s Collaborative, the VERUM Foundation, the Robert-Vogel-Foundation, the National Institute for Health and Care Research (NIHR) Sheffield Biomedical Research Centre (NIHR203321), the University of Cambridge and the Ludwig-Maximilians-University Munich Strategic Partnership within the framework of the German Excellence Initiative and Excellence Strategy and the European Commission under the Innovative Health Initiative program (project 101132356).

Author contributions

C.M designed and conducted most experiments (electrophysiological recordings, immunohistological staining, olfactory behaviour tests, imaging of mouse and human brain tissue, ELISA, data analysis and 3D reconstruction) and performed manuscript preparation.T.N and performed olfactory bulb window surgery and NA 2-photon in vivo measurements. P.F. performed olfactory bulb window surgery and NA 2-photon in vivo measurements and analysed olfactory sensitivity tests. F.S performed olfactory bulb microglia sequencing and subsequent data analysis. B.R performed human odour identification tests and subsequent analysis. J.G performed immunofluorescent staining, confocal imaging and 3D reconstruction. Y.T performed microglia isolation, phagocytosis assay and subsequent analysis. K.O performed microglia isolation for sequencing.K.W and G.B performed small animal PET study and analysis.J.W helped establish the MFG-E8 antibody stain. S.G., C.K. and M.S. performed human odour identification tests and subsequent analysis. R.B.B., G.L. and R.J.M. provided TSPO-KO mice. R.P performed human PET study and analysis. J.J.N. performed project planning. S.T performed project planning. M.B performed small animal PET study and analysis.J.H performed project planning and manuscript revision. L.P performed virus injections, project planning and supervision and wrote the manuscript with input from all authors. All authors provided comments and approved the manuscript.

Competing interests

M.B. received consulting/speaker honoraria from Life Molecular Imaging, GE healthcare, and Roche, and reader honoraria from Life Molecular Imaging. All other authors declare no competing interests.

Data availability

All data necessary (source data) for the conclusions of the study are provided with the Article. Further request should be addressed to the corresponding author.

Code availability

Any code generated within this study will be shared upon request.

Additional information

Supplementary Information is available for this paper. Correspondence and requests for materials should be addressed to the corresponding author.

