Aniruddha Yadav

Morphologic evidence for spatially clustered spines in apical dendrites of monkey neocortical pyramidal cells

The general organization of neocortical connectivity in rhesus monkey is relatively well understood. However, mounting evidence points to an organizing principle that involves clustered synapses at the level of individual dendrites. Several synaptic plasticity studies have reported cooperative interaction between neighboring synapses on a given dendritic branch, which may potentially induce synapse clusters. Additionally, theoretical models have predicted that such cooperativity is advantageous, in that it greatly enhances a neurons computational repertoire. However, largely because of the lack of sufficient morphologic data, the existence of clustered synapses in neurons on a global scale has never been established. The majority of excitatory synapses are found within dendritic spines. In this study, we demonstrate that spine clusters do exist on pyramidal neurons by analyzing the three‐dimensional locations of ∼40,000 spines on 280 apical dendritic branches in layer III of the rhesus monkey prefrontal cortex. By using clustering algorithms and Monte Carlo simulations, we quantify the probability that the observed extent of clustering does not occur randomly. This provides a measure that tests for spine clustering on a global scale, whenever high‐resolution morphologic data are available. Here we demonstrate that spine clusters occur significantly more frequently than expected by pure chance and that spine clustering is concentrated in apical terminal branches. These findings indicate that spine clustering is driven by systematic biological processes. We also found that mushroom‐shaped and stubby spines are predominant in clusters on dendritic segments that display prolific clustering, independently supporting a causal link between spine morphology and synaptic clustering. J. Comp. Neurol. 520:2888–2902, 2012.

Early fear memory defects are associated with altered synaptic plasticity and molecular architecture in the TgCRND8 Alzheimer's disease mouse model

Alzheimers disease (AD) is a complex and slowly progressing dementing disorder that results in neuronal and synaptic loss, deposition in brain of aberrantly folded proteins, and impairment of spatial and episodic memory. Most studies of mouse models of AD have employed analyses of cognitive status and assessment of amyloid burden, gliosis, and molecular pathology during disease progression. Here we sought to understand the behavioral, cellular, ultrastructural, and molecular changes that occur at a pathological stage equivalent to the early stages of human AD. We studied the TgCRND8 mouse, a model of aggressive AD amyloidosis, at an early stage of plaque pathology (3 months of age) in comparison to their wildtype littermates and assessed changes in cognition, neuron and spine structure, and expression of synaptic glutamate receptor proteins. We found that, at this age, TgCRND8 mice display substantial plaque deposition in the neocortex and hippocampus and impairment on cued and contextual memory tasks. Of particular interest, we also observed a significant decrease in the number of neurons in the hippocampus. Furthermore, analysis of CA1 neurons revealed significant changes in apical and basal dendritic spine types, as well as altered expression of GluN1 and GluA2 receptors. This change in molecular architecture within the hippocampus may reflect a rising representation of inherently less stable thin spine populations, which can cause cognitive decline. These changes, taken together with toxic insults from amyloid‐β protein, may underlie the observed neuronal loss. J. Comp. Neurol. 522:2319–2335, 2014.

Automated evolutionary optimization of ion channel conductances and kinetics in models of young and aged rhesus monkey pyramidal neurons

Conductance-based compartment modeling requires tuning of many parameters to fit the neuron model to target electrophysiological data. Automated parameter optimization via evolutionary algorithms (EAs) is a common approach to accomplish this task, using error functions to quantify differences between model and target. We present a three-stage EA optimization protocol for tuning ion channel conductances and kinetics in a generic neuron model with minimal manual intervention. We use the technique of Latin hypercube sampling in a new way, to choose weights for error functions automatically so that each function influences the parameter search to a similar degree. This protocol requires no specialized physiological data collection and is applicable to commonly-collected current clamp data and either single- or multi-objective optimization. We applied the protocol to two representative pyramidal neurons from layer 3 of the prefrontal cortex of rhesus monkeys, in which action potential firing rates are significantly higher in aged compared to young animals. Using an idealized dendritic topology and models with either 4 or 8 ion channels (10 or 23 free parameters respectively), we produced populations of parameter combinations fitting the target datasets in less than 80 hours of optimization each. Passive parameter differences between young and aged models were consistent with our prior results using simpler models and hand tuning. We analyzed parameter values among fits to a single neuron to facilitate refinement of the underlying model, and across fits to multiple neurons to show how our protocol will lead to predictions of parameter differences with aging in these neurons.

