From the perspective of the HO thalamus, L5tt projections provide a strong but sparse drive to individual neurons (discussed in the next section). Where does this information go after being further processed in HO thalamus? (Sherman and Guillery, 2011) suggested a significant revision to our understanding of thalamocortical processing with the “transthalamic” hypothesis, specifically that cortical L5tt neurons in one source region send information to a secondary cortical recipient region via HO thalamus. Recent anatomical and functional data suggest that transthalamic pathways paralleling “direct” cortico-cortical pathways could be a common feature of the thalamocortical system and that these pathways carry distinct rather than redundant information (Sherman, 2016; Bennett et al., 2019; Mo and Sherman, 2019; Blot et al., 2021). However, there is recent evidence that certain HO circuits can also be involved in recurrent “closed-loop” networks, in which the same cortical region providing L5tt input to HO thalamus is itself reciprocally innervated by HO thalamus (Wimmer et al., 2010; Guo et al., 2020). This reciprocal connectivity motif is seen in FO CTC networks, where L6A modulatory CT projections originate from the same cortical region innervated by FO TC projections. However, the transthalamic and recurrent network motifs are not mutually exclusive and may subserve different functions. As we highlight below, based on region-specific descending CT and ascending HO TC connectivity patterns, different combinations of recurrent and transthalamic communication are possible. An important focus for future studies is to understand the distinct signals transmitted by transthalamic and recurrent pathways, and to uncover the computational scheme integrating these information channels with cortico-cortical signals.

Currently, it is understood that HO nuclei, which represent a majority of the thalamus by volume, participate in the generation of activity in distinct cortico-thalamo-cortical (CTC) networks (Guillery, 1995; Sherman and Guillery, 2011; Sherman, 2016). Additionally, these CTC networks serve to integrate a diverse range of cortical and subcortical signals (Groh et al., 2013; Bickford, 2015). While a full assessment of this literature is beyond the scope of this review, we note that these “convergence zones” have been reported across modalities (Groh et al., 2013; Bickford, 2015). For example, projections from the primary visual cortex in the HO lateral posterior (LP) nucleus overlap with terminals originating in the superior colliculus (Li et al., 2003c; Masterson et al., 2009). Elsewhere, Bosch-Bouju et al. (2013) introduce the concept of a “super-integrator,” where motor regions of the thalamus integrate information from the cortex with information from the basal ganglia and cerebellum. Additional convergence zones have been identified by exploiting the differential distribution of type 1 and type 2 vesicular glutamate transporters, which are specific to cortical and subcortical inputs, respectively (Rovó et al., 2012; Bickford, 2015). Recent anatomical evidence points to a role for HO thalamus in integrating information from different cortical regions (Prasad et al., 2020), even at the level of single cells (Sampathkumar et al., 2021).

A central regulator of thalamic function is feedback inhibition via thalamic reticular nucleus (TRN), a thin layer of GABAergic neurons partially encapsulating the relay nuclei which project to cortex (Pinault, 2004). As well as TC afferents to cortex, thalamic relay neurons also send thalamoreticular projections to TRN which in turn provide feedback inhibition to relay neurons (Figure 1); the temporal scale of this inhibition is sensitive to spiking patterns (Figure 2), with high-frequency bursts triggering long-lasting IPSCs due to GABA “spillover” to extrasynaptic receptors, while tonic spiking patterns trigger shorter IPSCs (Halassa and Acsady, 2016). A recent pair of milestone studies in the somatosensory thalamus reveal that properties of HO and FO intrathalamic inhibitory circuitry differ significantly: HO nucleus POm excites and is inhibited by a discrete shell population of TRN neurons; furthermore, the synaptic dynamics of POm-TRN connections as well as the intrinsic properties of POm-connected TRN neurons are functionally distinct from those in VP-TRN circuits (Li et al., 2020; Martinez-Garcia et al., 2020). Thus, it may be that the dynamics of intrathalamic inhibition are matched to the distinct signal processing requirements of HO and FO circuits carrying L5tt and sensory information, respectively.

