The non-receptor tyrosine kinase Pyk2 modulates acute locomotor effects of cocaine in D1 receptor-expressing neurons of the nucleus accumbens

The striatum is critical for cocaine-induced locomotor responses. Although the role of D1 receptor-expressing neurons is established, underlying molecular pathways are not fully understood. We studied the role of Pyk2, a non-receptor, calcium-dependent protein-tyrosine kinase. The locomotor coordination and basal activity of Pyk2 knock-out mice were not altered and major striatal protein markers were normal. Cocaine injection increased Pyk2 tyrosine phosphorylation in mouse striatum. Pyk2-deficient mice displayed decreased locomotor response to acute cocaine injection. In contrast, locomotor sensitization and conditioned place preference were normal. Cocaine-activated ERK phosphorylation, a signaling pathway essential for these late responses, was unaltered. Conditional deletion of Pyk2 in the nucleus accumbens or in D1 neurons reproduced decreased locomotor response to cocaine, whereas deletion of Pyk2 in the dorsal striatum or in A2A receptor-expressing neurons did not. In mice lacking Pyk2 in D1-neurons locomotor response to D1 agonist SKF-81297, but not to an anticholinergic drug, was blunted. Our results identify Pyk2 as a regulator of acute locomotor responses to psychostimulants. They highlight the role of tyrosine phosphorylation pathways in striatal neurons and suggest that changes in Pyk2 expression or activation may alter specific responses to drugs of abuse, or possibly other behavioral responses linked to dopamine action.


Results
Pyk2 protein is enriched in ventral D1 SPNs. We used immunohistochemistry to evaluate the regional distribution of Pyk2 protein in the striatum. Pyk2 immunoreactivity appeared slightly stronger in the NAc than in the DS ( + 15%, Fig. 1A right panel, unpaired t test, t 34 = 4.3, p = 10 −4 , see Supplementary Table 1 for details of all statistical analyses), in agreement with a previous study of mRNA distribution 27 . Since Pyk2 striatal immunoreactivity displayed an irregular pattern, we assessed whether its distribution followed the neurochemically-defined patch/matrix compartmentalization [28][29][30] defined by calbindin, a matrix-enriched protein (Fig. 1B). Pyk2 appeared more expressed in calbindin-positive neurons indicating an enrichment of Pyk2 in matrix compartment (+21%, Fig. 1C, n = 20 regions from 3 mice, two-tailed Mann Whitney's test, U 20,20 = 12, p < 10 −4 ). To evaluate the relative expression of Pyk2 in neurons of the direct and indirect pathways, we then compared its decrease following conditional deletion in either of these populations, as compared to their matched controls. Pyk2 striatal protein levels were decreased by 67% in Pyk2 f/f;D1::Cre mice (Fig. 1D, n = 11 mice per group, Mann-Whitney test, U 11,11 = 0, p < 10 −4 ). This result should be taken with caution since the expression of D1R is wider during development of the striatum than in the adult 31,32 and could lead to a broader developmental action of D1::Cre and subsequent overestimation of the apparent enrichment of Pyk2 in D1 neurons. However, Pyk2 decrease in Pyk2 f/f;A2A::Cre mice was limited to 36%, as compared to matched Pyk2 f/f mice (Fig. 1E, Mann-Whitney test, U 7,5 = 4, p = 0.03), supporting the hypothesis of a relatively higher expression of Pyk2 protein in D1R-expressing SPNs than in A 2A Rpositive neurons. In conclusion, although Pyk2 is expressed in all SPNs it appears to be slightly more abundant in D1 neurons, and in the DS matrix and the NAc.