Why are pyramidal cell firing rates increased with aging, and what can we do about it?

Altered neuronal morphology and electrophysiological function in aged primates are correlated with cognitive deficits [1]. Recent experimental studies of young and aged layer 2/3 pyramidal neurons of the prefrontal cortex (PFC) of rhesus monkeys show an age related increase in both the somatic input resistance and action potential (AP) firing rate [1]. Aged cells display fewer apical dendrites, and reduced spine numbers, although the average spine in an aged cell is larger [2]. Figure 1 compares the morphology and firing patterns for typical young (left column) and aged (right column) pyramidal cells from the PFC of rhesus monkeys [2], illustrating that an aged cell fires at a higher frequency for the same 2s current injection (380 pA) at the soma, despite being similar in overall size and morphology to the young cell.

Functional consequences of age-related morphologic changes to pyramidal neurons of the rhesus monkey prefrontal cortex

Layer 3 (L3) pyramidal neurons in the lateral prefrontal cortex (LPFC) of rhesus monkeys exhibit dendritic regression, spine loss and increased action potential (AP) firing rates during normal aging. The relationship between these structural and functional alterations, if any, is unknown. To address this issue, morphological and electrophysiological properties of L3 LPFC pyramidal neurons from young and aged rhesus monkeys were characterized using in vitro whole-cell patch-clamp recordings and high-resolution digital reconstruction of neurons. Consistent with our previous studies, aged neurons exhibited significantly reduced dendritic arbor length and spine density, as well as increased input resistance and firing rates. Computational models using the digital reconstructions with Hodgkin-Huxley and AMPA channels allowed us to assess relationships between demonstrated age-related changes and to predict physiological changes that have not yet been tested empirically. For example, the models predict that in both backpropagating APs and excitatory postsynaptic currents (EPSCs), attenuation is lower in aged versus young neurons. Importantly, when identical densities of passive parameters and voltage- and calcium-gated conductances were used in young and aged model neurons, neither input resistance nor firing rates differed between the two age groups. Tuning passive parameters for each model predicted significantly higher membrane resistance (Rm) in aged versus young neurons. This Rm increase alone did not account for increased firing rates in aged models, but coupling these Rm values with subtle differences in morphology and membrane capacitance did. The predicted differences in passive parameters (or parameters with similar effects) are mathematically plausible, but must be tested empirically.Layer 3 (L3) pyramidal neurons in the lateral prefrontal cortex (LPFC) of rhesus monkeys exhibit dendritic regression, spine loss and increased action potential (AP) firing rates during normal aging. The relationship between these structural and functional alterations, if any, is unknown. To address this issue, morphological and electrophysiological properties of L3 LPFC pyramidal neurons from young and aged rhesus monkeys were characterized using in vitro whole-cell patch-clamp recordings and high-resolution digital reconstruction of neurons. Consistent with our previous studies, aged neurons exhibited significantly reduced dendritic arbor length and spine density, as well as increased input resistance and firing rates. Computational models using the digital reconstructions with Hodgkin-Huxley and AMPA channels allowed us to assess relationships between demonstrated age-related changes and to predict physiological changes that have not yet been tested empirically. For example, the models predict that in both backpropagating APs and excitatory postsynaptic currents (EPSCs), attenuation is lower in aged versus young neurons. Importantly, when identical densities of passive parameters and voltage- and calcium-gated conductances were used in young and aged model neurons, neither input resistance nor firing rates differed between the two age groups. Tuning passive parameters for each model predicted significantly higher membrane resistance (R m ) in aged versus young neurons. This R m increase alone did not account for increased firing rates in aged models, but coupling these R m values with subtle differences in morphology and membrane capacitance did. The predicted differences in passive parameters (or parameters with similar effects) are mathematically plausible, but must be tested empirically.

