Activity-dependent maturation of prefrontal gamma ...Gamma control of cognitive maturation...

Gamma control of cognitive maturation Bitzenhofer et al.

1

1

2

3

4

5

6

7

Activity-dependent maturation of prefrontal gamma 8

oscillations sculpts cognitive performance 9

10

Sebastian H. Bitzenhofer1,*,§, Jastyn A. Pöpplau1, Mattia Chini1, Annette 11

Marquardt1 & Ileana L. Hanganu-Opatz1,* 12

1 Developmental Neurophysiology, Institute of Neuroanatomy, University Medical Center 13

Hamburg-Eppendorf, Hamburg, Germany 14 §Current address: Center for Neural Circuits and Behavior, Department of 15

Neurosciences, University of California, San Diego, La Jolla, CA, USA. 16

17

* Corresponding author: Ileana L. Hanganu-Opatz 18

[email protected] 19

Falkenried 94, 20251 Hamburg, Germany 20

21

Sebastian Bitzenhofer 22



25

4 figures 26

Words: Abstract 70, Main 1883, 39 references 27

Supplementary Material: Methods, Supplementary Fig. 1-4, Supplementary Tab. 1 28

29

Running title 30

Gamma control of cognitive maturation 31

32

One sentence summary 33

Fast oscillatory activity in layer 2/3 sculpts the maturation of prefrontal cortex and 34

cognitive performance. 35

36

.CC-BY-NC 4.0 International licenseavailable under awas not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint (whichthis version posted February 22, 2019. ; https://doi.org/10.1101/558957doi: bioRxiv preprint

https://doi.org/10.1101/558957

http://creativecommons.org/licenses/by-nc/4.0/


2

Abstract 37

Gamma oscillations are the neural network attribute of cognitive processing. They 38

emerge early in life, yet their contribution to cortical circuit formation is unknown. We 39

show that layer 2/3 pyramidal neurons entrain mouse prefrontal cortex in fast oscillations 40

with increasing frequency across development. Chronic boosting of fast oscillations at 41

neonatal age reversibly alters neuronal morphology, but cause permanent circuity 42

dysfunction and impairs recognition memory and social interactions later in life. 43

44



https://doi.org/10.1101/558957



3

Main 45

Synchronization of neuronal activity in fast rhythms is a commonly observed feature in 46

the cerebral cortex 1. Although its contribution to cognitive performance is still a matter of 47

debate, high frequency oscillatory activity facilitates the communication between 48

neuronal ensembles and is thought to shape information processing in cortical networks 49

2-4. Oscillatory activity at frequencies within gamma band has been proposed to emerge 50

from reciprocal interactions of excitatory and inhibitory neurons 1. Fast inhibitory 51

feedback via soma-targeting parvalbumin (PV)-expressing inhibitory interneurons leads 52

to fast gamma (30-80 Hz) 5,6, whereas dendrite-targeting somatostatin (SST)-expressing 53

inhibitory interneurons contribute to beta/low gamma (20-40 Hz) activity 6,7. 54

Interneuronal dysfunction and ensuing abnormal gamma activity in the medial prefrontal 55

cortex (mPFC) has been related to impaired cognitive flexibility 8. Moreover, pyramidal 56

neurons critically contribute to the generation of fast oscillatory rhythms, beyond solely 57

providing the excitatory drive. As recently shown, their feed-forward excitation 58

determines oscillatory dynamics 9. 59

Despite substantial knowledge linking fast oscillations in the mature cortex and 60

cognitive abilities 8,10,11, their role during development is still largely unknown. Here we 61

examined the emergence of gamma activity in the mouse mPFC across development 62

and assessed the role of fast oscillations for the maturation of prefrontal networks and 63

cognitive abilities. Shifting the level of early oscillatory activity to a transient excessive 64

boosting of fast frequency events, we provide evidence that gamma entrainment of 65

neonatal circuits is critical for the prefrontal function and behavioral performance of 66

adults. 67



https://doi.org/10.1101/558957



4

Extracellular recordings in the mPFC of postnatal day (P) 5-40 mice revealed that 68

fast frequency oscillations occur spontaneously in awake and urethane-anesthetized 69

mice starting around P8 (Fig. 1a,b and S1a-d). While anesthesia reduced the oscillatory 70

power during development across a broad range of frequencies (Fig. S1e-g), it did not 71

affect the spectral composition of oscillations 12. In both states, fast frequency 72

oscillations increased in power and frequency with age, reaching stable values around 73

P25 with an adult-like gamma frequency peak at ~50 Hz (Fig. 1c,d). Using a recently 74

established protocol for optogenetic manipulation in neonatal mice 13,14, we interrogated 75

the neuronal substrate of fast oscillatory activity across development. To this end, we 76

transfected ~25% of pyramidal neurons in layers 2/3 (L2/3) by in utero electroporation 77

(IUE) with the light-sensitive channelrhodopsin 2 derivate E123T T159C (ChR2(ET/TC)) 78

and stimulated them with ramp light pulses. This stimulation type activates the network 79

without imposing a specific rhythm, thus enabling neurons to fire at their preferred 80

frequencies. We focused on neurons from L2/3 because our previous investigations 81

identified them as drivers of fast oscillatory rhythms at neonatal age 13. This layer-82

specificity seems to be a developmental feature of mPFC, since recent work in adult 83

sensory cortices has shown that gamma rhythms can be driven by pyramidal neurons 84

from all layers 15,16. We found that light-driven fast oscillatory activity increased in power 85

and frequency across development, similarly to what we observed for spontaneous 86

oscillations (Fig. 1e-k and S1h-l). This increase co-occurred with the maturation of PV-87

expressing interneurons (Fig. S2a,b) and the maturation of inhibitory synapses on 88

excitatory neurons 17,18. Therefore, a developmental transition from SST-dominated to 89

PV-dominated feedback inhibition might underlie the gamma frequency increase across 90

development. Alternatively, increased gamma frequency with age might result from 91



https://doi.org/10.1101/558957



5

alterations in the biophysical properties underlying excitability and maximal firing rate of 92

cortical pyramidal neurons 19. Indeed, we found that the firing rate and amplitude of 93

single unit activity (SUA) increased with age, whereas its half width decreased (Fig. S2c-94

e). To disentangle the role of interneurons from that of pyramidal neurons for the 95

generation of fast oscillations across development, we transfected L2/3 pyramidal 96

neurons with ChR2(ET/TC) by IUE in transgenic mice expressing archaerhodopsin3 97

(ArchT) in all interneurons. Silencing of interneurons led to a broadband increase of 98

activity in fast frequencies (>12 Hz) that was most pronounced in neonatal mice (Fig. 99

S2f-i). However, silencing of interneurons did not affect gamma activity driven by light 100

stimulation of L2/3 neurons, demonstrating a critical role of pyramidal neurons for the 101

generation of gamma rhythmicity. 102

The presence of neuronal ensembles capable of generating fast oscillatory 103

rhythms during early development raises the question of whether this pattern of activity 104

solely reflects or directly contributes to the maturation of cortical circuits and their 105

functions. Abundant literature documented the role of coordinated electrical activity for 106

dendrite formation, synaptic pruning and apoptosis 20-22. To elucidate the role of 107

neonatal fast oscillatory activity for prefrontal network maturation, we established here 108

an optogenetic protocol for chronic early stimulation (ES) of L2/3 pyramidal neurons at 109

the age when fast oscillations start to occur in the mPFC (Fig. 2a) 23. ES mice were 110

stimulated daily from P7-11 with blue light ramps (473 nm), whereas control animals 111

were similarly stimulated, but with yellow light ramps (594 nm) that do not activate 112