Pyk2 knockout does not impair motor coordination. To study the role of Pyk2 in the striatum, we first assessed motor coordination. Pyk2 +/+ and Pyk2 -/mice were trained on an accelerating rotarod for 3 days and their motor coordination was evaluated by measuring the time they were able to stay on the rod during each trial. Mice of both genotypes displayed similar performances, improving their ability to stay on the rod across trials ( Fig. 2A, two-way ANOVA, trial number effect, F 11,191 = 14.03, p < 10 −4 ; genotype effect, F 1,191 = 2.57, p = 0.11, see Supplementary Table 1 for details of all statistical analyses). Similarly, there was no alteration of rotarod performance following specific AAV-mediated Cre deletion of Pyk2 in the NAc (Fig. 2B, two-way ANOVA, trial as percentage of wild type mean density (6-22 mice per group). (H), Immunoblot of Gαolf, DARPP-32, synapsin 1, tyrosine hydroxylase (TH), and actin. (I), Results as in H were quantified and analyzed as indicated in G (6-7 mice per group). In A, C-E, G, and I, means ± SEM are indicated, 2-tailed Mann and Whitney test, *p < 0.05, **p < 0.01, ****p < 0.0001. See Supplementary Table 1 for all details of statistical analyses and  Supplementary Figures 2 and 3 for full length blots. Pyk2 is activated by cocaine and involved in its acute locomotor effects but not in their sensitization. Considering the relative enrichment of Pyk2 in the NAc, a region involved in reward response and addiction, we assessed a possible role of Pyk2 in the effects of cocaine. We immunoprecipitated Pyk2 from striatal extracts of mice killed 10 min after the injection of saline vehicle or cocaine (20 mg/kg), a time previously shown to allow detection of cocaine-induced increased tyrosine phosphorylation of SFK or NMDA-R 8 , and measured its tyrosine phosphorylation by immunoblotting (Fig. 3A). Tyrosine phosphorylation of Pyk2 was increased 10 min after cocaine injection, indicating its activation in response to cocaine injection (Fig. 3B, unpaired t test, t 12 = 2.49, p = 0.029). We then explored a possible role of Pyk2 in the acute response to cocaine by comparing locomotor activity of Pyk2 +/+ and Pyk2 -/mice after cocaine injection. Basal locomotor activity was similar in the two groups of mice, whereas cocaine-induced hyperlocomotion was decreased in Pyk2 -/mice (Fig. 3C, two-way ANOVA, genotype effect, F 1,306 = 39.05, p < 10 −4 , see Supplementary Table 1 for details of all statistical analyses). Repeated cocaine exposure increases behavioral responses due to a sensitization mechanism 36 , which is clearly visible following a single cocaine administration 37 . We therefore injected cocaine a second time, thirteen days later, to assess locomotor sensitization (Fig. 3D). Locomotor response to the second injection of cocaine was increased in both Pyk2 +/+ and Pyk2 -/mice but there was no difference between the two genotypes ( Fig. 3D, two-way ANOVA, genotype effect, F 1,306 = 1.59, p = 0.21). Paired analysis showed a significant increase in locomotor activity in both genotypes (Fig. 3E, two-way ANOVA, injection effect:  www.nature.com/scientificreports www.nature.com/scientificreports/ compared the sensitization ratio (total locomotion during the first 15 min after the 1 st injection/total locomotion in the same period after the second injection) in the 2 genotypes. These ratios were similar in the two groups ( Fig. 3F, unpaired t test, t 17 = 1.24, p = 0.23). To test the possibility that decreased locomotor activity was linked to increased time spent in stereotypies we quantified the number of rearings and the time spent grooming during 40 min after the first cocaine injection, in 6 randomly selected mice in each group ( Supplementary Fig. 1). There was no significant difference between Pyk2 +/+ and Pyk2 -/mice for these two parameters. The limited number of rearings (<100) and time spent grooming (<100 sec) during the 40-min after cocaine injection make it unlikely that these activities interfered with locomotor activity. Therefore our results indicate a predominant role of Pyk2 in the acute cocaine-induced locomotor response but not in its sensitization.

Pyk2 deletion in NAc, but not DS, recapitulates the effects of full KO on cocaine-induced locomotion.
