1 Developmental decrease of entorhinal gate disrupts 2 prefrontal-hippocampal communication in immune- 3 challenged DISC1 knockdown mice

31 The prefrontal-hippocampal dysfunction that underlies cognitive deficits in mental disorders 32 emerges during early development. The contribution of the lateral entorhinal cortex (LEC), a 33 gatekeeper of prefrontal cortex (PFC) and hippocampus (HP), to the early dysfunction is fully 34 unknown. Here we show that the poorer LEC-dependent associative recognition memory 35 detectable at pre-juvenile age is preceded by abnormal communication within LEC-HP-PFC 36 networks of neonatal mice mimicking the combined genetic and environmental etiology (GE) 37 of disease. The prominent entorhinal drive to HP is weaker in GE mice as a result of sparser 38 projections from LEC to CA1 and decreased efficiency of axonal terminals to activate the 39 hippocampal circuits. In contrast, the direct entorhinal drive to PFC is not affected in GE 40 mice, yet the PFC is indirectly compromised, as target of the under-activated HP. Thus, 41 already at neonatal age, the entorhinal function gating prefrontal-hippocampal circuits is 42 impaired in a mouse model of disease. 43

investigated the activity patterns within entorhinal-hippocampal-prefrontal networks in 157 neonatal (P8-10) CON (n = 12) and GE (n = 10) mice (Fig. 2a). Extracellular recordings of 158 LFP and MUA showed that discontinuous spindle-shaped oscillations with frequency 159 components peaking in theta band (4-12 Hz) intermixed with irregular low amplitude beta-160 gamma band components (12-30 Hz) were the dominant pattern of entorhinal network 161 activity of both groups of mice (Fig. 2a, supplementary fig. 4a). The discontinuous oscillatory 162 events classified as spindle-bursts were superimposed on a slow rhythm (2-4 Hz) that 163 continuously entrained the neonatal LEC (supplementary fig. 4a) and related to respiration 40 . 164 The occurrence and duration of discontinuous oscillatory events (4-30 Hz) were comparable 165 in CON and GE mice (supplementary fig. 4b). However, their power (4-30 Hz) was 166 significantly smaller (F (1,28) = 11.89, p = 0.002, one-way ANOVA) in GE mice (9.96 ± 1.20) 167 than CON mice (23.45 ± 3.39) (Fig. 2a). Given that single-hit E and G mice were 168 indistinguishable in their activity patterns from CON mice (supplementary fig. 4c), single-hit 169 models were not considered for the rest of investigations. The diminished network activity in 170 LEC was accompanied, as previously reported 25 , by the dysfunction of network activity in 171 both HP and PFC (Fig. 2b, c). 172 Next, we questioned whether the dampening of oscillatory activity in LEC, HP and 173 PFC during early development related to communication deficits within the limbic networks. 174 For this we firstly assessed the coupling by synchrony between LEC and PFC-HP pathway 175 in neonatal CON (n = 14) and GE (n = 14) mice by calculating the coherence of oscillatory 176 events and considering only the imaginary part that was not corrupted by volume 177 conductance 41 . A tight theta-beta band coupling of spindle-bursts between LEC and HP as 178 well as between HP and PFC was detected in neonatal CON mice (Fig. 2d). In contrast, the 0.010, F (1, 26) = 5.18, p = 0.03, one-way ANOVA; HP-PFC: 0.345 ± 0.013 vs. 0.284 ± 0.014, 181 F (1, 26) = 11.340, p = 0.002, one-way ANOVA) (Fig. 2d). The coherence between LEC and 182 PFC was much higher than the coherence calculated for the shuffled data in both CON and 183 GE mice, yet no frequency-specific coupling was detected (Fig. 2d). In a second step, to get 184 better insights into the directionality of information flow within entorhinal-hippocampal-185 prefrontal networks, we used the generalized partial directed coherence (gPDC), a measure 186 that reflects the directionality of network interactions in different frequency bands. In CON 187 mice, we confirmed the previously reported drive from HP to PFC as well as the stronger 188 information flow from LEC to HP than from LEC to PFC 32, 42 . The entorhinal drive to HP was 189 significantly decreased in GE mice (0.084 ± 0.006 vs. 0.112 ± 0.006, p = 0.0009, one-way 190 ANOVA), which, besides the previously reported local dysfunction in HP and PFC, might 191 further contribute to the reduction of hippocampal drive to PFC (0.069 ± 0.004 vs. 0.094 ± 192 0.006, p = 0.002, one-way ANOVA) (Fig. 2e) as well as HP-projecting neurons were similar in P8-10 CON (27 neurons from 10 mice) and 239 GE (12 neurons from 5 mice) mice (Table 1). All investigated neurons showed linear I-V 240 relationships and their firing increased in response to depolarizing current injection. The 241 active membrane properties (i.e. action potential (AP) threshold, AP amplitude, half-width, 242 Rheobase, firing frequency) of both groups of entorhinal neurons did not differ between CON 243 and GE mice (Table 1). These results suggest that circuit dysfunction of GE mice does not 244 mainly relate to cellular abnormalities of entorhinal PFC-projecting and HP-projecting 245

