During early development, a substantial proportion of central neurons undergoes programmed cell death. This fundamental homeostatic process controlling the ultimate number of neurons is essential for the proper structural and functional development of the brain. Neuronal activity has a major impact on neuronal death and survival rates in developing cortical networks. While blockade of neuronal activity profoundly increases cellular mortality rate, an increase in neuronal activity is generally associated with a decline in neuronal death. Although a better understanding of these activity-dependent regulatory mechanisms as well as their importance for brain development is emerging, many critical questions about this fundamental developmental process remain open. During the last two funding periods, we have shown how neuronal activity controls apoptosis induction in distinct neuronal cell types destined to die during early corticogenesis (e.g. Cajal-Retzius neurons). Moreover, we could show that electrical activity tunes the final number of surviving neurons in the developing cortex in a region-specific manner. Thus, activity homeostatically regulates the population size of developing neuronal networks in a region- and time-dependent manner. At the single-cell level, we demonstrated that not only action potential firing per se acts as a pro-survival factor in early development, but also the specific temporal discharge pattern controls neuronal survival. Synchronized spindle burst activity, which represents a characteristic feature of the perinatal cerebral cortex in all mammalian species studied so far (from mouse to humans), fullfil a physiologically relevant role in the control of cell survival vs. cell death. In summary, our previous studies showed that activity-dependent modulation of neuronal survival occurs (i) at the level of functional regions in the developing cortex, (ii) at the level of single neurons in cortical cultures and even (iii), as indicated by on-going experiments, at the subcellular level of the neuronal nucleus. In the third funding period, we propose to study (1) the molecular mechanisms that translate patterned electrical activity into cell death versus survival decisions and how they are reflected in nano-structural changes in the cell nucleus. At the network level, physiological versus non-physiological as well as local versus global changes in cortical activity determine the spatio-temporal extent of apoptosis. This implies that activity during early development directly affects ultimate population sizes in the mature cortex. Therefore, the second major aim of this proposal is to address (2) the impact of altered cell death on neuronal network activity in the developing somatosensory cortex and on the function of the adult cortex. In summary, the results of these planned experiments will provide a better understanding of activity-dependent apoptosis as an important homeostatic element at the network, cellular and subcellular level during early development. Moreover, these data will expand our knowledge on how this developmental process shapes both the structure and the function of the mature nervous system.

Adult neurogenesis contributes to the remodelling of the hippocampal circuitry throughout life. A key regulatory step in this process is the transition of quiescent neural stem cells (NSCs) to an active/proliferative state. Recent work in Drosophila has uncovered a critical function of the Hippo-Yap signalling pathway in regulating NSC activation, which motivated us to explore whether its role is evolutionary conserved in NSC biology across species. In the past funding period, we discovered that Yap1 is sufficient to drive quiescent mouse hippocampal NSCs into cell cycle. As shown by single cell RNAsequencing, Yap1 gain-of-function induces an activated NSC transcriptome signature. Conversely, deletion of Yap1 in the NSC compartment leads to a progressive reduction in active NSCs, demonstrating that Yap1 plays a physiological role in the control of NSC transition from quiescence to activation. In the new funding period, we aim to deepen our understanding of the physiological role of Yap1 in regulating the homeostasis between NSC quiescence and activation. In Aim 1, we will use clonal analyses in vivo to reveal the effect of Yap1 loss-of-function on NSC lineage progression, including self-renewal capacity. Furthermore, we will scrutinize the effect of deleting Yap1 on the frequency of active NSCs returning to quiescence or reentering cell-cycle. In Aim 2, we will employ single cell RNA sequencing to unveil the molecular underpinnings of Yap1 loss-of-function in promoting quiescence at the expense of activation. Conversely, in Aim 3 we will employ Yap1 gain-of-function to identify direct gene targets of YAP1 in adult hippocampal NSCs, and uncover how transcriptome changes are brought about by remodelling of the chromatin landscape. Finally, we will examine the hypothesis that Yap1 activity declines with age, potentially contributing to the decreased proportion of active NSCs during aging. Overall, this study will shed light on molecular and cellular mechanisms contributing to the tightly regulated balance between NSC quiescence and activation throughout lifetime.

Homeostatic processes require the synthesis of new transcripts and proteins. Among these processes, neural activity-dependent post-transcriptional regulation at the mRNA level has profound consequences on RNA structure, subcellular localization, and translation. 3’ untranslated regions (3’UTRs) are considered as central hubs for posttranscriptional modifications, including RNA stabilization and/or translational facilitation or inhibition. In the previous funding period, we unraveled that hippocampal memory and long-term potentiation are impaired in mice deficient for the RNA-binding protein Gadd45α (Growth arrest and DNA damage-inducible protein 45α). This phenotype is accompanied by reduced levels of memory-related mRNAs (e.g., Grin2a, Grin2b, Kcnq3, Grm5). The majority of the Gadd45α-regulated transcripts shows unusually long 3’UTRs that are destabilized in Gadd45a-deficient mice via a post-transcriptional mechanism, finally leading to reduced levels of the corresponding proteins in synaptosomes. Moreover, we characterized Gadd45α as an RNA-binding protein (RBP), specifically interacting with the memory-related mRNAs containing long 3’UTRs. Recently, we have also revealed that Gadd45α interacts with components of the Ccr4-Not complex, a machinery involved in RNA decay, a post-transcriptional process regulating mRNA stability, and thereby regulating RNA homeostasis in the cell. Our follow-up study will serve as entry point, and Gadd45a as a springboard for unraveling how extended 3’UTRs can function as hubs for post-transcriptional regulation of neural RNA homeostasis, via interaction with the Ccr4-Not complex, mediated by the RNA decay machinery. Specifically, we will characterize in detail the Gadd45α-RNA interaction, focusing on those mRNA species with long 3’UTRs and destabilized by loss of Gadd45a. We will further investigate the function of Ccr4-Not components in the regulation of above mRNA species with long 3’UTRs. Our study will provide novel insights into the unexpected role of Gadd45α as an RBP interacting with the RNA decay machinery and hence in RNA homeostasis.