To identify the striatal region implicated in the effects of Pyk2 deletion, we studied cocaine responses in Pyk2 f/f mice bilaterally injected with AAV-GFP-Cre in the NAc. The behavior was investigated 3 weeks after the injection and at the end of the behavioral tests, the mice were sacrificed and the position of the tip of the injection needle, Pyk2 immunoreactivity, and GFP expression were checked (Fig. 4A). In mice injected with AAV-GFP-Cre, Pyk2 immunoreactivity was decreased in the GFP-expressing area (Fig. 4A). Animals in which the injection was not correctly located were not included in the statistical analysis. In mice injected with AAV-GFP-Cre in the NAc, basal locomotor activity was unchanged, but acute locomotor response to cocaine was decreased as compared to mice injected with AAV-GFP (Fig. 4B, two-way ANOVA, Cre effect, F 1,342 = 12.94, p = 0.0004, see Supplementary Table 1 for details of all statistical analyses). Following a second injection of cocaine at day 14, the locomotor response increased in the two groups of mice and the difference between them was blunted (
Pyk2 knockout selectively blunts the acute locomotor effects of a D1R agonist but not a cholinergic antagonist. Since the consequences of the absence of Pyk2 on acute locomotor effects of cocaine appeared to result from Pyk2 deficit in D1 neurons, we examined whether the effects of D1R stimulation were Locomotor activity of Pyk2 f/f;DS,GFP (n = 8 mice) and Pyk2 f/f;DS,GFP-Cre mice (n = 7) after the first (G) and the second (H, 13 days later) injection of 20 mg/kg cocaine. I, Pairwise comparison of the distance traveled 0-15 minutes following the first and the second injection in Pyk2 f/f;DS,GFP and Pyk2 f/f;DS,GFP-Cre mice (from data in G and H). J, Sensitization ratios, as in E, in Pyk2 f/f;DS,GFP and Pyk2 f/f;DS,GFP-Cre mice. (B-E and G-J), Data are means ± SEM, indicated by a shaded area in B, C, G, and H, and horizontal bars in E and J. In B-D and G-I, data were analyzed with 2-way ANOVA followed by post-hoc Sidak's multiple comparisons test, and in E and I by Mann and Whitney test. ns, not significant, *p < 0.05, **p < 0.01, ***p < 0.001. See Supplementary Table 1    www.nature.com/scientificreports www.nature.com/scientificreports/ altered. We used SKF-81297, a D1R selective agonist, known to increase locomotor activity [38][39][40] . The acute locomotor response to SKF-81297 was slightly reduced in Pyk2 f/f;D1::Cre mice as compared to Pyk2 f/f mice (Fig. 6A, genotype effect, F 1,801 = 5.80, p = 0.016). These animals received a second injection of SKF-81297 13 days later that triggered a stronger locomotor response than the first injection, without difference between genotypes (Fig. 6B, genotype effect, F 1,752 = 0.36, p = 0.55). Interestingly, both the first and second injection of SKF-81297 induced a biphasic effect on locomotor activity, not previously reported to our knowledge. Inspection of the videos suggested the existence of stereotypies at the trough between the two peaks of locomotor activity. Quantification of grooming duration showed no significant difference between the two genotypes but an increased time spent grooming 40-45 min after the beginning of the recording (i.e. 10-15 min post-injection) as compared to 35-40 min after the beginning of the recording (i.e. 15-20 min post-injection, Fig. 6C) strongly indicating that the presence of stereotypy accounted for the transiently decreased locomotion. As reported in previous studies [41][42][43] , a sensitization of the response was observed following the second injection of SKF-81297. We analyzed the sensitization during the 45 minutes following injection (Fig. 6D and E). The distance traveled after the second injection of SKF-81297 was increased (Fig. 6D, two-way ANOVA, injection effect: F 1,42 = 83.86, p < 10 −4 ; Sidak's multiple comparisons post hoc tests: Pyk2 f/f , t 42 = 5.19, p < 10 −4 , Pyk2 f/f;D1::Cre , t 42 = 7.90, p < 10 −4 ) but the sensitization ratios did not differ between genotypes (Fig. 6E, Student's t test, t 42 = 1.61, p = 0.11).
We then sought to determine whether Pyk2 could decrease any kind of drug-induced locomotor activity by testing the effects of trihexyphenidine (THX), an anticholinergic substance that increases locomotor activity 44 . There was no difference in locomotor activity between Pyk2 f/f;D1::Cre and Pyk2 f/f mice following the first (Fig. 6F, two-way ANOVA, genotype effect: F 1,414 = 0.004, p = 0.95) or second injection of THX (Fig. 6G, F 1,396 = 0.44, p = 0.51). No sensitization of the response was observed in either genotype. These results indicate that deletion of Pyk2 in D1R-expressing neurons has a slight effect on the locomotor effects of the first injection of a D1 agonist, but not of an anticholinergic agent. In addition, we provide evidence of a biphasic effect of SKF-81297 and a sensitization of both components upon a second administration of the drug.

Discussion
In this study, we show the involvement of the non-receptor tyrosine kinase Pyk2 in the acute locomotor response to cocaine. We demonstrate that complete knockout of Pyk2 or its specific deletion in the NAc or in D1R-expressing neurons decreases acute cocaine-induced hyperlocomotion. Acute locomotor response to a selective D1 agonist but not to a cholinergic antagonist was also slightly altered. These results indicate a role of Pyk2 located in D1R-expressing neurons of the NAc in the D1R-induced acute locomotor response. In contrast, no alteration was observed in locomotor sensitization or cocaine-induced CPP.
