Beta amyloid (Aβ), a key component in the pathophysiology of Alzheimer’s disease (AD) is thought to target excitatory synapses early in the disease. We show that excessive GluA2 receptor endocytosis, mediated by a PDZ ligand-domain interaction with protein interacting with C kinase (PICK1) and protein kinase C alpha (PKCα), is required for Aβ to weaken synapses. In mice lacking PICK1 or PKCα, elevations of Aβ fail to depress synaptic transmission in cultured brain slices. We find that reintroducing PKCα in PKCα null mouse slices rescues the Aβ-induced depression. Moreover, the reintroduction of a PDZ mutant form of PKCα does not rescue the synaptic depression. In addition, analysis of whole genome sequencing (WGS) of late onset AD (LOAD) families’ data identified two rare highly penetrant variants in the PKCα gene (PRKCA)

that display increased signaling output relative to wild-type PKCα. We conclude that targeting these proteins could accelerate the development of treatments that could prevent, halt or reverse the course of the disease as well as improve its early detection.

Excitatory synapses possess a vast array of proteins, including glutamate receptors such as α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptors, which mediate fast synaptic transmission. There are four different AMPA receptor subunits: GluA1, GluA2, GluA3, and GluA4). AMPA receptors are generally found in the adult rodent brain in two heteromeric forms: GluA1/GluA2 receptors and GluA3/GluA2 receptors. Trafficking of GluA1/GluA2 receptors is activity-dependent, while GluA3/GluA2 heteromers are constitutively cycling and replacing synaptic GluA1/GluA2 and GluA3/GluA2 receptors. Given the importance of the GluA1 subunit in plasticity, on which the absence of GluA3 has little effect, GluA3 has not been well-studied. As it turns out, GluA3 is required for the expression of the synaptic dysfunction caused by amyloid-β (Aβ), a small peptide thought to be responsible for the pathogenesis of Alzheimer’s disease. In the absence of GluA3, Aβ does not cause synaptic depression, memory deficits, or block long-term potentiation (LTP). These results will be described in Chapter 2.

Studies have demonstrated that Aβ causes synaptic dysfunction by activating signaling pathways involved in long-term depression (LTD). Previous studies from our lab established successfully engineered associative fear memories by pairing optogenetic activation of the auditory cortex (AuC) and medial geniculate nucleus (MGN) with footshock in cued fear conditioning. Replacing the traditional conditioned stimulus (CS) with an optogenetically-delivered input (ODI) allowed us to utilize LTP and LTD protocols to activate and inactivate memories. To examine if Aβ in Alzheimer’s disease behaves as LTD in vivo, Aβ and a blue-shifted form of channelrhodopsin (ChR) known as oChIEF were co-expressed in the AuC and MGN. A class of Alzheimer’s disease therapeutics known as γ-secretase modulators (GSMs) were used to control Aβ levels, and a behaviorial paradigm was established to determine whether Aβ could remove inactivate memories like LTD, and later be recovered with LTP in the presence of GSMs. This study will be described in Chapter 3.

Lastly, given the importance of LTP in learning and memory, a new method for identifying recently potentiated synapses was investigated. One existing strategy for assessing learning-induced plasticity involves the use of two-photon laser scanning microscopy (TPLSM) in vivo, which can provide high visual and temporal resolution on synapses undergoing potentiation. However, a caveat to this technique is that TPLSM techniques are labor intensive and thus low throughput. Another strategy for brain regions involved in a learning task is examining the localization of immediate- early gene (IEG) transcripts and proteins. Despite this method being relatively high throughput, the correlation between synaptic strength and IEG expression is poorly understood. Thus, we established a new technique, which we call SYNaptic Proximity Ligation Assay. SYNPLA applies the antibody-based methodology of PLA to the diversity of transsynaptic proteins to label synapses. In Chapter 4, SYNPLA’s labeling of all synapses or recently potentiated synapses will be described.

Lateral habenula has been reported to carry the reward negative feature that distinguishes it from many other subcortical brain regions. Lateral habenula neurons are inhibited by rewarding events and activated by unpleasant events. Previous reports have revealed the role of LHb in encoding reward information. However, it is not clear how aversive stimulus affect reward processes in LHb neurons. Here we used two-photon excited fluorescence scanning microscopy to record calcium signal through GRIN lens implanted in mice with GCaMP6S expression in LHb. Based on neural responsive pattern to reward and reward omission, we categorized imaged neurons into reward-selective (RS), non-specific (NS), and non-responsive (NR) neurons. By inducing acute stress (tail shock), we observed that the reward response of RS neurons was inverted, and anticipation and consumption of reward significantly decreased. Later, we also found that the reward response in LHb neurons can be manipulated by switching between presence and absence of stress.