Murphy, C. Olfactory and other sensory impairments in Alzheimer disease. Nature Reviews Neurology 15, 11–24 (2019).
Guo, J. et al. Trajectories of olfactory identification preceding incident mild cognitive impairment and dementia: a longitudinal study. eBioMedicine 98, 104862 (2023).
Poe, G. R. et al. Locus coeruleus: a new look at the blue spot. Nature Reviews Neuroscience 21, 644–659 (2020).
Alzheimer’s disease facts and figures. Alzheimer’s Dement. 19, 1598–1695 (2023).
Busche, M. A. & Hyman, B. T. Synergy between amyloid-β and tau in Alzheimer’s disease. Nat. Neurosci. 23, 1183–1193 (2020).
Reardon, S. Alzheimer’s drug donanemab: what promising trial means for treatments. Nature 617, 232–233 (2023).
Sims, J. R. et al. Donanemab in Early Symptomatic Alzheimer Disease. JAMA 330, 512–527 (2023).
Braak, H., Thal, D. R., Ghebremedhin, E. & Tredici, K. D. Stages of the Pathologic Process in Alzheimer Disease: Age Categories From 1 to 100 Years. J. Neuropathol. Exp. Neurol. 70, 960–969 (2011).
Dintica, C. S. et al. Impaired olfaction is associated with cognitive decline and neurodegeneration in the brain. Neurology 92, 10.1212/WNL.0000000000006919 (2019).
Pacyna, R. R., Han, S. D., Wroblewski, K. E., McClintock, M. K. & Pinto, J. M. Rapid olfactory decline during aging predicts dementia and GMV loss in AD brain regions. Alzheimer’s Dement. 19, 1479–1490 (2023).
Woodward, M. R. et al. Validation of olfactory deficit as a biomarker of Alzheimer disease. Neurol.: Clin. Pr. 7, 5–14 (2017).
Sakakibara, Y. et al. Widespread Reduced Density of Noradrenergic Locus Coeruleus Axons in the App Knock-In Mouse Model of Amyloid-β Amyloidosis. J Alzheimer’s Dis 1–18 (2021) doi:10.3233/jad-210385.
Linster, C. & Escanilla, O. Noradrenergic effects on olfactory perception and learning. Brain Res. 1709, 33–38 (2019).
Yang, M. & Crawley, J. N. Simple Behavioral Assessment of Mouse Olfaction. Curr. Protoc. Neurosci. 48, 8.24.1-8.24.12 (2009).
Calafate, S. et al. Early alterations in the MCH system link aberrant neuronal activity and sleep disturbances in a mouse model of Alzheimer’s disease. Nat. Neurosci. 26, 1021–1031 (2023).
Feng, J. et al. A Genetically Encoded Fluorescent Sensor for Rapid and Specific In Vivo Detection of Norepinephrine. Neuron 102, 745-761.e8 (2019).
Kostka, J. K., Gretenkord, S., Spehr, M. & Hanganu‐Opatz, I. L. Bursting mitral cells time the oscillatory coupling between olfactory bulb and entorhinal networks in neonatal mice. J. Physiol. 598, 5753–5769 (2020).
Leng, F. & Edison, P. Neuroinflammation and microglial activation in Alzheimer disease: where do we go from here? Nat. Rev. Neurol. 17, 157–172 (2021).
Masuda, T. et al. Spatial and temporal heterogeneity of mouse and human microglia at single-cell resolution. Nature 566, 388–392 (2019).
Keren-Shaul, H. et al. A Unique Microglia Type Associated with Restricting Development of Alzheimer’s Disease. Cell 169, 1276-1290.e17 (2017).
Monasor, L. S. et al. Fibrillar Aβ triggers microglial proteome alterations and dysfunction in Alzheimer mouse models. Elife 9, e54083 (2020).
Sobue, A. et al. Microglial gene signature reveals loss of homeostatic microglia associated with neurodegeneration of Alzheimer’s disease. Acta Neuropathol. Commun. 9, 1 (2021).
Paolicelli, R. C. et al. Microglia states and nomenclature: A field at its crossroads. Neuron 110, 3458–3483 (2022).
Shi, Y. et al. Long-term diazepam treatment enhances microglial spine engulfment and impairs cognitive performance via the mitochondrial 18 kDa translocator protein (TSPO). Nat Neurosci 1–13 (2022) doi:10.1038/s41593-022-01013-9.
Fairley, L. H. et al. Mitochondrial control of microglial phagocytosis by the translocator protein and hexokinase 2 in Alzheimer’s disease. Proc. Natl. Acad. Sci. 120, e2209177120 (2023).
Banati, R. B. et al. Positron emission tomography and functional characterization of a complete PBR/TSPO knockout. Nat. Commun. 5, 5452 (2014).
Brown, G. C. & Neher, J. J. Microglial phagocytosis of live neurons. Nat. Rev. Neurosci. 15, 209–216 (2014).
Lv, Z. et al. Clearance of β-amyloid and synapses by the optogenetic depolarization of microglia is complement selective. Neuron 112, 740-754.e7 (2024).
Rueda‐Carrasco, J. et al. Microglia‐synapse engulfment via PtdSer‐TREM2 ameliorates neuronal hyperactivity in Alzheimer’s disease models. EMBO J. 42, e113246 (2023).
Kurematsu, C. et al. Synaptic pruning of murine adult-born neurons by microglia depends on phosphatidylserine. J. Exp. Med. 219, e20202304 (2022).
Wagner, J. et al. Medin co-aggregates with vascular amyloid-β in Alzheimer’s disease. Nature 612, 123–131 (2022).
Tzioras, M. et al. Human astrocytes and microglia show augmented ingestion of synapses in Alzheimer’s disease via MFG-E8. Cell Rep. Med. 4, 101175 (2023).
Sanchez-Padilla, J. et al. Mitochondrial oxidant stress in locus coeruleus is regulated by activity and nitric oxide synthase. Nature Neuroscience 17, 832–840 (2014).
Saito, T. et al. Single App knock-in mouse models of Alzheimer’s disease. Nature Neuroscience 17, 661–663 (2014).
Jacobs, H. I. L. et al. In vivo and neuropathology data support locus coeruleus integrity as indicator of Alzheimer’s disease pathology and cognitive decline. Sci Transl Med 13, eabj2511 (2021).
Bueichekú, E. et al. Spatiotemporal patterns of locus coeruleus integrity predict cortical tau and cognition. Nat. Aging 1–13 (2024) doi:10.1038/s43587-024-00626-y.
Liu, S. et al. Disparity of smell tests in Alzheimer’s disease and other neurodegenerative disorders: a systematic review and meta-analysis. Front. Aging Neurosci. 15, 1249512 (2023).
Dintica, C. S. et al. Impaired olfaction is associated with cognitive decline and neurodegeneration in the brain. Neurology 92, 10.1212/WNL.0000000000006919 (2019).
Pacyna, R. R., Han, S. D., Wroblewski, K. E., McClintock, M. K. & Pinto, J. M. Rapid olfactory decline during aging predicts dementia and GMV loss in AD brain regions. Alzheimer’s Dement. 19, 1479–1490 (2023).
Audronyte, E., Pakulaite-Kazliene, G., Sutnikiene, V. & Kaubrys, G. Odor Discrimination as a Marker of Early Alzheimer’s Disease. J. Alzheimer’s Dis. 94, 1169–1178 (2023).
Borghammer, P. et al. A postmortem study suggests a revision of the dual-hit hypothesis of Parkinson’s disease. npj Park.’s Dis. 8, 166 (2022).
Eckmeier, D. & Shea, S. D. Noradrenergic Plasticity of Olfactory Sensory Neuron Inputs to the Main Olfactory Bulb. Journal of Neuroscience 34, 15234–15243 (2014).
Audronyte, E., Pakulaite-Kazliene, G., Sutnikiene, V. & Kaubrys, G. Properties of odor identification testing in screening for early-stage Alzheimer’s disease. Sci. Rep. 13, 6075 (2023).
Igeta, Y., Hemmi, I., Yuyama, K. & Ouchi, Y. Odor identification score as an alternative method for early identification of amyloidogenesis in Alzheimer’s disease. Sci. Rep. 14, 4658 (2024).
Liu, D. et al. Olfactory deficit: a potential functional marker across the Alzheimer’s disease continuum. Front. Neurosci. 18, 1309482 (2024).
Rey, N. L. et al. Locus coeruleus degeneration exacerbates olfactory deficits in APP/PS1 transgenic mice. Neurobiology Of Aging 33, 426.e1–11 (2012).
Weinshenker, D. Long Road to Ruin: Noradrenergic Dysfunction in Neurodegenerative Disease. Trends in Neurosciences 1–13 (2018) doi:10.1016/j.tins.2018.01.010.
Ross, J. A., Reyes, B. A. S., Thomas, S. A. & Bockstaele, E. J. V. Localization of endogenous amyloid-β to the coeruleo-cortical pathway: consequences of noradrenergic depletion. Brain Struct Funct 223, 267–284 (2018).
Chalermpalanupap, T., Weinshenker, D. & Rorabaugh, J. M. Down but Not Out: The Consequences of Pretangle Tau in the Locus Coeruleus. Neural plasticity 2017, 1–9 (2017).
Kelly, L. et al. Identification of intraneuronal amyloid beta oligomers in locus coeruleus neurons of Alzheimer’s patients and their potential impact on inhibitory neurotransmitter receptors and neuronal excitability. Neuropath Appl Neuro 47, 488–505 (2021).

