Glucocorticoids rescue cell surface trafficking of R451C Neuroligin3 and enhance synapse formation

Neuroligins are synaptic cell adhesion proteins with a role in synaptic function, implicated in neurodevelopmental disorders. The autism spectrum disorder‐associated substitution Arg451Cys (R451C) in NLGN3 promotes a partial misfolding of the extracellular domain of the protein leading to retention in the endoplasmic reticulum (ER) and the induction of the unfolded protein response (UPR). The reduced trafficking of R451C NLGN3 to the cell surface leads to altered synaptic function and social behavior. A screening in HEK‐293 cells overexpressing NLGN3 of 2662 compounds (FDA‐approved small molecule drug library), led to the identification of several glucocorticoids such as alclometasone dipropionate, desonide, prednisolone sodium phosphate, and dexamethasone (DEX), with the ability to favor the exit of full‐length R451C NLGN3 from the ER. DEX improved the stability of R451C NLGN3 and trafficking to the cell surface, reduced the activation of the UPR, and increased the formation of artificial synapses between HEK‐293 and hippocampal primary neurons. The effect of DEX was validated on a novel model system represented by neural stem progenitor cells and differentiated neurons derived from the R451C NLGN3 knock‐in mouse, expressing the endogenous protein. This work shows a potential rescue strategy for an autism‐linked mutation affecting cell surface trafficking of a synaptic protein.


| INTRODUCTION
The neuroligin (NLGN) family of postsynaptic cell adhesion molecules plays a major role in regulating synapse formation and maturation during brain development. 16][7][8] Among the NLGNs, the highest number of ASDs-associated mutations have been found in the NLGN3 and NLGN4 genes, [9][10][11] although ASDs-mutations have been recently mapped in the NLGN1 and NLGN2 genes. 12,13It is noteworthy that most of the mutations result in residue substitutions impairing the folding of the extracellular domain, therefore causing the retention of the mutant protein in the endoplasmic reticulum (ER). 4Ultimately, this leads to a reduced trafficking of the mutant protein to the cell surface causing functional and behavioral alterations. 14ong the NLGNs variants, the R451C substitution in NLGN3 was one of the first mutations to be reported in association with ASDs.
The cysteine residue replacing the arginine at position 451 causes a local misfolding in the structure of the extracellular domain of the protein, leading to a slowdown in the maturation process of NLGN3. 15,16erefore, mutant R451C NLGN3 is retained in the ER through the persistent association with molecular chaperones, it is degraded at a high rate by the proteasome, 15,16 and consequently the amount of mutant protein reaching the cell surface is reduced.At the functional level, the purified mutant protein is unable to associate with neurexin 17 and when exposed on the surface of a heterologous system, it has reduced capacity to form synapses in transfected hippocampal neurons. 18In the knock-in mouse strain expressing the R451C mutation, mutant endogenous NLGN3 is highly unstable and is promptly degraded by the proteasome.Consequently, NLGN3 protein levels in the brain are reduced to around 10% compared with its levels in the parental mice strain. 19,20Consistent to the in vitro findings, the mutant protein is retained in the ER and degraded by the proteasome also in vivo. 21The accumulation of misfolded proteins in the ER can lead to a cellular stress condition and to the activation of the unfolded protein response (UPR), a signaling pathway that aims at reducing the stress in the organelle by enhancing protein folding capacity and proteasome-mediated degradation, while attenuating protein synthesis. 22The ER-retention of R451C NLGN3 activates the UPR, both in vitro and in vivo.Interestingly, in the R451C NLGN3 mouse brain, the activation of UPR is specific to the cerebellum, where the UPR modulates neurotransmission. 21,23In agreement with this, it has been recently shown that other ASDs mutations cause retention of mutant NLGN3 in the ER and UPR activation. 24In recent years, several studies have focused on developing treatments for correcting protein misfolding and impaired trafficking, and attenuating the UPR by using disease model systems both in vitro and in vivo.UPR signaling is crucial in several diseases, particularly in those involving highly secretory cell types, like neurons. 25,26Since the nature of the misfolding caused by the R451C substitution is confined to a superficial loop of the acetylcholinesterase-like domain, we have used this mutant protein to select compounds to restore NLGN3 trafficking to the cell surface.
In this work, using an FDA-approved small molecule drug library (2662 compounds), we have identified several compounds, some of which belonging to the glucocorticoids (GCs) family, that favor NLGN3 trafficking, both for the wild type (WT) and R451C forms.Specifically, we have shown that Dexamethasone (DEX) favors the exit of R451C NLGN3 from the ER and improves its trafficking to the cell surface.Our results showed that GCs mitigate ER stress and enhance the formation of artificial synapses.Isolated neural stem progenitor cells (NSPCs) can be propagated and differentiated in vitro, maintaining high neurogenic potential and representing a valuable tool for studying the mechanisms of several diseases. 27With the goal of confirming our results in a cell model that express either WT or R451C NLGN3 as endogenous proteins, we have isolated NSPCs from the dentate gyrus of the hippocampus of adult WT and knock-in R451C NLGN3 mouse strains and established a novel model system for studying the molecular mechanisms underlying a monogenic form of ASDs.Cell cultures derived from the R451C NLGN3 mice showed reduced levels of mutant NLGN3 due to proteasomal degradation, as previously observed in other cell-based systems.
Interestingly, we found that DEX treatment increased mutant NLGN3 protein levels both in proliferative and differentiated conditions and reduced ER stress, suggesting that the selected compound is also effective in a cell model expressing the endogenous protein of this form of ASDs.

