Abstract
Leucine zipper protein 1 (LUZP1) functions in the maintenance and dynamics of the cytoskeleton by interacting with actin and microtubules. Deficiency or mutation of LUZP1 is associated with brain developmental disorders; however, its precise role in brain function remains unclear. We showed that LUZP1 localizes to actin and is highly expressed in CaMKIIα-expressing neurons within the mouse hippocampal dentate gyrus. Depletion of LUZP1 impedes dendritic spine maturation, which is characterized by excess immature filopodia and loss of mature mushroom spines both in vitro and in vivo. LUZP1 knockdown reduces spontaneous electrical activity and synaptic plasticity in hippocampal neurons. Conditional deletion of LUZP1 in CaMKIIα-expressing neurons causes impaired learning and memory behavior in mice of both sexes. Mechanistically, LUZP1 control dendritic maturation by directly interacting with filamin A and modulating the Rac1-PAK1 signaling pathway. These findings shed light on the role of LUZP1 in regulating synaptic plasticity and brain function.
Significance Statement
Unraveling brain development and function is critical for understanding neurological disorders. The discovery of LUZP1 sheds light on cytoskeletal dynamics, crucial for brain health. LUZP1's interaction with actin and its link to brain disorders highlight its neural importance. LUZP1's presence in specific hippocampal neurons, notably CaMKIIα-expressing ones, provides insights into synaptic function. Depletion of LUZP1 hinders dendritic spine maturation, impacting synaptic activity and plasticity, crucial for neural connectivity. The association of LUZP1 with learning impairments upon deletion in neurons emphasizes its role in cognition. Its involvement in the Rac1-PAK1 pathway offers novel insights into dendritic maturation and synaptic plasticity regulation. These discoveries illuminate LUZP1's significant impact on synaptic plasticity and brain function, hinting at potential interventions for LUZP1-associated neurological disorders.
Introduction
Dendritic spines have been identified as the fundamental structures for neuroplasticity and neural circuit formation, playing a crucial role in the regulation of neuron function (Calabrese et al., 2006; Yoshihara et al., 2009). During neuronal maturation, dendritic spines undergo structural and functional changes, which are characterized by the transformation of immature filamentous-like filopodia dendritic spines into mature mushroom dendritic spines with enlarged heads and well-defined necks (Zuo et al., 2005; Yoshihara et al., 2009). This transformation is crucial for the establishment of stable and functional synaptic connections (Bourne and Harris, 2007; Arstikaitis, 2011). Mature mushroom spines are characterized by a bulbous head that contains various structures, including postsynaptic densities (PSDs), neurotransmitter receptors, signaling molecules, and molecular machinery necessary for synaptic function (Yoshihara et al., 2009; Runge et al., 2020; Stratton and Khanna, 2020). They serve as the primary sites of excitatory synaptic input, receive neurotransmitter signals from presynaptic neurons, and convert them into electrical signals in postsynaptic neurons (Hering and Sheng, 2001). An excessive number of immature spines or a delay in their maturation are implicated in abnormalities in synaptic plasticity, neural development, and cognition (Petrak et al., 2005; Kanjhan et al., 2016; Ozcan, 2017).
Studies have shown that abnormal polymerization or depolymerization of the actin cytoskeleton (Pyronneau et al., 2017; Hoi et al., 2020) as well as mutations in actin-binding proteins (Soria Fregozo and Pérez Vega, 2012; Costa et al., 2020; Khan et al., 2021; Speranza et al., 2022) can alter spine morphology and filopodia density in different mouse models of neurodevelopmental disorders (Duffney et al., 2015; Yan et al., 2016; Pyronneau et al., 2017). For instance, the absence of nArgBP2, an actin-binding protein enriched in dendritic spines, impairs dendritic complexity and maturation in hippocampal dentate gyrus (DG) granule cells (Duffney et al., 2015; Lee et al., 2016). Drebrin, which is also an actin-binding protein, plays a critical role in maintaining dendritic spine morphology and synaptic connectivity (Merriam et al., 2013; Martin et al., 2019), and its downregulation in the hippocampus has been implicated in Alzheimer's disease (AD; Pelucchi et al., 2020).
Leucine zipper protein 1 (LUZP1) is an actin-stabilizing protein that contains three leucine zipper motifs at the N-terminus and plays a role in orchestrating protein‒protein interactions (Gonçalves, 2022). LUZP1 participates in the maintenance and dynamics of cell structure by interacting with actin and microtubules, thus regulating several critical biological processes, such as cell migration, cell‒cell connections, and cell signaling (Wang and Nakamura, 2019; Bozal-Basterra et al., 2020, 2021; Gonçalves et al., 2020; Gonçalves, 2022). LUZP1 localizes to actin filaments as well as to the nucleus of neuronal cells (Sun et al., 1996). LUZP1 depletion is implicated in dysregulated ciliogenesis, cell migration, and centrosome function, which are associated with Townes–Brocks syndrome (TBS) and 1p36 deletion syndrome (Jordan et al., 2015; Bozal-Basterra et al., 2020). Moreover, deletion of LUZP1 causes abnormal neural tube development in mouse embryos and brain development abnormalities (Hsu et al., 2008; Jordan et al., 2015). These findings suggest that LUZP1 plays a critical role in regulating neural structure and function; nevertheless, its precise function has not been fully elucidated.
In the present study, we found that LUZP1 is highly expressed in CaMKIIα-expressing neurons in the granular cell layer (GCL) of the mouse hippocampus. Due to the embryonic lethality observed in LUZP1 knock-out mice, which restricts their use for studying synaptic and cognitive functions in the adult brain in vivo, we generated mice with conditional LUZP1 knock-out in CaMKIIα-expressing neurons. We found that the deletion of LUZP1 both in vitro and in vivo disrupted dendritic spine maturation in hippocampal neurons, which was characterized by an abnormal increase in filopodia-like protrusions but a decrease in mushroom-shaped spines. Aberrant neuronal structures and electrophysiological activities in the hippocampus due to LUZP1 dysfunction may attribute to deficits in learning and memory in mice. Collectively, our findings provide novel insights into the role of LUZP1 in regulating synaptic plasticity and brain function.
Materials and Methods
Compliance with ethical standards
All experimental procedures and use of the animals were conducted in accordance with the guidelines established by the Association for Assessment and Accreditation of Laboratory Animal Care and the Institutional Animal Care and Use Committee of Sichuan University. All efforts were made to minimize the suffering of the mice.
Animals
Adult male C57BL/6 mice (body weight 18-22 g) were obtained from Vital River Laboratory Animal Technology. The mice were housed in a controlled environment with a 12 h light/dark cycle, maintained at a constant temperature, and provided with ad libitum access to food and water. All experimental procedures were approved by the Institutional Animal Care and Use Committee of Sichuan University.
CaMKIIα-CreERT2 mice (GemPharmatech ) were mated with female LUZP1flox/flox mice (GemPharmatech) to generate LUZP1flox/flox-CaMKIIα-CreERT2+ mice, the mice with LUZP1 conditional knock-out in CaMKIIα-expressing neurons (LUZP1-CKO) after tamoxifen induction. LUZP1flox/flox-CaMKIIα-CreERT2− mice (LUZP1-C) were used as controls. Tamoxifen (75 mg/kg, catalog #T5648, Sigma-Aldrich) was administered via intraperitoneal (i.p.) injection for 5 consecutive days to conditionally delete LUZP1 in CaMKIIα-expressing neurons in the hippocampus of mice (6–8 weeks). The strategy of tamoxifen treatment was referred to https://www.jax.org/research-and-faculty/resources/cre-repository/tamoxifen. On the 14th day after the first tamoxifen injection, the mice were killed, and LUZP1 expression in the hippocampus was detected using quantitative reverse transcription-polymerase chain reaction (qRT-PCR) and Western blotting. Additionally, tail genotyping of mice was performed by PCR using the Quick Genotyping Assay Kit for Mouse Tail (catalog #d7283S, Beyotime Institute of Biotechnology). DNA extraction from tail clippings was performed using the Animal Genomic DNA Quick Extraction Kit (catalog #d0065S, Beyotime Institute of Biotechnology). The primer sequences are listed in Table 1.
