Male Sprague–Dawley rats (aged ~7 weeks, weight 250 g) were obtained from Monash Animal Services (MAS) and littermates randomly assigned to either Isolated (Isol.; n = 13) or Enriched (EE; n = 13) housing conditions, always in a 12 h light/dark cycle with ad libitum food and water, for 8–10 weeks; from week 8 onwards, animals were removed for terminal electrophysiological experiments conducted over a 2 week period to ensure that there was no confound from ageing. All experiments were conducted in accordance with guidelines from the National Health and Medical Research Council and received approval from the Monash University Standing Committee on Ethics in Animal Experimentation.

Rats in EE housing were housed in groups of 3 in a large (69 × 60 × 270 cm) run which featured a front-facing Plexiglas wall and 3 steel walls with a steel base and a wire mesh cage lid. The cage floor was covered with wood shavings and contained numerous plastic and metal toys and objects of various colors, sizes, and textures. The objects used for enrichment were kept constant but their locations were re-arranged every 2.5 days. The Isolated housing rats were housed individually in standard sized cages (31× 24 × 45 cm) with wood shavings and shredded paper, but no toys.

Cortical responses from animals in the two housing conditions were recorded 8–10 weeks after commencement of housing in the test condition. Extracellular recordings were taken from the posteromedial barrel subfield region of somatosensory cortex (PMBSF; the so-called barrel cortex), as detailed elsewhere (Rajan et al., 2006, 2007; Alwis et al., 2012). Briefly, animals were anesthetized with 5% halothane (Sigma Aldrich, USA) mixed in O₂ (O₂ flow rate = 1 ml/min) and tracheotomised for mechanical ventilation (2.5–3.5 mL tidal volume, 72–80 breaths/min; both dependent on animal size) to maintain anesthesia thereafter with 0.5–2% halothane mixed in O₂ (O₂ flow rate = 0.3 ml/min). A thermostatically-controlled heat pad with feedback from a rectal probe (Fine Science Tools Inc., U.S.A) was used to maintain body temperature at 37–38°C.

Once anesthesia was established, as determined from absent palpebral reflexes or responses to strong forepaw pinching, the skull was exposed and anchored to a head bar with a screw and dental cement. Then a section of skull ~5 mm in diameter, located above barrel cortex (~2 mm caudal of bregma and 6 mm lateral of the midline), was removed and the exposed cortex (dura intact) covered with silicone oil. Recordings were obtained using a parylene-coated tungsten microelectrode (2–4 MΩ resistance; FHC, ME, U.S.A) which was moved using a fast-stepping micro-drive (Kopf Instruments, California, U.S.A). Initially, the electrode was advanced to a depth between 600 and 800 μm from the surface to allow determination of the Principal Whisker (PW) by manual deflection of the whiskers using a hand held probe. Here the PW was classified as the whisker which produced the greatest neuronal firing response to manual whisker deflection (with electrode output monitored aurally through speakers as well as on an oscilloscope screen) and, in barrel cortex, was always unequivocally identifiable at this depth. Where drive was weak, or a result of multi-whisker activity, the electrode was retracted and a new penetration was made until a single PW could be identified; the electrode output was monitored (see below) after amplification and filtering, on an oscilloscope and through speakers. If strong PW drive was obtained (see Alwis et al., 2012 for details) the electrode was then retracted to the cortical surface under visual control and zeroed here. Then it was advanced systematically to record from neurons at a number of different depths, using stimuli delivered under computer control to the PW. The first recording was made at a depth of about 150 μm from the surface; thereafter recordings were made at regular intervals to ensure data was collected from all animals from the following layers: Layer 2 (150–300 μm); Upper Layer 3 (350–500 μm); Deep Layer 3 (550–700 μm); Layer 4 (750–1000 μm); and Layer 5 (1100–1400 μm). To ensure that data representation from deeper layers was not compromised by the later recordings perforce of this strategy, in some cases we advanced the electrode first to Layer 4 and collected data from that layer and then from Layer 5 before retracting back to the surface to advance systematically as above to obtain data from the other layers. We have previously demonstrated that recorded responses in halothane anesthetized animals, using the same experimental techniques outlined in the present study and in our previous work (see Rajan et al., 2006; Alwis et al., 2012; Johnstone et al., 2013), were comparable to those seen in awake animals, where the pattern of temporal and spatial responses to whisker deflection were similar in both the anesthetized and awake states (Rajan et al., 2006; Maravall et al., 2007).

