Presynaptic PTPσ regulates postsynaptic NMDA receptor function through direct adhesion-independent mechanisms

Synaptic adhesion molecules regulate synapse development and function. However, whether and how presynaptic adhesion molecules regulate postsynaptic NMDAR function remains largely unclear. Presynaptic LAR family receptor tyrosine phosphatases (LAR-RPTPs) regulate synapse development through mechanisms that include trans-synaptic adhesion; however, whether they regulate postsynaptic receptor functions remains unknown. Here we report that presynaptic PTPσ, a LAR-RPTP, enhances postsynaptic NMDA receptor (NMDAR) currents and NMDAR-dependent synaptic plasticity in the hippocampus. This regulation does not involve trans-synaptic adhesions of PTPσ, suggesting that the cytoplasmic domains of PTPσ, known to have tyrosine phosphatase activity and mediate protein-protein interactions, are important. In line with this, phosphotyrosine levels of presynaptic proteins, including neurexin-1, are strongly increased in PTPσ-mutant mice. Behaviorally, PTPσ-dependent NMDAR regulation is important for social and reward-related novelty recognition. These results suggest that presynaptic PTPσ regulates postsynaptic NMDAR function through trans-synaptic and direct adhesion-independent mechanisms and novelty recognition in social and reward contexts.

In the present study, we found that presynaptic PTPs trans-synaptically regulates the postsynaptic localization and function of NMDARs in the hippocampus. Surprisingly, this regulation does not involve trans-synaptic adhesion of PTPs, suggesting that the cytoplasmic domains of PTPs, possessing tyrosine phosphatase activity and mediating protein-protein interactions, are important. In line with this, a proteomic analysis revealed strong increases in phosphotyrosine (pTyr) levels in presynaptic proteins, including neurexins. Behaviorally, this trans-synaptic regulation is critical for novelty recognition in multiple assays.
Intriguingly, LTD induced by low-frequency stimulation (LFS-LTD) was also suppressed at Emx1-Cre;Ptprs fl/fl SC-CA1 synapses ( Figure 1G), a result that contrasts with a previous report that LFS-LTD at these synapses is normal in global PTPs-KO mice (Horn et al., 2012). In contrast, metabotropic glutamate receptor (mGluR)-dependent LTD induced by the group I mGluR agonist DHPG was normal at mutant synapses ( Figure 1H).
Because the concomitant decrease in HFS/TBS-LTP and LFS-LTD could involve decreased NMDAR-mediated synaptic transmission (Bliss and Collingridge, 1993;Malenka and Bear, 2004), we next tested if NMDAR-mediated synaptic currents were suppressed. Indeed, the ratio of NMDAR-EPSCs and AMPAR-EPSCs (NMDA/AMPA ratio) was decreased at Emx1-Cre;Ptprs fl/fl SC-CA1 synapses ( Figure 1I). This result, together with the normal levels of basal excitatory synaptic transmission and mEPSCs, mediated by AMPARs, suggest that NMDAR currents are selectively decreased.
The decay kinetics of mutant NMDAR EPSCs strongly suggest that the decrease in NMDAR currents is mediated by the GluN2B subunit of NMDARs ( Figure 1I). Indeed, levels of the GluN2B subunit were most strongly decreased in crude synaptosomal (P2) and PSD fractions, but not in the total lysates, of the Emx1-Cre;Ptprs fl/fl hippocampus ( Figure 1J). The GluN2A subunit also showed a trend toward a decrease, but this difference did not reach statistical significance.
The reduced synaptic plasticity (LTP and LTD) in Emx1-Cre;Ptprs fl/fl mice may be attributable to reduced NMDAR-mediated synaptic transmission or changes in the signaling pathways downstream of NMDAR activation. To test this possibility, we activated NMDARs in the Emx1-Cre;Ptprs fl/fl hippocampus using D-cycloserine (20 mM), a glycine-site NMDAR agonist. D-cycloserine fully rescued the NMDA/AMPA ratio and TBS-LTP at SC-CA1 synapses in Emx1-Cre;Ptprs fl/fl hippocampal slices without affecting the NMDA/AMPA ratio or TBS-LTP at WT SC-CA1 synapses ( Figure 1K,L).
These results collectively suggest that excitatory neuron-restricted deletion of PTPs leads to decreases in NMDAR-mediated synaptic transmission and NMDAR-dependent synaptic plasticity, without affecting AMPAR-mediated transmission, in the hippocampal CA1 region. In addition, considering the extents of the decreases in HFS-LTP, TBS-LTP, and LFS-LTD (~44%,~66%, and~53%,  ). (J) Decreased levels of the GluN2B, but not GluN1 or GluN2A, subunit of NMDARs in crude synaptosomal (P2) and PSD I fractions, but not in Figure 1 continued on next page respectively) and the decrease in the NMDA/AMPA ratio (~45%) at the mutant synapses under naïve and D-cycloserine rescue conditions ( Figure 1E-I and K,L), the decreased LTP and LTD seem to mainly involve decreased NMDAR currents rather than signaling pathways downstream of NMDAR activation. In addition, the decreased levels of GluN2B in the PSD fraction (~20%) may contribute partly to the decrease in NMDAR currents (~45%).
To this end, we deleted PTPs in presynaptic neurons in the hippocampal CA3 region and measured LTP at SC-CA1 synapses in the CA1 area by injecting AAV1-hSyn-Cre-eGFP into the CA3 region of Ptprs fl/fl mice~2.5 weeks prior to LTP measurement at~4 weeks ( Figure 2A). In control experiments, we injected AAV1-hSyn-Cre-eGFP into the CA1 (postsynaptic) area and measured LTP at SC-CA1 synapses. Specific expression of Cre recombinase in the CA3 or CA1 region was confirmed by monitoring EGFP (enhanced green fluorescent protein) signals. Reduced levels of PTPs protein (~20-40% of WT) were confirmed by immunoblot analyses of hippocampal samples from Cre-expressing CA3 and CA1 areas ( Figure 2B).
LTP experiments indicated that Cre-induced deletion of PTPs in the CA3 region suppresses TBS-LTP at SC-CA1 synapses in Ptprs fl/fl mice compared with control synapses expressing EGFP alone (no Cre) ( Figure 2C). In contrast, deletion of PTPs in the CA1 region had no effect on TBS-LTP. These results suggest that PTPs in the presynaptic (CA3) region, but not the postsynaptic (CA1) region, is important for normal LTP at SC-CA1 synapses.

