Published comment on ´Differences between dorsal and ventral striatum in the sensitivity of tonically active neurons to rewarding events` by K. Marche, A-C. Martel and P. Apicella and published in Frontiers in Systems Neuroscience 24th July 2017 doi.org/10.3389/fnsys.2017.00052
SUMMARY
Marche, Martel and Apicella investigated a small group of cholinergic interneurons found in the striatum called TANs. These interneurons are tonically active and even although they are only 3% of the total number of neurons present they act to modulate the local striatal output. The TANs are known to be involved in stimulus-reward events and can be identified by their distinctive firing characteristics ie. the stop in firing in response to stimulus and reward (termed pauses) and the following rebound to baseline activity (termed rebounds). In their article, the authors described their study into whether the TAN responses are uniformly distributed throughout the dorsal striatal region (motor and associative striatum) and ventral region (limbic striatum) and how TAN firing responses can differ.
In order to study the striatal TANs in a stimulus-reward type scenario, Marche, Martel and Apicella used a Pavlovian conditioning task and 2 adult male monkeys (F and T) as experimental subjects. The task consisted of a given visual stimulus (0.3secs) and the restrained monkey having to perform reaching arm movements to receive a liquid as reward response. Test conditions were: fixed reward timing (FRT) – the liquid reward was given at a fixed time interval (1sec) after the visual stimulus independent of any behaviour shown; variable reward timing (VRT) – the interval between the stimulus and reward was varied; and unpredicted (free) reward timing (URT) – where the reward was given independent of stimulus. Blocks of 30 to 40 tests with each trial lasting 6 secs were carried out with randomly alternating conditions. The locations of the total 62 TANs observed were plotted for each monkey (Monkey F, had 37; Monkey T, 25) and firing was recorded. The neuronal activity observed was analysed by detecting changes in the TAN firing. A test period of 100ms was monitored in 10ms time periods starting at the presentation of the stimulus or delivery of reward and the average spike counts within that interval was calculated across all trials. The onset of a modulation was taken to be the beginning of the first of at least 5 time periods that showed a significant difference in spike activity compared to the control. The offset of modulation was taken to be the first of 5 time periods when activity returned to control values and the magnitude of change in TAN activity was measured by counting the firing spikes between the onset and offset and expressed as a percentage above or below baseline activity. The activities of the populations of TANs were pooled across the samples for the different striatal regions corresponding to the functional territories conventionally defined for a primate. The striatal regions investigated were: the motor region (corresponded to part of the posterior putamen to anterior commissure) and contained a total of 26 neurons; an associative region (included dorsal pre-commissural parts of both the caudate nucleus and putamen) – 21 neurons; a limbic striatum region – the ventral part of the caudate nucleus and putamen rostral to the anterior commissure – 15 neurons.
Marche, Martel and Apicella found that the mean firing frequency of the TANs was similar in all three regions (approx. 5.3 spikes/sec). Their study also showed that 68% of the neurons had changes in activity after the presentation of the visual stimulus and 69% after the delivery of the reward in the FRT condition which indicated that the fractions of responsive TANs within the regions did not differ between the stimulus and reward. Twenty one out of the total 62 neurons responded to only one event, ie. either to the stimulus or reward. The proportion of TANs displaying a response to the visual stimulus were said not to vary significantly among the striatal regions (limbic striatum 80%; motor 58%; associative 71%), but the percentage of TANs responsive to the reward showed some variation with the limbic striatum having a higher proportion of firing TANs than the other two regions (100% limbic; 65% motor; 52% associative). The proportion of TANs responding to both stimulus and reward was higher in the limbic striatum, but the responses were not specifically related to the stimulus. This was shown by the excitatory component of firing not being influenced by the subsequent delivery of the reward.
Marche, Martel and Apicella also looked at the durations and magnitudes of the pauses and rebounds of the TAN responses they observed. They found that the duration of the pause responses to stimuli were significantly longer in the limbic region than in the motor and associative regions. The magnitudes of the pauses were also greater in the limbic and associative regions than the motor region. Investigation of the rebounds from the pauses initiated by the stimuli showed similar patterns with the durations of the rebounds being longer in the limbic and associative regions than the motor area, but the magnitudes did not significantly differ although it was said that there was a trend towards the limbic region. In the case of the pauses and rebounds of the TAN firing in response to the rewards, then few differences between the areas were shown. Only the durations of the pauses following reward gave a significant difference with the highest values being obtained in the limbic and associative areas.
