Impact of COVID-19 on the Nervous System

Post by Amanda McFarlan

What's the science?

The spread of the SARS-CoV-2 virus, better known as COVID-19, has become a global public health concern. Infection with COVID-19 is known to affect the respiratory system, causing fever, dry coughs, and in more serious cases, shortness of breath and chest pain. Although it is less common, some individuals infected with COVID-19 also display neurological symptoms such as headache, dizziness, and a loss of taste and smell. However, it remains unknown exactly what the COVID-19 virus is doing in the body to cause these neurological symptoms. This week in Cell, Iadecola and colleagues summarized the recent data describing the effects of COVID-19 on the nervous system.

What do we already know?

Neurological abnormalities, ranging from a headache to total delirium, have been observed in ~30% of hospitalized individuals infected with COVID-19. In some cases, COVID-19 has also been shown to have delayed neurological symptoms post-infection affecting both the central and peripheral nervous systems. Studies have shown that infection with COVID-19 increases an individual’s risk of having a stroke. Indeed, the incidence of stroke in hospitalized patients with COVID-19 is ~7 times higher compared to patients with influenza, which suggests that the COVID-19 virus is likely to be involved in causing vascular brain injuries. Additionally, the majority of studies testing the cerebrospinal fluid (CSF) and brain tissue samples of COVID-19 patients with neurological abnormalities did not show evidence of viral invasion, however, small quantities of the COVID-19 virus were detected in the CSF of at least two cases of hospitalized individuals with severe infectious encephalopathy (altered mental state due to damage to the brain). The peripheral effects of COVID-19 could potentially explain the occurrence of encephalopathy without evidence of brain invasion by the virus. For example, COVID-19 is known to act on the respiratory system and in severe cases, can cause lung damage which can lead to hypoxia (lack of oxygen). In line with this, post-mortem studies from COVID-19-related deaths revealed that there was brain damage in regions most susceptible to hypoxic brain injury: the neocortex, hippocampus, and cerebellum.

What’s new?

One major question to ask when investigating the effects of COVID-19 on the nervous system is: can the COVID-19 virus enter the brain, and if so, how? The main entry point for COVID-19 into human cells is through an enzyme that is located on the membrane of cells in the lungs, kidneys, arteries, heart, and intestines called the angiotensin converting enzyme-2 (ACE2). Recent data has shown that ACE2 was also expressed in the choroid plexus (brain cells that produce CSF) and in neocortical neurons. In support of these findings, it was shown in studies using organoids and transgenic mice expressing the human ACE2 gene that the COVID-19 virus can infect neurons and cause cell death in an ACE2-dependent manner. Since one of the common neurological symptoms associated with COVID-19 is a loss of smell and taste, the olfactory system may be a possible route for the virus to enter the brain. Indeed, studies have shown that olfactory epithelial cells had detectable levels of ACE2 protein and RNA, although it is unknown whether the virus could reach the olfactory neurons to enter the brain. Additionally, since COVID-19 has been described as being blood-borne, it could potentially enter the brain via the blood-brain-barrier or infected immune cells. Thus, there are many possible ways in which the COVID-19 virus could invade the brain, however, more conclusive data from CSF and post-mortem tissue is required to determine the exact route of entry. 

What’s the bottom line?

The evidence to date suggests that the COVID-19 virus can result in acute and prolonged neurological abnormalities that range in level of severity. Most of this data, however, is derived from case studies, which means that the acute and long-term neurological effects of COVID-19 at the population level are still unclear. The development of adequate experimental models as well as the collaboration between clinical and basic scientists around the world will be critical for answering these questions quickly and efficiently.

Iadecola et al. Effects of COVID-19 on the nervous system. Cell (2020). Access the original scientific publication here.

Coactivating Large Neuronal Populations Enhances Cortical Excitability

Post by Elisa Guma

What's the science?

Our brain’s ability to structurally and functionally remodel in response to sensory inputs is critical to our survival. Repetitive exposure to the same sensory stimulus can induce long-term changes in underlying neural circuits, which in turn can alter future responses to the same stimulus. Importantly, exposure to sensory stimuli typically coactivates large groups of cortical neurons, leading to correlated neuronal activity and increased efficacy of synaptic transmission and neuronal excitability. This week in PNAS, Zhang and colleagues used optogenetic methods to directly activate populations of cortical neurons in the mouse cortex to determine the mechanisms underlying enhanced neuronal excitability in response to repetitive stimulus exposure. 

How did they do it?