| GCs improve the trafficking of both soluble and full-length R451C NLGN3
A library of 2662 compounds (FDA-approved small molecule drug library) was screened for molecules that improve the trafficking of R451C NLGN3.We have used a fluorescent approach, based on HEK-293 cells engineered to produce a truncated form of NLGN3, lacking the transmembrane and intracellular domains, and C-terminally fused to the fluorescent protein Venus (NLGN3-Venus). 28The resulting cell lines produce a fluorescent NLGN3, either WT or R451C, which is secreted by the cells, therefore allowing the quantification of protein trafficking by measuring fluorescence levels in the cell culture medium.In this system, due to the retention of the mutant protein in the ER, the cell line expressing R451C NLGN3-Venus shows around 50% reduction in the amount of the secreted protein in comparison to cells expressing the WT protein, as shown before. 28Each compound was added to the medium 24 h after seeding, at the concentration of 2 μM, and the fluorescence signal was measured after 60 h, to allow the mutant and fluorescent protein to accumulate in the medium.The initial screening was conducted using only the cell line expressing R451C NLGN3-Venus.
After this first screening, 13 compounds (see Table 1) were identified for their ability to increase the fluorescence levels in the culture medium at least two-fold, in comparison to the untreated R451C NLGN3-Venus control condition (Figure 1A).Most of the compounds we identified belonged to the GCs family.
After this first screening, 4 out of the 13 compounds that showed activity were selected for further analysis: alclometasone dipropionate (AD), desonide (D), prednisolone sodium phosphate (PSP) and DEX.These GCs were further tested for their effect on the full-length NLGN3 protein form.HEK-293 cell lines stably expressing NLGN3, either WT or R451C, were generated and the GCs were administered at the concentration of 2 μM for 48 h.
Western blot analysis by using the anti-NLGN3/FLAG antibody (Anti-FLAG, Sigma-Aldrich no.F7425) detected a single band in the lysates expressing the WT form that corresponds to the fully processed form of the protein, while R451C NLGN3 appeared as two distinct bands representing the unprocessed and the mature forms as previously described in De Jaco et al. 29 For the densitometric analysis, the signal resulting from the two bands together, was always considered.
No significant increase in protein levels, in cell lysates, was observed for the WT protein after treatment (Figure 1B).Instead, fulllength R451C NLGN3 levels increased significantly after treatment with each of the four GCs, in line with what was previously observed for the secreted form (Figure 1C).Changes in the intracellular localization of R451C NLGN3 after GCs administration were assessed by coimmunostaining with the ER-resident protein calreticulin.HEK-293 cells expressing full-length WT NLGN3, which shows a physiological localization on the cell surface, were used as control.The decrease in colocalization of NLGN3 with calreticulin was used to quantify the activity of each compound (Figure 2A).The correlation coefficient for colocalization in cells expressing WT NLGN3 was half of that shown by untreated cells expressing the mutant protein (Figure 2B), suggesting a significantly higher retention of R451C NLGN3 in the ER, in agreement with our previous results. 29Interestingly, the treatment with each of the four GCs significantly reduced the correlation coefficient in cells expressing R451C NLGN3, when compared with the untreated condition (Figure 2B), suggesting a positive effect on protein trafficking.
To quantify the fraction of protein reaching the cell membrane resulting from the treatment with GCs, we used a cytofluorimetric analysis conducted on intact cells (Figure 2C).The intensity values relative to the fluorescence present on the cell membrane (FITC-A) are reported in Table 2.The average FITC-A signal in HEK-293 cells expressing either WT or R451C NLGN3 following the treatments was higher compared with the untreated cells for both genotypes.These results suggest an effect of GCs on the cell surface trafficking of NLGN3 and a rescue effect on the R451C protein with respect to the WT untreated condition.Moreover, we investigated the levels of the extracellular cleaved domain of NLGN3, in the culture medium by western blot, and found that treatment with the selected GCs increased the levels of cleaved R451C NLGN3 (Figure 2D,E).

| Enhanced R451C NLGN3 protein levels are dependent on the activation of the glucocorticoid receptor
The mechanism of action of GCs usually relies on pleiotropic effects of the GC receptor (GR) on several transcriptional pathways.GR activity is modulated by posttranslational modifications, such as phosphorylation at Ser211, which is increased after GCs administration. 30To T A B L E 1 FDA-approved compounds selected after a wide screening for increasing R451C NLGN3 protein levels.investigate the involvement of the GR in mediating the effects of GCs on R451C NLGN3 protein levels, we analyzed the phosphorylation of the receptor at Ser211 in lysates of cells expressing either WT or R451C NLGN3, untreated or treated with AD, D, PSP, or DEX.A significant increase in phosphorylation levels was observed in the lysates from both cell lines after treatment with GCs in comparison to untreated conditions (Figure 3A).The combined treatment with the GR antagonist mifepristone (RU486) and DEX significantly reduced the effect of DEX on R451C NLGN3 levels, while having no effect on WT NLGN3 (Figure 3B).As a control, we confirmed that the combination of RU486 and DEX decreased the activation of the GR in both cell lines (Figure 3C).The involvement of the GR in increasing trafficking of R451C NLGN3 was also assessed by cytofluorimetric analysis on intact cells after the combined treatment with RU486 and DEX.
We observed a reduced localization of R451C NLGN3 on the cell surface, compared with cells treated with DEX alone (Figure 3D).
Taken together, these results indicate a direct involvement of the GR in mediating the effects of GCs, and in particular of DEX, on the cell surface trafficking of R451C NLGN3.We selected DEX for a detailed characterization of the effects of GCs in our cellular model of ASDs.
First, we evaluated the effect of different concentrations and times of treatment on R451C NLGN3 protein levels and on the phosphorylation of the GR on Ser211.We observed a significant activation of the GR at all three tested concentrations of DEX (0.02, 0.2, and 2 μM, Figure 4A), as well as a similar increase in R451C NLGN3 protein levels (Figure 4B).An increase in the mutant protein levels was detectable only after 16 and 48 h of treatment with DEX at 2 μM (Figure 4C).However, the activation of the GR was observed from the earliest time point, with Ser211 phosphorylation detectable at similar levels at all time points (1, 8, 16, and 48 h; Figure 4D).DEX had no