Neural signals from the microelectrode were treated as described previously (see Rajan et al., 2006, 2007; Alwis et al., 2012 for details), being amplified, band-pass filtered (0.3–10 kHz) and displayed on an oscilloscope, and through speakers for aural monitoring of neural activity. A Schmitt trigger box was used to set a voltage trigger level, while monitoring on the oscilloscope at a level 2× noise level, for neuronal cluster activity recordings. Trigger crossings generated digital pulses that were fed to a PC; this computer used Spike 2 software (CED, UK) for stimulus generation and/or delivery (Alwis et al., 2012) and stimulus trigger events were stored along with the Schmitt trigger pulses. The Spike2 software was also used to generate on-line displays of rasters of spike occurrences and peristimulus time histograms (PSTHs). A minimally-filtered copy of the signals recorded by the electrode was also stored for any offline analysis if needed later (Rajan et al., 2006).

For quantitative data recording, the PW was threaded through a hole at the end of a motor-controlled lever arm system positioned 5 mm from the mystacial pad to deflect the PW in computer-controlled patterns. At each recording site, voltage triggers were set at a level 2× noise level to obtain responses from neuronal clusters (of 4–6 neurons, as determined by online spike sorting using Spike 2 software). Responses were first characterized for laminar location using a suite of 3 trapezoidal stimuli, where only the onset ramp velocity was varied (60, 150, 400 mm/s; Alwis et al., 2012). This suite of trapezoids was presented for 150–300 repetitions in a pseudo-random manner.

Then a series of four complex “naturalistic” whisker deflections, obtained from studies in which whisker motion was recorded in awake behaving rats, were played out from text files which stored stimulus characteristics. These stimuli have been described in detail in our previous studies (Alwis et al., 2012; Johnstone et al., 2013); the four waveforms were those seen in whisker motion across a smooth and rough surface (Ritt et al., 2008); when rats made contact with a rod placed in the path of the whiskers (Hartmann et al., 2003); and in head-fixed rats that were engaging in “free” whisking (Gao et al., 2001). Details of the acquisition and conversion of these waveforms to whisker deflection patterns have been previously described by our group (Alwis et al., 2012; Johnstone et al., 2013). Each suite consisted of 10 stimulus amplitudes, from the lowest amplitude of 0.2 mm, and then continuing from 0.4 to 3.6 mm in 0.4 mm steps. Each suite of 10 stimulus amplitude was presented 50 times, in pseudo-random order across successive presentations.

Cluster responses were segregated into lamina by depth (from the cortical surface) as noted above: Layer 2 (150–300 μm); Upper Layer 3 (350–500 μm); Deep Layer 3 (550–700 μm); Layer 4 (750–1000 μm); and Layer 5 (1100–1400 μm). Data were represented as firing rate (in spikes/s) in 1 ms bins over the period from 200 ms prior to stimulus onset until 100 ms post stimulus offset. The data collected in the 200 ms pre-stimulus bin was used to calculate spontaneous firing rates, which were subtracted from responses during the rest of the data collection period to correct for the spontaneous firing rate. The spontaneous activity-corrected firing rates were used for all subsequent analyses.

Offline analysis was conducted to generate population peri-stimulus time histograms (PSTHs) showing the pattern of population responses within a lamina, in animals housed in either Enrichment or Isolation. PSTHs were obtained by averaging cluster responses within a lamina across each presentation of a stimulus. A 5-point weighted moving average was then applied to the data to smooth out noise and a grand PSTH was produced by averaging the data across all multi-units.

For quantitative analysis for each stimulus, we used only data from multi-neuronal clusters considered to be responsive to that stimulus. Clusters were classified as responsive if their response rates were significantly greater (more than 1.4 SD >) than spontaneous firing rates at more than two consecutive stimulus velocities (for the trapezoidal stimuli) or at more than two consecutive stimulus amplitudes (for the four naturalistic stimuli). We then extracted the following metrics for each stimulus: peak firing rate, area under the curve, latency to peak and half-peak width. These calculations were done using specific counting windows for each stimulus: for the trapezoidal and object contact stimuli, a counting window from 5 to 50 ms after stimulus onset was used, for the surface texture discrimination stimuli, a 5–30 ms counting window, while the free whisking stimulus was analysed using a 5–200 ms counting window. Almost identical effects were seen for the firing rate measures (peak firing rate and area under the curve), and almost identical effects for the timing measures (latency to peak and half-peak width) and hence, we present data only from the peak response (PFR) and the latency to the PFR (L_PFR) for each stimulus in its designated analysis window.

