Prefrontal cortex neurons encode ambient light intensity differentially across regions and layers

Abstract

While light can affect emotional and cognitive processes of the medial prefrontal cortex (mPFC), no light-encoding was hitherto identified in this region. Here, extracellular recordings in awake mice revealed that over half of studied mPFC neurons showed photosensitivity, that was diminished by inhibition of intrinsically photosensitive retinal ganglion cells (ipRGCs), or of the upstream thalamic perihabenular nucleus (PHb). In 15% of mPFC photosensitive neurons, firing rate changed monotonically along light-intensity steps and gradients. These light-intensity-encoding neurons comprised four types, two enhancing and two suppressing their firing rate with increased light intensity. Similar types were identified in the PHb, where they exhibited shorter latency and increased sensitivity. Light suppressed prelimbic activity but boosted infralimbic activity, mirroring the regions' contrasting roles in fear-conditioning, drug-seeking, and anxiety. We posit that prefrontal photosensitivity represents a substrate of light-susceptible, mPFC-mediated functions, which could be ultimately studied as a therapeutical target in psychiatric and addiction disorders.

Fig. 1. mPFC neurons exhibiting transient and persistent responses to light.

a Two multielectrode array placements (one more medial), during extracellular recordings in awake mice during light stimulation of the eyes, superimposed with corresponding mPFC subregions on a coronal atlas plane. b Distribution of neurons exhibiting statistically-significant transient and/or persistent effect of light on FR, or no response. c Mapping of neurons identified in the MOs, in sagittal and coronal planes. d Incidence (mean ± SD, across recording sessions) of neurons in the mPFC demonstrating transient and persistent light-responsiveness was significantly higher than in the MOs (permutation t-test, one-sided; transient: p = 3E-5, effect size (ES): d = 1.327; persistent: p = 1E-4, ES d = 1.145; mPFC: 1682 neurons, 60 recording sessions, 20 mice; MOs: 218 neurons, 8 recording sessions, 2 mice). Here and elsewhere, values over bars represent number of neurons. e, f Mapping of neurons in the frontal cortex exhibiting transient and/or persistent light-response (f), and a close-up view along the sagittal and coronal planes (e). Dots, each representing a neuron, indicate whether the neuron is light-responsive across the ‘early’ (blue) or ‘steady-state’ (red) window, or during both (magenta). Non-responsive neurons are coloured in grey. Abbreviations: anterior cingulate, AC; prelimbic cortex, PL; infralimbic cortex, IL; dorsal peduncular area, DP; dorsal taenia tecta, dTT. g Probability of neurons to be assigned to each mPFC subregion (median, 1st and 3rd quartiles; number of neurons assigned to each subregion are presented). h Distribution (percentage out of recorded neurons) of transient and persistent light-responsive neurons across the mPFC’s five subregions. Incidence of transient and persistent neurons in the IL and PL was higher but did not vary significantly compared to the remaining mPFC regions [transient: χ²(1, 1675) = 3.59, p = 0.058, one-sided, ES φ = 0. 040; persistent: χ²(1, 1675) = 3.02, p = 0.082, one-sided, ES φ = 0.052]. i Incidence (mean ± SD) across recording sessions of neurons in the mPFC demonstrating transient and persistent light-responsiveness did not differ significantly across hemispheres [permutation t-test, two-sided; transient, p = 0.94; persistent, p = 0.99; n_left = 32 and n_right = 28 recording sessions], as did the incidence of neurons across animals [not presented, permutation t-test; transient, p = 0.50; persistent, p = 0.58; n_left = 18 and n_right = 15 animals]. Mapping of neurons utilized BrainRender https://github.com/brainglobe/brainrender. Source data are provided as a Source Data file.

Fig. 2. Four types of light-intensity-dependent persistent response in the mPFC.