Weaker responsiveness of HP to optogenetic activation of LEC in neonatal GE mice 247
To directly test the functional communication along axonal pathways within entorhinal-248 hippocampal-prefrontal networks of CON and GE mice, we monitored the responsiveness of 249 the three areas to the activation of LEC. For this, we selectively transfected pyramidal 250 neurons in LEC of CON (n = 11) and GE (n = 13) mice with a highly efficient fast-kinetics 251 double mutant ChR2E123T/T159C (ET/TC) 43 and the red fluorescent protein mCherry by 252 micro-injections performed at P1 (Fig. 4a, supplementary fig. 6). 253 First, we assessed the efficiency of light stimulation in evoking action potentials in 254 entorhinal pyramidal neurons in vivo. Blue light pulses (473 nm, 20-40 mW/mm 2 ) at a 255 frequency of 8 Hz led shortly (<10 ms) after the stimulus to precisely timed firing of 256 transfected neurons in the LEC of both P8-10 CON and GE mice (Fig. 4b). The used light 257 power did not cause local tissue heating that might interfere with neuronal spiking 44, 45 . The 258 efficiency of light stimulation in evoking entorhinal spikes was similar for both CON and GE 259 groups (Fig. 4b). From the second pulse on, the firing of neurons from both groups gradually 260 lost the precise timing to the stimulus and the response reliability, most likely due to the mice (supplementary fig. 7). In line with the unweighted projections from the LEC to the two 264 areas (Fig. 3), pulsed light stimulation in LEC led to rhythmic firing in HP, yet not in PFC (Fig.  265 4b). Quantification of the hippocampal firing probabilities upon stimulus revealed that CA1 266 neurons were reliably activated by light pulses in CON, yet not GE mice (Fig. 4b) coherence between LEC and HP (0.14 ± 0.05) but not between LEC and PFC (-0.01 ± 0.02) 273 ( Fig. 4c). Of note, frequency-specific boosting of HP through LEC activation caused an 274 indirect augmentation of synchrony within prefrontal-hippocampal networks in CON mice 275 (0.19 ± 0.06) (Fig. 4c). In GE mice, the stimulation-induced LEC-HP coherence increase was 276 of lower magnitude (0.04 ± 0.03 vs. 0.14 ± 0.05, F (1,17) = 4.10, p = 0.049) and consequently, 277 not sufficient to augment the hippocampal-prefrontal synchrony (-0.03 ± 0.04) (Fig. 4c). 278 Taken together, these results indicate that LEC has a critical role for the activation of 279 HP that on its turn boosts the entrainment of PFC. In contrast, the direct impact of entorhinal 280 activity on PFC is low, if any. The hub function of LEC persists in GE mice, yet the LEC-281 driven activation of HP is much weaker, being not further relayed to PFC. Already at neonatal age, entorhinal projections target the HP (Fig. 3) and, in line with 297 previous studies 46 , accumulate in stratum lacunosum of CA1 area (Fig. 6a). The density of 298 these projections significantly differs between CON and GE mice (Fig. 3, supplementary fig.  299 5). To test the function of entorhinal innervation of HP and whether the sparser projections in 300 GE mice cause the network and neuronal deficits described above, we performed multi-site 301 recordings of LFP and MUA in CA1 area during pulsed and ramp light stimulation of 302 entorhinal terminals in HP (Fig. 6a). The field response evoked by light pulses (3 ms, 473 nm, 303 8 Hz) in HP had a fast (~15 ms) onset in all investigated mice, yet a smaller amplitude 304 (23.37 ± 3.91 µV, p = 0.040) in GE mice than CON mice (50.63 ± 11.79 µV) (Fig. 6b). Light 305 stimulation of entorhinal terminals efficiently evoked hippocampal spikes of CON mice 306 (probability 0.72 ± 0.08) (Fig. 6c), which was in line with the dense entorhinal axonal 307 terminals (supplementary fig. 5). The firing probability upon stimulus was significantly (p = 308 0.002) lower in the HP of GE mice (0.32 ± 0.05) (Fig. 6c). Not only the firing efficiency 309 decreased but also the number of responsive hippocampal units was lower in GE mice (8 310 out 73, ~11%) when compared to CON mice (22 out of 81, ~27%) (Fig. 6d, e). These results 311 indicate that the function of entorhinal projections in HP is impaired in GE mice, their 312 efficiency to boost the hippocampal activity being decreased. 313 If the function of entorhinal projections to PFC but not to HP is largely intact in GE 314 mice, the question arises, whether the weaker entorhinal drive to HP is still sufficient to 315 entrain the neural activity in PFC. Light activation of entorhinal axonal terminals in HP (Fig.  316 7a) led to an increase of neuronal firing both in layer 5/6 (1.87 ± 0.18) and layer 2/3 (1.70 ± if any, effect in the PFC of GE mice (layer 5/6, 1.50 ± 0.12, p = 0.04; layer 2/3: 1.44 ± 0.10, p 319 = 0.04). Correspondingly, the prefrontal-hippocampal coupling augmented in CON mice 320 during stimulating entorhinal terminals in HP (relative change 0.09 ± 0.03), yet not in GE 321 mice (relative change, -0.02 ± 0.004, p = 0.049, Fig. 7d). (iv) the associative recognition memory that depends on entorhinal-hippocampal-prefrontal memory ("Where") 48 and LEC that codes context ("What") and temporal ("When") 349 information 49, 50 . In contrast to MEC, the LEC function has been less well dissected. It codes 350 for object features and context-related locations, being critical for the performance in In line with the density of axonal projections, the direct entorhinal drive to HP is stronger than 364 to PFC. On its turn, the hippocampal activation boosts the firing and oscillatory entrainment 365 of PFC. By these means, two entorhinal pathways, directly and indirectly activate the PFC. 366 They ensure the necessary level of neonatal excitation and oscillatory activation, which are 367 mandatory for adult prefrontal-related behavior 57 . This knowledge gain might be instrumental 368 for answering the question how non-sensory cortices, such as PFC, generate early patterns 369 of oscillatory activity. Spontaneous activity from the periphery travels along axonal 370 projections via brainstem and thalamic nuclei and boosts the entrainment of developing characteristic functional topographies 58, 59 . Such mechanisms are irrelevant for early 373 prefrontal oscillations; here, it seems that LEC drives the activation patterns. This 374 mechanism is not fully decoupled from sensory inputs, since the LEC receives direct inputs 375 from the olfactory bulb. The blind and deaf mouse pups that do not actively whisker at 376 neonatal age have already adult-like olfactory abilities 60 . The olfactory bulb not only 377 processes and forwards the odor information to the LEC, but also spontaneously generates 378 patterns of early oscillatory activity that activate the LEC 40, 61 . Therefore, at neonatal age, the 379 entorhinal direct drive to HP as well as direct and indirect to PFC, is controlled by the Hartig GmbH, Roedermark, Germany) was mounted 100 cm above the arena and connected 500 to a PC via PCI interface serving as frame grabber for video tracking software (Video Mot2 501 software, TSE Systems GmbH, Bad Homburg, Germany). 502