Statistical analysis was carried out using two-way repeated measures ANOVAs to determine laminar-specific differences, with peak firing rates or latencies being the dependent variable and the independent between-animal factors being housing condition and stimulus amplitude/velocity. When the ANOVA revealed a significant main factor effect of Group (housing condition) or Stimulus parameter (Amplitude/Velocity) or a significant interaction term between these two factors, post-hoc Bonferroni tests were used to where the differences lay.

Electrophysiological recordings were obtained from 13 rats housed in Isolation and 13 rats in EE housing, from multi-unit clusters in layers 2 (L2), Upper layer 3 (U3), Deep layer 3 (D3), Layer 4 (L4), and Layer 5 (L5), in response to simple and complex, “naturalistic” whisker motion patterns. Only data from responsive multi-neuronal clusters (see Materials and Methods for definition) were extracted for analysis. We present below, for each stimulus, laminar-specific data on the response patterns of clusters in the two housing conditions, and then data for two major metrics of the responses.

To visualize the laminar-specific response pattern of neuronal clusters to any specific stimulus, Grand Peri-Stimulus Time Histograms (GPSTHs), which are defined as histograms of firing rate against time from stimulus onset, were generated. For this, responses of each cluster during and after the stimulus period were corrected for the spontaneous activity measured in that cluster in the 200 ms period prior to stimulus onset. Then, for rats in each housing condition, responses from all responsive clusters in a specific lamina were averaged to generate the GPSTH of firing rate against stimulus period for that lamina. Finally, to quantify the effects, analysis windows were defined for each stimulus in relation to aspects of that stimulus (e.g., the onset period, or the post-stimulus offset period; defined in Materials and Methods for each stimulus), and for these windows we present two representative quantitative metrics from all responsive clusters: (1) peak excitatory firing rate (PFR) and (2) latency to the peak (L_PFR). The PFR was the across-clusters average peak firing rate during the stimulus period as seen in the population GPSTH and expressed as spikes/s, and the L_PFR was the time from stimulus onset to this peak in the GPSTH. Note that other firing rate and timing values were also calculated (see Materials and Methods) but the effects were the same across all firing rate measures or across all timing measures and hence only PFR and L_PFR are presented here.

EE is generally reported to increase sensory cortical responsiveness in sensory cortex studies restricted to recordings from some layers. Mainardi et al. (2010) found that evoked potentials were increased in layers 3/4 of visual cortex of rats exposed to EE and, at the single neuron level, enrichment resulted in no change in RF size or spontaneous activity (Beaulieu and Cynader, 1990a) but increased responsiveness, orientation tuning, and temporal contrast tuning (Beaulieu and Cynader, 1990b, layers unknown). In primary auditory cortex (A1), 8 weeks of auditory enrichment resulted in increased stimulus-evoked neuronal response rates, spontaneous rate, and response latency, in layer IV (Engineer et al., 2004), increased auditory evoked potential amplitudes (Engineer et al., 2004; Percaccio et al., 2005, 2007) and decreased ability to follow high stimulus repetition rates (Percaccio et al., 2005, 2007). We too found increases in driven and spontaneous activity but no change in response latency. In other studies in adult auditory cortex, different types of enhanced auditory experience distorts the tonotopic map (Pienkowski and Eggermont, 2009; Zhou et al., 2011) and the ability to follow high stimulus rates (Zhou et al., 2011). However, there is dispute as to whether it decreases neuronal frequency selectivity at all frequencies (Zhou et al., 2011) or increases it within the frequency band of the enrichment stimulus (Pienkowski and Eggermont, 2009). In non-primary sensory cortex (posterior auditory field) EE increased response strength and decreased RF size, leading to an increase in spectral and temporal selectivity (Jakkamsetti et al., 2012). In contrast, Polley et al. (2004) reported that naturalistic sensory experience in their EE conditions decreased neuronal responses and RF size in barrel cortex, but only in the upper layers and not in layer 4. However, (Guic et al., 2008) suggested that the effects in the Polley et al. (2004) study reflected the effects of habituation even despite EE, and a lack of active exploration; Guic et al. (2008) found that EE caused an increase in RF size in contrast to Polley et al. (2004). Megevand et al. (2009) report increased greater whisker-driven evoked potentials in barrel cortex after EE.