ChR2(ET/TC). Both intra- and transcranial ramp light stimulation acutely boosted 113

network oscillations in beta-low gamma frequency range in neonatal mice (Fig. 2b). 114



https://doi.org/10.1101/558957



6

First, we assessed the effects of ES on the morphology of L2/3 pyramidal 115

neurons. Immediately after stimulation, light-induced boosting of fast oscillatory activity 116

led to an exhaustive and premature growth of dendrites of L2/3 pyramidal neurons (Fig. 117

2c,d). At P11-12, stimulated neurons reached a dendritic length (3166 ± 632 µm) 118

significantly higher than of age-matched controls (2029 ± 244 µm, p=0.007) but 119

comparable to that of control animals at P23-25 (3356 ± 356 µm). These effects were 120

transient, such that the dendritic complexity and length of L2/3 pyramidal neurons was 121

largely unchanged in P23-25 (ES: 2906.5 ± 860.5 µm, p=0.631) and P38-40 (control: 122

4202 ± 625.5 µm; ES: 3746. 5± 630.5 µm, p=0.162) ES mice. Surprisingly, ES caused a 123

transient reduction of GABA-positive neurons (Fig. 2e). The survival of interneurons has 124

been shown to depend on the level of excitatory input during early postnatal 125

development 20,24. Therefore, the reduction of interneuron density might be explained by 126

the fact that optogenetic stimulation and subsequent augmented firing of transfected 127

neurons (~25% of total pyramidal neurons) reduces the overall level of activity, due to 128

surround suppression 7,15. 129

Next, we monitored the ES effects on functional network maturation. Extracellular 130

recordings from the mPFC of P11-12, P23-25 and P38-40 control and ES mice were 131

performed simultaneously with light stimulation of ChR2(ET/TC)-transfected L2/3 132

pyramidal neurons. Acutely stimulated activity was similar in control and ES mice at 133

P11-12, whereas at P23-25 the induced oscillatory peak slightly decreased, yet not at 134

significant level. At P38-40, the amplitude of driven oscillations in gamma band was 135

significantly reduced in both anesthetized (modulation index, P38-40: control: 0.72 ± 136

0.16; ES: 0.19 ± 0.24, p=0.025) (Fig. 3a-c) and awake (modulation index, P38-40: 137

control: 0.72 ± 0.06; ES: 0.50 ± 0.17, p=0.043) (Fig. S3f,g) ES mice when compared to 138



https://doi.org/10.1101/558957



7

controls. Bilateral recordings from all prefrontal layers in ES mice revealed that interlayer 139

as well as interhemispheric gamma frequency coherence was reduced at P38-40 when 140

compared to controls (Fig. 3d-i). In contrast, spontaneous activity of mPFC did not differ 141

between controls and ES mice across development (Fig. S3a-e), indicating that 142

perturbation of fast neonatal rhythms might solely disrupt the ability of local circuits to 143

respond to activation (i.e. under physiological conditions: stimulus, task). Of note, the 144

reduction of L2/3 driven gamma activity in ES mice occurred towards the end of juvenile 145

development, when dendritic morphology and interneuron number had largely recovered 146

(Fig. 2d,e). 147

Gamma activity is thought to affect information processing in adult mPFC and, 148

ultimately, behavior 8,10,11. Weaker gamma entrainment of prefrontal circuits after 149

interfering with fast oscillatory activity through ES at neonatal age might come along with 150

compromised cognitive abilities. We used a battery of tests to examine mPFC-151

dependent behavior of ES and control mice at P16-36. To avoid confounding effects, we 152

ascertained that ES mice have normal somatic and reflex development (Fig. S4a). Their 153

open field behavior was also normal (Fig. S4b). However, ES mice failed to distinguish 154

between objects in mPFC-dependent novel object and recency recognition tasks, but not 155

in a hippocampus-dependent object location recognition task (Fig. 4a,b and S4c). 156

Moreover, they showed abnormal social interactions as mirrored by reduced interaction 157

time with the mother (Fig. 4c). Spatial working memory was also impaired as revealed 158

by significant deficits of ES mice in 8-arm maze test (Fig. 4d). In contrast, spatial 159

alteration and tail suspension were not affected (Fig. S4d,e). Combined analysis of 160

behavioral performance throughout the different tests with support vector classification 161

and 5-fold cross-validation yielded a correct classification of control and ES mice in 77% 162



https://doi.org/10.1101/558957



8

in the training and 78% in the test data set (Fig. 4e). Thus, the cognitive deficits at 163

adulthood are reliable markers of functional imbalance within neonatal prefrontal 164

circuitry. 165

Gamma oscillations are highly relevant for adult cortical function 3,4,25 and 166

dysfunction 8,26, yet their development is still poorly understood. The results of the 167

present study (i) provide insights into the age-dependent mechanisms of early fast 168

oscillations and (ii) demonstrate their relevance for the functional maturation of prefrontal 169

networks as well as cognitive and social abilities. Throughout development, L2/3 170

pyramidal neurons have been identified as key players for the generation of fast 171

oscillations in the mPFC. Their activation led to activity patterns with similar features as 172

spontaneously-generated gamma oscillations. With age, the fast rhythms become more 173

prominent. The frequency and amplitude increase results from the maturation of 174

electrical properties of pyramidal neurons but also from the progressive embedding of 175

PV-expressing fast-spiking interneurons into circuits initially controlled by SST-176

expressing interneurons 27,28. 177

Do fast oscillations during development simply mirror neural maturation or do they 178

contribute to circuit refinement? This key question has been previously approached for 179

sensory systems with defined cortical topography. Neonatal fast oscillations have been 180

proposed to facilitate the formation of topographic units through input replay in 181

somatosensory thalamocortical circuits 29. In visual cortex, gamma activity depends on 182

visual experience and has been proposed as an indicator of cortical maturity 30. Being 183

not directly driven by corresponding environmental stimuli or sensory periphery, limbic 184

circuits follow distinct developmental rules. Emergence of oscillation-coupled prefrontal 185

ensembles at a developmental stage when the mice are blind, deaf and lack whisking 186



https://doi.org/10.1101/558957



9

and motor abilities, is driven by hippocampal theta oscillations 23,31. They result, on their 187

turn, from activation of lateral entorhinal cortex under the control of olfactory stimuli 32,33. 188

Excitatory hippocampal projections targeting L5/6 neurons promote interlaminar 189

interactions within mPFC centered on coordinated activation of L2/3 pyramidal neurons. 190

Boosting this neonatal process through chronic light stimulation of L2/3 pyramidal 191

neurons seems to push the system out of an optimum level of activity and synchrony. 192

The direct consequences are transient structural changes (exuberant dendritic 193

arborization, decreased density of interneurons) that are compensated before adulthood. 194

However, the function of prefrontal circuits appears permanently compromised after ES. 195

The weaker gamma entrainment of adult prefrontal circuits contributes to poorer 196

performance in behavioral tasks that require mPFC, such as novel and recency 197

recognition, working memory as well as social interaction. These results uncover the role 198

of neonatal oscillations for the maturation of limbic circuit function and cognitive abilities. 199

