In the present study, we showed how a simple paradigm of conditional learning influences the excitatory inputs and inhibitory outputs of three molecularly distinct GABAergic interneurons in the barrel cortex. Our key findings reveal that the pairing of tactile stimulation of whiskers with a tail shock enhances inhibitory outputs of SST-INs and PV-INs but not VIP-INs onto L4 excitatory neurons in the cortical representations of one row of whiskers stimulated during the conditioning protocol. On the other hand, this form of learning has no effect on global excitatory inputs to these three types of GABAergic interneurons in L4 of the barrel cortex. In conclusion, in this case, conditional learning controls the outputs but not the inputs of L4 GABAergic interneurons. Surprisingly, after pseudoconditioning, we observed an increased decay time of sEPSCs in PVINs, indicating that this form of learning may alter the excitatory drive to PVINs.
Previous studies based on the learning model used in our work have shown that conditioning increases the density of double positive cells for SST and glutamate decarboxylase 67 (GAD67) in barrels corresponding to the manipulated whiskers (Cybulska-Klosowicz et al. 2013), suggesting increased activity of SST-INs after conditioning. Recent studies have revealed that L4 SST-IN activity in the barrel cortex is crucial for the learning process and formation of association between CS and UCS in mice (Dobrzanski et al. 2022). Also, our recent study has shown specific learning-induced plastic changes in intrinsic excitability of all three molecular classes of GABAergic interneurons in the barrel cortex (Kanigowski and Urban-Ciecko 2024). We have shown that conditioning increases intrinsic excitability of SST-LTS, whereas pseudoconditioning decreases intrinsic excitability of PV-INs and VIP-AC (Kanigowski and Urban-Ciecko 2024).
Our current optogenetic experiments showed that conditioning increases the inhibition of L4 excitatory neurons by SST-INs and PV-INs in the barrel cortex. The increased inhibition of L4 excitatory cells by SST-INs may be caused by the strengthening of synapses between these cells. However, our previous experiments have shown that conditioning increases intrinsic excitability of SST-LTS (Kanigowski and Urban-Ciecko 2024). Thus, elevated inhibition of L4 excitatory neurons by SST-INs may also be related to increased intrinsic excitability of SST-LTS.
In terms of PV-INs, previous studies have not found any plastic changes in L4 PV-INs of the barrel cortex in the conditional learning model used in this study (Siucinska and Kossut 2006; Tokarski et al. 2007; Bekisz et al. 2010; Kanigowski and Urban-Ciecko 2024). However, our present optogenetic experiments revealed an increase in the inhibition of L4 excitatory neurons by PV-INs following conditioning. Since our previous study has not shown any changes in intrinsic excitability of L4 PV-INs after conditioning (Kanigowski and Urban-Ciecko, 2024), changes in intrinsic excitability of PV-INs are unlikely responsible for stronger inhibition of L4 excitatory cells by PV-INs. So, it seems that increased inhibition of L4 excitatory neurons by PV-INs might depend on changes in the strength of inhibitory synapses between PV-INs and excitatory cells.
For the first time, we demonstrated that L4 excitatory neurons in the barrel cortex are innervated by VIP-INs. Consistent with data obtained from L2/3 and L5 of the medial prefrontal cortex, auditory cortex, and visual cortex (Pfeffer et al. 2013; Pi et al. 2013), we showed that also in the barrel cortex, L4 excitatory neurons are sparsely innervated by VIP-INs and the strength of these connections is weak. For these reasons, only the percentage of L4 excitatory cells of the barrel cortex innervated by VIP-INs was analyzed. However, this analysis did not show any statistically significant differences between the groups of animals in the fraction of L4 excitatory cells innervated by VIP-INs. Considering that VIP-INs primarily establish disinhibitory circuits and connect with SST-INs and PV-INs in various brain regions (Caputi et al. 2009; Lee et al. 2013; Pfeffer et al. 2013; Jiang et al. 2015; Walker et al. 2016; Kullander and Topolnik 2021), it would be valuable to investigate whether similar disinhibitory connection patterns are preserved in L4 of the barrel cortex and whether learning influences the strength of these connections. However, it would require a double transgenic mouse line with a possibility for the manipulation or visualization of two district classes of GABAergic interneurons simultaneously.
Moreover, our current experiments did not reveal any changes in the amplitude, frequency, or kinetic of sEPSCs in L4 SST-LTS or both electrophysiological types of VIP-INs as a result of conditioning or pseudoconditioning. We observed an increase in the decay time of sEPSCs in PV-INs after pseudoconditioning, with no changes in the amplitude or frequency of these currents. The increased decay time of sEPSCs in PV-INs after pseudoconditioning might indicate changes in the kinetics of AMPA receptors on PV-IN synapses, more specifically, their prolonged deactivation time (Wall et al. 2002). Prolonged decay time of sEPSCs may lead to reduced firing precision and information transmission between cells (Rodriguez-Molina et al. 2007). This would be consistent with the assumption that pseudoconditioning might be the opposition to associative learning and may lead to a deterioration in the functioning of the local network (Kanigowski and Urban-Ciecko 2024). At the same time, we did not observe any changes in the parameters of sEPSCs in PV-INs after conditioning, consistently as it has been observed for FS interneurons (presumably PV-INs) in the previous study (Tokarski et al. 2007).