Dopamine (DA) neurons in the ventro-lateral substantia nigra (l-SN) are the clinically most relevant neuronal population that selectively degenerates in Parkinson's disease (PD) leading to substantial motor impairment. The mechanisms of this remarkable differential vulnerability, even within the dopamine midbrain system, are still not understood and there is currently no treatment that halts or at least slows down the progressive DA l-SN cell loss. More specifically in the context of this CRC, we do not yet know when homeostatic buffering of DA electrical function breaks down, so that a novel pathophysiological state is established, and moreover how that pathophysiology contributes to neurodegeneration. Recent pathological studies indicated that the time window between clinical diagnosis of PD and massive loss of DA cells is small – and almost complete within just 4 years (Kordower et al., 2013). This implies that loss of electrophysiological homeostasis of DA neurons is likely to take place in the long prodromal phase of PD. This assumption is in line with our own studies where altered electrical phenotypes were detected long before the onset of neurodegeneration (Chiu et al., 2020; Lasser-Katz et al., 2017; Subramaniam et al., 2014a). What exactly captures the homeostatic state of an electrically active DA midbrain neuron? We argue that rather than a single setpoint (e.g. mean firing rate) it is better described as an operational space that contains a set of different but each well-tuned neural activity patterns. As shown in Figure 1A, similar concept of firing and pattern rate homeostasis as the centerpiece of interconnected modes of neuronal homeostasis has recently been proposed (Frere and Slutsky, 2018). DA neurons are autonomous pacemaker cells that show continuous in vivo and in vitro baseline discharge in a narrow bandwidth between 1-8 Hz. In addition, they generate synaptically driven fast bursts of action potentials at higher frequencies (20-80 Hz) band in response to e.g. salient stimuli or rewards. This implies that DA neurons have at least two functional set-points; one for baseline pacemaker rate and another for fast frequency burst responses. In the last funding periods, we have shown that these two setpoints can be dissociated by molecular interventions (Schiemann et al., 2012) or in response to chronic alpha-synuclein overexpression (Subramaniam et al., 2014a), arguing for complementary homeostatic control mechanisms. Moreover, we demonstrated that these most vulnerable DA neurons in the lateral substantia nigra possess a significantly larger burst rate compared to neighboring more resistant DA neurons in the medial SN (Farassat et al., 2019). We also identified the underlying molecular biophysics that render l-SN DA neurons uniquely excitable in comparison to other DA subpopulations in the midbrain. We found that both in vitro autonomous pacemaker discharge, in vitro burst responses evoked in dynamic clamp as well as spontaneously occurring in vivo burst rates were driven by the voltage-gated calcium channel Cav1.3 in l-SN DA neurons. These results, together with an extended repertoire of methods including in vivo patch-clamp recordings of DA neurons developed in our lab (Otomo et al., 2020) and long-term in vivo monitoring of DA function using the genetically encoded DA sensor dLight established by others (Patriarchi et al., 2018), enable us to refine the central research question for the third funding period: how is firing and burst rate homeostasis selectively maintained in l-SN DA neuron population when challenged by toxic alpha-synuclein aggregates, cell loss and aging? This question will be addressed in the light of recent advances in the field of mechanistic understanding of homeostatic setpoint control in central neurons (Styr et al., 2019) that highlighted the role of mitochondrial metabolism and calcium handing. Organized across four main aims, we wish to define how Cav1.3 channel function contributes to electrical homeostasis of l-SN DA neurons (AIM 1)? How acute and chronic challenge of l-SN DA neurons with toxic alpha-synuclein aggregates affects their function (AIM 2)? How cell-loss via partial lesion alters homeostatic setpoints of l-SN DA neurons (AIM 3)? Finally, how are the particular functions of l-SN DA neurons affected in a genetic mouse model of early onset PD (22q11; AIM 4)? In combination, this research project might lead to a better understanding of the relevant pathomechanisms of selective vulnerability in Parkinson's disease and pave the way for novel therapeutic options.

