Enhanced LTP of population spikes in the dentate gyrus of mice haploinsufficient for neurobeachin

Deletion of the autism candidate molecule neurobeachin (Nbea), a large PH-BEACH-domain containing neuronal protein, has been shown to affect synaptic function by interfering with neurotransmitter receptor targeting and dendritic spine formation. Previous analysis of mice lacking one allele of the Nbea gene identified impaired spatial learning and memory in addition to altered autism-related behaviours. However, no functional data from living heterozygous Nbea mice (Nbea+/−) are available to corroborate the behavioural phenotype. Here, we explored the consequences of Nbea haploinsufficiency on excitation/inhibition balance and synaptic plasticity in the intact hippocampal dentate gyrus of Nbea+/− animals in vivo by electrophysiological recordings. Based on field potential recordings, we show that Nbea+/− mice display enhanced LTP of the granule cell population spike, but no differences in basal synaptic transmission, synapse numbers, short-term plasticity, or network inhibition. These data indicate that Nbea haploinsufficiency causes remarkably specific alterations to granule cell excitability in vivo, which may contribute to the behavioural abnormalities in Nbea+/− mice and to related symptoms in patients.

Neurobeachin (Nbea) belongs to the BEACH (beige and Chediak-Higashi) domain-containing protein family whose members are involved in the trafficking of membrane proteins (for review, see 1 ). Nbea is highly expressed in endocrine cells and neurons where it is enriched at tubulovesicular endomembranes and in the postsynaptic compartment, and plays a role in vesicle secretion, presumably via regulation of protein kinase A (PKA) [2][3][4] . Previous research established Nbea as an essential molecule in synapse function because homozygous Nbea knockout mice (Nbea −/− ) die immediately after birth from breathing paralysis due to a dramatic reduction of evoked synaptic transmission at the neuromuscular junction and in respiratory neurons in the brainstem 5,6 . Despite this dramatic phenotype, the exact role of Nbea is still incompletely understood as its effects are pleiotropic and not all phenotypic observations could be repeated in independent models and assays. While it is widely agreed that complete deletion of Nbea in mice leads to defects of spontaneous and evoked synaptic transmission at excitatory and inhibitory synapses [4][5][6][7] , only some of the defects can be attributed to reduced surface levels of postsynaptic AMPA, NMDA and GABA A receptor subunits 4,[7][8][9][10] . Moreover, while the overall density of asymmetric, presumably excitatory synaptic contacts and synaptic vesicle protein levels are diminished in acute slices from Nbeadeficient embryonic mice 5 , these parameters appear normal in primary neuronal cultures 4,8 . In contrast, the number of excitatory synapses on spines, small protrusions arising from dendritic shafts where most excitatory synapses reside, is reduced in both knockout cultures and brain tissue 8,9 . Additional research identified roles of Nbea in dense-core vesicle function in mouse pancreatic cells and platelets 3,11 and, possibly, in transcriptional regulation 12 . Collectively, these findings indicate that Nbea is a fascinating but elusive molecule that mandates the use of highly specific techniques to uncover its functions.
Nbea was identified as a candidate gene for autism spectrum disorder (ASD) based on a translocation of the Nbea gene found in a patient with ASD 13 . Deletions of the chromosomal regions on which Nbea resides have also been discovered in other patients with ASDs [14][15][16] and developmental/intellectual disability 17 . The Nbea gene is located across a common fragile site, making it more vulnerable to breakage during metaphase 18 . Compared to other genes, the human Nbea gene has a lower tolerance for disruptive mutations, which indicates that it is under purifying selection and thus a likely candidate gene for ASD 19 . To address this important aspect, several groups investigated heterozygous Nbea +/− mice, which exhibit a 30% reduction in forebrain Nbea protein levels 8 , for construct and face validity. In fact, autism-related abnormalities are present in Nbea +/− mice including altered social behaviours, increased self-grooming, delayed spatial learning and memory, increased conditioned fear responses, and impaired fear memory extinction 20,21 . Along with these findings, long-term potentiation (LTP) was enhanced in acute slices of the CA1 region of the hippocampus, an important site for spatial learning 20 . These intriguing results indicate that Nbea might play an important role in the regulation of synaptic plasticity in the hippocampus, but this finding has not been confirmed by in vivo experiments.
