LRP1 in Adult-Born Neural Stem Cells Modulates Neurogenesis and Hippocampal Memory

Highlights

What are the main findings?

• Using a mouse LRP1 knockout model, we show that LRP1 loss in neural stem cells leads to impairments in hippocampal-dependent memory.
• LRP1 knockout reduced dendritic complexity in newborn neurons and increased the total number of adult-born hippocampal neurons in 10-month-old mice.

What are the implication of the main findings?

• It reveals the essential role of LRP1 in neural stem cell biology and hippocampal neurogenesis.
• It implicates a role for LRP1 in hippocampal dysfunction that could contribute to multiple disorders, emphasizing the potential for targeting LRP1 in therapeutic strategies.

Abstract

(1) Background: Adult neurogenesis within the hippocampus modulates hippocampal memory and is often dysregulated in diseases that cause memory dysfunction, notably Alzheimer’s disease. We have discovered a novel modulator of hippocampal neurogenesis—low-density lipoprotein receptor-related protein 1 (LRP1). (2) Methods: Using an inducible knockout of LRP1, male and female mice were subject to loss of LRP1, specifically in adult-born neural stem cells at 3 months of age. (3) Results: After 6 months with the knockout, animals without LRP1 in adult-born neural stem cells displayed behavioral phenotypes consistent with deficits in working memory and hippocampal-mediated spatial memory. We also found that over time, increasing numbers of adult-born LRP1-knockout neurons were present, although those neurons were morphologically less complex with fewer dendrites than controls. Our data suggest that the increase in the total number of adult-born neurons 6 months after knockout is due to a subtle increase in hippocampal proliferation over time. (4) Conclusions: Altogether, our data suggest that LRP1 is an important and previously unknown regulator of hippocampal neurogenesis.

1. Introduction

The hippocampus plays a major role in memory formation by integrating signal input from other brain regions, such as the entorhinal cortex, prefrontal cortex, amygdala and hypothalamus. The hippocampus is one of two established sites in the adult brain that can generate new neurons; this adult neurogenesis is vital to hippocampal function [1,2]. Here, the neurogenic niche is located within the subgranular zone (SGZ) of the dentate gyrus [1]. Normal hippocampal function requires that new neurons arise, develop, migrate and incorporate into the hippocampal circuitry [3,4]. This process is tightly regulated and typified via progression from the largely quiescent adult neural stem cell (NSC or Type 1) pool to the highly proliferative precursor (transit-amplifying) cells (Type 2). The progeny from Type 2 cells develop into neuroblasts (Type 3 cells) when they lose the expression of stem cell markers, such as GFAP and Sox2, but gain expression of neuronal differentiation markers, such as Prox-1, doublecortin (DCX) and polysialylated neural cell adhesion molecule (PSA-NCAM). These neuroblasts then migrate into the granule cell layer (GCL) to become mature granule cells and integrate into the circuitry, receiving input from the entorhinal cortex and also projecting axons into the CA3 region of the hippocampus as mossy fibers. During these unique developmental stages, developing neurons can also undergo natural culling and have somewhat high natural rates of apoptosis [1].

Studies show that when the development and incorporation of newborn neurons are disrupted, hippocampal-based memory is negatively affected [5,6]. Moreover, enhancement of hippocampal neurogenesis improves and preserves some forms of memory retention [7,8]. Notably, the use of focal irradiation to deplete adult hippocampal neurogenesis resulted in impairments in contextual fear conditioning and long-term potentiation in the dentate gyrus [5]. Additionally, killing NSCs impairs spatial learning and memory [6]. In contrast, neurogenesis can be enhanced via environmental enrichment and exercise [9,10]. In rodents, environmental enrichment increased dentate gyrus neurogenesis, which correlated with better performance in the Morris Water Maze [11,12] and novel object recognition [13]. Furthermore, mice subjected to voluntary wheel running had improved adult hippocampal neurogenesis and memory, as measured by Morris Water Maze and spatial pattern separation tests [14,15]. Altogether, studies strongly support that active hippocampal neurogenesis is important for cognition.

We have discovered a novel regulator of adult hippocampal NSCs in mice. We previously described the effect of LRP1 knockout (KO) in adult NSCs within the context of ischemic damage [16]. We were specifically interested in LRP1 for its known links to neurodegenerative diseases [17,18,19,20] and ability to alter signal regulation [21,22]. Safina and others have shown the effects of LRP1 loss on NSCs in vitro [23], but we are the first to report how LRP1KO alters NSCs in vivo [16]. Given our previous findings within the subventricular zone, the overall goal of this study was to investigate the effects of LRP1 loss in adult hippocampal neurogenesis. Using a transgenic mouse model that utilizes a nestin-driven inducible Cre recombinase, we initiated LRP1KO in adult NSCs at 3 months of age. We discovered that 6 months following NSC LRP1KO, hippocampal-dependent memory deficits occurred. Further, at this timepoint we identified an elevation in neuronal lineage markers of LRP1KO cells, including both immature neurons expressing doublecortin (DCX) and neuronal nuclear marker (NeuN), as well as mature neurons only expressing NeuN. The mature neurons also displayed reduced morphological complexity, which could account for the perceived alterations in hippocampal-driven behavior. Consistent with this increased number of mature newborn neurons by 10 months of age, we observed that LRP1KO induced increased proportions of proliferating cells in the hippocampus. We similarly observed reduced levels of programmed cell death in adult-born cells 1 month after LRP1KO, but only within mice exposed to behavioral testing, suggesting a possible effect on stress-induced hippocampal cell death. Our findings suggest that LRP1 is a previously unidentified yet vital player in hippocampal neurogenic functions.

