The striatum, despite its relatively homogeneous appearance in simple cell stains, is made up of a mosaic of macroscopic zones, the striosomes and matrix, which differ in their input and output connections and are thought to allow specialized processing by physically modular groupings of striatal neurons (Crittenden et al., 2016; Fujiyama et al., 2011; Gerfen, 1984; Graybiel and Ragsdale, 1978; Jiménez-Castellanos and Graybiel, 1989; Langer and Graybiel, 1989; Lopez-Huerta et al., 2016; Salinas et al., 2016; Smith et al., 2016; Stephenson-Jones et al., 2016; Walker et al., 1993; Watabe-Uchida et al., 2012). Particularly striking among these modules are the striosomes (also called patches), which are distinct from the surrounding matrix and its constituent modules by differential expression of neurotransmitters, receptors and many other gene expression patterns, including those related to dopaminergic and cholinergic transmission (Banghart et al., 2015; Brimblecombe and Cragg, 2015; Brimblecombe and Cragg, 2017; Crittenden and Graybiel, 2011; Cui et al., 2014; Flaherty and Graybiel, 1994; Gerfen, 1992; Graybiel, 2010; Graybiel and Ragsdale, 1978). Striosomes in the anterior striatum have strong inputs from particular regions related to the limbic system, including parts of the orbitofrontal and medial prefrontal cortex (Eblen and Graybiel, 1995; Friedman et al., 2015; Gerfen, 1984; Ragsdale and Graybiel, 1990) and, at subcortical levels, the bed nucleus of the stria terminalis (Smith et al., 2016) and basolateral amygdala (Ragsdale and Graybiel, 1988). The striosomes are equally specialized in their outputs: they project directly to subsets of dopamine-containing neurons of the substantia nigra (Crittenden et al., 2016; Fujiyama et al., 2011) and, via the pallidum, to the lateral habenula (Rajakumar et al., 1993; Stephenson-Jones et al., 2016). By contrast, the matrix and its constituent matrisomes receive abundant input from sensorimotor and associative parts of the neocortex (Flaherty and Graybiel, 1994; Gerfen, 1984; Parthasarathy et al., 1992; Ragsdale and Graybiel, 1990), and project via the main direct and indirect pathways to the pallidum and non-dopaminergic pars reticulata of the substantia nigra (Flaherty and Graybiel, 1994; Giménez-Amaya and Graybiel, 1991; Kreitzer and Malenka, 2008), universally thought to modulate movement control (Albin et al., 1989; Alexander and Crutcher, 1990; DeLong, 1990).

This contrast in connectivity between striosomes and the surrounding matrix highlights the possibility that striosomes, which physically form three-dimensional labyrinths within the much larger matrix, could serve as limbic outposts within the large sensorimotor matrix. The question of what the actual functions of striosomes are, however, remains unsolved. Answering this question has importance for clinical work as well as for basic science: striosomes have been found, in post-mortem studies, to be selectively vulnerable in disorders with neurologic and neuropsychiatric features (Crittenden and Graybiel, 2016; Saka et al., 2004; Sato et al., 2008; Tippett et al., 2007). Ideas about the functions of striosomes have ranged from striosomes serving as the critic in actor-critic architecture models (Doya, 1999), to their generating responsibility signals in hierarchical learning models (Amemori et al., 2011), to their being critical to motivationally demanding approach-avoidance decision-making prior to action (Friedman et al., 2017, 2015), and to other functions (Brown et al., 1999; Crittenden et al., 2016). However, the technical difficulties involved in reliably identifying and recording the activity of striosomal neurons have been exceedingly challenging; striosomes are too small to yet be detected by fMRI, and their neurons have remained unrecognizable in in vivo electrophysiological studies with the exception of those identifying putative striosomes by combinations of antidromic and orthodromic stimulation (Friedman et al., 2017, 2015). With the development of endoscopic calcium imaging (Bocarsly et al., 2015; Carvalho Poyraz et al., 2016; Luo et al., 2011) and 2-photon imaging of deep-lying structures (Dombeck et al., 2010; Howe and Dombeck, 2016; Kaifosh et al., 2013; Lovett-Barron et al., 2014; Mizrahi et al., 2004; Sato et al., 2016), combined with the use of genetic mouse models that allow direct visual identification of selectively labeled neurons, identifying functions of these specialized striatal zones should be within reach.

Here we report that we have developed a 2-photon microscopy protocol for simultaneously examining the activity of striosomal and matrix neurons in the dorsal caudoputamen of behaving head-fixed mice in which we used fate-mapping to label preferentially striosomal neurons by virtue of their early neurogenesis relative to that of matrix neurons (Fishell and van der Kooy, 1987; Graybiel, 1984; Graybiel and Hickey, 1982; Hagimoto et al., 2017; Kelly et al., 2017; Newman et al., 2015; Taniguchi et al., 2011). Key to this work was achieving dense, permanent labeling of not only striosomal cell bodies, but also their striosome-bounded neuropil. We accomplished this differential labeling by pulse-labeling with tamoxifen during the generation time of the spiny projection neurons (SPNs) of striosomes using Mash1(Ascl1)-CreER;Ai14 driver lines with induction at embryonic day (E) 11.5 (Kelly et al., 2017). This method allowed striosomal detection based on the labeling of SPN cell bodies as well as the rich neuropil labeling of the striosomes, capitalizing on the fact that SPN processes of striosome and matrix compartments rarely cross striosomal borders (Bolam et al., 1988; Lopez-Huerta et al., 2016; Walker et al., 1993). Thus even though only a fraction of striosomal neurons were tagged, it was possible, because of the restricted neuropil labeling generated by their local processes, to identify neurons as being inside striosomes and, concomitantly, to identify clearly neurons as lying outside of the zones of neuropil labeling, in the matrix.

With this method, we compared the activity patterns of striosomal and matrix neurons related to multiple elementary aspects of striatal encoding as mice performed a classical conditioning task. By having cues signaling different reward delivery probabilities, we tested whether striosomes and matrix differentially encode changes in expected outcome and received rewards (Amemori et al., 2015; Bayer and Glimcher, 2005; Bromberg-Martin and Hikosaka, 2011; Friedman et al., 2015; Keiflin and Janak, 2015; Matsumoto and Hikosaka, 2007; Oyama et al., 2010, 2015; Schultz, 2016; Schultz et al., 1997; Stalnaker et al., 2012; Watabe-Uchida et al., 2017, 2012). By imaging day by day during the acquisition and overtraining periods of the task, we asked whether these patterns changed in systematic ways with experience. Finally, we tested the effect of reward history on the activity patterns of current trials, given reports that strong reward-history activity has been found in sites considered to be directly or indirectly connected with striosomes (Bromberg-Martin et al., 2010; Hamid et al., 2016; Tai et al., 2012).

We demonstrate that neurons visually identified as being within striosomes or within the extra-striosomal matrix have considerable overlap in their response properties during all phases of task performance. Thus, striosomes and matrix share common features related to simple reward processing and manifest acquisition of responses to different task events as a result of reward-based learning. The activities of neurons in the striosome and matrix compartments differed, however, in their relative emphases on different task epochs. Striosomal neurons more strongly encoded reward prediction, and matrix neurons more strongly encoded reward history. These findings suggest that neurons in striosomes and matrix can be differentially tuned by reinforcement contingencies both during learning and during subsequent performance. This work opens the opportunity for future functional understanding of striosome-matrix architecture by in vivo microscopy combined with selective tagging of neurons with known developmental origins, an opportunity that will be valuable conceptually in linking developmental programs to circuit function, and in the study of both normal animals and those representing models of disease states.

To detect striosomes, we performed experiments in Mash1-CreER;Ai14 mice, following the method of Kelly et al., 2017. This method takes advantage of the finding that Mash1 is a differential driver of the striosomal lineage during the ~E10-E13 window of neurogenesis of striosomes in mouse (Kelly et al., 2017). We injected pregnant Mash1-CreER;Ai14 dams with tamoxifen at E11.5, in the middle of this neurogenic phase of striosomal development. This treatment led to the permanent expression in the resulting offspring of tdTomato in cells being born at the time of induction. We found strong tdTomato labeling of striosomes in the striatal regions of the caudoputamen that we examined (Figure 1, Figure 1—figure supplement 1). Critically, this labeling marked not only the cell bodies of the striosomal neurons, but also their local processes, which were confined to the neuropil as confirmed histologically in initial immunohistochemical experiments (Figure 1). These experiments demonstrated that the clusters of labeled neurons and their neuropil corresponded to striosomes, as evidenced by the close match between the zones of tdTomato neuropil labeling and mu-opioid receptor 1 (MOR1)-rich immunostaining (Table 1) (Kelly et al., 2017; Tajima and Fukuda, 2013). We also observed sparsely distributed tdTomato-labeled neurons outside of MOR1-labeled striosomes, scattered in the extra-striosomal matrix, but they never exhibited patchy neuropil labeling.

