The rostral intralaminar nuclear complex of the thalamus supports striatally mediated action reinforcement

Essential revisions:

Reviewer #1 (Recommendations for the authors):

While overall this is a well-executed study that uses a variety of behavior tasks to demonstrate the activity of DS-projecting rILN neurons. Yet, the different parts of this study (electrophysiology, behavior, tracing) feel somewhat disjointed, and additional control experiments would be needed to substantiate some of the claims made.

– The statement that the authors "discovered that rILN neurons projecting to the DS integrated information from several subcortical and cortical sources" (abstract lines 39 – 41) is misleading and not fully supported by the presented data. While the authors demonstrate that rILN neurons can receive input projections from several subcortical and cortical areas, they do not show whether these projections converge onto the same individual neurons in rILN, and more importantly, they do not show whether "information" is carried by these projections or integrated into rILN neurons during behavior. Recordings and manipulation experiments of these subcortical and cortical inputs to rILN would be necessary to support such a claim.

Thank you for identifying this inadvertent misrepresentation in our conclusions. We have revised this sentence to state: “Here we discovered that rILN neurons projecting to the DS are innervated by a range of cortical and subcortical afferents and that rILNDS neurons stably signaled at two time points in mice performing an action sequence task reinforced by sucrose reward: action initiation and reward acquisition” (lines 39 – 42).

– While fiber photometry is a powerful and suitable approach to measuring neuronal calcium signals, the authors should at least discuss the possibility that measuring calcium photometry signals at the cell soma region might not reflect neuronal firing activity (Legaria et al., 2022), and thus this activity might not indicate rILN neuron signaling to the striatum. Fiber photometry recordings of the axons of these projection neurons in DS would provide more definitive evidence for this signaling.

Therefore, it would be helpful if the authors could tune down the conclusions related to these areas.

Thank you for bringing our attention to this publication, which is now cited in our discussion. We now include acknowledgement of this potential limitation with photometric recordings in our manuscript (lines 292 – 296).

Other suggestions include:

– Please clarify if DS projecting neurons from all rILN nuclei show the reported action initiation and reward acquisition activity or only CL neurons.

The arrangement of the rILN nuclei presents a technical challenge for experiments attempting to selectively record from or manipulate a single nucleus in this grouping. Based on our findings that the three nuclei do not differ in electrophysiological properties, we approached the in vivo experiments with the intent to target the rILN as a unit. As the reviewer points out, the medial-lateral coordinate for optic fiber placement tended to align above the CL and PC nuclei. However, variability in fiber placement and spread of light within tissue resulted in inclusion of CM activity as well. Given the spread of light through tissue (Shin, et al., 2016; PMID: 27895987), it would be very difficult to confidently determine from histology which photometry recordings were primarily obtained from CL vs PC vs CM neuronal activity. We agree with the reviewer that these three nuclei may differently signal during reward-driven behavior. Our di-synaptic tracing study supports this possibility as it revealed unique afferent connectivity to rILNDS projecting neurons. We now mention this limitation of our approach in the discussion (lines 324 – 330).

– Along similar lines, to what extent of rILN was targeted for optogenetic activation and inhibition? It seems that the authors implanted a total of 4 optic fibers, two on each side (please clarify in methods). What was the reasoning behind this? Please show that only rILN and not PF was activated/inhibited.

We apologize for the confusion in our description of this method. For our optogenetic experiments, we infused viruses at four locations (bilateral striatum and rILN) and implanted only two fibers (bilateral rILN) to selectively target striatally-projecting rILN neurons. We have added clarification on this detail to the methods section.

To prevent inadvertent modulation of Pf neurons, we used virus injection coordinates and volumes that prevented viral spread to the Pf and furthermore implanted the optic fibers in the more rostral regions of the rILN. We histologically confirmed viral expression and fiber placement for all mice and excluded any mice with viral spread to the Pf or off-target fiber placement. We include these criteria for post-hoc exclusion in the methods.

– The transsynaptic tracing experiments – the cell count quantifications in CM, PC, and CL are needed.

Thank you for this suggestion, we now include cell counts for 2 cases per investigated afferent (Figure 5 Supplement 2).

– Why is the injection site for the retrograde cre-dependent tdTomato AAV (Figure 5 middle left panels) showing expression? Please clarify. Is the cre coming through transsynaptic AAV1 from direct projections of each AAV1 injection site (AAV1 is not supposed to spread across a second synapse)? The diagrams suggest that not all regions (e.g. SUM or SC) have direct projections to DS.

We apologize for this confusion. The tdTomato fluorophore expression observed in the striatum may arise from several possible circuit configurations. To survey just a couple: (1) tdTomato expression in the DS arises from direct projections from the afferent bypassing the thalamus (e.g. ipsilateral ACCStriatum), which would result in labeled striatal somata (ACC pyramidal neurons delivering AAV1-cre to an MSN, and those local MSN collaterals retrogradely picking up rAAV-DIO-tdtomato) and ACC labeled axon terminals in the DS (ACC interneurons delivering AAV1-cre to DS-projecting ACC pyramidal neurons that pick up rAAV-DIO-tdtomato); (2) terminal projections arising from the labeled rILN neurons shown in the middle-right panels (i.e. ACCrILNStriatum).