2. Materials and Methods

2.1. Animals

All animal procedures were approved by the UT Health Institutional Animal Care and Use Committee (IACUC) in accordance with NIH guidelines. Vivarium conditions included 12 h light/dark cycle with ad libitum food and water. As previously described [16], adult NSC-specific LRP1 knockout (LRP1KO) mice were generated by crossing Nestin-CreERT2 (Jackson stock 016261, Bar Harbor, ME, USA) to induce Cre recombinase in NSCs, tdTomato Ai14 reporter mice (tdTomato-stopfl/fl stock 007914), and LRP1 floxed mice (stock 012604). Knockout was induced at 3 months of age with daily IP tamoxifen (100 μL of 20 mg/mL dissolved in corn oil) administered for 5 days. Both males and females were used in the experiments; however, occasionally, aggressive mice had to be housed separately. Given the known role that single housing has on neurogenesis [24], these mice were excluded from the study (12 single-housed mice of 83 total used (14%); 8 Control males, 1 Control female, 3 LRP1KO males). Except for single-housed mice, all mice assigned to experiments survived until the endpoint. At either 4 or 10 months of age, mice were euthanized via 5% isoflurane/oxygen before the brains were collected for harvesting.

2.2. Behavior

Prior to behavioral testing, all animals were habituated in their home cage to the behavioral testing room for at least one hour. AnyMaze software (v.7.44 Stoelting, Wood Dale, IL, USA) was used to capture and analyze all behavioral tests by a blinded female researcher. For each maze, the apparatus was thoroughly cleaned with 70% ethanol between mice and allowed to dry. Tests were conducted within a behavioral testing room illuminated to 596 lux using fluorescent ceiling lights unless otherwise specified.

2.2.1. Elevated Plus Maze

Mice were placed in the center of an Elevated Plus Maze (Stoelting) with supplemental illumination to 900 lux in the open arms (and the closed arms illuminated to 400 lux) to enable video recording. Mice were allowed to explore the maze uninterrupted for 5 min. Time spent in, distance traveled, and entries (defined as all four paws in the maze arm, full body excluding tail) into each arm of the maze were recorded and analyzed. Animals that fell off the maze, or mice that did not explore the maze, were excluded from subsequent behavioral analysis (7 total mice; 3 Control males, 1 Control female, 2 LRP1KO males, and 1 LRP1KO female).

2.2.2. Y-Maze

Mice were placed in the center of a Y Maze (Stoelting) and allowed to explore the maze uninterrupted for 5 min. Distance traveled, entries (including order) into each arm of the maze were recorded and analyzed. Percent spontaneous alternations (%SAP) were calculated by dividing the total number of successful alternations by the total number of arm entries and multiplying by 100. A successful alternation is considered any iteration of arm travel that does not include returning to an arm the animal was most recently in. Any animals that escaped the maze or failed to explore the maze were removed from the analysis (2 total mice; 1 LRP1KO male, and 1 LRP1KO female).

2.2.3. Barnes Maze

The Barnes Maze protocol was adapted based on Attar et al. (2013) but was optimized to determine the appropriate number of training sessions adequate for our transgenic mice to learn the location of the escape hole at 9 months of age [25]. The Barnes Maze was constructed in-house from a 48″ × 48″ × 0.5″ sheet of StarBoard^® in Dolphin Gray (Boedeker Plastics, Shiner, TX, USA) (see Supplementary Methods). For our experimental animals, we established a protocol that included one “habituation day” to the maze prior to 5 consecutive “training days” and two “Probes”—one at 48 h and another at 7 days after the last training. Room lighting and a loud buzzer (~90 db) were used as aversive stimuli to encourage mice to enter the escape hole. The buzzer was turned off once the mouse entered the escape hole.

Habituation

The mice were placed at the center of the maze under a 2000 mL glass beaker for 30 s before being slowly guided to the escape hole and allowed 3 min to find the escape box located directly underneath the hole. If the mice did not enter the hole in 3 min, the mice were gently nudged into the hole with the beaker and allowed to remain in the escape box for 1 min before returning to their home cage. This process was repeated once after an hour.

