Reliable Activation of Immature Neurons in the Adult Hippocampus

Neurons born in the adult dentate gyrus develop, mature, and connect over a long interval that can last from six to eight weeks. It has been proposed that, during this period, developing neurons play a relevant role in hippocampal signal processing owing to their distinctive electrical properties. However, it has remained unknown whether immature neurons can be recruited into a network before synaptic and functional maturity have been achieved. To address this question, we used retroviral expression of green fluorescent protein to identify developing granule cells of the adult mouse hippocampus and investigate the balance of afferent excitation, intrinsic excitability, and firing behavior by patch clamp recordings in acute slices. We found that glutamatergic inputs onto young neurons are significantly weaker than those of mature cells, yet stimulation of cortical excitatory axons elicits a similar spiking probability in neurons at either developmental stage. Young neurons are highly efficient in transducing ionic currents into membrane depolarization due to their high input resistance, which decreases substantially in mature neurons as the inward rectifier potassium (Kir) conductance increases. Pharmacological blockade of Kir channels in mature neurons mimics the high excitability characteristic of young neurons. Conversely, Kir overexpression induces mature-like firing properties in young neurons. Therefore, the differences in excitatory drive of young and mature neurons are compensated by changes in membrane excitability that render an equalized firing activity. These observations demonstrate that the adult hippocampus continuously generates a population of highly excitable young neurons capable of information processing.


Introduction
The dentate gyrus is the main gateway to the hippocampus and it constitutes a primary neurogenic niche of the adult brain. Neural progenitor cells of the subrgranular zone give rise to dentate granule cells (DGCs) that develop and mature over several weeks. A substantial fraction of those newly generated neurons become integrated in the hippocampal network and are then maintained throughout adulthood [1,2]. In recent years, we and others have utilized in-vivo retroviral labeling or transgenic technologies to express fluorescent reporters in new DGCs of the adult mouse hippocampus to study their anatomical and functional maturation [3][4][5][6]. It was thus shown that developing DGCs follow a precise sequence to establish their afferent connectivity and functional maturation. Developing neurons are initially contacted by GABAergic terminals and later by glutamatergic axons. In parallel, their membrane resistance decreases and excitability becomes mature [4,[7][8][9][10]. When fully developed, adult-born neurons achieve a functional profile that is indistinguishable from that of all other DGCs as reflected by their inputs, intrinsic membrane properties and firing behavior [11,12]. In addition, adult-born DGCs form functional glutamatergic synapses onto dentate gyrus interneurons and CA3 pyramidal cells [13,14]. These observations indicate that new neurons receive, process and convey information onto target neurons, and might therefore participate in hippocampal function.
The participation of newly generated neurons in hippocampal processing has been highlighted by recent reports whereby reduced or blocked neurogenesis impair performance in hippocampus-dependent learning tasks [15][16][17][18][19]. Studies using the expression of immediate early genes as indicators of neuronal activity suggest that new DGCs are preferentially active during behaviors that involve spatial processing and memory formation [20][21][22]. Yet, the developmental stage at which new neurons become functionally significant and how those unique properties might impact in their function remains unknown. Young neurons exhibit high input resistance and increased susceptibility for the induction of long-term potentiation (LTP) of entorhinal glutamatergic inputs [23][24][25]. Based on these properties it has been proposed that they might play unique roles in hippocampal function [26][27][28].
To put forward the hypothesis that young neurons behave as a distinct neuronal population within the active hippocampal networks it is crucial to determine when they become engaged in firing activity, and how their activity compares to the remaining units of the network. We have used retroviral labeling and electrophysiological tools to study intrinsic membrane properties and input-output conversion in developing DGCs of the adult mouse hippocampus. We found that three-to four-week-old neurons display weak glutamatergic inputs, yet spike reliably in response to perforant path stimulation. In these immature neurons ionic currents are efficiently converted into membrane depolarization due to their high input resistance. As neurons mature the membrane resistance decreases, with a consequent reduction in excitability. In addition, we found a developmental increase in the inward rectifier K + conductance (Kir) that highlights Kir channels as regulators of the excitability in newborn DGCs in the adult hippocampus.