| DEX improves the stability of R451C NLGN3
We  5B).In order to understand whether DEX has a similar effect on other misfolded proteins, we transfected HEK-293 cells with a plasmid encoding for NLGN3 carrying the G221R substitution, previously characterized for causing a global misfolding of the protein and leading to its degradation through the proteasomal pathway. 29DEX treatment had no effect on G221R NLGN3 protein levels compared with the untreated condition (Figure S1).
Finally, we analyzed the effect of DEX on a mutant variant of the neuronal secretory protein neuroserpin, delta neuroserpin (ΔNS), which is detectable in the cell lysates only after treatment with the proteasomal inhibitor MG132. 31,32Also in this case, treatment with DEX was not able to increase ΔNS protein levels compared with the untreated condition (Figure S1).These results suggest that DEX has an effect on the stability of only partially misfolded proteins.

| DEX reduces the ER stress induced by the R451C NLGN3 mutant protein
It was previously shown that R451C NLGN3 activates the UPR in vitro, in PC12 cells with inducible expression of mutant NLGN3, 23 and in vivo, in the cerebellum of the knock-in mouse R451C NLGN3. 21Previous studies support a role for GCs in ameliorating UPR by enhancing the correct folding and secretion of proteins. 33,34sed on this, we evaluated the effect of DEX on the levels of UPR sensors and mediators in our cell systems.We observed that DEX reduced the activation of protein kinase-like endoplasmic reticulum kinase caused by R451C NLGN3, as shown by the decrease in eIF2α phosphorylation in cells expressing mutant but not WT NLGN3 (Figure 6A).We also quantified the levels of the unprocessed form of activating transcription factor 6 (ATF6), which has been shown to be upregulated in ER stress conditions. 35We found that DEX treatment reverted the increase in ATF6 levels observed in lysates of cells expressing R451C NLGN3, with no effect on WT NLGN3-expressing cells (Figure 6B).Finally, we analyzed the activation of the inositolrequiring enzyme 1 (IRE1) branch of the UPR by analyzing the downstream splicing of XBP-1, using a reporter plasmid expressing XBP1 fused to the fluorescent protein Venus.Under ER stress conditions, spliced XBP1 mRNA is translated into a fluorescent protein that can be detected by fluorescence microscopy. 36As shown in Figure 6C, the treatment with DEX caused a reduction in the unconventional splicing of XBP1 in R451C NLGN3 cells, in comparison to the untreated cells.Among the target genes for the three UPR signaling cascades are several ER proteins such as immunoglobulin heavychain-binding protein (BiP), GRP94, and protein disulfide isomerase (PDI), involved in protein folding and ER quality control systems. 37By analyzing the levels of these ER chaperones, we observed a decrease in PDI protein levels after treatment with DEX only in cells expressing R451C NLGN3, in agreement with the decrease in UPR activation (Figure S2A).However, DEX had no effect on the levels of GRP94 and BiP in either the WT or R451C NLGN3 cells (Figure S2B).These results show that DEX alleviates ER stress by modulating all three branches of the UPR and specific ER proteins.

| DEX increases synapse formation in cocultures with primary hippocampal neurons
It has been described that mutant R451C NLGN3 no longer has synaptogenic activity in vitro due to inefficient transport to the cell surface 18 and decreased affinity for the partner protein neurexin. 17 next studied whether the treatment with DEX could rescue synapse formation.We analyzed the artificial synapses formed between HEK-293 cells, expressing NLGN3, and primary hippocampal neurons, cultured for 7 days before adding the HEK-293 cells.After 24 h, the  F I G U R E 3 Legend on next page.
NLGN3 signal (FLAG-staining) after DEX treatment, compared with untreated cells (Figure 7A).It is noteworthy that DEX treatment led to a comparable level of colocalized signals in cocultures of hippocampal neurons with HEK-293 cells expressing either WT or R451C NLGN3.Notably, when comparing the distribution of colocalized signal areas in untreated cultures, we observed a more pronounced degree of colocalization in the WT cocultures as opposed to the mutant ones (Figure 7B).
As a control, we showed that DEX had no effect on VGLUT1 staining in primary neurons compared with untreated neurons, both in cocultures with WT NLGN3 cells (data not shown) and with R451C NLGN3 cells (Figure S3).Our results thus support that the increase in colocalization, in the presence of DEX, was not due to an increase in the number of presynaptic puncta but rather to the increased trafficking of NLGN3 to the cell surface.

