Genetic deletion of β-arrestin 2 modulates LSD-stimulated behaviors in mice

Recent evidence suggests that psychedelic drugs can exert benecial effects on anxiety, depression, and ethanol and nicotine abuse in humans. However, their hallucinogenic side-effects often preclude their clinical use. Lysergic acid diethylamide (LSD) is a prototypical hallucinogen and its psychedelic actions are exerted through the 5-HT2A serotonin receptor (5-HT2AR). 5-HT2AR activation stimulates Gq- and β-arrestin- (βArr) mediated signaling. To separate these signaling modalities, we have used βArr1 and βArr2 mice. We nd that LSD stimulates motor activities to similar extents in WT and βArr1-KO mice, with non-signicant effects in βArr2-KOs. LSD robustly stimulates many surrogates of psychedelic drug actions including head twitches, grooming, retrograde walking, and nose-poking in WT and βArr1-KO animals. By contrast, LSD slightly stimulates head twitches in βArr2-KO mice, without effects on other surrogates. The 5-HT2AR antagonist MDL100907 (MDL) blocks these LSD effects. LSD also disrupts prepulse inhibition (PPI) in WT and βArr1-KOs; PPI is unaffected in βArr2-KOs. MDL restores PPI in WT mice, but this antagonist is without effect and haloperidol is required in βArr1-KOs. Collectively, these results reveal that LSD’s psychedelic drug-like actions appear to require βArr2. two-way ANOVAs detected signicant treatment effects for baseline rearing p<0.001] and stereotypical activities p<0.001]. To control for these baseline differences in rearing and stereotypy, these LSD post-injection activities were analyzed by ANCOVA. a LSD-stimulated locomotor activities in WT and βArr1-KO mice. A two-way ANOVA identied a signicant treatment effect [F(4,93)=18.916, p<0.001]. b LSD-stimulated rearing activities in βArr1 animals. An ANCOVA failed to nd any signicant differences. c LSD-stimulated stereotypical activities in βArr1 mice. An ANCOVA revealed a signicant treatment effect [F(4,92)=7.029, P=0.024]. N = 8-17 mice/group.


Background
Lysergic acid diethylamide (LSD) is a prototypical psychedelic drug and is one of the most potent drugs in this class 1 . LSD alters sensation, perception, thought, mood, sense of time and space, and consciousness of self in humans 1,2 . Since LSD-induced states bear many similarities to early acute phases of psychosis 2 and because serotonin (5-HT) and LSD both contain an indolamine moiety, Woolley and Shaw 3 proposed that aberrant 5-HT levels in brain may produce mental disturbances including psychosis. This suggestion gave rise to the 5-HT hypothesis for schizophrenia and stimulated researchers to study LSD in hopes of gaining a better understanding of the disorder. However, this research was largely curtailed when LSD was classi ed as a DEA Schedule I drug in the 1960's. Recent research has revealed that LSD has medicinal value in treating cluster headaches 4 , anxiety and depressive disorders in life-threatening conditions when combined with psychotherapy 5 , and it may have potential for studying human consciousness and substance abuse 6-7 .
LSD shares structural similarities to 5-HT 1 . Thus, it is not surprising that LSD has high a nities for all thirteen 5-HT G protein coupled receptors (GPCRs) 8- 10 . Besides 5-HT receptors, LSD activates other biogenic amine GPCRs 8 and this polypharmacolgy may contribute to LSD's many actions. One activity in particular regarding LSD is its hallucinogenic actions. This activity is ascribed to 5-HT 2A receptor (5-HT2AR) stimulation since in drug discrimination studies, potency is correlated with hallucinogenic potency in humans 11 . Because the same psychedelics produce head twitches in mice, this response is used as a proxy for hallucinations in humans 12 , even though non-psychedelic drugs like 5hydroxytryptophan induce robust head-twitch responses (HTRs) 13 . Hallucinogen-induced HTRs in rodents are blocked by the highly selective 5-HT2AR antagonist MDL100907 [14][15][16] and are absent in htr2A knockout (KO) mice [17][18] . In addition, human studies have shown the hallucinogenic actions of LSD are blocked with the 5-HT2AR preferring antagonist ketanserin 19 . Thus, the hallucinogenic effects of LSD appear to be mediated through the 5-HT2AR 20 .