Results
The activity of the hippocampal network can only be modified by neurons that can integrate incoming signals to produce a firing behavior and alter the state of postsynaptic targets. Firing is ultimately shaped by the concerted action of synaptic integration and excitability. To investigate the impact of developing DGCs in the adult hippocampal network, a retrovirus encoding GFP driven by a strong promoter was used to label adult-born DGCs and acute brain slices were prepared 18 to 29 days after retroviral injection (dpi), at which time neurons are still immature, yet afferent excitatory connections become established [4,9,29]. An additional experimental group of GFP + neurons that reached functional maturity was included for comparison (42 to 56 dpi). Intrinsic properties, excitatory inputs and spiking were characterized by electrophysiological recordings in GFP + neurons and compared with those of unlabeled mature neurons of the outer granule cell layer, that are mostly generated during perinatal development (''mature'' group; see Methods section).

Young neurons are highly excitable
The excitability of developing neurons in the adult dentate gyrus was investigated by monitoring passive and active membrane properties in whole-cell recordings. Injection of current steps of small amplitude readily generated action potentials in young DGCs, whereas increasingly larger currents were required to reach the membrane threshold for action potential in more mature neurons (Figure 1A-1C; Table 1). This is in agreement with previous observations that intrinsic excitability is higher in immature neurons [4,7,8,23,24,30]. The dynamic properties of spikes also displayed a marked time-dependent maturation. Eighteen to 20 dpi neurons (''19 dpi'' group) could only generate single action potentials with immature characteristics in response to prolonged membrane depolarizations, whereas older DGCs exhibited a progressive increase in the number of spikes. Interestingly, 24-29 dpi neurons fired with higher efficacy than younger or older DGCs ( Figure 1B).
The input resistance of a neuron (R in ) reflects the cell size and the density of ion channels open at resting. High values of R in are associated with an enhanced excitability, as small inward currents can elicit large membrane depolarizations. Young neurons displayed high values of R in (in the GV range) that decreased as they reached mature developmental stages ( Figure 1D). The agedependent decline in R in could, by itself, explain the larger current needed to elicit spiking in mature neurons. However, the increase in spiking frequency that accompanies neuronal maturation also suggests changes in spike properties, as addressed below.

Weak glutamatergic inputs elicit spikes in young neurons
To determine whether an excitatory drive can trigger action potentials in immature DGCs, postsynaptic responses to afferent glutamatergic stimulation were recorded in the presence of GABA receptor antagonists (picrotoxin 100 mM and CGP 55845 100 nM). Excitatory postsynaptic currents (EPSCs) were evoked by increasing stimulus intensities delivered to the medial perforant path (Figure 2A). Peak EPSC amplitudes were very weak throughout the stimulus range in 19 dpi neurons but increased substantially with age, reaching a maximal amplitude by 49 dpi ( Figure 2B and 2C). Therefore, all functional studies in developing neurons were carried out in DGCs aging 21 to 29 dpi, hereafter the ''young'' group. The small amplitudes of evoked EPSCs recorded in young neurons ( Figure 2D) might be due to presynaptic mechanisms such as a small number of synaptic contacts and/or low release probability, or to postsynaptic mechanisms such as reduced amplitude of unitary postsynaptic currents [31]. Recordings of miniature excitatory postsynaptic currents (mEPSCs) revealed a reduced frequency in young cells with no differences in amplitude, indicating presynaptic rather than postsynaptic differences ( Figure 2E-2G). Together with the morphological evidence of low spine density in immature DGCs (21 dpi: 0.7860.10 spines/mm, mature: 2.4060.06 spines/mm, N = 19 neurons for both; V.C. Piatti and A.F.S. unpublished observations; see also [29]) our observations indicate that the weak glutamatergic input is due to a limited number of afferent terminals impinging onto young neurons.
Stimuli that evoked the larger EPSC amplitudes were selected to assess spiking by an excitatory drive. DGCs were practically unable to spike by 19 dpi most likely due to their weak excitatory input combined with their limited ability to generate action potentials ( Figure 1B and Figure 2H). As afferent excitation strengthened (.21 dpi), the proportion of young DGCs exhibiting spikes in response to robust stimuli increased substantially. These observations indicate that, in principle, immature neurons of the adult hippocampus might be recruited by cortical afferent activity.

