Synapses are specialized adhesive junctions that allow neurons to transmit and process information by connecting to various neural circuits in the brain. The number of synapses in the brain changes throughout life via a dynamic process of synapse build-up and removal, which is essential for its normal development, function, and adaptation to changes in the environment (1, 2). The number of synapses decreases as people age and this decline is thought to contribute to age-related cognitive decline and the potential risk of neurodegenerative disorders such as Alzheimer's disease (AD) (3). Therefore, understanding the regulation of synapse build-and-removal dynamics may guide the development of effective therapeutic strategies for treating neurological and neurodegenerative diseases (4).
Synaptic adhesion molecules participate in regulating synapse numbers via molecular processes, playing a critical role in organizing synapses, preserving their functional integrity, and facilitating changes to pre-existing synapses (5–7). Leucine-rich repeat-containing 4s (LRRC4s, also known as Netrin-G ligands, NGLs), are a family of three postsynaptic adhesion molecules: LRRC4, LRRC4B, and LRRC4C(8, 9). The LRRC4 proteins share similar structures, including leucine-rich repeat (LRR) domains flanking the LRR N-terminus and C-terminus, an immunoglobulin-like C2-type (IgC) domain, a transmembrane domain, and a PDZ binding (PB) domain. LRRC4B can form excitatory synapses by transsynaptic interaction with presynaptic protein tyrosine phosphatase receptors (PTPR-F, -D, and -S), also known as LAR-RPTPs (10–12). It also, via its PB domain, binds to the postsynaptic scaffolding protein postsynaptic density-95 (PSD95) and promotes clustering of NMDA and AMPAR receptors at the postsynaptic site (13). Thus, LRRC4B plays a crucial role in various stages of excitatory synapse formation and function, such as inducing the initial transsynaptic connection between axons and dendrites, clustering postsynaptic proteins, and controlling synaptic plasticity (13). Lrrc4b knockout (KO) mice exhibit major impairments in brain development, synaptic transmission, and plasticity, which are accompanied by behavioral hyperactivity and reduced cognitive function (14). However, the exact mechanism by which LRRC4B regulates synapse dynamics is yet to be understood. Additionally, it is possible that LRRC4B interacts with other unknown proteins that regulate dynamic synapse formation and disassembly, apart from the known LRRC4B-interacting proteins.
The family with sequence similarity 19 member A (FAM19A, also known as TAFA) family consists of five paralogous members that encode small, secreted proteins (15). These proteins are predominantly expressed in the central nervous system (15). The amino acid (aa) sequence of the secreted FAM19A5 is highly conserved across vertebrate species, suggesting a physiologically relevant function in vertebrates (15, 16). FAM19A5 transcripts are abundant in the brain but less so in peripheral tissues (17). Advances in single-cell RNA-seq (scRNA-seq) analysis of mouse brain tissue have provided more detailed expression pattern, showing that Fam19a5 transcripts are mainly found in excitatory and inhibitory neurons in various regions including layers 2/3 and 5 of the cortex and in the CA1 and CA3 regions of the hippocampus in mice and humans (18, 19).
A recent study showed that FAM19As, other than FAM19A5, interact with neurexins (19), which play a role in synaptic adhesion. This suggests that FAM19A5, similar to its paralogous proteins, may act as a ligand for other synaptic adhesion molecules and be involved in synapse formation and function, thereby impacting neural function. This is supported by findings from Fam19a5 KO mice, which exhibited hyperactivity, depressive-like behaviors, and reduced spatial learning and memory, as well as a decrease in dendritic spine density, glutamate signaling, and neuronal activity (20). Genome-wide association studies have revealed the association of FAM19A5 with neurological disorders including AD, attention deficit hyperactivity disorder, and autism (21, 22). Furthermore, a single-nucleus RNA-seq (snRNA-seq) in human AD brain samples showed that FAM19A5 transcript levels in neurons were higher in those with early AD pathology than in those without (23). This result aligns with human serum assay results (24), which showed increased FAM19A5 levels in patients with vascular dementia compared to controls, suggesting a potential contribution of FAM19A5 to the progression of AD pathology. However, the exact role of FAM19A5 function in pathophysiological conditions remains elusive due to limited research on its interacting partner proteins. Previous studies proposed that FAM19A5 may interact with N-formyl peptide receptor 2 or sphingosine 1-phosphate receptor 2 (25, 26), but our recent study refutes these claims (27).