The 5-HT2AR is a rhodopsin family member of GPCRs that is coupled to G q protein and β-arrestin (βArr) mediated signaling [21][22][23][24] . Recent experiments reveal the 5-HT2AR preferentially activates G q family members, with moderate activity at G z , and minimal activities at G i -, G 12/13 -, and G s -family members 25 .
However, the 5-HT2AR binds to both βArr1 and βArr2 proteins in vitro and is complexed with these βArrs in cortical neurons in vivo 24 . While most GPCR agonists, like 5-HT, activate both G protein and βArr signaling, ligand binding can activate also G protein-dependent signaling while serving to activate or inhibit βArr-mediated signaling. Hence, a given ligand can act as an agonist at one pathway while inhibiting the other pathway or it can possess combinations of these actions. This property is termed functional selectivity or biased signaling 26-28 and ligands have been developed to exploit these signaling features 29 . Although LSD activates G protein signaling at many GPCRs 10 , this psychedelic stimulates βArr-mediated responses at most tested biogenic amine GPCRs 8 . Interestingly, LSD displays βArr-biased signaling at the 5-HT2AR 9-10,25 . Most 5-HT2AR-containing neurons express both βArr1 and βArr2 24 , and global βArr1 and βArr2 knockout (KO) mice have been generated [30][31] . Since LSD is βArr biased at the 5-HT2AR, the present investigations were conducted to determine whether LSD produces behavioral effects that were differential among the wild-type (WT) and βArr1-KO, and WT and βArr2-KO mice.

Results
Effects of Arrb1 or Arrb2 deletion on LSD-stimulated motor activities. LSD has been reported to stimulate, inhibit, or produce biphasic effects on a variety of motor activities in rodents 17,[32][33][34][35][36] . We examined responses to LSD in the global βArr1-KO and global βArr2-KO mice to determine whether disruption of either gene product could modify the behavioral responses to this hallucinogen and to test whether 5-HT2AR antagonism could block these effects. Locomotor, rearing, and stereotypical activities were monitored at 5-min intervals over the 120 min test in both the βArr1 and βArr2 genotypes (Supplementary Figs. S1-S2).
When cumulative baseline locomotion was examined in βArr1 mice, activity was not differentiated by genotype or by the pre-assigned treatment condition (Supplementary Table S1). Following LSD injection, only treatment effects were found (Fig. 1a). Here, locomotor activities were stimulated by LSD relative to control groups given the vehicle or 0.5 mg/kg MDL alone (p-values≤0.001). When administered with LSD, both doses of MDL blocked the locomotor-stimulating effects of this psychedelic (p-values≤0.001). It should be emphasized that no sex effects were detected in any experiments in this manuscript.
An examination of cumulative baseline rearing and stereotypical activities in the βArr1 mice found these responses to be signi cantly lower in some pre-assigned treatment groups than in others (p-values≤0.001) (Supplementary Table S1). To correct for these baseline differences in the subsequent LSD-post injection analyses, the rearing and stereotypical data were submitted separately to ANCOVA. No signi cant effects of LSD were observed for rearing (Fig. 1b). By comparison for stereotypical activities, ANCOVA revealed a signi cant treatment effect in βArr1 mice following LSD administration (p=0.024).