Re-expression of presynaptic PTPs rescues postsynaptic LTP in the hippocampus
To further test the hypothesis that presynaptic PTPs regulates postsynaptic LTP, we re-expressed PTPs in presynaptic CA3 neurons by locally injecting AAV-eIF1a-Ptprs into the CA3 region of Emx1-Cre;Ptprs fl/fl mice ( Figure 3A).
In addition to WT PTPs constructs, we used constructs of mutant PTPs with extracellular mutations that abrogate trans-synaptic interactions with postsynaptic/extracellular adhesion molecules (TrkC, Slitrk1, HSPG, and CSPG) (Figure 3B,C; Figure 3-figure supplement 1). Expression levels and molecular weights of these PTPs mutants were verified by immunoblot analysis of HEK293T cell lysates using two different PTPs antibodies (targeting N-and C-termini) ( Figure 3D).
Re-expression of WT PTPs in the CA3 region of Emx1-Cre;Ptprs fl/fl mice by local injection of AAV-eIF1a-Ptprs rescued TBS-LTP at SC-CA1 synapses, restoring it to levels comparable to those in Ptprs fl/fl (control) mice injected with control virus (AAV-hSyn-eGFP) ( Figure 3E). These results suggest that acute presynaptic re-expression of PTPs in CA3 rescues postsynaptic LTP in the CA1 region.

