During postnatal development endogenous neurosteroids influence GABA-ergic neurotransmission of mouse cortical neurons

As neuronal development progresses, GABAergic synaptic transmission undergoes a de ﬁ ned program of recon ﬁ guration. For example, GABA A receptor (GABA A R)-mediated synaptic currents, (miniature inhibitory postsynaptic currents; mIPSCs), which initially exhibit a relatively slow decay phase, become progressively reduced in duration, thereby supporting the temporal resolution required for mature network activity. Here we report that during postnatal development of cortical layer 2/3 pyramidal neurons, GABA A R-mediated phasic inhibition is in ﬂ uenced by a resident neurosteroid tone, which wanes in the second postnatal week, resulting in the brief phasic events characteristic of mature neuronal signalling. Treatment of cortical slices with the immediate precursor of 5 a -pregnan-3 a -ol-20-one (5 a 3 a ), the GABA A R-inactive 5 a -dihydroprogesterone, (5 a -DHP), greatly prolonged the mIPSCs of P20 pyramidal neurons, demonstrating these more mature neurons retain the capacity to synthesize GABA A R-active neurosteroids, but now lack the endogenous steroid substrate. Previously, such developmental plasticity of phasic inhibition was ascribed to the expression of synaptic GABA A Rs incorporating the a 1 subunit. However, the duration of mIPSCs recorded from L2/3 cortical neurons derived from a 1 subunit deleted mice, were similarly under the developmental in ﬂ uence of a neurosteroid tone. In addition to principal cells, synaptic GABA A Rs of L2/3 interneurons were modulated by native neurosteroids in a development-dependent manner. In summary, local neurosteroids in ﬂ uence synaptic transmission during a crucial period of cortical neurodevelopment, ﬁ ndings which may be of importance for establishing normal network connectivity. © 2015 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).


Introduction
The postnatal brain undergoes considerable neuronal plasticity to meet the changing demands of rapidly developing networks.During this critical time the duration of synaptic events mediated by GABA A Rs becomes progressively reduced, permitting postsynaptic neurons to respond to input from certain fast-spiking GABA-ergic interneurons and thereby appropriately influence the temporal window for postsynaptic excitation (Whittington et al. 2011;Deidda et al. 2014;Fritschy and Panzanelli, 2014).
Alterations to the subunit composition of synaptic GABA A Rs are implicated in producing these crucial changes to inhibitory postsynaptic current (IPSC) kinetics (Brickley et al. 1996;Okada et al. 2000;Vicini et al. 2001;Juttner et al. 2001;Goldstein et al. 2002;Bosman et al. 2005;Takahashi, 2005;Fritschy and Panzanelli, 2014;Deidda et al. 2014).GABA A Rs are members of the Cys-loop transmitter-gated ion channel family and in common with glycine, nicotinic acetylcholine and 5HT 3 receptors are composed of five subunits (Olsen and Sieghart, 2008).In mammals 19 subunit genes underpin the expression of ~20e30 native GABA A R subtypes, which display distinct pharmacological and physiological properties (Olsen and Sieghart, 2008).In the CNS, these GABA A R subtypes exhibit a heterogeneous expression pattern, which importantly in many neurons is known to change during neonatal development (Olsen and Sieghart, 2008;Fritschy and Panzanelli, 2014;Rudolph and Mohler 2014).In particular, an increased expression of receptors incorporating the a1 subunit (a1-GABA A Rs) is implicated in the appearance of short duration IPSCs (Okada et al. 2000;Vicini et al. 2001;Peden et al. 2008;Eyre et al. 2012;Deidda et al. 2014;Fritschy and Panzanelli, 2014).However, during development of thalamocortical inhibitory synapses, changes to IPSC kinetics occur prior to the temporal expression of the a1 subunit (Peden et al. 2008;Brown et al. 2015), implicating, at least in these neurons, additional factor(s) that influence GABA A R ion channel gating properties.
Certain naturally occurring neurosteroids act in a non-genomic manner as endogenous positive allosteric modulators of the GABA A R (Belelli and Lambert, 2005;Zorumski et al. 2013).The cortical levels of these neurosteroids change during neonatal development (Grobin and Morrow, 2001).Furthermore, the enzymes required to synthesize these GABA A R-active steroids are expressed in certain neurons, suggesting that these local neuromodulators may act as paracrine, or autocrine messengers, to locally influence neuronal inhibition (Agis-Balboa et al., 2006;Do Rego et al., 2009;Castelli et al. 2013;Brown et al. 2015).Here, we demonstrate for mouse cortical L2/3 pyramidal neurons and interneurons that during early (P7-15) neonatal development, their synaptic GABA A Rs are influenced by an endogenous neurosteroid tone, which consequently prolongs the duration of phasic GABAergic neurotransmission.During subsequent development this modulation wanes, such that by P20-24 it has dissipated, resulting in brief IPSCs, characteristic of mature inhibitory synapses.However, when provided with 5a-dihydroprogesterone (5a-DHP), the 5a3a precursor, these more mature neurons retain the capacity to synthesise GABA A R-active neurosteroids, suggesting that the developmental changes to GABAergic neurotransmission reflect a timed loss of steroid substrate, acting in concert with the established ontogenetic pattern of a1 subunit expression.Importantly, neurosteroid levels are not static, but are perturbed in a variety of physiological and pathophysiological conditions (Belelli and Lambert, 2005;Zorumski et al. 2013).Therefore, given the role GABA A Rs may play in a number of disorders including autism, schizophrenia, Fragile X and Down syndrome (Deidda et al., 2014;Rudolph and Mohler, 2014), these findings may not only be important in better understanding how phasic GABAergic neurotransmission changes to accommodate the demands of neuronal network activity during development, but may additionally allow new insights into the pathology of certain neurodevelopmental disorders.

