Synchronizing Brain Circuits Restores Working Memory in Older Adults

Post by Amanda McFarlan

What's the science?

Working memory is a type of short-term memory (important for immediate processing and integrating of an individual’s surroundings) that is known to decline with age. Theories of aging propose that decline in cognitive processes like working memory may be a result of desynchronization between cortical areas in the brain. Previous studies have shown that synchronization in the temporal cortex occurs during working memory tasks. This week in the Nature Neuroscience, Reinhart and colleagues investigated the role of temporal synchronization on working memory performance in aging.

How did they do it?

The authors recruited a total of 84 male and female participants (42 younger adults aged 20-29 years old and 42 older adults aged 60-76 years old) for their study. They used an experimental paradigm where the older adults participated in both the experimental and control conditions, while the younger adults only participated in the control condition. In the experimental condition, electroencephalography (EEG) activity and task performance levels were recorded during and after the administration of frontotemporal in-phase theta-tuned high-definition transcranial alternating-current stimulation (HD-tACS). In the control condition, the participants’ EEG activity and task performance levels were recorded during and after the administration of sham HD-tACS. Participants performed a working memory task and a control task while receiving HD-tACS. During the working memory task, participants were presented with an image of a real-world object that was followed by a delay. Then, they were presented with an image of another real-world object and had to determine whether this object was the same or different from the image they had previously been shown. For the control task, participants were presented with a real-world object, followed by a delay. Then, they were presented with a grated stimulus (see Figure) and they had to determine if the grating was titled clockwise or anti-clockwise. The participants alternated between the two tasks (working memory and control) 10 times during the administration of HD-tACS (total of 25 minutes) and 20 times in the post-stimulation period (total of 50 minutes).

What did they find?

The authors found that the older adults in the control condition were slower and less accurate when performing the working memory task compared to the younger adults, suggesting that older adults display deficits in working memory performance. They showed that younger adults displayed increased memory-specific coupling of theta-gamma rhythms in the left temporal cortex, while older adults showed no evidence of brain activity coupling. These findings suggest that coupling of theta-gamma rhythms in the temporal cortex are important for working memory and may be predictive of behavioural success. Next, they determined that younger adults, but not older adults, had significantly increased phase synchronization between the prefrontal cortex and left temporal cortex during the working memory task compared to the control task. However, they determined that there were no differences in synchronization between the temporal and occipital cortices between older and younger adults, suggesting that short-range communication between nearby sensory cortices remains intact in older adults, while long-range communication becomes less efficient. Next, the authors examined the effects of HD-tACS stimulation on working memory performance in older adults. They found that HD-tACS stimulation improved working memory performance in older adults to levels that were comparable to younger adults and that these effects were long-lasting. Additionally, HD-tACS stimulation increased theta-gamma coupling as well as increased theta-phased synchronization between the prefrontal and left temporal cortices during the working memory task, similar to younger adults. Together, these findings suggest that HD-tACS stimulation is sufficient to induce temporal synchronization in brain activity that improves working memory performance.

What's the impact?

This is the first study to show that cognitive decline may be a result of desynchronization between long-range frontotemporal connections. Importantly, the authors showed that stimulation with high-definition transcranial alternating-current is sufficient to improve deficits in working memory in older adults to levels that are indistinguishable from younger adults. Altogether, these findings highlight a non-invasive, non-pharmacological intervention that may be useful for treating and improving cognitive decline in aging or clinical populations.

Reinhart and Nguyen. Working memory revived in older adults by synchronizing rhythmic brain circuits. Nature Neuroscience (2019). Access the original scientific publication here.

Hippocampus-Amygdala Communication Supports Pattern Separation of Emotional Memories

Post by Shireen Parimoo

What's the science?

Pattern separation is a hippocampal-dependent process that allows us to discriminate between similar events or experiences - like remembering where we parked the car at work on a specific day. Emotional events are remembered vividly because they increase arousal, which enhances memory formation and retrieval. The strength of emotional memories is modulated by the amygdala. Memories for emotional events are often biased to central aspects of the event while peripheral details are forgotten, making it difficult to discriminate between similar emotional events. The precise neural mechanism supporting this process remains unclear. This week in Neuron, Zheng and colleagues investigated the neural mechanism underlying the pattern separation of emotional memories in the amygdala and hippocampus using intracranial recordings.

