Investigation of hippocampal synaptic transmission and plasticity in mice deficient in the actin-binding protein Drebrin

The dynamic regulation of the actin cytoskeleton plays a key role in controlling the structure and function of synapses. It is vital for activity-dependent modulation of synaptic transmission and long-term changes in synaptic morphology associated with memory consolidation. Several regulators of actin dynamics at the synapse have been identified, of which a salient one is the postsynaptic actin stabilising protein Drebrin (DBN). It has been suggested that DBN modulates neurotransmission and changes in dendritic spine morphology associated with synaptic plasticity. Given that a decrease in DBN levels is correlated with cognitive deficits associated with ageing and dementia, it was hypothesised that DBN protein abundance instructs the integrity and function of synapses. We created a novel DBN deficient mouse line. Analysis of gross brain and neuronal morphology revealed no phenotype in the absence of DBN. Electrophysiological recordings in acute hippocampal slices and primary hippocampal neuronal cultures showed that basal synaptic transmission, and both long-term and homeostatic synaptic plasticity were unchanged, suggesting that loss of DBN is not sufficient in inducing synapse dysfunction. We propose that the overall lack of changes in synaptic function and plasticity in DBN deficient mice may indicate robust compensatory mechanisms that safeguard cytoskeleton dynamics at the synapse.

increases 4 . This phase is followed by an increase in proteins that are known to stabilize the supra-structure of the actin cytoskeleton by bundling F-actin or linking F-actin to the PSD in the spine 4 . Of note, pathological changes associated with spine morphology and structural plasticity deficits occur in a variety of neurological disorders accompanied by cognitive decline such as seen in Alzheimer's disease 16 .
Drebrin (Developmentally-regulated brain protein, DBN) is a cytoplasmic actin-filament binding protein highly expressed in neurons and known to stabilize actin filaments 17 . Two isoforms have been identified, DBN E and DBN A, that are transcribed from a single gene through alternative splicing 18 . DBN E primarily promotes neurite-extension during development 19,20 whereas DBN A localizes to dendritic spines 21,22 and is implicated in shaping spine morphology 23 . Overexpression has been shown to cause spine elongation 24 , and, conversely, depletion of DBN using siRNA in neuronal cultures was reported to induce spine shrinkage coinciding with altered electrophysiological properties 25 .
The current picture of the interplay between synaptic activity and DBN dynamics is not completely understood. It has been suggested that LTP inducing stimuli induce a transient exodus and subsequently, re-entry of DBN into dendritic spines 4,26 resulting in a net-increase of DBN in potentiated spines 27 . Likewise blockage of NMDA receptors (NMDA-R) increases the proportion of DBN immuno-positive spines 28 .
Two mouse models of DBN deficiency have been previously published, one being deficient for the isoform conversion of DBN A from DBN E 29 and a DBN KO mouse generated by excision of DBN exons 4-7 30 . Both models show reduced hippocampal LTP and altered spine morphology in CA1 hippocampal neurons 30,31 . Additionally, DBN A deficient mice exhibit impaired context dependent fear conditioning 29 .
We generated novel conditional mouse alleles of the Dbn1 gene and analysed ubiquitous and cell-type specific knockout effects of the actin binding protein DBN. Surprisingly, DBN KO mice develop normally and show no obvious defects in brain development. Deletion of DBN did not alter the basic properties at hippocampal CA1 synapses or different forms of plasticity including LTD, LTP and homeostatic plasticity. Our data argue against a prominent role of DBN in mediating changes in synaptic strength at hippocampal excitatory synapses, at least in the healthy young adult brain.