Breeding of mice
All animal studies were approved by the University of Dundee Ethical Review Committee (Home Office Project Licenses 60/4005 and 70/8161, Dr. Belelli), and complied with Schedule 1 of the UK Government Animals (Scientific Procedures) Act, 1986.Transgenic a1 subunit 'knockout' (a1 À/À ) mice were generated on a mixed C57BL6-129SvEv background (Sur et al. 2001).Transgenic GAD 67-GFP "knock-in" mice were generated on a C57BL/6J background as described previously (Tamamaki et al. 2003).Electrophysiological experiments were performed on brain slices prepared from the first 2e3 generations of a1 À/À , GAD67-GFP, or corresponding WT offspring from heterozygous ( þ/À ) breeding pairs housed at the University of Dundee.

Drugs and reagents
For in vitro experiments, finasteride, indomethacin, 5a3a and 5a-DHP were prepared as concentrated stock solutions (1000x final concentration) in DMSO, whereas bicuculline methobromide and TTX were prepared as concentrated stock solutions in distilled water.Drug stock solutions were diluted to the final required concentration in ECS, whereas kynurenic acid was dissolved directly into the ECS.Similarly, a-CD and g-CD were dissolved directly into the extracellular and intracellular solution.
For acute studies with 5a3a (1 mM), the steroid was perfused directly in to the recording chamber, with the effect determined on mIPSCs acquired after ~7 min of drug contact with the slice preparation.To investigate the impact of prolonged exposure to either 5a3a (100 nM), or 5a-DHP (3 mM), the test steroid was preincubated with the cortical slice at room temperature for > 2 h, before the tissue was transferred to the recording chamber, where it was continuously perfused with ECS (see Section 2.3 above) containing the test steroid.Note for some experiments with 5a-DHP (3 mM) the cortical slice was co-incubated with indomethacin (100 mM).The CD studies employed two protocols: the first involved pre-incubating cortical slices in the holding chamber at room temperature with either a-, or g-CD (1 mM, >1 h).Recordings were then made with both ECS and ICS containing the CD (1 and 0.5 mM, respectively).In the second protocol, CD was included only in the recording pipette (0.5 mM).When the CD was applied to the intracellular compartment alone, mIPSCs were only included for analysis if they were recorded for at least 6 min after obtaining whole-cell access.To examine the influence of inhibiting the 5a-R enzyme, finasteride (50 mM) was pre-incubated with the cortical slice in a holding chamber containing oxygenated ECS (at room temperature) for > than 4 h prior to recording.Subsequent recordings from such slices were made either with a control intracellular pipette solution, or with the pipette containing g-CD, to determine the combined influence of intracellular g-CD and finasteride treatment.Note the final DMSO concentration (0.1%) had no effect on any of the mIPSC parameters measured.

Electrophysiological analysis
Digitized data was analysed offline using WinEDR/WinWCP software (Strathclyde University, UK).The mIPSCs were identified by an algorithmic detection protocol.To eliminate distal events, which may be affected by imperfect voltage-clamp, Gaussian distributions of 10e90% rise time were generated and mIPSCs falling outside the Gaussian limits were excluded.Individual mIPSCs were visually inspected and spurious events omitted.Typically, for each neuron data from 50, or more mIPSCs were analysed with respect to their peak amplitude, 10e90% rise time, and time taken to decay from peak by 50% (T50).Accepted mIPSCs recorded from a single neuron were averaged and fitted with either a mono-exponential (y(t) ¼ Ae (Àt/t) ), or bi-exponential (y(t) ¼ A 1 e (Àt/t1) þ A 2 e (Àt/t2) ) decay function, where y(t) is the current amplitude at time t, A is the current amplitude and t is the decay time constant.To compare goodness of fit between a mono-or bi-exponential decay, an F test was applied to the standard deviation of the residuals.The overwhelming majority of mIPSC decay times analysed were best fit by a bi-exponential function.Subsequently, a mean weighted decay constant (t w ) was calculated to accommodate the relative contribution of each decay component whereby: Here, t 1 and t 2 are the decay time constants for the first and second exponential functions, and P 1 and P 2 are the proportions of current amplitude described by each component i.e.
All reported data are expressed as mean values ± standard error of the mean (S.E.M.).To determine statistical significance, Student's t-tests (paired, or unpaired) and ANOVA (one or two-way, followed post-hoc by Tukey's HSD or independent samples t-test, SigmaStat, Systat Software Inc.San Jose, CA, USA) were used as appropriate.For comparison of cumulative probability distributions of mIPSC T50 values, the Kolmogorov-Smirnoff (KS) test was used (SPSS software, Chicago, IL, USA).