It is clear that for all cells, including neurons, the proteome is dynamically regulated to maintain cellular function and viability. In neurons, proteome regulation occurs in the cell body as well as the dendrites and axons, allowing the initiation of proteome remodeling at synapses, far from the cell body. While it is clear that changes in synaptic transmission involve extensive regulation of the synaptic proteome via the regulated synthesis and degradation of proteins, it is not well understood how these two processes are coordinately regulated to achieve the desired level of individual proteins at synapses. We plan to examine how protein degradation mediated by the ubiquitin proteasome system (UPS) is affected if protein synthesis is blocked. We will characterize the feedforward mechanisms that regulate UPS activity in response to protein synthesis inhibition. Our objective is to elucidate feedback mechanisms between the translation machinery and the UPS that ensure that the levels of individual synaptic proteins are maintained within a certain concentration range. The first experiments will examine whether there is regulation of the UPS upon inhibition of protein synthesis. Hippocampal cultured neurons or hippocampal slices will be treated with or without protein synthesis inhibitors (PSIs) and after different incubation times polyubiquitinated proteins will be immunoprecipitated with anti-poly-Ub antibodies (e.g. K48 ubiquitin-linked antibodies) and analyzed by mass spectrometry. Subsequent experiments will examine how protein synthesis inhibition brings about changes in UPS activity. In order to specifically evaluate changes in global proteasome activity following inhibition of translation we will use fluorogenic substrates (NSuc-LLVY, Z-LLE, and Boc-LRR fused to MCA or Aminoluciferin) on hippocampal cultured neurons. The proteasome comprises three catalytic subunits with different catalytic activities: β1 ("post-glutamyl peptide hydrolase" site), β2 ("trypsin-like" site) and β5 ("chymotrypsin-like" site). By using a combination of substrates with different specificities for these sites, we will be able to capture potential changes in the relative catalytic activity. Do changes in protein synthesis affect the activity of the proteasome locally? It has been previously shown that, in neurons, proteostatic regulation can occur via redistribution of the proteasome between cellular compartments and changes in local activity. In order to examine whether a similar process of proteasomal redistribution between neuronal compartments is initiated upon treatment of hippocampal cultures with PSIs, after blockade of translation we will use a reporter to visualize in situ proteasomal degradation over time in different neuronal compartments (soma, dendrites, axons and synapses).

Chemical synaptic transmission is a highly regulated process that is also plastic. It can be acutely modulated to adapt to different activity regimes, and it can be enhanced or reduced in the long-term, to change the output of a circuit and to potentiate connections during Hebbian learning. Last, it can homeostatically change to ensure steady circuit output, despite alteration of the functional properties of the neurons constituting it. Among mechanisms affecting presynaptic homeostatic plasticity (PHP) are differential responses of voltage-gated Ca2+ channels (VGCCs) to altered electrical input, as well as changing the rates by which synaptic vesicles (SVs) are primed for fusion, or recycled subsequently. We investigate the transmitter release machinery at the cholinergic neuromuscular junction (NMJ) of the nematode Caenorhabditis elegans, by electron microscopy (EM), electrophysiology, voltage imaging, functional assays, and novel optogenetic tools we developed. We found that regulation of the localization of VGCCs through RIM binding protein affects the rate of SV recycling. Classical PHP is not well-characterized in C. elegans, however, we found that its NMJs integrate systemic neuromodulatory signals to alter their output, in part by altering SV fusion rates, but also by loading SVs with different amount of transmitter: Synaptic output (quantal size and content) can be modulated by neuropeptide signalling in an autocrine fashion, involving regulation of dense core vesicle (DCV) fusion via synapsin. In the next funding period, we will develop and use optogenetic tools to understand how VGCC localization participates in the fine-control of SV fusion and recycling, and characterize novel proteins as missing links in PHP, connecting SV fusion and recycling mechanisms. We want to investigate the mechanisms leading to ‘neuromodulatory PHP’ in detail, by addressing which neuropeptides and which upstream signals contribute to this regulation, and how.

Firing rate homeostasis after entorhinal denervation of dentate granule cells is a prototypic form of homeostasis: following denervation the firing rate of granule cells temporarily decreases. However, within a week the denervated granule cells recover and return their firing rate to the pre-denervation baseline. During the first two funding periods, we identified three homeostatic mechanisms contributing to this phenomenon: synaptic strengthening, dendritic reorganization, and collateral sprouting. In the next funding period, we will (i) target previously identified homeostatic mechanisms to enhance regeneration using pharmacology and optogenetics and (ii) focus on the site of action potential generation itself, i.e. the axon initial segment (AIS), which has recently emerged as a major site of homeostatic plasticity fine tuning neuronal excitability. In preliminary work leading to this application, we could show that entorhinal denervation changes AIS morphology and expression levels of AIS-related molecules in denervated granule cells. These alterations mirror homeostatic changes in firing rate. In addition, our preliminary data indicate that not only the length of the AIS is homeostatically regulated but also the cisternal organelle (CO), a compartment-specific endoplasmic reticulum (ER) structure found within the AIS. The CO is hypothesized to be a local Ca2+ store and its formation critically depends on the plasticity-related protein Synaptopodin (SP). Based on these initial observations, which support a role of the AIS in firing rate homeostasis after denervation, we will use organotypic slice culture models of homeostasis (entorhinal denervation, tetrodotoxin (TTX)) and optogenetic stimulation of granule cells to monitor structural and molecular changes of the AIS and the CO. Reporter mice for the AIS-specific protein ankyrin G and a panel of SP-mutants (SP-KO, SPtg) will be employed to study the CO within the AIS using 2-photon time-lapse imaging. Patch-clamp (with project A11 Roeper) and multichannel recordings (with project A1 Luhmann/Sinning) will be performed to understand homeostasis on the single cell and the hippocampal network levels. Ca2+ imaging will be employed to understand the role of the CO as a local Ca2+ store under physiological and homeostatic plasticity conditions. (iii) In silico analysis: Data on denervation-induced homeostasis obtained in this project throughout the three funding periods, in particular the four homeostatic mechanisms, i.e. synaptic strengthening, sprouting, dendritic remodelling, and AIS remodelling, will be integrated into a recently published realistic granule cell model. This will help us to understand how granule cells homeostatically adapt to denervation and will provide insight into the non-linear interactions of the different homeostatic mechanisms. This approach will also help to identify the most promising targets for enhancing functional regeneration.

