Social play behavior is driven by glycine-dependent mechanisms

histology. The LEDs were controlled remotely via a smartphone using a self-made program. Optical activation of the WiChR was carried out with constant blue light bilaterally for 2 minutes.


In brief
Dvorzhak et al. show that the age-related decline in the social playfulness of rats depends in part on glycinergic synaptic transmission in the periaqueductal gray (PAG), a structure whose neuronal activity is required for social play.Glycine stimulates social play through NMDA receptors potentiation, which increases neuronal activity in the PAG.

INTRODUCTION
Social play is a repeated, incompletely functional behavior differing from more serious versions structurally, contextually, or ontogenetically, and is initiated voluntarily when the animal is in a relaxed or low-stress setting. 1 Social play behavior is widespread among young mammals and is also observed in other vertebrates and invertebrates, [1][2][3] but compared with other behaviors, it is still the least studied at the level of neural circuitry.Remarkably, we know very little about which brain areas are involved, and have no understanding of the molecular and cellular mechanisms underlying social play behavior. 1,4On the one hand, the fact that vertebrates and invertebrates play suggests that this type of behavior is independent of brain type and has some common mechanisms at the molecular and cellular level.On the other hand, the social play has a structure that requires coordination and, consequently, certain play centers.Social play behavior is evolutionarily ancient among vertebrates and is found even in sharks, 1 therefore play initiating centers should also be phylogenetically ancient.In addition, social play, such as rough and tumble play, is related to inter-partner communication and the expression of emotions. 5We hypothesized that the periaqueductal gray (PAG), a midbrain structure with a key role in the regulation of emotion and emotion-driven action selection, [6][7][8] as well as fight or flight behavior, 9,10 may also be crucial for the generation of social play behavior. 4ugh and tumble play (also called play fighting) is one of the most common forms of social play, 11 is particularly intense in juvenile mammals, and declines with age. 12,13To identify the neural mechanisms involved, we used the age dependence of social play to identify candidate mechanisms that co-correlate with the age-related decline in playfulness.Our analysis shows that glycinergic (Gly) synaptic transmission is causally linked to the age-related decline of social play behavior.

Social play
First, we established an animal model for social play behavior (rough and tumble play) in rodents. 5,146][17] Play activity is defined as the summed time of pouncing, pinning, and boxing divided by the total time of observation. 16,18Initially, we analyzed the behavior of mice-the most common animal model in neurosciences.However, mice showed a very limited amount of social play behavior (Figures 1B and 1C).In sharp contrast, play behavior was pronounced in male rats (Figures 1B and 1C, ANOVA F (2, 58) = 25.10,p < 0.0001).Non-playful social activity such as for example anogenital sniffing was comparable between male mice and rats (Figure 1D, Kruskal-Wallis statistic = 1.95, groups n = 3, p = 0.4).Therefore, we decided to continue our investigations on social (C) Juvenile (4-6 weeks) rats but not juvenile mice (5 weeks) show rough and tumble play, which is suppressed in adult rats (47-55 weeks).(D) Juvenile mice and rats as well as adult rats spend similar time for anogenital sniffing, corresponding to non-playful social interactions.(E) Based on the Paxinos atlas scheme of the implantation of the guiding cannulas (in blue) and the positioning of the injecting cannulas (in red).(F) PAG photo showing the distribution of the green dye (DiI) injected bilaterally according to the scheme in (D).The red dotted line indicates the boundaries of the PAG.(G) Bilateral injection of 2 mL of 2% lidocaine and 100 mM muscimol through guide cannulas into the PAG almost completely blocks play behavior in injected rat.The day after the muscimol injection, the rat's playfulness completely recovers, showing specificity of the effect to the drug but not to the PAG damage.CTRL means injection of isotonic 0.9% NaCl.(legend continued on next page) play behavior in male rats.Consistent with previous work, 13,19 we found a striking age-dependency of play (Figures 1B and 1C) with age-independent non-playful social activity (Figure 1D).Aggressive attacks, another non-playful social activity that may look like rough and tumble play but is distinguished from it by aggressive posturing, piloerection, serious biting, and so on, 20,21 are not observed in juvenile male rats, 13 but we also did not see them in our experiments with adults (Video S6).In the next set of experiments, we directly tested if the PAG is involved in social play.We used different pharmacological and optogenetic tools to analyze the role of the PAG in this type of behavior.Bilateral injection of the sodium channel blocker lidocaine through guide cannulas into the PAG almost completely blocks play behavior (Figures 1E-1G; Video S1, sum of signed ranks (W) = À21, p = 0.03).Lidocaine not only blocks spiking in the somatic and dendritic compartments but also the axonal propagation and could thus block play within the PAG in an unspecific manner.Indeed, lidocaine injection causes motor impairment, leading to a significant decrease in the distance traveled in the open field test (Figure 1H, Student's t test, t = 4, df = 5, p = 0.01).Therefore, we repeated the same type of experiment with the specific g-aminobutyric acid (GABA)-a-receptor agonist muscimol.Muscimol (100 mM) also efficiently blocks social play (Figures 1E-1G; Video S2, Friedman statistic = 11, groups n = 3, p = 0.0012), but does not cause motor impairment (Figure 1H, ANOVA (1.168, 7.006) = 0.13, p = 0.76).It is important to point out that these experiments were completely reversible (Figure 1G).To further increase specificity, we performed optogenetic experiments with the recently developed potassium-based inhibitory opsin called WiChR 22 (see also Figure S1).Following an AAV-hSyn-WiChR-mScarlet virus injection, we implanted wireless light-emitting diodes (LEDs).Remarkably, light stimulation of WiChR expressed in PAG almost completely abolished play activity-in a reversible manner (Figures 1I-1L; Video S3, Friedman statistic = 25.2, groups n = 5, p < 0.0001).