Phasic GABAergic transmission from P7 -8 cortical L2/3 pyramidal neurons is influenced by an endogenous neurosteroid tone
To test for endogenous modulation of GABAergic neurotransmission by neurosteroids, we utilized g-CD, a neurosteroid scavenger (Shu et al. 2007;Brown et al. 2015).For P7 -8 neurons, the g-CD pre-incubation protocol (>1 h.see Methods) had no effect on the mIPSC frequency, or amplitude (in both cases p > 0.05, one way ANOVA), but greatly reduced their duration (t W control ¼ 12.1 ± 0.3 ms, n ¼ 55, vs t W g-CD ¼ 8.5 ± 0.2 ms, n ¼ 20, p < 0.001, one way ANOVA, Fig. 2A, F, G Table 1).The structurally related a-CD is ineffective in sequestering pregnane steroids, as the pore diameter of the molecule is smaller (6 vs 8 cyclic sugars) than that of g-CD (Davis and Brewster, 2004;Shu et al. 2004Shu et al. , 2007;;Brown et al. 2015).Importantly, the equivalent treatment with a-CD had no effect on the mIPSC t W of P7 -8 neurons (p > 0.05, one way ANOVA, Fig. 2B, F, G).
To confirm that the reduced decay times observed following g-CD were due to neurosteroid sequestration, we pre-incubated P7 -8 WT cortical brain slices with the 5a-reductase inhibitor finasteride (50 mM > 4 h), which significantly reduced the mIPSC duration (t W control ¼ 12.1 ± 0.3 ms, n ¼ 55; t W finasteride ¼ 8.5 ± 0.3 ms, n ¼ 7, p < 0.001, one way ANOVA, Fig. 2C, F, G, Table 1).This effect on the mIPSC duration was indistinguishable from that produced by g- 05, one way ANOVA, Fig. 2F, G, Table 1).Collectively, these findings indicate that an endogenous neurosteroid tone influences synaptic GABA A Rs at P7 e 8.
Interpretation of the combined effects of finasteride and g-CD is potentially complicated, as given that finasteride is a steroid, it is conceivable that the extracellular g-CD may sequester this 5areductase inhibitor.We have previously demonstrated that intracellular g-CD alone is equi-effective in influencing the mIPSCs of developing thalamic neurons (Brown et al., 2015), suggesting a protocol to avoid this complexity.Therefore, we first investigated the effect on cortical mIPSCs of incorporating the membraneimpermeant g-CD (0.5 mM) solely in the recording pipette (ICS g-CD).This treatment (recordings made > 6 min after achieving the whole-cell recording configuration) significantly reduced the mIPSC t W (control ¼ 12.1 ± 0.3 ms, n ¼ 55; g-CD ICS: 9.2 ± 0.6 ms, n ¼ 6, p < 0.05, one way ANOVA, Fig. 2D, F, G), an effect indistinguishable from that of g-CD resulting from the pre-incubation protocol (g-CD pre-incubation: 8.5 ± 0.2 ms, n ¼ 20, p > 0.05, one way ANOVA, Fig. 2F, G) and not significantly different from that produced by finasteride (50 mM) treatment (p > 0.05, one way ANOVA t W ¼ 8.5 ± 0.3 ms, n ¼ 7, Fig. 2F, G).Finally, we now determined the combined effect of finasteride and g-CD treatment.
For P7 -8 neurons, treatment of the slice with finasteride (50 mM) for > 4 h, followed by intracellular g-CD (0.5 mM), resulted in mIPSCs with a significantly reduced duration (t w ¼ 7.3 ± 0.2 ms; n ¼ 6, p < 0.001, one way ANOVA), that was not significantly different from that produced solely by finasteride, or by intracellular g-CD alone (p > 0.05, one way ANOVA, Fig. 2E, F, G).

The neurosteroid influence on phasiceGABAergic transmission of cortical L2/3 pyramidal neurons changes during development
We next assessed whether neurosteroids contribute to the developmental changes in the duration of phasic GABAergic events by determining the effect of g-CD on the mIPSCs of neurons at different stages of development (Brown et al. 2015).Treatment with g-CD reduced the decay time of mIPSCs recorded from P7 -8, P10 and P15 neurons, relative to their respective controls (Fig. 3, Table 1).However, the developmental stage significantly influenced the effect of g-CD on the mIPSC decay time (Fig. 3, age Â treatment interaction, F 3,157 ¼ 6.15, p < 0.001, two-way ANOVA), such that by P20-24, g-CD had no significant effect (control tw ¼ 5.4 ± 0.2 ms n ¼ 25; g-CD ¼ 5.3 ± 0.6 ms; n ¼ 8; p > 0.05; independent samples t-test, Fig. 3).Further implicating a changing neurosteroid impact during development, the effect of finasteride (50 mM) was also significantly influenced by post-natal age (Fig. 3, age Â treatment interaction, F 2,127 ¼ 10.47, p < 0.001, two-way ANOVA), such that, in contrast to P7 e P8 recordings, finasteride had no significant effect on the mIPSC duration of P20 e P24 neurons (control t W ¼ 5.4 ± 0.2 ms; n ¼ 25; finasteride t W ¼ 5.3 ± 0.2 ms; n ¼ 6; p > 0.05, post-hoc Tukey HSD, Fig. 3).

