Layer-specific excitatory circuits differentially control recurrent network dynamics in the neocortex Riccardo Beltramo, Giulia D’Urso, Marco Dal Maschio, Pasqualina Farisello, Serena Bovetti, Yoanne Clovis, Davide De Pietri Tonelli, Tommaso Fellin#
Department of Neuroscience and Brain Technologies, Istituto Italiano di Tecnologia, Via Morego 30, 16163 Genova, Italy.
# Corresponding author: Tommaso Fellin, Department of Neuroscience and Brain Technologies, Istituto Italiano di Tecnologia, Via Morego 30, 16163 Genova, Italy , tel: +39 010 71781549, fax:+39 010 71781230, email:
[email protected]
Keywords: neocortex, cortical layers, pyramidal neurons, spontaneous activity, slow oscillations, up- and down-states.
Number of pages: 44 Number of figures: 7 Number of words in the abstract: 125
1
Abstract In the absence of external stimuli, the mammalian neocortex displays intrinsic network oscillations. These dynamics are characterized by translaminar assemblies of neurons whose activity synchronizes rhythmically in space and time. How different cortical layers influence the formation of these spontaneous cellular assemblies is poorly understood. Here we show that excitatory neurons in supragranular and infragranular layers play distinct roles in the regulation of intrinsic low-frequency oscillations in vivo. Optogenetic activation of infragranular neurons generates network activity that resembles spontaneous events, while photoinhibition of these same
neurons
significantly
attenuates
slow
ongoing
dynamics.
In
contrast,
light
activation/inhibition of supragranular cells has modest effects on spontaneous slow activity. This study represents the first causal demonstration that excitatory circuits located in distinct cortical layers differentially control spontaneous low-frequency dynamics.
Introduction The mammalian neocortex is a complex cellular network that is organized in up to 6 horizontal layers 1. Each of these layers contains neurons with distinct anatomical and functional properties 2-4
. Moreover, cortical cells are highly interconnected with each other through excitatory
projections both within and across cortical layers 5-8. One feature of this elaborate circuitry is that in the absence of inputs coming from other brain areas, it displays intrinsic dynamics
9-12
. A
prototypical example of this phenomenon is represented by sensory cortices that, even in the complete absence of incoming sensory signals (for example, during quiet wakefulness or sleep), show recurrent spontaneous activities
13-16
. A major component of this intrinsic dynamics is 2
represented by the slow oscillation individual neurons recordings
20
13, 19, 21
17-20
, a P30) Rbp4Cre mice. Newborn mice were deeply anaesthetized by hypothermia (total duration, 15-20 min) and their heads were immobilized on a custom neonatal stereotaxic apparatus maintained at 4°C during surgery. The skull was exposed by a skin incision and 300–400 nl of either ChR2 or a combination of Arch and eNpHR (1:1) viruses was injected at stereotaxic coordinates of 0 mm 17
bregma, 1.5 mm lateral to sagittal sinus, and 0.3 mm depth, by means of a glass micropipette. After the micropipette was removed, the skin was sutured. The pups were quickly revitalized under a heat lamp and subsequently returned to the dam. Adult animals were anesthetized with 2% isoflurane and positioned into a stereotaxic apparatus (Stoelting Co, Wood Dale, IL). Mice were maintained on a warm platform at 37°C to maintain the body temperature constant during anesthesia. The skull was exposed by a skin incision and a small hole was drilled through the skull at stereotaxic coordinates 1.4 mm posterior to bregma and 2.5 mm lateral to the sagittal sinus. Over a period of 30 min, 1 µl of ChR2 or a combination of Arch and eNpHR viruses (0.5 µl each) was injected at 700 µm below the surface of the scalp. After surgery, the animals were left under a heat lamp and monitored until recovery.
In utero electroporation Pregnant CD1 and C57BL/6J mice (Charles River, Calco, Italy) 15.5 days postcoitum (vaginal plug day was defined as 0.5) were anesthetized with 2 % isoflurane. The abdomen was shaved and swabbed with 70% ethanol and povidone-iodine. A midline ventral laparotomy (3 cm) was performed and the uterus was exposed and moistened with PBS pre-warmed at 37°C. Using glass micropipettes and PV820 Pneumatic picopump (WPI, Sarasota, FL), 1-2Ԝμl of DNA solution in PBS and 5% Fast Green FCF (Sigma-Aldrich, Milan, Italy) were injected through the uterine wall into the lateral ventricle of each embryo. Animals were injected with pCAGGS-mCherry 55 (0.8 μg/μl) along with either pCAGGS-ChR2-Venus
34
(0.8 μg/μl, Addgene #15753) or with a
combination of pCAGGS-Arch-GFP and pAAV-CAG-ArchT-GFP 56 (Addgene #29777) (each at concentration of 0.8 μg/μl). pCAGGS-Arch-GFP was subcloned from FCK-Arch-GFP
36
(Addgene #22217). Approximately 89 % of mCherry-positive cells were also opsin-positive.
