Lead Contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Ryuichi Shigemoto (ryuichi.shigemoto@ist.ac.at)

Materials Availability

This study did not generate any new reagents or mouse lines.

Data and Code Availability

Original data have been deposited to Mendeley Data:

https://doi.org/10.17632/p2spfxzf57.1

Virus vectors

AAV₅-hSyn-DIO-mCherry, AAV₅-hSynDIO-hM4Di-mCherry and AAV₅-hSyn-DIO-hM3Dq-mCherry vectors were obtained from University of North Carolina vector core and Addgene viral service. AAV₉-hSyn-Flex-GCaMP6s-WPRE-SV40 and AAV₅-hSyn-Flex-GCaMP6s-WPRE-SV40 were obtained from University of Pennsylvania Vector Core. pAAV₁-CAG-Vamp2-venus transfer plasmid was generated by subcloning VAMP2-venus gene cassette obtained from Hiromu Yawo into pAAV-CaMKII-GFP (Addgene). Plasmid AAV-CAG::FLEX-rev::ChR2-HA-2a-hM4D was obtained from Addgene. Viral titrations were between 8 × 10¹² and 7.5 × 10¹³ genome copy per ml. For inducing expression in Prox1-creERT2 animals, tamoxifen (T5648, Sigma) previously diluted in corn oil was injected i.p. (75 mg/kg body weight) for 5 consecutive days.

Stereotactic injections, cannula and GRIN lens implantation

Animals were anaesthetized using a mixture of 100 mg kg⁻¹ Ketamine / 10 mg kg⁻¹ Xylazine and injected 200 mg kg⁻¹ of Novalgin as an intraoperative analgesic. Animals were placed in a stereotaxic apparatus (Kopf). The skull was exposed and craniotomies were performed at the following coordinates: ventral hilus injections or GRIN lens/cannula implantation; anteroposterior (AP) −3.2 mm, ± 3.0 mm mediolateral (ML), dorsal dentate gyrus injections or GRIN lens/cannula implantation; −1.8 mm AP and ± 1.3 mm ML. Virus injections were performed at a controlled rate of 0.1 μl min⁻¹ using a water filled glass micropipette coupled to a 1 μl Hamilton microsyringe (7001; Hamilton). The pipette was slowly lowered to the target site and remained in place for another 5 min post-injection before being slowly withdrawn. For ventral hilus injections, 0.5 μl of undiluted virus was delivered at two dorso-ventral (DV) coordinates measured from the dura DV = −3.5 mm and DV = −3.7 mm. For dorsal dentate gyrus 0.7 μl of AAV₉-hSyn-Flex-GCaMP6s-WPRE-SV40 diluted 1:7 in PBS was injected at DV = −1.9 mm below the dura. For mice used in intracerebral CNO infusions, bilateral guide cannulas were implanted above the dorsal dentate gyrus (−1.7 mm AP; ± 1.0 mm ML; −1.7 mm DV) or two unilateral cannulas (PlasticsOne) were implanted above the ventral hilus on each hemisphere (−3.2 mm AP; ± 3.1 mm ML; −3.2 mm DV). For GRIN lens (DORIC lenses, Quebec) implantation four M1 screws were secured into the skull at the anterior and posterior edges of the surgical site to anchor the implant. After drying the surface of the skull, the lens was lowered slowly into the target, while imaging until the desired fluorescence was achieved. All cannulas and GRIN lens were secured to the skull using Super-bond dental cement (Sun medical, Japan). After surgery animals were injected with 5 mg kg⁻¹ of Metacam (meloxicam) subcutaneously as a postoperative anti-inflammatory and painkiller. The animals recovered in a heat pad and then transferred to their home-cages and monitored closely for any discomfort signals for the three following days. Due to the short working distance of the lens (80 μm), the dorsal GRIN lens implantation required a slight “push” of the outer and medial molecular layer to reach focus. We are aware that this could impinge damage over the GCs dendrites and fibers coming from the entorhinal cortex. Although we cannot exclude any damaging effect of the GRIN lens implantation, the reproducibility of our results, the preserved functional connectivity observed with the combined chemogenetic approach, and the normal sparse activity during environmental exploration indicate that any damage is not significant for the purpose of our study.

