Introduction

In the retina, photoreceptor cells convert light signals into electrical signals, which are then transmitted to the inner nuclear layer of the retina through synapses. Here, a series of cells, including bipolar cells, amplify, modulate, and integrate these signals before passing them on to the retinal ganglion cells. These signals are subsequently transmitted through the optic nerve axons to the visual cortex, where they undergo further processing and integration, ultimately forming vision. The output of these processes constitutes one of the primary sensory inputs of the brain1,2,3,4,5. The degeneration of retinal neurons is one of the leading causes of vision loss, drawing substantial attention in both hereditary diseases such as retinitis pigmentosa, Stargardt disease, and Leber congenital amaurosis, and in complex age-related diseases like age-related macular degeneration. In recent years, there has been substantial progress in understanding the mechanisms of retinal neurodegenerative diseases, leading to considerable advancements in various biomedical therapies, including gene therapy and stem cell therapy6. These developments have brought hope for effective treatment of retinal neurodegenerative diseases7,8,9,10.

In preclinical proof-of-concept studies on the mechanisms and treatments of retinal degenerative diseases, we typically use a series of rigorous visual electrophysiological tests to assess whether the visual function of the corresponding animal models has been effectively improved. These tests include methods such as flash electroretinography (FERG), pattern electroretinography (PERG), pattern visual evoked potential (PVEP), and flash visual evoked potential (FVEP). These tests provide a comprehensive evaluation of the entire visual pathway, from retinal photoreceptor cells to the visual cortex of the brain11,12,13. FERG and FVEP are used to evaluate the responses of the retina and visual cortex to flash stimuli, respectively, while PERG and PVEP assess responses to pattern stimuli, focusing on signal transmission from ganglion cells to the primary visual cortex14,15,16,17. By conducting statistical analysis of these test results, we can gain a comprehensive understanding of the transmission and processing of visual signals at various levels. This, in turn, helps in studying the specific functions of different types of cells in visual signal conduction.

Since animals cannot actively cooperate with testing procedures like humans, using anesthetics to reduce experimental errors caused by movement and stress during electrophysiological recordings becomes crucial. Commonly used anesthetics include ketamine/xylazine and chloral hydrate. However, ketamine/xylazine is strictly regulated or even banned in many countries, including China, making its acquisition process extremely cumbersome, complex, and sometimes illegal. Chloral hydrate was once used for anesthesia and euthanasia in large animals. However, due to its pronounced side effects, such as vascular and respiratory depression, its use in veterinary practice has been greatly reduced18. In the research field, especially in studies involving laboratory animals like mice and rats, the use of chloral hydrate has been strictly limited, as many consider it to have sedative rather than anesthetic effects19. In humans, while chloral hydrate serves as a sedative and anxiolytic, concerns have been raised regarding potential adverse effects, including respiratory depression and gastrointestinal discomfort20,21,22. For these reasons, we did not use chloral hydrate in this study. Therefore, alternatives such as ethyl carbamate (EC), avertin (AR), and pentobarbital sodium (PS) have become substitutes for these animal anesthetics. It is important to note that these anesthetics have different effects on electroretinogram (ERG) results. For instance, wild-type rats under PS anesthesia do not exhibit oscillatory potentials (OPs) similar to those observed with ketamine or avertin23,24. Furthermore, even under identical stimulation conditions, the FERG amplitude in wild-type mice varies substantially depending on the anesthetic used25,26. To avoid inaccuracies in evaluating visual function, it is essential to conduct a comprehensive comparison of the effects of these three anesthetics on visual electrophysiological signals within a consistent experimental framework. Additionally, the visual neural pathway encompasses not only the photoreceptor cells in the outer nuclear layer (ONL) and the cells in the inner nuclear layer (INL) involved in FERG but also the retinal ganglion cell (RGC) layer and the visual cortex. In contrast to the extensive research on FERG, there is a notable lack of detailed comparisons regarding the impact of these anesthetics on PERG, FVEP, and PVEP signals, which reflect the status of these neurons.

This study investigated the effects of three commonly used anesthetics: EC, AR, and PS, on the complete visual signal transduction pathway under identical conditions. Compared to PS, which suppresses oscillatory potentials and PERG waveforms, mice anesthetized with AR consistently exhibited stable and high-amplitude characteristic peaks across all visual electrophysiological tests. Although the PERG results in the EC group were comparable to those in the AR group, their FERG and FVEP results were lower than those of the other two groups. Most notably, typical PVEP results were observed only in mice anesthetized with AR, while they were absent under the other two anesthetics. By conducting a series of visual electrophysiological tests, including FERG, PERG, PVEP, and FVEP, we determined that AR is the most suitable anesthetic among the three tested for visual electrophysiological research.

Results

Comparative analysis of anesthetic effects on induction and duration of anesthesia in mice

To assess the differential impact of various anesthetics on mice, we categorized eighteen two-month-old mice, characterized by stable physiological states and full maturity, into three cohorts: those receiving ethyl carbamate (EC), pentobarbital sodium (PS), and avertin (AR) for anesthesia.

In this study, we conducted four visual electrophysiological experiments, each requiring approximately 60 min of recording time. During these experiments, even minor movements by the mice could cause electrode displacement or detachment, leading to substantial experimental errors. To prevent this, all recordings were conducted under deep anesthesia to ensure the accuracy of the results. Based on previous research and experimental requirements, the anesthetic doses were set at 1.6 g/kg for ethyl carbamate27,28,29,30,31, 0.30 g/kg for avertin32,33,34,35,36, and 0.115 g/kg for pentobarbital sodium37,38,39,40,41,42. Under these doses, the deep anesthesia time in mice was approximately 70 min (EC group: 71 ± 14.05 min, AR group: 79.88 ± 3.93 min, PS group: 65.47 ± 3.05 min), thereby ensuring sufficient anesthesia for approximately 60 min of recording. After 70 min of deep anesthesia, the mice began to show gradual physiological changes, including subtle head movements during thoracic or thoracoabdominal breathing, corneal reflexes, and head lifting and neck turning movements. Stable electrical signals could no longer be obtained in mice exhibiting these movements. Therefore, all experimental recordings were performed during the deep anesthesia phase. The recording time (i.e., deep anesthesia duration) was kept consistent across the three groups (approximately 70 min), ensuring comparable physiological states and, consequently, data accuracy and reliability.