18
Immediately after the injections, using BTX–ECM 830 electroporator (Harvard Apparatus, Holliston, MA), 6 square electrical pulses of 30 V amplitude and 50 ms duration were delivered to the embryo at 1 Hz through forceps-type circular electrodes (5 mm diameter) positioned at 0° angle with respect to the rostral–caudal axis of the head. After the electroporation, the uterine horns were returned to the peritoneal cavity and the abdominal wall and the skin were sutured.
Surgery and optical stimulation. Mice (> P45) were anesthetized with urethane (2 g/kg) and placed on a stereotaxic apparatus. Body temperature was measured with a rectal probe and kept at 37 °C with a heating pad. Depth of anesthesia was assured by monitoring respiration rate, heartbeat, eyelid reflex, vibrissae movements, reactions to tail and toe pinching. Oxygen saturation was controlled by a pulseoxymeter (MouseOx, Starr Life Sciences Corp., Oakmont, PA). All incisions were infiltrated with lidocaine. A craniotomy was performed over the neocortex (most experiments were performed in the sensory-motor areas) at least 300 μm away from the site of injection. The surface of the brain was continuously kept moist with normal HEPES-buffered artificial cerebrospinal fluid (ACSF). The dura was carefully dissected with a metal forceps only when extracellular recordings were performed. Photostimulation and photoinhibition were performed with a 473 nm (World Star Tech, Toronto, Canada) and a 590 nm (Cobolt, Vretenvägen, Sweden) continuous wave, solid-state laser sources, respectively. Lasers were controlled by a command voltage, either directly with a TTL signal for the 473 nm laser or via an analog signal using an acousto-optic modulator (Gooch & Housego, USA) for the 590 nm laser. Light pulses were delivered to the brain via an optical fiber (amsTechnologies, Milan, Italy) positioned close to the surface of the cortex and parallel to the recording electrode. This experimental configuration allowed the excitation of the
19
apical dendrites of opsin-expressing cells while minimizing the generation of laser artifacts in the LFP frequency band 57. Light intensity was adjusted with neutral density filters (ThorLabs, USA). For experiments described in supplementary fig 6, 8, 9 the fiber optic was placed few millimeters away from the cortical surface. Blue light was presented at different intensities ~ 0.2, ~ 2, ~ 5, ~ 7, ~ 13, ~ 18 mW (measured at the fiber tip) but, for most experiments involving stimulation of layer V, light power was between 2 and 7 mW. Blue light stimulation of layer II/III neurons in vivo was performed in the 0.2-18 mW power range, but quantification was performed only for higher light intensities (13-18 mW) to precisely control for the apparent weaker effect of layer II/III neurons compared to layer V. Yellow light was presented at intensity ~ 30 mW (measured at the fiber tip) in all experiments.
In vivo extracellular recordings and analysis. LFP and MU were recorded with custom-built electrodes (FHC Inc., Bowdoin, ME, distance between the tips ~ 200-250 μm)58. Electrodes were inserted normally to the brain surface with the tip of the more superficial electrode placed at depth ~ 450-550 μm. Electrical signals were filtered at 0.1 Hz - 5 kHz, amplified by an AMamplifier (AM-system, Carlsborg, WA) and digitized at 50 kHz. For LFP analysis, signals were re-sampled (1 KHz) and low-pass filtered (100 Hz). Light-induced artifacts in the LFP signal were measured at the end of the experiments. Animals were sacrificed with an intracardiac bolus of urethane eliminating brain electrical activity, while maintaining the relative positions between the recording electrode, the optical fiber and the brain unchanged. Previous stimulation protocols were repeated under these experimental conditions and “laser artifacts” were eliminated by subtraction. Analysis of LFP and MU signals was performed with custom code (Matlab, Natick, MA). Spontaneous down- to up-state transitions were identified as the time point in which the 20
LFP signal crossed a threshold set five times the standard deviation of the noise above baseline. Only transitions in which the LFP signal remained for more than 150 ms above threshold were considered. Up- to down-state transitions were identified as the time points in which the LFP signal crossed the threshold and remained below the threshold for longer than 100 ms. Only upstates preceded and followed by down-states of 300 ms or longer duration were selected for spectral and spike content analysis. Time windows (300 ms duration) preceding and following an identified up-state were called “pre” and “post”, respectively. The spectral content of the upstates was quantified by calculating the short-time Fourier transform over a moving Hamming window of 100 ms (overlap 99%). The average power within the frequency bands of 30-60 Hz (low gamma) and 60-90 Hz (high gamma) was extracted from the resulting spectrogram in the three different time windows before during and after the up-state (“pre”, “up” and “down”, respectively), excluding the power contribution in the first and the last 25% of each window duration. Power values were normalized to the total power in the “pre” period. For light-evoked events, three time windows were identified before, during and after the stimulation (“pre”, “stim” and “post”, respectively). The duration of the pre and post periods was set at 300 ms. For analysis in figures 5 and 6, only light stimulation episodes that were preceded by an ongoing upstate in the 300 ms before the stimulus were considered. In figures 3-4 and supplementary figures 8-9, power spectra were calculated by computing the Fast Fourier Transform in rectangular windows (window duration 4096 points, 50% overlap). When rhythmic laser pulses were applied, the power values calculated in the periods before, during and after the stimulation were normalized to the total power in the period before the stimulation. MU signals were high-pass filtered at 300 Hz. Data were first rectified and spike events were identified by binarizing the data according to a threshold set at five times the standard deviation of the noise (experimental
21
points with a value above the threshold were given value 1, otherwise 0). Spikes were identified using a first-derivative spike-sorting algorithm. The time axis was then binned (bin value 10 ms) and the peri-stimulus time histogram (PSTH) generated by plotting the number of spikes per bin over the recording time.