An investigation into the average activity of the entire population of TANs regardless of responsiveness to stimulus or reward found that there was clear modulation of the whole sample recorded in each striatal region after each task event. Population activity was found to be the same for all regions for the control period at 110-120ms after the onset of the stimulus, but the duration of the pause differed with 70 msecs for the motor region, 110msecs for the associative area and 160msecs for the limbic striatum. There were no differences in the duration of the rebound period. After the delivery of reward, again population activity appeared at 110-120msecs and the durations of the pauses were again area specific with the motor region giving a pause length of 70msecs and the associative and limbic regions both 120msecs.
Marche, Martel and Apicella also investigated whether there were differences in TAN responses between the regions when the timing of the reward was changed (ie. the VRT condition). In this case under VRT conditions the same number of neurons and the fraction responded to the stimulus and reward as in the FRT condition. Durations of the pauses in response to the stimuli were the same for all three striatal regions, but the magnitudes of the pause responses were significantly different in the VRT condition with a higher result obtained for the limbic region than the associative striatum. No differences in the durations or magnitudes of the TAN responses to reward were observed under the VRT condition for all 3 regions. When the population activity was investigated as a whole (ie. averaged values of all 27 TANs activities) then the durations of the pauses after stimuli were different with 30msecs for the motor striatum, 80 msecs for the associative striatum and 120msecs for the limbic striatum. Following the reward the population responses ranged from 50msecs for the motor striatum to 110msecs for both the associative and limbic regions.
The third condition investigated by Marche, Martel and Apicella was with the reward being delivered randomly in their conditioning task (URT). This condition ensured that any activity observed was linked to the reward and was independent of the stimulus. The authors found that in the URT condition approximately the same proportion of neurons responded to the reward as in the FRT condition. In the URT condition, the duration of the pause response to the random reward was significantly different for the studied regions ie. longer durations in limbic and associative striatum areas compared to the motor area. This was different to the results obtained for the FRT condition and was supported to some extent when the activities of the populations as a whole were investigated. The averaged population activity of the duration of the pauses in response to the rewards was found to be 170msecs for the limbic region, 80msecs for the associative area and 70msecs for the motor area. The limbic area neurons also produced the stronger responses.
The authors concluded their article by saying that in the striatum`s role of processing motivational information, the ventral striatal TANs exhibit stronger responses to rewarding stimuli than those found in the dorsal striatum. Not only was a higher proportion of neurons in the ventral striatum responsive the TANs present also demonstrated particular response features eg. the duration and magnitudes of the pause and rebound periods for stimulus onset and reward delivery were greater in the ventral area than in the dorsal striatum. This supported the view that the ventral TAN system is different to that found dorsally with possible greater local cholinergic input that could parallel the functional specialisation within the striatum. Marche, Martel and Apicella continued their discussion by noting the similarity of the TANs observed in rodents to those found in their primate subjects and went on to consider whether the afferent signals into the areas may be responsible for the pause response of the TANs. They also considered how TAN activity could be linked to reward predictability and action selection.
COMMENT
What makes Marche, Martel and Apicella`s article interesting is that it again brings to attention the fact that a brain area serving a particular function does not necessarily mean that all neuronal cells in that area act in the same way. Marche, Martel and Apicella describe in their article a small subset of neuronal firing cells called TANs that are located in the striatum and that are tonically active (ie. active without specific synaptic input), but have distinctive excitatory firing characteristics compared to the other excitatory cells present. This comment explores where TAN activity fits in with the cognitive functioning of the striatum and why the identifiable pauses and rebounds in firing are necessary for the neurochemical functioning of the area.
The experimental basis used by the authors to investigate TAN functioning was a conditioning task with a visual stimulus, motor movement (arm reaching) and reward (liquid). This type of experiment is ideal for looking at brain area and neurochemical functioning linked to learnt behaviour based on emotional value and expectations in relation to a reward given after a specific stimulus – core functions associated with the striatal region. The dorsal region (traditionally said to consist of the caudate and putamen) is linked with motor and associative functions (ie. cognition involving motor function), certain executive functions (eg. inhibitory control and impulsivity) and stimulus response learning. The ventral region (traditionally said to consist of the nucleus accumbens, hereafter termed NA, and the olfactory bulb) is said to be linked to mediating reward cognition, reinforcement, and motivational salience. There is a small degree of overlap as the dorsal striatum is also a component of the reward system that, along with the NA core, mediates the encoding of new motor programs associated with future reward acquisition (eg. the conditioned motor response to the reward cue). Before we continue it should be noted that the divisions of the striatum used by Marche, Martel and Apicella in their article do not agree with more common striatal partition. For example, the motor region is described by the authors as that area between part of the posterior putamen to the anterior commissure; the associative area as the dorsal pre-commissural parts of the caudate and putamen; and the limbic area as the ventral part of the caudate and putamen rostral to the anterior commissural. This means that the ´dorsal` term used here reflects the dorsal parts of the caudate and putamen and the ´ventral` term the ventral parts of the same areas. Cognitive roles attributed to the dorsal and ventral areas therefore relate to the caudate and putamen regions and not the more well-known NA.
