Anesthesia Lowers Spatial Frequency Preference in the Primary Visual Cortex

Reviewer #1 (Public review):

This manuscript characterizes the effects of isoflurane on visual processing in layer 2/3 of the mouse primary visual cortex (V1). General anesthesia, including isoflurane, has been reported to modulate various neural processes, such as size tuning, direction selectivity, and spatial selectivity in V1. Using two-photon calcium imaging, the authors monitored neural responses to visual stimuli under isoflurane anaesthesia and found that spatial frequency preferences are also affected across cell types, with the magnitude and direction of these effects varying between cell types.

The authors performed careful and rigorous comparisons of neuronal responses between the two conditions using well-chosen nonparametric statistics. At the same time, because two-photon calcium imaging can be combined with cell-type-specific labeling, the authors labelled inhibitory neurons with tdTomato, allowing them to distinguish GCaMP activities in excitatory and specific inhibitory cell classes. We also appreciated that the manuscript provides not only summary statistics but also example GCaMP traces (Figure 1), which makes it easy for readers to understand the quality of the raw data.

We believe that the manuscript could be improved by emphasizing the following three points.

(1) The analyses are limited to the neurons that responded to visual stimuli in both the anesthetized and awake states. According to Table S1, the proportion of visually responsive neurons that met such criteria is only 27.4% for the excitatory neurons. This raises the potential concerns that the reported effects of isoflurane may not fully reflect population-level changes in visual coding. We suggest that the authors repeat the same analyses, including average tuning curves and decoding analyses, for all recorded neurons in each condition.

(2) The manuscript would benefit from tuning curves of spatial frequency preference for individual neurons, as this would help readers assess whether the reported statistics are appropriate (Figures 2A-D). In addition, more in-depth single-neuron analyses would help distinguish between the two proposed hypotheses in Figure 5 that may not be evident from average responses alone. This is because, with the current analysis, it is not clear how the shape of the tuning curves will affect the estimation of spatial frequency preference. To address this potential concern and strengthen the interpretation of the results, we suggest:
a) repeating the analysis at the level of individual neuronal responses, instead of average responses, and
b) using simulated data to examine how changes in tuning-curve width could affect estimated spatial frequency preference.

For example, using the neuronal responses in the awake condition, one could broaden the tuning curves and recompute the preferred spatial frequency, then compare the resulting distribution with that observed under anesthesia.

(3) We believe the manuscript's overall framing is a little broader than what is directly supported by the data. In particular:

(a) the statement "reduced sensory perception during anesthesia is linked to a degradation in spatial resolution at the cellular level" in the Abstract is an unclear and unsupported claim. We suggest removing this sentence and more directly summarizing the findings.

(b) given the discrepancy between the effects of urethane and isoflurane as laid out in the discussion, the current title "Anesthesia Lowers Spatial Frequency Preference in the Primary Visual Cortex" appears overstated and should be revised to explicitly reflect the specific anesthetic tested: "Isoflurane Anesthesia Lowers Spatial Frequency Preference in the Primary Visual Cortex".

Reviewer #2 (Public review):

Summary:

The main objective of the study was to link the changes in brain state due to anesthesia to consequences on visual neural processing, particularly effects on spatial frequency tuning. This is accomplished by 2-photon imaging of excitatory and inhibitory neurons (separating PV- and SST-positive subtypes) in mouse visual cortex during full-field visual stimulation with gratings, and tracking neuronal tuning for spatial frequency before, during, and after isoflurane anesthesia. The main finding is that anesthesia induces lower spatial frequency preferences in excitatory neurons, and this leads to poorer population representations (decoding) of higher spatial frequency responses during anesthesia. A second main finding is that anesthesia impacts inhibitory neuron subtypes in distinct ways, with the most pronounced effects of anesthesia on somatostatin inhibitory neurons.

