Here we present a series of evidence that delineates the role of released ACh from those of VGLUT3-dependent glutamate release from striatal cholinergic interneurons. Our data provide a novel perspective on the function of striatal cholinergic neurons suggesting the possibility that they can use distinct neurotransmitters to regulate striatal circuitry. We found that elimination of VAChT in the striatum, without disruption of VGLUT3, did not cause overt disruptions or alterations in several behavioural tasks previously thought to be dependent on striatal ACh release, such as motor learning, sensorimotor gating, and spontaneous locomotor activity. However, we uncovered a novel form of regulation of MSNs by cholinergic tone, and found that selective silencing of striatal ACh release results in an increase in the responses to D1R and D2R agonists. In contrast to the effects of direct dopamine receptor agonists, we found that overall these mice do not show increased locomotor response to cocaine. Similarly, sensitization and rewarding effects of cocaine did not seem to be dependent on striatal release of ACh. Thus, our results significantly depart from previous studies in which the specific contributions of striatal ACh release (mediated by VAChT) were not separated from those of glutamate release (mediated by VGLUT3). These data suggest that VGLUT-3 dependent glutamate release may influence locomotor activity and responses to cocaine considerably more than VAChT-dependent ACh release. Our data suggest that targeted approaches aimed at inhibiting VAChT activity in the striatum may potentially provide a novel strategy to enhance dopaminergic signalling, without causing other major behavioural disturbances.
VAChTD2-cre-flox/flox Mice Have Normal Motor Performance, Sensorimotor Gating, and Motor Learning
Our studies in VAChTD2-cre-flox/flox
mice indicated that elimination of ACh release in the striatum does not seem to play a major role in motor function and motor learning, at least for acrobatic motor skills in the rotarod test. This observation is also in agreement with previous experiments in striatal cholinergic neuron-ablated mice that presented no deficiency in rotarod performance 
. However, we cannot completely exclude more subtle effects of ACh in fine motor tuning and motor tasks. For example, the chronic nature of elimination of ACh release in our experiments may lead to adaptations in motor behaviour. Future experiments using VAChTD2-Cre-flox/flox
mice and more sophisticated motor behavioural tests may be necessary to pinpoint possible roles for striatal ACh in motor learning and performance.
There are multiple lines of evidence that pharmacological modulation of cholinergic receptors regulates locomotor activity. It is known that muscarinic antagonists increase locomotor activity and M1 and M4 muscarinic receptor KO mice are hyperactive 
. Moreover, we have recently observed that mice with a significant decrease in VAChT expression in the whole forebrain show hyperactivity 
. The present work provides compelling evidence for more selective roles of the neurotransmitter ACh in the striatum, indicating that decreased striatal expression of VAChT does not cause overt motor consequences. These results may be of particular importance, since there have been reports that in Huntington's disease VAChT levels are decreased in the striatum 
. Our data suggest, however, that this alteration is unlikely to contribute to gross motor symptoms observed in Huntington's disease. Cholinergic neurotransmission in brain regions other than the striatum may still play a role in control of locomotion.
Previous attempts to assess the function of cholinergic neurons in the striatum were performed following the ablation of cholinergic neurons using immunotoxin-mediated cell targeting. Injection of toxin targeting transgenic cholinergic neuron in the accumbens led to an 80% decrease in ChAT-positive neurons 
. Elimination of cholinergic neurons in the accumbens by this means inhibited certain forms of reward-related learning; however, it also induced hyperactivity and increased sensitivity to the locomotor and the rewarding effects of cocaine, including increased sensitivity in the CPP test to low doses of cocaine 
. In contrast, recent experiments using an optogenetic approach to inactivate or activate cholinergic neurons in the accumbens found no effects of inactivation of these neurons on locomotor activity, albeit their silencing prevented the response to cocaine in a CPP test 
. Thus, elimination of cholinergic neurons in the accumbens seemed to increase sensitivity to cocaine-induced CPP 
, whereas optogenetics silencing of these neurons blocked cocaine-induced CPP 
. The reason for the different outcome in these two experiments is not entirely clear at the moment, but could be related to the chronic versus acute nature of the manipulations. Although in our experiments we have targeted the whole striatum, rather than only the accumbens, we did not detect major alterations in cocaine-induced CPP, suggesting that the above effects obtained with neuronal ablation or by optogenetics manipulation may be linked not to loss of cholinergic transmission per se but rather to suppression of glutamate release from cholinergic neurons.
While an optogenetic approach provides a novel paradigm to acutely activate or inactivate populations of neurons, it is unlikely that this method can separate VAChT from VGLUT3-dependent neurotransmission as selectively as that which can be achieved using VAChTD2-Cre-flox/flox
mice. Interestingly, recent data have shown that cholinergic neurons in the habenula secrete both ACh and glutamate (mediated by VGLUT1), and release of either of these neurotransmitters appears to depend on the frequency of stimulation 
. Basal forebrain neurons in culture release both ACh and glutamate 
. Importantly, recent work shows that optogenetics stimulation of striatal cholinergic neurons can evoke synaptic glutamatergic neurotransmission onto MSNs, with predominant activity over NMDA receptors 
. The co-release of glutamate with dopamine has also been described 
, suggesting that interpretation of the roles of dopaminergic neurons will also need to take into account glutamate co-release. Therefore, the co-release of glutamate with classical neurotransmitters may be a more common mechanism than previously appreciated and may have a broad impact in circuitry control. However, we cannot discard the possibility that other neuromodulators released from cholinergic neurons, such as ATP or peptides, could also play a role as co-transmitters.