| Generation of neural staminal progenitor cultures from adult mice brain
9][40] When cultured under proliferative conditions, NSPCs express the staminal marker nestin, as confirmed by immunofluorescence analysis (Figure S4); while, after 15 days of culture in our differentiating conditions, differentiated NSPCs express the neuronal marker MAP2 (Figure S4).To validate our new model system, levels of NLGN1, 2, and 3 in proliferating and in differentiated conditions were analyzed both in WT and in R451C NLGN3 cells (Figure S5A).As expected, NSPCs derived from the hippocampus of R451C NLGN3 mice showed a strong reduction in NLGN3 protein levels compared with the WT cells as previously observed. 19,21,41reover, mutant NLGN3 levels didn't increase after inducing differentiation (Figure S5A).We have previously shown that R451C NLGN3 is preferentially degraded by the proteasome, both in vitro and in vivo. 21,29In agreement with this result, when R451C NLGN3 cultures were treated with the proteasome inhibitor MG132 (2 μM for 6 h), a significant increase in NLGN3 levels was observed, in comparison to the untreated condition (Figure S5B), confirming that the mutant protein is degraded by the proteasome in this newly  NLGN3 and NLGN4 proteins. 4While some of the mutations severely disrupt the structural integrity of the extracellular neurexin-binding domain, others result in milder effects depending on the region they map in the structure. 45Among all the mutations found in the NLGNs, the R451C NLGN3 was the first to be identified in ASDs patients and it represents the reference mutation in the field because it has been extensively studied by our and several other groups over two decades. 9,21,23,29The R451C substitution causes a localized misfolding of the extracellular domain of NLGN3, which is therefore significantly retained in the ER with only a residual fraction of the protein reaching the cell surface. 19,29,46Although this mutation has been highly characterized for its effects in several model systems both in vitro and in vivo, few studies attempted to identify a possible rescue strategy for correcting impaired cell surface trafficking.For this reason, we have generated a simple but reliable cell system expressing a truncated and fluorescent form of R451C NLGN3 and screened an FDAapproved library of 2662 small chemical compounds.We identified 13 compounds able to increase the secretion of the truncated R451C NLGN3 in the culture medium 17,28,29,47 and we initially studied the effects of four of them (AD, D, PSP, and DEX), all belonging to the GCs family.We showed that they caused increased levels of fulllength R451C NLGN3 but did not affect levels of NLGN3 in cells expressing the WT form of the protein.With this analysis, we quantified total protein levels in the cells, including the fractions undergoing the maturation process, retained in the ER, and exposed on the cell surface.On the contrary, using fluorescence activated cell sorting (FACS) analysis on intact cells, we specifically measured the amount of protein exposed on the cell membrane and our results clearly showed that all compounds increased levels of both WT and mutant protein reaching the cell surface, thus having an effect in promoting the overall trafficking of NLGN3.In line with these results, we showed by immunocolocalization staining, that all four tested GCs were able to favor the exit of the R451C protein fraction retained in the ER.Generally, NLGN3 performs its functions tethered to the cell membrane, although it has been recently shown that its extracellular domain can be cleaved and secreted in the medium to regulate physiological processes, such as cell proliferation in glioma. 48,49The presence of ASDs-linked mutations specifically impairing folding and cell surface trafficking of NLGN3, also reduced the extracellular levels of the cleaved form in the culture medium. 24As expected, in our model system of HEK-293 cells stably overexpressing R451C NLGN3, we observed reduced levels of the cleaved protein form in the culture medium in comparison to cells expressing the WT protein, as previously observed by Quartier et al. 24 More interestingly, treatment with the four GCs led to a higher amount of cleaved R451C extracellular domain in the culture medium.This result indirectly supports an effect of the treatment in enhancing trafficking of the mutant protein from the ER to the cell surface.
To simplify the study, we continued our analysis of the molecular mechanisms elicited by the GCs, by using only one of the four compounds.Hence, DEX was chosen for its documented ability in promoting cell surface trafficking of other proteins in in vitro cellular systems. 50,51Moreover, the use of this drug offers the translational value of using it in in vivo studies.Nonetheless, the effect of DEX on protein trafficking has never been approached in the study of a synaptic protein or in the contest of ASDs.
Our results showed that DEX treatment prolonged the half-life of the mutant protein, from 23.4 to 55 h, therefore improving its stability.The effect of DEX may be related to the local nature of the  misfolding caused by the R451C mutation in the NLGN3 structure (De Jaco et al., 2010).In fact, DEX had no effect on improving trafficking of other misfolded neuronal proteins such as a mutant form of neuroserpin (Δneuroserpin) or NLGN3 protein carrying a more severe mutation (G221R).This is probably due to the severity of the defect introduced by these mutations in the respective protein structure. 29reover, the increased protein levels and trafficking of mutant NLGN3 to the plasma membrane are abolished when mifepristone (RU486), an antagonist of the GR receptor, is supplemented to the cell medium in combination with DEX, supporting a direct involvement of the signaling activated by this receptor in the rescue of the R451C NLGN3 trafficking.
The dual genetic and environmental nature characterizing ASDs represent the main obstacle to the search for effective therapies. 52wever, many pharmacological therapeutic approaches have been developed to modulate UPR effectors.Our group has previously shown that the retention of R451C NLGN3 causes ER stress and the activation of the UPR both in vitro and in vivo, which is correlated to neurotransmission alterations in the cerebellum. 21,234][55][56][57] Our results showed that DEX treatment in HEK-293 cells stably over-expressing the autism-linked mutant R451C NLGN3, decreased the levels of mediators (EIF2alpha, ATF6, and XBP-1) downstream the main branches of the UPR, repristinating the homeostasis in the ER.Furthermore, the analysis of UPR targets showed decreased levels of PDI, which plays a crucial role in protein folding by catalyzing disulfide bond formation. 58 understand whether the greater fraction of R451C NLGN3 protein present on the cell surface following DEX treatment, was able to interact with the physiological partner neurexin, we analyzed the formation of artificial contacts between HEK-293 overexpressing NLGN3 and primary hippocampal neurons.Our experiments showed a higher number of contacts after treating the cocultures with DEX, in comparison to the untreated control.This suggests that the fraction of protein reaching the cell surface is able to interact with neurexin despite the presence of the mutation.However, the effect of DEX does not seem to be specific for the R451C mutation, since it also increases the formation of artificial synapses between hippocampal neurons and HEK-293 cells expressing the WT form of NLGN3.These results are probably correlated to the increased trafficking of WT NLGN3 to the cell surface shown by the FACS analysis.Indeed, since NLGN3 is a cell-adhesion protein, higher amounts of the protein on the cell membrane lead to an increased number of synaptic contacts.
To validate our results in a physiological cell system, we chose to use NSPCs derived from the hippocampal region of adult knock in mice endogenously expressing the R451C mutation. 19NSPCs are advanced cell cultures for studying neurodevelopmental disorders, including ASDs and a valuable tool for studying cellular mechanisms underlying the effect of potential therapeutic drugs. 59NSPCs cultures express R451C NLGN3 as the endogenous protein and the regulation of the gene is under the proper gene promoter.We also derived NSPCs from the murine parental strain, as a control.WT NLGN3 was correctly synthesized and expressed in both proliferative and differentiated conditions, while the cultures expressing R451C NLGN3 showed poorly detectable levels of mutant NLGN3, thus reproducing the trafficking defects described in previous in vitro systems and in the in vivo knock in mouse model. 23,29On the contrary, NLGN1 and 2 protein levels were normal and progressively increased during neuronal differentiation in both WT and R451C NLGN3-expressing cells.
NSPCs cultures expressing R451C NLGN3 showed UPR activation as demonstrated by higher levels of the UPR target BiP in comparison to WT, in line with what we have previously observed in other in vitro and in vivo systems. 21,23Moreover, the R451C NLGN3 cultures showed a slower calcium response after depolarization, in comparison to the WT cells.This is in agreement with the dysregulation of Ca 2+ concentration that is associated with several neurodevelopmental disorders, including ASDs. 60 This could reflect a higher sensitivity of the NSPCs, where the endogenous mutant protein has a higher impact on the cellular homeostasis and the effect of DEX results in a stronger outcome.Unfortunately, it was not possible to evaluate whether DEX increased the localization of endogenous NLGN3 on the cell surface, therefore favoring the formation of synaptic contacts in differentiated cultures expressing R451C NLGN3, since the endogenous protein is not flag-tagged and there are no commercially available antibodies against NLGN3 to be used in immunostaining.However, to overcome this limitation, we analyzed the number of Synapsin 1 puncta in differentiated cultures and found that the presynaptic protein increased after DEX treatment in both genotypes, suggesting the formation of a higher number of synaptic contacts.
In this work, we present a novel cellular model of an ASDs-related mutation that recapitulates the effects of the R451C substitution in WT or carrying the R451C mutation. 17,28,29,47   food and water and were kept on a 12-h light/12-h dark cycle at the temperature of 20 ± 1 C.The experiments were done on 2-monthold mice (P60) mice.For genotyping, the R451C mutation was detected after PCR amplification of the genomic DNA, extracted from the mouse tail using primers previously described. 19The DNA was used for PCR amplification of the insertion site of the loxP site in the NLGN3 gene (FW 5 0 -TGTACCAGGAATGGGAAGCAG-3 0 ; RW
NSPCs were cultured in adherent condition on Poly-Ornithine (10 μg/mL, Sigma) and laminin (5-2.5 μg/mL, Corning) coated flasks or coverslips and maintained under proliferative conditions at 37 C in a 5% CO 2 atmosphere, half of medium was replaced with fresh medium each 3 days.

