Gamma activity accelerates during prefrontal development

Sebastian H Bitzenhofer; Jastyn A Pöpplau; Ileana Hanganu-Opatz

doi:10.7554/eLife.56795

. 2020 Nov 18;9:e56795. doi: 10.7554/eLife.56795

Gamma activity accelerates during prefrontal development

Sebastian H Bitzenhofer ^1,^†,^✉, Jastyn A Pöpplau ¹, Ileana Hanganu-Opatz ^1,^✉

Editors: Martin Vinck², Laura L Colgin³

PMCID: PMC7673781 PMID: 33206597

Abstract

Gamma oscillations are a prominent activity pattern in the cerebral cortex. While gamma rhythms have been extensively studied in the adult prefrontal cortex in the context of cognitive (dys)functions, little is known about their development. We addressed this issue by using extracellular recordings and optogenetic stimulations in mice across postnatal development. We show that fast rhythmic activity in the prefrontal cortex becomes prominent during the second postnatal week. While initially at about 15 Hz, fast oscillatory activity progressively accelerates with age and stabilizes within gamma frequency range (30–80 Hz) during the fourth postnatal week. Activation of layer 2/3 pyramidal neurons drives fast oscillations throughout development, yet the acceleration of their frequency follows similar temporal dynamics as the maturation of fast-spiking interneurons. These findings uncover the development of prefrontal gamma activity and provide a framework to examine the origin of abnormal gamma activity in neurodevelopmental disorders.

Research organism: Mouse

Introduction

Synchronization of neuronal activity in fast oscillatory rhythms is a commonly observed feature in the adult cerebral cortex. While its exact functions are still a matter of debate, oscillatory activity in gamma frequency range has been proposed to organize neuronal ensembles and to shape information processing in cortical networks (Singer, 2018; Cardin, 2016; Sohal, 2016). Gamma activity emerges from reciprocal interactions between excitatory and inhibitory neurons. In the visual cortex, fast inhibitory feedback via soma-targeting parvalbumin (PV)-expressing inhibitory interneurons leads to fast gamma activity (30–80 Hz) (Cardin et al., 2009; Chen et al., 2017), whereas dendrite-targeting somatostatin (SOM)-expressing inhibitory interneurons contribute to beta/low gamma activity (20–40 Hz) (Chen et al., 2017; Veit et al., 2017). A fine-tuned balance between excitatory drive and inhibitory feedback is mandatory for circuit function underlying cognitive performance. Interneuronal dysfunction and ensuing abnormal gamma activity in the medial prefrontal cortex (mPFC) have been linked to impaired cognitive flexibility (Cho et al., 2015). Moreover, imbalance between excitation and inhibition in cortical networks and resulting gamma disruption have been proposed to cause cognitive disabilities in autism and schizophrenia (Cho et al., 2015; Cao et al., 2018; Rojas and Wilson, 2014).

Despite substantial literature linking cognitive abilities and disabilities to gamma oscillations in the adult mPFC, the ontogeny of prefrontal gamma activity is poorly understood. This knowledge gap is even more striking considering that abnormal patterns of fast oscillatory activity have been described at early postnatal age in autism and schizophrenia mouse models (Chini et al., 2020; Richter et al., 2019; Hartung et al., 2016). Knowing the time course of prefrontal gamma maturation is essential for understanding the developmental aspects of mental disorders.

To this end, we performed an in-depth investigation of the developmental profile of gamma activity in the mouse mPFC from postnatal day (P) five until P40. We show that pronounced fast oscillatory activity emerges toward the end of the second postnatal week and increases in frequency and amplitude with age. While activation of layer 2/3 pyramidal neurons (L2/3 PYRs) drives fast oscillatory activity throughout development, the acceleration of its frequency follows the same dynamics as the maturation of inhibitory feedback and fast-spiking (FS) interneurons.

Results

Fast oscillatory activity in the prefrontal cortex accelerates during development

Extracellular recordings in the mPFC of anesthetized and non-anesthetized P5-40 mice revealed that oscillatory activity at fast frequencies (>12 Hz) can be detected at the beginning of the second postnatal week. The temporal dynamics of these fast oscillations are similar in the two states, yet their magnitude is higher in non-anesthetized mice, as previously described (Chini et al., 2019). The magnitude of fast oscillations increases with age (Mann-Kendall trend test, p=3.93×10⁻²², n = 114 recordings, tau-b 0.625) and can be detected as distinct peaks in power spectra at the end of the second postnatal week (Figure 1a,b). The peak frequency of these oscillations gradually increases with age (Mann-Kendall trend test, p=2.73×10⁻⁸, n = 114 recordings, tau-b 0.361), starting at ~ 20 Hz at P12 and reaching the characteristic gamma frequency of 50–60 Hz at P25 (Figure 1b–d). Both, peak strength and peak frequency, do not change after P25. A linear regression model of peak frequency and peak amplitude shows significant correlation with age (n = 114, df = 111, R² = 0.542, p=5.48×10⁻²⁰; ANOVA: peak frequency F(1,111)=17.8, p=4.86×10⁻⁵, peak amplitude F(1,111)=74.4, p=5.05×10⁻¹⁴).

Figure 1. — (a) Schematic of extracellular recordings in the mPFC of anesthetized and non-anesthetized P5-40 mice. (b) Characteristic examples of extracellular recordings of local field potentials (LFP) and multi-unit activity (MUA) at different ages after band-pass filtering (left) and the corresponding power spectra (right). (c) Z-scored average power spectra of spontaneous oscillatory activity for P5-40 mice (n = 114 recordings from 100 mice). (d) Scatter plot displaying peak frequencies of fast oscillations (12–100 Hz) during postnatal development of anesthetized (gray, n = 80 recordings/mice) and non-anesthetized mice (green, n = 34 recordings from 20 mice). Marker size displays peak strength. (See Supplementary file 1 for a summary of experimental conditions. See Supplementary file 2 for statistics.).

Figure 1—source data 1. Source data for Figure 1b.

elife-56795-fig1-data1.mat^{(1.6KB, mat)}

FS interneuron maturation resembles the time course of gamma development

FS PV-expressing interneurons have been identified as key elements for the generation of oscillatory activity in gamma frequency range in the adult cortex (Cardin et al., 2009). To assess whether the developmental gamma dynamics relate to FS PV-expressing interneuron maturation, we performed immunohistochemistry and single unit analysis in P5-40 mice.

First, immunostainings showed that PV expression in the mPFC starts at the end of the second postnatal week, increases until P25 and stabilizes afterwards (Mann-Kendall trend test, p=1.29×10⁻⁷, n = 38 mice, tau-b 0.623) (Figure 2a). These dynamics over age follow a similar trend as the changes in peak power and peak frequency of the fast oscillations described above. In contrast, the number of SOM positive neurons does not significantly vary along postnatal development (Mann-Kendall trend test, p=0.99, n = 39 mice, tau-b −0.003) (Figure 2b).

Figure 2. — (a) Left, examples of PV immunostaining in the mPFC at different ages. Right, scatter plot displaying the density of PV-immunopositive neurons in the mPFC of P5-40 mice (n = 38 mice). (b) Same as (a) for SOM-immunopositive neurons (n = 39 mice). (c) Example mean waveforms of extracellular recorded single units from P5-40 mice. (d) Schematic showing features classically used to distinguish RS and FS units in adult mice. (e) Left, scatter plot showing the first two components of a t-sne dimensionality reduction on the mean waveforms for all units recorded from P5-40 mice (n = 3554 units from 66 recordings/mice). Right, same as left with the first two clusters obtained by hierarchical clustering labeled in black and red. (f) Scatter plot of half width and trough-to-peak time for cluster 1 (black) and 2 (red). (g) Mean waveform for cluster 1 (black) and 2 (red). (h) Scatter plot showing the proportion of FS units for P5-40 mice. (i) Scatter plots showing classic spike shape features for P5-40 for cluster 1 (RS, black) and 2 (FS, red). (See Supplementary file 1 for a summary of experimental conditions. See Supplementary file 2 for statistics).

Figure 2—source data 1. Source data for Figure 2a,b.

elife-56795-fig2-data1.mat^{(795B, mat)}

Figure 2—source data 2. Source data for Figure 2f.

elife-56795-fig2-data2.mat^{(31.4KB, mat)}

Figure 2—source data 3. Source data for Figure 2g.

elife-56795-fig2-data3.mat^{(3.2KB, mat)}

Figure 2—source data 4. Source data for Figure 2h.

elife-56795-fig2-data4.mat^{(626B, mat)}

Figure 2—source data 5. Source data for Figure 2i.

elife-56795-fig2-data5.mat^{(57.7KB, mat)}

Figure 2—figure supplement 1. — (a) Left, examples of PV immunostaining in the mPFC at different ages. Right, scatter plot displaying the density of PV-immunopositive neurons in the mPFC of P5-40 mice (n = 38 mice). (b) Same as (a) for SOM-immunopositive neurons (n = 39 mice). (c) Example mean waveforms of extracellular recorded single units from P5-40 mice. (d) Schematic showing features classically used to distinguish RS and FS units in adult mice. (e) Left, scatter plot showing the first two components of a t-sne dimensionality reduction on the mean waveforms for all units recorded from P5-40 mice (n = 3554 units from 66 recordings/mice). Right, same as left with the first two clusters obtained by hierarchical clustering labeled in black and red. (f) Scatter plot of half width and trough-to-peak time for cluster 1 (black) and 2 (red). (g) Mean waveform for cluster 1 (black) and 2 (red). (h) Scatter plot showing the proportion of FS units for P5-40 mice. (i) Scatter plots showing classic spike shape features for P5-40 for cluster 1 (RS, black) and 2 (FS, red). (See Supplementary file 1 for a summary of experimental conditions. See Supplementary file 2 for statistics).

Figure 2—source data 1. Source data for Figure 2a,b.

elife-56795-fig2-data1.mat^{(795B, mat)}

Figure 2—source data 2. Source data for Figure 2f.

elife-56795-fig2-data2.mat^{(31.4KB, mat)}

Figure 2—source data 3. Source data for Figure 2g.

elife-56795-fig2-data3.mat^{(3.2KB, mat)}

Figure 2—source data 4. Source data for Figure 2h.

elife-56795-fig2-data4.mat^{(626B, mat)}

Figure 2—source data 5. Source data for Figure 2i.

elife-56795-fig2-data5.mat^{(57.7KB, mat)}

Second, to directly assess the functional maturation of FS putatively PV-expressing neurons, we used bilateral extracellular recordings and identified single unit activity (SUA). The classically used action potential features to distinguish adult FS and regular-spiking (RS) neurons (i.e. trough to peak time and half width) cannot be applied during early development, because of a strong overlap of these features. Therefore, we developed an algorithm to classify RS and FS units using dimensionality reduction with t-Distributed Stochastic Neighbor Embedding (t-sne) on the mean waveforms of all units recorded across development, followed by hierarchical clustering based on pairwise Euclidean distance (Figure 2c–e). This approach resulted in an unbiased detection of FS units across age (Figure 2f,g). Of note, a more detailed analysis revealed that the dimensionality reduction based on mean waveforms correlates with features like trough to peak and half width, but is less affected by age, cortical layer, and amplitude (Figure 2—figure supplement 1). The classification of units with this method revealed that FS units start to be detected at the end of the second postnatal week. Their number gradually increased until P25 (Mann-Kendall trend test, p=1.44×10⁻⁷, n = 66 recordings, tau-b 0.458) (Figure 2h). A comparison of the classical features across age showed that trough-to-peak duration (Mann-Kendall trend test, RS, p=4.57×10⁻⁵, n = 3172 units, tau-b −0.051; FS, p=9.23×10⁻¹¹, n = 382 units, tau-b −0.236), half width (Mann-Kendall trend test, RS, p=1.61×10⁻²⁸, n = 3172 units, tau-b −0.134; FS, p=5.17×10⁻¹⁷, n = 382 units, tau-b −0.295), and negative amplitude (Mann-Kendall trend test, RS, p=4.45×10⁻²², n = 3172 units, tau-b −0.117; FS, p=3.82×10⁻⁶, n = 382 units, tau-b −0.163) of RS and FS units gradually decreased from the end of the second postnatal week until P25. However, the most prominent changes were detected for through-to-peak duration and half width of FS units (Figure 2i). A linear regression model of single unit features shows significant correlation with age (n = 52, df = 44, R² = 0.550, p=2.31×10⁻⁷; see Supplementary file 2 for details). These results are consistent with a detailed description of the physiological development of prefrontal PV-expressing interneurons performed in brain slices (Miyamae et al., 2017).

Thus, in line with the immunohistochemical examination, the analysis of single units showed that FS putatively PV-positive interneurons show similar dynamics of maturation as fast oscillations recorded in the mPFC of P5-40 mice.

Activation of L2/3 pyramidal neurons drives fast oscillations with similar acceleration across development as spontaneous activity

Besides FS interneurons, L2/3 PYRs in mPFC have been found to induce fast oscillations in the mPFC of P8-10 mice. Their non-rhythmic activation (but not activation of L5/6PYRs) drives oscillatory activity peaking within 15–20 Hz range (Bitzenhofer et al., 2017a; Bitzenhofer et al., 2017b), similar to the peak frequency of spontaneous network activity at this age. To test if L2/3 PYRs-driven activity also accelerates with age, we optogenetically manipulated these neurons in P5-40 mice. Stable expression of the light-sensitive channelrhodopsin two derivate E123T T159C (ChR2(ET/TC)) restricted to about 25% of PYR in L2/3 of the mPFC was achieved by in utero electroporation (IUE) at embryonic day (E) 15.5 (Figure 3a). Optogenetic stimulation with ramps of steadily increasing light power (473 nm, 3 s, 30 repetitions) were performed during extracellular recordings in the mPFC. As previously shown, this type of stimulation activates the network without forcing a specific rhythm (Bitzenhofer et al., 2017a).

Figure 3. — (a) ChR2(ET/TC)−2A-RFP-expression in L2/3 PYRs in mPFC after IUE at E15.5 in a coronal slice of a P10 mouse. (b) Characteristic examples of extracellular recordings of LFP and MUA during ramp light stimulations (473 nm, 3 s) of prefrontal L2/3 PYRs at different ages (left) and the corresponding MI of power spectra (right). (c) Z-scored average MI of power spectra for P5-40 mice (n = 115 recordings from 101 mice). (d) Scatter plot displaying stimulus induced peak frequencies during postnatal development for anesthetized (gray, n = 80 recordings/mice) and non-anesthetized mice (green, n = 35 recordings from 21 mice). Marker size displays peak strength. (See Supplementary file 1 for a summary of experimental conditions. See Supplementary file 2 for statistics).

Figure 3—source data 1. Source data for Figure 3d.

elife-56795-fig3-data1.mat^{(1.6KB, mat)}

Figure 3—figure supplement 1. — (a) ChR2(ET/TC)−2A-RFP-expression in L2/3 PYRs in mPFC after IUE at E15.5 in a coronal slice of a P10 mouse. (b) Characteristic examples of extracellular recordings of LFP and MUA during ramp light stimulations (473 nm, 3 s) of prefrontal L2/3 PYRs at different ages (left) and the corresponding MI of power spectra (right). (c) Z-scored average MI of power spectra for P5-40 mice (n = 115 recordings from 101 mice). (d) Scatter plot displaying stimulus induced peak frequencies during postnatal development for anesthetized (gray, n = 80 recordings/mice) and non-anesthetized mice (green, n = 35 recordings from 21 mice). Marker size displays peak strength. (See Supplementary file 1 for a summary of experimental conditions. See Supplementary file 2 for statistics).