Strengths:

(1) A main strength is that the study is that it is straightforward, and reassuringly, the results confirm multiple previous studies showing anesthesia's effects on the amplitude of cortical responses: larger and less selective responses in excitatory neurons (versus awake responses); strongly reduced responses in somatostatin inhibitory neurons (versus awake responses) (Fig. 5I-L), with less differences across anesthetized and awake states on response amplitude of PV neurons.

(2) These confirmations of prior observations (on the amplitude of responses) establish good ground for the new results on spatial frequency tuning. For excitatory neurons, spatial frequency selectivity shifts to higher values in awake versus anesthetized conditions; this is because anesthesia induces larger responses to lower spatial frequencies. In somatostatin neurons, instead, wakefulness reduces the lower spatial frequency responses present in anesthesia, and dramatically increases the overall amplitude of responses and medium and higher spatial frequencies. This is consistent with prior work showing that in awake states, somatostatin neurons exert broad inhibition in V1; this study extends that finding to the tuning of spatial frequencies.

Weaknesses:

(1) A first weakness of the study is the lack of examination of changes to single neuron receptive field sizes and/or surround suppression across conditions, and how these may relate to the effects on spatial frequency tuning with full field gratings. There is a well-known relationship between the size of the receptive field and the resulting selectivity for spatial frequencies (i.e., large receptive fields prefer lower spatial frequency stimuli). Likewise, there are many studies showing how surround suppression / spatial integration is impacted by anesthesia (and arousal). A more detailed examination of all these related quantities on an individual neuron basis would provide a greater understanding of the factors underlying the effects on spatial frequency tuning. One could imagine that receptive field changes, and/or changes in surround suppression, influence the selectivity to full-field gratings.

(2) A second weakness is the lack of examination/insight into the temporal dynamics of the effects. The experimental paradigm records activity across control, anesthesia, and recovery epochs in a single duration (~40 mins) session. The epochs are simply binned together ("Awake", "Anes.", "Recover"). It is not clear how the start of the anesthesia bin is defined, nor is it clear how the recovery period is defined. It is also not clear what the changes are to motor tone, brain state, etc., that are also strong influences on visual responses in mouse V1. Presumably, these onset/offset effects are similar enough across mice and sessions that they affect all the bins in the same way, but greater examination of the temporal effects in excitatory, PV, and SOM neurons could shed light on interactions driving the changes. Is there some temporal dependence of anesthesia on selectivity changes across the cell types? For example, at the onset of anesthesia, are SOM neurons losing broadband frequency responses before the excitatory neurons gain low frequency responses? Do PV neurons also show effects after the changes in SOM neurons (suggesting strong SOM -> PV inhibition)? Such analysis might shed light on the timing/causality of the effects among these 3 neuron types.

(3) A third weakness concerns the interpretation of the low and high arousal conditions during awake states (Figure 6). It is not clear how movement (or lack of movement) impacts the high arousal epochs, nor is it clear how the low arousal condition compares to the brain state during anesthesia. For example, deep versus light anesthesia can lead to synchronized or asynchronous states, respectively, and low arousal in wakefulness can show strong low-frequency oscillations of activity, which could promote a lower excitability state than light anesthesia. Without some more detail about commonly measured brain state or body/face motion metrics, it is difficult to know what brain states are represented by the bins and how to interpret the comparisons.

Overall, the study uses adequate methods and experimental design to demonstrate solid support for the (somewhat narrow) central finding that anesthesia lowers the spatial resolution of mouse V1 responses.

Since this is a very well-examined topic, the findings here are not totally surprising, but confirmatory and slightly extend prior findings (a good thing). As such, the study will likely have most relevance to specialists in the mouse visual system, but if the study could address some of the remaining questions discussed above, this would potentially broaden the implications of the study to general insights about the operation of cortical circuitry.