The mechanisms described here might explain cognitive difficulties of preterm 200

born humans that experience excessive sensory stimulation in neonatal intensive care 201

unit (NICU) at a comparable stage of brain development (2nd-3rd gestational trimester) 202

34,35. These stressful stimuli might trigger premature gamma entrainment, perturbing the 203

activity-dependent maturation of cortical networks (Moiseev et al., 2015). Frontal regions 204

have been reported to be particularly vulnerable to NICU conditions 36 and 205

correspondingly, preterm children highly prone to frontally-confined impairment, such as 206

memory and attention deficits (Taylor and Clark, 2016). Thus, our findings lend support 207

to the concept that fast network oscillations have a central role for neurodevelopmental 208

disorders 37,38. 209

210



https://doi.org/10.1101/558957



10

References 211

1 Buzsaki, G. Rhythms of the Brain. (Oxford University Press, 2006). 212

2 Singer, W. Neuronal oscillations: unavoidable and useful? Eur J Neurosci 48, 2389-2398. 213

3 Cardin, J. A. Snapshots of the Brain in Action: Local Circuit Operations through the Lens 214 of gamma Oscillations. Journal of Neuroscience 36, 10496-10504. 215

4 Sohal, V. S. How Close Are We to Understanding What (if Anything) gamma Oscillations 216 Do in Cortical Circuits? Journal of Neuroscience 36, 10489-10495. 217

5 Cardin, J. A. et al. Driving fast-spiking cells induces gamma rhythm and controls sensory 218 responses. Nature 459, 663-667. 219

6 Chen, G. et al. Distinct Inhibitory Circuits Orchestrate Cortical beta and gamma Band 220 Oscillations. Neuron 96, 1403-1418 e1406. 221

7 Veit, J., Hakim, R., Jadi, M. P., Sejnowski, T. J. & Adesnik, H. Cortical gamma band 222 synchronization through somatostatin interneurons. Nat Neurosci 20, 951-959. 223

8 Cho, K. K. et al. Gamma rhythms link prefrontal interneuron dysfunction with cognitive 224 inflexibility in Dlx5/6(+/-) mice. Neuron 85, 1332-1343. 225

9 Nicholson, E., Kuzmin, D. A., Leite, M., Akam, T. E. & Kullmann, D. M. Analogue closed-226 loop optogenetic modulation of hippocampal pyramidal cells dissociates gamma 227 frequency and amplitude. Elife 7. 228

10 Cao, W. et al. Gamma Oscillation Dysfunction in mPFC Leads to Social Deficits in 229 Neuroligin 3 R451C Knockin Mice. Neuron 97, 1253-1260 e1257. 230

11 Bastos, A. M., Loonis, R., Kornblith, S., Lundqvist, M. & Miller, E. K. Laminar recordings 231 in frontal cortex suggest distinct layers for maintenance and control of working memory. 232 Proc Natl Acad Sci U S A 115, 1117-1122. 233

12 Chini, M. G., S.; Kostka, J.K.; Poepplau, J.A.; Cornelissen, L.; Berde, C.B.; Hanganu-234 Opatz, I.L.; Bitzenhofer S.H. Identifying neurophysiological features associated with 235 anesthetic state in newborn mice and humans. bioRxiv. 236

13 Bitzenhofer, S. H. et al. Layer-specific optogenetic activation of pyramidal neurons 237 causes beta-gamma entrainment of neonatal networks. Nat Commun 8, 14563. 238

14 Bitzenhofer, S. H., Ahlbeck, J. & Hanganu-Opatz, I. L. Methodological Approach for 239 Optogenetic Manipulation of Neonatal Neuronal Networks. Front Cell Neurosci 11, 239. 240

15 Adesnik, H. & Scanziani, M. Lateral competition for cortical space by layer-specific 241 horizontal circuits. Nature 464, 1155-1160. 242

16 Adesnik, H. Layer-specific excitation/inhibition balances during neuronal synchronization 243 in the visual cortex. J Physiol 596, 1639-1657. 244

17 Virtanen, M. A. et al. Development of inhibitory synaptic inputs on layer 2/3 pyramidal 245 neurons in the rat medial prefrontal cortex. Brain Struct Funct 223, 1999-2012. 246

18 Guan, W. et al. Eye opening differentially modulates inhibitory synaptic transmission in 247 the developing visual cortex. Elife 6. 248

19 Oswald, A. M. & Reyes, A. D. Maturation of intrinsic and synaptic properties of layer 2/3 249 pyramidal neurons in mouse auditory cortex. J Neurophysiol 99, 2998-3008. 250

20 Wong, F. K. et al. Pyramidal cell regulation of interneuron survival sculpts cortical 251 networks. Nature 557, 668-673. 252

21 Riccomagno, M. M. & Kolodkin, A. L. Sculpting neural circuits by axon and dendrite 253 pruning. Annu Rev Cell Dev Biol 31, 779-805. 254



https://doi.org/10.1101/558957



11

22 Anastasiades, P. G. & Butt, S. J. A role for silent synapses in the development of the 255 pathway from layer 2/3 to 5 pyramidal cells in the neocortex. J Neurosci 32, 13085-256 13099. 257

23 Brockmann, M. D., Poschel, B., Cichon, N. & Hanganu-Opatz, I. L. Coupled oscillations 258 mediate directed interactions between prefrontal cortex and hippocampus of the neonatal 259 rat. Neuron 71, 332-347. 260

24 Denaxa, M. et al. Modulation of Apoptosis Controls Inhibitory Interneuron Number in the 261 Cortex. Cell Rep 22, 1710-1721. 262

25 Kim, H., Ahrlund-Richter, S., Wang, X. M., Deisseroth, K. & Carlen, M. Prefrontal 263 Parvalbumin Neurons in Control of Attention. Cell 164, 208-218. 264

26 Uhlhaas, P. J. & Singer, W. Abnormal neural oscillations and synchrony in schizophrenia. 265 Nat Rev Neurosci 11, 100-113. 266

27 Tuncdemir, S. N. et al. Early Somatostatin Interneuron Connectivity Mediates the 267 Maturation of Deep Layer Cortical Circuits. Neuron 89, 521-535. 268

28 Mayer, C. et al. Developmental diversification of cortical inhibitory interneurons. Nature 269 555, 457-462. 270

29 Minlebaev, M., Colonnese, M., Tsintsadze, T., Sirota, A. & Khazipov, R. Early gamma 271 oscillations synchronize developing thalamus and cortex. Science 334, 226-229. 272

30 Chen, G., Rasch, M. J., Wang, R. & Zhang, X. H. Experience-dependent emergence of 273 beta and gamma band oscillations in the primary visual cortex during the critical period. 274 Sci Rep 5. 275

31 Ahlbeck, J., Song, L., Chini, M., Bitzenhofer, S. H. & Hanganu-Opatz, I. L. Glutamatergic 276 drive along the septo-temporal axis of hippocampus boosts prelimbic oscillations in the 277 neonatal mouse. Elife 7. 278

32 Hartung, H., Brockmann, M. D., Poschel, B., De Feo, V. & Hanganu-Opatz, I. L. Thalamic 279 and Entorhinal Network Activity Differently Modulates the Functional Development of 280 Prefrontal-Hippocampal Interactions. J Neurosci 36, 3676-3690. 281

33 Gretenkord, S. et al. Coordinated electrical activity in the olfactory bulb gates the 282 oscillatory entrainment of entorhinal networks in neonatal mice. PLoS Biol 17, e2006994. 283