Exploratory behavior in the open field.
Pre-juvenile mice (P16) were allowed to freely 503 explore the testing arena for 10 min. Additionally, the floor area of the arena was digitally 504 subdivided in 8 zones (4 center zones and 4 border zones) using the zone monitor mode of 505 the VideoMot 2 analysis software (VideoMot 2, TSE Systems GmbH). The time spent by 506 pups in center and border zones, as well as the running distance and velocity was quantified. mouse was placed into the arena containing two different objects and released with the back 511 to the objects. After 10 min of free exploration of objects the mouse was returned to a 512 temporary holding cage. Subsequently, the test trial was performed after a delay of 5 min 513 post-familiarization. In NOPd task, the mice were allowed to investigate one familiar and one 514 novel object with a different shape and texture for 5 min. The nature of this test is similar to 515 the novel object preference test, except that the test trial involves an association between 516 two different objects (an association of object-object). In OLP task, the mice were allowed to 517 investigate one familiar and a copy of the old object that was previously presented for 5 min. 518 This test examines whether animals recognize the location that was once occupied by a 519 particular object (an association of object-location). Object interaction during the first 4 min 520 was analyzed and compared between the groups. All trials were video-tracked and the 521 analysis was performed using the Video Mot2 analysis software. The object recognition 522 module of the software was used and a 3-point tracking method identified the head, the rear 523 end and the center of gravity of the mouse. Digitally, a circular zone of 1.5 cm was created 524 around each object and every entry of the head point into this area was considered as object 525 interaction. Climbing or sitting on the object, mirrored by the presence of both head and 526 center of gravity points within the circular zone, were not counted as interactions. 527 The object discrimination was computed for NOPd and OLP as (time at object2-time at 528 object1) / (time at object1+ time at object2) (Fig. 1). 529