Young DGCs fire with high efficacy
Since young and mature DGCs display distinct excitability, spike properties and input strength, it would be predictable that they are differentially recruited into an active network. To address this question spiking probability of young (21-29 dpi) and mature DGCs were measured in current clamp, while repetitive stimuli of increasing strength were delivered at low frequency to the medial perforant path ( Figure 3A). Surprisingly, young and mature neurons exhibited a similar spiking probability at all stimulus strengths ( Figure 3B). A more precise determination of the firing behavior was obtained by dual cell-attached recordings of young and mature neurons, where the firing probability was measured in pairs of neurons stimulated under the same conditions. This approach brings the advantages that: 1) the intracellular milieu remains unaltered by the recoding pipette; 2) the resting potential of the cells, which greatly influences firing behavior, remains at physiological values; 3) all data are collected in a paired fashion and systematic errors are minimized. In each experiment, action currents elicited by repetitive stimuli delivered at low frequency were recorded to simultaneously assess spiking probability in both neurons. Consistent with the results presented above, no differences in overall firing behavior were observed ( Figure 3C and 3D). Yet, differences were found in the stimulus-to-spike delay; young neurons displayed longer spike latency ( Figure 3E-3F), consistent with the slower EPSP rise time (Table 1). In addition, spike timing precision increased with neuronal maturation ( Figure 3G), consistent with previous observations in the developing brain [32].
It is surprising that young and mature DGCs have different functional properties but display similar firing behavior. To better understand this phenomenon we investigated how synaptic currents are transduced into membrane depolarization. Current-and voltage clamp recordings were combined to monitor subthreshold postsynaptic responses evoked by afferent stimulation with increasing strength. Thus, each stimulus generated an EPSP / EPSC pair whereby a given postsynaptic depolarization was associated to a particular synaptic current ( Figure 4A and 4B). Young neurons required about half the synaptic strength as mature DGCs to depolarize the membrane to a similar extent ( Figure 4C). This observation was corroborated in simultaneous double patch recordings, where the same stimulus elicited similar EPSPs but smaller EPSCs in young DGCs ( Figure 4D and 4E). We also examined whether the high efficacy of membrane depolarization was effective during suprathreshold depolarizations in young DGCs. Thus, EPSC amplitude and the corresponding spiking probability were monitored for each stimulus strength. Similarly to the subthreshold situation, young neurons demanded about half the synaptic strength to reach a similar spiking probability when compared to mature DGCs ( Figure 4F and 4G, Figure S1). Late onset of inward rectifier K + conductance in developing DGCs To better understand the observed differences in firing behavior we characterized action potentials, voltage-gated Na + and K + currents, and inward rectifier K + currents (Kir), which contribute to neuronal excitability in subthreshold conditions [33,34]. The membrane threshold for spiking was slightly hyperpolarized in young DGCs while spike amplitude, rise slope, and afterhyperpolarization (AHP) were more prominent in older cells ( Figure 5A, Figure S2, Table 1). The amplitude of voltageactivated currents also increased according to the maturation stage, yet a slower pace was observed in the development of Kir currents when comparing young vs mature neurons. Kir conductance increased by ,300% whereas voltage-gated Na + and K + currents reached near plateau levels in young neurons, increasing by only ,50% ( Figure 5B-5I, Figure S3). These observations suggest an ongoing homeostatic regulation that maintains a high degree of excitability in immature DGCs by boosting currents involved in spike generation while limiting the development of Kir currents.
The observations described above suggest that the activity of DGCs might be regulated by the level of Kir expression. To address this question neuronal excitability was studied in the presence of extracellular Ba 2+ (200 mM), a well known blocker of Kir channels [34][35][36]. Application of Ba 2+ effectively abolished the inward rectification current component in young and mature neurons ( Figure 6A, Figure S4). To evaluate Kir contribution to the firing behavior, depolarizing currents were delivered to young and mature neurons in the presence and absence of Ba 2+ . As shown above ( Figure 1B), young neurons displayed enhanced repetitive spiking for small injected current pulses. Ba 2+ induced a significant leftward shift in the spiking curve of mature neurons without altering the responsiveness of young cells ( Figure 6B). We then assessed how acute Kir blockade affects synaptic integration. Analysis of subthreshold depolarization in response to evoked synaptic currents revealed that Ba 2+ produced an increase in the EPSP/EPSC slope of mature DGCs toward values that are similar to those of young neurons ( Figure 6C and 6D). Therefore, Kir blockade elicits a young neuronal behavior in fully mature DGCs, equalizing their degree of excitability. To investigate the effects of accelerated Kir expression on the excitability of young neurons, Kir 2.1 was overexpressed by retroviral delivery into dividing progenitor cells [37]. Kir overexpression increased inward rectifying currents and decreased suprathreshold excitability, inducing mature-like properties in young neurons ( Figure 6E and 6F). These findings strongly suggest that young neurons of the adult dentate gyrus can spike with high efficacy due to the delayed onset of Kir channels.

