The functional connectivity of the cerebral cortex is shaped by experience during development, especially during a critical period early in life. In the prenatal and neonatal cortex, transient neuronal circuits are formed by a population of subplate neurons (SPNs). However, SPNs are absent in the adult cortex. While SPNs are crucial for normal development of the cerebral cortex and of thalamocortical synapses, little is known about how they are integrated in the developing thalamocortical circuit. We therefore investigated SPNs in vitro in thalamocortical slices of A1 and medial geniculate nucleus (MGN) in mouse from postnatal day 1 (P1) to P13. We found that SPNs can fire action potentials at P1 and that their intrinsic membrane properties are mature after P5. We find that SPNs receive functional excitatory inputs from the MGN as early as P2. The MGN projections to SPNs strengthen between P2 and P13 and are capable of inducing action potentials in SPNs. Selective activation of SPNs by photostimulation produced EPSCs in layer 4 neurons, demonstrating a functional excitatory connection. Thus, SPNs are tightly integrated into the developing thalamocortical circuit and would be a reliable relay of early spontaneous and sound-evoked activity. The role of SPNs in development likely results from their strong excitatory projection to layer 4, which might function to regulate activity-dependent processes that enable mechanisms required for the functional maturation and plasticity of the developing cortex and thereby contribute to the development of normal cortical organization.
Subplate neurons (SPNs) are among the earliest-appearing neurons in the cerebral cortex that are largely lost in the adult (Allendoerfer and Shatz, 1994; Kanold, 2004, 2009). Subplate neurons are located in the developing white matter of all cortical regions, including auditory cortex (ACX). In rodents, SPNs exist during the first 3 weeks of life, but some (15–39%) remain into adulthood (Vandevelde et al., 1996; Price et al., 1997; Reep, 2000; Clancy et al., 2001; Torres-Reveron and Friedlander, 2007). Subplate death occurs earlier at caudal than rostral areas (Price et al., 1997), thus more SPNs might be present in ACX than in visual cortex at a given age. Removing SPNs in visual cortex, however, prevents the segregation and functional maturation of thalamocortical and intracortical projections, suggesting a key role for SPNs in cortical development (Ghosh and Shatz, 1992; Kanold et al., 2003; Kanold and Shatz, 2006). SPNs receive input from lower brain centers, such as the thalamus, before their axons grow to their targets in the cortex (Friauf et al., 1990; Allendoerfer and Shatz, 1994; Hanganu et al., 2002; Higashi et al., 2002; Molnár et al., 2003; Kanold, 2004). Glutamatergic SPNs project into the overlying developing cortical plate and are thought to excite cortical neurons (Friauf et al., 1990; Finney et al., 1998; Piñon et al., 2009). Such an excitatory function of SPNs on layer 4 neurons combined with reliable transmission of thalamic activity through SPNs could control activity-dependent processes in layer 4 leading to mature cortical circuitry (Kanold et al., 2003; Kanold and Shatz, 2006; Kanold, 2009). However, direct evidence of such an excitatory connection has been lacking.
While SPNs are present before and during the onset of functional hearing [∼postnatal day (P)8–P9] (Alford and Ruben, 1963; Mikaelian and Ruben, 1965; Mikaelian et al., 1965; Shnerson et al., 1983; Friauf et al., 1990; Romand and Ehret, 1990; Catalano et al., 1991; Friauf and Shatz, 1991; Ehret and Romand, 1992; Molnár and Blakemore, 1995), little is known about the functional properties of SPNs in ACX. Given the potential key role of SPNs mediating thalamic excitation to the cortical plate we therefore investigated SPNs and their associated circuits by in vitro recordings in thalamocortical brain slices from neonatal mouse ACX.
We found that SPNs fire action potentials as early as P1. SPNs receive thalamic (MGN) inputs as early as tested (P2) and over postnatal development the number of MGN inputs to SPNs increases. MGN inputs can elicit action potentials in SPNs, aided by the high input resistance and depolarized resting potential of SPNs at young ages. In addition, we find that selective SPN stimulation results in EPSCs in layer 4 neurons, thus SPNs provide excitatory input to layer 4. Together, we demonstrate that SPNs are integrated into the developing thalamocortical microcircuit, receive functionally strong thalamic input, and provide a strong excitatory feedforward excitation to developing layer 4, the target layer of thalamic axons. This feedforward excitatory circuit might influence maturational processes in layer 4, leading to mature thalamocortical and intracortical connectivity.