Training

Each mouse was trained 3× per day, with at least an hour between training sessions. The mice were placed underneath a round opaque container in the center of the maze for 15 s before removing the container and allowing the mice to explore the maze for 2 min. If the mice did not find the escape box within that time, the experimenter guided them to the escape hole with a 2000 mL glass beaker and allowed them 3 more minutes to enter the escape box before gently nudging them into the box. The mice were allowed to stay within the escape box for one minute before returning them to the home cage.

Probe

After the training sessions were completed, mice were subjected to two probe phases to test memory. Probe 1 was administered at 48 h and Probe 2 at 7 days. Each probe had the same setup as the training days, except the escape box was removed. Mice were placed underneath the circular opaque container in the center of the maze for 15 s before the container was removed and the mice were allowed to freely explore the maze for 2 min. Afterwards, the mice were returned to their home cage. All activity was recorded and analyzed using AnyMaze. One male control mouse was removed from the analysis due to a lack of maze exploration (near-zero distance traveled).

2.3. Immunohistochemistry

On the day of harvest, mice were anesthetized with 5% isoflurane/oxygen before intracardial perfusion with PBS. Brains were extracted, and half were flash frozen for future experiments, while the other half were placed in 4% PFA overnight and further processed for histological analysis. The brains were then placed in 30% sucrose/phosphate-buffered saline (PBS) for at least 3 days before they were immersed in OCT (Optimal Cryo-Temp, Tissue Plus—Fisher HealthCare, Waltham, MA, USA) and frozen using isobutane surrounded by liquid nitrogen. Coronal slices (30 μm) of the half-brains were adhered to 1% gelatin coated slides, dried overnight and frozen at −80 °C for storage. Tissue slices were rehydrated using PBS and then permeabilized using 0.2% triton in PBS for 10 min, then washed with PBS (2 × 5 min). Slices were stained with Sudan Black (0.1% in 70% EtOH) for 10 min and washed until clear with PBS. Sections were incubated in a blocking solution of 5% bovine serum albumin (BSA) in PBS for 45 min before incubating with primary antibody (in blocking solution) overnight at 4 °C. The next day, sections were washed with PBS (4 × 15 min) and incubated in secondary antibody in blocking solution for 1 hr at room temperature. After washing with PBS (3 × 5 min), sections were incubated with DAPI (1 µg/mL in PBS), washed with deionized water, and allowed to dry overnight, protected from light at room temperature. Slides were coverslipped using Aquapolymount and imaged within days of coverslipping on a Zeiss LS710 Confocal with a 20X APOCHROMAT 20X/0.8 objective (Carl Zeiss AG, Oberkochen, Germany). Between 2 and 5 slices per animal were used to obtain counts by a blinded researcher.

2.4. Terminal Deoxynucleotidyl Transferase dUTP Nick End Labeling (TUNEL)

Immediately after doublecortin and post-secondary antibody incubation, sections were washed with PBS (3 × 5 min), and the tissue was fixed with 4% paraformaldehyde (PFA) for 10 min. Sections were washed with PBS (2 × 5 min) before being subjected to the TUNEL procedure, which was performed using DeadEnd™ Fluorometric TUNEL System (Promega G3250, Madison, WI, USA) according to the manufacturer’s protocol. Immediately following the TUNEL procedure, sections were stained with DAPI for 5 min, washed with ddH20 and allowed to dry overnight at room temperature. Coverslips were mounted using Aquapolymount and imaged within days. Sections were imaged using a Zeiss AXIO Imager. A1 fluorescence microscope at 10X EC-Plan NeoFluar 10X/0.3 objective (Carl Zeiss AG, Oberkochen, Germany). The ImageJ/FIJI v.2.160/1.54p/Java21.0.7 colocalization object counter plugin was used to count cells by a blinded researcher, which were obtained from 2–5 slices per mouse.

2.5. Morphology

After immunohistochemical staining for neuronal nuclear marker (NeuN) and Doublecortin (DCX), Z-stack confocal images were obtained with a 10 µm Z depth (1 µm increments) and used to perform morphological analyses via the Sholl technique [26]. Sholl analysis was performed using the ImageJ package, SNT v4.2.1 [27]. First, mature neurons dual-expressing tdTomato and NeuN, but not DCX, with clearly distinguishable projections without overlapping projections from other nearby neurons were isolated prior to generating tracings with the tdTomato signal and the SNT tracing tool. All tracings were then subjected to automated SNT Sholl analyses. An average of 4 cells were analyzed per animal with 10–13 animals per genotype (3–5 males/5–7 females). All analyses were performed blinded.