Age-related morphologic changes alter robustness of neuronal function

Aging and neurodegenerative disorders significantly alter both neuronal function and morphology. We have shown recently that, if uncompensated by altered ion channel distributions, morphologic changes in aging increase neuronal excitability [1]. Different combinations of active conductances can evoke similar firing patterns, giving rise to functional homeostasis. To elucidate possible mechanisms of homeostasis in the parameter space defined by ionic conductances, we recently utilized thermal random walks to identify regions (an ensemble) within which the functional output of a neuron is maintained. Typically a number of such regions, which are topologically disconnected from each other, are identified. The eigenvalues and eigenvectors of the ensemble covariance matrix serve as measures for its size and orientation, and the capacity for functional homeostasis or robustness is measured by the set of largest (robust) eigenvalues and corresponding eigenvectors. In this study, we explore how functional homeostasis might be affected by the kinds of morphologic changes that occur with aging or neurodegenerative disorders. To this end, we extract the change in orientation and extent of the ensembles found for the original model by subjecting it to three kinds of perturbations involving morphology and desired firing properties. First, we mimic age-related morphologic degeneration by reducing the number of spines in the original neuron. Second, we increase the number of spines in the original neuron, mimicking a younger one. The third perturbation also involves increased spines, however, the target firing rate is reduced so as to also mimic the electrophysiology of a younger neuron. We find that, for all the perturbations considered, the initial ensembles remain topologically disconnected, although they change in size and orientation. In particular, while the sensitive eigendirections remain unchanged, the robust eigendirections change in orientation, suggesting that functional robustness of a neuron is determined largely by its morphology. Consequently, possible homeostatic mechanisms, that enable a young neuron to maintain function, are able to do so less efficiently as the neuron undergoes age related morphologic changes .This is so because a homeostatic mechanism would be more effective if it were to orient itself along the robust eigendirections of the ensembles, and become less so as the robust directions change due to morphologic changes. Although morphologic perturbations drastically change robust eigendirections, preliminary evidence indicates that these directions may be restored through appropriate compensatory changes in functional output. In conclusion, the need to maintain functional robustness may be responsible for altered function as a neuron ages.

Network effects of age-related NMDA reduction in a model of working memory

Specific cognitive deficits in working memory tasks are associated with normal aging in humans and nonhuman primates, even in the absence of pathologies such as Alzheimers disease. Because normal aging does not involve widespread neuron death or gross morphological degeneration, the cause of these deficits remains unclear, although subtle anatomical and physiological effects are likely to play a causal role [1]. NMDA glutamate receptors in spines mediate synaptic communication and are critical in long-term potentiation (LTP) and learning. NMDAs dependence on postsynaptic depolarization has been proposed to enhance network stability through synaptic bistability [2]. Reduced NMDA expression is associated with aging in macaque monkeys and rats [3].

Functional consequences of age-related morphologic changes in pyramidal neurons of the rhesus monkey prefrontal cortex

In normal aging, neocortical pyramidal neuron dendrites and dendritic spines undergo significant changes [1,2], often with concomitant physiological changes. In layer 3 of the prefrontal cortex (PFC) of the rhesus monkey, aged pyramidal neurons have a significantly higher input resistance and higher action potential (AP) firing rates in vitro compared to young neurons [3]. Our multidimensional approach combines whole-cell patch clamp recording, confocal microscopy, 3D digital reconstruction, and computational modeling to explore structure/function relationships. We now have a unique database of electrophysiological recordings, morphologic reconstructions, and compartment models from six young and six aged layer 3 pyramidal neurons from the rhesus monkey PFC. As in prior studies [4], the length of individual dendritic branches were significantly shorter in aged than in young neurons, with fewer dendritic spines. These morphological changes significantly reduced the somatofugal and somatopetal dendritic voltage attenuation in aged versus young model neurons. However, they were insufficient to account for the increase in input resistance observed in vitro, even after including synaptic background activity constrained by cell-specific total spine number. This suggests that specific membrane resistance (Rm) is higher on average in aged neurons. Using our recently developed model [5], we conducted a systematic sampling of the parameter space of Hodgkin-Huxley maximal conductances for each of the twelve model neurons, to fit firing rates of each model to the mean young and aged rates recorded empirically. In both age groups, some model neurons had several good fits to empirical data, while others had no good fits; there was no difference in the number of best-fit parameter sets from young and aged models. When the same conductance parameters were applied to all models, the mean firing rates of young and aged model neurons did not differ. This result also held when different values of Rm were applied to each age group, and when a wider array of voltage- and calcium-gated ion channels were included. Overall these simulations predict that age-related morphologic differences do affect dendritic signal integration, but do not account for changes in neuronal excitability observed in in vitro recordings. Our modeling suggests that morphology, passive cable properties, and active channel conductances could trade off against one another, constraining neuronal excitability within a certain range for each age group. Even so, we predict that the membrane resistance and active channel conductances of PFC pyramidal cells are changed with aging. Such predictions begin to reveal how networks comprising these neurons may function differently in young and aged animals.