Materials and Methods
Mice (C57BL/6) from Jackson (jax.org) are deeply anesthetized with isofluorane (Halocarbon). A block of brain containing ACX and MGN is removed and thalamocortical slices (500–600 μm thick) are cut on a vibrating microtome (Leica) in ice-cold ACSF containing (in mm): 130 NaCl, 3 KCl, 1.25 KH2PO4, 20 NaHCO3, 10 glucose, 1.3 MgSO4, 2.5 CaCl2 (pH 7.35–7.4, in 95%O2-5%CO2). The cutting angle is ∼15 degrees from the horizontal plane (lateral raised) at older ages (Cruikshank et al., 2002) and somewhat steeper at younger ages to compensate for the lateral growth of cortex. Slices are incubated for 1 h in ACSF at 30°C and then kept at room temperature. For recording, slices are held in a chamber on a fixed stage microscope (Nikon FN1) and superfused (2–4 ml/min) with ACSF at room temperature. The location of the recording site in ACX was identified by landmarks (location of hippocampus) (see Fig. 1A) and validated by monosynaptic responses to electrical MGN stimulation.
Identification of subplate.
The subplate zone is readily identified as a relative cell sparse area between the large layer 6 pyramidal cells and the ventricular zone. The subplate encompasses horizontally oriented cells in an upper sublamina below layer 6 (future layer 6b) and deeper more scattered neurons in the future white matter. The upper sublamina is separated from layer 6 by a cell sparse zone. Compared to other neurons, the deeper subplate neurons in young rodents stain relatively poorly for markers such as NeuN but are apparent under differential interference contrast (DIC) where subplate neurons intermingled with horizontally traveling fiber can be readily seen due to their horizontal appearance (see Figs. 1A,B, 6).
Whole-cell recordings are performed with a patch-clamp amplifier (Multiclamp 700B, Molecular Devices) using pipettes with input resistance of 4–8 MΩ. Data are acquired with a Digidata AD board (Molecular Devices) under pClamp (v9 and v10) and analyzed offline using MATLAB (MathWorks). Electrodes are filled with (in mm) 110 K-gluconate, 4 KCl, 4 NaCl, 0.2 CaCl2, 10 HEPES, 1.1 EGTA, 2 Mg-ATP, 1 MgCl2 and 5 glutathione (pH 7.2, 300 mOsm). Membrane voltages are corrected for an estimated liquid junction potential of 14 mV. Some cells were recorded using a solution containing 115 mm cesium methanesulfonate (CsMeSO3) 5 NaF, 10 EGTA, 10 HEPES, 15 CsCl, 3.5 MgATP, 3 QX-314 (pH 7.25, 300 mOsm). We found no obvious differences in EPSC sizes with these two solutions. For LSPS all cells were recorded with this solution. Biocytin or neurobiotin (0.5%) is added to the electrode solution as needed. Electrical stimulation: We use a bipolar electrode (Microprobe, biphasic, 0.2 ms, at 0.03 Hz) coupled to a stimulus isolator (Cygnus). Photostimulation: 0.5–1 mm caged glutamate (N-(6-nitro-7-coumarylmethyl)-l-glutamate; Ncm-Glu) (Kao, 2006) is added to the ACSF. Without UV light, this compound has no effect on neuronal activity (N = 40 neurons) (Kao, 2006). UV laser light (500 mW, 355 nm, 100 kHz repetition rate, DPSS, 2–10 ms pulses) is split via a 33% beam splitter (CVI Melles Griot) and coupled via a fiber optical coupler (Oz Optics) into an optical fiber (25 μm core diameter, Oz Optics). The distal end of the fiber is held within a bent glass pipette to reduce beam elongation and apply focused light to a small area. We visualized the stimulated site for all cells by taking a picture of the light beam to ensure that stimulation was limited to the subplate. In preliminary experiments we varied laser power and determined that this power provided reliable direct activation of neurons recorded in whole-cell patch configuration. For laser-scanning photostimulation (LSPS), the other fraction of the split beam is coupled into a microscope via scan mirrors (Cambridge Technology) and a dichroic mirror (Shepherd et al., 2003). The laser beam in LSPS enters the slice axially through the objective (Olympus 4×, 0.16 NA/air) and has a diameter of ∼15 μm. Laser power at the sample is <20 mW. In preliminary experiments we varied laser power and determined that this power provided reliable activation of neurons. We typically imaged 