2.6. Flow Cytometry

Mice were injected with 5-ethynyl-2′-deoxyuridine (EdU, 50 mg/kg daily) for 7 days prior to humane euthanasia and dissection of the dentate gyrus and subventricular zone for further analysis. Analyses were performed on dentate gyrus tissue, with the subventricular zone cells serving as single-channel controls for compensation to confirm proper flow cytometry gating. The flow cytometry protocol was adapted from Velloso et al., 2021 [28]. Briefly, once dissected, tissue was minced and placed in a digestion buffer (0.20 U/mL liberase-DH, 100 μg/mL DNAse I, and PBS-Glucose-Mg²⁺ [0.6% dextrose, 1 mM MgCl₂]) for 15 min at 37 °C. Debris was removed by placing cells in Percoll (90% in PBS), followed by centrifugation (400× g, 4 °C, 10 min). The top debris layer was removed and the remaining cell suspension was washed and strained in a 40 um Corning^® Cell Strainer (Corning, NY, USA) before resuspension in PBS-Glucose (0.6% dextrose, 0.2% BSA). Non-specific background was blocked with mouse CD16/CD32 (FC receptor block, 1:50 for 10 min, 4 °C). Cells were pelleted (400× g for 5 min at 4 °C) and resuspended in 500 μL of PBS-Glucose (0.6% dextrose, 0.2% BSA). Living and dead cells were stained using LIVE/DEAD™ Fixable Violet Dead Cell Stain Kit (Invitrogen L34963, Carlsbad, CA, USA) according to the manufacturer’s instructions. Cells were incubated with antibodies for the neural progenitor marker APC CD133 (Invitrogen, Carlsbad, CA, USA, 1:100, 20 min 4 °C), pelleted (400× g, 4 °C, 10 min), washed (PBS), fixed in 4% PFA (20 min 4 °C), pelleted, and then washed again. Finally, cells were subjected to Click-iT™ Plus EdU labeling (Invitrogen, Carlsbad, CA, USA) according to the manufacturer’s instructions. Cells were resuspended in PBS-Glucose (0.6% dextrose, 0.2% BSA) and kept on ice until analysis on a BD^® LSR II cytometer (BD Biosciences, Franklin Lakes, NJ, USA). Data was analyzed using FlowJo software (V11, FlowJo, LLC, Ashland, OR, USA).

2.7. Statistical Analyses

Statistical evaluation was performed using GraphPad Prism (GraphPad Software v10.4.0), and power analysis for sample size calculations was performed in Prism using the Barnes maze results. Most data was analyzed via main effects two-way ANOVA, followed by Sidak’s test for individual comparisons, assuming normality and homoscedasticity unless otherwise specified. All tests used, along with statistical variables for ANOVA, are indicated in figure legends, while significant comparisons are indicated within the figures and reported with a p-value. Blinded analysis was performed by researchers who were unaware of the experimental group to which each mouse was assigned. Values are presented as mean ± SEM, and differences were considered statistically significant at p < 0.05.

2.8. Additional Supplementary Methods

Additional methods, including a detailed list of reagents and sources (Supplementary Table S1), are available in the Supplementary Materials.

3. Results

3.1. The Effect of LRP1 Knockout on Hippocampal NSCs of Adult Mice

Our previous study found significant alterations in NSCs’ response to ischemic stroke [16]. Given that the hippocampus is a major site of adult neurogenesis, we tested whether loss of LRP1 in adult NSCs caused any hippocampal phenotypes using our previously established mouse model, which expressed CreERt2 under control of the Nestin promoter. Mice were crossed with Ai4 mice, which expressed a stop-floxed tdTomato reporter, and with floxed LRP1 (LRP1KO) mice. As a control, LRP1KO mice are compared to those that express Nestin-CreERt2 and tdTomato, but do not have a floxed LRP1 cassette (Control mice). Treatment with tamoxifen in adults induces expression of CreERt2, which excises the stop codon to cause tdTomato expression, and in LRP1KO mice, removal of the LRP1 gene (Figure 1A). At 3 months of age, Cre expression was induced. Mice were aged an additional 1 or 6 months prior to behavioral testing, and mice were eventually harvested at 4 or 10 months of age (Figure 1B). In harvested tissues, no overt alterations in the dentate gyrus morphology were visualized (Figure 1C), and the heights of the dorsal granule cell layer and ventral granule cell layer were similar between LRP1KO and Control mice at each time point (Figure 1D,E).

3.2. Loss of LRP1 in Adult NSCs Induces Behavioral Deficits by 10 Months of Age

Because we were specifically interested in investigating the effect of LRP1KO in adult hippocampal neurogenesis, we employed behavioral assays that can be affected by hippocampal function [29,30]. It should be noted that no sex differences were observed in any behavioral analysis. The Y Maze was used to test for differences in working memory. When mice are placed in the center of the Y-shaped maze, they naturally tend to explore novel arms of the maze. Regardless of genotype, LRP1KO and Control mice explore similar distances in the Y maze equally (Figure 2A). The frequency with which mice turn into novel arms of the maze is measured via percent spontaneous alternations (%SAP). We found that LRP1KO mice displayed a significant decrease in %SAP compared to Control mice at 10 months of age (p = 0.026) but not at 4 months of age (Figure 2B).

