Abstract
Excitatory synapses and the actin-rich dendritic spines on which they reside are indispensable for information processing and storage in the brain. In the adult hippocampus, excitatory synapses must balance plasticity and stability to support learning and memory. However, the mechanisms governing this balance remain poorly understood. Tiam1 is an actin cytoskeleton regulator prominently expressed in the dentate gyrus (DG) throughout life. Previously, we showed that Tiam1 promotes dentate granule cell synapse and spine stabilization during development, but its role in the adult hippocampus remains unclear. Here, we deleted Tiam1 from adult forebrain excitatory neurons (Tiam1fKO) and assessed the effects on hippocampal-dependent behaviors. Adult male and female Tiam1fKO mice displayed enhanced contextual fear memory, fear extinction, and spatial discrimination. Investigation into underlying mechanisms revealed that dentate granule cells from Tiam1fKO brain slices exhibited augmented synaptic plasticity and N-methyl-D-aspartate–type glutamate receptor (NMDAR) function. Additionally, Tiam1 loss in primary hippocampal neurons blocked agonist-induced NMDAR internalization, reduced filamentous actin levels, and promoted activity-dependent spine remodeling. Notably, strong NMDAR activation in wild-type hippocampal neurons triggered Tiam1 loss from spines. Our results suggest that Tiam1 normally constrains hippocampal-dependent learning and memory in the adult brain by restricting NMDAR-mediated synaptic plasticity in the DG. We propose that Tiam1 achieves this by limiting NMDAR availability at synaptic membranes and stabilizing spine actin cytoskeleton and that these constraints can be alleviated by activity-dependent degradation of Tiam1. These findings reveal a previously unknown mechanism restricting hippocampal synaptic plasticity and highlight Tiam1 as a therapeutic target for enhancing cognitive function.
Significance Statement
The precise and dynamic regulation of excitatory synapses within hippocampal circuits is indispensable for learning and memory. These specialized connections must remain malleable to support the acquisition of relevant new information but stable enough to protect the loss of previously stored memories. Dysregulation of this intricate balance drives cognitive decline following central nervous system injury and disease. Here, we establish the Rac1 guanine nucleotide exchange factor Tiam1 as an essential regulator of this balance that functions in the adult hippocampus to limit learning and memory. Our findings identify a previously unknown mechanism for restricting hippocampal synaptic plasticity that represents a potential therapeutic target for improving cognitive function.
Introduction
Our ability to learn and form memories relies on precise and dynamic regulation of excitatory synapses (Bavelier et al., 2010; Kennedy, 2013). These specialized connections must maintain a balance between plasticity and stability that enables both the acquisition of new information and the protection of previously stored knowledge (Abraham and Robins, 2005; Bavelier et al., 2010). Dysregulation of these processes impairs cognitive function (Y. S. Lee and Silva, 2009; Fan et al., 2017; Forrest et al., 2018). A brain region strikingly vulnerable to synapse dysregulation is the hippocampus, which is critical for episodic learning and memory (Leal and Yassa, 2015; Fan et al., 2017). Numerous studies have demonstrated a connection between the gradual impairment of hippocampal synaptic plasticity and the cognitive decline accompanying aging and neurodegenerative diseases (Foster, 1999; Burke and Barnes, 2006; Cheung and Ip, 2011; Koffie et al., 2011; Leal and Yassa, 2015; Jang and Chung, 2016; Jackson et al., 2019). Thus, it is imperative to understand how hippocampal synaptic plasticity is maintained normally to develop therapies to treat cognitive decline.
The neurotransmitter glutamate mediates excitatory synaptic transmission primarily on actin-rich dendritic spines (Kennedy, 2013). Spines are vital postsynaptic signaling compartments whose functional properties strongly correlate with their morphologies (Yuste and Denk, 1995; Kennedy, 2013; Sala and Segal, 2014). These specialized protrusions undergo rapid activity-dependent structural and functional remodeling involving their growth or shrinkage and the insertion or removal of synaptic glutamate receptors [α-amino-3-hydroxy-5-methyl-4-ioxazolepropionic acid- (AMPARs) and N-methyl-D-aspartate–type receptors (NMDARs)]. Alterations in the underlying cytoskeleton drive these changes (Carlisle and Kennedy, 2005; Hlushchenko et al., 2016; Lei et al., 2016; Chidambaram et al., 2019). Rho-family small GTPases (e.g., Rac1, RhoA) and their regulators, guanine nucleotide exchange factors (GEFs, activators) and GTPase-activating proteins (inhibitors), play fundamental roles in mediating this synaptic plasticity by regulating actin cytoskeleton remodeling and influencing receptor surface levels at synaptic sites (Tolias et al., 2011; Duman et al., 2015, 2022).
Tiam1 is a Rac1-specific GEF enriched throughout life in the dentate gyrus (DG; Cheng et al., 2021), a subregion of the hippocampus crucial for learning, memory, spatial coding, and pattern separation (Leal and Yassa, 2015; McAvoy et al., 2016). We and others have established Tiam1 as a major regulator of dendritic spine and excitatory synapse development in hippocampal neurons (Tolias et al., 2005, 2007; H. Zhang and Macara, 2006; Lai et al., 2012; Duman et al., 2013; Um et al., 2014). More recently, Tiam1 was demonstrated to specifically promote the development of DG granule cell glutamatergic synapses ex vivo in hippocampal slices (Rao et al., 2019) and in vivo using Tiam1 global knock-out (Tiam1KO) mice developed by our lab (Cheng et al., 2021). However, despite Tiam1’s continued robust expression in the adult hippocampus, its role in the mature brain remains unclear. To address this gap in knowledge, we deleted Tiam1 from adult mouse forebrain excitatory neurons and assessed changes in hippocampal-dependent behaviors and DG granule cell plasticity. We find that Tiam1 limits hippocampal-dependent learning and memory and restricts synaptic plasticity and NMDAR function in the adult DG. Based on our results from primary hippocampal neurons, we propose that these effects may be due to activity-dependent Tiam1–mediated NMDAR trafficking and filamentous actin (F-actin) assembly in spines. Moreover, we suggest that these constraints on plasticity may be fine-tuned by targeting Tiam1 loss from spines.
Materials and Methods
Animals
Deletion of Tiam1 from postnatal forebrain excitatory neurons was achieved as previously described (Ru et al., 2022). Briefly, our Tiam1 floxed mice (Tiam1fl/fl; Cheng et al., 2021) were crossed with the CaMKIIα-Cre animals to generate Tiam1fl/+;CaMKIIα-Cre mice. The Tiam1fl/+;CaMKIIα-Cre mice were then crossed with Tiam1fl/fl mice to obtain Tiam1fl/fl;CaMKIIα-Cre (Tiam1fKO) and Tiam1fl/fl (Control or Con) littermates. Male and female 3.5 to 6 month adult mice were used for experiments unless otherwise indicated. Genotyping of mice was conducted by PCR from tail DNA. For Tiam1fl/fl mice, the following primers were used: P1, ACGTGTGTTAATTAGCCAGGTTTGATGG; P2, GATCCACTAGTTCTAGAGCGGCCGAA; and P3, CTACCCGGAGGAAGTGGAAGCACTACT. For CaMKIIα-Cre mice, the following primers were used: forward, GCATTACCGGTCGATGCAACGAGTGATGAG, and reverse, GAGTGAACGAACCTGGTCGAAATCAGTGCG. The Thy1-GPF line M (Jax stock #007788) was obtained from Jackson Laboratory (Feng et al., 2000). Mice were group housed under a standard 12 h light/dark cycle. Neuronal cultures from rats were prepared from Long–Evans Embryonic Day 18 (E18) timed-pregnant rats purchased from Envigo or Charles River Laboratories. All animals were maintained in the animal facilities at Baylor College of Medicine (BCM). All procedures involving experimental animals were conducted in strict accordance with the National Institutes of Health (NIH) guidelines and were approved by the Animal Care and Use Committee of BCM in accordance with applicable legislation. Every effort was made to minimize animal suffering.
DNA constructs
The following constructs have been described previously: pSUPER (Duman et al., 2013), pSUPER-Tiam1 RNAi (Tolias et al., 2005, 2007; Duman et al., 2013), pCI-SEP-GluN2B (Kopec et al., 2006), pCI-SEP-GluR1 (Kopec et al., 2006), pCMV-Flag-Tiam1(Tolias et al., 2005), pCMV-eGFP (Duman et al., 2019), and pCAG-ERT2-Cre-ERT2 (Matsuda and Cepko, 2007). pEF1α-tdTomato was purchased from ClonTech (631975). pAAV-EF1α-Cre-P2A-tdTomato-WPRE-polyA was obtained from A. Tolias (BCM). All constructs were verified by DNA sequencing.
Antibodies and reagents
The following antibodies were purchased and used according to vendor specifications: anti-Tiam1 (sc-872, Santa Cruz Biotechnology); anti-GAPDH (sc-32233, sc-25778, Santa Cruz Biotechnology); anti-BrdU (5-bromo-2′-deoxyuridine; OBT0030G, Accurate Chemical and Scientific); anti-doublecortin (DCX; ab18723, Abcam); anti-NeuN (neuronal nuclear protein; MAB377, Merck Millipore); anti-flag (14793, Cell Signaling Technology); anti-PAK1 (2602, Cell Signaling Technology); anti-CaMKII (50049, Cell Signaling Technology); anti-actin (MAB1501, Sigma-Aldrich); and anti-GFP (GFP-1020, Antibodies). The anti-pPAK1 was generated using the PAKα peptide PEHTKS(p)VYTRS(p)VIEP with phosphates added to serine residues 198 and 203 as previously done (Shamah et al., 2001). The anti-pPAK1 antibody was affinity purified using the antigen purified as a thrombin-cleaved GST fusion coupled to UltraLink Biosupport (Thermo Fisher Scientific). We used goat anti-rabbit or anti-mouse HRP-conjugated secondary antibodies (Merck Millipore) for Western blotting and Cy5- or Cy3-conjugated secondary antibodies (Jackson ImmunoResearch Laboratories) for immunostaining. The following reagents were purchased: (Z)-4-hydroxytamoxifen (4OHT; H7904, Sigma-Aldrich), L-glutamic acid (0218, Tocris Bioscience), picrotoxin (PTX; 1128, Tocris Bioscience), forskolin (1099, Tocris Bioscience), rolipram (0905, Tocris Bioscience), NMDA (0114, Tocris Bioscience), CNQX (C127, Merck Millipore), D-AP5 (0106, Tocris Bioscience), NH4Cl (A9434, Sigma-Aldrich), and Texas Red-phalloidin (T7471, Invitrogen).
Mouse behavioral tests
For behavioral experiments, female and male control and Tiam1fKO mice were used at 3.5 to 6 months-old. All behavioral tests were performed and analyzed with the experimenter blinded to the genotype.
Fear conditioning and extinction
Classical fear conditioning was conducted as previously (Cheng et al., 2021). For training, mice were placed in a training chamber (Context A) and allowed to explore ad libitum for 2 min, before being subjected to two pairings of a 30 s tone (85 dB, spaced by 2 min) each ending with electrical footshock (0.7 mA, 2 s). Contextual fear memory was tested 24 h after training by returning mice to Context A for 5 min and monitoring their freezing behavior with no shock or tone presented. Cued fear memory was tested 2 h later by placing mice a different chamber (Context B) with a novel environment (dim light, vanilla odor, different floor). After a 3 min exploratory period, mice were subjected to the tone for 3 min, and their freezing behavior was recorded and analyzed. For fear memory extinction, 24 h after fear memory test, mice were placed in Context A for 5 min without any tone or footshock. Extinction training continued for 8 d. To assess fear memory maintenance, we trained the mice in classical fear conditioning as described above, and the memory was tested for the first time at 48 h post-training. For fear conditioning and extinction, freezing behavior was recorded and analyzed using the FreezeFrame software (Actimetrics).