Discussion
DGCs generated in the adult hippocampus receive functional afferents, spike in response to an excitatory drive, and release glutamate onto postsynaptic target cells (this work, [3,4,9,11,13,14]). This complete set of functional properties endows newly generated neurons with the capacity to play a significant role in hippocampal function. In agreement with this view, exploratory behavior and spatial learning paradigms that involve various hippocampal areas displayed preferential recruitment of new neurons within the dentate gyrus [20,21]. In the present work we have used a retroviral labeling technique that allows reliable birth dating of newly generated neurons. Although fully mature neuronal properties were not observed until 49 dpi, DGCs that have developed for three to four weeks received weak excitatory inputs yet were capable of spiking in response to an afferent excitation. This is due to the fact that young neurons are very efficient in transducing ionic currents into membrane depolarization. Together with the recent observation that new neurons establish synapses onto target cells before reaching maturation [13,14], the data presented here supports the notion that young DGCs participate in information processing [21,27,28]. Whether or not they are preferentially recruited during hippocampus-dependent behaviors may depend on the integration of both excitatory and inhibitory inputs, an issue that remains to be investigated.
Kir channels are involved in the resting membrane potential and conductance, exerting a role in the regulation of cellular excitability [36]. They are conductive at the resting potential and, therefore, decrease the membrane resistance with a consequent reduction in excitability. Intrinsic excitability refers to the propensity of a neuron to fire action potentials when exposed to an input signal, and it is directly attributable to the suite of ion channels inserted into the plasma membrane [38,39]. Young DGCs display a high input resistance, low threshold potential for spiking, low threshold currents, and large EPSPs relative to the weak glutamatergic input. The expression of Kir in DGCs has been previously observed by in situ hybridization [40], and it is now demonstrated in young and mature neurons by I-V curves and pharmacological blockade. Kir currents strongly influenced the resting membrane conductance, since application of Ba 2+ increased membrane resistance of mature neurons close to young DGC values and enhanced their excitability.  Voltage-gated Na + and K + channels displayed a slight agedependent increase, consistent with the observed changes in action potential threshold and shape. This minor increment indicates that voltage-gated currents were close to plateau levels in young neurons. Consistent with this notion we observed only a minor increase in the rising slope of the action potential, a parameter that reflects the density of Na + channels [41]. Remarkably, Kir conductance was weak in young neurons and increased substantially in mature DGCs. The early expression of voltage-gated Na + and K + channels together with the delayed expression of Kir optimize excitability of young neurons. This mechanism might obey to a homeostatic requirement to maintain a firing pattern in neurons with immature glutamatergic inputs. Supporting this idea, a stimulus delivered to the medial perforant path elicited similar spiking probabilities in young and mature neurons. The involvement of homeostatic mechanisms in the control of intrinsic excitability has been investigated in detail [38,39]. In developing neurons, the strength of voltage-gated currents can be tailored by the synaptic drive to stabilize the dynamic range of neuronal output [42,43]. In addition, reduced inhibition was shown to be compensated by an increase in endogenous leak conductance thus maintaining firing activity unaltered [44]. Our results suggest a homeostatic compensation of intrinsic excitability mediated by Kir that equalizes firing properties of all active DGCs in the adult hippocampus. This mechanism may facilitate activity dependent competition for survival [45] or, alternatively, it may strengthen the integration of young neurons in the network at a time in which afferent excitatory synapses display enhanced plasticity [25].

Viral vectors
A replication-deficient retroviral vector based on the Moloney murine leukemia virus was used to express enhanced GFP (or Kir-IRES-GFP in experiments shown in Figure 6E and 6F) driven by a CAG promoter [11,29]. Mouse Kir 2.1 cDNA was kindly provided by G. Lanuza. Retroviral particles were assembled using three separate plasmids containing the capside (CMV-vsvg), viral proteins (CMV-gag/pol), and transgene (CAG-GFP). Plasmids were transfected into 293T cells using Lipofectamine 2000 (Invitrogen, Carlsbad, CA). Virus-containing supernatant was harvested 48 h after transfection and concentrated by two rounds of ultracentrifugation.