a Mean ± SEM FR vs. time for 7 intensities (left); steady-state FR fitted to a sigmoid and the RMSE (middle); RMSE null distribution (right). Neurons with RMSE_real< 5^th percentile RMSE_null were deemed intensity-encoding. b Distribution of neurons exhibiting statistically-significant transient and/or persistent intensity-encoding response (60 recording sessions, 20 mice). c Light-evoked FR (mean ± SEM) during the highest tested intensity, for four functional types. d Mean ± SEM baseline-subtracted steady-state FR, vs. light intensity, for the four types. e Threshold intensity (asterisk) at FR criterion of 0.1 Hz, varied by 0.8 log photons cm⁻² s⁻¹ across types (colours as in (c)). f ON peak latency. g Type-specific latency of late response (sample as in c). Latency differed across types [permutation ANOVA; F = 15.7, p < 0.001, effect size (ES) η² = 0.094; median (25^th and 75^th percentile): ‘ON-OFF-suppressed’: 0.4 (0.2, 1.0) sec, ‘ON-enhanced’: 1.95 (0.45, 3.95) sec, ‘ON-suppressed’: 1.0 (0.37, 3.0) sec]. Post-hoc comparisons, permutation t-test, two-sided: latency differed between ‘ON-OFF suppressed’ and the two ‘ON’ types (p = 3E-4, ES d = 0.609), but not across ‘ON’ types (p = 0.3206, ES d = 0.258). h Percent change in FR from baseline (median, 25^th and 75^th percentile) for all types (sample as in (c)). Error calculation included all data; data points exceeding ±150% change not plotted, to facilitate comparisons. i, Mapping across the mPFC. j Persistent intensity-encoding neurons incidence, was higher in the PL and IL [χ²(1, 1675) = 4.36, p = 0.036, one-sided, ES φ = 0.051]. k Incidence (mean ± SD) across recording sessions did not differ between hemispheres [premutation t-test, two-sided; p = 0.714, ES d = 0.095; n_left = 32, n_right = 28 recording sessions], as did the incidence across animals [not presented, premutation t-test, p = 0.162, ES d = 0.503; n_left = 18 and n_right = 15 mice]. l Distribution of the four types did not differ between hemispheres [χ²(9, 252) = 2.14, p = 0.543, one-sided, ES Cramer’s V = 0.092]. m ‘Enhanced’ vs. ‘suppressed’ types distribution across mPFC subregions [IL vs. PL: χ²(1, 93) = 22.90, p = 2E-6, one-sided, ES φ = 0.497; DP vs. dTT: χ²(1, 18) = 2.82, p = 0.09, one-sided, ES φ = 0.396]. n Median (25^th and 75^th percentile) light-evoked steady-state firing of intensity-encoding neurons, across mPFC subregions [sample as in (j)]. Light-evoked firing differed across subregions (permutation ANOVA; F = 9.28, p = 4E-6, ES η² = 0.131), and was higher in IL vs. PL (permutation t-test, one-sided, p = 1E-6, ES d = 1.152) and in DP vs. dTT (permutation t-test, one-sided, p = 0.0079, ES d = 1.477). Mapping of neurons utilized BrainRender https://github.com/brainglobe/brainrender. Source data are provided as a Source Data file.

Fig. 3. The activity of mPFC neurons continuously tracks the intensity of ambient light.

a Main plots, Light-evoked FR (mean ± SEM across neurons) in response to 10 s of the highest tested intensity, for the four functional types of intensity-encoding neurons. Insets, Steady-state FR (mean ± SEM across neurons) vs. light intensity for the four types. b FR modulations (mean ± SEM across neurons) in the four types in response to a bi-phasic stimulus. The secondary (top) abscissa and the grayscale gradient represent light intensity. A smoothed version (thick line; moving average, 5 s) of the raw, noisy, FR trace (thin line) is presented. c A sigmoid fitted to the mean FR along each phase, for each type. The transition point from ascending to descending intensity (30 s) is marked by a vertical dashed black line; slopes (n) of the sigmoid fitted to the two phases are indicated; RMSE_real and the 5^th percentile of the RMSE_null are indicated below plots. Orange: points in which FR change lagged behind the transition between light phases. Considering that each data point represents a 1 s interval, FR lagged behind the intensity transition by 1, 0, 3, and 5 s, for ON-OFF enhanced, ON-OFF suppressed, ON enhanced, and ON suppressed types, respectively. Local minima/maxima in FR (arrows) for all four types appeared along the ascending phase at intensity 13.8 – 14.2 log photons cm⁻² s⁻¹ (6–12 s from stimulus onset), and along the descending phase at intensities 14–14.3 log photons cm⁻² s⁻¹ (46–51 s from stimulus onset). Inset. Mean FR (spikes/sec) as a function of light intensity (log photons cm⁻² s⁻¹) for the ascending and descending phases, for each of the four types. Source data are provided as a Source Data file.