Spatial discrimination
This test was performed as described previously (Fu et al., 2019) to assess the ability of mice to discriminate between very similar spatial positions of two objects in an arena, an ability that requires the DG (Clelland et al., 2009; Inokuchi, 2011). Briefly, external spatial cues were present on a three-walled enclosure surrounding the arena. During the training phase, mice were introduced into an arena with two identical objects (25 ml Erlenmeyer flasks) in set positions and allowed to explore for three 3 min trials separated by 3 min intertrial intervals. In the test phase, which occurred 3 min after the last training trial, mice were reintroduced into the arena where one object remained stationary, designated as stationary object (S.O.) and the other object had been displaced by either 4 cm (Position 1) or 12 cm (Position 2), designated as displaced object (D.O.). The amount of time spent exploring each object was recorded using the ANY-maze software during the test trial. The following formula was used to calculate the percentage of time spent with the object: [(time spent with S.O. or D.O.) / (total time spent in exploring both objects)] × 100. A greater percentage of time spent exploring the D.O. compared with that exploring the S.O. indicates successful spatial discrimination.
Accelerating rotarod test
Mice were subjected to a horizontally oriented accelerating rotarod. Animals were tested four trials per day for 2 d, with an interval of 30 min between each trial (Mulherkar et al., 2017). During each trial, the rotation speed of the rotarod increased from 4 to 40 rpm and then maintained 40 rpm for a total of 5 min. The latency of mice to fall was used to assess their motor learning, coordination, and balance.
Western blot analysis
Immunoblotting was performed on protein samples collected from mouse hippocampi or days in vitro (DIV) 21 primary hippocampal neurons. The hippocampi of mice were dissected and homogenized in RIPA lysis buffer [50 mM Tris, 150 mM NaCl, 1% NP-40, 0.5% deoxycholic acid, 0.1% SDS, 1 mM EDTA, 1 mM DTT (GenDEPOT), 10 mM β-glycerol phosphate), pH 8.0, with 1 mM DTT, cOmplete Protease Inhibitor Cocktail (Roche), and XpertPhosphatase Inhibitor Cocktail (GenDEPOT). DIV 21 hippocampal neurons were lysed in NP40 lysis buffer (50 mM Tris, pH 7.5, 150 mM NaCl, 1% NP-40, 1 mM EDTA, pH 8.0, 5% glycerol) with 1 mM DTT, cOmplete protease inhibitor mixture, and XpertPhosphatase Inhibitor Cocktail. Protein concentrations were determined using the Pierce BCA protein assay kit (Thermo Fisher Scientific). Protein lysates were separated on SDS–PAGE gels and transferred to PVDF membranes (Merck Millipore). The membranes were blocked with OneBlock blocking buffer (Genesee Scientific) for 1 h at room temperature, incubated with primary antibody overnight at 4°C, and then incubated with HRP-conjugated secondary antibody for 2 h at room temperature. Western blots were visualized using enhanced chemiluminescence (Amersham) on the Odyssey imaging systems (LI-COR Biosciences) and quantified using the ImageJ software or Image Studio (LI-COR Biosciences). Quantification of the Western blots shows the relative density presented as the ratio of protein over GAPDH, normalized to their respective control.
F- to G-actin ratio
Actin is present in two forms, a filamentous form (F-actin) that is insoluble and a monomeric globular form (G-actin) that is soluble. F-actin to G-actin ratio was determined by Western blot, as previously described (Zeng et al., 2007; Huang et al., 2013). Briefly, the hippocampi of control and Tiam1fKO mice were isolated, homogenized in cold NP40 lysis buffer supplemented with 1 mM DTT, cOmplete protease inhibitor mixture, and XpertPhosphatase Inhibitor Cocktail and centrifuged at 15,000 × g for 30 min. Soluble actin (G-actin) was measured in the supernatant. The insoluble F-actin in the pellet was resuspended in lysis buffer plus an equal volume of buffer 2 (1.5 mM guanidine hydrochloride, 1 mM sodium acetate, 1 mM CaCl2, 1 mM ATP, 20 mM Tris–HCl), pH 7.5, and incubated on ice for 1 h to convert F-actin into soluble G-actin, with gentle mixing every 15 min. The samples were centrifuged at 15,000 × g for 30 min, and F-actin was measured in this supernatant. Samples from the supernatant (G-actin) and pellet (F-actin) fractions were proportionally loaded and analyzed by Western blotting, quantification shown as the relative density of F-/G-actin, relative to control.
Electrophysiology
Brain slices were prepared as previously described (Cheng et al., 2021). All the chemicals used in the electrophysiology experiments were purchased from Sigma-Aldrich, unless otherwise stated. Adult male animals were deeply anesthetized using 3% isoflurane prior to decapitation. The brain was removed and placed into cold (0−4°C) oxygenated N-methyl-D-glucamine (NMDG) solution (in mM: 93 NMDG, 93 HCl, 2.5 KCl, 1.2 NaH2PO4, 30 NaHCO3, 20 HEPES, 25 glucose, 5 sodium ascorbate, 2 thiourea, 3 sodium pyruvate, 10 MgSO4, and 0.5 CaCl2), pH 7.35. Parasagittal brain slices (300 μm thick) were cut with a microslicer and maintained at 37.0 ± 0.5°C in oxygenated NMDG solution for 10 min and then transferred to physiological solution (in mM: 125 NaCl, 2.5 KCl, 1.25 NaH2PO4, 25 NaHCO3, 1 MgCl2, 25 glucose, and 2 CaCl2; ACSF), pH 7.4, for ∼0.5−1 h. Finally, slices were equilibrated at room temperature for at least 30–45 min before being transferred to a submerged recording chamber constantly perfused with ACSF bubbled with 95% O2/5% CO2 at 33.0 ± 0.5°C. For recordings, borosilicate pipettes (5–6 MΩ) filled with intracellular solution were used. The DG was visualized under DIC infrared illumination.
To perform miniature excitatory postsynaptic current (mEPSC) measurements, an intracellular solution (in mM: 120 potassium gluconate, 10 HEPES, 4 KCl, 4 Mg-ATP, 0.3 Na3GTP, 10 sodium phosphocreatine) containing 0.5% biocytin, pH 7.25, was used. To block action potential-mediated neurotransmitter release and GABAA receptors, 0.5 μM tetrodotoxin and 50 μM PTX (Tocris Bioscience) were applied to the bath, respectively. For evoked EPSC measurements, we used an internal solution containing the following (in mM):135 CsCH3SO3, 10 HEPES, 8 NaCl, 0.25 EGTA, 2 MgCl2, 4 Mg-ATP, 0.3 Na3GTP, and 5 phosphocreatine, pH adjusted to 7.3 with NaOH. Neurons were held at −70 mV, and electrical stimuli were delivered to the perforant path (PP) by means of a current stimulus isolator (WPI) connected to a bipolar concentric stimulating electrode (FHC). The measurements were performed using a stimulation that yielded ≃50% of the maximal response. Paired-pulse ratio (PPR) was assessed by delivering pairs of stimuli at 50 ms interstimulus intervals, repeated at 0.05 Hz. To obtain the NMDAR/AMPAR current ratio, stimuli of identical amplitude were delivered at holding potentials of −60 and +40 mV, with a frequency of 0.05 Hz in the presence of 50 μM PTX (Tocris Bioscience), as described previously (Antonelli et al., 2016). Stable synaptic responses were first obtained at −60 mV, and the amplitude of these responses was taken as the AMPAR-specific component. Next, the holding potential was changed to +40 mV, and dual-component EPSCs were collected. At 50 ms poststimulus, the amplitude of these EPSCs was interpreted to be the NMDAR-specific component.
Field excitatory postsynaptic potential (fEPSP) recordings were performed in an interface chamber (Fine Science Tools). Oxygenated ACSF (95% O2/5% CO2) was warmed to 31°C and perfused into the recording chamber at a rate of 1 ml/min. Electrophysiological traces were amplified (Model 1800 amplifier, A-M Systems), digitized and stored (Digidata models 1200 and 1320A with the Clampex software, Molecular Devices). Extracellular stimuli were administered (Model 2200 stimulus isolator, A-M Systems) on the PP fibers using enameled, bipolar platinum-tungsten electrodes. fEPSPs were recorded with an ACSF-filled glass recording electrode (1–3 MΩ). The stimuli were set to an intensity that evoked a fEPSP that had a slope of 30–40% of the maximum fEPSP slope. Long-term potentiation (LTP) was induced by administering a weak, high-frequency stimulation protocol (two burst of 50 pulses, delivered at 100 Hz; 2 × 100 Hz, 30 s intertrain interval) in the presence of PTX (50 μM; Kannangara et al., 2015). Synaptic efficacy was monitored 20 min prior to and 1 h following induction of LTP by recording fEPSPs every 20 s.
In vivo morphological reconstruction
Morphology of neurons was reconstructed from slices after recordings as previously described (Cheng et al., 2021). Slices were fixed in freshly prepared 2.5% glutaraldehyde/4% paraformaldehyde (PFA) in 0.1 M phosphate-buffered saline at 4°C for 7 d. Neuronal morphology was revealed using the avidin–biotin–peroxidase method. Neurons located in the dorsal DG were reconstructed using a 100× oil-immersion objective lens and Camera Lucida System (Neurolucida, MicroBrightField). Dendritic arbor structure and spine density analysis were performed using Neurolucida.
Neuronal culture and transfection
Dissociated hippocampal neuron cultures were prepared from E18 rats as described (Duman et al., 2019). Neurons (3.0 × 105 cells/ml) were plated onto nitric acid-washed glass or cell-culture–treated plastic, coated overnight with 20 μg/ml poly-D-lysine (Corning) and 3 μg/ml laminin (Corning), in Neurobasal medium (Invitrogen) supplemented with B27 (Invitrogen), 2 mM glutamine (Thermo Fisher Scientific), and 100 U/ml penicillin/streptomycin (Thermo Fisher Scientific), and culture medium was changed on DIV 1. For live imaging experiments, rat hippocampal neurons were changed to phenol-red free Neurobasal medium supplemented as above. At DIV 6–7, neurons were transfected using the calcium phosphate method, as previously described (Duman et al., 2013). Mouse hippocampal neurons were prepared from Tiam1f/f P0 or P1 pups, dissociated in the same medium as rat neurons, and plated (5.0 × 105 cells/ml) in Basal Medium Eagle (Invitrogen) supplemented with 45% glucose, 10% heat inactivated bovine calf serum (HyClone), 2 mM glutamine (Thermo Fisher Scientific), and 100 U/ml penicillin/streptomycin (Thermo Fisher Scientific). At 24 h after plating, culture medium was replaced with the same medium as rat neurons, except using Neurobasal A medium (Invitrogen). For live imaging experiments, mouse hippocampal neurons were changed to phenol-red free Neurobasal A medium supplemented as above. Mouse neurons were transfected at DIV 4. To achieve late deletion of Tiam1 in cultures, Tiam1f/f mouse hippocampal neurons transfected with pCAG-ERT2-Cre-ERT2 were treated with 0.1 μM 4OHT at DIV 14. Early deletion was achieved in these neurons by adding 4OHT at DIV 7. For all Tiam1 deletion culture experiments, Tiam1 loss was confirmed by immunocytochemistry. All neuronal culture experiments were performed on DIV 21, unless otherwise stated.