Here, we found that LRRC4B interacts with FAM19A5 with high affinity. This binding suppresses the interaction of LRRC4B with the presynaptic adhesion molecule PTPRF, implicating the role of FAM19A5 in synapse assembly. FAM19A5 binds to a core domain within LRRC4B, hereafter referred to as FAM19A5 binding (FB) domain. Decoy peptides or proteins containing the FB domain (FB decoys) and an anti-FAM19A5 antibody called NS101 (27) can capture FAM19A5 and disrupt the endogenous FAM19A5-LRRC4B complex. The FB decoys and NS101 induced neurite outgrowth and synapse formation in mouse primary neurons. This led us to propose that FAM19A5 regulates neurite growth and synapse formation as a suppressor, by forming the FAM19A5-LRRC4B complex. Given that the FAM19A5-LRRC4B complex hinders new synapse formation, it may serve as a target for treating synapse loss in AD. In mouse models of AD, systemic treatment with NS101 facilitated transport of FAM19A5 from the brain into the blood, which may reduce the level of the FAM19A5-LRRC4B complex in the brain. This resulted in increased synaptic density, function, and plasticity in the hippocampus, as well as improved cognitive ability in the AD mice. Our findings suggest that blocking the formation of the FAM19A5-LRRC4B complex could be a promising therapeutic strategy for AD.
FAM19A5 binds to the FB domain of LRRC4B.
To uncover potential binding partners for FAM19A5, we searched the mouse brain scRNA-seq database (28) and found synapse adhesion molecules whose transcript levels showed a strong correlation with Fam19a5 transcript levels across 168 different types of neurons (Fig. 1A and fig. S1A). Notably, the expression of Lrrc4b was found to be highly correlated with Fam19a5 transcripts (Fig. 1A), suggesting a possible interaction between FAM19A5 and LRRC4B. To assess this possibility, we performed coimmunoprecipitation (co-IP) assays using various anti-FAM19A5 antibodies that target the N-terminal (N-A5-Ab) and C-terminal (C-A5-Ab) regions of FAM19A5 (29). V5-tagged LRRC4B transfected into HEK293 cells, treated with recombinant FAM19A5 (rcFAM19A5), was coimmunoprecipitated with C-A5-Ab (fig. S1B) and vice versa (fig. S1C). Two FAM19A5 transcripts, isoform 1 and isoform 2, produce two FAM19A5 isoform proteins that differ in their N-terminal sequences (25, 29). To assess the interactions between the FAM19A5 isoforms and LRRC4B, we cotransfected HEK293 cells with cDNA encoding the FLAG-tagged LRRC4B protein and either FAM19A5 isoform 1 or isoform 2. Immunofluorescence revealed that both FAM19A5 isoforms colocalized strongly with LRRC4B (Fig. S1D). Similarly, immunoprecipitates of LRRC4B from cell lysates using an anti-FLAG antibody also included both FAM19A5 isoforms (fig. S1E), and immunoprecipitates using C-A5-Ab showed interactions between LRRC4B and both FAM19A5 isoforms (fig. S1F). However, the anti-FAM19A5 antibody N-A5-Ab only weakly coimmunoprecipitated LRRC4B, as an epitope of FAM19A5 recognized by N-A5-Ab is involved in LRRC4B binding. LRRC4B can be released into the culture medium by cleavage at the juxtamembrane region (Fig. 1B) (30). When FAM19A5 isoform 1 was coexpressed with LRRC4B, there was less FAM19A5 in the culture medium, and more was retained in the cell (Fig. 1B), indicating that LRRC4B traps soluble FAM19A5 in the plasma membrane.
LRRC4B expressed in neurons can be secreted into the cerebrospinal fluid (CSF) of humans where FAM19A5 is also found (31–33). Therefore, the FAM19A5-LRRC4B complex can be present in the CSF, as seen in immunoblots of human CSF showing both FAM19A5 and LRRC4B (Fig. 1C). In addition, rcFAM19A5-treated primary cortical neurons showed FAM19A5 localization to neurites where LRRC4B is highly expressed (Fig. 1D).