Key resources table
Cell culture
Mouse neuroblasts Neuro-2a (N2a) and mouse hippocampus-derived neuronal cell line HT22 were maintained in high-glucose DMEM medium (catalog #11965092, Invitrogen), supplemented with 10% FBS, 1% Penicillin-Streptomycin-Glutamine (100×) (catalog #10378016, Invitrogen). Both cells were cultured at 37°C in a humidified atmosphere with 5% CO2 flow.
Primary hippocampal neurons were isolated from embryonic day 18 (E18) mouse embryos. The hippocampal neurons were cultured on 18 mm coverslips coated with poly-d-lysine (1 mg/ml, Sigma). The hippocampal neurons were cultured in culture dishes coated with poly-d-lysine (0.1 mg/ml) at a density of 1.5 × 106 cells per 35 mm dish for protein immunoblot analysis and qRT-PCR. The neurons were cultured in a Neurobasal medium supplemented with 2% B27 and 0.25% ʟ-glutamine (Invitrogen). Hippocampal neurons were collected at days in vitro (DIV) 21 after transfection with LUZP1 shRNA plasmid at DIV13 for RNA-seq analysis.
Transfection of primary hippocampal neurons
Primary neurons were transfected using Lipofectamine 2000 (Invitrogen). Briefly, 2 μg of DNA and 3 μl of Lipofectamine 2000 were diluted in 100 μl of neurobasal medium (Thermo) per well in a 12-well plate. The mixture was incubated at room temperature for 30 min, and conditioned medium of the neurons was transferred to a new 12-well plate and kept in a 37°C incubator. Then, 1 ml of incubation medium and 200 μl of DNA/Lipofectamine mixture were added to the neurons and incubated at 37°C for 45 min. The transfection mixture was replaced with a conditioned culture medium.
Western blot analysis and pull-down assay
The brain tissue was lysed, and protein extraction was performed using a mammalian cell and tissue extraction kit (K269-500, BioVision). RIPA buffer containing protease and phosphatase inhibitors was used to lyse the primary hippocampal neurons. The total protein concentration was determined using a Bradford assay kit (P0006, Beyotime). Protein samples (20 μg) were loaded onto a 10% SDS-PAGE gel for separation and transferred onto a PVDF membrane. The PVDF membrane was incubated overnight at 4°C with primary antibodies, followed by incubation with secondary antibodies for 90 min. Immunoreactivity was visualized using a chemiluminescence substrate and a chemiluminescence imaging system. The band density was quantified using Chemi analysis software (CLINX).
For the Rac1 activity assay and m7 GTP pull-down assay, lysate (0.5–1 mg) was incubated with PAK-PBD beads (20 μl, Cytoskeleton) or m7 GTP beads (20 μl, Jena Bioscience) at 4°C with rotation for 2 h. After incubation, the beads were centrifuged at 6,000 rpm for 1 min and washed three times with 600 μl of washing buffer. The beads were eluted with sample buffer for subsequent western blotting analysis.
Co-immunoprecipitation
Immunoprecipitation was performed using Dynabeads Protein G (10004D, Invitrogen) according to the manufacturer's instructions with minor modifications. Briefly, 50 μl of Dynabeads Protein G beads were first coated with 1–2 μg of primary antibody or normal mouse IgG1 (sc-3877, Santa Cruz Biotechnology). The mouse hippocampal tissue was lysed, and the supernatants were subsequently incubated with antibody-bound Dynabeads Protein G overnight at 4°C. After washing, the Dynabeads–antibody–protein complex was eluted with 50 mM glycine, pH 2.8, at RT for 2 min to dissociate the complex. The samples were then subjected to SDS-PAGE and immunoblotting.
Quantitative reverse transcription-PCR detection
Hippocampal neurons were processed according to the instructions of the Total RNA Isolation Kit (Axygen) to isolate total RNA. Purified RNA was used to prepare cDNA in a 20 μl reaction volume using the Bestar First Strand cDNA Synthesis Kit (DBI Bioscience). For qRT-PCR analysis, cDNA amplification was performed using SYBR Green (DBI Bioscience). Gene expression was analyzed using the 2−ΔΔCt method and the relative levels of the target gene were normalized to β-actin. The sequences of all primers were listed in Table 1.
RNA-sequencing
RNA-seq procedures were performed as previously described (Wang et al., 2023b). Briefly, total RNA was extracted from tissues using TRIzol reagent according to the manufacturer's specifications (Invitrogen). An RNA-seq transcriptome library was generated from 1 μg of total RNA using the TruSeq RNA sample preparation kit (Illumina). cDNA was generated using SuperScript double-stranded cDNA synthesis kit (Invitrogen) with random hexamer primers (Illumina). The paired-end RNA-seq library was subsequently sequenced using an Illumina HiSeq xten/NovaSeq 6000 sequencer (2 × 150 bp read length). Significance analysis was conducted (1.2-fold change and padjust < 0.05) to identify strongly up- or downregulated genes by contrasting each group with the control group. Functional annotations of genes were performed using the Kyoto Encyclopedia of Genes and Genomes (KEGG) database (https://www.kegg.jp). KEGG pathway enrichment analysis was carried out using R script on the gene/transcript set, and the q-value was used for adjustment through multiple comparisons.
Immunofluorescence staining
Primary hippocampal neurons were fixed with a solution containing 4% paraformaldehyde (PFA) and 4% sucrose for 15 min at room temperature. Neurons were permeabilized with 0.1% Triton X-100 for 10 min and incubated overnight at 4°C with primary antibodies, followed by incubation with Alexa Fluor-conjugated secondary antibodies for 1 h. Immunofluorescence images were captured using a confocal microscope (Leica SP8 X, Leica) equipped with LAS X software. Dendritic spine morphology was manually quantified and analyzed using MetaMorph software. Dendritic spines were classified based on previously reported morphological parameters including length (L), head width (H), and neck width (N) values (Lin et al., 2017). Accordingly, spines were defined as mushroom (H/N ratio > 1.5), stubby (H/N ratio ≤ 1 and L/N ratio ≤ 1), thin (1 < H/N ratio < 1.5 and 1.5 ≤ L/N ratio ≤ 3), or filopodia (H/N ratio < 1.2 and L/N ratio > 3). For spine volume analysis, GFP-labeled neurons were automatically reconstructed using the Filament Tracer module of Imaris software (version 9.5.1, Bitplane). Dendritic spines were subsequently classified into four categories using the “Imaris Spines Classifier” extension with the autoset criteria. All steps were performed blind to the experimental conditions.
Stereotaxic injection
Mice were anesthetized with sodium pentobarbital (60 mg/kg, i.p.) and placed on a stereotaxic apparatus. An injection cannula was positioned at the bregma as the starting point, and the coordinates of the DG area of hippocampus (AP, −1.7 mm; ML, ±1.2 mm; DV, −2.0 mm) were determined based on a stereotaxic atlas of the mouse brain. AAV viruses were bilaterally injected into the hippocampus using stereotactic guidance. The injection needle was left in place for 10 min, and mice were allowed to recover for 3 weeks before experiments.
Immunohistochemical staining
Immunohistochemical staining was performed as previously described (Jiang et al., 2022). The mice were anesthetized with sodium pentobarbital (60 mg/kg) and transcranial perfused with cold 4% paraformaldehyde solution. The brains were fixed, dehydrated, and sectioned into 10 μm slices using a cryostat. Brain slices were permeabilized and blocked with 0.25% Triton X-100 in 10% normal goat serum for 1 h, incubated overnight with primary antibodies, followed by staining with fluorescently labeled secondary antibodies (Thermo Fisher Scientific). Imaging was performed using a laser scanning confocal microscope (Leica SP8 X, Leica) equipped with a 40×/1.40 objective at a resolution of 1,024 × 1,024 pixels. A high-throughput fluorescent automatic scanning system (Olympus VS200) was used for whole-brain imaging of brain slices.
Transmission electron microscope
Mice hippocampi were fixed with cold 2% paraformaldehyde and 2.5% glutaraldehyde solution. Samples were washed three times with 0.1 M PBS for 15 min each and subsequently fixed using a solution of 1% osmium tetroxide and 1.5% potassium ferrocyanide. Hippocampi were stained and incubated with 1% uranyl acetate at room temperature for 1 h and then dehydrated using a serial ethanol gradient. Samples were then placed in a 1:1 epoxy propane:acetone mixture and were subjected to gradient infiltration of 1:1, 2:1, and 3:1 epon812:expoxy propane mixtures. Ultrathin sections of ∼60–90 nm were prepared using an ultramicrotome, mounted on glass slides, and then transferred onto copper grids. The sections were stained with uranyl acetate for 10–15 min, followed by lead citrate staining for 1–2 min. The copper grids were imaged using a JEM-1400FLASH transmission electron microscope (manufactured by JEOL). Each grid was initially observed at 6,000× magnification, and specific regions of interest were selected for image acquisition to observe the synapses.
