Pfn2 mutants are born in Mendelian ratio but in heterozygous breeding pairs we noted delayed growth of the Pfn2^−/− pups compared to Pfn^+/− and wt control littermates in larger size litters, suggesting a competitive disadvantage during rearing. Genotyping at weaning age (P21-P25) showed a reduced mendelian ratio to about 22% for Pfn2^−/− animals (Figure 1A), pointing to a mortality rate of Pfn2^−/− pups before weaning that was not observed in Pfn2^+/− and wt control littermates (Figure 1B).

These findings prompted us to investigate survival and life expectancy of Pfn2^−/− mice through a standard survival analysis. The Kaplan–Meier plot showed a significant mortality of Pfn2^−/− mice between 7 and 13 months (Peto-Prentice test, p < 0.001), with the median survival halved compared to wt littermate controls (Figure 1C). Interestingly, Pfn2 heterozygote mice displayed an intermediate phenotype, with significantly reduced survival compared to wt controls (p = 0.001), indicating a gene dosage effect on mouse survival. As a consequence, life expectancy (average life span) of both Pfn2^−/− and heterozygote mice was significantly reduced by 30 and 25%, respectively (Figure 1C, inset).

Premature mortality has been reported in autism spectrum disorder (Hirvikoski et al., 2016). One associated cause is epilepsy, in a recent study found in 12% of individuals with autism (Lukmanji et al., 2019). In a small-scale study (N = 70 Pfn2^−/− mice) we observed spontaneous, sensory induced, seizures in 12.5% of Pfn2^−/− mice, irrespective of their sex. The seizures were tonic or tonic–clonic, had an average length of 14 s in females and 15 s in males (Figure 1D), and the mice typically recovered within 60–120 s. The seizures’ onset was tendentially earlier in male mutant mice (8 months) than in females (11 months, Figure 1E). Since the onset of seizures coincides with the period of higher mortality of the Pfn2^−/− mice, we suspect that they might contribute to their reduced life expectancy.

The competitive disadvantage of Pfn2^−/− pups during rearing hinted at altered social behavior, either in the relationship of the mutant pups with the mother or with the other littermates for successful breastfeeding. To study sociability more in detail, we addressed two established behaviors in adult mice: maternal behavior of Pfn2^−/− mice toward the pups and social interactions of Pfn2^−/− male mice.

Maternal behavior was found severely affected in Pfn2^−/− mice. Nest building was almost absent in Pfn2^−/− females (Supplementary Figure 1) and when Pfn2^−/− mothers were allowed as single parents to rear their litters, all pups were lost irrespectively of the pups’ genotype (Table 1, top part). Interestingly, with the cohabitation of a wt or Pfn2^+/− father, all litters of Pfn2^−/− mothers could be rescued and survived. However, only about half of the litters survived when both parents were Pfn2^−/− animals (Table 1, bottom part). One possible reason for the rearing deficit of Pfn2^−/− mothers was a severely compromised pup retrieval behavior: when challenged to retrieve P7 pups dispersed in the cage, wt mothers quickly collected all pups within 7 min back into the nest (Figure 2A). Instead, Pfn2^−/− mothers never retrieved a single pup within the maximally allowed experimental time (30 min). Thus, Pfn2^−/− dams completely lacked pup rearing behavior.