To identify FAM19A5-binding sequences in LRRC4B, various FLAG-tagged deletion constructs for LRRC4B were generated (Fig. 1E). HEK293 cells expressing these constructs were treated with rcFAM19A5, and the cell lysates were coimmunoprecipitated with an anti-FLAG antibody. We found that all constructs containing the 484–498 aa sequence of LRRC4B were able to bind to FAM19A5, whereas deletion of the 484–498 aa sequence abolished FAM19A5 binding (Fig. 1E and fig. S2A). Since this 484–498 sequence is highly conserved among LRRC4 family members in mammals (Fig. 1F), we designated this core sequence the FAM19A5 binding (FB) domain. Indeed, FAM19A5 was found to bind LRRC4 and LRRC4C through the FB domain (fig. S2B and Fig. 2C). An enzyme-linked immunosorbent assay (ELISA) and surface plasmon resonance (SPR) revealed that LRRC4B(453–576) conjugated with human Fc [hereafter, LRRC4B(453–576)-hFc] had ~ 100 picomolar affinity by ELISA (Fig. 1G) and ~ 30 picomolar affinity for FAM19A5 by SPR (Fig. 1H). FB peptides inhibited binding between FAM19A5 and LRRC4B(453–576)-hFc/LRRC4B(36–576)-hFc in a competitive manner. However, peptides lacking one or more aa from the FB domain (positions 484–498) gradually lost their ability to inhibit the FAM19A5-LRRC4B interaction (fig. S2D).
Using a combination of AlphaFold2 and RoseTTAFold (34, 35), we modeled the structure of the FAM19A5-LRRC4B complex (Fig. 2, A to C and fig. S3A), showing that the FB sequence from Tyr484 to Glu496 of LRRC4B directly interacted with FAM19A5. In particular, the sequence of FB, which had a disordered structure before binding to FAM19A5 (Fig. 2A), became a β-strand constituting a β-sheet structure together with β-strands formed from FAM19A5 (Fig. 2, B and C). Then, we further identified the residues responsible for the interaction between FAM19A5 and FB via in silico Ala scanning where each non-Ala residue of the complex was replaced with Ala using Schrodinger Bioluminate®. This calculates changes in protein-peptide binding affinity based on a large increase in the binding free energy following the Ala mutation (fig. S3B). For FB, Ala substitution for residues that actively participate in β-sheet formation with FAM19A5 increased the binding free energy, which means weakened binding, except for Thr494 and Leu495 in FB. These two residues are unlikely to contribute significantly to the interaction, consistent with the biochemical results (Fig. 2D). In addition to residues participating in β-sheet formation, residues from Glu496 to Pro499 are likely involved in the interaction with FAM19A5 as substitution of residues to Ala increased the binding free energy (fig. S3B). Full-length LRRC4B(36–716) and LRRC4B(453–576) fragments in which Thr488 and Thr489 were substituted with Ala showed a very weak binding to FAM19A5 by a co-IP assay (fig. S3, C and D) and ELISA (Fig. 2E).
For FAM19A5, the basic residues Arg58, Arg59, Arg125, and Lys127 were found to be involved in the interaction with the FB domain via in silico Ala scanning (fig. S3E). These four residues from FAM19A5 participate in β-sheet formation through hydrogen bond interactions with the b-strand formed in the FB domain (fig. S3A). Furthermore, the anionic carboxylate (RCOO−) of Glu493 of LRRC4B and the cationic ammonium (RNH3+) from Lys127 can form a salt bridge (Fig. 2C). Indeed, a double mutation substituting Ala for Arg125 and Lys127 of FAM19A5 reduced the binding affinity to LRRC4B. Arg58 and Arg59 may also play a role in the interaction with FB, as their Ala substitution reduced affinity to LRRC4B, and quadruple mutations of Arg58, Arg59, Arg125, and Lys127 further reduced the binding affinity to LRRC4B (Fig. 2F). In particular, N-A5-Ab epitope mapping via Ala mutation indicated that Arg52, Pro57, Arg58, and Arg59 of FAM19A5 were found to be epitopes for N-A5-Ab (fig. S3F); thus, N-A5-Ab and LRRC4B partly compete at Arg58 and Arg59 for FAM19A5 binding (fig. S3, A and F).