Here, we performed electrophysiological recordings in the hippocampus of Nbea +/− mice in vivo to study the potentially important role of Nbea in synaptic transmission and plasticity in the intact brain. Field potentials evoked by stimulation of the perforant path were recorded in the granule cell layer of the dentate gyrus in anaesthetised mice. We found that Nbea +/− mice displayed enhanced LTP of the granule cell population spike, which shows that Nbea has a functional role in vivo that could help explain the behavioural phenotypes observed in haploinsufficient mice.

Results
The granule cell population spike is enhanced following LTP induction in Nbea +/− mice. Nbea deficiency is known to improve the induction and maintenance of long-term potentiation (LTP) at Schaffer collateral synapses in acute slices of the hippocampus in vitro 20 . Here, we tested whether LTP is also altered in intact animals, a prerequisite for drawing firm conclusions about the pathomechanism in human patients with Nbea haploinsufficiency 13 . LTP at perforant path-to-granule cell (PP-GC) synapses in Nbea +/− and wild-type littermate control mice was induced by theta-burst stimulation (TBS) protocols. Since Nbea +/− mutant mice lack only about 30% of Nbea protein 8 , we decided to use a previously established stimulation protocol 22 consisting of a weak TBS followed by strong TBS 30 min after the weak TBS to uncover even subtle differences between both groups (Fig. 1a-e). The stimulation intensities were set to elicit a stable population spike and were similar between groups (Fig. 1b, 4 wild-type and 7 Nbea +/− mice, 227.5 ± 45.71 µA vs. 227.4 ± 43.35 µA, unpaired Welch's t test, P = 0.9991). The pre-TBS population spike amplitude was also quite similar between groups (Fig. 1c, 5 wild-type and 7 Nbea +/− mice, 4.861 ± 0.9791 mV vs. 4.292 ± 0.3686 mV, unpaired Welch's t test, P = 0.5517). During the 30 min following the weak TBS, we observed an almost 40% higher potentiation of the population spike relative to the pre-TBS baseline in Nbea +/− compared to control mice (Fig. 1e, 135.6 ± 5.692% for wild-type and 173.8 ± 8.459% for Nbea +/− mice, unpaired Welch's t test, P = 0.0046). The strong TBS further increased the spike amplitude in wild-type mice, thus substantiating the efficacy of LTP induction by TBS in control mice. In contrast, the population spike amplitude was not further enhanced by the second TBS in Nbea +/− mice, revealing that a weak TBS is sufficient to induce maximal potentiation in Nbea +/− mice (Fig. 1d). There was no significant difference between groups in the spike amplitude potentiation following the strong TBS, neither in the first 30 min (171.2 ± 12.02% for Nbea +/− , 145.2 ± 9.004% for wild-type, unpaired Welch's t test, P = 0.1405, Fig. 1e) nor the last 30 min (162.8 ± 9.152% for Nbea +/− , 137.7 ± 9.792% for wild-type, unpaired Welch's t test, P = 0.0929, Fig. 1e).
In a separate set of experiments, we induced LTP using only the weak TBS protocol in order to determine whether the Nbea +/− mice also displayed an enhancement in synaptic efficacy as measured by the fEPSP slope ( Fig. 1f-j). The population spike was minimised to allow for accurate assessment of the initial fEPSP slope. Both the stimulation intensities (Fig. 1g, 3 wild-type and 4 Nbea +/− mice, 275.0 ± 43.30 µA vs. 323.8 ± 44.60 µA, unpaired Welch's t test, P = 0.4767) and the pre-TBS slopes (Fig. 1h, 0.5564 ± 0.07608 mV/ms for wild-type and 0.5433 ± 0.1521 mV/ms for Nbea +/− mice, unpaired Welch's t test, 0.9421) were similar between groups; however, the population spike amplitudes were highly variable and therefore not included in the analysis. Interestingly, there were no significant differences in the potentiation of the fEPSP slope in the first 30 min after wTBS (Fig. 1j, 4 Nbea +/− and 3 wild-type mice, 144.1 ± 12.57% vs. 157.1 ± 14.08%, unpaired Welch's t test, P = 0.5302). Together, these data demonstrate that (1) even a moderate reduction in Nbea protein levels is sufficient to alter long-term potentiation of population spikes in the dentate gyrus of living mice, but that (2) potentiation fEPSP slope is apparently not affected by Nbea.

