Representational drift gates critical-period plasticity in mouse visual cortex

SUMMARY

Brief monocular deprivation during a developmentally critical period, but not thereafter, shifts cortical responses toward the non-deprived eye. The characteristics of neural circuitry that permit this experience-dependent plasticity are poorly understood. Here, we performed repeated calcium imaging at cellular resolution to track the tuning properties of populations of excitatory layer 2/3 neurons in the visual cortex of juvenile mice during the critical period, adult mice after the critical period, and adult nogo-66 receptor (ngr1) mutant mice that retain critical-period plasticity. The instability of tuning for populations of neurons, termed “representational drift,” was significantly greater during the critical period than in adulthood. Adult ngr1 mutant mice displayed representational drift similar to that of juvenile mice. We propose that representational drift adapts the tuning of populations of neurons to recent experience during the critical period.

Graphical Abstract

In brief

Brown and McGee employ in vivo calcium imaging to measure representational drift during and after visual circuit maturation. The instability of neuronal tuning is greater during the developmental critical period, and the ngr1 gene increases neuronal tuning stability in adulthood. This tuning instability gates adaptation to recent visual experience.

RESULTS AND DISCUSSION

Experience refines the functions of neurons in the central nervous system, particularly during developmental “critical periods” when neural circuits are most plastic.¹ This experience-dependent plasticity is well studied in the visual system, where the tuning properties of neurons can be quantified from their responses to precise visual stimuli. Neurons in primary visual cortex (V1) are tuned for ocular dominance (OD) in many mammalian species, including primates, felines, and rodents. Brief abnormal visual experience that alters OD tuning is a classic model for experience-dependent plasticity confined to a critical period that is conserved across these species.^2–6 In the mouse, the critical period for OD plasticity spans postnatal day (P) 19 to ~32.² Closing one eye by lid suture (monocular deprivation, MD) for a few days only during this interval disrupts normal OD by shifting the responses of neurons in V1 toward the open (non-deprived) eye.² Many genes, signaling pathways, and forms of synaptic plasticity are required for OD plasticity.⁷ In contrast, the features of neuronal circuits that mediate this experience-dependent plasticity and/or confine it to the critical period remain poorly understood.

Previously, we determined how OD plasticity alters the tuning of excitatory neurons in layer (L) 2/3 of V1 by measuring their response properties both before and after MD during the critical period.⁸ MD shifts the overall OD of the neuronal population through a complex reorganization of the OD tuning of populations of neurons. The result is fewer neurons respond to the eye receiving MD and the magnitude of responses to that eye is reduced.

Recent studies have revealed a surprising instability of neuronal tuning properties in multiple brain regions, a phenomenon termed “representational drift.”^9,10 However, the physiologic role and significance of representational drift are unclear.¹¹ Here, we tested the hypothesis that greater representational drift is a defining circuit feature of plasticity during the critical period.

To determine the relationship between the instability of neuronal tuning in V1 and OD plasticity, first we confirmed that OD plasticity in response to brief MD is restricted to the critical period, as measured with calcium imaging at cellular resolution. We performed single-time-point imaging on the V1 of alert wild-type (WT) mice expressing the fluorescent calcium sensor GCaMP6s in excitatory cortical neurons.¹² We calculated the tuning properties of L2/3 neurons from responses to a battery of static sinusoidal gratings spanning both orientation (spaced at 30 degrees) and spatial frequency (SF) (spaced at half-octaves) (Figures 1A–1H and S1).^8,12,13 Visual responsiveness and orientation/SF tuning were determined for all neurons for each eye independently. Ocular dominance index (ODI) values were calculated for each neuron that was visually responsive to stimuli presented to either eye. MD during the critical period decreased the average ODI per mouse from values that reflect the typical bias of non-deprived mouse V1 for stimuli presented to the contralateral eye (Figures 1I–1K).^8,14 This plasticity was confined to the critical period and was not detectable in adult mice (P60–90), as ODI values following 4 days of MD in adults were similar to those of non-deprived mice (Figures 1I and1H). These results are consistent with published studies that examined the effects of brief MD on OD with multi-unit electrophysiologic recordings,² as well as more indirect methods such as visually evoked potentials (VEPs) and optical imaging of intrinsic signals.^15,16

Figure 1. Calcium imaging at neuronal resolution detects OD plasticity confined to the developmental critical period.

(A) An example cranial window 3 mm in diameter centered over visual cortex. Upper left corner, coordinate axes. Scale bar, 1 mm.