We next tested whether mice displayed an anxiety-like phenotype with the Elevated Plus Maze. At 4 months of age, LRP1KO mice explored the maze similarly to Control mice (Figure 2C–E, Supplementary Figure S1A,B). However, in 10-month-old mice, LRP1KO mice travelled less distance than Control counterparts (p = 0.024; Figure 2C), even though they exhibited a similar total number of entries (entries into open and closed arm but not center) (Figure 2D). Ten-month-old LRP1KO mice had a slight tendency to spend more time in the closed arm of the maze (p = 0.11; Figure 2E) but spent similar time in the open arms (Figure S1A) and had a similar ratio of open arm entries to Control mice (Figure S1B).

3.3. Adult NSC LRP1KO Causes Hippocampal-Dependent Memory Deficits

To definitively test whether hippocampal-driven behaviors are affected by LRP1 in NSCs, we employed a canonical behavioral measure of hippocampal function, the Barnes maze, a more sophisticated test of learning and memory that relies on the mouse learning the location of a target escape hole on a circular platform with dummy escape holes along the perimeter (Figure 3A) [25,29,30]. For the Barnes maze, we tested these mice 6 months after tamoxifen, at 9 months of age. Because genetic strain can influence memory acquisition in the Barnes maze [25,30,31,32], we initially optimized our protocol and determined that three daily trainings over the course of 5 days (15 total training sessions) allowed acquisition of the task. We measured the latency to identify the target hole during individual trainings (Figure 3B) and found that the latency to identify the target significantly decreased between the first training session and the last training session in both Control (p = 0.0003) and LRP1KO (p = 0.011) mice (Figure 3C), suggesting that LRP1KO mice learn the location of the escape hole similarly to Control mice. The mice were then probed for the ability to remember the escape location at 48 h (Probe 1) and 7 days (Probe 2) after training. The number of nose pokes over the holes in each quadrant of the maze was measured, and LRP1KO and Control mice had increased nose pokes over any hole within the target quadrant both at 48 h (Figure 3D) and 7 days (Figure 3G) post-training. We also tested the number of nose pokes over individual holes. At 48 h post-training, LRP1KO and Control mice have a similar number of nose pokes and time spent over the target hole (Figure 3E,F). However, by 7 days post-training, LRP1KO mice exhibit a deficit of nose pokes at the target hole, with an increase in nose pokes over holes surrounding the target hole (Figure 3H). In contrast, Control mice retain the location of the escape hole, with most nose pokes centered over the target hole (Figure 3H). Similarly, LRP1KO mice spent less time near the target escape hole compared to Control mice (p = 0.017, Figure 3I). These findings suggest the KO animals maintain similar spatial learning but lack the same spatial memory as control animals.

3.4. LRP1KO Does Not Affect Migration of Newborn Hippocampal Neurons, but Induces Morphological Alterations in Adult-Born Neurons

We previously showed that LRP1KO in the subventricular zone induces deficits in migration of NSCs toward ischemic lesions, putatively due to a loss of chemokine receptor 4 (CXCR4) expression [16]. CXCR4 is also a mediator of migration of NSCs from the SGZ into the granule cell layer [33]. To test whether altered migration could underlie behavioral deficits, we measured the localization of tdTomato-positive cells as a function of distance from the SGZ, both by measuring the total number of cells as well as the percentage of cells in each population at any given distance from the SGZ (Supplementary Figure S2). We found that regardless of genotype, the same total number of cells could be measured at any given distance in both LRP1KO and Control mice at 4 months of age (Supplementary Figure S2A,B). By 10 months of age, we observed an upward trend in the total number of tdTomato-positive cells localized 3–18 μm from the SGZ of LRP1KO mice compared to Control mice (p = 0.1039, Supplementary Figure S2C); however, measured as a percentage of the total cell population, the percentage of total cells localized any given distance from the SGZ was the same in LRP1KO and Control mice (Supplementary Figure S2D).

Liu et al. (2010) and others have shown that loss of LRP1 in mature neurons causes a reduction of dendrites [34,35]. To test whether adult-born hippocampal neurons had altered morphology, we employed Sholl analysis of tdTomato-positive neurons in 10-month-old Control (Figure 4A) and LRP1KO mice (Figure 4B). At greater distances from the soma, LRP1KO neurons displayed fewer intersections compared to Control neurons (p < 0.0001, Figure 4C). The Sholl decay coefficient, which measures the rate at which dendritic intersections decrease at further distances from the soma [36], was increased in LRP1KO mice (p = 0.031), suggesting a steeper decline in dendritic branch densities with distance from the soma (Figure 4D). Altogether, these results suggest that without LRP1, neurons fail to develop or maintain the appropriate dendritic architecture.