Long-term potentiation
Mice were anesthetized using 4% isoflurane and were rapidly decapitated to harvest the brain. Thick coronal slices of 400 μm were prepared using a Leica VT1200 Vibratome in a cold sucrose cutting solution equilibrated with 95% O2 and 5% CO2. The hippocampal slices were incubated for 1 h at 32–34°C and then in artificial cerebrospinal fluid at room temperature. Recording electrodes were made from borosilicate glass, and the intracellular solution was used to fill the electrodes, resulting in a resistance of 4–5.5 MΩ. The series resistance was monitored in all recordings. At the beginning of the experiment, single stimuli were applied to the brain tissue using the Sequencing Keys function. The initial stimulation intensity was set to 2 V with an increment of 1, and the interstimulus interval was 30 s. The maximum stimulus response was typically obtained within 8 V, at which point the single stimulation was stopped, and the stimulus intensity corresponding to half-maximal response (EC 50%) was calculated on the recording paper. This stimulus intensity was then used for all subsequent experiments. During the induction phase of LTP, the first 15 min consisted of single stimuli with an interstimulus interval of 30 s. A total of 30 responses were recorded during this period, and these 30 responses were expected to exhibit similar characteristics when given the same stimulus. If the response fluctuated significantly (outside of 10% of the overall response), the recording was discarded. LTP was induced by applying high-frequency stimulation (HFS), which consisted of four 1 s trains of 100 Hz stimulation with a 20 s interval between each train. The magnitude of LTP was quantified as the percentage change in field excitatory postsynaptic potential (fEPSP) slope (40%) within 60 min after LTP induction. Following the HFS stimulation, single stimulus recordings were conducted for an additional hour. Data were excluded from analysis if the series resistance exceeded 33 MΩ or changed by >10%. Recordings were acquired using a MultiClamp 700B microelectrode amplifier (Molecular Devices), digitized with a Digidata 1550B, and analyzed using pClamp 10 software (Molecular Devices). The signals were sampled at 10 kHz, amplified with a gain of 500, and bandpass filtered between 1 and 3,000 Hz.
Multielectrode array recording
Neuronal electrophysiological activity was recorded using a multielectrode array recording (MEA) system (Axion BioSystems). Primary cultured hippocampal neurons were seeded at a density of 1 × 105 cells per well onto 24-well MEA plates coated with poly-d-lysine (1 µg/µl). The MEA plates contained 16 embedded gold electrode arrays. The neurons were subsequently cultured in a humidified incubator at 37°C with 5% CO2. At DIV13, the hippocampal neurons were transfected, and at DIV21, spontaneous activity of the hippocampal neurons was recorded. The raw data files were recorded every 5 min using the Axion BioSystems Integrated Studio software (AxIS, version 3.0.2.1). All data were filtered using dual 200 Hz (high-pass) and 3,000 Hz (low-pass) Butterworth filters. An adaptive threshold spike detector was used to identify any amplitude exceeding six standard deviations (6×SD) above the estimated noise on each channel. The number of active electrodes and spikes within a 5 min recording period were quantified. Bursts were detected by measuring the interspike intervals (ISI). The threshold algorithm in the AxIS software required a minimum of five spikes per burst, with a maximum interspike interval set at 100 ms.
Animal behavioral tests
Animals were habituated to the behavior room 1 d prior to conducting experiments to alleviate stress. All behavior tests were video recorded and analyzed using computerized video tracking software (ANY-Maze or EthoVision 7.0).
Morris water maze test
The Morris water maze test was performed as previously described with slight modifications (Wang et al., 2023a). The mice were placed in a circular pool (diameter, 160 cm; height, 50 cm) containing a platform. Mice were allowed 90 s to reach a visible platform placed slightly above the water surface or were gently guided to the area given that they exceed the time limit. Four trials with random start quadrant positions were then conducted. Twenty-four hours after pretraining, the platform was submerged, and the mice had four trials per day for 5 d to find the platform. A probe test was subsequently conducted in which the submerged platform was removed, and the mice were given 60 s to find the target quadrant. The time spent in the target quadrant and the number of crosses over the previous platform location were recorded for analysis. Finally, the position of the platform was then changed to the opposite quadrant, and the tests were repeated.
Fear conditioning test
On Day 1, mice were placed in a conditioning chamber individually. A cue tone of 60 dB white noise (conditioned stimulus, CS) was presented for 20 s followed by a footshock (unconditioned stimulus, US) with an intensity of 0.5 mA for 2 s. The unconditioned stimulus is delivered within the last few seconds of the conditioned stimulus presentation. The CS–US pairing was conducted three times in total with a cycle interval of 1 min. After 24 h, the mice were placed back into the same conditioning chamber for 5 min without any stimulus for contextual fear recall. On the 3rd day, the mice were placed in an alternate conditioning chamber (with different visual and tactile cues) for 5 min in total, during which three tone presentations (60 dB white noise) were given for 20 s at an interval of 1 min between cue tones. Freezing response, defined as the absence of movement in any body part for 1 s, is automatically scored and used to measure fear memory.
Object location and recognition task
The objection location and recognition tests were performed as previously described (Wang et al., 2023a). Mice were habituated in the testing chamber for 10 min without any objects present on the first day. On the second day, the mice were placed in the testing chamber for 10 min containing: two identical objects facing the same chamber side for the location test or at opposite corner quadrants for the recognition test. On the third day, one of the familiar objects was moved to a diagonal position for the location test or was replaced with a new object for the recognition test. The mice were once again placed in the chamber and allowed to explore objects for 10 min. The time spent exploring the objects was then recorded.
T-maze test
The T-maze test was performed as previously described (Wang et al., 2023a). Mice were allowed to familiarize themselves with the T-maze (consisting of a start arm and two goal arms) for 5 min on the first day. On the second day, the mice were trained to move from the start arm to the right arm (familiar arm) and explore the area instead of the left arm (novel arm). On the third day, the mice were given the opportunity to explore both arms, and the time spent in the novel and familiar arms during a 5 min testing period was recorded.
Open field test
The open field test was performed as previously described to observe the mice exploratory behaviors in a novel environment (Wang et al., 2023b). The mice were allowed to explore the open field (48 cm × 48 cm × 30 cm) freely for 5 min. The total distance traveled and the time spent in the central zone of mice were then measured.
Elevated plus maze test
The elevated plus maze test was performed as previously described to observe the mice ambulatory behaviors in a novel environment (Wang et al., 2023b). Mice were placed in the plus-shaped maze that consisted of a central grid and enclosed arms for 5 min. The time spent and distance traveled in the open arms and closed arms, respectively, were recorded. The ratios of time spent in the open arms and time spent entering the open arms were evaluated.
Forced swimming test
The forced swim test was performed as previously described (Wang et al., 2023b). Mice were placed in a transparent cylindrical container (diameter, 15 cm; water depth, 15 cm) for 5 min. The immobility time was measured which reflects despair-like behaviors.
Tail suspension test
The mice were suspended in air by applying tape to the distal end of their tails. Mice were visually segregated from one another and were not allowed to escape or grab onto nearby surfaces. The time spent immobile within a 6 min period was then measured.
Quantification and statistical analysis
The number of animals and sample size of each experiment was listed in Table 2. Data are presented as mean ± SEM. The normal distribution of each dataset was determined using the Shapiro–Wilk normality test. For experimental groups that passed the normality test, statistical analysis was conducted using Student's t test when comparing two groups and one-way analysis of variance (ANOVA) followed by Tukey's post hoc test when comparing more than two groups. Statistical significance was defined as p < 0.05. Both detailed statistical methods and results were summarized in Table 3.