(B) Wide-field calcium imaging of neural activity in response to a horizontal bar 30 degrees wide and 2 degrees high drifting top to bottom at 10 degrees per second. White square encapsulates imaging field presented in (E) and (F). Scale bar, 1 mm.

(C) Illustration of the visual stimulus. Sinusoidal gratings (not square as depicted) at 30 degree intervals of orientation and 0.028 to 0.48 cycles per degree of SF spaced at half octaves, as well as an iso-luminant gray screen, are presented randomly at 4 Hz for 10 min. Each combination of orientation and SF is presented 40 times on average (range 29–56).

(D) Illustration of calcium imaging setup. Alert mice are head-fixed freely running on a spherical treadmill floating on a column of air and positioned with monitor 35 cm away centered on azimuth at zero elevation. An occluder (not depicted) is placed in front of the contralateral or ipsilateral eye in separate imaging sessions. A camera records pupil diameter.

(E) Example reference image of visual cortex from an experiment. The imaging field is 750 × 500 μm. Scale bar, 100μm.

(F) White circles correspond to manually identified ROIs in the reference image (E). The ROI colored green (lower left corner) is presented in (G). Scale bar, 100μm.

(G) A representative trace of fluorescence over time for an ROI (gray, top) and the corresponding inferred spike rate (ISR) (red, middle). The timing of presentations of the preferred stimuli (90 degrees, 0.21 cycles per degree, cpd) are represented by vertical lines (black, bottom). Scale bar, 1 min and 240 stimuli.

(H) Heatmap of the ISR for all combinations of orientation and SF.

(I) Timeline of imaging and monocular deprivation.

(J) Average ODI for WT critical period (CP) mice that were either non-deprived (ND) or received 4 days of monocular deprivation (4MD) (WT CP ND, mean ODI = 0.48, n = 17 mice; WT CP 4MD P32 mice, mean ODI = −0.12, n = 8 mice; one-way ANOVA with Sidak’s multiple comparisons test). Experimental groups augmented from Brown and McGee.⁹

(K) Average ODI for adult WT mice after closure of CP that were either ND or 4MD (WT ND mean ODI = 0.45, n = 15 mice; WT 4MD mice, mean ODI = 0.35, n = 7 mice, one-way ANOVA with Sidak’s multiple comparisons test).

See also Figure S1.

Next, we leveraged the phenotype of mice lacking a functional gene for the nogo-66 receptor 1 (ngr1) to test how instability of neuronal tuning relates to the critical period. The critical period does not close in ngr1 mutants and they retain OD plasticity in response to brief MD as adults, as well as recovery of normal OD and acuity after longer durations of MD. The OD plasticity displayed by adult constitutive ngr1 “knockout” (KO) mice is indistinguishable from that exhibited by WT mice during the critical period, including the time course, magnitude, laminar progression, sensitivity to barbiturates, and sensitivity to benzodiazepines.^14,17–20 Adult ngr1 KO mice also display intracortical disinhibition after 1–2 days of MD; this is the first known circuit adaptation following MD during the critical period.^18,21 Thus, comparing the neuronal tuning of adult WT and ngr1 KO mice permits isolating the circuit features associated with OD plasticity from other forms of plasticity that may be coincident but unrelated to this critical period during development.

We confirmed that adult ngr1 KO mice display OD plasticity to brief MD with single-time-point calcium imaging. Mice (P60–90) receiving 4 days of MD possessed lower mean ODI values than non-deprived mice (Figure 2A). In the mouse, OD plasticity during the critical period is driven by a depression of aggregate neuronal responses to visual stimuli presented to the eye receiving MD.^16,22 At cellular resolution, OD plasticity results from a decrease in the average response amplitude for neurons responding to visual stimuli presented to the contralateral eye, an increase in the percentage of neurons that are predominantly monocular and only respond to visual stimuli presented to the ipsilateral eye (ODI = −1), a decrease in the percentage of neurons that are predominantly monocular and only respond to visual stimuli presented to the contralateral eye (ODI = 1), and a modest reduction in the ODI values of binocular neurons⁸ (Figure S2A). Adult mice lacking ngr1 displayed these same cellular mechanisms of OD plasticity (Figures 2B, 2C, and S2B). In addition, MD altered the distribution of neurons for adult ngr1 KO mice when plotted in as histograms of ODI or viewed as fields of neurons per mouse (Figures 2D–2I). Then, we tracked the tuning properties of excitatory neurons in L2/3 of V1 across an interval of 4 days for WT and ngr1 KO mice. We calculated the OD tuning for hundreds of neurons from juvenile non-deprived WT mice during the critical period, adult non-deprived WT mice, adult non-deprived ngr1 KO mice, and adult ngr1 KO mice receiving MD after the first imaging time point (Figure 3). In addition to neurons that were consistently visually responsive, some neurons that were not visually responsive (non-responsive, NR) during the first imaging session became responsive to visual stimuli, whereas others that were previously visually responsive became non-responsive. The overall distribution of OD did not change between time points for non-deprived juvenile WT mice, adult WT mice, or adult ngr1 KO mice, as the ratios of predominantly contralateral monocular neurons (C), binocular neurons (B), and predominantly ipsilateral monocular neurons (I) were quite similar at both time points (Figures 3A–3C). In contrast, adult ngr1 KO mice receiving 4 days of MD displayed OD plasticity that reduced the percentage of C neurons and increased the percentage of I neurons, consistent with the results from the single-time-point calcium imaging experiments (Figure 2).