For the case of Figure 2J, in order to avoid expression of hM3Dq-mCherry in ventral CGs of the double transgenics Prox1CreERT2/Calb2-Cre, we first injected the virus carrying GCaMP6s into the dorsal DG. For the next 5 days, these animals received tamoxifen injections to allow the recombination and expression of GCaMP6. Two weeks later when the tamoxifen effect had been gone, we injected the virus carrying hM3Dq into the ventral hilus, during the same surgery for GRIN lens implantation.

Calcium imaging in behaving mice and data analysis

Recording sessions began 4 to 6 weeks after lens implantation. First, animals were adapted to the manipulation procedure and to carrying the microscope on their heads in their home-cages for 4 days. Once adapted they explored a square environment (35 cm × 35 cm) constructed of 40 cm high matt black painted walls, visual cues in one (colored tape) and removable floor (laboratory surface paper) for 20 min every day for 6 days. Three videos of 3 min each were taken at 0, 5 and 10 min after the beginning of the exploration. The analysis was performed for the first exploration period to be consistent with the times where the foot-shock was delivered in the subsequent fear conditioning experiments. During that time animals explored the novel and familiar environments in a similar fashion, the well-known reduction of locomotion along familiarization becomes evident only after that period. The paper floor remained in the environment during all familiarization sessions for each animal to keep their individual odor constant. On day 6, after exploring the familiar environment, animals were transferred to a different environment: a square box (20 cm × 20 cm) with two white and two vertical black and white striped walls. The floor had stainless steel rods of 5 mm diameter, separated by 1 cm. Animals explored this environment for 10 min. Two videos of 3 min each were recorded during this time. Environments were carefully cleaned with 70% ethanol between animals. For familiarization in the fear conditioning environment, animals explored the conditioning cage (17 × 17 cm, Ugo Basile SLR, Italy) for 10 min for 6 consecutive days. The cage was not cleaned in order to keep their individual odor constant. After the 10 min exploration during day 6, a single foot-shock 1 sec, 0,65 mA was delivered. Immediately after, animals explored a novel environment (circular environment 20 cm diameter, 40 cm walls). For the running wheel experiments, animals were housed with a running wheel after implantation. For familiarization, animals visited a square environment (35 cm × 35 cm) constructed of 40 cm high matt black painted walls, visual cues in one (colored tape) and removable floor (laboratory surface paper) including their running wheel for 20 min every day for 6 days. On day 6, after exploring the familiar environment, animals were transferred to a different environment: a circular environment (40 cm diameter) constructed of 40 cm high matt black painted wall, visual cues (colored tape) and white plastic floor, together with their running wheel. Calcium-imaging data was acquired at 20 Hz, constant excitation power and exposure to compare fluorescent changes across sessions and corrected for motion artifacts using DORIC Neuroscience Studio Software. The rare artifacts that could not be corrected were excluded from the analysis. Imaged fields of views were highly constant across days. Regions of interest from single cells were carefully drawn by hand based on videos taken on different recording days (Figure S4B&F). To denoise and estimate the most likely neuronal activity underlying the imaged traces, we used Markov Chain Monte Carlo method developed for spike inference in continuous time (MCMC-method) provided in the CaImAn toolbox.⁷³ We subsequently used this inference to generate the estimated calcium traces (Figure S4I&J). This approach is more precise than other denoising approaches and inherently defines uncertainty in all inferred transients. To compare the relative change across days, we normalized all inferred events for every drawn ROI across all days to the single maximum inferred ΔF (t)/F₀ measured for each ROI_. We then calculated the area under the curve (AUC) (as a measure of neuronal activity for the first three minutes during exploration of either familiar or novel environments. We took advantage of the inherent uncertainty estimation of the MCMC-analysis to determine which events are significant. AUCs were calculated for all sessions and ROIs. We identified a total of 229 independently active dorsal granule cell ROIs and 60 ventral mossy cell ROIs for the familiarization experiment. For visualization purposes, we normalized all AUCs to the maximal AUC measured of all cells (Figure 3B&G, Figure S4O-Q). To compare the global activity change across animals, we compared the average AUC for all cells per animal (Figure 3C&H). All traces were normalized to the maximal average AUC measured across days. Inter-event intervals and peak distribution were extracted from the MCMC inferred traces. Peaks were required to have at least 1% (from normalized maximum for each ROI) (Figure S4K-N). The same analysis was performed for the experiments in which dorsal GC were imaged during chemogenetic ventral MCs activation (n = 386 dorsal GC ROIs). Control (mCherry) and experimental (hM3Dq) animals were continuously recorded after CNO injection. A strong wave of neuronal activity (Figure 2K-L, Video S1) was recorded only in experimental animals 30-33 min after injection. As exemplified in Video S1, animals showed a normal behavior after CNO, except for one animal that showed signs of seizure after 60 min of CNO injection. This behavioral abnormality did not influence the results since these time points were not included in the analysis. To extract the locomotion from the behavioral videos, a custom written MATLAB (Mathworks) script was used to contrast enhance, background subtract and fit an ellipsoid on the tracked mouse. The center of mass of the ellipsoid was used as the tracking point (see Supplementary Videos – red square on mice) (Figure 3D, I, Figure S5K, L). We separately calculated the distribution of locomotion speeds for novel (data from day 1 and day 6 novel) and familiar (data from day 5 and day 6) environments in ventral MCs and dorsal GC recorded animals, respectively. To correlate locomotion and neuronal activity, we used all AUC events, including epochs of no activity, calculated for each ROI and binned them for different locomotion speeds. All novel (day 1 and day 6 novel) and familiar (day 5 and day 6) events were used (Figure 3E, J). For the running wheel analysis, only AUC during constant locomotion on the running wheel were used (seven epochs with a total duration of 153 s and 400 s for novel and familiar environments, respectively) (Figure 3K).

