Skip to main content
Elsevier Sponsored Documents logoLink to Elsevier Sponsored Documents
. 2024 Feb 7;112(3):421–440.e7. doi: 10.1016/j.neuron.2023.10.024

Reorganization of adolescent prefrontal cortex circuitry is required for mouse cognitive maturation

Jastyn A Pöpplau 1,∗, Timo Schwarze 1, Mariia Dorofeikova 1, Irina Pochinok 1, Anne Günther 1, Annette Marquardt 1, Ileana L Hanganu-Opatz 1,2,∗∗
PMCID: PMC10855252  PMID: 37979584

Summary

Most cognitive functions involving the prefrontal cortex emerge during late development. Increasing evidence links this delayed maturation to the protracted timeline of prefrontal development, which likely does not reach full maturity before the end of adolescence. However, the underlying mechanisms that drive the emergence and fine-tuning of cognitive abilities during adolescence, caused by circuit wiring, are still unknown. Here, we continuously monitored prefrontal activity throughout the postnatal development of mice and showed that an initial activity increase was interrupted by an extensive microglia-mediated breakdown of activity, followed by the rewiring of circuit elements to achieve adult-like patterns and synchrony. Interfering with these processes during adolescence, but not adulthood, led to a long-lasting microglia-induced disruption of prefrontal activity and neuronal morphology and decreased cognitive abilities. These results identified a nonlinear reorganization of prefrontal circuits during adolescence and revealed its importance for adult network function and cognitive processing.

Keywords: prefrontal cortex, development, adolescence, microglia, network activity, electrophysiology

Graphical abstract

graphic file with name fx1.jpg

Highlights

  • •

    Prefrontal synchrony and inhibition strength linearly increase during adolescence

  • •

    Adolescent mPFC activity is characterized by a peak of gamma and spiking activity

  • •

    Microglia mediate these non-linear activity dynamics via circuit refinement

  • •

    Microglia manipulation during adolescence disrupts adult prefrontal function


Pöpplau et al. uncover a non-linear trajectory of prefrontal development in mice with a major reorganization of activity and structure during adolescence. Interfering with these non-linear dynamics via microglia manipulation results in a long-lasting disruption of adult network function and mPFC-dependent cognitive abilities.

Introduction

Flexible adaptation to permanently changing internal and environmental demands is indispensable for daily life; however, cognitive flexibility develops late in life.1 Long after sensory-motor abilities are fully developed, cognitive skills continue to improve.2 It is commonly accepted that neuronal networks accounting for cognitive flexibility are centered in the prefrontal cortex (PFC).3,4 The protracted timeline of prefrontal development aligns with the delayed maturation of higher-order cognitive abilities.1,2,5 Similar to their nonlinear acquisition, the assembly of prefrontal circuits likely also undergoes fluctuations, which might be particularly relevant during specific periods of development. Critical periods controlling the development of sensory systems have been extensively documented, and a multifactorial framework of facilitating and terminating factors that govern critical period plasticity has been characterized.6,7,8 Similar developmental trajectories have been proposed for prefrontal development, but with a much broader and more protracted temporal profile than that described for sensory systems.7,8,9 Such prolonged windows of plasticity might be required for the emergence and calibration of the complex integrative functions of the PFC.2 However, they represent extended periods of increased vulnerability to prefrontal dysfunction, e.g., mental disorders.1,10 Indeed, several studies have revealed that interfering with prefrontal development at specific time points increases vulnerability to impaired cognitive functioning but bears equally good potential for rescuing pathological traits in disease models.11,12,13,14,15

Several mental disorders have their onset from adolescence to young adulthood, when the fine-tuning of cognitive functions occurs.2,16,17 In light of the recent progress in identifying and characterizing prodromal symptoms, adolescence might represent a window not only of vulnerability but also of opportunity. Targeted therapeutic interventions might foster the capacity of the immature PFC to recover. However, in the absence of mechanistic knowledge of late prefrontal development, such strategies are poorly developed.7,18,19 Several mediators of prefrontal circuit assembly during adolescence, such as changes in neuromodulatory systems, the strength of afferent projections, gene expression, microglia cell activity, and the emergence of pubertal hormones, have been identified and proposed as actively shaping cognitive abilities.7,13,14,20,21,22 However, the activity-dependent refinement and assembling of prefrontal circuitry that underlie these abilities are still largely unexplored. This knowledge is indispensable for understanding how dysfunctions of mediators disrupt circuit development, ultimately causing pathological trajectories.

Here, we address this knowledge gap by combining chronic extracellular monitoring of prefrontal activity and optogenetic manipulations with in-depth morphological assessments in mice during late development (i.e., postnatal day [P]16–60). We show that the maturation of fast prefrontal activity within the gamma frequency range and functional connectivity of putative layer 2/3 pyramidal neurons (L2/3 PYRs) follow nonlinear dynamics with a prominent increase during early adolescence (P28–P35), followed by a decrease during late adolescence (P36–P43), and a recurrent increase to achieve adult-like activity patterns. By contrast, the developmental profiles of prefrontal synchrony and inhibition follow linear trajectories and reach mature patterns only toward the end of adolescence. The timeline of these functional fluctuations is linked to microglia-mediated changes in the dendritic complexity and spine density of L2/3 PYRs during early adolescence. Microglia ablation during adolescence, but not adulthood, leads to a long-lasting disruption of prefrontal function. These results reveal that the dynamic reorganization of prefrontal circuits during adolescence is critical for adult cognitive performance.

Results

Prefrontal activity undergoes a nonlinear reorganization during adolescence

Electrical activity has been described as a suitable proxy of neuronal network function not only in the adult brain but also during early development.8,11,23,24 To gain initial insights into the assembling of the PFC during late development, we monitored the patterns of electrical activity throughout the time period of P16–P60. To accomplish this, we developed a protocol for chronic head-fixed multi-site extracellular recordings (Figure 1A). Recordings were performed in L2/3 of the prelimbic subdivision (PL) of the medial PFC (mPFC) of mice moving or resting on a custom-made spinning disc from P16, when eyes and ears have opened, whiskers have developed, and mice start weaning, to P60, when mice are considered young adults.1,2 In this study, the focus on L2/3 of the PL was based on previous data, identifying electrical activity in this subdivision during early development (i.e., the first to second postnatal weeks) as critical for adult prefrontal function and behavior.11,12,25,26 Implanted and non-implanted mice demonstrated a similar increase in weight. The locomotor behavior of implanted mice was similar across ages (Figures S1A–S1D). Major growth of the skull has been reported throughout the investigated period.27 Nonetheless, the chronic recording procedure neither affected the gross growth parameters of the brain nor caused cortical thickness alterations in the prefrontal subdivisions cingulate cortex (CG), PL, or infralimbic cortex (IL) in implanted mice compared with non-implanted mice. Only the cortical thickness of the adjacent motor areas within the recorded hemisphere was slightly reduced in implanted mice (Figures S1E–S1I). Moreover, the features of prefrontal activity acutely recorded in adult mice were similar to those of age-matched mice that were chronically implanted (Figures S1J and S1K). Thus, the surgery and recording protocol are well suited for chronically monitoring cortical activity throughout late development. However, confounding factors, such as the individual developmental trajectories of mice, might impact the observed age-dependent changes. To control for these nested effects, we used linear-mixed-effect models for statistical analysis and included parameters, such as sex, mouse identity, and recording number, as random effects in our statistical model (see STAR Methods; Tables S1 and S2 for details).

Figure 1.

Figure 1

Organization of prefrontal activity throughout late development

(A) Top, schematic of the recording setup and electrode positioning within L2/3 of the mPFC. Bottom, timeline of chronic recordings.

(B) Characteristic examples of LFP activity (middle) accompanied by color-coded wavelet spectra (top) and electromyography (EMG, bottom) of a mouse during resting phase recorded at P16, P30, P40, and P58, respectively.

(C) Color-coded power spectra, averaged for all recordings during resting periods from P16 to P60 mice (n = 195 recordings, 26 mice).

(D) Left, scatterplot displaying the LFP power (1–12 Hz) of mice recorded from P16 to P60 (n = 195 recordings, 26 mice). Data are fitted by a linear regression (red line) and a cubic-spline (blue line) model. Dashed gray lines indicate local minima and maxima of the spline fit. Violin plots display the same data when grouped by age (n = 146 recordings, 26 mice). Middle, same as left for 12–100 Hz. Right, same as left for the slow-fast frequency MI, quantifying the proportional change of slow vs. fast prefrontal activity.

(E) Color-coded age-averaged power spectra (left, n = 195 recordings, 26 mice) and example power spectra (right) during resting periods.

(F) Left, scatterplot displaying detected peak frequencies within the 12–100 (top) and 1–12 Hz band (bottom) of mice chronically recorded from P16 to P60 (n = 195 recordings, 26 mice). Data are fitted by a linear regression (red line) and a cubic-spline (blue line) model. Right, violin plots displaying the same data as left when grouped by age (n = 146 recordings, 26 mice).

(G) Same as (F) for amplitudes of detected peak frequencies.

Data of violin plots are presented as median with 25th and 75th percentile. ∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001. See also Figures S1–S3 and Tables S1 for detailed statistics and S3.

We collected a large dataset (for recording summary, see Table S3) of local field potential (LFP, n = 195 recordings from 26 mice) and single unit activity (SUA, n = 7,230 units). During the investigated time window of P16–P60, visual inspection of the age-aligned power spectra revealed that the resting-state LFP activity, especially within faster frequency ranges, underwent substantial fluctuations. During early adolescence, the progressive increase in prefrontal power with age was interrupted by a pronounced decrease in fast activity at P35–P45, followed by a power increase during early adulthood (Figures 1B and 1C). Frequency band-specific quantification of average power revealed distinct trajectories for slow (1–12 Hz) and fast (12–100 Hz) frequencies. Although the dynamic of slow activity in the PL was largely age-independent, power in the fast frequency range increased with age (Figure 1D). In line with these power changes, we considered the following four distinct age groups: pre-juvenile, with a pronounced increase in fast activity power (PreJuv, P16–P23); early adolescent, with a first peak of fast activity (EarlyAdo, P28–P35); late adolescent, with a decrease in fast activity (LateAdo, P36–P43); and adult (Adult, P53–P60). The nomenclature of the age groups matches previous classifications based on behavioral and physical markers of puberty and sexual maturity in rodents.17,22 For fast PFC activity and the slow-fast frequency modulation index (MI), the residuals of the EarlyAdo group showed a significant deviation from linear statistics. Moreover, a restricted cubic-spline model represented the dynamics of the data more precisely (i.e., reduction of root mean square error [RMSE] and a 3-fold increase in R2) than a linear regression and introduced a peak at P32 and a trough at P44 (Figure 1D; Table S1). No consistent peaks or troughs were identified for changes in locomotor behavior across age (Figures S1B–S1D). Thus, nonlinear changes characterized the development of fast oscillatory activity during adolescence, and they were largely independent of movement dynamics across ages.

In line with recent recommendations for studying neural oscillations,28 we detected peak frequencies of the most prominent amplitudes within each power spectrum. Similar to the activity recorded in anesthetized mice,26 fast peak frequencies resided within the gamma band (30–80 Hz) and increased linearly with age, from ∼30 Hz at P16 to ∼65 Hz in adulthood. Peak amplitudes also increased with age but peaked during early adolescence, dropped during late adolescence, and increased again until adulthood. By contrast, the slow activity showed no age-dependent changes in peak frequency or amplitude. Only the peaks, which were mainly confined to the theta range (4–10 Hz), gained better definition with increasing age. Although initially only one main peak at ∼5 Hz was detected, a bimodal distribution with peaks at ∼4 and ∼8 Hz characterized the slow activity of older mice (Figures 1E and 1G; Table S1).

During movement periods, the overall developmental dynamics of prefrontal activity did not change; however, the amplitude of LFP power was larger when compared with resting-state activity. Although slow activity during resting showed no age-dependent changes in peak amplitude and frequency, during running, slow peak frequencies increased in their amplitude and frequency from ∼4 Hz in PreJuv mice to ∼8 Hz during adolescence (Figures S2A–S2E). To exclude a potential influence of the applied recording methodology on the results, we developed a novel recording protocol relying on chronic implantation of silicon probes and monitoring of mice (LFP, n = 130 recordings from 4 mice) between P16 and P200. Under this freely moving condition, mice moved more and at higher speeds than under head-fixed conditions. Moreover, after pre-juvenile age, the time spent moving and the movement speed increased, most likely as a result of growing body size (Figures S3A–S3E). Despite these motility differences, the gamma development had a similar nonlinear dynamic as observed under head-fixed conditions, with a peak detected at P30 and a trough at P41 (reduction of RMSE and a 2-fold increase in R2 in restricted cubic-spline fit compared with linear fit). The slow activity increased in amplitude and frequency until adulthood, similar to the activity recorded during movement under head-fixed conditions. Notably, no differences were present between P53–P60 and P80–P200 mice, indicating that the gross patterns of prefrontal LFP activity reached maturity by P60 (Figures S3F–S3H; Table S1). Thus, age-dependent changes of slow activity, but not gamma activity, varied with state (i.e., resting vs. movement).

Adult-like prefrontal oscillatory activity and theta-gamma synchrony co-emerge toward the end of adolescence

Since the amplitude of fast activity is often modulated by the phase of slow oscillations,29,30 we hypothesized that the age-dependent changes in prefrontal slow and fast frequency oscillations affect prefrontal synchrony. To test this hypothesis, we quantified the phase-amplitude coupling (PAC) within the PL from P16 to P60. Coupling increased with age and showed an adult-like pattern31 in P53–P60 mice (Figures 2A–2C). Cross-frequency coupling has been reported to underlie information processing within and between areas in behavioral tasks29,30; thus, its late emergence might be one of the reasons mature cognitive performance is initially observed in young adulthood.

Figure 2.

Figure 2

Prefrontal theta-gamma synchrony throughout late development

(A) Color-coded PAC MI spectra averaged over age and calculated for the amplitude of 20–100 Hz activity to the averaged 3–6 Hz phase for mice recorded from P16 to P60 (n = 195 recordings, 26 mice).

(B) Left, scatterplot displaying the PAC MI calculated for the averaged amplitude of 80–100 Hz to the averaged 3–6 Hz phase for mice recorded from P16 to P60 (n = 195 recordings, 26 mice). Right, violin plots displaying the same data as left when grouped by age (n = 146 recordings, 26 mice).

(C) Color-coded PAC MI spectra when calculated for each age group (n = 146 recordings, 26 mice).

(D) Left, scatterplot displaying PPC for theta phase (4–10 Hz) of significantly locked units (n = 3,571 units, 190 recordings, 26 mice). Right, violin plots displaying the same data as left when grouped by age (n = 2,423 units, 146 recordings, 26 mice).

(E) Circular plot of the median theta phase of significantly locked units displayed for each investigated age group.

(F) Same as (D) for PPC values quantified for gamma phase (30–100 Hz), (left: 5,797 units, 191 recordings, 26 mice; right: 4,058 units, 144 recordings, 26 mice).

(G) Same as (E) for median gamma phase.