| Preparation of cell lysates
Lysates of HEK-293 cells were obtained using a lysis buffer (150 mM NaCl, 10 mM Tris pH 8.0, Nonidet P40 0.5%) supplemented with a protease inhibitor cocktail (Sigma-Aldrich).Lysis was performed for 15 min on ice, followed by centrifugation at 14 000g for 15 min at studied the decay of R451C NLGN3 protein over time by treating HEK-293 cells with the inhibitor of protein synthesis cycloheximide (CHX) for 24, 48, or 72 h.Treatment with CHX in combination with DEX significantly increased the levels of R451C NLGN3, both at 24 and 48 h, in comparison to CHX alone at the same time points (Figure 5A), suggesting that DEX enhances the stability of the mutant protein.Moreover, DEX treatment in combination with CHX increased the half-life of the mutant protein up to 55 h in comparison to CHX alone (23.4 h; Figure

cocultures were treated with 2
μM of DEX for 48 h.Immunofluorescence staining for the presynaptic marker VGLUT1 and NLGN3 (FLAG-staining) was followed by confocal microscopy analysis.As shown in Figure 7A, DEX treatment favored the formation of contacts between HEK-293 cells, either WT or R451C, and the neuronal synaptic terminals (Figure 7A), as shown by the increase in colocalization between VGLUT1 and NLGN3 on HEK-293 cells.These results were determined by assessing the proximity of VGLUT1 puncta to the T A B L E 2 FITC-A values in cells expressing WT or R451C NLGN3 following glucocorticoids treatments.