Excitatory transmission at PP-GC synapses is unchanged in Nbea +/− mice. Previous studies
revealed a strong reduction of evoked glutamatergic postsynaptic responses in the complete absence of Nbea protein in acute slices and primary neuronal cultures of homozygous mice 4,5 , establishing that Nbea is essential for basic synaptic transmission. However, slice recordings of fEPSP slopes from heterozygous neurons were remarkably normal 20 and spontaneous glutamatergic postsynaptic responses only showed a slight reduction in frequency and rise time but not in amplitude 8 . In order to explore the pre-and postsynaptic properties of haploinsufficient Nbea +/− mice in vivo, we probed the function of excitatory medial PP-GC synapses with different stimulation protocols.
To exclude the possibility that morphological changes such as a reduction in the number of synaptic contacts interfered with our analysis of functional plasticity, we examined the relevant ultrastructure of these synapses. Samples of the middle molecular layer of the hippocampal dentate gyrus, the very location of medial PP-GC synapses, were imaged in wild-type littermate control and Nbea +/− mice by transmission electron microscopy. No apparent structural alterations were detected concerning the shape of asymmetric (type 1) synapses (representative image in Fig. 2a) or their position on dendritic spine heads and dendritic shafts. Moreover, quantification of the area density of the number of asymmetric synapses in the middle molecular layer revealed no significant www.nature.com/scientificreports/  Normal synaptic ultrastructure and basal synaptic transmission at perforant path-to-granule cell synapses in Nbea +/− mice. (a) Representative electron micrograph of the middle molecular layer of the dentate gyrus from a Nbea +/− mouse, showing four asymmetric (type 1) synapses terminating on dendritic profiles (arrowheads). One bouton forms a spinous synapse on an unequivocal dendritic spine (arrow). Scale bar, 200 nm. (b) Quantitative comparison of the area density of asymmetric synapses in the middle molecular layer of the dentate gyrus from wild-type and Nbea +/− mice (unpaired Welch's t test, P = 0.4295). (c) Paired-pulse facilitation of the fEPSP amplitude elicited at stimulus intensities subthreshold for spike generation was not different in wild-type and Nbea +/− mice (two-way ANOVA with Bonferroni's multiple-comparison tests, P = 0.2876 for genotype, P = 0.2434 for interaction). Example traces show the response to PPF stimulation at an interpulse interval (IPI) of 15 ms. Inset: stimulation intensities used to elicit PPF were not significantly different between groups (unpaired Welch's t test, P = 0.1354). (d) The response of the fEPSP slope to perforant path stimulation at increasing stimulus intensities shows no difference between wild-type and Nbea +/− mice (two-way ANOVA with Bonferroni's multiplecomparison tests, P = 0.6143 for genotype, P = 0.7891 for interaction). All values reported as means ± SEM.

Scientific RepoRtS
| (2020) 10:16058 | https://doi.org/10.1038/s41598-020-72925-4 www.nature.com/scientificreports/ differences between control and mutant mice (Fig. 2b, n = 7 sections from 2 mice per genotype, 49.71 ± 6.221 for Nbea +/− and 55.71 ± 3.790 for wild-type mice, unpaired Welch's t test, P = 0.4295). Next, we measured paired-pulse facilitation (PPF), a form of presynaptic short-term plasticity which depends mostly on the condition of presynaptic terminals 23 . To elicit PPF, we stimulated the perforant path axons at increasing interpulse intervals (IPIs). To avoid activation of the postsynaptic membranes by the firing of granule cells, we used intensities that were subthreshold for population spike evocation. We then measured the augmentation of the fEPSP amplitude (PPF) as the ratio of the two fEPSP amplitudes induced by two successive stimuli at IPIs from 15 to 100 ms (Fig. 2c). As shown previously, PP-GC synapses express PPF at low IPIs which declines with increasing IPI 22,24 . The stimulation intensities (in µA) used to elicit PPF were not significantly different between groups (52.28 ± 5.526 for Nbea +/− and 68.30 ± 8.561 for wild-type, unpaired Welch's t test, P = 0.1354). The variation observed in the degree of PPF was not due to genotype (n = 19 Nbea +/− and 10 wild-type mice, P = 0.2876 determined by two-way ANOVA with Bonferroni's multiple-comparison tests) or the interaction between IPI and genotype (P = 0.2434), suggesting that properties of short-term plasticity are not altered by the reduction in Nbea protein levels.