Data of violin plots are presented as median with 25th and 75th percentile. ∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001. See also Figure S4 and Tables S1 for detailed statistics and S3.

Next, we investigated how changes in oscillatory activity related to the local firing of neurons by calculating the SUA phase locking to LFP. Across all ages, ∼40% of the recorded units were significantly locked to theta frequencies and ∼80% to gamma frequencies (Figure S4A). Locking strength to theta increased with age. Although prefrontal units of PreJuv mice preferentially fired toward the trough of the theta cycle, units of older mice showed a stronger locking and a phase shift toward the peak (Figures 2D, 2E, and S4B). For gamma frequencies, we observed a decrease in locking strength with age. Age group comparisons revealed the lowest pairwise phase consistency (PPC) values for the EarlyAdo and Adult age groups (Figures 2F and S4B). In all investigated mice, unit firing preferentially preceded the gamma cycle peak, with a minor shift toward the peak with age (Figure 2G).

Overall, these data revealed an age-dependent increase in theta modulation of fast activity in the mPFC, which most likely contributed to the emergence of oscillatory theta-gamma synchrony during the late stages of development. Moreover, the decrease in gamma phase locking with age suggests that the firing of individual units diversified while being more precisely paced by theta rhythms. Correspondingly, we found the highest joint entropy of spike trains, which reflects the spiking complexity or diversity, in the EarlyAdo and Adult age groups (Figure S4C). During early adolescence, the parameters of fast activity and joint entropy matched those of adult mice but lacked adult-like theta modulation, indicating a more disorganized local network within the adolescent mPFC.

Spiking of putative prefrontal L2/3 PYRs resembles the age-dependent dynamics of gamma oscillations

To dissect the cellular substrate of the identified age-dependent fluctuations, we relied on the extracellular waveform shape to distinguish regular-spiking (RS) units, largely corresponding to L2/3 PYRs, from fast-spiking (FS) units, representing putative parvalbumin-expressing (PV+) interneurons (Figure 3A).11 Overall, the firing rates of all prefrontal neurons were higher during movement than during resting periods. The firing rates of putative L2/3 PYRs increased with age (Figures S4D and S4E; Table S1), but not linearly; they reached an initial peak during early adolescence and a second peak in adulthood. Similar nonlinear dynamics were detected for the coefficients of variation (CVs) and inter-spike intervals (ISIs) (Figures 3B–3D). For FS PV+ interneurons, only minor and largely linear changes were detected. Only their changes in CV mirrored the nonlinear dynamics found for RS units (Figures 3B–3D and S4D–S4G).

Figure 3.

Figure 3

Spiking dynamics of putative prefrontal layer 2/3 pyramidal neurons throughout late development

(A) Left, scatterplot displaying trough to peak latency as a function of half-width of all units recorded from P16 to P60 mice (n = 7,230 units, 192 recordings, 26 mice). Red dotted lines correspond to classification thresholds for RS (black) and FS (red) units. Right, color-coded scatterplots of data shown left, displaying age (top), firing rate (middle), and CV (bottom).

(B) Violin plots displaying the firing rates of RS (left, n = 3,936 units, 143 recordings, 26 mice) and FS (right, n = 1,146 units, 135 recordings, 26 mice) units for investigated age groups.

(C) Same as (B) for CV.

(D) Same as (B) for average ISI during active periods.

(E) Left, schematic drawing indicating the position of recording sites along the dorsal-ventral axis of the mPFC. Right, color-coded correlation matrix of TC values quantified for 3 ms (top) and 100 ms (bottom) latency and averaged for each age group.

(F) Violin plots displaying the TC values between RS-RS unit-pairs for 3 ms (top) and 100 ms (bottom) latency (n = 85,422 unit-pairs, 143 recordings, 26 mice).

Data of violin plots are presented as median with 25th and 75th percentile. ∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001. See also Figure S4 and Tables S1 for detailed statistics and S3.

These age-dependent changes in neuronal activity of L2/3 PYRs might affect neuronal communication within the mPFC. To address this hypothesis, we calculated the spike-time tiling coefficient (TC), a firing rate-independent measure of correlated spiking activity.32 The correlated activity between putative L2/3 PYRs present in PreJuv mice augmented and broadened along the dorsal-ventral axis of the mPFC in EarlyAdo mice. During late adolescence, the degree of correlation dramatically decreased. Subsequently, TCs increased to reach the adult level in P53–P60 mice. Moreover, a precisely defined pattern of correlation was detected at this age within the PL and IL (Figures 3E and 3F). Overall, these results resembled the fluctuations found for gamma oscillations, indicating that L2/3 PYRs might contribute to nonlinear changes occurring in the adolescent mPFC.

The inhibitory effect of interneuronal firing in the mPFC gradually increases with age

Even if the firing properties of FS PV+ interneurons seem to relate poorly to the nonlinear reorganization of prefrontal activity during adolescence, their inhibitory effect might be altered. Moreover, waveform-based clustering mainly identifies FS PV+ interneurons, leaving out other subtypes of interneurons. Previous studies have identified a decrease in calretinin (CR) and an increase in PV expression, as well as a moderate increase in somatostatin (SOM) expression, in the mPFC during adolescence through early adulthood.7,18,26,33 To better monitor the developmental dynamics of these subtypes, we optogenetically manipulated the interneuronal activity in the developing mPFC. Thus, we expressed the excitatory channelrhodopsin 2 (ChR2) E123T/T159C variant selectively in Dlx5/6+ interneurons (Figure 4A), comprising mainly CR+, PV+, and SOM+ interneurons.34 From P16 to P60, 3 s long ramped light activation of interneurons led to a broadband decrease in the LFP power. The slope of individual optogenetic MI (OMI) spectra showed an age-independent positive tilt. By contrast, the intercept, a measure of inhibition strength, progressively shifted toward more negative values with age. When compared with baseline, the post-stimulus LFP activity increased; however, the modulation reached significance only in adulthood (Figures 4B–4E; Table S1).

Figure 4.

Figure 4

The inhibitory effect of interneuronal firing in prefrontal circuits throughout late development

(A) Left, drawing illustrating viral targeting of interneurons in mPFC and the viral construct. Right, coronal section from a P58 eYFP-expressing Dlx5/6-cre mouse.

(B) Characteristic examples of LFP activity during blue ramped light stimulation of Dlx 5/6+ interneurons of a mouse recorded at P17, P30, P40, and P58, respectively.

(C) OMI spectra of LFP activity pre vs. during (left) and pre- vs. post-ramp stimulation (right) of each age group (n = 34 recordings, 6 mice).

(D) Scatterplots displaying the slope (left) and the intercept (middle) of individual OMI spectra for mice recorded from P16 to P60 (n = 47 recordings, 6 mice). Right, violin plots displaying the same data as left when grouped by age (n = 34 recordings, 6 mice).

(E) Same as (D) for post-stimulation periods.

(F) Left, bar graph displaying the percentage of significantly modulated units. Right, color-coded raster plots displaying the Z score of firing rate change of all recorded single units for each age group.

(G) Line plots displaying firing rate change of inhibited (left, n = 945 units, 34 recordings, 6 mice) and activated (right, n = 530 units, 34 recordings, 6 mice) units for each age group. Inset, firing rate change 1 s post stimulation displayed at higher magnification. Color-striped bars indicate significant difference.

(H) Left, volcano plot displaying the OMI as a function of the p value of firing rate change for mice recorded from P16 to P60 (n = 2,058 units, 47 recordings, 6 mice). Right, violin plots displaying the p value of firing rate change when grouped by age (n = 1,476 units, 34 recordings, 6 mice).

(I) Same as (F) for effects 10 ms after 3 ms square pulse stimulation.

(J) Same as (G) for firing rate changes after 3 ms square pulse stimulation.

(K) Left, scatterplot displaying the post-stimulus time that single units require to reach pre-stimulus firing rates (n = 2,058 units, 47 recordings, 6 mice). Right, violin plots displaying the same data as left when grouped by age (n = 1,476 units, 34 recordings, 6 mice).

For line plots in (C), (G), and (J) data are presented as mean ± SEM. Data of violin plots are presented as median with 25th and 75th percentile. ∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001. See also Figures S5 and S6 and Tables S1 for detailed statistics and S3.

The light-induced modulation of LFP activity was accompanied by an age-dependent increase in the fraction of significantly modulated single units and the modulation strength (Figures 4F and 4H; Table S1). The inhibited single units (i.e., putative L2/3 PYRs) showed a progressive age-dependent increase in firing after the light stimulus (Figure 4G). Such rebound firing upon activation of interneurons has been previously reported in adults and has been shown to depend on postsynaptic GABAA receptor signaling.35 Thus, the present data suggest an augmented and more efficient inhibitory drive of prefrontal interneurons onto L2/3 PYRs with age. The activated portion of single units (i.e., putative interneurons) showed an age-dependent increase in firing rate during stimulation (Figure 4G), indicating that the ability of interneurons to follow the optogenetic stimulus increases with age.36

Besides ramp light stimulation, we used 3 ms long light pulses to investigate network responses to brief interneuronal firing. The pulsed stimulation activated a comparable proportion of interneurons across age groups, confirming an age-independent efficiency of optogenetic stimulation. By contrast, the percentage of significantly inhibited units increased until late adolescence (Table S1). Brief interneuronal firing was followed by a prolonged decrease in overall firing (Figures 4I and 4J). The strength of this inhibitory effect increased with age, as reflected by the progressively longer time neurons needed to recurrently reach their baseline firing rates (Figure 4K).

In response to the optogenetic ramp stimulation, most RS units were significantly inhibited, whereas most FS units were activated. However, a small proportion of FS units, present in all investigated age groups, showed a decrease in firing during ramp stimulation, which was followed by pronounced rebound firing (Figures S5A–S5F). To assess whether these responses arose from non-expressing FS units, we photo-tagged expressing FS and RS units with pulsed light stimulation and investigated their responses to ramped light stimulation. In the PreJuv group, tagged FS units mainly displayed an activation during ramp stimulation. However, with age, the proportion of inhibited tagged FS units increased, suggesting a strengthening of interneuron-interneuron connections (Figures S5G and S5H; Table S1).

To assess the non-specific effects of the optogenetic protocols used, we employed the same stimulation paradigm in Cre− control mice. Neither the LFP activity nor SUA was affected, with only a very small percentage of recorded units being significantly modulated (Figures S6A–S6C; Table S1). Moreover, we tested whether the observed developmental dynamics relied on age-dependent changes in viral expression. For this, mice were transfected with the viral construct containing the ChR2 at P43 and investigated from P53 to P60. Light activation of interneurons led to similar effects as those detected in chronically recorded mice of the Adult group (Figures S6D–S6L).

Thus, optogenetic activation of interneurons revealed that inhibition increased with age in the mPFC. However, the dynamics of these developmental changes contrasted with the nonlinear reorganization of fast network activity that occurred during the same time period. Therefore, we suggest that the L2/3 PYRs, and to a lesser extent the interneurons, control the activity changes in the adolescent mPFC.

Prefrontal L2/3 PYRs experience microglia-mediated structural remodeling during adolescence

The observed functional changes within the mPFC throughout adolescence might be the result of structural modifications.11,37 Correspondingly, we hypothesized that structural and synaptic changes of prefrontal L2/3 from P16 to P60 accounted for the described nonlinear maturation of prefrontal activity. To test this hypothesis, we tagged L2/3 PYRs with a fluorescent protein by using in utero electroporation (IUE) at embryonic age (E)15.5. As we previously showed, this strategy is suitable for sparse labeling of PYRs exclusively confined to prefrontal L2/3.11,12 Morphological investigation of labeled neurons revealed that the dendritic arborization complexity and the dendritic length of L2/3 PYRs peaked during early adolescence (P33–P35) (Figures 5A–5C). Moreover, EarlyAdo mice showed an augmented spine density, regardless of the dendrite type, compared with younger or older mice (Figures 5D and 5E). Thus, increased prefrontal activity during early adolescence is related to prominent dendritic and synaptic maturation of L2/3 PYRs at this age. Since most excitatory synapses are formed on dendritic spines,38 early adolescence is likely characterized by an increase in excitatory connectivity within the mPFC.

Figure 5.

Figure 5

Dendritic arborization complexity and spine density of prefrontal layer 2/3 pyramidal neurons throughout late development

(A) Characteristic photographs of tDimer2-expressing L2/3 PYRs within the PL displayed together with average heatmaps of all overlayed dendrites for each age group.

(B) Dendritic intersections within a 200 μm radius from the soma center of prelimbic L2/3 PYRs (n = 12 mice, 71 neurons). Color-striped bars indicate significant difference.

(C) Violin plots displaying the total branch length (left) and number of branches (right) of the cells analyzed in (B).

(D) Violin plots displaying spine densities of basal (i.e., originating from the soma; left, n = 17 mice, 60 neurons, 358 dendrites) and distal (i.e., originating from the apex/apical dendrite; right, n = 17 mice, 60 neurons, 251 dendrites) dendrites of L2/3 PYRs within the PL.

(E) Characteristic photographs of basal dendrites from each age group.

For line plots in (B), data are presented as mean ± SEM. Data of violin plots are presented as median with 25th and 75th percentile. ∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001. See also Figure S6 and Table S1 for detailed statistics.

(F) Characteristic photographs of Iba1 immunostainings in the PL of each age group.

(G) Violin plot displaying the densities of Iba1-positive cells within the PL (n = 25 mice, 800 stacks).

(H) Violin plots displaying roundness, cell spread, cell area, and cell perimeter for Iba1-positive cells within the PL (n = 25 mice, 26,707 cells).

(I) Characteristic reconstructions of processed raw images of Iba1-positive cells (green) containing CD68- (blue) and PSD95- (red) positive puncta of L2/3 PYRs from each age group. Pink corresponds to an overlay of all three colors.

(J) Bar diagrams displaying the percentage of microglia with colocalized puncta (n = 17 mice, 625 cells).

(K) Violin plots displaying volume of vesicles (left) and inclusions (right), quantified on colocalized Iba1-, CD68- or Iba1-, CD68-, and PSD95-positive pixel, respectively (n = 17 mice, 625 cells).

(L) Same as (K) for the number of inclusions.

(M) Violin plot displaying the phagocytic ratio (i.e., MI of colocalized PSD95 vs. solely CD68 puncta, n = 17 mice, 625 cells).

For line plots in (B), data are presented as mean ± SEM. For (K) and (L), data were normalized to Iba1 volume. Data of violin plots are presented as median with 25th and 75th percentile. ∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001. See also Table S1 for detailed statistics.

