Psilocybin exerts distinct effects on resting state networks associated with serotonin and dopamine in mice

Hallucinogenic agents have been proposed as potent antidepressants; this includes the serotonin (5-HT) receptor 2A agonist psilocybin. In human subjects, psilocybin alters functional connectivity (FC) within the default-mode network (DMN), a constellation of inter-connected regions that is involved in self-reference and displays altered FC in depressive disorders. In this study we investigated the effects of psilocybin on FC in the analogue of the DMN in mouse, with a view to establishing an experimental animal model to investigate underlying mechanisms. Psilocybin effects were investigated in lightly-anaesthetized mice using resting-state fMRI. Dual-regression analysis identified reduced FC within the ventral striatum in psilocybin-relative to vehicle-treated mice. Refinement of the analysis using spatial references derived from both gene expression maps and viral tracer projection fields revealed two distinct effects of psilocybin: it increased FC between 5-HT-associated networks and elements of the murine DMN, thalamus, and midbrain; it decreased FC within dopamine (DA)-associated striatal networks. These results suggest that interaction between 5-HT- and DA-regulated neural networks contributes to the neural and therefore psychological effects of psilocybin. Furthermore, they highlight how information on molecular expression patterns and structural connectivity can assist in the interpretation of pharmaco-fMRI findings.


41
Psychiatric disorders are associated with changes in the status of specific neurotransmitter systems, 42 including those of serotonin (5-hydroxytryptamine, 5-HT) and dopamine (DA). Recently, psilocybin, a 43 psychedelic compound derived from "magic mushrooms" with high affinity to the 5-HTA receptor (Nichols,44 2004; Halberstadt and Geyer, 2011), has gained in research interest due to its potential to alleviate 45 depression and anxiety (Grob et al., 2011;Griffiths et al., 2016;Carhart-Harris et al., 2017a). Whilst the 46 primary target of its active metabolite psilocin is the 5-HT2A receptor and it binds to a lesser extent to 47 receptor 5-HT1A , psilocybin has also been shown to induce DA release in the nucleus accumbens in rats 48 (Sakashita et al., 2015) and to reduce 11 C-raclopride binding to the D2 receptor in the caudate-putamen in 49 humans (Vollenweider et al., 1999).

50
The direct and indirect actions of psilocybin on specific monoamine neurotransmitter systems result in 51 downstream metabolic and physiological effects on the brain. Thus, psilocybin increases global energy 52 metabolism indicated by elevated glucose utilization rates in the frontomedial, frontolateral and anterior 53 cingulate cortices as well as in basal ganglia (Vollenweider et al., 1997). Also, somewhat counterintuitively,

78
To test these hypotheses, we acquired resting-state fMRI data from mice treated with psilocybin or vehicle.

79
Conventional data analysis did not reveal any drug-induced alterations within elements of the rodent DMN. St Isle, France) weighing 25 ± 1.2 g were studied. Animals were maintained in standard housing with food 94 and water available ad libitum and in a reversed 12:12 day/night cycle (light off: 07:00-19:00 h). Anaesthesia 95 was induced with isoflurane 3.5% in a mixture of 20% O2 and 80% air. Isoflurane was reduced to 2%, 96 animals were then intubated endotracheally and ventilated mechanically at 80 breaths/min, positioned on 97 an MR-compatible cradle, and a cannula was positioned in the tail vein. Psilocybin was dissolved in sterile 98 water with tartaric acid and infused i.v. during 5 min with either 1 mg/kg (n = 13) or 2 mg/kg (n = 12) or 99 vehicle only (n = 15). These doses were chosen because they were demonstrated previously to elicit To test for potential confounding cardiovascular-mediated effects of psilocybin on brain FC, heart rate was 187 monitored in a separate group of mice using electrocardiogram recordings and applying the same protocol 188 as for the MRI experiment. Heart rate was 302 ± 30 beats per minute (bpm) in mice (n = 5) that received2 189 mg/kg psilocybin and 296 ± 27 bpm in mice (n = 5) that received vehicle; psilocybin was without a significant

276
To differentiate the effects of psilocybin on 5-HT-and DA-associated networks across the whole brain, a 277 dual-regression analysis was carried with brain-wide Drd2 and Htr2a gene expressions used as spatial 278 reference maps (Figure 4a and b, Figure S3a). The averaged resting BOLD time series from the regions 279 delineated in these reference maps were extracted and used to establish FC between these reference 280 regions and the remaining voxels across the whole-brain. The dual-regression analysis revealed reduced 281 FC between the Drd2 expressing regions and the striatum, pallidum and thalamus in psilocybin versus 12 vehicle-treated mice, comparable to the results obtained previously (Figure 1). A weak opposite effect, i.e. 283 increased FC due to psilocybin, was found between Htr2a expressing regions and the left dorsal striatum, 284 although this effect did not survive statistical correction in the subsequent ROI analysis (Figure 4b).

285
Analyses carried out for other 5-HT receptor gene-expression maps did not reveal significant clusters. To 286 confirm these results, we used viral tracer maps, obtained by virus injection targeting DA neurons in the 287 VTA or 5-HT neurons in the DRN, as spatial references in the dual-regression analysis (Figure 4c

331
The DMN is considered to be one of the most important systems underlying higher cognitive functions 332 (Raichle et al., 2001), and within-and between-region FC is increased in depression (Sheline et al., 2010) 333 and animal models thereof (Grandjean et al., 2016)

381
We are aware of potential limitations associated with our approach. Firstly, psilocin administration has been 382 associated with decoupling between the hemodynamic response and neuronal activity in rats during short-383 duration, high frequency (>10 Hz) whisker sensory stimulation (Spain et al., 2015); however, this was not 384 the case for lower frequency paradigms. Moreover, this effect would not be expected to be confined to 385 specific RSNs, hence the relevance of this observation in the context of resting-state FC remains moot.