Chapter 1: The neural mechanisms conferring reduced motivation, as observed in depressed individuals, is poorly understood. Here we examine in rodents if reduced motivation to exert effort is controlled by transmission from the lateral habenula (LHb, a nucleus overactive in depressed-like states) to the rostromedial tegmental nucleus (RMTg, a nucleus that inhibits dopaminergic neurons). In an aversive test wherein immobility indicates loss of effort, LHb→RMTg transmission increased during transitions into immobility; driving LHb→RMTg increased immobility; and inhibiting LHb→RMTg produced opposite effects. In an appetitive test, driving LHb→RMTg reduced effort exerted to receive a reward, without affecting the reward’s hedonic property. Notably, LHb→RMTg stimulation had an effect on only specific aspects of motor tasks and promoted avoidance, indicating that LHb→RMTg activity does not generally reduce movement but appears to carry a negative valence that reduce effort. These results indicate that LHb→RMTg activity controls the motivation to exert effort and may contribute to the reduced motivation in depression.

Chapter 2: Co-release of neurotransmitters has been detected in numerous brain regions under physiological and pathological conditions. However, if the opponent neurotransmitters γ-aminobutyric acid (GABA) and glutamate are normally released from single vesicles is not well established. A recent exhaustive study demonstrated that GABA and glutamate are released from individual presynaptic neurons originating at several nuclei targeting the lateral habenula (LHb). Their anatomical techniques indicated that GABA-containing and glutamate containing presynaptic vesicles are segregated. Here we use electrophysiology to characterize a different synaptic projection onto the lateral habenula, and find that the majority of individual quantal events contain a dual glutamate and GABA composition. The lateral habenula may therefore have multiple means to balance excitation and inhibition by co-released GABA and glutamate.

Due to the firing of lateral habenula neurons (LHb) biasing a negative perspective of the world, an increase in LHb activity has been implicated in causing depression. Inputs to the LHb from higher-level cognition areas are likely providing the necessary signals for the LHb to respond to aversive stimuli and subsequently increase in activity. Projections to the LHb from the medial prefrontal cortex (mPFC) that processes negative valence events, have been identified, but the functions and characteristics of mPFC inputs to LHb still remain unknown. Here, we have identified and characterized mPFC inputs to LHb to be excitatory, glutamatergic, and under selective neuromodulation by norepinephrine (NE). Inputs from two other brain regions that may be instructing the LHb to process aversive stimuli, the ventral tegmental area (VTA) and the lateral hypothalamic area (LHA), were not affected by the addition of NE. We therefore propose that the lack of effect by NE on both VTA and LHA inputs may be due to the activation of adrenergic receptors at mPFC terminals, but not at VTA or LHA terminals, facilitating increased glutamate release.

The miniature excitatory post synaptic current (mEPSC) is a fundamental electrophysiological measurement of quantal vesicular release of transmitter at synapses. Typically, for excitatory transmission at central nervous system neurons, due to the small size of the electrical signal relative to the noise, recordings containing mEPSCs are analyzed by eye in conjunction with a basic peak/template finding algorithm. Such methods require expertise and considerable time, and generally require elimination of signals smaller than ~8-10 pA to avoid counting noise (i.e. false positives). Here we describe an entirely automated approach to detect mEPSCs using a machine learning based tool. This method eliminates inter observer bias, and permits accurate detection of spontaneous mEPSCs smaller than 10 pA from brain slice neurons. Using whole cell recordings from the soma of CA1 hippocampal slice neurons this tool can detect more than 95% of mEPSCs identified by a trained human observer. Importantly, this tool identifies few events as mEPSCs in a recording of the same cell when excitatory postsynaptic receptors are blocked, indicating a low false positive rate. Such recordings permit calculation of a true (96.6 ± 0.9%; n=10) and false (1.4 ± 0.1%; n=10) positive detection rate for mEPSCs. Using this new tool mEPSCs can be rapidly and accurately measured in an unbiased manner, and permits analysis of previously undetected mEPSCs.