Extracellular regions of PTPs are not important for postsynaptic LTP regulation
We next tested whether trans-synaptic adhesions of PTPs are important for postsynaptic LTP regulation using mutant PTPs proteins that lack HSPG/CSPG or Slitrk1/TrkC interactions (K68A/K69A/ K71A/K72A and Y224S, respectively) (see Supplementary file 1 for details) (Coles et al., 2014;Coles et al., 2011;Han et al., 2018;Um et al., 2014;Won et al., 2017). Surprisingly, re-expression of either PTPs mutant in the CA3 region by local AAV injection rescued TBS-LTP to an extent similar to that of WT PTPs injection ( Figure 3E), suggesting that HSPG/CSPG and Slitrk1 interactions are not important for the rescue of LTP.  However, these PTPs mutants might not cover as yet unknown trans-synaptic or extracellular binding partners of PTPs and their contribution to postsynaptic LTP regulation. We thus generated PTPs mutants carrying small deletions of Ig1+2, Ig3, or FNIII1-2 domains ( Figure 3C) (Supplementary file 1). However, all of these PTPs mutants rescued TBS-LTP at SC-CA1 synapses in the Emx1-Cre;Ptprs fl/fl hippocampus when re-expressed in the CA3 region, although the PTPs mutant containing an Ig3 deletion produced only partial rescue ( Figure 3F). These results collectively suggest that the extracellular domains or regions of PTPs are not important for PTPs-dependent postsynaptic regulation of LTP.
Synaptic proteins with altered pTyr levels in the Emx1-Cre;Ptprs fl/fl cortex and hippocampus Presynaptic PTPs-dependent regulation of postsynaptic LTP does not involve extracellular regions of PTPs, suggesting that the cytoplasmic region of PTPs is important. This region contains the D1 and D2 domains, which are known to possess tyrosine phosphatase activity and mediate interactions with cytoplasmic proteins, respectively, with the latter (D2 domain) potentially linking the D1 domain with its pTyr substrates. We thus employed a proteomic approach to perform an unbiased search of proteins with altered pTyr levels in the Emx1-Cre;Ptprs fl/fl cortex and hippocampus using anti-pTyr antibodies followed by liquid chromatography tandem mass spectrometry (LC-MS/MS) to pull down and identify pTyr proteins ( Figure 4A).
This proteomic analysis revealed that, of 1549 proteins (3894 pTyr motifs), 57 proteins (80 pTyr motifs) showed significant changes in pTyr levels in mutant mice compared with WT mice (p<0.05; fold change >1.5), as indicated by a volcano plot ( Figure 4B; Supplementary file 2). These significant changes included both upregulation (29 proteins/33 motifs) and downregulation (29 proteins/ 47 motifs) of pTyr levels. Among the proteins with significantly changed pTyr levels in mutant mice was LRRTM3, which showed both up and downregulation of phosphorylation at different pTyr motifs.
Proteins of particular interest among those that were differentially tyrosine phosphorylated include the GluN2B subunit of NMDARs with altered phosphorylation at tyrosine residues 985, 997, 1004, 1367 and 1369. Given the decreased P2 and PSD, but not total, levels of GluN2B and decreased NMDAR currents in the Emx1-Cre;Ptprs fl/fl hippocampus ( Figure 1J), these changes in pTyr, which have not been previously reported, suggest that altered phosphorylation of GluN2B may regulate the synaptic localization or function of GluN2B. Another protein of interest was TrkC, a postsynaptic adhesion partner of PTPs (Takahashi et al., 2011) that was found to be differentially phosphorylated at tyrosine residues 597 and 604. Although these pTyr residues have not been studied previously, these findings suggest that maintenance of normal tyrosine phosphorylation of TrkC at these residues is dependent on presynaptic PTPs. Other proteins of interest included neurexins, which regulate NMDAR-and AMPAR-mediated synaptic responses (Dai et al., 2019); Neto1, a trans-membrane protein that regulates synaptic localization of NMDARs and kainate receptors, and regulates LTP and learning and memory (Cousins et al., 2013;Molnár, 2013;Ng et al., 2009;  . Samples from three WT or KO mice were pooled for each analysis. Note that a single protein can be linked to multiple dots (different pTyr motifs); conversely, the same peptide can belong to two different proteins (e.g., Erbin and LRRC7). (C) Functional analysis (DAVID GO analysis) of proteins from Emx1-Cre;Ptprs fl/fl mice with significantly altered pTyr levels. Note that synapse-related GO terms are strongly enriched. (D-F) Sunburst plots from SynGO analyses of specific pre-and postsynaptic localizations of proteins with altered pTyr levels (total, upregulated, and downregulated) from the Emx1-Cre;Ptprs fl/fl cortex and hippocampus. Note that upregulated proteins tend to be those that localize to presynaptic sites, whereas downregulated proteins tend to be those that localize to postsynaptic sites. (G) Volcano plots showing that presynaptic proteins, determined by SynGO analyses, showed mainly increased pTyr levels, whereas postsynaptic proteins showed mainly decreased pTyr levels. (H) Largely normal synaptic levels of PTPs-related proteins in the hippocampus of Emx1-Cre;Ptprs fl/fl mice (3 weeks), as shown by immunoblot analyses of crude synaptosomes. (n = 6 mice for WT and cKO and for some, 6, five mice for WT and cKO, *p<0.05, ns, not significant, onesample t-test). The online version of this article includes the following figure supplement(s) for figure 4:  Wyeth et al., 2014); and APP, which associates with and regulates the trafficking of NMDARs (Cousins et al., 2013;Snyder et al., 2005).
A DAVID Gene Ontology (GO) analysis (http:// david.ncifcrf.gov) of proteins with significant pTyr levels showed that GO terms in the cellular compartment module with the strongest scores were synapse related, and included 'synapse', 'excitatory synapse' and 'postsynaptic specialization' ( Figure 4C). Other strong GO terms included 'kinase binding' and 'scaffold protein binding', in the protein binding module, and 'protein/receptor localization to synapse' and 'dendritic spine development', in the molecular function module. Therefore, proteins with altered pTyr levels were those that are strongly associated with synapse organization and protein-protein interactions.
Application of SynGO analysis, a recently reported set of expert-curated GO terms for synaptic proteins (https://www.syngoportal.org/; Koopmans et al., 2019), to proteins with altered pTyr levels from Emx1-Cre;Ptprs fl/fl mice showed that 29 of the 57 proteins with significant pTyr changes corresponded to synaptic proteins in the SynGO database. These proteins fell into diverse functional categories; notably, proteins exhibiting upregulated pTyr levels were linked to 'synapse organization' and 'presynapse' functions, whereas those with downregulated pTyr levels were linked to 'synapse organization' and 'postsynapse' functions ( Additional SynGO analyses of protein localization showed that these pTyr proteins were localized to both pre-and postsynaptic sites ( Figure 4D). Intriguingly, upregulated proteins were more enriched at presynaptic than postsynaptic sites (9 presynaptic and four postsynaptic) ( Figure 4E), whereas downregulated proteins were more strongly enriched at postsynaptic than presynaptic sites (15 postsynaptic and 0 presynaptic) ( Figure 4F).
Volcano plot displays of these pre-and postsynaptic proteins, determined based on the SynGO analysis, further highlighted the fact that presynaptic proteins showed mainly increased pTyr levels, whereas postsynaptic proteins showed mainly decreased pTyr levels ( Figure 4G). These results collectively suggest that excitatory neuron-restricted deletion of PTPs leads to strong changes in pTyr levels in synaptic proteins, and primarily increases pTyr levels of presynaptic proteins and decreases pTry levels of postsynaptic proteins. These findings further indicate that upregulated presynaptic pTyr proteins might represent potential pTyr substrates of PTPs.
Changes in pTyr levels in the abovementioned synaptic proteins may influence their synaptic function or localization. These changes may also occur in PTPs-interacting presynaptic proteins, such as liprin-a and caskin (Bomkamp et al., 2019;Serra-Pagès et al., 1998). In addition, deletion of PTPs may lead, directly or indirectly, to the loss of postsynaptic partners of PTPs or the PTPs-dependent regulation of postsynaptic plasticity.
To test these possibilities, we investigated whether synaptic levels of PTPs-related proteins are decreased by performing immunoblot analyses of crude synaptosomes from WT and Emx1-Cre; Ptprs fl/fl mice. We found no changes in the synaptic levels of D2 domain-interacting proteins (liprin-a 1/2 and Caskin 1/2) or known substrates of PTPs (N-cadherin and b-catenin) (Siu et al., 2007; Figure 4H). Moreover, there were no changes in the synaptic levels of postsynaptic scaffolding proteins (PSD-95, PSD-93, SynGAP, and Shank3) or postsynaptic binding partners of PTPs (SALM5 and NGL-3), although there was a moderate increase in NGL-3 levels. These results suggest that deletion of PTPs has minimal impacts on the synaptic localization of PTPs-related proteins.
Suppressed novelty recognition in Emx1-Cre;Ptprs fl/fl mice Hippocampal NMDAR-dependent LTP and LTD have been linked to multiple types of learning and memory behaviors (Bliss et al., 2003;Malenka and Bear, 2004). We thus first subjected Emx1-Cre; Ptprs fl/fl mice to novel object-recognition tests, in which a subject mouse is exposed to two identical objects on day 1, and one of the two objects is replaced with a new object on day 2. Unlike WT mice, Emx1-Cre;Ptprs fl/fl mice failed to recognize the novel object on day 2 ( Figure 5A). The increase in baseline object exploration (~2 folds) in the mutant mice, partly attributable to the increased locomotion and object exploration (~20% and~30%, respectively, n = 16 mice), is less likely to affect the relative exploration of familiar and novel objects.
To test if this lack of novel-object preference is specific for an object but not for a novel mouse, we next subjected mice to a three-chamber test, designed to test for social approach and social-novelty recognition (Silverman et al., 2010). Emx1-Cre;Ptprs fl/fl mice showed normal social approach, as indicated by the preference for a social target (stranger mouse) over an object, but failed to show (n = 16 mice for WT and cKO, *p<0.05, ns, not significant, Student's t-test). (B) Normal social approach, but suppressed social-novelty recognition, in Emx1-Cre;Ptprs fl/fl mice (2-3 months) in the three-chamber test. S1, first/initial social stranger; O, object; S2, second/new social stranger. (n = 19 [WT] and 15 [cKO], **p<0.01, ***p<0.001, ns, not significant, RM two-way ANOVA with Sidak's test). (C) Normal social recognition and habituation, but suppressed social-novelty recognition, in Emx1-Cre;Ptprs fl/fl mice (2-3 months) in a modified three-chamber test, in which the subject mouse was exposed to the initial stranger mouse for four consecutive days and introduced to a new stranger mouse on day 5. (D) Suppressed reward-arm recognition in the reversal, but not initial, phase of the Y-maze in Emx1-Cre;Ptprs fl/fl mice (2-3 months The online version of this article includes the following figure supplement(s) for figure 5: Figure 5 continued on next page normal social novelty preference, similarly exploring new and old stranger mice ( Figure 5B). These results suggest that Emx1-Cre;Ptprs fl/fl mice fail to recognize both a novel object and a novel mouse.
To further explore this phenotype, we performed a modified social interaction test in which a subject mouse was repeatedly exposed to the initial stranger mouse for four consecutive days and then was introduced to a new stranger mouse on day 5 (Bariselli et al., 2018). Emx1-Cre;Ptprs fl/fl mice spent increasingly less time with the initial stranger mouse over the first four days, indicative of normal social cognition and habituation, but spent less time with a novel mouse on day five compared with WT mice (Figure 5C), further confirming the decrease in social-novelty recognition.
In the Y-maze test, where the reward arm on day one was switched to another arm on day 2, Emx1-Cre;Ptprs fl/fl mice displayed less efficient switching to the novel arm containing the reward on day 2 ( Figure 5D). Lastly, in the Morris water maze test, Emx1-Cre;Ptprs fl/fl mice showed normal learning and memory in the initial phase. However, in the reversal phase, in which the location of the hidden platform was switched to a new quadrant, the mutant mice were less efficient in switching to the novel platform location ( Figure 5E). In other behavioral tests, Emx1-Cre;Ptprs fl/fl mice showed normal levels of motor coordination, repetitive behaviors, and prepulse inhibition ( Figure 5-figure  supplement 1). Notably, these mice showed moderately decreased locomotor activity in a familiar environment (Laboras cages) and moderately increased locomotor activity in a novel environment (open-field apparatus). In addition, these mice showed moderately increased open-field center time and strongly increased open-arm time in the elevated plus-maze test, suggestive of anxiolytic-like behavior, although they performed normally in the light-dark chamber test.
These results indicate that Emx1-Cre;Ptprs fl/fl mice display decreases in the ability to recognize a novel object (novel-object recognition test), a novel social target (three-chamber test), a novel reward-arm location (Y-maze test) and a novel platform location (Morris water maze), collectively suggesting that the mutant mice have suppressed novelty recognition.