Immunocytochemistry and microscopy
For rat and mouse primary neurons experiments, cells were fixed with 4% PFA and 20% sucrose and immunostained overnight at 4°C in 5% bovine serum albumin (BSA), 15% goat serum (Invitrogen), and 0.3% Triton X-100. Following incubation in primary antibodies, fluorescently conjugated secondary antibodies were diluted in the same blocking buffer as indicated above and added to neurons for 2 h at room temperature. Fixed neuronal cultures were mounted in FluorSave (EMD Millipore). Images of dendrites and spines from cultured neurons were obtained with a Zeiss LSM 880 microscope. For whole dendritic arbor images, the microscope was operated in confocal mode, and images were acquired using a 10× objective. For dendritic spines, the microscope was operated in a FastAiryscan mode, and images were acquired using a 63× oil objective.
Surface NMDAR subunit analyses
Rat neurons were transfected with empty pSUPER vector, pSUPER-Tiam1 RNAi, or pCMV-Flag-Tiam1 in combination with pCI-SEP-GluN2B or pCI-SEP-GluR1. At DIV17–18, fluorescent images were taken of transfected neurons to assess the surface level of the SEP-receptor subunit before and 5 min after a brief chemical long-term depression (cLTD) stimulation (3 min, 30 μM NMDA; Oh et al., 2006). Neurons were then treated with NH4Cl (50 mM), pH 8.5, and imaged to determine the total level of the transfected SEP-subunit. SEP-receptor subunit endocytosis was calculated as follows: [(mean intensity of the SEP-subunit before cLTD − mean intensity of the SEP-subunit after cLTD) / mean intensity of total SEP-subunit]. Values were normalized to respective controls and converted into a percentage. Images were collected using a Zeiss AxioObserver.1 epifluorescence microscope attached to an Apotome with a 40× oil-immersion objective and analyzed using the ImageJ software. The efficiency of Tiam1 knockdown in neurons by pSUPER-Tiam1 RNAi has been well established in previous work (Duman et al., 2013; Tolias et al., 2005, 2007).
Live imaging of Tiam1 late deletion neurons
For all live imaging experiments, cells were maintained at 35°C and 5% CO2 and imaged with a Zeiss LSM 880 microscope operated in FastAiryscan mode using a 63× oil objective. To measure spine formation and elimination rates of control and Tiam1 late deletion neurons, we imaged the dendritic segments from live cells 1 h apart. For chemical LTP (cLTP) experiments, we used a previously validated protocol (Otmakhov et al., 2004; Oh et al., 2006) that has been reported to drive an increase in spine density (Franchini et al., 2019). Briefly, DIV 21 neurons were incubated in 1.5× ACSF with low magnesium (in mM: 186 NaCl, 4.5 KCl, 1 MgCl2, 4.5 CaCl2, 15 D-glucose, and 15 HEPES), pH 7.4, for 30 min, followed by 1 min of stimulation with 50 μM forskolin, 100 μM PTX, and 0.1 μM rolipram in ACSF without MgCl2 to induce NMDAR-dependent cLTP. After stimulation, neurons were allowed to recover for 1 h in regular 1.5× ACSF.
Adult neurogenesis and immunohistochemistry
Adult neurogenesis experiments were conducted as previously described (Cheng et al., 2021). Briefly, control and Tiam1fKO (3.5- to 4-month-old) were intraperitoneally injected with 200 mg/kg BrdU (Sigma-Aldrich) once every 24 h for 4 d. Mice were transcardially perfused with 4% PFA 14 d after the first injection to study the production of adult newborn neurons or 28 d after first injection to study the survival of the adult newborn neurons in the DG. Following transcardial perfusion, brains were postfixed in 4% PFA and subsequently cryoprotected in 30% sucrose at 4°C. Brains were cryosectioned (30 μm thick), and then free-floating brain sections were incubated first in antigen retrieval solution (Vector Laboratories) at 80°C for 40 min, treated with 2 N HCl at 37°C for 25 min, and then blocked in blocking solution (3% BSA, 10% goat serum and 0.1% Triton X-100 in phosphate-buffered saline) at room temperature for 1 h. Following blocking, sections were incubated in primary antibody at 4°C for 24–36 h and secondary antibody at room temperature for 2 h. Sections were costained with anti-BrdU and DCX (14 d) or NeuN (28 d). The sections were then mounted in the VECTASHIELD antifade mounting medium with DAPI (Vector Laboratories). Sections were imaged using Zeiss AxioObserver.1 epifluorescence microscope attached to an Apotome with a 10× objective. We obtained a Z series of 10 images taken at 1 μm intervals, and the maximum intensity projection of the images acquired using the AxioVision microscopy software (Zeiss) was analyzed using the ImageJ software. The 4.5- to-5-month-old GFP–expressing control and Tiam1fKO mice were also transcardially perfused and cryosectioned (40 μm thick), as above. Free-floating sections from these mice were blocked in a blocking solution at room temperature for 1 h, incubated with anti-GFP primary antibody at 4°C for 16–24 h, and secondary antibodies at room temperature for 2 h. Staining with Texas Red-phalloidin was performed at the same time as incubation with secondary antibody. GFP-expressing DG granule cells from control and Tiam1fKO mouse hippocampal sections were imaged with a Zeiss LSM 880 microscope operated in a FastAiryscan mode using a 63× oil objective. We obtained a Z series at 0.17 μm intervals, masked individual segments using the Surface tool in the Imaris software, and the sum slice projection of the masked images was analyzed using the ImageJ software.
Data analysis
Experimental conditions of samples were blinded for data collection and analysis. Spine measurements from neuronal cultures were obtained from 3D confocal stacks using the Imaris software (Bitplane Scientific Software). Spines were visualized by expressing eGFP or tdTomato in neurons to allow morphological measurements. Spine data were collected from secondary and tertiary segments of dendrite (80–100 μm dendrite analyzed/neuron) using the Imaris Filament Tracer tool. Spine formation and elimination rates in live neurons were analyzed using ImageJ (NIH). Over the time course of the experiment, all formation and elimination events were counted, and the data from each cell were normalized for length of dendrite analyzed. Data are presented normalized to the average of the vehicle control for that day. Dendric arbor analysis of neuronal cultures was performed on single-plane images in ImageJ (NIH) using the NeuronJ and Sholl plugins. All intensity analysis of images was performed in ImageJ (NIH), unless otherwise indicated. To assess the levels of Tiam1 or CaMKIIα in spines and dendrites in transfected neurons treated with vehicle or glutamate, we drawn ROIs over individual structures identified as spines or dendrites using the eGFP channel, and these ROIs were used to measure the intensity values in channels containing the respective stains (for Tiam1 or CaMKIIα). Results were normalized to vehicle-treated control neurons. Spine enrichment values were calculated for individual segments by dividing the mean intensity value at spines by the mean intensity value within the dendritic shaft, again normalized to vehicle-treated neurons. All representative images of primary neuronal cultures are masked to remove axons and dendrites from other neurons. Representative images of some dendritic segments were straightened using ImageJ (NIH).
Statistical analysis
Data presented in the manuscript are mean ± SEM (standard error of the mean). Prism GraphPad was used to perform statistical analyses. Details of statistical analysis are described in the figure legends. Briefly, we used Student's t test when comparing two independent groups and ANOVA of appropriate dimensionality (one- or two-way) when comparing greater numbers of independent groups. Tukey’s post hoc test was used for pairwise comparisons when appropriate to compare multiple groups. p < 0.05 is considered statistically significant. The following symbols were used: n.s. (not significant, p > 0.05); *p < 0.05; **p < 0.01; ***p < 0.001; and ****p < 0.0001.
Results
Late postnatal deletion of Tiam1 from forebrain excitatory neurons enhances contextual fear memory, extinction, and DG-dependent spatial discrimination
We previously reported that global embryonic deletion of Tiam1 resulted in Tiam1 knock-out (Tiam1KO) mice that exhibited enhanced contextual fear memory (Cheng et al., 2021). However, it remained unclear whether Tiam1 plays an ongoing role in regulating adult hippocampal-dependent memory functions or whether the memory enhancements observed in Tiam1KO mice were driven by circuit compensation resulting from the early developmental loss of Tiam1. Moreover, the cell types that contribute to the memory phenotype in Tiam1KO mice were not known due to global Tiam1 deletion. To elucidate the role of Tiam1 in the adult mammalian brain, we genetically ablated Tiam1 from postnatal forebrain excitatory neurons by crossing our Tiam1 floxed mice (Tiam1fl/fl) with CaMKIIα-Cre transgenic mice (Tsien et al., 1996; Dragatsis and Zeitlin, 2000), resulting in Tiam1fKO (Tiam1fl/fl;CaMKIIα-Cre) animals (Ru et al., 2022; Li et al., 2023). Importantly, as we previously showed (Ru et al., 2022), Tiam1fKO mice are viable and fertile and exhibit normal motor function (Extended Data Fig. 1-1A). Since Tiam1 is highly expressed throughout life in the DG region of the hippocampus (Extended Data Fig. 1-1B; Cheng et al., 2021), we assessed Tiam1 loss by performing Western blot analyses on hippocampal lysates from different aged Tiam1fKO animals and littermate controls (Tiam1fl/fl mice). We found that Tiam1 levels in the hippocampi of Tiam1fKO mice were similar to those of control mice between 1 and 2.5 months of age (Extended Data Fig. 1-1C,D) but were significantly reduced by 3 months of age (Con, 1.00 ± 0.16; Tiam1fKO, 0.34 ± 0.04; Fig. 1A), and Tiam1 loss persisted in older Tiam1fKO animals (Extended Data Fig. 1-1E,F). This late ablation of Tiam1 from forebrain excitatory neurons in adult mice is consistent with other studies utilizing CaMKIIα-Cre mice (Huang et al., 2013; I. H. Kim et al., 2014; Mulherkar et al., 2017; P. J. Zhu et al., 2018; Johnson et al., 2019).