Given its role in gating thalamocortical transmission as well as its positional and physiological properties, the TRN has been implicated in the regulation of attention in the “searchlight hypothesis” (Crick, 1984; Crabtree, 2018). Regions in the TRN show increased activity in response to attentional stimuli, and the specific region in which this response is found is modality-dependent (McAlonan et al., 2000, 2006). Moreover, limbic TRN projections correlate with arousal states, while sensory TRN projections are suppressed by attentional states (Halassa et al., 2014). Work by Halassa et al. (2011) demonstrates TRN-dependent control of thalamocortical firing mode and state regulation, where selective drive of TRN causes a switch from tonic to burst firing and generates state-dependent neocortical spindles (Halassa et al., 2011).

Likewise, there is evidence for an attentional role of HO thalamus. For example, the MD is activated in humans during tasks requiring a rule-dependent shift in attentional allocation (i.e., set-shifting), such as the Wisconsin card-sorting task (Monchi et al., 2001; Halassa and Kastner, 2017). Human and monkey studies also point to a role of the pulvinar in visual attention. Pulvinar lesions in patients result in impairments in filtering distracting information, while pulvinar inactivation in monkey impairs spatial attention (Danziger et al., 2004; Snow et al., 2009; Wilke et al., 2010; Halassa and Kastner, 2017). In addition, Yu et al. (2008) describe the pulvinar’s role in sustained attention, employing the five-choice serial reaction time task to show that half of recorded units in this nucleus were attention-modulated (Yu et al., 2018). However, TRN control of HO thalamus in the context of attention and arousal has yet to be systematically investigated.

Thalamic reticular nucleus inhibition is also influenced by topographically organized CT projections from L6A which innervate both relay neurons and neurons in TRN (Lam and Sherman, 2010; Figure 1). Thus, these CT inputs simultaneously provide direct excitatory input to relay neurons and exert additional top-down control on inhibition via TRN, providing a cortical signal for task or attention-specific modulation of the thalamus. Overall, the net functional impact of L6A CT signals on thalamic relay neurons is determined by an excitation-inhibition balancing act of rate-dependent synaptic depression and facilitation (Crandall et al., 2015); this dynamic balance is seen in vivo in both HO and FO nuclei (Kirchgessner et al., 2020). It should be noted that HO and FO nuclei are targeted by partially disjoint populations of neurons in L6A (Thomson, 2010; Whilden et al., 2021; Figure 1), perhaps in keeping with the distinct reticular inhibitory circuits discussed above. Finally, aside from the L6A CT pathways, recent studies have identified a HO-thalamus specific pathway from layer 6B (Figure 1) which does not send collaterals to TRN and appears to exert strong excitatory influences on relay neurons (Hoerder-Suabedissen et al., 2018; Ansorge et al., 2020), although detailed physiological measures of these inputs are currently lacking. Notably, L6B neurons are orexin-sensitive (Bayer et al., 2004), suggesting that the cortex may provide HO thalamus with descending excitation based on wakefulness.

Another key differentiator of HO thalamic circuits is the presence of additional subcortical inhibition from sources of “extrathalamic inhibition” (ETI) (Figures 1, 2), such as the zona incerta (ZI) and the anterior pretectal nucleus (APT) [reviewed in Halassa and Acsady (2016)]. Anatomical and experimental studies have suggested that these ETI sources exert a significantly more powerful inhibitory effect on the thalamus than do intrathalamic inhibition sources (Barthó et al., 2002; Lavallée et al., 2005; Park et al., 2018). In contrast to the TRN, which targets all thalamic nuclei, ETI inputs display greater target selectivity, a property that is evolutionarily highly conserved (Halassa and Acsady, 2016). Across sensory modalities, studies have demonstrated that HO, but not FO nuclei, receive ETI input, and this appears to serve as the inhibitory counterpart of L5tt excitatory inputs. This topographical restriction means that these ETI projections do not interact with ascending sensory relays, nor do they directly impact FO thalamus activity. Overall, ET inhibition of HO nuclei is notably strong, rapid, and precise, rendering it capable of influencing the timing of individual spikes in target cells (Halassa and Acsady, 2016). Further spatial selectivity has been established within subregions of the ZI. For example, Lavallée et al. (2005) found that the posterior (PO) HO nucleus receives inhibitory input from the ventral division of the ZI, which is in fact the same ZI subregion that receives input from the APT (Giber et al., 2008; Murray et al., 2010). Further underscoring the importance of ETI nuclei in sculpting HO thalamus processing of L5tt inputs, ZI and APT are also strongly driven by L5tt collaterals (Figure 1)–thus, HO nuclei are embedded in an additional feedforward inhibitory loop similar in motif to the L6A-TRN loop, but specific to HO nuclei and with different synaptic properties (Barthó et al., 2002, 2007; Halassa and Acsady, 2016).