Fig. 4. Transmission from PHb affects light-evoked mPFC firing rate.

a Viral labelling of mPFC-projecting PHb neurons in WT mice, by bilateral injection of retrograde Cre/GFP-expressing AAV in the mPFC (shown in the AC, PL), and Cre-dependent AAV-hM4D(Gi) DREADD/mCherry in the PHb. Light-evoked firing of the same neurons was recorded following saline and CNO subcutaneous infusion. b, c GFP/Cre-positive somata and axons in the mPFC (b) and mCherry/DREADDs-positive somata of mPFC-projecting PHb neurons (c); reproduced in 4 mice in which all four injections were accurate. d Incidence of light-responsive mPFC neurons following saline and CNO, in mice expressing DREADDs in the PHb, and in mice that do not (control). Incidence differed between saline and CNO in the DREADDs group [χ²(1, 78) = 6.68, p = 0.009, one-sided, effect size (ES) φ = 0.102] but not in controls [χ²(1, 290) = 3.02, p = 0.08, ES φ = 0.293]. Only 3 intensity-encoding neurons were identified after saline, none retained intensity-encoding following CNO. e WT mice bilaterally injected in the PHb with an inhibiting constitutive light-activated eOPN3, to optogenetically inhibit axon terminals of mPFC-projecting PHb neurons using an optotrode, while recording mPFC neuronal firing. Control: mice injected with a constitutive fluorescent reporter. f eOPN3/mScarlet florescence in the PHb, with slight expression in the dentate gyrus (molecular layer) and habenula (both not reported to project to IL/PL). Labeled terminals dominated in the IL/PL (targets of mPFC-projecting PHb neurons). g Absolute steady-state FR (mean ± SEM) of identified neurons, vs. light intensity, with fitted sigmoid, in mScarlet and eOPN3 mice (mScarlet, n = 145 neurons, 4 sessions, 2 mice; eOPN3, n = 221 neurons, 5 sessions, 3 mice). FR in response to the 3 highest intensities in eOPN3 mice was lower than in mScarlet mice (permutation t-test, one-sided, corrected for multiple comparisons; p = 0.028, 0.005, and 0.006, ES d = 0.251, 0.401, and 0.381, for the highest to 3rd highest intensities). h Incidence across recording sessions (mean ± SD) of light-responsive and intensity-encoding neurons in the eOPN3 group (n = 5 sessions) was lower than in controls (mScarlet; n = 4 sessions; permutation t-test, one-sided; p = 0.012, ES d = 0.794 for light-responsive neurons, p = 0.008, ES d = 1.216 for intensity-encoding neurons), as did the incidence across animals (not presented, permutation t-test, one-sided; p = 0.016, ES d = 4.468 for light-responsive neurons, p = 9E-6, ES d = 4.567 for intensity-encoding neurons). i Incidence of the four mPFC intensity-encoding types in eOPN3 and mScarlet mice. Source data are provided as a Source Data file.

Fig. 5. The PHb harbours light-intensity-encoding neurons.