Reviewer #3 (Public review):

Summary:

This manuscript is focused on studying the spatial frequency selectivity of individual neurons in the mouse primary visual cortex (V1) in the anesthetized and awake brain states using 2-photon calcium imaging. Although previous studies have demonstrated that anesthesia decreases both size tuning and spatial selectivity in V1 neurons, the strength of this study is its focus on characterization of the same neurons in awake and anesthetized states in combination with transgenic mouse lines selectively labeling pan-inhibitory neurons and also more specific neuronal subtypes, including parvalbumin-positive (PV+) or somatostatin-positive (SOM+) interneurons. A combination of these methodologies allows for a more in-depth mechanistic study of the properties of different types of neurons. The main findings suggest that in excitatory neurons, anesthesia leads to a shift in preferred SF and broadening of SF tuning, with no changes in orientation and direction selectivity. Downward shift in preferred SF was more pronounced in both SOM+ and PV+ interneurons.

Strengths:

(1) 2-photon calcium imaging with single-cell resolution.

(2) Characterization of excitatory and two types of inhibitory neurons.

Weaknesses:

(1) VIP interneurons are critical to the neural circuit, and their characterization would be critical to the mechanistic understanding of this process, but is missing.

(2) Unfortunately, the manuscript does not lead to an additional insight into the nature of this anesthesia-induced shift in SF preference.

(3) Furthermore, it also doesn't help understand how SF preference is encoded in V1.

(4) Finally, some critical histological controls are missing.

Author response:

Thank you for the eLife assessment and the constructive reviews. We appreciate the reviewers’ valuable insights and the time they dedicated to providing such thoughtful feedback on our manuscript. The reviewers highlighted the technical rigor of our study, specifically the tracking of individual neurons across both anesthetized and awake states using two-photon imaging. They also emphasized the importance of our cell-type-specific analysis (excitatory, PV, and SOM neurons) and noted that the study provides solid evidence for isoflurane-induced shifts in preferred spatial frequency (SF).

Based on our team's evaluation of the reviewers' comments, we would like to outline our planned revisions.

(1) Expanded Population and Single-Neuron Analysis

We will re-analyze our dataset to include all neurons that were responsive under anesthesia, in the awake state, or both. This will ensure our findings accurately represent the entire population of visually responsive neurons. We will also provide examples of individual tuning curves to clarify the relationship between tuning shape and SF shifts in individual neurons.

(2) Addressing Methodological Scope and Behavioral Metrics

Receptive Field Size and Dynamics: While we did not utilize a stimulus set specifically designed to map receptive field (RF) sizes, we intend to examine how other functional parameters co-varied with the shift in preferred SF within each cell type. Furthermore, although characterizing the precise temporal dynamics during anesthesia onset presents technical challenges, we will attempt to analyze the time-dependence of the observed changes to provide deeper insight into the transition between states.

Behavioral Metrics: While pupil size is a well-established proxy for brain state, we will explore the inclusion of other available behavioral parameters.

(3) Cell-type Specificity (SOM, PV, and VIP)

SOM vs. PV Comparison: We will perform a detailed comparison of preferred SFs between SOM and PV interneurons, including those responsive only under anesthesia or only in the awake state.

VIP Neurons: While VIP neurons are known to play critical roles in cortical circuits, such as disinhibition, we have decided not to conduct new recordings for VIP interneurons in the present study. Based on existing literature, the proportion of visually responsive VIP cells is too low to yield statistically reliable conclusions for this specific study (de Vries et al., Nature Neuroscience 23, 138-151, 2020). Additionally, we intend to focus our analysis on inhibitory interneuron subtypes that provide direct input to pyramidal cells.

Histology: We will provide additional histological validation.

(4) Refined Framing

As suggested, we will focus the manuscript strictly on isoflurane anesthesia. This includes updating the title and abstract to reflect this specificity and discussing how our results compare with other anesthetics like urethane. Furthermore, we will substantially deepen our discussion on the potential mechanisms by which anesthesia induces a downward shift in preferred spatial frequency.

We believe these additions will significantly strengthen the manuscript.