Animals. Mice, both male and female (1-6 months of age) were used and held on a 12-h light/dark cycle with food and water ad libitum. The APP^NL-G-Fmouse line is a knock-in model, were pathogenic Aβ is elevated by inserting 3 different mutations, associated with AD³⁴. Crossing APP^NL-G-Fmice with Dbh-Cre was used to manipulate the locus coeruleus-noradrenergic system. Dbh-Cre mice express the Cre recombinase under the dbh (dopamine beta hydroxylase) promotor⁵¹. APP^NL-G-Fmice were also crossed with TSPO-KO²⁶ mice to access the effect of a TSPO knock-out on the noradrenergic system. As control animals, C57BL/6J mice were used, purchased from the Jackson Laboratory (Maine, United States). All animal experiments were approved by the Government of Upper Bavaria and followed the regulations of the Ludwig Maximilian University of Munich.

Immunostaining: Mouse brain tissue. Mice were deeply anesthetized and transcardially perfused with phosphate-buffered saline (PBS) and 4% paraformaldehyde (PFA). Brains got fixed by immersion in PFA at 4°C for 16 h. 50 µm thick slices were cut in a coronal plane using a vibratome (VT1200S, Leica Biosystems). Each 4 slices per animal containing the olfactory bulb, piriform cortex, hippocampus and locus coeruleus were used for an immunostaining analysis. Staining was performed on free-floating sections. Slices were blocked with blocking solution (10 % normal goat serum and 10 % normal donkey serum in 0.3 %Triton and PBS) for 2 hours at RT. Primary antibodies were incubated over-night at 4°C, followed by washing and secondary antibody incubation for 2 hours at RT, protected against light. Slices were mounted and cover slipped with mounting medium, containing DAPI (Dako, Santa Clara, USA).Primary antibodies used were: rabbit anti-NET (1:500, Abcam, ab254361), mouse anti-NET (1:1000, Thermo Fisher, MA5-24547), guinea pig anti-Iba1 (1:500, Synaptic Systems, 234308), chicken anti-TH (1:1000, Abcam, ab76442), mouse anti-Aß (NAB228) (1:500, Santa Cruz, sc-3277), rat anti-CD68 (1:500, BioRad, MCA1957), goat anti-MFG-E8 (1:500, R&D Systems, AF2805), rabbit anti-C1q (1:1000, Abcam, ab182451), chicken anti-GFP (1:1000, Abcam, ab13970), rabbit anti-GFP (1:1000, Thermo Fisher, A21311), rabbit, HA-tag (1:500, Sigma, H6908), Streptavidin 488 (1:1000, Invitrogen, S32354), Streptavidin 647 (1:1000, Invitrogen, S32357).

Image acquisition. Three-dimensional images were acquired with a Zeiss LSM900 confocal microscope (Carl Zeiss, Oberkochen).

NET fibre quantification. For the quantification of the NET fibre density as well as Iba1-microglia and NAB288-Aβ-plaque area, a 10x objective (8-bit stacks of 101.41 µm x 101.41 µm x 25 µm) was used. The staining density (area %) was analysed with ImageJ. After a manual brightness/contrast adjustment, a threshold was set to calculate the perceptual area of NET-positive LC fibres, Iba1-positive microglia and NAB288-positive Aβ plaques. Results from 4 sections per animal from 4-8 animals per groups were averaged and reported as mean ± s.e.m.

Colocalisation analysis. For the engulfment of NET in microglia, airyscan images were taken with a 63x/1.4x NA oil immersion objective. Z-stack images were acquired of 8 microglia per mouse from 3 animals per group in the external plexiform layer, covering 30 µm at 0.14 µm intervals. Colocalisation of Iba1⁺ microglia- NET⁺ LC axon contact points was analysed on 15 µm z-stack images (40x/1.3x magnification, 0.3 µm intervals) of 6 pictures per mouse, 3 mice per genotype. Colocalisation of PS on NET⁺ LC axon was analysed on 15 µm z-stack images (40x/0.7x magnification, 0.3 µm intervals) of 7 pictures per mouse, 3 mice per genotype. Colocalisation of C1q on NET⁺ LC axon was analysed on 6 µm z-stack images (63x/1.4x magnification, 0.18 µm intervals) of 5 pictures per mouse, 2 mice per genotype. Colocalisation of MFG-E8 on NET⁺ LC axon was analysed on 15 µm z-stack images (40x/0.7x magnification, 0.3 µm intervals) of 6 pictures per mouse, 4 mice per genotype. All images were 3-D reconstruction in IMARIS (Bitplane, 9.6.1) using the Surface module. Colocalisation was measured in volume and normalized to the NET axon density.

