An Open-Source Restraint System for Magnetic Resonance Imaging in Awake Rats

Synthesis

Both reviewers were in agreement that this work is a valuable contribution to the field of awake rodent fMRI. Reviewers also commented positively on the practicality and utility of this head restraint method for eliminating anesthesia, especially for longitudinal studies. They had several suggestions for the manuscript, listed below, which would further improve clarity and reproducibility.

Reviewer 1

Major Comments

1. The habituation protocol is currently provided only in narrative form, which makes it difficult to follow step-by-step. Since the training schedule is critical for achieving stable awake imaging, I recommend adding a simple visual workflow diagram summarizing the day-by-day progression (e.g., exploration → body restraint → head fixation → mock scanner exposure) along with session durations. A timeline figure would make the protocol considerably easier for others to implement accurately.

2. Because the primary aim of the system is to enable stable awake scanning, it would be helpful to include a brief performance comparison with scans acquired under anesthesia using the same hardware and acquisition parameters. This does not require additional functional interpretation. A small dataset (e.g., n = 2-3) demonstrating differences in mean framewise displacement and susceptibility-related distortion / anatomical alignment would clearly show the practical advantage of the awake preparation. This comparison would strengthen the methodological contribution without expanding the scope of the study.

3. Motion traces are currently shown for a single animal. Providing a simple summary table with mean {plus minus} SD of motion metrics (e.g., FD, translation, rotation) across animals would help define the typical performance range of the system and improve reproducibility for labs adopting the method.

4. The manuscript describes the habituation schedule, but it does not specify how the authors determined when an animal was considered fully habituated. For reproducibility, it would be helpful to state the behavioral criteria used to define habituation completion.

Additionally, since corticosterone is commonly used in the awake fMRI field to assess long-term stress burden, the authors may consider reporting corticosterone levels across the habituation period. This would further support the usability and translational relevance of the method.

Minor Comments

1. It would be helpful to clarify that the neck plate rests at the shoulder girdle / thoracic inlet, rather than on the cervical airway. A brief clarification in the text would reduce potential misinterpretation of restraint positioning.

2. If data are available, providing an approximate proportion of animals completing the full habituation duration would assist readers in planning timeline and throughput expectations.

3. The functional connectivity matrix could be made easier to interpret by adding region labels or modular grouping, although this is not essential to the core contribution.

Since the gradient insert bore is relatively small, it would be helpful for the authors to briefly comment on the scope for integrating concurrent behavioral tasks during awake scanning. In particular, clarifying which types of head-fixed compatible behaviors are feasible within the current spatial constraints would help readers understand the potential for future behavioral extensions of the setup.

Reviewer 2

Major:

1. Several times throughout the manuscript, the authors mention the reduced stress by the introduced protocol. Although, it might be true that stress levels are reduced by the fact that anesthesia was not used, it has not been proved by showing any physiological stress indicators or by referring to any previous article showing the correlation between stress and anesthesia use. Therefore, I would encourage the authors to either downplay the claim of reduced stress (by the fact of not using anesthesia) or show any measured physiological markers or add references to show the link between anesthesia and reduced stress.

2. Habitutation schedule of 11 days was used. Was it based on previous literature or observed motion level of rats?

3. Based on the correlation matrix, very weak resting state correlation is observed. The authors state this in the discussion, but in my opinion should be discussed further where it arises. Typically, resting state correlations between cortical regions are order of magnitude greater in awake animals than reported here (see DOIs: 10.1016/j.neuroimage.2021.118413 and 10.3389/fnins.2018.00548). It raises some concerns e.g. if the signal is influenced by some unwanted noise (system related or biological), which could be a reason for low correlation values. Which kind of filtering was used for the raw fMRI signal before generating correlation matrix? More details about the analysis should be added to method section. Could authors provide some raw fMRI signals from brain areas used for the analysis to be included for the response or potentially to the supplementary material?

4. The authors mention the method being suitable for behavioral studies using a treadmill (I assume with some modifications done first to the body tube). Have authors tried the rigidity of the head restraint alone without fully restricting the body or having a bigger body tube? With the neck support used is the position of rat in a natural position so that it can be potentially walking on top the treadmill comfortably. Some clarifications to these could help with the potential researchers thinking about the suitability of the system for their studies.

5. What was the sound pressure of the T2-weighted anatomical scans? Did the rats react to this sequence differently compared to GE-EPI sequence for example by moving?

6. For the males, Sprague Dawley and Long Evans strains were used, but any explanation for the reasons to use of two different strains is not given. Did the strains behave differently in the restraint setup? Please add some explanation or discuss further in the discussion about the potential behavioral differences.