The deep anesthesia state was determined by the toe pinch reflex and further confirmed through stable EEG results. Once the mice exhibited no response to the toe pinch, the stability of the EEG was used to confirm the anesthesia state (Fig. S1). The moment EEG stabilization was achieved marked the beginning of the measurements. The induction time was defined as the interval from anesthetic administration to the loss of the toe-pinch reflex. The deep anesthesia time was as the physiological state in which the mice exhibit no response to external stimuli and complete loss of consciousness. The anesthesia time encompassed the entire period from the onset of induction until the restoration of consciousness and motor function. All assessment was based on both the toe-pinch reflex and EEG monitoring (Fig. S1). Statistical analysis revealed that the AR group exhibited the most rapid anesthesia induction, whereas the EC group demonstrated the longest induction and overall anesthesia duration, with substantial intra-group variability (Table 1).

Table 1 Anesthesia timetable. Representative data are presented as mean ± SEM; n = 6 independent biological replicates.
Full size table

Assessment of scotopic response and ERG parameters in mice administered with various anesthetics

To objectively evaluate the scotopic responses of 18 two-month-old mice, we recorded their reactions under conditions of dark adaptation with incrementally intensified flash luminance (Fig. 1A). Considering that the frequency of oscillatory potentials (OPs) varies among species, and that in mice, as well as in humans, OPs typically exhibit frequencies between 75 and 80 Hz and 130–150 Hz43, we used a 30 Hz low-frequency filter in our methodology to isolate lower frequency components. This approach was intended to mitigate the effect of high-frequency filtering on the a-wave and b-wave, thus preserving their amplitudes for more precise fitting using established mathematical models (Fig. 1B). Subsequently, we harnessed the capabilities of the Matlab Statistical Toolbox to fit the scotopic ERG a-wave (Fig. 1C–E), extracting key parameters such as Rmax, Td, and S from the Lamb and Pugh models (Fig. 1F-H).

Fig. 1
figure 1

Scotopic ERG and a-wave Fitting. (A) Representative scotopic and photopic ERG responses in mice treated with different anesthetics. (B) The scotopic and photopic ERG waveforms from different groups were processed with a band-pass filter of 0.3–30 Hz to remove the influence of oscillatory potentials. (C–E) A-wave fitting waveforms for EC (C), AR (D), and PS (E). (F–H) Quantitative statistics of the a-wave related parameters, Rmax (F), S (G), and Td (H), derived from fitting with the Lamb and Pugh model. Representative data are presented as mean ± SEM; n = 6 independent biological replicates. Two-way ANOVA tests with Tukey’s multiple comparisons. *, P < 0.05; **, P < 0.01; ***, P < 0.001. All experiments were independently performed at least three times to ensure repeatable results.

Full size image

Dissecting the scotopic a-wave analysis, marked disparities were evident among the groups in terms of Rmax (F (2, 15) = 12.17, P < 0.001), S (F (2, 15) = 3.823, P = 0.05), and Td (F (2, 15) = 22.71, P < 0.001). Specifically, the a-wave Rmax of the EC group (2.581 ± 0.02244 log μV) was inferior to those of the AR and PS groups (2.750 ± 0.02365 and 2.773 ± 0.04052 log μV, respectively) (Fig. 1F), i suggesting that EC may suppress rod phototransduction signals.

Contrastingly, the S value, indicative of sensitivity to variations in light luminance, did not differ substantially across the EC, AR, and PS groups, with respective readings of 0.006748 ± 0.0002511, 0.006303 ± 0.0003810, and 0.009598 ± 0.001517. This uniformity underscores that the trio of anesthetics did not substantially influence the rod cells’ responsiveness to light stimuli (Fig. 1G).

Departing from the trends noted for Rmax and S, the Td parameter for the EC group (8.760 ± 0.1937 ms) substantially exceeded that of the AR (5.605 ± 0.3914 ms) and PS (7.586 ± 0.3813 ms) groups (Fig. 1H), indicating a markedly slower response to light in cells treated with EC. It is noteworthy that the PS group manifested the second-lengthiest Td value, hinting that PS may obstruct the signal transduction within the rod cells44.

Analysis of scotopic b-wave amplitude variations in ERG under different anesthetic treatments

In our investigation, we employed the Naka-Rushton equation to analyze the b-wave amplitude (R) in dark-adapted electroretinograms, revealing a dependency of R on the stimulus illumination. The b-wave fitting data for the dark-adapted ERG indicated significant intergroup variations in Rmax (F (2, 15) = 25.24, P < 0.001), N (F (2, 15) = 16.47, P < 0.001), and K (F (2, 15) = 94.78, P < 0.001) (Fig. 2).

Fig. 2
figure 2

Fitting of scotopic ERG b-waves. (A–C) Representative scotopic ERG b-waves of EC (A), AR (B) and PS (C). ERGs were recorded in dark-adapted mice with the light intensities increasing from − 3.6 to 2.15 log cd s m−2. (D–F) B-wave fitting waveforms of EC (D), AR (E) and PS (F). (G–I) Statistical analysis of the b-wave parameters, Rmax (G), N (H), and K (I), obtained by applying the Naka-Rushton equation. Representative data are presented as mean ± SEM; n = 6 independent biological replicates. Two-way ANOVA tests with Tukey’s multiple com-parisons. *, P < 0.05; **, P < 0.01; ***, P < 0.001. All experiments were independently performed at least three times to ensure repeatable results.