Figure 3—source data 1. Source data for Figure 3d.

elife-56795-fig3-data1.mat^{(1.6KB, mat)}

Similar to spontaneous activity, activating L2/3 PYRs induced oscillatory activity with a gradually increasing frequency during development (Figure 3b–d). Consistent peaks in the modulation index (MI) of power spectra were detected at 15–20 Hz at the beginning of the second postnatal week and increased in frequency (Mann-Kendall trend test, p=7.69×10⁻⁶, n = 115 recordings, tau-b 0.288) and amplitude (Mann-Kendall trend test, p=1.04×10⁻⁹, n = 115 recordings, tau-b 0.392) until reaching stable values within 50–60 Hz at P25. A linear regression model of peak frequency and peak amplitude shows significant correlation with age (n = 115, df = 112, R² = 0.364, p=3.72×10⁻¹²; ANOVA: peak frequency F(1,112)=13.9, p=2.95×10⁻⁴, peak amplitude F(1,112)=43.8, p=1.26×10⁻⁹). Control stimulations with light that does not activate ChR2(ET/TC) (594 nm, 3 s, 30 repetitions) did not induce activity and led to the detection of unspecific peak frequencies (Mann-Kendall trend test, p=0.09, n = 111 recordings, tau-b 0.111) and low amplitudes (Mann-Kendall trend test, p=0.74, n = 111 recordings, tau-b 0.022) (Figure 3—figure supplement 1).

Peak frequency and amplitude of activity induced by ramp light stimulation are significantly correlated with peak frequency and amplitude of spontaneous activity at the level of individual recordings (peak frequency n = 114, df = 112, R² = 0.16, p=6.00×10⁻⁶; peak amplitude n = 114, df = 112, R² = 0.206, p=2.30×10⁻⁷). Thus, L2/3 PYR-driven activity in the mPFC follows the same developmental dynamics as spontaneous activity indicating the importance of L2/3 PYRs for gamma maturation.

The rhythmicity of pyramidal cell and interneuron firing follows similar development as accelerating gamma activity in mPFC

To assess the contribution of distinct cell types to the emergence of gamma during postnatal development, we compared the firing of RS units, mainly corresponding to PYRs, and FS units, mainly corresponding to PV-expressing interneurons, during ramp light stimulations of L2/3 PYRs in P5-40 mice (Figure 4a,b, Figure 4—figure supplement 1).

Figure 4—figure supplement 1. — (a) Raster plots and peri-stimulus time histograms for activated RS example units in response to ramp light stimulation (3 s, 473 nm, 30 repetitions) of prefrontal PYRs at different ages. (b) Same as (a) for FS units. (c) Average firing rate change of RS (black, n = 1824 units from 66 recordings/mice) and FS (red, n = 226 units from 66 recordings/mice) units in response to ramp light stimulation of prefrontal L2/3 PYRs for different age groups. (d) Line plot displaying the average firing rate changes of RS and FS units during ramp light stimulation for different age groups. (e) Histograms of the MI of firing rates in response to ramp light stimulation for RS and FS units. (Average data is displayed as mean ± sem. See Supplementary file 1 for a summary of experimental conditions. See Supplementary file 2 for statistics.).

Figure 4—source data 1. Source data for Figure 4d.

elife-56795-fig4-data1.mat^{(508B, mat)}

The average firing rate of RS and FS units in the stimulated hemisphere in the mPFC increased in response to ramp stimulation (Figure 4c). While ramp-induced firing rate changes averaged for RS units (Mann-Kendall trend test, p=0.07, n = 7 age groups, tau-b 0.619) became more prominent at older age, the average firing rate changes were stable for FS units (Mann-Kendall trend test, p=0.88, n = 7 age groups, tau-b 0.047) (Figure 4d). At the level of individual units, most RS units showed positive modulation of their firing rates in response to stimulation at P5-10, whereas at older age the proportion of positively modulated units decreased (Mann-Kendall trend test, p=1.52×10⁻¹⁴, n = 1821 units, tau-b −0.123) (Figure 4e). In contrast, individual FS units showed both positive and negative modulation of activity throughout development (Mann-Kendall trend test, p=0.91, n = 225 units, tau-b −0.005), yet the low number of FS units at young age precluded clear conclusions. Thus, during early postnatal development most RS units are activated by ramp light stimulations but only moderately increase their firing rate. During late postnatal development some RS units strongly increase their firing rate, whereas others reduce their firing rate.

Next, we tested whether RS and FS units engage in rhythmic activity and calculated autocorrelations and spike-triggered LFP power of individual units. While no clear peaks of rhythmicity were found during spontaneous activity before stimulations (Figure 5—figure supplement 1), autocorrelations showed that a subset of RS and FS units fire rhythmically in response to ramp light stimulation of prefrontal L2/3 PYRs (Figure 5a). The power of autocorrelations revealed that prominent rhythmic firing starts at about P15 and increases in frequency before it stabilizes at about P25 (Figure 5b). A multifactorial ANOVA shows significant effects for age group (F(6,153093) = 25.4, p=2.21×10⁻³⁰) and frequency (F(99,153093) = 492.7, p=0.000), but not between RS and FS units (F(1,153093) = 1.56, p=0.211). Next, we calculated the power of averaged spike-triggered LFPs to examine the interaction of single unit rhythmicity with oscillatory LFP activity. The development of spike-triggered LFP power is consistent with the development of single unit autocorrelation power (multifactorial ANOVA: unit type F(1,814825) = 277.2, p=3.12×10⁻⁶², age group F(6,814825) = 847.4, p=0.000, frequency F(400, 814825)=614.5, p=0.000), indicating that single unit rhythmicity of local neurons is reflected in the prefrontal LFP. Consistent results were obtained for inter-spike intervals and pairwise phase consistency (Figure 5—figure supplement 2), as well as crosscorrelations between simultaneously recorded unit pairs (Figure 5—figure supplement 3). RS units show higher values at lower instantaneous frequencies for inter-spike intervals than FS units, suggesting that they tend to skip cycles of induced gamma activity (Figure 5—figure supplement 2d). Overall, the dynamics of RS and FS rhythmicity are similar to the development of spontaneous and stimulated gamma activity, indicating close interactions between RS and FS units during fast oscillations.

Figure 5. — (a) Color-coded autocorrelations of prefrontal RS (top) and FS (bottom) units during ramp light stimulation (3 s, 473 nm) for different age groups. Each row represents one unit (only units firing > 1 Hz are included). (b) Average autocorrelation power of RS (black) and FS (red) units during ramp light stimulation for different age groups. (c) Average power of mean spike-triggered LFP of RS (black) and FS (red) units during ramp light stimulation for different age groups. (Average data is displayed as mean ± sem. See Supplementary file 1 for a summary of experimental conditions. See Supplementary file 2 for statistics.).

Figure 5—figure supplement 1. — (a) Color-coded autocorrelations of prefrontal RS (top) and FS (bottom) units during ramp light stimulation (3 s, 473 nm) for different age groups. Each row represents one unit (only units firing > 1 Hz are included). (b) Average autocorrelation power of RS (black) and FS (red) units during ramp light stimulation for different age groups. (c) Average power of mean spike-triggered LFP of RS (black) and FS (red) units during ramp light stimulation for different age groups. (Average data is displayed as mean ± sem. See Supplementary file 1 for a summary of experimental conditions. See Supplementary file 2 for statistics.).

Inhibitory feedback maturation resembles the dynamics of gamma development

Stimulation of prefrontal L2/3 PYRs with short light pulses (3 ms, 473 nm) at different frequencies was used to test the maximal firing frequencies of RS and FS units in P5-40 mice. Pulse stimulations induced a short increase of firing for both RS and FS units (Figure 6a). Confirming previous results (Bitzenhofer et al., 2017a), RS units did not follow high stimulation frequencies in mice younger than P11 and showed strong attenuation in the response to repetitive pulses. With ongoing development, this attenuation at high stimulation frequencies became less prominent for RS and FS units (Figure 6b), yet the low number of FS units at young age precluded clear conclusions. Inter-spike intervals of individual units revealed several peaks at fractions of the stimulation frequency for RS and FS, especially at higher stimulation frequencies (Figure 6—figure supplement 1). These data suggest that individual units do not fire in response to every light pulse in the pulse train but skip some pulses.

Figure 6. — (a) Firing rate changes of prefrontal RS (black, n = 1824 units from 66 recordings/mice) and FS (red, n = 226 units from 66 recordings/mice) units in response to repetitive pulse light stimulation (3 ms, 473 nm) of 4, 8, 16, 32, and 64 Hz averaged for different age groups. (b) Line plots displaying the ratio of firing rate change in response to the 10^th versus the 1^st pulse for different frequencies and age groups (multifactorial ANOVA: unit type F(1,7698) = 0.39, p=0.530, age group F(6, 7698)=18.7, p=1.00×10⁻²¹, stimulation frequency F(4, 7698)=36.0, p=6.29×10⁻³⁰). (c) Firing rate changes of RS and FS units in response to pulse light stimulation (3 ms, 473 nm) of L2/3 PYRs averaged for different age groups. (d) Line plot displaying the average firing rate change of RS and FS units 0–10 ms after pulse light stimulation for different age groups. (e) Same as (c) displayed at longer time scale. (f) Same as (d) for 10–50 ms after pulse start. (Average data is displayed as mean ± sem. See Supplementary file 1 for a summary of experimental conditions. See Supplementary file 2 for statistics).

Figure 6—source data 1. Source data for Figure 6b.

elife-56795-fig6-data1.mat^{(1.3KB, mat)}

Figure 6—source data 2. Source data for Figure 6d.

elife-56795-fig6-data2.mat^{(514B, mat)}

Figure 6—source data 3. Source data for Figure 6f.

elife-56795-fig6-data3.mat^{(516B, mat)}

Figure 6—figure supplement 1. — (a) Firing rate changes of prefrontal RS (black, n = 1824 units from 66 recordings/mice) and FS (red, n = 226 units from 66 recordings/mice) units in response to repetitive pulse light stimulation (3 ms, 473 nm) of 4, 8, 16, 32, and 64 Hz averaged for different age groups. (b) Line plots displaying the ratio of firing rate change in response to the 10^th versus the 1^st pulse for different frequencies and age groups (multifactorial ANOVA: unit type F(1,7698) = 0.39, p=0.530, age group F(6, 7698)=18.7, p=1.00×10⁻²¹, stimulation frequency F(4, 7698)=36.0, p=6.29×10⁻³⁰). (c) Firing rate changes of RS and FS units in response to pulse light stimulation (3 ms, 473 nm) of L2/3 PYRs averaged for different age groups. (d) Line plot displaying the average firing rate change of RS and FS units 0–10 ms after pulse light stimulation for different age groups. (e) Same as (c) displayed at longer time scale. (f) Same as (d) for 10–50 ms after pulse start. (Average data is displayed as mean ± sem. See Supplementary file 1 for a summary of experimental conditions. See Supplementary file 2 for statistics).

Figure 6—source data 1. Source data for Figure 6b.

elife-56795-fig6-data1.mat^{(1.3KB, mat)}

Figure 6—source data 2. Source data for Figure 6d.

elife-56795-fig6-data2.mat^{(514B, mat)}

Figure 6—source data 3. Source data for Figure 6f.

elife-56795-fig6-data3.mat^{(516B, mat)}

To assess the development of inhibitory feedback, we examined the firing rate changes of RS and FS units from P5-40 mice in response to individual 3 ms-long light pulses. The firing rate of RS and FS units transiently increased after pulse stimulation (Figure 6c). This effect significantly increased with age for RS units (Mann-Kendall trend test, p=0.04, n = 7 age groups, tau-b 0.714), but not for FS units (Mann-Kendall trend test, p=0.07, n = 7 age groups, tau-b 0.619) (Figure 6d). Next, we analyzed the delays of light-induced firing peaks for the two populations of units. The similar delays observed for RS and FS units suggest that the majority of RS units are non-transfected neurons, that are indirectly activated. The initial peak of increased firing was followed by reduced firing rates for RS and FS units only during late postnatal development (Figure 6e,f). The magnitude and duration of this firing depression gradually augmented with age and reached significance for RS units (Mann-Kendall trend test, RS, p=6.9×10⁻³, n = 7 age groups, tau-b −0.905; FS, p=0.07, n = 7 age groups, tau-b −0.619). Thus, the inhibitory feedback in response to L2/3 PYR firing in the mPFC increases with age.

FS characteristics and rhythmicity of single units relate to the frequency of fast oscillatory activity

We used stepwise linear regression models to identify the most important correlations of single unit measures with gamma activity across development at the level of individual mice. The models included the following predictor variables (average values for each mouse): (i) proportion of FS units, action potential half width, trough-to-peak, and amplitude for RS and FS units (see Figure 2), (ii) firing rate change, peak power of autocorrelation, and peak power of spike-triggered LFP for RS and FS units during ramp light stimulation (see Figure 5), and (iii) firing rate change for RS and FS units 0–10 ms and 10–50 ms after pulse stimulation (see Figure 6). Stepwise linear regression to predict the peak frequency during ramp light stimulation identified action potential half width of FS units, as well as peak power of autocorrelation of RS and FS units as best predictor variables (n = 42, df = 38, R² = 0.136, p=0.036; see Supplementary file 2 for detailed statistics). Stepwise linear regression to predict the peak amplitude during ramp light stimulation identified firing rate change of RS units during ramp light stimulation, peak power of spike-triggered LFP for RS units, as well as RS firing rate change 0–10 ms after pulse stimulation and FS firing rate change 10–50 ms after pulse stimulation as best predictor variables (n = 48, df = 43, R² = 0.716, p=4.35×10⁻¹²; see Supplementary file 2 for detailed statistics).

Thus, peak amplitude during ramp light stimulation is correlated with firing rate changes in response to ramp and pulse stimulations, indicating the importance of stimulation efficacy for peak amplitude. On the other hand, FS characteristics and rhythmicity of single units show the strongest correlation with the peak frequency of LFP power.