Figure 1 with 1 supplement see all

Download asset Open asset

For in-vivo experiments, we used 2-photon microscopy to image the striatum of 5 striosome-labeled mice that had received unilateral intrastriatal injections of AAV5-hSyn-GCaMP6s and had been implanted with cannula windows and a headplate (Figure 2A). Each mouse was trained on a classical conditioning task in which two auditory tones (1.5 s duration each) were associated with reward delivery by different probabilities (tone 1, 80% vs tone 2, 20%) (Figure 2B). Inter-trial intervals were 7 ± 1.75 s. With training, mice began to lick in anticipation of the reward, and the amount of this anticipatory licking became greater when cued by the tone indicating a high probability (80%) of reward (Figure 2C). We calculated a learning criterion based on the anticipatory lick rates during the two cues and the subsequent delay period (0.5 s). Mice exhibiting a divergence in anticipatory licking for the two cues for at least two out of three consecutive sessions were considered as trained (Figure 2D). We performed imaging during training (n = 3; task acquisition) and after this criterion had been reached (n = 5; criterion). Two mice were trained for an additional five sessions (overtraining), in which we imaged the same fields of view as in the criterion phase.

Figure 2

Download asset Open asset

Imaging of striosomes

Clusters of tdTomato-positive neurons were clearly visible in vivo in the 2-photon microscope at 40x magnification, and the neuropil of these neurons delimited zones in which many dendritic processes could be identified (Figure 3). We simultaneously recorded transients in striosomal and matrix neurons from fields of view with clear striosomes. In all animals, we could see at least two different striosomes, from which we imaged at least five different non-overlapping fields of view. In some instances, we could see two different striosomes in one field of view. In the entire data set, we imaged 1867 neurons in striosomes and 4453 in the matrix. Because striosomes form parts of extended branched labyrinths, it was possible to follow some striosomes through ±100 µm in depth, and across ±800 µm in the field of view. During training, we rotated through the fields of view, but after the training criterion had been reached, we recorded activity in unique non-overlapping fields of view (2704 neurons, of which 727 were in striosomes; between 252 and 782 neurons per mouse; Table 2).

Figure 3

Download asset Open asset

Table 2

			Mouse
	1	2	3	4	5	Total
Number of neurons	587	782	252	426	657	2704
Striosomal neurons	218 (37.1 %)	214 (27.4 %)	41 (16.3 %)	77 (18.1 %)	177 (26.9 %)	727 (26.9 %)
Matrix neurons	369 (62.9 %)	568 (72.6 %)	211 (83.7 %)	349 (81.9 %)	480 (73.1 %)	1977 (73.1 %)
tdTomato-positive neurons in striosomes	33 (5.6 %)	21 (2.7 %)	11 (4.4 %)	13 (3.1 %)	33 (5.0 %)	111 (4.1 %)
tdTomato-negative neurons in striosomes	182 (31.0 %)	191 (24.4 %)	30 (11.9 %)	60 (14.1 %)	134 (20.4 %)	597 (22.1 %)
tdTomato-positive neurons outside of striosomes	3 (0.5 %)	2 (0.3 %)	0 (0.0 %)	4 (0.9 %)	10 (1.5 %)	19 (0.7 %)

To control for small but significant differences in GCaMP6s expression (Table 3) between striosomes and matrix, we calculated ΔF/F as: ΔF/F = F_t – F₀ / F₀ (F_t: fluorescence at time t; F₀: baseline fluorescence). We quantified the mean, standard deviation and maximum values of the ΔF/F signal during the baseline periods to test for potential differences in the signal-to-noise ratio of our recordings, but did not observe differences between striosomal and matrix neurons (Table 3).

Table 3

	Cell type
	Striosomal	In striosomal neuropil	tdTomato labeled	Matrix
Baseline fluorescence	290.0 (8.5) ***	274.9 (8.1) ***	337.2 (27.9)	364.5 (6.8)
ΔF/F baseline mean	11.3 (0.7)	11.9 (0.7)	9.3 (1.9)	11.9 (0.4)
ΔF/F baseline standard deviation	37.2 (1.3)	38.2 (1.4)	33.5 (3.5)	38.5 (0.7)
ΔF/F baseline maximum	250.6 (9.9)	259.3 (11.2)	216.2 (22.8)	255.2 (6.0)

1. ***p<0.001.

Striatal neurons exhibit heightened activity during different task epochs

As an initial approach to our data, we analyzed the overall fluorescence for every session in trained animals by averaging the frame-wide fluorescence (Figure 4A). Both cues evoked large responses in the neuropil signal, which were calculated as z-scores based on the mean signal and its standard deviation during a 1 s period before cue onset. These signals were larger for the high-probability cue. After reward delivery, there was a prolonged, strong activation that peaked at around 3 s after reward delivery (Figure 4B). To determine more precisely the nature of this activation, we aligned neuronal responses in the rewarded trials to the tone onset, to the first lick after reward delivery and to the end of the licking bout (Figure 4B). This analysis demonstrated that, in addition to the tone response, there was an additional increase in the signal during the post-reward licking period, and that this signal increased over time, peaked at the time of the last lick, and then subsided.

Figure 4 with 1 supplement see all

Download asset Open asset

Next, we analyzed single-cell activity to investigate the neural dynamics of task encoding by the striatal neurons. In particular, we asked whether the prolonged activation seen in the frame-wide fluorescence signal was also visible in single neurons, or whether individual neurons were active during specific task events. Neuronal firing as indicated by the calcium transients was sparse during the task, but we found that individual neurons were active for particular events during the task (Figure 4D,G). For instance, the red color-coded neuron illustrated in Figure 4C and D became active soon after tone onset, whereas the neuron color-coded in gray fired during the post-reward licking period. The timing of their activities with respect to specific trial events seemed relatively stable, resembling what has been reported before for neurons in the striatum of behaving rodents by recording and analyzing spike activity (Bakhurin et al., 2017; Barnes et al., 2011; Gage et al., 2010; Jog et al., 1999; Rueda-Orozco and Robbe, 2015). To determine task encoding by single neurons at a population level, we defined task-modulated neurons as those that were significantly active, according to Wilcoxon sign-rank tests during the cue, reward licking and post-licking epochs of the task (see Materials and methods). Altogether, 38.2% of the striatal neurons imaged in our samples were task-modulated. Of these, most (85%) were active during only one of the three task epochs. Among task-modulated neurons, most were selectively active during the post-reward licking period (57%), but substantial numbers of neurons were also active during the tone presentation (17%) or after the licking had stopped (11%, Figure 4E,F).

For population analyses, we calculated z-scores for the neuronal responses using the mean and the standard deviation of the 1 s baseline period preceding tone onsets. Analysis of session-averaged population responses of neurons selectively active during these three epochs demonstrated a similar sequence of neuronal events as the sequence that we found with analysis of the frame-wide fluorescence signals. The activation of a small group of neurons after cue onset was followed by a prolonged increase in the responses of neurons active during the post-reward licking period (Figure 4F,G). This population activity ramped up until mice stopped licking, then quickly subsided (Figure 4F). The analysis of single-cell responses also identified a group of neurons that became maximally active just after the end of licking. Grouping neurons based on the epoch during which they were active and sorting responses within each group by the timing of their peak session-averaged activity exposed a tiling of task time by neurons active in each of the three epochs (Figure 4G).

To determine the temporal specificity of responses during the post-reward licking period for individual neurons, we compared them to the same responses shuffled for each neuron by substituting responses in a given trial with the response in the same trial from a randomly selected task-modulated neuron recorded simultaneously during the same session (Figure 4—figure supplement 1A). To quantify the trial-to-trial variability in responses, we computed a reliability index as the mean correlation of responses in all pairwise combinations of trials (Rikhye and Sur, 2015). Shuffling the data decreased response reliability, without affecting the mean peak responses, and increased the standard deviation of peak times (Figure 4—figure supplement 1B–D). In addition, we measured the ridge-to-background ratio, which quantifies the mean response magnitude surrounding response peaks relative to other time points (Harvey et al., 2012). We found that the ratio was higher for observed data as compared to the shuffled data (Figure 4—figure supplement 1E). Together, these analyses indicate temporal specificity in the responses of individual neurons and suggest that the prolonged ramping of population activity observed during the post-reward licking period was produced by individual neurons being active within different specific time intervals during licking, and not by them being active throughout the licking period.

Encoding of reward-predicting tones is stronger in striosomes than in the matrix

To dissociate the specific contributions of striosomes and matrix to task encoding, we again first compared aggregate GCaMP6s neuronal responses in both striatal compartments. We drew regions of interest (ROIs) around striosomes defined by tdTomato neuropil labeling and around nearby regions of the matrix in the same field of view with similar overall intensity of fluorescence and size, and compared the total amount of fluorescence from these regions. Both striosomes and matrix exhibited qualitatively similar responses, but there was a significantly stronger tone-evoked activation in striosomes than in the nearby matrix regions sampled (Figure 5A) (ANOVA main effect p<0.001). Moreover, the high-probability tone cue evoked a larger response than the low-probability tone cue (p<0.001), and there was a trend for an interaction between compartment and tone (p=0.055).