The authors injected the adult mouse primary visual cortex with adeno-associated virus vector that would transfect pyramidal neurons with channelrhodopsin 2 (ChR2), a blue light sensitive channel, and implanted them with multi-unit electrodes in layers 2 and 3 of the visual cortex. With this set up, the authors were able to stimulate the visual cortex (using a laser to activate ChR 2) and measure neuronal activity in discrete response areas (RAs) at each electrode in anesthetized mice.

In their first experiment, the authors examined whether repetitive synchronous coactivation (compared to random stimulation) of multiple neuronal populations in 4 identified RAs of the visual cortex could alter the spiking rate of the neurons in these areas in response to a test (light) stimulus. The authors characterized the duration of the enhancement of evoked spiking following neuronal coactivation, as well as the dependence of the enhancement of the number of coactivation stimuli, and the size of the neuronal population coactivated. The experiments were repeated in the motor cortex (to understand whether they were generalizable or specific to the visual cortex. To better understand how synaptic transmission might affect neuronal response to synchronous coactivation, the authors infused the visual cortex with an antagonist for the AMPA or NMDA subtype of glutamate receptors and measures stimulus-evoked synaptic activity using in vivo whole cell recording.

Next, the authors wanted to know if the enhancement of neuronal excitability could be achieved at a longer range. First, they examined cross-cortical enhancement between the visual cortex and somatosensory cortex, and then examined whether the cross-hemispheric spread of spiking enhancement could occur between the right and left visual cortices. The size of coactivated populations and distance between regions were considered to be modulating factors. Finally, for contralateral communication to occur between hemispheres, neuronal activity passes through subcortical regions. To understand the underlying mechanisms of this pathway, the authors also chemogenetically inhibited neuronal activity in the midline/intralaminar thalamus, as there is evidence for it being connected to the visual cortex and measured whether spiking enhancement between hemispheres could still be achieved.

What did they find?

First, the authors confirmed that repetitive activation of neuronal populations caused enhancement of neuronal spiking rates in all RAs recorded, which was not observed if the sequence was random, suggesting that the synchronous coactivation of cortical neurons is required for the observed enhancement of neuronal firing. The enhancement induced by synchronous coactivation was long-lasting (persisted up to 100 minutes following the coactivation), saturable (plateaued after 3 episodes of coactivation), and dependent on the size of the coactivation area (smaller area only show enhancement of closer RAs). Similar coactivation was observed in the motor cortex suggesting that enhancement is a general property of the cerebral cortex. AMPA receptor blockade led to a reduction or elimination of spiking of neurons, whereas NMDA receptor blockade abolished enhanced test stimulus-evoked spiking highlighting the critical role these glutamate receptor subunits play in inducing neuronal excitability. Many of the features observed thus far are reminiscent of long-term potentiation.

To achieve cross-cortical neuronal enhancement, the size of coactivated neuronal populations needed to be increased; no enhancement was observed if only one region within the visual and somatosensory cortices were stimulated, however, there was a significant elevation of test-stimulus-evoked spiking if two visual RAs and one somatosensory RA (or vice versa) were coactivated. Finally, the authors found that repeatedly coactivating four RAs in the left visual cortex led to enhanced neuronal spiking in both hemispheres. This was markedly reduced by silencing the midline/intralaminar thalamus, suggesting that it plays a critical role in mediating contralateral coactivation.

What's the impact?

This study shows that repetitive coactivation of large populations of cortical neurons results in persistent enhancement of neuronal spiking evoked by optogenetic stimulation. This effect was dependent on NMDA receptor activity and could be observed in local cortical areas, distance cortical areas, and even in different hemispheres. Finally, neuronal activity in the thalamus may be involved in the enhancement of cortical excitability. This work elucidates some of the mechanisms underlying cortical excitability and may help in understanding disorders in which cortical excitability is aberrant, such as epilepsy.

Zhang et al. Global enhancement of cortical excitability following coactivation of large neuronal populations. PNAS (2020). Access the original scientific publication here

Cortical-Like Dynamics Emerge as a Result of Sampling-Based Inference

Post by Cody Walters

What’s the science?

The brain’s cortical dynamics are characterized by a variety of motifs such as stimulus-dependent oscillations, transients, and changes in variability. The functional role of these phenomena are unclear, and to date there is no unifying framework to account for these distinct features of cortical activity. This week in Nature Neuroscience, Echeveste et al. provide evidence that these dynamics might emerge as a result of how the brain represents uncertainty.

How did they do it?