3.5. LRP1KO Increases the Total Number of Adult-Born Neurons in the Hippocampus by 10 Months of Age

Safina et. al. showed that in culture, loss of LRP1 in NSCs derived from the subventricular zone displays a decrease in neurogenesis and oligodendrogenesis but an increase in astrogliogenesis [23]. To test whether the differentiation of adult-born neurons was affected by LRP1KO in the hippocampus in vivo, we measured the number of tdTomato-positive hippocampal cells co-labelled for either DCX, an early marker of developing neurons [1] or NeuN, a marker that is acquired by mature neurons [1] (Figure 5A,B). We found that by 10 months, LRP1KO mice had an increased total number of tdTomato-positive, DCX/NeuN-negative cells compared to Control mice (p = 0.0008, Figure 5(a6) pink arrow, Figure 5C). Indeed, while the total number of tdTomato-positive cells decreased from 4 to 10 months in Control mice (p = 0.0002), LRP1KO mice did not display the same decrease in tdTomato-positive cells, suggesting a preservation of this pool (Supplementary Figure S3A). The total number of tdTomato, DCX-positive early neurons in Control mice (Figure 5(a5) green arrow) decreased from 4 to 10 months (p < 0.0001, Figure 5D), consistent with age-related changes observed in neurogenesis [37]. Interestingly, the total number of these tdTomato/DCX double-positive early neurons was decreased in LRP1KO mice compared to Controls at 4 months (p = 0.0001); however, it did not decrease further with aging, leaving levels similar to Control mice by 10 months (Figure 5D). Indeed, the changes in the whole population of DCX-positive cells (both tdTomato-positive and tdTomato-negative) was similar, with a decrease in the total number of DCX-positive cells in Control mice over time (p < 0.0001) and an early reduction in LRP1KO mice compared to Controls (p = 0.004), which did not reduce further with age (Figure S3B). There may have been some compensation by the non-tdTomato-expressing cells in LRP1KO mice in the DCX-positive cell population, because LRP1KO mice have an increased total number of tdTomato-negative, DCX-positive cells compared to Control mice at 4 months (p = 0.006, Figure S3D).

The number of intermediate maturing neurons triple expressing tdTomato/DCX/NeuN (Figure 5(b4,b5) yellow arrow) were also counted, and we found that Control mice had reduced numbers of these cells at 10 months compared to 4 months (p = 0.0004, Figure 5E), further suggesting age-related reductions in neurogenesis. In contrast, LRP1KO mice did not display such a reduction in the triple-expressing tdTomato/DCX/NeuN cells with age, such that LRP1KO mice had more of these cells than Control mice by 10 months of age (p = 0.005, Figure 5E). As a whole, the DCX/NeuN-positive cell population (with and without tdTomato) was affected by the alteration in the tdTomato-positive pool in LRP1KO mice. While both Control and LRP1KO mice had age-related decreases in the DCX/NeuN intermediate cells with age (p < 0.0001), in LRP1KO mice, the decrease was less severe, so that by 10 months, LRP1KO mice had a greater number of these cells compared to Control mice (p = 0.019, Figure S3C). This effect on the whole cell population of intermediate maturing neurons was likely attributable mainly to tdTomato-positive cells, as tdTomato-negative, DCX/NeuN-positive cells were consistent between Control and LRP1KO mice at both 4 and 10 months (Figure S3E).

Ultimately, the described alterations in the maturing hippocampal neuron pool resulted in LRP1KO mice developing a greater number of mature tdTomato-positive NeuN-expressing neurons at 10 months of age compared to Control mice (p < 0.0001, Figure 5(a6,b4,b5)-blue arrows, Figure 5F).

3.6. Increased Neuronal Development May Be Associated with Increased Proliferation

An increase in proliferation or conversely a decrease in cell death could account for the increased number of hippocampal neurons developing in LRP1KO mice. Normally, NSCs undergo asymmetric division to create progenitor cells and preserve the stem cell pool [1]. To test whether altered proliferation could account for the increased flux of cells differentiating into neurons, mice were injected with EdU for 7 days prior to harvest to label dividing cells in the hippocampus (Supplementary Figure S4A). Using immunohistochemistry, we found that regardless of genotype, the total number of proliferating cells was similar between LRP1KO and Control mice at both 4 months and 10 months of age. Similarly, in Control (p = 0.016) and LRP1KO mice (p = 0.005), the number of proliferating cells decreased with age, consistent with age-related reductions others have observed in neurogenesis [38] (Supplementary Figure S4B). Given that the total number of proliferating hippocampal cells is low, results from histology may be underpowered to detect differences in proliferation. Thus, we also performed flow cytometry as an alternative, potentially more sensitive way to measure proliferating cells [39] (Figure 6A–E; gating strategy can be found in Figure S5). Along with tdTomato and EdU, we used a Live/Dead kit and labelled for CD133, a cell-surface neural progenitor marker, to exclude astrocyte progenitor cell contamination. Using flow cytometry, we found the ratio of proliferating and non-proliferating cells was similar in Control and LRP1KO mice at 4 months of age (Figure 6A,B,E). Conversely, by 10 months of age, we observed an increased ratio of proliferating cells to non-proliferating cells in LRP1KO mice (Figure 6C–E).