Discussion

Gamma oscillations (30–80 Hz) result from a fine-tuned interplay between excitation and inhibition in the adult brain (Atallah and Scanziani, 2009). For example, in sensory cortices, they arise through inhibitory feedback from PV-expressing interneurons (Cardin et al., 2009; Chen et al., 2017). Specifically, fast synaptic inhibition of the perisomatic region of pyramidal neurons by PV-expressing interneurons is critical for the generation of gamma oscillations (Cardin, 2016). Suppressing PV-expressing interneurons in the adult mPFC reduces the power in gamma frequency and impairs interhemispheric synchronization and cognitive abilities (Cho et al., 2015; Cho et al., 2020). SOM-expressing interneurons contribute to oscillatory activity at the lower end of the gamma frequency range (20–40 Hz) (Chen et al., 2017; Veit et al., 2017). In contrast, the mechanisms controlling the emergence of gamma activity during development are still poorly understood. Here, we reveal that the fast oscillatory activity in the mouse mPFC emerges during the second postnatal week and increases in frequency and amplitude before it stabilizes in gamma frequency range (30–80 Hz) during the fourth postnatal week. Further, we show that the functional maturation of FS PV-expressing interneurons and single unit rhythmicity of RS and FS units are best correlated with the accelerating gamma activity in the LFP. While activation of L2/3 PYR drives fast oscillatory activity throughout development, the acceleration toward higher frequencies relates to the maturation of inhibitory feedback and of FS interneurons. The time course of the maturation of FS units, putatively PV-expressing interneurons, is consistent with a recent characterization of PV-expressing interneuron physiology in the mPFC in vitro (Miyamae et al., 2017). These results suggest that the interplay between excitatory and inhibitory neurons is not only critical for the generation of adult gamma activity but also for its emergence during postnatal development.

The L2/3 PYRs-driven oscillatory activity was strongly correlated with spontaneously occurring fast frequency oscillations. However, broader peaks in power spectra for spontaneous activity indicate that the stimulation-induced activity might not cover the full diversity of spontaneous gamma activity. The broadness of spontaneous activity gamma power might also explain the absence of clear locking and rhythmicity of single units in the absence of prefrontal stimulation or task-induced activation of the mPFC.

Starting with the first electroencephalographic recordings, adult brain rhythms have been defined according to their frequencies and related to a specific state or task (Buzsáki and Draguhn, 2004). These ‘classical’ frequency bands (i.e. delta, theta, alpha, beta, gamma) are largely preserved between different mammalian species (Buzsáki and Draguhn, 2004; Buzsáki et al., 2013). However, how they emerge during development is still largely unknown. Synchronization of cortical areas in fast oscillatory rhythms starts during the first postnatal week (Brockmann et al., 2011). These low-amplitude patterns are detected in the rodent mPFC as early as P5,1–2 days later compared to primary sensory cortices (S1, V1) (Minlebaev et al., 2011; Yang et al., 2013; Dupont et al., 2006; Shen and Colonnese, 2016; Yang et al., 2009). However, this neonatal fast activity has a relative low frequency (around 20 Hz) (Brockmann et al., 2011; Yang et al., 2009). It is organized in infrequent short bursts and its detection is hampered by the lack of a clear peak in LFP power spectra. We previously showed that in the developing mPFC the detection of prominent fast oscillations with frequencies above 12 Hz coincides with the switch from discontinuous to continuous activity (Brockmann et al., 2011). These oscillations are initially within 15-20 Hz frequency range that was classically defined as beta range. The present results indicate that these rhythms progressively increase their frequency and amplitude with age until they stabilize in gamma frequency range at 50–60 Hz during the fourth postnatal week. A similar increase in gamma frequency has been previously described after eye opening in the visual cortex (Hoy and Niell, 2015). Therefore, identification of oscillatory patterns in developing circuits according to ‘classical’ frequency bands established for adults should be avoided.

Adult gamma activity in the cerebral cortex relies on FS PV-expressing interneurons (Cardin et al., 2009). To test whether this mechanism underlies the fast rhythms in the developing brain, we developed an unbiased approach to detect FS units corresponding to putatively PV-expressing interneurons. Since PV expression and FS characteristics do not completely overlap (Ma et al., 2006; Onorato et al., 2020), the method has the same drawback as that typically used for the distinction of adult RS and FS and cannot identify RS interneurons. For this, clustering of prefrontal neurons from mice of all investigated ages was performed based on a dimensionality reduction of their mean waveforms and not on pre-defined waveform features. To validate this approach, we compared the results to pre-defined waveform features typically used to identify FS units and found that they largely agree for adult mice. We demonstrate that FS units are detected in the mPFC during the second postnatal week and progressively mature until the fourth postnatal week, consistent with PV interneuron maturation (Okaty et al., 2009). The similar dynamics of FS interneuron maturation and acceleration of fast oscillatory activity, as well as the correlation of FS characteristics with induced fast oscillation frequencies across age supports the hypothesis that FS interneurons are key elements for prefrontal gamma development. In the absence of FS interneurons at early age, inhibitory feedback from SOM neurons – important for slow gamma activity in the adult cortex (Veit et al., 2017) – might contribute to early oscillatory activity at frequencies within 12–20 Hz range.

While we only found minor age-dependent changes in the extracellular waveforms of RS units, an in-depth investigation of prefrontal PYRs during development had identified prominent changes in their dendritic arborization, passive and active membrane properties, as well as excitatory and inhibitory inputs (Kroon et al., 2019). These changes, even though not detected with extracellular recordings, most likely contribute to the maturation of pyramidal-interneuronal interactions and finally, of gamma activity. Indeed, we found that the maturation of inhibitory feedback in response to prefrontal L2/3 PYRs stimulation follows the same dynamics as the development of gamma oscillations. Furthermore, rhythmicity of single RS and FS units were identified as good predictors for the frequencies of oscillatory activity.

GABAergic transmission in the rodent cortex matures during postnatal development, reaching an adult-like state toward the end of the fourth postnatal week (Le Magueresse and Monyer, 2013; Butt et al., 2017; Lim et al., 2018). Shortly after birth, GABA acts depolarizing due to high intracellular chloride in immature neurons expressing low levels of the chloride cotransporter KCC2 relative to NKCC1 (Rivera et al., 1999; Lim et al., 2018). However, this depolarization is not sufficient to trigger action potential firing and results in shunting inhibition (Kirmse et al., 2018). The switch of GABA action from depolarizing to hyperpolarizing has been reported to occur during the second postnatal week (Ben-Ari et al., 2012), coinciding with the emergence of gamma band oscillations. Moreover, the composition of GABA_A-receptor subunits changes during postnatal development, causing a progressive decrease of decay-time constants of inhibitory postsynaptic currents (IPSCs) until they reach adult-like kinetics in the fourth postnatal week (Okaty et al., 2009; Bosman et al., 2005; Laurie et al., 1992). Simulations of neuronal networks proposed that increasing IPSCs kinetics in FS interneurons results in increasing gamma frequency (Doischer et al., 2008). The gradual increase of prefrontal gamma frequency from the second to the fourth postnatal week provides experimental evidence for this hypothesis.

In-depth understanding of the dynamics and mechanisms of gamma activity in the developing cortex appears relevant for neurodevelopmental disorders, such as schizophrenia and autism. Both, in patients and disease mouse models, gamma oscillations have been reported to be altered, likely to reflect abnormal pyramidal-interneuronal interactions (Cho et al., 2015; Cao et al., 2018; Rojas and Wilson, 2014). These dysfunction seems to emerge already during development (Chini et al., 2020; Richter et al., 2019; Hartung et al., 2016). Elucidating the developmental dynamics of cortical gamma activity might uncover the timeline of disease-related deficits.

Materials and methods

Key resources table.

Reagent type (species) or resource	Designation	Source or reference	Identifiers	Additional information
Antibody	Rabbit polyclonal-anti-parvalbumin	Abcam	ab11427	(1:500)
Antibody	Rabbit polyclonal-anti-somatostatin	Santa Cruz	sc13099	(1:250)
Antibody	Goat-anti-rabbit secondary antibody, Alexa Fluor 488	Invitrogen-Thermo Fisher	A11008	(1:500)
Chemical compound, drug	Isoflurane	Abbott	B506
Chemical compound, drug	Urethane	Fluka analytical	94300
Strain, strain background (mouse, both genders)	C57Bl/6J	Universitätsklinikum Hamburg-Eppendorf –Animal facility	C57Bl/6J	https://www.jax.org/strain/008199
Recombinant DNA reagent	pAAV-CAG-ChR2(E123T/T159C)−2AtDimer2	Provided by T. G. Oertner	pAAV-CAG-ChR2(E123T/T159C)−2AtDimer2	http://www.oertner.com/
Software, algorithm	Matlab R2018b	MathWorks	Matlab R2018b	https://www.mathworks.com/
Software, algorithm	Kilosort2	MouseLand		https://github.com/MouseLand/Kilosort2
Software, algorithm	ImageJ	ImageJ		https://imagej.nih.gov/ij/
Other	Arduino Uno SMD	Arduino	A000073
Other	Digital Lynx 4SX	Neuralynx	Digital Lynx 4SX	http://neuralynx.com/
Other	Diode laser (473 nm)	Omicron	LuxX 473–100
Other	Electroporation device	BEX	CUY21EX
Other	Electroporation tweezer-type paddles	Protech	CUY650-P5
Other	Recording electrode (one-shank, 16 channels)	Neuronexus	A1 × 16 5 mm
Other	Recording electrode (four-shank, 16 channels)	Neuronexus	A4 × 4 5 mm

Open in a new tab

Animals

All experiments were performed in compliance with the German laws and the guidelines of the European Community for the use of animals in research and were approved by the local ethical committee (G132/12, G17/015, N18/015). Timed-pregnant mice from the animal facility of the University Medical Center Hamburg-Eppendorf were housed individually at a 12 hr light/12 hr dark cycle and were given access to water and food ad libitum. The day of vaginal plug detection was considered embryonic day (E) 0.5, the day of birth was considered postnatal day (P) 0. Experiments were carried out on C57Bl/6J mice of both sexes.

In utero electroporation (IUE)

Pregnant mice (C57Bl6/J, The Jackson Laboratory, ME, USA) received additional wet food daily, supplemented with 2–4 drops Metacam (0.5 mg/ml, Boehringer-Ingelheim, Germany) one day before until two days after in IUE. At E15.5, pregnant mice were injected subcutaneously with buprenorphine (0.05 mg/kg body weight) 30 min before surgery. Surgery was performed under isoflurane anesthesia (induction 5%, maintenance 3.5%) on a heating blanket. Eyes were covered with eye ointment and pain reflexes and breathing were monitored to assess anesthesia depth. Uterine horns were exposed and moistened with warm sterile PBS. 0.75–1.25 µl of opsin- and fluorophore-encoding plasmid (pAAV-CAG-ChR2(E123T/T159C)−2A-tDimer2, 1.25 µg/µl) purified with NucleoBond (Macherey-Nagel, Germany) in sterile PBS with 0.1% fast green dye was injected in the right lateral ventricle of each embryo using pulled borosilicate glass capillaries. Electroporation tweezer paddles of 5 mm diameter were oriented at a rough 20° leftward angle from the midline of the head and a rough 10° downward angle from the anterior to posterior axis to transfect precursor cells of medial prefrontal layer 2/3 PYRs neurons with five electroporation pulses (35 V, 50 ms, 950 ms interval, CU21EX, BEX, Japan). Uterine horns were placed back into the abdominal cavity. Abdominal cavity was filled with warm sterile PBS and abdominal muscles and skin were sutured with absorbable and non-absorbable suture thread, respectively. After recovery from anesthesia, mice were returned to their home cage, placed half on a heating blanket for two days after surgery. Fluorophore expression was assessed at P2 in the pups with a portable fluorescence flashlight (Nightsea, MA, USA) through the intact skin and skull and confirmed in brain slices postmortem.

Electrophysiology

Acute head-fixed recordings

Multi-site extracellular recordings were performed unilaterally or bilaterally in the mPFC of non-anesthetized or anesthetized P5-40 mice. Mice were on a heating blanket during the entire procedure. Under isoflurane anesthesia (induction: 5%; maintenance: 2.5%), a craniotomy was performed above the mPFC (0.5 mm anterior to bregma, 0.1–0.5 mm lateral to the midline). Pups were head-fixed into a stereotaxic apparatus using two plastic bars mounted on the nasal and occipital bones with dental cement. Multi-site electrodes (NeuroNexus, MI, USA) were inserted into the mPFC (four-shank, A4 × 4 recording sites, 100 µm spacing, 125 µm shank distance, 1.8–2.0 mm deep). A silver wire was inserted into the cerebellum and served as ground and reference. For non-anesthetized and anesthetized recordings, pups were allowed to recover for 30 min prior to recordings. For anesthetized recordings, urethane (1 mg/g body weight) was injected intraperitoneally prior to the surgery.

Acute head-fixed recordings with chronically implanted head-fixation adapters

Multisite extracellular recordings were performed unilaterally in the mPFC of P23-25 and P38-40 mice. The adapter for head fixation was implanted at least 5 days before recordings. Under isoflurane anesthesia (5% induction, 2.5% maintenance), a metal head-post (Luigs and Neumann, Germany) was attached to the skull with dental cement and a craniotomy was performed above the mPFC (0.5–2.0 mm anterior to bregma, 0.1–0.5 mm right to the midline) and protected by a customized synthetic window. A silver wire was implanted between skull and brain tissue above the cerebellum and served as ground and reference. 0.5% bupivacaine/1% lidocaine was locally applied to cutting edges. After recovery from anesthesia, mice were returned to their home cage. After recovery from the surgery, mice were accustomed to head-fixation and trained to run on a custom-made spinning disc. For non-anesthetized recordings, craniotomies were uncovered and multi-site electrodes (NeuroNexus, MI, USA) were inserted into the mPFC (one-shank, A1 × 16 recording sites, 100 µm spacing, 2.0 mm deep).

Extracellular signals were band-pass filtered (0.1–9000 Hz) and digitized (32 kHz) with a multichannel extracellular amplifier (Digital Lynx SX; Neuralynx, Bozeman, MO, USA). Electrode position was confirmed in brain slices postmortem.

Optogenetic stimulation

Ramp (linearly increasing light power) light stimulation and pulsed (short pulses of 3 ms) light stimulation at different frequencies was performed using an Arduino uno (Arduino, Italy) controlled laser system (473 nm / 594 nm wavelength, Omicron, Austria) coupled to a 50 µm (four- shank electrodes) or 105 µm (one-shank electrodes) diameter light fiber (Thorlabs, NJ, USA) glued to the multisite electrodes, ending 200 µm above the top recording site. Each type of stimulation was repeated 30 times. At the beginning of each recording, laser power was adjusted to reliably trigger neuronal spiking in response to light pulses of 3 ms duration.

Histology

Mice (P5-40) were anesthetized with 10% ketamine (aniMedica, Germanry)/2% xylazine (WDT, Germany) in 0.9% NaCl (10 µg/g body weight, intraperitoneal) and transcardially perfused with 4% paraformaldehyde (Histofix, Carl Roth, Germany). Brains were removed and postfixed in 4% paraformaldehyde for 24 hr. Brains were sectioned coronally with a vibratom at 50 µm for immunohistochemistry.

Immunohistochemistry

Free-floating slices were permeabilized and blocked with PBS containing 0.8% Triton X-100 (Sigma-Aldrich, MO, USA), 5% normal bovine serum (Jackson Immuno Research, PA, USA) and 0.05% sodium azide. Slices were incubated over night with primary antibody rabbit-anti-parvalbumin (1:500, #ab11427, Abcam, UK) or rabbit-anti-somatostatin (1:250, #sc13099, Santa Cruz, CA, USA), followed by 2 hr incubation with secondary antibody goat-anti-rabbit Alexa Fluor 488 (1:500, #A11008, Invitrogen-Thermo Fisher, MA, USA). Sections were transferred to glass slides and covered with Fluoromount (Sigma-Aldrich, MO, USA).