34 Moiseev, A., Doesburg, S. M., Herdman, A. T., Ribary, U. & Grunau, R. E. Altered 284 Network Oscillations and Functional Connectivity Dynamics in Children Born Very 285 Preterm. Brain Topogr 28, 726-745. 286

35 Altimier, L. & Phillips, R. The Neonatal Integrative Developmental Care Model: Advanced 287 Clinical Applications of the Seven Core Measures for Neuroprotective Family-centered 288 Developmental Care. Newborn and Infant Nursing Reviews 16, 230-244. 289

36 Als, H. et al. Early experience alters brain function and structure. Pediatrics 113, 846-290 857. 291

37 Peiker, I. et al. Perceptual Integration Deficits in Autism Spectrum Disorders Are 292 Associated with Reduced Interhemispheric Gamma-Band Coherence. Journal of 293 Neuroscience 35, 16352-16361. 294

38 Chini, M. L., C.; Poepplau, J. A.; Xu, X.; Ahlbeck, J.; Bitzenhofer, S. H.; Mulert, C.; 295 Hanganu-Opatz, I. L. Microglia inhibition rescues developmental hypofrontality in a 296 mouse model of mental illness. bioRxiv. 297

298

299



https://doi.org/10.1101/558957



12

Acknowledgments 300

We thank P. Putthoff, A. Dahlmann, and K. Titze for excellent technical assistance. This 301

work was funded by grants from the European Research Council (ERC-2015-CoG 302

681577 to I.L.H.-O.) and the German Research Foundation (Ha 4466/10-1, Ha4466/11-303

1, Ha4466/12-1, SPP 1665, SFB 936 B5 to I.L.H.-O.). 304

I.L. H.-O. is founding member of FENS Kavli Network of Excellence. 305

306

Author contributions: S.H.B. and I.L.H.-O. designed the experiments, S.H.B., J.A.P 307

and A.M. carried out the experiments, S.H.B., J.A.P. and M.C. analyzed the data, 308

S.H.B., I.L.H.-O., M.C and J.A.P interpreted the data and wrote the manuscript. All 309

authors discussed and commented on the manuscript. 310

311



https://doi.org/10.1101/558957


312

313

314

315

316

317

318

319

320

321

322

323

Gamma

Figures

Fig. 1.

Schema

anesthe

mPFC o

corresp

oscillato

displayi

anesthe

strength

IUE at

a control of

s

Age dep

atic of the

etized (hea

of awake m

ponding po

ory activity

ing the age

etized (n=8

h. (e) ChR2

E15.5 in a

f cognitive

pendence

e recording

ad-fixed) P

mice at diffe

ower spect

y averaged

e-depende

80) and aw

2(ET/TC)-2

a coronal s

maturation

of fast rh

g paradigm

P5-40 mice

erent ages

tra (right).

d over ea

ence of pea

wake mice

2A-RFP-ex

lice of a P

n

hythmic a

ms in awak

e. (b) Exa

s after band

(c) Color

ach age fo

ak frequen

(n=20, 35

xpression i

P10 mouse

activity in

ke (head-fi

ample loca

d-pass filte

r-coded po

or P5-40 m

ncies of fas

recording

n L2/3 pyra

. (f) Exam

B

the deve

ixed on a

l field pote

ering (left) a

ower spect

mice (n=80

st oscillatio

s). Marker

amidal neu

ple LFPs d

Bitzenhofer

eloping m

movable d

entials (LF

accompani

tra of spo

0). (d) Sc

ons (12-10

r size displ

urons in mP

driven by r

r et al.

13

mPFC. (a)

disk) and

FPs) from

ed by the

ntaneous

atter plot

00 Hz) for

ays peak

PFC after

ramp light



https://doi.org/10.1101/558957



14

stimulations (473 nm, 3 s) of L2/3 pyramidal neurons in mPFC at different ages (left) 324

accompanied by the corresponding modulation index (MI) of power spectra normalized 325

to pre-stimulus (right). (g) Color-coded normalized (during-to-before stimulation) MI of 326

power spectra averaged over age for P5-40 mice (n=80). (h) Scatter plot displaying the 327

age-dependence of stimulus induced peak frequencies across development for 328

anesthetized (n=80) and awake mice (n=20, 35 recordings). Marker size displays peak 329

strength. (i-k) Same as F-H for control ramp light stimulations (594 nm, 3s). See 330

Supplementary Tab. 1 for detailed information on statistics. Average data is presented 331

as median ± MAD. 332



https://doi.org/10.1101/558957


333

334

335

336

337

338

339

340

341

342

343

344

345

346

347

Gamma

Fig. 2.

mPFC.

behavio

(right)

togethe

Exampl

control

dendriti

the som

(conditi

group;

(e) Left

mice at

positive

a control of

Structura

(a) Schem

oral assess

ramp light

er with the

le pictures

(left) and

ic intersect

ma center

on p=2.2e

ES 18 cell

t, GABA an

t P11-12,

e and GAB

f cognitive

l alteration

matic timeli

sment. (b)

t stimulatio

e correspon

and avera

ES mice

tions avera

for contro

-16) and P

ls of 3 mic

nd CaMKII

P23-25 an

BA-positive

maturation

ns in mPF

ne of mani

Example o

ons of L2

nding MI o

age heat m

(right) at

aged for L2

ol and ES

P38-40 (co

ce / age gr

immunost

nd P38-40

e neuron d

n

FC as resu

ipulations a

of LFPs driv

2/3 pyramid

of power s

maps of IUE

P11-12,

2/3 pyramid

S mice at

ndition p=2

roup). Thin

tainings of

0. Right, vi

density in

ult of ES o

as well as

ven by intr

dal neuron

spectra (bo

E-transfecte

P23-25 an

dal neuron

P11-12 (c

2.4e-5) (co

lines corr

prefrontal

iolin plots

prefrontal

B

of L2/3 pyr

morpholog

racranial (le

ns in mPF

ottom) for

ed L2/3 py

nd P38-40

s within a

condition p

ontrol, 18 c

respond to

neurons f

of RFP-tra

L2/3 of co

Bitzenhofer

ramidal ne

gical, functi

eft) and tra

FC (top) d

a P11 m

yramidal ne

0. (d) Line

250 μm ra

p=2.2e-16)

cells of 3 m

individual

rom contro

ansfected,

ontrol and

r et al.

15

eurons in

onal, and

anscranial

displayed

ouse. (c)

eurons for

e plots of

dius from

), P23-25

mice / age

neurons.

ol and ES

CaMKII-

ES mice



https://doi.org/10.1101/558957



16

(right). *P<0.05, **P<0.01 and ***P<0.001. See Supplementary Tab. 1 for detailed 348

information on statistics. Average data is presented as median ± MAD. 349



https://doi.org/10.1101/558957


350

351

352

353

354

355

356

357

358

359

360

361

362

363

Gamma

Fig. 3.

