Here, we describe a standard protocol for quantifying the optokinetic reflex. It combines virtual drum stimulation and video-oculography, and thus allows precise evaluation of the feature selectivity of the behavior and its adaptive plasticity.
The optokinetic reflex (OKR) is an essential innate eye movement that is triggered by the global motion of the visual environment and serves to stabilize retinal images. Due to its importance and robustness, the OKR has been used to study visual-motor learning and to evaluate the visual functions of mice with different genetic backgrounds, ages, and drug treatments. Here, we introduce a procedure for evaluating OKR responses of head-fixed mice with high accuracy. Head fixation can rule out the contribution of vestibular stimulation on eye movements, making it possible to measure eye movements triggered only by visual motion. The OKR is elicited by a virtual drum system, in which a vertical grating presented on three computer monitors drifts horizontally in an oscillatory manner or unidirectionally at a constant velocity. With this virtual reality system, we can systematically change visual parameters like spatial frequency, temporal/oscillation frequency, contrast, luminance, and the direction of gratings, and quantify tuning curves of visual feature selectivity. High-speed infrared video-oculography ensures accurate measurement of the trajectory of eye movements. The eyes of individual mice are calibrated to provide opportunities to compare the OKRs between animals of different ages, genders, and genetic backgrounds. The quantitative power of this technique allows it to detect changes in the OKR when this behavior plastically adapts due to aging, sensory experience, or motor learning; thus, it makes this technique a valuable addition to the repertoire of tools used to investigate the plasticity of ocular behaviors.
In response to visual stimuli in the environment, our eyes move to shift our gaze, stabilize retinal images, track moving targets, or align the foveae of two eyes with targets located at different distances from the observer, which are vital to proper vision1,2. Oculomotor behaviors have been widely used as attractive models of sensorimotor integration to understand the neural circuits in health and disease, at least partly because of the simplicity of the oculomotor system3. Controlled by three pairs of extraocular muscles, the eye rotates in the socket primarily around three corresponding axes: elevation and depression along the transverse axis, adduction and abduction along the vertical axis, and intorsion and extorsion along the anteroposterior axis1,2. Such a simple system allows researchers to evaluate the oculomotor behaviors of mice easily and accurately in a lab environment.
One prime oculomotor behavior is the optokinetic reflex (OKR). This involuntary eye movement is triggered by slow drifts or slips of images on the retina and serves to stabilize retinal images as an animal's head or its surroundings move2,4. The OKR, as a behavioral paradigm, is interesting to researchers for several reasons. First, it can be stimulated reliably and quantified accurately5,6. Second, the procedures of quantifying this behavior are relatively simple and standardized and can be applied to evaluate the visual functions of a large cohort of animals7. Third, this innate behavior is highly plastic5,8,9. Its amplitude can be potentiated when repetitive retinal slips occur for a long time5,8,9, or when its working partner vestibular ocular reflex (VOR), another mechanism of stabilizing retinal images triggered by vestibular input2, is impaired5. These experimental paradigms of OKR potentiation empower researchers to unveil the circuit basis underlying oculomotor learning.
Two non-invasive methods have primarily been used to evaluate the OKR in previous studies: (1) video-oculography combined with a physical drum7,10,11,12,13 or (2) arbitrary determination of head turns combined with a virtual drum6,14,15,16. Although their applications have made fruitful discoveries in understanding the molecular and circuit mechanisms of oculomotor plasticity, these two methods each have some drawbacks which limit their powers in quantitatively examining the properties of the OKR. First, physical drums, with printed patterns of black and white stripes or dots, do not allow easy and quick changes of visual patterns, which largely restricts the measurement of the dependence of the OKR on certain visual features, such as spatial frequency, direction, and contrast of moving gratings8,17. Instead, tests of the selectivity of the OKR to these visual features can benefit from computerized visual stimulation, in which visual features can be conveniently modified from trial to trial. In this way, researchers can systematically examine the OKR behavior in the multi-dimensional visual parameter space. Moreover, the second method of the OKR assay reports only the thresholds of visual parameters that trigger discernible OKRs, but not the amplitudes of eye or head movements6, 14,15,16. The lack of quantitative power thus prevents analyzing the shape of tuning curves and the preferred visual features, or detecting subtle differences between individual mice in normal and pathological conditions. To overcome the above limitations, video-oculography and computerized virtual visual stimulation had been combined to assay the OKR behavior in recent studies5,17,18,19,20. However, these previously published studies did not provide enough technical details or step-by-step instructions, and consequently it is still challenging for researchers to establish such an OKR test for their own research.