Fear conditioning and novel and familiar environment exploration

Animals were acclimated to handling for three days before behavioral assays. A fear conditioning system (Ugo Basile SLR, Italy), coupled to Ethovision software (Noldus, Wageningen) was used for contextual fear conditioning and analysis. During pre-exposure, animals explored the conditioning chamber for 3 min. 24 h later, animals were reintroduced to the conditioning cage explored for 3 min. Then, three consecutive 1 s foot shocks of 0.65 mA with a 1 min interval between them were delivered. 24 h later, animals were placed again in the conditioning chamber for a 3 min exploration period. 24 h later, animals explored a distinct novel environment B, which consisted of an opaque black painted plastic tube of 35 cm diameter and 40 cm high. This environment was located in a different soundproof box than the conditioning chamber. Before every context exploration, the conditioning chamber was odorized with 1% acetic acid, and context B was odorized with 0.25% benzaldehyde. Both environments were carefully cleaned with ethanol 70% after every exploration. For conditioning in a familiar environment, animals explored the conditioning cage (17 × 17 cm, Ugo Basile SLR, Italy) for 10 min for 6 consecutive days. In order to keep the animal’s individual odor constant, a piece of laboratory paper was kept under the grid floor during all the familiarization and conditioning procedures. During day 6, animals received 0.75 mg/kg of CNO i.p. and returned to their home cages for 40 min. Then, animals explored the conditioning box for 10 min. After this period, three consecutive 1 s foot shocks of 0.65 mA with a 1 min interval between them were delivered. 24 h later, animals explored the conditioning box for 3 min and 24 h later, they explored a novel environment (circular environment 20 cm diameter, 40 cm high) for 3 min. Freezing was measured with activity tracking in Ethovision, with a threshold of 500 ms for immobility detection. For measuring locomotion in familiar environments (Figure S5C, D, I, J), animals explored a circle (38.5 cm diameter) with black wall (Figure S5C) or a square (40 cm × 40 cm) with white wall (Figure S5D) for 10 min every day for 5 days before measuring distance traveled for 5 min in the same environments on day 6. The distance traveled in the circle (38.5 cm diameter) for 5 min on day 1 was used as locomotion in a novel environment (Figure S5B). For novel environment exploration and following perfusion for c-Fos immunolabelling, animals were handled for 3 days before experiments to acclimatize to the experimenter and procedures. After CNO injection, animals were kept in their home-cages for 40 min before being transferred to a square box of 20 cm by 20 cm, with two white walls, and two walls with vertical black and white lines with white plastic floor. Inside the cage there was a running wheel and a styrofoam box 5 cm by 7 cm and 3 cm high. Animals explored this environment for 30 min, then returned to their home-cage and remained there for another 40 min before perfusion.