Figure 2. OD plasticity of adult ngr1 KO mice at neuronal resolution.

(A) Average ODI per mouse for adult WT and ngr1 KO non-deprived (ND) mice and after 4 days of MD (4MD): (WT ND ODI = 0.45, n = 15 mice; WT 4MD mice, mean ODI = 0.35, n = 7 mice; ngr1 KO mean ODI = 0.45, n = 12 mice; ngr1 KO 4MD mice mean = 0.09 n = 10 mice, one-way ANOVA with Sidak’s multiple comparisons test). Horizonal lines represent means. WT mice are presented from Figure 1K for comparison

(B) dF/F of contralateral neurons from WT ND, WT 4MD, ngr1 KO ND, and ngr1 KO 4MD groups: (WT ND = 0.15, n = 1,222 neurons; WT 4MD = 0.16, n = 475 neurons; ngr1 KO ND = 0.17, n =1,100 neurons; ngr1 KO 4MD = 0.11 n = 550 neurons; one-way ANOVA Sidak’s multiple comparisons test).

(C) ODI values of binocular neurons from WT ND, WT 4MD, ngr1 KO ND, and ngr1 KO 4MD groups: (WT ND = 0.15, n = 391 neurons; WT 4MD = 0.16, n = 220 neurons; ngr1 KO = 0.17, n = 415 neurons; ngr1 KO 4MD mice mean = 0.08, n = 226 neurons; one-way ANOVA with Sidak’s multiple comparisons test).

(D) Histogram of neuronal ODI values for WT ND and 4MD groups: (WT ND mean ODI = 0.51, 1,403 neurons; WT 4MD mean ODI = 0.36, 564 neurons).

(E) Histogram of neuronal ODI values for ngr1 KO ND and 4MD groups: (ngr1 KO mean ODI = 0.46, 1,276 neurons; ngr1 KO 4MD mean ODI = 0.17, 767 neurons).

(F–I) Example fields of neurons for WT ND, WT 4MD, ngr1 KO ND, and ngr1 KO 4MD mice. Neurons are color coded by ODI. Red neurons respond to contralateral eye. Blue neurons respond to the ipsilateral eye. Other colors respond to stimuli from both eyes. The mean ODI for the imaging field is stated in the lower right corner. Scale bar, 100μm.

See also Figure S2.

Figure 3. Reorganization of tuning for binocularity for longitudinally tracked neurons.

(A) Sankey plots of ODI values for neurons imaged at P28 and again at P32 for CP ND animals. Vertical bars on left side represent neuronal ODI at first imaging session. The length of the bar is proportional to the size of the population with that tuning preference. The gray bar represents non-visually evoked neurons (NR), the red bar represents neurons predominantly contralateral (C), the green bar represents neurons that are binocular (B), and the blue bar represents predominantly ipsilateral (I) neurons. The lines connect to the tuning properties of neurons at second imaging session 4 days later (n = 494 neurons, n = 5 mice). The thickness of the line is proportional to the size of the population.

(B–D) Sankey plots of ODI values of adult WT ND (n =780 neurons, n = 6 mice), adult ngr1 KO ND mice (n = 595 neurons, n = 4 animals), and adult ngr1 KO 4MD mice (n = 1,062 neurons, n = 7 animals).