Role of glycine in social play
In the next set of experiments, we investigated the molecular, cellular, and circuit mechanisms of social play.Since older rats are much less playful than younger ones, we recorded physiological activity from principal neurons within the PAG of juvenile and adult animals (see STAR Methods) and analyzed up to 178 different parameters of intrinsic and synaptic properties (Table S1) as well as their correlation with the playfulness of juvenile male rats.Although most parameters were not significantly different, we found a striking difference in firing properties of neurons of juvenile (playful) and adult animals (non-playful).Neurons from younger animals showed a spontaneous firing rate of up to 10 Hz (average: 1.39 ± 0.33 Hz, n = 46 neurons from 10 rats), whereas neurons from adult animals were almost silent (0.40 ± 0.18 Hz, n = 26 neurons from 6 rats) (Figures 2B and 2C; Table S1, t = 2.136, F (1, 70) = 4.6, p = 0.036).This spontaneous firing can be completely blocked by glutamatergic ionotropic receptor antagonists (L-(+)-2-Amino-5-phosphonopentanoic acid [APV] with 6,7-dinitroquinoxaline-2,3-dione [DNQX]) indicating the extrinsic nature of this firing as a projection of neural network activity in PAG (Figures S2A-S2C).Assessment of the tetrodotoxin (TTX)-sensitive fraction of spontaneous postsynaptic currents (sPSCs) in PAG slices shows a 39% suppression of excitatory neuronal activity in PAG networks (Figures S2D-S2F; Table S1), which cannot be explained by the inversion of GABA action (Figures S5A-S5C).Next, we wondered how these differences come about and if they are causally linked to social play behavior.In our detailed analysis of the physiological parameters (see above), which show significant age dependence and correlate with the playfulness of juvenile male rats in the same direction, the most significant difference between playful and non-playful male rats was found in Gly transmission (Table S1; Figure 2D).The frequency of Gly miniature inhibitory postsynaptic currents (Gly-mIPSCs) has a positive correlation with the playfulness of juvenile male rats (Table S1) and an age-related decrease from 1.16 ± 0.16 Hz to almost a complete loss of transmission (Figure 2E, nested t = 3.103, F (1, 32) = 9.6, p = 0.004).The very few remaining Gly-mIPSCs had a similar quantal amplitude (17.5 ± 1.9 pA, n = 21, Figure 2F; Table S1, nested t = 0.197, F (1, 31) = 0.039, p = 0.85).Moreover, Gly-IPSCs evoked by minimal electrical stimulation (Gly-eIPSCs) (Figures 2G-2I) show a 61% age-related decrease of amplitude (Figure 2J, Mann-Whitney U = 28, p = 0.0015) and a 68% age-related increase in the coefficient of variation (Figure 2L, t = 2.139, df = 25, p = 0.042) at age-independent paired pulse ratio (PPR; Figure 2K, Mann-Whitney U = 72, p = 0.38).All of this points to a reduced readily releasable pool (RRP) size of Gly inputs in adult male rats.Indeed, assessment of RRP based on the depletion of Gly inputs 23 with a 40 Hz minimal stimulation (Figures 3A and  3B), followed by plotting of a cumulative Gly-eIPSCs amplitude curve (see STAR Methods; Figures 3C and 3D), shows a 63% age-related decrease in RRP (Figure 3F), an age-independent quantal response size (Figure 3E) and age-independent release probability (Figure 3G).
Another reason for the decrease in Gly transmission could be the loss of Gly neurons.Indeed, immunohistochemical staining of PAG slices with antibodies against glycine and NeuN followed by confocal imaging (Figures S3A-S3F) reveals a 31% reduction of glycine-rich neurons in adult male rats (Figures S3B, S3C,  and S3G).
We were puzzled by this strong decrease in Gly transmission, and wondered whether it might be related to the observed decrease in spontaneous spiking (Figures 2B, 2C, and S2D-S2F).Indeed, somewhat surprisingly, raising the level of extracellular glycine via the application of the glycine transporter 1 (GlyT1) inhibitor ALX-5407 triggered an increase in excitatory inputs on PAG neurons (Figures S2G and S2H).Furthermore, this effect of ALX-5407 was inhibited by blocking action potential firing with TTX and also by the application of the N-methyl-Daspartate (NMDA) receptor antagonist MK-801 (Figure S2I).To understand the mechanisms of glycine action, we tested how the glycine receptors antagonist strychnine and the blocker of the glycine-binding site on NMDA receptors HA-966 influence neuronal activity in PAG slices.Strychnine does not alter the frequency of spontaneous excitatory postsynaptic currents (sEPSCs) (Figures S2K-S2M), although it does reduce the frequency of spontaneous IPSCs (sIPSCs) (Figures S2K, S2L, and S2N).By contrast, HA966 reduces the frequency of both sEPSCs and sIPSCs (Figures S2O-S2T).The suppression of sEPSCs by HA966 decreases with age, whereas the suppression of sIPSCs is age independent (Figures S2O-S2R and S2U).Therefore, HA966 reduces the ratio between excitation and inhibition (E/I) in PAG slices only from juvenile, but not from adult male rats (Figures S2O-S2R and S2V).Taken together, an age-dependent decrease in firing rate correlates with a collapse of Gly activitymost likely via the glycine-binding site of NMDA receptors.
To test a causal relationship of glycine and social play behavior, we established in vivo experiments with pharmacological manipulations and behavioral readouts.Indeed, blocking of astrocytic glycine transporters (GlyT1), with ALX-5407 strongly enhanced social play (Figures 4A1 and 4A3; Video S4, F (2.311, 16.18) = 7.44, p = 0.004).However, GlyT1 can operate both in the uptake mode and release glycine from astrocytes. 24o resolve this uncertainty, we applied the neurotransmitter itself, which resulted in an up to 3-fold overall increase of social play (Figures 4A1 and 4A3; Video S4).The unidirectional effects of the GlyT1 blocker itself and together with glycine demonstrate the uptake mode of transporters in rest.In both cases, general motor activity (Figures 4A2 and 4A4, Friedman statistic = 1.86, p = 0.6) and percent of time in the central area (Figure S4G) in the open field test were not altered, nor was anogenital sniffing (Figures S4A and S4B), an example of non-playful social activity.Consistent with the NMDA-dependent mechanism suggested by our in vitro data, this effect could be blocked by the NMDA receptor antagonist MK-801 and also by the blocker of the glycine-binding site on the NMDA receptor HA-966, but not by the glycine receptor antagonist strychnine (Figures 4B1-4B4 If indeed the age-dependent reduction in glycine signaling we observed in vitro is responsible for the reduction of playfulness in  older animals, we wondered whether it might be possible to make older rats playful again by manipulating glycine concentration in the PAG.Strikingly, this was indeed the case: locally applied ALX-5407, when combined with glycine, made nonplayers play again-an effect which was NMDA receptor dependent (Figures 5, S4E, S4F, and S4I; Video S6).

