Faculty Bibliography

Sensory perception depends on neocortical computations that contextually adjust sensory signals in different internal and environmental contexts. Neocortical layer 1 (L1) is the main target of cortical and subcortical inputs that provide "top-down" information for context-dependent sensory processing. Although L1 is devoid of excitatory cells, it contains the distal "tuft" dendrites of pyramidal cells (PCs) located in deeper layers. L1 also contains a poorly characterized population of GABAergic interneurons (INs), which regulate the impact that different top-down inputs have on PCs. A poor comprehension of L1 IN subtypes and how they affect PC activity has hampered our understanding of the mechanisms that underlie contextual modulation of sensory processing. We used novel genetic strategies in male and female mice combined with electrophysiological and morphological methods to help resolve differences that were unclear when using only electrophysiological and/or morphological approaches. We discovered that L1 contains four distinct populations of INs, each with a unique molecular profile, morphology, and electrophysiology, including a previously overlooked IN population (named here "canopy cells") representing 40% of L1 INs. In contrast to what is observed in other layers, most L1 neurons appear to be unique to the layer, highlighting the specialized character of the signal processing that takes place in L1. This new understanding of INs in L1, as well as the application of genetic methods based on the markers described here, will enable investigation of the cellular and circuit mechanisms of top-down processing in L1 with unprecedented detail.SIGNIFICANCE STATEMENT Neocortical layer 1 (L1) is the main target of corticocortical and subcortical projections that mediate top-down or context-dependent sensory perception. However, this unique layer is often referred to as "enigmatic" because its neuronal composition has been difficult to determine. Using a combination of genetic, electrophysiological, and morphological approaches that helped to resolve differences that were unclear when using a single approach, we were able to decipher the neuronal composition of L1. We identified markers that distinguish L1 neurons and found that the layer contains four populations of GABAergic interneurons, each with unique molecular profiles, morphologies, and electrophysiological properties. These findings provide a new framework for studying the circuit mechanisms underlying the processing of top-down inputs in neocortical L1.

Neuronal hyperexcitability is one of the major characteristics of Fragile X syndrome (FXS), yet the molecular mechanisms of this critical dysfunction remain poorly understood. Here we report a major role of voltage-independent K⁺ channel dysfunction in hyperexcitability of CA3 pyramidal neurons in Fmr1 KO mice. We observed a reduction of voltage-independent small conductance Ca²⁺-activated K⁺ (SK) currents in both male and female mice leading to decreased AP threshold and reduced medium afterhyperpolarization (mAHP). These SK channel-dependent deficits led to markedly increased AP firing and abnormal input-output signal transmission of CA3 pyramidal neurons. The SK current defect was mediated, at least in part, by loss of FMRP interaction with the SK channels (specifically the SK2 isoform), without changes in the channel expression. Intracellular application of selective SK channel openers or a genetic reintroduction of an N-terminal FMRP fragment lacking the ability to associate with polyribosomes normalized all observed excitability defects in CA3 pyramidal neurons of Fmr1 KO mice. These results suggest that dysfunction of voltage-independent SK channels is the primary cause of CA3 neuronal hyperexcitability in Fmr1 KO mice and support the critical translation-independent role for FMRP as a regulator of neural excitability. Our findings may thus provide a new avenue to ameliorate hippocampal excitability defects in FXS.SIGNIFICANCE STATEMENTDespite two decades of research, no effective treatment is currently available for Fragile X syndrome (FXS). Neuronal hyperexcitability is widely considered as one of the hallmarks of FXS. Excitability research in the FXS field has thus far focused primarily on voltage-gated ion channels, while contributions from voltage-independent channels have been largely overlooked. Here we report that voltage-independent SK channel dysfunction causes hippocampal neuron hyperexcitability in the FXS mouse model. Our results support a major role for translation-independent FMRP function in regulating ion channel activity, and specifically the SK channels, in hyperexcitability defects in FXS. Our findings may thus open a new direction to ameliorate hippocampal excitability defects in FXS.

Neuronal signals in the prefrontal cortex have been reported to predict upcoming decisions. Such activity patterns are often coupled to perceptual cues indicating correct choices or values of different options. How does the prefrontal cortex signal future decisions when no cues are present but when decisions are made based on internal valuations of past experiences with stochastic outcomes? We trained rats to perform a two-arm bandit-task, successfully adjusting choices between certain-small or possible-big rewards with changing long-term advantages. We discovered specialized prefrontal neurons, whose firing during the encounter of no-reward predicted the subsequent choice of animals, even for unlikely or uncertain decisions and several seconds before choice execution. Optogenetic silencing of the prelimbic cortex exclusively timed to encounters of no reward, provoked animals to excessive gambling for large rewards. Firing of prefrontal neurons during outcome evaluation signals subsequent choices during gambling and is essential for dynamically adjusting decisions based on internal valuations.