Interestingly, there was considerable reorganization of the OD tuning of individual neurons (C, B, or I) for non-deprived mice, irrespective of age or genotype, despite the conserved overall distribution of OD across the population of visually responsive neurons (Figure 3). Many neurons maintained stable tuning for OD at both imaging time points (stable), but other neurons displayed different tuning (plastic). We calculated the percentage of stable and plastic neurons for these four groups (Figure 4A). The percentage of plastic neurons with tuning instability was significantly greater in juvenile WT mice than adult WT mice. This finding is consistent with our prior study measuring the stability of OD tuning for juvenile WT mice following 4 days of MD and another prior study measuring the stability of tuning exclusively binocular neurons (B)²³ (Figure S3A). This circuit reorganization is only detectable with repeat imaging because the distribution of OD tuning is conserved between time points in non-deprived mice (Figure S3B). Interestingly, adult ngr1 KO mice exhibited a significantly greater percentage of plastic neurons than adult WT mice.

Figure 4. Response properties of longitudinally imaged neurons.

(A) The proportion of neurons with stable OD tuning (S) verses unstable OD tuning (plastic, P) from CP WT ND mice (n = 494; S = 150, P = 344), WT ND (n = 780; S = 290, P = 490), ngr1 KO ND (n = 594; S = 157 S, P = 431), and ngr1 KO 4MD (n = 1,029; S = 221 S, P = 808). Chi-squared tests.

(B–I) Tuning properties for stable and plastic predominantly contralateral neurons from WT ND (S = 175, P = 153), ngr1 KO ND (S = 87, P = 164), and ngr1 KO 4MD (S = 110, P = 298) groups from A.

(B) Signal:noise ratio (SNR): (WT ND S = 2.15, P = 2.05; ngr1 KO ND 2.36, P = 1.95; ngr1 KO 4MD S = 2.18, P = 2.12). Horizonal lines represent means.

(C) dF/F: (WT ND S1 = 0.14, S2 = 0.13, P = 0.12; ngr1 KO ND S1 = 0.18, S2 = 0.14, P = 0.13; ngr1 KO 4MD S1 = 0.13, S2 = 0.10, P = 0.12); mixed effects analysis for repeated measures and Brown-Forsythe and Welch ANOVA. Lines connect individual stable neurons. Horizonal lines represent means. (D) SF preference: (WT ND S = 0.045, P = 0.041; ngr1 KO ND S = 0.037, P = 0.038; ngr1 KO 4MD S = 0.036, P = 0.036). Horizonal lines represent medians. (E) Change in SF preference measured in octaves for stable neurons between time points (WT ND = 0.25; ngr1 KO ND = 0.21; ngr1 KO 4MD = 0.50; Brown-Forsythe and Welch ANOVA with Dunnett’s multiple comparisons test). Horizonal lines represent medians. (F) Scatterplot for preferred SF for stable neurons at both time points presented in E. (G) Orientation preference in degrees: (WT ND S = 92, P = 92; ngr1 KO ND S = 90, P = 92; ngr1 KO 4MD S = 94, P = 85). Horizonal lines represent medians. (H) Change in orientation preference measured in degrees for stable neurons between time points (WT ND = 18; ngr1 KO ND = 21; ngr1 KO 4MD = 48; Brown-Forsythe and Welch ANOVA with Dunnett’s multiple comparisons test). Horizonal lines represent means. (I) Scatterplot for orientation preference for stable neurons at both time points presented in (H).

See also Figure S3.

Last, we explored the features of neurons with OD tuning instability within the response strength and orientation/SF tuning of these populations. We formulated several discrete hypotheses before analyzing the data: (1) tuning instability reflects our criteria for visual responsiveness, (2) our measurements lack sensitivity to detect meaningful differences, and (3) plastic neurons represent only a subset of preferred orientation or SF. We confined our analysis to predominantly contralateral monocular neurons (C) because this was the population with the best sampling.

The inclusion criteria we employed to discriminate visually responsive neurons from non-responsive neurons is an intersectional method based on the ratio of the number of responses relative to presentations (spike ratio) and the ratio of signal to noise (SNR) for the visual stimulus encompassing the preferred SF and orientation for each neuron.^8,24 These criteria overlap with those applied by other groups measuring these tuning properties for excitatory neurons in mouse V1 with calcium imaging using sinusoidal gratings and the same calcium sensor GCaMP6s.^23,25,26 If tuning instability were a consequence of neurons straddling the inclusion criteria for visual responsiveness, then they would be clustered near these threshold values. However, neurons with OD tuning instability (plastic) displayed similar mean spike ratio and SNR to neurons with stable tuning (Figures 4B and S3). Critically, these plastic neurons, independent of genotype or deprivation, spanned the same range of magnitude for spike ratio, SNR, preferred SF, preferred orientation, and the average response amplitude (normalized change in fluorescence, dF/F) as stable neurons (Figures S3D–S3F).