Microglia have been previously identified to shape neuronal morphology in an activity-dependent manner via active phagocytic mechanisms.39,40,41 Since they might also account for the structural and synaptic changes observed for L2/3 PYRs, we investigated the microglia in prefrontal L2/3 of P16–P60 mice. Microglia density decreased until late adolescence. Moreover, adolescent microglia presented the most rounded and least ramified morphology when compared with the other age groups (Figures 5F and 5H). To better relate these findings to the structural modification of L2/3 PYRs, we next tagged the postsynaptic density protein PSD95 of these neurons with a fluorophore by using IUE at E15.5, enabling the quantification of phagocytosed spines. Although a moderate percentage of microglia showed phagocytic activity in the pre-juvenile and adult stages, this level was augmented during adolescence (Table S1). Not only were more microglia phagocytically active throughout adolescence but also the analyzed cells showed enlarged lysosomal vesicles containing an increased amount of postsynaptic material. Notably, the volume of inclusions, as well as the phagocytic ratio, increased during early adolescence, indicating more densely packed lysosomal vesicles or the engulfment of larger compartments (Figures 5I–5M). Thus, these data revealed that microglia contributed to the structural remodeling of L2/3 PYRs and, consequently, might shape prefrontal circuits that generate fast network activity.

Interfering with adolescent microglia-driven remodeling causes a disruption of prefrontal network activity

Next, we asked whether the nonlinear pattern of fast network activity development is critical for adult circuit function and cognitive abilities. To accomplish this, we disrupted microglia activity during early adolescence and monitored the consequences of manipulation at the neuronal, circuit, and behavioral levels (Figure 7A). We focused on microglia because (1) as shown above, they are active contributors to structural and functional changes of L2/3 PYRs during adolescent development and (2) their ablation during distinct stages of development has been shown to result in diverse morphological and behavioral alterations in an age- and brain area-dependent manners.13,42,43,44 To ablate microglia, we used the diet-administered colony-stimulating factor 1 receptor inhibitor, PLX3397,44,45,46 which causes a brain-wide clearance of microglia. Previous studies have shown that this approach causes no lasting changes in behavior or neuronal activity in adult mice, even after weeks of application.45,46 Here, PLX3397 administration during adolescence (P28–P37) strongly reduced microglia density. Moreover, the surviving portion of cells already showed an increased ramification and a less rounded morphology 3 days after administration onset. However, these effects were transient, as 4 days after PLX3397 treatment, microglia started to repopulate the tissue and showed a morphological phenotype comparable with the cells of control mice (Figures 6A and 6B).

Figure 7.

Figure 7

Adult cognitive abilities and prefrontal morphology after microglia manipulation during adolescence

(A) Left, color-coded heatmaps of the time spent at different locations within the open-field arena by Control I (left) and PLX I (right) P56–P60 mice. Right, violin plots displaying relative time spent in the center (given as %, left) and relative resting time (given as %, right) in the open-field arena (n = 41 mice).

(B) Violin plots displaying grooming duration (left) and frequency (right) in the open-field arena (n = 41 mice).

(C) Left, color-coded heatmaps of the time spent at different locations within the eight-arm maze by Control I (left) and PLX I (right) P56–P60 mice during all trials. Right, line plot displaying the number of working memory (left) and reference (right) errors for all 12 trials. Violin plots displaying the normalized slope of the fitted working memory (left) and reference memory (right) performance of individual mice (n = 14 mice). Color-striped bars indicate significant difference.

(D) Left, schematic of the experimental protocol during odor discrimination. Right, violin plots displaying the number of trials (left) and average trial latency (right) required by each mouse to reach criterium (n = 27 mice).

(E) Left, schematic of the experimental protocol during odor reversal. Right, violin plots displaying the number of trials (left), average trial latency (middle), and total errors (given as %, right) required by each mouse to reach criterium (n = 27 mice).

(F) Characteristic photographs of tDimer2-expressing L2/3 PYRs within the PL of Control I (left) and PLX I (right) P56–P60 mice.

(G) Left, dendritic intersections within a 200 μm radius from the soma center of prelimbic L2/3 PYRs (n = 6 mice, 37 neurons). Color-striped bars indicate significant difference. Right, violin plots displaying the total branch length (left) and the number of branches (right) of the cells analyzed left.

(H) Violin plots displaying spine densities of basal (i.e., originating from the soma; left, n = 6 mice, 24 neurons, dendrites) and distal (i.e., originating from the apex/apical dendrite; right, n = 6 mice, 24 neurons, dendrites) dendrites of L2/3 PYRs within the PL.

(I) Violin plots displaying the densities (left, n = 6 mice, 200 stacks) and roundness (right, n = 6 mice, 5504 cells) of Iba1-positive cells within the PL.

(J) Characteristic reconstructions of processed raw images of Iba1-positive cells (green) containing CD68- (blue) and PSD95- (red) positive puncta of L2/3 PYRs from Control I (left) and PLX I (right) P58–P60 mice. Pink corresponds to an overlay of all three colors.

(K) Left, violin plots displaying volume of inclusions (left) and the number of inclusions (right), quantified on colocalized Iba1-, CD68-, and PSD95-positive pixel (n = 6 mice, 270 cells). Right, bar diagrams displaying the percentage of microglia with colocalized puncta.

For line plots in (C) and (G), data are presented as mean ± SEM. Data of violin plots are presented as median with 25th and 75th percentiles. ∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001. See also Figures S7 and S8 as well as Table S1 for detailed statistics.

Figure 6.

Figure 6

Remodeling of prefrontal but not somatosensory network activity after microglia manipulation during adolescence

(A) Characteristic photographs of Iba1 immunostainings in the PL of control mice (left), PLX3397-treated mice (3, 6, 10 days after the diet onset, middle) as well as 10 day-long PLX3397-treated mice 4 days after application termination (right).

(B) Violin plots displaying the density of Iba1-positive cells (left, n = 348 stacks, 16 mice) as well as the cell perimeter and area (right, n = 4,970 cells, 16 mice) of control and PLX-treated mice.

(C) Schematic of the experimental procedure.

(D) Left, schematic of recording conditions. Right, digital photomontage of a coronal section showing the electrode position in the mPFC.

(E) Left, power spectra of LFP activity recorded in the PL from P56 to P60 Control I and PLX I mice. Right, scatterplots of peak frequencies as a function of peak amplitudes for the 1–12 and 12–100 Hz activity, respectively (n = 28 recordings, 14 mice).

(F) Left, scatterplot displaying trough to peak latency as a function of half-width of all units recorded in mPFC of P56–P60 Control I and PLX I mice. Red dotted lines correspond to classification thresholds for RS (black) and FS (red) units (n = 882 units, 28 recordings, 14 mice). Right, violin plots displaying the firing rates of RS and FS units, respectively (RS: n = 730 units, 28 recordings, 14 mice; FS: n = 152 units, 27 recordings, 14 mice).

(G) Left, line plot displaying the distance between RS-RS unit-pairs recorded in mPFC as a function of the TC value quantified for 3 ms latency. Violin plots displaying the TC values for RS-RS unit-pairs recorded from P56 to P60 Control I and PLX I mice (n = 9,839 unit-pairs, 28 recordings, 14 mice). Right, same as left for FS-FS unit-pairs (n = 407 unit-pairs, 27 recordings, 14 mice).

(H) Left, digital photomontage of a coronal section showing the electrode position in the S1. Right, same as (E) for power spectra of LFP activity recorded in the S1 (n = 28 recordings, 14 mice).

(I) Left, firing rates of units recorded in the S1 of P56–P60 Control I and PLX I mice (n = 644 units, 28 recordings, 14 mice). Right, same as (G) for unit-pairs recorded in the S1 (n = 7,728 unit-pairs, 28 recordings, 14 mice).

(J) Left, gPDC spectrum of information flow from the mPFC to the S1 and quantifications of gPDC peak frequencies as a function of peak amplitudes for the 12–100 Hz band (n = 28 recordings, 14 mice). Right, same as left for information flow from the S1 to the mPFC (n = 28 recordings, 14 mice).

(K) Left, line plot displaying spike-spike cross-covariance calculated for 400 ms latencies between units simultaneously recorded in the S1 and the mPFC (n = 342 driver-units, 28 recordings, 14 mice). Right, bar plot displaying peak count of individual spike-spike cross-covariance spectra shown left. Color-striped bars indicate significant difference.

For line plots in (E), (G), (H), (I), (J), and (K) data are presented as mean ± SEM. Data of violin plots are presented as median with 25th and 75th percentiles. ∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001. See also Figures S7 and S8 as well as Table S1 for detailed statistics.

To control for the specificity of PLX3397-induced effects, we also investigated the primary somatosensory cortex (S1) as a cortical control area. In young adult mice (P56–P60), extracellular recordings revealed major changes in the prefrontal network activity in PLX3397-treated mice (PLX I) compared with controls (Control I). Although theta peak power significantly decreased, gamma peak power increased and shifted toward faster frequencies (Figures 6D and 6E). On the SUA level, waveform-based classification revealed that RS units (i.e., putative L2/3 PYRs), but not FS units (i.e., putative FS PV+ interneurons), showed an increase in firing rates and a stronger correlation of spiking activity in PLX I mice. Notably, these results mirrored the circumstances present during early adolescence and suggested that PLX3397-treated mice lacked the previously identified developmental fluctuations. By contrast, neither the network nor neuronal spiking activity in S1 was affected by PLX3397 administration (Figures 6H and 6I). To assess whether the disrupted prefrontal activity in PLX I mice affected the information flow between the PFC and S1, we quantified the generalized partial directed coherence (gPDC), a measure that reflects the frequency-dependent directionality of area communication.47 Although the information flow S1 → mPFC did not differ between PLX I and Control I mice, the top-down control mPFC → S1 was reduced and, in line with the gamma activity, shifted toward faster frequencies (Figure 6J). Moreover, we assessed whether prefrontal firing matched the spiking in S1 and the effects of the PLX3397 diet on this relationship. In Control I mice, prefrontal spiking preceded S1 firing by ∼200 ms, confirming the presence of a top-down drive from mPFC → S1. By contrast, in PLX I mice, no directionality was present, and most peaks were randomly distributed (Figure 6K), suggesting the loss of prefrontal top-down drive.

Taken together, these results indicate that microglia ablation during the developmental time window of major circuit reorganization profoundly affected the activity and oscillatory coupling of mPFC and shifted the networks toward a more immature state. Although the PLX3397-induced microglia ablation was brain-wide, these effects were not detected in S1, suggesting a particular vulnerability of the mPFC during adolescence.

Adolescent microglia ablation leads to poor cognitive abilities and morphological alterations in adulthood

Since microglia activity during adolescence drives the reorganization of prefrontal circuits, their ablation during adolescence might impair or delay the emergence of PFC-dependent behavioral abilities. To test this hypothesis, we conducted several behavioral tests on P56–P60 mice.

First, monitoring spontaneous behaviors revealed that PLX I mice showed no alterations in open-field behavior, but the duration and frequency of their self-grooming increased, suggesting a preference for repetitive behaviors (Figures 7A and 7B). Notably, such behavioral traits are typical for adolescent but not adult rodents,2,7 suggesting that PLX I mice exhibit immature behavior. Second, we monitored working memory using a food-baited eight-arm maze. Although the performance of Control I mice gradually increased across trials, PLX I mice improved to a lesser extent, making more working and reference memory errors over time (Figure 7C). Third, we assessed decision-making abilities during an attentional four-choice set-shifting task in which mice had to associate an odor with a reward during a discrimination and a reversal phase. PLX I mice required more trials to learn the task, as well as more time and trials during rule reversal, and made more errors overall (Figures 7D and 7E).

Functional and behavioral alterations in adult PLX I mice were accompanied by increased basal spine densities (i.e., on dendrites originating from the soma) of L2/3 PYRs, whereas no differences in dendritic complexity were detected. By contrast, at the same age, microglia showed increased phagocytic activity but reduced density, most likely limiting the overall effect of microglia dysfunction on neuronal morphology (Figures 7F–7K). Thus, PLX3397 treatment during adolescence altered prefrontal network activity and morphology, which might have contributed to poorer cognitive performance.

Do these functional and behavioral deficits persist throughout life, or are they compensated for? To address this question, we recorded the same mice, which were defined as Control II and PLX II (P98–P102), ∼70 days after the PLX3397 administration (Figure 8A). The lack of changes in the neuronal and network activity of S1 persisted. However, prefrontal activity was still impaired, but in a different way. Although, in young adults, gamma power was augmented as a result of the PLX3397 diet, in PLX II mice, it was significantly lower (Figures 8B and 8C). This effect might have been due to an overshoot of compensatory mechanisms48 or hypofunctional circuit maturation (i.e., delayed pruning prevents developmental myelination49). Regardless of the contrasting changes in prefrontal activity, behavioral disabilities persisted to a similar extent, including a reduction in spontaneous alternations within a Y-maze and slower discrimination and reversal learning during the attentional set-shifting task. Only the increased grooming behavior present in PLX I mice returned to normal levels in PLX II mice (Figures 8D–8G). Morphological investigation revealed that L2/3 PYRs had decreased dendritic complexity and basal spine densities, most likely due to microglia dysfunction, i.e., increased densities and phagocytic activity (Figures 8H–8M). Thus, the PLX3397 diet might delay developmental pruning during adolescence, which is followed by a lasting overactivation of microglia cells, ultimately leading to disturbed prefrontal functioning.

Figure 8.

Figure 8

Long-lasting functional, behavioral, and morphological consequences of microglia manipulation during adolescence

(A) Schematic of the experimental procedure.

(B) Left, power spectra of LFP activity recorded in the PL of P98–P102 Control II and PLX II mice. Right, scatterplot of peak frequencies as a function of peak amplitudes for 1–12 and 12–100 Hz bands, respectively (n = 11 mice, 22 recordings).

(C) Same as (B) for LFP activity recorded in the S1 (n = 11 mice, 22 recordings).

(D) Top, color-coded heatmaps of the time spent at different locations within the open-field arena by Control II (left) and PLX II (right) P98–P102 mice. Bottom, violin plots displaying the relative time spent in the center (given as %, left), relative resting time (given as % middle), and grooming frequency (right) in the open-field arena (n = 24 mice).

(E) Left, schematic of the experimental protocol of spontaneous alternations monitoring. Right, violin plots displaying the number of entries (left) and percentage of alternations (right) in the Y-maze (n = 24 mice).

(F) Left, schematic of the experimental protocol during odor discrimination. Right, violin plots displaying the number of trials (left) and average trial latency (right) required by each mouse to reach criterium (n = 24 mice).

(G) Left, schematic of the experimental protocol during odor reversal. Right, violin plots displaying the number of trials (left), average trial latency (middle), and total errors (given as %, right) required by each mouse to reach criterium (n = 24 mice).

(H) Characteristic photographs of tDimer2-expressing L2/3 PYRs within the PL of Control II (left) and PLX II (right) P98–P102 mice.

(I) Left, dendritic intersections within a 200 μm radius from the soma center of prelimbic L2/3 PYRs (n = 6 mice, 36 neurons). Color-striped bars indicate significant difference. Right, violin plots displaying the total branch length (left) and number of branches (right) of the cells analyzed left.

(J) Violin plots displaying spine densities of basal (i.e., originating from the soma; left, n = 6 mice, 24 neurons, 121 dendrites) and distal (i.e., originating from the apex/apical dendrite; right, n = 6 mice, 24 neurons, 101 dendrites) dendrites of L2/3 PYRs within the PL.

(K) Violin plots displaying the densities (left, n = 6 mice, 200 stacks) and roundness (right, n = 6 mice, 6,149 cells) of Iba1-positive cells within the PL.