Minor:

7. Please add to the methods, what was the reward used after habituation sessions.

8. In the introduction, the reliance of custom hardware is mentioned as a limitation for awake rodent fMRI. This is somewhat contradictory statement as custom RF-coil was used in the presented manuscript. The use of custom build RF-coil should be also added to the limitations chapter in the discussion. Also, as authors already mentioned, one other limitation of the restraint method is the use of anchoring system that is not suitable for multimodal imaging such as EEG/fMRI. Authors could refer to the already published manuscript (doi: 10.1016/j.neuroimage.2022.118924) that allows simultaneous EEG/fMRI measurement.

9. It should be noted that zero echo time sequence presented by MacKinnon et al., 2025 is not the original reference to show benefits of ZTE approach for awake imaging. The original works implementing zero echo time sequences as a quiet sequence for awake animal imaging, should be used as references (dois: 10.1016/j.neuroimage.2022.118924, 10.7554/eLife.90318, 10.1162/imag_a_00384), while work by MacKinnon can be mentioned as refinement of method using slightly different gradient waveform, resulting marginally smaller sound pressure than the original work using MB-SWIFT or ZTE approaches.

10. For the figure 5, please add bregma coordinates for the image slices presented for temporal standard deviation and temporal SNR. Also please add units/explanation for the color bar.

11. In lines 78 and 216, the sentences (with "XXX"-markings) should be corrected.

12. Typo in the line 161 (remove the "the rat to"?).

Author Response

Dear Editor, We thank the reviewers and the editorial team for their constructive comments and feedback. Below, we have responded to the reviews provided by each referee, and we have correspondingly updated the manuscript for resubmission. We believe these changes have significantly improved the manuscript and we hope that it will be deemed suitable for publication in eNeuro. We will continue to build upon the findings from this manuscript, and look forward to incorporating the suggestions from the reviewers in future studies.

We confirm that neither the manuscript nor any parts of its content are currently under consideration or published in another journal. All authors have approved the manuscript and agree with its submission to eNeuro.

Sincerely, Jibran Khokhar Responses to Editor and Reviewers:

Thank you to the editor and reviewers for their continued insight and help with our manuscript. We have addressed the comments below.

Synthesis of Reviews:

Computational Neuroscience Model Code Accessibility Comments for Author (Required):

N/A Synthesis Statement for Author (Required):

Both reviewers were in agreement that this work is a valuable contribution to the field of awake rodent fMRI. Reviewers also commented positively on the practicality and utility of this head restraint method for eliminating anesthesia, especially for longitudinal studies. They had several suggestions for the manuscript, listed below, which would further improve clarity and reproducibility.

Reviewer 1 Major Comments 1. The habituation protocol is currently provided only in narrative form, which makes it difficult to follow step-by-step. Since the training schedule is critical for achieving stable awake imaging, I recommend adding a simple visual workflow diagram summarizing the day-by-day progression (e.g., exploration → body restraint → head fixation → mock scanner exposure) along with session durations. A timeline figure would make the protocol considerably easier for others to implement accurately.

Response We have added a visual workflow diagram to the "Steps for habituation to restraint, head fixing, and scanner noise" section as Figure 3, as suggested by the reviewer, seen below.

We have also referenced this figure throughout the manuscript on lines 164, 169, 171, 192, 205, and 220.

Figure 3. Visual flow diagram summarizing the step-by-step procedure for habituating a rat to the restraint system using the habituation protocol.

2. Because the primary aim of the system is to enable stable awake scanning, it would be helpful to include a brief performance comparison with scans acquired under anesthesia using the same hardware and acquisition parameters. This does not require additional functional interpretation. A small dataset (e.g., n = 2-3) demonstrating differences in mean framewise displacement and susceptibility-related distortion / anatomical alignment would clearly show the practical advantage of the awake preparation. This comparison would strengthen the methodological contribution without expanding the scope of the study.

Answer To compare performance between awake and anesthetized rats, we have included data on translation, rotation, and framewise displacement during scanning for both groups in Table 1, as shown below.

We have updated the text on head motion in the results section as follows on lines 302 to 309: "As shown in Fig. 7A and B (head motion and signal quality metrics for a representative anesthetized and awake rat, respectively), both subjects exhibited minimal head motion throughout the scan. Translational head motion along the x and y axes were significantly lower, but significantly higher along the z axis, in anesthetized rats compared to awake rats (all p < 0.0001, n = 5/group), while rotational displacements along all three axes were significantly lower in anesthetized rats compared to awake rats (all p < 0.0001), as shown in Table 1. Framewise displacement values were also significantly lower in anesthetized rats compared to awake rats (p < 0.0001)." Table 1: Comparison of head-motion parameters, including translation and rotation along the x, y, and z axes, as well as framewise displacement between anesthetized and awake rats.