When we are looking at the striatum and it`s roles in motivationally salient events we can see that it is linked to reward value, the expectation of that reward and the updating of the value of that reward. The input and interpretation of the information relating to the reward itself is not the responsibility of the striatum alone with multiple brain areas involved including sensory systems for informational input and associative areas for assessment, eg. the ventral tegmentum (VTA) with its dopaminergic system (work by Mortig et al, Wittman et al). In the VTA case, the lateral hypothalamus is said to send projections to the VTA which are believed to be important for behaviour and corticotrophin releasing factor (CRF) acts on the VTA to decrease the motivation to work for rewards also by regulating the DA activity (Tyree et al. and Wanat et al.). Also the prefrontal cortex (PFC) plays a major role with immediate and delayed rewards separately represented in the dorsal medial PFC (mPFC) and compared in the ventral PFC in order to guide decisions (Wang et al.). The orbitofrontal cortex (OFC) is known to be linked to the waiting for rewards and their subjective value (McGuire et al). The activity of this area is believed to be controlled by various neurotransmitters such as 5HT, GABA and glutamate eg. citalopram (a 5HT inhibitor) pretreatment is said to lead to increased OFC responses to reward (Delben et al).
The importance of the reward lies not just in its characteristics, but the value placed upon it by the individual. The determination and storage of values also require the inter-connectivity and inter-functionality of multiple brain areas. For example, the encoding and retrieval of values occurs through the firing and connectivity of particular brain areas such as the basal amgydala and the insular cortex. Connectivity here mediates encoding and retrieval of outcome values with the amygdala encoding and the insular cortex retrieving such values in guide choices (Parkes et al.). The PFC again has a number of different functions eg. the mapping of the value of events on a common scale (Gross et al.) and particularly the OFC which links events to reward values (Winecoff et al., Favonik et al.). This area allows the representation of contexts that guide memory retrieval and the comparison of the subjective value of outcome (Hornick t al.). The anterior cingulate cortex also plays a role and is important for encoding competitive effort ie. learning the value of actions.
It is in the role of values and the subtleties of the value system that the striatum shows like the PFC, different functionality for different local areas. The traditionally termed dorsal striatum is said to comprise of the caudate and putamen. In the case of the caudate, activity relating to values was said to rely on dopaminergic functioning eg. Tai et al. found that stimulation of the dorsal striatal dopamine D1 and D2 receptor–expressing neurons during decision-making in mice introduced opposing biases in the distribution of choices. The effect of that stimulation on choice was dependent on recent reward history and mimicked an additive change in the action value. Other researchers found such subtleties in the dorsal striatal functioning relating to values. For example Wunderlich et al. found that the computational processes underlying forward planning are expressed in the anterior caudate nucleus as values of the individual options in a decision tree, ie. competing values of available options. In contrast, the team found that values represented in the putamen pertained solely to the values learnt during extensive training. During the actual choice stage, both of these areas demonstrated functional coupling to the ventromedial PFC which was consistent with it acting as a value comparator integrating the outputs of the two striatal regions. This link with the PFC functioning was further corroborated by Campbell-Meiklejohn et al. who showed that values computed from choices were weighted by their associated confidence and specifically represented in the ventromedial area of the PFC. The tendency to self-monitor predicted a selectively enhanced response to accordance with other`s results in the right temporal-parietal junction and therefore, indicated that using cues of the reliability of other peoples’ knowledge to enhance expectation of personal success generated value correlates that were anatomically distinct from those concurrently computed from direct, personal experience. It also indicated that representation of decision values in the ventromedial PFC was sub-organized according to computed values. Meder et al. found another subtlety of the value system. They found that when gathering valued goods, risk and reward are often coupled and escalated over time. This requires increasing activity and connectivity of a cortico-subcortical ´braking` network that increased with gains and included the caudate as well as the pre-supplementary motor area, inferior frontal gyrus and subthalamic nucleus. The putamen was also linked with value computation and definition although less so than the caudate and more in the area of information processing and storage. Jang et al. reported a link between the putamen and hippocampus with the inter-regional connectivity associated with binding of more abstract information such as attentional state, emotional state that could be recorded with the specific data information to the stimulus and reward in order to increase the chance of perception and appropriate behavioural response.