Suspecting a general social interaction deficit upon reduction of profilin 2 levels, we applied to adult male mice the Social Interaction paradigm in the 3-chambered maze to assess adult sociability (Figure 2B) (Moy et al., 2007). When social interaction between the adult test mouse and a stranger juvenile was assessed, Pfn2^−/− mice showed a mild reduction of sociability. Only wt controls showed a significant preference for the Social interaction (SI) zone in terms of number of visits (SI: 8 ± 1 vs. Empty: 5 ± 1), while both Pfn2^+/− (SI: 9 ± 1 vs. Empty: 8 ± 1) and Pfn2^−/− (SI: 12 ± 2 vs. Empty: 10 ± 2) mice failed to do so (Figure 2C). Wt control mice also showed the highest preference for the social compartment when measuring the time spent in the SI zone compared to the Empty zone (Figure 2D), although also Pfn2^−/− and Pfn2^+/− mice showed some degree of preference for the SI compartment. The average visit duration of Pfn2^−/− mice to the stranger juvenile mouse was tendentially reduced compared to wt controls (Figure 2E), as well as the average sniffing time (Figure 2F). In addition, the interaction of Pfn2^−/− mice with the stranger juvenile mouse was qualitatively very different from control mice: direct nose contacts of Pfn2^−/− mice with the juvenile mouse were rare, with the mutant mice mostly running around and over the metal enclosure. In a previous study (Pilo Boyl et al., 2007) we reported a striking increase of novelty-seeking behavior in Pfn2^−/− mice, which spent more time than wt control littermates to explore novel objects. We suspect that this phenotype might have masked the social behavior in our set-up, where the juvenile mouse has represented an interesting novel object due to its immature sociability and its motility, triggering a novelty-seeking behavior in the Pfn2^−/− mice.

Ritualistic and repetitive behavior, accompanied by resistance to changes, is the second core symptom that characterizes ASD and can be measured in mouse models for ASD as stereotypic behavior (Peça et al., 2011; Schmeisser et al., 2012). We therefore assessed an array of stereotypic behaviors in control and Pfn2 mutant mice after transfer into a novel cage. In this context, no significant differences were observed for self-grooming and digging. However, Pfn2^−/− animals showed significantly higher occurrence of circling, and wall leaning stereotypies compared to Pfn2^+/− and wt control mice (Figure 3A), indicating a stronger susceptibility to repetitive behavior. Also jerking, a tic-like stereotypic behavior that can have both physiologic or pathologic origin, was significantly increased in Pfn2^−/− mice (Figure 3A).

Complementary to this general compulsive behavior, we evaluated the resistance to changes in Pfn2^−/− mice, using a more sophisticated paradigm. For this purpose, we tested the animals in a Y-maze set-up, where mice typically explore the arms of the maze in sequential order (spontaneous alternation, SPA). Pfn2^−/− mice displayed decreased SPA (Figure 3B), and while in our experiment no control mouse ever returned to the same arm (SAR), about 50% of the tested Pfn2^−/− mice showed this repetitive behavior (Figure 3C). Pfn2^−/− mutants also showed a marked deficit in decision making, as suggested by the longer latency to initiate exploration (Figure 3D). These data suggest a propensity to repetitive behavior and an increased resistance to changes in Pfn2^−/− mice.

ASD typically emerge within the first 2 years after birth and crying behavior of babies is the first way of communication in humans. Atypical crying behavior has been reported in incipient ASD infants, often resulting in a negative response of the mother (Esposito and Venuti, 2009). In mouse newborns, up to 8–9 days after birth, “crying” behavior is an important communication pathway, triggered by separation distress. Separated pups emit ultrasonic calls in the 30–110 kHz frequency range to attract their mother’s attention. We monitored ultrasonic vocalizations (USV) from P7 pups and found increased number of USVs in Pfn2^−/− pups compared to wt controls (Figure 4A, medians Pfn2^−/− males: 592, wt males: 165; Pfn2^−/− females: 990, wt females: 311 calls/5 min). Similar observations have been reported in other mouse models of autism (Scattoni et al., 2009; Tsai et al., 2012). Detailed analysis of the USV traces revealed that, despite Pfn2^−/− and wt control littermates displaying a collection of vocalizations of similar shape and frequency, Pfn2^−/− pups communicated with a more monotonous vocalization pattern, with persistent use of long (20–90 ms), sometimes interrupted, calls in the middle range of US frequencies (<70–75 kHz) and rather flat frequency modulation (examples shown in Figure 4B). Pfn2^−/− male and female pups showed, respectively, a 6- and 4-fold median increase of this type of “flat calls” compared to wt control littermates (Figure 4C, medians Pfn2^−/− males: 144, wt males: 22.5; Pfn2^−/− females: 216, wt females: 49 flat calls/5 min). In Pfn2^−/− mutants these “flat calls” were often arranged in particularly long trains of calls that were uniformly spaced (with a regular inter-call interval of 150 ± 10 ms, see Supplementary Figure 2) and more intense (<−50 dB, darker gray tone). In comparison to wt control littermates, Pfn2^−/− male and female pups presented significantly higher number of long trains of calls composed of more than 10 of these monotonous “flat calls” (Figure 4D). About 50% of Pfn2 mutant pups presented one or more very long trains of over 20 “flat calls,” while none of the control mice showed this behavior (call train example in Supplementary Figure 2). Vocalization patterns are usually made of groups of 2 or 3 closely spaced calls (with short inter-call intervals, <50 ms), within which frequency variations and harmonics are seen. We observed an increased group size and extended group of calls duration in both Pfn2^−/− males and females compared to wt control littermates (Figures 4E,F).