2 . 7 |
generated system as well.No effect was observed in the NSPCs expressing WT NLGN3.For a detailed characterization of the physiological properties of the neuronal cultures at 15 days of differentiation, we studied the calcium flux in either WT or R451C NLGN3 differentiated cultures by evaluating the calcium transients evoked by a depolarizing stimulus, represented by a local application of 50 mM KCl. Endogenous calcium flux was visualized by loading either WT or R451C NLGN3 cells with Fluo4, a calcium-sensitive fluorescent dye.42 R451C NLGN3 neurons showed a decreased amplitude after the KCl stimulus in comparison to WT neurons (FigureS5C), suggesting a different functional phenotype in comparison to WT cultures.DEX increases endogenous R451C NLGN3 levels, decreases UPR activation and increases the formation of synaptic puncta Proliferating and differentiated NSPCs, derived from either WT or R451C knock-in mice, were treated with DEX (2 μM for 48 h).In line with our previous results, a significant increase in the R451C NLGN3 protein levels was observed after DEX treatment in both proliferating and differentiated conditions, while no effect was found in cells expressing the WT protein (Figure8A,B).To assess whether the increase in the R451C NLGN3 levels elicited by DEX in NSPCs lysates in proliferative conditions was due to a transcriptional regulation, the mRNA for NLGN3 was quantified by real-time polymerase chain reaction (RT-PCR).No changes were found in the relative amounts of either R451C or WT NLGN3 mRNA after DEX treatment in comparison to the untreated conditions (FigureS6), supporting a posttranslational effect caused by the compound on the stability of the mutant protein.In agreement with our published data, proliferating and differentiated cultures expressing R451C NLGN3 showed higher levels of the chaperone BiP in comparison to WT cultures (Figure8C,D).21,23Interestingly, DEX treatment decreases BiP levels in NSPCs expressing R451C NLGN3, demonstrating a reduction of the UPR that is not detectable in WT cultures (Figure8C,D).Since our previous results in HEK-293 cells showed that higher levels of NLGN3 on the cell surface were able to induce artificial synapses by recognizing the presynaptic partner protein, we decided to analyze the expression of Synapsin1, a marker of synaptic specialization.43,44Our data indicate that the number of synaptic puncta increases after DEX treatment in both WT and R451C NLGN3 differentiated NSPCs cultures (Figure9).

F I G U R E 3
The activation of the glucocorticoid receptor (GR) is responsible for the stabilization of intracellular R451C NLGN3 caused by dexamethasone (DEX) treatment.(A) Representative western blot and densitometric analysis of phosphorylated GR levels in either wild type (WT) or R451C NLGN3 expressing cells after treatment with alclometasone dipropionate (AD), desonide (D), DEX or prednisolone sodium phosphate (PSP; 2 μM for 48 h), normalized to total GR and to the corresponding untreated condition (mean ± SEM; n = 3; unpaired t-test; **p < 0.01, ***p < 0.001).(B) Densitometric analysis and representative western blot of NLGN3 levels in lysates from WT and R451C NLGN3 expressing cells either untreated or treated with DEX (2 μM for 48 h), RU486 (20 μM for 48 h) or both compounds, normalized to the corresponding untreated condition (mean ± SEM; n = 3; unpaired t-test; **p < 0.01).(C) Representative western blot and densitometric analysis of phosphorylated GR (P-GR) and total GR levels in lysates from WT and R451C NLGN3 expressing cells untreated or treated with DEX (2 μM for 48 h), RU486 (20 μM for 48 h) or both compounds (mean ± SEM; n = 3; unpaired t-test *p < 0.05, **p < 0.01, ***p < 0.001).(D) Flow cytometry analysis of surface NLGN3 in HEK-293 expressing either WT or R451C NLGN3, untreated or treated with DEX (2 μM for 48 h) or with a combination of DEX and RU486 (RU486 20 μM for 48 h) as indicated.Cell adhesion presynaptic and postsynaptic proteins, including the NRXNs and the NLGNs, have a role in the maturation, function, and specification of the synapse.8Alterations in protein processing and cell surface trafficking of the NLGNs have been described for several ASDs-associated mutations, mainly deriving from amino acid substitutions causing misfolding of the extracellular domain in the F I G U R E 4 Dose-response and time-course analysis of dexamethasone (DEX) effects on NLGN3 protein levels and glucocorticoid receptor (GR) activation.(A) HEK-293 stably expressing full-length NLGN3, either wild type (WT) or R451C, were untreated (0) or treated with DEX at different concentrations (0.02, 0.2, or 2 μM) for 48 h and cell lysates were analyzed by western blot for NLGN3 normalized to actin (B), western blot analysis for phosphorylated GR (P-GR) and total GR levels, followed by densitometric analysis.HEK-293 stably expressing full-length NLGN3, either WT or R451C, were treated with DEX 2 μM for different hours (0, 1, 8, 16, and 48) and cell lysates analyzed by western blot for NLGN3 (C, normalized to 0 h; D) western blot analysis for phosphorylated GR (P-GR) and total GR levels, normalized to 0 h, followed by densitometric analysis.Results are expressed as the mean ± SEM; n = 3; unpaired t-test *p < 0.05, **p < 0.01.