The authors optimized a recurrent neural network (a form of machine learning) with biological constraints to perform sampling-based probabilistic inference. In order to train the network, they employed a known generative model (a Gaussian scale mixture model) to capture the statistics of small segments of natural images. A generative model is a probabilistic model of the world that describes how sensory observations were likely generated. Specifically, a well-calibrated generative model allows you to infer the state of latent (i.e., unobserved) variables that most likely caused a given stimulus. The authors used a generative model that describes each image patch as a weighted sum of gabor filters (which are commonly used to approximate the receptive fields found in visual cortex) with varying orientations and intensities. Using Bayes rule, they ‘inverted’ their generative model in order to probabilistically infer the configuration of latent variables (in this case, the set of gabor filter intensities) that were the most likely cause of a given image patch. This method is referred to as the ‘ideal observer’ because it performs the task optimally (from a Bayesian perspective). The distribution of inferred latent variable values (i.e., gabor filter intensities) given a specific stimulus (i.e., image segment) are known as the posterior in Bayesian modeling, and the weights in the recurrent network were trained in order to mimic the posterior distributions produced by the generative model. 

The architecture of the recurrent network mirrored the structure of the generative model such that each excitatory neuron in the network corresponded to a specific latent variable (i.e., gabor filter intensity). Thus, as the firing rate patterns between neurons evolved over time, they traced out regions in membrane potential space that reflected a probability distribution (with each point in that region of neural activity representing a ‘sample’ from the posterior). Under this assumption, the mean of the neural activity is related to the estimated peak of the posterior (i.e., the most likely configuration of latent variables given an image segment), while variability in the neural activity is related to the uncertainty of that estimate (i.e., the width of the posterior). 

What did they find?

The authors first confirmed that the mean, variance, and covariance of the posterior distributions obtained by the ideal observer for five different stimulus (i.e., gabor filter) contrast levels of varying orientation were closely matched by the distribution of network activity profiles, thus demonstrating that the network had been successfully optimized to perform sampling-based inference. In order to determine whether their network could generalize to stimuli it hadn’t been trained on, the authors tested its performance on novel image patches. Once again, they found that the network activity profiles were able to closely match the ideal observer posterior distributions. 

The authors then examined how the neural responses in the optimized network compared to actual neural data obtained from macaques and mice on a similar visual task. Interestingly, they found that the network replicated four fundamental properties of cortical dynamics: (1) the tuning curves of individual neurons in the network closely matched experimental data, (2) network activity exhibited variability quenching (i.e., a reduction in firing rate variability at stimulus onset), (3) the network exhibited gamma oscillations, and the peak of those gamma oscillations shifted to higher frequencies as stimulus contrast levels increased, and (4) the network displayed stimulus-dependent inhibitory transients that scaled with stimulus contrast. Furthermore, they found that these cortical-like features of the network disappeared when it was only trained to match the mean (and not the variance and covariance) of the ideal observer posterior distributions. This suggests that the cortical-like dynamics they observed were not an inevitable consequence of the structure of the network, but rather that these dynamical features arose due to the network being optimized for a specific computational objective (i.e., to match the mean, variance, and covariance of the ideal observer posterior distributions). 

Next, the authors examined the potential functional role of oscillations in the network, finding that gamma oscillations allowed the network to more rapidly converge with the target distribution that it was attempting to match through sampling. Indeed, gamma oscillations appeared to allow the network to more quickly traverse terrain in neural state space (and thus get a more accurate estimate of the target distribution) as evidenced by the fact that the first principle component of the neural activity contained much of the gamma power. This result suggests that the network uses gamma oscillations to speed-up neural state space sampling in the direction that captures most of the variance of the target distribution. 

Finally, the authors explored the potential functional role of transients. They found that overshoots seen at stimulus onset (when the network has to suddenly switch from representing one target mean to a different target mean) allowed neurons in the network to more accurately approximate the new target distribution (as compared to approaching the new target mean exponentially). The reason for this is that the transient overshoots seen during the first ~100 milliseconds after stimulus onset average out to match the target mean, thus conferring the ability to perform more rapid inference. This form of averaging-out made specific predictions about overshoots, namely (1) that they should scale with the magnitude of the difference between the pre-stimulus onset mean and the post-stimulus onset mean, and (2) that they should be orientation-tuned (i.e., larger for a given neuron’s preferred stimulus orientation). The authors then provided experimental confirmation for both of these predictions through a novel analysis of awake macaque data.

What’s the impact?

Echeveste et al. showed that a recurrent network with biological constraints exhibited various dynamical features commonly seen in cortical structures such as gamma oscillation peak-shifting, variability quenching, and transients. Importantly, none of these dynamics were directly encoded into the network; rather, they emerged indirectly as a consequence of the computational goal the network was optimized to perform: sampling-based inference. These results highlight sampling-based inference as a candidate unifying framework that parsimoniously explains a suite of dynamical features exhibited by sensory cortices. 

Cortical-like dynamics in recurrent circuits optimized for sampling-based probabilistic inference, (2020). Access the publication here.