Decreased neurosteroid synthesis contributes to the changes to phasic GABAergic transmission evident in P20-24 cortex
The loss of neurosteroid influence on phasic inhibition of P20 -24 pyramidal neurons, inferred by both the finasteride and g-CD experiments, may be due to the synaptic GABA A Rs becoming neurosteroid-insensitive (Koksma et al. 2003), or alternatively a  g-CD consequence of a decreased neurosteroid synthesis.The neurosteroid interaction with GABA A Rs may be influenced by factors such as subunit composition and phosphorylation status (Koksma et al. 2003;Belelli and Lambert, 2005).We therefore investigated whether GABA A Rs expressed by P20 -24 L2/3 pyramidal neurons retained neurosteroid sensitivity.The acute bath application of exogenous 5a3a (1 mM, whereby mIPSCs were analysed before, and after ~7 min application of the steroid) resulted in mIPSCs with a significantly prolonged decay phase (control t W : 5.5 ± 0.5 ms vs 1 mM 5a3a t W : 8.7 ± 1.7 ms, n ¼ 7, p < 0.05, paired t-test).For isolated single cell studies, acutely applied 5a3a acts at nM aqueous concentrations to enhance GABA A R function (Pistis et al. 1997;Belelli and Lambert, 2005).Therefore, the relatively limited effect of acutely applied 5a3a at the relatively high concentration of 1 mM might suggest that these cortical synaptic GABA A Rs are relatively insensitive to the neurosteroid by P20-24.Alternatively, the effect of the steroid when applied acutely to a brain slice may be underestimated.In support of the latter, the general anesthetics etomidate and propofol, which in common with neurosteroids are lipophilic and efficacious GABA A R modulators, require several hours to approach equilibrium within in vitro brain slice preparations (Gredell et al. 2004;Benkwitz et al. 2007).To ascertain whether the GABA modulatory effects of 5a3a are underestimated when applied acutely to a cortical slice, we determined the effect of a lower concentration (100 nM) of 5a3a on the mIPSCs of P20-24 L2/3 pyramidal neurons, but now pre-incubated (>2 h), before being continuously applied during the recording.Employing this protocol, the 10 fold lower concentration of 5a3a (100 nM) produced a clear and large prolongation of the mIPSC decay (control t W ¼ 5.4 ± 0.2 ms, n ¼ 25 vs 5a3a 100 nM t W ¼ 12.6 ± 0.9 ms, n ¼ 5,  p < 0.001, one way ANOVA; post-hoc Tukey HSD Fig. 4A, C).Importantly, this experiment establishes that P20 -24 cortical synaptic receptors retain sensitivity to nM aqueous concentrations of this neurosteroid and consequently, the change in mIPSC decay at this stage of development is not due to neurosteroid-insensitive synaptic GABA A Rs.
To determine whether more mature (P20 -24) L2/3 cortical pyramidal neurons retain the capacity to synthesize GABA A R-active neurosteroids, we investigated the influence of the GABA A R-inactive steroid 5a-DHP, the immediate precursor of 5a3a (Brown et al. 2015).We had previously shown that a prolonged incubation (>2 h), but not a short incubation (30e60 min) of thalamic slices with 5a-DHP prolonged the mIPSC decay phase of VB neurons (Brown et al. 2015).Here, pre-incubation of the cortical slice with 5a-DHP (3 mM) for > 2 h, followed by continuous perfusion of this steroid during the recording (see Methods), resulted in greatly prolonged mIPSCs (control t W ¼ 5.4 ± 0.2 ms, n ¼ 25 vs 5a-DHP t W ¼ 16.1 ± 0.6 ms, n ¼ 7, p < 0.001, one way ANOVA -see Fig. 4B, C).
This effect was markedly reduced by co-incubation with the 3a-HSD inhibitor indomethacin (100 mM; t W ¼ 6.3 ± 0.4 ms; n ¼ 6; post hoc Tukey HSD p < 0.001), or by intracellular (0.5 mM) g-CD (t W ¼ 8.7 ± 0.3 ms; n ¼ 6, post hoc Tukey HSD p < 0.001) -Fig.4B, C. Therefore, when provided with the immediate precursor, cortical tissue from P20-24 mice retains the ability to synthesize GABA A Ractive neurosteroids.Furthermore, these data provide additional evidence that their synaptic GABA A Rs remain neurosteroid sensitive.