Full size image

Aligning with the a-wave findings, the Rmax for the dark-adapted b-wave in the EC group (3.012 ± 0.02970 log μV) was subpar compared to the AR (3.176 ± 0.02866 log μV) and PS (3.351 ± 0.04141 log μV) groups (Fig. 2G). The scotopic ERG b-wave mainly reflected the function of inner nuclear layer cells, which aggregates signals from photoreceptors before relaying them to amacrine and ganglion cells45. Given that the amplitude of the photoreceptor signals can influence those generated by bipolar cell transmission, the direct link between a-wave reduction and the subsequent b-wave diminishment remains to be elucidated.

Moreover, the N and K parameters for the dark-adapted b-wave in the AR group (N, 6.207 ± 0.3509; K, 5.152 ± 0.04133) were notably less than those in the EC and PS groups (N and K, 7.985 ± 0.2212 and 5.556 ± 0.01385 for EC group; 8.035 ± 0.1590 and 5.828 ± 0.04194 for PS group) (Fig. 2H, I), implying superior cell sensitivity in the AR group. It is important to observe that PS did not diminish the Rmax of the scotopic ERG a-wave (Fig. 1F) nor was the S value for the dark-adapted a-wave lowest in the PS group (Fig. 1G), indicating that PS did not exert marked inhibitory effects on rod-initiated signals. Nonetheless, the elevated K value of the dark-adapted b-wave in the PS group, the highest among the studied groups (Fig. 2I), suggests that PS might exert a modulatory influence on the sensitivity of bipolar cells46.

Characterization of oscillatory potential amplitudes and implicit times in scotopic ERG responses under anesthesia

To accurately delineate the peak amplitudes of OPs within the 75–80 Hz and 130–150 Hz frequency ranges, we processed the scotopic ERG response through a 65 Hz high-pass filter coupled with a 300 Hz low-pass filter. We defined the amplitude from the nadir of the initial negative wave to the zenith of the first positive wave as OP1, with subsequent peaks labeled consecutively as OP2 through OP5 (Fig. 3A). The cumulative OP amplitude (OPs) was calculated as the aggregate of OP1-OP5, while the implicit times (Ti) were gauged from the inception of each OP to the precise moment of stimulation.

Fig. 3
figure 3

Record of the oscillatory potentials. (A) Representative waveforms of oscillatory potentials (OPs) in three groups were recorded under dark adaptation, in response to a flash with a luminance of 1.4 log cd s m−2. (B) The frequency-___domain waveforms of oscillatory potentials (OPs) for three groups: EC (gray), AR (blue), and PS (green). (C, D) Quantitative statistics of the implicit time (C) and amplitude (D) of OPs across three groups. Representative data are presented as mean ± SEM; n = 6 independent biological replicates. Two-way ANOVA tests with Tukey’s multiple comparisons. *, P < 0.05; **, P < 0.01; ***, P < 0.001. All experiments were independently performed at least three times to ensure repeatable results.

Full size image

The frequency response spectra of the three anesthetic groups exhibit uniformity. Nonetheless, the peak responses of the EC and PS groups are markedly reduced compared to the AR group, suggesting that while EC and PS induce a certain suppression of frequencies, they still maintain the principal spectral components (Fig. 3B). In the within-group analysis of OP implict times, a significant interaction effect was observed (F(10, 90) = 98.35, P < 0.001), along with highly significant effects for both the row factor (F(5, 90) = 34,477, P < 0.001) and the column factor (F(2, 90) = 466.6, P < 0.001). Similarly, in the analysis of OP peak amplitudes, a significant interaction effect was detected (F(10, 90) = 41.79, P < 0.001), with both the row factor (F(5, 90) = 438.3, P < 0.001) and the column factor (F(2, 90) = 142.9, P < 0.001) also showing highly significant results. Our analysis revealed that the Ti values for the PS group (214.00 ± 1.84 ms) exhibited delays relative to the EC and AR groups (186.35 ± 1.98 and 185.42 ± 2.73 ms, respectively) (Fig. 3C), insinuating that PS impose inhibitory effects on OPs production. the OP peak amplitudes of the AR groups (2132.47 ± 116.86 μV) were substantially increased compared to the EC and PS group (907.03 ± 256.25 and 839.71 ± 176.86 μV, respectively) (Fig. 3D).

Analysis of cone function and photopic ERG response to anesthetics

To evaluate the effects of anesthetics on cone function, we performed photopic ERG against a white background illumination of 1.48 log cd s m−2. The PS group exhibited a singular peak profile, in contrast to the EC and AR groups, which each displayed a tri-phasic peak pattern (Fig. 4A–C).

Fig. 4
figure 4

Analysis of photopic ERG b-waves. (A–C) Exemplary photopic ERG b-waves were recorded from EC (A), AR (B), and PS (C) mice, with the light intensities progressively increasing from − 0.6 to 1.4 log cd s m−2 against a background illumination of 1.48 log cd s m−2. (D–F) Curves depicting the fitted b-wave responses for EC (D), AR (E), and PS (F). (G–I) Quantitative evaluation of b-wave parameters, such as R (G), N (H), and K (I), derived through the application of the Naka-Rushton equation. Representative data are presented as mean ± SEM; n = 6 independent biological replicates. Two-way ANOVA tests with Tukey’s multiple com-parisons. *, P < 0.05; **, P < 0.01; ***, P < 0.001. All experiments were independently performed at least three times to ensure repeatable results.

Full size image

To more comprehensively analyze the differences in b-wave of photopic ERG among different groups, we fitted the b-wave using the Naka-Rushton equation and analyzed the related parameters (Fig. 4D–F). The analytical fit utilizing the Naka-Rushton equation showed that Rmax (F (2, 15) = 7.759, P = 0.0049), K (F (2, 15) = 29.92, P < 0.001), and N (F (2, 15) = 5.717, P = 0.0143) differed substantially among the groups (Fig. 4G–I). Specifically, the Rmax for the b-wave during the AR group’s photopic ERG (2.977 ± 0.05866 log μV) exceeded those recorded for the EC (2.678 ± 0.08031 log μV) and PS (2.654 ± 0.05187 log μV) groups (Fig. 4G). Additionally, the AR group exhibited a notably lower N value (5.664 ± 0.2764) when contrasted with the PS group (8.978 ± 0.6912) (Fig. 4H). Relative to the EC and AR groups (4.844 ± 0.03474 and 4.756 ± 0.03853, respectively), the PS group presented the lowest K value in photopic ERG (4.455 ± 0.03853) (Fig. 4I).