We next tested whether we could measure any differences in cell death, as many immature cell types, including Type 2 and Type 3 progenitor cells, undergo apoptosis as a normal part of neurogenesis [1]. To measure the number of Type 2b and 3 cells undergoing programmed cell death in the hippocampus, we performed terminal deoxynucleotide transferase dUTP nick end labeling (TUNEL) and DCX co-labeling (Figure 7A–F), as DCX expression becomes present once neural stem cells develop into Type 2b cells, remains present through Type 3, and loses DCX expression once cells reach a mature postmitotic stage [1,37]. We first performed this analysis in mice harvested after behavioral testing in the Elevated Plus Maze. In tdTomato-positive, DCX-negative cells, we found that LRP1KO mice had a reduced percentage of TUNEL-positive cells compared to Control mice at 4 months of age (p = 0.0048), and that these differences were no longer evident at 10 months of age, likely because the percentage of cells undergoing apoptosis in Control mice had reduced by this time point (p = 0.0127), whereas the percentage remained consistent in LRP1KO mice (Figure 7G). In tdTomato-positive, DCX-positive presumptive Type 2b/3 cells, we observed no differences in TUNEL between LRP1KO and Control mice at either time point (Figure 7H). In the population of tdTomato-negative cells (non-neuronal cells), LRP1KO mice exhibited an increase in total TUNEL with age (p = 0.044), but the levels were not significantly different to Control mice at either 4- or 10-months of age (Figure 7I). We repeated our analysis using new 4-month-old mice that did not spend time in the Barnes maze or Elevated Plus Maze, and interestingly, we did not observe a significant difference in TUNEL between the Control mice and the LRP1KO mice in the tdTomato-positive, DCX-negative cell population (Figure 7J).

4. Discussion

In summary, our data suggests that the lack of LRP1 in adult NSCs impairs hippocampal spatial memory and working memory and increases anxiety-like behavior in 10-month-old LRP1KO mice. Adult-born neurons without LRP1 display elevated maturation markers coupled with decreased dendritic complexity. LRP1 loss also resulted in increased proliferation and diminished cell death. Together, our results demonstrate the importance of LRP1 in adult hippocampal neurogenesis—the implications of which are likely to extend beyond the basic biology of learning and memory into the memory and behavioral deficits associated with neurodegenerative and other diseases.

To date, no studies have investigated the importance of LRP1 in adult hippocampal NSCs. We are the first to discover a significant role for LRP1 in mediating hippocampal NSC biology. Of note is the behavioral phenotype identified in our Barnes maze testing. One week after training, LRP1KO mice display a loss of explicit target hole identification yet maintain the ability to identify the target quadrant. This suggests a loss in pattern separation—the ability to identify similar, but distinct contexts. Pattern separation relies on hippocampal neurogenesis [40,41,42]. Further, Nakashiba and colleagues found a distinct role for young granule cells versus older granule cells in regulating pattern separation [43]. Although our data do not supply definitive evidence for a deficit in pattern separation, the appearance of this memory phenotype in the older mice elicits questions about what specific stage of neurogenesis is most affected by LRP1 loss.

To identify potential causes of the behavioral phenotype observed in LRP1KO mice, we examined the effect on NSC development within the hippocampus. Even though developing neurons appear to migrate normally through the granule cell layer (Figure S2), neurons lacking LRP1 have reduced dendritic complexity (Figure 4), suggesting that impaired neuronal function could account for behavioral deficits. May et al. (2023) found that deletion of LRP1 in adult synapsin I-positive neurons induced significant motor and neurological deficits and altered synaptic function [34,44]. Liu et al. (2010) also previously described the effect of LRP1 knockout in mouse forebrain neurons and discovered that such neurons exhibited a host of deficiencies, including age-related degeneration, synaptic loss, and degeneration of neurites [34]. The Liu study (2010) found a degeneration of LRP1 knockout neurons; however, our study does not differentiate whether observed morphological differences were due to the degeneration of dendrites or a failure to properly form in the first place [34]. However, given the documented role of LRP1 in proper synaptic function, it is entirely possible that the deficits in behavior we observe are due to impaired neuronal function in newborn neurons caused by LRP1 knockout. Nevertheless, future studies will necessitate determining whether behavior could be rescued by re-expressing LRP1 in adult neurons, or if the effects of LRP1 on neuronal development irreversibly alter neuronal connectivity and function within the hippocampus.