Furthermore, to exclude the possibility that input-output (I-O) relations were changed in Nbea +/− neurons, I-O curves were measured to probe differences in basal excitatory transmission of PP-GC synapses. We stimulated the perforant path across a range of stimulation intensities (30-800 µA, Fig. 2d) and analysed the slope of the resulting fEPSPs, since the slope is an indicator of synaptic strength 25 . We found that the variation in the I-O relationship was not caused by the genotype (n = 14 Nbea +/− and 10 wild-type mice, two-way ANOVA with Bonferroni's multiple-comparison tests, P = 0.6143), or the interaction between stimulation intensity and genotype (P = 0.7891), confirming results obtained from slice recordings at the Schaffer collateral terminals 20 . Together, these results indicate that a moderate reduction in Nbea protein levels by about 30% in haploinsufficient mice 8 does not affect properties of basal excitatory transmission at hippocampal synapses in vivo.
EPSP-spike coupling is slightly reduced in Nbea +/− mice. One advantage of our recording method is that the synaptic strength and the excitability of the granule cell population can be assessed separately. The population spike is superimposed on the fEPSP and represents the summed firing activity of the granule cells 26 . Therefore, we studied the excitability of granule cells in control versus Nbea +/− mice by measuring the population spike amplitudes across a range of stimulus intensities from 30 to 800 µA. Comparing the results from both groups (Fig. 3a), there were no significant differences in the ability of Nbea haploinsufficient granule cells to produce action potentials compared to their wild-type littermates (n = 14 Nbea +/− and 10 wild-type mice, two-way ANOVA with Bonferroni's multiple-comparison tests, P = 0.0676). The interaction between stimulation intensity and genotype was also not a significant factor in explaining the variation (P = 0.4168). These data suggest that the action-potential-generating capability of GCs is not altered by deletion of one allele of the Nbea gene.
To further examine the relationship between the synaptic input strength and firing ability of granule cells, we performed a fEPSP-population spike (E-S) coupling analysis by plotting fEPSP slopes against population spike amplitudes (Fig. 3b). The coupling between synaptic drive and the action potential firing ability reflects the intrinsic excitability of the neurons 27 as well as the excitation/inhibition balance of the local circuits 28 . This analysis revealed a rightward shift and a lower top parameter of the Boltzmann-fitted Nbea +/− E-S curve (Fig. 3c, n = 7 wild-type and 13 Nbea +/− mice, 13.96 ± 0.8924 vs. 10.62 ± 1.044, unpaired Welch's t test, P = 0.0261) with respect to the wild-type curve. The rightward shift indicates that heterozygous mice require a stronger synaptic input than their wild-type littermates to produce a similar spike response, i.e. reduced E-S coupling. The reduction in the top parameter reflects the lower maximal spike amplitude in heterozygotes. Thus, our in vivo recordings revealed a reduced tendency of Nbea +/− neurons to translate changes in synaptic input into changes in firing. We therefore tested whether this reduced E-S coupling was due to differences in the balance of excitation and inhibition in the dentate gyrus of Nbea +/− mice by examining network inhibition.