We have also included additional text in the discussion section as follows on lines 357 to 366: "Overall, framewise displacement, translation, and rotation were significantly higher in awake rats than in anesthetized rats during scanning, with the exception of translation along the z-axis, which was lower in awake animals. This pattern is expected, as anesthetized rats are largely immobilized. The reduced z-axis translation in awake rats likely reflects the head-fixation setup, which restricts vertical movement, whereas anesthetized rats can exhibit small z-axis shifts due to breathing. Importantly, although motion was greater in awake rats than in anesthetized rats, the magnitude of movements observed with our restraint system was lower than that reported in previous studies using other restraint systems (Stenroos et al., 2018). This highlights the effectiveness of our design in minimizing motion-related artifacts and improving fMRI signal quality in awake rats." To extend the comparison between awake and anesthetized rats, we compared functional connectivity in awake and anesthetized rats as well. We have replaced the previous functional connectivity matrix heat map (Figure 8) with a new heat map representing the differences (changes) in functional connectivity between anesthetized and awake rats, as seen below.

Figure 8. Average Z-scored functional connectivity matrix showing the difference (awake - anesthetized) in pairwise functional connectivity between awake and anesthetized rats. The matrix shows pairwise Z-transformed Pearson correlation coefficients between BOLD signals across 59 brain regions. Warmer colors indicate stronger positive connectivity in awake rats, and cooler colors indicate weaker or negative connectivity in awake rats compared to anesthetized rats.

The results of this analysis have also been described in the results section on lines 316 to 326 as follows: "We compared Z-scored functional connectivity (FC) matrices derived from resting-state fMRI data in awake and anesthetized rats (Fig. 8). This analysis revealed widespread FC differences between groups, with some region pairs showing increased and others showing decreased connectivity. Among the increases observed in awake rats relative to anesthetized rats, the largest effects were seen between the left and right primary somatosensory cortex (L-S1-R-S1; z-score: l +0.4), ventral Cornu Ammonis 1 and right auditory parietal cortex (CA1v-R-Par (Aud); z-score: +0.4), and the hypothalamus and insula (Hyp1-Ins2; z-score: +0.3). In contrast, the largest decreases in FC in awake rats were observed between the cingulate and prelimbic cortices (Cg2-PrL; z-score: -0.3), retrosplenial and motor cortices (RSC1-M1/M2; z-score: / -0.3), and the right visual cortex and hypothalamus (R-V1/V2-Hyp1; z-score: / -0.3). However, none of these differences survived correction using NBS," and has been discussed in the discussion section as follows on lines 334 to 336: "Moreover, functional connectivity in awake rats was stronger between some region pairs and weaker between others compared with anesthetized rats," and on lines 367 to 379: "Z-scored FC comparisons revealed widespread differences in regions that form part of the rat default mode network (DMN) (Lu et al., 2012). Specifically, awake rats showed higher connectivity between hippocampal and hypothalamic regions and lower connectivity involving the retrosplenial, prefrontal, and visual cortices compared with anesthetized rats. These patterns are consistent with prior studies demonstrating anesthesia-related alterations in similar DMN regions, particularly when contrasting awake and isoflurane conditions (Paasonen et al., 2018; Hori et al., 2020; Tsurugizawa and Yoshimaru, 2021; Gutierrez-Barragan et al., 2022). Collectively, these findings reinforce evidence that resting-state FC differs markedly between awake and anesthetized conditions and that such effects depend on both anesthetic type and dose (Liu et al., 2013; Jonckers et al., 2014). These anesthesia-dependent variations might confound findings from preclinical fMRI studies. The proposed restraint system and protocol enable awake imaging without anesthesia, helping preserve native FC patterns and improve data quality and replicability." 3. Motion traces are currently shown for a single animal. Providing a simple summary table with mean {plus minus} SD of motion metrics (e.g., FD, translation, rotation) across animals would help define the typical performance range of the system and improve reproducibility for labs adopting the method.

Answer We have added a table (Table 1) summarizing the mean W SD of motion metrics (translation, rotation, and framewise displacement) for all awake and anesthetized rats, illustrating the typical performance range of the restraint system. Corresponding descriptions have been added to the Results section and further addressed in the Discussion, as detailed above.

We have also updated Figure 6 by including a second motion trace for a representative anesthetized rat for comparison.

Figure 7. Head motion and image quality metrics during fMRI scanning. A. Head motion and image quality metrics in an anesthetized rat. B. Head motion and image quality metrics in an awake rat. Left panel: Top: Time course of translational head motion parameters, Middle: Rotational motion parameters, and Bottom: Framewise displacement across the scanning session. Middle panel: Temporal standard deviation maps displayed in sagittal (top), coronal (middle), and axial (bottom) views. Right panel: Temporal signal-to-noise ratio maps for sagittal (top), coronal (middle), and axial (bottom) views.