Intrinsic thalamic contributions to a corticothalamic code from L5tt to HO thalamus are expected to be considerable, given the nonlinear bursting properties of thalamic neurons (Llinás and Jahnsen, 1982) and increased appreciation for bursting as a significant mode of information processing (Zeldenrust et al., 2018b). In brief, across modalities, HO neurons display the well-known burst-tonic excitability switch characteristic of thalamic relay neurons (Figure 2). Burst spiking–high-frequency sodium APs superimposed on a transient calcium plateau or “low-threshold spike”–arises from the interplay of slow voltage-dependent currents, namely I_T, carried by the Cav3 subfamily of calcium channels, with some contribution from I_H, the hyperpolarization-activated cation current (Jahnsen and Llinas, 1984; Destexhe et al., 1993; Sherman, 2001). In particular, the phenomenon of bursting in thalamic neurons is well-studied, but a limitation to understanding HO bursting is that most detailed biophysical measurements come from FO neurons, typically from the visual system.

Several reports indicate that bursting properties differ between HO and FO nuclei: (Ramcharan et al., 2005) report more bursting in primate HO MD and HO pulvinar; cortico-recipient LP in rat shows enhanced bursting (Li et al., 2003a) and likewise, in tree shrew pulvinar (Wei et al., 2011), increased bursting was associated with increased expression of Cav3.2 and SK2, an afterhyperpolarization-generating K+ channel which can support repetitive firing. The somatosensory system appears to be an exception, as (Landisman and Connors, 2007) found that HO POm spikes at lower frequencies relative to FO VPM, both within bursts and during tonic spiking, associated with a higher voltage threshold for AP initiation in HO POm. A recent comparison across sensory modalities (Desai and Varela, 2021) provides a mechanistic framework for understanding some of these differences based on electrophysiology and simulations. Rebound burst size (number of APs) in HO nuclei was comparable across sensory modalities. However, differences between FO and HO bursting were modality specific and in agreement with data previously reported for single modalities: in visual and auditory thalamus, FO nuclei were less bursty than HO nuclei, while in somatosensory thalamus, the FO nucleus was more bursty than HO. This spectrum of bursting properties could be reproduced by changing the voltage-dependence and maximal conductance of I_T. Taken together, these studies suggest functionally relevant differences in fast and slow intrinsic excitability mechanisms in HO and FO thalamic neurons.

Indeed, a recent comprehensive report of the thalamus transcriptome (Phillips et al., 2019) found that HO and FO nuclei show distinct transcriptomic profiles based on expression of genes tightly linked to neuronal identity (ion channels, receptors), in line with distinct intrinsic excitability of HO neurons. These points underscore the need for additional basic functional data from HO neurons, particularly voltage-clamp studies of biophysical properties necessary for computational studies, and at minimum, caution in using FO models as stand-ins for HO neurons in simulations.

We emphasize that careful consideration of bursting mechanisms is key to understanding how HO thalamus performs computations on descending cortical signals, as these mechanisms are responsible for nonlinear transformation of L5tt-evoked EPSCs to EPSPs. In the simple case of synaptic input after a period of inactivity, e.g., the beginning of a cortical Up state–large undepressed EPSCs will be further enhanced by activation of I_T; functionally, this means that a single L5tt spike can evoke bursts of spikes in a post-synaptic HO neuron (supralinear corticothalamic spike transfer) (Groh et al., 2008; Seol and Kuner, 2015). In the more complicated case of higher input rates, synaptic depression will decrease EPSC size and preceding depolarization will simultaneously inactivate I_T, supporting a more linear EPSC→ EPSP transformation. Here, integration of coincident inputs from different L5tt neurons can be required to drive output spikes (Groh et al., 2008). Although L5tt neurons’ drive of HO neurons at the single-neuron level has been most studied in the somatosensory system, the parallel effect of synaptic depression paired with inactivation of bursting is also apparent in data from other modalities (Li et al., 2003b; Collins et al., 2018).