Structural plasticity is a fundamental process accompanying both Hebbian and homeostatic synaptic plasticity and it is necessary for proper circuit wiring and maintenance. Hebbian and homeostatic forms of plasticity have been traditionally seen as independent cellular events although recent evidence supports an interplay of these two types of plasticity to stabilize neuronal networks. Homeostatic synaptic plasticity is considered now a metaplasticity mechanism to integrate changes in neuronal activity and support optimal Hebbian learning. Therefore, identifying common molecular players in these forms of plasticity will help us to understand the processes that underlie their interplay to sufficiently allow flexibility but also maintain stable activity. In the first two funding periods of the CRC1080, we have characterized signaling complexes at synaptic and perisynaptic neuronal membranes involved in Hebbian and homeostatic structural plasticity with a focus on AMPA receptor (AMPAR) dynamics and the remodelling of dendritic trees and dendritic spines. Thus, we identified the interaction of the glutamate receptor interacting protein (GRIP)-1 with ephrinB2 and the apolipoprotein E receptor 2 (ApoER2) as well as the complex formed by ephrinB2 and Vascular endothelial growth factor receptor (VEGFR)-2 to be essential for processes such as dendritic arborization and AMPAR membrane insertion during synaptic and structural plasticity. Our studies using expansion microscopy also allowed for the first time the quantitative assessment of dynamic changes of AMPAR distribution during hippocampal homeostatic plasticity. We showed that ephrinB2 signaling was also able to control the homeostatic relocation of AMPARs to the surface of spines to induce mushroom spine recovery after hippocampal denervation. Recent research has made increasingly clear the crucial role of the vasculature in brain formation and maintenance, not only by providing the necessary oxygen and nutrients matching the continuously high demand of the brain, but also by expressing signaling cues essential for the proper spatio-temporal development of the neuronal and glial network, thus regulating processes such as differentiation, migration or the establishment of connectivity. However, the molecular players mediating the intricate crosstalk between neuronal networks and the vasculature have been hitherto largely unexplored. We will focus on two signaling pathways we have recently uncovered to be at the interface between the vascular and nervous systems: VEGF/VEGFR2 and Reelin/ApoER2/Dab1 signaling pathways. We will now investigate how these pathways regulate the communication between vessels, glia and neurons to achieve circuitry development as well as structural and functional homeostatic plasticity.

Mutations in single genes like TSC1, TSC2 or SHANK3 lead to syndromic forms of autism spectrum disorder (ASD) and intellectual disability (ID). Patients mostly develop symptoms along a time scale that depends on the mutation. However, clinical variability is high and not all patients show the full spectrum of symptoms. This suggests that at some point – depending on the affected gene and individual ability to compensate – effective homeostatic regulation in the brains of individuals is overstrained, leading to the development of disease symptoms. In the past funding period, we have longitudinally analyzed a mouse model for tuberous sclerosis carrying a heterozygous mutation in Tsc2. In behavioral batteries carried out at several timepoints we found that, while changes in grooming behavior developed already in 2-month old mice, deficiencies in social behavior were only seen at a later timepoint of 3-4 months of age. Careful cognitive testing revealed subtle cognitive aberrations in Tsc2+/- mice in aging animals of 8-10 months. By extended protein expression analysis of cortical synaptosome fractions at different timepoints from postnatal day (P) 10 to 8-10 months, we interestingly found that a reduction in Tsc2 protein expression was seen in the Tsc2+/- animals only until P17, suggesting full compensation of the primary defect by the intact Tsc2 allele at later stages and the development of the behavioral phenotype only later on, independently of the primary defect. Furthermore, after instable expression of several excitatory and inhibitory receptor units in the synaptic compartments at early stages, stable upregulation of Neuroligin/Neurexin molecules from P28 onwards was detected. We hypothesize that during early stages of postnatal brain development, when Tsc2 expression is reduced, there is a first window of vulnerability. In this context, in the next funding period we will re-balance the system and lessen the need for homeostatic regulation by stimulating gene expression from the intact Tsc2 allele in these early time periods through epigenetic modulators. We aim to reduce or – in the ideal case – rescue the phenotype starting from 2-4 months of age. Furthermore, we hypothesize that, while active homeostatic regulation tries to balance the system, overstraining at the time point of phenotype development overwhelms the system, which, as our data suggest, results in overexpression of Neurexin/Neuroligins, molecules that are closely linked to autism spectrum disorder (ASD). We will thus interfere with Neuroligin/Neurexin overexpression using AAV-mediated shRNA expression and/or CRISPR/Cas9-based genome editing and subsequently analyse effects on the behavioral phenotype. Finally, we will study a second well-established mouse model of syndromic ASD and ID – Shank3+/- mutants – to see if strict homeostatic regulation and spillover of the system after overstraining leading to the development of ASD-like behavior, is a phenomenon more general to monogenic forms of ASD.