Network inhibition is unaltered in Nbea +/− mice. Recordings in intact mice allow the assessment of distinct properties of neuronal network behaviour that are often difficult to replicate in acute slices or cultures, such as the influence of inhibition on firing activity 29 . Earlier studies have reported that inhibitory synapse function was severely compromised in both acute slices and primary neuronal cultures of null-mutant Nbea neurons 4,5,7,8 . We assessed network inhibition in the dentate gyrus of Nbea +/− mice in vivo by measuring pairedpulse inhibition (PPI) at short IPIs. In addition, we determined disinhibition (PPDI) at longer IPIs. PPI reflects the recruitment of GABAergic interneurons mediating granule cell inhibition by both feedforward and feedback mechanisms 26,30,31 , and PPDI reflects the inhibition of these interneurons 32 . We first applied stimulation intensities triggering population spikes of approximately 1 mV but could not detect significant differences in PPI between Nbea +/− and wild-type mice. PPI/PPDI curves were nearly congruent (Fig. 4a, n = 15 Nbea +/− and 12 wild-type mice, two-way ANOVA with Bonferroni's multiple-comparison tests, P = 0.6760 for genotype, P = 0.3239 for interaction) and the IPI at which PPI switched to PPDI was similar in both groups (37.09 ± 1.937 ms for Nbea +/− and 34.27 ± 2.453 ms for wild-type mice, Welch's t test, P = 0.3766). To determine whether there was a difference between wild-type and Nbea +/− mice at a higher stimulus intensity, we applied 800 μA/0.2 ms stimuli to recruit as much feedback inhibition as possible. The genotype effect was not significant (Fig. 4b, n = 15 Nbea +/− and 12 wild-type mice, two-way ANOVA with Bonferroni's multiple-comparison tests, P = 0.4264), but the interaction between genotype and IPI was significant (P = 0.0057). This indicates that there might have been an effect of the genotype at some IPIs, but there were no major functional consequences to this discrepancy since the PPI/PPDI shift was not significantly different between groups (40.30 ± 1.995 ms for Nbea +/− and 39.11 ± 2.241 ms for wildtype mice, unpaired Welch's t test, P = 0.6948). These results suggest that the haploinsufficiency of Nbea does www.nature.com/scientificreports/ not primarily manifest itself at inhibitory synapses, although more subtle changes or effects at distinct types of inhibitory neurons cannot be excluded based on our finding of decreased E-S coupling.

Discussion
Using in vivo field potential recordings from the mouse hippocampus, we provide evidence for an involvement of the neuronal BEACH-domain containing protein Nbea in granule cell excitability in the intact brain. While LTP of the population spike was enhanced in the dentate gyrus of Nbea +/− mice, LTP of the fEPSP slope as well as basal synaptic transmission, short-term presynaptic plasticity, and network inhibition were unaffected. These are important findings because they demonstrate that even a moderate reduction of Nbea protein level 8 can cause functional abnormalities in distinct neural circuitry.
Our results indicate an enhanced response to LTP stimulation in Nbea +/− mice. Following LTP induction with a weak TBS protocol in vivo, we found a comparable increase in synaptic potentiation and a significantly larger increase in the granule cell spike response to perforant path stimulation in Nbea +/− mice. A previous study reported enhanced LTP in the CA1 region in acute slices from female Nbea +/− mice following a similar TBS protocol 20 , indicating that Nbea plays a functional role in different hippocampal subregions and in different aspects of LTP. What is the behavioural relevance of enhanced LTP? Although LTP is widely believed to be essential for synaptic modifications underlying memory formation, enhanced LTP does not necessarily lead to improved memory formation. In fact, multiple studies have reported an association between impairments in learning and memory and augmented LTP [33][34][35][36][37][38] . In line with the inverse correlation between spatial memory and LTP, Nbea +/− mice showed impaired learning and memory in behavioural tests 20 . The learning deficits were most pronounced in delayed spatial learning tested in the Morris water maze task, which is highly sensitive to hippocampal lesions 39   www.nature.com/scientificreports/ rats in the Morris water maze, whereas LTP of the EPSP was not correlated 40 . This suggests that potentiation of the population spike, and not the EPSP, forms the physiological basis for delayed spatial learning. Furthermore, while both CA1 and the dentate gyrus are important sites for spatial learning 41,42 , the dentate gyrus is also important for the discrimination of similar environments (pattern separation) 43 . In line with this, an increase in LTP has been shown to coincide with less specific hippocampal place fields 44 . During spatial exploration, the capability for enhanced LTP of the population spike may lead to overactivation of the granule cells and