The description given above for the role of the caudate and putamen relates to the traditionally called dorsal striatum although in the case of Marche, Martel and Apicella`s study this covers both the limbic and associative striatum areas. Although they do not include the NA in their studies this particular area (traditionally termed the ventral striatum and linked with the olfactory bulb) is known to play an important role in values and again relies on the activity of the dopaminergic neurochemical system. Wieland et al. found that phasic dopamine is involved in the assignment of values to stimuli and synaptic plasticity was specific depending on input. West et al. also found that NA neurons encode features of stimulus and behaviour learning and selection associated with rewards. Cue-selective encoding during training by the NA core neurons reliably predicted subsequent behaviour whereas NA shell neurons significantly decreased cue-selective encoding in the devalued condition compared with the non-devalued condition. This indicated that even the NA demonstrated regional specificity to value.
In addition to its role in encoding and retrieving value, the striatum has also been shown to play a role in expectation of reward and the subsequent updating of information on the basis of feedback. These functions can be adequately investigated using conditioning experiments, one of which was carried out by Marche, Martel and Apicella. For example, the striatum is known to play a key role in reinforcement learning specifically in the encoding of the teaching signals such as reward prediction errors (RPEs). The attribution of incorrect values is associated with impaired coding of RPE and an increased turnover of dopamine in the striatum (particularly in the ventral striatal region) and prefrontal cortex connectivity (Boehme et al.). Both negative and positive expectations are linked with striatal activity. For example, negative expectations influence behaviour with activity of the head of the caudate plus NA. In addition activity in the ACC, PFC (including mPFC and OFC), left hippocampus, insular cortex, insular cortex and amygdala plus others is observed whereas positive expectations are linked with the striatum and particularly the NA as well as the dorsolateral PFC, ACC and frontal operculum. The list shows that multiple areas are involved and hence, it is difficult to distinguish unique roles for the dorsal and ventral striatal regions alone. Therefore, higher levels of research are carried out on the other strong cognitive areas such as the ventromedial PFC, VTA and ACC.
However, it is known that the dopaminergic system is involved in the expectation and receipt of expected reward as well as unexpected reward timing and character and striatal areas are known to have dopaminergic functioning. The ability to predict favourable outcomes is said to require dopamine release once the conditioned sequence of events is learnt. At first, dopamine release occurs with the presentation of the reward leading to signaling a predictive cue onset after learning (Day et al.). Eschel et al. found marked homogeneity among individual dopamine neurons with their responses to both unexpected and expected rewards following the same function, but just scaled up or down. As a result, they were able to describe both individual and population responses using just two parameters. Research narrows this type of functioning to the traditionally termed ventral striatum and particularly the NA, which was not part of the experimental model used by Marche, Martel and Apicella. However, NA neurons demonstrated a subtlety of the system with subsets of NA neurons being responsible for particular responses. Owesson-Wright et al. found that DA neurons that projected from the VTA to the NA fired in response to predicted and unpredicted rewards or cues. Both cue presentation and action caused dopamine release that predicted reward delivery, but distinct populations of NA neurons were found to encode the behavioral events at the same specific locations selectively. This was thought possible through different dopamine receptor population activity. A subset of DA-2 receptors mediated responses to the cue whereas the dopamine responses acting after the performance of the required behavioural action was found to be the DA-1 receptor.
This capability of expectation can also be linked to the opposite, that of unexpected reward whether relating to timing or character. Again, such a function is correlated to dopaminergic system functioning in the striatum such as in unpredicted reward in conditioning experiments (Redgrave et al. and Steinberg et al.). The signal requires a reporting of errors in reward prediction (Niv et al.) which Steinberg et al. found required strong activation of midbrain dopamine neurons. The phasic signal which signals the discrepancies between actual and expected outcomes (the so-called reward prediction error) represented an opportunity for new learning (ie. updating) which is well known in conditioning experiments, as that occurring in extinction for example. Again it appears that the VTA-NA connectivity is important. Chowdhury et al. found that abnormal expected values resulted in healthy elderly subjects exhibiting incomplete RPEs in the NA and this signal was tightly coupled to inter-individual differences in the connectivity between the VTA and NA as indicated above. In their experiments the administration of the dopamine precursor L-DOPA increased task performance as it restored the necessary dopaminergic system. Such errors in reward prediction or unexpected rewards whether due to timing or character require the updating of stored information and this too, appears to require striatal involvement particularly the caudate (Chiu et al.) and its dopamine functioning (Diederen et al.).