In summary, Pfn2^−/− pups appeared to be in distress when separated from the mother and engaged in a despair calling behavior with increased calling and reduced frequency modulation (monotony) compared to control pups.

In addition to the core symptoms and the possible communication deficits, impaired motor coordination is often found as a comorbidity in ASD (Fournier et al., 2010). We tested basic motor performance and coordination in Pfn2^−/− mice using a fixed rotation speed RotaRod paradigm, while we assessed motor learning by constantly increasing the rotation speed. In older Pfn2 mutants, motor performance and coordination at fixed low rotation speed was significantly impaired (Figures 5A,B), while younger mutant mice did not show significant deficits (Figures 5C,D). Motor learning, on the other hand, was unaffected at all ages in Pfn2^−/− mice, in line with previous findings in the conditioned learning paradigm (Pilo Boyl et al., 2007). Thus, Pfn2^−/− mice showed an age dependent impairment of motor performance that could depend on impaired coordination.

Coordination skills were further investigated with the Hanging test, which addresses both muscle strength and coordination between legs and body. The performance of Pfn2^−/− mice was markedly impaired at both young (Figure 6A) and older age (Figure 6B). In the group of older mice, the Pfn2^+/− animals showed a gene dosage-dependent intermediate phenotype.

To discriminate between muscle strength and coordination, we measured the mice grip strength. While gripping only with the forelimbs depends exclusively on the muscle control and strength, when the mice are allowed to grip with all four limbs, the coordination between fore- and hindlimbs affects the gripping strength. Both Pfn2^−/− and Pfn2^+/− mice showed about 20% less pulling strength with their forelimbs (Figure 6C, left), suggesting a mild deficit in muscle strength. However, grip strength with all four limbs was reduced only in Pfn2^−/− mice (Figure 6C, right), while Pfn2^+/− animals were comparable to wt controls, suggesting that the main motor function affected by loss of PFN2 was coordination.

Coordination issues could be spotted already at very early age (i.e., at P10) in Pfn2^−/− mice by the characteristic hindlimb clasping when gently lifted by the tail instead of the normal hindlimb outward splaying (Supplementary Figure 3A), a phenotype typically present in mouse models with motor coordination and/or ataxia issues and cerebellar dysfunction (Lalonde and Strazielle, 2011; Yerger et al., 2022). Interestingly, the gait quality was only mildly affected in a foot-printing assay, where we only observed a slightly shorter pace (Supplementary Figure 3B).

Since motor coordination largely depends on cerebellar function, we studied Purkinje cell physiology in Pfn2^−/− mice.