(a) Top

ES mic

corresp

right, sa

mice (c

(d) Sca

of imag

control

Scatter

imagina

control

Same a

a control of

Prefronta

left, MI of

ce at P1

ponding MI

ame as top

control n=1

atter plots d

ginary cohe

and ES m

plots disp

ary cohere

and ES m

as (d,e) for

f cognitive

l dysfunct

power spe

1-12 (con

power pe

p for contro

0, ES n=1

displaying

erence bet

mice for s

playing the

ence betwe

mice for st

r P23-25 m

maturation

tion as res

ectra of ligh

trol n=10,

ak strength

ol stimulatio

1). (c) Sam

the peak s

tween ligh

stimulation

peak stren

een light-dr

timulation

mice. (h,i) S

n

sult of ES

ht-driven ac

, ES n=1

h as a func

on (ramp, 5

me as (a) f

strength as

t-driven L2

(473 nm)

ngth as a f

riven L2/3

(473 nm)

Same as (d

S of L2/3 p

ctivity (ram

0). Top r

ction of pe

594 nm, 3s

for P38-40

s a function

2/3 and ips

and cont

function of

and contr

and contro

d,e) for P3

B

pyramidal

mp, 473 nm

right, scat

ak frequen

s). (b) Sam

mice (con

n of peak fr

silateral L5

trol stimula

f peak freq

ralateral L2

ol stimulat

38-40 mice

Bitzenhofer

neurons i

m, 3s) for co

tter plot d

ncy. Bottom

me as (a) fo

ntrol n=9, E

requency f

5/6 in the

ation (594

quency for

2/3 in the

tion (594 n

. *P<0.05,

r et al.

17

n mPFC.

ontrol and

displaying

m left and

or P23-25

ES n=12).

for the MI

mPFC of

nm). (e)

the MI of

mPFC of

nm). (f,g)

**P<0.01



https://doi.org/10.1101/558957



18

and ***P<0.001. See Supplementary Tab. 1 for detailed information on statistics. 364

Average data is presented as median ± MAD. 365



https://doi.org/10.1101/558957


366

367

368

369

370

371

372

373

374

375

376

377

378

379

380

381

Gamma

Fig. 4.

pyrami

Bottom

control

Bottom

control

Bottom

(n=19)

memory

in an 8

matrix (

machin

combin

conditio

informa

a control of

Poorer co

idal neuro

, discrimin

(n=28) an

, discrimin

(n=28) an

, discrimina

and ES m

y (conditio

8-arm radia

(top) and d

e classifica

ed behav

on. *P<0.0

ation on sta

f cognitive

ognitive a

ons in mP

nation ratio

nd ES mic

ation ratio

nd ES mic

ation ratio

mice (n=2

n p=0.001)

al maze w

decision sp

ation used

ioral perfo

5, **P<0.0

atistics. Ave

maturation

nd social

PFC. (a) T

o of interac

ce (n=30).

of interact

ce (n=30).

of interacti

1). (d) Sp

) performa

ith 4 baite

pace (botto

d to predic

ormance.

01 and ***P

erage data

n

performa

Top, schem

ction time

(b) Top,

tion time w

(c) Top, s

ion time wi

patial work

nce for P2

ed arms ov

om) as well

t if an anim

Fill color

P<0.001. S

a is present

ance of mi

matic of n

with a no

schematic

with a less-

schematic

ith a mothe

king (condi

3-36 contr

ver 14 con

l as single

mal belong

of single

See Suppl

ted as med

B

ice experi

novel objec

ovel-to-fam

of recenc

-to-more re

of matern

er-to-empty

ition p=0.0

rol (n=10) a

nsecutive d

data point

gs to contr

data poin

ementary

dian ± MAD

Bitzenhofer

encing ES

ct recognit

miliar object

cy recognit

ecent objec

nal interact

y bin for P2

007) and r

and ES mic

days. (e) C

ts for suppo

rol or ES b

nts represe

Tab. 1 for

D.

r et al.

19

S of L2/3

tion task.

t for P17

tion task.

ct for P22

tion task.

21 control

reference

ce (n=12)

Confusion

ort vector

based on

ents true

r detailed



https://doi.org/10.1101/558957



20

List of supplementary materials 383

384

Methods 385

386

Fig. S1 (related to Fig. 1): Spontaneous and L2/3-driven activity in awake and 387

anesthetized mice across development 388

389

Fig. S2 (related to Fig. 1): Neurochemical profile and firing patterns in PFC across 390

development. 391

392

Fig. S3 (related to Fig. 3): ES effects on spontaneous and L2/3 pyramidal neuron-393

driven activity in anesthetized and awake head-fixed mice across development 394

395

Fig. S4 (related to Fig. 4): Developmental milestones and behavioral performance after 396

ES of L2/3 pyramidal neurons in the mPFC 397

398

Supplementary Tab. 1: Statistics summary 399

400



https://doi.org/10.1101/558957



1

Supplementary materials for: 1

2

3

Activity-dependent maturation of prefrontal gamma 4

oscillations sculpts cognitive performance 5

Sebastian H. Bitzenhofer1,*, Jastyn A. Pöpplau1, Mattia Chini1, Annette Marquardt1, 6

Ileana L. Hanganu-Opatz1,* 7

1 Developmental Neurophysiology, Institute of Neuroanatomy, University Medical Center 8

Hamburg-Eppendorf, Hamburg, Germany 9

10

11

* Corresponding authors: Ileana L. Hanganu-Opatz 12



15

Sebastian H. Bitzenhofer 16



19



https://doi.org/10.1101/558957



2

List of supplementary materials 20

21

Methods 22

23

Supplementary Fig.1 (related to Fig. 1): Spontaneous and L2/3-driven activity in 24

awake and anesthetized mice across development 25

26

Supplementary Fig.2 (related to Fig. 1): Neurochemical identity and firing patterns in 27

PFC across development. 28

29

Supplementary Fig.3 (related to Fig. 3): ES effects on spontaneous and L2/3 30

pyramidal neuron-driven activity in anesthetized and awake head-fixed mice across 31

development 32

33

Supplementary Fig.4 (related to Fig. 4): Developmental milestones and behavioral 34

performance after ES of L2/3 pyramidal neurons in the mPFC 35

36

Supplementary Tab. 1: Statistics summary 37

38



https://doi.org/10.1101/558957



3

Methods 39

Animals 40

All experiments were performed in compliance with the German laws and the guidelines 41

of the European Community for the use of animals in research and were approved by 42

the local ethical committee (G132/12, G17/015, N18/015). Experiments were carried out 43

on C57Bl/6J mice of both sexes. To achieve interneuron-specific expression of 44

archaerhodopsin-3 (ArchT), Gad2 driver line (Gad2-IRES-Cre knock-in, The Jackson 45

Laboratory, ME, USA) was crossed with Ai40 reporter line (Ai40(RCL-ArchT/EGFP)-D, 46

The Jackson Laboratory, ME, USA). Timed-pregnant mice from the animal facility of the 47

University Medical Center Hamburg-Eppendorf were housed individually at a 12 h 48

light/12 h dark cycle and were given access to water and food ad libitum. The day of 49

vaginal plug detection was considered E0.5, the day of birth was considered P0. 50

In utero electroporation 51

Pregnant mice received additional wet food on a daily basis, supplemented with 2-4 52

drops Metacam (0.5 mg/ml, Boehringer-Ingelheim, Germany) one day before until two 53

days after in utero electroporation. At E15.5, pregnant mice were injected 54

subcutaneously with buprenorphine (0.05 mg/kg body weight) 30 min before surgery. 55