Therefore, we can see that the striatum is important in the encoding and retrieval of values and particular regions have specific functions. We can assume that these areas function like other brain regions and have a number of excitatory and inhibitory cells whose activity is linked to the demands of the area`s function at that time. There is a lot of research on excitatory neuronal cells particularly the glutamate and dopamine regulated cells and also certain inhibitory cells particularly the GABA regulated ones. Marche, Martel and Apicella instead investigated a particular subset of excitatory neurons found in the striatum, which we have to assume are important for the correct functioning of those areas. Therefore, we have to ask what are these TANs there for particularly when there are other excitatory mechanisms in place? However, before we continue we have to discuss the work on TANs of this region by others, particularly that of Garr (2016) because not all Marche, Martel and Apicella`s conclusions support the views of others. Garr also described a subset of tonically active interneurons (TANs) in the striatum that are also cholinergic and were said to modulate medium spiny neurons (MSN) excitability and sensitivity to cortical input. Just like Marche, Martel and Apicella, these interneurons were found at a population level, to respond to motivationally relevant stimuli with a pause in firing followed by a subsequent increase in firing above baseline (the rebound). In the case of Marche, Martel and Apicella this population was found to be just 3% of the total striatal neuronal cell population and only 62 cells responded in total with 21 responding to only one event. Both Garr and the authors here appear to agree that the patterns observed in the population may not accurately represent the responses of the individual cells and Marche, Martel and Apicella compared both in their study.
On closer investigation of the conditioning experiments used by both sets of investigators, certain discrepancies appear. In the case of Garr, the action of the TANs was investigated using a conditioning task where the two monkeys used were trained to press a lever for a water reward in the presence of the following two cues: one that signaled the amount of force required for a successful response (high or low force) and one that signaled the magnitude of the reward to be earned (high or low reward). The experimental set-up gave four cue combinations: low force/high reward; high force/high reward; low force/low reward; or high force/low reward and after training the results collected from the TANs showed that the majority of them were located in the caudate nucleus. This correlates to the results of Marche, Martel and Apicella who found that limbic and associative regions both containing caudate regions had higher levels of TANs (71, 80% for stimuli response and 52%, 100% for reward response) than the motor region that consisted only of the putamen region.
The TANs in both studies demonstrated the typical TAN firing responses of pause followed by increase in firing (rebound) in response to both the cue and reward. However, in Garr`s study the TANs activity was investigated further and they were found to be separate groups modulated during the two task demands ie. the TANs that paused in response to the cue presentation were generally not the same cells that paused in response to reward delivery, and the same was true for increases in firing. Garr found that few TANs showed a pause during cue presentation, but many showed an elevation response. This could indicate that other excitatory TAN cells are fired rather than just those exhibiting the distinctive firing pattern. These firing cells could relate to the information processing capability of the striatum in binding information of the event with the activity of the hippocampus as given above. Following the delivery of the reward, a similar number of TANs responded with pauses and rebounds which indicate that these are the interneurons associated with reward value and expectation as indicated above to elicit striatal motivational salience function. This supports Garr`s interpretation of results which showed that the majority of pauses preceded the lever press (behavioural action required in this conditioning experiment) and the majority of rebounds coincided with the lever press. If the above view is correct then the ´rebounds` would coincide with the assessment of the value of performing the lever press to obtain the reward (expected and feedback) and may not reflect the same cell firing as that experienced by the reward response.
Garr`s experiments also looked at the effect of effort in performing the action (ie. force modulation) against the size of the reward (eg. high or low water). He found that the magnitude of the pauses and rebounds during the cue and reward stages were dependent on the trial type with separate groups of TANs modulated by the amount of required force and the amount of reward. Only few TANS responded to the cue with pauses, but more were found to rebound. The vast majority was modulated only during the high-force trials and these were seen to be separate from those showing reward magnitude-modulated rebounds during cue presentation where the vast majority of these were modulated during low-reward trials. Therefore, Garr interpreted his results as the presence of two separate groups: one firing for high-force trials, and one for low-reward trials. This correlates to the interpretation given above that the TANS ´rebounds` would correlate to value assessment of the lever press reward.