Reviewer #3 (Recommendations for the authors):

1) The final neuroanatomical tracing studies are not very well integrated with the rest of the study. The experiments are introduced as a comparison between the striatum- and cortex-projecting neurons, but the latter experiment was not done. Without distinguishing these two populations or possibly comparing the inputs to intralaminar versus parafascicular nuclei, the experiment does not offer a lot to the overall conclusions. The authors should either do the additional work to make a meaningful conclusion from these experiments or exclude them from the current manuscript.

Thank you for identifying this weakness in our introduction for the tracing study. We agree that it was inadvertently misleading to reference the rILNcortical projection without including comparable experimental results. As such, we have rewritten our rationale for the experiment in the manuscript (lines 220-222).

We do feel that this experiment assessing rILNstriatal afferents has merit and sets the foundation to assess differences between rILN subpopulations with unique postsynaptic targets. Given that our manuscript exclusively characterizes the rILNstriatal projection in both physiology and behavior, we thought it was fitting to include a qualitative survey of afferents that may contribute to the behaviorally dynamic activity of this circuit. Whereas future investigators may seek to compare the connectivity and function of the rILN with other thalamic nuclei, such as the parafascicular nucleus, we respectively feel that it extends beyond the scope of the current study.

2) In line 290, the authors state that "in addition to dopamine terminal signaling coinciding with reward presentation (Howe & Dombeck, 2016); an activity that is not reflected in cell body firing", however, there are plenty of examples of reward-responsive VTA dopamine neurons from single-unit recordings in SNc/VTA (e.g., PMID: 22258508). The authors should also be very careful in their interpretation of the cited work. The Howe and Dombek paper do not claim that movement and reward-related signals are not encoded in the midbrain, but rather that they are differentially encoded by distinct dopamine neurons subpopulations (see their more recent work: bioRxiv: 2022.06.20.496872) arguing that the differences between axonal and somatic dopamine cell responses is due to genetic identity rather than terminal modulation. Undoubtedly terminal modulation is a key player, and this idea is recently gaining a lot of traction, but the dopamine signal in the striatum likely reflects a combination of somatic activity and terminal modulation, and not just the latter.

Thank you for pointing out our over-interpretation of these studies, we have modified our discussion to emphasize that both dopamine neuron somatic activity and terminal modulation likely contribute to reward-related signaling (lines 283 – 289).

3) The authors show that activity in rILN→striatum projection neurons correlates with movement in S2, and their previous study showed that chemogenetically inhibiting all rILN neurons suppresses movement. How do the optogenetic manipulations here affect spontaneous movement and might this explain their observed effects on goal-directed operant responding in Figure 4?

This a great question. We analyzed a number of parameters in the FR5 task to examine whether optogenetic inhibition altered movement. 2-way repeated measures ANOVAs (factors of light delivery and virus expression) did not reveal significant effects for the average FR5 inter-press interval (light: F(1,26)=1.45, P=0.24; virus: F(1,26)=0.33, P=0.57), total FR5 duration (light: F(1,26)=0.001, P=0.97; virus: F(1,26)=0.22, P=0.65), or reward retrieval latency (light: F(1,26)=0.80, P=0.34; virus: F(1,26)=0.99, P=0.33). We did find a significant effect of light delivery on the latency to initiate the FR5 sequence (an average increase of 100 ms for NpHR-eYFP mice and 160 ms for eYFP control mice; light: F(1,26)=17.37, P=0.0003; virus: F(1,26)=0.05, P=0.83), but only the eYFP control cohort yielded a significant post-hoc test (P = 0.0033; NpHR-eYFP: P = 0.054).

We think two factors contribute to this apparent discrepancy in motor impairment between the present experiment and the published Cover et al., 2019 Cell Reports finding. The 2019 experiment chemogenetically suppressed somatic and terminal activity of all rILN neurons, whereas here we transiently inhibited somatic activity of striatally projecting rILN neurons. Beyond this methodological difference, however, we suspect that the context underlying movement may play a significant role. The 2019 study monitored spontaneous locomotion in an environment free of rewards and meaningful sensory information. In contrast, the present study assessed performance on an external cue-driven paradigm in motivated (food-restricted) subjects.

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

Reviewer #3 (Recommendations for the authors):

In Figure 5C (left), it isn't clear that the fluorogold/AAV1 injection was actually into the supramamillary nucleus. It's hard to tell where the injection was but it seems to be dispersed throughout the midbrain.

The core of the injection site does appear in the area of the SUM, as evidenced by the dense fluorogold labeling. The purplish, scattered off-target labeling appears associated with histological imperfections resulting from our mounting method. The prominent red (tdtomato) labeling lateral to the injection site (midbrain/substantia nigra) are tdtomato-positive axons originating from the dorsal striatum. After further assessment, we now cannot be certain that the injection site did not also involve the lateral hypothalamus, however, which sits laterally to the SUM. As such, we have revised the manuscript and associated figure/figure legends to reflect this interpretation. Where we used SUM to abbreviate for supramammillary nucleus, we now used S/L to abbreviate for supramammillary nucleus/lateral hypothalamus. Thank you for pointing this out

Also, in Figures 5A, B, and D, the way the authors have schematized ipsi/contralateral projections is confusing, in particular, showing three circles with the seed nucleus. Perhaps the authors can think of a better way to convey laterality.

We have revised the way ipsi/contralateral projections are schematized. We now just show two circles: one depicting the ipsilateral seed nucleus and the other the contralateral seed nucleus. We revised the figure legend accordingly. We agree that this streamlines conceptual understanding for the reader. Thank you for the suggestion.