PAG is necessary for play
The present study investigated the mechanisms of age-related decline in social play behavior, in which we focused on the PAG.A variety of observations point to non-forebrain structures as driver of social play behavior, e.g., decorticated rats 25 or rats with damaged amygdala 26 or basal ganglia 27,28 remain playful, which does not exclude modulatory function of the forebrain on social play. 29,30The involvement of different midbrain structures in playful behavior has previously been shown by c-fos gene expression. 31Among these structures, PAG attracts special attention because it is the output, decision-making structure of neural networks determining many social behaviors. 32Moreover, neural activity is increased in the PAG during laughter or tickling in humans 33 and hand chasing play in rats. 4This study presents evidence for the necessity of PAG for social play behavior.Thus, specific suppression of PAG neuron activity by lidocaine, muscimol, or optogenetic methods almost completely blocks rough and tumble play.Nevertheless, as previously shown, PAG is involved in the formation of instinctive responses that themselves may influence social play behavior, such as anxiety, 34 fear, 35 social fear, 32,36 etc.These collateral reactions are antagonistic to social play behavior, and hence when PAG is silenced they should also be suppressed, which in turn should reinforce social play behavior, but we observe the opposite.Moreover, most of these collateral reactions cause disturbance of general motor activity (e.g., freezing or anxiety) or non-playful social activity (social fear), but we do not observe a significant change between different phases of illumination or after muscimol injections in the open field tests (Figures 1E-1L and S1G-S1J).Since the PAG is critically involved in the initiation of social play behavior, age-related changes in PAG are possibly associated with age-related declines in playfulness.

Intrinsic PAG factors leading to social play decline with age
Next, to identify the characteristics of PAG networks related to social play, we determined the electrophysiological correlates of social play and examined the age dependence of these correlates using the patch-clamp method.Despite many parameters correlating with social play and being age-dependent, only one parameter (Gly-mIPSCs frequency), characterizing Gly synaptic transmission, is significantly positively correlated with social play and decreases with age (Table S1).Of course, we do not claim that this is the only parameter related to social play behavior, but it is an important physiological response characteristic that impacts on the operation of PAG networks.In line with this idea, we demonstrate that the inhibitory mediator glycine shifts the balance of E/I in PAG networks toward excitation using NMDA receptors via the glycine-binding domain (Figures S2G-S2V), thereby increasing general neuronal activity of PAG networks (Figures 2A-2C and S2D-S2F) and stimulating rats to engage in social play (Figures 4 and 5).
Our data also show a strong age-related decline in Gly synaptic transmission (Figures 2D-2F), leading to decreased NMDA receptor activation via Gly-binding domain (Figures S2U and S2V), reduced neuronal activity (Figures S2S and S2T) in the PAG and resulting in decreased social play activity (Figure 1C).Moreover, there are no significant age-related changes in basal membrane properties (Table S1) or amplitude of Gly-mIPSCs (Figure 2F), which is mainly determined by postsynaptic receptors.An agerelated decline in NMDA receptor expression could also explain the decrease in their activation, but such a mechanism is inconsistent with radio-ligand labeling data. 37In addition to decreased NMDA receptors activation with age, neuronal activity may decrease due to age-related changes in firing properties, properties of synaptic inputs, inversion of chlorine currents, and strength of tonic GABA currents.Indeed, among other unchanged firing properties, rheobase significantly increases with age, which, on the background of unchanged membrane properties, can also be a result of decreased activation of NMDA receptors (Table S1).Analysis of miniature currents in the presence of TTX also indicates the absence of age-related changes in the properties of synaptic inputs (Table S1).Likewise, the strength of tonic GABAergic currents in PAG slices is age independent (Table S1; Figures S5D and S5E).Although the exact age of chloride current inversion in the PAG remains understudied, it is to be expected within the first 3 weeks of postnatal development as in other parts of the brain. 38Indeed, bicuculline applied to PAG slices from male rats at 4 weeks of age induces an increase in sEPSCs (Figures S5A-S5C), indicating the inhibitory nature of GABA already in the juvenile period.Thus, we can conclude that one of the key intrinsic PAG factors determining the age-dependent decrease in neuronal PAG network activity is reduced NMDA receptor activation by glycine.
The most likely reason for the age-related decline in Gly neurotransmission is a reduction in the total RRP (Figure 3).Gly neurons create varicose terminals on which synapses occur predominantly 39 and likely have multiple contacts with postsynaptic cells, which are sensed by this cells as one large synapse.The age-dependent RRP reduction cannot be attributed to intrasynaptic maturation or synaptic pruning.This scenario seems unlikely, since a strong depletion of RRP at each individual synapse would have to lead to a change in PPR, 40 which we do not observe.In addition, we observe a 30% reduction in the number of glycine neurons in PAG slices with age (Figure S3), whereas the density of NeuN+ neurons remained unchanged (Figure S3H).It is also possible that ''dilution'' of glycine neurons and synaptic pruning together are responsible for the decrease in the number of Gly synapses per postsynaptic cell in the PAG with age.
Role of glycine and NMDA receptors in social play PAG injection experiments in vivo confirmed a causal link between glycine-induced NMDA-receptors potentiation and social play behavior.However, glycine and GlyT1 blocker can stimulate social play through attenuation of antagonistic responses, such as anxiety and social fear.However, after the respective injections we do not see significant change in the time of anogenital sniffing, which inversely characterizes social fear (Figures S4A-S4D), as well as in the percentage of time animals spend in the central area in the open-field test which characterizes anxiolytic effects (Figures S4G and S4H).If anything, Gly and ALX-5407 show an anxiogenic effects.
It should also be noted that injected NMDA receptor blockers did not completely suppress the social play behavior in all cases.On the one hand, this may reflect imperfect injections in which the blockers did not reach all receptors.On the other hand, it may point to NMDA-independent mechanisms of social play behavior.However, experiments with intraperitoneal administration of the NMDA receptor blocker MK801 completely and specifically suppresses social play activity in rats, 41 contradicting such a hypothesis.
Most surprisingly, we were able to recover playful behavior in adult animals via pharmacological manipulation of the PAG glycine level (Figures 5, S4E, S4F, and S4I; Video S6).On the one hand, this suggests the functionality of social play networks in the PAG of adult rats and the modulating role of glycine in it.On the other hand, the failure of glycine to fully restore playfulness suggests other age-dependent factors acting on social play behavior.
However, the age at which rats become fully adult is not certain; many researchers believe that adulthood begins at 2-3 months of age. 42,43Nevertheless, although rats appear physiologically and behaviorally to be adults after the third months of life, they continue to develop socially and form hierarchies, which is reflected in changes in play content. 44Adult social play differs from adolescent play not only in form but also in essence and offers adults far fewer benefits than adolescents. 45The choice of the adult age for rats in our work (6 months), is due to the publication of Sengupta 46 who related 6-month-old rats with 18 years of human life.
Other age-related factors that reduce social play may include the hormonal rise during the peri-puberty.Indeed, the sex component of playfulness reverses with age after the onset of sexual differentiation. 13,19Nevertheless, the fact that play behavior in rats is sex-independent in early adolescence, 19 when sex hormones are at the same low level in both sexes, indicates a principal similarity in the structure of neuronal play networks between the sexes. 13Testosterone, which is absent in females and at about 30% of its maximum in male rats throughout adolescence, 13 apparently stimulates playfulness. 47,48The agedependent sex-specific suppression of playfulness in rats is possibly associated with increased levels of female hormones during adolescence.This is supported by the suppressive effect of estradiol 49,50 and progesterone on NMDA receptors. 51Nevertheless, the age dependence of the sex difference of playfulness requires further studies.
Opioids and cannabinoids have been shown to stimulate social play behavior via the reward system. 52,53Dopamine stimulates the motivation for social play. 53,54Noradrenaline, serotonin, and vasopressin systems have a complex and not yet fully explored pharmacology of modulation of social play. 53However, the social play pharmacology of these neurotransmitters has been studied at a systemic level or in relation to forebrain structures (modulatory structures in relation to social play), making it difficult to understand their mechanisms of action on the initiation of social play.In all of this glycine is a novel and probably very significant pharmacological determinant of social play behavior.
Social play activity peaks in the juvenile to prepubertal period 13 roughly coinciding with the peak of neural network shaping and synaptic pruning 55 guided by synaptic activity. 56lycine has the unique ability to support active synapses via NMDA receptors and in parallel increase neuronal threshold via glycine receptors, thereby cutting off weak inputs, which generally facilitates synaptic selection.Consistent with our findings, glycine increases neuronal activity, at least in PAG networks, and stimulates social play behavior.Glycine has also been shown to act as an attractant 57 and activator of microglia, 58 the main ''gardener'' in brain tissue. 59Taken together, our data suggest that glycine coordinates playful behavior.