Presynaptic PTPs-dependent regulation of postsynaptic NMDARs is important for novelty recognition
The impaired novel recognition in Emx1-Cre;Ptprs fl/fl mice may involve PTPs-dependent regulation of postsynaptic LTP. To assess this possibility, we first tested whether pharmacological activation of NMDARs could rescue the impaired social-novelty recognition in Emx1-Cre;Ptprs fl/fl mice.
Acute treatment with the NMDAR agonist D-cycloserine (20 mg/kg; i.p.) rescued social-novelty recognition deficits in Emx1-Cre;Ptprs fl/fl mice, as assessed using the three-chamber test, without affecting social-novelty recognition in WT mice ( Figure 6A). In addition, D-cycloserine had no effect on social approach in WT or Emx1-Cre;Ptprs fl/fl mice.
We next tested whether presynaptic, but not postsynaptic, deletion of PTPs affects novelty recognition by injecting AAV1-hSyn-Cre-eGFP or control AAV1-hSyn-DCre-eGFP into the CA3 or CA1 region of Ptprs fl/fl mice (8-11 weeks). Cre-induced PTPs deletion in the CA3 region resulted in impaired social-novelty recognition in Ptprs fl/fl mice in the three-chamber test compared with control Ptprs fl/fl mice expressing EGFP alone, without affecting social approach ( Figure 6B,C). In contrast, Cre-induced PTPs deletion in the CA1 region had no effect on social-novelty recognition or social approach.
Cre-induced deletion of PTPs in the CA3 region also impaired recognition of the novel reward arm location in the Y-maze test in Ptprs fl/fl mice compared with control Ptprs fl/fl mice expressing EGFP alone ( Figure 6D). These results suggest that deletion of presynaptic, but not postsynaptic, PTPs impairs social novelty and novel reward-arm recognition in adult mice.
To more directly test whether NMDAR function in the postsynaptic CA1 area is important for social novelty and reward-arm recognition, we acutely knocked down expression of the GluN1 subunit of NMDARs in the CA1 region of Ptprs fl/fl mice (8-11 weeks) and monitored its impacts on novelty recognition. WT mice (C57BL/6J) injected in the CA1 region with AAV-pU6-shGluN1 displayed suppressed social-novelty recognition, but normal social approach, in the three-chamber test ( Figure 6E,F). In contrast, control WT mice injected with AAV-pU6-shCtrl displayed normal socialnovelty recognition and social approach.
In addition, WT mice injected in the CA1 region with AAV-pU6-shGluN1 failed to recognize the novel reward-arm location in the Y-maze test, whereas control WT mice similarly injected with AAV-pU6-shCtrl showed no such change ( Figure 6G). Decreased expression of GluN1 was verified by immunoblot analysis of the GluN1 protein expressed in the infected hippocampus ( Figure 6H). Therefore, normal expression of NMDARs in the CA1 region is important for social novelty and novel reward-arm recognition.
These results collectively suggest that presynaptic PTPs-mediated regulation of postsynaptic NMDAR currents and NMDAR-dependent LTP is important for social novelty and novel reward-arm recognition in mice.
Our data also suggest mechanisms by which presynaptic PTPs regulates postsynaptic NMDAR currents and LTP. Our data suggest that the decreased synaptic levels of GluN2B at mutant excitatory synapses, supported by immunoblot analysis of synaptic proteins and the faster decay kinetics of NMDAR currents, may partly contribute to the decreased NMDAR currents, which in turn seems to underlie the reduced LTP, based on quantitative comparisons. In addition, point mutations or small deletions in the extracellular region of PTPs that disrupt trans-synaptic adhesions of Ig1-3 or FNIII1-2 domains with TrkC, SALM5, Slitrk1, CSPG/HSPG, and NGL-3 do not affect the PTPs-dependent regulation of postsynaptic LTP, suggesting that the cytoplasmic regions of PTPs, containing the tyrosine phosphatase activity and mediating presynaptic protein-protein interctions, may be important.