Impact of anesthetic variation on mouse PERG through analysis of amplitude and latency

To evaluate the impact of anesthetics on ganglion cell signaling, we performed pattern electroretinograms (PERG) (Fig. 5A), which revealed distinct differences among the mice in various groups (Fig. 5B).

Fig. 5
figure 5

Characterization of PERG under varied anesthetic conditions in mice. (A) Schematic of the PERG experiment. (B) The waveforms of PERG results in mice under the influence of three different anesthetics: EC, AR, and PS, are illustrated. (C–D) The implicit times (C) and amplitude (D) for N1, P1, and N2 waves among the groups are quantified. (E–M) Statistical analysis of parameters following frequency-___domain analysis, including magnitude (E–G), phase (H–J), and power spectral density (PSD) (K–M). Compared to the other two groups, the PS group mice (G, J, M) demonstrated insensitivity to graphic stimuli when compared with the EC (E, H, K) and AR (F, I, L) groups. Representative data are presented as mean ± SEM; n = 6 independent biological replicates. Two-way ANOVA tests with Tukey’s multiple com-parisons. *, P < 0.05; **, P < 0.01; ***, P < 0.001. All experiments were independently performed at least three times to ensure repeatable results.

Full size image

The P1 and N2 waves exhibited significant intragroup variability in both implicit time and amplitude characteristics. For the P1 component, the analysis revealed F(2,15) = 40.58 with P < 0.001 for implicit time and F(2,15) = 45.98 with P < 0.001 for amplitude. Similarly, the N2 component showed F(2,15) = 5.708 with P = 0.01 for implicit time and F(2,15) = 230.1 with P < 0.001 for amplitude. In contrast, the N1 wave showed no significant intergroup differences in either implicit time (F(2,15) = 1.109, P = 0.36) or amplitude (F(2,15) = 2.658, P = 0.10). The N1 waves of the three groups of mice showed almost no differences, except that the implicit time of the EC group (20.61 ± 4.331 ms) was slightly higher than that of the AR group (13.13 ± 2.949 ms) (Fig. 5C).

Regarding the latency and amplitude of the P1 peak, the AR group exhibited the shortest implicit time (114.00 ± 3.824 ms) and the highest amplitude (17.28 ± 0.7070 μV), followed by the EC group with a latency of 110.2 ± 1.821 ms and an amplitude of 14.05 ± 0.9080 μV. The PS group demonstrated the longest latency (140.3 ± 1.397 ms) and the lowest amplitude (7.406 ± 0.5744 μV) (Fig. 5C, D).

The distinction among the three groups was most pronounced at the N2 peak, where the EC group’s performance was notably superior to that of the other two. The EC group exhibited the most robust N2 peak amplitude (23.44 ± 0.2596 μV) coupled with the shortest latency (229.7 ± 4.456 ms), a substantial enhancement compared to the AR group’s amplitude (23.34 ± 0.4336 μV) and latency (236.4 ± 3.264 ms), and the PS group’s amplitude (7.333 ± 0.4959 μV) and latency (245.1 ± 0.9631 ms) (Fig. 5C, D).

We conducted a frequency ___domain analysis of the results from three anesthetic groups (Fig. 5E–M). Although the phase results for all groups remained between − 4 and 4 rad, the magnitude (Fig. 5E–G) and Power Spectral Density (PSD) (Fig. 5K–M) for the PS group were substantially lower than those of the other two groups. This phenomenon corresponds with the time-___domain analysis results of the three groups’ PERG. In conclusion, the use of PS substantially prolonged the latency across all waves while also diminishing their amplitudes, suggesting a negative impact on the signal transmission process of ganglion cells.

Assessing the impact of anesthetics on visual signal transmission via FVEP

FERG and PERG are both visual electrophysiological experiments used to measure retinal signals. We conducted additional FVEP testing to evaluate the electrophysiological responses of the visual cortex to simple flash stimuli. The typical FVEP waveform consists of two positive peaks, P1 and P2, and one negative peak, N1, all of which were observed in the FVEP results across three groups (Fig. 6A).

Fig. 6
figure 6

The effects of different anesthetics on FVEP differ between anesthetics. (A) Representative FVEP waveforms from distinct groups: EC (bottom), AR (middle), and PS (top), elicited under a luminance of − 0.3 log cd s m−2. (B, C) The density plots show the density distribution of N1 (B) and P2 (C) amplitudes across various groups in their FVEP responses. (D–F) Latencies of P1 (D), N1 (E), and P2 (F) in different groups. (G–I) Peak amplitudes for P1 (G), N1 (H), and P2 (I) in different groups. Representative data are presented as mean ± SEM; n = 6 independent biological replicates. Two-way ANOVA tests with Tukey’s multiple com-parisons. *, P < 0.05; **, P < 0.01; ***, P < 0.001. All experiments were independently performed at least three times to ensure repeatable results.

Full size image

The temporal analysis indicates that the distribution of P1 implicit time shows no significant intergroup differences (F(2,15) = 0.8872, P = 0.43). In contrast, both the implicit time of N1 and P2 exhibit significant intergroup variations, with F(2,15) = 14.25, P < 0.001 for N1 and F(2,15) = 138.1, P < 0.001 for P2. Regarding amplitude analysis, the P1, N1, and P2 waves all demonstrate significant intergroup differences. Specifically, the P1 wave exhibits highly significant differences in amplitude (F(2,15) = 44.84, P < 0.001), while the N1 and P2 wave differences, though slightly less pronounced, remain statistically significant (N1: F(2,15) = 6.318, P = 0.01; P2: F(2,15) = 6.107, P = 0.01). These findings collectively suggest substantial intergroup variability in the amplitudes.