Late postnatal deletion of Tiam1 from forebrain excitatory neurons enhances contextual fear memory, contextual fear extinction, and spatial discrimination. Crossing Tiam1fl/fl mice with CaMKIIα-Cre mice results in deletion of Tiam1 from forebrain excitatory neurons. For all figures, mice are abbreviated as Tiam1fl/fl (control or Con) or Tiam1fl/fl;Cre (Tiam1fKO). A, Immunoblots and quantification of hippocampal lysates from 3-month-old control and Tiam1fKO mice showing Tiam1 loss. Remaining Tiam1 protein is likely due to Tiam1 expression in other cell types, including astrocytes (N = 6 mice per genotype for quantification). Two-tailed Student's t test (t(10) = 4.165; p = 0.0019). B, Adult control and Tiam1fKO mice were subjected to classical fear conditioning (pairing of tone with footshock in Context A), and then tested for (C) contextual fear memory (exposure to Context A 24 h after training) and (D) cued fear memory (exposure to tone in Context B 26 h after training) by recording freezing behavior. Freezing behavior in (C) Context A prior to fear conditioning (naive) and (D) Context B before tone serve as controls for contextual and cued fear memory, respectively. Tiam1fKO mice displayed enhanced contextual fear memory and normal cued fear memory compared with control mice. (control, N = 31; Tiam1fKO, N = 35). Two-way RM ANOVA (contextual memory: main effect genotype, F(1,64) = 8.715; p = 0.0044; main effect training, F(1,64) = 224.5; p < 0.0001; genotype × training interaction, F(1,64) = 10.78; p = 0.0017; cued memory: main effect genotype, F(1,64) = 0.1613; p = 0.6893; main effect training, F(1,64) = 1250; p < 0.0001; genotype × training interaction, F(1,64) = 0.0005; p = 0.9825). Tukey's post hoc test showed a significant difference between Con and Tiam1fKO in Context A; p < 0001. E, Adult control and Tiam1fKO mice were subjected to extinction trials after fear conditioning (i.e., repetitive exposure to Context A without footshock). Fear memory extinction was defined as freezing less than during the initial test 24 h after training (Context A). Tiam1fKO mice extinguished the fear memory by Day 2 (+), whereas control animals did so starting on Day 6 (#; control, N = 31; Tiam1fKO, N = 35). Two-way RM ANOVA main effect genotype F(1,64) = 2.404; p = 0.1260; main effect extinction, F(8,512) = 35.30; p < 0.0001; genotype × extinction interaction, F(8,512) = 3.033; p = 0.0024. F, Control and Tiam1fKO mice were assessed in a spatial discrimination task where mice were exposed to two identical objects and then one object was moved to a new position (P1 or P3). Mice typically spend more time with the D.O. than the S.O. G, This was true for both control and Tiam1fKO mice following a large displacement to P3. H, However, only Tiam1fKO mice spent more time with the displaced object following a small displacement to P1, suggesting Tiam1fKO mice have enhanced spatial discrimination (P3, control N = 19; Tiam1fKO N = 17; P1, control N = 17; Tiam1fKO N = 21). Two-way ANOVA (P3, main effect genotype, F(1,68) = 0.000; p > 0.999; main effect object set, F(1,68) = 21.33; p < 0.0001; genotype × object set interaction, F(1,68) = 0.7587; p = 0.3868; P1, main effect genotype, F(1,72) = 0.000; p > 0.9999; main effect object set, F(1,72) = 24.23; p < 0.0001; genotype × object set interaction, F(1,72) = 13.38; p = 0.0005. Tukey's post hoc test showed a significant difference between S.O. and D.O. for Con (p = 0.0397) and Tiam1fKO (p = 0.0019) mice in P3 and Tiam1fKO mice in P1 (p < 0.0001). Data are ±SEM. *p < 0.05; **p < 0.01; ****p < 0.0001. Not significant (ns), p > 0.05. For additional details, see Extended Data Figure 1-1.
Figure 1-1
(A) 3.5-5-month-old control and Tiam1fKO mice were subjected to rotarod task showing normal motor function (Control N = 13, Tiam1fKO N = 10). Two-way RM ANOVA Main effect Genotype: F(1,21) = 0.9957, p = 0.3297, Main effect Training: F(3.793,79.66) = 32.33, p < 0.0001, Genotype x Training Interaction: F(7,147) = 0.2488, p = 0.9719. (B) In situ hybridization images (Top) of Tiam1 mRNA in the hippocampus of 18-month-old and 24-month-old mice. Intensity-coded log-expression summary images (Bottom) show Tiam1 expression. Image credit: Allen Institute. Immunoblots and quantification of hippocampal lysates from (C) 1-month, (D) 2.5-month, (E) 3.5-4-month-old control, and (F) 4.5-5-month Tiam1fKO mice showing late Tiam1 loss in adult animals (N = 3-6 per genotype). Two-tailed Student’s t-test (1 month: t(4) = 0.3063, p = 0.7747; 2.5 month: t(4) = 0.2628, p = 0.8057; 3.5-4 month: t(8) = 3.846, p = 0.0049; 4.5-5 month: t(4) = 5.366, p = 0.0058). (G) Comparison of post-shock freezing in Context A versus Context B 24 hr after training (Control N = 31, Tiam1fKO N = 35). Two-way RM ANOVA Main effect Genotype: F(1,64) = 4.934, p = 0.0299, Main effect Context: F(1,64) = 56.45, p < 0.0001, Genotype x Context Interaction: F(1,64) = 9.539, p = 0.0030. Tukey’s post hoc test showed a significant difference between Con and Tiam1fKO in Context A (p = 0.0006), Con Context A vs. Con Context B (p = 0.0034), and Tiam1fKO Context A vs. Tiam1fKO Context B (p < 0.0001). (H) Contextual fear memory stability was determined by subjecting control and Tiam1fKO mice to classical fear conditioning and testing contextual fear memory 48 hr after training (Control N = 19, Tiam1fKO N = 19). Two-way RM ANOVA Main effect Genotype: F(1,36) = 0.5231, p = 0.0282, Main effect Training: F(1,36) = 135.7, p=<0.0001, Genotype x Training Interaction: F(1,36) = 6.020, p = 0.0191. Tukey’s post hoc test showed a significant difference between Con and Tiam1fKO in Context A, p = 0.0013. Data are ± SEM. **p < 0.01, ***p < 0.001, ****p < 0.0001. Not significant (ns), p > 0.05. Download Figure 1-1, TIF file.
To determine whether Tiam1 regulates learning and memory in adult animals, we subjected 3.5- to 5-month-old Tiam1fKO mice and control littermates to classical fear conditioning (J. J. Kim and Jung, 2006; Cheng et al., 2021; Fig. 1B). Animals were trained to associate a neutral environmental cue (i.e., chamber, tone) with an aversive stimulus (i.e., footshock). Twenty-four hours after training, memory formation was tested by measuring freezing behavior upon re-exposure to the cues (LeDoux, 2000; Curzon et al., 2009). Contextual fear conditioning, which assesses the learned association between training context and footshock, depends on both hippocampal and amygdala function, whereas cued fear conditioning, which tests the association between tone and footshock, does not require the hippocampus (LeDoux, 2000; J. J. Kim and Jung, 2006). We found that Tiam1fKO mice froze substantially more than control mice in the contextual fear memory test (Con, 26.23 ± 3.17%; Tiam1fKO, 40.03 ± 3.05%; Fig. 1C), whereas similar levels of freezing were observed in the cued fear memory test (Con, 80.23 ± 1.56%; Tiam1fKO, 81.16 ± 1.64%; Fig. 1D). Importantly, relative to control mice, Tiam1fKO animals displayed a greater freezing response in Context A but a comparable response in Context B (Extended Data Fig. 1-1G), suggesting intact contextual discrimination abilities. These results indicate that Tiam1fKO mice have enhanced contextual fear memory.
Altered learning and memory may affect cognitive flexibility, which is the ability to adapt to changing conditions (Chaby et al., 2019). To determine if this enhancement in fear memory reflects deficits in cognitive flexibility as in post-traumatic stress disorder (Popescu et al., 2023), we tested the ability of Tiam1fKO mice to extinguish a contextual fear memory. During extinction, the conditioned fear response typically decreases as a new memory forms that suppresses the trained fear memory (Tang et al., 1999; H. Lee and Kaang, 2023). Fear extinction was conducted by repeatedly re-exposing animals to the training chamber (Context A) without footshock. Compared with their initial freezing response 24 h after training, Tiam1fKO mice showed significantly reduced freezing by Day 2 of the extinction trials (Tiam1fKO: 27.79 ± 2.70%), whereas control animals did not extinguish their fear memory until Day 6 (Con, 14.96 ± 1.97%; Fig. 1E). Notably, even on Day 1 of the extinction trials, Tiam1fKO mice (Context A 40.03 ± 3.05% vs Day 1 31.27 ± 2.79%), but not control mice (Context A 26.23 ± 3.17% vs Day 1 25.75 ± 3.26%), appeared to display reduced freezing relative to their initial response (Fig. 1E), suggesting that just one exposure to Context A without footshock was sufficient to initiate fear memory extinction in the Tiam1fKO mice. An alternative explanation for this rapid decline in freezing is that contextual fear memory is less stable in the Tiam1fKO mice. To test this possibility, we subjected a different cohort of mice to classical fear conditioning and tested their contextual fear memory for the first time 48 h post-training. Compared with control animals, Tiam1fKO mice showed a similar enhancement in contextual fear memory at 48 h after training (Con, 30.95 ± 4.71%; Tiam1fKO, 46.28 ± 5.48%; Extended Data Fig. 1-1H) as they did at 24 h after training (Con, 26.23 ± 3.17%; Tiam1fKO, 40.03 ± 3.05%; Fig. 1C), suggesting that Tiam1fKO mice do not passively lose their enhanced fear memory during this time frame. These results indicate that Tiam1fKO mice require fewer extinction trials to suppress the conditioned fear memory compared with control animals, suggesting greater cognitive flexibility.
Since Tiam1 is highly expressed in the DG throughout life (Cheng et al., 2021; Extended Data Fig. 1-1B), we specifically assessed DG function in Tiam1fKO mice by subjecting them to a spatial discrimination task (Wimmer et al., 2012; Fu et al., 2019). Mice were exposed repetitively to two identical objects during training, and then one object was moved to a new position (P1–3 of increasing distance), and the animals’ times spent exploring the displaced (D.O.) and stationary (S.O.) objects were recorded (Fig. 1F). Since mice are interested in novelty, they typically spend more time with the displaced object if they are able to detect the displacement (Wimmer et al., 2012; Fu et al., 2019). Both control and Tiam1fKO animals successfully recognized a large object displacement to P3 (Con, S.O. 43.97 ± 3.07%, D.O. 56.03 ± 3.07%; Tiam1fKO, S.O. 41.17 ± 3.37%, D.O. 58.83 ± 3.37%; Fig. 1G). Given the enhancement of hippocampal function in Tiam1fKO mice, we next tested a small displacement distance to P1 that wild-type (WT) animals typically cannot detect (Fu et al., 2019). As expected, using this short displacement distance, control mice spent equal time exploring the displaced and stationary objects, but Tiam1fKO mice showed a significant preference for exploring the displaced object (Con, S.O. 48.06 ± 3.60%, D.O. 51.94 ± 3.60%; Tiam1fKO, S.O. 36.82 ± 2.61%, D.O. 63.18 ± 2.61%; Fig. 1H). These results indicate that Tiam1fKO mice have enhanced spatial discrimination ability. Together, our findings establish that Tiam1 plays an ongoing critical role in restricting contextual learning and memory and DG-dependent spatial discrimination in the adult brain and that these effects are due to Tiam1’s function in forebrain excitatory neurons.
DG granule cell dendrites, spines, and excitatory synapses appear relatively normal in adult Tiam1fKO mice
Developmental loss of Tiam1 results in DG granule neurons that form normally through Postnatal Day (P)21 but fail to stabilize their dendrites and spines, resulting in mature granule neurons with simplified dendritic arbors, decreased dendritic spine density, and reduced basal excitatory synaptic transmission by P42 (Cheng et al., 2021). Based on these findings, we proposed that Tiam1 is required for the stabilization and maintenance of spines and dendrites in vivo during a period of activity-dependent refinement (Cheng et al., 2021). To assess the ongoing requirement of Tiam1 in the maintenance of DG granule cell dendrites and spines outside of this critical window of postnatal development, we performed morphological analysis of these structures on biocytin-filled neurons from adult control and Tiam1fKO mice ∼15–30 d after Tiam1 loss was initially detected in the hippocampus (Fig. 1A; Extended Data Fig. 1-1C–F). DG granule cells from 3.5 to 4-month-old Tiam1fKO mice had similar dendritic arbor complexity to neurons from control littermates, as indicated by the number of total Sholl crossings (Con, 164.41 ± 11.76 µm; Tiam1fKO, 179.18 ± 15.03 µm; Fig. 2A–C). Likewise, no change was detected in total dendritic length (Con, 2201.44 ± 136.60 µm; Tiam1fKO, 2454.96 ± 188.78 µm; Fig. 2D), average dendrite length (Con, 190.1 ± 8.90 µm; Tiam1fKO, 190.0 ± 6.45 µm; Fig. 2E), or dendritic arbor angle (Con, 71.85 ± 6.27°; Tiam1fKO, 62.46 ± 3.37°; Fig. 2F). Analysis of dendritic spines of DG granule cells in Tiam1fKO mice similarly showed no difference in spine densities when compared with DG granule cells from control mice (Con, 1.06 ± 0.04 spine/µm; Tiam1fKO, 1.15 ± 0.05 spine/µm; Fig. 2G,H). To assess potential changes in basal excitatory synaptic transmission, we recorded mEPSCs from DG granule cells in acute hippocampal slices from 3.5- to 4-month-old control and Tiam1fKO mice. Consistent with the lack of changes in spine density, we detected no difference in mEPSC frequency (Con, 0.77 ± 0.08 Hz; Tiam1fKO, 0.70 ± 0.10 Hz) or amplitude (Con, −5.87 ± 0.36 pA; Tiam1fKO, −6.33 ± 0.29 pA) in DG granule cells (Fig. 2I–K), indicating that basal excitatory synaptic transmission was unaffected by the adult loss of Tiam1. Our results demonstrate that during the timeframe we investigated, Tiam1 is not essential for maintaining DG granule cell dendrites, spines, or excitatory synapses in the adult hippocampus.