Surgery was performed under isoflurane anesthesia (induction 5%, maintenance 3.5%) 56

on a heating blanket. Eyes were covered with eye ointment and pain reflexes and 57

breathing were monitored to assess anesthesia depth. Uterine horns were exposed and 58

moistened with warm sterile PBS. 0.75-1.25 µl of opsin- and fluorophore-encoding 59

plasmid (pAAV-CAG-ChR2(E123T/T159C)-2A-tDimer2, 1.25 µg/µl) purified with 60

NucleoBond (Macherey-Nagel, Germany) in sterile PBS with 0.1% fast green dye was 61

injected in the right lateral ventricle of each embryo using pulled borosilicate glass 62



https://doi.org/10.1101/558957



4

capillaries. Electroporation tweezer paddles of 5 mm diameter were oriented at a rough 63

20° leftward angle from the midline of the head and a rough 10° downward angle from 64

the anterior to posterior axis to transfect precursor cells of medial prefrontal L2/3 65

pyramidal neurons with 5 electroporation pulses (35 V, 50 ms, 950 ms interval, CU21EX, 66

BEX, Japan). Uterine horns were placed back into the abdominal cavity. Abdominal 67

cavity was filled with warm sterile PBS and abdominal muscles and skin were sutured 68

with absorbable and non-absorbable suture thread, respectively. After recovery from 69

anesthesia, mice were returned to their home cage, placed half on a heating blanket for 70

two days after surgery. Fluorophore expression was assessed at P2 in the pups with a 71

portable fluorescence flashlight (Nightsea, MA, USA) through the intact skin and skull 72

and confirmed in brain slices post mortem. 73

Early stimulation (ES) 74

A stimulation window was implanted at P7 for chronic transcranial optogenetic 75

stimulation in mice transfected by in utero electroporation. Mice were placed on a 76

heating blanket and anesthetized with isoflurane (5% induction, 2% maintenance). 77

Breathing and pain reflexes were monitored to assess anesthesia depth. The skin above 78

the skull was cut along the midline (3 mm) at the level of the mPFC and gently spread 79

with a forceps, before covering the incision with transparent tissue adhesive (Surgibond, 80

SMI, Belgium). Mice were returned to the dam in the home cage after recovery from 81

anesthesia. From P7-11 mice were stimulated daily under isoflurane anesthesia (5% 82

induction, 2% maintenance) with ramp stimulations of linearly increasing light power 83

(473 nm wavelength, 3 s duration, 7 s interval, 180 repetitions, 30 min total duration). 84

Light stimulation was performed using an Arduino uno (Arduino, Italy) controlled laser 85

system (Omicron, Austria) coupled to a 200 µm diameter light fiber (Thorlabs, NJ, USA) 86



https://doi.org/10.1101/558957



5

positioned directly above the tissue adhesive window. Light power attenuation was set to 87

reach 10 mW in the brain, adjusted for measured light attenuation by the tissue adhesive 88

(~30%) and by the immature skull (~25%). Control animals were treated identical, but 89

stimulated with light of 594 nm wavelength that does not activate the expressed opsin 90

ChR2(ET/TC). 91

Optogenetics and electrophysiology in vivo 92

Acute recordings. Multi-site extracellular recordings of local field potential (LFP) and 93

multi-unit activity (MUA) were performed unilaterally or bilaterally in the mPFC of non-94

anesthetized or anesthetized P5-40 mice. Pups were on a heating blanket during the 95

entire procedure. Under isoflurane anesthesia (induction: 5%; maintenance: 2.5%), a 96

craniotomy was performed above the mPFC (0.5 mm anterior to bregma, 0.1-0.5 mm 97

lateral to the midline). Neck muscles were cut and 0.5% bupivacaine / 1% lidocaine was 98

locally applied to cutting edges. Pups were head-fixed into a stereotaxic apparatus using 99

two plastic bars mounted on the nasal and occipital bones with dental cement. Multi-site 100

electrodes (NeuroNexus, MI, USA) were inserted into the mPFC (four-shank, A4x4 101

recording sites, 100 µm spacing, 125 µm shank distance, 1.8-2.0 mm deep). A silver 102

wire was inserted into the cerebellum and served as ground and reference. Pups were 103

allowed to recover for 30 min prior to recordings. For anesthetized recordings, urethane 104

(1 mg/g body weight) was injected intraperitoneally prior to the surgery. Extracellular 105

signals were band-pass filtered (0.1-9,000 Hz) and digitized (32 kHz) with a multichannel 106

extracellular amplifier (Digital Lynx SX; Neuralynx, Bozeman, MO, USA). Electrode 107

position was confirmed in brain slices post mortem. 108

Chronic recordings. Multisite extracellular recordings were performed in the mPFC of 109

P23-25 and P38-40 mice. The adapter for head fixation was implanted at least 5 days 110



https://doi.org/10.1101/558957



6

before recordings. Under isoflurane anesthesia (5% induction, 2.5% maintenance), a 111

metal head-post (Luigs and Neumann, Germany) was attached to the skull with dental 112

cement and a craniotomy was performed above the mPFC (0.5-2.0 mm anterior to 113

bregma, 0.1-0.5 mm right to the midline) and protected by a customized synthetic 114

window. A silver wire was implanted between skull and brain tissue above the 115

cerebellum and served as ground and reference. 0.5% bupivacaine / 1% lidocaine was 116

locally applied to cutting edges. After recovery from anesthesia, mice were returned to 117

their home cage. After recovery from the surgery, mice were accustomed to head-118

fixation and trained to run on a custom-made spinning disc. For recordings, craniotomies 119

were uncovered and multi-site electrodes (NeuroNexus, MI, USA) were inserted into the 120

mPFC (one-shank, A1x16 recording sites, 100 µm spacing, 2.0 mm deep). Extracellular 121

signals were band-pass filtered (0.1-9000 Hz) and digitized (32 kHz) with a multichannel 122

extracellular amplifier (Digital Lynx SX; Neuralynx, Bozeman, MO, USA). Electrode 123

position was confirmed in brain slices post mortem. 124

Optogenetic stimulation. Pulsed (light on-off) and ramp (linearly increasing light power) 125

light stimulation was performed using an Arduino uno (Arduino, Italy) controlled laser 126

system (473 nm / 594 nm wavelength, Omicron, Austria) coupled to a 50 µm (4 shank 127

electrodes) or 105 µm (1 shank electrodes) diameter light fiber (Thorlabs, NJ, USA) 128

glued to the multisite electrodes, ending 200 µm above the top recording site. 129

Histology 130

P5-40 mice were anesthetized with 10% ketamine (aniMedica, Germanry) / 2% xylazine 131

(WDT, Germany) in 0.9% NaCl (10 µg/g body weight, intraperitoneal) and transcardially 132

perfused with 4% paraformaldehyde (Histofix, Carl Roth, Germany). Brains were 133

removed and postfixed in 4% paraformaldehyde for 24 h. Brains were sectioned 134



https://doi.org/10.1101/558957



7

coronally with a vibratom at 50 µm for immunohistochemistry or 100 µm for examination 135

of dendritic complexity. 136

Immunohistochemistry. Free-floating slices were permeabilized and blocked with PBS 137

containing 0.8% Triton X-100 (Sigma-Aldrich, MO, USA), 5% normal bovine serum 138

(Jackson Immuno Research, PA, USA) and 0.05% sodium azide. Slices were incubated 139

over night with primary antibody rabbit-anti-GABA (1:1000, #A2052, Sigma-Aldrich, 140

MMO, USA), rabbit-anti-Ca2+/calmodulin-dependent protein kinase II (1:200, #PA5-141