CNO delivery

Clozapine-N-oxide (CNO, Tocris, Catalog number 4936) was dissolved in dimethylsulfoxide (DMSO) at 1% final concentration, and then diluted with PBS to 1 mg/mL for intraperitoneal (i.p.) injections and with ACSF at 3 μM for intracerebral infusions. For i.p. injections, we delivered 3 mg/kg or 0.75 mg/kg of CNO, followed by behavior 40 min after the injection. For intracranial delivery, CNO was infused bilaterally into the dorsal or ventral dentate gyrus with 0.3 μl per hemisphere at a concentration of 3 μM using a 26-gauge stainless steel internal cannula (PlasticsOne). The internal cannula was connected with a 1 μl Hamilton syringe (7001; Hamilton) to control the injection rate at 100 nL min⁻¹. The injection cannula was left connected for 5 min before removal to allow for diffusion. Finally, all behavior was performed 15 min following the drug infusion.

Imunohistochemistry and in situ hybridization

For immunohistochemistry, mice were overdosed with 750–1000 mg kg⁻¹ mixture of Ketamine, Xylazine and perfused transcardially with PBS, followed by 4% cold paraformaldehyde (PFA) in PB. Extracted brains were kept in 30% sucrose in PB at 4°C overnight, then transferred to PBS. A sliding microtome (Leica SM2000R) was used to cut 40-μm coronal slices in PBS. Slices were washed with PBS-T (PBS + 0.2% Triton X-100), then incubated with PBS-T + 20% normal donkey serum, at room temperature for 1 h for blocking. Then, slices were incubated with one or more primary antibodies at 4°C for 24 h (rabbit anti-calretinin AB5054, Millipore, dilution 1:1000; goat anti c-Fos SC-52, Santa Cruz Biotechnology, dilution 1:500; rabbit anti-PV, Abcam, dilution 1:1000; or for electrophysiology slices, Streptavidin Alexa Fluor 488 conjugated (1:1000); rabbit anti HA-tag AB3724, Cell Signaling Technology, dilution 1:1000). Three washes of PBS-T for 10 min each were performed on the slices before 1 h incubation with secondary antibody at 1:1000 dilution (A-21206; A-21447, ThermoFisher Scientific). Slices were washed three more times in PBS-T for 10 min each, stained with 4′,6-diamidino-2-phenylindole (DAPI) to label cell nuclei and mounted with Prolong antifade reagent (Thermofisher) onto microscope slides. For in situ hybridization, Calb2-cre animals were injected hilus with pAAV₅-hSynDIO-mCherry in the ventral hilus. Three weeks later, they were anesthetized with isofluorane, decapitated and their brains removed immediately. The ventral hippocampus was dissected, placed in a Cryomold (Sakura, Japan) filled with O.C.T. Tissue-Tek compound (Sakura, Japan), and stored at −80 °C until 14 μm sections were cut with a cryostat (Microm HM 560, Thermo) and mounted in Superfrost plus slides (Thermo scientific). The whole in situ hybridization procedure was performed with and according to the protocol of the RNAscope Fluorescent Multiplex Assay (ACD) using probes for vGluT1 and mCherry. Slides were imaged in an upright Zeiss LSM 700 confocal microscope. Cytoarchitecture borders for cell counting were defined in accordance with the Allen Brain Institute Atlas, http://atlas.brain-map.org.