16 × 16 or 32 × 32 stimulation sites spaced 20–50 μm enabling us to image areas of 1mmx1 mm. Stimuli are applied at 0.5–1 Hz. Data acquisition is performed by National Instruments AD boards and custom software (Ephus) (Shepherd et al., 2003). Ephus (available online at http://research.janelia.org/labs/display/ephus/Ephus) is written in MATLAB and adapted to our setup. Analysis was performed similar to methods described previously (Shepherd et al., 2003) by using custom software written in MATLAB. To detect monosynaptically evoked EPSCs we detected peak EPSC amplitudes in a ∼60 ms time window after the stimulation. Traces containing a short-latency (<15–20 ms “direct”) response were discarded from the analysis (see Fig. 6, black patches) as were traces that contained longer latency inward currents of long duration (>100 ms). (Fig. 6, gray patches) These currents could be sometimes seen in locations surrounding (<50 μm) areas that gave a direct response. Occasionally, some of the direct responses contained synaptically evoked responses that we did not separate out, leading to an underestimation of local short-range connections. Cells that only showed direct responses were excluded from the analysis.
We use NBQX (50 μm) to block AMPA currents and TTX (1 μm) to block action potentials. All chemicals and drugs were obtained from Sigma.
Results are plotted as means ± SD, compared with a Students t test, and deemed significant if p < 0.05.
We investigated the physiological characteristics of subplate neurons in mouse auditory cortex over three age ranges (P1–4, P5–P9, and P10–P13), approximately corresponding to distinct functional states of the auditory system. Hair cells do not become functional before P5 and thus only spontaneous evoked activity is present in the auditory system. After onset of hair cell function but before the opening of the ear canal, spontaneous and high-threshold sound-evoked responses might occur. In mice of the C57BL/6 strain the ear canal opens at P9 (Shnerson and Pujol, 1983; Pujol et al., 1997; Ruebsamen and Lippe, 1997). After ear opening, low-threshold hearing develops, followed by an early critical period (∼P11–13 in rat) (de Villers-Sidani et al., 2007).
The intrinsic properties of subplate neurons mature over the postnatal period
We recorded from SPNs (N = 140) in thalamocortical slices of A1 over the first 2 postnatal weeks using whole-cell patch-clamp recordings (Fig. 1A–C). SPNs are located in the future white matter below layer 6 and have diverse morphology (Allendoerfer and Shatz, 1994; Luhmann et al., 2000) (Fig. 1B–D). In response to current injections, SPNs at all ages were able to fire action potentials (APs) (Fig. 1E). However, in response to depolarization, the pattern of APs and the number of spikes evoked varied strongly with age. At early ages (P1/4) sustained current injection resulted in only one AP in 11 of 13 SP neurons, while this firing pattern was rare at ages older than P5 (4/37 cells). Figure 1F shows that maximal number of evoked action potentials to a 250 ms depolarization increased from P1/4 to P10/13 (P1/4: 1.1 ± 0.3, P5/9: 3.6 ± 1.8, P10/13: 3.6 ± 1.6, both p < 0.01). However, action potential threshold remained constant over this age range (−48.4 ± 6 mV vs −50.9 ± 4 mV, p > 0.1).
One feature of SPNs at P1/4 is a sustained depolarization after the AP (Fig. 1E), which might be due to low expression levels of K-channels. To test the availability of K-channels, we measured the membrane resistance (Rm) of SPNs by linear fits to the I–V curve below spike threshold and found that Rm decreased between P1/4 and P10/13 (P1/4: 757 ± 284 MΩ, P5/9: 554 ± 260 MΩ, P10/13: 285 ± 95 MΩ, both p < 0.05) (Fig. 1G). In addition, in the first 2 postnatal weeks, the resting membrane potential showed a steady hyperpolarizing trend (P1/4: −56.9 ± 13.9 mV, P5/9: −63.7 ± 8.4 mV, P10/13: −68.5 ± 6.3 mV; P1/4 to P5/8 p = 0.13; P10/13 p < 0.05) (Fig. 1H). These results also indicate that SPNs at early ages require less depolarization (8.5 mV at P1/4 vs 17.6 mV at P10/13) to trigger an action potential—that is, SPNs at early ages can more easily respond to synaptic inputs.