Because FB-containing peptides/proteins bound to FAM19A5 with high affinity, we determined whether they compete with LRRC4B to dissociate the FAM19A5-LRRC4B complex (Fig. 2G). The cell-free ELISA revealed that WT LRRC4B(453–576)-hFc (Fig. 2H) and FB-containing peptides (Fig. 2I) resulted in a dose-dependent decrease in the binding between FAM19A5 and LRRC4B. However, the mutant (MT) peptide in which Ala was substituted for Thr488 and Thr489 in MT LRRC4B(453–576)-hFc only slightly inhibited binding (Fig. 2H). Treatment with WT LRRC4B(453–576)-hFc for 30 min caused a substantial dissociation of FAM19A5 from full-length LRRC4B in HEK293 cells expressing both FAM19A5 isoform 2 and LRRC4B (Fig. 2J). In contrast, in HEK293 cells treated with MT LRRC4B(453–576)-hFc, FAM19A5 remained largely bound to LRRC4B (Fig. 2J), suggesting that FB-containing peptides and proteins can act as decoys to steer FAM19A5 away from the complex.
FAM19A5 inhibits the interaction between LRRC4B and PTPRF via the FB domain.
We then examined the role of FAM19A5 in regulating LRRC4B function. LRRC4B is known to interact physically with the presynaptic protein PTPRF, facilitating synapse formation between neurons (36). Hence, it is likely that alterations in FAM19A5 levels during synapse formation or elimination can affect the dynamic equilibrium of LRRC4B and PTPRF interaction (Fig. 3A). PTPRF(30-1263)-hFc bound to HIS-TEV-LRRC4B(36–576) in a dose-dependent manner. Preincubation of LRRC4B-coated plates with 1 µM WT rcFAM19A5 significantly reduced the binding, whereas the binding was not inhibited by MT FAM19A5(R58A, R59A, R125A, K127A), which weakly binds to the FB domain (Fig. 2F and Fig. 3B). Increasing doses of WT rcFAM19A5 competitively inhibited the binding, but MT rcFAM19A5 was less effective in doing so (Fig. 3C). We have further investigated whether FAM19A5 can affect synapse formation in primary cultured neurons. Treatment of FAM19A5 twice at 3 and 6 DIV significantly decreased the number of puncta colabeled for the presynaptic marker synaptophysin (SYP) and PSD95 at 7 DIV (Fig. 3D). These results suggest that FAM19A5, through the formation of the FAM19A5-LRRC4B complex, may function as a suppressor of synapse formation.
FB decoys promote neurite growth and synapse formation.
We addressed the effect of the FB decoy in neurons. We first measured the transcript levels of various synapse-related genes, including Fam19a5, Lrrc4b, and Ptprf in cultured neurons by RNA-seq. Fam19a5 transcript levels in mouse hippocampal neurons were already high at day in vitro 1 (DIV 1) compared to those of other Fam19a family members (Fig. 4A). Lrrc4b and Ptprf transcripts were also notably high in hippocampal neurons at DIV 1 compared to other synaptic molecules, and the Lrrc4b transcript level remained high until DIV 16 (Fig. 4A). High expression levels of Fam19a5, Lrrc4b, and Ptprf in early stages of cultured neurons suggest their involvement in neurite outgrowth and synaptogenesis, potentially via formation of the FAM19A5-LRRC4B complex. Therefore, we studied the effects of the FB decoy LRRC4B(453–576)-hFc on neurite and synapse development in cultured neurons. Treatment with WT LRRC4B(453–576)-hFc led to a dose-dependent increase in neurite length and the number of primary neurites and branching points in primary cortical neurons (Fig. 4B). In contrast, treatment with MT LRRC4B(453–576)-hFc had no impact on neurite growth and was comparable to the control group (Fig. 4B).