To assess the sensitivity of our measurements, we compared the average response magnitude (dF/F) to the preferred stimuli for neurons with stable OD tuning to neurons with plastic OD tuning. Again, the distribution of the dF/F values were similar across genotypes (Figure 4C). Plastic neurons displayed lower average dF/F, but the effect size was modest. In contrast, stable neurons from ngr1 KO mice receiving 4 days of MD displayed the reduced dF/F amplitudes associated with critical-period OD plasticity, confirming the sensitivity of these measurements for detecting significant differences.⁸ Neurons with tuning instability for OD also possessed the same range of preferred SF and orientation, although MD increased the average magnitude of difference in preferred SF and orientation between time points for stable neurons in adult ngr1 mutant mice (Figures 4D–4I). Thus, the instability of tuning for OD cannot be attributed to neurons with weaker responses nor predicted from tuning for orientation or SF.

In summary, the tuning of a significant fraction of excitatory neurons in L2/3 of mouse V1 is unstable. The fraction of neurons with tuning instability is larger during the critical period for both non-deprived mice and those receiving MD, as well as in adult mice lacking a gene required to close the critical period. In the presence of consistent visual experience, the magnitude of tuning instability is not evident because the distribution of tuning properties is unaltered at the population level despite the interchange of tuning by individual neurons. Yet, perturbing vision with MD reveals OD plasticity in mice with greater tuning instability. We propose that this representational drift adapts the tuning of populations of neurons to recent experience during the critical period.

Beginning at eye opening (~P14), visual experience alters the tuning of neurons in V1 to increase the populations of neurons with predominantly ipsilateral tuning or binocular tuning.²⁶ The proportions of contralateral neurons, binocular neurons, and ipsilateral neurons approach adult levels by the close of the critical period.²⁷ During the critical period, binocular matching of preferred orientation improves as binocular neurons with poor matching lose responsiveness to one eye to become predominantly monocular while previously predominantly monocular neurons gain responsiveness to the other eye with more similar tuning for orientation.²³ This is also the period of heightened sensitivity for OD plasticity, when brief abnormal visual experience caused by MD alters OD tuning.⁸ The lower magnitude of representational drift present in adult mice may explain why they require longer durations of MD to display OD plasticity.^28,29 What distinguishes cortical neurons with tuning instability from neurons with more stable tuning remains unclear. Neurons with OD tuning instability were not evident from response strength or the stability of tuning for SF or orientation. Perhaps a transcriptional signature or signaling pathway regulates the stability of neuronal tuning. NGR1 is a leucine-rich repeat protein attached to the neuronal membrane by a glycosylphosphatidylinositol (GPI) lipid anchor, and how NGR1 may signal to limit representational drift and OD plasticity is unclear.³⁰ Interestingly, NGR1 binds to molecules enriched in both peri-neuronal nets and myelin,^31,32 both of which have been implicated as factors closing the critical period for OD plasticity.^33,34 Identifying a molecular profile for tuning instability may benefit from a better understanding of the function of NGR1.

Understanding whether tuning stability is conserved at the neuronal level or at the population level will require tracking the tuning properties of neurons over longer time courses. If tuning stability is conserved at the neuronal level, then whether neurons are stable or plastic for eye dominance should be conserved across multiple imaging sessions. Alternatively, neurons may each have some range of plasticity that is coordinated at the population level. Some evidence for this latter model is that the ngr1 gene does not operate in a cell-autonomous manner to limit OD plasticity; conditional deletion of ngr1 restricted to either layer 4 excitatory neurons or parvalbumin-positive interneurons sustains OD plasticity for neurons across layers in adult mice^14,35

We predict that other strategies for promoting experience-dependent plasticity, including removing perineuronal nets,³³ reducing myelination,³⁴ altering cholinergic tone,³⁶ or treating with psychedelics,³⁷ also increase tuning instability. An important test of the relationship between OD plasticity and representational drift will be to assess whether pharmacologic or genetic manipulations that block OD plasticity attenuate representational drift.^22,38 However, in the case of a negative result, where OD plasticity is occluded but representational drift is normal during the critical period, one caveat would be that it may prove difficult to rule out that these manipulations prevent MD from engaging representational drift to reorganize eye dominance. These future directions will refine this circuit model of experience-dependent plasticity: the magnitude of representational drift gates plasticity that adapts populations of neurons to recent experience by directing the instability of neuronal tuning.