The information processing role of bursts is an active area of study, and our understanding of thalamic burst coding continues to be refined. Recent studies (Elijah et al., 2015; Mease et al., 2017; Zeldenrust et al., 2018a; Park et al., 2019) make it clear that various properties of thalamic bursts (number of spikes, timing of spikes, burst onset, etc.) can convey information about presynaptic inputs, so it may be that the distinct bursting properties of HO thalamus subserve a particular functional role. In HO POm, we found that intrinsic bursting and high spiking threshold of POm neurons provides a mechanism for “multiplexed coding” of low- and high-frequency (∼5 Hz and >100 Hz) information, and that high-frequency encoding channel showed information-preserving adaptation (Mease et al., 2017). Furthermore, the exact bandwidth of the “slow” encoding channel was tuned by depolarization. This finding suggests that POm spike trains could carry information both in burst size and precise spike timing within bursts. However, the implications of this intrinsic code in combination with strong L5tt-HO synaptic depression have yet to be assessed. While it is not clear if these findings can be generalized to other nuclei, the observation that HO bursts seem to show less variation across modalities (Desai and Varela, 2021) suggests that further investigation of a common HO thalamic computational scheme for encoding L5tt inputs may provide new insights beyond those gleaned from studies focusing on computation in FO thalamus.

A final unexplored factor in HO thalamic encoding of L5tt inputs is neuromodulation. As membrane potential is a critical mediator of bursting properties, any modulatory input altering baseline depolarization or hyperpolarization would be expected to exert strong control over HO thalamic encoding. While neuromodulation specific to HO thalamus has not been widely studied, evidence suggests that it may dynamically regulate HO CTC network activity. For example, Stitt et al. (2018) suggest that activity in the HO thalamus may be state-dependent and influenced by ongoing levels of arousal. Elsewhere, it was found that the brainstem cholinergic system preferentially suppresses spontaneous activity in the POm-targeting region of ZI, thereby enhancing whisker responses in the POm (Masri et al., 2006; Trageser et al., 2006). Furthermore, Varela and Sherman found neuromodulation may exert differential effects on FO and HO neurons: while all FO and most HO neurons are depolarized by muscarinic and serotonergic activation, a significant fraction (15–20%) of HO neurons are hyperpolarized (Varela and Sherman, 2007, 2009). Taken in combination with attentional control of TRN and arousal-dependent activation of L6B discussed in the previous section, these studies tentatively point to an unexplored state-dependence of HO thalamus’s processing of L5tt drive. Future studies of this topic are warranted, particularly incorporating insight from recent molecular profiling work (Phillips et al., 2019) which may provide specific candidate mechanisms for state-dependent control.

The active, nonlinear dendritic properties of L5tt neurons [reviewed in Larkum (2013); Ramaswamy and Markram (2015)] provide a single-neuron substrate for the integration of “top-down” and “bottom-up” information streams arriving at different lamina. Excitation of either basal or apical dendrites leads to sparse spiking, but near-coincident excitation of both regions triggers a burst of high-frequency (>100 Hz) APs which depends on a backpropagating AP in the soma triggering a calcium plateau in the apical tuft (Larkum et al., 1999; Larkum, 2013). In this framework, bursts indicate a temporal alignment of internal representation and novel external information. Such high-frequency bursts are characteristic of L5tt neurons in the primary somatosensory (Larkum et al., 1999; de Kock and Sakmann, 2008), visual (Shai et al., 2015), and auditory (Llano and Sherman, 2009; Williamson and Polley, 2019) cortices, and are associated with perception (Takahashi et al., 2016, 2020) and exploratory whisker touch (de Kock et al., 2021). While bursting appears to be an important nonlinear computational property of L5tt neurons, it remains to be determined if these spiking patterns are relevant for postsynaptic targets such as HO thalamus, i.e., are these high-frequency patterns faithfully transferred across the L5tt-HO thalamus synapse?