Cell quantification

Images of immunofluorescence in the right mPFC were acquired with a confocal microscope (DM IRBE, Leica, Germany) using a 10x objective (numerical aperture 0.3). Immunopositive cells were automatically quantified with custom-written algorithms in ImageJ environment. The region of interest (ROI) was manually defined over L2/3 of the mPFC. Image contrast was enhanced before applying a median filter. Local background was subtracted to reduce background noise and images were binarized and segmented using the watershed function. Counting was done after detecting the neurons with the extended maxima function of the MorphoLibJ plugin.

Data analysis

Electrophysiological data were analyzed with custom-written algorithms in Matlab environment. Data were band-pass filtered (500–9000 Hz for spike analysis or 1–100 Hz for local field potentials (LFP)) using a third-order Butterworth filter forward and backward to preserve phase information before down-sampling to analyze LFP. Each type of optogenetic stimulation (ramps or pulses at different frequencies) was repeated 30 times for each recording. Each recording contributes a single data point for Figures 1 and 3. Most mice were recorded once. Only mice with chronically implanted head fixation were recorded more than one time (13 recordings from six mice at P24-26 and 12 recordings from five mice at P37-40), but all recordings were with acutely inserted electrodes. All mice for single unit analysis were recorded only once. (See Supplementary file 1 for a summary of experimental conditions.)

Power spectral density

For power spectral density analysis, 2 s-long windows of LFP signal were concatenated and the power was calculated using Welch’s method with non-overlapping windows. Spectra were multiplied with squared frequency.

Modulation index

For optogenetic stimulations, MI was calculated as (value stimulation - value pre- stimulation) / (value stimulation + value pre-stimulation).

Peak frequency and strength

Peak frequency and peak strength were calculated for the most prominent peak in the spectrum defined by the product of peak amplitude, peak half width and peak prominence.

Single unit analysis

Spikes were detected and sorted with Kilosort2 in Matlab. t-sne dimensionality reduction was applied on mean waveforms of all units. Hierarchical clustering was performed to identify FS and RS units for all ages simultaneously. Autocorrelations of single units with a minimum firing rate of 1 Hz were calculated before and during optogenetic ramp stimulation. Power spectral densities of mean autocorrelations were calculated per unit. Crosscorrelations of simultaneously recorded RS-RS, FS-FS, and RS-FS unit pairs were calculated during optogenetic ramp stimulation. Power of the mean spike-triggered LFP, pairwise phase consistency (Vinck et al., 2010), coefficient of variation, C_V2 (Holt et al., 1996), and inter-spike intervals were calculated for individual units before and during optogenetic ramp stimulation.

Statistics

No statistical measures were used to estimate sample size since effect size was unknown. Data were tested for consistent trends across age with the non-parametric Mann-Kendall trend test. Mann-Kendall coefficient tau-b adjusting for ties is reported. Multifactorial ANOVAs were used to compare main effects. Linear regression models were used to test for significant links of averaged single unit and LFP activity measures with age per recording. Stepwise linear regression was used to identify significant links from averaged single unit activity measures (proportion FS units, as well as half width, trough to peak, amplitude, rate change for ramp light stimulation, autocorrelation peak during ramp light stimulation, spike-triggered LFP peak frequency during ramp stimulation, and rate change early (0–10 ms) and late (10–50 ms) after pulse stimulation for RS and FS units) with LFP activity (ramp light-induced frequency peak and amplitude). See Supplementary file 2 for detailed statistics.

Acknowledgements

We thank M Chini for helpful discussions and comments on the manuscript as well as A Marquardt, P Putthoff, A Dahlmann, and K Titze for excellent technical assistance. This work was funded by grants from the European Research Council (ERC-2015-CoG 681577 to ILH-O) and the German Research Foundation (Ha 4466/10–1, Ha4466/11-1, Ha4466/12-1, SPP 1665, SFB 936 B5 to ILH-O).

Funding Statement

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Contributor Information

Sebastian H Bitzenhofer, Email: sbitzenhofer@ucsd.edu.

Ileana Hanganu-Opatz, Email: ileana.hanganu-opatz@zmnh.uni-hamburg.de.

Martin Vinck, Ernst Strüngmann Institute (ESI) for Neuroscience in Cooperation with Max Planck Society, Germany.

Laura L Colgin, University of Texas at Austin, United States.

Funding Information

This paper was supported by the following grants:

H2020 European Research Council ERC-2015-CoG 681577 to Ileana Hanganu-Opatz.
Deutsche Forschungsgemeinschaft Ha4466/10-1 to Ileana Hanganu-Opatz.
Deutsche Forschungsgemeinschaft Ha4466/11-1 to Ileana Hanganu-Opatz.
Deutsche Forschungsgemeinschaft Ha4466/12-1 to Ileana Hanganu-Opatz.
Deutsche Forschungsgemeinschaft SPP 1665 to Ileana Hanganu-Opatz.
Deutsche Forschungsgemeinschaft SFB 936 B5 to Ileana Hanganu-Opatz.

Additional information

Competing interests

No competing interests declared.

Author contributions

Conceptualization, Data curation, Software, Formal analysis, Supervision, Investigation, Visualization, Methodology, Writing - original draft, Writing - review and editing.

Data curation, Formal analysis, Investigation, Methodology, Writing - review and editing.

Conceptualization, Supervision, Funding acquisition, Project administration, Writing - review and editing.

Ethics

Animal experimentation: All experiments were performed in compliance with the German laws and the guidelines of the European Community for the use of animals in research and were approved by the local ethical committee (G132/12, G17/015, N18/015).

Additional files

Supplementary file 1. Recording summary.

Table summarizing the recordings for each experimental condition.

elife-56795-supp1.docx^{(35.7KB, docx)}

Supplementary file 2. Detailed statistical results.

Table summarizing the statistical results.

elife-56795-supp2.docx^{(55KB, docx)}

Transparent reporting form

elife-56795-transrepform.pdf^{(219.2KB, pdf)}

Data availability

The authors declare that all data and code supporting the findings of this study are included in the manuscript. LFP and SUA data for all recordings is available at the following open-access repository: https://doi.org/10.12751/g-node.heyl6r.

The following dataset was generated:

Bitzenhofer SB, Pöpplau A, Hanganu-Opatz IL. 2020. Ephys data associated with the paper *Gamma activity accelerates during prefrontal development*. g-node.

References

Atallah BV, Scanziani M. Instantaneous modulation of gamma oscillation frequency by balancing excitation with inhibition. Neuron. 2009;62:566–577. doi: 10.1016/j.neuron.2009.04.027. [DOI] [PMC free article] [PubMed] [Google Scholar]
Ben-Ari Y, Khalilov I, Kahle KT, Cherubini E. The GABA excitatory/inhibitory shift in brain maturation and neurological disorders. The Neuroscientist. 2012;18:467–486. doi: 10.1177/1073858412438697. [DOI] [PubMed] [Google Scholar]
Bitzenhofer SH, Ahlbeck J, Wolff A, Wiegert JS, Gee CE, Oertner TG, Hanganu-Opatz IL. Layer-specific optogenetic activation of pyramidal neurons causes beta-gamma entrainment of neonatal networks. Nature Communications. 2017a;8:14563. doi: 10.1038/ncomms14563. [DOI] [PMC free article] [PubMed] [Google Scholar]
Bitzenhofer SH, Ahlbeck J, Hanganu-Opatz IL. Methodological approach for optogenetic manipulation of neonatal neuronal networks. Frontiers in Cellular Neuroscience. 2017b;11:239. doi: 10.3389/fncel.2017.00239. [DOI] [PMC free article] [PubMed] [Google Scholar]
Bosman LW, Heinen K, Spijker S, Brussaard AB. Mice lacking the major adult GABAA receptor subtype have normal number of synapses, but retain juvenile IPSC kinetics until adulthood. Journal of Neurophysiology. 2005;94:338–346. doi: 10.1152/jn.00084.2005. [DOI] [PubMed] [Google Scholar]
Brockmann MD, Pöschel B, Cichon N, Hanganu-Opatz IL. Coupled oscillations mediate directed interactions between prefrontal cortex and Hippocampus of the neonatal rat. Neuron. 2011;71:332–347. doi: 10.1016/j.neuron.2011.05.041. [DOI] [PubMed] [Google Scholar]
Butt SJ, Stacey JA, Teramoto Y, Vagnoni C. A role for GABAergic interneuron diversity in circuit development and plasticity of the neonatal cerebral cortex. Current Opinion in Neurobiology. 2017;43:149–155. doi: 10.1016/j.conb.2017.03.011. [DOI] [PubMed] [Google Scholar]
Buzsáki G, Logothetis N, Singer W. Scaling Brain Size, Keeping Timing: Evolutionary Preservation of Brain Rhythms. Neuron. 2013;80:751–764. doi: 10.1016/j.neuron.2013.10.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
Buzsáki G, Draguhn A. Neuronal oscillations in cortical networks. Science. 2004;304:1926–1929. doi: 10.1126/science.1099745. [DOI] [PubMed] [Google Scholar]
Cao W, Lin S, Xia QQ, Du YL, Yang Q, Zhang MY, Lu YQ, Xu J, Duan SM, Xia J, Feng G, Xu J, Luo JH. Gamma oscillation dysfunction in mPFC leads to social deficits in neuroligin 3 R451C knockin mice. Neuron. 2018;97:1253–1260. doi: 10.1016/j.neuron.2018.02.001. [DOI] [PubMed] [Google Scholar]
Cardin JA, Carlén M, Meletis K, Knoblich U, Zhang F, Deisseroth K, Tsai LH, Moore CI. Driving fast-spiking cells induces gamma rhythm and controls sensory responses. Nature. 2009;459:663–667. doi: 10.1038/nature08002. [DOI] [PMC free article] [PubMed] [Google Scholar]
Cardin JA. Snapshots of the brain in action: local circuit operations through the Lens of γ oscillations. Journal of Neuroscience. 2016;36:10496–10504. doi: 10.1523/JNEUROSCI.1021-16.2016. [DOI] [PMC free article] [PubMed] [Google Scholar]
Chen G, Zhang Y, Li X, Zhao X, Ye Q, Lin Y, Tao HW, Rasch MJ, Zhang X. Distinct inhibitory circuits orchestrate cortical beta and gamma band oscillations. Neuron. 2017;96:1403–1418. doi: 10.1016/j.neuron.2017.11.033. [DOI] [PMC free article] [PubMed] [Google Scholar]
Chini M, Gretenkord S, Kostka JK, Pöpplau JA, Cornelissen L, Berde CB, Hanganu-Opatz IL, Bitzenhofer SH. Neural correlates of anesthesia in newborn mice and humans. Frontiers in Neural Circuits. 2019;13:00038. doi: 10.3389/fncir.2019.00038. [DOI] [PMC free article] [PubMed] [Google Scholar]
Chini M, Pöpplau JA, Lindemann C, Carol-Perdiguer L, Hnida M, Oberländer V, Xu X, Ahlbeck J, Bitzenhofer SH, Mulert C, Hanganu-Opatz IL. Resolving and rescuing developmental miswiring in a mouse model of cognitive impairment. Neuron. 2020;105:60–74. doi: 10.1016/j.neuron.2019.09.042. [DOI] [PMC free article] [PubMed] [Google Scholar]
Cho KK, Hoch R, Lee AT, Patel T, Rubenstein JL, Sohal VS. Gamma rhythms link prefrontal interneuron dysfunction with cognitive inflexibility in Dlx5/6+/− mice. Neuron. 2015;85:1332–1343. doi: 10.1016/j.neuron.2015.02.019. [DOI] [PMC free article] [PubMed] [Google Scholar]
Cho KKA, Davidson TJ, Bouvier G, Marshall JD, Schnitzer MJ, Sohal VS. Cross-hemispheric gamma synchrony between prefrontal parvalbumin interneurons supports behavioral adaptation during rule shift learning. Nature Neuroscience. 2020;23:892–902. doi: 10.1038/s41593-020-0647-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
Doischer D, Hosp JA, Yanagawa Y, Obata K, Jonas P, Vida I, Bartos M. Postnatal differentiation of basket cells from slow to fast signaling devices. Journal of Neuroscience. 2008;28:12956–12968. doi: 10.1523/JNEUROSCI.2890-08.2008. [DOI] [PMC free article] [PubMed] [Google Scholar]
Dupont E, Hanganu IL, Kilb W, Hirsch S, Luhmann HJ. Rapid developmental switch in the mechanisms driving early cortical columnar networks. Nature. 2006;439:79–83. doi: 10.1038/nature04264. [DOI] [PubMed] [Google Scholar]
Hartung H, Cichon N, De Feo V, Riemann S, Schildt S, Lindemann C, Mulert C, Gogos JA, Hanganu-Opatz IL. From Shortage to Surge: A Developmental Switch in Hippocampal–Prefrontal Coupling in a Gene–Environment Model of Neuropsychiatric Disorders. Cerebral Cortex. 2016;26:4265–4281. doi: 10.1093/cercor/bhw274. [DOI] [PMC free article] [PubMed] [Google Scholar]
Holt GR, Softky WR, Koch C. Comparison of discharge variability visual cortex neurons . Journal of Neurophysiology. 1996;75:1806–1814. doi: 10.1152/jn.1996.75.5.1806. [DOI] [PubMed] [Google Scholar]
Hoy JL, Niell CM. Layer-specific refinement of visual cortex function after eye opening in the awake mouse. Journal of Neuroscience. 2015;35:3370–3383. doi: 10.1523/JNEUROSCI.3174-14.2015. [DOI] [PMC free article] [PubMed] [Google Scholar]
Kirmse K, Hübner CA, Isbrandt D, Witte OW, Holthoff K. GABAergic transmission during brain development: multiple effects at multiple stages. The Neuroscientist. 2018;24:36–53. doi: 10.1177/1073858417701382. [DOI] [PubMed] [Google Scholar]
Kroon T, van Hugte E, van Linge L, Mansvelder HD, Meredith RM. Early postnatal development of pyramidal neurons across layers of the mouse medial prefrontal cortex. Scientific Reports. 2019;9:5037. doi: 10.1038/s41598-019-41661-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
Laurie DJ, Wisden W, Seeburg PH. The distribution of thirteen GABAA receptor subunit mRNAs in the rat brain. III. embryonic and postnatal development. The Journal of Neuroscience. 1992;12:4151–4172. doi: 10.1523/JNEUROSCI.12-11-04151.1992. [DOI] [PMC free article] [PubMed] [Google Scholar]
Le Magueresse C, Monyer H. GABAergic interneurons shape the functional maturation of the cortex. Neuron. 2013;77:388–405. doi: 10.1016/j.neuron.2013.01.011. [DOI] [PubMed] [Google Scholar]
Lim L, Mi D, Llorca A, Marín O. Development and functional diversification of cortical interneurons. Neuron. 2018;100:294–313. doi: 10.1016/j.neuron.2018.10.009. [DOI] [PMC free article] [PubMed] [Google Scholar]
Ma Y, Hu H, Berrebi AS, Mathers PH, Agmon A. Distinct Subtypes of Somatostatin-Containing Neocortical Interneurons Revealed in Transgenic Mice. Journal of Neuroscience. 2006;26:5069–5082. doi: 10.1523/JNEUROSCI.0661-06.2006. [DOI] [PMC free article] [PubMed] [Google Scholar]
Minlebaev M, Colonnese M, Tsintsadze T, Sirota A, Khazipov R. Early γ oscillations synchronize developing thalamus and cortex. Science. 2011;334:226–229. doi: 10.1126/science.1210574. [DOI] [PubMed] [Google Scholar]
Miyamae T, Chen K, Lewis DA, Gonzalez-Burgos G. Distinct physiological maturation of Parvalbumin-Positive neuron subtypes in mouse prefrontal cortex. The Journal of Neuroscience. 2017;37:4883–4902. doi: 10.1523/JNEUROSCI.3325-16.2017. [DOI] [PMC free article] [PubMed] [Google Scholar]
Okaty BW, Miller MN, Sugino K, Hempel CM, Nelson SB. Transcriptional and electrophysiological maturation of neocortical fast-spiking GABAergic interneurons. Journal of Neuroscience. 2009;29:7040–7052. doi: 10.1523/JNEUROSCI.0105-09.2009. [DOI] [PMC free article] [PubMed] [Google Scholar]
Onorato I, Neuenschwander S, Hoy J, Lima B, Rocha K-S, Broggini AC, Uran C, Spyropoulos G, Klon-Lipok J, Womelsdorf T, Fries P, Niell C, Singer W, Vinck M. A Distinct Class of Bursting Neurons with Strong Gamma Synchronization and Stimulus Selectivity in Monkey V1. Neuron. 2020;105:180–197. doi: 10.1016/j.neuron.2019.09.039. [DOI] [PubMed] [Google Scholar]
Richter M, Murtaza N, Scharrenberg R, White SH, Johanns O, Walker S, Yuen RKC, Schwanke B, Bedürftig B, Henis M, Scharf S, Kraus V, Dörk R, Hellmann J, Lindenmaier Z, Ellegood J, Hartung H, Kwan V, Sedlacik J, Fiehler J, Schweizer M, Lerch JP, Hanganu-Opatz IL, Morellini F, Scherer SW, Singh KK, Calderon de Anda F. Altered TAOK2 activity causes autism-related neurodevelopmental and cognitive abnormalities through RhoA signaling. Molecular Psychiatry. 2019;24:1329–1350. doi: 10.1038/s41380-018-0025-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
Rivera C, Voipio J, Payne JA, Ruusuvuori E, Lahtinen H, Lamsa K, Pirvola U, Saarma M, Kaila K. The K+/Cl- co-transporter KCC2 renders GABA hyperpolarizing during neuronal maturation. Nature. 1999;397:251–255. doi: 10.1038/16697. [DOI] [PubMed] [Google Scholar]
Rojas DC, Wilson LB. γ-band abnormalities as markers of autism spectrum disorders. Biomarkers in Medicine. 2014;8:353–368. doi: 10.2217/bmm.14.15. [DOI] [PMC free article] [PubMed] [Google Scholar]
Shen J, Colonnese MT. Development of Activity in the Mouse Visual Cortex. Journal of Neuroscience. 2016;36:12259–12275. doi: 10.1523/JNEUROSCI.1903-16.2016. [DOI] [PMC free article] [PubMed] [Google Scholar]
Singer W. Neuronal oscillations: unavoidable and useful? European Journal of Neuroscience. 2018;48:2389–2398. doi: 10.1111/ejn.13796. [DOI] [PubMed] [Google Scholar]
Sohal VS. How close are we to understanding what (if anything) γ oscillations do in cortical circuits? Journal of Neuroscience. 2016;36:10489–10495. doi: 10.1523/JNEUROSCI.0990-16.2016. [DOI] [PMC free article] [PubMed] [Google Scholar]
Veit J, Hakim R, Jadi MP, Sejnowski TJ, Adesnik H. Cortical gamma band synchronization through somatostatin interneurons. Nature Neuroscience. 2017;20:951–959. doi: 10.1038/nn.4562. [DOI] [PMC free article] [PubMed] [Google Scholar]
Vinck M, van Wingerden M, Womelsdorf T, Fries P, Pennartz CM. The pairwise phase consistency: a bias-free measure of rhythmic neuronal synchronization. NeuroImage. 2010;51:112–122. doi: 10.1016/j.neuroimage.2010.01.073. [DOI] [PubMed] [Google Scholar]
Yang JW, Hanganu-Opatz IL, Sun JJ, Luhmann HJ. Three patterns of oscillatory activity differentially synchronize developing neocortical networks in vivo. Journal of Neuroscience. 2009;29:9011–9025. doi: 10.1523/JNEUROSCI.5646-08.2009. [DOI] [PMC free article] [PubMed] [Google Scholar]
Yang JW, An S, Sun JJ, Reyes-Puerta V, Kindler J, Berger T, Kilb W, Luhmann HJ. Thalamic network oscillations synchronize ontogenetic columns in the newborn rat barrel cortex. Cerebral Cortex. 2013;23:1299–1316. doi: 10.1093/cercor/bhs103. [DOI] [PubMed] [Google Scholar]