RESOURCE AVAILABILITY

Lead contact

Requests for further information and resources should be directed to and will be fulfilled by the lead contact, Aaron W. McGee (awmcgee@arizona.edu).

Materials availability

This study did not generate new materials

Data and code availability

Data for all figures of the manuscript are available on Mendeley Data: https://doi.org/10.17632/j2z3h9rtys.1 and are publicly available as of the date of publication.

This study did not generate new code.

Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

STAR★METHODS

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

All procedures were approved by University of Louisville Institutional Animal Care and Use Committee (IACUC) protocol 22105 and were in accord with guidelines set by the US National Institutes of Health. Mice were anesthetized by isoflurane inhalation and killed by carbon dioxide asphyxiation or cervical dislocation following deep anesthesia in accordance with approved protocols. Mice were housed in groups of 5 or fewer per cage in a 12/12 light–dark cycle. Animals were naive subjects with no prior history of participation in research studies. A total of 85 mice, both male (44) and female (41) were used in this study. The following male and female mice were assigned at random to the following groups: one time point calcium imaging, P28–32 non-deprived WT, 9 males and 8 females; P32 4MD WT, 4 males and 4 females; P60–90 non-deprived WT, 7 males and 8 females; P60–90 4MD WT, 4 males and 3 females; P60–90 non-deprived ngr1 KO, 6 males and 6 females; P60–90 4MD ngr1 KO, 5 males and 5 females; two time point repeat calcium imaging, P60–90 non-deprived WT, 4 males and 2 females; P60–90 non-deprived ngr1 KO repeat imaging, 2 males and 2 females; P60–90 4MD ngr1 KO, 3 males and 4 females.

Imaging was performed on double transgenic mice expressing GCaMP6s in forebrain excitatory neurons. The CaMKII-tTA (stock no. 007004) and TRE-GCaMP6s (stock no. 024742) transgenic mouse lines were obtained from Jackson Labs^12,39 Mice were genotyped with primer sets suggested by Jackson Labs. These mice were crossed onto the ngr1 KO background.⁴⁰

METHOD DETAILS

Cranial windows

Wide field epi-fluorescent calcium imaging and two-photon calcium imaging were performed though a cranial window.⁴¹ In brief, mice were administered carprofen (5 mg/kg) and buprenorphine (0.1 mg/kg) for analgesia and anesthetized with isoflurane (4% induction, 1% to 2% maintenance). The scalp was shaved and mice were mounted on a stereotaxic frame with palate bar and their body temperature maintained at 37°C with a heat pad controlled by feedback from a rectal thermometer (TCAT-2LV, Physitemp). The scalp was resected, the connective tissue removed from the skull, and a custom aluminum headbar affixed with C&B metabond (Parkell). A circular region of bone 3 mm in diameter centered over left visual cortex was removed using a high-speed drill (Foredom). Care was taken to not perturb the dura. A sterile 3 mm circular glass coverslip was sealed to the surrounding skull with cyanoacrylate (Pacer Technology) and dental acrylic (Ortho-ject, Lang Dental). The remaining exposed skull likewise sealed with cyanoacrylate and dental acrylic. Mice recovered on a heating pad and returned to standard housing for at least 2 days prior to 2-photon imaging.

Wide field epi-fluorescent calcium imaging

After implantation of the cranial window and before 2-photon imaging, the binocular zone of visual cortex was identified with wide field calcium imaging similar to our method for optical imaging of intrinsic signals.^42,43 In brief, mice were anesthetized with isoflurane (4% induction), provided a low dose of the sedative chlorprothixene (0.5 mg/kg IP; C1761, Sigma) and secured by the aluminum headbar. The eyes were lubricated with a thin layer of ophthalmic ointment (Puralube, Dechra Pharmaceuticals). Body temperature was maintained at 37°C with heating pad regulated by a rectal thermometer (TCAT-2LV, Physitemp). Visual stimulus was provided through custom-written software (MATLAB, Mathworks). A monitor was placed 25 cm directly in front of the animal and subtended +40 to −40 degrees of visual space in the vertical axis. A horizonal white bar (2 degrees high and 20 degrees wide) centered on the zero-degree azimuth drifted from the top to bottom of the monitor with a period of 8 seconds. The stimulus was repeated 60 times. Cortex was illuminated with blue light (475 ± 30 nm) (475/35, Semrock) from a stable light source (Intralux dc-1100, Volpi). Fluorescence was captured utilizing a green filter (HQ620/20) attached to a tandem lens (50 mm lens, Computar) and camera (Manta G-1236B, Allied Vision). The imaging plane was defocused to approximately 200 μm below the pia. Images were captured at 10 Hz as images of 1,024 × 1,024 pixels and 12-bit depth. Images were binned spatially 4 × 4 before the magnitude of the response at the stimulus frequency (0.125 Hz) was measured by Fourier analysis.