2008), SynGAP
Our data also suggest that presynaptic PTPs regulates not only postsynaptic NMDAR currents and LTP, but also behavioral novelty recognition. This hypothesis is supported by the observations that 1) D-cycloserine rescues social novelty deficits in Emx1-Cre;Ptprs fl/fl mice, 2) acute presynaptic (CA3) deletion of PTPs impairs social novelty and novel reward-arm recognition, and 3) acute postsynaptic (CA1) knockdown of the GluN1 subunit of NMDARs impairs social novelty and novel reward-arm recognition. Previous studies have implicated the hippocampus and hippocampal NMDARs in the regulation of novelty recognition in various contexts (Hitti and Siegelbaum, 2014;Kitanishi et al., 2015;Leroy et al., 2017;Rondi-Reig et al., 2001). In addition, hippocampal LTD but not LTP has been associated with novel-object/feature recognition in a space (Dong et al., 2012;Kemp and Manahan-Vaughan, 2004;Kemp and Manahan-Vaughan, 2007). Our data, in particular those from CA3-PTPs-KO and CA1-GluN1-knockdown experiments, extend previous findings by demonstrating that a presynaptic adhesion molecule-PTPs-can regulate social and reward-arm novelty recognition through trans-synaptic regulation of postsynaptic NMDAR currents and NMDAR-dependent LTP in the hippocampus. Whether our results would also involve the rescue of NMDAR-dependent LTD, which is impaired in the Emx1-Cre;Ptprs fl/fl hippocampus, remains to be determined. In addition, care should be taken in interpreting our results because Emx1-Cre;Ptprs fl/fl mice lack Ptprs expression not only in the hippocampus but also in other brain regions such as the prefrontal cortex, known to be involved in social novelty cognition in mice and rats (Cao et al., 2018;Finlay et al., 2015;Liang et al., 2018;Niu et al., 2018;Watson et al., 2012).
Notably, Emx1-Cre;Ptprs fl/fl mice also display anxiolytic-like behavior as supported by moderately increased center time in the open-field test and strongly increased open-arm time in the elevated plus-maze test, although these mice acted normally in the light-dark test. Whether the anxiolytic-like behavior involves suppressed NMDAR function in the hippocampus or other brain regions is an open question. Previous studies have shown that anxiety involves various brain regions, including anterior cingulate cortex, hippocampus, and amygdala (Adhikari, 2014;Apps and Strata, 2015;Barthas et al., 2015;Calhoon and Tye, 2015;Duval et al., 2015;Kim et al., 2011;Tovote et al., 2015).
Lastly, a recent paper has reported that deletion of all three LAR-RPTPs (PTPs, PTPd, and LAR) minimally affects synapse development but critically regulate postsynaptic NMDAR, but not AMPAR, responses by a trans-synaptic mechanism (Sclip and Südhof, 2020). These results are in line with our results that PTPs deletion in mice selectively decreases NMDAR, but not AMPAR, currents. This study also extends our study by finding that PTPd and LAR, in addition to PTPs, participate in the regulation of trans-synaptic NMDAR regulation. An obvious direction for follow-up studies based on these results would be to identify specific mechanisms underlying the trans-synaptic but indirect NMDAR regulation.
In conclusion, our results suggest that presynaptic PTPs regulates postsynaptic NMDAR currents and NMDAR-dependent LTP in the hippocampus through trans-synaptic adhesion-independent mechanisms, and suggest that this regulation may be important for novelty recognition in socialand reward-related contexts.  (Figure 2 and