Therefore, Garr concluded that there were different populations of TANs in the striatum responding to different stages of the condition task and even to different conditions (eg. effort and reward size). Marche, Martel and Apicella did not go into the same depth of conditions that Garr did preferring to look at reward timing, but they did show to some extent a difference in TAN populations that went beyond the limbic and associative striatal location. They found that in relation to their fixed interval reward trials that the proportion of TANs responding to the stimulus was on average 70% and to reward 72% and indicated that the same cells are likely to respond to both even though they showed that a third of their TANs only responded to one or the other. They showed that the proportion of TANs responding to the stimuli did not vary significantly with region (58-80%) whereas they did for reward (65% motor, 52% associative and 100% limbic). In each case the subtle differences may be masked by averaging. For example, a closer look at the results shows that more cells in the associative region respond to the stimuli than to the reward (71% to 52%) and conversely, more cells respond to the reward in the limbic system than to the stimuli (80% to 100%). The duration and magnitudes of the responses were the same so it can be said that the studied responses are TAN cell responses. The results obtained indicate a difference in caudate and putamen functioning dependent on location and task demand. Dorsally located TANs in the caudate and putamen are more active during stimulus presentation and stimulus response and ventrally located TANs in the caudate and putamen are more active during reward presentation and reward response. This may be loosely translated to the functions attributed to dorsal and ventral striatal regions given above although we must assume that term ´ventral` relates to caudate and putamen areas and not just nucleus accumbens as given in the more traditional definitions. Dorsal areas are linked to the computing of options and personal reward value as well as binding of abstract information that may aid in the assessment. Therefore, this definition of function supports the observation that TANs are more active in the dorsal caudate and putamen (associative areas) during the stimulus presentation ie. when value of cue and learnt reward are assessed. The observation that ventrally located TANs in the caudate and putamen are more active during reward presentation and response may support the role of ventral areas in expectation of reward, but may be indicative of specific location of function as given by Garr in his study above where cells responded to different levels of effort and reward.
So we have discussed why the TANs are there functionally, but why are they present from a neurochemical perspective? Why not just a set of normal-acting firing excitatory neurons or interneurons rather than cells that fire, stop firing and then fire again? We know that the mechanism for firing is acetylcholine binding and this results in dopamine release and we have to assume that the pauses are brought about by mechanisms responsible for hyperpolarization of firing cells (eg. GABA binding or chloride ion/potassium ion channel opening). These characteristics are shared by other neuronal cell types that are also present in the striatum. In fact, the principle type of cell in this area constituting 95% of the population are medium spiny neurons (MSN) which are GABAergic inhibitory neurons. They too can be influenced by dopamine and this leads to 2 subpopulations of neurons depending on the presence of DA1 type and DA2 type receptors with 40% of cells having both. The interneuron population consists of excitatory and inhibitory cells. The inhibitory interneurons are GABAergic and are of many different types. The best known are the parvalbumin expressing interneurons (also known as fast-spiking interneurons) which are responsible for fast feedforward inhibition of the principle neurons, but there are also types responding to tyrosine hydroxylase, somatostatin, nitric oxide synthase and neuropeptide-y. The excitatory interneurons of which TANs are a subset also influence striatal cell firing and are influenced by both dopamine (via DA5 receptor) and acetylcholine. For example, large aspiny interneurons release acetylcholine and respond to the salient environment with stereotypical responses which are temporally aligned with responses of DA neurons of the substantia nigra.