A variety of mechanisms may contribute to cortical plasticity after EE (reviews by Van Praag et al., 2000; Nithianantharajah and Hannan, 2006), including structural and molecular changes like changes in dendritic structure and function and enhanced synaptic plasticity (Yang et al., 2009; Fu and Zuo, 2011; Jung and Herms, 2012), increases in certain neurotransmitters (Giovannini et al., 2001; Naka et al., 2002) or in synaptic release probability (Percaccio et al., 2005), and shifts in cortical E/I ratios (Coq and Xerri, 1998; Engineer et al., 2004; Polley et al., 2004; Baroncelli et al., 2011), given that a decrease in inhibition can trigger plasticity, even in non-critical periods (Maya Vetencourt et al., 2008; Southwell et al., 2010; Maya-Vetencourt et al., 2012), through changes in neuronal connectivity (Baroncelli et al., 2010, 2012). Our results can most parsimoniously be explained by this last factor (not excluding that the other mechanisms may also contribute to this factor)—that EE-induced plasticity in our case was due to a shift in the E/I balance to favor E. The stability of neuronal connections in the adult brain is held to be due to maturation of cortical inhibitory interneurons which results in decreased plasticity (Fagiolini and Hensch, 2000; Hensch and Fagiolini, 2005). Induction of adult cortical plasticity appears then to be brought about primarily through a decrease in cortical inhibitory activity (Hensch and Fagiolini, 2005; Sale et al., 2007; Benali et al., 2008; Baroncelli et al., 2010, 2011; Luz and Shamir, 2012). Consistent with this idea, EE causes a reduction in the number of visual cortical neurons expressing GAD67 (Scali et al., 2012; Tognini et al., 2012), and a decrease in extracellular basal GABA levels, coupled with a restoration of white matter long-term potentiation (WM-LTP), suggesting decreased cortical inhibition (Sale et al., 2007). Decreases in GABAergic inhibition after EE, through a reduction in expression of GABA_A receptor subunits, have been demonstrated in adult rat auditory cortex (Zhou et al., 2011), and in cat visual cortex enrichment decreases the number of inhibitory synapses (Beaulieu and Colonnier, 1987). Treatment with IGF-1, a peptide thought to play a role in EE-induced plasticity by enhancing synaptic plasticity (Torres-Aleman, 1999; Aberg et al., 2000), causes a decrease in basal inhibitory neurotransmitter levels with no overall changes in basal glutamate levels (Maya-Vetencourt et al., 2012) and a shift in ocular dominance as measured by visual evoked potentials, indicating visual cortical plasticity in adult rats that had undergone monocular deprivation (Maya-Vetencourt et al., 2012).

The hypothesis that EE simply decreases cortical inhibition is not universally accepted: Nichols et al. (2007) found that GABA_A receptor-mediated inhibitory postsynaptic currents do not change in supragranular auditory cortex after EE but there was an increase in AMPA-mediated current amplitudes. Further, the group had previously shown that EE results in a decreased ability of AI neurons to follow rapid stimuli (Percaccio et al., 2005), inconsistent with effects predicted by a decrease in inhibition. Similarly, Zhou et al. (2011) reported that enhanced auditory experience decreased expression of GABA_A and NMDA receptors, but had no effect on the AMPA glutamate receptor.