Figure 5 with 1 supplement see all

Download asset Open asset

We tested for the selectivity of the responses to the high- and low-probability tone cues by quantifying the area under the receiver operating characteristic curve (AUROC) for these responses. Both striosomal and matrix responses displayed significant selectivity for the high-probability tone (p<0.05), but there was no difference in selectivity between the striosomes and matrix at this stage of learning. We also performed an AUROC analysis to estimate the selectivity for rewarded trials (Figure 5B). Both striosomal and matrix neuropil had elevated activity in rewarded trials, compared to non-rewarded trials, with both high- and low-probability tones. Repeated measures ANOVA showed that striosomes had a higher selectivity for rewarded trials than did the matrix (ANOVA main effect p<0.001). The selectivity for reward was larger in low-probability tone trials than in high-probability tone trials for both compartments (ANOVA main effect p<0.001), but there was no interaction between cell-type and selectivity for reward (Figure 5B). Thus, both striosomes and matrix were more activated when the reward was less expected. We also tested how the beginning and end of licking were reflected by activity in the two compartments (Figure 5C). The striosomal activation was higher than matrix activation (ANOVA main effect p<0.001), and the activation was larger at the end of licking than at the beginning (Figure 5C) (±1 s around event, ANOVA main effect p<0.001). Thus, the overall fluorescence shows that both striosomes and matrix are active during the trials, but that striosomes are more strongly activated, particularly during the cue period, and also at the end of the post-reward licking period. However, we note that summing activity over populations of neurons makes it impossible to dissociate the reward-related activation from the carry-over effects of the tone-related activation.

Next, we analyzed single-cell calcium responses of striosomal and matrix neurons during the task. Individual neurons in both compartments were active during the task (Figure 5D). We found a higher proportion of striosomal neurons (42.9%; 312 out of 712 neurons) than matrix neurons (36.5%; 721 out of 1977 neurons) that were task-modulated (p<0.005, Fisher’s exact test; Figure 5E). Among the task-modulated neurons, a higher percentage of striosomal neurons was active during the cue epoch of the task (23.7% of striosomal, 13.7% of matrix, p<0.001, Fisher’s exact test). By contrast, we found no differences in the percentages of striosomal and matrix neurons that were active during the post-reward licking or post-licking period (p>0.05). In plots of session-averaged activity sorted by the timing of peak responses, we observed that striosomal and matrix neuron activities similarly spanned each of the three task epochs, as though they tiled the temporal space of the task (Figure 5F). We found no differences in the trial-to-trial reliability of striosomal and matrix responses during the task epochs (Figure 5—figure supplement 1). To compare responses of all task-modulated striosomal and matrix neurons during these epochs, we analyzed population-averaged activity aligned to different task events (Figure 5G–J). As in our neuropil analyses, we found that individual striosomal neurons were more robustly active than individual matrix neurons during the cue epoch of the task (Figure 5G, ANOVA main effect p<0.001). Moreover, the high-probability tone elicited a higher response than the low-probability tone (p<0.001).

We used an AUROC analysis to compare activity in trials that were rewarded (responses aligned to first lick after reward delivery) or unrewarded (responses aligned to 2 s after cue onset, a time period matching that for the rewarded trial analysis). We found that striosomal neurons were more selective for rewarded trials (Figure 5H, p<0.001). The selectivity for reward was greater for low-probability than for high-probability tone trials (p<0.01). Although neurons in the two compartments responded similarly during post-reward licking (Figure 5I, p>0.05), striosomes had a higher response during the post-licking period (Figure 5J, p<0.01). Together, these findings demonstrate that neurons in both the striosome and matrix compartments are task-modulated in relatively similar patterns in this appetitive classical conditioning task, that neurons in striosomes are more strongly task-modulated than neurons in the nearby matrix, and that they are particularly more active during reward-predicting cues.

Striosomal tone-evoked responses are acquired during learning

To determine how these responses were shaped by training, we analyzed striatal activity during the acquisition period of the task. To quantify levels of learning, we tested for significance in the difference between anticipatory licking for the high- and low-probability cues during the tone presentation and the reward delay. If mice exhibited a significant difference on 2 out of 3 consecutive days, we considered them as being trained. Sessions performed before this criterion was met were categorized as acquisition sessions. This categorization allowed us to ask whether the strong striosomal cue-related response was a sensory feature, or whether it was an acquired response related to the meaning of the stimulus. Of the five mice studied, two were initially trained on a three-tone version of this task and were therefore excluded from the analysis of the initial training period (Table 4). The three mice included and the two mice excluded from the training data set had similar baseline ΔF/F values and percentages of task-modulated and tone-modulated neurons. Activity measures for the neuropil signals during training for all sessions before the mice reached the learning criterion (n = 33) were compared with the signals in the sessions after criterial performance had been met (n = 20). The striosomal responses to the tones were much stronger after animals learned the task (Figure 6A). The neuropil signal in striosomes was significantly higher after the task performance reached the training criterion than before this point (p<0.05). Such a difference was not observed for the matrix (p>0.05; ANOVA interaction p<0.001). In order to perform a similar analysis for single cells, we grouped responses in consecutive three-day bins. The results indicated that during training, the percentage of task-modulated neurons increased steadily (Figure 6B), but that when mice reached the learning criterion (sessions 11 and 12 for the mice shown in Figure 6B,C), there was a rapid increase in the proportion of cue-modulated neurons (Figure 6C), most notably among striosomal neurons.

Figure 6

Download asset Open asset

Table 4

			Mouse
	1	2	3	4	5
Number of acquisition sessions	12	11	10	19 + 2 *	12 + 4 *
Number of criterion sessions	7	9	4	5	8
Number of overtraining sessions	5	5	0	0	0
Mean baseline ΔF/F	11.2 (0.7)	10.8 (0.7)	12.4 (1.4)	12.1 (1.0)	12.7 (0.6)
Standard deviation baseline ΔF/F	35.5 (1.3)	41.3 (1.3)	30.6 (1.6)	37.9 (1.7)	39.8 (1.2)
Maximum baseline ΔF/F	237.2 (9.6)	296.3 (11.2)	148.8 (9.3)	236.1 (13.2)	270.3 (9.7)
Task-modulated neurons in striosomes	54.1%	22.4%	31.7%	66.2%	46.3%
Task-modulated neurons in matrix	46.9%	20.8%	19.4%	56.4%	40.6%
Tone-modulated neurons in striosomes	21.6%	14.0%	14.6%	16.9%	7.3%
Tone-modulated neurons in matrix	13.0%	9.2%	3.3%	9.5%	4.4%

1. *Two mice were initially trained on a more complex version of the task with three tones instead of two (numbers of sessions trained on the two versions indicated by the first and second number, respectively). Data of these mice were excluded from acquisition analyses.

We further tested whether there was a sudden step-like increase in striosomal tone signaling during training. We averaged the z-scores of the activity of all task-modulated neurons for each of the last five sessions before criterial performance, for the session in which the learning criterion was met, and for each of the first five sessions after the criterial session (Figure 6D). Comparing these values indicated a clear increase in striosomal signaling during the tone (Figure 6E) when the mice began to exhibit differential licking responses to the two tones. This increase in striosomal activity was significant (Figure 6F, ANOVA, training main effect p<0.001; interaction p<0.05). In addition to this development of tone responsiveness, there was a tone-related activation in striosomes in the sessions in which the animals were first exposed to the task (Figure 6C), perhaps reflecting a surprise or novelty signal effect. This tone-related activation disappeared after 1–3 sessions and then reemerged later as mice learned the task. There was also an increase in the percentage of neurons that responded during the post-reward licking period (Figure 6G), and the average activity of all task-modulated neurons during training increased in the period after reward delivery (Figure 6H). In contrast to the increases in tone response, this reward-period increase occurred several sessions before mice learned the task.

During overtraining, tone-related responses of striosomal neurons intensify and become increasingly selective for high-probability tones

To investigate further the relationship between neuronal responses and learning, two mice were trained for an additional five sessions. In these overtraining sessions, we imaged again the same fields of view from which we had collected movies during the criterion phase (Figure 7). The tone-evoked aggregate response became notably higher and sharper during this phase (Figure 7A). The increase in responses related to the tone during overtraining was particularly strong in striosomes. By contrast, the reward period activation immediately following the peak of the cue-evoked response was reduced. The signal initially dropped compared to the earlier sessions but subsequently reached the same magnitude, thus resembling previously reported task-bracketing patterns (Barnes et al., 2005; Jin and Costa, 2010; Jog et al., 1999; Smith and Graybiel, 2013; Thorn et al., 2010). This pattern contrasted with the cue-evoked licking response (Figure 7B), which remained high after the high-probability cue, when the bracketing-like effect in the ΔF/F signal was greatest. We compared the peak responses of the striosomal and matrix samples during the tone presentation period for the acquisition, post-criterion and overtraining sessions (Figure 7C), and found a highly significant interaction (ANOVA interaction p<0.005). In the trained and overtrained mice, striosomes had significantly higher tone-evoked responses than did the matrix (paired t-test, trained mice p<0.01 and overtrained mice p<0.05). The striosomal neuropil responses also became more selective for the high-probability cue during overtraining (Figure 7D), so that during overtraining the striosomal selectivity was significantly larger than the matrix response (paired t-test p<0.05).