Nevertheless, Bonferroni post-hoc analyses only identi ed a trend between the group treated with LSD and the group given MDL alone (p=0.062) (Fig. 1c). Collectively, these results indicate that LSD stimulates locomotor activities to similar extents in the WT and βArr1-KO animals, and the 5-HT2AR antagonist blocks these responses. Rearing and stereotypical activities are unaffected by LSD in either genotype When baseline motor activities were evaluated in the βArr2 mice, no signi cant differences were found (Supplementary Table S2). Effects of LSD in the βArr2-KO mice were quite different from those of the WT animals. LSD was more potent in stimulating cumulative locomotor activities in the WT than in the βArr2-KO mice (p-values<0.001) (Fig. 2a). When locomotion was analyzed within WT animals, the LSDstimulated responses were higher than those in the vehicle and MDL controls, as well as in the treatment groups administered MDL with LSD (p-values<0.001). Hence, all three doses of the 5-HT2AR antagonist were e cacious in suppressing the LSD-induced hyperlocomotion. Although LSD increased locomotor activity in βArr2-KO mice, it was not signi cantly different from any other treatment group.
Similar to locomotion, LSD also stimulated rearing activities to a greater extent in WT compared to βArr2-KO mice (p-values<0.001) (Fig. 2b). In WT animals, rearing activities were increased with LSD over that of the vehicle and MDL controls (p-values<0.001). When 0.1 or 0.5 mg/kg MDL was given with LSD, both doses reduced the LSD-stimulated rearing activities to control levels (p-values≤0.001). By comparison, LSD was without effect in the βArr2-KO mice.
An assessment of stereotypical activities failed to nd any genotype differences between the βArr2 mice ( Fig. 2c). Nonetheless, treatment effects were evident with LSD stimulating stereotypical activities over that of the vehicle and MDL controls (p-values≤0.013). Notably, 0.5 mg/kg MDL abrogated the LSD effects (p=0.003). Together, these results indicate that LSD stimulates motor responses to similar extents in the WT βArr1 and WT βArr2 mice, and in the βArr1-KO animals. The 5-HT2AR antagonist blocks these LSD-stimulated activities. By striking comparison, LSD exerts minimal effects on these same responses in the βArr2-KO mice where none of their motor activities were signi cantly increased above that of controls.
LSD effects on additional behaviors. LSD modi es a number of behaviors in mice 12,17,[37][38][39][40][41] that include, at least, HTRs, grooming, and retrograde walking. When these responses were examined in the βArr1 mice, no genotype differences were noted, although overall treatment effects were evident. Relative to the vehicle and MDL controls, LSD stimulated HTRs in the WT and βArr1-KO mice (p-values<0.001) (Fig. 3a). When 0.1 or 0.5 mg/kg MDL was administered with LSD, both doses of the 5-HT2AR antagonist blocked the LSD effects by restoring the numbers of HTRs to those of controls. Aside from HTRs, LSD augmented also grooming over that of the controls (p-values<0.001) (Fig. 3b). When 0.1 or 0.5 mg/kg MDL was given with LSD, both doses of the 5-HT2AR antagonist normalized the LSD stimulatory effects to those of controls (p-values<0.001).
Besides HTRs and grooming, LSD was e cacious in potentiating retrograde walking in the WT and βArr1-KO mice compared to the vehicle and MDL controls (p-values<0.001) (Fig. 3c;). With LSD, both 0.1 and 0.5 mg/kg MDL depressed retrograde walking (p-values<0.001). Nose poking behaviors were examined also. Here, LSD increased nose-poking over that of controls (p-values<0.001) (Fig. 3d). When MDL was administered with LSD, both doses of the 5-HT2AR antagonist normalized the LSD stimulated nosepoking behaviors (p-values<0.001).