How did they do it?

Seven pre-surgical epilepsy patients completed an episodic memory task with emotional stimuli while local field potentials were recorded from the amygdala and the hippocampus using intracranial stereo-electroencephalography. The memory task consisted of two phases: (i) an encoding phase, in which participants had to rate the emotional valence of positive, negative, and neutral images, and (ii) a retrieval phase, in which they had to make old/new judgments about the same, novel, or perceptually similar (‘lure’) images. The lures were presented to create interference between previously presented target images and the novel lures. To examine pattern separation, the authors computed the oscillatory power (squared amplitude), phase synchrony (phase coupling strength of signals in different regions of the brain), and phase-amplitude coupling (coupling between the phase of one signal and the amplitude of another signal) in different frequency bands. To determine whether hippocampal oscillations were driving oscillatory activity in the amygdala or vice versa, they computed the phase-slope index and performed a Granger causality analysis. Finally, they trained a classifier using the occurring phases of high-frequency activity (HFA) relative to the theta (2-7 Hz) and alpha bands (7-14 Hz) as inputs to predict whether participants would false alarm or correctly reject the lures.

What did they find?

Participants performed well on the memory task with 80% accuracy overall. However, they were worse at discriminating emotional lures than neutral lures, with poorer discrimination of negatively-valenced lures than positive lures. Oscillatory activity was observed in the theta and alpha frequency bands when participants made recognition judgments. Specifically, in the hippocampus there was greater theta activity on trials where participants correctly identified the lure images as being new, which was accompanied by a simultaneous reduction in alpha power. In the amygdala, on the other hand, there was a lag between increased theta activity and a subsequent reduction in alpha power. Alpha activity increased on false alarm trials where participants incorrectly categorized images as being old. Oscillatory activity in the amygdala and the hippocampus was synchronized in the theta band on trials where the participant correctly identified a lure and in the alpha band on false alarm trials. This pattern of activity was present for both emotional and neutral images, but stronger for emotional images  only on lure trials. This suggests that theta and alpha oscillations in the amygdala and hippocampus are involved in pattern separation, which is modulated by emotional valence.

Bidirectional communication between the hippocampus and the amygdala in the theta band was observed on correctly identified lure trials, and unidirectional communication was observed in the alpha band from the amygdala to the hippocampus on lure false alarm trials. Similarly, there was greater phase-amplitude coupling between theta oscillations in the amygdala and HFA in the hippocampus — as well as the reverse — on correctly identified lure trials. There was also a modulation of hippocampal HFA by alpha oscillations in the amygdala on lure false alarm trials. The trough of the theta cycle was associated with HFA in the amygdala (descending slope) and the hippocampus (ascending slope) on correctly identified lure trials. Such theta phase-dependent modulation of HFA was not present on lure false alarm trials. Moreover, the influence of amygdala theta oscillations on the hippocampus was stronger with emotional stimuli than neutral images, especially for negatively-valenced images. However, this emotional modulation of directional connectivity was not observed in the reverse direction (hippocampus to amygdala) or on lure false alarm trials. Thus, bidirectional communication between the amygdala and the hippocampus in the theta band results in the correct discrimination of lures from targets, whereas a disproportionately amygdala-driven interaction in the alpha band results in poor discrimination.

What's the impact?

This study is the first to show that bidirectional theta oscillations between the hippocampus and the amygdala at memory retrieval are necessary for the pattern separation of emotional events. These findings are consistent with the literature demonstrating the role of low-frequency oscillations in memory formation and provide further insight into the mechanism that allows for the resolution of interference between highly similar emotional events at retrieval.

Zheng et al. Multiplexing of theta and alpha rhythms in the amygdala-hippocampal circuit supports pattern separation of emotional information. Neuron (2019). Access the original scientific publication here.

New Gene Expression Technologies Improve Understanding of ALS

Post by Flora Moujaes

What's the science?