Our extracellular recordings do not allow direct measurements of inhibition but the large increase in driven and spontaneous response strength strongly supports the inference that EE in our study also caused a shift in the cortical E/I ratio to favor excitation over inhibition. Sensory deprivation also induces cortical plasticity through a decrease in inhibition (Welker et al., 1989; Buonomano and Merzenich, 1998). Kelly et al. (1999) reported that whisker trimming leads to a loss of inhibitory input from surrounding whiskers, as well as a net decrease in tonic cortical inhibition, resulting in increased barrel cortex responses to the PW. In the mature barrel cortex, deprivation-induced plasticity decreases L4 feed-forward excitation to L2/3 inhibitory neurons but with improved inhibition to L2/3 pyramidal cells (House et al., 2011), resulting in E/I balance being maintained in L2/3. We found increased activity in all layers, indicating that an E/I balance was not maintained in L2/3. Thus, EE, in our case, appears to exert global cross-lamina effects primarily, if not solely, directed to favoring excitatory over inhibitory inputs. This does not mean that large changes in excitation need have occurred. It has been elegantly demonstrated that during normal processing in somatosensory cortex, excitation and inhibition are highly synchronized and well-correlated in time and strength, in a continuous manner (Okun and Lampl, 2008); thus, even small changes in timing and strength of the two processes could result in a very large increase in excitation as we see (Isaacson and Scanziani, 2011).

Only certain forms of inhibition may be affected after exposure to EE since, even in our extracellular recordings, we do see some indications of inhibition after EE, with decreases in post-stimulus responses to levels below pre-stimulus spontaneous activity (see Figure 4A). Jakkamsetti et al. (2012) suggest that EE affects basal levels of GABA and evoked GABA differently, such that stimulus-evoked inhibition remains intact, while basal inhibitory activity is suppressed after EE, which could explain our evidence of intact post-stimulus inhibition but increased spontaneous activity.

Cortical inhibition is heterogeneous, either acting to affect neuronal responses (Wehr and Zador, 2003; Haider et al., 2013) by increasing response selectivity and hence, RF sizes, or altering response gain (Isaacson and Scanziani, 2011; Wilson et al., 2012), including acting through “silent” shunting inhibition (Isaacson and Scanziani, 2011). Selective loss of some forms of cortical inhibition may be all that is required to obtain the EE-induced increases in cortical spontaneous and driven responses. In auditory cortex, loss of surround inhibition can occur without loss of stimulus-driven within-field inhibition (Rajan, 1998, 2001)and then results in no change cortical in map topography (Rajan, 1998) but an increase in bandwidth of cortical neurons (Rajan, 2001), and thereby likely an increase in the overall representation of any specific frequency and increase response rates to the “best” frequency (Rajan, 1998). This parallels the observations that that post-stimulus inhibition is preserved in EE animals (our present study), and that EE does not change the exclusive representation area of a whisker in barrel cortex though there is an expansion of the total representation area of that whisker (Guic et al., 2008).

Our finding of a global increase in neuronal responsiveness in barrel cortex after EE without response timing changes suggests an intra-cortical locus for plasticity (cf. Fox et al., 2002; Polley et al., 2004), especially given this pattern of effects even in the thalamo-recipient Layer 4 responses. Long-term whisker deprivation increases the functionality of specific GABA_A receptors on Layer 4 cells such that there is faster decay of inhibition in these cells (Li et al., 2009). In this case, a normal thalamic input could produce larger responses in Layer 4 as we find with EE. Our findings of an inverse relationship between cortical depth and the amount of EE induced increase in responses can be explained by postulating that a Layer 4 change is amplified by local mechanisms when information flows from Layer 4 to upper layers (Armstrong-James et al., 1992; Jellema et al., 2004; Lubke and Feldmeyer, 2007; Megevand et al., 2009), as has been suggested for experience-dependent plasticity in barrel cortex (Fox et al., 2002). This is consistent with the fact that supra-granular cortical layers show longer-lasting and greater amounts of plasticity than layers 4 or 5 (Glazewski and Fox, 1996; Wallace and Fox, 1999; Fox et al., 2002; Nichols et al., 2007). It is also consistent with the postulate (Nichols et al., 2007) that layer 2/3 is a “privileged substrate” for consolidating EE-induced cortical plasticity, possibly through spike-timing-dependent plasticity, or even structural plasticity, with studies reporting an EE-induced increase in dendritic branching in supragranular layers of the occipital cortex (Volkmar and Greenough, 1972; Greenough et al., 1973). A similar depth-dependent effect has been seen for plasticity evoked by rhythmic whisker stimulation in anesthetized animals, a paradigm suggested to engage the same intra-cortical plasticity mechanisms as EE (Megevand et al., 2009).