Results and Discussion
Generation of DBN KO mice. The Dbn1 KO mice were established from ES cell clone EPD0211_3_A05, obtained from the supported KOMP Repository (www.komp.org) 32 generated by the Wellcome Trust Sanger Institute. Following removal of the trapping cassette by crossing with a Flp deleter strain, the resulting pre-conditional allele was validated by PCR and sequencing (data not shown). DBN KO mice were obtained by crossing homozygote pre-conditional mice (Dbn1 loxp/loxp ) with a Cre/loxP-deleter strain, which induced the expected excision of exon 1-6 between loxP sites (Fig. 1a,b; Fig. S1a). As a result of homozygosity for the null allele, full length DBN protein or DBN fragments of any lower molecular weight were not detected (Fig. 1d). This validated the use of our genetically modified mice as a proper tool to study the physiological effects of DBN depletion.
DBN KO mice are viable and fertile, and Mendelian distribution of DBN deficient progeny was as expected ( Fig. 1c; number of litters analysed: 18; total number of mice: wild type (WT): 32, heterozygous (HET): 46, knockout (KO): 30). Overall, brain sizes between the two genotypes were comparable and gross brain morphology of WT and DBN KO littermates at P37 revealed no changes in cresyl violet stained brain slices ( Fig. 1e upper panel;  Fig. S1b). Similarly, the anatomy of the corpus callosum, which is composed of axons that cross the midline and form connections in the contralateral side of the brain, was well preserved in DBN KO brains (Fig. S1b). To further examine neuritogenesis and axon formation, control and DBN KO hippocampal neurons were cultured in vitro, which revealed no differences in overall neurite formation, neuronal polarization or axon branching ( Fig. 1e second panel). Neurons obtained from DBN KO brains also showed no obvious differences in dendritic spine formation, dendritic spine size or spine number (Fig. 1e lower panel). These results contrast previous studies in neuronal cell culture which linked DBN A as an essential actin regulator to the control of neuritogenesis 19,20 and the establishment of dendritic spine morphology and density [33][34][35] .
These results indicate that gross brain anatomy and dendritic spine protein composition is not impaired in this newly generated DBN deficient mouse model.

Basic synaptic transmission is normal in DBN KO mice.
To investigate the hypothesis that DBN deficiency causes defects in synaptic properties, we first tested whether the loss of DBN leads to alterations in postsynaptic responses. To this end, we performed electrophysiological recordings in acute hippocampal slices in young (P13-P16) and young adult (P34-P39) mice. We stimulated the Schaffer collaterals in area CA1 and recorded postsynaptic field potentials (fEPSPs) in the stratum radiatum of area CA1. Comparing the size of the afferent fibre volley (FV) as a measure for the excitation of presynaptic fibers, with the slope of the fEPSP representing the postsynaptic responsiveness, we found no significant differences between the two groups in their input-output relationship. (Fig. 2a   To study presynaptic effects that may occur as a consequence of DBN deficiencies, we analysed paired-pulse facilitation (PPF). Using stimulation intervals at 50, 100 and 500 ms, we found that PPF is not significantly changed in DBN KO mice (

