Regulation of excitatory presynaptic activity by Ambra1 protein 1 determines neuronal networks in sex-dimorphic manner

18 Heterozygous mutation of Ambra1 , known as a positive autophagy regulator, produces autism- 19 like behavior in mice and autistic phenotypes in humans in a female-specific manner. However, 20 the substantial roles of the Ambra1 mutation in neurons are still unknown. We find that Ambra1 21 heterozygotes display a moderate decrease in excitatory synaptic release in-vitro and ex-vivo 22 exclusively in females without autophagy activity, resulting in significant alterations in γ- 23 oscillation power and seizure susceptibility by excitatory/inhibitory (E/I) imbalance. Specifically, 24 Ambra1 deficiency has no effect on neurogenesis and morphogenesis, but selectively 25 decreases excitatory synaptic activity without changes in synapse number, quantal size, 26 synaptic release probability, and synaptic plasticity. Therefore, the limited excitatory 27 synaptopathy by Ambra1 expression levels ultimately determines E/I imbalance in global 28 neural networks leading to the female-specific ASD.


Introduction 34
Autism-spectrum disorder (ASD) is a neurodevelopmental disorder mainly characterized by 35 deficits in social interaction/communication and restricted/repetitive patterns of behavior 36 (Association, 2013). Epidemiological studies estimated that ASD has been diagnosed in more 37 than 1% of the world's population (Elsabbagh et al., 2012;Vos et al., 2016) and described as 38 a sexually-dimorphic disease, with four times more males than females being diagnosed 39 8 Ambra1 gt/gt neurons could be compared simultaneously, it enables us to understand the 247 substantial role of Ambra1 in neurons. As a result, regardless of sex, Ambra1 deficiency had 248 no effect on neuronal development and synapse formation, and decrease selectively the 249 number of functional glutamatergic synapses. Moreover, surprisingly, the cultured Ambra1 +/gt 250 neurons also showed a female-specific decrease in eEPSC size, as the change in mEPSC 251 frequency in Ambra1 +/gt female brain acute slice ( Figure 6A-D). Thus, ASD caused by E/I 252 imbalance in Ambra1 heterozygous mutation is a neuron-intrinsic property of Ambra1 by sex 253 difference without any environmental factors. And our previous study, the decrease in relative 254 Ambra1 expression level in female Ambra1 +/gt brain, compared to one in male ( Dere et al., 255 2014), may help to understand the female-specific synaptic phenotype. Thus, we can suggest 256 that the size of EPSCs would be determined according to the expression level of Ambra1 257

protein. 258
In Ambra1 gt/gt neurons, the comparable mEPSC amplitudes and the eEPSC or mEPSC 259 frequencies reduced in half lead to the novel fact that there are Ambra1-dependent and -260 independent synapses. In particular, it can be speculated that Ambra1 deficiency makes 261 Ambra1-dependent synapses into silence ( Figure 6). And, although Ambra1-dependent 262 synapses maintain silence, the fact that the reduction of glutamate-induced response is less 263 than that of synaptic responses such as eEPSC, mEPSC frequency and RRP size raise two 264 possibilities ( Figure 6). Firstly, depletion of functional synaptic receptors that may be induced 265 by Ambra1 deficiency can lead to an increase in the number of extrasynaptic receptors. The 266 other possibility is that the Ambra1 gt/gt neurons display the complete ablation of synaptic 267 release in Ambra1-dependent synapses. To elucidate the impairment of synaptic release and 268 its relationship to synaptic receptors, we recall our previous work . The 269 number of functional synaptic glutamate receptors was reduced by approximately 40% in 270 Munc13-deficient synapses in which synaptic transmission from presynaptic terminal is 271 completely impaired. So, the difference between reduction ratios in synaptic parameters such 272 as the eEPSC size and mEPSC frequency, and glutamate induced responses in Ambra1 gt/gt 273 can be attributed to the complete impairment of glutamate release. In order to support the two 274 hypotheses, it can be inferred that the Ambra1-dependent synapses accounts for about 50% 275 of the total synapses. Considering our data evaluating the expression levels of synaptic 276 receptors and scaffolding proteins ( Figure S1), we highly appreciate the latter possibility. We 277 figured out that the most critical factor of E/I imbalance in Ambra1 heterozygous brain is the 278 contribution of Ambra1 in the activity of glutamatergic synapses, regardless of autophagy. In 279 the end, we discovered that the novel function of Ambra1 in neuronal cells play an important 280 factor in ASD manifestation. 281 excitatory synaptic activity, and that its sex-dimorphic expression of protein level modulates 283 the degree of glutamate release depending on sex (Dere et al., 2014). However, the 284 mechanism for the sex-dimorphic expression of Ambra1 protein level is still an important piece 285 to be studied further. This sex-dimorphic reduction of synaptic transmission by Ambra1 286 haploinsufficiency might manifest female-specific phenotypes, such as autistic-like behaviors, 287 increased seizure propensity and aberrant gamma oscillations (Dere et al., 2014) For sex determination of embryos, PCR analyses of Y chromosomes were performed using 311 GoTaq® G2 Flexi DNA polymerase, forward primer 5'-GGT GTG GTC CCG TGG TGA GAG-312 3', and reverse primer 5'-GAG GCA ACT GCA GGC TGT AAA ATG-3' to generate a 270 bp 313 fragment (94°C/1 min, 33 cycles with 94°C/1 min, 63°C/30 s, 72°C/30 s, and 72°C/7 min). PCR 314 products were analyzed on a 1.5% agarose gel in Tris-Acetate-EDTA buffer, which were 315 stained with HDGreen® Plus Safe DNA Dye (Intas). 316 mRNA expression of Ambra1 from data of Allen Brain Atlas 317 Data of mRNA expression level in different brain regions were extracted from Allen Mouse 318 Brain Atlas (Figure 1a, http://mouse.brain-map.org/) 51,52 . mRNA expression level in regions of 319 interests (ROIs) of in situ hybridization was calculated by multiplying expression density and 320 intensity. 321