The absence of changes in response timing, as would be expected for plasticity evoked sub-cortically, is consistent with studies showing that thalamic plasticity is limited in the adult brain (Fox et al., 1996, 2002), except in the case of peripheral nerve injury (Li et al., 1995; Jones and Pons, 1998). However, it must be recognized that the debate on sub-cortical contributions to cortical changes in EE is not yet resolved. EE is capable of enhancing thalamocortical transmission in adult rat visual cortex (Mainardi et al., 2010), while EE in an experimental model of adult monocular amblyopia causes an increase in presynaptic thalamo-cortical activity, hence increasing postsynaptic stimulation of the visual cortex (Tognini et al., 2012). In the somatosensory cortex, whisker deprivation increases neuronal response rates in cortical Layer 4, with no changes in response rates in thalamic VPM (Wallace and Fox, 1999). In the auditory system it has been argued that cortical changes induced by enriched auditory experience are likely to occur in thalamus but be modified by local mechanisms in cortex (Pienkowski and Eggermont, 2009).

We are currently conducting further studies to investigate the exact mechanisms underlying the EE-induced changes in neuronal plasticity reported in the present study, using slice preparations and immunohistochemical techniques.

Our results show a stimulus amplitude-dependent increase in peak firing rates, where for all stimuli (except one) EE induced greater increases in excitatory responses at lower stimulus amplitudes than at higher stimulus amplitudes. The effect may simply reflect response saturation at higher levels; a multiplier effect will produce a larger change at lower stimulus amplitudes where responses are not saturated than it would at higher levels where responses are closer to saturation. This is consistent with the observation of Engineer et al. (2004) that stimuli that elicited smaller responses which did not reach saturation were more suitable to reveal whether EE has an effect on auditory cortical activity.

The odd one out of the stimuli in this inverse relationship between EE-induced increases and stimulus parameter (amplitude or velocity) was the stimulus waveform mimicking whisker motions seen in exploratory, free-whisking behavior. For this stimulus, the EE effect was non-monotonic with greatest effects at intermediate amplitudes than at higher amplitudes where EE effects either plateaued or even decreased in relative potency compared to the EE effects at intermediate amplitudes. EE influences exploratory behavior (Dell'omo et al., 2000; Frostig, 2006), thus making a stimulus that mimics exploratory whisker motion most relevant for investigation of EE effects. It is interesting to speculate that the maximization of EE effects to particular amplitudes of free, exploratory whisking may be linked to the effects of EE on exploratory behaviors—does EE promote rat behavior such that the animals are more likely to explore objects when the object is at a particular distance from the animal, i.e., does the animal tend to position itself so that it can best explore an object at a certain distance away? Szwed et al. (2006) have shown that spike rate is particularly important in encoding (in trigeminal ganglion neurons) radial distance of an object and whisker exploration of environmental objects and the EE-induced optimization of firing rate at particular amplitudes may be seen to support this hypothesis. However, against this is that an object a certain fixed distance from the animal will be at different radial distances along the length of whiskers arrayed across the rostro-caudal axis of the mystacial pad. Thus, the relevance of the EE-induced increase in firing rate at particular deflection amplitudes for this specific stimulus alone remains unknown.

One final point that merits consideration is that it is generally held that novel objects in EE housing should be changed regularly to prevent habituation. We did not do so but still found highly elevated response rates, especially in supragranular layers, whereas Polley et al. (2004) reported that EE led to decreased responses in upper layers of barrel cortex (but not in layer 4), and a decrease in RF sizes. As noted above, Guic et al. (2008) suggested that the Polley et al. (2004) results reflected the effects of habituation and a lack of active exploration even despite EE conditions. If this criticism is valid, it would indicate that, in our case, the constancy of the objects in the EE housing was not a deterrent to active exploration of those objects, insofar as such active exploration led to the observed increase in response rates to all stimuli. In part this may reflect the fact that, in our study, when cages were cleaned and food and water replenished, the experimenters made a point of bringing back to the surface all objects that had become submerged under the bedding material. It is possible that this re-appearance of the objects contributed to maintaining active exploration of the objects and thereby contributed to the effects observed in this study. Under such conditions the effects of EE appear to be remarkably well-expressed in barrel cortex, across all layers, stimulus types, response patterns, and response durations.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.