Purkinje cells (PC) are large inhibitory neurons of the cerebellar cortex organized in a monolayer between the inner granule cells and the external molecular layer, in which they extend their large dendritic arbors. These arbors receive extensive excitatory input from parallel and climbing fibers that generate dense trains, respectively, of simple and complex spikes (Sugimori and Llinás, 1990; Ghez and Thach, 2010). The central part of the cerebellum comprising the vermis region, the spinocerebellum, has been shown to regulate motor execution (Ghez and Thach, 2010).

We performed electrophysiological studies on Purkinje cells. Spontaneous excitatory post-synaptic currents (sEPSCs) recorded from PCs showed a significant leftward shift of the cumulative probability of the inter-event intervals in Pfn2^−/− mice, indicating a significant increase of the sEPSCs frequency (Figure 7A), while their amplitude distribution did not change significantly (not shown). Moreover, the paired-pulse ratio (PPR) between pairs of stimuli on parallel fibers with increasing inter-stimulus interval (ISI) was significantly reduced in Pfn2^−/− cells between 10 and 50 ms paired-pulse (PP) intervals (Figure 7B), indicative of an altered short-term synaptic plasticity that we can attribute to higher release probability of glutamatergic vesicles, since we have shown in a previous work that synaptic density is not altered in Pfn2^−/− mice (Pilo Boyl et al., 2007). To confirm the increased excitatory input from parallel fibers onto PCs, two additional experiments were performed. First, we studied the effect of PFN2 depletion when manipulating the release probability by increasing the extracellular [Ca²⁺] from 2 mM, in standard control ACSF conditions, to 4 mM. In these conditions, the increase of EPSCs amplitude was 22% larger in Pfn2^−/− than wt neurons (Figure 7C, Pfn2^−/−: 2.75 ± 0.25 vs. wt: 2.27 ± 0.24-fold increase), indicating that PCs in Pfn2^−/− animals experience higher excitatory stimulation. In a second experiment, frequency-dependent plasticity of the parallel fiber-PC synapse was analyzed. Increasing the frequency of stimulation from 0.05 to 0.2 Hz produced a mild potentiation of the EPSCs in both Pfn2^−/− and wt control cells (Figure 7D), however, an additional frequency increase up to 1 Hz produced potentiation only in wt control neurons but not in Pfn2^−/− cells that reached a plateau at much lower stimulation frequency than wt controls (Figure 7D, linear regressions, dotted lines). These results support the previous findings of a higher basal probability of glutamate release in Pfn2^−/− glutamatergic neurons.

In summary, our results showed that PFN2 is regulating cerebellar physiology by restraining the glutamatergic input in PCs from afferent projections. The findings in the cerebellum are in line with our previous observations in striatum (Pilo Boyl et al., 2007), strengthening the point that PFN2 has a critical function in limiting glutamatergic release in the CNS.

PCs of the cerebellum are a class of neurons particularly vulnerable to excessive glutamatergic input due to their double connectivity with parallel and climbing fibers. Excitotoxicity is predominantly a glutamate-dependent event, mediated by NMDA receptors, which, in conditions of dysregulated glutamate homeostasis, cause excessive and/or prolonged rises in intracellular calcium that trigger neuronal cell death pathways (for a review, see Armada-Moreira et al., 2020). The increased glutamatergic output observed in profilin 2 knock-out mice led us to investigate the structure of the PC layer in 4 and 10 months old Pfn2^−/− mice in spinocerebellar slices, the time points aligning with the motor coordination experiments previously described. At 4 months the PC layer in Pfn2^−/− mice was essentially normal (Figure 8A), showing no significant difference in the linear density of PCs compared to wt controls (Figure 8B). At this age motor performance was not affected. However, at 10 months of age we observed severe changes in Pfn2 mutant mice: the PC layer in the central part of the cerebellum was highly disorganized, with long stretches completely devoid of PC cell bodies (Figure 8C), resulting in significantly reduced linear PC density (Figure 8D, Pfn2^−/−: 2.27 ± 0.23 vs. wt: 4.60 ± 0.17 cells/100 μm). Severe motor coordination and performance deficits were indeed present at this age. These findings also suggest that PCs in Pfn2^−/− mice are slowly lost throughout life.