Mice
We received ES cells containing a Ptprs-targeted allele from KOMP (RRID:MGI:5797751; Ptrps tm1a (KOMP)Mbp ), and transgenic mice were generated through ES injection. We backcrossed it with C57BL/6J strains for more than five generations before we conduct experiments. After mating with Protamine-Flp, the resulting Ptprs fl/+ mice were crossed with Emx1-Cre mice (JAX #005628) to produce Emx1-Cre;Ptprs fl/fl mice. For Ptprs global knockout mice (Ptprs gKO mice), we treated fertilized eggs at the two-cell embryo stage with purified HTNC, a cell-permeable Cre recombinase (Histidine-TAT-Nuclear localization-Cre fusion peptide (Peitz et al., 2002), in a media at the final concentration of 0.3 mM for 30-40 mins. Emx1-Cre;Ptprs fl/fl mice were genotyped by polymerase chain reaction (PCR) using the following primer sets: Ptprs allele, forward, 5'-CTCCTTCCTCTCCAAACGG-3', reverse, 5'-TGAGCGTCTGAATGGAGCAC-3', Cre allele, forward, 5'-GATCTCCGGTATTGAAAC TCCAGC-3', reverse, 5'-GCTAAACATGCTTCATCGTCGG-3'. Appropriate expression patterns of Emx1-Cre was confirmed by crossing with ROSA-tdTomato mice (JAX #7909). All mice were housed and bred at the mouse facility of Korea Advanced Institute of Science and Technology (KAIST) and maintained according to the Animal Research Requirements of KAIST. All animals were fed ad libitum and housed under 12 hr light/dark cycles (light phase during 1 am to one pm). We crossed Ptprs fl/fl male mice and Emx1-Cre;Ptprs fl/fl female mice to produce littermate pairs of wild-type (WT) and KO mice. Mice were weaned at the age of postnatal day 21, and mixed-genotype littermates in the same gender were housed together until experiments. All procedures were approved by the Committee of Animal Research at KAIST (KA2016-33).