(L) Characteristic reconstructions of processed raw images of Iba1-positive cells (green) containing CD68- (blue) and PSD95- (red) positive puncta of L2/3 PYRs from Control II (left) and PLX II (right) P98–P102 mice. Pink corresponds to an overlay of all three colors.

(M) Left, violin plots displaying volume of inclusions (left) and the number of inclusions (right), quantified on colocalized Iba1-, CD68-, and PSD95-positive pixel (n = 6 mice, 252 cells). Right, bar diagrams displaying the percentage of microglia with colocalized puncta.

For line plots in (B), (C), and (I), data are presented as mean ± SEM. Data of violin plots are presented as median with 25th and 75th percentiles. ∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001. See also Table S1 for detailed statistics.

Finally, to test whether the microglia-driven reorganization of the mPFC was confined to adolescence, we applied the same PLX3397 treatment protocol to pre-juvenile mice (Figure S7A). This microglia ablation did not affect the neuronal or network activity of S1 or the mPFC in P56–P60 mice (Figures S7B and S7C). Correspondingly, open-field and spontaneous behaviors, as well as cognitive abilities during attentional set-shifting, were similar in controls and PLX3397-treated mice. Only the behavioral performance in the spontaneous alternation task in the Y-maze showed improvement in mice treated with PLX3397 at the pre-juvenile age (Figures S7D–S7F). Similarly, when the PLX3397 treatment was applied to adults, neither functional nor behavioral consequences were apparent ∼30 days after the treatment (Figures S8A–S8F). Thus, transient microglia ablation during the adolescent stage, but not the pre-juvenile or adult stages, had extensive and long-lasting consequences on prefrontal network activity and function.

Discussion

Late development, broadly defined as the transitional period between childhood (i.e., early development) and adulthood, is evolution’s solution for bringing the capacity of complex brain functions to fruition. Since the structural changes during this period are modest compared with those occurring during early development, the functional remodeling of circuitry is considered a marker of the late developmental stage. Especially in the PFC, this process has been hypothesized to control the emergence or optimization of most cognitive abilities, such as working memory and decision making.17,50 It has been postulated that poor cognitive performance, a hallmark of several neuropsychiatric disorders, results from a disruption of this remodeling process during late development. However, the functional remodeling of prefrontal circuitry at this age is poorly understood. Here, we fill this knowledge gap and show that (1) although prefrontal synchrony, SUA locking to the theta phase, and inhibition progressively increase with age, gamma band activity and spiking complexity fluctuate nonlinearly, (2) these fluctuations resemble the firing patterns and dendritic arborization and spine density of L2/3 PYRs that undergo pruning from early to late adolescence as a result of phagocytic activity of microglia, and (3) ablation of microglia during the developmental time window of circuit reorganization, but neither before nor after, alters prefrontal function, morphology, and cognitive maturation. These results indicate that, in line with the aforementioned long-standing hypothesis, major remodeling of local prefrontal circuits occurs during late development. The pre-structure resulting after early development disorganizes during adolescence. Subsequently, more adult-like focal patterns are present.

Previous studies demonstrated the critical role of electrical activity during the early stage of development (i.e., neonatal period) for adult circuit function and behavioral performance.8,11,51 Correspondingly, during this period, the circuits are particularly vulnerable to insults. For example, we previously showed that abnormal patterns of oscillatory activity are present during the neonatal period in the PFC of a mouse model of mental disorders as a result of impairment of L2/3 PYRs and contribute to cognitive deficits in adulthood.12 This study complemented those findings and identified the adolescent period during late development as another potential “window of sensitivity” during which circuits are profoundly reorganized and thus vulnerable to endogenous and external factors. Notably, a similar acceleration of gamma frequency, as well as a reduction in spike-field coherence, has been previously identified in the visual cortex during critical period plasticity.52

Although age-dependent changes in gamma activity were largely state-independent, activity in slow frequency ranges showed distinct dynamics across behavioral states. The observed increase in peak amplitude and accelerating of peak frequencies from initially 4–8 Hz in adulthood during movement, when compared with the resting state, most likely resulted from an increased hippocampal drive.53,54 The prefrontal-hippocampal network has been shown to reach functional and anatomical maturities toward the end of adolescence.55 We hypothesize that the development of prefrontal-hippocampal communication contributes to the observed acceleration and augmentation of prefrontal theta activity.

To date, only a few studies have investigated prefrontal neuronal and network activities during late development. In the PFC subdivision anterior cingulate cortex (ACC), the firing rates of top-down projection neurons and the frequency of excitatory postsynaptic currents (EPSCs), but not inhibitory postsynaptic currents (IPSCs), increased during adolescence, boosting the refinement of long-range connectivity.15 Similarly, in the mPFC, the EPSC frequency of L2/3 PYRs peaked at the same developmental stage.56 Moreover, in the orbitofrontal cortex, adolescent neurons showed higher firing variability and excitability during motivated behaviors.57 Although such studies mainly focus on broad age groups and lack systematic monitoring of electrical activity patterns throughout development, they provide support for the increased excitatory activity in the mPFC during adolescence before the onset of reorganization. The subsequent breakdown of connectivity (i.e., a disorganized network) has also been identified in humans, with electroencephalogram (EEG)/magnetoencephalography studies detecting a lower signal-noise ratio in children and adolescents than in adults.58 In addition, although EEG stimulus-induced gamma power was found to undergo major fluctuations from childhood to adulthood, with a peak during early adolescence, neuronal synchrony and cognitive performance increased until adulthood.59,60 These results are in line with our findings and suggest an enhancement of coordinated local processing from adolescence to adulthood, finally leading to the emergence of mature cognitive functioning.57,59

The observed age-dependent changes in prefrontal synchrony might be the result of inhibitory network maturation throughout late development, as previous studies identified inhibition as critical for the coordination and synchronization of neuronal activity.61 We discovered a progressive increase in prefrontal inhibitory strength, most likely promoted by the emergence of mature physiological abilities of L2/3 PYRs and, subsequently, interneurons. These findings are in line with previous studies that identified a multifactorial cascade of developmental milestones, which most likely contribute to the strengthening of prefrontal inhibition.2,18,36,57,62 These inhibitory milestones might depend on proper microglia function during development, according to recent findings in the barrel cortex.63 Although inhibitory development seems to have similar dynamics in distinct cortical areas, the PFC appears to be the last to reach full maturity,1,64,65 with prefrontal L2/3 showing the slowest development.36,66 However, such progressive inhibitory development contrasts with the identified nonlinear dynamics of gamma band maturation. Our results identified L2/3 PYRs as potential mediators of nonlinear changes throughout adolescence. Their firing patterns and morphological development closely resembled changes in fast prefrontal oscillations.

The nonlinear dynamics of prefrontal development, which involve making, breaking, and rebuilding connectivity, might be surprising at first glance. However, a long-standing dogma of brain development is the activity-dependent pruning of inappropriate connections to enable the formation and fine-tuning of mature circuits.67 For the PFC, it has been suggested that an initial overproduction of synapses is followed by activity-dependent microglia-mediated pruning during adolescence, resulting in the strengthening of remaining connections.1,2,68 Moreover, pruning deficits during this time were proposed to precede the onset of mental disorders.69 However, experimental evidence for these hypotheses is sparse, and the corresponding studies were mainly conducted in different species, prefrontal subdivisions, and age groups.70,71,72,73,74,75 The present longitudinal monitoring of electrical activity from P16 to P60 provides a full characterization of the functional development of the mPFC and identifies the cellular elements and mechanisms that coordinate the reorganization of circuitry.

An intriguing question that remains is what triggers the observed microglia-dependent dynamics of making, breaking, and rebuilding in the mPFC. Several complementary mechanisms might underlie these processes. Active neurons release adenosine triphosphate, which recruits microglia for synapse remodeling and triggers the release of brain-derived neurotrophic factor (BDNF), which, in turn, has been shown to increase network excitability.40,76 Notably, frontal BDNF levels increase from P28 to P35 onward in mice and from adolescence to young adulthood in humans.7 Less active synapses are tagged by so-called “eat-me” factors and promote phagocytic microglia activity.40 Such eat-me factors have been found to be enriched in the adolescent mPFC compared with the pre-adolescent mPFC.13 The local specification of active vs. non-active synapses might result from long-term depression (LTD) mechanisms that develop during adolescence via the strengthening of afferent projections and the emergence of more focal inhibition. Although dopaminergic projections gradually innervate the developing mPFC, dopaminergic receptor (DR) densities peak during adolescence.7 Particularly, DRs of type 2 (DR2) are suggested to promote prefrontal LTD and have been found to regulate pruning in the ACC.77,78 For instance, DR2-deficient mice show hyper-glutamatergic function but a decreased likelihood of changes in inhibitory synapse development.78 Moreover, hippocampal-mediated LTD in the mPFC emerges only by late adolescence.21 Microglia, in turn, potentially foster LTD by trogocytosis, incomplete engulfment, or nibbling of synapses, leading to a weakening of synaptic strength.79 Thus, multiple and complex microglia-neuron interactions might underlie the functional fluctuations detected during adolescence.

Transient microglia ablation during adolescence, but not during the pre-juvenile or adult periods, extensively disrupted neuronal morphology as a result of microglia dysfunction, perturbed the prefrontal network activity, and impaired behavioral abilities. By contrast, similar manipulations in S1 had no effect on network activity. Sensory cortices seem to exhibit distinct time windows of sensitivity to microglia manipulation. Although PLX3397 application during the first 2 postnatal weeks, but not during later development, led to altered synaptic connectivity in S1, in the primary visual cortex (V1), these changes were only observed when continuing PLX3397 application until P25.43 Similarly, PLX3397 application from P14 to P28 increased EPSC frequency and spine density as well as caused the disruption of connectivity and critical period plasticity in V1.44 In line with these findings, we identified behavioral alterations in pre-juvenile PLX-treated mice. However, neither the prefrontal nor the somatosensory network activity was perturbed. Thus, observed differences might emerge from alterations in other brain areas, such as V1, which might become more apparent in visual tasks. In a recent study, microglia were ablated with bilateral clodronate injections into the mPFC during adolescence. Decreased dendritic complexity and synaptic development of PYRs, increased IPSCs onto PYRs, and disturbed behavioral abilities were reported 40 days post treatment.13 Despite slightly differing manipulation protocols (e.g., injection at P42, only male mice), the described functional, morphological, and behavioral deficits are largely similar to those reported in this study during adulthood (P98–P102). Moreover, we used mice of both sexes and tracked sex-dependent effects in our statistical model (see STAR Methods; Table S2). Particularly, age-dependent changes in microglia morphology and pruning were affected by sex. Considering that females reach sexual maturity before males80 and that steroid hormones are thought to drive synaptic pruning,81 time windows of vulnerability may differ between the sexes. However, fluctuating functional dynamics of prefrontal activity showed no sex dependence, suggesting only modest sex-dependent effects in the context of adolescent-typical behaviors.22,80

Since our previous investigations showed that L2/3 PYRs, but not L5/6 PYRs, are key players in the generation of fast oscillations and synchronization in the PL25 and are the cellular substrate of disorganized prefrontal activity in a model of mental disorders,12 this study focused on the upper prefrontal layers. Moreover, we previously showed that manipulating L2/3 PYR activity during early development (P7–P11) disturbs prefrontal processing and behavioral abilities during late development.11 However, it remains unknown whether the revealed processes are specific to the upper prefrontal layers or represent general phenomena in all layers of the frontal or limbic cortices. As predicted by the first hypothesis, L2/3 PYRs have been found to exhibit more excitatory than inhibitory spontaneous synaptic events until ∼P30, whereas L5/6 PYRs exhibit an equal amount of both.62 Laminar developmental differences might also account for observed L2/3-specific alterations in mental diseases, such as decreased spine densities and an increased number of differentially expressed genes in patients suffering from schizophrenia.82,83 Moreover, L2/3 PYRs have been identified as particularly vulnerable to microglia-mediated over-pruning12,56,84; however, it remains to be investigated if this sensitivity is restricted to frontal cortices. Recent data also revealed distinct laminar distributions of microglia states and densities in S1.85 Finally, even if microglia appear to be key mediators of prefrontal development, additional events interfering with mechanisms underlying the functional trajectories revealed here might result in the abnormal assembly of adolescent prefrontal circuits.14,57

Despite the structural differences between the mouse and human PFC and the species-characteristic cognitive complexity levels, it is likely that the nonlinear making, breaking, and rebuilding of prefrontal connectivity that we identified in mice are also present in humans during late development.1,86,87 Notably, anatomical post-mortem studies in humans have reported major reorganization of brain architecture during adolescence that is most extensive (i.e., temporal increase in gray matter volume, gyrification, and spine densities) in frontal regions.71,72,88,89 Identifying the dynamics of functional development and the mechanisms underlying the adolescent reorganization of connectivity is essential, not only to gain a better understanding of the PFC as a hub of cognitive processing but also to open new clinical research perspectives, especially in the context of targeting specific time periods of vulnerability as potential therapeutic avenues for mental diseases.

STAR★Methods

Key resources table

REAGENT or RESOURCE SOURCE IDENTIFIER
Antibodies

Alexa Fluor-488 goat anti-rabbit IgG secondary antibody Thermo Fisher Scientific Cat# A32731, RRID: AB_2633280
Rat monoclonal primary antibody against mouse CD68 Bio-Rad Cat# MCA1957GA, RRID: AB_324217
Rabbit polyclonal primary antibody against Iba1 FUJIFILM Cat# 019-19741, RRID: AB_839504
Alexa Fluor-555 goat anti-rabbit IgG secondary Thermo Fisher Scientific Cat# A32732, RRID: AB_2633281
Alexa Fluor-405 goat anti-rat IgG secondary antibody Abcam Cat# ab175671, RRID: AB_2890626

Bacterial and virus strains

AAV9-Ef1alpha-DIO-hChR2(ET/TC)-eYFP https://www.nature.com/articles/nmeth.1808 Addgene AAV9; 35507-AAV9
pCAG_PSD95.FingR-eGFP-CCR5TC https://www.sciencedirect.com/science/article/pii/S089662731300319X Addgene AAV9; 46295-AAV9

Chemicals, peptides, and recombinant proteins

Pexidartinib (PLX3397) Hölzel Diagnostika Cat# HY-16749
Clodronate disodium salt Millipore Cas# 88416-50-6

Deposited data

Extracellular recording data and sorted single unit data This paper https://doi.org/10.12751/g-node.sq1cxy
Custom analysis code This paper https://doi.org/10.12751/g-node.sq1cxy

Experimental models: Organisms/strains

Mouse: C57Bl/6J Universitätsklinikum Hamburg-Eppendorf– Animal facility Mouse: C57Bl/6J
Mouse: Dlx5/6-Cre (Tg(dlx5a-cre)1Mekk/J Jackson Laboratory RRID: IMSR_JAX:008199

Recombinant DNA

pAAV-CAG- 2AtDimer2 Donated by the Oertner lab N/A

Software and algorithms

Matlab R2019a, R2021a MathWorks RRID:SCR_001622
ImageJ 2.0.0 ImageJ RRID:SCR_003070
R Statistical Software 3.5.1 R Project for Statistical Computing RRID:SCR_001905
Cheetah 6 Neuralynx http://neuralynx.com/
Anaconda 1.9.6 Conda RRID:SCR_018317
Spyder 5.1.5 Spyder RRID:SCR_017585
Jupyter 6.0.0 Jupyter Notebook RRID:SCR_018315
Video Mot2 TSE Systems https://www.tsesystems.com/productdetails/videomot
uEye Cockpit iDS Imaging https://en.ids-imaging.com/ids-software-suite.html
phy Cortex lab https://github.com/cortex-lab/phy
klusta Cortex lab https://github.com/kwikteam /klusta
ezTrack Cai lab https://github.com/denisecailab/ezTrack
Custom analysis code This paper https://doi.org/10.12751/g-node.sq1cxy

Other

Arduino Uno SMD Arduino A000073
Digital Lynx 4SX Neuralynx http://neuralynx.com/
Diode laser (473 nm) Omicron LuxX® 473-100
Electroporation device BEX CUY21EX
Electroporation tweezer-type paddles Protech CUY650-P5
Recording optrode (1 shank, 16 channels) Neuronexus A1x16-5mm-100-703-OA16LP
Recording electrode (1 shank, 16 channels) Neuronexus A1x16-5mm-50-703 LP
Recording electrode (1 shank, 16 channels) Neuronexus A1x16-5mm-100-703 LP
Recording electrode (1 shank, 16 channels) Neuronexus CM1x16-3mm-100-703 LP
Head-stage Neuralynx HS-16-N2T-CNR-MDR50
Confocal microscope Zeiss LSM 700
Mobile HomeCage® Neurotar Mobile HomeCage Large® with tracking capability
Floating foam maze Neurotar Assembled foam sociability cage
Lightweight stainless-steel head plates Neurotar Model 3, round 6,2 mm opening
Headpost stand Luigs & Neumann 200-200 500 1100
Head post - S (for mice) Luigs & Neumann 200-100 500 2203

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Ileana L. Hanganu-Opatz (hangop@zmnh.uni-hamburg.de)

Materials availability

The study did not generate new unique reagents.