Modeling predicts that parameters shaping action potentials and synaptic responses differ in pyramidal neurons of the visual and prefrontal cortices

Pyramidal neurons in the prefrontal cortex integrate inputs that are greater in number and diversity than pyramidal neurons of the primary visual cortex [1]. We have recently characterized the morphology and physiology of layer 3 neurons from area 46 of the dorsolateral prefrontal cortex (dlPFC) and from visual area V1 in the rhesus monkey. Ultra-high resolution confocal imaging and 3D reconstruction revealed that the dendritic arbors of V1 neurons are much smaller and less complex than those of the dlPFC, and possess far fewer dendritic spines. Physiologically, whole cell patch clamp recordings demonstrated that V1 neurons have twice as high mean input resistance and significantly increased action potential firing rates compared to dlPFC neurons. Further, AMPAR-mediated spontaneous excitatory postsynaptic currents (sEPSCs) exhibit significantly faster kinetics and smaller mean amplitudes in V1 compared to dlPFC neurons. We have used compartment modeling with the NEURON simulation environment [2] to investigate the extent to which morphology accounts for the observed physiological differences. We first applied a relatively simple model, including only Hodgkin-Huxley sodium and potassium currents [3], to representative 3D reconstructions of neurons from dlPFC and V1. This model predicts that morphology alone leads to major differences in the attenuation of electrical signals coming into, and leaving, the soma in neurons from the two brain areas. It also predicts that morphology largely accounts for the increased excitability of V1 cells, but not the differences in kinetics and amplitudes of sEPSCs. The model predicts that parameters controlling passive voltage spread, the density of Hodgkin-Huxley channels, and synaptic inputs likely differ between dlPFC and V1 neurons. Yet, the simple model on which these predictions are based reproduces neither the H-current dependent ‘sag’ response during hyperpolarization, nor spike frequency adaptation, both of which are observed in dlPFC and V1 neurons. We have extended the model to include the H current [4], plus a calcium-dependent potassium channel with an associated one-pool source of intracellular calcium [5]. We have tuned the model with a combination of manual and automatic optimization techniques to match the resting potential, input resistance, and overall firing rates of the recorded dlPFC and V1 neurons. As in the simpler model, we find that morphology alone cannot account for observed physiological differences. Therefore, we predict that parameters shaping action potentials and synaptic input differ between dlPFC and V1.

Improved parameter fitting for models of young and aged neurons.

The parameters of neuronal compartment models must be determined carefully in order to match experimental data. Optimization algorithms can simplify this task by automatically searching the multidimensional parameter space to identify combinations of parameters that best fit experimental data, as measured by a fitness function that represents salient differences between simulated and experimental data. The success of automated parameter fitting depends critically on many issues, including the choice of parameters selected for fitting, the parameter search method, and the design of the fitness function. Previously we developed a fitness function that explicitly quantifies the shape of action potentials and afterhyperpolarizations [1], and the time-varying firing rate [2]. The present project arises from our study of neocortical pyramidal cells from young and aged rhesus monkeys in vitro [3]. In response to current clamp protocols that evoked action potentials, aged neurons fired at significantly higher rates than young ones did. Yet, both young and aged neurons displayed an initial fast phase of firing rate adaptation followed by a slower one. Neither parameter optimization with our existing fitness function, nor the popular phase-plane method [4], sufficiently captured the two-phase adaptation. We have extended our fitness function to fit firing rate time series with two single exponentials, separated by an automatically-chosen cutoff. The fitness function also includes a term to quantify action potential backpropagation, whereby voltage decays approximately exponentially with distance from the soma [5]. The fitness function is designed for use with NEURON, with the parameter search guided by MATLAB’s Global Optimization Toolbox. We used the fitness function and search method to optimize morphologically accurate compartmental models of young and aged neurons to their respective physiological data. We compared results from our original and improved fitness functions against the phase-plane fitness function, separately optimizing either ion channel densities or channel kinetics. All searches identified multiple combinations of parameters that fit the data, with our new fitness function having the best qualitative matches to empirical data. These parameter fits will be used to identify biological mechanisms that may be responsible for the physiological differences between young and aged neurons.