hence to a seemingly paradoxical decrease of information content in the hippocampal neuronal network, which may be associated with learning deficits 33 . Therefore, our finding that Nbea +/− mice displayed enhanced population spike LTP in the dentate gyrus can help explain their impairments in spatial memory. From this we predict that future behavioural tests might be successful in revealing deficits in pattern separation tasks in Nbea +/− mice. Despite the difference in spike LTP, we found that basal synaptic function was unaffected by Nbea deficiency. There were no significant changes in the input-output relationship of fEPSP slopes to the stimulation intensity  www.nature.com/scientificreports/ in Nbea +/− mice as compared to their wild-type littermates. Therefore, the basal glutamate-mediated synaptic transmission is likely not affected by the haploinsufficiency of Nbea. These results are consistent with previous work comparing input-output curves in acute hippocampal slices between wild-type and Nbea +/− mice where no differences were detected at low stimulation intensities 20 . However, previous experiments using Nbea −/− neurons found that in contrast to spontaneous release events, the amplitudes of action-potential induced excitatory postsynaptic currents were significantly affected by the complete lack of Nbea 9 . This can be explained directly by the impaired AMPA receptor subunit trafficking to postsynaptic membranes 9 which limits the number of glutamate receptors. While the amount of receptors in Nbea −/− neurons might suffice to respond to the neurotransmitter release by single vesicles, they are more than saturated by an action potential induced glutamate release 9 . Unaltered evoked glutamatergic responses in Nbea +/− mice indicate that either AMPAR-subunit trafficking is less impaired in haploinsufficient animals or they have compensatory mechanisms to generate equally strong responses as wild-type mice. Presynaptic short-term plasticity at perforant path-granule cell synapses was not affected by a reduction in Nbea levels as revealed by unchanged paired-pulse facilitation (PPF). This finding is in line with in vitro experiments using both heterozygous 20 as well as knockout 4 neurons. In acute hippocampal slices, PPF at CA1 synapses was not affected 20 . In recordings from synaptically connected pairs of neurons in culture, it was found that the synaptic phenotype of Nbea knockout was only observed if the responding postsynaptic neuron was Nbea-deficient, indicating that presynaptic short-term plasticity mechanisms are unaltered 4 . While these results are seemingly in opposition to the earlier results indicating a strong presynaptic role for Nbea 5,6 , this discrepancy might reflect a difference in Nbea function between hippocampal and cortical neurons where Nbea acts mostly postsynaptically 4,20 and brainstem neurons 5 and neuromuscular junctions 6 where Nbea also has an important presynaptic function.
In contrast to the measures of synaptic transmission, action potential generation (measured by the population spike amplitude) in the granule cells was slightly, but non-significantly, reduced in Nbea +/− mice. The E-S analysis, in which the population spike amplitudes are plotted against the fEPSP slopes for each stimulation intensity, revealed diminished granule cell firing in response to an EPSP of a given size in Nbea +/− mice as compared to wild-type mice, suggesting that Nbea might regulate the intrinsic excitability of granule cells (albeit at very high stimulation intensities that may not reflect physiological levels of synaptic activity). A decrease in E-S coupling could reflect a higher level of feed-forward inhibition in Nbea +/− mice, which is mediated by basket cells 45 or it could be due to changes in voltage-gated conductances that increase the intrinsic excitability of the granule cells. Further investigations using paired recordings are necessary to answer this question definitively.
To our surprise, we were unable to observe any differences in network inhibition between Nbea +/− and wild-type mice as assessed by PPI/PPDI measurements. This was unexpected since Nbea was shown to have a greater effect on inhibitory than excitatory synaptic transmission in brainstem acute slices, which may result in an increased excitation-inhibition (E/I) ratio 5 . In line with this, other studies implicated Nbea in the trafficking of glycine and GABA A receptors to postsynaptic membranes 4,7,46 and it was shown that Nbea −/− cells had a 40% reduction in cell surface levels of GABA A receptors 4 . Therefore, we expected PPI to be impaired by a reduction in Nbea. However, most of these previous studies were performed using dissociated neuronal cultures 4,7,46 , where the ratio of GABAergic to glutamatergic cells and the connectivity can differ from the in vivo conditions 47 . Another reason for unchanged network inhibition might be that the loss of one allele in Nbea +/− mice is insufficient to reduce GABA A receptor levels.