Behavioral protocols for quantifying the cFos expression in mouse doing NOPd or 530
OLP task. P16 CON mice were randomly divided into 2 groups (n = 4 mice / group). The 531 mice were allowed to freely explore the arena containing two different objects for 10 mins. 532 This familiarization process continued for 3 days with 2 trials per day. On the third day (P18), 533 task and one mouse to perform the familiarization trial. The mice were perfused ~90 min 537 after the last behavioral trial. Fisher Scientific, USA) was used. Mice were injected at P7 with BDA unilaterally into LEC 552 using iontophoresis. The bone above LEC was carefully removed using a syringe. A glass 553 capillary (~30 µm tip diameter) was filled with ~1 µL of 5% BDA diluted in 0.125 M 554 phosphate buffer by capillary forces, and a silver wire was inserted such that it was in 555 contact with the BDA solution. The positive pole of the iontophoresis device was attached to 556 the silver wire, the negative one was attached to the skin of the neck. Iontophoretically 557 injection by applying anodal current to the pipette (6 s on/off current pulses of 6 µA) was 558 done for 10 min. Following injection, the pipette was left in place for at least 5 min and then 559 slowly retracted. The scalp was closed by application of tissue adhesive glue and the pups 560 were left on a heating pad for 10-15 min to fully recover before they were given back to the 561 mother. The pups were perfused at P10. Second, to locate the innerved neurons in PFC by volume of 0.1 µl of WGA (2.5% in PBS) was delivered via a 10 µl microsyringe pump 565 controller. The slow injection speed (0.05 µl/min) and the maintenance of the syringe in 566 place for at least 8 min ensured an optimal diffusion of the tracer. 40 hours after the injection, 567 the pups were perfused. 568 Histology and staining protocols. Histological procedures were performed as previously 569 Microscopic stacks were acquired as 2048×2048 pixel images (pixel size, 78 nm; Z-step, 596 500 nm). All images were similarly processed and analyzed using ImageJ software. presented as mean ± sem. Significance levels of p < 0.05 (*), p < 0.01 (**) or p < 0.001 (***) 649 were tested. Statistical parameters can be found in the main text. 650

Code availability 651
All the codes used in the current study are available from the corresponding author upon 652 request. 653

Data availability 654
The data supporting the findings of this study is available with the article and its 655 Supplementary Information file, or is available from the corresponding author upon request.

58.
Khazipov R, Sirota A, Leinekugel X, Holmes GL, Ben-Ari Y, Buzsáki G. Early motor 32 848 Fig. 1 The performance of pre-juvenile GE mice in associative recognition memory 849 tasks. (a) Schematic of the protocol for NOPd task (top) and violin plots displaying the 850 discrimination ratio in familiarization and test trials when averaged for CON and GE mice 851