We have also added a corresponding text to the Results section as follows on lines 302 to 309: "As shown in Fig. 7A and B (head motion and signal quality metrics for a representative anesthetized and awake rat, respectively), both subjects exhibited minimal head motion throughout the scan. Translational head motion along the x and y axes were significantly lower, but higher along the z axis, in anesthetized rats compared to awake rats (all p < 0.0001; n=5/group), while rotational displacements along all three axes were significantly lower in anesthetized rats compared to awake rats (all p < 0.0001), as shown in Table 1. Framewise displacement values were also significantly lower in anesthetized rats compared to awake rats (p < 0.0001)." 4. The manuscript describes the habituation schedule, but it does not specify how the authors determined when an animal was considered fully habituated. For reproducibility, it would be helpful to state the behavioral criteria used to define habituation completion.

Answer While these behaviors were not formally quantified, we relied on consistent qualitative indicators that the animals reached a stable, habituated state during restraint. By approximately the 11th day of habituation, all animals met the following criteria: they could be placed into the restraint and head-fixation apparatus without resistance; they displayed minimal or no tail flicks, body twisting, or escape attempts during immobilization; and they reliably consumed the food rewards provided at the end of each session. We have added the following text to the section "Steps for habituation to restraint, head fixing, and scanner noise" on lines 224 to 228: "By the final day of habituation, rats were introduced into the restraint and head-fixation setup without resistance, and exhibited minimal or no tail flicks, body twisting, or escape attempts. They also produced fewer fecal pellets during this session and consistently consumed the rewards offered after each session. These behaviors were used as qualitative criteria to determine completion of habituation." 5. Additionally, since corticosterone is commonly used in the awake fMRI field to assess long-term stress burden, the authors may consider reporting corticosterone levels across the habituation period. This would further support the usability and translational relevance of the method.

Answer We agree that corticosterone is a valuable biomarker for assessing long-term stress in awake fMRI paradigms. Unfortunately, we did not collect blood samples during this study and therefore cannot report corticosterone levels across the habituation period. Instead, we relied on qualitative behavioral indicators, such as minimal or no tail flicks, body twisting, or escape attempts, number of fecal pellets produced, and consistent consumption of rewards by the final days of habituation, as indirect signs of reduced stress. We acknowledge that these behavioral observations do not substitute for an objective biochemical measure, and we have added this point to the limitations section of the manuscript as follows on lines 433 to 438: "Corticosterone levels, albeit confounded by blood draw procedures, were not measured during habituation and scanning, which limits our ability to objectively quantify stress across the protocol. Although reductions in agitation-related behaviors and consistent reward consumption were used as indirect indicators of reduced stress, future studies should incorporate serial corticosterone measurements to provide an objective assessment of stress burden." Minor Comments 1. It would be helpful to clarify that the neck plate rests at the shoulder girdle / thoracic inlet, rather than on the cervical airway. A brief clarification in the text would reduce potential misinterpretation of restraint positioning.

Answer We thank the reviewer for this helpful suggestion. The neck plate does not rest on or apply pressure to the cervical airway. Instead, when lowered, it sits at the level of the thoracic inlet/shoulder girdle, where the shape of the plate and connector creates an adjustable circular opening that accommodates the rat's neck without constriction. This positioning ensures that the airway remains unobstructed and that the animals can breathe comfortably during immobilization and head fixation. We have added the following clarification to the "Steps for habituation to restraint, head fixing, and scanner noise" section on lines 183 to 187: "When lowered, the neck plate engages at the level of the shoulder girdle/thoracic inlet. The plate and connector form an adjustable circular opening that fits around the neck without compressing the cervical airway, ensuring unobstructed breathing during restraint." 2. If data are available, providing an approximate proportion of animals completing the full habituation duration would assist readers in planning timeline and throughput expectations.

Answer We thank the reviewer for this suggestion. In our dataset, 100% of the rats that began the habituation protocol (n = 5) successfully completed the full habituation schedule. Although animals show the greatest variability in resistance during the first immobilization and head-fixation sessions, all rats reliably adapted over repeated sessions, and by the final day no meaningful differences in compliance were observed. We have added the following clarification to the "Steps for habituation to restraint, head fixing, and scanner noise" section on lines 228 to 229: "Across all animals exposed to the protocol, 100% (n = 5) successfully completed the full habituation protocol." 3. The functional connectivity matrix could be made easier to interpret by adding region labels or modular grouping, although this is not essential to the core contribution.

Answer We have updated the figure for the functional connectivity matrix by making the region labels clearer. We have also referenced the atlas from which the definitions for these labels can be obtained in the section on "MRI Experiments" as follows on lines 267 to 270: "Image processing was performed using the Rodent Automated Bold Improvement of EPI Sequences (RABIES) software in conjunction with the SIGMA rat brain template (Barrière et al., 2019; Desrosiers-Grégoire et al., 2024)." Figure 8. Average Z-scored functional connectivity matrix showing the difference (awake - anesthetized) in pairwise functional connectivity between awake and anesthetized rats. The matrix shows pairwise Z-transformed Pearson correlation coefficients between BOLD signals across 59 brain regions. Warmer colors indicate stronger positive connectivity, and cooler colors indicate weaker or negative connectivity.