Single cell DBN KO does not have an impact on AMPA-R mEPSCs. To differentiate between network
and single-cell phenotype, the dbn1 gene was deleted in a subset of hippocampal pyramidal cells that normally express DBN protein. This was achieved by viral mediated co-expression of Cre-recombinase with a GFP-tagged nuclear localisation signal in DBN loxp/loxp mice. Injection led to sparse labelling of CA1 hippocampal neurons ( Fig. 3a), whereas the Schaffer collaterals forming synapses onto DBN deficient CA1 pyramidal cell derived from WT neurons. Subsequent recordings from GFP + (DBN KO genotype) and GFP − (WT genotype) CA1 pyramidal neurons in acute hippocampal slices were performed at P21 ± 3 days. Miniature excitatory postsynaptic currents (mEPSCs) were recorded in the presence of the sodium channel blocker tetrodotoxin (TTX), the AMPA-R desensitization inhibitor cyclothiazide, the GABA A receptor antagonist gabazine (SR 95531), blocking any contribution of fast inhibitory synaptic currents, and APV, a selective NMDA-R antagonist. CA1 pyramidal cell identity was verified post-hoc by biocytin staining of the recorded cell ( Fig. 3a,b). Infected neurons showed spontaneous activity demonstrating that neurons were vital and fully integrated in the neuronal network. Sequential recordings from a GFP + and a GFP − CA1 pyramidal cell (or vice versa) showed neither significant differences in mEPSC frequency ( Fig. 3c; GFP − 2.6. ± 0.2 Hz; GFP + 2.2 ± 0.2 Hz; N = 15, n = 32; P = 0.17; paired t-test) nor in the mean amplitude ( Fig. 3d; GFP − 16.1 ± 0.5 pA; GFP + 16.2 ± 0.6 pA; N = 15, n = 32; P = 0.85; paired t-test). The cumulative probability plots of the mEPSC amplitudes were not significantly shifted ( Fig. 3e; P > 0.05, KS normality test, Wilcoxon Signed Rank Test), indicating that DBN deficiency does not impact AMPAR-mediated transmission. In combination with the fact that frequency remained unchanged, we conclude that neurotransmitter release is not altered and thus presynaptic mechanisms are not affected by loss of DBN.
Long-term plasticity is not impaired in DBN KO mice. Two recent studies reported that DBN and DBN A deficient mice exhibit impaired LTP 30,31 and hippocampus-dependent learning tasks 30,32 .
Induction of LTP and LTD requires NMDA-R activation [41][42][43] whereas continued expression depends on recruitment and internalisation of AMPA-R 43 . As DBN was reported to facilitate synaptic targeting of NMDA-R 44,45 , we tested if loss of DBN might cause impairments in various forms of synaptic plasticity.
To study effects on long-term synaptic plasticity in WT and DBN KO mice, we recorded a 10-15 min baseline of stable fEPSPs in area CA1 before applying plasticity induction protocols. To induce LTD, the Schaffer collaterals were stimulated with 900 pulses at 1 Hz for 15 minutes. No significant differences between WT and DBN KO littermates were found in the fEPSP slope 30-35 min after induction of LTD (Fig. 4a,c; WT 74% ± 5%, N = 3 n = 9; KO 78% ± 2%, N = 3 n = 9; P = 0.41, unpaired t-test).  Next, we tested if loss of DBN affects the continuous process of induction and expression of LTP. Maintenance of LTP is highly dependent on external parameters such as oxygenation and temperature 46,47 . Therefore we tested different conditions that varied the temperature (potentially influencing actin dynamics) and the flow rate of recording solution (potentially affecting oxygenation and/or the washout of secreted factors 48 ), or the composition of the recording solution (addition of glycine and D-serine as co-agonist of NMDA receptors). In the first approach we recorded at room temperature, a perfusion rate of 5 ml min −1 to avoid anoxia and provided glycine (10 μ M) and D-serine (20 μ M) in the medium. No significant differences in potentiation were detectable between WT and DBN KO slices in the fEPSP slope 40-45 minutes after induction of LTP (Fig. 5a,c; WT 139% ± 6%, N = 6 n = 10; KO 138% ± 5%, N = 5 n = 9; P = 0.84, unpaired t-test). In the second approach, we applied a protocol that had previously been reliably used to investigate long-lasting forms of LTP (l-LTP; > 2 h) consisting of elevated temperature (29 ± 0.5 °C) and a perfusion rate of 2 ml min −1 . Again, the level of potentiation did not differ between slices from WT and KO mice 40-45 min post LTP induction ( Fig. 5d; WT 161% ± 13%, N = 4, n = 5; KO 148% ± 10%, N = 3, n = 4; P = 0.48, unpaired t-test). Recordings from DBN KO mice also showed stable l-LTP (WT 137% ± 4%, N = 5 n = 5; KO 134% ± 8%, N = 5 n = 5; P = 0.72, unpaired t-test) indicating no additional role of DBN in extending synaptic enhancement from a short-lasting to a long-lasting form.
Taken together our results show that neither LTD nor LTP are altered in DBN KO mice. DBN was shown to be involved in the spatiotemporal reorganization during LTP as a solidifier 4 , yet our data show that the loss of DBN is dispensable for the long-lasting maintenance of LTP at the Schaffer-collateral-CA1 synapse.
Homeostatic plasticity is not impaired in DBN deficient primary neuronal cultures. Finally, we tested DBN deficient neurons for another form of synaptic plasticity: Homeostatic synaptic plasticity affects synaptic networks globally within a desired range in a compensatory manner and guards the network from hyper-excitation or silencing 49 . The process involves postsynaptic trafficking of glutamate receptors 49 . Since DBN is able to remodel the actin cytoskeleton -the scaffold for traffic of vesicular cargo -we hypothesised that DBN deficiency could interfere with transport of AMPA-R-loaded vesicles during homeostatic scaling in response to chronic synaptic silencing. Furthermore, DBN was reported to facilitate activity dependent NMDA-R insertion 44 whereas depletion of DBN A diminishes the homeostatic synaptic increase in NR2A upon NMDA-R inhibition 50 . TTX, a potent Na + channel blocker and D-APV, a selective NMDA-R antagonist were added to neuronal cultures for 24 ± 4 hours (h) to block action potentials and chronically silence network activity. Under these conditions, induced homeostatic synaptic plasticity is evidenced by an increase in mEPSC amplitudes 49,51 . To isolate AMPA-R-mediated mEPSCs from sister DBN loxp/loxp cultures infected with Cre or control virus (Fig. 6a), we included TTX and picrotoxin in the recording solution to block action potentials and GABAergic currents, respectively.  Untreated DBN positive neurons had an average amplitude of 21.1 ± 0.8 pA (N = 4, n = 24), which was significantly increased after chronic silencing of synaptic activity (Fig. 6b left panel; 27.4 ± 1.8 pA, N = 4, n = 24; P < 0.05 compared to silenced DBN positive neurons; one-way ANOVA, Bonferroni post-hoc test). However, the silencing effect was not significantly different in Cre-infected, DBN deficient neurons (Fig. 6b right panel; basal 20.9 ± 1.7 pA, N = 4 n = 20; silenced 27.2 ± 2.2 pA, N = 4, n = 23; one-way ANOVA, Bonferroni post-hoc test). Analysis of the cumulative probability of amplitudes did not reveal any significant changes either ( Fig. 6c; P > 0.05, KS normality test, Wilcoxon Signed Rank Test). The frequency of mEPSCs remained unchanged for all conditions ( Fig. 6d; basal, DBN + 4.1 ± 0.5 Hz; silenced, DBN + 4.1 ± 0.5 Hz; basal DBN − 4.1 ± 0.4 Hz; silenced DBN − 4.8 ± 0.5 Hz; one-way ANOVA, Tuckey's Multiple Comparison Test).
These results demonstrate that chronic synaptic silencing did not affect DBN KO cultures differently than WT cultures. Depletion of DBN does not perturb the physiological expression of increased synaptic strength as a result of chronic TTX and APV induced homeostatic synaptic plasticity in primary neuronal cultures.