Compared to the other two groups, the AR group exhibited a more concentrated distribution of N1 and P2 peak amplitudes within a narrower range, indicative of stability in the results (Fig. 6B, C). There was no substantial difference in the latency of the P1 among the three FVEP groups (Fig. 6D), while the latency of the N1 characteristic peak in the PS group (36.22 ± 0.1776 ms) was higher than that of the EC group (34.31 ± 0.9270 ms) and the AR group (34.14 ± 0.8919 ms) (Fig. 6E).However, the peak of the PS group’s P2 was mostly concentrated around 60 ms and 75 ms, which differs from the distribution of the peak times for the P2 in the EC and AR groups (Fig. 6F).

The FVEP outcomes for the EC group were inferior compared to those of the other two groups, despite the P1 peak amplitude of the PS group (17.33 ± 1.926 μV) being lower than those recorded for the AR and EC group (60.48 ± 4.515 μV and 35.17 ± 2.710 μV respectively) (Fig. 6G). Specifically, the EC group exhibited substantially lower mean amplitudes for both N1 (274.6 ± 15.01 μV) and P2 (341.6 ± 17.94 μV) compared to the AR group, which demonstrated higher amplitudes of 340.1 ± 8.701 μV for N1 and 430.6 ± 15.28 μV for P2, and to the PS group, which had a mean N1 amplitude of 288.1 ± 16.36 μV and a P2 amplitude of 389.1 ± 20.45 μV (Fig. 6H, I).

Assessing the impact of anesthetics on visual signal transmission via PVEP

To further investigate the effects of anesthetics on the patterned visual signal pathway, we subsequently conducted PVEP assessments. This evaluation focused on analyzing the electrophysiological responses of the primary visual cortex to patterned stimuli, aiming to provide a more comprehensive analysis of the effects at the central nervous system level. Substantially, PVEP waveforms were exclusive to the AR group (Fig. 7A). In the PS group, no distinct waveform was observed, which may be related to the diminished signal conduction of RGCs. Conversely, the lack of substantial waveforms in the EC group stood in stark contrast to their definitive PERG outcomes. This suggests potential EC-mediated disruption in the signal transduction from RGCs to the visual cortex.

Fig. 7
figure 7

Evaluation of PVEP Outcomes. (A) Schematic representation of PVEP waveforms in EC (left), AR (center), and PS (right) mice. (B–C) Statistical analysis of the N1 (B) and P2 amplitudes (C) in the PVEP of AR mice. (D–I) Frequency ___domain analysis of the PVEP results for the AR group, including magnitude (D, E), phase (F, G), and power spectral density (PSD) (H, I). Frequency ___domain analysis results are statistically analyzed for recordings at a depth of 800 μm (D, F, H) and across all depths (E, G, I). Representative data are presented as mean ± SEM; n = 6 independent biological replicates. Two-way ANOVA tests with Tukey’s multiple com-parisons. *, P < 0.05; **, P < 0.01; ***, P < 0.001. All experiments were independently performed at least three times to ensure repeatable results.

Full size image

At a depth of 800 µm, PVEP N1 and P2 waves in the AR group reached their zenith, with amplitudes of 68.65 ± 4.89 µV and 69.57 ± 4.78 µV, respectively ((Fig. 7B, C). Fourier Transform analysis of the PVEP signals in AR mice elucidated the relationship between magnitude, phase, and PSD across frequencies (Fig. 7D–I). Similar to its PERG frequency ___domain analysis results, the AR group’s PVEP magnitude (Fig. 7D, E) and PSD (Fig. 7H, I) primarily respond within the 0–10 Hz range across virtually all depths. Similarly, the AR group’s phase is also concentrated between − 4 to 4 radians (Fig. 7F, G).

In summary, Fourier Transform analysis has unveiled primary PVEP signal frequencies at 0–10 Hz with notable amplitude and power, and a complex phase response.

Discussion

This study conducts a comprehensive comparison of the efficacy of three anesthetics—AR, EC, and PS—in recording electrophysiological signals from the retina and primary visual cortex in mice. In this research, 2–2.5 m pigmented C57BL/6J mice were selected as the experimental subjects because retinal development and light response vary greatly between pigmented and non-pigmented laboratory animals47,48. Compared to C57BL/6J mice, non-pigmented C57BL/6J-c2J mice exhibit substantial retinal degeneration. During development and early adulthood, these mice have abnormally high photoreceptor death rates, leading to substantially reduced ERG amplitudes, sometimes even less than half of those observed in C57BL/6J mice49. Additionally, albino mice show alterations in phototransduction gain and slower recovery after photobleaching50,51. Based on this, we selected C57BL/6J mice aged 2–2.5 months, as their retinas are fully developed at this stage52, and no substantial sex differences have been found in their visual electrophysiological signals53,54. Therefore, the results for male and female mice were not analyzed separately in this study. However, this method may not be applicable to older animals. As age increases, the protective effect of estrogen on vision becomes more evident, leading to less pronounced visual decline in older female animals compared to males55,56.

The research conducted a comprehensive comparative analysis of 35 parameters across four types of visual electrophysiological experiments (FERG, PERG, PVEP, and FVEP). The results showed no substantial differences among the three groups for five of these parameters (Table S1). However, in the remaining 30 parameters, the AR group outperformed the other two groups in 25 of them. Particularly in the PVEP-related parameters, only the AR-anesthetized mice exhibited typical responses, whereas no such responses were recorded in the EC and PS groups.

In addition to the AR group’s strong performance in 25 parameters and the five parameters showing no substantial difference among the three groups, the EC and PS group displayed slight advantages in the remaining five parameters. However, these advantages were not substantial enough to impact the overall conclusions of this study. For instance, the differences between the PS and AR groups did not result in qualitative changes. In the scotopic ERG b-wave Rmax, the PS group had the best performance (3.351), closely followed by the AR group (3.176), with a difference of only 0.175. This small variation was insufficient to alter the overall assessment. Similar trends were observed in other parameters, such as the N value of the scotopic ERG b-wave and the K value of the photopic ERG b-wave.