Interestingly, we also observed in Pfn2^−/− cerebella the presence of calbindin positive cells in the internal granular layer (Figures 8A,C right panels), which hint to a late neurodevelopmental migration defect, conceivable according to the proven function of profilins in cell motility (Zhou et al., 2019).

The etiology of ASD remains largely unknown, but one hypothesis suggests it could be based on a shift of the excitation/inhibition (E/I) ratio in the brain or in select neuronal circuits (Rubenstein and Merzenich, 2003; Lee et al., 2017). We previously documented increased glutamatergic transmission in Pfn2^−/− mice, through higher glutamatergic vesicle exocytosis (Pilo Boyl et al., 2007), and we provided additional findings in this work supporting the possibility that a general increase of the glutamatergic neurotransmission not counterbalanced by an increase of GABAergic inhibition could underlie the autistic-like traits of Pfn2^−/− mice. In order to prove this, inhibition needs to be measured. The CA3-CA1 Schaffer collaterals in the hippocampus are a well-characterized brain circuit to address the balance between excitatory and inhibitory synaptic inputs. We first analyzed excitatory transmission in this circuit using three different paradigms. We recorded miniature excitatory post-synaptic currents (mEPSCs) from CA1 pyramidal neurons, and then, while blocking GABA_A receptors, we measured the paired-pulse ratio (PPR) and calculated the input–output (I-O) relation in the Schaffer collaterals pathway. Glutamatergic transmission, similarly to all other previously tested circuits, was increased also in the hippocampus of Pfn2^−/− mice: mEPSCs were significantly more frequent in Pfn2^−/− mice compared to wt controls, as indicated by the leftward shift of the cumulative curve for the inter-event intervals (Figure 9A), with only a mild reduction in their amplitude (Supplementary Figure 4A). By means of extracellular field recordings we showed that the paired-pulse ratio at 40 ms ISI was reduced by 40% (Figure 9B, Pfn2^−/− 3.08 ± 0.30 vs. wt 1.80 ± 0.18) while the I-O relation was significantly increased in Pfn2^−/− mice compared to control littermates (Figure 9C), indicating that for a similar number of stimulated pre-synaptic fibers, Pfn2^−/− mice exhibited an increased post-synaptic response. There was no difference in the stimulation intensity to obtain the same fiber volley in the two genotypes (Supplementary Figure 4B), therefore similar numbers of pre-synaptic fibers responded to the same stimulation level.

Next, we studied inhibitory transmission in the same circuit, measuring miniature inhibitory post-synaptic currents (mIPSCs) in CA1 pyramidal neurons and calculating the I-O relation while blocking AMPA receptors. Contrary to the excitatory drive that was clearly increased, the inhibitory transmission was unchanged or mildly reduced. In the input–output relation experiment, Pfn2^−/− pyramidal neurons showed a tendential loss of GABA-dependent currents (Figure 9D), which was confirmed by a significant decrease in the amplitudes of mIPSCs (Figure 9E) with no change in mIPSCs frequencies (Figure 9F), suggesting a prevalent post-synaptic defect in CA1 pyramidal neurons, where PFN2 was shown to interact with gephyrin scaffolds and was proposed to regulate receptor densities (Giesemann et al., 2003; Murk et al., 2012). Overall our data indicate that in Pfn2^−/− mice excitatory synaptic inputs are increased while inhibitory synaptic inputs are reduced, thereby suggesting a shift of the E/I balance toward excessive excitation.

The Pfn2 knock-out (Pfn2^−/−) mouse model was previously described (Pilo Boyl et al., 2007). All mice used in the experiments were littermates generated by crossing heterozygous parents. Mice were socially housed with a standard 12 h light/dark cycle at 22 °C and 50–55% humidity, with free access to water and food pellets. All experiments were performed according to EU regulations (Licenses n. 19/2005-B and AZ 84–02.04.2013.A233).