Detailed information on the values used for quantification in this study
Detailed information on the statistical methods and results for individual figures in this study
Results
LUZP1 is predominantly expressed in CaMKIIα-expressing neurons in the GCL of the DG
The importance of LUZP1 in the central nervous system (CNS) was primarily recognized due to the severe neural tube closure defects observed in LUZP1 knock-out embryos during brain development (Hsu et al., 2008). Previous studies have demonstrated that LUZP1 localizes to various subcellular locations, including actin stress fibers, where it plays a role in forming actomyosin bundles (Bozal-Basterra et al., 2020; Gonçalves et al., 2020; Yano et al., 2021; Wang et al., 2024). However, the expression profile and physiological function of LUZP1 within the brain remain to be defined. Here, we first examined the expression profile of LUZP1 in the mouse brain using the antibody targeting the LUZP1 protein. Analysis of the mammalian sequence of LUZP1 (https://www.ensembl.org/Homo_sapiens/Gene/Summary?db=core,g=ENSG00000169641,r=1:23084030-23178152 and https://www.ensembl.org/Mus_musculus/Gene/Summary?db=core,g=ENSMUSG00000001089,r=4:136197072-136282091) showed that the amino acid sequence of the LUZP1 protein in both humans and mice is encoded by two exons (Extended Data Fig. S1A, upper panel). The immunogen sequences of the anti-LUZP1 antibodies from Sigma (HPA028542) and Proteintech (17483-1-AP) are encoded by the same exon (Extended Data Fig. S1A, lower panel). Consistent with our previous results in U2OS cells (Wang et al., 2024), both antibodies recognize one single band in mouse N2a (also known as Neuro-2a) cells and hippocampal tissue, with very few weak nonspecific bands (Extended Data Fig. S1B). The immunofluorescent staining of endogenous LUZP1 in mouse hippocampal neuronal HT22 cells, using the LUZP1 antibody from Sigma, revealed that LUZP1 localized to actin stress fibers in the cytosol and the actin meshwork around the plasma membrane (Extended Data Fig. S1C). Hence, the LUZP1 antibody from Sigma was primarily used for the subsequent analysis, with the LUZP1 antibody from Proteintech applied in some Western blotting assays.
By dissecting different mouse brain regions, LUZP1 was shown to be ubiquitously expressed in the brain (Fig. 1A). Subsequent immunohistochemical staining of mouse whole-brain slice further confirmed the ubiquitous expression of LUZP1 in the mouse brain, with particularly high expression within the dentate gyrus (DG) region of the hippocampus (Fig. 1B). The DG of the hippocampus is composed of four distinct layers, the molecular layer (ML), mature granule neuron-enriched granular cell layer (GCL), neural progenitor cell-enriched subgranular zone (SGZ), and hilus (Rao and Shetty, 2004). To determine the sublayer distribution of LUZP1 in the DG region, the coexpression of LUZP1 with the SGZ protein Pax6, SGZ-to-inner GCL protein Doublecortin (Dcx), and the GCL protein NeuN were visualized (Kuipers et al., 2015). The immunostaining results revealed specifically high expression of LUZP1 within the excitatory granule neuronal densely enriched GCL, a brain region known for prominent CaMKIIα expression (Wang et al., 2013; Shah et al., 2016) and vital for regulating synaptic function (Fig. 1C,D; Lazarov and Hollands, 2016). Moreover, the expression of LUZP1 in primary hippocampal neurons was further examined at DIV21 through immunofluorescence staining. The results confirmed the expression of LUZP1 in neurons, along with neuronal dendrites and dendritic spines (Fig. 1E). Additionally, the partial colocalization of LUZP1 puncta with the synaptic proteins PSD95 and SV2 confirmed the presence of LUZP1 at synaptic terminals (Fig. 1E). Collectively, these data reveal that LUZP1 is ubiquitously expressed in the brain, with particularly high expression in CaMKIIα-expressing neurons in the GCL of the DG (Fig. 1F).
Localization of LUZP1 in the mouse brain. A, Expression of LUZP1 in cortex, striatum, caudate nucleus, and hippocampus of wild-type mice. B, Immunofluorescence staining of LUZP1 (green) and nuclear marker DAPI (blue) in whole brain slice of mice. Scale bar, 1,000 μm. C, Immunofluorescence staining of LUZP1 (green); marker proteins of SGZ and GCL, including Pax6, DCX, and NeuN (magenta); and DAPI (blue) in the hippocampal DG region with different layers shown, including ML, GCL, SGZ, and Hilus. Scale bar: 1,000 μm (DG region) or 50 μm (magnification). D, Immunofluorescence staining of LUZP1 protein (green), CaMKIIα (red), and DAPI (blue) in the DG of the hippocampus. Scale bar, 25 μm. E, Immunofluorescence images showing endogenous LUZP1 expression and colocalization with synaptic proteins in primary hippocampal neurons at DIV21. Scale bar, 10 μm. Line graph below showing the fluorescence intensity profiles of the indicated proteins. F, A schematic diagram showing the enriched distribution of LUZP1 in the GCL layer of the hippocampal DG region.
The expression level of LUZP1 coincides with neuronal maturation
To investigate the expression pattern of LUZP1 during the development of the central nervous system, LUZP1 protein levels in the hippocampus were assessed by Western blot analysis at various developmental stages, starting from embryos. The results revealed an upregulation of LUZP1 expression during mouse hippocampal development, particularly from the embryonic (E16 and E18) to postnatal (P0–5) stages (Fig. 2A,B). This increasing trend of LUZP1 expression in the hippocampus persisted significantly throughout postnatal development (Fig. 2C,D). Additionally, immunofluorescence staining of brain sections further confirmed the dynamic expression pattern of LUZP1 during hippocampal maturation, displaying a similar increased trend in the hippocampus after birth (Fig. 2E,F). Taken together, these results indicate that the temporal expression of LUZP1 coincides with the progression of hippocampal maturation.
Expression of LUZP1 coincides with dendritic spine maturation. A, B, Expression of LUZP1 was upregulated in the hippocampus at different developmental stages in wild-type embryos period (E16 and E18) and postnatal days (P0–5). The intensity of LUZP1 was normalized with that of α-Tubulin. Results were pooled from three independent experiments. Data were mean ± SEM, ***p < 0.001 for the comparison between E16 and P1, E16 and P3, or E16 and P5, one-way ANOVA, Tukey's multiple-comparison test. C, D, Expression of LUZP1 was upregulated in the hippocampus at different developmental stages in postnatal weeks (PW1–PW8). The intensity of LUZP1 was normalized with that of α-Tubulin. Results were pooled from three independent experiments. Data were mean ± SEM, ***p < 0.001 for the comparison between PW1 and PW2 or PW3 or PW4 or PW6 or PW8, one-way ANOVA, Tukey's multiple-comparison test. E, Immunofluorescence staining was performed to visualize LUZP1 protein (green) and DAPI (blue) in the DG region of the hippocampus at different developmental stages in postnatal days (P0 and P1) and weeks (PW1 and PW2). Scale bars: 100 μm (left) and 50 μm (right). F, The intensity of LUZP1 was calculated. Results were pooled from three independent experiments. Data were mean ± SEM, ***p < 0.001 for the comparison between P0 and PW1, P0 and PW2, one-way ANOVA, Tukey's multiple-comparison test. Sample size and statistical tests are reported in detail in Tables 2 and 3.
LUZP1 regulates dendritic spine maturation both in vitro and in vivo
The transition from filopodia to mushroom spines is an important hallmark of dendritic maturation (Khanal and Hotulainen, 2021). To investigate the role of LUZP1 in dendritic maturation, we generated a short hairpin RNA (shRNA) targeting mouse LUZP1 and assessed its knockdown efficiency in primary cultured hippocampal neurons. Western blot analysis revealed a significant decrease in the LUZP1 protein level after LUZP1-shRNA transfection (Fig. 3A). To examine the effect of LUZP1 knockdown on spine morphology, we transfected LUZP1-shRNA into primary hippocampal neurons at DIV13. Furthermore, an RNAi-resistant LUZP1 construct was introduced to assess whether its expression could rescue the aberrant phenotype induced by LUZP1-shRNA. After neuronal maturation at DIV21, we analyzed the densities of different spine types. Interestingly, transfection of LUZP1-shRNA resulted in a significant reduction in the density of mushroom spines but an increase in the density of filopodia in neurons. Compared with the control spines, the volume and diameter of the spine head decreased, while the length of the spine neck extended in the LUZP1-silenced neurons. This suggests a shrunken mushroom head with an extended protrusion. Moreover, this alteration in spine morphology can be partially rescued by re-expressing LUZP1 in the LUZP1-silenced neurons (Fig. 3B). These results indicate that LUZP1 deficiency can disturb the spine morphology of hippocampal neurons.