Previous studies have elucidated the substantial impact of anesthetics on systemic effects, highlighting the potential adverse effects of certain anesthetics57,58,59,60. Among the three anesthetics, avertin exhibited physiological effects most similar to those of ketamine/xylazine (KX), which has long been considered the gold standard for mouse experiments. Avertin had a milder impact on heart rate compared to ketamine/xylazine and sodium pentobarbital in experimental animals61,62,63. Although animals under EC anesthesia displayed physiological characteristics remarkably similar to the sleep state64, rats anesthetized with ethyl carbamate showed abnormal cardio-pulmonary patterns, including a marked decrease in heart rate during inspiration. Additionally, only mice anesthetized with pentobarbital returned to normal glucose levels within 60 min post-injection, while blood glucose remained elevated under the influence of the other anesthetics. The most pronounced increase was induced by ketamine/xylazine, followed by avertin, with EC causing a relatively lower rise65,66.

When comparing the effects of the three anesthetics on visual electrophysiological outcomes, AR’s performance was also the closest to KX. Based on previous studies67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102,103,104,105,106, we compiled amplitude for various waves from four visual electrophysiology experiments conducted using KX (Table S2). It is important to note that while some wave peaks, such as the scotopic ERG b-wave and PERG, were slightly lower in the KX group compared to the AR group, this may be attributed to differences in experimental conditions, electrode materials, and recording positions. Therefore, evaluating the efficacy of anesthetics solely through amplitude comparisons across studies is not comprehensive. Notably, unlike the EC and PS groups, mice anesthetized with KX did not exhibit any absence of typical responses in visual electrophysiological tests. Thus, compared to EC and PS, AR appears to be the most suitable alternative to KX as an anesthetic.

One of the particularly interesting phenomena observed in this study is the substantially reduced OPs amplitude in the scotopic ERG experiments for both the EC and PS groups. While the precise origin of OPs remains elusive, intravitreal injection of gamma-aminobutyric acid (GABA) specifically abolishes late OP components without substantially affecting early ones, a pattern closely mirroring the changes observed under PS and EC treatment107. PS affects the central nervous system through a series of complex mechanisms, including enhancing gamma-aminobutyric acid type A receptor function, inhibiting glutamate receptors, reducing voltage-gated sodium channel activity, activating potassium channels, and inhibiting acetylcholine receptors108,109,110,111,112. These combined actions contribute to its anesthetic effect. Studies on isolated rat retinas have shown that PS substantially reduces the activity of depolarizing bipolar cells without affecting photoreceptors. Moreover, secondary components of the ERG, including the scotopic threshold response and OPs, are strongly suppressed by PS109. These findings are consistent with our observations in PS-anesthetized mice, where the OPs in FERG were substantially suppressed and PERG, a key indicator of ganglion cell activity, was notably reduced. These results further confirm the profound impact of sodium pentobarbital on the electrical signals of specific retinal cells.

EC enhances the function of nicotinic acetylcholine, gamma-aminobutyric acid, and glycine receptors, while inhibiting N-methyl-D-aspartate receptor and alpha-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor receptors in a concentration-dependent manner113,114,115. Due to EC’s complex mechanism of action, accurately recording genuine electrical signals in mice under this anesthesia becomes exceptionally challenging. This difficulty arises from the extensive influence of the anesthetic on the mice’s nervous system, potentially leading to signal distortion or alteration, which makes it difficult to accurately reflect the true physiological state of the mice’s visual system.

AR indirectly activates gamma-aminobutyric acid type A and glycine receptors by modulating their agonists116. Compared to the direct mechanisms of EC and PS, this indirect regulation substantially reduces the potential impact on visual electrophysiological signal transmission. This likely explains why the FERG and PERG results in the AR group of mice were superior to those in the other two groups.

Another noteworthy observation is that, while FERG and PERG showed typical responses, the characteristic peaks of PVEP were absent in both the EC and PS groups, whereas FVEP performance remained normal. This may be related to the differences in the pathways involved in the processing of visual signals between pattern stimuli and flash stimuli. Pattern stimuli involve the cooperation of multiple brain regions, with the layers of the V1 region (L1 to L6) each playing distinct roles in this process, whereas the brain regions activated by flash stimuli are primarily concentrated in the V1 area. Specifically, pattern-reversal stimuli uniquely activate the cortex surrounding the intraparietal sulcus and the ventral occipital cortex, regions crucial for the analysis and integration of complex visual information 117,118. The intraparietal sulcus cortex is part of the dorsal visual pathway, responsible for processing spatial ___location, motion, and visually guided behaviors, with L5 of the V1 region interacting closely with this area. The ventral occipital cortex belongs to the ventral visual pathway, which handles object recognition, color, and shape analysis, with L2/3 and L6 of the V1 region interacting primarily with this area119. The PVEP results from the AR group, with peak values first increasing and then decreasing across depths, reflect this layer-specific functionality.

However, in the EC and PS groups, characteristic peaks were absent across all V1 depths, which may be related to the regulation of the brain’s GABAergic inhibitory system by PS and EC. This system plays a critical role in limiting the spread of layer-specific excitatory activity and maintaining topological projections120,121. When this system is disrupted, functional projections between layers may be affected, leading to weakened or absent VEP signals. Additionally, EC may suppress AMPA and NMDA receptors, thereby disrupting inter-layer functional projections and reducing neuronal population activity122. AR, on the other hand, indirectly activates GABA-A and glycine receptors by modulating their agonists116. This indirect regulation, as opposed to the direct mechanisms of EC and PS, likely explains why PVEP results in the AR group were unaffected.

Resting-state functional magnetic resonance imaging (rs-fMRI) further supports this conclusion, with AR and ketamine/xylazine showing favorable seed-based, global, and interhemispheric connectivity results, comparable to those of awake animals123. The connectivity results for ethyl carbamate were moderate117,118, and this suppression of brain activity was similarly observed in PS-anesthetized mice124.

