Superior colliculus projections drive dopamine neuron activity and movement but not value

Abstract

To navigate dynamic environments, animals must rapidly integrate sensory information and respond appropriately to gather rewards and avoid threats. It is well established that dopamine (DA) neurons in the ventral tegmental area (VTA) and substantia nigra (SNc) are key for creating associations between environmental stimuli (i.e., cues) and the outcomes they predict. Critically, it remains unclear how sensory information is integrated into dopamine pathways. The superior colliculus (SC) receives direct visual input and is positioned as a relay for dopamine neuron augmentation. We characterized the anatomical organization and functional impact of SC projections to the VTA and SNc in rats. First, we show that neurons in the deep layers of SC synapse densely throughout the ventral midbrain, interfacing with projections to the striatum and ventral pallidum, and these SC projections excite dopamine and GABA neurons in the VTA/SNc in vivo. Despite this, cues predicting SC→VTA/SNc neuron activation did not reliably evoke behavior in an optogenetic Pavlovian conditioning paradigm, and activation of SC→VTA/SNc neurons did not support primary reinforcement or produce place preference/avoidance. Instead, we find that stimulation of SC→VTA/SNc neurons evokes head turning. Focusing optogenetic activation solely onto dopamine neurons that receive input from the SC was sufficient to invigorate turning, but not reinforcement. Turning intensity increased with repeated stimulations, suggesting that this circuit may underlie sensorimotor learning for exploration and attentional switching. Together our results show that collicular neurons contribute to cue-guided behaviors by controlling pose adjustments through interaction with dopamine neurons that preferentially engage movement instead of reward.

Fig 1.. Superior colliculus projections to the ventral midbrain.

A) Viral approach for targeting SC neurons. B) Injection of a GFP-expressing virus (green) into the SC resulted in expression throughout the intermediate and deep layers, with terminals visible throughout the ventral midbrain in the VTA and SNc. C) Tissue was counterstained for tyrosine hydroxylase (TH, red), demonstrating dense intermingling of SC projections with DA neurons in the VTA and SNc. D) Viral approach for visualizing monosynaptic connections between SC and the ventral midbrain. E) Injection of a virus coding for membrane-bound GFP and mRuby conjugated to the synaptophysin protein in the SC demonstrated strong innervation of the VTA and SNc. F) Dense mRuby puncta, indicating synaptic contacts between SC terminals and ventral midbrain neurons, were seen in VTA and SNc, G) closely associated with GFP-expressing terminals. H) Viral approach for transsynaptic tracing of SC-forebrain circuits. An AAV1 virus delivering cre recombinase was injected into the SC, combined with injection of a cre-dependent virus coding for mCherry into the VTA/SNc. I) Via the anterograde transsynaptic transport of AAV1-cre, we visualized VTA/SNc neurons receiving monosynaptic inputs from the SC (top: sagittal view; bottom: coronal view). J) mCherry fibers from VTA/SNc neurons receiving SC input were evident throughout the striatum and ventral pallidum in the forebrain.

Fig 2.. Superior colliculus projections activate VTA/SNc dopamine neurons in vivo.

A) Approach to target SC terminals with the red-shifted excitatory opsin ChrimsonR and dopamine neurons with a cre-dependent GCaMP8-coding virus in TH-cre rats, for simultaneous optogenetic stimulation and photometry recordings in the VTA or SNc. B) Histology image showing targeting of GCaMP to TH-positive neurons, and fiber placements in the midbrain. C) SC terminal stimulation evoked locomotion with increased intensity at 20 versus 5 Hz laser delivery (N=12). D) Z-score average trace of DA neuron fluorescence for all rats time locked to 5-Hz stimulation onset (N=11). Robust phasic DA neuron activation was seen, as measured by signal E) peak and F) area under the curve (AUC) measures. VTA (N=5) and SNc (N=6) DA neuron activity differed in G) peak signal, but not H) AUC. I) Z-scored trace of DA neuron fluorescence time locked to 20-Hz stimulation (N=11), which produced robust sustained DA neuron activation, as measured by signal J) peak and K) AUC measures. VTA (N=6) and SNc (N=5) DA neuron activity at 20 Hz did not differ in L) peak signal or M) AUC. N) Rats (N=7, 4 VTA, 3 SNC) consumed sucrose reward from a port in the chamber. O) Both sucrose and 20-Hz SC terminal stimulation evoked strong dopamine neuron responses. P) Sucrose evoked a larger peak dopamine signal, Q) but a similar overall response based on AUC. **p<.01 (unpaired t test), *p<.05. ###p<.001 (one sample t test vs 0), ##p<.01 (vs 0), #p<.05 (vs 0). Error bars depict subjects mean +/−SEM.

Fig 3.. Superior colliculus projections activate VTA GABA neurons in vivo.

Approach to target SC terminals with the red-shifted excitatory opsin ChrimsonR and GABA/GAD1+ neurons with a cre-dependent GCaMP8-coding virus in wild type rats, for simultaneous optogenetic stimulation and photometry recordings in the VTA. B) Histology image showing GAD-GCaMP targeting in TH negative neurons and fiber placements in the midbrain. C) Z-score averaged trace of GABA neuron fluorescence time locked to 5-Hz stimulation (N=8). Robust phasic GABA neuron activation was seen, as measured by signal D) peak and E) AUC measures. F) Z-scored trace of DA neuron fluorescence time locked to 20-Hz stimulation (N=7). Robust sustained GABA neuron activation was seen, as measured by signal G) peak and H) AUC measures. ##p<.01 (one sample t test vs 0), ####p<.0001 (vs 0). Error bars depict subject mean +/− SEM.