38239, Thermo Fisher, MA, USA; 1:500, #ab52476, Abcam, UK), rabbit-anti-142

parvalbumin (1:500, #ab11427, Abcam, UK) or rabbit-anti-somatostatin (1:250, 143

#sc13099, Santa Cruz, CA, USA), followed by 2 h incubation with secondary antibody 144

goat-anti-rabbit Alexa Fluor 488 (1:500, #A11008, Invitrogen-Thermo Fisher, MA, USA) 145

or goat-anti-rat Alexa Fluor 488 (1:750, #A11006, Invitrogen-Thermo Fisher, MA, USA). 146

Sections were transferred to glass slides and covered with Fluoromount (Sigma-Aldrich, 147

MO, USA). 148

Cell quantification. Images of immunofluorescence in the right mPFC as well as IUE-149

induced tDimer2 expression were acquired with a confocal microscope (DM IRBE, 150

Leica, Germany) using a 10x objective (numerical aperture 0.3). tDimer2-positive and 151

immunopositive cells were automatically quantified with custom-written algorithms in 152

ImageJ environment. The region of interest (ROI) was manually defined over L2/3 of the 153

mPFC. Image contrast was enhanced before applying a median filter. Local background 154

was subtracted to reduce background noise and images were binarized and segmented 155

using the watershed function. Counting was done after detecting the neurons with the 156

extended maxima function of the MorphoLibJ plugin. 157



https://doi.org/10.1101/558957



8

Dendritic complexity. Image stacks of tDimer2-positive neurons were acquired with a 158

confocal microscope (LSN700, Zeiss, Germany) using a 40x objective. Stacks of 6 159

neurons per animal were acquired as 2048x2048 pixel images (voxel size 156*156*500 160

nm). Dendritic complexity was quantified by Sholl analysis in ImageJ environment. 161

Images were binarized using auto threshold function and the dendrites were traced 162

using the semi-automatic simple neurite tracer plugin. The geometric center was 163

identified and the traced dendritic tree was analyzed with the Sholl analysis plugin. 164

Behavior 165

Mice were handled and adapted to the room of investigation daily starting two days prior 166

to behavioral examination. Arenas and objects were cleaned with 0.1% acetic acid 167

before each trial. Animals were tracked automatically (Video Mot2, TSE Systems GmbH, 168

Germany). 169

Developmental milestones. Somatic and reflex development was examined from P2-20 170

at a 3-day interval. Weight, body length and tail length were measured. Grasping reflex 171

was assessed by touching front paws with a toothpick. Vibrissa placing was measured 172

as head movement in response to gently touching the vibrissa with a toothpick. Auditory 173

startle was assessed in response to finger snapping. The days of pinnae detachment 174

and eye opening were monitored. Surface righting was measured as time to turn around 175

after being positioned on the back (max 30 s). Cliff avoidance was measured as time 176

until withdrawing after being positioned with forepaws and snout over an elevated edge 177

(max 30 s). Bar holding was measured as time hanging on an toothpick grasped with the 178

forepaws (max 10 s). 179

Open field. Each mouse was positioned in the center of a circular arena individually (34 180

cm in diameter) at P16 for 10 min. Behavior was quantified by measuring: discrimination 181



https://doi.org/10.1101/558957



9

index of time spent in the center and the border of the arena ((time in surround - time in 182

center) / (time in surround + time in center)), grooming time, average velocity and 183

number of rearing, wall rearing and jumping. 184

Object recognition. Novel object recognition (NOR, P17), object location recognition 185

(OLR, P18) and recency recognition (RR, P21) were performed in the same arena as 186

the open field examination. Mouse center, tail and snout position were tracked 187

automatically. Object interaction was defined as the snout being within <1 cm distance 188

from an object. For NOR, each mouse explored two identical objects for 10 min during 189

the sample phase. After a delay period of 5 min in a break box, the mouse was placed 190

back in the arena for the test phase where one of the objects was replaced by a novel 191

object. Behavior was quantified as discrimination index of time spent interacting with the 192

novel and familiar object ((time novel object - time familiar object) / (time novel object + 193

time familiar object)). OLR was performed similarly, but one object was relocated for the 194

test phase instead of being exchanged. For RR, each mouse explored two identical 195

objects during the first sample phase for 10 min, followed by a delay phase of 30 min, 196

and a second sample phase of 10 min with two novel identical objects. After a second 197

break of 5 min, time interacting with an object of the first sample phase (old) and an 198

object from the second sample phase (recent) was assessed during the test phase for 2 199

min. Behavior was quantified as discrimination index of time spent interacting with the 200

novel and familiar object ((time old object – time recent object) / (time old object + time 201

recent object)). 202

Maternal interaction. Maternal interaction was performed at P21 in the same arena as 203

the open field examination with two plastic containers, one empty and one containing the 204

dam of the mouse pup examined. Small holes in the containers allowed the pup and the 205



https://doi.org/10.1101/558957



10

mother to interact. Behavior was quantified as discrimination index of time spent 206

interacting with the empty container and the container containing the mother ((time 207

mother container – time empty container) / (time mother container + time empty 208

container)). 209

Spontaneous alteration. At P15, 18 and 21, each mouse was positioned on the central 210

arm of an elevated t-maze. After 1 min, the mouse had access to one of the other arms. 211

The mouse was placed back in the start arm for 1 min, before a second run. Behavior 212

was quantified as alteration or no alteration between the two arms in the first and second 213

run. 214

Tail suspension. Mice were fixed with their tail on a bar 30 cm above ground for 5 min at 215

P21. Behavior was quantified as time spent inactive, passively hanging. 216

Spatial working memory. Mice were positioned in the center of an elevated 8-arm radial 217

maze daily from P23-36. Arms were without walls and 4 arms contained a food pellet at 218

the distal end (baited). On the first day, mice were allowed to examine the maze for 20 219

min or until all arms were visited. During the following days, mice were allowed to 220

examine the maze until all baited arms were visited, but for max 20 min and arm entries 221

were assessed. Visit of a non-baited arm was considered as reference memory error, 222

repeated visit of the same arm in one trial as working memory error. 223

Data analysis 224

In vivo data were analyzed with custom-written algorithms in Matlab environment. Data 225

were band-pass filtered (500-9000 Hz for spike analysis or 1-100 Hz for LFP) using a 226

third-order Butterworth filter forward and backward to preserve phase information before 227

down-sampling to analyze LFP. 228



https://doi.org/10.1101/558957



11

Power spectral density. For power spectral density analysis, 2 s-long windows of LFP 229

signal were concatenated and the power was calculated using Welch’s method with non-230

overlapping windows. Spectra were multiplied with squared frequency. 231

Imaginary coherence. The imaginary part of complex coherence, which is insensitive to 232

volume conduction, was calculated by taking the absolute value of the imaginary 233

component of the normalized cross-spectrum. 234

Modulation index. For optogenetic stimulations, modulation index was calculated as 235

(value stimulation - value pre stimulation) / (value stimulation + value pre stimulation). 236

Peak frequency and strength. Peak frequency and peak strength were calculated for the 237

most prominent peak in the spectrum defined by the product of peak amplitude, peak 238

half width and peak prominence. 239

Single unit analysis. Spikes were detected and sorted with Ultra Mega Sort 2000 240

software in Matlab. 241

Support vector classification. Model training and performance evaluation were carried 242

out using the scikit-learn toolbox in Python. The set was iteratively (n=500) divided using 243