Neurites grow and differentiate into dendrites or axons which eventually form a synapse (37, 38). Therefore, we determined whether disruption of the FAM19A5-LRRC4B complex promoted synapse formation. The addition of WT LRRC4B(453–576)-hFc, but not MT LRRC4B(453–576)-hFc, increased the intensities of SYP and PSD95 (Fig. 4C). In addition, the number of SYP and PSD95 colocalized puncta, indicative of pre- and postsynaptic connections (39, 40), increased in response to WT LRRC4B(453–576)-hFc (Fig. 4C). Together, these results suggest that FB decoys, by disrupting the FAM19A5-LRRC4B complex, promote the neural development process from neurite growth to the formation of initial contacts between dendrites and axons.
NS101, the anti-FAM19A5 antibody, enhances synapse formation.
Antibody therapeutics have high specificity and affinity for their targets and offer a unique advantage over peptides and Fc-conjugated proteins with their prolonged half-life in the blood (41), thereby providing more opportunities for the drug to reach the brain with a single systemic administration. Therefore, we developed NS101, an optimized anti-FAM19A5 antibody derived from N-A5-Ab (27), NS101 was more potent than WT LRRC4B(453–576)-hFc in blocking FAM19A5-LRRC4B complex formation (Fig. 5A), and its ability to dissociate the FAM19A5-LRRC4B complex was corroborated by a co-IP assay using HEK293 cells expressing FLAG-tagged LRRC4B treated with rcFAM19A5 in the presence of NS101 (fig. S4A). NS101 significantly increased the release of FAM19A5 into the culture medium from HEK293 cells cotransfected with LRRC4B and FAM19A5 isoform 1 or 2 but not from cells transfected only with FAM19A5 in a time- and dose-dependent manner (Fig. 5B and fig. S4B). Considering that FAM19A5 isoform 1 was mostly captured by LRRC4B in the plasma membrane (Fig. 1D), NS101 can release FAM19A5 into the medium through dissociation of the FAM19A5-LRRC4B complex. NS101 separated FAM19A5 isoform 2 from LRRC4B in living cells (fig. S4C). Additionally, NS101 increased the release of endogenous FAM19A5 into the culture medium from primary hippocampal neurons (Fig. 5C).
We evaluated the ability of systemically administered NS101 to capture FAM19A5 proteins in the brain and transport them to the peripheral circulation. We measured the baseline levels of FAM19A5 in the blood and CSF of mice, rats, monkeys, and humans and found that FAM19A5 concentrations were lower in plasma than in CSF in all species (Fig. 5D). This was confirmed using samples from FAM19A5 KO mice (fig. S5, A and B). Following a single intravenous (iv) administration of NS101 in rats, the plasma FAM19A5 levels promptly increased, reaching the highest point at 24–36 hours and declining gradually over the next 28 days (Fig. 5E). An increased dose of NS101 led to higher plasma FAM19A5 levels (fig. S5C). The peak time and overall change in concentrations of both CSF NS101 and plasma FAM19A5 were comparable after iv administration of NS101 (Fig. 5E). This correlation indicates that the increase in plasma FAM19A5 levels was likely due to the action of NS101 in the brain and CSF. To visualize the NS101-FAM19A5 complex, we used disuccinimidyl glutarate to irreversibly crosslink proteins in CSF samples and detected them in immunoblots using N-A5-Ab. The NS101-FAM19A5 complex was found to be greater than 250 kDa in size, while the FAM19A5 protein bands were 12 and 15 kDa (fig. S5D). The formation of the complex in the CSF started immediately after an iv administration of NS101, peaked at 24 hours, and gradually decreased until 672 hours (Fig. 5F). This shows a correlation between the formation of the complex and the residual concentration of NS101 in the CSF. We then further traced FAM19A5 migration in vivo using a version of FAM19A5 in which Ala was substituted for Arg58 and Arg59 [FAM19A5(R58A, R59A)], which binds poorly to NS101 but strongly to LRRC4B (fig. S5E). WT and MT FAM19A5 were intracranially injected into the mouse brain and NS101 was administered iv one day later. Plasma FAM19A5 levels were significantly higher in WT FAM19A5-treated mice than in the sham group (Fig. 5G). However, plasma FAM19A5 levels were not significantly different between the FAM19A5(R58A, R59A)-treated and sham groups (Fig. 5G). Therefore, it is likely that NS101 dissociates FAM19A5 from LRRC4B, forms an NS101-FAM19A5 complex in the brain, and migrates to the CSF, where the complex drains into the peripheral blood.