4. Since the gradient insert bore is relatively small, it would be helpful for the authors to briefly comment on the scope for integrating concurrent behavioral tasks during awake scanning. In particular, clarifying which types of head-fixed compatible behaviors are feasible within the current spatial constraints would help readers understand the potential for future behavioral extensions of the setup.

Answer We agree that the relatively small gradient insert bore limits the complexity and range of behavioral tasks that can be performed during awake scanning. We have modified the text to clarify the types of behaviors that are feasible within the existing spatial limitations on lines 395 to 403: "Another key advantage is the system's potential compatibility with behavioral paradigms. Because most of the rat's face, including the eyes, remains accessible during restraint, additional behaviors such as pupil tracking, requiring only an MRI-compatible camera positioned near the head, and lick-based tasks, using a small MRI-compatible lickometer (Frie and Khokhar, 2024) for liquid reward delivery or cue-driven licking, can be integrated during scanning. Although some locomotor behaviors requiring substantial limb movement, such as treadmill locomotion, are constrained by the current bore diameter, we are developing an expanded version of the system that will incorporate a low-profile roller-based treadmill, similar to that used in a previous study (Vasilev et al., 2022)." Reviewer 2 Major:

1. Several times throughout the manuscript, the authors mention the reduced stress by the introduced protocol. Although, it might be true that stress levels are reduced by the fact that anesthesia was not used, it has not been proved by showing any physiological stress indicators or by referring to any previous article showing the correlation between stress and anesthesia use. Therefore, I would encourage the authors to either downplay the claim of reduced stress (by the fact of not using anesthesia) or show any measured physiological markers or add references to show the link between anesthesia and reduced stress.

Answer We thank the reviewer for this important clarification. Because we did not collect physiological stress markers (e.g., corticosterone), we revised the manuscript to avoid making any unsubstantiated claims. Specifically, we removed language suggesting that the protocol "reduces stress" and instead restricted our statements to objectively observed behaviors during habituation (e.g., decreases in escape attempts and body movements). These revisions have been applied throughout the Significance Statement (lines 30 to 32: "By reducing motion, this simple restraint system allows for investigation of neural activity and connectivity without confounds from sedation or anesthesia."), Introduction (lines 69 to 71: "Our approach combines skull-screw-less head post implantation with a structured 11-day habituation regimen to minimize motion without the use of anesthesia."), and Methods (lines 164 to 165: "It is designed to minimize movement while preserving physiologically relevant brain states."), to ensure the text no longer implies that stress reduction was demonstrated.

2. Habituation schedule of 11 days was used. Was it based on previous literature or observed motion level of rats? Answer The 11-day habituation schedule was not based on prior literature but was determined from our observations during the habituation sessions. Although we did not formally quantify behavioral or physiological indicators, we noted progressive reductions in resistance to placement in the restraint system and minimal spontaneous movements (e.g., tail flicks, body twisting, escape attempts), as well as fewer fecal pellets by day 11, at which point animals appeared reliably acclimated. We agree that objective, quantifiable behavioral criteria would strengthen reproducibility. We have added the following text to the section "Steps for habituation to restraint, head fixing, and scanner noise" on lines 224 to 228: "By the final day of habituation, rats were introduced into the restraint and head-fixation setup without resistance, and exhibited minimal or no tail flicks, body twisting, or escape attempts. They also produced fewer fecal pellets during the restraint session and consistently consumed the rewards offered after each session. These behaviors were used as qualitative criteria to determine completion of habituation." 3. Based on the correlation matrix, very weak resting state correlation is observed. The authors state this in the discussion, but in my opinion should be discussed further where it arises. Typically, resting state correlations between cortical regions are order of magnitude greater in awake animals than reported here (see DOIs: 10.1016/j.neuroimage.2021.118413 and 10.3389/fnins.2018.00548). It raises some concerns e.g. if the signal is influenced by some unwanted noise (system related or biological), which could be a reason for low correlation values. Which kind of filtering was used for the raw fMRI signal before generating correlation matrix? More details about the analysis should be added to method section. Could authors provide some raw fMRI signals from brain areas used for the analysis to be included for the response or potentially to the supplementary material? Answer We thank the reviewer for highlighting this important issue. We agree that the initially reported resting-state correlations were substantially weaker (-0.03 to 0.03) than those typically observed in awake rodents and warranted closer examination. Upon re-evaluating our analysis pipeline, we identified that the functional connectivity matrices were not z-score normalized as intended due to an error in the analysis script. As a result, the values originally reported reflected unnormalized correlation coefficients. After correcting this issue, and regenerating the connectivity matrices using Fisher z-transformation, the magnitude of the difference between awake and anesthetized correlations was between -0.5 to 0.5; this is now comparable to values reported in prior awake rodent fMRI studies, including those cited by the reviewer, as seen below.