Here, we present a protocol to precisely quantify the visual feature selectivity of OKR behavior under photopic or scotopic conditions with the combination of video-oculography and computerized virtual visual stimulation. Mice are head-fixed to avoid the eye movement evoked by vestibular stimulation. A high-speed camera is used to record the ocular movements from mice viewing moving gratings with changing visual parameters. The physical size of the eyeballs of individual mice is calibrated to ensure the accuracy of deriving the angle of eye movements21. This quantitative method allows comparing OKR behavior between animals of different ages or genetic backgrounds, or monitoring its change caused by pharmacological treatments or visual-motor learning.
All experimental procedures performed in this study were approved by the Biological Sciences Local Animal Care Committee, in accordance with guidelines established by the University of Toronto Animal Care Committee and the Canadian Council on Animal Care.
1. Implantation of a head bar on top of the skull
NOTE: To avoid the contribution of VOR behavior to the eye movements, the head of the mouse is immobilized during the OKR test. Therefore, a head bar is surgically implanted on top of the skull.
2. Setup of the virtual drum and video-oculography
3. Calibration of eye movements
NOTE: Rotational eye movements are calculated based on movements of the pupil and the radius of the orbit of the pupillary movements (Rp, the distance from the center of the pupil to the center of the eyeball). For each individual mouse, this radius is measured experimentally21.
4. Record eye movements of the OKR
5. Analysis of eye movements of the OKR with the eye analysis software
With the procedure detailed above, we evaluated the dependence of the OKR on several visual features. The example traces shown here were derived using the analysis codes provided in Supplementary Coding File 1, and the example traces raw file can be found in Supplementary Coding File 2. When the drum grating drifted in a sinusoidal trajectory (0.4 Hz), the animal's eye automatically followed the movement of the grating in a similar oscillatory manner (Figure 3B top panel), which is characteristic of the OKR behavior2,5,8. The amplitude of OKR eye movements in the horizontal axis was derived with fast Fourier transform (Figure 3C&D), and the OKR gain was calculated as the ratio of the amplitude of eye movements to the amplitude of grating movement (Figure 3C). The OKR gain varied with the values of spatial frequency, oscillation frequency, and the direction of movement grating (Figure 4A). First, the spatial frequency tuning curve of the OKR behavior had an inverted V-shape and peaked at an intermediate spatial frequency of 0.16 cpd (Figure 4A left). Second, the oscillation frequency tuning curve monotonically declined as the oscillation frequency of the drum grating increased (Figure 4A middle), indicating that the OKR behavior works best in response to the low-speed visual motion4. Both the amplitude and shape of the oscillation frequency tuning curve varied when presenting gratings of different spatial frequencies17. Third, the horizontal OKR could also be induced by gratings moving in different directions (Figure 4A right). The strongest horizontal OKR behavior was elicited by the temporo-nasal motion (0°). The OKR gain dropped to ~80%, or ~30% of the maximum, when the grating moved in oblique angles 30° or 60° deviated from the temporo-nasal direction (both upward and downward), respectively, and the horizontal OKR disappeared when the grating moved vertically upward or downward (90° and 270°). Moreover, the shapes of the tuning curves were influenced by the level of luminance. For example, animals performed the OKR behavior well in response to spatial frequencies of 0.16 and 0.32 cpd under the photopic condition, but the spatial frequency tuning curve shifted leftward under the scotopic condition (Figure 4B). To analyze the shape of tuning curves, we fit them with appropriate mathematical functions. For example, the Gaussian function was used to estimate the peak and bandwidth of spatial frequency tuning (Figure 4C). With this analysis, we found that the tuning curve under the scotopic condition had a lower value in the preferred spatial frequency in comparison to the photopic condition. The procedure detailed above can also be used to quantify the plasticity of OKR behavior. Following 45 min of continuous OKR stimulation, the amplitude of the OKR behavior was significantly potentiated (Figure 4D), consistent with previous reports. These results demonstrate the applications of this protocol in examining the oculomotor behaviors and potentials in understanding the brain circuits involved in these behaviors.