Staining: Human brain tissue. Human brain tissue from 7 healthy control subjects, 7 prodromal AD subjects and 6 AD patients was provided from the Munich brain bank. Demographic details of the subjects are listed in Supplementary Table 3. Paraffin embedded brain sections (5 µm) of the olfactory bulb were cut in a horizontal plane, using a microtome (Leica SM2010R) and mounted on glass slides until further processing. Sections were deparaffinized with xylene and rehydrated through a series of descending alcohol concentrations. For the DAB staining, an automated IHC/SH slice staining system (Ventana BenchMark ULTRA) was used. On separate slices, NET 1:200, Aß 1:5000 and Tau 1:400 was stained and visualized with an upright Bridgefield microscope. Each 4 pictures per subject (20x magnification) were acquired and analysed regarding their perceptual density of NET⁺ LC axons.

Microglia isolation. Primary microglia were isolated from the olfactory bulb of 2-month-old C57BL/6J and APP^NL-G-Fmice using MACS technology (Miltenyi Biotec) according to manufacturer’s instructions. Briefly, mice were perfused with PBS and the brain washed in ice cold HBSS (Gibco) supplemented with 7 mM HEPES (Gibco). Chopped tissue pieces were incubated with digestion medium D-MEM/GlutaMax high glucose and pyruvate (Gibco) supplemented with 20 U papain per ml (Sigma P3125) and 0.01 \% L-Cysteine (Sigma) for 15 min at 37 C in a water bath. Subsequently, enzymatic digestion was stopped using blocking medium 10 \% heat-inactivated FBS (Sigma) in D-MEM/GlutaMax high glucose and pyruvate. Mechanical dissociation was gently but thoroughly performed by using three fire-polished, BSA-coated glass Pasteur pipettes with decreasing diameter. Subsequently, microglia were magnetically labelled with CD11b microbeads (Miltenyi Biotec, 130-097-678) in MACS buffer (0.5 \% BSA, 2 mM EDTA in 1x PBS, sterile filtered) and the suspension loaded onto a pre-washed LS-column (Miltenyi Biotec, 130-042-401). Following washing with 3x1 ml MACS buffer, magnetic separation resulted in a CD11b enriched and a CD11b depleted fraction. To increase purity further, the microglia-enriched fraction was loaded onto another LS-column. Total numbers of obtained microglia fractions were quantified using C-Chip chambers (Nano EnTek, DHC-N01). Isolated primary microglia were washed twice with 1x PBS (Gibco) and immediately processed for sequencing or plated for a phagocytosis assay.

Phagocytosis assay. Synaptic Protein was enriched using the Syn-PER™ Synaptic Protein Extraction Reagent (Thermo Fisher) according to manufacturer’s protocol and published previously⁵². In brief, fresh brains from C57BL/6J mice at 4 months of age were isolated and homogenized in 10mL/g of brain tissue of Syn-PER™ reagent substituted with protease and phosphatase inhibitor. The homogenate was then centrifuged at 1200 x g at 4°C for 10 minutes. The supernatant containing the synaptic fraction was then transferred into a new tube and spun at 15.000 x g at 4°C for 20 minutes. The supernatant was aspirated and the pellet of synaptic protein was resuspended in 1mL of Syn-PER™ reagent containing 5 \% (v/v) DMSO per gram tissue originally used. Synaptosome extracts were then stored at -80° before further usage.Synaptic Protein was labelled with the pHrodo™ Red succinimidyl ester (Thermo Fisher Scientific), which emits a red fluorescent signal only in acidic environments. Labelling was performed as previously described⁵³. In brief, synaptic protein was washed in 100 mM sodium bicarbonate, pH 8.5 and spun down (17,000 x g for 4 min at 4C). pHrodo™ dye was dissolved in 150 µL DMSO per 1 mg dye to a concentration of 10 mM. The pHrodo™ stock solution was added to the synaptic protein at a concentration 1 µl pHrodo per 1 mg of synaptic protein. After incubating at room temperature for 2 hours, protected from light, the labelled protein was washed twice in DPBS and spun down (at 17,000 x g for 4 min at 4C). After resuspending synaptic protein with 100 mM sodium bicarbonate, pH 8.5 to a concentration of 1000 µg/ml, it was aliquoted and stored at -80°C before usage.Primary microglia were cultured in tissue culture treated 96-well plates in microglia-medium adding freshly 10 ng/ml GM-CSF (R&D Systems) for three days in vitro (DIV) at 37°C, 5 \% CO2, changing medium at DIV 1. For the phagocytic uptake assay, medium was replaced with medium in which pHrodo™ labelled synaptic protein was resuspended at the desired concentration (2.5 µg/mL). For the Cytochalasin D (CytoD) control, cells were treated with 10 µM CytoD (Sigma) for 30 minutes, before adding medium with labelled synaptic protein and CytoD. Immediately after adding the substrates the cells were placed in an Incucyte™ S3 Live-Cell Analysis System (Sartorius). Scans were performed every hour with 20x magnification and both phase contrast and red fluorescent channels, acquiring a minimum of three images per well and scan. Quantification was done using the cell-by-cell adherent analysis. Phagocytic index was calculated using the total integrated intensity (RCU x µm2/Image) normalized to the number of cells per image.