LUZP1 regulates dendritic spine maturation both in vitro and in vivo. A, Knockdown of LUZP1 by shRNA in cultured primary hippocampal neurons. Results were pooled from three independent experiments. Data were mean ± SEM, **p < 0.01, one-way ANOVA, Tukey's multiple-comparison test. B, Hippocampal neurons were cotransfected with the LUZP1-shRNA or control shRNA with or without the RNAi-resistant LUZP1 construct. Results were pooled from three independent experiments. The density of the individual spine type was quantified, and 38 dendrites in each group were quantified. The head volume, head diameter, and neck length of dendritic spines were analyzed in Imaris, and spine from 25–26 dendrites in each group were quantified. Data were mean ± SEM, *p < 0.05, ***p < 0.001, one-way ANOVA, Tukey's multiple-comparison test. Scale bar, 10 μm. C, Schematic diagram depicting the LUZP1 conditional knock-out strategy. The purple rectangles correspond to exon 1, exon 2, exon 3, exon 4, and exon 5 of the LUZP1 gene, respectively. The red triangles indicate the LoxP sites flanking exons 3 and 4. D, Expression of LUZP1 in the hippocampus of LUZP1flox/flox-CaMKIIα-creERT− control (LUZP1-C) and LUZP1flox/flox-CaMKIIα-creERT+ conditional knock-out (LUZP1-CKO) mice. Results were pooled from three independent experiments. Data were mean ± SEM, ***p < 0.001, Student's t test. E, Representative immunofluorescence images of LUZP1 expression in coronal sections of brain slices from LUZP1 conditional knock-out mice and control mice. Scale bars: 1,000 μm (whole coronal section) and 500 μm (magnification). F, AAV-sparse-NCSP-YFP-2E5 carrying GFP-expressing construct was stereotaxically injected into hippocampal DG area of LUZP1-C and LUZP1-CKO mice. Representative images of the GFP-positive secondary apical dendrites from hippocampal DG neurons of the injected mice (6–8 weeks old) were shown. Scale bar, 5 μm. Significant difference in mushroom spine density was observed between LUZP1-C and LUZP1-CKO mice. Results were obtained from three separate experiments, with three mice per condition per experiment. A total of 36 dendrites were quantified for each condition. Data were presented as mean ± SEM, **p < 0.01, ***p < 0.001, Student's t test. Sample size and statistical tests are reported in detail in Tables 2 and 3.
Since homozygous LUZP1 knock-out is embryonically lethal due to abnormal neural tube development (Hsu et al., 2008), we generated LUZP1 conditional knock-out (LUZP1flox/flox) mice using a Cre/loxP gene-targeting strategy to investigate the function of LUZP1 in vivo. As both CaMKIIα and LUZP1 are highly expressed in the neurons of the GCL within the DG (Tsien et al., 1996; Dragatsis and Zeitlin, 2000; Wang et al., 2013; Shah et al., 2016), we used a CaMKIIα promoter-driven CreERT2 transgenic line (CaMKIIα-CreERT2) and LUZP1flox/flox to generate LUZP1 conditional knock-out (LUZP1-CKO) mice (LUZP1flox/flox-CaMKIIα-CreERT2+), in which LUZP1 was selectively knocked out in CaMKIIα-expressing neurons after tamoxifen injection. LUZP1 control (LUZP1-C) mice (LUZP1flox/flox-CaMKIIα-CreERT2−) were also generated (Fig. 3C). We performed qRT-PCR, Western blotting, and immunofluorescent staining assays to validate the knock-out efficiency (Fig. 3D,E). To assess the impact of LUZP1 conditional knock-out on spine morphology in vivo, we examined the spine morphology of the secondary apical dendrites of hippocampal DG neurons in 8-week-old mice. Although the densities of filopodia spines were similar between LUZP1-C and LUZP1-CKO mice, the densities of mushroom spines were notably decreased in LUZP1 knock-out neurons (Fig. 3F). These findings both in vivo and in vitro indicate that LUZP1 can affect the dendritic spine maturation of hippocampal neurons, thus suggesting LUZP1 may be involved in the maturation of spines.
Impaired synaptic structure and plasticity in LUZP1 conditional knock-out mice
Spine maturation is critical for maintaining normal synaptic structure and plasticity in neurons and is often correlated with stronger and more stable synaptic connections (Holtmaat and Svoboda, 2009; Kasai et al., 2021). To investigate the changes in synaptic architecture at the subcellular level resulting from LUZP1 depletion, we analyzed the morphology of neuronal synapses in the DG of LUZP1-C and LUZP1-CKO mice by transmission electron microscopy (Fig. 4A). We observed a significant decrease in postsynaptic density (PSD) thickness and length (Fig. 4B,C), indicating impaired structural integrity of the postsynaptic compartment. Additionally, the width of the synaptic cleft was significantly increased (Fig. 4D), suggesting an expansion of the gap between the pre- and postsynaptic elements.
Impaired synaptic structure and plasticity in LUZP1 conditional knock-out mice. A, Represented transmission electron microscopy image of synapse of the hippocampus neurons in the DG of LUZP1flox/flox-CaMKIIα-creERT− control (LUZP1-C) and LUZP1flox/flox- CaMKIIα-creERT+ conditional knock-out (LUZP1-CKO) mice. Scale bar, 500 nm. B, D, Cumulative frequency curve showing significant decrease in the PSD length, thickness, and synaptic cleft width of hippocampal neuron spines in LUZP1-C and LUZP1-CKO mice. Four mice were imaged for each condition, and the total number of synapses for quantification is shown in the graph. *p < 0.05, ***p < 0.001, Student's t test. E, Circuitry of the hippocampus during LTP. For this model of LTP, medial perforant path (MPP) acted as presynaptic portion, while DG served as the postsynaptic region. F, Represented pictures of the field potential recording of the two groups of animals before and after HFS. G, The normalized fEPSP slope versus time in the DG area of the hippocampal slices of the two groups of animals (n = 8). H, The normalized average fEPSP slopes for the first 10 min after HFS stimulation in both groups of animals. LTP induction rate was 1.564 ± 0.048 in LUZP1-C mice and 1.421 ± 0.043 in LUZP1-CKO mice, p = 0.0442. Data were presented as mean ± SEM, Student's t test. Sample size and statistical tests are reported in detail in Tables 2 and 3.
Spine maturation is closely linked to synaptic plasticity and is more likely to undergo synaptic potentiation, such as long-term potentiation (LTP; Lüscher et al., 2000; Paulsen and Sejnowski, 2000). We then measured LTP in hippocampal slices to investigate the changes in synaptic plasticity resulting from LUZP1 conditional knock-out. Here, the perforant path serves as the presynaptic component, while the DG functions as the postsynaptic component (Fig. 4E). Field potential recordings were performed before and after high-frequency stimulation (HFS), and noticeable differences were observed between the two groups (Fig. 4F). Compared with that in the control group, the slope of the field excitatory postsynaptic potential (fEPSP) versus time in the DG in the LUZP-CKO group was significantly altered (Fig. 4G), and the LTP altered ratio was significantly reduced (Fig. 4H). These results show that the loss of LUZP1 can impair synaptic plasticity and specifically hinder LTP in synapses.
LUZP1 knockdown reduces the spontaneous electrical activity of hippocampal neurons
Reduced synaptic maturation and plasticity can diminish excitatory or inhibitory synaptic strength, resulting in imbalanced regulation of neural activity (Desai, 2003). To evaluate the impact of LUZP1 knockdown on the electrophysiological properties of neurons, we performed MEA recordings to capture the spontaneous electrical activity of primary hippocampal neurons in vitro. Hippocampal neurons isolated from mouse embryos were evenly seeded in the wells of 24-well MEA plates. Neurons were transfected with LUZP1-shRNA or control shRNA at DIV13 (Fig. 5A). At DIV21, the spontaneous electrical activity of the neurons was recorded. A heatmap of the real-time spike firing rate data revealed a decreasing trend in LUZP1-knockdown neurons (Fig. 5B). We further quantitatively assessed the electrophysiological activity of the neurons. LUZP1 knockdown significantly reduced the number of spikes, number of active electrodes, burst duration, network burst frequency and percentage, and number of network bursts in primary hippocampal neurons (Fig. 5C), indicating that LUZP1 knockdown can decrease the frequency of synchronous firing and the propensity for burst firing patterns. Furthermore, this effect also extends to the entire neuronal network, impacting burst dynamics and firing patterns across interconnected neurons. Consequently, the synchrony index for all active neurons within a time window was also evidently decreased and the representative raster plots displayed a noticeable decrease in spiking activity at all electrodes in LUZP1-knockdown neurons (Fig. 5D). These findings further support the notion that LUZP1 downregulation has a profound impact on spontaneous neuronal network activity, which are consistent with all our in vitro studies. Overall, the impaired neuronal electrophysiological activity observed upon LUZP1 depletion highlights the critical role of LUZP1 in neuronal functional plasticity.