Figure 8. Average Z-scored functional connectivity matrix showing the difference in pairwise functional connectivity between awake and anesthetized rats. The matrix shows pairwise Z-transformed Pearson correlation coefficients between BOLD signals across 59 brain regions. Warmer colors indicate stronger positive connectivity, and cooler colors indicate weaker or negative connectivity.

To address the reviewer's question regarding preprocessing, the raw fMRI time series were processed using the RABIES pipeline, including motion correction, distortion correction, nuisance regression (motion parameters, white matter, CSF signals), global signal regression, and band-pass filtering (0.01-0.1 Hz; Grandjean et al., 2023; Desrosiers-Grégoire et al., 2024). These details can be found in the supplementary materials section of our recent paper (Frie et al, 2024).

We have also included representative raw fMRI signals from the brain regions used in the analysis as Fig. 6 in the main manuscript, as requested by the reviewer. We also added the following text to the Methods section on lines 271 to 274: "Labeled images in native space produced during the RABIES preprocessing stage were used to extract region-specific masks. Raw time series from each mask were averaged across all voxels within the region and plotted. Each time series was normalized to a 0-1 range." In addition, we added the following text to the Results section on lines 290 to 300: "Representative raw BOLD time series were extracted from regions used in the functional connectivity analysis to qualitatively assess signal quality (Fig. 6). Shown are normalized BOLD signals from 12 functional brain regions (CA1v, R-Par, L-S1, R-S1, Hyp1, Ins2, Cg2, PrL, RSC1, M1/M2, L-dHip, and Str), obtained by averaging voxel-wise signals within each region's mask in both anesthetized and awake rats. Across all regions, the time-series exhibited stable fluctuations throughout the scan duration, with no evidence of signal dropout, saturation, or prolonged drift. Following normalization, signal amplitudes remain within a consistent dynamic range, and fluctuations occur on timescales characteristic of low-frequency BOLD activity typically analyzed in resting-state fMRI. Together, these features support preservation of BOLD temporal structure and adequate signal stability, supporting the suitability of the data for subsequent functional connectivity analyses." We also added the following text to the Discussion section on lines 349 to 356: "Representative raw BOLD time series from multiple brain regions further support the quality of the fMRI data acquired with this awake imaging setup. The signals show stable low-frequency fluctuations across the entire scan, with no abrupt discontinuities, sustained baseline shifts, or large-amplitude spikes indicative of scanner artifacts, physiological instability, or excessive motion. This stability suggests that the head-fixation and habituation procedures were sufficient to preserve signal integrity during awake scanning. Although awake imaging in our study involved greater motion and physiological variability than anesthesia, these factors did not dominate or obscure the neural signals of interest." Figure 6. Representative average raw time series from multiple regions in anesthetized and awake rats. Left: masks of the corresponding regions overlaid on the SIGMA template anatomical image. Middle: average raw time series from an anesthetized rat. Right: average raw time series from an awake rat. CA1v, ventral Cornu Ammonis 1; R-Par, right auditory parietal cortex; L-S1, left primary somatosensory cortex; R-S1, right primary somatosensory cortex; Hyp1, hypothalamus 1; Ins2, insular cortex 2; Cg2, cingulate cortex 2; PrL, prelimbic cortex; RSC1, retrosplenial cortex; M1/M2, primary and secondary motor cortex; L-dHip, left dorsal hippocampus; Str, striatum.

4. The authors mention the method being suitable for behavioral studies using a treadmill (I assume with some modifications done first to the body tube). Have authors tried the rigidity of the head restraint alone without fully restricting the body or having a bigger body tube? With the neck support used is the position of rat in a natural position so that it can be potentially walking on top the treadmill comfortably. Some clarifications to these could help with the potential researchers thinking about the suitability of the system for their studies.

Answer We evaluated the rigidity of the head fixation independently of full-body restraint by testing a configuration in which only the head was fixed while the rat was positioned on a treadmill. In these tests, the head-post provided sufficient rigidity and stability for head fixation alone. However, in the current design, the neck-plate geometry positions the head and neck lower than would be required for a natural posture, which prevents comfortable or sustained walking. As noted by the reviewer, the restricted bore diameter and body-tube dimensions further limit limb movement. Consequently, while head fixation alone is mechanically stable, as shown below, the present configuration is not yet suitable for treadmill-based behavioral tasks. We are actively developing modified versions of the restraint system that incorporate a larger body tube, elevated head and neck support to better align the animal in a natural locomotor posture, and roller grid-based low-profile treadmills. These modifications are intended to enable proper stepping mechanics and facilitate future integration of treadmill walking during awake fMRI.