Data and code availability

  • •

    LFP and SUA data for all chronic recordings are uploaded to an open-access repository and are publicly available. The DOI is listed in the key resources table. Microscopy images and behavioral videos reported in this paper will be shared by the lead contact upon request.

  • •

    All original or customized code is deposited to an open-access repository and is publicly available. The DOI is listed in the key resources table.

  • •

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

Experimental model and study participant details

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 (N18/015, N19/121). Experiments were carried out on C57BL/6J and Dlx5/6-Cre (Tg(dlx5a-cre)1Mekk/J, RRID: IMSR_JAX:008199, Jackson Laboratory) mice of both sexes. Pregnant dams were housed individual and weaned offspring was housed with at least two cage-mates on a 12 h light / 12 h dark cycle. The humidity and temperature were kept constant (40% relative humidity; 22°C). Food and water were available ad libitum for mice that were not treated with PLX3397. The day of vaginal plug detection was considered E0.5, the day of birth was considered P0. Experiments were performed on P5 to P100 mice. The effect of sex on the data distribution was estimated by linear-mixed effect models with sex as a random parameter and can be found in Table S2.

Method details

Microglia depletion diet

PLX3397 (Pexidartinib, Cat# HY-16749, Hölzel Diagnostika, Germany) diet was administered for 10 days to mice either between P15 and P24, P28 and P37, or P70 and P79. In line with previous studies45,46 a concentration of 290 ppm (290 mg/kg) wet food was applied daily / mouse. Control mice were fed with wet food containing no PLX3397. No differences in weight, liquid, or food uptake between treated and control mice were observed.

In utero electroporation

Pregnant mice 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 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 2.5%) on a heating blanket. Eyes were covered with eye ointment (Vidisic, Bausch + Lomb, Berlin, Germany) and pain reflexes and breathing were monitored to assess anesthesia depth. Uterine horns were exposed and moistened with warm sterile phosphate buffered saline (PBS). Solution containing 1.25 μg/μl of fluorophore-encoding plasmid DNA (pAAV-CAG-2A-tDimer2 (Oertner lab, Hamburg, Germany) or pCAG_PSD95.FingR-eGFP-CCR5TC (Plasmid #46295, Addgene, MA, USA)) in sterile PBS with 0.1% fast green dye was injected at a volume of 0.75–1.25 μl 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 L2/3 PYRs with 5 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 (Electron microscopy sciences, Hatfield, USA) through the intact skin and skull and confirmed in brain slices postmortem before the animals were used for further analysis.

Virus injections

At P5 or P43, 100 nl virus (titer, ≥ 1×1013 vg/mL) of Cre-dependent AAV9-Ef1alpha-DIO-hChR2(ET/TC)-eYFP (#35507, Addgene, MA, USA) was injected into the mPFC (0.5 mm anterior to bregma, 0.1-0.5 mm lateral to the midline, 1.6 mm deep) of Dlx5/6-Cre+ and Cre- mice under isoflurane anesthesia (induction 5%, maintenance 2.5%). For this, mice were placed on a heating blanket. A NanoFil needle and syringe (NANOFIL, NF35BV, World Precision Instruments, USA) connected to a micropump (Micro4, WPI, Sarasota, FL, USA) were used to inject the viral construct at a rate of 50 nl/min. Following injection, the syringe was left in place for at least 3 min to avoid reflux of the virus fluid. Incisions were closed with a surgical glue (Surgibond, SMI, Belgium) and disinfected with Betaisodona. Pups were maintained on the heating blanket until full recovery and returned to their dam. The first recordings combined with optogenetic stimulation were performed not earlier than 10 days after injection, to enable reliable viral expression.

In vivo Electrophysiology

Chronic extracellular head-fixed recordings

Multisite extracellular head-fixed recordings were performed unilaterally in the mPFC or S1 of P16–60 and P98–102 mice. The adapter for head fixation was implanted at least 2 days before recordings. Buprenorphine (0.05 mg/kg body weight) was subcutaneously injected 30 min before surgery. Mice were anesthetized with isoflurane anesthesia (5% induction, 2.5% maintenance) and placed on a heating blanket to maintain the body temperature. Eyes were covered with eye ointment (Vidisic, Bausch + Lomb, Berlin, Germany) to prevent drying. Prior to surgery, the depth of anesthesia and analgesia was evaluated with a toe pinch to test the paw withdrawal reflex. The skin of the head was disinfected with Betaisodona. After removing the skin, 0.5% bupivacaine/1% lidocaine was locally applied to cutting edges. A metal head-post (Luigs and Neumann, Germany or Neurotar, Helsinki, Finland) 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 S1 (1.0–1.5 mm posterior to bregma, 2.5–3.0 mm right to the midline) and protected by a customized synthetic window filled with Kwik-Cast sealant (World Precision Instruments, Friedberg, Germany). A silver wire was implanted between skull and brain tissue above the cerebellum and served as ground and reference. Mice were placed in a home cage on a heating blanket and returned to their cage-mates after full recovery from anesthesia. Metacam (0.5 mg/ml, Boehringer-Ingelheim, Germany) mixed into soft food was provided for 2 days after surgery. After recovery from the surgery, mice were accustomed to head-fixation and trained to move on a custom-made spinning disc or in the Mobile HomeCage recording system (Neurotar, Helsinki, Finland). For recordings, craniotomies were uncovered and multi-site electrodes (NeuroNexus, MI, USA) were stereotaxically inserted into the mPFC (one-shank, A1x16 recording sites, 100 μm spacing, 2.0 mm deep) and S1 (one-shank, A1x16 recording sites, 50 μm spacing, 1.0 mm deep). Recordings were acquired for 30-60 min. 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). After recordings, the craniotomy was closed, and mice were returned to their home cage. Care was taken that mice were not kept away from their cage-mates for more than 1.5 h at a time. The same mice were recorded up to 18 times. Acutely recorded P53–60 mice were recorded for 1–2 times. Electrode position was confirmed in brain slices postmortem. For LFP analysis of prefrontal activity a recording channel within the PL and for LFP analysis of primary somatosensory activity a recording channel within L2/3 was selected.

Chronic extracellular freely moving recordings

Multisite extracellular freely moving recordings were performed unilaterally in the mPFC of P16–200 mice. Buprenorphine (0.05 mg/kg body weight) was subcutaneously injected 30 min before surgery. Mice were anesthetized with isoflurane anesthesia (5% induction, 2.5% maintenance) and placed on a heating blanket to maintain the body temperature. Eyes were covered with eye ointment (Vidisic, Bausch + Lomb, Berlin, Germany) to prevent drying. Prior to surgery, the depth of anesthesia and analgesia was evaluated with a toe pinch to test the paw withdrawal reflex. The skin of the head was disinfected with Betaisodona. After removing the skin, 0.5% bupivacaine/1% lidocaine was locally applied to cutting edges. 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 a multi-site electrode (NeuroNexus, MI, USA) was stereotaxically inserted into the mPFC (one-shank, CM1x16 recording sites, 100 μm spacing, 2.0 mm deep) with a motorized manipulator (Scientifica, NJ, USA) at a speed of 5 μm/min. Reference and ground wires were implanted between skull and brain tissue above the cerebellum. Mice were placed in a home cage on a heating blanket and returned to their cage-mates after full recovery from anesthesia. Metacam (0.5 mg/ml, Boehringer-Ingelheim, Germany) mixed into soft food was provided for 2 days after surgery. After recovery from the surgery, mice were connected to the recording head-stage within a squared open field arena (50x50 cm). Recordings were acquired for ∼ 30 min while the behavior of the mice was video-tracked. 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). After recordings, the mice were disconnected from the head-stage and returned to their cage-mates. The same mice were recorded up to 37 times. Electrode position was confirmed in brain slices postmortem.

Acute optogenetic stimulation

Ramp (3 s linearly increasing light power) and pulsed (short pulses of 3 ms) light stimulation was performed using an Arduino uno (Arduino, Italy) controlled laser system (473 nm wavelength, Omicron, Austria) coupled to a 105 μm diameter light fiber (Thorlabs, NJ, USA) glued to the multisite electrodes, ending 200 μm above the top recording site. Laser power was adjusted to reliably induce neuronal firing and set between 2mW-5mW/mm2 at fiber tip. Responses to stimulations were averaged over 30–60 repetitions. Dlx5/6-Cre+ and Cre- mice underwent the same stimulation protocol. For analysis of Cre- mice recordings from mice of all ages were pooled.

Histology

Perfusion and brain slicing

Non-recorded mice of 4 age groups (PreJuv P16-18, EarlyAdo P33–35, LateAdo P41–43, and Adult P58–60) as well as P60 and P100 adolescent PLX-treated and control mice were anesthetized with 10% ketamine (aniMedica, Germany) / 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 h. Brains were sectioned coronally with a vibratome at 50 μm for immunohistochemistry and at 100 μm for morphological investigations of tDimer2-expressing L2/3 PYRs, respectively. A similar perfusion protocol was used for chronically recorded mice and 100 μm-thick coronal brains sections were sectioned with a vibratome for post-mortem electrode reconstruction.

Immunohistochemistry

Free-floating slices were washed 3 times for 15 min with PBS. For the second and third washing step PBS was supplemented with 0,03% hydrogen peroxide (Sigma-Aldrich, Darmstadt, Germany). Washed slices were permeabilized and blocked with PBS containing 0.5% Triton X-100 (Sigma-Aldrich, Darmstadt, Germany) and 5% normal goat serum (Cell Signaling Technology, Leiden, Netherlands) for 1 h. Slices were incubated over night with primary antibody rabbit-anit-Iba1(1:500, Cat# 019-19741, RRID: AB_839504, FUJIFILM, Neuss, Germany) and rat-anti-mouse CD68 (1:500, Cat# MCA1957GA, RRID: AB_324217, Bio-Rad, Feldkirchen, Germany) and afterwards washed for 3 times with PBS containing 0.2% Trition X-100. This was followed by 2 h of incubation with secondary antibody goat-anti-rabbit Alexa Fluor 488 (1:1000, Cat# A32731, RRID: AB_2633280, Thermo Fisher Scientific, MA, USA), goat-anti-rat Alexa Fluor 405 (1:1000, Cat# ab175671, RRID: AB_2890626, Abcam, Cambridge, UK) and/or goat-anti-rabbit Alexa Fluor 555 (1:1000, Cat# A32732, RRID: AB_2633281, Thermo Fisher Scientific, MA, USA) in PBS containing 0.5% Triton X-100 and 1% normal goat serum. After incubation, sections were 3 times washed for 15 min with PBS. Sections were transferred to glass slides and covered with Vectashield mounting medium (LINARIS, Dossenheim, Germany).

Dendritic complexity and spine quantification

Image stacks of tDimer2-positive PYRs were acquired on a confocal microscope (LSN700, Zeiss, Germany) using a 40x objective with oil. Stacks of 5 to 7 neurons per animal were acquired as 2048x2048 pixel images with an interval of 0.5 μm, resolution of 7.69 pixels/μm, and scan speed of 6. Dendritic complexity was quantified by Sholl analysis in ImageJ (RRID:SCR_003070). Stacks were binarized using auto threshold function and the dendrites were traced along each image of the stack using the semi-automatic Simple Neurite Tracer plugin. Only connected pixels were assumed to belong to the targeted neuron. The geometric center was identified, and the traced dendritic tree was analyzed with an interval of 5 μm with the Sholl analysis plugin. Example photographs, shown in the figures, were generated by considering the maximum intensity Z-projection of single stacks. For spine quantification, dendrite of interest (Distal dendrites: Apical, proximal oblique or secondary apical; Basal dendrites: Primary basal or secondary basal) was manually measured and spines along the measured compartment were counted (Straight Line, Segmented Line Tool, Point Picker tool).

Microglia quantification and morphological analysis

Image stacks of Iba1-positive cells were acquired on a confocal microscope (LSN700, Zeiss, Germany) using a 20x objective. Stacks of 2 regions (ventral, dorsal) within the mPFC of both hemispheres of 4 to 8 slices per animal were acquired as 512x512 pixel images with an interval of 0.75 μm, resolution of 1.60 pixels/μm, and scan speed of 9. The morphology and density of microglial cells was assessed on thresholded, maximum intensity Z-projections performed in ImageJ environment and imported in the Matlab (RRID:SCR_001622, MathWorks, MA, USA) environment, using previously reported criteria.12,90 Briefly, images were automatically thresholded (graythresh and im2bw functions) and putative microglia were identified as objects between 200 and 1500 pixels (bwareaopen). Around the center of mass of each of the isolated cells, a region of interest (ROI) was computed and visually examined. If the ROI contained a properly segmented, single microglia cell, its features (area, perimeter) were quantified (regionprops). ROIs in which the microglia touched the boundaries of the image or in which more than one cell was included were discarded. Further, cell spread (analogous to process length) was computed as the average distance between the center of mass and the “extrema” of the cell; roundness was defined as the ratio between 4∗pi∗area and the square of the perimeter of the cell. The number of identified objects, that were visually confirmed as microglia cells, was used for cell density quantifications.