LUZP1 knockdown reduces the spontaneous electrical activity of primary hippocampal neurons. A, Example diagram of microelectrode arrays (MEA). Neurons are plated and cultured over the electrodes. B, The primary hippocampus neurons were transfected with LUZP1-shRNA or control shRNA. Heatmap was generated to illustrate the representative real-time spike firing rate during MEA recording. The firing rate was represented by color-coding, where white or red indicated a high firing frequency, while blue or black indicated a low firing frequency. C, The number of spikes, number of active electrodes, burst duration, network burst frequency and percentage, and number of network bursts were quantified per well during a 300 s recording period. Values were presented as the mean ± SEM of three individual experiments. Significance level: *p < 0.05, **p < 0.01, when compared with control group by Student's t test. Sample size and statistical tests are reported in detail in Tables 2 and 3. D, Raster plots displaying representative spikes were generated for a 30 s recording period (from 0 to 30 s) during a total recording time of 300 s. Each vertical line on the plot represents an active spike, with all the vertical lines in each row representing the detected spikes from one electrode. A total of 16 electrodes were used in each well, resulting in 16 rows being displayed.
Impaired learning and memory abilities in LUZP1 conditional knock-out mice
Synaptic plasticity is a fundamental process underlying learning, memory formation, and cognitive function in which the hippocampus plays a crucial role (Kim and Diamond, 2002). We assessed the influence of LUZP1 conditional knock-out in hippocampal CaMKIIα-expressing neurons on the learning and memory functions of mice through a battery of behavioral paradigms. In addition, we implemented another approach to reduce hippocampal LUZP1 expression by infusing Lenti-LUZP1 shRNA into the DG. Through Western blotting and qRT-PCR, we confirmed the knockdown of LUZP1 in 8-week-old mice (Extended Data Fig. S2A). The Morris water maze test showed that compared with control mice, both LUZP1-CKO mice and LUZP1-knockdown (abbreviated as LUZP1.sh) mice spent more time finding the platform (Fig. 6A–F, Extended Data Fig. S2C). In the fear conditioning test, both types of LUZP1-deficient mice exhibited shorter freezing times in both the context and cue tone tests, indicating impaired fear learning and memory (Fig. 6G–J, Extended Data Fig. S2D). In the object recognition task test, LUZP1-deficient mice exhibited reduced discrimination and less preference for the novel object (Fig. 6M,N; Extended Data Fig. S2E); moreover, the exploration time for objects in novel positions was also decreased (Fig. 6K,L; Extended Data Fig. S2F). In the T-maze test, LUZP1-deficient mice also exhibited impaired spatial memory, as they spent less time in the novel open arm (Fig. 6O, Extended Data Fig. S2G). Collectively, these results indicate that LUZP1 deficiency in the hippocampus significantly impaired spatial and recognition memory, highlighting the importance of LUZP1 in synaptic plasticity and cognitive processes.
Impaired learning and memory abilities in LUZP1 conditional knock-out mice. A, The performance of the mice in the Morris water maze test. The latencies to find the platform in the two groups over 5 consecutive training days (D1–5). Two-way ANOVA test. B, Representative swimming paths of LUZP1flox/flox-CaMKIIα-creERT− control (LUZP1-C) and LUZP1flox/flox-CaMKIIα-creERT+ conditional knock-out (LUZP1-CKO) mice during the training days. C, MWM probe test showed percent time spent in different quadrants of LUZP1-C and LUZP1-CKO mice. Two-way ANOVA followed by Bonferroni's multiple-comparisons test. The distance traveled in the target quadrant and the latency for two groups of mice to reach the target quadrant on the test day (D7). Student's t test. D, After changing the platform to the opposite position, the latencies to find the platform in the two groups over five consecutive reversal training days (D8–12). Two-way ANOVA test. E, Representative swimming paths of the LUZP1-C and LUZP1-CKO mice during the reversal training days. F, MWM probe test showed percent time spent in different quadrants of LUZP1-C and LUZP1-CKO mice. Two-way ANOVA followed by Bonferroni's multiple-comparisons test. The distance traveled in the target quadrant and the latency for two groups of mice to reach the target quadrant on the test day (D14). Student's t test. n = 9 for each group. G, Schematic of the fear-conditioning protocol. H, Freezing response to the training context. No significant difference was found between LUZP1-C and LUZP1-CKO mice. Student's t test. I, Freezing response to the conditioned room context. The percentage of freezing time of LUZP1-CKO mice was significantly decreased compared with the control group. Student's t test. J, Freezing response to the conditioned cue tone context. LUZP1-CKO mice froze significantly more than the control group. Student's t test. n = 10 for each group. K, Diagram of object location task. L, Decreased discrimination index (DI) of LUZP1-CKO mice compared with the control group in object location task. Student's t test. n = 10 for each group. DI was used to evaluate the scores of exploration time for the old location or new location object as DI = (TN − TO) / (TN + TO) × 100%. M, Diagram of object recognition task. N, Reduced DI of LUZP1-CKO mice compared with the control group in object recognition task. Student's t test. n = 10 for each group. DI was used to evaluate the scores of exploration time for the familiar or novel object as DI = (TN − TF) / (TN + TF) × 100%. O, In the T-maze test, time and distance of LUZP1-CKO mice spent in new arms was significantly reduced compared with the control group. Student's t test. n = 10 for each group. Data were presented as mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001. Sample size and statistical tests are reported in detail in Tables 2 and 3.
To further explore the anxiety- or depression-related phenotypes of LUZP1-deficient mice, we also performed the elevated plus maze test, open field test, forced swimming test, and tail suspension test. However, LUZP1-deficient mice showed no substantial changes in these behaviors (Extended Data Fig. S3A–J). Taken together, these data indicate that the enriched expression of LUZP1 in the hippocampus is pivotal for spatial and recognition-related cognition.
LUZP1 interacts with FLNA and modulates the Rac1-PAK1 signaling pathway to regulate spine maturation
To further elucidate the molecular mechanism of impaired synaptic structure and plasticity resulting from LUZP1 depletion, we employed RNA-seq analysis to explore the signaling pathways involved in LUZP1 knockdown in hippocampal neurons. The Venn diagram shows the number of coexpressed and uniquely expressed genes between the two groups (Fig. 7A). Gene expression levels were normalized using fragments per kilobase of transcript per million mapped reads (FPKM; Trapnell et al., 2010), and significance analysis (1.2-fold change and padjust < 0.05) was conducted to identify differentially expressed genes (DEGs; Table 4, Extended Data Table S1-4). Among a total of 1,130 genes, 281 genes were upregulated, while 849 genes were downregulated in the LUZP1-knockdown neurons (Fig. 7B,C). A heatmap was used to visualize the extent of changes in these significantly altered genes (Fig. 7D). To gain insight into the functions of these up- and downregulated DEGs, we performed functional enrichment analysis using KEGG pathway analysis. The results revealed a significant enrichment of genes related to the nervous system and neurodegenerative diseases. Specifically, the cAMP signaling pathway, axon guidance, and cell adhesion molecules, including filamin alpha (Flna), p21-activated kinase 1 (Pak1), Rac family small GTPase 1 (Rac1), and cofilin (Cfl), were strongly represented (Fig. 7E,F). Moreover, changes in these genes were additionally identified in primary hippocampal neurons after knockdown of LUZP1. Compared with the control shRNA-treated neurons, the mRNA level of Flna was found to be remarkably downregulated in the LUZP1 shRNA-treated neurons (Fig. 7G).