NA Elisa. In order to measure potential difference in the noradrenaline concentration between C57BL/6J mice and APP^NL-G-Fmice, a noradrenaline ELISA was carried out. Mice were deeply anesthetized and perfused with PBS and their brains got rapidly removed. The olfactory bulb was dissected and snap frozen using liquid nitrogen. The tissue was homogenized in 0.01M HCl in the presence of 0.15 mM EDTA and 4 mM sodium metabisulfite, before being processed with an ELISA kit (BA E-5200) according to the manufacturer’s protocol.

RNA sequencing and Bioinformatics. RNA was isolated from microglial cell pellets using the RNeasy Plus Micro kit (Qiagen, 74034). Briefly, samples were lysed with RLT Plus lysis buffer containing beta-Mercaptoethanol, genomic DNA was removed by passing the lysate through gDNA eliminator columns, and the eluate was applied to RNeasy spin columns. Contaminants were removed with repeated Ethanol washes before RNA was eluted with 20 µL molecular grade water. All steps were carried out automatically on a Qiacube machine. RNA was quantified on a Qubit Fluorometer (Invitrogen, Q33230) and 6 ng of total RNA were used as input for library preparation with the Takara SMART-seq Stranded kit (Takara, 634444) following the manufacturer’s instructions. Fragmentation time was kept at 6 minutes and AMPure XP beads (Beckman Coulter, A63880) were used for all clean-up steps. Library QC using a Bioanalyzer revealed average insert sizes around 350 bps. The molarity of each of the 16 libraries was determined by using the ddPCR Library Quantification Kit for Illumina TruSeq (Bio-Rad, 1863040) according to the manufacturer’s instructions. Libraries were then diluted to 4 nM and pooled in an equimolar fashion. Paired-end sequencing was carried out for 150 cycles on a NextSeq 550 sequencer (Illumina, 20024907) using a High-Output flow cell. After sample demultiplexing, reads were aligned using STAR v2.7.8 to a customized genome based on the GRCm39 assembly and the gencode vM32 primary annotation that additionally contained sequences and annotations for the human APP gene. Group assignments were verified by manually inspecting alignments to the (human) APP sequence and checking for presence of the NL-, G- and F- mutations in transgenic animals. The count matrix produced by STAR v2.7.8 was used as an input for differential expression testing using edgeR. The count matrix was filtered to retain genes with at least 5 counts in at least 50% of samples and quasi-likelihood tests were conducted after fitting appropriate binomial models. Differential expression was considered significant if FDR < 0.1 and if the absolute log-fold-change exceeded 0.5. Gene lists were annotated with the enrichR package. All analyses made heavy use of the tidyverse and ggplot2 packages and were performed on a server running Arch Linux, R version 4.3.2 and Rstudio Server 2023.03.0.

Behavioural olfactory tests. All behavioural experiments were conducted during the light-phase of the animals and were performed in a blinded manner. To evaluate possible differences in odour performance, C57BL/6J and APP^NL-G-Fmice at 1, 3 and 6 months of age underwent a buried food test. One day before the test, animals got food deprived for 18 hours. On the test day, animals got acclimated to the new environment for at least 30 minutes in a fresh cage with increased bedding volume. The test begins with placing the animal in the test cage with a food pellet buried in the bedding. The time it takes for the animals to reach the food pellet was analysed based on a video recording. The mean search time that the two groups took to find the food pellet was calculated and compared by an unpaired student’s t-test. The sensitivity test evaluates whether mice can perceive odours even at weak concentrations. At the beginning of the experiment, the animals got acclimated to the odour applicator (a dry cotton swab without odour) for 30 minutes to exclude the applicator itself as a potential source of error and a new, interesting object. For the test, a pleasant-smelling odour “vanilla” got applied to a cotton swab in two ascending concentrations (1:1000 and 1:1 in water), and each concentration got presented to the mouse for 2 minutes consecutively, with 1 min break in between to change the odorant. Water, in which all odours are dissolved, was used as a control. Mice were filmed from the top and side with 2 synchronized cameras, and their nose was segmented and tracked offline in both videos using 2 S.L.E.A.P. networks (PMID: 35379947). A python code was used to track the 3D position of the nose relative to the odour dispersing cotton tip, and to quantify the time spent interacting with the different odour concentrations (investigation zone < 2 cm nose to cotton tip).