Figure 7 with 1 supplement see all

Download asset Open asset

The percentage of task-modulated neurons grew with training, then slightly dropped during overtraining (Figure 7E, left; striosomes: 12.2% during acquisition, 42.9% after criterion and 37.7% during overtraining; matrix: 8.6% during acquisition, 36.6% after criterion and 27.5% during overtraining). At all stages, there were a higher percentage of striosomal task-modulated neurons than matrix neurons responding in the task (Fisher’s exact test, p<0.01). By contrast, the percentage of cue-modulated neurons (Figure 7E, second panel) grew further during overtraining (striosomes: 4.1% during acquisition, 15.0% after criterion and 21.1% during overtraining; matrix: 2.3% during acquisition, 8.1% after criterion and 13.9% during overtraining). There were more tone-modulated neurons in striosomes than in matrix during acquisition, after criterion and during overtraining (Fisher exact test, p<0.05). The percentage of cells that were active during the post-reward licking period (Figure 7E, third panel) increased during training but went down during overtraining (striosomes: 7.0% during acquisition, 28.9% after criterion and 14.9% during overtraining; matrix: 5.6% during acquisition, 27.0% after criterion and 13.3% during overtraining), but there were no differences between striosomes and matrix (Fisher’s exact test, p>0.05). The proportion of neurons activated after the end of licking remained stable for both striosomal and matrix neurons during overtraining (Figure 7E, fourth panel; striosomes: 1.9% during acquisition, 5.6% after criterion and 6.6% during overtraining; matrix: 1.1% during acquisition, 7.3% after criterion and 6.8% during overtraining). We found no differences between striosomes and matrix at any training stage (Fisher’s exact test, p>0.05).

The limited number of significantly modulated neurons in these two mice was too small to make further statistical comparisons between the neuronal responses. Nevertheless, the findings for the entire performance period of the mice collectively demonstrate that the activity patterns observed after training were largely acquired during training, that the strengthening of the tone response was greater for striosomes than for matrix, that this response emerged at the time the animals began differentially responding to the tones, and that this response developed further during overtraining, becoming larger and more selective for the high-probability cue. Because the definition of the training phases was by necessity somewhat arbitrary, and the behavioral performance of the mice could fluctuate across days, we used linear regression to test how well the behavioral performance could predict the ΔF/F activation in striosomes and matrix. For every session, we calculated the mean and standard deviation of the baseline period (1 s preceding the cue onset) and then calculated tone-evoked licking and ΔF/F responses of the neuropil signal in z-scores. We found that in sessions in which the tone-evoked licking was greater, the neuropil response was also greater (Figure 7—figure supplement 1). To test this relationship, we first made two separate models for striosomes and matrix. The regression coefficients for licking were significant for both compartments for the high-probability cue responses but not for the low-probability cue responses (Table 5). When we tested how well the difference in licking during both cue periods could predict the difference in ΔF/F activation during the two cue types, we found a significant regression coefficient for striosomes, but only a trend for the matrix. Next, we made a combined model accounting for ΔF/F activation as a function of licking and quantified the residuals for striosomes and matrix. For both cues and for the difference between them, we found that the striosomal residuals were significantly bigger than those for the matrix. Together, these linear regression analyses demonstrate that in sessions in which the behavioral performance was better, the neuronal response was larger, especially in striosomes. Thus, the behavior was predictive of the neural response, particularly for striosomes.

Table 5

			Trial type
		High-probability cue	Low-probability cue	Difference (high − low)
Striosome model	Regression coefficient	0.069 ***	0.057	0.074 **
	R-squared	0.180	0.029	0.095
Matrix model	Regression coefficient	0.042 *	0.016	0.056
	R-squared	0.081	0.003	0.053
Combined model	Regression coefficient	0.056 ***	0.037	0.065 **
	R-squared	0.118	0.014	0.072
	Residual for striosomes (mean ± SEM)	0.053 ± 0.021 ***	0.031 ± 0.021 ***	0.022 ± 0.023 ***
	Residual for matrix (mean ± SEM)	−0.053 ± 0.021 ***	−0.031 ± 0.018 ***	−0.022 ± 0.024 ***

1. *p<0.05, **p<0.01, ***p<0.001

Matrix responses are more sensitive to recent outcome history than are striosomal responses

In the classical conditioning task employed in this study, mice used the auditory tone presented during the cue epoch to guide their expectation for receiving a reward in the current trial. We examined their licking responses as a proxy for such expectation in order to ask whether, in addition to the information provided by the cue, the mice used the outcome of the previous trial to tailor their reward expectation in the current trial. In trials following rewarded trials, mice showed increased anticipatory licking during the cue and reward delay (Figure 8A, left and right; n = 33 sessions from five mice; p<0.001, Wilcoxon signed-rank test), but licking during the post-reward period was unaffected by outcome in the previous trial (Figure 8A, middle and right; p>0.05, Wilcoxon signed-rank test).

Figure 8

Download asset Open asset

To determine whether the task-related activity of the striatal neurons in our sample was also modulated by outcome history, we compared the activities in trials preceded by a rewarded or unrewarded trial, regardless of the cue type (high- or low-probability) presented in the current trial. We first analyzed the effect of reward history on the cue-period responses of single task-modulated neurons and found that activity was slightly greater when the previous trial was rewarded (mean z-scores: 0.21 ± 0.01 vs. 0.17 ± 0.01 for previously rewarded and unrewarded; p<0.01). However, when we analyzed the effect of outcome history on neural responses observed during post-reward licking in currently rewarded trials, we found that the activity of a subset of striatal neurons was highly sensitive to outcome in previous trials (Figure 8B). Activity during post-reward licking was enhanced when the previous trial was unrewarded, compared to when the previous trial was rewarded. Similarly, population-averaged responses of task-modulated neurons were significantly higher when the previous trial was unrewarded, as compared to when it was rewarded (p<0.001, Wilcoxon rank-sum test). Importantly, post-reward licking behavior was invariant to previous trial outcome (Figure 8A), making it unlikely that the observed changes in neural activity were related to changes in the motor output during reward consumption.

To determine how far back in time we could detect an outcome history effect, we computed a history modulation index (see Materials and methods) for currently rewarded trials with two types of reward history. In the first group, we separated rewarded trials based on whether the previous trial was rewarded or unrewarded (one trial back). For the second group, we disregarded the outcome status in the immediately preceding trial and separated trials depending on the outcome status of two trials in the past (two trials back). This analysis showed that recent reward history has a stronger influence on post-reward licking responses of task-modulated neurons than trials farther back in the past (Figure 8C, p<0.001, Wilcoxon signed-rank test).

We asked whether this history modulation effect was detectable for both striosomal and matrix neurons (Figure 9). Examination of both population-averaged responses (Figure 9A) and single-cell responses (Figure 9B) suggested that both striosomal and matrix neurons were modulated by previous reward history, but that matrix neurons were more sensitive to this modulation. Quantification of this comparison by calculating the history modulation indices for striosomal and matrix neurons confirmed that the matrix responses were more influenced by previous reward history than the responses of striosomal neurons (Figure 9C,D; p<0.01, Wilcoxon rank-sum test).

Figure 9

Download asset Open asset

Our findings demonstrate that 2-photon calcium imaging can be used to identify the activity patterns of subpopulations of neurons distinguished as being in either the striosome or matrix compartments of the striatum. Even with the use of a simple classical conditioning task involving cues predicting high or low probabilities of receiving reward, we could detect in all mice many task-related striatal neurons, altogether 38% of the 2704 neurons successfully imaged in the post-training phase. We found a remarkable parallel in many of the responses of neurons in striosomes and neurons in the matrix. Yet we also found clear differences in the responses of the striosomal and matrix neurons during cue presentation, found contrasts in the timing and selectivity of striosomal and matrix responses during learning and overtraining, and found that the responses of the two compartments were differentially affected by reward history. These findings, based on direct visual detection of striosomes by their birthdate-labeled neuropil and cell bodies, demonstrate that neurons of the two main compartments of the striatum, even though sharing many basic features of neuronal responses during reward-based conditioning, have distinguishable response properties that hint at distinct encoding functions of striosomal and matrix neurons related to reinforcement learning and performance. These findings suggest that in vivo imaging of striosomes and matrix could succeed in solving long-standing questions about the functions of these two major compartments of the striatum.

All experiments were conducted in accordance with the National Institutes of Health guidelines and with the approval of the Committee on Animal Care at the Massachusetts Institute of Technology (MIT).