Therefore, the first possible reason for the presence of TANs in the striatum is the fine control of firing within the striatum itself and between it and other brain areas. The striatum must perform neurochemically to fulfil its cognitive demands which in the case of conditioning means reacting to sensory input, informational processing in the working memory state of input and value of reward, long term storage of information and its subsequent recall on repetition of trial conditions plus monitoring of expected reward. The striatum must also be linked to other areas that perform these functions. Interneuron functioning allows control of overall striatal functioning by switching on and switching off adjoining neurons and interneurons and this type of modulatory control can be seen in other areas. For example the firing of the GABA interneurons in the hippocampus is linked to short term responses to repeated stimuli, also a requirement in the learning the task in the conditioning experiments carried out in this article. In the hippocampus, pyramidal neurons represent the major postsynaptic target of most interneurons, whereas a small fraction of synaptic contacts (5–15%) from interneurons is made onto other GABAergic cells. The GABA interneurons that innervate each other (the so-called interneuron-specific interneurons) are controlled via specific inhibitory mechanisms. In this example, information reaches the pyramidal cells of the CA3 region in the hippocampus via mossy fibre synapses made by dentate gyrus (DG) cell axons whereas small terminals and filopodial extensions target GABA containing interneurons. These synapses are unusual since they exhibit low basal release probability, pronounced frequency facilitation and exhibit a lack of involvement of the glutamate receptor, NMDA receptor in long-term potentiation (LTP). Synaptically released glutamate mediates both negative and positive feedback acting on presynaptic metabotropic glutamate receptors and kainite receptors, but no post-synaptic activation of NMDA receptors. LTP at the mossy fibre pyramidal synapses occurs through the increase in neurotransmitter release, presynaptic binding, presynaptic calcium ion release and activation of presynaptic adenylate cyclase. The mossy fibre interneuron interaction range of short-term responses to repetitive stimulation goes from pronounced depression to modest facilitation of the firing. This is also seen in the excitatory cholinergic interneuron firing of the TAN subset of cells in the striatum. In the case of the hippocampus cells, glutamate released here activates post synaptic AMPA receptors and this calcium-permeable AMPA receptor activation leads to NMDA receptor-independent LTP and presynaptic decrease in neurotransmitter release whereas activation of calcium-impermeable AMPA receptors shows robust NMDA receptor-dependent LTP and down regulation of post-synaptic population of AMPA receptors (Nicoll et al.).
The TAN firing can also control activation that stems from outside the region. For example in the case of hippocampal long-range input interneurons, the activity of these cells can be controlled by extrinsic GABAergic projections: one arising from the medial septum (MS) and the other from the medial entorhinal cortex (MEC). Activation of the septal GABAergic afferents produces a silencing of interneurons and is associated with rescinding inhibition in the adjoining pyramidal cells. Two distinct populations of MS interneurons have been identified for this function: fast-firing and burst-firing cells. A subset of these cells expresses hyperpolarization-activated and cyclic-nucleotide-gated non-selective cation channels and exhibits firing and rebound spiking in response to rhythmic inhibition. These MS interneurons show a different phase preference during hippocampal theta activity and are thought to target different types of hippocampal interneurons cells that are active at the positive peak of the theta oscillation control dendritic inhibition of the CA1 pyramidal cells.
Therefore, TANs may be a unique subset of interneurons that control firing in the striatum via their own short-range interactions or long-range connectivity. The second reason for having TANs is that the TANs may keep the firing of the area ´ticking over` when there is no outside stimulus since the interneuronal subset also demonstrates tonic activity. The experiments of Marche, Martel and Apicella were performed with a conditioning task and therefore, there were definite responses to stimuli and reward. However, TANs can fire spontaneously and this has been observed with other neurons and interneurons within the striatum. Yorgason and colleagues showed that certain neurons in the NA (part of the traditional ventral striatum) demonstrate spontaneous DA release which is regulated by a number of factors including voltage-gated ion channels, DA2-autoreceptors, and nicotinic acetylcholine receptors (also seen in TAN cells) on cholinergic interneurons. The spontaneous release was described as infrequent (0.3 per minute), but the rate and amplitude of the firing were increased after blocking the potassium Kv channels. The firing of these cholinergic interneuron cells could be increased by a number of different factors eg. blocking glutamate reuptake, but it was found that only the effect on potassium kv channels influenced dopamine release. This could indicate that spontaneous dopamine release via cholinergic interneurons in the NA area are independent of stimulus activated firing and could be responsible as suggested above for ´ticking-over` like functionality, a role played by TANs in the caudate and putamen striatal regions observed by Marche, Martel and Apicella. However, this hypothesis is unlikely since the area is inclined to be permanently active to some degree because of its functioning (eg. value assessment) and connectivity to the major sensory input and processing brain areas (eg. VTA, PFC).