Visual stimuli and two-photon calcium imaging

Visual stimulus presentation and image acquisition were both performed according to our published methods which were modified from published studies.^8,23,24,44 In brief, a battery of static sinusoidal gratings was generated in real time with custom software (Processing, MATLAB). Stimulus presentation was synchronized to the imaging data by time stamping the presentation of each visual stimulus to the image acquisition frame number a transistor–transistor logic (TTL) pulse generated with an Arduino at each stimulus transition. Orientation was sampled at equal intervals of 30 degrees from 0 to 150 degrees (6 orientations). SF was sampled in 8 steps on a logarithmic scale at half-octaves from 0.028 to 0.48 cpd. An iso-luminant grey screen was included (blank) was provided as a ninth step in the SF sampling as a control. Spatial phase was equally sampled at 45-degree intervals from 0 to 315 degrees for each combination of orientation and SF. Gratings with random combinations of orientation, SF, and spatial phase were presented at a rate of 4 Hz on a monitor with a refresh rate of 60Hz. Imaging sessions were 10 minutes (2,400 gratings presented in total). Consequently, each combination of orientation and SF was presented 40 times on average (range 29 to 56). The monitor was centered on the zero azimuth and elevation 35 cm away from the mouse and subtended 45 (vertical) by 80 degrees (horizontal) of visual space.

Imaging was performed with a resonant scanning 2-photon microscope controlled by Scanbox image acquisition and analysis software (Neurolabware). The objective lens was fixed at vertical for all experiments. Fluorescence excitation was provided by a tunable wavelength infrared laser (Ultra II, Coherent) at 920 nm. Images were collected through a 16× water-immersion objected (Nikon, 0.8 NA). Images (512 × 796 pixels, 520 × 740 μm) were captured at 15.5 Hz at depths between 150 and 300 μm. Eye movements and changes in pupil size were recorded using a Dalsa Genie M1280 camera (Teledyne Dalsa) fitted with 50 mm 1.8 lens (Computar) and 800 nm long-pass filter (Edmunds Optics). Imaging was performed on alert mice positioned on a spherical treadmill by the aluminum head bar affixed to the skull. Eye movements and changes in pupil size were recorded using a Dalsa Genie M1280 camera (Teledyne Dalsa) fitted with a 740nm long-pass filter. The visual stimulus was presented to each eye separately by covering the fellow eye with a small custom occluder.

Image processing

Image processing was performed as described previously.^8,23 Imaging series for each eye were motion corrected with the SbxAlign tool. Regions of interest (ROIs) corresponding to excitatory neurons were selected manually with the SbxSegment tool following computation of pixel-wise correlation of fluorescence changes over time from 350 evenly spaced frames (~4%). ROIs for each experiment were determined by correlated pixels the size similar to that of a neuronal soma. The fluorescence signal for each ROI and the surrounding neuropil were extracted from this segmentation map.

Image analysis

Image analysis was performed as described previously with minor modifications.⁸ The fluorescence signal for each neuron was extracted by computing the mean of the calcium fluorescence within each ROI and subtracting the median fluorescence from the surrounding perimeter of neuropil.^23,45 An inferred spike rate (ISR) was estimated from adjusted fluorescence signal with the Vanilla algorithm.⁴⁶ A reverse correlation of the ISR to stimulus onset was used to calculate the preferred stimuli.^8,23,44,45 Neurons that satisfied 3 criteria were categorized as visually responsive: (1) the ISR was highest with the optimal delay of 4 to 9 frames following stimulus onset. This delay was determined empirically for this transgenic GCaMP6s mouse^8,23; (2) the SNR was greater than 1.3. The signal is the mean of the spiking standard deviation at the optical delay between 4 and 9 frames after stimulus onset and the noise this value at frames −2 to 0 before the stimulus onset or 15 to 18 after it.^23,44 (3) and neuron responded to at least 13% of the presentations of the preferred stimulus. Visual responsiveness for every neuron was determined independently for each eye. The visual stimulus capturing the preferred orientation and SF was the determined from the matrix of all orientations and SFs presented as the combination with highest average ISR.