We have updated the text in the Discussion section as follows on lines 396 to 403: "Because most of the rat's face, including the eyes, remains accessible during restraint, additional behaviors such as pupil tracking, requiring only an MRI-compatible camera positioned near the head and lick-based tasks, using a small MRI-compatible lickometer (Frie and Khokhar, 2024) for liquid reward delivery or cue-driven licking, can be integrated during scanning. Although behaviors requiring substantial limb movement, such as treadmill locomotion, are constrained by the current bore diameter, we are developing an expanded version of the system that will incorporate a low-profile roller-based linear treadmill, similar to that used in a previous study (Vasilev et al., 2022)." 5. What was the sound pressure of the T2-weighted anatomical scans? Did the rats react to this sequence differently compared to GE-EPI sequence for example by moving? Answer The peak acoustic output of the MRI system used in this study is approximately 120 dB, and this applies to both the T2-weighted anatomical and GE-EPI functional sequences. During habituation, rats were repeatedly exposed to recorded scanner noise of comparable intensity to acclimate them to both sequence types. Although animals were not quantitatively monitored during mock scanner sessions, they were continuously observed during actual MRI acquisition. We did not observe any overt behavioral differences or increases in motion during the T2-weighted anatomical scans compared to the GE-EPI sequence. In particular, no sequence-specific increases in head movement, struggling, or agitation were apparent during anatomical imaging. While we acknowledge that these observations were qualitative and not quantitative, they suggest that the T2-weighted sequence did not elicit a stronger behavioral or motion response than the GE-EPI sequence in habituated animals and vice versa. This information is provided in the section "Steps for habituation to restraint, head fixing, and scanner noise" on lines 206 to 218 as follows: "After immobilization and head fixation, place the rat with the restraint system inside the mock MRI chamber and expose it to recorded MRI (T2-weighted anatomical and GE-EPI acquisition) sounds. i. Using a speaker capable of producing sounds up to 120 dB sound pressure level (SPL; the peak intensity of sounds produced by the MRI system used during both T2-weighted anatomical and GE-EPI sequences), play the recorded MRI sounds-starting at a lower intensity on day 7 and gradually increasing the intensity each day, reaching full intensity by day 11. ii. From days 7 to 11, immobilize the rat with head-fixing in the mock MRI chamber for progressively longer durations: 30, 45, 60, 90, and 120 minutes, respectively. No overt behavioral differences, such as head movement, struggling, or agitation, were observed during exposure to MRI sounds generated during either T2-weighted anatomical scans or GE-EPI sequence." 6. For the males, Sprague Dawley and Long Evans strains were used, but any explanation for the reasons to use of two different strains is not given. Did the strains behave differently in the restraint setup? Please add some explanation or discuss further in the discussion about the potential behavioral differences.

Answer Two male rat strains (Sprague Dawley and Long Evans) were included to test the broader applicability of the design across multiple rat strains. We acknowledge that the inclusion of multiple strains is a potential confound for strain-specific behavioral or stress-related differences during habituation and restraint. We have now added text to address this limitation to the Discussion section as follows on lines 438 to 444: "Another potential confound is the inclusion of two rat strains, which on one hand shows broader applicability of this design, but also introduces potential confounds. Long Evans and Sprague Dawley rats are known to differ in behavior and stress responsiveness (Schwarz et al., 2010). Although we observed no clear strain-related differences during restraint or scanning, the small sample size limits detection of subtle effects. Future studies with strain-balanced cohorts are needed to evaluate strain-dependent differences in habituation and motion during awake fMRI." Minor:

7. Please add to the methods, what was the reward used after habituation sessions.

Answer Banana-flavored sugar pellets were used as rewards after the habituation sessions (Bio-Serv, 45 mg, banana flavor; F0024). This information has been added to the Methods section as follows on lines 87 to 88: "Rats received banana-flavored sugar pellets (Bio-Serv, 45 mg, banana flavor, NJ, USA) as a reward after each habituation session." 8. In the introduction, the reliance of custom hardware is mentioned as a limitation for awake rodent fMRI. This is somewhat contradictory statement as custom RF-coil was used in the presented manuscript. The use of custom build RF-coil should be also added to the limitations chapter in the discussion. Also, as authors already mentioned, one other limitation of the restraint method is the use of anchoring system that is not suitable for multimodal imaging such as EEG/fMRI. Authors could refer to the already published manuscript (doi: 10.1016/j.neuroimage.2022.118924) that allows simultaneous EEG/fMRI measurement.