F I G U R E 5
Dexamethasone (DEX) treatment stabilizes the intracellular R451C NLGN3 protein.(A) HEK-293 cells expressing R451C NLGN3 were treated with cycloheximide (CHX, 70 μM) in the presence or absence of DEX (2 μM) for the times indicated in the figure.The figure shows a representative western blot analysis of the cell lysates and the quantification of NLGN3 R451C bands by densitometry, normalized to actin and the untreated condition (mean ± SEM; n = 3; unpaired t-test *p < 0.05).(B) Band intensities of the three experiments in (A) were quantified and normalized to the corresponding untreated condition to obtain the curves representing the degradation rates in the presence or absence of DEX; from these curves, the half-life of R451C NLGN3 was calculated as 23.4 h without DEX and 55 h with DEX.F I G U R E 6 Dexamethasone (DEX) treatment reduces the activation of the three branches of the unfolded protein response (UPR).(A) Representative western blot and densitometric analysis of phosphorylated eIF2α (P-eIF2α) normalized to total eIF2α levels in the lysates of wild type (WT) or R451C NLGN3 expressing cells, either untreated or treated with DEX (2 μM for 48 h; mean ± SEM; n = 3; unpaired t-test *p < 0.05).(B) Representative western blot and densitometric analysis of ATF6 protein levels in the lysates of WT or R451C NLGN3 expressing cells, either untreated or treated with DEX (2 μM for 48 h), normalized to actin (mean ± SEM; n = 3; unpaired t-test *p < 0.05).(C) Detection of spliced XBP1 by using a fluorescence reporter (pCAX-F-XBP1ΔDBD-Venus, green) transiently expressed in WT or R451C NLGN3 cells, either untreated or treated with DEX (2 μM for 48 h).DAPI (blue) was used to stain the nuclei.The histograms show the ratio between green fluorescent cells (spliced XBP1) and total cells (DAPI) normalized to the untreated condition (mean ± SEM; n = 4; unpaired t-test *p < 0.05).

F I G U R E 7
Dexamethasone (DEX) treatment enhances the number of synapses in cocultures of hippocampal neurons and HEK-293 cells expressing wild type (WT) or R451C NLGN3.(A) Representative confocal immunofluorescence images of NLGN3-FLAG (red), MAP2 (cyan), and VGLUT1 (green) staining of cocultured hippocampal neurons and HEK-293 cells expressing WT or R451C NLGN3, untreated or treated with DEX (2 μM for 48 h; scale bar: 20 μm, DAPI for nuclei visualization in blue).The inserts are zoomed images showing the neuronal presynaptic protein VGLUT1 on the surface of HEK-293 cells expressing WT or R451C NLGN3 (scale bar: 5 μm).The images show a maximum intensity projection of a region of interest.The histograms show the quantification of NLGN3 and VGLUT1 colocalization, expressed as the percentage of VGLUT1 puncta in proximity to the HEK-293 area.Values are expressed as mean ± SEM (n = 11 fields from three independent experiments; Mann Whitney ****p < 0.0001).(B) The graph illustrates the cumulative distribution of colocalized area for untreated WT (blue) and R451C NLGN3 (yellow) cocultures.Statistical significance was assessed by the Kolmogorov Smirnov test (***p < 0.001).F I G U R E 8 Legend on next page.
Treatment of NSPCs cultures expressing endogenous R451C NLGN3, with DEX, improved mutant protein levels, both in proliferative and differentiated conditions, confirming the results obtained in the HEK-293 cells.Moreover, the drug decreased the ER stress by reducing the levels of BiP and therefore repristinating the homeostasis in the ER.Although UPR mediators are reduced in both cell model systems, the decrease of the UPR-target gene BiP was only observed in the lysates from NSPCs cultures but not in the lysates of HEK-293 over-expressing the mutant protein.

NLGN3 and allows us 4 . 2 |
to study the molecular mechanisms linked to a monogenic form of ASDs in physiological conditions since the F I G U R E 8 Dexamethasone (DEX) treatment increases R451C NLGN3 endogenous protein levels and reduces UPR activation in proliferating and differentiated neural stem progenitor cells (NSPCs).(A,B) Representative western blot and densitometric analysis of NLGN3 levels (normalized to actin) in proliferating (A) or differentiated for 7 days (B, Neu) NSPCs expressing endogenous wild type (WT) or R451C NLGN3, treated or not with DEX (2 μM for 48 h; mean ± SEM; n = 3; unpaired t-test *p < 0.05, **p < 0.01).(C,D) Densitometric analysis and representative western blot of BiP protein levels (normalized to actin) in proliferating (C) or differentiated for 7 days NSPCs expressing endogenous WT or R451C NLGN3 treated or not with DEX (2 μM for 48 h; mean ± SEM; n = 3; unpaired t-test *p < 0.05).F I G U R E 9 Legend on next page.expression of the mutant protein is endogenous.Moreover, our results indicate that the treatment with DEX is a valuable approach to increase the stability of R451C NLGN3 and promote its trafficking to the cell membrane, where it can play its role in inducing the formation of synaptic contacts.This work is part of a new and intriguing field of research, which aims to find possible strategies to improve altered protein trafficking for diseases correlated with protein misfolding, most of which have been untreatable so far.The use of FDA-approved GCs could enhance the secretion of proteins and alleviate the UPR in different cell systems, as we show here for a mutant variant of NLGN3 both in overexpression and in endogenous conditions.Moreover, our work shows and confirms that the use of GCs may be beneficial to many forms of human disease involving protein misfolding and ER stress.33In summary, this study shows a potential strategy to reduce the accumulation of misfolded proteins in the ER, enhancing surface trafficking, secretion, and synapse formation.If these effects translate in vivo, these drugs may have positive effects in contrasting the pathological processes of neurodevelopmental disorders caused by genetic mutations related to ASDs. 4 | MATERIALS AND METHODS 4.1 | Transient transfection HEK-293 cells were transfected using poly-ethylenimine (PEI; Sigma-Aldrich-Merck, Milano, Italy).Cells were seeded in a 6-well plate and transfected, the day after the seeding, with the pcDNA3.1 plasmid encoding full-length NLGN3 WT, NLGN3-R451C, or NLGN3-G221R.Alternatively, HEK-293 cells were transfected using a pTP6-NS plasmid encoding for ΔNS.Stable transfection and clonal selection HEK-293 cells were stably transfected using the pcDNA3.1 plasmid encoding full-length NLGN3, either WT or R451C, or the pcDNA3.1 plasmid encoding the extracellular domain of NLGN3-Venus either