In contradistinction to βArr1 mice, genotype differences were identi ed between the βArr2 animals. HTRs were signi cantly increased with LSD in WT relative to βArr2-KO mice (p<0.001) (Fig. 4a). Genotype effects were noted also in the 0.05 mg/kg MDL plus LSD group (p<0.001). In WT mice, HTRs were stimulated by LSD and they were still enhanced when 0.05 MDL was given with LSD relative to the vehicle and MDL controls (p-values<0.001). Notably, both 0.1 and 0.5 mg/kg MDL signi cantly reduced the LSD-stimulated responses (p-values≤0.002) -with the higher MDL dose being the more e cacious in suppressing HTRs to control levels (p<0.001). In the βArr2-KO mice, the LSD (p-values≤0.023) and 0.05 mg/kg MDL plus LSD treatments (p-values≤0.006) increased HTRs compared to the vehicle and MDL controls. Only 0.5 mg/kg MDL was su cient to normalize this LSD-stimulated response in the βArr2-KO mice (p=0.019).
For grooming, the durations of responding were higher in WT than in the βArr2-KO groups administered LSD alone, 0.05 mg/kg MDL plus LSD, or 0.5 mg/kg MDL with LSD (p-values≤0.016) (Fig. 4b). In WT mice, LSD augmented grooming relative to the vehicle and MDL controls (p<0.001). While 0.05 mg/kg MDL failed to block the LSD effects, both of the 0.1 and 0.5 mg/kg doses were e cacious in normalizing the responses (p-values<0.001). In βArr2-KO animals, the duration of grooming to LSD was not signi cantly different from the vehicle and MDL controls. Nevertheless, grooming was enhanced in the group administered 0.05 mg/kg MDL plus LSD compared to all groups (p-values≤0.013), except those given LSD alone.
Since LSD can induce alterations in tactile perception 42 , we examined grooming in detail as it has a chained organization of responses in rodents 43 . Note, that since the WT βArr1 and WT βArr2 mice responded identically to the different treatment conditions, only one of the WT strains is represented. Analyses of the video-recordings con rmed that all genotypes engaged in a normal sequence of grooming beginning with the face, progressing down the body, and ending at the feet or tail (Movie 1). When LSD was administered, the sequence of grooming in the WT and βArr1-KO mice became abbreviated, non-sequential, and/or restricted to one area of the body (Movies 2-3). By sharp comparison, the grooming sequence was complete and rarely perturbed in the βArr2-KO animals (Movie 4). When the 5-HT2AR antagonist MDL was administered alone, the organization of grooming was intact in the WT and βArr1-KO mice (Movie 5). By comparison, with MDL the βArr2-KO animals often paused in grooming bouts and/or displayed twitching of the neck and back muscles; however, they would nish the grooming sequence (Movie 6). The patterns of grooming among the genotypes administered MDL plus LSD were divergent. In WT mice given MDL plus LSD, the organization of grooming was restored (Movie 7). When the βArr1 mutants received the same treatment, they began the grooming sequence, engaged in focal grooming of a part of the body, and then completed the sequence (Movie 8). When this same drug combination was administered to βArr2-KO mice, they usually began the sequence appropriately, but at some mid-or later-point they would become focused on one area of grooming (Movie 9). Nevertheless, they usually completed the grooming sequence Aside from abnormalities in the organization of grooming, LSD also induced retrograde walking and stimulated nose-poking behaviors. No signi cant genotype effects were obtained for retrograde walking (Fig. 4c). In WT mice, LSD potentiated the incidences of retrograde walking compared to the MDL and vehicle controls (p<0.001). Although 0.05 mg/kg MDL was ineffective in decreasing this LSD-stimulated behavior, both 0.1 and 0.5 mg/kg MDL suppressed this response (p-values<0.001). By contrast, LSD was without any signi cant effect on retrograde walking in the βArr2-KO animals. Similar to retrograde walking, no genotype effects were observed for nose poking behavior (Fig. 4d). In WT mice, LSD stimulated nose-poking behaviors relative to all other groups (p-values<0.001). All doses of the 5-HT2AR antagonist reduced the LSD-stimulated nose poking to the levels of the vehicle and MDL controls. No treatment effects were noted among the βArr2-KO animals.