Neuronal communication is modulated by adaptive responses to changes in the environment. The visual system has been a model for the investigation of homeostatic regulation of neuron and network homeostasis. Thus, changes in visual experience such as visual deprivation alter synaptic function in the visual cortex, in particular during a critical developmental time window. A key mechanism underlying such modulation is up- or downscaling of synaptic strength. AMPA receptors (AMPARs) mediate most of the fast excitatory transmission in the central nervous system (CNS). AMPARs comprise four core subunits and several auxiliary subunits, many of which exert a strong influence on receptor function. We and others have observed that expression of auxiliary subunits is regulated in an activity-dependent manner in different brain areas. For example, visual deprivation (keeping mice in a dark box for 7 days) increases the expression of the auxiliary subunit CKAMP44 in the cortex and lateral geniculate nucleus (LGN) of adult mice (i.e. after the critical period of visual system development). We tested whether CKAMP44 upregulation is required for homeostatic plasticity of synapse function in the LGN using CKAMP44 knockout mice and found that visual deprivation alters the function of retinogeniculate synapses by rendering short-term depression stronger. This effect is mediated by postsynaptic (increased number of synaptic CKAMP44-bound AMPARs) and presynaptic (increased release probability) mechanisms. In the next funding period, we will investigate whether and how neuronal communication changes in the visual system due to the observed changes in synapse function. It has been shown that the mean firing rate of LGN neurons does not change during visual deprivation despite strongly altered input (i.e. decorrelated firing at lower rates of retinal ganglion cells) (Linden et al., 2009). We hypothesize that increased release probability and CKAMP44-bound AMPAR number in retinogeniculate synapses are homeostatic plasticity mechanisms that strengthen fidelity of information transfer from retina to LGN. These changes in synapse function would allow LGN relay neurons to sustain activity when retinal ganglion cells fire desynchronized and at low frequency. In addition, based on preliminary in vivo recording data, we hypothesize that the stronger short-term depression in retinogeniculate synapses of visually deprived mice affects processing of visual information by increasing low-pass filter properties relay neurons. We will test these hypotheses by recording responses of LGN and visual cortex neurons to visual stimuli in head-fixed non-anesthetized mice using Neuropixels probes. To investigate whether information processing is affected by visual deprivation, we will compare response amplitudes, transfer rates, and tuning curves in visually deprived mice with that in control mice. To test for the relevance of CKAMP44 upregulation, we will also perform in vivo recording experiments with CKAMP44-/- mice. By correlating data from in vitro and in vivo electrophysiology with computational modeling, we will reveal how homeostatic changes in synapse function of LGN neurons influences computation of neurons in the visual system.

Tight spatial and temporal control of activity-dependent Ca2+ signals in distinct subcellular domains is essential for synapse, and thus, brain function. At presynaptic terminals Ca2+ triggers synaptic vesicle (SV) release, affects SV recycling, and regulates homeostatic as well as experience-dependent plasticity, all of which are required for normal neural circuit function and operate in different but partially overlapping time and space domains. Although considerable knowledge exists with regard to each of these Ca2+ dependent process at the presynapse, mechanisms how the different Ca2+ signals are separated in time and space are less clear. This project addresses cellular and molecular principles that allow for parallel but separate intracellular computation of Ca2+ mediated processes within single presynaptic terminals, with strict focus on mechanisms underlying presynaptic membrane homeostasis through SV endocytosis and presynaptic excitability homeostasis. By combining Drosophila genetics with electrophysiology and imaging we found spatial and functional separation of two voltage gated calcium channels (Cav1 and Cav2) at the neuromuscular synapse. Both contribute equally to presynaptic Ca2+ signal amplitude. Cav2 localized to active zones is required for neurotransmitter release, whereas Cav1 localized around active zones fine tunes SV release and augments vesicle recycling. Knockdown of the membrane bound calcium ATPase, PMCA, reveals that it protects Cav2 triggered release against coincident Cav1 mediated Ca2+ influx. PMCA has high buffering capacity, it can significantly speed Ca2+ clearance on millisecond timescales, and PMCA function can be dynamically regulated. Its strategic localization in between active zones protect Ca2+ nanodomains from spill-over of Ca2+ influx through Cav1 channels, while leaving the spatiotemporal properties of Ca2+ nanodomains formed by Cav2 unaffected. Instead, PMCA ensures stable release probability in the face of presynaptic Ca2+ signals that augment SV recycling. PMCA activity is adjusted by numerous factors, including activity-dependent changes in Ca2+ concentration, pH, and ER Ca2+ sensors. This provides effective means to dynamically regulate homeostatic plasticity and SV recycling rates. We will now integrate our analysis of the functional interplay of different voltage gated calcium channels and PMCA with additional key players in presynaptic Ca2+ signalling to address their roles in tuning SV endocytosis to different activity demands for maintaining presynaptic membrane homeostasis (aim 1), in presynaptic homeostatic plasticity at the Drosophila NMJ (aim 2), and in activity dependent mid- and long term presynaptic homeostatic plasticity at central synapses (aim 3).