DG granule cell dendrites, spines, and excitatory synapses appear normal in adult Tiam1fKO mice. A, Reconstructed morphologies of biocytin-filled DG granule cells from 3.5- to 4-month-old control and Tiam1fKO mice. B, Summary of Sholl crossings, (C) total Sholl intersections, (D) total length, (E) average length, and (F) dendritic angle measurements of DG granule cell dendrites (N = 3 mice per genotype; control n = 17 cells; Tiam1fKO n = 18 cells). Two-tailed Student's t test (total Sholl, t(33) = 0.8260; p = 0.4147; total length, t(33) = 1.077; p = 0.2892; average length, t(33) = 0.0096; p = 0.9924; dendritic angle, t(33) = 1.340; p = 0.1894) (G), Representative images of dendritic spines on biocytin-labeled DG granule cells from 3.5- to 4- month-old control and Tiam1fKO mice and (H) quantification of dendritic spine density (scale bar, 5 μm; N = 3–4 mice per genotype; control n = 26 segments; Tiam1fKO n = 25 segments). Two-tailed Student's t test (t(49) = 1.462; p = 0.1500). I, Representative traces and summary graphs of mEPSC (J) frequency and (K) amplitude recorded from DG granule cells from brain slices of 3.5- to 4-month-old control and Tiam1fKO mice (N = 3–4 mice per genotype; control n = 24 cells; Tiam1fKO n = 27 cells). Two-tailed Student's t test (frequency, t(49) = 0.5107; p = 0.6119; amplitude, t(49) = 1.014; p = 0.3158). No significant difference between control and Tiam1fKO mice is seen for any reported measure. Data are ±SEM. Not significant (ns), p > 0.05.
Tiam1 restricts NMDAR-dependent synaptic plasticity in the DG
We previously reported that Tiam1KO mice exhibited increased survival of adult-born neurons in the DG (Cheng et al., 2021). To test whether altered neurogenesis contributes to enhance hippocampal-dependent learning and memory in Tiam1fKO mice, we intraperitoneally injected 3.5- to 4-month-old Tiam1fKO mice with the thymidine analog BrdU that acts as a tracer for adult newborn neurons by incorporating into dividing cells during DNA synthesis (Wojtowicz and Kee, 2006). Then, 14 or 28 d after BrdU-labeling, brains were collected, sectioned, and immunostained for the immature neuronal marker DCX or the mature neuronal marker NeuN, respectively. Unlike Tiam1KO mice, Tiam1fKO mice did not exhibit differences in either the proliferation (Fig. 3A,B) or the survival (Fig. 3C,D) of adult newborn neurons. Since adult neurogenesis appears normal in Tiam1fKO mice, it is unlikely to be the mechanism responsible for the enhanced hippocampal-dependent memory.
Adult loss of Tiam1 does not affect the production or survival of newborn DG granule cells. The production of adult-born granule cells was monitored by labeling with BrdU and the immature neuronal marker DCX 14 d after BrdU injection. A, Representative immunohistochemistry images of adult-born granule cells from the DG of control and Tiam1fKO mice and (B) quantification of neurons labeled with BrdU with or without DCX showing no change in newborn neuron production (N = 3 mice per genotype, 16 hippocampal sections were analyzed per mouse). Two-tailed Student's t test (BrdU in SGZ: t(4) = 1.066, p = 0.3465; BrdU and DCX in SGZ: t(4) = 0.9163, p = 0.4113; BrdU in GCL: t(4) = 0.5692, p = 0.5997; BrdU and DCX in GCL: t(4) = 1.1260, p = 0.0.3232). Adult-born neuron survival was determined by labeling with BrdU and the mature neuronal marker NeuN 28 d after BrdU injection. C, Representative images of adult-born granule cells from the DG of control and Tiam1fKO mice and (D) quantification of neurons labeled with BrdU with or without NeuN showing no change in newborn neuron survival (N = 3 mice per genotype, 16 hippocampal sections were analyzed per mouse). Two-tailed Student's t test (BrdU in SGZ: t(4) = 1.3420, p = 0.2508; BrdU and NeuN in SGZ: t(4) = 1.681, p = 0.1681; BrdU in GCL: t(4) = 1.037, p = 0.3582; BrdU and NeuN in GCL: t(4) = 1.109, p = 0.3298). SGZ, Subgranule zone; GCL, granule cell layer. Data are ± SEM. Not significant (ns), p > 0.05.
LTP is widely considered to be a cellular correlate of learning and memory (Martinez and Derrick, 1996; Whitlock et al., 2006; Nicoll, 2017). To determine if changes in LTP account for the enhanced hippocampal-dependent learning and memory in Tiam1fKO mice, we prepared acute hippocampal brain slices from 3.5- to 5-month-old control and Tiam1fKO mice and recorded fEPSPs in the DG evoked by PP fiber stimulation. To assess synaptic plasticity, we induced LTP in the DG of control and Tiam1fKO hippocampal slices using a high-frequency stimulation protocol (2 × 100 Hz; Kannangara et al., 2015). We found that Tiam1fKO mice showed a marked enhancement in DG LTP relative to control mice (Fig. 4A,B). Importantly, no difference was detected in input–output curves or PPR, indicating that basal and presynaptic functions were unaltered in Tiam1fKO mice (Extended Data Fig. 4-1A–C). LTP depends on the action of NMDARs, which induce Ca2+-dependent signaling that drives synaptic structural and functional remodeling (Lüscher and Malenka, 2012), and AMPARs, which mediate fast excitatory synaptic transmission and increase at synapses during LTP (Herring and Nicoll, 2016). To determine if the augmented plasticity observed in Tiam1fKO mice is due to altered NMDAR or AMPAR function, we measured isolated NMDAR- and AMPAR-mediated EPSCs from DG granule cells at holding potentials of −60 and +40 mV, respectively. The NMDAR/AMPAR ratios of synaptic currents evoked in DG granule cells by PP stimulation were calculated in hippocampal slices from 4- to 5-month-old control and Tiam1fKO mice. Notably, we observed an increase in the NMDAR/AMPAR ratio in DG granule cells from Tiam1fKO mice compared with control littermates (Con, 0.58±; Tiam1fKO, 1.22 ± 0.18; Fig. 4C,D), suggesting enhanced NMDAR activity and/or level relative to AMPARs in the DG of Tiam1fKO mice. Together, these results suggest that Tiam1 normally limits NMDAR-dependent synaptic plasticity in the DG by restricting NMDAR function.
Tiam1 restricts NMDAR-dependent synaptic plasticity in the DG, facilitates activity-dependent NMDAR internalization in primary hippocampal neurons, and promotes F-actin assembly/stabilization in the spines and dendrites of DG granule cells. A, High-frequency stimulation (HFS)-induced LTP in the DG from Tiam1fKO hippocampal slices is enhanced compared with control slices (N = 5 mice; n = 8 slices per genotype). Two-way RM ANOVA main effect genotype, F(1,14) = 10.59; p = 0.0058; main effect time, F(4.306,60.29) = 2.924; p = 0.0252; genotype × time interaction, F(59,826) = 1.817; p = 0.0003. (B) Traces of fEPSPs at the baseline and 50–60 mins after HFS. C, Traces of DG granule cell NMDAR- and AMPAR-mediated EPSCs evoked by PP stimulation in slices from adult control and Tiam1fKO mice. D, Quantification indicated an increase in the NMDA/AMPA ratio in Tiam1fKO mice (N = 3 mice; n = 9 cells per genotype). Two-tailed Student's t test (t(16) = 03.133; p = 0.0064). E, Surface fluorescence of SEP-tagged GluN2B before and 5 min after a cLTD treatment (30 µM NMDA, 3 min) in control, Tiam1 knockdown (KD), and Tiam1 overexpressing (OE) rat primary hippocampal neurons at DIV 17–18. Quantification of SEP-GluN2B internalization [expressed as SEP-GluN2B endocytosis (% rel. to Con)] showed that (F) Tiam1 KD reduced (N = 3 independent sets; control n = 47 cells; KD n = 35 cells) and (G) Tiam1 OE increased (N = 3 independent sets; control n = 49 cells; OE n = 31 cells) the internalization of GluN2B-containing NMDARs relative to control neurons. Two-tailed Student's t test (KD, t(76) = 2.342; p = 0.0218; OE, t(78) = 2.137; p = 0.0357). H, Phalloidin staining showing basal F-actin in dendritic segments from brain sections of eGFP-expressing adult control and Tiam1fKO mice (N = 3 mice; 45 segments per genotype). Quantification showed lower levels of F-actin in (I) spines and (J) dendrites of Tiam1fKO animals compared with control mice. Two-tailed Student's t test (spines, t(88) = 4.822; p < 0.0001; dendrite, t(88) = 3.159; p = 0.0022). Data are ±SEM. *p < 0.05; **p < 0.01; ****p < 0.0001. Not significant (ns), p > 0.05. See Extended Data Figure 4-1 for additional analyses.
Figure 4-1
(A) fEPSC input-output curves of medial perforant path inputs to the DG were similar between control and Tiam1fKO mice (n = 8 slices, N = 5 mice per genotype). Two-way RM ANOVA (Main effect Genotype: F(1,15) = 0.2630, p = 0.6155, Main effect Stimulation: F(1.846,27.69) = 61.45, p < 0.0001, Genotype x Stimulation Interaction: F(14,210) = 0.3146, p = 0.9918) followed by Tukey’s post hoc test. Similarly, no differences were detected in EPSC (B) input-output curves or (C) paired-pulse ratios of perforant path inputs between control and Tiam1fKO mice (n = 17 neurons, N = 3 mice per genotype). Two-way mixed-model ANOVA (Main effect Genotype: F(1,33) = 1.452, p = 0.2368, Main effect Stimulus: F(8,259) = 89.88, p < 0.0001, Genotype x Stimulus Interaction: F(8,259) = 1.786, p = 0.0800) and Two-tailed Student’s t-test (t(33) = 1.329, p = 0.1930), respectively. Tiam1 KD in primary hippocampal neurons did not alter basal surface SEP-GluN2B puncta (D) size, (E) density, or (F) average intensity (N = 3 independent sets; control n = 51 cells, KD n = 34 cells). Two-tailed Student’s t-test (size: t(83) = 0.7851, p = 0.4347; density: t(83) = 1.853, p = 0.0674; intensity: t(83) = 0.7802, p = 0.4375). (G) Surface SEP-AMPAR before and 5 min after cLTD stimulation in control and Tiam1 knockdown (KD) hippocampal neurons showing no difference in SEP-GluR1 endocytosis (N = 3 independent sets; control n = 32, Tiam1 KD = 35). Two-tailed Student’s t-test (t(65) = 1.527, p = 0.1317). Data are ± SEM. Not significant (ns), p > 0.05. (H) Western blot analysis and quantification (I) of F- and G-actin ratio from hippocampal lysates of 5.5-6-month-old mice showed comparable levels between control and Tiam1fKO mice. (Control N = 4 mice, Tiam1fKO mice N = 6 mice). Two-tailed Student’s t-test (t(8) = 0.4052, p = 0.6959). (J) Hippocampal lysates from 4-5-month-old control and Tiam1fKO mice showed equivalent levels of (K) pPAK (active) and (L) total PAK (N = 6 mice per genotype). Two-tailed Student’s t-test (pPAK/PAK:t(10) = 0.5323, p = 0.6061; PAK/GAPDH: t(10) = 0.8965, p = 0.3910). Data are ± SEM. Not significant (ns), p > 0.05. Download Figure 4-1, TIF file.