We then examined the effects of NS101 on neurite outgrowth and synapse formation. NS101 significantly promoted neurite growth in terms of length, primary neurite number, and branching point number of cultured neurons at DIV 3 compared to untreated or control human immunoglobulin G (hIgG)-treated neurons (fig. S6A and B). At DIV 7, the contribution of NS101 to synapse formation was revealed by the high immunoreactivity of SYP and PSD95 and the increased number of SYP/PSD95 colocalized puncta (fig. S6, C and D). To determine whether NS101 is also involved in structural changes at the synapse, and not just a secondary effect of increased neurite growth, we calculated the number of spines per unit length of dendrites (i.e., spine density) using the classic DiI staining method to visualize dendritic spines (42). Our findings showed a significant increase in spine density after NS101 treatment (Fig. 5, H and I), indicating that NS101 contributes to synaptic connectivity by both guiding neurite formation and promoting synapse formation. We then tested whether NS101 could enhance synaptic connectivity under pathological conditions where Aβ1–42 oligomers reduce synaptic connections between neurons in vitro (43). To do this, NS101 was administered to mouse hippocampal neurons treated with Aβ1–42 oligomers at DIV 15 and DIV 18. Immunohistochemistry using SYP and PSD95 antibodies showed that Aβ1–42 significantly reduced the number of synapses, however, the addition of NS101 prevented this reduction, particularly the number of puncta that were colabeled by SYP and PSD95 (Fig. 5, J and K).
NS101 restores synapse number, function, and plasticity in the hippocampus in AD mouse models.
Treatment of hippocampal neurons with rcFAM19A5 reduced neurite growth and synapse formation (Fig. 3C), raising the possibility that FAM19A5 can serve as a risk factor for neurodegenerative diseases. To further explore this idea, we measured FAM19A5 levels in human CSFs and found a high correlation with age (Fig. 6A) and with total tau (t-tau) and phosphorylated tau (p-tau) (Fig. 6B). Given that t-tau and p-tau levels are closely associated with synapse pathology (4, 44), we investigated the effect of NS101 in mouse models of AD. AD is characterized by the loss of synapses outpacing the acquisition of new synapses, which is attributed to external environmental factors and/or internal neuronal factors (4, 44–46). A recent article emphasized the significance of restoring synapse function in the amyloidopathy mouse model, even without removing amyloid β (Aβ). The study showed that using a synthetic synaptic organizer that physically connects presynapses and postsynapses can help restore neural circuits and cognitive function in an AD animal model (47). This study also supports the notion that blocking a synaptic suppressor such as FAM19A5 can help to restore neural circuits in AD patients.
As seen in rat models (Fig. 5E), the systemic iv administration of NS101 to AD mouse models led to the release of FAM19A5 from the brain into the blood (fig. S7), suggesting a reduction in the FAM19A5-LRRC4B complex in the brain. In the AD-relevant APP/PS1 murine model of amyloidopathy, there was a significant reduction in SYP immunoreactivities in both CA1 and CA3 of the hippocampus compared to WT mice (Fig. 6C and fig. S9D) (48, 49). PSD95 immunoreactivity was markedly reduced in CA1 but less so in CA3. The iv administration of NS101 once a week for eight consecutive weeks to ten-month-old APP/PS1 mice resulted in the restoration of SYP and PSD95 immunoreactivity to level comparable to those seen in WT mice (Fig. 6C and fig. S9D). Similarly, NS101 significantly increased the intensity of synaptic vesicle glycoprotein 2A (SV2A), an essential synaptic vesicular protein and a biomarker of synaptic density (50), in the CA1 region of APP/PS1 (fig. S9C). Collectively, these results suggest that NS101 enhances synaptic density by promoting structural synaptic connections.