Histological and Immunohistochemical Analyses 322
Mice were perfused transcardially with Ringer solution followed by 4% paraformaldehyde (PFA) 323 in 0.1 M phosphate buffer (PBS, pH=7.4). Brains were post-fixed at 4°C in 4% PFA in PBS for 324 2 h for X-galactosidase (X-gal) histochemical staining, or post-fixed overnight, followed by 325 cryo-protection in 30% sucrose solution in PBS and in liquid nitrogen for immunohistochemistry. 326 X-gal histochemical staining was performed with brains of 9 weeks old mice. Coronal brain 327 sections (50 μm) were cut using Leica VT1000S vibrotome (Leica Biosystems) and incubated were obtained using an Axiophot microscope (Carls Zeiss Microscopy GmbH). 332 Ambra1 WT and Het female mouse brain were cut into coronal sections (30 μm) with Leica 333 CM1950 instrument (Leica Biosystems). Sections were blocked with 10% normal horse serum 334 (NHS) and 0.2% Triton-X-100 in PBS for 1 h at room temperature (RT). PBS with 5% NHS and 335 0.2% Triton-X-100 was also used for the primary and secondary antibody dilution. Incubation 336 of the primary antibodies was carried out at 4°C for 1-3 nights, followed by incubation of 337 secondary antibodies (1:500) for 2 h and DAPI (1:10,000, D9542, Sigma-Aldrich) in PBS for 5 338 min at RT. Washing was performed between every step and sections were mounted using 339 Aqua-Poly/mount. The following antibodies were used for immuohistochemistry: mouse anti- (mEPSCs/mIPSCs) were recorded in the presence of 1 μM TTX, mixed with 10 μM bicuculine 449 methiodide for measuring mEPSCs or with 10 μM NBQX for measuring mIPSCs with washing 450 with 1 μM TTX for 15 mins between. An EPC-10 amplifier with Patchmaster v2X80 software 451 was used for data acquisition (HEKA/Harvard Bioscience). Subsequently, slices were fixed 452 using 4% PFA in PBS for two hours at RT and washed by PBS. 453

Immunohistochemistry of Biocytin-filled Neurons 454
After being washed in PBS, blocked and permeabilized for 1h with blocking solution (5% NGS 455 and 0.5% Triton X-100 in PBS), fixed brain slices obtained from patching were stained with 456 Alexa-Fluor-555-labeled streptavidin (1:1000; S32355, Thermo Fisher Scientific.) and DAPI 457 (1:10,000) in blocking solution. After washing, the slices were mounted on glass slides and 458 covered with cover slips in Aqua-Poly/Mount. CA1 pyramidal neurons in hippocampus were 459 scanned using a Leica SP5 confocal microscope with 100 x/1.44 NA oil objective at 0.126 μm 460 z-intervals. The basal and apical part of pyramidal neurons 3D Gaussian-filtered (σx,y 0.7, σz 461 0.7,), using custom-written macros to handle the large data sets. Only cells with a pyramidal 462 shape and a location in CA1 were used for further analysis. 463