Electrophysiology
For electrophysiological experiments for the hippocampus, sagittal hippocampal slices (400 mm thickness for extracellular recordings and 300 mm for intracellular recordings) from Emx1-Cre;Ptprs fl/ fl mice, or virus-injected mice, and their appropriate controls (see each figures) were prepared using a vibratome (Leica VT1200) in ice-cold dissection buffer containing (in mM) 212 sucrose, 25 NaHCO 3 , 5 KCl, 1.25 NaH 2 PO 4 , 0.5 CaCl 2 , 3.5 MgSO 4 , 10 D-glucose, 1.25 L-ascorbic acid and 2 Na-pyruvate bubbled with 95% O 2 /5% CO 2 . For virus-injected samples, slices with fluorescence signals derived from co-injected AAV1-hSyn-eGFP were used. The slices were recovered at 32˚C for 1 hr in normal ACSF (in mM: 125 NaCl, 2.5 KCl, 1.25 NaH 2 PO 4 , 25 NaHCO 3 , 10 glucose, 2.5 CaCl 2 and 1.3 MgCl 2 oxygenated with 95% O 2 /5% CO 2 ). For electrophysiological recordings, a single slice was moved to and maintained in a submerged-type chamber at 28˚C, continuously perfused with ACSF (2 ml/min) saturated with 95% O 2 /5% CO 2 . Stimulation and recording pipettes were pulled from borosilicate glass capillaries (Harvard Apparatus) using a micropipette electrode puller (Narishege). For extracellular recordings, mouse hippocampal slices at the age of postnatal days 16-33 were used (for the exact ages of each experiment, see S1_Table). fEPSPs were recorded in the stratum radiatum of the hippocampal CA1 region using pipettes filled with ACSF (1 MW). fEPSPs were amplified (Multiclamp 700B, Molecular Devices) and digitized (Digidata 1440A, 1550 Molecular Devices) for analyses. The Schaffer collateral pathway was stimulated every 20 s with pipettes filled with ACSF (0.3-0.5 MW). The stimulation intensity was adjusted to yield a half-maximal response, and three successive responses were averaged and expressed relative to the normalized baseline. To induce LTP or LTD, high-frequency stimulation (100 Hz, 1 s), theta-burst stimulation (40 trains of pulses, each train is composed with 4 stimuli in 100 Hz; 40 trains are divided by four bursts, each containing 10 trains with 1 s inter-burst interval; 170 ms inter-train-interval), or low-frequency stimulation (1 Hz, 15 min) were applied after a stable baseline was acquired. To induce mGluR dependent LTD, DHPG (50 mM) was added to ACSF for 5 min after acquiring a stable baseline. The paired-pulse ratio was measured across a range of inter-stimulus intervals of 25,50,75,100,200, rescue, 20 mM DCS-containing ACSF were used to perfuse slices from the beginning of baseline recording.
Whole-cell patch-clamp recordings of hippocampal CA1 pyramidal neurons were made using a MultiClamp 700B amplifier (Molecular Devices) and Digidata 1440A, 1550 (Molecular Devices). During whole-cell patch-clamp recordings, series resistance was monitored each sweep by measuring the peak amplitude of the capacitance currents in response to short hyperpolarizing step pulse (5 mV, 40 ms); only cells with a change in <20% were included in the analysis. For afferent stimulation of hippocampal pyramidal neurons, the Schaffer collateral pathway was selected. For NMDA/AMPA ratio experiments, mouse hippocampal slices (P19-23) were used. Recording pipettes (2.5-3.5 MW) were filled with an internal solution containing the following (in mM): 100 CsMeSO 4 , 10 TEA-Cl, 8 NaCl, 10 HEPES, 5 QX-314-Cl, 2 Mg-ATP, 0.3 Na-GTP, and 10 EGTA, with pH 7.25, 295 mOsm. CA1 pyramidal neurons were voltage clamped at À70 mV, and EPSCs were evoked at every 15 s. AMPAR-mediated EPSCs were recorded at À70 mV, and 20 consecutive responses were recorded after stable baseline. After recording AMPAR-mediated EPSCs, the holding potential was changed to +40 mV to record NMDAR-mediated EPSCs. NMDA component was measured at 60 ms after the stimulation. The NMDA/AMPA ratio was determined by dividing the mean value of 20 NMDAR EPSCs by the mean value of 20 AMPAR EPSC peak amplitudes.

Brain imaging
To examine the gross morphology of the brain, coronal sections (50 mm) of mouse brains were prepared using a vibratome (Leica) and mounted on DAPI-containing Vectashield (Vector Laboratory). For X-gal staining, coronal sections (100 mm) of mouse brains were prepared using a vibratome (Leica) followed by X-gal staining for 30 min (20 mg/mL X-gal; in 2 mM MgCl 2 , 5 mM K 4 Fe (CN) 6 .3H 2 O(Sigma #P-8131), 5 mM K 3 Fe(CN) 6 , 0.01% DOC, 0.02% NP-40 in 1 x PBS). For immunofluorescence imaging of brain sections after electrophysiological and behavioral experiments, coronal brain sections (50 mm) were prepared and used for image acquisition without staining using a confocal microscope (LSM-780, Zeiss).

Phosphoscan proteomic analysis
Changes in phospho-tyrosine levels in proteins from Emx1-Cre;Ptprs fl/fl mice were determined using PhosphoScan service (Cell Signaling Technology). Briefly, mouse brain samples containing the cortex and hippocampus were dissected in ice-cold dissection buffer (see the Materials and methods for electrophysiology) with protease/phosphatase inhibitor cocktails. Brain samples from three different mice were pooled to make n number of one. Brain samples were snap-frozen in liquid nitrogen were protease-digested and fractionated by solid-phase extraction. The fractionated peptides were incubated with designated immobilized PTM (post-translational modification)-motif antibodies, and the peptides containing the corresponding PTM-sequences were eluted and analyzed using LC-MS/MS. Mass spectra were assigned to peptide sequences using Socerer program. Finally, the peptide sequence assignment was linked to parent ion peak intensities to measure approximate foldchanges in validated peptides between paired samples.

Novel object recognition test
Novel object recognition test was performed in the open-field box. On day 1, mice were allowed to explore a novel object (white cylinder) On day 2, mice explored two identical objects (blue cylinder or silver-colored box) for 20 min. On day 3, mice were placed in the same box where one of the two objects was replaced with a new object (blue cylinder and silver-colored box). Sniffing time for each object was measured. Object exploration was defined by the mouse's nose being oriented toward the object and came within 2 cm of the target as measured by EthoVision XT12 program (Noldus).

Three-chamber test
The three-chambered apparatus, designed to measure social approach and social novelty recognition (Silverman et al., 2010), had the dimensions of 40 cm W x 20 cm H x 26 cm D with a center chamber of 12 cm W and side chambers of 14 cm W. In the first session, the mouse could freely move around the whole three-chambered apparatus with two small containers in the left or right corner for 10 min (Session #1). The mouse was then gently guided to the center chamber while a novel 'Object' and a wild-type stranger mouse 'Stranger 1 (129Sv strain)' were placed in the two plastic containers. The subject mouse was then allowed to freely explore all three chambers for 10 min (Session #2). In the third session, the subject mouse was again gently guided to the center chamber while the 'Object' was replaced with a wild-type 'Stranger 2' mouse. The subject mouse again freely explored all three chambers for 10 min (Session #3). Object/Stranger exploration was defined by the mouse's nose being oriented toward the target and came within 2 cm of it as measured by EthoVision XT 12 program (Noldus). Three-chamber tests over 5-consecutive days were performed as described previously (Bariselli et al., 2018). For this experiment, we used mice that did not experience other behavioral tests to minimize potential confounds. The same stranger was exposed to the subject mouse during the first four days. Minor differences in this test, compared with the above mentioned three-chamber test, were the lack of session #3, the use of empty space instead of an object, and 5-min-long session #1 during days 2-5 (10 min for session #1 on day 1).