eLife. doi: 10.7554/eLife.56795.sa1

Decision letter

Editor: Martin Vinck¹

Reviewed by: Quentin Perrenoud², Christopher I Moore³

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Acceptance summary:

Bitzenhofer et al., show that fast network oscillations in the γ-frequency range become progressively stronger during early development, and stabilize around the fourth postnatal week. This development shows a similar temporal progression and is correlated with the maturation of fast spiking interneurons, a cell class that is known to regulate fast network oscillations. These findings provide important new insights into the development of γ oscillations and will be valuable for understanding the development of neuropsychiatric diseases linked to GABAergic dysfunction and abnormal oscillatory activity.

Decision letter after peer review:

Thank you for submitting your article "Γ activity accelerates during prefrontal development" for consideration by eLife. Your article has been reviewed by four peer reviewers, one of whom is a member of our Board of Reviewing Editors, and the evaluation has been overseen by Laura Colgin as the Senior Editor. The following individuals involved in review of your submission have agreed to reveal their identity: Quentin Perrenoud (Reviewer #3); Christopher I Moore (Reviewer #4).

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

We would like to draw your attention to changes in our revision policy that we have made in response to COVID-19 (https://elifesciences.org/articles/57162). Specifically, when editors judge that a submitted work as a whole belongs in eLife but that some conclusions require a modest amount of additional new data, as they do with your paper, we are asking that the manuscript be revised to either limit claims to those supported by data in hand, or to explicitly state that the relevant conclusions require additional supporting data.

Our expectation is that the authors will eventually carry out the additional experiments and report on how they affect the relevant conclusions either in a preprint on bioRxiv or medRxiv, or if appropriate, as a Research Advance in eLife, either of which would be linked to the original paper.

Summary:

Bitzenhofer et al., examined the developmental trajectory of γ-band oscillations during early development in mice prefrontal cortex. The authors used extracellular recordings and optogenetic stimulations in mice aged P5-40. Rhythmic activity in the prefrontal cortex became more prominent during the second postnatal week and increased in frequency. Developmental modifications in the strength of γ-band oscillations correlated with activation of layer 2/3 pyramidal neurons while the acceleration of their frequency followed similar temporal dynamics as the maturation of fast-spiking interneurons. The current findings provide important data on the maturation of rhythmic activity and the underlying mechanisms in prefrontal circuits. Given the relative scarcity of data in this field, the findings are potentially important and of general interest.

The reviewers made many positive comments on your manuscript. They think the study fills an important gap in terms of developmental studies on oscillations and provides mechanistic insight into the emergence of γ rhythms. They agree that the dataset is rich and contains many challenging experiments providing valuable data that is scarce in the field. They also consider that the findings fit well with the existing literature.

Essential revisions:

The reviewers were generally appreciative of the quality and richness of the experiments. Their main concerns were the analysis and interpretation of the data. The reviewers assess that this work can be done without further experiments, but by further analysis.

1) The greatest concern pertains to the claims about the mechanisms underlying the development of rhythms and their statistical support.

This refers to claims like, "These results demonstrate that the interplay between excitatory and inhibitory neurons is not only critical for the generation of adult γ activity but also for its emergence during postnatal development."

A core argument is that developmental modifications in high-frequency activity mirror the changes observed in both PV/FS-activation and L2/3 pyramidal neurons. The authors refer frequently to the fact the patterns of electrophysiological activity "mirror" those observed, for example, for the maturation of FS interneurons. The authors should investigate these relationships more formally, rather stating that there are similar trends in the data, for example, through analysis of individual changes.

Overall, the authors make little effort to test this link statistically and their claim is only backed by observational and potentially subjective similarities between developmental trends.

Specifically, reviewers commented that all the data presented in the manuscript, except for the immunohistochemistry, come from the same data-set. It should thus be possible to test whether a significant link exists across recordings between the power or the mode of high frequency activity and the various metrics used by the authors to quantify RS and FS firing. A variety of methods exist for this purpose and would be acceptable here (a stepwise multi-linear regression for instance).

The authors should also characterize the development of synchrony and phase-locking over time. How does synchrony amongst FS evolve, for example, over time?

In general, the study makes strong claims like "demonstrate that the interplay between excitatory and inhibitory neurons is not only critical for the generation of adult γ activity but also for its emergence during postnatal development"(Discussion section). Most of the observation provided by the author are based on correlations. Thus, we encourage the author to be more cautious in their conclusions.

2) Important analyses about neural activity are missing that are critical for the interpretation of the data. Overall, the analysis of the spiking and rhythmicity of RS and FS unit is somewhat limited. The reviewers feel that adding these analyses are necessary to make the paper more complete and impactful.

Requests:

i) Pairwise analyses of synchrony amongst FS over time. Important from many perspectives, including relating these data to the slice literature where younger animals are often used e.g. to examine coupling frequency.

ii) The authors should see whether other spiking metrics evolve over time: CV2 is one among many, ISI probability (as distinct from frequency analysis of the spike train--these do give a different perspective).

iii) The manuscript does not provide of direct measure of the phase-locking of RS and FS units to high-frequency activity. The sole metric used for this purpose is the autocorrelation which is rather a measure of intrinsic rhythmicity than a measure of the entrainment to LFP signals. It is also important to note that, while γ activity is often conceived as a sustained oscillation, studies have shown that in most cortical regions it patterns in short bouts (Siegle et al., 2014) having dynamics that are similar to filtered white noise (Burns et al., 2011). Thus, an increase in γ activity is generally not accompanied by an increase of rhythmicity in the autocorrelation of RS and FS units (Perrenoud et al., 2016). A more appropriate approach would be to estimate the phase of high-frequency activity using a wavelet or a Hilbert transform (Bruns, 2004) and to quantify the phase locking of RS and FS units using either spike-field coherence, or if firing bias is a concern, with the Pairwise Phase Consistency.

iv) The authors should do a more sensitive quantification of spike-field coherence for spontaneous firing, because autocorrelations will be highly insensitive when the firing rates are low. Otherwise, what does the γ in the PFC reflect, if not the firing of individual neurons?

v) Given the richness of the dataset and the policies of eLife with respect to data publication, we strongly encourage the authors to published their raw data which may allow other aspects of this dataset to be analysed in the future.

3) Spectral analyses. The reviewers requested a more refined spectral analysis:

i) The authors examine spontaneous activity (Figure 1) and compare these data to optogenetically driven oscillations (Figure 3). However, inspection of spectrograms (Figure 1 and Figure 3B) may reveal some differences. The spontaneous high-frequency activity appears much more broadbanded (Figure 1B) than the optogenetically driven activity (Figure 3B). This raises the question whether the patterns are actually reflecting the same process and thus potentially involving different mechanisms (broad-band modulation vs. narrow-banded oscillation). Similarly, changes in peak-frequency vs. increases in power may reflect quite distinct phenomena that should be disentangled. In this context, did the authors examine also changes in lower frequency power across development? Close inspection of Figure 1B reveals a clear peak in the δ/theta-range at P40 which is not present in other groups.

ii) Reviewers commented that there is very little low frequency power for the P5-10 group. Reviewers would like to see what happens if the authors normalize the power differently (using a standard division by the total power), and if there would not be more γ power at high frequencies?

Related to this, reviewers commented that the spontaneous firing rates seem to be very low for the P5-10 group (see Figure 4) and were concerns that comparing absolute power across animals amounts to comparing mean firing rates.

iii) The lower peak seems to shift to a lower frequency with age, rather than to a higher frequency. The authors should discuss this.

iv) There appear to be artifacts in the spectrum shown in Figure 1, column 3, second row P12 – what are these artifacts? Artifacts are peaks separated by about 10Hz, meaning they are 0.1s artifacts, this could be due to anesthesia apparatus? Could this have influenced the quantification of γ?

v) Figure 3C: From this figure, the developmental change in spectral power is not clearly visible. The authors may want to use a different way of highlighting changes in spectral power across development.

4) Effect of behavioural state. The authors should address the issue of behavioural state.

In general comparisons are made between animals where it is not clear that the behavioural state is the same. For example, chronic recordings were done only in P25 and P40 groups, but the behavioural state of the younger (and the conditions of recordings) during wakefulness are completely unclear. It is not clear whether changes in γ observed might be due to behavioural state (see also Hoy and Niell, 2015). Was the behavioral state the same between mice? Did mice at different ages run more than at other ages? How did running impact FS or RS firing?

The reviewers further wondered how the time in the session impacted likelihood of seeing different rhythms (did lower frequencies predominate later in the session, as many prior studies would predict)?

Overall, the authors feel that the data in this sense is underused and that there may be a lot of useful developmental richness that would make the paper a more important contribution.

The Materials and methods section says some wake animals were recorded from a stereotaxis apparatus. What does this apparatus imply for behavioral state?

The conditions of the recordings need to be clearly described.

5) Richer analyses of the adaptation functions is highly warranted, given interesting work (e.g., by Llampl and many others) in the adaptation of inhibitory versus excitatory currents.

Reviewers further wondered if the difference in apparent adaptation cannot be simply explained by the overall difference in activation to light trains?

6) While FS units are generally assumed to correspond to PV interneurons, pyramidal neurons (Onorato et al., 2019) and Somatostatin interneurons (Ma et al., 2006) can also display these characteristics. Thus, the FS units identified in this study might not only include PV interneurons. The authors should discuss this.

The authors wondered if the authors can give a strong argument or piece data that the "fast-spiking firing properties" are also altered, in the sense of the neurons' input-output functions (how they respond to current or synaptic input). That is fast spiking needs to be more clearly defined and interpreted.

7) In general reviewers agreed that the dataset (nature of the statistical units, mice, recording and state – awake, anesthetized, sex) is not described clearly enough. These points would greatly improve the paper.

i) Reviewers noted it is generally difficult to figure out how the authors exactly analyzed their data, what data points went in the figures, etc. Tables summarizing the dataset would be very useful.

ii) Chronic recordings were done only in P23-25 and P38-40 – are these difference mice?