SPNs receive functionally strong thalamic inputs
SPNs are thought to receive thalamic inputs, as has been demonstrated directly in the cat visual and rodent somatosensory cortex by CSD analysis and optical recordings (Friauf et al., 1990; Allendoerfer and Shatz, 1994; Hanganu et al., 2002; Higashi et al., 2002; Molnár et al., 2003). We therefore investigated if SPNs in ACX receive inputs from the auditory thalamus [medial geniculate nucleus (MGN)] and how these inputs change over the course of development. MGN fibers reach the cortical plate before birth (Gurung and Fritzsch, 2004). Electrical stimulation of the MGN as early as P2 resulted in short-latency EPSCs in SPNs (Fig. 2A), indicating that monosynaptic transmission from thalamus to cortex is already present before the onset of hearing. Occasionally, we observed larger, long-latency EPSCs, which might reflect disynaptic inputs (Fig. 2A, red traces). We focused our analysis on the monosynaptically evoked EPSC, which was taken as either the only evoked EPSC, or the first (short-latency) EPSC if multiple EPSCs were observed.
Increasing stimulation amplitudes resulted in corresponding increases in EPSC amplitude (Fig. 2A). In some cells, especially at young ages, these increases occurred in a stepwise manner (Fig. 2A, top), suggesting the activation of separate synapses, whereas at older ages such steps were less clear (Fig. 2A, bottom, B). The minimally evoked EPSC amplitude was similar at P1/4 and P10/13, while the maximally evoked EPSCs were larger at older ages (Fig. 2C). The ratio of maximal to minimal EPSC is an estimate of the number of thalamic fibers innervating SPNs, and the increased ratio (Fig. 2C) suggests that the thalamo-subplate synaptic connection does not prune but increases its numbers of connections during the postnatal period, even though SPNs are starting to disappear. EPSC latencies (time to peak) were similar at all ages (Fig. 2D).
However, we did not observe any MGN stimulation evoked outward currents under our recording conditions at any age, which suggests that MGN inputs might not engage inhibitory circuits synapsing on SPNs.
Since layer 6 neurons are located close to SPNs, we investigated if layer 6 neurons receive MGN inputs at these ages. We recorded from layer 6 neurons (N = 8) (Figs. 1D, 3A,C) in the same slice in which SPN showed MGN inputs (Fig. 3B,D). In these slices and by stimulating at the same stimulation site, layer 6 cells did not show short latency inputs (Fig. 3B,D). However, 2/8 layer 6 neurons showed long latency inputs (Fig. 3D). Comparison of the layer 6 responses with SPN responses to stimulation at the same site (Fig. 3D) shows the large differences in latencies. These data might also indicate that SPNs receive MGN inputs and drive cortical circuits that give rise to layer 6 inputs.
The high input resistance at young ages suggests that even small EPSCs might have a large effect on the membrane potential of SPNs. Given the relatively depolarized resting potential and the smaller depolarization required to reach action potential threshold, it is possible that small EPSCs might produce large EPSPs and action potentials. Figure 4, A and B, shows sequential voltage and current-clamp recordings in a cell at P7 using stimuli of identical intensities. Note that even at the lowest intensities, which produced only a small EPSC (Fig. 4A), a large, long-lasting EPSP was generated (Fig. 4B) (cell held hyperpolarized to prevent action potentials). If the cells were held close to their normal resting potentials (Fig. 1), these large EPSPs were able to produce action potentials (N = 3 cells) (Fig. 4C). These large and long-lasting EPSPs might partially be due to the high input resistance and long time constant of SPNs, but also could indicate recruitment of NMDA receptors by sufficient depolarization. We therefore investigated if NMDA receptors are present at the MGN-SPN synapse by holding SPNs at two different potentials, −60 mV and +40 mV. Holding SPNs between P4 and P13 at depolarized potentials (+40 mV) revealed a large long lasting outward current in 15/15 SPNs tested, which was abolished by holding the neuron at ∼0 mV (the reversal potential of glutamate receptors) (Fig. 4D). This indicates that this current was carried by glutamate. In addition, the lack of inward currents at 0 mV indicates that MGN stimulation did not recruit GABAergic inputs to SPNs. The conductance ratio of the NMDA and AMPA component of the glutamatergic current gNMDA/gAMPA (calculated from the peak currents by dividing by the driving force) was 3.4 ± 1.5. Together, these data show that synaptic transmission from MGN to SPN is reliable and functionally strong at young ages, and includes both AMPA and NMDA receptor-mediated components. Thus, SPNs can reliably relay MGN activity to the developing cortex.