Microglia pruning quantification of excitatory postsynaptic compartments

Image stacks of Iba1-positive cells, PSD95 positive puncta of IUE targeted L2/3 PYRs, and CD68 stained lysosomes were acquired on a confocal microscope (LSN700, Zeiss, Germany) using a 40x objective with oil. Stacks of 2 regions (ventral, dorsal) within the mPFC of both hemispheres of 6–10 slices per animal were acquired as 1024x1024 pixel images with an interval of 0.5 μm, resolution of 12.82 pixels/μm, and scan speed of 4. Overlays of Iba1-positive cells and PSD95 positive puncta, as well as CD68 positive lysosomal pixel were assessed using custom-written Matlab routines that were designed according to previously reported criteria.12,90 Briefly, the background noise of PSD95 and CD68 stacks was reduced in the ImageJ environment using the subtract background (rolling ball radius, 2 pixels) and despeckle functions. Stacks were passed through a maximum filter (radius, 2 pixels), thresholded (auto threshold) and segmented (watershed). Further processing was carried out in the Matlab environment. Puncta were labeled (bwlabeln, connectivity = 8) and their volume was measured (histcounts). Microglia stacks were entirely processed in the Matlab environment. Stacks were passed through a 3D hysteresis filter (hysteresis3d function; lower threshold = 0.1, upper threshold = 0.5, connectivity = 26) and a 3D median filter (ordfilt3D function). Puncta of PSD95 and CD68 stacks showing 100% overlap in xyz direction with each other and the processed Iba1 signal were then quantified and normalized by total Iba1 volume of each analyzed cell. To generate representative overview images, stacks of processed Iba1, PSD95, and CD68 signals were z-projected and merged. Phagocytic ratio was quantified as (number connected PSD95 pixel - number connected CD68 pixel) / (number connected PSD95 pixel + number connected CD68 pixel). Microglia with more than 6 overlaying puncta were considered as phagocytic active. The customized Matlab code including all subfunctions has been uploaded to the repository to code/morphology.

Behavior

Quantification of grooming and open field behavior

Mice were handled and adapted to the behavioral investigation room two days prior to behavioral examination. Mice were positioned in the center of a circular arena (34 cm in diameter) and allowed to explore for 10 min. The arena was cleaned with 0.1% acetic acid between mice. Grooming was tracked online using the video Mot2 software (Video Mot2, TSE Systems GmbH, Germany). Animal’s position was tracked offline using the Python-based tracking system ezTrack86.

Spontaneous alteration

Each mouse was positioned in the start arm of an elevated Y-maze. Visited arms during free exploration were monitored for 10 min. The Y-maze was cleaned with 0.1% acetic acid between mice. Percentage of alternations was calculated as number of alternations / (entries - 2).

Attentional set-shifting

The four-choice odor discrimination and reversal task was conducted as previously described.91 During the habituation session, four dishes (6 cm in diameter) with wooden shavings and a piece of Cocoa Krispies were placed in the rectangular maze with 4 compartments (30x30 cm). During pre-training, one dish with increasing amounts of shavings covering the cereal reward was put in alternating quadrants. For testing the four-choice odor discrimination and reversal, the wooden shavings were scented by mixing them with dried spices (FUCHS Gewürze GmbH) (garlic powder, black pepper, rosemary, thyme - 0.1% by volume and clove - 0.05% by volume). During the initial phase of the test, the animal had to discriminate among four odors and learn which one is associated with the buried food reward. Each trial was terminated as soon as the mouse made the choice by digging with front paws or nose into the dish and, in case of a correct choice, the food reward was eaten. Once criterion of 8 out of 10 consecutive correct trials was met in the discrimination phase, the mouse was moved to the reversal phase immediately. One previously unrewarded odor became the newly rewarded odor, and the other one was switched for a novel odor. To complete the reversal, the mouse had to reach criterion by completing 8 out of 10 consecutive trials correctly.

Spatial working memory

At P58–60, adolescent PLX treated and control mice were positioned in the center of an elevated eight-arm radial maze. 4 arms contained a food pellet at the distal end (baited). On the first day, mice were allowed to examine the maze for 20 min or until all arms were visited. During the following 12 trials (4 trials daily), mice were allowed to examine the maze until all baited arms were visited (for max 20 min) and arm entries were assessed. Visit of a non-baited arm was considered as reference memory error, repeated visit of the same arm in one trial as working memory error. Throughout behavioral testing, mice had restricted access to food. The amount of food was set to preclude weight loss. Animals were tracked offline using the python-based tracking system ezTrack92 (https://github.com/denisecailab/ezTrack). To quantify the behavioral performance across trials for individual mice, single values across trials were fitted by a robust linear regression model and the resulting slope was normalized by its intercept (slope / intercept2).

Quantification and statistical analysis

Electrophysiological data analysis

LFP analysis

In vivo data were analyzed with custom-written algorithms in Matlab environment. 100 randomly selected10 s-long signal segments were extracted from baseline recordings with a typical duration of 30–45 min and band-pass filtered (0.5–100 Hz) using a third-order Butterworth filter forward and backward to preserve phase information before down-sampling to analyze the LFP.

Power spectral density

For power spectral density analysis, the 10 s-long, filtered segments of LFP signal were concatenated and the power was calculated for 1 s-long segments using Welch’s method with non-overlapping windows and followed by median averaging for resting and running periods separately. Spectra were multiplied with squared frequency. Time-frequency plots of power were calculated with a continuous wavelet transformation (Morlet wavelet). Slow-fast frequency MI was calculated as the spectral density power ratio of frequencies 1–12 Hz and 12–100 Hz according to following formula: (power 1–12 Hz - power 12–100 Hz) / (power 1–12 Hz + power 12–100 Hz). The function getMI has been uploaded to the repository to code/LFP.

Optogenetic modulation index

For optogenetic stimulations, OMI was calculated with the getMI function as (power value stimulation or post - power value pre stimulation) / (power value stimulation or post + power value pre stimulation) for spectral densities of 2 s-long, centered segments before, during, or post ramped light stimulation. Slope and intercept of OMI spectra were quantified using the robustfit function in Matlab.

Peak frequency and amplitude

Squared frequency multiplied spectra from either 1–12 Hz or 12–100 Hz were smoothed and peaks with a minimum width of 5 Hz were detected. The most prominent peak in the spectra was defined by the product of peak amplitude, peak half width, and peak prominence (i.e., how much the peak stands out due to its intrinsic height and its location relative to other peaks). The peak frequency corresponded to the resulting x-axis value and the peak amplitude to the average value of the y-axis between +/- 1 Hz of the peak frequency. Peak detection was carried out on the powerspectrum of each recording with similar parameters across age.

Phase-amplitude coupling

PAC between mPFC theta phase and gamma amplitude was calculated on filtered 10 s-long segments using the PAC toolbox based on the normalized modulation index measure.93 Range of phase vector was set to 1 to 12 Hz, and range of amplitude vector was set to 20–100 Hz. Significant coupling was calculated in comparison to a shuffled dataset. Non-significant values were rejected. The customized Matlab code including all subfunctions has been uploaded to the repository to code/LFP/PAC.

Generalized partial directed coherence

gPDC represents a frequency-domain representation based on the concept of Granger causality and was quantified on 1s-long LFP segments. The segments were denoised using the Matlab wavelet toolbox and further processed according to a previously described algorithm to quantify gPDC.47 The customized Matlab code including all subfunctions has been uploaded to the repository to code/LFP/gPDC.

Single unit analysis

Spikes were detected with klusta94 (https://github.com/kwikteam/klusta). Automatically-obtained clusters were then manually curated using phy (https://github.com/cortex-lab/phy). Data were imported and analyzed using custom-written routines in Matlab. Firing rate, CV, and ISI were calculated on spike-times during baseline periods. For CV and ISI calculation, only intervals smaller than 1 s were considered. The threshold for significantly modulated units during optogenetic experiments were set to p < 0.001 with a MI > 0 corresponding to activated and MI < 0 to inhibited units. For ramped light stimulations, the significance was calculated by comparing the firing rate 3 s pre stimulation onset vs. 3 s during stimulation for all stimulation trials. For 3 ms pulsed light stimulations, a time window of 10 ms post and pre stimulation onset was considered. To identify photo-tagged units after pulsed light stimulation a shorter time window of 5 ms post and pre stimulation onset was considered and the threshold for MI was set to > 0.2.

Classification of RS and FS units

RS and FS units were distinguished by setting a threshold based on spike halfwidth and trough to peak latency (halfwidth < 0.31 ms, trough to peak < 0.64 ms). We previously found by using PV-Cre mice that this threshold mainly identifies PV+ interneurons in L2/3 of the mPFC.11

Spike-time tiling coefficient

TC, a firing rate-independent measure, was calculated according to a previously published algorythm.32 Briefly, the spike times of two spike trains are correlated. For this, the proportion of spikes from spike train A that fall within ± Δt of a spike from spike train B within 3 or 100 ms is quantified, respectively. The resulting value is subtracted by the proportion of time that occurs within ± Δt of spikes from spike train B. This is then divided by 1 minus the product of these 2 values. The same is then applied after inverting spike train A and B, and the mean between the 2 values is kept which ranges between -1 and 1. The function getTC has been uploaded to the repository to code/SUA.

Spike-LFP coupling

Phase locking of spiking units to network oscillations was assessed using the firing rate-independent PPC measure according to a previously published algorythm95 after identifying significantly locked single units. Briefly, the LFP was filtered for theta (4–10 Hz) and gamma (30–100 Hz) using a third-order Butterworth filter. The instantaneous phase was extracted using the Hilbert transform on the filtered signal. The coupling between spikes and network oscillations was tested for significance using the Rayleigh test for non-uniformity. Only units that showed a significant phase locking to theta or gamma phases were used for PPC analysis. The function computePPC has been uploaded to the repository to code/SUA.

Spike-spike cross-covariance

Cross-covariance of spike trains was calculated as described previously.96 Only pairs of units with firing rates > 0.05 Hz and significant standardized cross variance were considered (two-tailed critical z value at level α = 0.01). The mean cross-covariance for one driver unit was calculated by averaging all significant cross-covariance spectra and normalizing it by the number of spectra.

Joint entropy

The joint entropy between two spike trains was calculated as z=H(x,y). The function getJointEntropy has been uploaded to the repository to code/SUA.

Statistical analysis

The statistical details for each figure panel, including used statistical test, exact n number, representation of n number (e.g., number of animals, number of cells, etc.), and measures to estimate effect size can be found in Table S1. The used method for visualizing the average (e.g., median, SEM, etc.) in each figure panel can be found in the figure legend.

Statistical analyses were performed in Matlab, R Statistical Software (RRID:SCR_001905, R Project for Statistical Computing, Austria), and Python (RRID:SCR_018317, Python Software Foundation, Wolfeboro Falls, New Hampshire, USA) by using the Spyder (RRID:SCR_017585) development environment. Non-nested data were tested for significant differences (∗p < 0.05, ∗∗p < 0.01 and ∗∗∗p < 0.001) using non-parametric Wilcoxon rank sum test for unpaired and Wilcoxon signed rank test for paired data or Kruskal-Wallis test with Bonferroni corrected post hoc analysis in Matlab. Circular data were tested for significant difference using the circ_cmtest function in Matlab, a non-parametric multi-sample test for equal medians and similar to Kruskal-Wallis test for linear data. Nested data were analyzed with linear-mixed-effect models. Parameter estimation was done using the lmer function implemented in the lme4 R package. To test the significance of age or condition in the model, a likelihood ratio test against a reduced model was performed in which the age or condition was removed (lmerTest R package). Post hoc analysis with Tukey’s multiple comparison correction was carried out using the glhp function of the multcomp R package. Depending on the experiment, random effects included animal, recording number, sex, unit-pair distance, slice, cell / neuron, or / and prefrontal subregion (see Table S1 for detailed statistics). For statistical analysis of sex differences, sex as random effect was removed from the model to investigate its statistical impact on age or condition (see Table S2 for detailed statistics of sex effects).

To explore the nonlinear patterns present in the data across investigated ages, we used the restricted (natural) cubic splines, as proposed by Stone and Koo,97 and compared the results with those obtained from linear regression. The restricted version of cubic splines is constrained to be linear in the tails, handling potential irregular behavior of cubic spline functions before the first knot and after the last knot. In the spline fit, the locations of knots were predetermined and not considered as free parameters for estimation, meaning that the model did not infer the knot positions. Following the recommendation by Harrell,98 we placed the knots at fixed equally spaced percentiles of a predictor’s marginal distribution (in our case, age distribution). For many datasets, using four knots offers an adequate fit of the model, achieving a desirable trade-off between flexibility and overfitting. For larger datasets (n ≥ 100), five knots are sometimes preferrable.98 Following these suggestions, we implemented restricted cubic spline models with both four and five knots, and also extended the analysis to include fits with three and six knots. For every dataset, four restricted cubic spline models with a varying number of knots (k, 3, 4, 5, or 6) were fit. The knots were placed in the following percentiles: k = 3, percentiles = [0.1, 0.5, 0.9]; k = 4, percentiles = [0.05, 0.35, 0.65, 0.95]; k = 5, percentiles = [0.05, 0.275, 0.5, 0.725, 0.95]; k = 6, percentiles = [0.05, 0.23, 0.41, 0.59, 0.77, 0.95]. Apart from the nonlinear model, we conducted linear regression for each dataset. The evaluation and comparison of all fitted models were based on calculations of AIC (Akaike’s information criterion), BIC (Bayesian information criterion), RMSE, and R2. Using the model with the best fit, we computed the local minima and maxima, which served to determine the turning points in investigated dynamics. The modified version of the published code has been uploaded to the repository to code/statistics.

No statistical measures were used to estimate sample size, since the effect size was unknown. Expect for Sholl analysis and spine density quantifications of investigated age groups, investigators were blinded to the group allocation or unsupervised analysis software was used to preclude investigator biases.

Acknowledgments

We thank A. Dahlmann and P. Putthoff for excellent technical assistance and M. Hnida and A. Offermanns for helpful discussions and comments on the manuscript.

Funding: German Research Foundation Ha4466/11-1 (302153259) to I.L.H.-O.; German Research Foundation SFB 936 B5 (178316478) to I.L.H.-O.; German Research Foundation FOR5159 TP1 (437610067) to I.L.H.-O.; European Research Council ERC-2015-CoG (681577) to I.L.H.-O.; Horizon 2020 DEEPER (101016787) to I.L.H.-O.; Landesforschungsförderung Hamburg LFF73 and LFF76 to I.L.H.-O.

Author contributions

Conceptualization, J.A.P. and I.L.H.-O.; methodology, J.A.P., T.S., I.P., and I.L.H.-O.; investigation, J.A.P., T.S., A.G., M.D., I.P., and A.M.; visualization, J.A.P. and T.S.; writing – original draft, J.A.P. and I.L.H.-O.; writing – review & editing, J.A.P., T.S., A.G., M.D., I.P., and I.L.H.-O.; supervision, J.A.P. and I.L.H.-O.

Declaration of interests

The authors declare no competing interests.