Conclusion
In summary, in none of the five sets of electrophysiological experiments presented in this study we observed changes between WT and KO animals in glutamatergic transmission as a result of altered synaptic drive. Our results appear to contradict previous in vitro 25 , ex vivo 30,31 and in vivo 30,32,52 experiments, which might be due to different methodological approaches. However, we effectively controlled for the validity of our KO model and carefully exercised different protocols to confirm our findings.
The finding of no discernible phenotype might be due to two possibilities 53 : (i) the intrinsic feature of genetic robustness in actin binding proteins enables genetic buffering by alternative pathways or functional complementation of other genes, or (ii) the abnormal phenotype will only become evident under specific conditions, for example ageing, disease or stress.
The lack of phenotypic changes in glutamatergic synaptic hippocampal synapses of DBN KO mice is possibly caused by compensatory mechanisms as functional redundancy and overlapping functions have been reported for several actin binding proteins 54 .
It is worth speculating that under certain conditions, the loss of DBN cannot be compensated, unmasking a specific DBN KO phenotype. In this regard, cellular stress 55 or aging might render the dendritic spine vulnerable to the loss of the actin binding protein DBN. However, at this point, our results support the idea that DBN deficiency alone is not sufficient in causing synaptic dysfunction.

Methods
All animals used were handled in accordance with the relevant guidelines and regulations. Protocols were approved by the 'Landesamt für Gesundheit und Soziales' (LaGeSo; Regional Office for Health and Social Affairs) in Berlin and animals reported under the permit number T0347/11 and G0189/14.