Subplate neurons excite layer 4 neurons
SPNs axons project into the developing cortical plate, especially radially to layer 4 (Friauf et al., 1990; Allendoerfer and Shatz, 1994; Piñon et al., 2009). One outcome of SPN removal is the lack of maturational processes that are thought to depend on neuronal activity, such as the upregulation of KCC2 (Kanold et al., 2003; Kanold and Shatz, 2006). Thus, the strong maturational effect of SPNs on layer 4 might be mediated by strong excitatory projections from SPNs to layer 4. To investigate the functional effect of SPNs on layer 4, SPNs need to be selectively stimulated. However, common techniques such as electrical stimulation of the white matter will excite both SPNs and thalamic axons. We therefore selectively activated SPNs by photostimulation. Selective stimulation of SPNs was achieved by bathing the slice in caged glutamate solution and locating an optical fiber tip in the subplate (Fig. 5A). The illuminated area has an elliptical shape, with the long axis being <100 μm and the short axis being <50 μm, and is entirely contained in the subplate. We verified that the activation area is in the subplate by visualizing the beam (Fig. 5A). In the presence of caged glutamate in the bath a brief laser pulse led to short-latency inward currents in SPNs, but only if the SPN was in the center of the illuminated area (Fig. 5B, black trace) (N = 5). These photostimulated inward currents were readily blocked by NBQX and were thus glutamatergic (Fig. 5B, red trace). Current-clamp recordings showed that these currents were sufficient to evoke action potentials in SPNs at all ages studied (Fig. 5C,D). The latency of the evoked action potential was consistent across trials (Fig. 5E). We then investigated whether photostimulation of SPNs evokes synaptic currents in layer 4 cells located radially above SPNs. Following photostimulation of SPNs, layer 4 cells (30/32) showed an inward current (Fig. 5F, black trace), while layer 2/3 cells (2/2) did not show any evoked currents following SPN stimulation. The inward current in layer 4 cells was blocked by TTX (Fig. 5F, red trace), indicating that neuronal spiking, presumably from SPNs, was involved, and that the current did not originate by direct activation of glutamate receptors on the layer 4 neuron by the uncaged glutamate. The onset latency of the photostimulation-evoked EPSC matched the latency of SPN firing (Fig. 5F), and was longer than the inward currents produced by direct activation of the SPNs, further suggesting that it was generated by SPN activation. Following SPN stimulation, EPSC amplitudes in layer 4 ranged from 10 to 170 pA (Fig. 5G). However, the average amplitude did not change over development (P5/8: 34 ± 29.3 pA N = 12, P9/13: 48.9 ± 52 pA N = 15; p > 0.1). Because input resistances of layer 4 cells ranged from 310 to 1184 MΩ (mean 717 ± 255 MΩ), these EPSCs should strongly depolarize layer 4 neurons, just as MGN inputs depolarize SPNs. Indeed, current-clamp recordings (N = 3 cells) revealed large EPSPs in layer 4 following photostimulation of SPNs (Fig. 5H), indicating a strong functional impact of SPNs on layer 4 neurons. Together, these data indicate a strong functional excitatory connection between SPNs and layer 4 neurons.
We next investigated the spatial pattern of SPN inputs to layer 4 by using laser-scanning photostimulation (Shepherd et al., 2003). The laser beam was targeted on a 16 × 16 to 32 × 32 pixel array covering an area of the slice (Fig. 6A). Cell-attached patch recordings from SPNs (N = 3) showed that laser pulses were able to trigger short latency action potentials in SPNs (Fig. 6B,C). Plotting the spatial location of stimulation sites that cause SPN spiking showed that photostimulation had to be performed within ∼100 μm of the SPN soma (Fig. 6B,C). Plotting the latency to the first spike (Fig. 6C) shows that activation close to the soma caused spikes with the shortest latency. Thus, only photostimulation of the soma or proximal dendrites is sufficient to activate SPNs.