Inclusion and diversity

We support inclusive, diverse, and equitable conduct of research.

Published: November 17, 2023

Footnotes

Supplemental information can be found online at https://doi.org/10.1016/j.neuron.2023.10.024.

Contributor Information

Jastyn A. Pöpplau, Email: jastyn.poepplau@zmnh.uni-hamburg.de.

Ileana L. Hanganu-Opatz, Email: hangop@zmnh.uni-hamburg.de.

Supplemental information

Document S1. Figures S1–S8
mmc1.pdf (2.4MB, pdf)
Table S1. Detailed statistical results, related to STAR Methods
mmc2.xlsx (96.7KB, xlsx)
Table S2. Detailed statistical results for the effect of sex, related to STAR Methods
mmc3.xlsx (35.4KB, xlsx)
Table S3. Recording summary of chronically recorded mice, related to STAR Methods
mmc4.xlsx (20.9KB, xlsx)
Document S2. Article plus supplemental information
mmc5.pdf (13MB, pdf)

References

  • 1.Chini M., Hanganu-Opatz I.L. Prefrontal cortex development in health and disease: lessons from rodents and humans. Trends Neurosci. 2021;44:227–240. doi: 10.1016/j.tins.2020.10.017. [DOI] [PubMed] [Google Scholar]
  • 2.Klune C.B., Jin B., DeNardo L.A. Linking mPFC circuit maturation to the developmental regulation of emotional memory and cognitive flexibility. eLife. 2021;10 doi: 10.7554/eLife.64567. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Miller E.K., Cohen J.D. An integrative theory of prefrontal cortex function. Annu. Rev. Neurosci. 2001;24:167–202. doi: 10.1146/annurev.neuro.24.1.167. [DOI] [PubMed] [Google Scholar]
  • 4.Koechlin E., Ody C., Kouneiher F. The architecture of cognitive control in the human prefrontal cortex. Science. 2003;302:1181–1185. doi: 10.1126/science.1088545. [DOI] [PubMed] [Google Scholar]
  • 5.Spencer-Smith M., Anderson V. Healthy and abnormal development of the prefrontal cortex. Dev. Neurorehabil. 2009;12:279–297. doi: 10.3109/17518420903090701. [DOI] [PubMed] [Google Scholar]
  • 6.Hensch T.K. Critical period plasticity in local cortical circuits. Nat. Rev. Neurosci. 2005;6:877–888. doi: 10.1038/nrn1787. [DOI] [PubMed] [Google Scholar]
  • 7.Larsen B., Luna B. Adolescence as a neurobiological critical period for the development of higher-order cognition. Neurosci. Biobehav. Rev. 2018;94:179–195. doi: 10.1016/j.neubiorev.2018.09.005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Reh R.K., Dias B.G., Nelson C.A., 3rd, Kaufer D., Werker J.F., Kolb B., Levine J.D., Hensch T.K. Critical period regulation across multiple timescales. Proc. Natl. Acad. Sci. USA. 2020;117:23242–23251. doi: 10.1073/pnas.1820836117. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Pinto J.G., Jones D.G., Murphy K.M. Comparing development of synaptic proteins in rat visual, somatosensory, and frontal cortex. Front. Neural Circuits. 2013;7:97. doi: 10.3389/fncir.2013.00097. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Kolb B., Mychasiuk R., Muhammad A., Li Y., Frost D.O., Gibb R. Experience and the developing prefrontal cortex. Proc. Natl. Acad. Sci. USA. 2012;109(Suppl 2):17186–17193. doi: 10.1073/pnas.1121251109. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Bitzenhofer S.H., Pöpplau J.A., Chini M., Marquardt A., Hanganu-Opatz I.L. A transient developmental increase in prefrontal activity alters network maturation and causes cognitive dysfunction in adult mice. Neuron. 2021;109:1350–1364.e6. doi: 10.1016/j.neuron.2021.02.011. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Chini M., Pöpplau J.A., Lindemann C., Carol-Perdiguer L., Hnida M., Oberländer V., Xu X., Ahlbeck J., Bitzenhofer S.H., Mulert C., et al. Resolving and rescuing developmental miswiring in a mouse model of cognitive impairment. Neuron. 2020;105:60–74.e7. doi: 10.1016/j.neuron.2019.09.042. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Schalbetter S.M., von Arx A.S., Cruz-Ochoa N., Dawson K., Ivanov A., Mueller F.S., Lin H.Y., Amport R., Mildenberger W., Mattei D., et al. Adolescence is a sensitive period for prefrontal microglia to act on cognitive development. Sci. Adv. 2022;8:eabi6672. doi: 10.1126/sciadv.abi6672. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Benoit L.J., Holt E.S., Posani L., Fusi S., Harris A.Z., Canetta S., Kellendonk C. Adolescent thalamic inhibition leads to long-lasting impairments in prefrontal cortex function. Nat. Neurosci. 2022;25:714–725. doi: 10.1038/s41593-022-01072-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Nabel E.M., Garkun Y., Koike H., Sadahiro M., Liang A., Norman K.J., Taccheri G., Demars M.P., Im S., Caro K., et al. Adolescent frontal top-down neurons receive heightened local drive to establish adult attentional behavior in mice. Nat. Commun. 2020;11:3983. doi: 10.1038/s41467-020-17787-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Solmi M., Radua J., Olivola M., Croce E., Soardo L., Salazar de Pablo G., Il Shin J., Kirkbride J.B., Jones P., Kim J.H., et al. Age at onset of mental disorders worldwide: large-scale meta-analysis of 192 epidemiological studies. Mol. Psychiatry. 2022;27:281–295. doi: 10.1038/s41380-021-01161-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Spear L.P. The adolescent brain and age-related behavioral manifestations. Neurosci. Biobehav. Rev. 2000;24:417–463. doi: 10.1016/s0149-7634(00)00014-2. [DOI] [PubMed] [Google Scholar]
  • 18.Caballero A., Tseng K.Y. GABAergic function as a limiting factor for prefrontal maturation during adolescence. Trends Neurosci. 2016;39:441–448. doi: 10.1016/j.tins.2016.04.010. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Sakurai T., Gamo N.J. Cognitive functions associated with developing prefrontal cortex during adolescence and developmental neuropsychiatric disorders. Neurobiol. Dis. 2019;131:104322. doi: 10.1016/j.nbd.2018.11.007. [DOI] [PubMed] [Google Scholar]
  • 20.Bhattacherjee A., Djekidel M.N., Chen R., Chen W., Tuesta L.M., Zhang Y. Cell type-specific transcriptional programs in mouse prefrontal cortex during adolescence and addiction. Nat. Commun. 2019;10:4169. doi: 10.1038/s41467-019-12054-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Zimmermann K.S., Richardson R., Baker K.D. Maturational changes in prefrontal and amygdala circuits in adolescence: implications for understanding fear inhibition during a vulnerable period of development. Brain Sci. 2019;9 doi: 10.3390/brainsci9030065. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Varlinskaya E.I., Vetter-O'Hagen C.S., Spear L.P. Puberty and gonadal hormones: role in adolescent-typical behavioral alterations. Horm. Behav. 2013;64:343–349. doi: 10.1016/j.yhbeh.2012.11.012. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Jung F., Carlén M. Neuronal oscillations and the mouse prefrontal cortex. Int. Rev. Neurobiol. 2021;158:337–372. doi: 10.1016/bs.irn.2020.11.005. [DOI] [PubMed] [Google Scholar]
  • 24.Kirischuk S., Sinning A., Blanquie O., Yang J.W., Luhmann H.J., Kilb W. Modulation of neocortical development by early neuronal activity: physiology and pathophysiology. Front. Cell. Neurosci. 2017;11:379. doi: 10.3389/fncel.2017.00379. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Bitzenhofer S.H., Ahlbeck J., Wolff A., Wiegert J.S., Gee C.E., Oertner T.G., Hanganu-Opatz I.L. Layer-specific optogenetic activation of pyramidal neurons causes beta-gamma entrainment of neonatal networks. Nat. Commun. 2017;8:14563. doi: 10.1038/ncomms14563. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Bitzenhofer S.H., Pöpplau J.A., Hanganu-Opatz I. Gamma activity accelerates during prefrontal development. eLife. 2020;9 doi: 10.7554/eLife.56795. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Wei X., Thomas N., Hatch N.E., Hu M., Liu F. Postnatal craniofacial skeletal development of female C57BL/6NCrl mice. Front. Physiol. 2017;8:697. doi: 10.3389/fphys.2017.00697. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Donoghue T., Schaworonkow N., Voytek B. Methodological considerations for studying neural oscillations. Eur. J. Neurosci. 2022;55:3502–3527. doi: 10.1111/ejn.15361. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Buzsáki G., Wang X.J. Mechanisms of gamma oscillations. Annu. Rev. Neurosci. 2012;35:203–225. doi: 10.1146/annurev-neuro-062111-150444. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Sirota A., Montgomery S., Fujisawa S., Isomura Y., Zugaro M., Buzsáki G. Entrainment of neocortical neurons and gamma oscillations by the hippocampal theta rhythm. Neuron. 2008;60:683–697. doi: 10.1016/j.neuron.2008.09.014. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Karalis N., Sirota A. Breathing coordinates cortico-hippocampal dynamics in mice during offline states. Nat. Commun. 2022;13:467. doi: 10.1038/s41467-022-28090-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Cutts C.S., Eglen S.J. Detecting pairwise correlations in spike trains: an objective comparison of methods and application to the study of retinal waves. J. Neurosci. 2014;34:14288–14303. doi: 10.1523/JNEUROSCI.2767-14.2014. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Du X., Serena K., Hwang W.J., Grech A.M., Wu Y.W.C., Schroeder A., Hill R.A. Prefrontal cortical parvalbumin and somatostatin expression and cell density increase during adolescence and are modified by BDNF and sex. Mol. Cell. Neurosci. 2018;88:177–188. doi: 10.1016/j.mcn.2018.02.001. [DOI] [PubMed] [Google Scholar]
  • 34.Wang Y., Dye C.A., Sohal V., Long J.E., Estrada R.C., Roztocil T., Lufkin T., Deisseroth K., Baraban S.C., Rubenstein J.L. Dlx5 and Dlx6 regulate the development of parvalbumin-expressing cortical interneurons. J. Neurosci. 2010;30:5334–5345. doi: 10.1523/JNEUROSCI.5963-09.2010. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Chang M., Dian J.A., Dufour S., Wang L., Moradi Chameh H., Ramani M., Zhang L., Carlen P.L., Womelsdorf T., Valiante T.A. Brief activation of GABAergic interneurons initiates the transition to ictal events through post-inhibitory rebound excitation. Neurobiol. Dis. 2018;109:102–116. doi: 10.1016/j.nbd.2017.10.007. [DOI] [PubMed] [Google Scholar]
  • 36.Miyamae T., Chen K., Lewis D.A., Gonzalez-Burgos G. Distinct physiological maturation of parvalbumin-positive neuron subtypes in mouse prefrontal cortex. J. Neurosci. 2017;37:4883–4902. doi: 10.1523/JNEUROSCI.3325-16.2017. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Vaillant A.R., Zanassi P., Walsh G.S., Aumont A., Alonso A., Miller F.D. Signaling mechanisms underlying reversible, activity-dependent dendrite formation. Neuron. 2002;34:985–998. doi: 10.1016/s0896-6273(02)00717-1. [DOI] [PubMed] [Google Scholar]
  • 38.Berry K.P., Nedivi E. Spine dynamics: are they all the same? Neuron. 2017;96:43–55. doi: 10.1016/j.neuron.2017.08.008. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Schafer D.P., Lehrman E.K., Kautzman A.G., Koyama R., Mardinly A.R., Yamasaki R., Ransohoff R.M., Greenberg M.E., Barres B.A., Stevens B. Microglia sculpt postnatal neural circuits in an activity and complement-dependent manner. Neuron. 2012;74:691–705. doi: 10.1016/j.neuron.2012.03.026. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Ball J.B., Green-Fulgham S.M., Watkins L.R. Mechanisms of microglia-mediated synapse turnover and synaptogenesis. Prog. Neurobiol. 2022;218:102336. doi: 10.1016/j.pneurobio.2022.102336. [DOI] [PubMed] [Google Scholar]
  • 41.Morizawa Y.M., Matsumoto M., Nakashima Y., Endo N., Aida T., Ishikane H., Beppu K., Moritoh S., Inada H., Osumi N., et al. Synaptic pruning through glial synapse engulfment upon motor learning. Nat. Neurosci. 2022;25:1458–1469. doi: 10.1038/s41593-022-01184-5. [DOI] [PubMed] [Google Scholar]
  • 42.Paolicelli R.C., Bolasco G., Pagani F., Maggi L., Scianni M., Panzanelli P., Giustetto M., Ferreira T.A., Guiducci E., Dumas L., et al. Synaptic pruning by microglia is necessary for normal brain development. Science. 2011;333:1456–1458. doi: 10.1126/science.1202529. [DOI] [PubMed] [Google Scholar]
  • 43.Favuzzi E., Huang S., Saldi G.A., Binan L., Ibrahim L.A., Fernández-Otero M., Cao Y., Zeine A., Sefah A., Zheng K., et al. GABA-receptive microglia selectively sculpt developing inhibitory circuits. Cell. 2021;184:4048–4063.e32. doi: 10.1016/j.cell.2021.06.018. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44.Ma X., Chen K., Cui Y., Huang G., Nehme A., Zhang L., Li H., Wei J., Liong K., Liu Q., et al. Depletion of microglia in developing cortical circuits reveals its critical role in glutamatergic synapse development, functional connectivity, and critical period plasticity. J. Neurosci. Res. 2020;98:1968–1986. doi: 10.1002/jnr.24641. [DOI] [PubMed] [Google Scholar]
  • 45.Liu Y.J., Spangenberg E.E., Tang B., Holmes T.C., Green K.N., Xu X. Microglia elimination increases neural circuit connectivity and activity in adult mouse cortex. J. Neurosci. 2021;41:1274–1287. doi: 10.1523/JNEUROSCI.2140-20.2020. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46.Elmore M.R., Najafi A.R., Koike M.A., Dagher N.N., Spangenberg E.E., Rice R.A., Kitazawa M., Matusow B., Nguyen H., West B.L., et al. Colony-stimulating factor 1 receptor signaling is necessary for microglia viability, unmasking a microglia progenitor cell in the adult brain. Neuron. 2014;82:380–397. doi: 10.1016/j.neuron.2014.02.040. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47.Baccalá L.A., Sameshima K. Partial directed coherence: a new concept in neural structure determination. Biol. Cybern. 2001;84:463–474. doi: 10.1007/PL00007990. [DOI] [PubMed] [Google Scholar]
  • 48.Vinogradov S., Chafee M.V., Lee E., Morishita H. Psychosis spectrum illnesses as disorders of prefrontal critical period plasticity. Neuropsychopharmacology. 2023;48:168–185. doi: 10.1038/s41386-022-01451-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 49.Milbocker K.A., Campbell T.S., Collins N., Kim S., Smith I.F., Roth T.L., Klintsova A.Y. Glia-driven brain circuit refinement is altered by early-life adversity: behavioral outcomes. Front. Behav. Neurosci. 2021;15:786234. doi: 10.3389/fnbeh.2021.786234. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 50.Steinberg L. Risk taking in adolescence: what changes, and why? Ann. N. Y. Acad. Sci. 2004;1021:51–58. doi: 10.1196/annals.1308.005. [DOI] [PubMed] [Google Scholar]
