Prefrontal PV interneurons facilitate attention and are linked to attentional dysfunction in a mouse model of absence epilepsy

  1. Brielle Ferguson
  2. Cameron Glick
  3. John R Huguenard

  • Curated by eLife

    eLife logo

    Evaluation Summary:

    The manuscript by Ferguson and Huguenard examined the mechanism underlying attentional deficits in the haploinsufficient Scn8a mouse model of absence epilepsy. Using a detection behavior paradigm where detection of the cue was dependent on the attention level, they show that Scn8a+/- mice perform worse than controls when cues are of intermediate duration. In the medial prefrontal cortex (mPFC), high gamma-band power, which is presumed to be mediated by activity in parvalbumin-positive (PV+) interneurons, was reduced in Scn8a+/- mice, but optogenetic stimulation of these neurons at low gamma-frequency normalized performance on the cue-based attention task. The results of this study identify mPFC PV dysfunction, rather than seizure activity, as a potentially important cellular substrate, for attention during cue-presentation.

    (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1 agreed to share their name with the authors.)

Reviewed by

Read the full article See related articles

Listed in

Log in to save this article

Abstract

Absence seizures are characterized by brief periods of unconsciousness accompanied by lapses in motor function that can occur hundreds of times throughout the day. Outside of these frequent moments of unconsciousness, approximately a third of people living with the disorder experience treatment-resistant attention impairments. Convergent evidence suggests prefrontal cortex (PFC) dysfunction may underlie attention impairments in affected patients. To examine this, we use a combination of slice physiology, fiber photometry, electrocorticography (ECoG), optogenetics, and behavior in the Scn8a +/- mouse model of absence epilepsy. Attention function was measured using a novel visual attention task where a light cue that varied in duration predicted the location of a food reward. In Scn8a +/- mice, we find altered parvalbumin interneuron (PVIN) output in the medial PFC (mPFC) in vitro and PVIN hypoactivity along with reductions in gamma power during cue presentation in vivo. This was associated with poorer attention performance in Scn8a +/- mice that could be rescued by gamma-frequency optogenetic stimulation of PVINs. This highlights cue-related PVIN activity as an important mechanism for attention and suggests PVINs may represent a therapeutic target for cognitive comorbidities in absence epilepsy.

Article activity feed

  1. Version published to 10.7554/elife.78349 on eLife
  2. eLife

    Evaluation Summary:

    The manuscript by Ferguson and Huguenard examined the mechanism underlying attentional deficits in the haploinsufficient Scn8a mouse model of absence epilepsy. Using a detection behavior paradigm where detection of the cue was dependent on the attention level, they show that Scn8a+/- mice perform worse than controls when cues are of intermediate duration. In the medial prefrontal cortex (mPFC), high gamma-band power, which is presumed to be mediated by activity in parvalbumin-positive (PV+) interneurons, was reduced in Scn8a+/- mice, but optogenetic stimulation of these neurons at low gamma-frequency normalized performance on the cue-based attention task. The results of this study identify mPFC PV dysfunction, rather than seizure activity, as a potentially important cellular substrate, for attention during cue-presentation.

    (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1 agreed to share their name with the authors.)

  3. eLife

    Reviewer #1 (Public Review):

    Overall the main conclusions of the papers are supported and justified by the author's data with one potential weakness on the gamma-power calculations.

    Strengths
    1. The authors adopted a novel cue-based detection behavior paradigm where animal detection of cue is dependent on the attention level. This task is essentially crucial to the author's claim that Scn8a heterozygotes have attention deficits and subsequent tests to unravel the mechanisms.
    2. The authors then take advantage of these validated tasks and carried out a series of optogenetic 'gain and loss of function' experiments and simultaneous neural recordings to understand the mechanisms involved.
    3. The authors rigorously examined specific reliance of PV neuron activity during attention related cue presentation, they used machine learning based algorithms to mine their data and their trained classifier can correctly identify the correct trial with 92% precision.
    4. Very interestingly, optogenetic activation of PV neurons at gamma-frequency increased the animal's ability on the cue-based attention task, together with their earlier results that gamma-band power was reduced in SCN8A heterozygotes, and the well-documented involvement of PV neurons in the gamma-oscillations, this experiment supports the author's claim that PV cells activities are important cellular substrate for attention during the cue-presentation.