In summary, responses to LSD across these LSD-stimulated behaviors were similar between the βArr1 genotypes and the 5-HT2AR antagonist reduced these responses to levels of the vehicle and MDL controls. Importantly, the WT mice responded quite differently than the βArr2-KO animals. HTRs and grooming to LSD were signi cantly higher in WT than in βArr2-KO mice. LSD did not signi cantly increase retrograde walking or nose poking behaviors in the βArr2-KO animals. Notably, LSD disrupted the sequences of grooming in WT and in βArr1-KO mice; the βArr2-KO animals were unaffected. Nonetheless, divergent responses to MDL alone or MDL plus LSD were observed among the genotypes, indicating actions required by 5-HT2AR activation.
LSD and MDL100907 effects on prepulse inhibition. LSD disrupts PPI in both rats and humans and the response can be restored with 5-HT2AR antagonists 37,44 . βArr1 mice were pre-treated with the vehicle or with 0.1 or 0.5 mg/kg MDL as controls. Subsequently, they were administered the vehicle or 0.3 mg/kg LSD and tested in PPI. No signi cant genotype or treatment effects were observed for null activity or in response to the 120 dB startle stimulus ( Supplementary Fig. S3a-b). In contrast, genotype effects were found in PPI where 0.1 and 0.5 mg/kg MDL normalized PPI in WT mice, whereby these same doses were ineffective in the βArr1-KO animals (p-values≤0.018) (Fig. 5a). As anticipated, LSD depressed PPI in both βArr1 genotypes relative to their MDL and vehicle controls (p-values≤0.002). Thus, LSD depressed PPI in both WT and βArr1-KO mice, while MDL only restored PPI in WT animals.
Since haloperidol can normalize PPI in mouse models 45 , we tested whether this antipsychotic drug could normalize the LSD-disrupted PPI in the βArr1-KO mice. For null activity, no genotype effects were evident ( Supplementary Fig. S3c). Overall treatment effects were found in the βArr1 animals where null activities were higher in the 0.1 mg/kg haloperidol plus LSD group than in mice treated with the vehicle or haloperidol alone (p-values≤0.009). An assessment of startle activity revealed that responses were lower overall in the WT relative to βArr1-KO mice (p=0.028) ( Supplementary Fig. S3d). For PPI, responses were reduced overall in the βArr1-KO compared to the WT animals (p=0.008) (Fig. 5b). Treatment effects were observed also, where LSD suppressed PPI relative to all other treatment conditions (p-values≤0.002). Here, haloperidol normalized the LSD-disrupted PPI in both WT and βArr1-KO mice.
PPI responses in the βArr2 mice were examined also. No signi cant genotype effects were reported for null or startle activities. Overall null activity was decreased in the 0.1 mg/kg MDL plus LSD group compared to the vehicle control and the LSD group (p-values≤0.003) (Supplementary Fig. S4a). No signi cant effects were detected for startle activity (Supplementary Fig. S4b). Nevertheless, striking genotype differences were evident for PPI (Fig. 6). Here, responses to LSD and to the 0.05 MDL plus LSD treatments were reduced in WT relative to the βArr2-KO mice (p-values≤0.001). In WT animals, LSD suppressed PPI compared to the MDL and vehicle controls (p=0.001). PPI was normalized when 0.1 mg/kg MDL was given with LSD. By dramatic comparison, LSD was completely without effect in the βArr2-KO mice. Collectively, these ndings show that LSD disrupts PPI in both genotypes of the βArr1 mice. PPI was aberrant also the WT βArr2 animals. The 5-HT2AR antagonist restored PPI in both WT strains, whereby haloperidol was required to normalize it in βArr1-KO mice. By contrast, PPI in βArr2-KO mice was unaffected by LSD.