Particular patterns of L5tt synchronization often involve inhibitory feedback interactions with interneurons, reviewed in Naka and Adesnik (2016). In particular, L5tt neurons excite both Martinotti cells (a subset of somatostatin-positive interneurons) which inhibit L5tt apical tufts, and PV neurons, which provide perisomatic inhibition to L5tt neurons. Inhibition from Martinotti neurons seems particularly tuned to L5tt bursting, as Martinotti neurons are preferentially recruited by high-frequency inputs from L5tt due to synaptic facilitation. This “frequency-dependent disynaptic inhibition” is a mechanism linking L5tt bursting to population synchrony, as bursting in a handful of L5tt neurons preferentially inhibits and synchronizes neighboring L5tt neurons (Silberberg and Markram, 2007; Hilscher et al., 2017).

Synchronous activity in particular frequency bands is an important mechanism by which cortical circuits transfer information (Buzsáki and Draguhn, 2004) and recent work has begun to relate understanding of cortical rhythms to specific neuronal cell types. For example, Adesnik and Scanziani (2010), Otsuka and Kawaguchi (2021) report that stimulation of L2/3 pyramidal neurons evokes beta/gamma band activity in L5tt neurons. Corticothalamic synchrony has been explored extensively from the perspective of CTL6 inputs to thalamus (e.g., Contreras et al., 1996; Bal et al., 2000; Béhuret et al., 2013), but how L5 synchrony might propagate to the HO thalamus is only beginning to be studied. For example (Stitt et al., 2018) find that alpha and theta band coherence is prominent in deep-layer PPC and pulvinar interactions; while this study did not specifically isolate layer 5, the involvement of HO nuclei suggest that L5tt input could participate in driving these corticothalamic oscillations. In sum, L5 synchronous population oscillations over a wide range (alpha, beta, theta, and gamma) of frequencies suggest that L5tt-HO thalamus synapses will have synchronized activation at diverse temporal scales that will engage different degrees of synaptic depression.

Recent efforts have begun to explore how pyramidal single-cell properties and population activity in the cortex are linked to implement network-level information coding strategies. The recently proposed Burst Ensemble (Naud and Sprekeler, 2018) theory suggests that ensemble event rate in L5tt (spike or burst count/time) reflects somatic input, burst probability reflects apical input, and burst rate reflects coincident somatic and apical input, while a simple spike rate code cannot disambiguate these combinations of inputs. Lipshutz et al. (2020) explored how simple model networks of pyramidal neurons can implement canonical correlation analysis, finding the features in apical and basal inputs which have maximal correlation; in this framework bursts indicate maximal alignment between the two input sources. Lankarany et al. (2019) demonstrate synchrony-division multiplexing: S1 neurons receiving common input can use the rate of asynchronous spiking to encode the intensity of low-contrast stimulus features while using the timing of synchronous spikes to encode the occurrence of high-contrast features. These studies highlight the importance of simultaneously considering cortical bursts and synchronous spikes as putatively informative signals for postsynaptic targets such as HO thalamus.

In sum, L5tt spike trains appear to carry information in spike count and timing and population synchrony; it is not well understood to what degree these information streams are disentangled and further transformed in HO thalamus. Groh et al. (2008) show that stimulating L5tt-POm synapses with in vivo L5tt spiking patterns resulted in single L5tt spikes driving POm spiking or bursting after long periods of silence in contrast to subthreshold EPSPs evoked at higher presynaptic L5tt rates sufficient to induce depression. Similarly, Collins et al. (2018) report that PFC L5tt EPSCs depress and only drive spiking in HO MD at the onset of 10 Hz stimulation. However, Groh et al. (2008) found that coincident activation of separate L5tt inputs served to overcome synaptic depression, and suggested a role for HO thalamus in detecting synchronous firing of L5tt neurons. Within the HO thalamus multiplexing framework (Mease et al., 2017), such coincident L5tt upstream activity could be encoded by the timing and spike count of POm bursts. Such coincidence detection may also work similarly in the case of integration of cortical and subcortical drivers, as in anesthetized animals, POm output reflects the latency between L5 activation and whisker stimulation (Groh et al., 2013).