Tiam1 mediates activity-dependent NMDAR internalization in hippocampal neurons
How might Tiam1 restrain NMDAR-dependent function and synaptic plasticity? NMDARs are dynamically regulated via membrane trafficking during development and synaptic plasticity processes (Scott et al., 2004; Lau and Zukin, 2007; Dupuis et al., 2014). Given the ability of Tiam1 to bind to NMDARs (Tolias et al., 2005) and promote ligand-stimulated endocytosis of other associated receptors (Palamidessi et al., 2008; Yoo et al., 2010; Boissier et al., 2013; Um et al., 2014; Gaitanos et al., 2016), we hypothesized that Tiam1 may affect hippocampal synaptic plasticity by modulating NMDAR surface localization. To investigate this possibility, we utilized rat primary hippocampal neuron cultures to test the ability of Tiam1 to regulate NMDAR activity-dependent trafficking. We transfected DIV 6–7 hippocampal neurons with super-ecliptic pHluorin (SEP)-tagged NMDAR subunit GluN2B. SEP is a pH-sensitive GFP whose fluorescence can be detected when located on the plasma membrane surface but is quenched in acidic intracellular compartments, such as endocytic vesicles (Miesenböck et al., 1998; Kopec et al., 2006). To reduce Tiam1 expression, we transfected DIV 6–7 hippocampal neurons with a shRNA that effectively knocks down (KD) Tiam1 in neuronal and acute slice cultures (Tolias et al., 2005, 2007; Duman et al., 2013; Rao et al., 2019). On DIV 17–18, we assessed the surface level of SEP-GluN2B between control and Tiam1 KD neurons before and 5 min after inducing cLTD, which drives NMDAR internalization (H. K. Lee et al., 1998). Compared with control neurons, Tiam1 KD neurons displayed reduced levels of GluN2B endocytosis following cLTD stimulation (Con, 100 ± 7.72%; Tiam1 KD, 77.92 ± 7.47%; Fig. 4E,F). Moreover, overexpression of Tiam1 enhanced the internalization of GluN2B after cLTD (Con, 100 ± 7.48%; Tiam1 OE, 125.40 ± 9.15%; Fig. 4E,G). In contrast, Tiam1 KD did not affect the basal SEP-GluN2B puncta size, density, or intensity (Extended Data Fig. 4-1D–F). To test whether Tiam1-driven receptor internalization is specific to NMDARs, we conducted the SEP internalization assay in hippocampal neurons expressing the SEP-tagged AMPA receptor subunit GluR1 (Extended Data Fig. 4-1G). We found no difference in the level of endocytosed SEP-GluR1 between control and Tiam1 KD neurons after cLTD stimulation, indicating the effects are specific to NMDARs. These results show that Tiam1 can promote the activity-dependent internalization of NMDARs in hippocampal neurons. We propose that this mechanism may act in vivo to restrict NMDAR function in the adult hippocampus.
Tiam1 promotes F-actin assembly in the DG and constrains activity-dependent spine remodeling in hippocampal neurons
Synaptic plasticity is propelled by actin-dependent processes in dendritic spines (Hotulainen and Hoogenraad, 2010; Spence and Soderling, 2015). Since the Rac1-GEF Tiam1 is an established regulator of the actin cytoskeleton (Tolias et al., 2005; Duman et al., 2013; Um et al., 2014), we next asked how Tiam1 deletion affects actin dynamics in the adult hippocampus. Actin exists in two forms, polymerized F-actin and monomeric G-actin. To test for global changes in F-actin levels in the hippocampus of Tiam1fKO mice, we performed a Western blot analysis of F- and G-actin isolated from the hippocampal tissue of these mice. However, compared with control mice, we found no difference in the F- to G-actin ratio (Extended Data Fig. 4-1H,I). Similarly, hippocampal lysates from Tiam1fKO mice showed normal levels of phosphorylated PAK (Extended Data Fig. 4-1J–L), a key downstream effector of Rac1 involved in actin regulation (Nikolić, 2008). Since global levels of F-actin and Rac1-PAK signaling in the hippocampus of Tiam1fKO mice appeared normal, we focused our analysis on the spines and dendrites of DG granule cells. To visualize these structures, we generated eGFP-expressing Tiam1fKO and control mice by crossing our animals with Thy1-GFP (M line) mice, which express eGFP sparsely in neurons throughout the brain (Feng et al., 2000). Hippocampal sections from eGFP-expressing Tiam1fKO and control mice were stained with phalloidin to assess F-actin levels in dendritic structures. We found that DG granule cells in Tiam1fKO mice possessed lower levels of F-actin in both spines (Con, 1.00 ± 0.03; Tiam1fKO, 0.78 ± 0.03) and dendrites (Con, 1.00 ± 0.04; Tiam1fKO, 0.84 ± 0.03) compared with control mice (Fig. 4H–J). These results indicate that Tiam1 promotes the assembly and/or stabilization of F-actin in spines and dendrites of DG granule cells in the adult hippocampus.
Actin cytoskeletal dynamics drive rapid spine remodeling during synaptic development and in the adult brain following the induction of synaptic plasticity (e.g., during learning; Spence and Soderling, 2015). Given the decrease in F-actin levels observed in the spines and dendrites of DG granule cells in Tiam1fKO mice, we wondered whether the structural plasticity of spines may be altered in the absence of Tiam1. To address this question, we utilized primary hippocampal neuronal cultures from P0 Tiam1fl/fl mouse pups to model late developmental Tiam1 ablation in vitro and visualize spine dynamics in live neurons. As previously seen with early developmental knockdown of Tiam1 in vitro (Tolias et al., 2005) and global genetic ablation in vivo (Cheng et al., 2021), early Tiam1 loss achieved by transfecting Tiam1fl/fl hippocampal neurons at DIV 4 with a constitutive Cre-expressing plasmid resulted in simplified dendritic arbors (Vector, 217.10 ± 15.46 µm; Cre, 130.30 ± 13.92 µm) and reduced spine densities (Vector, 0.94 ± 0.07 spine/µm; Cre, 0.65 ± 0.05 spine/µm; Fig. 5A–D). Likewise, transfecting Tiam1fl/fl hippocampal neurons with a 4OHT-inducible Cre construct and then adding 4OHT to cultures at DIV 7 to induce early Tiam1 loss resulted in dendrite and spine deficits (Extended Data Fig. 5-1). In contrast, transfecting Tiam1fl/fl hippocampal neurons with a 4OHT-inducible Cre construct and then adding 4OHT to cultures at DIV 14 resulted in the late developmental loss of Tiam1 with no alterations in their dendritic arbors (Veh, 175.90 ± 12.86 µm; 4OHT, 175.80 ± 17.62 µm) or spine densities (Veh, 1.19 ± 0.05 spine/µm; 4OHT, 1.14 ± 0.04 spine/µm) relative to control (vehicle-treated) neurons (Fig. 5E–H). Importantly, similar to the adult loss in vivo, late deletion of Tiam1 in cultures resulted in hippocampal neurons with a marked decrease in F-actin (Veh, 1.00 ± 0.08; 4OHT, 0.72 ± 0.04; Fig. 6A). These results indicate that late developmental loss of Tiam1 from hippocampal neuron cultures recapitulates the dendrite and spine phenotypes displayed in the adult hippocampus of Tiam1fKO mice (Figs. 2A–H, 4H–J).
Late Tiam1 deletion from hippocampal neuron cultures resembles adult Tiam1 loss in vivo. A, Early deletion of Tiam1 from hippocampal cultures was achieved by transfecting Tiam1fl/fl neurons with a constitutive Cre (Cre) construct and tdTomato as fill. B, Immunocytochemistry was used to quantify Tiam1 loss at DIV 21 (N = 3 independent sets; vector n = 27 cells; Cre n = 26). Two-tailed Student's t test (t(51) = 6.026; p < 0.0001). As expected, early deletion of Tiam1 resulted in a decrease in (C) dendrite complexity, as indicated by a reduction in total Sholl crossings, dendritic length, and dendritic tips, and (D) spine density. Two-tailed Student's t test (total Sholl, t(51) = 4.165; p = 0.0001; total length, t(51) = 6.012; p < 0.0001; total dendritic tips, t(51) = 6.055; p < 0.0001; average length, t(51) = 1.069; p = 0.2901; primary dendrites, t(51) = 0.2651; p = 0.7920; spine density, t(51) = 3.464; p = 0.0011). E, Late deletion of Tiam1 in hippocampal cultures from Tiam1fl/fl mice was accomplished by transfecting neurons with a 4OHT-inducible Cre and eGFP (fill) and treating neurons with 4OHT or vehicle (Veh, as control) at DIV 14, a timepoint where dendrites and spines have formed. F, Similar to early deletion, representative immunocytochemistry image and quantification indicated 4OHT-treated neurons showed Tiam1 loss by DIV 21 (N = 3 independent sets; vehicle n = 21 cells; 4OHT n = 19). Two-tailed Student's t test (t(38) = 7.132; p < 0.0001). Late loss of Tiam1 from cultures did not affect (G) dendrite complexity or (H) spine density. Two-tailed Student's t test (total Sholl, t(38) = 0.0054; p = 0.9958; total length, t(38) = 0.2852; p = 0.7770; total dendritic tips, t(38) = 0.0483; p = 0.9617; average length, t(38) = 0.5134; p = 0.6106; primary dendrites, t(38) = 0.0494; p = 0.9608; spine density, t(38) = 0.6415; p = 0.5250). Data are ±SEM. **p < 0.01; ***p < 0.001. Not significant (ns), p > 0.05. Representative dendritic segments in panels D and H were straightened. For early deletion of Tiam1 using the 4OHT-inducible Cre, see Extended Data Figure 5-1.
Figure 5-1
(A) Early deletion of Tiam1 from hippocampal cultures could also be achieved by transfecting Tiam1fl/fl neurons with a 4-hydroxytamoxifen (4OHT)-inducible Cre and eGFP (Fill) on DIV 4 and treating neurons with 4OHT or vehicle (Veh, as control) on DIV 7. (B) Representative immunocytochemistry images and quantification indicate 4OHT-treated neurons showed Tiam1 loss by DIV 21 (N = 3 independent sets; Vehicle n = 14 cells, 4OHT n = 12). Two-tailed Student’s t-test (t(24) = 2.886, p = 0.0081). Early loss of Tiam1 from Tiam1fl/fl cultures treated with 4OHT on DIV 7 resulted in (C) aberrant dendrite complexity and (D) a decrease in spine density by DIV 21, similar to results using constitutive Cre. N = 3 independent sets; Vehicle n = 13 cells, 4OHT n = 12). Two-tailed Student’s t-test (total Sholl: t(23) = 4.683, p = 0.0001; total length: t(23) = 4.203, p = 0.0003; total dendritic tips: t(23) = 4.412, p = 0.0002; average length: t(23) = 0.1347, p = 08940; primary dendrites: t(23) = 0.9957, p = 0.3298; spine density: t(23) = 6.795, p < 0.0001). Data are ± SEM. **p < 0.01, ***p < 0.001, ****p < 0.0001. Not significant (ns), p > 0.05. Representative dendritic segments in panels (D) were straightened. Download Figure 5-1, TIF file.