Phasic GABAergic transmission of L2/3 pyramidal neurons is influenced both by an endogenous neurosteroid tone and by the subunit composition of synaptic GABA A Rs
Despite treatment with g-CD, the mIPSC decay time still decreased with development, with a similar trend observed for finasteride-treated neurons (Fig. 3, Table 1).These observations suggest that factors additional to neurosteroids influence phasic inhibition during development.Numerous studies have implicated changes to the subunit composition of synaptic GABA A Rs to be important in this respect, with a particular emphasis on the role of the a1 subunit (Rovira and Ben-Ari, 1993;Tia et al. 1996;Hollrigel and Soltesz, 1997;Dunning et al. 1999;Kapur and Macdonald, 1999;Vicini et al., 1999Vicini et al., , 2001;;Okada et al. 2000;Ortinski et al. 2004).To investigate whether a1-GABA A Rs influence mIPSCs during the development (P7 e 24) of L2/3 pyramidal neurons, cortical brain slices were prepared from mice engineered to lack the a1 subunit (a1 À/À ).The decay phase of a1 À/À mIPSCs was prolonged in comparison to their WT counterparts, but, importantly, this occurred at all ages examined here (two-way ANOVA, age Â genotype interaction, F 3,139 ¼ 5.75, p ¼ 0.001; for post-hoc WT vs a1 À/À comparisons, p < 0.001 for P7 -8, P10, P15 and P20, Fig. 5).However, in common with WT neurons, the a1 À/À mIPSC decay phase became faster with development (p < 0.001, one way ANOVA, Fig. 5; Table 2), suggesting that factors other than increased expression of a1-subunit containing GABA A Rs must contribute to the developmental profile.In agreement, and further implicating a role for Fig. 4. The synaptic GABA A Rs of P20 e 24 L2/3 cortical pyramidal neurons are neurosteroid sensitive and furthermore, P20 e 24 cortical brain slices can synthesise GABA A Ractive neurosteroids.A).Averaged, superimposed mIPSCs normalised with respect to peak amplitude, recorded from representative WT L2/3 cortical pyramidal neurons from brain slices derived from P20 -24 mice, under control conditions and following incubation (>120 min) with 100 nM 5a3a.B) Averaged, superimposed P20-24 mIPSCs normalised with respect to peak amplitude, recorded under control conditions and following incubation (>120 min) with 3 mM 5a-DHP; co-incubation with 100 mM indomethacin (Indo) and 3 mM 5a-DHP (>120 min); or incubation with 3 mM 5a-DHP (>120 min) followed by 0.5 mM ICS g-CD.C) Summary bar graph illustrating the increase in the mIPSC t w in response to incubation of 100 nM 5a3a; 5a-DHP incubation and the much reduced 5a-DHP effect following either co-incubation with 100 mM indomethacin (Indo), or addition of 0.5 mM g-CD to the ICS (n ¼ 5e25 cells).*** ¼ p < 0.001 vs control, one way ANOVA with Tukey post hoc analysis.
Therefore, phasic inhibition of L2/3 interneurons, in common with pyramidal neurons, changes with development.Furthermore, early in development a neurosteroid tone is experienced by the synaptic GABA A Rs of both GABA-ergic interneurons and principal neurons, which by P20 -24 dissipates, resulting in brief phasic inhibitory events.

Endogenous neurosteroids prolong the mIPSCs of cortical L2/3 neurons during development
Postnatal development is marked by periods of considerable plasticity within cortical circuitry, wherein GABAergic neurotransmission is driven towards rapid and effective phasic inhibition, capable of supporting the complexity of mature cortical processing.During this period, several mechanisms may contribute to the mIPSC decay time, including: alterations in the subunit composition of synaptic GABA A Rs (Takahashi, 2005;Eyre et al. 2012), posttranslational modifications of synaptic proteins (Vithlani et al. 2011), the extent of receptor clustering (Petrini et al. 2003), changes in the kinetics of GABA release and alterations to GABA uptake (Mozrzymas, 2004).The latter is influenced by the activity Fig. 5.The effect of a1 subunit deletion on the t W of mIPSCs recorded from L2/3 neurons during development.A bar graph comparing t W of mIPSCs recorded from WT and a1 À/À L2/3 neurons during postnatal development (n ¼ 4e13 a1 À/À neurons).
The WT data is adapted from Fig. 1.Note that the decay time of mIPSCs of both WT and a1 À/À L2/3 pyramidal neurons, becomes reduced with age.However, at each developmental stage, mIPSCs recorded from a1 À/À neurons exhibit a slower decay c.f. those recorded from equivalent WT neurons (*** ¼ p < 0.001, post-hoc independent samples t-test following two-way ANOVA, with postnatal age and genotype as the independent variables).Illustrated above each developmental time point are corresponding averaged and superimposed mIPSCs, normalised with respect to peak amplitude, obtained from representative WT (black line) and a1 À/À pyramidal neurons (grey line).
0.5 ± 0.1 0.5 ± 0.1 0.5 ± 0.1 0.5 ± 0.1 0.5 ± 0.1 0.5 ± 0.1 0.5 ± 0.1 0.5 ± 0.1 0.5 ± 0.1 0.4 ± 0.1 t W (ms) 18.7 ± 2.0 12.2 ± 1.0 yy 12.9 ± 0.9 y 16.0 ± 1.3 9.2 ± 1.1*** 13.5 ± 0.9 9.8 ± 0.4** 7.9 ± 0.4 7.9 ± 0.6 8.8 ± 0.4 Frequency (Hz) ND ND ND 0.8 ± 0.3 1.5 ± 0.3 3.9 ± 0.7 3.5 ± 1.0 12.5 ± 1.6 8.5 ± 1.2 ND and location of the various GABA transporters.However, whereas the effects of transporter inhibitors on the time course of responses to iontophoretically applied GABA, or on IPSCs evoked by repetitive nerve stimulation are quite evident, they have relatively little effect on the amplitude or kinetics of mIPSCs (Keros and Hablitz, 2005;Scimemi, 2014).Our results indicate that in addition to a possible involvement of such factors, during postnatal development the mIPSCs of L2/3 pyramidal neurons become reduced in duration, at least in part due to a programmed loss of the influence of endogenous neurosteroids upon synaptic GABA A Rs.The validity of our conclusions is partly dependent on the specificity of g-CD in sequestering neurosteroids.A previous study reported that b-CD (0.5e1.5 mM), when applied to hippocampal neurons prolonged the decay of macroscopic currents mediated by GABA A Rs (Pytel et al. 2006).However, g-CD treatment of thalamic VB neurons, cortical L2/3 interneurons and pyramidal cells induced a marked reduction in the mIPSC decay time early in development, but had no effect on this parameter at later developmental time-points (e.g.P20-24; Brown et al., 2015).Furthermore, pre-incubation with a-CD had no effect on any of the mIPSC properties at any age studied here and most importantly the effects of g-CD on P7-8 neurons were recapitulated by pre-incubation with the 5aÀR inhibitor finasteride.A parsimonious explanation for the decreased duration of immature L2/3 mIPSCs following g-CD treatment posits that the steroid-sequestering molecule is effective in forming inclusion complexes with endogenous neurosteroids, whereas the observed insensitivity to a-CD reflects the hydrophobic inner cavity being too small to accommodate steroids (Szejtli, 1998;Shu et al. 2004;Brown et al. 2015).
Theoretically, the loss of neurosteroid influence on phasic inhibition with development may result from the synaptic GABA A Rs becoming insensitive to this endogenous modulator (Koksma et al. 2003).However, acute application of 5a3a (1 mM) clearly prolonged mIPSCs recorded from P20-24 cortical L2/3 pyramidal neurons.
Furthermore by pre-incubating the tissue with a lower, aqueous concentration (100 nM) of 5a3a, we demonstrated these synaptic GABA A Rs to be highly sensitive to the neurosteroid at this stage of development.Alternatively, a change in steroid enzyme expression, or a lack of steroid substrate(s) may be implicated in this developmental plasticity.We previously demonstrated that incubation of mouse thalamic slices with 5a-DHP, the immediate precursor of 5a3a, greatly increased the duration of the mIPSCs of VB neurons (Brown et al. 2015).Similarly here, incubation of P20 -24 cortical tissue with 5a-DHP greatly prolonged the mIPSCs of cortical pyramidal neurons and in common with thalamic neurons, this effect was prevented by co-incubation with the 3a-HSD inhibitor indomethacin, or reversed by intracellular g-CD (Brown et al. 2015).
Collectively these results suggest that the developmental change to phasic inhibition of cortical pyramidal neurons occurring between P7 and P20 results in a part from a lack of steroid substrate.