Conclusions
The critical role of the PAG in rat social play is consistent with previous work showing that the PAG is a critical gating structure for a range of emotion-driven behaviors. 6,7,32Glycine-mediated modulation of NMDA receptors has previously been linked to social behavior 60 and is widely distributed in all parts of the mammalian brain and invertebrate ganglia. 61,627][68][69] Finally, although play has long been neglected as a topic of serious scientific study, we believe a deeper understanding of playfulness may be a key to understanding human nature and culture (too crazy to cite e.g.Huizinga, ''Homo ludens'' 70 ).Our study is not only one of the first analyses of the synaptic mechanisms of social play, but will also hopefully trigger further investigations on the functional role of social play and of the PAG.

Lead contact
Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Dietmar Schmitz (dietmar.schmitz@charite.de).

Materials availability
This study did not generate new unique reagents.

Data and code availability
This study did not generate any unique datasets or code.All data on figures with detailed statistical analysis are available in Data S1.Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS
Animal maintenance and experiments were in accordance with the respective guidelines of local authorities and approved by the state government of Berlin (LAGESO, licenses: T0073/04, G0072/21) as well as followed the German animal welfare act and the European Council Directive 2010/63/EU on protection of animals used for experimental and other scientific purposes.5-week-old male C57BL/6NRj mice, 4-6 (juvenile) and 20-30 (adult) week-old male Long-Evans rats were provided by Janvier Labs at 3 weeks of age.Rats were kept in pairs in 59 cm x 38 cm x 26 cm cages filled with wooden chips and containing thin papers, plastic shelters (usually pipes) and a wooden bar.Animals had permanent access to food and water and were maintained at 22 0 C with a 12/12 hour light/dark cycle (lights on at 6 am).
The required number of animals determination for behavioral experiments was based on the publication 73 using a two-tailed paired t-test assuming of significance level a = 0,05, power (1-b) = 80% and 20% drop outs, as well as approved by the state government of Berlin (LAGESO, licenses: G0072/21).

METHOD DETAILS Surgery
Surgery was performed on 3-4 weeks old juvenile (weight 60-80 g) and 20-23 weeks old adult (330-400 g) rats under isoflurane (1.5% in oxygen) anesthesia.To prevent post-operative pain and inflammation, the rats were given metamizole dissolved in drinking water (200 mg/kg body weight) the day before and three days after the operation and carprofen (5mg/kg body weight) was injected subcutaneously 30 min before the operation.As an additional pain killer, 2% lidocaine was injected locally subcutaneously.Since in juvenile rats the distance between the bregma and the interaural line varies from 7.2 to 8.1 mm, stereotactic coordinates taken from the atlas 74 for adult rats, where this parameter is 9 mm, were scaled proportionally for each animal individually.The non-scaled coordinates are given in mm relative to bregma.
All animals that successfully survived postoperative recovery (1 week for deep brain injections, 2 weeks for optogenetics) and showed daily weight gain were used for behavioral experiments.