Male mice 3–5-months old (“young”) were used in behavioral experiments, except where differently indicated. Mice indicated as “old” were aged 6–10 months. Maternal behavior: pregnant females were single-housed a week before delivery. Litter survival was assessed 10 days post-partum (P10). In a second approach, pregnant female and male were left together to allow cross-fostering. Maternal behavior, pup retrieval: P5 pups from Pfn2^−/− and wt control females were dispersed in the cage and time for retrieval of the first pup and of the entire litter was scored up to 30 min. Social interaction: the test chamber was built according to Moy et al. (2007). The experiment consisted of two 10 min trials: first, the test animal explored a tripartite chamber containing two empty cages in the outer compartments; second, the same test animal was confronted with one empty cage and one cage containing the stranger mouse, with a modification of the original test to ensure a non-aversive context for Pfn2^−/− mice that are hyper-excitable (Pilo Boyl et al., 2007): we used juvenile mice of 3.5 weeks (P25) of age as social partners. Stereotypic and repetitive behavior: self-grooming, digging, jerking, circling, and wall-leaning behavior was scored as number of events in 300 s after transfer into a novel cage. Y-maze: mice were allowed to freely explore the Y maze for 5 min and percent (%) of SPA (spontaneous alternation) and SAR (same arm return) was calculated with respect to total arm entries. Ultrasound vocalizations (USV): USVs were measured in P5-P9 male and female pups (Branchi et al., 2001). Pups were singularly taken from the mother and placed in a soundproof chamber at room temperature (22 °C) under a condenser ultrasound microphone (Avisoft Bioacoustics CM16/CMPA) connected to an UltraSoundGate 116Hb and data were collected for 300 s with the Avisoft Recorder 2.76 in 16-bit format. Analysis of USVs was performed using the Avisoft SASLab Pro software (Avisoft Bioacoustics, Berlin DE). The low frequency (<70–75 kHz) long calls (>20 ms) and the call trains (series of calls with a regular spacing of 150 ± 10 ms) were manually scored. RotaRod: an automated apparatus (TSE Systems, Germany) was used to test young (2.5–5 months) and older (5.5–8 months) mice. On the first day mice were subjected to a constant speed (3 rpm) protocol for 4 sequential trials of 300 s each with inter-trial interval of 1 h 30 min. Second, for five consecutive days mice were subjected to an incremental rotation speed paradigm (3–30 rpm) in 300 s, three times a day with inter-trial interval of 1 h 30 min; the best performance of the day was selected for every single mouse to assess motor ability, coordination and learning. Hanging: mice were hanged with the forelegs to a thin metal bar and stay time was measured. Grip: a grip strength meter (Harvard Apparatus) was used to test muscle strength of the animals. Mice were allowed to grab a metal grid with the forepaws only or with all 4 paws and were gently and constantly pulled by the tail until they lost the grip. The pulling strength was measured in Newton. Three measurements were taken for each mouse and the average was used for statistical purposes. Epileptic seizures: visible tonic and tonic–clonic seizures were triggered in a number of Pfn2^−/− mice by some sudden stimulus or stressful situation. The observations in this work, for statistical reasons, come exclusively from the mice that showed the seizures upon the opening of the cage for the regular observation of the mice or the changing of the bedding. When the epileptic mouse started to become rigid, arching the back, the timer was started. Depending on the type of seizure, the mouse might remain rigid and eventually fall on one side or start convulsive movements and jumps. The timer was stopped when the mouse started to recover, as signaled by straightening and self-grooming.

4- and 10- to12-months old mice were anesthetized and perfused with 4% formaldehyde. Brains were post-fixed O/N at 4 °C. Sagittal cerebellar slices 300 μm thick were cut with a VT1000 Vibratome (Leica Microsystems, Germany) in cold PBS. Slices were stained with antibodies following standard procedures. Anti-calbindin monoclonal antibody (Sigma, Germany, 1: 500), anti-mouse Alexa546-conjugated secondary antibody (Molecular Probes, Germany, 1:400) and the nuclear dye Hoechst (Roche, Germany, 0.5 μg/mL) were used for the staining. Imaging was performed with a Leica TSE SPE Confocal Microscope (Germany) acquiring 40 μm Z-stacks.