Hippocampal neurons lacking Tiam1 are primed to undergo NMDAR-dependent synaptic remodeling. A, Phalloidin staining showing basal F-actin in eGFP-expressing vehicle and 4OHT-treated hippocampal mouse neurons indicating decreased F-actin in Tiam1-lacking neurons (N = 3 independent sets; vehicle n = 35 cells, 4OHT n = 30). Two-tailed Student's t test (t(63) = 2.898; p = 0.0052). Live cell imaging of (B) spine addition and (C) elimination events in vehicle- and 4OHT-treated neurons expressing inducible Cre and eGFP (fill) show no difference in basal spine remodeling (N = 4 independent sets; vehicle n = 41 cells; 4OHT n = 41). Two-tailed Student's t test (additions, t(80) = 5.18; p = 0.1329; eliminations, t(80) = 1.6963; p = 0.0944). D, Representative images of spines from vehicle- (Veh, control) and 4OHT-treated (Tiam1 null) Tiam1fl/fl mouse hippocampal neurons expressing inducible Cre and eGFP (fill) before (pre) and 60 min after (post) a cLTP stimulation. Quantification shows an augmented increase in spine density in 4OHT-treated neurons, compared with control (vehicle-) treated neurons (N = 3 independent sets; vehicle n = 15 cells; 4OHT n = 19 cells). Two-tailed Student's t test (t(32) = 4.115; p = 0.0003). Data are ±SEM. **p < 0.01; ***p < 0.001. Not significant (ns), p > 0.05. Representative dendritic segments in panel A were straightened. cLTD was also tested in late Tiam1 deletion neurons (Extended Data Fig. 6-1).
Figure 6-1
(A) Representative images of spines from Control (Sham treated) neurons and neurons 60 min after cLTD stimulation. Quantification shows (B, C) spine shrinkage in both vehicle and 4OHT-treated neurons, but (D) only an increase in spine density in 4OHT-treated neurons lacking Tiam1 (N=3 independent sets; Vehicle n=37-58 cells, 4OHT n=28-50 cells). Two-way ANOVA (spine length – Main effect Genotype: F(1,169)=1.008, p=0.3167, Main effect Treatment: F(1,169)=173.5, p<0.0001, Genotype x Treatment Interaction: F(1,169)=7.619, p=0.0064; max spine diameter – Main effect Genotype: F(1,169)=0.0008, p=0.9777, Main effect Treatment F(1,169)=88.31, p<0.0001, Genotype x Treatment Interaction: F(1,169)=0.0042, p=0.9487; spine density – Main effect Genotype: F(1,169)=12.55, p=0.0005, Main effect Treatment: F(1,169)=49.49, p<0.0001, Genotype x Treatment Interaction: F(1,169)=16.29, p<0.0001) followed by Tukey’s post hoc test. (E) Phalloidin staining showing an increase in F-actin 60 min post-cLTD stimulation in eGFP-expressing 4OHT-treated neurons compared to control neurons (N=3 independent sets; Vehicle n=36 cells, 4OHT n=35 cells). Two-tailed Student’s t-test (t(67)=5.036, p<0.0001). Data are ( SEM. **p<0.01, ****p<0.0001. Not significant (ns), p>0.05. Representative dendritic segments in panels (A) and (E) were straightened. Download Figure 6-1, TIF file.
Using this late Tiam1 deletion model, we first investigated whether Tiam1 loss alters spine dynamics in unstimulated neurons by monitoring spine addition and elimination events in live DIV 21 control (vehicle-) and 4OHT-treated hippocampal neurons. Consistent with their spine density phenotype, 4OHT-treated neurons displayed no difference in the basal rates of spine addition (Veh, 1.00 ± 0.13; 4OHT, 0.76 ± 0.09) or elimination (Veh, 1.00 ± 0.10; 4OHT, 0.76 ± 0.10) compared with control neurons (Fig. 6B,C). Next, we asked whether Tiam1 regulates activity-induced spine remodeling by subjecting DIV 21 control (vehicle-) and 4OHT-treated hippocampal neurons to a cLTP treatment that typically causes an increase in spine density (Franchini et al., 2019). Sixty minutes after receiving the cLTP stimulation, Tiam1 late deletion neurons showed an augmented increase in spine density compared with control neurons (Veh, 6.39 ± 1.46%; 4OHT, 16.99 ± 1.98%; Fig. 6D). We also tested the effects of a cLTD stimulation that typically induces spine shrinkage (He et al., 2011; Rajgor et al., 2018). As expected, 60 min following cLTD treatment, the spines of both control and 4OHT-treated neurons displayed a decrease in length and maximum head diameter (Extended Data Fig. 6-1B,C). However, unexpectedly, cLTD stimulation also induced an increase in both spine density (Extended Data Fig. 6-1D) and F-actin (Extended Data Fig. 6-1E) in 4OHT-treated neurons that were not present in control neurons. These results suggest that Tiam1 stabilizes F-actin in hippocampal neurons, and in its absence, spines and F-actin are in a primed state to remodel following activity-dependent stimulation.
Robust NMDAR activity leads to Tiam1 degradation
Molecular mechanisms that support learning and memory include posttranslational modifications, gene regulation, and protein synthesis and degradation (Kandel, 2012; Mayford et al., 2012; Ortega-de San Luis and Ryan, 2022). The role of regulated proteolysis has increasingly been shown to serve essential roles in synapse remodeling and learning and memory (Bingol and Sheng, 2011; Hegde, 2017; Patrick et al., 2023). The activity of NMDARs is required for this process, in part through the recruitment of the ubiquitin-proteasome system to synapses (Bingol and Schuman, 2006; Bingol et al., 2010). Tiam1 is an established target of protein degradation outside the nervous system (Woodcock et al., 2009a,b; Boissier and Huynh-Do, 2014; Magliozzi et al., 2014; G. Zhu et al., 2014; Genau et al., 2015; Vaughan et al., 2015), and in neurons, it binds to and signals downstream of NMDARs (Tolias et al., 2005). Given the enhancements in hippocampal-dependent learning, memory, and plasticity observed in Tiam1fKO mice, we hypothesized that Tiam1 levels may be regulated by neuronal activity such that strong activity that drives synaptic plasticity and memory formation causes Tiam1 loss. To test this, we monitored Tiam1 protein levels in WT DIV 21 hippocampal neurons at different time points after glutamate stimulation, which we previously reported drives NMDAR-dependent Tiam1 phosphorylation (Tolias et al., 2005). We found that 15 and 60 min glutamate stimulation caused a decrease in Tiam1 levels that was blocked by the NMDAR inhibitor AP5 but not by the AMPAR inhibitor CNQX (Veh, 1.00 ± 0.00; Glut., 5 min 0.71 ± 0.06, 15 min 0.45 ± 0.08, 60 min 0.28 ± 0.08; Glut. + AP5, 1.02 ± 0.16; Glut. + CNQX, 0.40 ± 0.13; Fig. 7A,B). Notably, glutamate-induced Tiam1 loss could also be prevented by the proteasome inhibitor MG-132 (Veh, 1.00 ± 0.00; Glut. 60 min, 0.32 ± 0.12; Glut. 60 min + MG-132, 0.79 ± 0.11; Fig. 7C,D). Tiam1 is present throughout neurons, localizing to the soma, dendritic shafts, and spines (Tolias et al., 2005; Duman et al., 2013). To determine whether Tiam1 loss occurs at spines, we transfected WT neurons with trace amounts of Flag-tagged full–length Tiam1, stimulated them with vehicle or glutamate, and monitored Tiam1 loss by immunostaining (Fig. 7E–G). We found that a 15 min glutamate stimulation resulted in neurons with lower Tiam1 levels at spines (Con, 1.00 ± 0.15; Glut. 0.61 ± 0.08) and a decrease in Tiam1 spine enrichment (i.e., relative level of Tiam1 in spines vs dendrites; Con, 1.00 ± 0.04; Glut. 0.81 ± 0.04). Similar results were observed following a 60 min glutamate treatment (Extended Data Fig. 7-1B–D). Importantly, 15 min glutamate stimulation increased the levels of endogenous CaMKIIα at spines (Con, 1.00 ± 0.07; Glut. 4.63 ± 0.55) and resulted in greater CaMKIIα spine enrichment relative to control neurons (Con, 1.00 ± 0.04; Glut. 1.51 ± 0.09; Fig. 7H–J). A similar synaptic accumulation of CaMKIIα has been reported for other stimulations that drive neuronal activity in neuronal cultures (Y. P. Zhang et al., 2008; Tullis and Bayer, 2024). These results suggest that glutamate treatment is not causing a general loss of synaptic molecules from spines. Together, our findings reveal that NMDAR activation can promote Tiam1 degradation and its loss from dendritic spines. We propose such a decrease in Tiam1 levels could destabilize synaptic actin and enhance NMDAR function, as shown with late genetic ablation of Tiam1 (Figs. 4, 6), enabling activity-dependent synaptic remodeling in the hippocampus following strong NMDAR activation that supports learning and memory.
Robust NMDAR activity can induce Tiam1 degradation. A, Representative immunoblot assessing endogenous Tiam1 levels from rat hippocampal neurons treated with vehicle (Veh) or glutamate (50 µM) for increasing time periods. GAPDH was used as loading control. B, Quantification of Western blots showing significant Tiam1 loss at 15 and 60 min postglutamate treatment. Preincubation with the NMDAR inhibitor AP5 (100 µM) but not the AMPAR inhibitor CNQX (100 µM) blocked Tiam1 loss induced by 60 min glutamate treatment (N = 3 independent sets). ANOVA (F(5,12) = 9.854; p = 0.0006) followed by Tukey's post hoc test. C, D, Similarly, the proteasome inhibitor MG-132 (10 µM) also blocked Tiam1 loss induced by 60 min glutamate treatment (N = 4 independent sets). ANOVA (F(2,9) = 14.34; p = 0.0016). Tukey's post hoc test showed a difference between Veh versus Glut. (p = 0.0014) and Glut. versus Glut. + MG-132 (p = 0.0142). E, Immunocytochemistry of DIV 21 hippocampal neurons expressing eGFP (fill) and Flag-tagged Tiam1 and treated with vehicle (Veh) or glutamate for 15 min. Neurons were fixed and stained with anti-Flag antibodies. F, G, Quantification showed glutamate-induced Tiam1 loss from spines (N = 3 independent sets; Veh n = 24 cells; Glut. N = 25 cells). Quantification measures of (F) relative levels of Tiam1 in spines (Spine Tiam1) and (G) relative levels of Tiam1 in spines versus dendrites (Tiam1 spine enrichment). Two-tailed Student's t test (spine, t(47) = 2.396; p = 0.0206; spine enrichment, t(47) = 3.746; p = 0.0005). H, Representative images and (I, J) quantification showing that 15 min glutamate treatment resulted in the enrichment of endogenous CaMKIIα in spines, in contrast to Tiam1 (N = 3 independent sets; Veh n = 47 cells; Glut. N = 39). Two-tailed Student's t test (spine, t(84) = 7.198; p < 0.0001; spine enrichment, t(84) = 5.442; p < 0.0001). K, Proposed model. Tiam1 normally restrict the plasticity of DG granule cell synapses by limiting the availability of NMDARs at synaptic membranes and promoting actin filament assembly/stabilization at spines. Strong neuronal activity induces Tiam1 degradation, priming synapses for NMDAR-dependent remodeling. Data are ±SEM. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001. Not significant (ns), p > 0.05. Model made using BioRender. For additional analyses, see Extended Data Figure 7-1.