1. 1. Albin RL
   2. Young AB
   3. Penney JB

   (1989) The functional anatomy of basal ganglia disorders

   Trends in Neurosciences 12:366–375.

   https://doi.org/10.1016/0166-2236(89)90074-X

   • PubMed
   • Google Scholar
2. 1. Alexander GE
   2. Crutcher MD

   (1990) Functional architecture of basal ganglia circuits: neural substrates of parallel processing

   Trends in Neurosciences 13:266–271.

   https://doi.org/10.1016/0166-2236(90)90107-L

   • PubMed
   • Google Scholar
3. 1. Amemori K
   2. Gibb LG
   3. Graybiel AM

   (2011) Shifting responsibly: the importance of striatal modularity to reinforcement learning in uncertain environments

   Frontiers in Human Neuroscience 5:47.

   https://doi.org/10.3389/fnhum.2011.00047

   • PubMed
   • Google Scholar
4. 1. Amemori K
   2. Graybiel AM

   (2012) Localized microstimulation of primate pregenual cingulate cortex induces negative decision-making

   Nature Neuroscience 15:776–785.

   https://doi.org/10.1038/nn.3088

   • PubMed
   • Google Scholar
5. 1. Amemori K
   2. Amemori S
   3. Graybiel AM

   (2015) Motivation and affective judgments differentially recruit neurons in the primate dorsolateral prefrontal and anterior cingulate cortex

   Journal of Neuroscience 35:1939–1953.

   https://doi.org/10.1523/JNEUROSCI.1731-14.2015

   • PubMed
   • Google Scholar
6. 1. Apicella P
   2. Scarnati E
   3. Ljungberg T
   4. Schultz W

   (1992) Neuronal activity in monkey striatum related to the expectation of predictable environmental events

   Journal of Neurophysiology 68:945–960.

   https://doi.org/10.1152/jn.1992.68.3.945

   • PubMed
   • Google Scholar
7. 1. Bakhurin KI
   2. Goudar V
   3. Shobe JL
   4. Claar LD
   5. Buonomano DV
   6. Masmanidis SC

   (2017) Differential encoding of time by prefrontal and striatal network dynamics

   The Journal of Neuroscience 37:854–870.

   https://doi.org/10.1523/JNEUROSCI.1789-16.2016

   • PubMed
   • Google Scholar
8. 1. Balleine BW
   2. Delgado MR
   3. Hikosaka O

   (2007) The role of the dorsal striatum in reward and decision-making

   Journal of Neuroscience 27:8161–8165.

   https://doi.org/10.1523/JNEUROSCI.1554-07.2007

   • PubMed
   • Google Scholar
9. 1. Banghart MR
   2. Neufeld SQ
   3. Wong NC
   4. Sabatini BL

   (2015) Enkephalin disinhibits mu opioid receptor-rich striatal patches via delta opioid receptors

   Neuron 88:1227–1239.

   https://doi.org/10.1016/j.neuron.2015.11.010

   • PubMed
   • Google Scholar
10. 1. Barnes TD
    2. Kubota Y
    3. Hu D
    4. Jin DZ
    5. Graybiel AM

    (2005) Activity of striatal neurons reflects dynamic encoding and recoding of procedural memories

    Nature 437:1158–1161.

    https://doi.org/10.1038/nature04053

    • PubMed
    • Google Scholar
11. 1. Barnes TD
    2. Mao JB
    3. Hu D
    4. Kubota Y
    5. Dreyer AA
    6. Stamoulis C
    7. Brown EN
    8. Graybiel AM

    (2011) Advance cueing produces enhanced action-boundary patterns of spike activity in the sensorimotor striatum

    Journal of Neurophysiology 105:1861–1878.

    https://doi.org/10.1152/jn.00871.2010

    • PubMed
    • Google Scholar
12. 1. Bayer HM
    2. Glimcher PW

    (2005) Midbrain dopamine neurons encode a quantitative reward prediction error signal

    Neuron 47:129–141.

    https://doi.org/10.1016/j.neuron.2005.05.020

    • PubMed
    • Google Scholar
13. Software

    1. Bloem B

    (2017) Elife_2017_scripts, version dc18a30

    Github.

    https://github.com/bloemb/eLife_2017_scripts
14. 1. Bocarsly ME
    2. Jiang WC
    3. Wang C
    4. Dudman JT
    5. Ji N
    6. Aponte Y

    (2015) Minimally invasive microendoscopy system for in vivo functional imaging of deep nuclei in the mouse brain

    Biomedical Optics Express 6:4546–4556.

    https://doi.org/10.1364/BOE.6.004546

    • PubMed
    • Google Scholar
15. 1. Bolam JP
    2. Izzo PN
    3. Graybiel AM

    (1988) Cellular substrate of the histochemically defined striosome/matrix system of the caudate nucleus: a combined Golgi and immunocytochemical study in cat and ferret

    Neuroscience 24:853–875.

    https://doi.org/10.1016/0306-4522(88)90073-5

    • PubMed
    • Google Scholar
16. 1. Brimblecombe KR
    2. Cragg SJ

    (2015) Substance P weights striatal dopamine transmission differently within the striosome-matrix axis

    Journal of Neuroscience 35:9017–9023.

    https://doi.org/10.1523/JNEUROSCI.0870-15.2015

    • PubMed
    • Google Scholar
17. 1. Brimblecombe KR
    2. Cragg SJ

    (2017) The striosome and matrix compartments of the striatum: a path through the labyrinth from neurochemistry toward function

    ACS Chemical Neuroscience 8:235–242.

    https://doi.org/10.1021/acschemneuro.6b00333

    • PubMed
    • Google Scholar
18. 1. Bromberg-Martin ES
    2. Matsumoto M
    3. Nakahara H
    4. Hikosaka O

    (2010) Multiple timescales of memory in lateral habenula and dopamine neurons

    Neuron 67:499–510.

    https://doi.org/10.1016/j.neuron.2010.06.031

    • PubMed
    • Google Scholar
19. 1. Bromberg-Martin ES
    2. Hikosaka O

    (2011) Lateral habenula neurons signal errors in the prediction of reward information

    Nature Neuroscience 14:1209–1216.

    https://doi.org/10.1038/nn.2902

    • PubMed
    • Google Scholar
20. 1. Brown J
    2. Bullock D
    3. Grossberg S

    (1999)

    How the basal ganglia use parallel excitatory and inhibitory learning pathways to selectively respond to unexpected rewarding cues

    Journal of Neuroscience 19:10502–10511.

    • PubMed
    • Google Scholar
21. 1. Carvalho Poyraz F
    2. Holzner E
    3. Bailey MR
    4. Meszaros J
    5. Kenney L
    6. Kheirbek MA
    7. Balsam PD
    8. Kellendonk C

    (2016) Decreasing striatopallidal pathway function enhances motivation by energizing the initiation of goal-directed action

    The Journal of Neuroscience 36:5988–6001.

    https://doi.org/10.1523/JNEUROSCI.0444-16.2016

    • PubMed
    • Google Scholar
22. 1. Chen TW
    2. Wardill TJ
    3. Sun Y
    4. Pulver SR
    5. Renninger SL
    6. Baohan A
    7. Schreiter ER
    8. Kerr RA
    9. Orger MB
    10. Jayaraman V
    11. Looger LL
    12. Svoboda K
    13. Kim DS

    (2013) Ultrasensitive fluorescent proteins for imaging neuronal activity

    Nature 499:295–300.

    https://doi.org/10.1038/nature12354

    • PubMed
    • Google Scholar
23. 1. Crittenden JR
    2. Graybiel AM

    (2011) Basal Ganglia disorders associated with imbalances in the striatal striosome and matrix compartments

    Frontiers in Neuroanatomy 5:59.

    https://doi.org/10.3389/fnana.2011.00059

    • PubMed
    • Google Scholar
24. Book

    1. Crittenden JR
    2. Graybiel AM

    (2016)

    Disease-associated changes in the striosome and matrix compartments of the dorsal striatum

    In: Steiner H, Tseng K. Y, editors. Handbook of Basal Ganglia Structure and Function. Amsterdam: Elsevier. pp. 801–821.

    • Google Scholar
25. 1. Crittenden JR
    2. Tillberg PW
    3. Riad MH
    4. Shima Y
    5. Gerfen CR
    6. Curry J
    7. Housman DE
    8. Nelson SB
    9. Boyden ES
    10. Graybiel AM

    (2016) Striosome-dendron bouquets highlight a unique striatonigral circuit targeting dopamine-containing neurons

    PNAS 113:11318–11323.

    https://doi.org/10.1073/pnas.1613337113

    • PubMed
    • Google Scholar
26. 1. Cui G
    2. Jun SB
    3. Jin X
    4. Pham MD
    5. Vogel SS
    6. Lovinger DM
    7. Costa RM

    (2013) Concurrent activation of striatal direct and indirect pathways during action initiation

    Nature 494:238–242.

    https://doi.org/10.1038/nature11846

    • PubMed
    • Google Scholar
27. 1. Cui Y
    2. Ostlund SB
    3. James AS
    4. Park CS
    5. Ge W
    6. Roberts KW
    7. Mittal N
    8. Murphy NP
    9. Cepeda C
    10. Kieffer BL
    11. Levine MS
    12. Jentsch JD
    13. Walwyn WM
    14. Sun YE
    15. Evans CJ
    16. Maidment NT
    17. Yang XW

    (2014) Targeted expression of μ-opioid receptors in a subset of striatal direct-pathway neurons restores opiate reward

    Nature Neuroscience 17:254–261.

    https://doi.org/10.1038/nn.3622

    • PubMed
    • Google Scholar
28. 1. DeLong MR

    (1990) Primate models of movement disorders of basal ganglia origin

    Trends in Neurosciences 13:281–285.

    https://doi.org/10.1016/0166-2236(90)90110-V

    • PubMed
    • Google Scholar
29. 1. Dombeck DA
    2. Harvey CD
    3. Tian L
    4. Looger LL
    5. Tank DW

    (2010) Functional imaging of hippocampal place cells at cellular resolution during virtual navigation

    Nature Neuroscience 13:1433–1440.

    https://doi.org/10.1038/nn.2648

    • PubMed
    • Google Scholar
30. 1. Doya K

    (1999) What are the computations of the cerebellum, the basal ganglia and the cerebral cortex?