Our immunofluorescence data in the striatum agree with previous results suggesting the enrichment of Pyk2 in the ventral part of the striatum 27 . We also provide evidence for a relative enrichment in D1R-expressing SPNs, which form the 'direct pathway' 48 , and in the matrix compartment of the striatum, which plays a specific yet incompletely characterized role in striatal function 49 . The dorsal striatum is implicated in the control of movement and in learning skilled motor tasks 50,51 . We assessed motor coordination of mice with accelerating rotarod training 52 , which can be impaired by alteration of several brain areas, including basal ganglia, motor cortex, and cerebellum 53 . We did not observe any difference between any of the complete or conditional Pyk2 mutant mice we generated and their respective controls, suggesting that Pyk2 is not required for the motor function of the dorsal striatum or other brain regions. Pyk2 appears to be more expressed in the NAc, which is a key component in brain circuitry underlying drug-evoked behaviors 54,55 . We demonstrated that tyrosine phosphorylation of Pyk2 www.nature.com/scientificreports www.nature.com/scientificreports/ www.nature.com/scientificreports www.nature.com/scientificreports/ in the striatum was increased following a single injection of cocaine. Pyk2 is activated by Ca 2+ -induced autophosphorylation of Tyr-402, followed by recruitment of SFKs which in turn phosphorylate Pyk2 on other tyrosine residues, although in some cells activation of SFKs can be the triggering mechanism (see 17 for a review). Cocaine injection in naïve mice increases the concentration of calcium in D1 SPNs 56 providing a possible basis for Pyk2 activation. In addition, cocaine-induced activation of Fyn has been reported in the striatum and proposed to be involved in NMDA receptors regulation and their synergism with D1R to activate the ERK pathway 8 . Although these results suggested a possible role of Pyk2 in ERK activation, ERK phosphorylation was still observed in Pyk2-deficient mice, underlining the redundancy in the mechanisms controlling the ERK pathway.
Concerning the functional consequences of Pyk2 knock-out, we found that the acute cocaine locomotor response was decreased, whereas locomotor sensitization and CPP, which are known to involve synaptic plasticity and ERK activation (see 7 for a review), were not altered. Although our results did not indicate effects of Pyk2 mutation on other acute effects of cocaine or D1 receptor agonist such as stereotyped behavior, a more detailed study would be necessary to evaluate minute specific alterations. The acute locomotor response was also impaired in mice specifically lacking Pyk2 in the NAc or in D1R-expressing neurons, but not in those devoid of Pyk2 in the DS or in A 2A R-expressing neurons. In mice lacking Pyk2 in D1 neurons a decreased locomotion was even observed after the second cocaine injection. This combination of results implicates an alteration induced by the absence of Pyk2 in the D1R-expressing neurons of the NAc. This conclusion is in line with the role of these neurons in the acute locomotor effects of psychostimulants [57][58][59] . A decreased locomotor response without alteration in locomotor sensitization or CPP was previously observed in Gαolf heterozygous (Gnal +/− ) mutant mice in which cAMP production is decreased 60 . These similarities could indicate a modulation of the D1R/Gαolf/cAMP pathway by Pyk2, a hypothesis supported by the attenuated locomotor response to the D1 agonist SKF-81297 in Pyk2 f/f;D1::Cre mice. However, Gαolf levels were not changed in the mutant mice (see Fig. 1H and I) and phosphorylation of cAMP-dependent protein kinase substrates was not altered (de Pins and Girault, unpublished observation). The only effect we observed was a blunting of GluN2B receptor phosphorylation at Tyr1472, in agreement with the role of Pyk2 in the phosphorylation of NMDAR 11 , but unlikely to explain by itself the effects on locomotor activity. Therefore we hypothesize that Pyk2 is also implicated in the regulation of other signaling mechanisms important for D1R-mediated regulation of locomotor activity. Further work will be necessary to explore this hypothesis. Our study clearly identifies the significant but circumscribed contribution of Pyk2 in NAc D1 neurons to the acute locomotor effects of cocaine. It raises the question of its role in other dopamine-mediated actions and opens novel avenues for investigating the role of tyrosine phosphorylation and non-receptor tyrosine kinases signaling in striatal neurons.

Materials and Methods
Animals. Floxed Pyk2 mice (Pyk2 f/f ) were generated by the insertion of LoxP sequences (Gen-O-way, Lyon, France) surrounding PTK2B exons 15b-18 coding for the kinase domain 61 . Homologous recombination was carried out in C57/Bl6 embryonic stem cells and germline transmission of the mutated allele was achieved in the same background. Floxed Pyk2 mice were initially bred to Cre-deleter mice to generate constitutive knockout www.nature.com/scientificreports www.nature.com/scientificreports/ mice, or to Drd1::Cre mice, Tg(Drd1a-cre)EY262Gsat 62 , or Adora2a::Cre mice, Tg(Adora2a-cre)2MDkde 63 , to generate conditional KO mice (Pyk2 f/f;D1::Cre and Pyk2 f/f;A2A::Cre , respectively). Mice were housed at 19-22 °C with 40-60% humidity, under a 12:12 h light/dark cycle, and had ad libitum access to food and water. Animal experiments and handling were in accordance with ethical guidelines of Declaration of Helsinki and NIH, (1985-revised publication no. 85-23, European Community Guidelines), and French Agriculture and Forestry Ministry guidelines for handling animals (decree 87849, licence A 75-05-22) and approval of the Charles Darwin ethical committee APAFIS#8861-2016111620082809. Mice used in this study were 3-6-month-old males.