Virus injections. Different viral injection into the LC region or olfactory bulb were carried out in this study. For injections into the olfactory bulb the following coordinates were used: right OB (AP: 5.00, ML: -1.07, DV: 2.57) and left OB (AP: 4.28, ML: 0.41, DV: 2.45), while injection into the LC region were made using the following coordinate: left LC (AP: -5.44, ML: -0.89, DV: 4.07) and right LC (AP:-5.44, ML: -0.99, DV: 3.99). Adjustments were made if blood vessels were right on top of the injection location. AAV-hSyn-DIO-h3MDGs / AAV1-Syn-GCamp8f; Chemogenetic activation of LC neurons was carried out to investigate if an increase in noradrenaline release could rescue the impaired olfaction in APP^NL-G-F x Dbh-Cre mice. 5-month-old mice were bilaterally injected in the LC with AAV-hSyn-DIO-h3MDGs or the control AAV1-Syn-GCamp8f. To activate H3MDGs 1 month post injection, mice were injected i.p. with 1 mg/kg CNO 30 min before undergoing the buried food test. For patch clamp recordings, a concentration of 3 µM was used. AAV5-Flex-hSyn1-APP^NL-G-F-P2A-HA / AAV-5-Flex-Ef1α-EYFP; To investigate APP^NL-G-F expression exclusively in the LC, we designed a custom-build Cre-dependent AAV virus. It is a mammalian FLEX conditional gene expression AAV virus (Cre-on) with the full vector name: pAAV[FLEXon]-SYN1>LL:rev({hAPP(KM670/671NL,I716F)}/P2A/HA):rev(LL):WPRE (Vector ID: VB230525-1787fff). The virus is flagged with an HA-tag for post-hoc virus expression validation.

Chronic olfactory bulb window implantation. To study pathology dependent norepinephrine release in the olfactory bulb, 2-month-old APP^NL-G-Fmice (n=3) and C57BL/6J (n=3) control animals were fitted with cranial windows. In short, mice were anesthetized with a mixture of Medetomidin, Midazolam and Fentanyl at 0.5, 5 and 0.05 mg/kg bodyweight respectively. Dexamethason was injected i.p. at 100 mg/kg to reduce inflammatory responses and the animal got headfixed in a stereotactic frame. The skin was cut vertically to expose lambda, bregma and the olfactory bulb and give adequate adherence space for the headbar. Surface edging was performed by scoring the skull lightly with a scalpel and applying a UV light curing mildly corrosive agent (IBond Self Etch, Kulzer 66046243). After locating the rostral rhinal vein, running just posterior of the olfactory bulb, a 3mm biopsy punch was used to indicate the craniotomy location just anterior of the vein. The Neurostar surgical robot was the used to drill the marked circle until the skull disk could be removed. The dura mater was removed on the exposed part of the left olfactory bulb. The norepinephrine sensor pAAV-hSyn-GRAB_NE1m was injected into the centre of the bulb (450 nl at 45 nl/min) at a depth of 400 µm. After injection the area was cleaned and a 3mm circular cover slip fitted over the craniotomy area. The window was fixed in place with tissue adhesive glue (Surgibond tissue adhesive, Praxisdienst, 190740). The entire area with exposed skull was subsequently filled with dental cement (Gradia Direct Flo BW, Spree Dental, 2485494) and a headbar suitable for the later utilized 2P-microscope quickly placed over the window. The cement was cured with UV. After surgery the mice received 5 mg/kg Enrofloxacin as an antibiotic, 25 mg/kg Carprofen to reduce inflammation and 0.1 mg/kg Buprenorphin as an analgesic. A mixture of Atipamezol and Flumazenil (2.5 and 0.5 mg/kg) was used to antagonize the anaesthesia.

2-photon imaging. One month after surgery all mice were trained on the wheel used for awake in vivo imaging, their windows cleaned and the injection site checked for expression. A delivery method for a vanilla scent was established by combining a tube connected to a picospritzer system (PSES-02DX) with a vial containing vanilla aroma (Butter-Vanille, Dr. Oetker, 60-1-01-144800). The tube opening was placed at a fixed distance of roughly 4cm in front of the mouse and a vacuum pump placed slightly behind the head to ensure quick dispersion of the scent after an airpuff was delivered. The two photon microscope system was the Femptonix system ATLAS with a Coherent Chameleon tunable laser set at 920nm. Three locations were imaged per mouse at depths between 30 and 60 µm below the surface with an 16x objective. Over three minutes a z-stack of 120x120x30 µm with a pixel size of 0.22 µm and a z step of 1 µm was recorded at 1.13 Hz. After one minute of baseline recording, 10 seconds of a vanilla delivering airpuff were administered. After each three-minute recording 20 minutes of waiting time separated the subsequent recording and ensured the dispersion of the odour inside of the imaging setup.For an additional long term trial, one WT mouse was imaged for 18 minutes with the above mentioned settings. Here, vanilla airpuffs at 10 seconds of length were applied at 5, 10 and 15 minutes. The recordings were loaded into Fiji and each z-stack projected with a summation of all 30 slices. Afterwards the EZCalcium Motion Correction (based on NoRMCorre) (PMID: 32499682) was used to reduce motion artefacts. For each individual recording the frame brightness was normalized to the average of the baseline frames 20-67 before the vanilla airpuff and the average of the three adjusted curves calculated. The first 20 frames were removed to account for inconsistencies at the start of each recording, such as startling of the animal. For the 18 minute recording the average was taken from frames 20-300. Heatmaps were created with the Python Seaborn distribution.