Figure 7-1
(A) Western blot of Tiam1 levels in WT hippocampal neurons 60 min after glutamate (50 µM) stimulation. Representative blots show that loss of Tiam1 induced by glutamate can be blocked to a similar extent by increasing amounts of the proteasome inhibitor MG-132. (B) Immunocytochemistry of DIV 21 hippocampal neurons expressing eGFP (Fill) and Flag-tagged Tiam1 and treated with vehicle (Veh) or glutamate for 60-min. Neurons were fixed and stained with anti-Flag antibodies. Quantification showed glutamate-induced loss of Tiam1 from (C) spines and (D) as well as a decrease in the spine enrichment of Tiam1. (N=3 independent sets, n=28 cells per condition). Two-tailed Student’s t-Test (spine: t(54)=3.241, p=0.0020; spine enrichment: t(54)=2.796, p=0.0072). Data are ( SEM. **p<0.01. Not significant (ns), p>0.05. Representative dendritic segments in panels (B) were straightened. Download Figure 7-1, TIF file.
Discussion
Precise control of hippocampal synaptic plasticity is critical for learning and memory (Kennedy, 2013; Donato et al., 2021; Ortega-de San Luis and Ryan, 2022). While there is extensive knowledge of mechanisms that promote this plasticity, much less is known regarding mechanisms that limit it. In the current study, we identify the actin cytoskeleton regulator Tiam1 as a molecule that restricts synaptic plasticity in the DG of the adult hippocampus. We previously established Tiam1 as a key mediator of hippocampal development that promotes spine growth and stabilizes dendrites, spines, and excitatory synapses during a postnatal critical window (Tolias et al., 2005; Cheng et al., 2021). Here, we show that Tiam1 is a major regulator of ongoing hippocampal function in adult animals. However, in contrast to development, Tiam1 is not essential for maintaining adult excitatory synapses during the timeframe of our investigation. Instead, we find that Tiam1 restricts NMDAR-mediated synaptic plasticity in the DG and constrains hippocampal-dependent learning and memory. Our results suggest that Tiam1 may regulate these processes by controlling NMDAR activity-dependent trafficking and F-actin stabilization at dendritic spines. Furthermore, we report that neuronal activity may modulate these constraints on synaptic plasticity by regulating Tiam1 protein levels. We propose a model where Tiam1 acts downstream of NMDAR activation in the DG to promote synaptic stability (Fig. 7K). Under conditions of strong neuronal activity, plasticity could be fine-tuned by targeting Tiam1 for degradation to allow synaptic structural and functional remodeling.
In humans and rodents, the hippocampus is vital for acquiring and encoding episodic memory, including associative and spatial memory (Y. S. Lee and Silva, 2009; Allen and Fortin, 2013). The DG subregion of the hippocampus regulates learning and memory, spatial navigation, and pattern separation (Lein et al., 2007). In this study, we uncover the role of Tiam1 in restricting hippocampal-mediated learning and memory and DG-dependent spatial discrimination in the adult brain. These findings align with a previous report from our lab demonstrating that global embryonic ablation of Tiam1 enhances contextual fear memory in mice (Cheng et al., 2021). However, unlike global Tiam1 deletion (i.e., Tiam1KO mice), the removal of Tiam1 postnatally from forebrain excitatory neurons (i.e., Tiam1fKO mice) does not affect basal hippocampal excitatory synaptic transmission, neural circuitry, or adult neurogenesis. Thus, the memory enhancements exhibited by Tiam1fKO mice cannot be explained by developmental alterations in hippocampal circuits or augmented adult neurogenesis. In addition to enhanced contextual fear memory, Tiam1fKO animals display greater contextual fear extinction and DG-dependent spatial discrimination than control mice. Given Tiam1’s predominant expression in the DG and altered DG function in Tiam1fKO mice, we propose that Tiam1 restricts learning and memory in the adult brain, at least in part, by acting in the DG. Under physiological conditions, limiting hippocampal-dependent learning may support a healthy memory system by ensuring that only relevant events and contexts are stored as memories rather than irrelevant events or noise. In the context of diseases where memory is impaired, this mechanism may exacerbate memory problems. Since Tiam1 serves an ongoing role in limiting hippocampal-dependent functions in the adult brain and its loss does not appear to affect memory stability, targeting Tiam1 activity or levels in disease could serve as a potential therapeutic avenue to enhance cognitive function.
In the current study, we sought to elucidate the physiological role of Tiam1 in adult hippocampal synaptic plasticity and function. In the DG, we find that Tiam1 restricts synaptic plasticity and NMDAR function in excitatory neurons and promotes F-actin assembly/stabilization in spines. In contrast to the hippocampus, recent work from our lab revealed that Tiam1 promotes chronic pain-associated maladaptive synaptic plasticity in the anterior cingulate cortex and the spinal dorsal horn (Ru et al., 2022; Li et al., 2023). In both cases, Tiam1 acts in excitatory neurons to promote nerve injury-induced F-actin assembly and synaptic NMDAR stabilization that drives maladaptive plasticity. These findings suggest that while Tiam1 controls synaptic plasticity by regulating F-actin remodeling and NMDAR function in all cases, it does so in a context-dependent manner. Tiam1 also promotes persistent structural LTP in CA1 pyramidal neurons from hippocampal organotypic slice cultures (Saneyoshi et al., 2019), suggesting that even within the hippocampus, the actions of Tiam1 are regulated in a cell-type–specific manner. Together, these findings implicate Tiam1 as a critical regulator of synaptic plasticity whose specific function may vary depending on the brain region, cell type, and developmental context.
Rho-family small GTPases, including Rac1, are essential regulators of actin dynamics. Tiam1 is a Rac1-GEF that promotes F-actin assembly by controlling the spatial and temporal activity of Rac1 and downstream signaling pathways (Boissier and Huynh-Do, 2014; Duman et al., 2022). While we were unable to detect any difference in the global levels of F-actin or phosphorylated PAK in hippocampal lysates from Tiam1fKO mice compared with control animals, we did find that DG granule cells from adult Tiam1fKO mice have reduced F-actin levels in their dendrites and spines, similar to what we observed in late Tiam1 deletion cultured hippocampal neurons. These results suggest that Tiam1 regulates specific sub-pools of F-actin within hippocampal neurons. The observation that phospho-PAK levels are relatively normal in Tiam1fKO hippocampal lysates could also indicate that Tiam1 promotes F-actin assembly and/or stability in hippocampal neurons via alternative Rac1 pathways such as the WAVE regulatory complex (IRSp53, WAVE, Arp2/3; Choi et al., 2005; Soderling et al., 2007; M. H. Kim et al., 2009; Sawallisch et al., 2009) or the Par polarity complex (Par3, Par6, aPKC; H. Zhang and Macara, 2006, 2008; Duman et al., 2013), both of which directly associate with Tiam1 (Chen and Macara, 2005; Connolly et al., 2005; Nishimura et al., 2005; Ten Klooster et al., 2006; Narayanan et al., 2013). While other neuronal Rac1-GEFs may be able to compensate for Tiam1 loss by activating Rac1, they likely do not assemble the same actin regulatory complexes required to rescue Tiam1-mediated F-actin assembly and/or stabilization (Duman et al., 2022). Moreover, they may not activate Rac1 signaling in the same spatiotemporal manner in response to stimulation. Future studies are required to assess how Tiam1 specifically regulates Rac1-dependent actin remodeling in the adult hippocampus.
To determine the potential mechanism(s) by which Tiam1 regulates synaptic plasticity, we utilized primary hippocampal neuronal cultures. We found that Tiam1 promotes agonist-induced NMDAR internalization, increases synaptic F-actin levels, and restricts activity-dependent structural remodeling of spines. Although Tiam1-null neurons have lower basal F-actin levels, cLTP stimulation resulted in an augmented increase in spine density in these neurons compared with control cells. Given our finding that Tiam1 facilitates NMDAR internalization, neurons lacking Tiam1 may have increased NMDAR surface expression, which could explain their altered response to NMDAR stimulation. However, analysis of the surface levels of SEP-GluN2B puncta between Tiam1 KD and control cultured neurons before NMDAR stimulation found no difference, suggesting that basal NMDAR surface levels are not affected by Tiam1 loss (Extended Data Fig. 4-1E,G). Alternatively, mature neurons lacking Tiam1 may undergo enhanced NMDAR-dependent spine remodeling due to a dearth of stabilized F-actin. The actin cytoskeleton plays a crucial role in maintaining synapse stability, while its remodeling drives synaptic structural and functional plasticity (Star et al., 2002; Bosch et al., 2014; Spence and Soderling, 2015; Chazeau and Giannone, 2016; Obashi et al., 2019; Okabe, 2020). Our results argue that in the adult hippocampus, Tiam1 limits the ability of spines to undergo activity-dependent remodeling by stabilizing the actin cytoskeleton.
We previously found that Tiam1 is phosphorylated and activated following brief NMDAR activation in a calcium-dependent fashion (Tolias et al., 2005). Here, we report that prolonging this same stimulation drives Tiam1 loss. Importantly, outside of the central nervous system, phosphorylation of Tiam1 can result in its degradation (Woodcock et al., 2009a,b; Magliozzi et al., 2014) via the ubiquitin-proteasome system (Magliozzi et al., 2014; Genau et al., 2015; Vaughan et al., 2015). Based on our findings, we propose that NMDAR-dependent activity may modulate brakes on plasticity within the hippocampus by targeting Tiam1 for degradation. In support of this, in the nucleus accumbens (NAc), downregulation of Tiam1 protein has been reported to prime NAc medium spiny neurons (MSNs) to undergo long-term changes in plasticity induced by repeated cocaine administration (Dietz et al., 2012). In rats that self-administered cocaine, lower Tiam1 protein levels were also reported within the NAc (Chandra et al., 2013). Importantly, repeated optogenetic activation of NAc D1-MSNs also resulted in the downregulation of Tiam1, and cocaine-induced Tiam1 loss was blocked by the optogenetic inhibition of these neurons (Chandra et al., 2013). In addition to acutely regulating spine plasticity, this mechanism may function at longer timescales to modulate the global plasticity of hippocampal neurons. Indeed, Tiam1 was previously identified among a group of proteins that showed decreased synthesis both during homeostatic up- and downscaling in hippocampal neurons, suggesting that Tiam1 may act as a “general scaling protein” that is downregulated in response to global changes in neuronal activity to enable synapse remodeling (Schanzenbächer et al., 2016). Together with these reports, our results indicate that the regulation of Tiam1 protein levels is a fundamental mechanism for controlling synaptic remodeling throughout the adult brain at multiple timescales.
Footnotes
We thank J. G. Duman for the helpful discussion and critical reading of the manuscript and S. Mulherkar, S. Veeraragavan, and other Tolias Lab members for the technical advice and support. This work was supported by National Institutes of Health (NIH) Grants F31NS122427 (F.A.B), NS062829 (K.F.T.), MH137505 (K.F.T.), NS124141 (K.F.T.), NS085171 (J.X.C.), and NS086965 (J.C.) and the Mission Connect-TIRR Foundation (K.F.T.). We also received technical assistance and resources from the Baylor College of Medicine Neuropathology and Behavioral IDDRC Cores (supported by NIH NICHD Grant U54 HD083092).
The authors declare no competing financial interests.
- Correspondence should be addressed to Kimberley Tolias at tolias{at}bcm.edu.