Effects of Arrb1 or Arrb2 deletion on 5-HT2AR expression. We examined whether deletion of Arrb1 or Arrb2 could alter 5-HT2AR expression by radioligand binding with brains from WT and βArr1-KO, and WT and βArr2-KO littermates. When [ 3 H]-ketanserin competition binding was examined, displacement with DOI and Ki values were found to be very similar with membranes from the WT and βArr1-KO and the WT and βArr2-KO brains (Fig. 7a). We examined also 5-HT2AR immuno uorescence in βArr1 and βArr2 brain sections (Fig. 7b-e). Here, we detected no apparent alterations in the relative receptor distributions among the genotypes. Together, these results are consistent with the hypothesis that neither global Arrb1 nor global Arrb2 genetic deletion decreases 5-HT2A receptor expression.

Discussion
In the present study, we analyzed whether global deletion of Arrb1 or Arrb2 was involved in LSDstimulated responses in mice. In many cases, we found that LSD modi ed behaviors in both strains of WT mice, as well as in the βArr1-KO animals. By contrast, LSD-induced responses in the βArr2-KO animals were either minimal or non-existent. Collectively, these results suggest the LSD-stimulated responses require βArr2. In this regard, βArr2 is reported to play a similar role in morphine-stimulated hyperlocomotion 46 and amphetamine-stimulated locomotor and rearing activities in βArr2 mice 47 .
While we found LSD stimulates locomotion in mice, in rats it has been reported to decrease ambulation 35 or increase locomotion 32,33,36 . While an inhibitory response to 0.2 mg/kg LSD was observed in rats, we only saw stimulatory effects with 0.3 mg/kg LSD and in pilot studies, doses of 0.1 to 0.5 mg/kg LSD were all stimulatory. An absence of LSD inhibitory effects could be attributed to differences in species tested, test environment and apparatus, and/or test procedure. For instance, in humans LSD's behavioral effects can be context speci c 1,2 and our 30 min habituation to the open eld prior to LSD administration may have reduced emotionality in our mice, such that only the stimulatory effects of LSD were evident.
To determine whether the locomotor-stimulating effects of LSD were due to 5-HT2AR activation, MDL was used as an antagonist. When used alone, this antagonist exerted no effects on motor performance in either βArr mouse strain. Importantly, 0.1 and 0.5 mg/kg MDL blocked the locomotor-stimulating effects of LSD in both WT strains and in the βArr1-KO animals. A similar effect has been observed in rats 36 . Hence, the present results indicate that the LSD-induced hyperactivity in βArr mice is promoted through the 5-HT2AR.
Besides motor activity, we examined the effects of LSD on HTRs, grooming, retrograde walking, and nosepoking behaviors. LSD and other psychedelics are well-known to stimulate HTRs in mice 17,38,41 and this behavior has been proposed as a proxy for hallucinations in humans 12 . Compared to vehicle, LSD stimulated HTRs to similar extents in WT and βArr1-KO mice. In βArr2-KO animals, this response was severely blunted compared to the WT controls. These results were unexpected since the individual competition binding curves could be superimposed among the different genotypes. Regardless, in both βArr1 and βArr2 mice, MDL reduced HTRs to levels of the vehicle controls. These ndings are consistent not only with the known action of MDL on blocking HTRs to various hallucinogens [14][15][16] , but also on the inability of LSD and other psychedelics to induce this response in the htr2A homozygous mutant mice 17,18,38 .
Aside from HTRs in rodents, LSD accentuates grooming behaviors in cats 48 and it can stimulate or inhibit grooming in mice 39,40 . In our investigations, LSD augmented grooming in both WT strains, and in βArr1-KO animals. By comparison, this psychedelic was ineffective in βArr2-KO mice. In both WT strains and in βArr1-KO animals, 0.1 and 0.5 mg/kg MDL returned the LSD-stimulated grooming to control levels. Thus, antagonism of the 5-HTR2A was su cient to restore LSD-induced grooming to baseline.