In vivo patch-clamp recordings and analysis. Patch-clamp recordings were performed as described in
58, 59
. 3-5 MΩ glass pipette were used as recording electrodes and filled with the
following intrapipette solution: K-gluconate 140, MgCl2 1, NaCl 8, Na2ATP 2, NaGTP 0.5, HEPES 10, phophocreatine 10 to pH 7.2 with KOH. In some experiments, biocytin 3 mg/ml was included in the pipette solution for subsequent cell identification. Electrical signals were amplified by a Multiclamp 700 B, filtered at 10 kHz, digitized at 50 kHz with a digidata 1440 and stored with pClamp 10 (Axon instruments, Union City, CA). Spontaneous down- to up-state transitions were identified as membrane depolarizations crossing a threshold set at the average resting membrane potential plus 7 mV, remaining above the threshold for longer than 150 ms. Up- to down-state transition were identified as membrane hyperpolarization for which the membrane voltage crossed the threshold and remained below it for longer than 100 ms. The start and end times of the up-state were defined as the time coordinates at which the membrane potential crossed the threshold. The area of the membrane depolarization and the spike number during the up-state transition were calculated between the start and the end time points of the upstate. For light-evoked up-states, start and end time points coincided with the times of the beginning and end of the light stimulus, respectively. Only light responses that were preceded by a period of neuronal silence (down-state) were considered in the analysis. Latencies of lightevoked responses were measured as the time point at which the line fitting the resting membrane 22
potential (baseline) intersected the linear fit of the rising phase of the light-evoked depolarization. The latency distribution displayed a clear bimodal distribution with some cells having submillisecond latencies (min. 0.4 ms, max. 0.8 ms; N = 5) and the remaining cells having supramillisecond latencies (min. 2.2 ms, max. 5.6 ms; N = 22). Cells were defined as ChR2-positive or ChR2-negative on the basis of the latency of their response to light (ChR2positive cells with latencies < 1 ms).
Slice electrophysiology: Coronal slices (300 µm thick) were prepared from injected mice (16-22 days old) with standard procedures
60
. Briefly, the animals were anesthetized with urethane
(Sigma, Milan, Italy) and decapitated. The brain was rapidly removed and placed in ice-cold cutting solution containing (in mM): 125 NaCl, 2.5 KCl, 1.25 NaH2PO4, 25 NaHCO3, 2 MgCl2, 1 CaCl2, 20 glucose, 2 Na-pyruvate, 3 Myo-inositol and 0.4 ascorbic acid (pH 7.4, 95% O2 / 5% CO2). Slices were cut with a vibratome (VT1000S, Leica Microsystems, GmbH, Wetzlar, Germany) and incubated in cutting solution at 34°C for 30 min and then at room temperature for additional 30 minutes before use. Slices were transferred to a submerged recording chamber and continuously perfused with artificial cerebrospinal fluid (aCSF) bubbled with 95%O2/5% CO2 (pH 7.4). The aCSF contained (in mM): 125 NaCl, 2.5 KCl, 1.25 NaH2PO4, 25 NaHCO3, 2 MgCl2, 2 CaCl2, 20 glucose. Bath temperature was maintained at 30-32 °C by an inline solution heater and temperature controller (TC-344B, Warner Instruments, Hamden, CT, USA). Recordings were obtained from epifluorescence-identified neurons in layer II/III or layer V in slices from electroporated cortices and from Rbp4-Cre x tdTomato mice injected with AAV, respectively. Experiments were performed in the whole-cell, current-clamp configuration and neurons were filled with biocytin (5 mg/mL) for subsequent anatomical reconstruction and 23
analysis (see below). Pipettes (tip resistance, 3-4 MΩ) were filled with internal solution containing (in mM): K-gluconate 140, MgCl2 1, NaCl 8, Na2ATP 2, NaGTP 0.5, HEPES 10, phosphocreatine 10 to pH 7.2 with KOH. Access resistance was monitored throughout the recordings and was typically