The absence of characteristic peaks across all depths in the EC and PS groups suggests that these anesthetics may weaken VEP signals by disrupting the complex network connections between brain regions. Notably, the widespread nature of this effect—manifested by the lack of characteristic peaks across different depths—further strengthens the idea that EC and PS exert a nonspecific inhibitory effect on not only the visual system but also broader brain function.

Retinal dysfunction can extend to brain regions involved in visual information processing31,125,126,127,128,129, highlighting the importance of studying the visual system as an integrated whole. With advances in technology and deeper exploration, research in visual science has gradually expanded from focusing solely on the retina to encompassing the entire visual pathway, especially the brain’s visual processing areas. This multi-level research approach helps to uncover the complete mechanisms of visual information processing. However, the improper selection of anesthetics may suppress the function of brain regions responsible for visual processing, potentially leading to abnormal experimental outcomes. Therefore, the choice of anesthetics is particularly critical when conducting visual electrophysiological studies.

This study comprehensively evaluated the effects of three anesthetics on the visual signal pathway from the retina to the brain’s visual cortex under identical conditions. The responses to different stimulation patterns, including light and pattern stimuli, were analyzed in detail. The results indicate that EC and PS suppress visual signal transmission, particularly in the brain’s visual processing centers. AR proved to be the most suitable anesthetic among the three for studying visual electrophysiological signals, displaying distinct electrophysiological characteristics and yielding superior experimental results.

Materials and methods

Animals

In this study, 54 adult C57BL/6J mice, aged 2–2.5 months and weighing 20–26 g, were obtained from the Shanghai Model Organisms Center, Inc. The mice were housed under optimal laboratory conditions, with free access to water and timely provision of feed, in an environment maintained at a temperature range of 20–24 °C and relative humidity between 40 and 60%. A strict 12-h light/dark cycle was implemented to simulate natural conditions. Subjects were randomized into three study groups, in alignment with the 3R principles (Reduction, Replacement, and Refinement) that guide ethical animal research practices. The 54 mice were randomly divided into three experimental groups, each containing 18 mice. In each group, six mice were anesthetized using EC, AR, and PS, respectively, with no repeat anesthesia performed on the same mice. The study design for each group is as follows:

  1. (1)

    Group 1: Focused on a comprehensive assessment of retinal function, including scotopic and photopic ERG, as well as OPs detection. The sequence of experiments was scotopic ERG first, followed by photopic ERG, with OPs data collected synchronously during the scotopic ERG experiment. The data presented in Table 1 were derived from this group. The instrumentation employed for the experiments on mice in this group was the RetiMINER-C visual electrophysiology system (AiErXi Medical Equipment Co., Ltd., Chongqing, China);

  2. (2)

    Group 2: Examined the primary visual cortex’s response to simple flash stimuli, using FVEP detection. The equipment utilized for this group was the Celeris System (Diagnosys, Lowell, MA, USA);

  3. (3)

    Group 3: Focused on the retinal and visual cortex responses to patterned stimuli, with PERG and PVEP recorded simultaneously. The instrument used for this group’s investigations was also the Celeris System (Diagnosys, Lowell, MA, USA).

The experimental order followed a progression from light stimuli to patterned stimuli, with Group 1 conducted first, followed by Group 2, and then Group 3. Within each group, the anesthesia order followed the pattern: "EC, AR, PS, AR, PS, EC, PS, EC, AR." Specifically, the first mouse in each group was anesthetized with EC and the experiment was conducted; after completing the experiment, the second mouse in the group was anesthetized with AR for the next experiment; then, the third mouse was anesthetized with PS to complete its experiment, and so on, following the pre-determined order for anesthetizing and testing each mouse in the group. This approach maximized the simulation of randomization to minimize potential bias. Each recording session rigorously maintains the animal’s core body temperature between 37 °C and 38 °C. Ethical approval for the research protocol was granted by the Ethics Committee of Wuhan University of Science and Technology (wust202013). Throughout the study, all pertinent institutional and governmental standards concerning the ethical utilization of animals were rigorously observed. All experiments were conducted in accordance with the ARRIVE guidelines for the use of animals in ophthalmic and vision research.

Flash electroretinogram

Prior to electrophysiological recordings, mice were dark-adapted overnight and subsequently handled under subdued red illumination.We anesthetized the mice using intraperitoneal injections of ethyl carbamate (EC, 0.2 g/ml, Sigma-Aldrich, #U2500), avertin (AR, also known as 2,2,2-tribromoethanol, 0.02 g/ml, Sigma-Aldrich, #T48402), and pentobarbital sodium (PS, 0.01 mg/ml, Sigma-Aldrich, #P3761). The doses were 1.6 g/kg, 0.3 g/kg, and 0.115 g/kg, respectively. The total injection volume for each mouse ranged from 0.18 to 0.39 μL. Following the induction of anesthesia, the state of anesthesia was preliminarily assessed via a toe pinch reflex test. Needle electrodes were then strategically placed on the skin between the tail and both ears to capture and monitor the electrical signals (Fig. S1). In the formal experiment, the placement of the ground and reference electrodes will effectively shield the EEG and ECG signals, ensuring they do not interfere with the recorded visual electrophysiological signals. This electrophysiological feedback was utilized to confirm the attainment of a deep anesthetic state. Unless otherwise specified, all subsequent experiments will adhere to the standardized anesthesia confirmation protocol established in this study.

Upon verification of adequate anesthesia, mydriasis was induced using an ophthalmic solution containing tropicamide, phenylephrine hydrochloride, and atropine sulfate monohydrate. Hydroxypropyl methylcellulose was used to ensure that the mouse cornea remained hydrated throughout the experiment. A single application of hydroxypropyl methylcellulose can maintain corneal moisture for approximately 40 min. Since the experiment lasts about 60 min, hydroxypropyl methylcellulose was reapplied around the 35-min mark to ensure continuous corneal hydration. After application, the mice were given a 3-min adaptation period to allow sufficient absorption of the moisturizer by the eyes and to regain stability before the experiment continued. The flash electroretinogram was conducted by applying a silver wire loop electrode directly onto the corneal surface, a reference electrode on the forehead, and a ground clip on the tail. The impedance of the recording electrod was maintained within the range of 1.5 and 2 kΩ. Scotopic electroretinogram were elicited using a spectrum of white light flashes from − 3.6 to 2.15 log cd s/m2, with a flash duration of 1 ms and an average of three flashes. The interstimulus interval was adjusted based on light intensity, exceeding 10 s for lower intensities and 30 s for higher ones. A- and b-wave amplitudes were quantified from baseline to trough and trough to peak, respectively.