a, b Schematic (a) and image (b) of an extracellular probe in the PHb. c, Incidence of PHb transient and persistent types, among light-responsive and intensity-encoding neurons. d Mean ± SEM FR across responsive and unresponsive neurons in the PHb. e Light-evoked FR (mean ± SEM) for the PHb’s four types (colour) and their PFC counterparts (grey). Insets, FR at 10 ms instead of 100 ms binning; the ON peak in the PHb (47.5 ± 5 ms) arrives 40 ms earlier than in the mPFC (87.5 ± 9.6 ms), for all types except ‘ON-suppressed’. f For all types, amplitude (mean ± SEM) of light-evoked FR in the PHb vs. mPFC differed for a subset of the 7 tested intensities (permutation t-test, two-sided; asterisks: p < 0.05, corrected for multiple comparisons; ‘ON-OFF-enhanced’, p = 0.0076, 0.0246, and 0.001 for the highest, and 2^nd and 4^th highest intensities; ‘ON-OFF-suppressed’, p = 0.0054, 0.0048, and 0.0006 for the highest, and 4^th and 5^th highest intensities; ‘ON-enhanced’, p = 0.0180 and 0.008 for the highest and 2^nd highest intensities; ‘ON-suppressed’, p = 0.0114 and 0.0372 for the highest and 2^nd highest intensities). g Decay time (mean ± SEM) of light-evoked firing in the PHb vs. mPFC was longer for all types, and statistically significant for the ‘ON-OFF-suppressed’, ‘ON-enhanced’, and ‘ON-suppressed’ (permutation t-test, one-sided; asterisks: p = 0.0084, 0.0302, and 0.0011; effect size (ES) Cohen’s d = 0.88, 0.63 and 1.06; sample as in (e)). h Light sensitivity for all types in the PHb (thick lines) and mPFC (thin lines). A 0.1 Hz response threshold criterion was attained at 9.67, 11.4, and 12.36 log photons cm⁻² s⁻¹ in the PHb, compared to 12.4, 13.02, 13.05, and 13.6 log photons cm⁻² s⁻¹ in the mPFC. For ‘ON-OFF-suppressed’, quality of fit was low (high RMSE); thus no estimation of response threshold was attempted. Inset, PHb sensitivity was 2.73-, 1.65-, and 1.24 log photons cm⁻² s⁻¹ higher than in the mPFC, for ‘ON-OFF-enhanced’, ‘ON-enhanced’, and ‘ON-suppressed’, respectively. i, Baseline FR (mean ± SEM) in the PHb was higher than in the mPFC for all types; asterisks: p < 0.05; (permutation t-test, one-sided; p = 0.0001, 0.0011, 0.0002, and 0.0004; ES Cohen’s d = 1.12, 1.37, 1.63, and 1.21; for ‘ON-OFF-enhanced’, ‘ON-OFF-suppressed’, ‘ON-enhanced’ and ‘ON-suppressed’, sample as in (e)). j Light-evoked change in FR relative to baseline (median; box: 25^th, 75^th percentile; error bars: 10^th, 90^th percentile) was smaller in the PHb than in the mPFC for all types except for ‘ON-OFF-suppressed’ (permutation t-test, one tail, p = 0.027, 0.125, 0.043, and 0.0019; ES Cohen’s d = 0.505, 0.617, 0.661 and 1.164 for ‘ON-OFF-enhanced’, ‘ON-OFF-suppressed’, ‘ON-enhanced’, and ‘ON-suppressed’; asterisks: p < 0.05). k Distribution of PHb types differed significantly from that of their mPFC counterparts [χ²(9, 306) = 17.69, p = 5E-4, one-sided, ES Cramer’s V = 0.241]. Source data are provided as a Source Data file.

Fig. 6. Chemogenetic inhibition of ipRGCs disrupts mPFC light responsiveness and intensity-encoding.

a Chemogenetic inhibition of ipRGCs in the Opn4^Cre/+ mouse. b Example retina with DREADD/mCherry-positive ipRGCs. c Absolute FR change relative to baseline (mean ± SEM; in response to the highest intensity), of light-responsive neurons (n = 101 and 56) in DREADD-expressing vs. mCherry-expressing mice, following infusion of saline vs. CNO. Error calculation included all data; data points exceeding 100% change not plotted, to facilitate comparisons. Effect of CNO vs. saline on the relative change in FR was evident in DREADD-expressing, but not in mCherry-expressing mice (permutation ANOVA; F = 6.2, p = 4E-4, effect size (ES) η² = 0.094). DREADD-CNO differed from all other groups (permutation t-test, two-sided; DREADD-CNO vs. DREADD-saline, p = 0.002, ES d = 0.555; DREADD-CNO vs. control-CNO, p = 1E-4, ES d = 0.713; DREADD-CNO vs. control-saline, p = 1E-4, ES d = 0.953). DREADD-saline, control-CNO, and control-saline groups did not differ between one another (permutation t-test, two-sided; DREADD-saline vs. control-CNO, p = 0.837, ES d = 0.164; DREADD-saline vs. control-saline, p = 0.819, ES d = 0.300; control-CNO vs. control-saline, p = 0.545, ES d = 0.111). d Absolute steady-state FR (mean ± SEM) of intensity-encoding neurons (all four types pooled) as a function of light intensity, along with the fitted sigmoid, in DREADD-expressing (n = 36 neurons) vs. mCherry-expressing (n = 31 neurons) mice, following saline vs. CNO infusion. FR in DREADD-expressing mice in response to the 4 highest intensities differed significantly between CNO and saline infusion (permutation t-test, two-sided, corrected for multiple comparisons; p = 0.004, 0.002, 0.001, and 0.002; ES d = 0.556, 0.859, 0.797, and 0.911; for the highest to 4^th highest intensity). FR in mCherry-expressing mice in response to the 3 highest intensities differed significantly between CNO and saline infusion (permutation t-test, corrected for multiple comparisons; p = 0.045, 0.002, and 0.003; ES d = 0.391, 0.533, 0.412; for the highest to 3^rd highest intensity). Thus, ES for the highest 3 intensities was higher in DREADD-expressing than mCherry-expressing mice. e Number of neurons assigned to each type, following CNO, as compared to saline. Incidence of neurons retaining intensity-encoding firing following CNO is presented above each type’s bar (n = 20, 10, 3 and 3 neurons). f Steady-state FR (mean ± SEM) of the four types of persistent, intensity-encoding neurons, per light intensity (9.4−15.4 log photons cm⁻² s⁻¹) and sigmoid fit, following CNO (grey) vs. saline (coloured). Numbers of neurons is indicated above plots. Source data are provided as a Source Data file.