Evidence is emerging that immune responses play a vital role in central nervous system (CNS) homeostasis and brain reserve, and that our cognitive abilities require a fine-tuned equilibrium of immune responses (Pape et al., 2019; Larochelle et al., 2018). In the previous funding period, we have shown that the immune cytokine interleukin-4 (IL-4) is able to improve remaining disability after an inflammatory attack to the CNS. In fact, IL-4 acts directly on neurons inducing outgrowth in vivo without affecting the immune cell infiltration during the chronic disease phase. Mechanistically, we identified a fast direct IL-4 receptor (IL-4R) signaling pathway in neurons, which leads to modulation of the axonal cytoskeleton (Vogelaar et al., 2018). In preliminary experiments on cortex and hippocampus, we found that the IL-4R colocalizes with markers for glutamatergic and GABAergic synapses. We were able to modify synaptic transmission by interfering with the IL-4/IL-4R pathway. In line with this, preliminary data indicate cognitive consequences upon IL-4R deficiency. Interestingly, we also found a link between another immune cytokine, tumor necrosis factor alpha (TNFα), and neuronal hyperexcitability resulting in an anxiety phenotype (Ellwardt et al., 2018), indicating that cytokine balance is relevant for neuro-immune interactions and CNS homeostasis. Following these observations, we now hypothesize that ILC01 4R signaling plays a role in synaptic plasticity and could possibly regulate neuronal excitability. In this project, we aim to further unravel the complex neuro-immune crosstalk in keeping the balance of synaptic processes, focusing on IL-4 as a potential beneficial cytokine. We want investigate the effects of IL-4 treatment and IL-4R knock out (KO) on vesicle release, synaptic signaling proteins and neuronal networks, also taking into account transcriptional effects and the role of non-neuronal cells (Wasser et al., 2020). In order to explore whether the regulatory cytokine IL-4 might be able to counteract neuronal hyperexcitability, in which the pro-inflammatory cytokine TNFα plays a role, we plan to utilize experimental neuroinflammation and traumatic brain injury models, which both display changes in neuronal activity. These studies contribute to the overall aim to establish the IL-4/IL-4R pathway as a major player in homeostatic CNS processes during health and disease.

Traumatic brain injury (TBI) is a major cause of death and disability and survivors often suffer from long-term impairment of motor functions, cognition and psychic health. TBI leads to damage of axons and may increase the risk of neurodegenerative diseases. Using the controlled-cortical impact (CCI) model of the motor/somatosensory cortex in mice we disclosed neuronal hyperactivity in acute brain slices exclusively in the contralateral cortical hemisphere indicating the presence of a trans-hemispheric diaschisis. Contalateral hyperactivity had manifested already at 24h after TBI and was associated with an imbalance of excitatory versus inhibitory synaptic strength but recovered around day 3 after TBI. This recovery coincided at the molecular proteomic level with a switch in the expression of two subunits of L-type voltage gated calcium channels (VGCC, Cav1.3 occurred, Cav1.2 disappeared) in contralateral GABAergic interneurons. We therefore hypothesize that this switch in VGCCs orchestrates the strength of GABAergic inhibition, thereby stabilizing the neuronal network excitability in the contralateral hemisphere. In addition, electrophysiology studies identified alterations of intrinsic membrane properties specifically in ipsilateral parvalbumin positive (PV) interneurons owing to a TBI-induced emergence of hyperpolarization-activated cyclic nucleotide–gated (HCN) currents. Although the homeostatic electrical balance recovered within three days, behaving mice presented with sustained non-goal directed nighttime hyperactivity up to six months after TBI suggesting a long-lasting overactivity and irritability. The aim for the next period is to unravel allostatic and re-balancing mechanisms, which hinder or promote a stabilization of the dysfunctional cortical networks after TBI and assess putative therapeutic implications and functions of subpopulations of GABAergic interneurons, which include parvalbumin (PV, ̴40%), somatostatin (SST, ̴30%) and vasointestinal peptide (VIP, ̴15%) positive interneurons. (I) We plan to perform ex vivo/in vitro electrophysiological studies in transgenic GABAergic interneuron specific tdTomato reporter mice driven by PV-Cre, SST-Cre and VIP-Cre to further characterize the observed contralateral adaptive switch of VGCCs and assess putative therapeutic effects of the VGCC blocker isradipine (coop. with A11). Using a similar approach, we plan to address the functional relevance of the observed HCN currents in ipsilateral PV interneurons post TBI. (II) We then advance to in vivo recordings with silicon probes in the cortex to observe time-dependent alterations in multi-unit activity and oscillatory changes. Miniaturized in vivo endoscopy shall be used to monitor neuronal activity over weeks by chronic calcium imaging ("Miniscope"). (III) We plan to study post-TBI behavior with IntelliCages to observe long-term alterations of cognitive, social and circadian behavior in social groups of mice under home-cage conditions in enriched environments. We further plan to study effects of medication with isradipine (to block VGCC Cav1.3) or beta-blocker (to prevent 􀀀-adrenoceptor mediated enhancement of Cav1.2) or with optogenetic stimulation of cortical interneurons to revert the TBI-induced effects on electrophysiology and behavior and test their therapeutic potential. The behavioural studies shall use the reporter mice as above to observe interneuronal pathology and gene expression patterns in final tissue. Within the CRC1080, we collaborate with B03 to study spine morphology and remodelling of inhibitory interneurons, with A12N to assess dendritic architecture and amyloid pathology after TBI and with C01 to assess regulation of neuronal activity via IL4R. Our studies shall lead to a deeper understanding of TBI evoked allostasis of cortical neuronal networks and explore preventive and therapeutic startegies to modulate neuronal excitability and TBI-evoked lasting behavioral hyperactivity.