  • 51.Minlebaev M., Ben-Ari Y., Khazipov R. Network mechanisms of spindle-burst oscillations in the neonatal rat barrel cortex in vivo. J. Neurophysiol. 2007;97:692–700. doi: 10.1152/jn.00759.2006. [DOI] [PubMed] [Google Scholar]
  • 52.Chen G., Rasch M.J., Wang R., Zhang X.H. Experience-dependent emergence of beta and gamma band oscillations in the primary visual cortex during the critical period. Sci. Rep. 2015;5:17847. doi: 10.1038/srep17847. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 53.Gordon J.A. Oscillations and hippocampal-prefrontal synchrony. Curr. Opin. Neurobiol. 2011;21:486–491. doi: 10.1016/j.conb.2011.02.012. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 54.Hyman J.M., Zilli E.A., Paley A.M., Hasselmo M.E. Medial prefrontal cortex cells show dynamic modulation with the hippocampal theta rhythm dependent on behavior. Hippocampus. 2005;15:739–749. doi: 10.1002/hipo.20106. [DOI] [PubMed] [Google Scholar]
  • 55.Caballero A., Thomases D.R., Flores-Barrera E., Cass D.K., Tseng K.Y. Emergence of GABAergic-dependent regulation of input-specific plasticity in the adult rat prefrontal cortex during adolescence. Psychopharmacol. (Berl.) 2014;231:1789–1796. doi: 10.1007/s00213-013-3216-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 56.Druart M., Nosten-Bertrand M., Poll S., Crux S., Nebeling F., Delhaye C., Dubois Y., Mittag M., Leboyer M., Tamouza R., et al. Elevated expression of complement C4 in the mouse prefrontal cortex causes schizophrenia-associated phenotypes. Mol. Psychiatry. 2021;26:3489–3501. doi: 10.1038/s41380-021-01081-6. [DOI] [PubMed] [Google Scholar]
  • 57.Sturman D.A., Moghaddam B. The neurobiology of adolescence: changes in brain architecture, functional dynamics, and behavioral tendencies. Neurosci. Biobehav. Rev. 2011;35:1704–1712. doi: 10.1016/j.neubiorev.2011.04.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 58.Segalowitz S.J., Santesso D.L., Jetha M.K. Electrophysiological changes during adolescence: a review. Brain Cogn. 2010;72:86–100. doi: 10.1016/j.bandc.2009.10.003. [DOI] [PubMed] [Google Scholar]
  • 59.Uhlhaas P.J., Roux F., Rodriguez E., Rotarska-Jagiela A., Singer W. Neural synchrony and the development of cortical networks. Trends Cogn. Sci. 2010;14:72–80. doi: 10.1016/j.tics.2009.12.002. [DOI] [PubMed] [Google Scholar]
  • 60.Uhlhaas P.J., Roux F., Singer W., Haenschel C., Sireteanu R., Rodriguez E. The development of neural synchrony reflects late maturation and restructuring of functional networks in humans. Proc. Natl. Acad. Sci. USA. 2009;106:9866–9871. doi: 10.1073/pnas.0900390106. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 61.Jang H.J., Chung H., Rowland J.M., Richards B.A., Kohl M.M., Kwag J. Distinct roles of parvalbumin and somatostatin interneurons in gating the synchronization of spike times in the neocortex. Sci. Adv. 2020;6:eaay5333. doi: 10.1126/sciadv.aay5333. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 62.Kroon T., van Hugte E., van Linge L., Mansvelder H.D., Meredith R.M. Early postnatal development of pyramidal neurons across layers of the mouse medial prefrontal cortex. Sci. Rep. 2019;9:5037. doi: 10.1038/s41598-019-41661-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 63.Gesuita L., Cavaccini A., Argunsah A.Ö., Favuzzi E., Ibrahim L.A., Stachniak T.J., De Gennaro M., Utz S., Greter M., Karayannis T. Microglia contribute to the postnatal development of cortical somatostatin-positive inhibitory cells and to whisker-evoked cortical activity. Cell Rep. 2022;40:111209. doi: 10.1016/j.celrep.2022.111209. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 64.Chini M., Pfeffer T., Hanganu-Opatz I. An increase of inhibition drives the developmental decorrelation of neural activity. eLife. 2022;11 doi: 10.7554/eLife.78811. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 65.Kalemaki K., Velli A., Christodoulou O., Denaxa M., Karagogeos D., Sidiropoulou K. The developmental changes in intrinsic and synaptic properties of prefrontal neurons enhance local network activity from the second to the third postnatal weeks in mice. Cereb. Cortex. 2022;32:3633–3650. doi: 10.1093/cercor/bhab438. [DOI] [PubMed] [Google Scholar]
  • 66.Rinetti-Vargas G., Phamluong K., Ron D., Bender K.J. Periadolescent maturation of GABAergic hyperpolarization at the axon initial segment. Cell Rep. 2017;20:21–29. doi: 10.1016/j.celrep.2017.06.030. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 67.Katz L.C., Shatz C.J. Synaptic activity and the construction of cortical circuits. Science. 1996;274:1133–1138. doi: 10.1126/science.274.5290.1133. [DOI] [PubMed] [Google Scholar]
  • 68.Kolk S.M., Rakic P. Development of prefrontal cortex. Neuropsychopharmacology. 2022;47:41–57. doi: 10.1038/s41386-021-01137-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 69.Bian W.J., Miao W.Y., He S.J., Qiu Z., Yu X. Coordinated spine pruning and maturation mediated by inter-spine competition for Cadherin/Catenin complexes. Cell. 2015;162:808–822. doi: 10.1016/j.cell.2015.07.018. [DOI] [PubMed] [Google Scholar]
  • 70.Koss W.A., Belden C.E., Hristov A.D., Juraska J.M. Dendritic remodeling in the adolescent medial prefrontal cortex and the basolateral amygdala of male and female rats. Synapse. 2014;68:61–72. doi: 10.1002/syn.21716. [DOI] [PubMed] [Google Scholar]
  • 71.Petanjek Z., Judaš M., Šimic G., Rasin M.R., Uylings H.B., Rakic P., Kostovic I. Extraordinary neoteny of synaptic spines in the human prefrontal cortex. Proc. Natl. Acad. Sci. USA. 2011;108:13281–13286. doi: 10.1073/pnas.1105108108. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 72.Huttenlocher P.R., Dabholkar A.S. Regional differences in synaptogenesis in human cerebral cortex. J. Comp. Neurol. 1997;387:167–178. doi: 10.1002/(sici)1096-9861(19971020)387:2<167::aid-cne1>3.0.co;2-z. [DOI] [PubMed] [Google Scholar]
  • 73.Anderson S.A., Classey J.D., Condé F., Lund J.S., Lewis D.A. Synchronous development of pyramidal neuron dendritic spines and parvalbumin-immunoreactive chandelier neuron axon terminals in layer III of monkey prefrontal cortex. Neuroscience. 1995;67:7–22. doi: 10.1016/0306-4522(95)00051-j. [DOI] [PubMed] [Google Scholar]
  • 74.Mallya A.P., Wang H.D., Lee H.N.R., Deutch A.Y. Microglial pruning of synapses in the prefrontal cortex during adolescence. Cereb. Cortex. 2019;29:1634–1643. doi: 10.1093/cercor/bhy061. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 75.Zuo Y., Lin A., Chang P., Gan W.B. Development of long-term dendritic spine stability in diverse regions of cerebral cortex. Neuron. 2005;46:181–189. doi: 10.1016/j.neuron.2005.04.001. [DOI] [PubMed] [Google Scholar]
  • 76.Ferrini F., De Koninck Y. Microglia control neuronal network excitability via BDNF signalling. Neural Plast. 2013;2013:429815. doi: 10.1155/2013/429815. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 77.Tseng K.Y., O'Donnell P. Dopamine modulation of prefrontal cortical interneurons changes during adolescence. Cereb. Cortex. 2007;17:1235–1240. doi: 10.1093/cercor/bhl034. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 78.Zhang Y.Q., Lin W.P., Huang L.P., Zhao B., Zhang C.C., Yin D.M. Dopamine D2 receptor regulates cortical synaptic pruning in rodents. Nat. Commun. 2021;12:6444. doi: 10.1038/s41467-021-26769-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 79.Weinhard L., di Bartolomei G., Bolasco G., Machado P., Schieber N.L., Neniskyte U., Exiga M., Vadisiute A., Raggioli A., Schertel A., et al. Microglia remodel synapses by presynaptic trogocytosis and spine head filopodia induction. Nat. Commun. 2018;9:1228. doi: 10.1038/s41467-018-03566-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 80.Vetter-O'Hagen C.S., Spear L.P. Hormonal and physical markers of puberty and their relationship to adolescent-typical novelty-directed behavior. Dev. Psychobiol. 2012;54:523–535. doi: 10.1002/dev.20610. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 81.Delevich K., Thomas A.W., Wilbrecht L. Adolescence and “late blooming” synapses of the prefrontal cortex. Cold Spring Harb. Symp. Quant. Biol. 2018;83:37–43. doi: 10.1101/sqb.2018.83.037507. [DOI] [PubMed] [Google Scholar]
  • 82.Kolluri N., Sun Z., Sampson A.R., Lewis D.A. Lamina-specific reductions in dendritic spine density in the prefrontal cortex of subjects with schizophrenia. Am. J. Psychiatry. 2005;162:1200–1202. doi: 10.1176/appi.ajp.162.6.1200. [DOI] [PubMed] [Google Scholar]
  • 83.Wu X., Shukla R., Alganem K., Zhang X., Eby H.M., Devine E.A., Depasquale E., Reigle J., Simmons M., Hahn M.K., et al. Transcriptional profile of pyramidal neurons in chronic schizophrenia reveals lamina-specific dysfunction of neuronal immunity. Mol. Psychiatry. 2021;26:7699–7708. doi: 10.1038/s41380-021-01205-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 84.Yilmaz M., Yalcin E., Presumey J., Aw E., Ma M., Whelan C.W., Stevens B., McCarroll S.A., Carroll M.C. Overexpression of schizophrenia susceptibility factor human complement C4A promotes excessive synaptic loss and behavioral changes in mice. Nat. Neurosci. 2021;24:214–224. doi: 10.1038/s41593-020-00763-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 85.Stogsdill J.A., Kim K., Binan L., Farhi S.L., Levin J.Z., Arlotta P. Pyramidal neuron subtype diversity governs microglia states in the neocortex. Nature. 2022;608:750–756. doi: 10.1038/s41586-022-05056-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 86.Carlén M. What constitutes the prefrontal cortex? Science. 2017;358:478–482. doi: 10.1126/science.aan8868. [DOI] [PubMed] [Google Scholar]
  • 87.Le Merre P., Ährlund-Richter S., Carlén M. The mouse prefrontal cortex: unity in diversity. Neuron. 2021;109:1925–1944. doi: 10.1016/j.neuron.2021.03.035. [DOI] [PubMed] [Google Scholar]
  • 88.Glantz L.A., Gilmore J.H., Hamer R.M., Lieberman J.A., Jarskog L.F. Synaptophysin and postsynaptic density protein 95 in the human prefrontal cortex from mid-gestation into early adulthood. Neuroscience. 2007;149:582–591. doi: 10.1016/j.neuroscience.2007.06.036. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 89.Klein D., Rotarska-Jagiela A., Genc E., Sritharan S., Mohr H., Roux F., Han C.E., Kaiser M., Singer W., Uhlhaas P.J. Adolescent brain maturation and cortical folding: evidence for reductions in gyrification. PLoS One. 2014;9 doi: 10.1371/journal.pone.0084914. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 90.Bellesi M., de Vivo L., Chini M., Gilli F., Tononi G., Cirelli C. Sleep loss promotes astrocytic phagocytosis and microglial activation in mouse cerebral cortex. J. Neurosci. 2017;37:5263–5273. doi: 10.1523/JNEUROSCI.3981-16.2017. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 91.Johnson C., Wilbrecht L. Juvenile mice show greater flexibility in multiple choice reversal learning than adults. Dev. Cogn. Neurosci. 2011;1:540–551. doi: 10.1016/j.dcn.2011.05.008. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 92.Pennington Z.T., Dong Z., Feng Y., Vetere L.M., Page-Harley L., Shuman T., Cai D.J. ezTrack: an open-source video analysis pipeline for the investigation of animal behavior. Sci. Rep. 2019;9:19979. doi: 10.1038/s41598-019-56408-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 93.Onslow A.C., Bogacz R., Jones M.W. Quantifying phase-amplitude coupling in neuronal network oscillations. Prog. Biophys. Mol. Biol. 2011;105:49–57. doi: 10.1016/j.pbiomolbio.2010.09.007. [DOI] [PubMed] [Google Scholar]
  • 94.Rossant C., Kadir S.N., Goodman D.F.M., Schulman J., Hunter M.L.D., Saleem A.B., Grosmark A., Belluscio M., Denfield G.H., Ecker A.S., et al. Spike sorting for large, dense electrode arrays. Nat. Neurosci. 2016;19:634–641. doi: 10.1038/nn.4268. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 95.Vinck M., van Wingerden M., Womelsdorf T., Fries P., Pennartz C.M. 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]
  • 96.Gretenkord S., Kostka J.K., Hartung H., Watznauer K., Fleck D., Minier-Toribio A., Spehr M., Hanganu-Opatz I.L. Coordinated electrical activity in the olfactory bulb gates the oscillatory entrainment of entorhinal networks in neonatal mice. PLoS Biol. 2019;17 doi: 10.1371/journal.pbio.2006994. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 97.Stone C., Koo C. American Statistical Association; 1985. Additive Splines in Statistics. [Google Scholar]
  • 98.Harrell J., Frank E. Springer; 2015. Regression Modeling Strategies. [DOI] [Google Scholar]

Associated Data

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

Supplementary Materials

Document S1. Figures S1–S8
mmc1.pdf (2.4MB, pdf)
Table S1. Detailed statistical results, related to STAR Methods
mmc2.xlsx (96.7KB, xlsx)
Table S2. Detailed statistical results for the effect of sex, related to STAR Methods
mmc3.xlsx (35.4KB, xlsx)
Table S3. Recording summary of chronically recorded mice, related to STAR Methods
mmc4.xlsx (20.9KB, xlsx)
Document S2. Article plus supplemental information
mmc5.pdf (13MB, pdf)

Data Availability Statement

  • •

    LFP and SUA data for all chronic recordings are uploaded to an open-access repository and are publicly available. The DOI is listed in the key resources table. Microscopy images and behavioral videos reported in this paper will be shared by the lead contact upon request.

  • •

    All original or customized code is deposited to an open-access repository and is publicly available. The DOI is listed in the key resources table.

  • •

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

RESOURCES