Implantation of guide cannulas and deep brain injections
6][77] Sharpened G26 needle tips (7 mm long) were used as guiding cannulas.On the blunt side, 1 cm polyethylene tubing was glued to the cannulas as required for positioning, protection against bacterial contamination of the cannulas and to simplify the insertion of the injection cannula.After sterilization with 100% EtOH overnight guide cannulas were implanted bilaterally at a 9 angle to the vertical (in the coronal plane) to prevent damage to the superior sagital sinus and glandula pineale.The stereotactic coordinates for implantation (relative to Bregma: anteroposterior (AP) -7.68, mediolateral (ML) 1.63; relative to dura mater: dorsoventral (DV) 2.46) were chosen so that the injection cannula would hit the PAG, while the guiding cannula would not touch it (Figure 1D).After implantation, the guide cannulas were fixed with acrylic glue and the polyethylene tubing was sealed.For better fixation of the cannulas, a 3D-printed plastic ring with an external thread (internal diameter = 7 mm, external diameter = 1 0 mm, height = 4 mm) was glued.The entire interior space of the ring was then filled with Hoffmann's dental phosphate cement pasta.At the end a plastic cap (height = 11 mm) was put on the ring and additionally secured with a screw to cover the cannulas with polyethylene tubes from the rats' paws.After a one-week recovery period, the day before the experiment, the rat's cap was removed, the polyethylene tubes were cut and a control isotonic 0.9% NaCl solution was injected with a Hamilton syringe and 33 Ga x 10 mm needle (NanoFil #NF33BV) through guiding cannulas, 2 ml into each hemisphere.This pre-injection helps to reduce the suppression of play by the increased stress of the very first injection.In general, the bilateral deep brain injection procedure lasts between 30 sec and 1 min and does not induce visible motor disturbances or behavioral abnormalities, but suppresses playfulness by 30%.The distribution of the injected substances in the PAG was assessed by the distribution of the green dye DiR injected postmortem (Figure 1E).
For further optical activation of the WiChR 22 construct, 470 nm blue LEDs (0.65 mm x 0.35 mm x 0.2 mm 0402 SMD chip Kingbright APG0603VBC-A1-5MAV) were bilaterally implanted immediately after the injections in a vertical manner (AP = -7.68,ML = 2, DV = 5.5; Figure 1H).The blue light from the LEDs (16 mW) was completely absorbed by the 2 mm white and 3 mm grey matter of the brain.The LEDs were soldered to copper wires (diameter 50 mm) coated with lacquer and placed inside a polyethylene tube with a diameter of 250 mm.The contact points were electrically insulated from the environment with cyanoacrylate glue.The wires connected the LEDs to a self-made wireless Bluetooth device, which was encapsulated in a 3D-printed housing (with maximum dimensions of 21 mm x17 mm x13 mm) together with the battery (CR1632), that continuously powered the device for over 5 hours.The Bluetooth device (JDY-16) was attached to the skull by a printed plastic ring that was separately glued to the skull.To ensure the fixation of the optical stimulator the inner space of the fixation ring with screws and LEDs wires was completely filled with Hoffmann's phosphate cement.The one-week recovery period was also sufficient for the expression of WiChR, whose distribution in PAG as well as location of LEDs were checked in postmortem histology.The LEDs were controlled remotely via a smartphone using a self-made program.Optical activation of the WiChR was carried out with constant blue light bilaterally for 2 minutes.

Behavior
Assessment of rough-and-tumble play (RTP) in male Long-Evans rats aged 4-6 or 25-30 weeks was based on the protocol of VanRyzin et al. 18 and conducted during the light phase of a 12:12 h reverse light:dark cycle, but in the dark using 750 nm infrared illumination (IR).The open plexiglass cage with dimensions of 55 cm x 37 cm x 50 cm was placed in a large non-transparent container (131 cm x 96 cm x 81 cm) and partially filled with wood bedding (2-3 cm).The video recording of the RTP was done using a Raspberry Pi camera (OV5647) connected to Raspberry Pi A and located with 2 IR-LEDs approximately 1 m above the playing arena (Figure 1A2).The videos were acquired using internal commands of the Raspbian GNU/Linux 10 (buster) operating system (Cambridge, UK).To habituate the rats to the experimental conditions, starting two days before the experiment, the animals were individually placed for at least 10 min every day on a play arena under experimental conditions.To increase the frequency of RTP, rats were placed in single cages 24 hours before the experiment.For darkness habituation, rats were placed in a dark experimental room for at least 20 min before the experiment and during habituation.The experiment always started with a 10-minute open field test on the play arena followed by a 3 min rest in a home cage, after that a pair of rats was put together to observe RTP.The rats were given approximately 1 min to acclimate before video recording started.After 5 or 10 min (depending on the type of experiment) of RTP video recording, the rats were returned to their home cage.The test rat (injected or with activated LEDs) was returned to the playing arena after a 3 min rest for the 10 min open field test, while the other rat (CTRL) remained in its cage until the next playing session.Control experiments showed that this protocol reduces the RTP playfulness by 50% in CTRL rats after 3 or 4 five-minute lasting RTP cycles, while in test rats we saw this reduction after 4-5 cycles (data not shown).
Behavior was scored offline by an experimenter and his children randomly and blind to the experimental conditions.A detailed definition and video of behavioral events can be found in the corresponding bio-protocol of VanRyzin et al. 18 To assess rat playfulness, each video was viewed frame-by-frame and the beginning and end of the playful event (pinning, pouncing and boxing) were registered.The RTP playfulness was defined as the percentage of the cumulative duration of pouncing, pinning and boxing events relative to the total video recording time.When comparing RTP scoring under control conditions (isotonic NaCl injections) there is no significant difference between RTP assessors (2w-ANOVA F(3,57)=0.09,p=0.96, Figure S1I), while there is a difference between CTRL and injected rats (2w-ANOVA F(1,57)=4.69,p=0.03, Figure S1I).There is also no statistical interaction between assessors and rat type (2w-ANOVA F(3,57)=1.25,p=0.3, Figure S1I).The 10-min pathway in the open field test was measured by tracking rats using DeepLabCut.