Based on our data, we conclude that the increased potentiation of the population spike we observed in Nbea +/− mice is due to changes in the intrinsic excitability of granule cells, rather than synaptic mechanisms. This is suggested by our finding that the increase in the population spike of Nbea +/− mice is greater than the fEPSP slope increase, which was comparable to that of wild-type mice following the single weak TBS protocol. The potentiation of the population spike that cannot be accounted for by the potentiation of the EPSP is known as E-S potentiation 48,49 . Since network inhibition measured by PPI was unchanged in Nbea +/− mice, differences in the intrinsic excitability of granule cells are the likeliest cause for the difference in E-S potentiation. Several mechanisms, such as an increase in the conductance of voltage-gated sodium channels 50 or a reduction in A-type potassium current 51 have been proposed to underlie E-S potentiation, though calcium influx appears to be essential 49 . Importantly, the action potential threshold does not necessarily correlate with E-S potentiation 49 , so the difference we observed in (pre-TBS) E-S coupling (which is a reflection of the firing threshold and synchrony) is not an indication of the degree of (post-TBS) E-S potentiation in the same cells. The exact mechanism by which Nbea might affect intrinsic excitability is unknown, but it is conceivable that Nbea might regulate the membrane trafficking of A-type potassium channels via its PH (pleckstrin homology)-BEACH domain in a similar manner to its trafficking of GluA2 AMPAR subunits 9 . Another possibility lies in the function of Nbea as an A-kinase anchoring protein 2 . It was previously shown that A-kinase anchoring proteins interact with Kv4.2 channels, which mediate the backpropagation of action potentials, in order to enhance their surface expression 52 and that the turnover of Kv4.2 channels is regulated by PKA in distal dendrites of CA1 pyramidal cells 53 . The downregulation of Kv4.2 channels has also been proposed as a potential mechanism for the changes in intrinsic excitability observed after LTP induction in the dentate gyrus in vitro 54 . Thus, Nbea might influence the intrinsic excitability of granule cells by trafficking and/or regulating the turnover of voltage-gated ion channels in dendrites.
ASD is often described as a synaptopathy due to many animal models exhibiting changes in synaptic function 55,56 . However, changes in intrinsic neuronal excitability might have stronger impact on the firing activity of neurons that exhibit high levels of dendritic voltage attenuation such as dentate granule cells 57 . Therefore, our finding that the autism candidate gene Nbea affects the intrinsic plasticity of granule cells highlights the importance of studying non-synaptic mechanisms of plasticity. Further research will be necessary to unravel the precise molecular mechanisms of this proposed function, but the difference in population spike LTP we observed suggests that there are important functional consequences to this difference in excitability. Input-output recordings determine granule cell responses to different stimulation intensities and allow insight into properties of basal excitatory transmission in the dentate gyrus. Three responses were collected and averaged for each stimulus intensity ranging from 30 to 800 µA (0.1 ms stimulus duration). The amplitude of the population spike was defined as the average of the amplitude from the first positive peak to the antipeak and the amplitude from the antipeak to the second positive peak. Only those mice that exhibited a population spike at a stimulation intensity of 300 µA or less were included in the analysis to avoid including mice in which the stimulation electrode was not optimally placed. For the analysis of the slope of the fEPSP, only the early component of the waveform, which is less affected by the population spike, was used. In the analysis relating the fEPSP slope to spike amplitude (E-S plot) each curve was fitted using a Boltzmann function.

Scientific
In order to investigate the conditions of the presynaptic membrane at perforant path-granule cell synapses, paired-pulse facilitation (PPF) of the fEPSP was measured. To achieve PPF of the fEPSP, two subsequent pulses at an intensity below the population spike threshold (between 20 and 120 µA/0.2 ms), with interpulse intervals (IPIs) varying from 15 to 100 ms were applied. Six paired-pulse responses at each IPI were collected and averaged, and the percentage of facilitation was calculated as relative potentiation of the second fEPSP to the first fEPSP.
Paired-pulse inhibition (PPI) and disinhibition (PPDI) of the population spike were examined to assess the efficacy of inhibition in the hippocampal network. Two pulses at weak stimulation intensities (evoking approximately 1 mV population spikes) were delivered with an IPI ranging from 20 to 100 ms. A total of six paired-pulse responses were recorded at each IPI and averaged. The percentage of change of the population spike amplitude was calculated as the relative change of the spike amplitude following the second pulse in comparison to the spike amplitude following the first one. A Boltzmann equation was used to fit PPI/PPDI curves, hereby obtaining the mean IPI at which the amplitudes of both population spikes are expected to be equal.