4. 3 | 4 . 4 | 4 . 5 |
Cell treatments AD, D, DEX, or PSP were added in the cell culture medium 24 h after seeding.All tested compounds were dissolved in 100% dimethyl sulfoxide (DMSO) at the concentration of 3.84 mM, except for DEX which was prepared at a concentration of 50 μM and added in the cell culture medium at the final concentration of 2 μM for 48 h.In the dose/response experiments the final concentrations tested were 0.02, 0.2, and 2 μM.CHX (at concentration 70 μM) was added to the cell culture medium for 24, 48, or 72 h.The GR antagonist mifepristone (RU486, at concentration 20 μM) was added to the cell culture medium for 48 h.The proteasome inhibitor MG132 (at concentration 2 μM) was added to the cell culture medium for 16 h (Z-Leu-Leu-Leual, no.C2211 Sigma-Aldrich).Screening of the FDA-approved compounds For the screening of FDA-approved compounds, we used stable HEK-293 cell lines expressing a truncated and fluorescent form of NLGN3 (NLGN3-Venus), either WT or R451C.Cells were plated at a density of 75000 cells in 96-well plate, in 150 μL Optimem medium added with 2% FBS.The compounds of the library were dissolved in 10% DMSO (Sigma-Aldrich) in H 2 O and used at the final concentration of 2 μM.Compounds were added in the cell culture medium 24 h after seeding and fluorescence levels were measured after 60 h of treatment.Quantification of protein secretion by fluorimetric measurement NLGN3-Venus protein trafficking was evaluated by transferring 100 μL of medium from cells either untreated or treated with selected compounds in a black 96-well plate and fluorescence was measured with the Promega Glomax fluorometer (excitation wavelength 485 nm, emission wavelength 535 nm).

5 0 -
GGTCAGAGCTGTCATTGTTCAC-3 0 ).Amplified DNA shows a 40 bp size shift for the R451C-NLGN3 genomic DNA in comparison with WT.Mouse sacrifice was done according to the animal protocol and to the current Italian law (D.lgs.26/2014).All experimental procedures and protocols were approved by the Ethical Committee for Animal Research of the Italian Ministry of Public Health (Authorization N. 541/2016-PR; 351/2021-PR).

4. 8 |
Immunocytochemistry of NSPCs and differentiated culturesCells were fixed with 2% PFA in phosphate buffered saline (PBS) for 5 min and then with 4% PFA for 15 min.After fixation, cultures were washed with potassium-PBS (KPBS), pH 7.4, and then incubated in KPBS containing 5% of the appropriate normal serum and 0.025% Triton X-100 (preincubation solution) for 1 h at room temperature.Subsequently, cultures were incubated at 4 C overnight in a preincubation solution containing the following primary antibodies: anti-Nestin (Millipore, MAB5326 1:500), anti-MAP2 (chicken, 1:200, Abcam ab5392), anti-Synapsin 1 (Cell Signaling, D12G5, 1:200).Secondary antibodies were incubated for 1 h at room temperature in Cy3-conjugated or Alexa 488-conjugated (Jackson Immunoresearch) secondary antibodies at dilution 1:200 in PBS and cover-slipped with DAKO fluorescent mounting medium (Dako S3023).Images were obtained by fluorescence microscopy (Nikon Eclipse TE300).For MAP2 and Synapsin 1 analysis, fluorescence images of 2048 Â 2048 pixels (6.5 μm/pixel) were acquired with oil objective 60Â/NA 1.42 (Olympus) on an Olympus iX73 microscope equipped with an X-Light V3 spinning disc head (CrestOptics), an LDI laser illuminator (89 North), a PRIME cMOS camera and a MetaMorph software (Molecular Devices).Stack images (z-step of 0.3 μm) were analyzed with the ImageJ software, flattened in a maximum intensity Z-projection of 13 slices.MAP2 staining was used to identify segments of 50 μm.Synapsin 1 puncta was counted within each segment.Results were exported in an Excel file and data was plotted in GraphPad Prism version 8.0 for Mac (GraphPad Software, San Diego, California, USA; www.graphpad.com).The analysis was performed on 10 MAP2-positive segments from two independent batches of cultures.

4 C.
Protein concentration was determined by the Bradford assay.4.10 | Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and western blotFifty μg of proteins or equal amounts of cell culture medium from NLGN3 (WT or R451C) expressing cells were subjected to SDS-PAGE