We next investigated the origin of synaptic inputs to layer 4 by whole-cell recordings from layer 4 neurons during LSPS (N = 21). Photostimulation caused large amplitude short latency inward currents (labeled direct in Fig. 6D1), which reflect the direct activation of the cell under study (Shepherd et al., 2003; Barbour and Callaway, 2008). We also observed smaller amplitude long latency inward currents (labeled “evoked” in Fig. 6D1). These currents likely reflect EPSCs evoked by photostimulation of presynaptic cells. Plotting the amplitude of the pEPSC as function of stimulation location showed that direct responses (Fig. 6D2, pink boxes) were clustered around the cell body (Fig. 6D2, red circle). Evoked responses were also clustered but in a spatially distinct region (Fig. 6D2, yellow boxes). To identify the laminar source of the evoked response, we plotted the amplitude of the responses and overlayed this graph with the DIC image of the slice (Fig. 6D3). We find that the source of the excitatory inputs was located the subplate but not layer 6 (Fig. 6D3). The source region for excitatory inputs in the subplate is relatively small (<200 × 200 μm). Comparing the activation region in subplate (Fig. 6B,C) with projection regions in layer 4 (Fig. 6D3) suggests that each layer 4 neuron receives input from 2 to 3 SPNs. The latency of the pEPSCs (Fig. 6D3, bottom) is similar to those of SPN spiking (Fig. 6C). The results of LSPS of a layer 4 neurons at P13 are shown in Figure 6E. In addition to subplate inputs (13/21 cells), layer 4 neurons receive inputs locally from layer 4, from layer 2/3, and in some cases projections from layer 6 (9/21 cells) (Fig. 6, compare D, E).
In the adult layer 6, neurons can send some projections into layer 4, even though this projection might be relatively weak in auditory cortex (Barbour and Callaway, 2008). Since layer 6 neurons can send dendrites into the subplate it is possible that our stimulation activates layer 6 neurons. To investigate if layer 6 neuron activation can contribute to the observed EPSCs in layer 4 neurons, we mapped directly the activation of layer 6 neurons (N = 2) and found that layer 6 neurons were readily excited by stimulation in a large area surrounding their soma but also in an area extending radially toward the pia (Fig. 6F). The latter area likely coincides with the prominent apical dendrite of layer 6 pyramidal cells (Fig. 1D). This large activation area is similar to those reported in layer 2/3 and layer 5 pyramidal cells (Shepherd et al., 2003). Thus, layer 6 neurons can be activated from a much larger area than SPNs. The large activation area of layer 6 neurons is larger than the size of the source of excitatory connections to layer 4 neurons. In particular, the region that gives rise to EPSCs in layer 4 (Fig. 6D,E) does not include the large activation regions of layer 6 cells (Fig. 6F), suggesting that there is a excitatory connection between SPNs and layer 4 neurons.
Our data show that SPNs are a strong excitatory relay of thalamic information to the developing cortical layer 4 before and after the onset of hearing. We also show that SPNs provide strong excitatory input to layer 4.
We report that SPNs receive MGN inputs and that these inputs are strong enough to trigger action potentials in SPNs. This strong effect of relatively weak synaptic inputs may be mediated by the high input resistances of SPNs. Thus, in effect, the “immature” intrinsic characteristics of SPNs (high Rm) might compensate for the immature synaptic inputs (small amplitudes). High input resistances in early development are not unique to SPNs: for example, layer 2/3 cells show a similar progression ∼10 d later (Oswald and Reyes, 2008). The decrease in input resistance and resting potential of SPNs over development are consistent with the increased expression of K-channels during the postnatal period. Importantly, the functional compensation of weak immature synapses by coincident development of intrinsic mechanisms points to a general mechanism whereby neural circuits can be functional even though their components have not matured. One feature of immature SPNs is that the action potential latency is very short, allowing SPN firing to preserve the timing information present in the thalamic inputs. Such preservation of timing information has been shown to be important in computational models of the role of SPNs in cortical development (Kanold and Shatz, 2006).