Figure 1C – are these different mice? If so, can we compare them to each other? What data goes exactly in this figure in terms of acute recordings and chronic recordings for 1A-C. It seems for the wake group the authors compare chronic to non-chronic conditions. That comparison seems problematic given the effect on behavioural state.

iii) N = 114 recordings, but n=80 and n=20,35 in the figure legends what does it refer to? The n's used for the analysis and the number of mice should be clear everywhere in the paper. Is every mouse contributing a separate data point for the statistics, or is every channel, or every recording a data point? If so, isn't the analysis problematic because you are inflating your statistical degrees of freedom massively. For example, p=2.73*10-8 seems to be a p-value that is suffering from inflation due to dependent measures. Similar comments apply to the rest of the paper.

It is unclear how sample size impact the inferences we should take from different plots and analyses--many more cells are sampled at certain days. For example, in Figure 6B, there seem to be differences in the adaptation functions, but it's not clear whether this is the case.

iv) Figure 1B panel right -there are no error bars on this figure, how reliable are these differences across mice? Figure 1 Heatmap – it seems that there is a lot variability, perhaps across different mice? The authors should show how consistent the effects are across mice that are recorded under similar conditions.

v) The authors should describe the intensity of the laser, and say how the intensity was calibrated and determined. The reviewers wondered whether the laser is actually driving activity in the pups, because this is not clearly visible from the plots, perhaps due to extremely low spontaneous firing rates. It appears that high frequency activity (which is dominated by spiking) does not seem to be affected by the laser. Single examples of neurons with raster plots would be very helpful to understand this. In general, providing raster plots with single spikes would be extremely helpful for the reader to assess the reliability of optogenetics simulation and the meaning of various statistics computed. The authors should ideally show an example of a single neuron that is positively modulated by light stimulation for the young pups to understand what is going on.

The text should clearly distinguish between absolute rate changes and modulation. If spontaneous firing rates are very low, then neurons might show a large relative change, but they could still be largely unresponsive to the light.

vi) The following statement is not clear. The first p-value doesn't seem to be significant.

"While ramp-induced firing rate changes of RS units (Mann207 Kendall trend test, p=0.07, n=7 age groups, tau-b 0.619) became more prominent at older 208 age, the firing rate changes were stable for FS units (Mann-Kendall trend test, p=0.88, 209 n=7 age groups, tau-b 0.047) (Figure 4B)"

vii) "whereas at older age the number 211 of activated and inactivated RS units got more balanced (Mann-Kendall trend test, p=1.52*10-14 212 , n=1821 units, tau-b -0.123)"

Please explain what "more balanced" means and how balanced is quantified.

viii) Reviewers commented that in Figure 5, it is unclear what the quantification or statistical test is. It looks like only the P36-40 is qualitatively different, and the P5-10. But the intermediate ages do not seem to show a consistent pattern.

(viiii) A variety of methods are available for hierarchical clustering. However, a reviewer did not understand which of these the authors use for the clustering of single units. Please make sure it is well detailed in the Materials and methods sections.

Comments on literature:

i) The authors highlight in this paper the importance of early development for the maturation of neural oscillations. It would be useful to expand the literature review to indicate that development of rhythmic activity as well as the underlying mechanism extend beyond P40.

ii) In the Introduction, please specify that the results obtained by Chen et al., 2017; Veit et al., 2017 were acquired in the visual cortex as the type of β activity observed in these studies has not yet been observed in other regions.

iii) Discussion and comparison with existing literature (Hoy and Niell, 2015 – which was not cited) is necessary. That study contains several findings reported here as well.

eLife. 2020 Nov 18;9:e56795. doi: 10.7554/eLife.56795.sa2

Author response

Essential revisions:

The reviewers were generally appreciative of the quality and richness of the experiments. Their main concerns were the analysis and interpretation of the data. The reviewers assess that this work can be done without further experiments, but by further analysis.

We thank the reviewers for the constructive feedback and most helpful comments.

1) The greatest concern pertains to the claims about the mechanisms underlying the development of rhythms and their statistical support.

This refers to claims like, "These results demonstrate that the interplay between excitatory and inhibitory neurons is not only critical for the generation of adult γ activity but also for its emergence during postnatal development."

We rephrased and toned down our statements.

A core argument is that developmental modifications in high-frequency activity mirror the changes observed in both PV/FS-activation and L2/3 pyramidal neurons. The authors refer frequently to the fact the patterns of electrophysiological activity "mirror" those observed, for example, for the maturation of FS interneurons. The authors should investigate these relationships more formally, rather stating that there are similar trends in the data, for example, through analysis of individual changes.

We used linear regression models to test for significant links between average measures of individual recordings and age and added the results to the Results section and Supplementary file 2. Further, we performed stepwise linear regressions to identify significantly links between these measures and oscillatory activity in the local field potential (LFP) (see below).

Overall, the authors make little effort to test this link statistically and their claim is only backed by observational and potentially subjective similarities between developmental trends.

Specifically, reviewers commented that all the data presented in the manuscript, except for the immunohistochemistry, come from the same data-set. It should thus be possible to test whether a significant link exists across recordings between the power or the mode of high frequency activity and the various metrics used by the authors to quantify RS and FS firing. A variety of methods exist for this purpose and would be acceptable here (a stepwise multi-linear regression for instance).

We thank the reviewers for the suggestion. We performed stepwise linear regressions on average values for individual mice to examine links between single unit measures and oscillatory activity in the LFP. We added the results to the Results section and to Supplementary file 2.

The authors should also characterize the development of synchrony and phase-locking over time. How does synchrony amongst FS evolve, for example, over time?

In line with the suggestion, we further analyzed the development of synchrony using cross-correlations of simultaneously recorded FS-FS, RS-RS, and RS-FS unit pairs and added the results as Figure 5—figure supplement 3. We analyzed phase locking using pairwise phase consistency and power of average spike-triggered LFP for spontaneous activity and during ramp light stimulation and added the results to Figure 5, Figure 5—figure supplement 1, and Figure 5—figure supplement 2.

In general, the study makes strong claims like "demonstrate that the interplay between excitatory and inhibitory neurons is not only critical for the generation of adult γ activity but also for its emergence during postnatal development"(Discussion section). Most of the observation provided by the author are based on correlations. Thus, we encourage the author to be more cautious in their conclusions.

We rephrased and toned down the statements.

2) Important analyses about neural activity are missing that are critical for the interpretation of the data. Overall, the analysis of the spiking and rhythmicity of RS and FS unit is somewhat limited. The reviewers feel that adding these analyses are necessary to make the paper more complete and impactful.

Requests:

i) Pairwise analyses of synchrony amongst FS over time. Important from many perspectives, including relating these data to the slice literature where younger animals are often used e.g. to examine coupling frequency.

We analyzed cross-correlations of simultaneously recorded RS-RS, FS-FS, and RS-FS unit pairs during ramp light stimulation across age. We added the new results to the manuscript (Discussion section) and as Figure 5—figure supplement 3. The additional analyses showed no rhythmic interactions of RS-RS unit pairs at young age, but at older age with increasing strength and frequency. Similar results were obtained for FS-FS and RS-FS unit pairs, but data were less clear due to the low numbers of simultaneously recorded FS unit pairs. No consistent time lag was detected for RSFS unit pairs interactions.

ii) The authors should see whether other spiking metrics evolve over time: CV2 is one among many, ISI probability (as distinct from frequency analysis of the spike train--these do give a different perspective).

In line with the suggestion, we complemented the rhythmicity analysis using autocorrelations with two other measures: (i) the coefficient of variation (CV) of inter-spike intervals, C_V2 and (ii) interspike intervals for RS and FS units. We added the new results to the manuscript (subsection “The rhythmicity of pyramidal cell and interneuron firing follows similar development as accelerating γ activity in mPFC”) and displayed the data as Figure 5—figure supplement 2. We show that RS units of older mice have longer inter-spike intervals (at lower instantaneous frequency), indicating that they have a higher tendency to skip cycles of fast oscillatory activity.

iii) The manuscript does not provide of direct measure of the phase-locking of RS and FS units to high-frequency activity. The sole metric used for this purpose is the autocorrelation which is rather a measure of intrinsic rhythmicity than a measure of the entrainment to LFP signals. It is also important to note that, while γ activity is often conceived as a sustained oscillation, studies have shown that in most cortical regions it patterns in short bouts (Siegle et al., 2014) having dynamics that are similar to filtered white noise (Burns et al., 2011). Thus, an increase in γ activity is generally not accompanied by an increase of rhythmicity in the autocorrelation of RS and FS units (Perrenoud et al., 2016). A more appropriate approach would be to estimate the phase of high-frequency activity using a wavelet or a Hilbert transform (Bruns, 2004) and to quantify the phase locking of RS and FS units using either spike-field coherence, or if firing bias is a concern, with the Pairwise Phase Consistency.

In line with the suggestion, we analyzed the power of average spike-triggered LFP, as well as pairwise phase consistency for RS and FS units during spontaneous activity and ramp light stimulations across age. We added the new results to the manuscript (subsection “The rhythmicity of pyramidal cell and interneuron firing follows similar development as accelerating γ activity in mPFC”) and displayed them in Figure 5—figure supplement 1 and Figure 5—figure supplement 2. These novel findings are consistent with the idea that γ activity in the LFP reflects rhythmic activity of single units in the medial prefrontal cortex (mPFC).

iv) The authors should do a more sensitive quantification of spike-field coherence for spontaneous firing, because autocorrelations will be highly insensitive when the firing rates are low. Otherwise, what does the γ in the PFC reflect, if not the firing of individual neurons?

As recommended, we added additional measures to quantify spike-field coherence (see point 2 (iii)). Autocorrelations, pairwise phase consistency, and power of average spike-triggered LFP did not show obvious rhythmicity during spontaneous activity. We assume this might be due to the short and low power γ activity in the absence of a task engaging the mPFC. We added a short discussion of this topic to the text (Discussion section).

v) Given the richness of the dataset and the policies of eLife with respect to data publication, we strongly encourage the authors to published their raw data which may allow other aspects of this dataset to be analysed in the future.

We made the dataset available online. https://doi.org/10.12751/g-node.heyl6r

3) Spectral analyses. The reviewers requested a more refined spectral analysis:

i) The authors examine spontaneous activity (Figure 1) and compare these data to optogenetically driven oscillations (Figure 3). However, inspection of spectrograms (Figure 1 and Figure 3B) may reveal some differences. The spontaneous high-frequency activity appears much more broadbanded (1B) than the optogenetically driven activity (3b). This raises the question whether the patterns are actually reflecting the same process and thus potentially involving different mechanisms (broad-band modulation vs. narrow-banded oscillation). Similarly, changes in peak-frequency vs. increases in power may reflect quite distinct phenomena that should be disentangled. In this context, did the authors examine also changes in lower frequency power across development? Close inspection of Figure 1B reveals a clear peak in the δ/theta-range at P40 which is not present in other groups.

We compare the data set corresponding to spontaneous activity and optogenetically-driven activity and did not detect a shift in the frequency of slow (i.e. δ/theta) oscillations. The misleading impression of differences between the two conditions in this frequency range might results from the fact that Figure 1B and Figure 3B show individual traces and power spectra or its modulation index. Only Figure 1C,D and Figure 3C,D summarize the full set of recordings. We changed Figure 1C and Figure 3C (z-scaled results per age) to highlight the increase in high-frequency activity, which is the topic of the present study.

We agree with the reviewer that the spontaneous high frequency activity appears to be more broadband than the optogenetically-driven one. This might result from the fact that activation of 25-30% layer 2/3 pyramidal neurons with non-specific ramp stimulations is most likely one but not the unique mechanisms of γ generation in the developing mPFC. However, the similar dynamics supports the suggestion that optogenetically-driven activity reflects at least certain aspects of spontaneous activity. We added a discussion of the relationship between spontaneous and optogenetically-driven fast activity to the manuscript (Discussion section).

As suggested, we tested for significant correlations of peak frequency and amplitude for baseline and ramp periods at the level of individual recordings. Both, peak frequency and amplitude between baseline and ramp stimulation are significantly correlated. We added the results to the Results section and addressed the issue in the Discussion section. To disentangle the changes in peak frequency vs. power increase, we identified the best set of predictors of the two parameters using stepwise linear regression. We found that, while peak frequency correlated with fast-spiking properties and single unit rhythmicity, peak amplitude was best predicted by optogenetically-driven firing rate changes, indicating an important effect of stimulation efficacy on peak amplitude. We included the new results of analysis in the text (Discussion section).

ii) Reviewers commented that there is very little low frequency power for the P5-10 group. Reviewers would like to see what happens if the authors normalize the power differently (using a standard division by the total power), and if there would not be more γ power at high frequencies?

Related to this, reviewers commented that the spontaneous firing rates seem to be very low for the P5-10 group (see Figure 4) and were concerns that comparing absolute power across animals amounts to comparing mean firing rates.

Cortical LFP power is generally low in young mice, due to low and discontinuous activity. In this manuscript we focus on the development of γ activity and therefore chose a scaling that enables optimal visualization of γ band activity. The development of theta activity in the mPFC has been previously addressed in several studies of our lab (Brockman et al., 2011; Bitzenhofer et al., 2015; Hartung et al., 2016). Figure 4 c-e shows firing rate changes, not absolute firing rates. Low firing rates (together with low rhythmicity and network coupling) in neonatal cortex most likely contribute to low LFP power at P5-P10. Further, we see a similar increase in γ power and frequency induced with ramp light stimulations, where we compare power change (modulation index) instead of absolute power (Figure 3).

iii) The lower peak seems to shift to a lower frequency with age, rather than to a higher frequency. The authors should discuss this.

The power spectra shown in Figure 1B are examples of power spectra of 4 individual recordings. We clarified this issue in the figure legend and changed the plot summarizing all data in Figure 1C for better illustration. Instead of the average power spectra, we show z-scored average power spectra for each age to increase the visibility of low power levels at young age.

iv) There appear to be artifacts in the spectrum shown in Figure 1, column 3, second row P12 – what are these artifacts? Artifacts are peaks separated by about 10Hz, meaning they are 0.1s artifacts, this could be due to anesthesia apparatus? Could this have influenced the quantification of γ?

The artifacts are not due to anesthesia apparatus, since mice have been anesthetized through intraperitoneal injections of urethane. In line with previous results, the artifacts might mirror the breathing of pups. Due to the thin and instable skull bone these artifacts are difficult to avoid, especially in non-anesthetized animals. They appear rather large in the spectra because the LFP power level is low. We analyzed the data and confirmed that in all but one pup, which we excluded from the analysis, the artifacts did not preclude reliable peak detection.

v) Figure 3C: From this figure, the developmental change in spectral power is not clearly visible. The authors may want to use a different way of highlighting changes in spectral power across development.

In line with the reviewer’ suggestion, we changed the plots in Figure 3C, Figure 3—figure supplement 1B, and Figure 1C by z-scoring the data for each age to increase the visibility of lower power levels/changes at young age.

4) Effect of behavioural state. The authors should address the issue of behavioural state.