RNA-seq analysis of LUZP1-deficient hippocampal neurons. A, Primary hippocampal neurons were transfected with LUZP1-shRNA or control shRNA. To visualize the overlapping relationship of all genes between the two groups, a Venn diagram was generated. B, C, The log2 fold change of RefSeq gene exons was assessed, and the corresponding significance values are represented as -log10 (P adjust). Genes that are upregulated (shown in red) or downregulated (shown in blue) by LUZP1 shRNA are depicted. D, The heatmap displays gene types that have undergone significant modifications (VIP ≥ 1, p < 0.05), and the log2-fold change ratio for each gene is represented by colored bars. E, The up- and downregulated DEGs within the gene set were classified according to the pathways they are involved in or the functions they perform using the KEGG database. F, When the adjusted p value (p adjust) <0.05, it is considered that there is a significant enrichment of this KEGG pathway function. G, Relative expression levels of selected LUZP1-regulated genes from the RNA-seq dataset for LUZP1 shRNA-treated and control neurons. Data are presented as the means ± SEM (n = 3), *p < 0.05, **p < 0.01, two-way ANOVA followed by Bonferroni's multiple-comparisons post hoc test. H, Co-immunoprecipitation (co-IP) showing the interaction between LUZP1 and FLNA in hippocampal tissue. LUZP1 immunoprecipitation pulled down FLNA, and vice versa, confirming their bidirectional interaction. I, Western blot analysis demonstrating that LUZP1 knockdown reduces FLNA expression, while LUZP1 overexpression rescues FLNA protein levels in hippocampal neurons. Results were pooled from three independent experiments. Data were presented as mean ± SEM, **p < 0.01, ***p < 0.001, one-way ANOVA, Tukey's multiple-comparison test. J, Left, Immunofluorescence images showing endogenous LUZP1 colocalization with Filamin A in primary hippocampal neurons at DIV21. Actin was visualized by the phalloidin probe (magenta). Scale bar, 10 μm. Right, Line graphs showing the fluorescence intensity profiles of the LUZP1 and Filamin A. Sample size and statistical tests are reported in detail in Tables 2 and 3.
To directly demonstrate the interaction between LUZP1 and FLNA in the hippocampus, we performed a co-IP experiment, which further confirmed their interaction (Fig. 7H). Also, consistent with the downregulated transcriptional level of Filamin A (abbreviated as FLNA), the protein level of FLNA was also profoundly decreased in the LUZP1 shRNA-treated neurons compared with control neurons, and LUZP1 overexpression restored FLNA protein levels (Fig. 7I). As an actin cross-linking protein, FLNA has been reported to interact with LUZP1 for mechanosensing and mechanotransduction (Wang and Nakamura, 2019). Consistent with previous reports, the immunostaining results showed the colocalization of LUZP1 with FLNA in primary hippocampal neurons and mouse hippocampal neuronal HT22 cells (Fig. 7J, Extended Data Fig. S4A). Taken together, these data suggest that LUZP1 may interact with FLNA to play a role in synaptic maturation of neuron.
Next, to investigate the role of FLNA in dendritic spine maturation, primary hippocampal neurons were treated with FLNA-shRNA. FLNA-shRNA efficiently knocked down the expression of FLNA at both the mRNA and protein levels (Fig. 8A). LUZP1 knockdown reduced mature spines and increased immature ones as before. However, LUZP1 overexpression failed to restore maturation when FLNA was also knocked down, confirming their cooperative function (Fig. 8B). We then cotransfected either FLNA-shRNA or control shRNA into primary hippocampal neurons at DIV13. Importantly, FLNA-shRNA markedly reduced the density of mushroom spines but increased the density of filopodia in neurons at DIV21. Moreover, to investigate whether these changes in neurite morphology were caused by the loss of FLNA, we introduced an RNAi-resistant FLNA (FLNA-shRNAir) construct to neurons treated with FLNA-shRNA. Similar to LUZP1 knockdown neurons, the volume and diameter of the spine head decreased upon the loss of FLNA. Additionally, FLNA knockdown resulted in a significant elongation of the spine neck, further supporting a transition toward a more immature spine phenotype. In contrast to the filopodia-like spines observed in FLNA-knockdown neurons, FLNA re-expression partially reversed spine morphology, restoring filopodia-like spines to a more mature mushroom-like structure (Fig. 8C,D). These consistent findings indicate that LUZP1 may regulate synaptic structure and function at least in part through an FLNA-dependent mechanism.
LUZP1 interacts with FLNA and modulates Rac1-PAK1 signaling pathway to regulate spine maturation. A, Knockdown of FLNA by shRNA in cultured hippocampal neurons. Results were pooled from three independent experiments. Data were presented as mean ± SEM, ***p < 0.001, Student's t test. B, Representative images of hippocampal neurons co-transfected with either LUZP1-shRNA or control shRNA, with or without an RNAi-resistant LUZP1 construct and FLNA-shRNA. Despite LUZP1 overexpression, spine maturation was not restored when FLNA was simultaneously knocked down. Results were pooled from three independent experiments. The head volume, head diameter, and neck length of dendritic spines were analyzed in Imaris, and spine from 30 dendrites in each group were quantified. Data were presented as mean ± SEM, *p < 0.05, **p < 0.01, ***p < 0.001, one-way ANOVA, Tukey's multiple-comparison test. Scale bar, 5 μm. C, D, Hippocampal neurons were cotransfected with the FLNA-shRNA or control shRNA with or without the RNAi-resistant FLNA construct. Results were pooled from three independent experiments. The density of the individual spine type was quantified, 40 dendrites in each group were quantified. The head volume, head diameter, and neck length of dendritic spines were analyzed in Imaris, and spine from 25 dendrites in each group were quantified. Data were presented as mean ± SEM, *p < 0.05, **p < 0.01, ***p < 0.001, one-way ANOVA, Tukey's multiple-comparison test. Scale bar, 10 μm. E, Increased PAK1 phosphorylation in LUZP1 knockdown neurons. Western blot showing the phosphorylated PAK1 (p-PAK1/2) and total PAK1 in two groups. Results were pooled from three independent experiments. Data were mean ± SEM, **p < 0.01, Student's t test. Western blot analysis was conducted to assess the levels of activated Rac1 (Rac1-GTP) and total Rac1 in neurons transfected with LUZP1-shRNA or control shRNA. An increased ratio of Rac1-GTP to total Rac1 was observed in the LUZP1-shRNA group compared with the control shRNA group. Results were pooled from three independent experiments. Data were presented as mean ± SEM, **p < 0.01, Student's t test. Neurons transfected with LUZP1-shRNA exhibited an increased phosphorylated cofilin (P-cofilin) to cofilin ratio compared with those transfected with control shRNA. Results were pooled from three independent experiments. Data were mean ± SEM, ***p < 0.001, Student's t test. F, Analysis of total phospho-cofilin intensity in the absence and rescue of LUZP1. Top, Representative example of dendrites contained for phosphorylated cofilin (P-cofilin). Scale bars: 5 μm. Bottom, Quantification of average phosphorylated cofilin concentrations in dendritic spine heads integrated immunostaining intensity within each spine was divided by spine area. a.u., arbitrary units. Results were pooled from three independent experiments. Spines from 30 dendrites in each group were quantified. Data were presented as mean ± SEM, *p < 0.05, **p < 0.01, one-way ANOVA, Tukey's multiple-comparison test. G, A schematic diagram showing the mechanism underlying LUZP1 modulation in the dendritic spine. LUZP1 contains binding domains: F-actin binding site (purple region, 400–500 aa) and two FLNA-binding sites (green region, residues 833–883). Sample size and statistical tests are reported in detail in Tables 2 and 3.
The morphology and synaptic function of spines are closely modulated by the concerted actions of multiple actin-binding proteins, including PAK1 (Hoi et al., 2020), one of the proteins that interact with FLNA (MacPherson and Fagerholm, 2010). We continued to investigate whether downregulated FLNA would further affect PAK1-related signaling pathways in LUZP1 knock-out neurons. Interestingly, the levels of phosphorylated PAK1, phosphorylated cofilin, and LIM kinase-Rac1-GTP (Fig. 8E) were significantly increased. We further confirmed an increase in p-Cofilin levels in dendritic spines following LUZP1 knockdown (Fig. 8F), suggesting that enhanced actin stabilization promotes the formation of filopodia-like spine morphology. These findings showed that the loss of LUZP1 and the downregulation of FLNA may affect the Rac1-PAK1 signaling pathway, thus dysregulating synaptic structure and plasticity (Fig. 8G).