Figure 1: Setup of the OKR rig. (A) Dimensions of the head bar. (B) Back view (left) and top view (middle) of the virtual drum system. Three monitors are mounted orthogonally to each other. The head of a mouse is placed at the center of the square enclosure and faces forward. The period of visual stimulation (space between stripes) is varied based on the azimuth to the animal's eye to ensure the projection of the grating on the eye has a constant spatial frequency. In other words, the spatial frequency of the grating is perceived as constant throughout the visual field, as if the grating drifts along the surface of a virtual drum (right). (C) Setup of IR video-oculography. Left: the position of the camera when fixed at the center during OKR recording. Blue arrow: along the optical axis. Green arrow: perpendicular to the optical axis. Right: the rotation of the camera during eye calibration. (D) Position of the X-CR and Y-CR LEDs fixed on the camera. (E) Levels of the left and right eyes (left), and the nasal and temporal corners of the eyes are aligned horizontally (right) by rotating the horizontal rod or the headplate adapter, respectively (red arrows). (F) Photo of the OKR station. Note that the OKR rig is placed inside a customized enclosure with a black curtain. Abbreviations: IR = infrared; CR = corneal reflection. Please click here to view a larger version of this figure.
Figure 2: Calibration and measurement of eye positions in video-oculography. (A) Schematic of the calibration. The radius of pupil rotation (Rp) is estimated by rotating the camera to the leftmost position (-10°, left panel) and to the rightmost position (10°, right panel). Red dots indicate the positions of corneal reflection X-CR when the camera is placed to the leftmost and rightmost positions. Blue dots indicate the centers of the pupils. Green bars indicate the distances between the corneal reflection and pupil center viewed in the camera video (PCR – PP). (B) Dependence of Rp on pupil size. Top: schematics of eyeballs with a small or large pupil. Bottom: relationship between Rp and the diameter of the pupil of an example mouse. The pupil size is changed by varying the luminance (10 values in the range of 0-160 cd/m2) to the mouse. Black dots: the data that are used for linear fitting. Blue dots: outliers that are excluded from linear fitting. Red curve: the best-fit line in linear regression. Note that the Rp is inversely proportional to the pupil diameter. (C) Calculation of the angle of eye position when the eye has moved to the right or left sides of the optic axis. Red dots, blue dots, and green bars have the same meanings as in A. Please click here to view a larger version of this figure.
Figure 3: Calculation of OKR gain. (A) Top: snapshots of nasal (N; left) and temporal (T; right) eye positions taken during OKR stimulation. Red ellipses: fitting to pupil profile. Red crosses: pupil centers. White arrows: corneal reflection of a reference LED. Middle: the trajectories of eye movements with (black) or without (red) a median filter (filter window = 0.05 s) to remove high-frequency noise. Bottom: estimation of the eye velocity by calculating the first order derivative of eye movements. Saccades (red arrows) are detected with a velocity threshold of 50 °/s. (B) The trajectory of slow eye movements of the OKR after removing the saccades/nystagmus (black) overlaid with drum trajectory22. Top: oscillatory drum motion with an amplitude of 5° and an oscillation frequency of 0.4 Hz. Bottom: unidirectional (temporo-nasal) drum motion with a constant velocity of 6.25 °/s. (C) Left: the cycle-average of the eye trajectory in B top. Right: the frequency analysis of the eye or drum grating movements by fast Fourier transform. It should be noted that the drum grating oscillates at 0.4 Hz, and thus the amplitudes of eye and drum grating movement peak at 0.4 Hz (star marks). OKR gain is the ratio of the amplitudes of eye and drum grating movement at 0.4 Hz. (D) Top: curve fitting of the eye trajectory in B top with sinusoidal function. Bottom: relationship of the amplitude of eye movements derived by the fast Fourier transform method and derived from sinusoidal curve fitting. Yellow dot: example in the top. Please click here to view a larger version of this figure.