While we observed a strengthening of the maximal MGN EPSC over development, we also observed a large increase in variance in EPSC amplitudes (Fig. 2C). We interpret this variability as originating from our preparation, since growth over development prevents us from preserving the entire thalamocortical pathway in a slice preparation at older ages. It seems likely that the maximal observed EPSC at older ages represents a lower estimate of the true total MGN input onto SPNs.
Thalamic input has been implicated in the survival of SPNs (Price and Lotto, 1996) in that removal of thalamic innervation from SPNs after their ingrowth and strengthening in layer 4 would cause the death of SPNs by depriving them of their excitatory inputs. However, we observe an increase rather than a decrease in the number of MGN inputs to SPNs during a period when thalamic inputs mature. Thus, retraction of thalamic inputs is unlikely to be the “death signal” for SPN neurons. Imaging experiments show that in S1 thalamic activation of subplate seems to decline by P10 (Higashi et al., 2005). This is consistent with our observation of a decrease in input resistance and thus excitability, which can counteract increases in synapse number. Thus, SPN death could be caused by reduced functional activation of SPNs.
In some areas of the brain synaptic circuits refine by initial exuberant growth and later pruning, while in other areas the numbers of synaptic inputs increase without evidence of hyperinnervation (LeVay et al., 1978; Bureau et al., 2004; Chen and Regehr, 2000; Friedlander and Martin, 1989; Price and Blakemore, 1985). We here find that thalamic synapses onto SPNs strengthen over development. However, compared to other areas of the brain there are distinct differences in SPN circuits, in that the majority of SPNs die. Thus, it might be that remaining SPNs “inherit” MGN inputs from SPNs targeted for programmed cell death and that SPNs at older ages showing smaller MGN input are the ones targeted for “cell death.” Such a mixed population of SPNs could also contribute to the large variability of maximal EPSCs seen in our data (Fig. 2C).
Following MGN stimulation, we occasionally observed disynaptic EPSCs in SPNs, especially with higher stimulation intensities. Thus, MGN stimulation likely activates excitatory SPNs, which appear to synapse on other SPNs (Fig. 7). Electrical stimulation of the white matter (which activates SPNs) activates GABAergic inputs on SPNs, thus it is likely that GABAergic SPNs innervate other SPNs (Voigt et al., 2001; Hanganu et al., 2002). However, we did not observe any IPSCs after MGN stimulation, even when holding SPNs at depolarized potentials. Thus, GABAergic SPNs might not be strongly driven by MGN inputs, but might be driven by input from other sources (Fig. 7).
SPN axons project to and ramify in cortical layer 4 (Friauf et al., 1990; Allendoerfer and Shatz, 1994; Piñon et al., 2009), but a functional connection between SPNs and layer 4 had not been directly demonstrated. Since the large majority of SPN that project to layer 4 are found to be glutamatergic (Finney et al., 1998), an excitatory connection had been hypothesized. Our results here show directly that SPNs provide strong excitatory inputs to layer 4 neurons throughout the postnatal period. While in many cortical regions in the adult layer 6 neurons project to layer 4 (Douglas and Martin, 2004), our data suggests that such projections might be weak or sparse at early ages in the auditory cortex. This is consistent with data in ferret visual cortex showing a late development of the axonal projection from layer 6 to layer 4 (Callaway and Lieber, 1996). In addition, recent results suggest that in auditory cortex the strength of layer 6 to layer 4 projections might be low (Barbour and Callaway, 2008), consistent with our findings. Our observation of projections from layer 2/3 to layer 4 in ACX confirms recent observation in older animals using similar techniques (Barbour and Callaway, 2008).
In the adult, remaining subplate neurons (layer 6b) provide input to cortical layer 6 (Torres-Reveron and Friedlander, 2007). This is consistent with our data showing that at older ages subplate activity might drive cortical circuits that excite layer 6 neurons (Fig. 3). Together, our data indicates that the developing cortical circuit operates in a feedforward and not recurrent manner. This is consistent with results in the developing barrel cortex (Bureau et al., 2004) showing early columnar pattern of connections. The presence of long latency activity in layer 6 following MGN stimulation is also consistent with the view that SPNs act as a generator or amplifier of cortical oscillatory activity (Dupont et al., 2006; Hanganu et al., 2009).