    Neural Networks 12:961–974.

    https://doi.org/10.1016/S0893-6080(99)00046-5

    • PubMed
    • Google Scholar
31. 1. Eblen F
    2. Graybiel AM

    (1995)

    Highly restricted origin of prefrontal cortical inputs to striosomes in the macaque monkey

    Journal of Neuroscience 15:5999–6013.

    • PubMed
    • Google Scholar
32. 1. Fishell G
    2. van der Kooy D

    (1987)

    Pattern formation in the striatum: developmental changes in the distribution of striatonigral neurons

    Journal of Neuroscience 7:1969–1978.

    • PubMed
    • Google Scholar
33. 1. Flaherty AW
    2. Graybiel AM

    (1994)

    Input-output organization of the sensorimotor striatum in the squirrel monkey

    Journal of Neuroscience 14:599–610.

    • PubMed
    • Google Scholar
34. 1. Friedman A
    2. Homma D
    3. Gibb LG
    4. Amemori K
    5. Rubin SJ
    6. Hood AS
    7. Riad MH
    8. Graybiel AM

    (2015) A corticostriatal path targeting striosomes controls decision-making under conflict

    Cell 161:1320–1333.

    https://doi.org/10.1016/j.cell.2015.04.049

    • PubMed
    • Google Scholar
35. 1. Friedman A
    2. Homma D
    3. Bloem B
    4. Gibb LG
    5. Amemori K-ichi
    6. Hu D
    7. Delcasso S
    8. Truong TF
    9. Yang J
    10. Hood AS
    11. Mikofalvy KA
    12. Beck DW
    13. Nguyen N
    14. Nelson ED
    15. Toro Arana SE
    16. Vorder Bruegge RH
    17. Goosens KA
    18. Graybiel AM

    (2017) Chronic stress alters striosome-circuit dynamics, leading to aberrant decision-making

    Cell 171:1191–1205.

    https://doi.org/10.1016/j.cell.2017.10.017

    • Google Scholar
36. 1. Fujiyama F
    2. Sohn J
    3. Nakano T
    4. Furuta T
    5. Nakamura KC
    6. Matsuda W
    7. Kaneko T

    (2011) Exclusive and common targets of neostriatofugal projections of rat striosome neurons: a single neuron-tracing study using a viral vector

    European Journal of Neuroscience 33:668–677.

    https://doi.org/10.1111/j.1460-9568.2010.07564.x

    • PubMed
    • Google Scholar
37. 1. Gage GJ
    2. Stoetzner CR
    3. Wiltschko AB
    4. Berke JD

    (2010) Selective activation of striatal fast-spiking interneurons during choice execution

    Neuron 67:466–479.

    https://doi.org/10.1016/j.neuron.2010.06.034

    • PubMed
    • Google Scholar
38. 1. Gerfen CR

    (1984) The neostriatal mosaic: compartmentalization of corticostriatal input and striatonigral output systems

    Nature 311:461–464.

    https://doi.org/10.1038/311461a0

    • PubMed
    • Google Scholar
39. 1. Gerfen CR

    (1992) The neostriatal mosaic: multiple levels of compartmental organization in the basal ganglia

    Annual Review of Neuroscience 15:285–320.

    https://doi.org/10.1146/annurev.ne.15.030192.001441

    • PubMed
    • Google Scholar
40. 1. Giménez-Amaya JM
    2. Graybiel AM

    (1991)

    Modular organization of projection neurons in the matrix compartment of the primate striatum

    Journal of Neuroscience 11:779–791.

    • PubMed
    • Google Scholar
41. 1. Graybiel AM
    2. Ragsdale CW

    (1978) Histochemically distinct compartments in the striatum of human, monkeys, and cat demonstrated by acetylthiocholinesterase staining

    PNAS 75:5723–5726.

    https://doi.org/10.1073/pnas.75.11.5723

    • PubMed
    • Google Scholar
42. 1. Graybiel AM
    2. Hickey TL

    (1982) Chemospecificity of ontogenetic units in the striatum: demonstration by combining [3H]thymidine neuronography and histochemical staining

    PNAS 79:198–202.

    https://doi.org/10.1073/pnas.79.1.198

    • PubMed
    • Google Scholar
43. 1. Graybiel AM

    (1984) Correspondence between the dopamine islands and striosomes of the mammalian striatum

    Neuroscience 13:1157–1187.

    https://doi.org/10.1016/0306-4522(84)90293-8

    • PubMed
    • Google Scholar
44. Book

    1. Graybiel AM

    (2010) Templates for neural dynamics in the striatum: Striosomes and matrisomes

    In: Shepherd G, Grillner S, editors. Handbook of Brain Microcircuits. New York: Oxford University Press. pp. 120–126.

    https://doi.org/10.1093/med/9780195389883.003.0012

    • Google Scholar
45. 1. Hagimoto K
    2. Takami S
    3. Murakami F
    4. Tanabe Y

    (2017) Distinct migratory behaviors of striosome and matrix cells underlying the mosaic formation in the developing striatum

    Journal of Comparative Neurology 525:794–817.

    https://doi.org/10.1002/cne.24096

    • PubMed
    • Google Scholar
46. 1. Hamid AA
    2. Pettibone JR
    3. Mabrouk OS
    4. Hetrick VL
    5. Schmidt R
    6. Vander Weele CM
    7. Kennedy RT
    8. Aragona BJ
    9. Berke JD

    (2016) Mesolimbic dopamine signals the value of work

    Nature Neuroscience 19:117–126.

    https://doi.org/10.1038/nn.4173

    • PubMed
    • Google Scholar
47. 1. Harvey CD
    2. Coen P
    3. Tank DW

    (2012) Choice-specific sequences in parietal cortex during a virtual-navigation decision task

    Nature 484:62–68.

    https://doi.org/10.1038/nature10918

    • PubMed
    • Google Scholar
48. 1. Hikosaka O
    2. Kim HF
    3. Yasuda M
    4. Yamamoto S

    (2014) Basal ganglia circuits for reward value-guided behavior

    Annual Review of Neuroscience 37:289–306.

    https://doi.org/10.1146/annurev-neuro-071013-013924

    • PubMed
    • Google Scholar
49. 1. Howe MW
    2. Tierney PL
    3. Sandberg SG
    4. Phillips PE
    5. Graybiel AM

    (2013) Prolonged dopamine signalling in striatum signals proximity and value of distant rewards

    Nature 500:575–579.

    https://doi.org/10.1038/nature12475

    • PubMed
    • Google Scholar
50. 1. Howe MW
    2. Dombeck DA

    (2016) Rapid signalling in distinct dopaminergic axons during locomotion and reward

    Nature 535:505–510.

    https://doi.org/10.1038/nature18942

    • PubMed
    • Google Scholar
51. 1. Jiménez-Castellanos J
    2. Graybiel AM

    (1989) Compartmental origins of striatal efferent projections in the cat

    Neuroscience 32:297–321.

    https://doi.org/10.1016/0306-4522(89)90080-8

    • PubMed
    • Google Scholar
52. 1. Jin X
    2. Costa RM

    (2010) Start/stop signals emerge in nigrostriatal circuits during sequence learning

    Nature 466:457–462.

    https://doi.org/10.1038/nature09263

    • PubMed
    • Google Scholar
53. 1. Jog MS
    2. Kubota Y
    3. Connolly CI
    4. Hillegaart V
    5. Graybiel AM

    (1999) Building neural representations of habits

    Science 286:1745–1749.

    https://doi.org/10.1126/science.286.5445.1745

    • PubMed
    • Google Scholar
54. 1. Kaifosh P
    2. Lovett-Barron M
    3. Turi GF
    4. Reardon TR
    5. Losonczy A

    (2013) Septo-hippocampal GABAergic signaling across multiple modalities in awake mice

    Nature Neuroscience 16:1182–1184.

    https://doi.org/10.1038/nn.3482

    • PubMed
    • Google Scholar
55. 1. Keiflin R
    2. Janak PH

    (2015) Dopamine prediction errors in reward learning and addiction: from theory to neural circuitry

    Neuron 88:247–263.

    https://doi.org/10.1016/j.neuron.2015.08.037

    • PubMed
    • Google Scholar
56. 1. Kelly SM
    2. Raudelas R
    3. He M
    4. Lee J
    5. Kim Y
    6. Gibb LG
    7. Wu P
    8. Matho K
    9. Osten P
    10. Graybiel AM
    11. Huang ZJ

    (2017)

    Radial glial lineage progression and differential intermediate progenitor amplification underlie striatal compartments and circuit organization

    Neuron in Revision.