Subjects and stereotaxic surgery
Female C57Bl/6J mice, 6-7 weeks of age, were anesthetized (100 mg ketamine/10 mg xylazine in 10 ml saline/g). CAG-GFP expressing retrovirus was infused (0.9 ml in 7 min) into the dorsal area of the right dentate gyrus (coordinates from bregma: anteroposterior = 22 mm, lateral = 1.5 mm, ventral = 1.9 mm) using a microcapillary calibrated pipette (Drummond Scientific, Broomall, PA) as described previously [4]. Housing, treatments, surgery and euthanasia were carried out under conditions that fully comply with the National Institutes of Health (NIH, USA) guidelines.
Whole-cell voltage-clamp recordings were performed at a holding potential (V h ) of 270 mV, unless otherwise noted. Criteria to include cells in the analysis were co-labeling with Alexa Fluor 594 or visual confirmation of GFP in the pipette tip and absolute leak current ,100 pA at V h . Series resistance was typically 10-20 MV, and experiments were discarded if higher than 25 MV. Membrane capacitance and input resistance were obtained from current traces evoked by a hyperpolarizing step of 10 mV. In current-clamp recordings the resting membrane potential was kept at 270 mV by passing a holding current. The threshold current for spiking was assessed by successive depolarizing current steps (5 or 10 pA; 500 ms) to drive the membrane potential (V m ) from resting to 0 mV. Action potential threshold was defined as the point at which the derivative of the membrane potential dV/dt deviated from the mean baseline value by .2 standard deviations [46], and was calculated using routines developed in our laboratory ( Figure S2). Voltage-dependent Na + and K + currents were measured after leak subtraction using a p/26 protocol and automatic detection of the fast inward peak and the late outward plateau. Kir currents were measured as the steady state inward deflection in response to hyperpolarizing voltage steps (from 2130 to 290 mV). Kir conductance was calculated after subtracting the leak conductance measured between 260 and 240 mV, where Kir channels are closed. Kir conductance calculated in this manner did not differ significantly from the one measured as the fraction of inward conductance with sensitivity to blockade by 200 mM Ba 2+ .
Extracellular stimulation of the medial perforant path was performed using concentric bipolar electrodes (50 mm diameter; Frederick Haer Company, Bowdoinham, ME) and an Iso-Flex stimulator (A.M.P.I.; Jerusalem, Israel). The stimulation electrode was placed orthodromically on the middle third of the molecular layer at .250 mm from the recorded cell. Functional inputs were assessed at stimulus strengths of 0.1-1.3 mA (50 ms) repeated at 15-s intervals. Excitatory postsynaptic currents (EPSCs) and potentials (EPSPs) were recorded in the presence of picrotoxin (100 mM) and CGP 55845 (100 nM), antagonists of GABA A and GABA B receptors. Peak amplitudes were calculated from the average of 5 traces (EPSCs) and 10-20 traces (EPSPs). Simultaneous cell-attached recordings were carried out under voltage clamp at 0 mV using pipettes with a high tip resistance (10)(11)(12)(13)(14). Stimulus intensity was gradually increased (0.3-2 mA, 50 ms) and, for each cell, spiking probability was obtained after 10-20 stimuli. For each pair of cells, the lower stimulus intensity at which both (young and mature) neurons fired was selected for statistical analysis of spiking probability. Whole-cell recordings were carried out at the end of all experiments to verify granule cell phenotype by spiking properties and morphology after filling with a fluorescent dye. Only experiments in which spiking was detected in both neurons were considered for analysis. Recordings were discarded if the seal resistance reached values below 8 GV. The integrity of the cell-attached patch was further confirmed by the absence of fluorescent dye in the cytoplasm. Figure S1 Spiking probability vs. input strength. Spiking probability vs. EPSC amplitude measured in young and mature DGCs. Input strength was binned into three categories according to the EPSC amplitude. (*) and (**) denote p,0.05 and p,0.01 by two-way ANOVA revealing a significant effect by age, with N = 23 (both).