Acute slice electrophysiology (perforated-patch-clamp). Acute brain slice recordings were performed as previously described^54–56. Mice were anaesthetized with isoflurane and subsequently decapitated, before the brain was rapidly removed and stored in cold (4°C) glycerol aCSF. 300 µm thick slices containing the region of the locus coeruleus and the olfactory bulb were cut in carbogenated (95% O2 and 5% CO2) glycerol aCSF (230 mM Glycerol, 2.5 mM KCl, 1.2 mM NaH2PO4, 10 mM HEPES, 21 mM NaHCO3, 5 mM glucose, 2 mM MgCl2, 2 mM CaCl2 (pH 7.2, 300-310 mOsm), using a vibration microtome (Leica VT1200S, Leica Biosystems, Wetzlar, Germany). Slices were immediately transferred into a maintenance chamber with warm (36°C) carbogenated aCSF (125 mM NaCl, 2.5 mM KCl, 1.2 mM NaH2PO4, 10 mM HEPES, 21 mM NaHCO3, 5 mM glucose, 2 mM MgCl2, 2 mM CaCl2 (pH 7.2, 300-310 mOsm)). After 50 min recovery, slices were kept at room temperature (~22°C) waiting for recordings. For electrophysiological recordings, slices were individually transferred into a recording chamber and perfused with carbogenated aCSF at a flow rate of 2.5 ml/min. The temperature was controlled with a heat controller and set to 26 °C. Perforated patch-clamp recordings were obtained from LC neurons and OB mitral cells visualized with an upright microscope, using a 60x water immersion objective. Biocytin labelling and post-hoc immunohistochemistry was used to confirm the right cell type. Patch pipettes were fabricated from borosilicate glass capillaries (outer diameter: 1.5 mm, inner diameter: 0.86 mm, length: 100 mm, Harvard Apparatus) with a vertical pipette puller (Narishige PC-10, Narishige Int. Ltd., London, UK). When filled with internal solution (tip-filled with potassium-D-gluconate intracellular pipette solution 1: 140 mM potassium-D-gluconate, 10 mM KCl, 10 mM HEPES, 0.1 mM EGTA, 2 mM MgCl2 (pH 7.2, ~290 mOsm) and back-filled with potassium-D-gluconate intracellular pipette solution 2: 140 mM potassium-D-gluconate, 10 mM KCl, 10 mM HEPES, 0.1 mM EGTA, 2 mM MgCl2, 0.02% Rhodamine Dextran, ~200 mg/ml Amphotericin B (dissolved in DMSO) and if needed 1% biocytin (pH 7.2, ~ 290 mOsm), they had a resistance of 4-5 MOhm. All experiments were performed using an EPC10 patch clamp (HEKA, Lambrecht, Germany) and controlled with the software PatchMaster (version 2.32; HEKA). The liquid junction potential (~14.6 mV) was compensated prior to seal formation and recordings were always compensated for series resistance and capacity. All executed protocols were recorded with Spike 2 (version 10a, Cambridge Electronic Design, Cambridge, UK). Data were sampled with 10 to 25 kHz and low-pass filtered with a 2 kHz Bessel filter.

Human TSPO-PET imaging acquisition and analysis. For PET imaging an established standardized protocol was used^57–59. All participants were scanned at the Department of Nuclear Medicine, LMU Munich, using a Biograph 64 PET/CT scanner (Siemens, Erlangen, Germany). Before each PET acquisition, a low-dose CT scan was performed for attenuation correction. Emission data of TSPO-PET were acquired from 60 to 80 minutesafter the injection of 187 ± 11 MBq [¹⁸F]GE-180 as an intravenous bolus, with some patients receiving dynamic PET imaging over 90 minutes. The specific activity was >1500 GBq/μmol at the end of radiosynthesis, and the injected mass was 0.13 ± 0.05 nmol. All participants provided written informed consent before the PET scans. Images were consistently reconstructed using a 3-dimensional ordered subsets expectation maximization algorithm (16 iterations, 4 subsets, 4 mm gaussian filter) with a matrix size of 336 × 336 × 109, and a voxel size of 1.018 × 1.018 × 2.027 mm. Standard corrections for attenuation, scatter, decay, and random counts were applied. The 60-80 min p.i. images of all patients and controls were analysed.

Small animal TSPO μPET. All small animal positron emission tomography (μPET) procedures followed an established standardized protocol for radiochemistry, acquisition and post-processing^60,61. In brief, [¹⁸F]GE-180 TSPO μPET with an emission window of 60-90 mins post injection was used to measure cerebral microglial activity. APP^NL-G-F and age-matched C57BL/6 mice were studied at ages between two and twelve months. The TSPO µPET signal in the cortex and the hippocampus was previously reported in other studies^62–64. All analyses were performed by PMOD (V3.5, PMOD technologies, Basel, Switzerland).Normalization of injected activity was performed by the previously validated myocardium correction method⁶⁵. TSPO μPET estimates deriving from predefined volumes of interest of the Mirrione atlas⁶⁶ were used: olfactory bulb (xx mm³) and cortical composite (xx mm³). Associations of TSPO µPET estimates with age and genotype as well as the interaction of age*genotype were tested by a linear regression model. We performed all PET data analyses using PMOD (V3.9; PMOD Technologies LLC; Zurich; Switzerland). The primary analysis used static emission recordings which were coregistered to the Montreal Neurology Institute (MNI) space using non-linear warping (16 iterations, frequency cutoff 25, transient input smoothing 8x8x8 mm³) to a tracer-specific template acquired in previous in-house studies. Intensity normalization of all PET images was performed by calculation of standardized uptake value ratios (SUVr) using the cerebellum as an established pseudo-reference tissue for TSPO-PET (9).