Acute hippocampal brain slices were obtained from P16-P24 mice. Animals were anesthetized with isoflurane, decapitated, the brains removed from the skull and placed into cold (4 °C) artificial cerebrospinal fluid (ACSF: 130 mM NaCl, 2.75 mM KCl, 1.43 mM MgSO₄, 2.5 mM CaCl₂, 1.1 mM NaH₂PO₄, 28.8 mM NaHCO₃, 11 mM Glucose), pH 7.3–7.4, 315–325 mOsm/L and gassed with carbogen (95% O₂, 5% CO₂). The hippocampi were dissected and transversally cut in 400 μM slices with a vibratome (Leica VT1200S) in chilled and carbogenated ACSF. Slices were then equilibrated in slice-chambers containing ACSF continuously gassed with carbogen for 30 min at 32 °C and subsequently stored at room temperature in ACSF. All recordings were done at room temperature (21–23 °C) in a recording chamber continuously perfused by carbogenated ACSF. If necessary, drugs were added to the ACSF. Recordings were obtained using 2–4 MΩ glass electrodes (Science Products GB150TF-8P, Hoffenheim, Germany) filled with intracellular solution and a Multiclamp 700B amplifier (Axon Instruments), and data were digitized using a Digidata 1322A (Axon Instruments). mIPSCs: miniature inhibitory post-synaptic currents (mIPSCs) were recorded from CA1 pyramidal neurons at −70 mV holding potential for 10 min and analyzing 2 min (ca. 120 events) in the middle region of the recording, using a high chloride intracellular solution containing (in mM): 90 CsCl, 20 Cs-gluconate, 8 NaCl, 2 MgCl₂, 10 HEPES, 1 EGTA, and 2 QX-314, pH 7.2 (290 mOsM/L). mIPSCs were recorded in ASCF supplemented with NBQX (10 μM) to block AMPA receptors and TTX (0.2 μM) to block action potentials. Evoked IPSCs: IPSCs were evoked with a stimulation electrode placed ca. 100 μm away from the soma of the recorded pyramidal cell. The position of the stimulation electrode was adjusted such that the smallest stimulus (5 μA) elicited a current of ca. 500 pA. mEPSCs: mEPSCs were recorded from CA1 pyramidal neurons in acute hippocampal brain slices at a holding potential of −70 mV, for at least 10 min to make sure that 120 events could be analyzed using the following intracellular solution (in mM): 150 Cs-gluconate, 8 NaCl, 2 MgATP, 10 HEPES, 0.2 EGTA and 5 QX-314 ([2-[(2,6-dimethylphenyl)amino]2-oxoethyl]-triethylazanium bromide), pH 7.2 (290 mOsM/L). mEPSCs were recorded in ASCF supplemented with picrotoxin (PTX, 100 μM) to block GABAA receptors, tetrodotoxin (TTX, 0.2 μM) to block action potentials and trichlormethiazide (TCM, 250 μM) to increase mEPSC frequency. fEPSCs: fEPSP were recorded from acute hippocampal slices, disconnecting the Schaffer collaterals cutting with a scalpel in the CA3 area to prevent recurrent spontaneous excitation. Schaffer collaterals after the cut were stimulated and synaptic responses were recorded in the stratum radiatum of the CA1 region, using glass microelectrodes filled with ACSF. fEPSCs were recorded in ASCF supplemented with picrotoxin (PTX, 100 μM) to block GABAA receptors. Before the experiment was started, recordings were registered for 20 min until fEPSP reached a stable baseline. For PPF experiments Schaffer collaterals were stimulated with a stimulus interval of 40 ms. For the input–output relation, the input was the amplitude of the fiber volley representing the strength of the action potential arriving from the Schaffer collaterals; the output was defined as the slope of the resulting fEPSP. Fiber volley amplitudes of 0.05, 0.1, 0.15, 0.2 and 0.25 mV were evoked and the resulting fEPSC plotted. All data were analyzed using Clampfit 10.2 (Molecular Devices), a custom written Matlab routine (MathWorks), and GraphPad Prism 5 (GraphPad Software). Values are given as means ± SEM.