Fig. 7. Terminal ablation of ipRGCs diminishes mPFC light responsiveness and intensity-encoding.

a Terminal ablation of ipRGCs in the Opn4^Cre/+ mouse. b Example retina following injection of an AAV inducing Cre-dependent expression of dtA and constitutive expression of mCherry. c Incidence (mean ± SEM) of light-responsive and intensity-encoding neurons per recording session. The dtA (15 recording sessions), mCherry (6), and WT (60) groups differed in incidence of light-responsive neurons but not of intensity-encoding neurons (permutation ANOVA, F = 4.71 and 2.29, p = 0.012 and 0.108, ES η² = 0.110 and 0.072, respectively). Pairwise comparisons: incidence across sessions of light-responsive neurons and intensity-encoding neurons in dtA mice was lower than in WT and mCherry mice (permutation t-test, one-sided, light-responsive: p = 0.002 and 0.034, ES d = 0.895 and 1.020; intensity-encoding: p = 0.034 and 0.032, ES d = 0.699 and 1.005), and WT and mCherry mice did not differ between one another (light-responsive: p = 0.982, ES d = 0.068, intensity-encoding: p = 0.8950, ES d = 0.181). The three groups did not differ in the incidence across mice of light-responsive or intensity-encoding neurons (permutation ANOVA, p > 0.07; number of mice: 6, 3, and 20 in the dtA, mCherry, and WT groups). d Absolute steady-state FR (mean ± SEM) of all identified neurons as a function of light intensity, along with the fitted sigmoid, in the WT, mCherry, and dtA groups (n = 1681, 158, and 473 neurons). FR in response to the 2 highest intensities in the WT group was significantly higher than in the dtA group (permutation t-test, one-sided, corrected for multiple comparisons; p = 0.001 and 0.006; ES d = 0.218 and 0.150; for the highest and 2nd highest intensities). e Incidence of the four intensity-encoding types, in the dtA vs. WT groups. Source data are provided as a Source Data file.

Fig. 8. The postulated neural network underlying mPFC photosensitivity.

a, b In the mouse, ipRGCs innervate PHb excitatory relay neurons and inhibitory neurons, which comprise both local interneurons and long-range projecting inhibitory neurons. c, Within the PHb, local interneurons synapse on excitatory relay neurons. d, e PHb projecting inhibitory neurons innervate the thalamic reticular nucleus (TRN) (d), which in turn, may provide inhibitory feedback to some or all of the PHb neuronal types (e), as it does to other dorsothalamic areas. f PHb excitatory relay neurons innervate the mPFC’s IL, PL, and possibly the rostral AC (cg1), perhaps with axon collaterals to the TRN, as often exhibited by neurons in other medial thalamic nuclei. g, h The AC receives additional retinal input indirectly through the visual cortex (g), which blends input from both ipRGCs and conventional RGCs, transmitted through the dorsal division of the lateral geniculate nucleus (dLGN) (h). i–l The IL and DP receive additional inhibitory input from the ipRGC-recipient central amygdala (CeA) (I, j), and all five mPFC subregions appear to receive additional excitatory input from the ipRGC-recipient lateral hypothalamus (LH) (k, l). m Corticothalamic feedback from the mPFC to the PHb, including axon collaterals to the TRN, may also exist. Selected light-modulated behaviours, and the different mPFC regions and intensity-encoding neuronal types that may underlie these behaviours, are indicated. The mFPC harbours four distinct functional types of intensity-encoding neurons, two types enhance and another two types suppress their FR with increasing light intensity. Roughly 80% of the neurons belonging to each type are excitatory neurons, and the rest are inhibitory interneurons (the E_on type deviates from this trend as it includes only excitatory neurons). The mPFC includes two pairs of neighbouring subregions that overall, oppositely react to light exposure – light overall enhances IL firing but suppresses PL firing, and similarly, light overall enhances DP firing but suppresses dTT firing. The AC displays equal fractions of ‘enhanced’ and ‘suppressed’ types. The PHb harbours four functional types resembling their mPFC counterparts in general form, but exhibit shorter latency, larger amplitude, higher sensitivity, and longer decay time.