Discussion
Although previous studies have shown that LUZP1 likely plays a pivotal role in the brain, its precise function remains largely unknown. Our study revealed that LUZP1 is highly coexpressed with actin in hippocampal CaMKIIα-expressing neurons and that its expression increases temporally throughout neuronal development. To explore the function of LUZP1, by using the Cre-loxP approach, we established a mouse model in which LUZP1 was specifically knocked out in CaMKIIα-expressing neurons. By delineating the function of LUZP1 via shRNA or gene knock-out, we demonstrated the critical role of LUZP1 in regulating dendritic spine maturation and neuronal activity both in vivo and in vitro. Our findings provide compelling evidence highlighting the importance of LUZP1 in synaptic maturation, plasticity, and cognition in the brains of mice.
During early development of the hippocampus after birth, dendritic spines gradually mature from filopodia-like protrusions to enlarged mushroom spines (Fiala et al., 1998; Kasai et al., 2021). The distinct morphologies of dendritic spines play crucial roles in determining their properties and functions. Mushroom spines, characterized by larger PSDs, are associated with enhanced synaptic strength, whereas thin spines exhibit dynamic characteristics and are transient in nature (Berry and Nedivi, 2017). In response to learning paradigms, thin spines can remodel into mushroom spines and contribute to synaptic specificity for learning and memory (Bourne and Harris, 2007). We noticed that LUZP1 knockdown decreased mushroom spine density but increased filopodia density in hippocampal neurons, indicating that LUZP1 can structurally regulate neurodevelopment by promoting dendritic spine maturation. LUZP1 has been reported to be involved in embryonic development, and LUZP1 KO embryos exhibit cranial neural tube closure defects, resulting in embryonic death (Lee et al., 2001; Hsu et al., 2008). Moreover, LUZP1 is implicated in certain developmental disorders, such as 1p36 deletion syndrome (Jordan et al., 2015). Indeed, brain developmental abnormalities, the phenotypes presented in LUZP1 KO mice, align with the clinical manifestations of patients with 1p36 deletion syndrome (Hsu et al., 2008; Jordan et al., 2015).
The maturation of dendritic spines is intricately linked to the development and maintenance of synaptic structures (Ethell and Pasquale, 2005; Yoshihara et al., 2009). Spines undergo significant structural changes during maturation, including the stabilization and enlargement of their heads with an increased PSD where neurotransmitter receptors are located (Yoshihara et al., 2009; Bosch and Hayashi, 2012). This facilitates the formation of precise and functional excitatory synapses between neurons, thereby ensuring efficient neurotransmission and synaptic plasticity (Hering and Sheng, 2001; Harris and Kater, 2003). We observed significant alterations in the synaptic structure of hippocampal neurons lacking LUZP1; these neurons exhibited decreased thickness and length of the postsynaptic density of dendritic spines but increased width of the synaptic cleft within the synapse. The aberrant maturation of dendritic spines may result in both a reduced surface area available for synaptic transmission and dysregulated communication between hippocampal neurons. Indeed, we observed a significant reduction in LTP in DG neurons upon LUZP1 knock-out.
Reduced synaptic maturity and plasticity can impact spontaneous electrical activity as well as the overall activity dynamics of neurons (Lohmann and Kessels, 2014). We found that depletion of LUZP1 markedly reduced synchronous firing and burst firing in hippocampal neurons. There are two possible reasons for this phenomenon. First, mature and plastic synapses enhance signal transmission efficiency. In contrast, decreased maturity and plasticity, as observed here in LUZP1-knockdown neurons, can weaken synaptic strength, thus decreasing spontaneous electrical activity (Gonzalez-Islas and Wenner, 2006). Second, decreased maturity and plasticity may disrupt the excitation–inhibition balance, leading to changes in activity patterns and spontaneous electrical activity (Gafarov, 2018).
CaMKIIα plays a vital role in synaptic plasticity and memory (Zalcman et al., 2018; Tao et al., 2021) and is highly expressed in the pyramidal neurons of the DG (Tsien et al., 1996; Dragatsis and Zeitlin, 2000; Wang et al., 2013; Shah et al., 2016). We observed that the knock-out of LUZP1 in CaMKIIα-expressing neurons impaired the cognitive functions of mice but did not induce anxiety or depression. Consistent with our findings, mice lacking the actin-binding proteins cortactin or drebrin A also exhibited impaired hippocampus-dependent spatial and recognition memory but displayed normal anxiety-like or depression-like behaviors (Cornelius et al., 2021). Since LTP strengthens synaptic connections and facilitates the formation and consolidation of memories in the hippocampus (Kullmann and Lamsa, 2007), the decreased LTP may be attributed to the impaired learning and memory observed in LUZP1 CKO mice in this study.
LUZP1 is able to bind to FLNA through its binding sites between residues 833 and 883, and its interaction with FLNA was previously discovered in mouse embryonic fibroblasts (Wang and Nakamura, 2019) as well as human cells (Gonçalves et al., 2020). Interestingly, the reported phenotypes of FLNA conditional knock-out mice (Feng et al., 2006), including brain development abnormalities, are similar to those observed in LUZP1 KO mice. FLNA is known to interact with the cytoplasmic tails of integrins, which are crucial for synapse development and function in the nervous system (MacPherson and Fagerholm, 2010). This interaction also plays a role in synaptic plasticity and is implicated in learning, memory, and cognitive processes (MacPherson and Fagerholm, 2010; Park and Goda, 2016). Here, we showed that LUZP1 interacted with FLNA in neurons and that LUZP1 knockdown decreased the expression of FLNA; moreover, FLNA knockdown led to immature dendritic spine morphology in primary hippocampal neurons. These results suggested that LUZP1 may regulate synapse maturation through an FLNA-dependent mechanism. We hypothesized that the knockdown of LUZP1 may weaken its interaction with FLNA, causing instability of the FLNA or FLNA-orchestrated cytoskeleton, thus ultimately resulting in a reduction in FLNA levels. The mechanisms underlying the interaction between LUZP1 and FLNA warrant further investigation.
The large dimeric actin-binding protein FLNA interacts with multiple proteins, including transmembrane cell receptor integrin and cytosol-localized PAK1 (Popowicz et al., 2006; MacPherson and Fagerholm, 2010). These interactions are critical for cytoskeletal rearrangement and membrane dynamics (MacPherson and Fagerholm, 2010), and one candidate we investigated was Rac1-PAK1 signaling. FLNA interacts with PAK1 at membrane ruffles, and this interaction is essential for orchestrating actin cytoskeletal rearrangement (MacPherson and Fagerholm, 2010). In LUZP1-deficient hippocampal neurons, we observed notable increases in the levels of phosphorylated PAK1, along with its downstream factor, phosphorylated cofilin, an actin-binding protein involved in cAMP signaling, and its upstream factor, Rac1-GTP. Indeed, dysregulation of the Rac1-PAK1 signaling pathway is associated with the excessive presence of elongated filopodia dendritic protrusions observed in neurodegenerative and neurodevelopmental diseases, such as AD and Fragile X syndrome (Heredia et al., 2006; Kashima et al., 2016; Pyronneau et al., 2017). These results are consistent with our findings, which showed that LUZP1 knockdown leads to the aberrant formation of immature dendritic spines on hippocampal neurons. Although clinical evidence of LUZP1 mutations in humans has not been reported, our results suggest that the dysregulation of filopodia spines in LUZP1-deficient hippocampal neurons may attribute to impaired neurodevelopment and cognitive ability. However, further experiments are needed to clarify the role of Rac1-PAK1 signaling in mediating the effect of LUZP1. In conclusion, our findings reveal the important role of LUZP1 in regulating dendritic spine maturation and synaptic function.
Footnotes
This work was partially supported by the National Natural Science Foundation of China (Grants 82371498, 82071494, 81272459), the “1·3·5 Project for Disciplines of Excellence (ZYGD23011), West China Hospital, Sichuan University,” the China Postdoctoral Science Foundation (No. 2021M702362), the Post-Doctor Research Project, West China Hospital, Sichuan University (No. 2020HXBH010), and Sichuan Science and Technology Program (No. 23NSFSC2884). We are super grateful to Yixiao Jiang for his encouragement and support throughout the research.
↵*X.W. and L.W. contributed equally to this work.
The authors declare no competing financial interests.
- Correspondence should be addressed to Xiaobo Cen at xbcen{at}scu.edu.cn.