To perform computations, a neuron must keep its synapses functional and supply them continually with new proteins. The homeostatic forces responsible for this task balance protein production, decay and protein transport such that the synaptic protein numbers are kept within a specific range that is necessary for synaptic function. For example, each excitatory synapse contains on average ~60 AMPA receptors (Nair et al., 2013), and this number can be rapidly upregulated or downregulated during long-term plasticity to serve increased demand (Ehrlich & Malinow, 2004). Experiments have shown that a breakdown of homeostatic protein regulation is associated with cognitive or neurological dysfunctions, e.g. Huntington’s disease (Krauss et al., 2013). However, we currently do not fully understand how the homeostatic set-point that regulates synaptic protein numbers is established, how it is dynamically maintained, and what role it plays for synaptic plasticity and network function. The goal of our project is to build a new model framework describing the stability of the homeostatic set-point of protein copy numbers and mathematically describe its dynamics during plasticity. Specifically, we will study two processes involved in the regulation of copy numbers. First, we will consider the homeostatic balance between degradation and synthesis observed in B01. Secondly, we will consider the homeostatic balance between exocytosis and endocytosis observed in B04. We will address the ability of both processes to respond to transient protein demand changes that are an integral part of synaptic plasticity. We will perform spiking network simulations to understand how the homeostatic set-point and its properties constrain synaptic plasticity rules and shape network activity.

We are interested in the mechanisms underlying sleep in vertebrates and their evolution. To this end, we study brain activity during rapid eye movement (REM) and nonREM sleep in a lizard, the Australian dragon Pogona vitticeps, with a special interest in the mechanisms underlying the alternation between REM and non-REM, and their regulation. Our work started with the discovery of REM sleep in this lizard (Shein-Idelson et al., 2016), a first demonstration of REM sleep in non-avian reptiles, suggesting that the apparition of REM sleep predates the divergence of amniotes 320M years ago. Having identified the electrophysiological markers of REM and non-REM, we searched for their site of origin in the forebrain. A combined electrophysiological, viral tract-tracing and single-cell transcriptomic approach led us to identify the source of sharpwave ripples (SWR) produced by the forebrain during non-REM sleep as a reptilian claustrum (Norimoto et al., 2020; Tosches et al., 2018). This finding was doubly interesting. First, it established the existence of a claustrum in reptiles, indicating again that the claustrum is an ancient brain structure at least as old as the amniotes. Second, it implicated the claustrum not as required for the generation of non-REM sleep, but as a relay between the brainstem and the forebrain, necessary for the electrophysiological phenotype of non-REM sleep activity, sharp wave ripples, and for their distribution throughout the forebrain, consequence of the extensive connectivity of the claustrum.

Our approach builds up on a combined molecular, transcriptomic and functional approach initiated to understand function from an evolutionary perspective. The logic of this approach will be developed in the summary report below. In this next funding period, we will aim to characterize and understand some aspects of sleep homeostasis. To this end, we need to understand better the mechanisms of sharp-wave ripple production, regulation and function; we will continue to exploit the lizard sleep model system, advantageous for its reliability, short sleep cycle (2-3mins) and approx. 50% duty cycle. Based on our past experience and on the fact that we work with a system that has not been explored extensively in the past (by us or by others), unexpected findings have tended to preempt precise long-term planning. For this reason, we describe only the main lines of our future plans, anticipating that they will probably require correction as we proceed.

The human brain is composed of approximately 1011 neurons, connected by approximately 1014 synapses. The behavior that it generates is the outcome of the collective activity of a very large number of neurons. Therefore, it is tempting to think that random heterogeneities in the connectivity are averaged out and do not substantially affect behavior. Indeed, this is the case in many theoretical models of brain function. However, in the previous CRC funding period we discovered that surprisingly, this intuition is wrong and that “idiosyncratic choice bias”, a common idiosyncratic tendency to choose one alternative over others in the absence of an identified reason, naturally emerges from these random fluctuations in connectivity and activity. Our aim in the next funding period is to study the role of neural small-scale heterogeneities in cortical representation and behavior. To that goal, (1) we will study the long-term dynamics of idiosyncratic choice bias in humans and probe its homeostatic stability using feedback; (2) we will study how the targeted ablation of several to dozens of neurons in the auditory cortex affect auditory representation and how homeostatic mechanisms compensate for the loss of these neurons; (3) we will develop an experimental paradigm that will allow us to mechanistically probe idiosyncratic choice bias in mice and compare it to idiosyncratic choice bias in humans. Together, our results will lay the foundation for mechanistic studies that link the heterogeneity of individual neurons and the dynamics of local networks to behavior.