Effects of LSD were examined also for the organization of grooming behavior. Under vehicle treatment, all mice displayed similar patterns of grooming that began with the face, progressed to the anks, and ended with the feet or tail. LSD disturbed this sequence of events in both WT strains and in βArr1-KO mice. By comparison, grooming in the βArr2-KO mice was largely unaffected by LSD. MDL did not alter grooming in the WT and βArr1-KO mice, whereas it prolonged grooming and promoted twitching of the neck and back muscles in βArr2-KO animals. This 5-HT2AR antagonist blocked the LSD-disrupting effects on the organization of grooming in WT mice and it mostly restored it in βArr1-KO animals. The MDL-LSD combination in βArr2-KO animals produced some disturbances, but the mice typically completed the grooming sequence. Together, these results suggest that additional receptor systems may be involved in the LSD-induced grooming responses. LSD effects on retrograde walking and nose-poking responses were examined also. We found LSD to stimulate these behaviors in WT animals from both strains, as well as in the βArr1-KO mice. However, LSD promoted neither response in βArr2-KO animals. Nevertheless, in the other genotypes MDL restored retrograde walking and nose-poking to the levels of vehicle controls. Hence, this 5-HT2AR antagonist normalized these LSD-stimulated behaviors.
LSD-induced states share many similarities with the early acute phases of psychosis 2 . PPI is abnormal in individuals diagnosed with schizophrenia 49 and LSD disrupts PPI in rats 36,39,44 . In βArr1 mice, LSD impaired PPI in both genotypes without affecting startle or null activities. Both 0.1 and 0.5 mg/kg MDL restored the LSD-disrupted PPI, but only in WT mice; an effect consistent with the action of the 5-HT2AR antagonist MDL11,939 in rats 44 . By comparison, MDL was ineffective in blocking the LSD effects in βArr1-KO animals. Since LSD activates human dopamine D2 receptors 8,50 , we used haloperidol as a D2 antagonist. We found this antagonist to restore the LSD-disrupted PPI in the βArr1-KO mice. Parenthetically, both 0.1 and 0.2 mg/kg haloperidol failed to rescue PPI in rats given 0.1 mg/kg LSD (s.c.) 36 ; the possible reasons for this discrepancy in mice versus rats are unclear. When βArr2 mice were tested, LSD disrupted PPI selectively only in WT mice. Notably, βArr2-KO mice were completely unresponsive to this psychedelic. As with WT animals from the βArr1 strain, MDL also normalized the LSD-disrupted PPI in the WT βArr2 mice. Thus, the LSD effects on PPI in the βArr mice are complex, with restoration of PPI with MDL in both strains of WT mice, normalization of PPI with haloperidol in βArr1-KO animals, and without any discernable effect in βArr2-KO subjects.
LSD and other psychedelics are well-known for their hallucinogenic actions 1 and these responses have been attributed to 5-HT2AR agonism 11 . We observed LSD to stimulate motor activity, head twitches, grooming, retrograde walking, and nose-poking in both strains of WT mice and in βArr1-KO animals. LSD also disrupted PPI in these same genotypes. The LSD-elicited responses in βArr2-KO mice were either signi cantly attenuated or completely absent. In conditions where LSD produced changes in behavior, these alterations were blocked with the 5-HT2AR antagonist MDL. While these results suggest that the 5-HT2AR is an essential component for all these responses, it should be recalled that LSD exerts a plethora of actions at many GPCRs 8-10 and, aside from HTRs, other behaviors are inconsistently affected by hallucinogens 17 . Hence, it is possible that LSD's effects on the 5-HT2AR are involved in a cascade of many GPCR-signaling events mediating these varied responses.