    • Google Scholar
57. 1. Kim EJ
    2. Ables JL
    3. Dickel LK
    4. Eisch AJ
    5. Johnson JE

    (2011) Ascl1 (Mash1) defines cells with long-term neurogenic potential in subgranular and subventricular zones in adult mouse brain

    PLoS One 6:e18472.

    https://doi.org/10.1371/journal.pone.0018472

    • PubMed
    • Google Scholar
58. 1. Klaus A
    2. Martins GJ
    3. Paixao VB
    4. Zhou P
    5. Paninski L
    6. Costa RM

    (2017) The spatiotemporal organization of the striatum encodes action space

    Neuron 95:1171–1180.

    https://doi.org/10.1016/j.neuron.2017.08.015

    • PubMed
    • Google Scholar
59. 1. Kreitzer AC
    2. Malenka RC

    (2008) Striatal plasticity and basal ganglia circuit function

    Neuron 60:543–554.

    https://doi.org/10.1016/j.neuron.2008.11.005

    • PubMed
    • Google Scholar
60. 1. Langer LF
    2. Graybiel AM

    (1989) Distinct nigrostriatal projection systems innervate striosomes and matrix in the primate striatum

    Brain Research 498:344–350.

    https://doi.org/10.1016/0006-8993(89)91114-1

    • PubMed
    • Google Scholar
61. 1. Lopez-Huerta VG
    2. Nakano Y
    3. Bausenwein J
    4. Jaidar O
    5. Lazarus M
    6. Cherassse Y
    7. Garcia-Munoz M
    8. Arbuthnott G

    (2016) The neostriatum: two entities, one structure?

    Brain Structure and Function 221:1737–1749.

    https://doi.org/10.1007/s00429-015-1000-4

    • PubMed
    • Google Scholar
62. 1. Lovett-Barron M
    2. Kaifosh P
    3. Kheirbek MA
    4. Danielson N
    5. Zaremba JD
    6. Reardon TR
    7. Turi GF
    8. Hen R
    9. Zemelman BV
    10. Losonczy A

    (2014) Dendritic inhibition in the hippocampus supports fear learning

    Science 343:857–863.

    https://doi.org/10.1126/science.1247485

    • PubMed
    • Google Scholar
63. 1. Luo Z
    2. Volkow ND
    3. Heintz N
    4. Pan Y
    5. Du C

    (2011) Acute cocaine induces fast activation of D1 receptor and progressive deactivation of D2 receptor striatal neurons: in vivo optical microprobe [Ca2+]i imaging

    Journal of Neuroscience 31:13180–13190.

    https://doi.org/10.1523/JNEUROSCI.2369-11.2011

    • PubMed
    • Google Scholar
64. 1. Madisen L
    2. Zwingman TA
    3. Sunkin SM
    4. Oh SW
    5. Zariwala HA
    6. Gu H
    7. Ng LL
    8. Palmiter RD
    9. Hawrylycz MJ
    10. Jones AR
    11. Lein ES
    12. Zeng H

    (2010) A robust and high-throughput Cre reporting and characterization system for the whole mouse brain

    Nature Neuroscience 13:133–140.

    https://doi.org/10.1038/nn.2467

    • PubMed
    • Google Scholar
65. 1. Matsumoto M
    2. Hikosaka O

    (2007) Lateral habenula as a source of negative reward signals in dopamine neurons

    Nature 447:1111–1115.

    https://doi.org/10.1038/nature05860

    • PubMed
    • Google Scholar
66. 1. Mink JW

    (1996) The basal ganglia: focused selection and inhibition of competing motor programs

    Progress in Neurobiology 50:381–425.

    https://doi.org/10.1016/S0301-0082(96)00042-1

    • PubMed
    • Google Scholar
67. 1. Mizrahi A
    2. Crowley JC
    3. Shtoyerman E
    4. Katz LC

    (2004) High-resolution in vivo imaging of hippocampal dendrites and spines

    Journal of Neuroscience 24:3147–3151.

    https://doi.org/10.1523/JNEUROSCI.5218-03.2004

    • PubMed
    • Google Scholar
68. 1. Nelson AB
    2. Kreitzer AC

    (2014) Reassessing models of basal ganglia function and dysfunction

    Annual Review of Neuroscience 37:117–135.

    https://doi.org/10.1146/annurev-neuro-071013-013916

    • PubMed
    • Google Scholar
69. 1. Newman H
    2. Liu FC
    3. Graybiel AM

    (2015) Dynamic ordering of early generated striatal cells destined to form the striosomal compartment of the striatum

    Journal of Comparative Neurology 523:943–962.

    https://doi.org/10.1002/cne.23725

    • PubMed
    • Google Scholar
70. 1. Niv Y
    2. Daw ND
    3. Joel D
    4. Dayan P

    (2007) Tonic dopamine: opportunity costs and the control of response vigor

    Psychopharmacology 191:507–520.

    https://doi.org/10.1007/s00213-006-0502-4

    • PubMed
    • Google Scholar
71. 1. Oyama K
    2. Hernádi I
    3. Iijima T
    4. Tsutsui K

    (2010) Reward prediction error coding in dorsal striatal neurons

    Journal of Neuroscience 30:11447–11457.

    https://doi.org/10.1523/JNEUROSCI.1719-10.2010

    • PubMed
    • Google Scholar
72. 1. Oyama K
    2. Tateyama Y
    3. Hernádi I
    4. Tobler PN
    5. Iijima T
    6. Tsutsui K

    (2015) Discrete coding of stimulus value, reward expectation, and reward prediction error in the dorsal striatum

    Journal of Neurophysiology 114:2600–2615.

    https://doi.org/10.1152/jn.00097.2015

    • PubMed
    • Google Scholar
73. 1. Packard MG
    2. Knowlton BJ

    (2002) Learning and memory functions of the Basal Ganglia

    Annual Review of Neuroscience 25:563–593.

    https://doi.org/10.1146/annurev.neuro.25.112701.142937

    • PubMed
    • Google Scholar
74. 1. Parthasarathy HB
    2. Schall JD
    3. Graybiel AM

    (1992)

    Distributed but convergent ordering of corticostriatal projections: analysis of the frontal eye field and the supplementary eye field in the macaque monkey

    Journal of Neuroscience 12:4468–4488.

    • PubMed
    • Google Scholar
75. 1. Ragsdale CW
    2. Graybiel AM

    (1988) Fibers from the basolateral nucleus of the amygdala selectively innervate striosomes in the caudate nucleus of the cat

    The Journal of Comparative Neurology 269:506–522.

    https://doi.org/10.1002/cne.902690404

    • PubMed
    • Google Scholar
76. 1. Ragsdale CW
    2. Graybiel AM

    (1990) A simple ordering of neocortical areas established by the compartmental organization of their striatal projections

    PNAS 87:6196–6199.

    https://doi.org/10.1073/pnas.87.16.6196

    • PubMed
    • Google Scholar
77. 1. Rajakumar N
    2. Elisevich K
    3. Flumerfelt BA

    (1993) Compartmental origin of the striato-entopeduncular projection in the rat

    The Journal of Comparative Neurology 331:286–296.

    https://doi.org/10.1002/cne.903310210

    • PubMed
    • Google Scholar
78. 1. Redgrave P
    2. Prescott TJ
    3. Gurney K

    (1999) The basal ganglia: a vertebrate solution to the selection problem?

    Neuroscience 89:1009–1023.

    https://doi.org/10.1016/S0306-4522(98)00319-4

    • PubMed
    • Google Scholar
79. 1. Rikhye RV
    2. Sur M

    (2015) Spatial correlations in natural scenes modulate response reliability in mouse visual cortex

    Journal of Neuroscience 35:14661–14680.

    https://doi.org/10.1523/JNEUROSCI.1660-15.2015

    • PubMed
    • Google Scholar
80. 1. Rueda-Orozco PE
    2. Robbe D

    (2015) The striatum multiplexes contextual and kinematic information to constrain motor habits execution

    Nature Neuroscience 18:453–460.

    https://doi.org/10.1038/nn.3924

    • PubMed
    • Google Scholar
81. 1. Saka E
    2. Goodrich C
    3. Harlan P
    4. Madras BK
    5. Graybiel AM

    (2004) Repetitive behaviors in monkeys are linked to specific striatal activation patterns

    Journal of Neuroscience 24:7557–7565.

    https://doi.org/10.1523/JNEUROSCI.1072-04.2004

    • PubMed
    • Google Scholar
82. 1. Salamone JD
    2. Correa M

    (2012) The mysterious motivational functions of mesolimbic dopamine

    Neuron 76:470–485.

    https://doi.org/10.1016/j.neuron.2012.10.021

    • PubMed
    • Google Scholar
83. 1. Salinas AG
    2. Davis MI
    3. Lovinger DM
    4. Mateo Y

    (2016) Dopamine dynamics and cocaine sensitivity differ between striosome and matrix compartments of the striatum

    Neuropharmacology 108:275–283.

    https://doi.org/10.1016/j.neuropharm.2016.03.049

    • PubMed
    • Google Scholar
84. 1. Samejima K
    2. Ueda Y
    3. Doya K
    4. Kimura M