Location of neurosteroid synthesis and action
Previous histochemical studies support the concept of a local Fig. 6.A developmentally-regulated neurosteroid tone influences the decay time of mIPSCs recorded from a1 ¡/¡ L2/3 neurons.A bar graph showing the effect of g-CD pre-incubation (1 mM) on the t W of mIPSCs recorded from a1 À/À L2/3 pyramidal neurons during development (n ¼ 4e13 cells).In common with WT L2/3 pyramidal neurons, treatment of a1 À/À cortical brain slices with g-CD results in faster decaying mIPSCs at P7 -8, P10 and P15, but not at P20-24.For comparison, the effect of finasteride (50 mM) pre-incubation at P7 -8 and P20 -24 is also shown (n ¼ 7e14 cells).yy ¼ p < 0.01, y ¼ p < 0.05 vs control, post hoc Tukey HSD test following a two-way ANOVA.** ¼ p < 0.01, *** ¼ p < 0.001, vs control, post-hoc independent samples ttest following a two-way ANOVA.n.s.¼ not significant.The independent variables for the two-way ANOVAs were postnatal age and treatment.Illustrated above each developmental time point is the corresponding averaged and superimposed mIPSCs, normalised with respect to peak amplitude, obtained from representative pyramidal neurons of WT and a1 À/À in the absence (control, black line) and following 1 mM g-CD pre-incubation (grey line).Fig. 7.The effect of intracellular g-cyclodextrin (0.5 mM) on the decay kinetics of mIPSCs recorded from P7 and P20 -24 L2/3 cortical GAD67-GFP þ neurons.A, C).Superimposed, averaged mIPSCs normalised with respect to peak amplitude from representative P7 (A) and P20 (C) control L2/3 GFP þ neurons (black line) and L2/3 GFP þ neurons in which the intracellular solution contained 0.5 mM g-CD (grey line) at P7 (A) and P20 (C).B, D).Cumulative probability plots of the mIPSC T50 for P7 (B) and P20 -24 neurons (D) control (black line) and ICS þ g-CD (grey line).In each case events were pooled from n ¼ 8 GFP þ cells (P7 control), n ¼ 8 GFP þ cells (P7 ICSþ g-CD), n ¼ 14 GFP þ cells (P20 -24 control), and n ¼ 7 GFP þ cells (P20 -24 ICSþ g-CD).For P7, but not for P20 -24 cells, the mIPSC T50 distribution is left-shifted indicating that all mIPSCs recorded from P7 GFP þ cells treated with intracellular g-CD, exhibited faster decay kinetics, compared with control (p < 0.001, KS-test).E).A summary bar graph showing a significant decrease in the mIPSC t W following intracellular g-CD treatment at P7, but not at P20 -24.n.s.¼ not-significant; *** ¼ p < 0.001 vs control, post-hoc independent samples t-test following two-way ANOVA, with postnatal age and treatment as the independent variables.neurosteroid synthesis in cortex.In mouse cortex the staining for mRNA encoding for the 5a3a synthesising enzymes 5a-reductase Type I (5a-R I) and 3a-hydroxysteroid dehydrogenase (3a-HSD), was co-located in layer 2/3/5 pyramidal neurons (Agis-Balboa et al. 2006).The 5a-R I staining co-localised with that for the vesicular glutamate transporter (VGLUT1), a marker of glutamatergic neurons (Agis-Balboa et al. 2006).By contrast, the 5a-R I, or the 3a-HSD staining did not co-localize with a marker for GABA-ergic neurons, or for glia (Agis-Balboa et al. 2006).In apparent agreement, an antibody raised against 5a3a revealed staining for this GABA A Ractive steroid in rat cortical L2-6 pyramidal neurons, but not in cells that had the appearance of GABA-ergic interneurons, or glia, but note were not categorically identified by specific neurochemical markers of interneuron subtypes (Saalmann et al. 2007).However, a recent study identified expression of 5aR Type II in cortical GABAergic cells, suggesting that neurosteroid synthesis and action may not always be confined to principal excitatory neurons (Castelli et al. 2013).
Whether cortical principal cells or interneuron populations are the locus of neurosteroid synthesis is not directly addressed by our finasteride, or intracellular g-CD experiments.If at P7 -8 the mode of neurosteroid action is exclusively autocrine, then a ubiquitous expression of steroid-synthesizing enzymes across different neuronal populations would be required.Alternatively, the local steroid concentration present during neonatal development may be sufficient to impact upon GABA-ergic and/or principal neurons, which are incapable of neurosteroid synthesis, thereby inferring a paracrine mode of action.Clearly, further studies are required to clarify the relative contribution of autocrine and paracrine neurosteroids to the inhibitory plasticity of the developing cortex.
Irrespective of the locus of synthesis, the lipophilic steroid is considered to access the synaptic GABA A Rs by lateral diffusion via the plasma membrane, a mechanism congruent with the proposed transmembrane neurosteroid binding site on the receptor (Hosie et al. 2006;Chisari et al. 2010).Numerous in vitro electrophysiological studies report enhancement of GABA A R function by low nM aqueous concentrations of 5a3a (see Belelli and Lambert, 2005), advocating the presence of a relatively high affinity binding site on the GABA A R. However, such neurosteroids are highly lipophilic, permitting much greater local concentrations to accumulate in the vicinity of the receptor, obviating the requirement for a high affinity binding site.Indeed, the proposed low affinity binding site (Chisari et al. 2010) is consistent with our observation that even when g-CD was applied exclusively to the cytosolic compartment, it efficiently removed the neurosteroid influence on synaptic GABA A Rs of immature cortical pyramidal neurons.