Animals use vision to interact with their environment. Interestingly, visual systems handle a range of luminances that far exceed the dynamic range of individual neurons. This is generally thought to be achieved through adaptation, which allows visual perception to scale with changes in luminance, or contrast, irrespective of background illumination. Visual systems thus function at dusk, dawn, or in daylight. However, visual perception is challenged when adaptation is not fast enough to deal with sudden changes in illumination, e.g., when gaze follows a moving object from sunlight into a shaded area. Under such conditions, contrast-sensitive neurons underestimate their input. We have recently identified a strategy that allows the visual system of Drosophila to achieve stable behavioral responses in dynamically changing environments, defined as sensory homeostasis. Previous work has shown that two first-order interneurons of the OFF pathway, the lamina neurons L2 and L3, are sensitive to contrast and luminance, respectively. Contrast-sensitive neuronal responses alone cannot drive contrast-constant behavioral responses in sudden dim light, which is aided by the luminance-sensitive pathway via L3. Thus, we identified a homeostatic mechanism acting through a luminance-sensitive neuron that ensures a robust behavioral response to contrast in changing visual environments. This achieves a set point of ‘contrast-constant’ responses in visual processing, meaning that flies show the same behavioral response to contrast irrespective of fast changes in illumination. Adaptation, another well-characterized mechanism to achieve stable responses to contrast in different light regimes, is alone insufficient for this robustness at the fast timescales of luminance change used in our paradigms. We now propose to study if luminance-based scaling of contrast as a sensory homeostasis mechanism is also present in the ON pathway, and thus represents a general strategy. We further want to understand how information from luminance- and contrast-sensitive outputs are combined in both ON and OFF pathways to define a contrast-coding set point. This will also identify the homeostatic circuit elements in this process. We hypothesize direction-selective T4 and T5 cells downstream in the visual processing pathway to constitute the cellular substrate to ensure fast network homeostasis with respect to contrast sensitivity. Finally, a precise balance between adaptation and other homeostatic mechanisms must exist. It is especially puzzling that the luminance-sensitive L3 neurons do not appear to adapt. Work in other sensory systems has suggested homeostatic mechanisms that feed the sensory output back into the periphery. We will further explore adaptation in L3, its downstream neurons, and explore potential feedback mechanisms.

Neural circuits in sensory cortices are highly plastic during development and young adulthood, especially during so-called critical periods. This plasticity occurs after sensory organs mature and is strongly driven by the sensory experience of the animal in its surrounding world. We still do not understand how the manifold of mechanisms – at the single neuron and network level – involved in this plasticity interact with each other and with the experience of the animal, serving it to adapt to its environment. An important role of these mechanisms is to maintain stable function, but at the same time allow for flexibility so that new memories can be encoded and new stimuli can be learned. Normal experience during the critical period in the visual system is characterized by the maturation of various functional circuit properties. This maturation is commonly believed to be driven by different forms of Hebbian plasticity, which alter synaptic connectivity in a synapse-specific manner based on highly coordinated patterns of pre- and postsynaptic activity, and operate at both excitatory and inhibitory synapses. While under normal conditions the different plasticity mechanisms work in a tightly integrated manner, perturbing sensory experience can have devastating effects on circuit function. Therefore, homeostatic mechanisms are necessary for maintaining stable function by adjusting overall synaptic strengths, neuronal excitability and the ratio between excitation (E) and inhibition (I), referred to as excitatory-inhibitory (E-I) balance.

In our recent work, we have identified specific network properties in the visual cortex which are homeostatically regulated following prolonged monocular deprivation (the deprivation of vision in one eye), by linking experimental measurements of individual synaptic strengths and large-scale network dynamics (Torrado Pacheco et al. 2019, Miska et al. 2018). We now propose to investigate the intricate interaction between diverse forms of Hebbian and homeostatic plasticity mechanisms, which regulate these network properties in the presence or absence of sensory input using extensive computational modeling and data analysis. At the same time, we present theoretical arguments for specific computations performed by the networks under the control of these mechanisms that enable them to efficiently and reliably encode sensory information, as well as discriminate between different stimuli. To achieve this, we will be strongly guided by experimental data from existing and novel collaborations in the context of the CRC, which will help us constrain the modeled mechanisms, as well as provide opportunities to test predictions from the models in real biological networks.

First, we will investigate how the interaction of excitatory and inhibitory plasticity shapes network dynamics under conditions of normal vision, and enables the restoration of different aspects of network dynamics, including firing rates and correlations, when sensory input is removed. We will build on existing work to incorporate plasticity in the feed-forward pathway from the thalamus to the visual cortex, and investigate the specificity of inhibitory tuning relative to excitatory tuning in the recurrent cortical circuit. We will apply information theoretic analysis to quantify the rate of information transfer between the thalamus and the cortex in experimental data, and compare it to our modeling outcomes to dissect the relative importance of the two pathways (feedforward vs. recurrent) in representing and transmitting sensory stimuli most efficiently. Second, we will investigate the homeostatic regulation of an important parameter for normal network function, the E-I ratio, which gradually increases from development to adulthood, enabling excitation and inhibition to become co-tuned. We will investigate how E-I balance is defined locally, and model the mechanisms for establishing this balance at a specific set-point under normal sensory input and for maintaining it during sensory deprivation. For this process, we hypothesize the relative importance of a fast-acting homeostatic mechanism – heterosynaptic plasticity – which modifies connections between synapses which are not activated during the induction of classical Hebbian, also called homosynaptic, plasticity. Importantly, we will investigate what a homeostatic E-I set-point might imply for the emergence of computations in cortical circuits, including the encoding/learning of new stimuli and the stable representation and recall of already learned stimuli. Third, we will study the role of different cell types, and especially inhibitory interneuron types, in shaping non-random cortical circuit connectivity and relate that to emergent structured network activity that connects feedforward information from the sensory world with internally generated signals. Specifically, we will study how feed-forward information from the thalamus conveyed by parvalbumin-positive interneurons interacts with feedback information from the local recurrent network conveyed by somatostatin-positive interneurons. We hypothesize that these different types of interneurons play distinct roles in the discrimination vs. reliability of encoding of different sensory stimuli, and will explore the potentially different types of homeostatic plasticity at different synapses in the presence or absence of sensory input.