Abundant evidence highlights L5tt bursts as somewhat privileged spiking patterns, both in terms of selective encoding of inputs and intracortical impact. From the point of view of HO thalamus, bursts would be translated into temporally discrete EPSCs of decreasing size. We hypothesized that the intrinsic properties of POm neurons may allow these EPSCs to influence spike timing within HO bursts, a situation which would preserve much of the temporal information with L5tt bursts (Mease et al., 2017), but this possibility remains to be tested particularly in awake animals during which HO thalamus bursting is less pronounced. An alternative hypothesis is that subsequent spikes in L5tt bursts evoke EPSCs too small to drive spiking in HO thalamus, or require coincident bursts from multiple presynaptic L5tt neurons to drive HO spiking. Combining cell-type-specific approaches with depth-resolved high-yield recordings in cortex (e.g., Senzai et al., 2019) and HO thalamus (e.g., Kirchgessner et al., 2020) will likely provide data to test these hypotheses.

Cell-type-specific circuit interventions have revealed roles for HO thalamus in conveying signals important for learning, perception, and behavioral salience. Recent in vivo studies in behaving animals are also providing insight into the qualitative content of these HO thalamus signals, supporting roles for HO TC in promoting awake cortical behavior patterns associated with learning, cognitive flexibility, perception, and even consciousness.

In S1, studies by Holtmaat and colleagues provide a strong line of evidence for HO POm in generating long-term potentiation of intracortical synapses onto L2/3 pyramidal neurons via NMDA-dependent plateau potentials which depend on VIP-mediated disinhibition (Gambino et al., 2014; Williams and Holtmaat, 2019); these plateau potentials also provide a mechanism for cortical map plasticity (Pagès et al., 2021). In a multi-whisker sensory association task, Audette et al. (2019) found that training sequentially induces plasticity at HO POm TC synapses onto L5 and L2 pyramidal neurons, thereby increasing POm-driven spiking without changes in cortical single-cell properties. In an auditory associative learning task, Pardi et al. (2020) also demonstrated learning-related HO TC plasticity, finding that HO TC input to A1 transmits memory-related information which reflects task-specific relevance of sensory stimuli. El-Boustani et al. (2020) find that HO thalamus inputs show goal-directed modulation in mice trained in a whisker discrimination task; similarly (LaTerra et al., 2020) find that HO POm axons in S1 signal correct performance during goal-directed behavior and that inhibition of POm impedes task performance.

In the context of cognition, it is suggested that a generic role for the thalamus may be to coordinate and maintain cortical representations relevant for particular cognitive tasks (Halassa and Kastner, 2017; Nakajima and Halassa, 2017). For example, in an auditory-visual cue-switching attentional task, Schmitt et al. (2017) find that HO MD maintains PFC ensemble representations of task rules by control of functional connectivity. In the same behavioral paradigm, Rikhye et al. (2018) found that interactions between the PFC and HO MD provide a mechanism for cognitive flexibility to switch cortical representations, with MD thalamus encoding behavioral context. Although the specific contributions of L5 inputs to such cognitive tasks remain to be assessed, recent evidence suggests that L5-CTC loops may be key to conscious perception. For example, Suzuki and Larkum (2020) show that HO POm TC input enables L5tt dendro-somatic coupling necessary for awake activity patterns and robust somatic spiking, and that general anesthesia blocks this coupling. Although not specific to L5, Redinbaugh et al. (2020) find that in primates, activity in central lateral (CL) HO thalamus and deep-layer cortical neurons correlate with consciousness level; indeed, gamma stimulation in CL could rouse monkeys from anesthesia. These studies point to diverse but unexplored functions for HO nuclei in higher-level cognitive computations in the cortex.

We emphasize here that a core component of disentangling HO thalamus’s role in critical cortical function is quantification of how patterns of L5tt spiking are selected and transmitted back to the cortex. Finally, understanding the impact of specific patterns of HO neuron activity in cortex will require combining emerging knowledge of cell-type-specific TC synaptic dynamics with the state-dependent, nonlinear dendritic integration properties of cortical neurons, along with detailed microcircuit connectivity patterns.