The role of GABA A R subunit composition in developmental plasticity of phasic GABAergic neurotransmission
The a1 subunit mRNA and protein is present in the cortex early in development, albeit at low levels, whereas the converse is true for a2/3 subunits, which are highly expressed early in life, before they decrease to lower levels in the mature cortex (Laurie et al. 1992;Fritschy et al. 1994;Pirker et al. 2000).Here the mIPSCs obtained from a1 À/À L2/3 pyramidal neurons at P7 -8, P10, P15 and P20 -24 exhibited slower decay kinetics c.f. WT controls at each developmental time-point.Since a1-GABA A Rs are associated with fast decay kinetics (Picton and Fisher, 2007), this finding suggests that a proportion of synaptic a1-GABA A Rs are present even at P7 -8 in L2/3 pyramidal neurons.However, a developmental decrease of the mIPSC decay time of a1 À/À L2/3 pyramidal neurons was still evident, implying the presence of additional factors, as previously suggested (Bosman et al. 2005).Moreover, the developmental profile of g-CD sensitivity for mIPSCs from P7 -8, P10 and P15 a1 À/À L2/3 pyramidal neurons was indistinguishable to WT.In summary, these findings indicate that although the duration of the mIPSCs is influenced by the a1 subunit, it is not solely responsible for the developmental changes that occur in the postnatal period P7 e 20.The results presented here reveal the waning impact during development of the endogenous neurosteroid tone is an additional important factor in influencing phasic GABAergic neurotransmission of both WT and a1 À/À L2/3 pyramidal neurons.However, a comparison of the mIPSC decay of P7 -8 and P20 -24 a1 À/À neurons, when treated with either finasteride, or g-CD, reveals an additional, as yet unidentified factor(s) that influences phasic GABAergic neurotransmission during postnatal development.