Following the scotopic electroretinogram trials, the mice underwent a 10-min adaptation under a 30 cd/m2 white light. Subsequently, their retinal responses to escalating light stimuli (− 0.6–1.4 log cd s/m2) were assessed under the same background illumination. The stimuli were applied at 5-s intervals, with each response quantified as the mean of three responses.

The photic stimuli were delivered using a bespoke LED apparatus, designed to uniformly project a homogenously distributed field of light onto the retina, with the white LED wavelength calibrated at 505 nm. The resultant bioelectrical signals were subjected to amplification through a precision direct current amplifier (DC Amplifier, GIMAmp, Chongqing IRC Medical Equipment Co., Ltd), followed by band-pass filtration within the frequency range of 0.3–300 Hz. The data were digitized at a sampling rate of 2000 Hz and processed using the RetiMINER-C visual electrophysiology system (AiErXi Medical Equipment Co., Ltd., Chongqing, China). Prior to each experimental session, calibration of the illumination intensity and amplifier adjustments were meticulously conducted using a photometric power meter (AQ2170, YOKOGAWA) and an electronic signal generator (FNIR-SISG-004A, FNIRSI), respectively.

Pattern electroretinogram and pattern visual evoked potential

Both PERG and PVEP experiments utilize the Celeris System (Diagnosys, Lowell, MA, USA) for consistent stimulation: a black and white horizontal striped stimulus with 2 Hz phase reversals, a spatial frequency of 0.06 cycles/degree, 100% contrast, and an intensity of 50 cd/m2. Each result is averaged over 200 repetitions. Following anesthesia, mydriasis in mice is induced with an ophthalmic solution of tropicamide, phenylephrine hydrochloride, and atropine sulfate monohydrate, while hydroxypropyl methylcellulose maintains corneal hydration. In PERG experiments, electrodes for recording, reference, and grounding are placed subcutaneously at the cornea, nasal bone, and tail, respectively. For PVEP, the recording electrode is placed in the occipital region, with reference and grounding electrodes matching the PERG setup.The impedance of the recording electrod was maintained within the range of 1.5 and 2 kΩ.

Flash visual evoked potential

The flash visual evoked potential (FVEP) recordings were conducted using the Celeris System (Diagnosys, Lowell, MA, USA), adhering strictly to the standardized procedures outlined in previous research130. Following the induction of anesthesia and confirmation of a deep anesthetic state, the cranial fur was meticulously shaved, and the skin was precisely incised and retracted to expose the lambdoidal suture. A microdrill with a 0.5 mm diameter was then used to carefully perforate the skull at the exact intersection of the sagittal and lambdoidal sutures.

The recording electrodes were inserted into the cerebral cortex at an optimized angle of 30–45 degrees through a hole until the impedance stabilized between 1.5 and 2 kΩ. At this point, the vertical depth (i.e., Z-axis depth) that the recording electrode penetrated into the cortex was approximately 1.5–1.6 mm. Ground and reference electrodes were positioned subcutaneously at the tail and snout, respectively. Throughout data acquisition, the impedance of the ground, reference, and recording electrodes was maintained within the 1.5–2 kΩ range to ensure stable and reliable signal quality.

A standardized flash intensity of 0.05 cd s/m2 was applied uniformly across all experimental sessions, with each eye independently exposed to 100 white light flashes at a frequency of 1 Hz via corneal stimulators. The elicited responses were efficiently captured within a 300 ms window, with a sampling rate of 2000 Hz to detect subtle potential changes. During signal processing, carefully selected filtering parameters were employed, including a low-frequency cutoff of 1.25 Hz and a high-frequency cutoff of 100 Hz, effectively eliminating noise and enhancing the signal-to-noise ratio and resolution.

Curve fitting

To more precisely and quantitatively characterize the electrophysiological differences, the electroretinogram (ERG) data were analyzed using the Lamb and Pugh model (1)131 and the Naka-Rushton Eq. (2)46. In the Lamb and Pugh model, the parameters R, I, S, T, and Td denote ERG amplitude, stimulus retinal illumination, response sensitivity, time since stimulus onset, and a minor delay before response onset, respectively; Rmax is the saturating amplitude of ERG, reflecting the maximal response capacity of retinal photoreceptors and indicating their light sensitivity. In the Naka-Rushton equation, K represents the half-saturation constant, indicating the light intensity required to elicit a half-maximal response, while N dictates the slope of the function, implying the retina’s dynamic range of response to varying light intensities132,133.

$$R=\left\{\left.1-{e}^{\left[-I\times S\times \left.{\left(T-\right.\left.{T}_{d}\right)}^{2}\right]\right.}\right\}\times {R}_{max}\right.$$
(1)
$$R=\frac{{I}^{N}}{{I}^{N}+{K}^{N}}\times {R}_{max}$$
(2)

Statistical analyses

In MATLAB, perform precise fitting of scotopic and photopic ERG experimental data to extract key parameters of the Lamb-Pugh model and the Naka-Rushton equation, such as the maximum response (Rmax), time delay (Td), and the half-saturation constant (S). Data were presented as mean ± standard error of the mean (SEM), processed using GraphPad Prism and Origin software. Variance homogeneity was verified via an F-test to determine the independence of group datasets (DFn = 2, DFd = 15). An F-value exceeding 3.682 was interpreted as statistically significant, warranting subsequent t-test comparisons. Differences among groups were considered statistically significant at *p < 0.05, **p < 0.01, ***p < 0.001, as determined by two-way ANOVA tests with Tukey’s multiple comparisons.