Slices preparation
Rats (P23-41, males) were decapitated following isoflurane anesthesia.Brains were removed and transferred to ice-cold, cholinebased artificial cerebrospinal fluid (chACSF) containing (in mM) 125 Choline-Cl, 2.5 KCl, 1.25 NaH2PO4, 25 NaHCO3, 10 Glucose, 0.5 CaCl2, 7 MgCl2, 0.5 Na-Pyruvate, 2.8 Na-Ascorbat, saturated with 95% O 2 , 5% CO 2 , pH 7.4 and adjusted to osmolarity 302 mosmol/l with water.Slices (300 mm) were cut in a coronal plane on a vibratome (VT1200S; Leica) and stored in a submerged chamber filled with chACSF at room temperature continuously oxygenized with carbogen.After 30 min of incubation in chACSF slices were transferred to a chamber with artificial cerebrospinal fluid (ACSF) containing (in mM) 126 NaCl, 25 NaHCO 3 , 11 glucose, 2.5 KCl, 2.4 CaCl 2 , 1.2 MgCl 2 , 1.4 NaH 2 PO 4 .The equilibration time was at least 1 hour after preparation before they were transferred to the recording chamber.Patch-clamp recording Recordings were performed in ACSF at 32-34  C in a submerged-type recording chamber.Cells in the PAG were identified using infrared differential contrast microscopy (BX51WI, Olympus).A Femto2D 2-photon laser scanning system (Femtonics, Budapest, Hungary) equipped with a femtosecond pulsed Ti:Sapphire laser tuned to l = 805 nm and power 0.5W(Cameleon, Coherent, Santa Clara, CA, US) controlled by the MATLAB-based MES software package (Femtonics, Budapest, Hungary) was used to acquire images of fluorophore-filled cells.Fluorescence was acquired in epifluorescence mode with a water immersion objective (LUMPLFL 603/1.0NA or UMPlanFL 103/0.3NA, Olympus, Hamburg, Germany).Transfluorescence and transmitted infrared were detected using an oil immersion condenser (Olympus).We performed somatic whole-cell patch-clamp recordings (pipette resistance 2-6 MU).In experiments where was used the current-clamp mode the intracellular solution contained (in mM) 155 K-Gluconate, 4 KCl, 5 Na-Gluconate, 0.5 CaCl 2 , 5 EGTA, 25 HEPES, 2 Mg-ATP, 0.3 Na-GTP.The pH was adjusted to 7.2 with KOH.In experiments where spontaneous postsynaptic currents were analyzed in the voltage-clamp mode the intracellular solution contained (in mM) 155 Cs-CH 3 SO 3 , 4 CsCl, 5 Na-Gluconate, 0.5 CaCl 2 , 5 EGTA, 25 HEPES, 2 Mg-ATP 0.3 Na-GTP, 0.03 Alexa-594.The pH was adjusted to 7.2 with CsOH.Recordings were performed using Multiclamp 700A/B amplifiers (Molecular Devices).Signals were filtered at 3 kHz, sampled at 10 kHz and digitized at 16 bit resolution using the Digidata 1550 (Molecular Devices) and WinWCP v5.5.2 (University of Strathclyde Glasgow, UK) or TIDA 4.11 software (HEKA Elektronik, Lambrecht, Germany).The experimentally determined reversal potential of GABA(A)-induced chloride currents (E Cl ) was -60 mV.To record excitatory postsynaptic currents (EPSC) in voltage-clamp the holding potential was set to -60 mV, to record inhibitory postsynaptic currents (IPSC) it was set to 0 mV.Only recordings with a series resistance below 20 MU were accepted (average series resistance amounted to 11 MU).Series resistance compensation was not applied.Access resistance was controlled by regularly (1/60 s) applying hyperpolarizing pulses of 10 mV.Cells exhibiting more than 20% changes in the access resistance during an experiment were discarded.Spontaneous and miniature postsynaptic currents were analyzed using a PeakCount V3.2 (courtesy of Dr Christian Hennenberger).The program employs a derivative threshold-crossing algorithm to detect individual PSCs with visual inspection of each event.Spontaneous firing activity of neurons was recorded in current clamp mode at holding current 0 mV.Electrical stimulation Evoked glycinergic inhibitory postsynaptic currents (eIPSCs) were elicited by focal electrical stimulation through a glass pipette filled with ACSF (about 10 MU) in presence of 20 mM.An isolated stimulation unit (Axon Instruments Isolator-11) was used to generate rectangular electrical pulses with 0.5 ms duration.Pulse intensity was adjusted to activate a unitary synaptic input (minimal stimulation).Stimulation was accepted as minimal if the following criteria were satisfied: (1) eIPSC latency remained stable (<20% fluctuations); (2) lowering stimulus intensity by 20% resulted in a complete failure of eIPSCs; and (3) an increase in stimulus intensity by 20% changed neither mean eIPSC amplitude nor eIPSC shape.Typical pulse intensity required for minimal stimulation was between 1 and 2 mA.

Estimation of readily releasable pool size
9][80] Repetitive stimulation leads to a decrease in the eIPSC amplitudes.9][80] Namely, stimulation trains of 100 pulses delivered at 40 Hz were applied and cumulative eIPSC amplitudes were plotted versus stimulus number (Figure 3).Because repetitive stimulation causes a compound postsynaptic response (synchronous and asynchronous), PSCs that peaked within a 3 ms interval following the end of a stimulus pulse were selected as stimulus-locked eIPSCs.After 50 pulses, the cumulative eIPSCs reached a steady state, as indicated by the linear slope dependence of the cumulative eIPSC amplitude on the pulse number (Figure 3C).Assuming that (1) the number of release sites remains constant throughout the experiment and (2) the linear component reflects vesicle recycling, the cumulative IPSC amplitude in the absence of pool replenishment can be estimated by back-extrapolation to the start of the train (Y-intercept, Figures 3C and 3D).Note Y-intercept represents the product of the RRP and the quantal amplitude.Since the synapse releases vesicles in a quantal manner after depletion, the size of the RRP is the Y-intercept normalized by the mean amplitude of late eIPSCs, which are defined as synchronous responses to the last 50 stimuli excluding failures (Figures 3D-3F).Indeed, during 40 Hz trains the mean late eIPSC amplitude has been shown to be close to the median miniature IPSC amplitude in this preparation (Figures 2D, 2F, and 3E).It should be mentioned that only late eIPSCs but not asynchronous IPSCs 79 were selected to obtain an estimate of the quantal amplitude for the projection of interest.Release probability (Pr) was calculated using the binomial model approximation, namely, Pr = mean eIPSC-amplitude/RRP/Quantal size.

Slices preparation
To prepare slices for immunohistochemical studies the animals were euthanized with Isoflurane and transcardially perfused with 0.9% saline followed by 4% paraformaldehyde (PFA) in PBS.Brains were harvested and immersion-fixed in 4% PFA for 24 h.After fixation, brains were placed in PBS+0.1%NaN3solution for PFA washing and storage.Brains were cut into 40 mm coronal sections using a vibratome (VT1200S; Leica) and 5% agar block.The free-floating sections were stored in PBS containing 0.1% sodium azide in 24-well cell culture plates at 4 C.

Staining
The sections were washed 3 3 10 min with 0.1M PBS and incubated at room temperature for 3 h in immunobuffer containing 0.1 M PBS, 1% Triton X-100 (Sigma-Aldrich) and 5% normal goat serum.The primary antibodies were diluted in immunobuffer and applied to the sections for 48 h incubation with light agitation at 4 0 C temperature.The primary antibodies and their dilutions were as follows: rabbit anti-Glycine 1:500, mouse anti-NeuN 1:1000, guinea pig anti-GABA 1:500, rabbit anti-Glutaminase.After incubation, following 2 3 10 min washes in 0.1M PBS, appropriate fluorophore-conjugated secondary antibodies diluted 1:500 in immunobuffer were applied and the sections were incubated for 3 h with light agitation at room temperature.The secondary antibodies were: goat anti-guinea pig Alexa Fluor 488, goat anti-rabbit Alexa Fluor 488, goat anti-rabbit Alexa Fluor 555, goat anti-mouse Alexa Fluor 647.Following incubation with the secondary antibodies, brain tissues were washed 3 3 10 min in PBS, mounted onto glass slides with DAPI containing mountain media (Fluorshield-Dapi; F6057-20ML; Sigma-Aldrich) and covered.