Fig 4.. Activation of SC projections to the VTA/SNc does not drive Pavlovian cue conditioning.

A) Approach for targeting of SC terminals in the ventral midbrain. ChR2-YFP expressed in deep layer SC neurons, and B) an optic fiber was implanted over the ipsilateral VTA (SC-VTA, n = 6), SNc (SC-SNc, n = 8), or a control region (Control, n = 8). C) Optogenetic Pavlovian conditioning paradigm, where a neutral cue was paired with optogenetic activation of SC terminals. D) Instances of cue approach behavior were measured via inspection of video recordings during conditioning. E) During the cue-only period, corresponding to the first 2-sec of cue presentations, minimal approach behavior occurred across all groups. F) During the laser period, corresponding to the final 5 sec of the cue when laser was also delivered, approach was similarly low for all groups. Error bars depict SEM.

Fig 5.. SC terminal stimulation in the VTA and SNc evokes head/neck turning independent of cue conditioning.

A) Schematic illustrating craning behavior, which was defined as a head deflection of at least 90 degrees contralateral to the stimulation hemisphere. B) Craning behavior did not occur during the cue-only period (before laser onset). C) During the laser period, craning was robust in the SC-VTA (n=6) and SC-SNc (n=8) groups relative to controls (n=8). D) Unpaired optogenetic conditioning procedure, where cue and laser presentations were separated by a variable interval, for a separate cohort of rats (n = 5). E) Unpaired rats exhibited a similar probability of craning as paired subjects across training, and F) when data was collapsed across all analysis sessions. G) In addition to head craning, full body rotation contralateral to the stimulation hemisphere was quantified. (H,I) Rotations during the laser period occurred on a subset of trials, increasing in probability across conditioning sessions for SC-VTA and SC-SNc rats. ****p<.0001, *p<.05. Error bars depict SEM.

Fig 6.. SC projections to VTA/SNc transiently shape head turning.

A) Rats’ body parts and static elements of the chamber were labeled in video frames from behavior sessions to train a neural network and acquire pose data using the DeepLapCut pipeline. B) A directionality vector was interpolated and tracked across video frames for calculation of turning angle. C) Example turn, showing a rat at three sequential timepoints during a laser stimulation event. Turning examples for D) SC-VTA, E) SC-SNc, and F) control subjects. The blue shaded zone corresponds to the laser stimulation window. A positive slope corresponds to an increase in the cumulative angle of deflection of the head vector contralateral to the hemisphere of laser stimulation. G) Averaged turning traces for paired SC-VTA/SNc rats (n=12) on the first and last day of conditioning versus controls. On the H) first and I) last sessions, turning occurred only during the laser period in VTA/SNc rats, and not in controls. J) Laser-evoked turning intensity increased for paired rats from the first to last day of conditioning. K) Average cumulative angle traces for unpaired VTA/SNc rats (n=5) show no turning during unpaired cue presentations, but L) robust turning during the laser windows. M) As with paired rats, turning was specific to the laser epoch. N) Unpaired rat turning increased qualitatively, but was not significantly higher on the last day of conditioning. ****p<.01, *p<.05. Error bars depict SEM.

Fig 7.. SC terminal stimulation does not support reinforcement or signal valence.

A) Intracranial self stimulation (ICSS) set up. B) Nose behavior was low overall (n=8). C) Plotted separately, SC-VTA (n=4), SC-SNc (n=4), and control (n=2) groups exhibited low responding, failing to discriminate between active and inactive nose pokes. Active = A, Inactive = I. D) Real time place assay paradigm. E) No preference or avoidance was observed for any group. Error bars depict SEM.

Fig 8.. Dopamine neurons receiving input from the SC promote turning but not reinforcement.

A) Viral approach for targeting dopamine neurons receiving direct input from the SC. In TH-cre rats, an AAV1-flp vector was injected into deep layer SC, followed by injection of a cre and flp-dependent virus coding for ChR2-YFP into the ipsilateral VTA/SNc. B) This resulted in strong expression of YFP in dopamine neurons, C) including those innervating the striatum. D) Zoomed in images showing YFP expression with TH counterstain in the dorsal and ventral striatum. E) Rats (N=9) with ChR2 expressed in dopamine neurons receiving SC input completed the F) optogenetic conditioning paradigm, where a cue predicted laser delivery. Turning behavior was quantified as the cumulative angle reached during cue/laser presentations. G) Averaged cumulative angle for trial epochs for the Day 1 versus 12 of training, where an increase during the laser period was evident. H) Total turning marginally increased from Day 1 to 12. I-L) Averaged cumulative angle traces across the trial period for Day 1, 4, 8, and 12, showing a steady increase in turning during laser. M) Rats were given the opportunity to nose poke for optogenetic stimulation of dopamine neurons that receive input from the SC. N) Across 4 training sessions, active and inactive nose pokes remained at a low level. O) Fiber optic placements. **p<.01. Error bars depict SEM.