Cortico-striatal action control inherent of opponent cognitive-motivational styles

Reviewer #1 (Public review):

Summary:

The authors measured glutamate transients in the DMS of rats as they performed an action selection task. They identified diverse patterns of behavior and glutamate dynamics depending on the pre-existing behavioral phenotype of the rat (sign tracker or goal tracker). Using pathway-specific DREADDs, they showed that these behavioral phenotypes and their corresponding glutamate transients were differentially dependent on input from the prelimbic cortex to the DMS.

Strengths:

Overall there are some very interesting results that make an important contribution to the field. Notably, the results seem to point to differential recruitment of the PL-DMS pathway in goal-tracking vs sign-tracking behaviors.

Weaknesses:

There is a lot of missing information and data that should be reported/presented to allow a complete understanding of the findings and what was done. The writing of the manuscript was mostly quite clear, however, there are some specific leaps in logic that require more elaboration, and the focus at the start and end on cholinergic neurons and Parkinson's disease are, at the moment, confusing and require more justification.

Reviewer #2 (Public review):

Summary:

The authors aimed to determine whether goal-directed and cue-driven attentional strategies (goal- and sign-tracking phenotypes) were associated with variation in cued motor responses and dorsomedial striatal (DMS) glutamate transmission. They used a treadmill task in which cues indicated whether rats should turn or stop to receive a reward. They collected and analyzed several behavioral measures related to task performance with a focus on turns (performance, latency, duration) for which there are more measures than for stops. First, they established that goal-trackers perform better than sign-trackers in post-criterion turn performance (cued turns completed) and turn initiation. They used glutamate sensors to measure glutamate transmission in DMS. They performed analyses on glutamate traces that suggest phasic glutamate DMS dynamics to cues were primarily associated with successful turn performance and were more characteristic of goal-trackers (ie. rats with "goal-directed" attentional strategy). Smaller and more frequent DMS glutamate peaks were associated with other task events, cued misses (missed turns), cued stops, and reward delivery and were more characteristic of sign-trackers (i.e. rats with "cue-driven" attentional strategies). Consistent with the reported glutamate findings, chemogenetic inhibition of prelimbic-DMS glutamate transmission had an effect on goal-trackers' turn performance without affecting sign-trackers' performance in the treadmill task.

Strengths:

The power of the sign- and goal-tracking model to account for neurobiological and behavioral variability is critically important to the field's understanding of the heterogeneity of the brain in health and disease. The approach and methodology are sound in their contribution to this important effort.

The authors establish behavioral differences, measure a neurobiological correlate of relevance, and then manipulate that correlate in a broader circuitry and show a causal role in behavior that is consistent with neurobiological measurements and phenotypic differences.

Sophisticated analyses provide a compelling description of the authors' observations.

Weaknesses:

It is challenging to assess what is considered the "n" in each analysis (trial, session, rat, trace (averaged across a session or single trial)). Representative glutamate traces (n = 5 traces (out of hundreds of recorded traces)) are used to illustrate a central finding, while more conventional trial-averaged population activity traces are not presented or analyzed. The latter would provide much-needed support for the reported findings and conclusions. Digging deeper into the methods, results, and figure legends, provides some answers to the reader, but much can be done to clarify what each data point represents and, in particular, how each rat contributes to a reported finding (ie. single trial-averaged trace per session for multiple sessions, or dozens of single traces across multiple sessions).

Representative traces should in theory be consistent with population averages within phenotype, and if not, discussion of such inconsistencies would enrich the conclusions drawn from the study. In particular, population traces of the phasic cue response in GT may resemble the representative peak examples, while smaller irregular peaks of ST may be missed in a population average (averaged prolonged elevation) and could serve as a rationale for more sophisticated analyses of peak probability presented subsequently.

Reviewer #3 (Public review):

Summary:

Avila and colleagues investigate the role of glutamate signaling in the dorsomedial striatum in a treadmill-based task where rats learn to turn or stop their walking based on learning cue-associations that allow them to acquire rewards. Phenotypic variation in Pavlovian conditioned sign and goal-tracking behavior was examined, where behavioral differences in stopping and turning were observed. Glutamate signals in the DMS were recorded during the treadmill task and were related to features of cue-controlled movement, with a stronger relationship seen for goal trackers. Finally, chemogenic inhibition of prelimbic neurons projecting to the DMS (the predicted source of those glutamate signals), preferentially affected cued movement in goal trackers. The authors couch these experiments in the context of cognitive control-attentional mechanisms, movement disorders, and individual differences in cue reactivity.

Strengths:

Overall these studies are interesting and are of general relevance to a number of research questions in neurology and psychiatry. The assessment of the intersection of individual differences in cue-related learning strategies with movement-related questions - in this case, cued turning behavior - is an interesting and understudied question. The link between this work and growing notions of corticostriatal control of action selection makes it timely.