Water-based Y-maze
The Y-maze test was performed as described previously (Trinh et al., 2012). The Y-maze apparatus was composed of three identical arms (35cm-long, 10cm-wide, 25 cm high). The Y-maze apparatus was placed at the center of a water tank (120 cm diameter) and the platform was placed in one of the three arms and hidden 2 cm under the water (20-22˚C) made invisible by white paint. On day 1, a subject mouse was placed in an arm without the hidden platform and allowed to freely swim until it finds the platform. Mice that cannot find the platform in 2 min were guided to the platform. Each session consisted of 5 swim trials, and four sessions were performed on each day. On day 2, a subject mouse was tested for the memory of the platform location for one session. Only the mice that were successful in identifying the correct arm over 80% of the time were used for the following experiments, where the platform location was changed to the opposite arm that was empty on the previous day. The day two experiments consisted of four sessions (five swims per session).

Morris water maze
Mice were trained to find the hidden platform (10 cm diameter) in a white plastic tank (120 cm diameter). Mice were given three trials per day with an inter-trial interval of 30 min. Experiments for the learning phase of the water maze were performed for seven consecutive days, followed by the probe test on day eight where mice were given 1 min to find the removed platform. For reversal training (days 9-13), the location of the platform was switched to the opposite position from the previously trained location, and mice were allowed to re-learn the new position of the platform. Target quadrant occupancy and the exact number of crossings over the former platform location during the probe test were measured using EthoVision XT12 program (Noldus).

Laboras test
For long-term measurements of mouse movements, we used the LABORAS system (Metris) (Quinn et al., 2006), designed to detect and analyze vibrations delivered from a cage with a mouse to a carbon-fiber vibration-sensitive plate placed underneath the cage. Each mouse was placed in the LABORAS cage without habituation, and its movements were recorded for 72 consecutive hours. The data during the last 48 hr, a period after full habituation to the environment, were analyzed by the software.

Open-field test
Mice were placed in an open field box (40Â40Â40 cm) and recorded with a video camera for 60 min. The center zone lines were 10 cm apart from the edge. The testing room was illuminated at~50 lux or 0 lux. Mice movements were analyzed using EthoVision XT12 program (Noldus).

Rotarod test
Mice were placed on the rotating rod for 10 s, followed by the start of rod rotation. The rotating speed of rod was gradually increased from 4 to 40 rpm over 5 min. The assay was performed for five consecutive days, while measuring the latencies of mice falling from the rod or showing 360-degree rotation on the rod.

Elevated plus-maze test
The elevated plus-maze consisted of two open arms, two closed arms, and a center zone, and was elevated to a height of 50 cm above the floor. Mice were placed in the center zone and allowed to explore the space for 8 min. The data were analyzed using EthoVision XT12 program (Noldus).

Light-dark test
The light-dark apparatus consisted of light (~200 lux) and dark (~0 lux) chambers adhered to each other. The size of the light chamber was 20Â30Â20 cm, and that of the dark chamber was 20Â13Â20 cm. An entrance enabled mice to freely move across the light and dark chambers. Mice were introduced to the center of the light chamber and allowed to explore the apparatus freely for 5 min. The time spent in dark and light chambers and the number of transitions were measured using EthoVision XT12 program (Noldus).

Prepulse inhibition
A subject mouse was placed in a startle chamber (SR-LAB). For acclimation, a background noise of 65 dB pulse was given for 5 min. After acclimation, 57 testing sound pulses with varying inter-trial intervals (7-23 s) were given. The testing sound pulses consist of 4 pulses (4 Â 120 ms, 120 dB) in the beginning and end stage of the test, and seven pulses (120 ms, 120 dB each) paired with prepulses (20 ms 100 ms prior to) at 70, 75, 80, 85 and 90 dB (total 35 paired pulses).

Statistics
For statistical comparison of two samples (e.g., WT vs. cKO), Student's t-test or Mann-Whitney test was used. The normality of data distributions was tested using the D'Agostino and Pearson normality test or Shapiro-Wilk normality test. Mann-Whitney tests were used for any column in either of the two tests in which the p-value was less than 0.05. For immunoblot and qRT-PCR results, a one-sample t-test was used. For results with one independent variable [e.g., cKO-eGFP vs. cKO-Ptprs(4A) vs. cKO-Ptprs(Y224S)], one-way analysis of variance (ANOVA) with Dunnett's multiple comparison test was used. For results with two independent variables (e.g., WT-Veh vs. WT-DCS vs. cKO-Veh vs. cKO-DCS), two-way ANOVA with Sidaks' multiple comparison test was used. For additional information on gender and number of mice/samples, detailed test information and statistical results, see Supplementary file 3. GraphPad Prism seven was used for all statistical analyses, except for the DAVID analysis (http://david.ncifcrf.gov).
The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication. . Transparent reporting form Data availability All data generated or analysed during this study are included in the manuscript and supporting files.