    (2005) Representation of action-specific reward values in the striatum

    Science 310:1337–1340.

    https://doi.org/10.1126/science.1115270

    • PubMed
    • Google Scholar
85. 1. Sato K
    2. Sumi-Ichinose C
    3. Kaji R
    4. Ikemoto K
    5. Nomura T
    6. Nagatsu I
    7. Ichinose H
    8. Ito M
    9. Sako W
    10. Nagahiro S
    11. Graybiel AM
    12. Goto S

    (2008) Differential involvement of striosome and matrix dopamine systems in a transgenic model of dopa-responsive dystonia

    PNAS 105:12551–12556.

    https://doi.org/10.1073/pnas.0806065105

    • PubMed
    • Google Scholar
86. 1. Sato M
    2. Kawano M
    3. Yanagawa Y
    4. Hayashi Y

    (2016) In vivo two-photon imaging of striatal neuronal circuits in mice

    Neurobiology of Learning and Memory 135:146–151.

    https://doi.org/10.1016/j.nlm.2016.07.006

    • PubMed
    • Google Scholar
87. 1. Schultz W
    2. Dayan P
    3. Montague PR

    (1997) A neural substrate of prediction and reward

    Science 275:1593–1599.

    https://doi.org/10.1126/science.275.5306.1593

    • PubMed
    • Google Scholar
88. 1. Schultz W

    (2016) Dopamine reward prediction-error signalling: a two-component response

    Nature Reviews Neuroscience 17:183–195.

    https://doi.org/10.1038/nrn.2015.26

    • PubMed
    • Google Scholar
89. 1. Slotnick B

    (2009) A simple 2-transistor touch or lick detector circuit

    Journal of the Experimental Analysis of Behavior 91:253–255.

    https://doi.org/10.1901/jeab.2009.91-253

    • PubMed
    • Google Scholar
90. 1. Smith KS
    2. Graybiel AM

    (2013) A dual operator view of habitual behavior reflecting cortical and striatal dynamics

    Neuron 79:361–374.

    https://doi.org/10.1016/j.neuron.2013.05.038

    • PubMed
    • Google Scholar
91. 1. Smith JB
    2. Klug JR
    3. Ross DL
    4. Howard CD
    5. Hollon NG
    6. Ko VI
    7. Hoffman H
    8. Callaway EM
    9. Gerfen CR
    10. Jin X

    (2016) Genetic-based dissection unveils the inputs and outputs of striatal patch and matrix compartments

    Neuron 91:1069–1084.

    https://doi.org/10.1016/j.neuron.2016.07.046

    • PubMed
    • Google Scholar
92. 1. Stalnaker TA
    2. Calhoon GG
    3. Ogawa M
    4. Roesch MR
    5. Schoenbaum G

    (2012) Reward prediction error signaling in posterior dorsomedial striatum is action specific

    Journal of Neuroscience 32:10296–10305.

    https://doi.org/10.1523/JNEUROSCI.0832-12.2012

    • PubMed
    • Google Scholar
93. 1. Stephenson-Jones M
    2. Yu K
    3. Ahrens S
    4. Tucciarone JM
    5. van Huijstee AN
    6. Mejia LA
    7. Penzo MA
    8. Tai LH
    9. Wilbrecht L
    10. Li B

    (2016) A basal ganglia circuit for evaluating action outcomes

    Nature 539:289–293.

    https://doi.org/10.1038/nature19845

    • PubMed
    • Google Scholar
94. 1. Tai LH
    2. Lee AM
    3. Benavidez N
    4. Bonci A
    5. Wilbrecht L

    (2012) Transient stimulation of distinct subpopulations of striatal neurons mimics changes in action value

    Nature Neuroscience 15:1281–1289.

    https://doi.org/10.1038/nn.3188

    • PubMed
    • Google Scholar
95. 1. Tajima K
    2. Fukuda T

    (2013) Region-specific diversity of striosomes in the mouse striatum revealed by the differential immunoreactivities for mu-opioid receptor, substance P, and enkephalin

    Neuroscience 241:215–228.

    https://doi.org/10.1016/j.neuroscience.2013.03.012

    • PubMed
    • Google Scholar
96. 1. Taniguchi H
    2. He M
    3. Wu P
    4. Kim S
    5. Paik R
    6. Sugino K
    7. Kvitsiani D
    8. Kvitsani D
    9. Fu Y
    10. Lu J
    11. Lin Y
    12. Miyoshi G
    13. Shima Y
    14. Fishell G
    15. Nelson SB
    16. Huang ZJ

    (2011) A resource of Cre driver lines for genetic targeting of GABAergic neurons in cerebral cortex

    Neuron 71:995–1013.

    https://doi.org/10.1016/j.neuron.2011.07.026

    • PubMed
    • Google Scholar
97. 1. Thorn CA
    2. Atallah H
    3. Howe M
    4. Graybiel AM

    (2010) Differential dynamics of activity changes in dorsolateral and dorsomedial striatal loops during learning

    Neuron 66:781–795.

    https://doi.org/10.1016/j.neuron.2010.04.036

    • PubMed
    • Google Scholar
98. 1. Tippett LJ
    2. Waldvogel HJ
    3. Thomas SJ
    4. Hogg VM
    5. van Roon-Mom W
    6. Synek BJ
    7. Graybiel AM
    8. Faull RL

    (2007) Striosomes and mood dysfunction in Huntington's disease

    Brain 130:206–221.

    https://doi.org/10.1093/brain/awl243

    • PubMed
    • Google Scholar
99. 1. Tseng Q
    2. Wang I
    3. Duchemin-Pelletier E
    4. Azioune A
    5. Carpi N
    6. Gao J
    7. Filhol O
    8. Piel M
    9. Théry M
    10. Balland M

    (2011) A new micropatterning method of soft substrates reveals that different tumorigenic signals can promote or reduce cell contraction levels

    Lab on a Chip 11:2231–2240.

    https://doi.org/10.1039/c0lc00641f

    • PubMed
    • Google Scholar
100. 1. Walker RH
     2. Arbuthnott GW
     3. Baughman RW
     4. Graybiel AM

     (1993) Dendritic domains of medium spiny neurons in the primate striatum: relationships to striosomal borders

     The Journal of Comparative Neurology 337:614–628.

     https://doi.org/10.1002/cne.903370407

     • PubMed
     • Google Scholar
101. 1. Watabe-Uchida M
     2. Zhu L
     3. Ogawa SK
     4. Vamanrao A
     5. Uchida N

     (2012) Whole-brain mapping of direct inputs to midbrain dopamine neurons

     Neuron 74:858–873.

     https://doi.org/10.1016/j.neuron.2012.03.017

     • PubMed
     • Google Scholar
102. 1. Watabe-Uchida M
     2. Eshel N
     3. Uchida N

     (2017) Neural circuitry of reward prediction error

     Annual Review of Neuroscience 40:373–394.

     https://doi.org/10.1146/annurev-neuro-072116-031109

     • PubMed
     • Google Scholar
103. 1. Yin HH
     2. Knowlton BJ

     (2006) The role of the basal ganglia in habit formation

     Nature Reviews Neuroscience 7:464–476.

     https://doi.org/10.1038/nrn1919

     • PubMed
     • Google Scholar

Author details

1. Bernard Bloem

   1. McGovern Institute for Brain Research, Massachusetts Institute of Technology, Cambridge, United States
   2. Department of Brain and Cognitive Sciences, Massachusetts Institute of Technology, Cambridge, United States

   Contribution

   Conceptualization, Software, Formal analysis, Funding acquisition, Investigation, Visualization, Methodology, Writing—original draft, Writing—review and editing

   Contributed equally with

   Rafiq Huda

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-0930-580X

2. Rafiq Huda

   1. Department of Brain and Cognitive Sciences, Massachusetts Institute of Technology, Cambridge, United States
   2. Picower Institute for Learning and Memory, Massachusetts Institute of Technology, Cambridge, United States

   Contribution

   Conceptualization, Software, Formal analysis, Funding acquisition, Investigation, Visualization, Methodology, Writing—original draft, Writing—review and editing

   Contributed equally with

   Bernard Bloem

   Competing interests

   No competing interests declared

3. Mriganka Sur

   1. Department of Brain and Cognitive Sciences, Massachusetts Institute of Technology, Cambridge, United States
   2. Picower Institute for Learning and Memory, Massachusetts Institute of Technology, Cambridge, United States

   Contribution

   Resources, Supervision, Funding acquisition, Writing—review and editing

   Competing interests

   No competing interests declared

4. Ann M Graybiel

   1. McGovern Institute for Brain Research, Massachusetts Institute of Technology, Cambridge, United States
   2. Department of Brain and Cognitive Sciences, Massachusetts Institute of Technology, Cambridge, United States

   Contribution

   Conceptualization, Resources, Supervision, Funding acquisition, Methodology, Writing—original draft, Writing—review and editing

   For correspondence

   graybiel@MIT.EDU

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-4326-7720

• 4,616

  views

• 681

  downloads

• 65

  citations

Views, downloads and citations are aggregated across all versions of this paper published by eLife.