5-fold cross-validation in a training (4/5) and a test (1/5) set. The value of the model 244

regularization parameter “C” was tuned on the training set, which was further split using 245

3-fold cross-validation. Model prediction was assessed on the test set. Performance was 246

stable across a wide range of regularization parameter values. To plot the classifier 247

decision space, we used t-sne to reduce the feature space to two dimensions, while 248

preserving the hyper-dimensional structure of the data. The decision space was then 249

approximated by imposing a Voronoi tessellation on the 2d plot, using k-nearest 250

regression on the t-sne coordinates of the predicted classes of the mice. 251



https://doi.org/10.1101/558957



12

Statistics. Statistical analyses were performed in the Matlab environment or in R 252

Statistical Software (Foundation for Statistical Computing, Austria). Data are presented 253

as median ± median absolute deviation (MAD). Data were tested for significant 254

differences (*P<0.05, **P<0.01 and ***P<0.001) using non-parametric Wilcoxon rank 255

sum test for unpaired and Wilcoxon signed rank test for paired data or Kruskal-Wallis 256

test with Bonferroni corrected post hoc analysis. Nested data were analyzed with linear 257

mixed-effect models considering within animal variance with Turkey multi comparison 258

correction for post hoc analysis. More information about statistic results are provided in 259

Supplementary Tab. 1. 260



https://doi.org/10.1101/558957



13

Supplementary figures 261

262

Supplementary Fig.1 (related to Fig. 1). Spontaneous and L2/3-driven activity in 263

awake and anesthetized mice across development. (a) Setup for recordings from 264

awake head-fixed mice on a spinning disc with one shank 1x16 electrode in L2/3. (b) 265

Example LFPs from mPFC of non-anesthetized mice at different ages. (c) Setup for 266

recordings from anesthetized head-fixed mice with four shank 4x4 electrode. (d) 267

Example LFPs from mPFC of anesthetized mice at different ages. (e) Left, average 268

power spectra of prefrontal activity for awake (n=6, 6 recordings) and anesthetized 269

(n=10) P11-12 mice. Right, scatter plot displaying corresponding peak strength as a 270

function of peak frequency. (f) Same as (e) for P23-25 mice (awake n=5, 13 recordings; 271



https://doi.org/10.1101/558957



14

anesthetized n=10). (g) Same as (e) for P38-40 mice (awake n=5, 12 recordings; 272

anesthetized n=9). (h) Example prefrontal LFPs driven by ramp light stimulation of L2/3 273

pyramidal neurons in non-anesthetized mice at different ages. (i) Example prefrontal 274

LFPs driven by ramp light stimulation of L2/3 pyramidal neurons in anesthetized mice at 275

different ages. (j) Scatter plots displaying peak strength as a function of peak frequency 276

for the MI of power spectra during-to-before stimulation (ramp, 473 nm or 594 nm) in the 277

mPFC of awake (n=6, 6 recordings) and anesthetized (n=10) P11-12 mice. (k) Same as 278

(j) for P23-25 mice (awake n=5, 13 recordings; anesthetized n=10). (l) Same as (j) for 279

P38-40 mice (awake n=5, 12 recordings; anesthetized n=9). *P<0.05, **P<0.01 and 280

***P<0.001. See Supplementary Tab. 1 for detailed information on statistics. Average 281

data is presented as median ± MAD. 282



https://doi.org/10.1101/558957



15

283

Supplementary Fig.2 (related to Fig. 1). Neurochemical identity and firing patterns 284

in PFC across development. (a) Left, examples of SST immunostaining of prefrontal 285

neurons from P6, P15, P26, and P34 mice. Right, scatter plot displaying the density of 286

SST-immunopositive neurons in the mPFC of P5-40 mice (n=39). (b) Same for PV-287

immunopositive neurons (n=38). (c) Scatter plot displaying prefrontal SUA firing rate for 288

P5-40 mice (806 units of 71 mice). (d) Same for SUA amplitude. (e) Same for unit half 289

width. (f) Example of EGFP-positive (arrow) and RFP-positive cells showing no overlap 290



https://doi.org/10.1101/558957



16

in expression. (g) Top, MI of power spectra during-to-before stimulation (ramp, 473 nm) 291

of L2/3 pyramidal neurons (ramp, 473 nm), inhibition of interneurons (square pulse, 594 292

nm) or a combination of both in the mPFC of P11-12 mice (n=5). Bottom, scatter plot 293

displaying corresponding peak strength as a function of peak frequency. (h) Same as (g) 294

for P23-25 mice (n=6). (i) Same as (g) for P38-40 mice (n=6). *P<0.05, **P<0.01 and 295





https://doi.org/10.1101/558957



17

298

Supplementary Fig.3 (related to Fig. 3). ES effects on spontaneous and L2/3 299

pyramidal neuron-driven activity in anesthetized and awake head-fixed mice 300

across development. (a) Left, average power spectra of spontaneous mPFC activity for 301

anesthetized head-fixed P11-12 control (n=10) and ES mice (n=10). Right, scatter plot 302

displaying corresponding peak strength as a function of peak frequency. (b) Same as (a) 303

for P23-25 mice (control n=10, ES n=11). (c) Same as (a) for P38-40 mice (control n=9, 304

ES n=12). (d) Same as (b) for awake head-fixed control (n=6, 13 recordings) and ES 305

mice (n=5, 14 recordings) on a spinning disc. (e) Same as (c) for awake head-fixed 306

control (n=5, 12 recordings) and ES mice (n=5, 12 recordings) on a spinning disc. (f) 307

Top left, MI of power spectra during-to-before stimulation (ramp, 473 nm) of L2/3 308

pyramidal neurons in the mPFC of P23-25 awake control (n=6, 13 recordings) and ES 309

mice (n=5, 14 recordings). Top right, scatter plot displaying corresponding peak strength 310



https://doi.org/10.1101/558957



18

as a function of peak frequency. Bottom, same as top for control stimulation (ramp, 594 311

nm) (g) Same as (f) for P38-40 mice (control n=5, 12 recordings; ES n=5, 12 312

recordings). *P<0.05, **P<0.01 and ***P<0.001. See Supplementary Tab. 1 for detailed 313

information on statistics. Average data is presented as median ± MAD. 314



https://doi.org/10.1101/558957



19

315

Supplementary Fig.4 (related to Fig. 4). Developmental milestones and behavioral 316

performance after ES of L2/3 pyramidal neurons in the mPFC. (a) Age dependence 317

of developmental milestones for control (n=11) and ES mice (n=11). (b) Left, schematic 318

of open field task (top) and discrimination ratio of time spent in border-to-center in an 319

open field (bottom). Right, behavioral quantification during exploration averaged for P16 320

control (n=28) and ES mice (n=30). (c) Top, schematic of object location recognition 321

task. Bottom, discrimination ratio of interaction time with an object in a novel-to-familiar 322

location averaged for P18 control (n=28) and ES mice (n=30). (d) Spontaneous 323

alteration in t-maze test for control (n=19) and ES mice (n=21) at P15, 18 and 21. (e) 324

Tail suspension test for P21 control (n=19) and ES mice (n=21). *P<0.05, **P<0.01 and 325





https://doi.org/10.1101/558957


Activity-dependent maturation of prefrontal gamma ...Gamma control of cognitive maturation...

Documents

Transcript of Activity-dependent maturation of prefrontal gamma ...Gamma control of cognitive maturation...