Parasagittal cerebellar slices (250 μm) were prepared from 1-month old mice and immersed in ice-cold gassed (95% O₂, 5% CO₂) sucrose-based artificial cerebrospinal fluid (ACSF) containing: 87 mM NaCl, 75 mM sucrose, 2 mM KCl, 7 mM MgCl₂, 0.5 mM CaCl₂, 25 mM NaHCO₃, 1.2 mM NaH₂PO₄ and 10 mM glucose, pH 7.4, 300–305 mOsm. Slices were then allowed to recover for at least 1 h at RT in standard ACSF (125 mM NaCl, 2 mM KCl, 1.2 mM MgCl₂, 2 mM CaCl₂, 25 mM NaHCO₃, 1.2 mM NaH₂PO₄ and 10 mM glucose) before measurement. In high [Ca²⁺] experiments, [Ca²⁺] concentration was raised to 4 mM and Mg²⁺ concentration was adjusted to 1 mM. Patch-clamp recording: whole-cell patch-clamp recordings on PCs were performed with a Multiclamp 700B amplifier (Molecular Devices, USA) using glass electrodes (3–4 MΩ) pulled with a vertical puller (PC-10, Narishige, Japan). Intracellular solution (140 mM Cs-Methanesulfonate, 10 mM Hepes, 0.5 mM EGTA, 2 mM MgATP, 0.3 mM Na₃GTP, 2 mM MgCl₂) was adjusted to 295–300 mOsm, pH 7.2. Signals were acquired with the DigiData-1440A (AutoMate Scientific, USA) amplifier using pCLAMP-v10 software and analyzed off-line with Clamp-fit 10 program (Axon Instruments, USA). Excitatory post-synaptic currents (EPSCs) were recorded clamping the cell at −70 mV. Spontaneous EPSCs were recorded during an initial 10 min baseline period followed by application of bicuculline (10 μM, Sigma Aldrich, USA) for 15 min. A concentric bipolar stimulating electrode (SNE-100 × 50 mm long Hugo Sachs Elektronik - Harvard Apparatus GmbH, Germany) was placed in the molecular layer of the cerebellar cortex for parallel fiber stimulation. Series resistance (Rs) was not compensated during voltage-clamp experiments to avoid increased electrical noise in the trace. Rs was constantly monitored over time and recordings in which it changed more than 20% were discarded. For Paired-Pulse experiments, pairs of stimuli were applied every 20 s. Stimulus was delivered through an A320R Isostim Stimulator/Isolator (WPI, USA). PPR was calculated as the ratio between the amplitude evoked by the second stimulus (A2) and the amplitude of the first (A1; A2/A1).

In behavioral tests one-way ANOVA or two-way ANOVA with repeated-measures in one factor were applied to assess statistical differences, followed by Tukey’s multiple comparisons post hoc test. Unpaired two-tailed t-Student’s test was used on two sets of data with normal distributions, while the non-parametric Mann–Whitney test for independent measures was used if distributions were not normal. Input–output experiments were subjected to regression analysis. Events and amplitudes frequency distributions were compared with the non-parametric Kolmogorov–Smirnov two-sample test by stochastically sampling a select number of values from each cell of the same genotype to produce a single distribution per genotype and reiterating the process 20 times, selecting at the end the highest p value resulting from all the comparisons. Kaplan-Mayer survival curves were analyzed applying the non-parametric Peto-Prentice generalized Wilcoxon test.