c-Fos counting

For counting c-Fos-positive granule cells in the dentate gyrus, three 40 μm sections were taken from dorsal DG and three for ventral DG from each animal if required. These sections were separated by 160 μm, thus representing the different antero-posterior levels of the whole dorsal or ventral DG. Note: the apparent reduced number of c-Fos positive GCs in control conditions compared to other studies is due to the thin sections used. This was required because after the activation experiments of ventral MCs, the high number of c-Fos positive GCs would make thicker sections uninterpretable. Pictures were taken in an upright Zeiss LSM 700 confocal microscope with the same magnification, laser power and exposure times. Then, the total area of the granule cell layer was measured, and cells positive for c-Fos within this area were counted using Fiji (https://fiji.sc). Finally, the density of cells was calculated dividing the number of cells by the area.

Electron microscopy

Animal anesthetization, fixation and pre-embedding immunoelectron microscopy (EM) were performed as described previously.⁷⁴ Briefly, mice were deeply anesthetized and trans-cardially perfused at a flow rate of 7 mL/min, first with 25 mM PBS for 1 min, followed by 10 min of fixative containing 4% paraformaldehyde, 15% picric acid and 0.05% glutaraldehyde in 0.1 M PB. Brains were removed and kept in PB until coronal sections of 50 μm thickness were cut in a vibratome (Linear Slicer Pro7, D.S.K, Japan). Slices were washed in PB twice, cryo-protected, and freeze-thawed before adding blocking solution (10% NGS in 2% BSA) for 90 min. Double immunolabeling was done by mixing primary antibodies in 2% BSA, incubated for 3 nights at 4°C. For the ventral mossy cell samples, we mixed rabbit anti-PV (1:1000, Abcam) with mouse anti-RFP IgG1k (1:1000, MBL). For the dorsal mossy cell samples, we mixed rabbit anti-PV (1:1000, Abcam) with mouse anti-GFP IgG2a (1:3000, Wako). After washing, sections were incubated overnight at 4°C in a mixture of secondary antibodies, 1.4 nm gold conjugated goat anti-rabbit (1:200, Nanoprobes) and goat anti-mouse IgG1k biotinylated (1:400, Life Technologies) antibodies for ventral mossy cell samples, and goat anti-mouse IgG2a biotinylated (1:10000, Life Technologies) antibodies for dorsal mossy cell samples. After 10 min post-fixation in 1% glutaraldehyde we performed silver enhancement of immunogold particles (HQ Silver enhancement kit, Nanoprobes) and immunoperoxidase reaction, followed by 15 min fixation with 0.5% osmium tetroxide and counterstaining with 1% uranylacetate for 30 min. Slices were dehydrated in an ascending ethanol series, propylene oxide and then infiltrated in resin (Durcupan ACM, Sigma) at room temperature overnight. Slides were flat-embedded and resin was polymerized for 48 h at 60°C. We have chosen the areas to be examined by the presence of the PV+ dendrites that were located in the inner one third of the molecular layer in a middle part of the upper blade of the dorsal dentate gyrus. The selected regions were re-embedded, and ultrathin serial sections at 70 nm thickness were cut with an ultramicrotome (Leica EM UC7) for examination in a transmission electron microscope (Tecnai 10, Philips, the Netherlands). The vast majority of DAB-positive synapses were made on GCs spines (Table S1, > 97% for ventral MCs, > 92% for dorsal MCs: Note that these values are minimum estimates because we targeted PV+ dendrites, making positive biases for synapses made on these dendrites over those on GCs) as described previously.²² We used a total number of 450 sections covering a total examined area of 16,700 mm². To estimate the relative contribution of ventral and dorsal MCs to synaptic innervation onto PV+ dendrites and GCs spines, we first calculated % of DAB-positive synapses in all synapses made on PV+ dendrites examined, and % of DAB-positive synapses in all synapses made on GCs spines in surrounding areas with similar depth from the section surface (Table S1). Then, these % values on PV+ dendrites were normalized by the % values on GCs spines in each animal to calculate an index Ip/g (Table S1), which indicates innervation strength on PV+ dendrites relative to that on GCs by MCs. The Ip/g values are independent of the fraction of transfected MCs. We compared Ip/g values between ventral and dorsal MCs samples with Student’s t test (n = 4 animals for each).