We then determined whether NS101-induced structural recovery is accompanied by improved basal synaptic transmission and function of neuronal circuits in hippocampal slices of APP/PS1 mice. NS101-treated APP/PS1 mice showed an increased frequency of miniature excitatory postsynaptic currents (mEPSCs) in CA1 pyramidal neurons compared to the hIgG-treated group (Fig. 6E). Furthermore, NS101 restored the mean frequency to a level comparable to that of normal mice, demonstrating enhanced excitatory synaptic transmission. The amplitude of mEPSCs remained unchanged in all groups, suggesting that NS101 did not impact excitatory postsynaptic receptor responsiveness (Fig. 6E). Field extracellular recordings verified that the excitatory postsynaptic potentials (fEPSPs) of Schaffer collateral (SC)-CA1 circuits, which play a major role in memory formation and emotional networks, were reduced in hippocampal slices from hIgG-treated APP/PS1 mice (Fig. 6F), as previously reported (51). NS101 significantly restored the diminished strength of SC-CA1 circuits (Fig. 6F). These findings align with NS101's impact on synaptic connections and indicate that it may normalize the reduced excitatory synaptic transmission of CA1 pyramidal cells to levels found in WT mice.
Long-term potentiation (LTP) is a sustained enhancement of synaptic strength after repetitive or strong stimulation (52, 53). LTP is linked to the creation of new dendritic spines and associated synapses in the hippocampus (54, 55). Thus, the LTP impairment observed in various AD mouse models has been suggested as a potential cause of cognitive deficits (56). To measure LTP at SC-CA1 synapses, we triggered SC fibers with the theta burst stimulation and monitored the changes in fEPSP responses. NS101 restored the reduced amplitude of fEPSPs (during the last 5 min) in APP/PS1 slices, similar to WT (Fig. 6G). In addition, post-tetanic potentiation (PTP), a form of presynaptic plasticity and an indicator of the probability of neurotransmitter release (57), was unchanged in the test groups (Fig. 6H), suggesting no alteration in presynaptic release probability (58). Thus, NS101 restored LTP by stabilizing and fortifying synaptic connections, potentially aiding in the recovery of cognitive deficits in AD mice.
NS101 ameliorates cognitive deficits in mouse models of AD.
Since NS101 has the ability to reinstate synapse formation under pathological conditions (Fig. 5, J and K), the new synapses formed in the hippocampus may help to recover cognitive abilities. APP/PS1 mice treated with iv hIgG once a week for 8 weeks showed a significant short-term memory decline compared to WT mice, as measured by the reduced alteration index in the Y-maze test. In contrast, NS101-treated APP/PS1 mice showed increased spontaneous alteration compared to hIgG-treated mice (Fig. 7A), indicating an improvement in short-term memory. Spatial learning and memory were evaluated using the Morris water maze (MWM). APP/PS1 mice treated with hIgG learned poorly in the training phase (fig. S8, A and B), with memory impairment in the probe test (fig. S8, C and D), compared to WT mice. Despite the positive effect of NS101 on short-term memory in the Y-maze (Fig. 7A), NS101 partially improved the degraded spatial learning and memory of APP/PS1 mice in the MWM test, as indicated by the parameters of spatial memory (fig. S8). APP/PS1 mice underwent a passive avoidance test to assess associative learning and memory after receiving a foot shock during the training period. hIgG-treated mice showed a significant reduction in latency compared to WT mice, suggesting a problem in associative learning and memory. NS101-treated APP/PS1 mice displayed a tendency to recover latency compared to hIgG-treated mice (Fig. 7B). Taken together, these results demonstrated that NS101 improved short-term memory and had partial effects on associative learning and memory in APP/PS1 mice. We administered iv NS101 to P301S mice, a transgenic model of tauopathy, to determine whether it restored cognitive function. NS101 significantly improved spatial learning and memory without affecting locomotion, as shown by the total distance mice moved in the MWM (Fig. 7, C to E). NS101 also increased the immunoreactivity of the synapse molecules SYP and PSD95 in the hippocampus compared to hIgG-treated control in both P301S and APP/PS1 mice (fig. S9, A and B). These results suggest that NS101 may ameliorate cognitive decline in AD pathology, including amyloidopathy and tauopathy.