Our results suggest that associative conditioning does not influence the excitatory connections of any class of L4 GABAergic interneurons in the barrel cortex. However, it is also possible that potential changes in excitatory inputs are unique and specific to a given connection type and that these changes are invisible in the global spontaneous activity.
Finally, our research did not reveal any changes in the spontaneous firing of all studied types of L4 GABAergic interneurons after conditioning or pseudoconditioning.
What is the impact of observed plastic changes on the local network? It has been shown that conditioning leads to increased intrinsic excitability of L4 excitatory neurons and elevated frequency of spikes recorded at the threshold potential (Bekisz et al. 2010). Likely, the increase in the inhibition of L4 excitatory cells by SST-INs and PV-INs after conditioning observed in our study is a response to elevated activity of glutamatergic cells. Thus, changes in synaptic inputs following conditioning would be a form of homeostatic plasticity maintaining a balance between excitation and inhibition (E/I balance) in the L4 local network of the barrel cortex and preventing excitatory neurons from excessive activity, which would otherwise result in epileptic activity (Le Roux et al. 2008; Turrigiano 2011). Increased inhibition of excitatory neurons by SST-INs and PV-INs may regulate glutamatergic neuron inputs and outputs and provide a balance that is necessary for synaptic plasticity and proper learning (Gandal et al. 2012; Campanac et al. 2013; Ntim et al. 2020; Toader et al. 2020). Many studies on mouse models of nervous system disorders have shown a disturbed E/I balance leading to learning disabilities of these animals (Costa et al. 2002; Souchet et al. 2014; Haji et al. 2020). Furthermore, transcranial magnetic stimulation in rats can modulate the ratio between excitation and inhibition in the cortex, which influences the responses of L4 neurons in the barrel cortex to vibrissae manipulation (Thimm and Funke 2015). Such magnetic stimulation of the cortex can also positively influence associative learning related to tactile stimuli recognized by the vibrissae (Mix et al. 2010). The abovementioned examples indicate that the E/I balance in L4 of the barrel cortex is fundamental for associative learning. Considering the ratio of SSTINs to PV-INs in the cortex, especially in L4, and the frequency and strength of connections formed between these interneurons and glutamatergic cells, the maintenance of the balance between excitation and inhibition would be primarily provided by PV-INs (Beierlein et al. 2003; Pfeffer et al. 2013; Tremblay et al. 2016). Also, stronger inhibition of L4 excitatory neurons by PV-INs than SST-INs was observed in our study (Fig. 1, 2).
The observed enhanced inhibition mediated by SST-INs and PV-INs may also increase the accuracy of information transfer between L4 glutamatergic cells in the barrel cortex cells by increasing the time accuracy of their discharges (Wlodarczyk et al. 2013). Moreover, increased inhibition from GABAergic interneurons may theoretically influence the accuracy of encoding information about the tactile stimulus sensed by the whiskers (Yu et al. 2016). It has been shown that the complexity of a somatosensory stimulus is encoded by the firing pattern of the cell population (Arabzadeh et al. 2006). Other analyses also have revealed that the timing of individual AP discharges potentially has a high information encoding capacity (Petersen et al. 2002). Thus, AP timing carries more information about texture of the surface than spike rates do. Also, AP timing better predicts animal behavior in a texture discrimination task (Zuo et al. 2015). Increased temporal precision of excitatory neuron discharges may positively impact on sensory information processing and encoding. Interestingly, as a result of conditioning, a simultaneous increase in inhibition from both SST-INs and PV-INs was observed. Due to their location in the network, both classes of interneurons are involved at different levels of information encoding. The increase in IPSCs originating from PV-INs may have dual effects, affecting both the initial inhibition of signals reaching L4 from the thalamus (due to the feed-forward inhibition) and the subsequent stages of sensory information processing in this layer by the feed-back inhibition (Beierlein et al. 2003; Gabernet et al. 2005; Inoue and Imoto 2006; Cruikshank et al. 2010; Koelbl et al. 2015; Sermet et al. 2019). Elevated inhibition of L4 excitatory neurons by PVINs increases the precision of the L4 glutamatergic cell responses to excitation coming from the thalamus. In a subsequent stage, SST-INs which receive only negligible excitation from thalamocortical axons and are activated by L4 glutamatergic neurons provide delayed feedback inhibition (Beierlein et al. 2003; Cruikshank et al. 2010; Ma et al. 2012; Sermet et al. 2019) and thus minimizing recurrent activity (M. L. Beierlein et al., 2002; Kapfer et al., 2007). Considering that SST-INs innervate distal dendrites and PV-INs innervate cell bodies and proximal dendrites of glutamatergic neurons, a simultaneous increase in inhibition from both classes of interneurons should enhance the control of information both at the input and output of excitatory cells. This two-step control should contribute to greater accuracy of correlated discharges of excitatory cells and lead to higher precision of information exiting from L4 to layers 2/3.