Slice electrophysiology

Calb2-cre and Drd2-cre mice (4.4 to 12 months old) were injected bilaterally with AAV expressing hM4Di and ChR2 in a Cre dependent manner. Three weeks post-injection they were anesthetized via i.p. injection of ketamine/xylazine (ketamine: 100 mg/kg; xylazine 10 mg/kg) and transcardially perfused with oxygenated (95% O₂ / 5 % CO₂) ice-cold artificial cerebrospinal fluid (ACSF) containing (in mM): 118 NaCl, 2.5 KCl, MgSO₄ 1.5, NaH₂PO₄1.25, CaCl₂ 2, Myo-inositol 3, Glucose 10, Sucrose 40, NaHCO₃ 25, pH = 7.4. The brain was rapidly excised and placed in a cutting chamber. Coronal brain slices were prepared with a D.S.K. Linear Slicer Pro7 (Dosaka, Japan) at 250 – 350 μm thickness for recordings of ventral MCs. For recordings of optogenetically evoked currents in dorsal GCs, parasagittal slices were prepared at a 15-20°C angle and at 1 – 1.5 mm thickness.⁷⁵ Slices were left to recover for 20 min at 35°C, thereafter heating was turned off and slices recovered to room temperature over the course of one h. One slice was transferred to the recording chamber and superfused with recording ACSF (2 mM CaCl₂) at a rate of 2-4 mL/min. Whole-cell patch clamp recordings were performed using an EPC 10 USB amplifier with PatchMaster software (HEKA) at a sampling rate of 20 – 50 kHz and filtered at 2.9 kHz. MCs or dentate gyrus GCs were identified morphologically. Patch pipettes were pulled with a Model P-97 horizontal puller (Sutter Instrument, USA) at a resistance of 3 – 5 MΩ and filled with internal solution containing (in mM): K-Gluconate 140, MgCl₂ 2, HEPES 10, EGTA 0.5, MgATP 2, NaGTP 0.2, Sucrose 32, pH = 7.4 adjusted with KOH. In addition, 0.25 - 0.3% biocytin was added to the internal solution for labeling of recorded neurons after slice fixation (4% paraformaldehyde). For optogenetic experiments, blue light (465 nm) was emitted from an LED (Plexon) and transmitted onto the slice through an optical fiber. Light pulses (5 ms duration, 15 mW) were applied at 10 Hz (for 10 s at 30 s interval) for current clamp experiments in MCs and at 0.1 - 0.5 Hz for voltage-clamp experiments in dorsal GCs. In voltage-clamp experiments, cells were held at −70 mV and light-evoked synaptic currents were measured by averaging traces of 50 - 100 light-stimulations before and 3 - 10 min after the start of the application of 3 μM CNO. To test the effect of hM4Di activation on light-evoked action potential firing, cells were hyperpolarized (∼-60-70 mV) to minimize spontaneous firing. Series resistance was monitored throughout all measurements and recordings with changes in series resistance exceeding 20% were discarded. Electrophysiological data was analyzed using Patchmaster software (HEKA), Clampfit 10 (Molecular Devices) and Prism 6 (GraphPad).