Viral vectors and stereotaxic injection.
For deletion of Pyk2 in the NAc or dorsal striatum (DS), 3-month Pyk2 f/f mice were stereotaxically injected with AAV expressing an enhanced green fluorescent protein (EGFP) Cre recombinase fusion protein (AV-9-PV2521, AAV9.CamKII.HI.eGFP-Cre.WPRE.SV40, Perelman School of Medicine, University of Pennsylvania, USA), here referred to as AAV-GFP-Cre. As control, we injected AAVs expressing GFP (AV-9-PV1917, AAV9.CamKII0.4.eGFP.WPRE.rBG, same source), referred to as AAV-GFP. Following anesthesia with pentobarbital (30 mg kg −1 ), we performed bilateral stereotaxic injections of AAV-GFP or AAV-GFP-Cre (2.6 × 10 9 GS per injection) in the NAc at the following coordinates 64 from the bregma (millimeters), anteroposterior, 1.3, lateral, ±1.3, and dorsoventral, −4.5 or in the DS, anteroposterior, 0.9, lateral, ±1.5, and dorsoventral, −2.75. AAV injection was carried out in 2 min. The cannula was left in place for 5 min for complete virus diffusion before being slowly pulled out of the tissue. Mice were placed on a warm plate for 2 h after surgery, received a subcutaneous injection of a non-steroidal anti-inflammatory drug (meloxicam, 2 mg/kg) during 3 days, and allowed to recover for 3 weeks before starting behavioral experiments.
Behavioral experiments. Rotarod. 4-month mice were trained at accelerating speed (4-40 rpm in 5 min), with four sessions per day for three consecutive days and the latency to fall was recorded.
Locomotor activity. Mice were placed either in a open-field chamber (50 cm × 50 cm, L x W) for cocaine response or in a 20-cm diameter cylinder for SKF-81297 [SKF] or trihexyphenidyl [THX] response. After 30 minutes, mice were i.p. injected with cocaine (20 mg/kg), SKF (3 mg/kg), or THX (15 mg/kg) and placed back in the chamber for 1 hour. Locomotion was recorded using an overhead digital camera. The distance traveled was measured in 5-min bins using EthoVision software (Noldus, Wageningen, the Netherlands).
Rearing and grooming. These two behaviors were manually scored during visualization of the videos recorded to measure locomotor activity during the period indicated in the Results.
Conditioned place preference. Conditioned place preference (CPP) was performed in two compartments of a Y-shaped maze (Imetronic, Pessac, France) with different wall textures and visual cues as follows. (i) Pretest: day-0, mice were placed in the center of the apparatus and allowed to explore freely both compartments for 20 min. The time spent in each compartment was recorded and the preferred and un-preferred compartments deduced for every mouse. (ii) Conditioning: day-1, mice were injected with saline and placed immediately in the preferred compartment for 15 min. The next day, they were placed in the other closed compartment after cocaine injection (15 mg/kg). This was repeated twice (3 saline-, 3 cocaine-pairings in total). (iii) Test: time spent in each compartment was measured on day-7 during 20 min. The CPP-score was calculated as the time spent in the cocaine-paired compartment during the test minus the time spent in this compartment during the pre-test.
Immunobloting. For the analysis of striatal proteins untreated mice were euthanized by cervical dislocation, striata dissected out, frozen using CO 2 pellets and stored at −80 °C until use. For pharmacological responses mice were i.p. injected with cocaine (20 mg/kg) or saline and placed in a 43 cm × 27 cm cage. After 10 minutes, mice were euthanized and heads were dipped in liquid nitrogen for 12 seconds. The frozen heads were cut into 210-μm-thick slices with a cryostat, and 10 frozen microdisks (1.4 mm diameter) were punched out bilaterally from the striatum and stored at −80 °C until use. Tissue samples were sonicated in 10 g/L SDS and 1 mM sodium orthovanadate in water, and placed at 100 °C for 5 min. Extracts (15 μg