Confocal microscopy
Mounted slices were imaged using a confocal microscope (Leica DMI 6000), equipped with a 20x immersion objective (HC PL APO 20x/ 0.70 IMM).Images were acquired with the LAS X 3.1.5.16308 software (Leica Microsystems, Germany).A 405 nm diode was used to excite DAPI, a 488 nm OPSL laser was used to excite Alexa Fluor 488, a 552 nm OPSL laser was used for Alexa Fluor 555 or mScarlet, and a 638 nm diode was used for Alexa Fluor 647.Emission was detected by 4 photomultiplier tubes: 410-530 nm for DAPI, 497-577 nm for Alexa Fluor 488, 560-630 nm for Alexa Fluor 555 and 650-750 nm for Alexa Fluor 647.The pinhole was 60.5 mm.Laser power and gain were kept the same for all imaged slices and conditions.Each image was acquired by averaging of 3 frames, where each line was an average result of 2 scan lines.Each image channel was acquired independently of each other.The final slice image was obtained by stitching together and aligning the brightness of the individual square fragments.

QUANTIFICATION AND STATISTICAL ANALYSIS
The results are presented as mean ± S.E.M.In all figures the error bars indicate that the S.E.M. normality of the data distributions was assessed using the Kolmogorov-Smirnov test (KS-t).Considering that the comparison of means could be influenced by inter-animal variance in slice experiments, 81 we performed multilevel (''nested data'') analysis, when necessary.Differences between means were determined by the nested t-test (nested t-t), Student's paired t-test (pt-t) or its corresponding non-parametric analog (Wilcoxon matched-pairs signed rank test, W-t), and the unpaired non-parametric Mann-Whitney test (MW-t), as appropriate.Differences between groups were tested using one-way repeated measures ANOVA summary (1-way-RM-ANOVA) or an equivalent nonparametric method (Friedman test), followed by a multiple comparison corrected Tukey's test (Tukey-MCT) or a corresponding non-parametric method (Dunn's-MCT).All statistical tests used are two-sided.All data points in the figures connected by straight gray lines represent paired measurements made on the same rat or cell.P values of <0.05 were considered statistically significant.Significance levels were marked by asterisks, as follows: *-p<0.05,**-p<0.01,and ***-p<0.001.The abbreviation ''ns'' in the figures means ''not significant''.Cohen's d ("d" in the figures) = jMean 2 À Mean 1 j = ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ffi ðSD 1 2 Ã ðn 1 À 1Þ+SD 2 2 Ã ðn 2 À 1ÞÞ=ðn 1 +n 2 À 2Þ q was used to estimate effect sizes.In figures describing electrophysiological or immunohistochemical experiments, the numbers on the bars in the numerator indicate the number of measured units (slices or cells, as indicated in figure legends or tables), and in the denominator the number of groups (animals).In figures describing behavioral experiments the numbers on the bars show the number of used animals.Each behavioral experiment was performed with each rat only once.

Figure 1 .
Figure 1.PAG is required for play behavior (A) Rats play activity (A1, drawing; A2, photos) is defined as the accumulated time spent pouncing (left), pinning (middle), and boxing (right) divided by the total time of observation.(B) Timeline histograms of different play activities and their sums for juvenile mice (top), juvenile (middle), and adult (bottom) rats.(C) Juvenile (4-6 weeks) rats but not juvenile mice (5 weeks) show rough and tumble play, which is suppressed in adult rats (47-55 weeks).(D) Juvenile mice and rats as well as adult rats spend similar time for anogenital sniffing, corresponding to non-playful social interactions.(E) Based on the Paxinos atlas scheme of the implantation of the guiding cannulas (in blue) and the positioning of the injecting cannulas (in red).(F) PAG photo showing the distribution of the green dye (DiI) injected bilaterally according to the scheme in (D).The red dotted line indicates the boundaries of the PAG.(G) Bilateral injection of 2 mL of 2% lidocaine and 100 mM muscimol through guide cannulas into the PAG almost completely blocks play behavior in injected rat.The day after the muscimol injection, the rat's playfulness completely recovers, showing specificity of the effect to the drug but not to the PAG damage.CTRL means injection of isotonic 0.9% NaCl.(H) In contrast to muscimol, lidocaine injection causes motor impairment, leading to a significant decrease in the distance traveled in the open field test.(I) Scheme of LEDs implantation (the blue color indicates the approximate area of light irradiation) and AAV-hSyn-WiChR-mScarlet virus injection sites (injection cannulas are marked in red).(J) Two-photon-image of PAG shows mScarlet expression 2 weeks after virus (AAV-hSyn-WiChR-mScarlet) injection.The arrows show the footprints of the implanted LEDs.
(H) In contrast to muscimol, lidocaine injection causes motor impairment, leading to a significant decrease in the distance traveled in the open field test.(I) Scheme of LEDs implantation (the blue color indicates the approximate area of light irradiation) and AAV-hSyn-WiChR-mScarlet virus injection sites (injection cannulas are marked in red).(J) Two-photon-image of PAG shows mScarlet expression 2 weeks after virus (AAV-hSyn-WiChR-mScarlet) injection.The arrows show the footprints of the implanted LEDs.