In addition to being the greatest genetic risk factor for Alzheimer's disease, expression of the ɛ4 allele of apolipoprotein E can lead to cognitive decline during ageing that is independent of Alzheimer's amyloid-β and tau pathology. In human post-mortem tissue and mouse models humanized for apolipoprotein E, we examined the impact of apolipoprotein E4 expression on brain exosomes, vesicles that are produced within and secreted from late-endocytic multivesicular bodies. Compared to humans or mice homozygous for the risk-neutral ɛ3 allele we show that the ɛ4 allele, whether homozygous or heterozygous with an ɛ3 allele, drives lower exosome levels in the brain extracellular space. In mice, we show that the apolipoprotein E4-driven change in brain exosome levels is age-dependent: while not present at age 6 months, it is detectable at 12 months of age. Expression levels of the exosome pathway regulators tumor susceptibility gene 101 (TSG101) and Ras-related protein Rab35 (RAB35) were found to be reduced in the brain at the protein and mRNA levels, arguing that apolipoprotein E4 genotype leads to a downregulation of exosome biosynthesis and release. Compromised exosome production is likely to have adverse effects, including diminishing a cell's ability to eliminate materials from the endosomal-lysosomal system. This reduction in brain exosome levels in 12-month-old apolipoprotein E4 mice occurs earlier than our previously reported brain endosomal pathway changes, arguing that an apolipoprotein E4-driven failure in exosome production plays a primary role in endosomal and lysosomal deficits that occur in apolipoprotein E4 mouse and human brains. Disruption of these interdependent endosomal-exosomal-lysosomal systems in apolipoprotein E4-expressing individuals may contribute to amyloidogenic amyloid-β precursor protein processing, compromise trophic signalling and synaptic function, and interfere with a neuron's ability to degrade material, all of which are events that lead to neuronal vulnerability and higher risk of Alzheimer's disease development. Together, these data suggest that exosome pathway dysfunction is a previously unappreciated component of the brain pathologies that occur as a result of apolipoprotein E4 expression.

We determined whether the structural and functional integrity of amacrine cells (ACs), the largest cohort of neurons in the mammalian retina, are affected in glaucoma. Intraocular injection of microbeads was made in mouse eyes to elevate intraocular pressure as a model of experimental glaucoma. Specific immunocytochemical markers were used to identify AC and displaced (d)ACs subpopulations in both the inner nuclear and ganglion cell layers, respectively, and to distinguish them from retinal ganglion cells (RGCs). Calretinin- and γ-aminobutyric acid (GABA)-immunoreactive (IR) cells were highly vulnerable to glaucomatous damage, whereas choline acetyltransferase (ChAT)-positive and glycinergic AC subtypes were unaffected. The AC loss began 4 weeks after initial microbead injection, corresponding to the time course of RGC loss. Recordings of electroretinogram (ERG) oscillatory potentials and scotopic threshold responses, which reflect AC and RGC activity, were significantly attenuated in glaucomatous eyes following a time course that matched that of the AC and RGC loss. Moreover, we found that it was the ACs coupled to RGCs via gap junctions that were lost in glaucoma, whereas uncoupled ACs were largely unaffected. Our results suggest that AC loss in glaucoma occurs secondary to RGC death through the gap junction-mediated bystander effect. J. Comp. Neurol., 2016. © 2016 Wiley Periodicals, Inc.

Learning to detect content-independent transformations from data is one of the central problems in biological and artificial intelligence. An example of such problem is unsupervised learning of a visual motion detector from pairs of consecutive video frames. Rao and Ruderman formulated this problem in terms of learning infinitesimal transformation operators (Lie group generators) via minimizing image reconstruction error. Unfortunately, it is difficult to map their model onto a biologically plausible neural network (NN) with local learning rules. Here we propose a biologically plausible model of motion detection. We also adopt the transformation-operator approach but, instead of reconstruction-error minimization, start with a similarity-preserving objective function. An online algorithm that optimizes such an objective function naturally maps onto an NN with biologically plausible learning rules. The trained NN recapitulates major features of the well-studied motion detector in the fly. In particular, it is consistent with the experimental observation that local motion detectors combine information from at least three adjacent pixels, something that contradicts the celebrated Hassenstein-Reichardt model.

A genetic pathway that times development works together with the sex-determination pathway to control the timing of sexually dimorphic neural development in C. elegans.

Over the past two decades, mathematicians and neuroscientists at New York University have developed several large-scale computational models of a layer of macaque primary visual cortex. Here we provide an overview of these models, organized by the specific questions about cortical processing that each model addressed. Each model was founded upon the available anatomical and physiological data; and not by building into the model network assumptions about theoretical mechanisms specifically designed to enable the network to produce desired response properties. Also, our aim was to use one comprehensive network, with a fixed architecture and one set of parameters, to model all experiments. The response properties of individual neurons and populations of neurons then emerge from this experimentally constrained model. This overview is dedicated to Professor David Cai, who played a leading role in several of the models described here. We are very fortunate to have had the opportunity to work with him over the past two decades.