In Utero Electroporation and Immunohistochemistry for Sholl Analysis 464
E14.5 mouse embryos from WT mothers bred with Het males were subjected to IUE (permit 465 number 33.19-42502-04-13/1052), as previously described 57,58 . DNA solution with pFUGW 466 (0.1 mg/mL) and pCX::myrVENUS (0.1 mg/mL) for the myrVenus construct 59,60 were used to 467 sparsely label CA1 pyramidal neurons in hippocampus. 468 In utero electroporated mice were perfused, and brains were post-fixed and cryo-protected 469 (15%-30% sucrose in PBS) at P28. Coronal brain sections (230 μm thickness) at -1.06 mm to 470 -2.46 mm from Bregma were collected using a Leica VT 1000S vibrotome. For 471 immunofluorescence staining, PFA was quenched by 1 mg/mL NaBH4 in PBS for 5 min 472 followed byh thorough washing in PBS. Brain sections were incubated in blocking solution (5% 473 normal goat serum (NGS) and 0.5 % Triton-X-100 in PBS) for 1 h at RT followed by incubation 474 in 0.2% Tween-20 and 10 μg/mL heparin in PBS for 1.5 h to improve the penetration of 475 antibody in thick brain sections. The blocking solution was used for diluting primary and 476 secondary antibodies (1:1000 dilution). The sections were incubated with polyclonal rabbit anti-477 GFP antibody (598, MBL) for 4 days at 4°C and with Alexa-Fluor (AF) 488 goat anti-rabbit IgG 478 (R37116, Thermo Fisher Scientific) overnight at RT followed by DAPI staining (1:10,000) and 479 mounted on slides. Images of CA1 pyramidal neurons in hippocampus were acquired with 1.02 480 washing. After DAPI staining, coverslips were mounted on slides. 529 Single autaptic neurons were imaged by Leica SP2 confocal microscope using 40x objective 530 (resolution: 1024 x 1024 pixels) with 1 μm z-step and analyzed using Image J software, as 531 referenced from previous publication 56 . Briefly, for the quantification of pre-and post-syanptic 532 puncta, the fluorescent signals of vGlut and PSD95 were thresholded and binarized followed 533 by being watersheded. The number of their puncta was analyzed using 'Analyze 534 particle' option. For the colocalization of preand post-synaptic marker, the Manders' overlap 535 coefficient was calculated by Intensity Correlation Analysis plugin. 536

Statistical Analysis 537
All data were analyzed separately for males and females. Statistical methods are described in 538 figure legends. All statistics were performed with Excel (Microsoft), GraphPad Prism 5 software 539 (GraphPad software) and SPSS 17 (SPSS Inc. The funders had no role in study design, data collection and interpretation, or the decision to 551 submit the work for publication. Blot between cortical homogenates of both genotypes (n=8 for each group) in female mice. 626 The data was normalized to the average value of wild-type group. I Comparison of LC3-II and 627 LC3-I intensities between two genotypes. The bar graphs represent mean ± S.E.M and 628 statistical analysis was performed by two-tailed unpaired t-test with significance level p < 0.05. glutamate-induced response between three genotypes in males (C) and females (D), 651 separately. E Short-term synaptic depression was monitored after application of 50 stimuli at 652 25 10 Hz. F Comparison of normalized eEPSC amplitude acquired from 2 nd stimuli (Paired pulse 653 ratio (PPR), left) and averaged during the steady state (36-40 th stimuli, right) from 10 Hz 654 stimulation experiment (E). G The bar graphs of eIPSC amplitude, RRP size, Pvr and 655 amplitudes and frequency of mIPSC and a dot graph of short-term synaptic depression by 50 656 stimuli at 10 Hz in GABAergic autaptic neurons from P0 striatum of Ambra1 +/+ and Ambra1 +/gt 657 females. Neuron numbers of each group are written within the bar or next to the legends from 658 2-4 independent experiments and the bars and dots in graphs are presented as mean ± S.E.M. 659 Statistical analysis of bar graphs from three groups (C, D, and F) was performed by one-way 660 ANOVA followed by Bonferroni or Tukey post-hoc test showing significance as asterisk (*, 661 p≤0.05; **, p≤0.01; ***, p≤0.001) and from two groups (G) by two-tailed unpaired t-test with 662 significance below 0.05. Short-term synaptic depression (E, G) was analyzed by Repeated 663 Measures of ANOVA. M, Fuoco C, Ucar A, Schwartz P, Gruss P, Piacentini M, Chowdhury K, Cecconi F. 2007.