While further work is required to assess the underlying mechanisms and functionality of L5-originating CTC networks, recent studies have emphasized their potential clinical relevance (see Table 2). Among these, an increasing number of studies highlight the importance of these CTC networks in processing nociceptive information in both acute and chronic pain states. For example, spared nerve injury (SNI) in rodents- a model of neuropathic pain–results in a three-fold increase in Ca²⁺ activity in the somata of L5 pyramidal neurons in S1, as well as an increase in dendritic spine Ca²⁺ activity (Cichon et al., 2017). However, this study did not assess the impact of enhanced cortical activity on HO thalamic nuclei. Indeed, the development of chronic pain is associated with several maladaptive alterations in S1 and other cortical regions (e.g., hyperexcitability, somatotopic reorganization), but there remains a need to characterize how these alterations specifically influence CTC networks, especially given that these cortical alterations typically correlate with the degree of mechanical allodynia (i.e., a painful sensation resulting from typically innocuous mechanical stimuli) (Tan and Kuner, 2021). The relevance of these cortical alterations, and their potential impact on HO nuclei and CTC networks, is further exemplified by the fact that targeting S1 alterations beneficially alters pain trajectories (Flor et al., 2001; Moseley and Flor, 2012; Cichon et al., 2017). Hyperactive states in L5 neurons have also been observed in other cortical areas under inflammatory states–for instance, a study employing a peripheral inflammatory mouse model found transient hyperactivity in L5 pyramidal neurons in the frontal/motor cortex (Odoj et al., 2021). However, this study did not observe L5 hyperactivity in S1, so it is unclear whether the frontal/motor cortex alterations are pain-relevant.

Cichon et al. (2017) further investigated the role of inhibitory circuits in the development of S1 alterations, specifically, the source of L5 hyperactivity in chronic pain. It was shown that 1 month following SNI induction, somatostatin (SOM) interneuron activity, which regulates somatic and dendritic pyramidal cell activity, was reduced by half. Likewise, PV interneuron activity was reduced. In part, these findings were the result of a 90% increase in vasoactive intestinal peptide (VIP)-expressing interneuron activity in SNI animals, which directly inhibit SOM and PV interneurons. The SOM contribution to L5 pyramidal cell hyperactivity was confirmed through selective SOM cell activation, which decreased L5 dendritic and somatic Ca²⁺ activity and prevented the development of mechanical allodynia (Cichon et al., 2017). However, activation of PV interneurons did not alter mechanical thresholds for pain, perhaps because these cells predominantly synapse perisomatically, or because they provide only brief inhibition of somata (Cichon et al., 2017). Given the specific interaction of HO TC inputs with PV and VIP interneurons covered in the previous section and the plasticity of these connections (Audette et al., 2017, 2019; Williams and Holtmaat, 2019), it may be productive to further assess how HO TC inputs contribute to maladaptive cortical plasticity related to chronic pain.

It may be that L5 alterations in part contribute to changes observed in HO nuclei during chronic pain. For example, the PO nucleus, which is involved in pain processing, displays higher spontaneous firing rates and greater responses to both noxious and innocuous peripheral stimuli in a chronic pain state (Perl and Whitlock, 1961; Whitlock and Perl, 1961; Casey, 1966; Mao et al., 1993; Masri et al., 2009; Park et al., 2018). Another relevant HO nucleus is the MD, which mediates affective aspects of pain and is similarly hyperactive in chronic pain states. However, as with the PO, it is not understood if these changes in part stem from CTC alterations at the level of the cortex (Rinaldi et al., 1991; Wang et al., 2007; Whitt et al., 2013; Mitchell, 2015). Reinforcing its role in affective components of pain, optogenetic activation of MD inputs in the anterior cingulate cortex (ACC) elicits a conditioned place-aversion in the SNI model and chemotherapy-induced neuropathy, but intriguingly, direct inhibition of L5tt ACC projection neurons to the MD nucleus also produces this effect (Meda et al., 2019). In addition, inactivation and lesioning of the MD nucleus both resulted in a reduction in thermal and mechanical hyperalgesia in a rodent model of neuropathic pain (Saadé et al., 2007). Overall, while further effort is required to assess the overall role of CTC networks in pain, evidence suggests that both HO-projecting L5tt cells and HO thalamic nuclei are altered in pain states, and that targeting these alterations may serve to benefit pain trajectories.