Generation of DBN KO Mice.
Heterozygous mice harbouring a promoter-driven Knockout First allele (Dbn1tm1a(KOMP)Wtsi) were obtained from the KOMP Repository. The trapping cassette creating a constitutive null mutation was removed by crossing with a Flp deleter strain and the resulting pre-conditional allele was validated by PCR and sequencing (data not shown) and subsequently bred to homozygosity. Primary neuronal cultures of homozygote pre-conditional embryos were infected with Cre-recombinase and loss of DBN protein was apparent (data not shown). Subsequently, homozygote pre-conditional mice were crossed with Cre/ loxP-deleter mice (B6.C-Tg(CMV-cre)1Cgn/J, Jackson Laboratories). Conversion of pre-conditional alleles into null-alleles was monitored by PCR-based genotyping. Subsequently, wild-type and knockout littermates from heterozygous crossings were used for experiments.

Electrophysiology.
Recordings were performed with a MultiClamp 700B (Axon Instruments, Union City, CA, USA) amplifier, signals were filtered at 2 kHz and digitized (National Instruments, BNC-2090) at 5 kHz, recorded and analysed with custom-made software in IGOR Pro (WaveMetrics Inc., OR, USA). Slice recordings were performed submerged in ACSF equilibrated with carbogen (95% O 2 , 5% CO 2 ) at room temperature perfused at 5 ml min −1 unless stated otherwise. fEPSPs were recorded with low-resistance patch-clamp electrodes filled with ACSF in the CA1 stratum radiatum. For extracellular fibre stimulation, pipettes were placed in the CA1stratum radiatum at the same height as the recording pipette 200 μ m apart.
Schaffer collaterals were stimulated with 100 μ s pulses at 0.05 Hz. fEPSP magnitude was determined by analysing 20-80% of the amplitude slope. Recordings were only analysed if the fibre volley remained constant throughout the recording. Paired-pulse facilitation was investigated by analysing the ratio of the second to the first fEPSP. For plasticity induction, the simulation intensity was set to elicit 40-50% of the maximum amplitude. Following a baseline of at least 10 min, LTD was induced with 900 pulses at 1 Hz for 15 min, LTP was induced by four tetany of 100 Hz for 1 second in 20 s intervals. LTD was studied in P13-P16 mice to reliably induce NMDA-R dependent synaptic depression 3 , LTP was induced in P34-39 mice.
Whole cell patch-clamp recordings were performed at 31-32 °C. Patch-clamp electrodes (resistance 2-5 MΩ) were filled with internal solution containing (in mM) 130 KMSO 3 , 10 KCl, 4 NaCl, 10 HEPES, 4 Mg-ATP, 0.5 Na 3 GTP, 5 phosphocreatine, 0.25% biocytin; pH adjusted to 7.3 with KOH. Access resistance (6-20 MΩ) was continuously monitored throughout the experiment. Recordings were discarded when the series resistance changed more than 20%. No R s compensation was used. mEPSCs were isolated at −60 mV in the presence of 1 μ M TTX, 1 μ M gabazine, 50 μ M D-APV and 100 μ M cyclothiazide. Only slices were analysed, in which mEPSCs were recorded from both infected and uninfected neurons. Putative interneurons were excluded from analysis based on morphology after visualization of biocytin.
Cells were identified using infrared differential contrast microscopy (BX51WI, Olympus). Fluorescently labelled cells were identified using fluorescence microscopy (XM10, Olympus) and adequate fluorescence filters visualizing GFP-tagged or RFP-tagged neurons.
Data were analysed with Igor plug-in NeuroMatic (www.neuromatic.thinkrandom.com). For mEPSC analysis, signals were detected automatically and additionally sorted manually by visual inspection to exclude false positive events.
Drugs and Chemical Compounds. D-APV, D-serine, gabazine, glycine, PTX and TTX were added from aqueous stock solutions; cyclothiazide was administered from a stock solution in DMSO (final concentration of DMSO < 0.1%). Drugs were purchased from Tocris/BioTrends GmbH or Biozol GmbH (Enching, Germany) and Sigma-Aldrich. ACSF compounds from Roth (Karlsruhe, Germany). Cell culture reagents were purchased from Life technologies (Grand Island, NY, USA).