Like other GPCRs, agonist actions at the 5-HT2AR can lead to G protein-dependent and -independent signaling, the latter of which involves βArr [22][23][24] . While both βArr1 and βArr2 are expressed ubiquitously in adult rodent brain, expression of βArr2 mRNA is much higher than that for βArr1--except in selected brain areas 51 . Thus, it may be surprising that the LSD-elicited responses were less disturbed in the βArr1-KO than in the βArr2-KO mice, because in the βArr1-KO animals βArr2-mediated signaling is still retained. In this regard, it is especially intriguing that LSD-induced HTRs were much more robust in both WT strains and in the βArr1-KO animals, than in the βArr2-KO mice. Our results with LSD suggest that βArr2 may be essential for the expression of hallucinogenic-like actions at the 5-HT2AR.

Methods
Subjects. Adult male and female WT and βArr1-KO, and WT and βArr2-KO mice were used in these experiments [30][31] . All mice had been backcrossed onto a C57BL/6J genetic background. Heterozygotes were used to generate the respective WT and KO animals. The mice were housed 3-5/cage in a temperature-and humidity-controlled room on a 14:10 h (lights on at 0600 h) light-dark cycle with food and water provided ad libitum. All experiments were conducted with an approved protocol from the Duke University Institutional Animal Care and Use Committee and all experiments and methods were performed in accordance with the relevant regulations and ARRIVE guidelines. Head twitch, grooming, and retrograde walking. These behaviors were lmed during assessment of motor activity. The responses were scored by blinded observers over the rst 30 min following injection of the vehicle or LSD after collection of baseline activity. The data are expressed as the numbers of head twitches, duration of grooming, and incidences of retrograde walking.
Nose-poking responses. Nose-pokes were monitored in a 5-choice serial reaction-time apparatus (Med Associates Inc., St. Albans, VT) 52 . Each chamber had ve LED-illuminated 1.24 cm 2 nose-poke apertures with infrared diodes to register nose pokes. No food or liquid reward was available. Mice were injected with the vehicle or different doses of MDL and returned to their home-cages. Thirty min later, the animals were injected with the vehicle or different doses of LSD and were placed immediately into the operant chambers for 30 min. The data are depicted as the numbers of head pokes.
Prepulse inhibition (PPI). PPI of the acoustic startle response was conducted using SR-LAB chambers (San Diego Instruments, San Diego, CA) as reported 45 . Mice were injected with vehicle or different doses of MDL or with 0.1 mg/kg haloperidol and returned to their home cages. Fifteen min later the animals were given the vehicle or different doses of LSD and were placed into the apparatus. After 10 min of habituation to a white noise background (64 dB), testing began. Each test consisted of 42 trials with 6 null trials, 18 pulse-alone trials, and 18 prepulse-pulse trials. Null trials comprised the white noise background, pulse trials consisted of 40 ms bursts of 120 dB white-noise, and prepulse-pulse trials were composed of 20 ms pre-pulse stimuli that were 4, 8, or 12 dB above the white-noise background (6 trials/dB), followed by the 120 dB pulse stimulus 100 ms later. Testing commenced with 10 pulse-alone trials followed by combinations of the prepulse-pulse and null trials, and it terminated with 10 pulse-alone trials. PPI responses were calculated as %PPI = [1-(pre-pulse trials/startle-only trials)]*100.
Statistics. All statistical analyses were performed with IBM SPSS Statistics 27 programs (IBM, Chicago, IL). The data are presented as means and standard errors of the mean. No sex effects were detected in any experiments. Hence, this variable was collapsed. All data were normally distributed. One-or two-way ANOVA, repeated measures ANOVA (RMANOVA), or analyses of covariance (ANCOVA) were used to analyze the data, followed by Tukey or Bonferroni post-hoc analyses. A p<0.05 was considered signi cant. All results were plotted using GraphPad Prism.

Competing interests
No competing interests for this work by any of the authors.

Data availability
Data that support this study are available from the corresponding authors upon reasonable request.  Supplementary  Table S1. A two-way ANOVA failed to identify any signi cant effects for baseline locomotion; separate     Effects of LSD and MDL100907 on behavioral responses in β-arrestin 2 mice. A description of the experimental design is presented in the Figure 3 legend