Weaknesses:

The clarity of the manuscript could be improved in several places, including in the graphical visualization of data. It is sometimes difficult to interpret the glutamate results, as presented, in the context of specific behavior, for example.

Author response:

Reviewer #1 (Public Review):

Strengths:

Overall there are some very interesting results that make an important contribution to the field. Notably, the results seem to point to differential recruitment of the PL-DMS pathway in goal-tracking vs sign-tracking behaviors.

Thank you.

Weaknesses:

There is a lot of missing information and data that should be reported/presented to allow a complete understanding of the findings and what was done. The writing of the manuscript was mostly quite clear, however, there are some specific leaps in logic that require more elaboration, and the focus at the start and end on cholinergic neurons and Parkinson's disease are, at the moment, confusing and require more justification.

In the revised paper, we provide additional information in support of results and clarify procedures and findings. Furthermore, we expand the discussion of the proposed interpretational framework that suggests that the contrasts between the cortical-striatal processing of movement cues in sign- versus goal trackers are related to previously established, parallel contrasts in the cortical cholinergic detection of attention-demanding cues.

Reviewer #2 (Public review):

Strengths:

The power of the sign- and goal-tracking model to account for neurobiological and behavioral variability is critically important to the field's understanding of the heterogeneity of the brain in health and disease. The approach and methodology are sound in their contribution to this important effort.

The authors establish behavioral differences, measure a neurobiological correlate of relevance, and then manipulate that correlate in a broader circuitry and show a causal role in behavior that is consistent with neurobiological measurements and phenotypic differences.

Sophisticated analyses provide a compelling description of the authors' observations.

Thank you.

Weaknesses:

It is challenging to assess what is considered the "n" in each analysis (trial, session, rat, trace (averaged across a session or single trial)). Representative glutamate traces (n = 5 traces (out of hundreds of recorded traces)) are used to illustrate a central finding, while more conventional trial-averaged population activity traces are not presented or analyzed. The latter would provide much-needed support for the reported findings and conclusions. Digging deeper into the methods, results, and figure legends, provides some answers to the reader, but much can be done to clarify what each data point represents and, in particular, how each rat contributes to a reported finding (ie. single trial-averaged trace per session for multiple sessions, or dozens of single traces across multiple sessions).

Representative traces should in theory be consistent with population averages within phenotype, and if not, discussion of such inconsistencies would enrich the conclusions drawn from the study. In particular, population traces of the phasic cue response in GT may resemble the representative peak examples, while smaller irregular peaks of ST may be missed in a population average (averaged prolonged elevation) and could serve as a rationale for more sophisticated analyses of peak probability presented subsequently.

Figures 5c-f depict individual data from all rats and trials. For all major analyses, the revised manuscript consolidates information about the number of rats per phenotype and sex, and the number of trials contributed by individual rats, in the result section.

As detailed in the section on statistical methods, and as mentioned by the reviewer under Strengths, we used advanced statistical methods to assure that data from individual animals contribute equally to the overall result, and to minimize the possibility that an inordinate number of trials obtained from just one or a couple of rats biased the overall analysis.

As the reviewer correctly pointed out, we have chosen not to show trial- or subject-averaged traces to illustrate glutamate release dynamics across trials. The present analyses focus on peak glutamate concentrations, the number of peaks, and the timing of peaks relative to a task cue or a behavioral event. Within a response bin, such as the 2-s period following turn cues, glutamate peaks – as defined in Methods - occur at variable times relative to cue onset. Averaging traces over a population of rats or trials would “wash-out” the phenotype- and task event-dependent patterns of glutamate peaks, yielding, for example, a single, nearly 2-s long plateau for cue-locked glutamate recordings from STs (Figure 5b). Thus, subject- or trial-averaged traces would not illustrate the major findings described in this paper and would rather be uninformative. As already mentioned, individual data from all subjects and trials are shown in Figs 5c-f.

Reviewer #3 (Public review):

Strengths:

Overall these studies are interesting and are of general relevance to a number of research questions in neurology and psychiatry. The assessment of the intersection of individual differences in cue-related learning strategies with movement-related questions - in this case, cued turning behavior - is an interesting and understudied question. The link between this work and growing notions of corticostriatal control of action selection makes it timely.

Thank you.

Weaknesses:

The clarity of the manuscript could be improved in several places, including in the graphical visualization of data. It is sometimes difficult to interpret the glutamate results, as presented, in the context of specific behavior, for example.

We appreciate the reviewer’s concerns about the complexity of some of the graphics, particularly the results from the arguably innovative analysis illustrated in Figure 6. Figure 6 illustrates that the likelihood of a cued turn can be predicted based on single and combined glutamate peak characteristics. The revised legend for this figure provides additional information and examples to ease the readers’ access to this figure.