Answer We agree with the reviewer and have revised the manuscript accordingly. To address the apparent contradiction in the Introduction, we clarified that reliance on custom hardware is a general limitation of awake rodent fMRI rather than a flaw specific to our implementation. The revised text in the Introduction now reads on lines 56 to 59: "Despite its advantages, widespread adoption of awake rodent fMRI has been hindered by technical barriers, including procedural complexity, lack of standardized protocols, and reliance on custom hardware, which is often required to achieve adequate signal quality and head stabilization." We explicitly added the use of a custom-built RF coil to the Limitations section of the Discussion, as follows on lines 418 to 421: "This study employed a custom-built RF coil, which was necessary to achieve compatibility with the head-fixation system. However, this may limit reproducibility and direct comparability with studies using other coil designs, a constraint common to many awake rodent fMRI implementations." We also expanded the discussion of the system's limited compatibility with multimodal studies as follows on lines 413 to 418: "This limitation could be addressed by stereotaxically implanting MRI-compatible electrodes or fibers before cementing the head post, as demonstrated previously (Paasonen et al., 2022), by designing head posts that anchor around the skull periphery while leaving the central area accessible for multimodal recordings, by incorporating non-ferromagnetic inserts into the head post design (Quansah Amissah et al., 2023), or by implanting skull surface EEG electrode arrays under the head-post (Ding et al., 2024)." 9. It should be noted that zero echo time sequence presented by MacKinnon et al., 2025 is not the original reference to show benefits of ZTE approach for awake imaging. The original works implementing zero echo time sequences as a quiet sequence for awake animal imaging, should be used as references (dois: 10.1016/j.neuroimage.2022.118924, 10.7554/eLife.90318, 10.1162/imag_a_00384), while work by MacKinnon can be mentioned as refinement of method using slightly different gradient waveform, resulting marginally smaller sound pressure than the original work using MB-SWIFT or ZTE approaches.

Answer We have updated the Discussion section and cited the reviewer-recommended references as follows on lines 428 to 433: "A promising strategy to mitigate acoustic noise and enhance suitability for awake imaging is the use of zero echo time (ZTE)-based and related quiet pulse sequences, which employ near-constant gradient activity to substantially reduce acoustic noise (Paasonen et al., 2022; Dvořáková et al., 2024; Stenroos et al., 2024), minimizing stress-induced movement and signal distortions while improving image quality." 10. For the figure 5, please add bregma coordinates for the image slices presented for temporal standard deviation and temporal SNR. Also please add units/explanation for the color bar.

Answer We have revised the legend for Figure 7 to include both bregma coordinates and clear explanations of the color bars, including units. Specifically, we added the following text to the figure legend on lines 644 to 651: "Slice locations are reported in millimeters relative to bregma. Sagittal slices (mediolateral): 4.6, 3.4, 1.9, and 0.4 mm; coronal slices (anteroposterior): −7.30, −6.04, −1.30, and 2.70 mm; transverse slices (dorsoventral): −7.10, −6.10, −4.28, and −3.38 mm. Temporal standard deviation maps represent voxel-wise signal variability (arbitrary signal units), with higher values indicating lower temporal variability and lower values indicating higher temporal variability. Temporal signal-to-noise ratio (tSNR) maps are unitless and represent the ratio of the mean signal to its temporal standard deviation, with higher values indicating greater signal stability over time." 11. In lines 78 and 216, the sentences (with "XXX"-markings) should be corrected.

Response We used "XXX" as placeholders because the journal's double-blind review policy requires that no identifying information be included in the manuscript.

12. Typo in the line 161 (remove the "the rat to"?).

Answer We have modified the text as follows on lines 176 to 177: "Wrap the napkin around the rear of the tube and the rear of the rat to encourage it to move forward through." References Desrosiers-Grégoire G, Devenyi GA, Grandjean J, Chakravarty MM (2024) A standardized image processing and data quality platform for rodent fMRI. Nature Communications 15:6708 Grandjean J, Desrosiers-Gregoire G, Anckaerts C, Angeles-Valdez D, Ayad F, Barrière DA, Blockx I, Bortel A, Broadwater M, Cardoso BM (2023) A consensus protocol for functional connectivity analysis in the rat brain. Nature neuroscience 26:673-681.

Frie JA, McCunn P, Eed A, Hassan A, Luciani KR, Chen C, Tyndale RF, Khokhar JY (2024) Factors influencing JUUL e-cigarette nicotine vapour-induced reward, withdrawal, pharmacokinetics and brain connectivity in rats: sex matters. Neuropsychopharmacology 49:782-795 Schwarz, C., Hentschke, H., Butovas, S., Haiss, F., Stüttgen, M. C., Gerdjikov, T. V., ... &Waiblinger, C. (2010). The head-fixed behaving rat-procedures and pitfalls. Somatosensory &motor research, 27(4), 131-148.