As a measure of synaptic plasticity and long-term potentiation (LTP), we compared responses with baseline stimulation prior to theta-burst stimulation (TBS) with responses subsequent to TBS. TBS is an effective LTP induction protocol, as the optimal repetition rate corresponds to the frequency of the hippocampal theta rhythm, an EEG pattern previously related indirectly to memory storage processes 60 . LTP was induced using weak TBS (i.e. three series of six trains of six pulses at 400 Hz, with 0.2 s between trains and 20 s between series) alone or a combination of weak TBS followed by strong TBS (i.e. six series of six trains of six pulses at 400 Hz, with 0.2 s between trains and 20 s between series) 30 min later 22  www.nature.com/scientificreports/ TBS were doubled in comparison to baseline recordings. The intensity of the baseline stimulus was set to generate a reliable population spike in order to examine the differences in LTP of the population spike. Since the potentiated population spike was often so large as to obscure the initial slope of the fEPSP, we performed another set of experiments in which the baseline stimulation intensity was set to elicit a minimal population spike. The maximum allowable baseline stimulus intensity was 400 µA. The potentiation of the population spike/fEPSP slope was expressed as a percentage change relative to the pre-TBS baseline.
Histology. Following the electrophysiological recordings, some mice were transcardially perfused in deep anaesthesia with 4% paraformaldehyde (PFA) in 0.1 M PBS. Brains were removed and post-fixed overnight in 4% PFA/0.1 M PBS at 4 °C. After embedding in Agar, 50 µm thick coronal sections were cut with a vibratome (Leica VT 1000 s, Wetzlar, Germany) and stored in cryoprotection solution (30% ethylene glycol, 20% glycerin in PBS) at − 20 °C until further use. Hippocampal sections were stained with 0.1% toluidine blue aqueous solution, differentiated in ethanol, and mounted on glass slides with DPX (all chemicals from Merck, Darmstadt, Germany). Finally, the sections were viewed under a microscope (Olympus BX40, Tokyo, Japan) to verify the correct placement of the stimulation and recording electrodes (Supplementary Figure S1).
Electron microscopy. Brain tissue from wild-type and Nbea +/− mice was embedded in epon resin (Electron Microscopy Science, EMS, Hatfield, USA). For embedding, anaesthetised adult male mice were transcardially perfused with 25 ml of 2% glutaraldehyde (Roth, Karlsruhe, Germany) and 2% PFA (Merck, Darmstadt, Germany) in 0.1 M PB at 37 °C, and postfixed at 4 °C overnight. Blocks of hippocampal tissue were contrasted in 1% osmium tetroxide for 2 h at room temperature. Following washes with distilled water and dehydrating, tissue was incubated with propylene oxide (EMS) for 45 min, infiltrated with propylene oxide/epon (1:1) for 1 h, in pure epon overnight, and hardened at 60 °C for 24 h. Additional contrasting of thin sections from brains was done on Formvar-coated copper grids with a saturated solution of 12% uranyl acetate and lead citrate. For direct comparison with electrophysiological analysis, coronal samples containing the middle molecular layer (MML) of the dorsal hippocampal dentate gyrus 1.9 mm posterior to bregma, representing the major termination area of the perforant pathway, were investigated. Ultrastructural analysis was done with a transmission electron microscope (Libra 120, Zeiss) at 80 kV, and images taken with a CCD camera (Tröndle, Moorenweis, Germany). For quantifying the density of asymmetric synapses, tissue areas of 225 µm 2 were reconstructed from 9 individual images and 7 of these reconstructions (total area of 1575 µm 2 ) analysed per genotype (n = 7 reconstructions from 2 animals per genotype). Asymmetric (type 1) synapses were defined as contacts with a visible synaptic cleft, a distinct postsynaptic density and at least three synaptic vesicles.
Statistical analysis. All statistical analyses were performed using the software Prism 7 for Windows or Mac (GraphPad Software, La Jolla, CA, USA). Electrophysiological and ultrastructural data were tested for statistical significance using an unpaired Student's t test (with Welch's correction if the variance or group sizes differed) or a two-way analysis of variance (ANOVA) with Bonferroni's multiple-comparison post-tests. The Shapiro-Wilk test was used to determine whether the samples were drawn from a normal distribution. A twotailed P-value lower than 0.05 was considered to be significant. Group values are reported as means ± standard error of the mean (SEM).

Data availability
All data will be made available upon request.