Amyotrophic lateral sclerosis (ALS) is a type of motor neuron disease. Motor neuron diseases are caused by the gradual deterioration of motor neurons found in the brain and spinal cord. As the motor neurons die, they can no longer control voluntary movement and the muscles begin to atrophy. The cause of the disease and the mechanisms that underlie ALS are still poorly understood, though previous work has indicated genetic factors play a role. This week in Science, Maniatis and colleagues combined new technologies for mapping gene expression that take into account spatial location (where the genes are expressed), and computational modelling, to provide new insights into the mechanisms that contribute to ALS.

How did they do it?

Researchers can determine whether a gene is activated or ‘expressed’ in a cell by examining messenger RNA (mRNA). mRNA is essentially a template for DNA: it is a set of molecules copied from the DNA code that tells the cell how to behave. There are thousands of RNA molecules per cell, and together they make up the mRNA, or ‘transcriptome’. Thus, the transcriptome reflects the genes that are being actively expressed in a cell at any given time. Standard methods to examine mRNA involve extracting all the RNA molecules from a tissue biopsy, and analysing them together. This results in an average representation of all the genes expressed in that tissue, however, spatial information of where the particular genes are expressed within the tissue is lost. Spatial transcriptomics enables you to capture the mRNA in the tissue while retaining the information about where in the tissue this mRNA is expressed. The method involves placing tissue samples on a glass slide covered in tiny spots. These tiny spots contain DNA strands with built in address labels that 1) capture and copy the mRNA and 2) encode a unique ‘barcode’ in the copy that corresponds to the spatial location of the spot on the slide.

The researchers used spatial transcriptomics to examine gene expression in 1) 56 ALS mice, 2) 11 control mice, and 3) post-mortem spinal cord tissue from seven deceased ALS patients. They collected over 76,000 spatial gene expression measurements (or ‘spots’) from 1,165 mouse tissue sections, and over 60,000 spatial gene expression measurements from 80 human tissue sections. The mice were examined at three disease time points: onset, symptomatic, and end-stage, which enabled the researchers to track gene expression over time. The human samples were collected from either end of the spine, which allowed the researchers to examine how ALS pathology spreads, as ALS symptoms usually appear in one part of the body before going on to cause widespread paralysis. The researchers then used computational methods to combine the spatial location and gene expression information.

What did they find?

The researchers produced a multidimensional gene expression atlas which they made available through an interactive data exploration portal at https://als-st.nygenome.org/ .

One of the main hypotheses explored in this study was that even though motor neurons are the most vulnerable in ALS, neighbouring cells such as microglial cells play a key role in the disease. Microglial cells are of particular interest to ALS as they are involved in removing damaged neurons. This study indicated that microglia dysfunction occurs well before the onset of ALS symptoms. This is important as it helps us understand how mutations may disrupt the function of both neuronal and non-neuronal cells, and how the impaired interactions between the different cell types in the nervous system may then lead to motor neuron loss in ALS. To further explore the spatiotemporal dynamics of microglial activation, the researchers examined two genes that had previously been indicated in ALS: TREM2 and TYROBP. They found that TREM2 and TYROBP were both expressed at higher levels by microglial cells; in particular spinal cord regions of mice with ALS symptoms. Thus, they were able to refine our understanding of such gene expressions, showing TREM2- and TYROBP-mediated signalling is an early step in disease-relevant changes in microglial gene expression.

What's the impact?

This is the largest study to examine spatial transcriptomics in mice and humans with ALS. This study highlights the value of examining gene expression in individual cell types in relation to their wider spatial context. The vast dataset generated by this study may aid with the development of therapeutic interventions for ALS and earlier diagnostic markers. This is especially important given that ALS currently affects over 200,000 patients worldwide and results in a life expectancy of up to 5 years following diagnosis.  In addition, the methods used in this study could be extended to examine other neurodegenerative diseases including Alzheimer’s, Parkinson’s, and Huntington’s disease.

Maniatis et al. Spatiotemporal dynamics of molecular pathology in amyotrophic lateral sclerosis. Science (2019). Access the original scientific publication here.