The physiological role of GABA A R-active neurosteroids during development
This study focused on L2/3 cortical neurons and found that the synaptic GABA A Rs expressed on both pyramidal and interneuron populations are developmentally influenced by neurosteroids to modulate the duration of GABAergic synaptic transmission.These changes in neurosteroid influence are occurring during an intense period of synaptogenesis, which may be reflected by our observation that the frequency of mIPSCs increases during postnatal development.Furthermore, at this time GABA may exert a depolarizing effect due to the dominance of the chloride importer NKCC1 (Owens et al. 1996;Ben-Ari et al. 2007).Hence, long duration synaptic events may be suited to recruiting voltage-gated calcium channel activation, which in turn may initiate various Ca 2þ -dependent processes in the neuron.In this scenario, the emergence of mature (hyperpolarizing) GABA-ergic signalling would inversely correlate with the decline of neurosteroid production.Adding complexity, a recent report has demonstrated that neonatal administration of 5a3a influences the hippocampal expression of the K þ Cl À co-transporter KCC2 (Modol et al., 2014).
We recently reported a similar developmentally controlled neurosteroid tone in somatosensory thalamocortical neurons (Brown et al. 2015).During development, the window of neurosteroid influence on GABA-ergic transmission in the somatosensory thalamus is shorter than that described here for cortex.In thalamic neurons the neurosteroid tone was absent by P10 (Brown et al. 2015), whereas in cortex this form of endogenous modulation persisted through to P15, but was no longer present by P20 -24.The reasons for the distinct temporal regulation between thalamus and cortex are unknown.In addition to thalamus and cortex, previous studies have identified a similar neurosteroid tone in spinal 0.3 ± 0.1 0.2 ± 0.1 2.7 ± 0.5 2.1 ± 0.9 neurons of the dorsal horn (Keller et al. 2004), which also exhibits a differential developmental profile for lamina II and lamina III/IV neurons (Inquimbert et al. 2008).Hence, the existence of a neurosteroid tone at multiple tiers of the CNS suggests a widespread role during postnatal development.Deciphering the interplay between endogenous neurosteroid synthesis, GABA A R potentiation, and neuronal maturation should therefore be a focus for future studies.

Fig. 1 .
Fig. 1.The properties of mIPSCs recorded from WT L2/3 pyramidal neurons during postnatal development.A).Traces showing typical current recordings from L2/3 pyramidal neurons derived from WT mice at P7 (top), P10 (middle) and P20 (bottom).Note the increase in the frequency of synaptic events with development.B).Averaged, superimposed mIPSCs normalised with respect to peak amplitude, recorded from representative WT L2/3 pyramidal neurons of P7, P10, P15 and P20 mice.The decay time decreases progressively with development.C).A graph showing both the net decrease in t W , and the concomitant increase in mIPSC frequency (n ¼ 14e55 neurons) occurring with development.Symbols represent the mean ± S.E.M.

Fig. 2 .
Fig.2.GABAergic synaptic currents of L2/3 pyramidal neurons are influenced by an endogenous neurosteroid tone during early postnatal development.A -E). Averaged mIPSCs, superimposed and normalised with respect to peak amplitude from a representative P7 cortical pyramidal neuron during control conditions (black line) and following treatment (grey lines) with either, A) 1 mM g-CD, B) 1 mM a-CD, C) 50 mM finasteride (Fin), D) 0.5 mM g-CD in the ICS only, or E) 50 mM finasteride þ 0.5 mM g-CD in the ICS.F).A cumulative probability plot of the mIPSC T50 for control P7 -8 L2/3 pyramidal neurons (events pooled from 55 cells) and P7 -8 L2/3 pyramidal neurons following treatment conditions described above.All treatments apart from a-CD resulted in a significant leftward shift in the T50 distribution indicating that these treatments resulted in mIPSCs with reduced decay times compared with control (in all cases p < 0.001, KS-test).G) Summary bar graph depicting the mean t w values for P7 -8 mIPSCs for control and after cyclodextrin/ finasteride treatments (n ¼ 6e55 cells).* ¼ p < 0.05, *** ¼ p < 0.001, vs control.n.s.¼ not significant, one way ANOVA with Tukey post hoc test.

Fig. 3 .
Fig. 3.The cortical neurosteroid tone is developmentally regulated.A bar graphshowing the effect of g-CD pre-incubation (1 mM) on the t W of mIPSCs recorded from WT L2/3 pyramidal neurons during development (n ¼ 8e55 cells).Note that at P7 -8, P10 and P15 the g-CD pre-incubation resulted in a significant reduction in t W vs control.For comparison, the effect of finasteride (50 mM) preincubation at P7 -8 and P20 -24 is also shown (n ¼ 6e55 cells).yyy ¼ p < 0.001, vs control, post hoc Tukey HSD test following two-way ANOVA.* ¼ p < 0.05, *** ¼ p < 0.001, vs control, post-hoc independent samples t-test following two-way ANOVA.n.s.¼ not significant.The independent variables for the two-way ANOVAs were postnatal age and treatment (g-CD, or finasteride).Illustrated above each developmental time point are the corresponding averaged and superimposed mIPSCs, normalised with respect to peak amplitude, obtained from representative pyramidal neurons in the absence (control, black line) and following 1 mM g-CD pre-incubation (grey line).

Table 1 A
summary of the impact of development and g-CD, or finasteride preincubation on the mIPSC properties of WT L2/3 pyramidal neurons.*¼ p < 0.05,*** ¼ p < 0.001, vs control, unpaired t-test.yyy ¼ p < 0.001, vs control, one way ANOVA with Tukey post hoc test.

Table 3 A
summary of the effects of intracellular g-CD treatment on the properties of mIPSCs of L2/3 GAD 67 GFP þ neurons.***¼ p < 0.001, vs control, unpaired t-test.