Human olfactory test. For detecting decreased olfactory performance due to neurodegenerative diseases, the "Sniffin' Sticks - Screening 12" test was employed. Developed in collaboration with the Working Group "Olfactology and Gustology" of the German Society for Otorhinolaryngology, Head and Neck Surgery, the test provides a preliminary diagnostic orientation and can be conveniently used in everyday settings. It classifies individuals as anosmics (no olfactory ability), hyposmics (reduced olfactory ability), or normosmics (normal olfactory ability)⁶⁷. The participants are presented with 12 familiar scents (health-safe aromas, mostly used in food as flavourings) separately, in succession. Both nostrils are assessed simultaneously. Each scent is presented with a multiple-choice format, where participants choose one of four terms that best describe the scent, even if they perceive no smell. During testing, no feedback is provided to ensure unbiased responses. Demographic details of the subjects are listed in Supplementary Table 3.

Statistics

All statistical analyses were performed in GraphPadPrism (version 10.1.1). Data are reported as mean ± s.e.m. Significance was set at P < 0.05 and expressed as *P < 0.05, **P < 0.01, ***P < 0.001 and ****P<0.0001. Statistical details of every experiment are explained in Supplementary Table 1 and 2.

Tillage, R. P. et al. Elimination of galanin synthesis in noradrenergic neurons reduces galanin in select brain areas and promotes active coping behaviors. Brain Structure and Function 225, 785–803 (2020).
Speed, H. E. et al. Autism-Associated Insertion Mutation (InsG) of Shank3 Exon 21 Causes Impaired Synaptic Transmission and Behavioral Deficits. J. Neurosci. 35, 9648–9665 (2015).
Lehrman, E. K. et al. CD47 Protects Synapses from Excess Microglia-Mediated Pruning during Development. Neuron 100, 120-134.e6 (2018).
Paeger, L. et al. Antagonistic modulation of NPY/AgRP and POMC neurons in the arcuate nucleus by noradrenalin. eLife 6, 166 (2017).
Paeger, L. et al. Energy imbalance alters Ca2+ handling and excitability of POMC neurons. eLife 6, e25641 (2017).
Jais, A. et al. PNOCARC Neurons Promote Hyperphagia and Obesity upon High-Fat-Diet Feeding. Neuron (2020) doi:10.1016/j.neuron.2020.03.022.
Xiang, X. et al. Microglial activation states drive glucose uptake and FDG-PET alterations in neurodegenerative diseases. Sci. Transl. Med. 13, eabe5640 (2021).
Rauchmann, B. et al. Microglial Activation and Connectivity in Alzheimer Disease and Aging. Ann. Neurol. 92, 768–781 (2022).
Finze, A. et al. Individual regional associations between Aβ-, tau- and neurodegeneration (ATN) with microglial activation in patients with primary and secondary tauopathies. Mol. Psychiatry 28, 4438–4450 (2023).
Brendel, M. et al. Glial Activation and Glucose Metabolism in a Transgenic Amyloid Mouse Model: A Triple-Tracer PET Study. J. Nucl. Med. 57, 954–960 (2016).
Overhoff, F. et al. Automated Spatial Brain Normalization and Hindbrain White Matter Reference Tissue Give Improved [18F]-Florbetaben PET Quantitation in Alzheimer’s Model Mice. Front. Neurosci. 10, 45 (2016).
Sacher, C. et al. Longitudinal PET Monitoring of Amyloidosis and Microglial Activation in a Second-Generation Amyloid-β Mouse Model. J. Nucl. Med. 60, 1787–1793 (2019).
Biechele, G. et al. Pre-therapeutic microglia activation and sex determine therapy effects of chronic immunomodulation. Theranostics 11, 8964–8976 (2021).
Biechele, G. et al. Glial activation is moderated by sex in response to amyloidosis but not to tau pathology in mouse models of neurodegenerative diseases. J Neuroinflamm 17, 374 (2020).
Deussing, M. et al. Coupling between physiological TSPO expression in brain and myocardium allows stabilization of late-phase cerebral [18F]GE180 PET quantification. NeuroImage 165, 83–91 (2018).
Ma, Y. et al. A three-dimensional digital atlas database of the adult C57BL/6J mouse brain by magnetic resonance microscopy. Neuroscience 135, 1203–1215 (2005).
Hummel, T., Kobal, G., Gudziol, H. & Mackay-Sim, A. Normative data for the “Sniffin’ Sticks” including tests of odor identification, odor discrimination, and olfactory thresholds: an upgrade based on a group of more than 3,000 subjects. Eur. Arch. Oto-Rhino-Laryngol. 264, 237–243 (2007).

Yes there is potential Competing Interest. Prof. Matthias Brendel (M.B.) received consulting/speaker honoraria from Life Molecular Imaging, GE healthcare, and Roche, and reader honoraria from Life Molecular Imaging. All other authors declare no competing interests.

SupplementaryTable120240816426.xlsx
Dataset 1
SupplementaryTable220240816426.xlsx
Dataset 2
SupplementryTable320240816426.xlsx
Dataset 3
ExtendedDataTable120240816426.xlsx
Extended Data Table 1
Extendeddatafigures.docx

Download PDF

Version 1

posted

You are reading this latest preprint version

Noradrenergic axon loss drives olfactory dysfunction in Alzheimer’s disease

Status:

Version 1

Abstract

Figures

Introduction

Discussion

Declarations

References

Methods

Additional Declarations

Supplementary Files

Status:

Version 1