In general comparisons are made between animals where it is not clear that the behavioural state is the same. For example, chronic recordings were done only in P25 and P40 groups, but the behavioural state of the younger (and the conditions of recordings) during wakefulness are completely unclear. It is not clear whether changes in γ observed might be due to behavioural state (see also Hoy and Niell, 2015). Was the behavioral state the same between mice? Did mice at different ages run more than at other ages? How did running impact FS or RS firing?

We agree with the reviewer on the high relevance of behavioral state. For this, we added the Supplementary file 1 and rephrased the text for clarification of recording conditions. Out of 115 recordings considered for the present study, 80 recordings were performed in P5-40 mice under anesthesia with acute head fixation (1 recording per mouse), 10 recordings were performed in P7-12 mice without anesthesia with acute head fixation (1 recording per mouse), and 25 recordings were performed without anesthesia with chronically implanted head fixation (6 mice at P24-26 and 5 mice at P37-40). All recordings were performed from head-fixed mice using acutely inserted electrodes.

We previously investigated in detail the patterns of oscillatory activity in the presence of anesthetics and their relationship to sleep (Chini et al., 2019). Mice at young age spent a large portion of time sleeping and no significant differences in the activity patterns recorded up to 1 hour were detected. Their poor motor abilities at this age made an investigation on a spinning disk superfluous. Only non-anesthetized mice at older age (P24-26 and P37-40) with chronic head fixation implants were free to sit or run on a spinning disc. Running increased γ power in these mice. To reduce the variability due to different behavioral states, recordings from non-anesthetized mice with chronic head fixation implants were excluded from single unit analysis.

The reviewers further wondered how the time in the session impacted likelihood of seeing different rhythms (did lower frequencies predominate later in the session, as many prior studies would predict)?

Recordings followed a specific protocol, such that the time in the session does not affect the results. As exemplified in Figure 4A,B and Figure 4—figure supplement 1 the spontaneous activity and optogenetic activation of individual units across trials was rather consistent. An in-depth investigation of modulatory factors of γ activity, such as anesthesia depth, tiredness, stress exceeds the aims of the current manuscript and will be the topic of future studies.

Overall, the authors feel that the data in this sense is underused and that there may be a lot of useful developmental richness that would make the paper a more important contribution.

We agree with the reviewer that the data contain further interesting information about the development of prefrontal activity, that may be addressed in future studies. We decided to focus, in the present manuscript solely on the development of fast oscillatory rhythms in the mPFC. To backup this goal, we exploited the richness of data to perform additional analyses addressing rhythmicity of single units and locking of single units to LFP. The new results have been added to the manuscript (subsection “Fast spiking interneuron maturation resembles the time course of γ development”). Of note, the development of slower oscillations in the mPFC has been addressed in previous studies (see point 3 (ii)).

The methods say some wake animals were recorded from a stereotaxis apparatus. What does this apparatus imply for behavioral state?

We developed a novel recording paradigm aiming to monitor the cortical activity of young mice. Due to the size of the animals as well as fragility and growth-induced changes of the skull, the “classical” tetrodes or telemetric recordings cannot be used for this purpose. Therefore, we built a moving disk on which non-anesthetized head-fixed mice are free to move. To improve the clarity of Materials and methods section, we rephrased the text and added a Supplementary file 1 summarizing the experimental conditions. We added to exemplary images to illustrate the two recording condition below:

The conditions of the recordings need to be clearly described.

As requested, we rephrased the text (Materials and methods section, Results section, figure legends) and added Supplementary file 1 summarizing experimental conditions.

5) Richer analyses of the adaptation functions is highly warranted, given interesting work (e.g., by Llampl and many others) in the adaptation of inhibitory versus excitatory currents.

Reviewers further wondered if the difference in apparent adaptation cannot be simply explained by the overall difference in activation to light trains?

We performed multifactorial ANOVA with the main factors RS/FS, age group, and stimulation frequency to compare adaptation functions and added the results to the figure legend and Supplementary file 2. RS and FS units were not significantly different, but adaptation functions changed significantly with age and stimulation frequency. Stronger adaptation at young age and high frequencies likely reflects an inability of neurons in the mPFC of neonatal mice to spike at high frequencies (see also Bitzenhofer et al., 2017). It is not clear to us how lower activation by light trains would explain a change in the ratio from early to late activation. We suggest that the lower activation in response to pulsed light for young age groups mainly reflects reduced connectivity and thereby reduced activation of not-opsin-expressing neurons.

6) While FS units are generally assumed to correspond to PV interneurons, pyramidal neurons (Onorato et al., 2019) and Somatostatin interneurons (Ma et al., 2006) can also display these characteristics. Thus, the FS units identified in this study might not only include PV interneurons. The authors should discuss this.

We added a brief discussion of the topic suggested by the reviewer to the Discussion section.

The authors wondered if the authors can give a strong argument or piece data that the "fast-spiking firing properties" are also altered, in the sense of the neurons' input-output functions (how they respond to current or synaptic input). That is fast spiking needs to be more clearly defined and interpreted.

In line with the suggestion, we refer to previously published data that characterized the development of PV interneurons in the mPFC in vitro (Miyamae et al., 2017).

7) In general reviewers agreed that the dataset (nature of the statistical units, mice, recording and state – awake, anesthetized, sex) is not described clearly enough. These points would greatly improve the paper.

i) Reviewers noted it is generally difficult to figure out how the authors exactly analyzed their data, what data points went in the figures, etc. Tables summarizing the dataset would be very useful.

We added the requested information to the text (Materials and methods section, Results section, Figure legends) and summarized the experimental conditions in a new table (Supplementary file 1).

ii) Chronic recordings were done only in P23-25 and P38-40 – are these difference mice?

Figure 1C – are these different mice? If so, can we compare them to each other? What data goes exactly in this figure in terms of acute recordings and chronic recordings for Figure 1A-C. It seems for the wake group the authors compare chronic to non-chronic conditions. That comparison seems problematic given the effect on behavioural state.

As mentioned above, we rephrased the text to improve the clarity of our statements and added the Supplementary file 1 summarizing experimental conditions. Especially, we aim to specify that the “chronic recordings” in young mice are not similar to those performed in adults but correspond to acute recordings in mice with chronically implanted head fixation adapters. Each of these mice was recorded 1-3 times either between P24-26 or P37-40. Thus, all recordings were acute head-fixed recordings with electrodes inserted prior to the recording session, but 11 mice at P24-26 and P37-40 had chronically implanted head fixation adapters. These 11 mice were added to show that the developmental increase in γ frequency does not depend on anesthesia. They were excluded from the analysis of single units to avoid the comparison of different behavioral states.

iii) N = 114 recordings, but n=80 and n=20,35 in the figure legends what does it refer to? The n's used for the analysis and the number of mice should be clear everywhere in the paper. Is every mouse contributing a separate data point for the statistics, or is every channel, or every recording a data point? If so, isn't the analysis problematic because you are inflating your statistical degrees of freedom massively. For example, p=2.73*10-8 seems to be a p-value that is suffering from inflation due to dependent measures. Similar comments apply to the rest of the paper.

We thank the reviewer for highlighting the error in figure legend of Figure 1. For Figure 1, the total was n=114 recordings, 80 recordings from anesthetized mice (one per mouse) and 34 recordings from 20 non-anesthetized mice that underwent multiple recordings (1 non-anesthetized recording was removed from baseline LFP analysis because recording artifacts interfered with γ peak detection). For Figure 3, the total was n=115 recordings, 80 recordings from anesthetized mice (one per mouse) and 35 recordings from 21 non-anesthetized mice. We corrected the mistake. Throughout the manuscript, we aimed to clarify the sample size and corresponding statistics by rephrasing the text and adding the Supplementary file 1 summarizing the experimental conditions.

It is unclear how sample size impact the inferences we should take from different plots and analyses--many more cells are sampled at certain days. For example, in Figure 6B, there seem to be differences in the adaptation functions, but it's not clear whether this is the case.

Sample sizes are low for FS at young age groups because of the late maturation of FS interneurons. We added statistical comparisons throughout the manuscript to identify statistically significant differences. Low numbers of FS at young age groups prevent strong conclusions.

iv) Figure 1B panel right -there are no error bars on this figure, how reliable are these differences across mice? Figure 1 Heatmap – it seems that there is a lot variability, perhaps across different mice? The authors should show how consistent the effects are across mice that are recorded under similar conditions.

The lack of error bars is due to the fact that Figure 1B panel right shows example power spectra of 4 individual recordings. The full dataset is displayed in Figure 1C and D. We modified the figure legend for better clarity. Moreover, we replaced the plot summarizing all data in Figure 1C as average power spectra by z-scored average power spectra for each age. Each data point in Figure 1D represents an individual recording, showing their variability, but a consistent increase of γ frequency across age.

v) The authors should describe the intensity of the laser, and say how the intensity was calibrated and determined. The reviewers wondered whether the laser is actually driving activity in the pups, because this is not clearly visible from the plots, perhaps due to extremely low spontaneous firing rates. It appears that high frequency activity (which is dominated by spiking) does not seem to be affected by the laser. Single examples of neurons with raster plots would be very helpful to understand this. In general, providing raster plots with single spikes would be extremely helpful for the reader to assess the reliability of optogenetics simulation and the meaning of various statistics computed. The authors should ideally show an example of a single neuron that is positively modulated by light stimulation for the young pups to understand what is going on.

The text should clearly distinguish between absolute rate changes and modulation. If spontaneous firing rates are very low, then neurons might show a large relative change, but they could still be largely unresponsive to the light.

In line with the reviewer’s suggestions, we added a description of laser intensity calibration to the Materials and methods section. We replaced the plot in Figure 3C showing the modulation index of induced power by the z-scored modulation index for each age to enable a better visualization of effects at young age. Induced activity is lower in young mice, presumably due to lower connectivity and weaker synchronization of non-opsin-expressing neurons.

We added raster plots of example RS and FS units at different ages positively modulated by optogenetic stimulation to Figure 4 as well as example units that were inactivated or not affected by the stimulation from the same recordings to Figure 4—figure supplement 1.

We rephrased the text to avoid confusions about absolute rate changes and modulation.

vi) The following statement is not clear. The first p-value doesn't seem to be significant.

"While ramp-induced firing rate changes of RS units (Mann207 Kendall trend test, p=0.07, n=7 age groups, tau-b 0.619) became more prominent at older 208 age, the firing rate changes were stable for FS units (Mann-Kendall trend test, p=0.88, 209 n=7 age groups, tau-b 0.047) (Figure 4B)"

We rephrased the sentence.

vii) "whereas at older age the number 211 of activated and inactivated RS units got more balanced (Mann-Kendall trend test, p=1.52*10-14 212 , n=1821 units, tau-b -0.123)"

Please explain what "more balanced" means and how balanced is quantified.

We rephrased the sentence.

viii) Reviewers commented that in Figure 5, it is unclear what the quantification or statistical test is. It looks like only the P36-40 is qualitatively different, and the P5-10. But the intermediate ages do not seem to show a consistent pattern.

To address the query, we performed multifactorial ANOVA with the main factors RS/FS, age group, and frequency to compare the autocorrelation power. We added the results to the text (Discussion section) and Supplementary file 2.

viiii) A variety of methods are available for hierarchical clustering. However, a reviewer did not understand which of these the authors use for the clustering of single units. Please make sure it is well detailed in the Materials and methods sections.

We specified that Pairwise Euclidean distance was used for hierarchical clustering (subsection “Fast spiking interneuron maturation resembles the time course of γ development”).

Comments on literature:

i) The authors highlight in this paper the importance of early development for the maturation of neural oscillations. It would be useful to expand the literature review to indicate that development of rhythmic activity as well as the underlying mechanism extend beyond P40.

According to our data, peak frequency and amplitude of γ activity stabilize around P25 and no major changes occur afterwards. We added a brief discussion of the γ activity and underlying mechanisms in mice beyond P40 (typically considered as young adult) (Discussion section).

ii) In the Introduction, please specify that the results obtained by Chen et al., 2017; Veit et al., 2017 were acquired in the visual cortex as the type of β activity observed in these studies has not yet been observed in other regions.

We rephrased accordingly.

iii) Discussion and comparison with existing literature (Hoy and Niell, 2015 – which was not cited) is necessary. That study contains several findings reported here as well.

We included the mentioned paper in the Discussion section.

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Data Citations

Bitzenhofer SB, Pöpplau A, Hanganu-Opatz IL. 2020. Ephys data associated with the paper *Gamma activity accelerates during prefrontal development*. g-node. [DOI] [PMC free article] [PubMed]

Supplementary Materials

Figure 1—source data 1. Source data for Figure 1b.

elife-56795-fig1-data1.mat^{(1.6KB, mat)}

Figure 2—source data 1. Source data for Figure 2a,b.

elife-56795-fig2-data1.mat^{(795B, mat)}

Figure 2—source data 2. Source data for Figure 2f.

elife-56795-fig2-data2.mat^{(31.4KB, mat)}

Figure 2—source data 3. Source data for Figure 2g.

elife-56795-fig2-data3.mat^{(3.2KB, mat)}

Figure 2—source data 4. Source data for Figure 2h.

elife-56795-fig2-data4.mat^{(626B, mat)}

Figure 2—source data 5. Source data for Figure 2i.

elife-56795-fig2-data5.mat^{(57.7KB, mat)}

Figure 2—figure supplement 1—source data 1. Source data for Figure 2—figure supplement 1b–d.

elife-56795-fig2-figsupp1-data1.mat^{(142.9KB, mat)}

Figure 3—source data 1. Source data for Figure 3d.

elife-56795-fig3-data1.mat^{(1.6KB, mat)}

Figure 3—figure supplement 1—source data 1. Source data for Figure 3—figure supplement 1c.

elife-56795-fig3-figsupp1-data1.mat^{(1.6KB, mat)}

Figure 4—source data 1. Source data for Figure 4d.

elife-56795-fig4-data1.mat^{(508B, mat)}

Figure 6—source data 1. Source data for Figure 6b.

elife-56795-fig6-data1.mat^{(1.3KB, mat)}

Figure 6—source data 2. Source data for Figure 6d.

elife-56795-fig6-data2.mat^{(514B, mat)}

Figure 6—source data 3. Source data for Figure 6f.

elife-56795-fig6-data3.mat^{(516B, mat)}

Supplementary file 1. Recording summary.

Table summarizing the recordings for each experimental condition.

elife-56795-supp1.docx^{(35.7KB, docx)}

Supplementary file 2. Detailed statistical results.

Table summarizing the statistical results.

elife-56795-supp2.docx^{(55KB, docx)}

Transparent reporting form

elife-56795-transrepform.pdf^{(219.2KB, pdf)}

Data Availability Statement

The following dataset was generated:

Bitzenhofer SB, Pöpplau A, Hanganu-Opatz IL. 2020. Ephys data associated with the paper *Gamma activity accelerates during prefrontal development*. g-node.