The role of VGLUT3 in striatal function is far from being fully understood 
. Interestingly, with respect to striatum-related behaviour, VGLUT3-null mice show hyperactivity and increased response to the locomotor effects of cocaine 
. Therefore, mice lacking VGLUT3 show a phenotype that is remarkably similar to that of mice in which cholinergic neurons in the accumbens were targeted by an immunotoxin 
. Experiments in VGLUT3-null mice suggested that the absence of VGLUT3 causes a decrease in striatal cholinergic tone. VGLUT3 is used by the striatal vesicles to facilitate VAChT-mediated ACh storage in synaptic vesicles 
. However, measurements of ACh release in VGLUT3-null mice have indicated only a modest reduction, by 30% to 40% 
, compared to almost 100% inhibition in VAChTD2-Cre-flox/flox
mice. It is unlikely that 40% reduction in ACh release observed in VGLUT3-null mice can be responsible for the hyperactive phenotype. Indeed, independent mouse lines with a 50% decrease in VAChT expression, and concomitant reduction of ACh release 
, did not present increased locomotor activity in the open field 
. We conclude that the locomotor phenotypes observed previously in striatal cholinergic neuron-ablated mice 
and in VGLUT3-null mice 
are either a consequence of the disruption of VGLUT3-mediated neurotransmission or the combination of reducing both glutamatergic and cholinergic activity simultaneously from these neurons. Future experiments using VAChTD2-Cre-flox/flox
mice, VGLUT3 floxed mice, and double knockouts will be necessary to provide an assessment of independent effects of VGLUT3-mediated neurotransmission in the striatum.
Although we have focused on striatal-related behaviours, the extent by which alterations in VAChT expression in other brain regions in VAChTD2-cre-flox/flox
mice may contribute to these phenotypes should also be taken into account. We did not detect Cre-expression in cholinergic neurons in the penduculopontine area, for example (Figure S2
), which harbours groups of cholinergic neurons that project to the midbrain and thalamus and could influence striatal function. However, we cannot completely exclude the possibility that cholinergic neurons in other brain regions would not be targeted in our mouse line. At the same time, as the phenotypes described here seem to be mainly striatal specific and cholinergic interneurons provide the almost exclusive source of cholinergic tone in the striatum, it is unlikely that other groups of cholinergic neurons would have contributed to the observed behaviours.
Elimination of cholinergic neurotransmission in the striatum did not cause hyperlocomotion, however the responses to direct activation of dopamine receptors were substantially increased. Both behavioural and ph
MRI analysis indicated an increased response to D1R agonist. Western blot analysis also showed selective increase of D2R expression in the striatum. Moreover, in addition to the increased D2R levels in the striatum, which likely reflect a combination of pre- and post-synaptic receptors, we also uncovered increased D2-like receptor pre-synaptic activity, revealed by the increased sensitivity of VAChTD2-Cre-flox/flox
mice to low doses of quinpirole. Certainly, we cannot rule out that changes at the level of receptors play a more complex role in regulating locomotor activity in VAChTD2-Cre-flox/flox
mice. Indeed, GPCRs may have agonist-independent activity 
. The locomotor effects of cocaine seem to depend mainly on inhibition of the dopamine transporter 
. However, acetylcholine can affect release of dopamine via distinct nicotinic receptors 
, as well as regulate both dopamine release and activity of MSNs, via distinct muscarinic receptors 
. The fact that both D1R and D2R had increased expression in the striatum would suggest that VAChTD2-Cre-flox/flox
mice should be more responsive to dopamine and might present increased spontaneous locomotor activity or cocaine-induced locomotion or CPP. However, this was not the case. It is likely that cell-autonomous compensatory mechanisms related to disrupted cholinergic function significantly altered striatal circuitry, preventing such a simple relationship. For example, because M4 muscarinic receptors seem to specifically regulate D1R-mediated signalling 
, it is possible that the increased expression of M4 receptors we detected in the striatum could counterbalance D1R-mediated responses in vivo, leading to unaltered locomotor activity. Moreover, because D2-like pre-synaptic receptors may be more active in VAChTD2-Cre-flox/flox
mice, elimination of ACh release in the striatum may also affect pre-synaptic control of dopamine release. The slightly decreased turnover of dopamine in mice without striatal VAChT supports the notion of direct consequences of reduced cholinergic tone at the level of dopaminergic terminals. Thus, behavioural analysis of VAChTD2-Cre-flox/flox
mice indicates that control of locomotor function and response to cocaine mediated by dopamine might become more complex in the absence of cholinergic tone. Future experiments will be needed to evaluate direct consequences of elimination of either acetylcholine or glutamate neurotransmission originating from striatal cholinergic neurons on dopamine transmission.