Our LSPS data show that layer 4 neurons receive SPN input from locations in the superficial subplate. Many of these superficial SPNs survive and form layer 6B (Torres-Reveron and Friedlander, 2007) while many of the deeper SPNs disappear, except for some sometimes termed “white matter neurons” (Torres-Reveron and Friedlander, 2007). Thus, the subplate may contain multiple circuits of which one projects to layer 4. Recent advances in molecularly typing SPNs (Hoerder-Suabedissen et al., 2009; Osheroff and Hatten, 2009) might be a tool to allow the functional dissection of these different circuits.
Cell culture studies had suggested that GABAergic SPNs play a role in organizing cortical activity (Voigt et al., 2001), but we did not find evidence of GABAergic inputs from SPNs to layer 4 in our recordings. Thus, GABAergic SPNs might locally organize the activity of excitatory SPNs, or project outside of layer 4. We did not observe MGN-driven inhibition in SPNs, suggesting that MGN inputs do not drive GABAergic SPNs. Since we did not identify the transmitter phenotype of the recorded SPNs it is possible that some recorded SPNs are GABAergic. In fact we occasionally recorded SPNs that did not receive MGN input. While the lack of MGN input could be due to slicing artifacts, it is also possible that we recorded from GABAergic SPN that did not receive MGN inputs. However, together with the absence of MGN-driven inhibition in SPNs, it can be proposed that there are distinct parallel circuits within the subplate, one providing feedforward excitatory connections as studied here, and another containing GABAergic neurons.
The excitatory projection from SPNs to layer 4 neurons forms a substrate whereby SPNs can relay or influence early spontaneous activity in cortex (Dupont et al., 2006). While SPNs can also be coupled to other SPNs via electrical synapses (Dupont et al., 2006), we did not see evidence of electrical coupling to layer 4 neurons in experiments in which spiking was blocked by TTX (Fig. 5F). Gap junction coupling would have been revealed by short-latency, TTX-resistant depolarization or inward current, which we did not observe. However, it is possible that electrical synapses between SPNs and other cortical neurons outside of layer 4 exist. It might also be the case that gap junction coupling might exert a local effect on dendritic processing that we cannot observe via somatic whole-cell recordings from layer 4 neurons due to insufficient voltage clamp. Figure 7 summarizes our current view of the SPN associated circuit in A1.
Together, results here show directly that SPNs provide strong excitatory inputs to layer 4 neurons throughout the postnatal period. This excitatory projection can interact with learning rules present at layer 4 synapses to enable strengthening of thalamocortical synapses and plasticity during the critical period (Kanold et al., 2003; Kanold and Shatz, 2006). For example, by depolarizing layer 4 neurons, SPNs can enhance mechanisms of heterosynaptic LTP at the thalamocortical synapse or increase correlations between thalamic and cortical activity so that the thalamocortical synapse can be strengthened via spike-timing-dependent plasticity mechanisms (Kanold and Shatz, 2006).
Our results show that SPNs are tightly integrated in the developing thalamocortical circuit. SPNs receive and relay thalamic information to the developing cortical layer 4. Via these feedforward excitatory connections, SPNs may control activity levels and patterns in layer 4 that ultimately lead to the functional maturation of layer 4. Loss of this excitatory feedforward circuit caused by injury in development (McQuillen and Ferriero, 2005) could prevent activity-dependent maturational processes and leave layer 4 in an immature state, as has been observed experimentally following SPN ablation (Kanold et al., 2003; Kanold and Shatz, 2006). Thus, SPNs form a key circuit during cortical development, tightly linking the thalamus to the developing cortical layer 4 and enabling thalamocortical development.
P.O.K. is supported by National Institutes of Health (NIH) Grant R01DC009607, and by the International Cerebral Palsy Research foundation; J.P.Y.K. is supported by NIH Grant R01 GM056481. We thank Dr. Karel Svoboda (Howard Hughes Medical Institute) for assistance in implementing LSPS and Julie Zemskova and Aminah Sheikh for histological assistance.
- Correspondence should be addressed to Patrick O. Kanold at the above address.