[bib1] Atallah BV, Scanziani M. Instantaneous modulation of gamma oscillation frequency by balancing excitation with inhibition. Neuron. 2009;62:566–577. doi: 10.1016/j.neuron.2009.04.027. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib2] Ben-Ari Y, Khalilov I, Kahle KT, Cherubini E. The GABA excitatory/inhibitory shift in brain maturation and neurological disorders. The Neuroscientist. 2012;18:467–486. doi: 10.1177/1073858412438697. [DOI] [PubMed] [Google Scholar]

[bib3] Bitzenhofer SH, Ahlbeck J, Wolff A, Wiegert JS, Gee CE, Oertner TG, Hanganu-Opatz IL. Layer-specific optogenetic activation of pyramidal neurons causes beta-gamma entrainment of neonatal networks. Nature Communications. 2017a;8:14563. doi: 10.1038/ncomms14563. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib4] Bitzenhofer SH, Ahlbeck J, Hanganu-Opatz IL. Methodological approach for optogenetic manipulation of neonatal neuronal networks. Frontiers in Cellular Neuroscience. 2017b;11:239. doi: 10.3389/fncel.2017.00239. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib5] Bosman LW, Heinen K, Spijker S, Brussaard AB. Mice lacking the major adult GABAA receptor subtype have normal number of synapses, but retain juvenile IPSC kinetics until adulthood. Journal of Neurophysiology. 2005;94:338–346. doi: 10.1152/jn.00084.2005. [DOI] [PubMed] [Google Scholar]

[bib6] Brockmann MD, Pöschel B, Cichon N, Hanganu-Opatz IL. Coupled oscillations mediate directed interactions between prefrontal cortex and Hippocampus of the neonatal rat. Neuron. 2011;71:332–347. doi: 10.1016/j.neuron.2011.05.041. [DOI] [PubMed] [Google Scholar]

[bib7] Butt SJ, Stacey JA, Teramoto Y, Vagnoni C. A role for GABAergic interneuron diversity in circuit development and plasticity of the neonatal cerebral cortex. Current Opinion in Neurobiology. 2017;43:149–155. doi: 10.1016/j.conb.2017.03.011. [DOI] [PubMed] [Google Scholar]

[bib8] Buzsáki G, Logothetis N, Singer W. Scaling Brain Size, Keeping Timing: Evolutionary Preservation of Brain Rhythms. Neuron. 2013;80:751–764. doi: 10.1016/j.neuron.2013.10.002. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib9] Buzsáki G, Draguhn A. Neuronal oscillations in cortical networks. Science. 2004;304:1926–1929. doi: 10.1126/science.1099745. [DOI] [PubMed] [Google Scholar]

[bib10] Cao W, Lin S, Xia QQ, Du YL, Yang Q, Zhang MY, Lu YQ, Xu J, Duan SM, Xia J, Feng G, Xu J, Luo JH. Gamma oscillation dysfunction in mPFC leads to social deficits in neuroligin 3 R451C knockin mice. Neuron. 2018;97:1253–1260. doi: 10.1016/j.neuron.2018.02.001. [DOI] [PubMed] [Google Scholar]

[bib11] Cardin JA, Carlén M, Meletis K, Knoblich U, Zhang F, Deisseroth K, Tsai LH, Moore CI. Driving fast-spiking cells induces gamma rhythm and controls sensory responses. Nature. 2009;459:663–667. doi: 10.1038/nature08002. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib12] Cardin JA. Snapshots of the brain in action: local circuit operations through the Lens of γ oscillations. Journal of Neuroscience. 2016;36:10496–10504. doi: 10.1523/JNEUROSCI.1021-16.2016. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib13] Chen G, Zhang Y, Li X, Zhao X, Ye Q, Lin Y, Tao HW, Rasch MJ, Zhang X. Distinct inhibitory circuits orchestrate cortical beta and gamma band oscillations. Neuron. 2017;96:1403–1418. doi: 10.1016/j.neuron.2017.11.033. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib14] Chini M, Gretenkord S, Kostka JK, Pöpplau JA, Cornelissen L, Berde CB, Hanganu-Opatz IL, Bitzenhofer SH. Neural correlates of anesthesia in newborn mice and humans. Frontiers in Neural Circuits. 2019;13:00038. doi: 10.3389/fncir.2019.00038. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib15] Chini M, Pöpplau JA, Lindemann C, Carol-Perdiguer L, Hnida M, Oberländer V, Xu X, Ahlbeck J, Bitzenhofer SH, Mulert C, Hanganu-Opatz IL. Resolving and rescuing developmental miswiring in a mouse model of cognitive impairment. Neuron. 2020;105:60–74. doi: 10.1016/j.neuron.2019.09.042. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib16] Cho KK, Hoch R, Lee AT, Patel T, Rubenstein JL, Sohal VS. Gamma rhythms link prefrontal interneuron dysfunction with cognitive inflexibility in Dlx5/6+/− mice. Neuron. 2015;85:1332–1343. doi: 10.1016/j.neuron.2015.02.019. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib17] Cho KKA, Davidson TJ, Bouvier G, Marshall JD, Schnitzer MJ, Sohal VS. Cross-hemispheric gamma synchrony between prefrontal parvalbumin interneurons supports behavioral adaptation during rule shift learning. Nature Neuroscience. 2020;23:892–902. doi: 10.1038/s41593-020-0647-1. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib18] Doischer D, Hosp JA, Yanagawa Y, Obata K, Jonas P, Vida I, Bartos M. Postnatal differentiation of basket cells from slow to fast signaling devices. Journal of Neuroscience. 2008;28:12956–12968. doi: 10.1523/JNEUROSCI.2890-08.2008. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib19] Dupont E, Hanganu IL, Kilb W, Hirsch S, Luhmann HJ. Rapid developmental switch in the mechanisms driving early cortical columnar networks. Nature. 2006;439:79–83. doi: 10.1038/nature04264. [DOI] [PubMed] [Google Scholar]

[bib20] Hartung H, Cichon N, De Feo V, Riemann S, Schildt S, Lindemann C, Mulert C, Gogos JA, Hanganu-Opatz IL. From Shortage to Surge: A Developmental Switch in Hippocampal–Prefrontal Coupling in a Gene–Environment Model of Neuropsychiatric Disorders. Cerebral Cortex. 2016;26:4265–4281. doi: 10.1093/cercor/bhw274. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib21] Holt GR, Softky WR, Koch C. Comparison of discharge variability visual cortex neurons . Journal of Neurophysiology. 1996;75:1806–1814. doi: 10.1152/jn.1996.75.5.1806. [DOI] [PubMed] [Google Scholar]

[bib22] Hoy JL, Niell CM. Layer-specific refinement of visual cortex function after eye opening in the awake mouse. Journal of Neuroscience. 2015;35:3370–3383. doi: 10.1523/JNEUROSCI.3174-14.2015. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib23] Kirmse K, Hübner CA, Isbrandt D, Witte OW, Holthoff K. GABAergic transmission during brain development: multiple effects at multiple stages. The Neuroscientist. 2018;24:36–53. doi: 10.1177/1073858417701382. [DOI] [PubMed] [Google Scholar]

[bib24] Kroon T, van Hugte E, van Linge L, Mansvelder HD, Meredith RM. Early postnatal development of pyramidal neurons across layers of the mouse medial prefrontal cortex. Scientific Reports. 2019;9:5037. doi: 10.1038/s41598-019-41661-9. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib25] Laurie DJ, Wisden W, Seeburg PH. The distribution of thirteen GABAA receptor subunit mRNAs in the rat brain. III. embryonic and postnatal development. The Journal of Neuroscience. 1992;12:4151–4172. doi: 10.1523/JNEUROSCI.12-11-04151.1992. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib26] Le Magueresse C, Monyer H. GABAergic interneurons shape the functional maturation of the cortex. Neuron. 2013;77:388–405. doi: 10.1016/j.neuron.2013.01.011. [DOI] [PubMed] [Google Scholar]

[bib27] Lim L, Mi D, Llorca A, Marín O. Development and functional diversification of cortical interneurons. Neuron. 2018;100:294–313. doi: 10.1016/j.neuron.2018.10.009. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib28] Ma Y, Hu H, Berrebi AS, Mathers PH, Agmon A. Distinct Subtypes of Somatostatin-Containing Neocortical Interneurons Revealed in Transgenic Mice. Journal of Neuroscience. 2006;26:5069–5082. doi: 10.1523/JNEUROSCI.0661-06.2006. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib29] Minlebaev M, Colonnese M, Tsintsadze T, Sirota A, Khazipov R. Early γ oscillations synchronize developing thalamus and cortex. Science. 2011;334:226–229. doi: 10.1126/science.1210574. [DOI] [PubMed] [Google Scholar]

[bib30] Miyamae T, Chen K, Lewis DA, Gonzalez-Burgos G. Distinct physiological maturation of Parvalbumin-Positive neuron subtypes in mouse prefrontal cortex. The Journal of Neuroscience. 2017;37:4883–4902. doi: 10.1523/JNEUROSCI.3325-16.2017. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib31] Okaty BW, Miller MN, Sugino K, Hempel CM, Nelson SB. Transcriptional and electrophysiological maturation of neocortical fast-spiking GABAergic interneurons. Journal of Neuroscience. 2009;29:7040–7052. doi: 10.1523/JNEUROSCI.0105-09.2009. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib32] Onorato I, Neuenschwander S, Hoy J, Lima B, Rocha K-S, Broggini AC, Uran C, Spyropoulos G, Klon-Lipok J, Womelsdorf T, Fries P, Niell C, Singer W, Vinck M. A Distinct Class of Bursting Neurons with Strong Gamma Synchronization and Stimulus Selectivity in Monkey V1. Neuron. 2020;105:180–197. doi: 10.1016/j.neuron.2019.09.039. [DOI] [PubMed] [Google Scholar]

[bib33] Richter M, Murtaza N, Scharrenberg R, White SH, Johanns O, Walker S, Yuen RKC, Schwanke B, Bedürftig B, Henis M, Scharf S, Kraus V, Dörk R, Hellmann J, Lindenmaier Z, Ellegood J, Hartung H, Kwan V, Sedlacik J, Fiehler J, Schweizer M, Lerch JP, Hanganu-Opatz IL, Morellini F, Scherer SW, Singh KK, Calderon de Anda F. Altered TAOK2 activity causes autism-related neurodevelopmental and cognitive abnormalities through RhoA signaling. Molecular Psychiatry. 2019;24:1329–1350. doi: 10.1038/s41380-018-0025-5. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib34] Rivera C, Voipio J, Payne JA, Ruusuvuori E, Lahtinen H, Lamsa K, Pirvola U, Saarma M, Kaila K. The K+/Cl- co-transporter KCC2 renders GABA hyperpolarizing during neuronal maturation. Nature. 1999;397:251–255. doi: 10.1038/16697. [DOI] [PubMed] [Google Scholar]

[bib35] Rojas DC, Wilson LB. γ-band abnormalities as markers of autism spectrum disorders. Biomarkers in Medicine. 2014;8:353–368. doi: 10.2217/bmm.14.15. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib36] Shen J, Colonnese MT. Development of Activity in the Mouse Visual Cortex. Journal of Neuroscience. 2016;36:12259–12275. doi: 10.1523/JNEUROSCI.1903-16.2016. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib37] Singer W. Neuronal oscillations: unavoidable and useful? European Journal of Neuroscience. 2018;48:2389–2398. doi: 10.1111/ejn.13796. [DOI] [PubMed] [Google Scholar]

[bib38] Sohal VS. How close are we to understanding what (if anything) γ oscillations do in cortical circuits? Journal of Neuroscience. 2016;36:10489–10495. doi: 10.1523/JNEUROSCI.0990-16.2016. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib39] Veit J, Hakim R, Jadi MP, Sejnowski TJ, Adesnik H. Cortical gamma band synchronization through somatostatin interneurons. Nature Neuroscience. 2017;20:951–959. doi: 10.1038/nn.4562. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib40] Vinck M, van Wingerden M, Womelsdorf T, Fries P, Pennartz CM. The pairwise phase consistency: a bias-free measure of rhythmic neuronal synchronization. NeuroImage. 2010;51:112–122. doi: 10.1016/j.neuroimage.2010.01.073. [DOI] [PubMed] [Google Scholar]

[bib41] Yang JW, Hanganu-Opatz IL, Sun JJ, Luhmann HJ. Three patterns of oscillatory activity differentially synchronize developing neocortical networks in vivo. Journal of Neuroscience. 2009;29:9011–9025. doi: 10.1523/JNEUROSCI.5646-08.2009. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib42] Yang JW, An S, Sun JJ, Reyes-Puerta V, Kindler J, Berger T, Kilb W, Luhmann HJ. Thalamic network oscillations synchronize ontogenetic columns in the newborn rat barrel cortex. Cerebral Cortex. 2013;23:1299–1316. doi: 10.1093/cercor/bhs103. [DOI] [PubMed] [Google Scholar]

PERMALINK

Gamma activity accelerates during prefrontal development

Sebastian H Bitzenhofer

Jastyn A Pöpplau

Ileana Hanganu-Opatz

Roles

Abstract

Introduction

Results

Fast oscillatory activity in the prefrontal cortex accelerates during development

Figure 1. Development of gamma activity in the mouse mPFC.

FS interneuron maturation resembles the time course of gamma development

Figure 2. Development of FS interneurons in the mouse mPFC.

Figure 2—figure supplement 1. t-sne dimensionality reduction.

Activation of L2/3 pyramidal neurons drives fast oscillations with similar acceleration across development as spontaneous activity

Figure 3. Development of L2/3 PYR-driven gamma in the mPFC.

Figure 3—figure supplement 1. Control stimulations of L2/3 PYRs in the mPFC.

The rhythmicity of pyramidal cell and interneuron firing follows similar development as accelerating gamma activity in mPFC

Figure 4. Development of RS and FS unit activity during L2/3 PYR-driven gamma in the mPFC.

Figure 4—figure supplement 1. RS and FS unit activity during L2/3 PYR-driven gamma in the mPFC.

Figure 5. Rhythmicity of RS and FS units across age.

Figure 5—figure supplement 1. Rhythmicity of RS and FS units during spontaneous activity.

Figure 5—figure supplement 2. Additional measures for rhythmicity of RS and FS units.

Figure 5—figure supplement 3. Crosscorrelations of RS and FS units during spontaneous activity.

Inhibitory feedback maturation resembles the dynamics of gamma development

Figure 6. Firing of RS and FS units in response to pulse light stimulation.

Figure 6—figure supplement 1. Inter-spike intervals of RS and FS units during pulse light stimulation.

FS characteristics and rhythmicity of single units relate to the frequency of fast oscillatory activity

Discussion

Materials and methods

Key resources table.

Animals

In utero electroporation (IUE)

Electrophysiology

Acute head-fixed recordings

Acute head-fixed recordings with chronically implanted head-fixation adapters

Optogenetic stimulation

Histology

Immunohistochemistry

Cell quantification

Data analysis

Power spectral density

Modulation index

Peak frequency and strength

Single unit analysis

Statistics

Acknowledgements

Funding Statement

Contributor Information

Funding Information

Additional information

Competing interests

Author contributions

Ethics

Additional files

Data availability

References

Decision letter

Roles

Author response

Author response image 1.

Associated Data

Data Citations

Supplementary Materials

Data Availability Statement

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles

Links to NCBI Databases