Figure 4: The visual feature selectivity and plasticity of OKR behaviors. (A) The visual feature selectivity of OKR gain under the photopic condition. Left: spatial frequency tuning curve from one animal (oscillation frequency: 0.4 Hz; trajectory: oscillating horizontally; mean luminance: 40 cd/m2; n = 15). Middle: oscillation frequency tuning curve from one animal (spatial frequency: 0.08 or 0.16 cpd; trajectory: oscillating horizontally; mean luminance: 40 cd/m2; n = 15). Right: direction tuning curve from one animal (spatial frequency: 0.16 cpd; temporal frequency: 1 Hz; mean luminance: 45 cd/m2; n = 24). The red arrow and bar indicate the temporo-nasal direction. For spatial and temporal/oscillation frequency tuning, a vertical drum grating presented on three monitors moves horizontally at a constant velocity or in an oscillatory manner. For direction tuning, a grating presented on only the right monitor moves in one of 12 directions at a constant velocity. Thickness: standard error of the mean (SEM). (B) Spatial frequency tuning curve of OKR gain from one animal under the scotopic or photopic condition. Photopic: oscillatory motion; oscillation frequency: 0.2 Hz; mean luminance: 40 cd/m2; n = 15. Scotopic: linear motion with a constant velocity; temporal frequency: 0.25 Hz; mean luminance: 8 x 10-5 cd/m2; n = 16. The scotopic condition is achieved by covering the monitors with five layers of Lee filter (299 1.2 ND). Thickness: SEM. (C) Gaussian fitting of spatial frequency tuning of OKR gain under photopic and scotopic conditions. Photopic: oscillatory motion; oscillation frequency: 0.2 Hz; mean luminance: 40 cd/m2; n = 15. Scotopic: linear motion with a constant velocity; temporal frequency: 0.25 Hz; mean luminance: 8 x 10-5 cd/m2; n = 16. (D) OKR potentiation of one mouse induced by 45 min of continuous OKR stimulation. Spatial frequency: 0.1 cpd; oscillation frequency: 0.4 Hz; mean luminance: 35 cd/m2; n = 40. Top: cycle averaged OKR trajectories before and after OKR potentiation. Thickness: SEM. Please click here to view a larger version of this figure.
Supplementary Coding File 1: Analysis code used for generating the example traces. Please click here to download this File.
Supplementary Coding File 2: Example traces generated with the software. Please click here to download this File.
The method of the OKR behavioral assay presented here provides several advantages. First, computer-generated visual stimulation solves the intrinsic issues of physical drums. Dealing with the issue that physical drums do not support the systematic examination of spatial frequency, direction, or contrast tuning8, the virtual drum allows these visual parameters to be changed on a trial-by-trial basis, thus facilitating a systematic and quantitative analysis of the feature selectivity of the OKR behavior (Figure 4A); while physical drums suffer from a non-uniform illumination by an external light source23, the virtual drum can readily provide homogeneous luminance across its surface; with the help of ND filters and a luminance meter, the computer-generated visual stimulation allows the OKR measurement at various well-controlled luminance levels from scotopic to photopic (Figure 4B), which is difficult to be done with physical drums. Without the limitation in the acceleration of physical drums due to their inertial mass, the virtual visual stimulation can achieve ideally accurate trajectories, especially at high acceleration and high velocity. Moreover, computer-generated visual stimulation allows the creative design of other types of visual stimulation, such as coherently moving dots, which helps investigate mechanisms of various oculomotor behaviors. Second, our procedure is standardized and thus requires minimal efforts to monitor the progress of behavioral recording, which provides an opportunity to examine multiple mice simultaneously. Therefore, it is suitable for studies involving a large cohort of animals (tens to hundreds of animals). Third, the high precision and quantitative power make it possible to compare repetitive OKR measurements of the same mice in longitudinal studies24, under different pharmacological treatments10, or under neural circuit perturbations5. Last, the Fourier transform-based analysis in the frequency domain5,7,9 gives equivalent results in the amplitude of oscillatory eye movements to the fitting-based analysis in the temporal domain12,25,26 (Figure 3D bottom), demonstrating that the analysis method presented here is both accurate and precise.
Our method also provides an opportunity to study OKR plasticity, a wildly used paradigm to investigate the mechanisms of oculomotor learning. When presenting continuous OKR stimulation to a mouse or surgically lesioning its vestibular organ, the amplitude of the OKR can be potentiated8,9. The OKR assay presented here is sensitive enough to capture small changes in eye movements happening in OKR potentiation (Figure 4D). The quantitative power of this method makes it possible to correlate behavioral changes with neural circuit dynamics, which is essential to reveal the mechanisms underlying oculomotor learning5,8,9,13.