The preferred orientation for each neuron was calculated as:

$$orientation=\frac{\arctan \left(\sum _n{O}_n\ast e^{i\ast 2\ast \pi \ast {\theta }_n/180}\right)}{2}$$

${\text{O}}_{\text{n}}$ is a 1 × 6 array of the mean z-scores associated with the calculation of the ISR at orientations ${\mathrm{\Theta }}_{\text{n}}$ (0 to 150 degrees, spaced every 30 degrees). Orientation calculated with this formula is in radians and was converted to degrees. The tuning width was the full width at half-maximum of the preferred orientation.

The preferred SF for each neuron was calculated as:

$$SF=10\frac{{\sum }_{k}S{f}_{\kappa }\ast lo{g}_{10}{\omega }_{\kappa }}{{\sum }_{k}S{f}_{\kappa }}$$

${\text{Sf}}_{\text{k}}$ is a 1 × 8 array of the mean z-scores at SFs ${\text{ω}}_{\text{k}}$ (8 equal steps on a logarithmic scale from 0.028 to 0.481 cpd). Tails of the distribution were clipped at 25% of the peak response. The tuning width was the full width at half-maximum of the preferred SF in octaves.

Ocular Dominance Index (ODI)

Neuronal ODI was calculated as (C − I) / (C + I), where C and I are the mean normalized change in fluorescence (dF/F) for the preferred visual stimulus for the contralateral eye and ipsilateral eye, respectively. In cases where neurons were determined to not be visually-responsive to stimuli presented to one eye (see Image Analysis), they were considered monocular for the other eye and assigned ODI values of 1 (contralateral) and −1 (ipsilateral).⁴⁷

Repeated two-photon calcium imaging

Each imaging session was segmented independently, and every ROI was assigned a unique number. No geometric transformations were performed to match segmentation masks for ROIs from the 2 imaging sessions. The segmentation masks for the 2 imaging sessions were then compared and ROIs with at least 50% overlap were considered the same neuron. A perimeter of neurons with overlapping ROIs and tuning properties that did not change between imaging sessions (a difference in orientation preference of less than 30 degrees and SF preference of less than an octave) defined the matched imaging plane.⁸ To determine the SNR values of lost neurons, the segmentation masks were exchanged between time points and the SNR from ROIs for the corresponding neurons at the other time point were calculated.

Monocular deprivation by lid suture

The right eye was sutured closed with a single mattress suture with 6–0 prolene monofilament (Ethicon 8709).³⁵ Prior to imaging, mice were briefly (<5 minutes) anesthetized with isoflurane (4% induction, 1% to 2% maintenance), the suture removed with Vannas scissors (Fine Science Tools). The eye was flushed with sterile saline and examined for corneal abrasions with a stereomicroscope. The mouse was then immediately head-fixed for imaging and allowed to recover for no less than 45 minutes. The occluder was positioned over one eye as soon as the mouse was head-fixed and occluded one of the eyes at all times. At no point during the experiment were mice permitted unobstructed binocular vision.

QUANTIFICATION AND STATISTICAL ANALYSIS

No statistical methods were used to predetermine sample size. All statistical analyses were done using Prism 8 software (GraphPad Software). Multiple comparisons were tested with analysis of variance (ANOVA) tests and paired tests with Mixed-effects analysis.

Supplementary Material

Supplemental information can be found online at https://doi.org/10.1016/j.cub.2025.07.026.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Deposited data
Calculated tuning properties for all neurons	N/A	N/A
Experimental models: Organisms/strains
Mouse: B6;DBA-Tg(tedO-GCaMP6s)2Niell/j	The Jackson Laboratory	RRID: ISMR_ JAX: 024742
Mouse: B6;Cg-Tg(Camk2a-tTA)1Mmay/DboJ	The Jackson Laboratory	RRID: ISMR_ JAX: 007004
Mouse: nogo-66 receptor (ngr1/rtn4r) constitutive mutant mice	N/A	Kim et al.1
Software and algorithms
MATLAB	Mathworks	https://www.mathworks.com/
Processing2	Processing	https://processing.org/

Highlights.

• Stability of neuronal tuning is higher after the critical period

• The ngr1 gene increases tuning stability in adulthood

• Tuning instability gates adaptation to recent visual experience