    Weaknesses
    The main weakness has to do with the claim of reduced Gamma powers during AET (Fig 6). 1) The results described a 'rel. power' in each of the panels (Fig 6A-F), However, I could not find a description of how the 'rel. power' is calculated. In the methods section (Line 546 to 549), it was described as the theta or beta band was extracted. I don't know what extracted means, please clarify. 2) Additionally, only the high-gamma-band (i.e. 60-90Hz) showed significant reduction in the Int and Long cues, so maybe the conclusion should be more specific on the high-gamma power? 3) there are appears to be two sets of statistics for the power calculation (Fig 6 legend), and one P value for the long is larger than 0.05 (0.0597 ), I am confused with these p values and I hope the authors can clarify it for me. 4) The representative traces presented in panel B of Fig 6 appears to show that the amplitude of high -gamma, but not low-gamma was reduced during the cue presentation phase compared with the pre-cue phase, is this correct?

  4. eLife

    Reviewer #2 (Public Review):

    In this manuscript, Ferguson and Huguenard use Scn8a+/- mice to investigate the mechanism underlying attentional deficits seen in absence epilepsy. They report that Scn8a+/- mice perform worse than controls when informational cues about the location of rewards are of intermediate duration. Interestingly, this attentional deficit seems unrelated to acute seizure activity because a) seizure activity in the S1 was lower during the task than in home cage and did not differ with performance and b) valproic acid reduced seizure activity but did not improve attentional performance. Stimulating mPFC GABA neurons throughout the cue in control mice also disrupts performance (across all cue lengths), indicating that this task depends on the mPFC. Moreover, long and intermediate cues increased calcium activity in PV interneurons on correct trials, and this feature was attenuated in Scn8a+/- mice. Similarly, mPFC gamma power (thought to be mediated by PV interneurons) was reduced during intermediate cues in the Scn8a+/- mice. Finally, optogenetic stimulation of PV neurons at 40 Hz improved the performance of Scn8a+/- mice to the intermediate cues.

    Overall, I think this well-written manuscript makes a compelling argument that mPFC PV dysfunction contributes to attentional deficits in Scn8a+/- mice, which is of translational interest. While I think the overall contours of the argument make sense, some of the details of the results and analyses need clarification.

    1. From the presented data, the link between cue-evoked PV activity and disrupted Scn8a+/- attentional performance is unclear. This lack of clarity may stem from the way data is presented. What one would have expected the data to look like is an increase in PV activity (either in time to peak, peak, or average response) in controls that correlates with performance and then an attenuation in that parameter in the Scn8a+/- mice in the condition where they show attenuated performance. Instead, a) PV activity seems to increase even in conditions with low attentional demands, suggesting that cue-evoked PV activity is not necessarily linked to attention. b) Animals are collapsed across genotype to reveal the parameter that correlates with performance (time to peak and peak, but not average). This analysis is potentially circular since the animals that perform poorly tend to be the Scn8a+/- mice. It would seem from the text that a similar concern impacts the machine learning/ROC analyses as well. C) Oddly, only one of these parameters that correlates with performance (peak) is reduced in Scn8a+/- mice. Moreover, it is also reduced at cue length with low attentional demands, yet has no behavioral impact. D) Instead, one of the parameters that does not correlate with performance (amplitude) is attenuated in Scn8a+/- mice at the intermediate cue. These 4 discrepancies weaken the argument that cue-evoked PV activity necessary for attention in control mice is disrupted in Scn8a+/- mice.
    2. The authors show that high, but not low, gamma power is reduced in Scn8a+/- mice at the intermediate cue associated with poor performance. However, they then rescue with low gamma stimulation. Does this low gamma stimulation increase high gamma power?
    3. Is it possible that seizure activity in S1 and PFC are not synchronized? The authors have ecog data from PFC. What happens to seizures during the task there?
    4. Questions about the ChR2 stimulation: perhaps it is a typo, but the methods say that 0.5 mW power were used. This is well below the effective stimulation power for ChR2. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2995811/ Similarly, the authors use continuous stimulation for 5 seconds to disrupt mPFC activity. Is this stimulation effective for the entire period? ChR2 desensitizes with continuous stimulation. https://www.eneuro.org/content/7/1/ENEURO.0222-18.2019

  5. Version published to 10.1101/2022.03.17.484733v1 on bioRxiv