To ensure the accuracy of OKR measurement, there are a few critical steps. First, during the surgery, extra care is needed to avoid superglue and dental acrylic touching the eyelid, which may damage the cornea or partially occlude the eye opening. Second, the strength of the OKR is affected by the behavioral state of mice27,28. Thus, a few rounds of accommodation are recommended to minimize the impact of stress on OKR measurement; also, the disturbance caused by odorants, noise, or light should be prevented during the recording. Finally, the heads of the mice should be oriented properly so that the line connecting the two corners of eyes is parallel to the horizontal axis. This guarantees that the direction of visual motion is aligned with axis of adduction and abduction eye movements. Thirdly, the uniform illumination of the eye is key to generating a sharp image of the pupil and in turn assuring high-quality eye tracking.
It is worth noting that there are a few limitations of the methods presented here. First, when an animal's eye blinks or opaque eye discharge blocks the pupil, the video-oculography loses track of the eye momentarily or permanently. Similarly, it cannot be used to monitor eye movements when the eyelids are sutured. Second, the temporal resolution of the video-oculography is limited by the frame rate of cameras to a range of 4-20 ms. Lastly, the head-fixed preparation does not allow monitoring the ocular behaviors of freely moving animals.
The video-oculography and virtual drum presented here had been successfully applied to characterize the visual feature selectivity and the plasticity of OKR behavior, and to understand the retinal and central circuits involved in mediating and adaptively modulating this behavior. In addition, they can also benefit studies in which other ocular behaviors are either the topics or even confounding factors of the neural phenomena. For example, the video-oculography can be used to monitor pupil dilation29 and saccade-like eye movements30,31, which are indicative of alertness and brain state32,33,34,35. Moreover, the calibration and measurement procedures outlined here are universally applicable for monitoring eye movements with a head-mounted camera in freely moving mice.
The authors have nothing to disclose.
We are thankful to Yingtian He for sharing data of direction tuning. This work was supported by grants from the Canadian Foundation of Innovation and Ontario Research Fund (CFI/ORF project no. 37597), NSERC (RGPIN-2019-06479), CIHR (Project Grant 437007), and Connaught New Researcher Awards.
2D translational stage | Thorlabs | XYT1 | |
Acrylic resin | Lang Dental | B1356 | For fixing headplate on skull and protecting skull |
Bupivacaine | STERIMAX | ST-BX223 | Bupivacaine Injection BP 0.5%. Local anesthesia |
Carprofen | RIMADYL | 8507-14-1 | Analgesia |
Compressed air | Dust-Off | ||
Eye ointment | Alcon | Systane | For maintaining moisture of eyes |
Graphic card | NVIDIA | Geforce GTX 1650 or Quadro P620. | For generating single screen among three monitors |
Heating pad | Kent Scientific | HTP-1500 | For maintaining body temperature |
High-speed infrared (IR) camera | Teledyne Dalsa | G3-GM12-M0640 | For recording eye rotation |
IR LED | Digikey | PDI-E803-ND | For CR reference and the illumination of the eye |
IR mirror | Edmund optics | 64-471 | For reflecting image of eye |
Isoflurane | FRESENIUS KABI | CP0406V2 | |
Labview | National instruments | version 2014 | eye tracking |
Lactated ringer | BAXTER | JB2324 | Water and energy supply |
Lidocaine and epinephrine mix | Dentsply Sirona | 82215-1 | XYLOCAINE. Local anesthesia |
Luminance Meter | Konica Minolta | LS-150 | for calibration of monitors |
Matlab | MathWorks | version xxx | analysis of eye movements |
Meyhoefer Curette | World Precision Instruments | 501773 | For scraping skull and removing fascia |
Microscope calibration slide | Amscope | MR095 | to measure the magnification of video-oculography |
Monitors | Acer | B247W | Visual stimulation |
Neutral density filter | Lee filters | 299 | to generate scotopic visual stimulation |
Nigh vision goggle | Alpha optics | AO-3277 | for scotopic OKR |
Photodiode | Digikey | TSL254-R-LF-ND | to synchronize visual stimulation and video-oculography |
Pilocarpine hydrochloride | Sigma-Aldrich | P6503 | |
Post | Thorlabs | TR1.5 | |
Post holder | Thorlabs | PH1 | |
PsychoPy | open source software | version xxx | visual stimulation toolkit |
Scissor | RWD | S12003-09 | For skin removal |
Superglue | Krazy Glue | Type: All purpose. For adhering headplate on the skull |