Figure 2 .
Figure 2. Age-related decrease of neural firing and synaptic Gly release in the PAG (A) Midbrain coronal slice image showing the location of the recorded cell filled with Alexa-594.(B) Traces of membrane potentials recorded in CC-mode (I holding = 0 pA) in dlPAG neurons from juvenile (top) and adult rats (bottom).(S1 and C2) The spontaneous neuronal firing frequency statistics show a significant age-dependent decrease of neuronal activity in PAG.(S1) shows the comparison of mean values; (C2) shows the cumulative distribution.(D) Traces of glycinergic inhibitory postsynaptic membrane currents recorded at 0 mV holding potential in the presence of 1 mM TTX and 20 mM bicucullin (BMI) in PAG neurons from juvenile (left) and adult (right) rat slices.The insets show an enlarged Gly-IPSC marked with an arrow.(E) The frequency of miniature Gly-IPSCs demonstrate a significant age-dependent decrease.(F) The amplitude of miniature Gly-IPSCs show that the quantal size of synaptic glycine release is independent of age.(G-I) Examples of glycinergic IPSCs (Gly-IPSCs) evoked by minimal paired at 20 Hz electrical stimulation and recorded with low Cl intracellular pipette solution at 0 mV in the presence of 20 mM BMI in caudal-dlPAG neurons of juvenile (G) and adult (H) rats.The gray thin traces show responses to individual stimuli (60 entries).The thicker traces (black for juvenile, red for adult) represent the average traces.(I) shows the same averaged traces, but scaled by the first response.The arrows indicate the removed stimulation artifacts.(J) The averaged Gly-eIPSCs amplitudes show a significant age-related decrease.(K) The paired pulse ratio (PPR) of Gly-eIPSCs does not change significantly with age.(L) The coefficient of variation (CV) of Gly-eIPSCs significantly increases with age, indicating, together with the PPR (E), a reduction in the readily releasable pool (RRP) of glycinergic synaptic vesicles with age.ns, p R 0.05; *p < 0.05; **p < 0.01.See also Figures S2 and S5 and Table S1 and Data S1.

Figure 3 .
Figure 3. Assessment of the readily releasable pool size for glycinergic inputs in the PAG (A and B) Examples of Gly-IPSCs induced with high-frequency (40 Hz) electrical stimulation (100 pulses) of caudal-dlPAG neurons from juvenile (A) and adult (B) rats.The upper traces show stimulation entirely.The bottom left traces show the first 10 stimuli, the bottom right traces show the last 10 stimuli.(C) The plot of cumulative amplitudes shows a linear increase after 50 pulses, indicating depletion of the readily releasable pool (RRP) for both the juvenile (white circles) and adult (gray circles) rat groups.The red lines represent these linear trends.(D) Y intercepts of the linear components in the cumulative amplitude curves indicate a significant decrease in the product of quantal size and RRP in adult rats.t = 2.6, df = 20, p = 0.02, Cohen's d = 1.(E) The quantal response size, defined as the average amplitude of synchronous Gly-IPSCs over the last 50 pulses, shows no age dependence.Mann-Whitney U = 34, p = 0.09, Cohen's d = 0.68.(F) The readily releasable pool (RRP) size, calculated as the ratio of Y-intercept to quantal size, is significantly reduced in adult rats t = 2.8, df = 20, p = 0.01, Cohen's d = 1.2.(G) The release probability (Pr) calculated as the ratio of the average amplitude 1 to the Y-intercept is negatively correlated with PPR, showing a good relationship between the two different methods of estimating real Pr.The horizontal lines showing the mean ± SE for Pr show no age-related changes, as well as the mean PPR values shown by the vertical lines.For release probability Mann-Whitney U = 58, p = 0.9.**p < 0.01; ***p < 0.001.See also Figure S3 and Data S1.

Figure 4 .
Figure 4. Increasing glycinergic transmission boosts play via the glycinergic site of the NMDA receptor (A) Glycine stimulates rough and tumble play in juvenile rats.Rats were injected into PAG via guide cannulas with 0.9% NaCl control solution (CTRL), 2 mM ALX-5407, a mixture of ALX-5407with 100 mM glycine, and a mixture of the previous solution with an NMDA receptor blocker, 10 mM MK-801.(A1) Timeline histograms of play activity for different rats after injections.(A2) Pathways traveled by the rat indicated with the asterisk in (A1) in a 10-min open-field test after injections.(A3) The statistics for the data in (A1) show a significant and unidirectional increase in play activity of juvenile rats following injections of the inhibitor GlyT1 (ALX-5407) and its mixture with glycine (Gly).This effect can be completely blocked by the NMDA receptor blocker (MK-801).(A4) The statistics of the 10 min traveled path in the open field test show no significant effect of injecting all the above substances into the PAG on the rats' motor activity.(B) Glycine stimulates play via NMDA receptors.Rats were injected into PAG via guide cannulas with control solution (CTRL), a mixture of 2 mM ALX-5407 with 100 mM glycine, a mixture of the previous substances with 2 mM strychnine (Str), Gly-receptor blocker and a mixture of all of the above with 400 mM HA-966, a blocker of Gly-binding site on NMDA receptor.(B1) Timeline histograms of play activity after injections.(B2) Pathways traveled by the rat indicated with the asterisk in (B1) in a 10-min open-field test after injections.(B3) The statistics for the data in (B1) show that the glycine-induced increase of play activity can be blocked by blocking of NMDA receptors, but not by blocking glycine receptors.(B4) The statistics of the traveled path in the open field test show no significant effect of injecting the substances mentioned above into the PAG on the motor activity of juvenile rats.The gray lines on (A3), (A4), (B3), and (B4) indicate paired measurements.Color code: pouncing, black; pins, red; boxing, green.ns, p R 0.05; *p < 0.05; **p < 0.01.See also Figure S4, Videos S4 and S5 and Data S1.

Figure 5 .
Figure 5. Glycine stimulation of play activity in adult rats(A) Timeline histograms of play activity for adult rats immediately after injections in PAG via guide cannulas of 0.9% NaCl control solution (CTRL), 2 mM ALX-5407, GlyT1 blocker, a mixture of the previous substance with 100 mM glycine (Gly), and a mixture of all of the substances mentioned above with an NMDA receptor blocker, 10 mM MK-801.Asterisk points to the rat whose data are shown in (B).Color code: pouncing, black; pins, red; boxing, green.(B) The statistics for the data in (A) show a significant increase in play activity of adult rats following injections of the inhibitor GlyT1 (ALX-5407) mixed with glycine (Gly) but not after injection of ALX-5407 alone.This effect of Gly can be completely blocked by the NMDA receptor blocker (MK-801).Friedman statistic = 11, p = 0.006.(C) Pathways of the adult rat labeled in (A) by asterisk in 10-min open-field tests, 10 min after injections of 0.9% NaCl solution (CTRL), 2 mM ALX-5407, a mixture of the previous substance with 100 mM glycine (Gly), and a mixture of all of the above with 10 mM MK-801.(D) The statistics of the 10 min path in the open field test show no significant effect of injecting all the above substances into the PAG on the motor activity of adult rats.F (1.210, 6.052) = 0.28, p = 0.7.The gray lines on (B) and (D) indicate paired measurements.ns, p R 0.05; *p < 0.05; **p < 0.01.See also Figure S4 and Video S6 and Data S1.