In the auditory system, comparable to the findings in chronic pain, Asokan et al. (2018) observed a hyperactive state of L5 projection cells in the auditory cortex following noise-induced damage to cochlear afferents, and this effect was sustained for several weeks (Asokan et al., 2018). Specifically, this L5 potentiation–which represents a form of compensatory plasticity–was observed in projections to the IC, but the same study also found axon collaterals to the HO MGBd nucleus (Asokan et al., 2018). Alterations in inhibitory networks are also implicated in this L5 pathology–for instance, PV interneuron-mediated intracortical inhibition is significantly reduced for at least 45 days following cochlear synaptopathy (Resnik and Polley, 2017). Increased sensory gain is a characteristic finding across noise-induced hearing-loss pathologies (e.g., tinnitus), and it is proposed that L5 projection neurons are responsible for driving hyperexcitability and strong coupling across tinnitus-associated brain networks (Asokan et al., 2018). As with the pain example discussed above, the impact of such pathological L5 activity on auditory HO nuclei remains to be assessed.

In addition to the apparent involvement of CTC networks in sensory modalities, recent work has alluded to their relevance to pathologies characterized by cognitive and behavioral dysfunction, such as schizophrenia (SZ) and Alzheimer’s disease (AD). Studies have demonstrated the involvement of the MD in these dysfunctions, and there is human, rodent and primate evidence that they can be elicited by damaging this thalamic nucleus (Mitchell et al., 2007, 2014; Mitchell and Gaffan, 2008; Mitchell, 2015; Perry et al., 2021). For instance, monkeys with damage in the MD display impairments in complex associated learning tasks (Mitchell et al., 2007). MD damage-associated impairment has been postulated to result from disruption of the influence of the MD nucleus on the PFC, since the MD nucleus and PFC are reciprocally connected within specific subdivisions (e.g., the pars parvicellularis of the MD is reciprocally connected to the dorsolateral PFC), but this view is contested (Mitchell, 2015; Collins et al., 2018). Specifically, the MD has been shown to activate cortico-cortical projections in layers 2 and 3 in the PFC (Collins et al., 2018). Further work on thalamic innervation of associative brain structures has shown that enhanced excitability in the MD elicits suppression of PFC excitatory neurons (Mukherjee et al., 2020).

Moreover, it was found that hypoxic-like damage to the PFC results in enhanced theta-frequency coherence between the MD and the PFC, as well as an increase in the frequency of bursting in MD neurons, while subsequent knockdown of T-type calcium channels (Cav3.1) in the MD nucleus decreased theta-frequency coherence and attenuated associated symptoms (e.g., frontal lobe seizures and locomotor hyperactivity). The authors propose that the observed neurological and behavioral abnormalities result from impaired thalamocortical feedback between the PFC and the MD, driven by the activation of thalamic T-type calcium channels (Kim et al., 2011). Moreover, abnormalities have been reported in HO nuclei in SZ patients, including reductions in the volume and activity of the MD nucleus and the PuV. This may disrupt transthalamic networks and account for schizophrenia-associated cognitive impairments (Sherman, 2017). In fact, Parnaudeau et al. (2013) show that pharmacogenetic inhibition of the MD nucleus disrupts MD-PFC synchrony in the beta range, causing cognitive impairment with relevance to SZ. Similarly, there is evidence that CT dysfunction contributes to cognitive and behavioral impairments observed in AD (Jagirdar and Chin, 2019).

In addition to these findings, it is increasingly understood that dendritic integration in pyramidal neurons, which plays essential roles in sensory processing and cognition, is disrupted in a range of neurodevelopmental disorders (Nelson and Bender, 2021). For example, autism spectrum disorder is associated with genetic changes that elicit functional, morphological, and organizational alterations in L5, but it is not understood if and how these cortical changes affect the rest of the CTC network (Nelson and Bender, 2021).