Another possible reason for the existence of TANs is that their activity may also compensate for the refractory periods of activated neuronal cells during task performance. This would allow continuity of firing that could satisfy the conditions required for learning (eg. sustained firing) or information processing (eg. theta wave synchrony of multiple brain areas). This hypothesis could be supported by the fact that the pauses come after the stimulus or reward at times when there is peak activity from other firing cells within the striatum and established connectivity to other areas and rebounds when those cells reach their refractory periods and stop firing in order to replenish their neurochemical stores and restore firing capability. In the case of LTP, a condition for long-term memory required in the correct performance of conditioning tasks, the continued activation of the area may require not only synaptic plasticity of the neuronal cells, but also of the interneurons too. LTP and long-term depression (LTD) have been found at GABAergic interneuron synapsing onto other cells eg. the pyramidal cells of the PFC, but the mechanisms of plasticity at these GABAergic synapses may differ significantly from those formed onto neuronal cells. For example, it has found that LTP occurs in both RAD interneurons and CA1 pyramidal cells and can be induced by theta-burst synaptic stimulation, but they are regulated differently. In the pyramidal cells, LTP is mediated by the activation of both GABA B receptors and a group of metabotropic glutamate receptors, whereas in interneurons, neither is required. This compensation for refractory periods can also be linked to the maintenance of firing so that synchronous firing of multiple brain areas can occur. We have already seen the extent of connectivity of the striatal regions in the establishment of values and their assessment and this has been linked to theta brain wave activity. For example, Murty and colleagues found that there was increased connectivity of category-selective visual cortex with both the VTA and the anterior hippocampus that predicted associative memory for high- but not low-reward memories and therefore, maintenance of synchronous firing is important for the value assessment process.
A fourth reason for the presence of TANs and their distinctive firing pattern relates again to their possible role in switching on or off neurons by aiding the establishment of temporal coordination of firing. In conditioning tasks and others where there is a distinctive order eg. stimulus-to-behaviour, the switching on-off ´pulsing` of interneuron firing could be akin to that seen in Morse code signaling. The switch-off periods would stop firing for a period after the stimulus or reward as given in the case of Marche and colleagues experiments, hence providing a distinctive period for informational content for the two parts of the task. This would of course be contrary to the third reason for their presence given above that of sustaining firing during the refractory periods since it would require the shut-down signal to be shared simultaneously by other cells in the area and this has to date not been shown. However, temporal coordination of neuronal activity via several types of cortical GABAergic interneurons has been shown in the hippocampus and extra-hippocortical areas. For example, Unal et al. suggested that oscillatory septal neuronal firing at delta, theta, and gamma synchrony frequencies during stimulus may phase interneuron activity and Jacobs et al. has also linked hippocampus cells to informational timing. Allen et al. showed that hippocampal activity differed depending on the temporal context of items. Salz et al. found that CA3 cells exhibited robust temporal modulation similar to the pattern of timed cell activity in the CA1 and the same populations of cells also exhibited typical place coding patterns in the same task. Middleton et al. investigated further and found that silencing CA3 cells disrupted temporal coding in the CA1 with gamma synchrony important for information binding temporal context and background timing was given by theta synchrony. The hippocampus may, because of its natural spiral physical structure and forced order of firing (eg. DG firing to CA3 to CA1), have natural conditions for timing and order that cannot be achieved through the striatum`s physiology and therefore, the TANs may establish the firing conditions necessary for temporal order of information.
Therefore, we can conclude that TANs although only a small subset of excitatory interneurons in the striatal regions must have a function important to the role that these brain areas have in event salience. They again, confirm the complexity of neurochemical systems in the brain and the need for detailed investigation and understanding of not only neuronal firing, but internal and external brain area connectivity. It is unlikely that manipulation by external means of such a small subset of neurons is possible in order to gain cognitive advantage not only because of their population size but also because they appear to work and be regulated by the ´giants` of the neurochemical world such as acetylcholine and dopamine. However, investigation does lead to a greater understanding of how brain area cognitive functioning relates to neurochemical mechanisms and should be continued.
Since we`re talking about the topic…………………………….
………can we assume that conditioning experiments of the type performed by the authors, but demonstrating successive approximation learning or extinction would lead to no change in the distinctive TAN firing pattern for either if the function of the TANs is to exert fine control over general striatal firing or compensate for refractory periods of other firing cells so sustained firing is achieved?
………if fear conditioning experiments were performed instead of positive reward ones would striatal firing be reduced and hence, a reduction in TAN firing population be observed because of an increase in firing to other brain areas such as the amygdala rather than the striatum?
……..would the administration of the GABA blocker etomidate establish a role of GABA in TAN firing in the conditioning experiment?
…….sleep deprivation is said to increase motivation for reward. Therefore, if the conditioning experiments were performed with sleep deprived subjects would an increase in TAN population number be observed and if so, would this reflect that the extent of TAN firing itself reflecting personal values rather than being a product of firing of other excitatory cells present in the striatal areas?