During normal regulation of hippocampal neurogenesis, NSCs give rise to progenitor cells, which proliferate to create an excess of newborn neurons, most of which undergo apoptosis [1]. Those that survive follow a distinctive maturation process in which they migrate a short distance to become part of the granular cell layer to integrate into the hippocampal circuitry [1,2,45]. We were surprised to observe that LRP1KO mice had a greater number of tdTomato-positive neurons developing within the SGZ (Figure 5), and that this increased number of neurons is associated with an early reduction in TUNEL as a marker of cell death, specifically in DCX-negative, tdTomato-positive cells of younger mice, but only in those mice that underwent behavioral testing (Figure 7). Notably, DCX-negative, tdTomato-negative cells had increased levels of cell death by 10 months in LRP1KO mice, possibly suggesting that the tdTomato-negative pool slowly compensated for the reduced cell death in the tdTomato-positive pool, which could also explain why the size of the granule cell layer did not increase in 10-month-old LRP1KO mice (Figure 1). Interestingly, in the tdTomato-positive pool, DCX-positive cells had similar amounts of cell death in Control and LRP1KO mice. These results suggest that LRP1 influences cell death of adult-born neurons at a specific stage of development, and possibly only under stressed conditions, as mice that did not undergo behavioral testing showed no differences in cell death. Importantly, Kim et al. (2009) found that impairing cell death via Bax deletion caused an accumulation of hippocampal neurons with impaired synaptic function [46], suggesting programmed cell death is essential in enabling quality control in newborn neurons. It has also been observed that acute stress can result in an increase in cell death in adult-born neurons [47,48]. Future experiments would require interrogation of the effects of acute stress on adult-born neurons lacking LRP1 as well as exploration of the mechanisms and stages of differentiation most affected by LRP1KO, and to what extent putative evasion of cell death influences differentiation and function of these neurons. While there are several potential mechanisms that could explain the changes observed within the LRP1KO mice, one potential mechanism to explore is LRP1-driven internalization and localization of NMDA receptors. LRP1 plays a role in the internalization of NMDA receptors [49,50], and NMDA receptors are involved in cell death pathways following stress [51,52] and contribute to dendritic complexity [53]. Thus, it is plausible that knocking out LRP1 can alter NMDA internalization and activity, leading to several changes that account for our observations.

Consistent with our observation of reduced markers of cell death in LRP1KO mice, we also found that age did not reduce neurogenesis markers, where such reductions were clearly observable in Control mice, altogether resulting in more total mature neurons in the 10-month-old LRP1KO mice. Several interesting points are raised, which require further investigation. The first is whether LRP1 truly enables developing neurons to evade cell death—an intriguing possibility that may allow targeting of NSCs to enhance neurogenesis without increasing the number of cell divisions of activated stem cells that are normally thought to deplete the NSC pool. Another question is the influence LRP1 has on the rate and timing of neurogenesis. The increase in tdTomato-positive neurons in LRP1KO mice was insufficient to significantly alter the total size of the granule cell layer in older mice (Figure 1), suggesting that the observed effect was either transient or slow, such that differences were measurable only with time and accumulation of the effects. Similarly, we did not measure whether there were any changes in the ratio of cells with differing cell fates, such as glial cells. This is certainly a possibility, considering Andreas Faissner’s group has shown that alterations in cell fate definitively occur due to loss of LRP1 [23,54].

We also observe increased hippocampal cellular proliferation, which our results would contradict results reported by Safina et al. (2016), who found NSCs derived from the subventricular zone had impaired proliferation in vitro [23], and support our own previous study, which found an increase in proliferation of subventricular zone NSCs in vivo [16]. Given the technical difficulties of measuring proliferation in such a small population of cells in vivo, we were unable to resolve this apparent contradiction in this study. However, differences in vivo vs. in vitro settings and cellular niche of origin could account for discrepancies between this study and that of Safina et al. (2016) [23]. Regardless, increased proliferation is very likely to at least partially account for the increase in mature LRP1KO adult-born neurons we observe in these mice.

Altogether, we have shown that LRP1 knockout in adult NSCs induces memory impairment in as little as 6–7 months post-knockout via reduced nose pokes within the Barnes maze target hole, indicating deficits in spatial memory in addition to impairments in working memory by decreased spontaneous alternations in the Y maze. We have observed multiple differences in cellular phenotypes that might account for such differences in behavior, including alterations in cell death, flux of developing neurons, adult neuronal morphology, and proliferation. Pinpointing whether any or all of the observed phenotypes cause the behavioral deficits remains an exciting challenge and increases our understanding of the influence of LRP1 on adult neurogenesis. While our findings are currently specific to LRP1 function within adult NSCs, our data exemplify the profound impact of targeting these mechanisms. The dentate gyrus exerts significant influence over hippocampal function, meaning even minor disruptions could have widespread effects on overall brain function. First, LRP1KO mice exhibit a measurable behavioral phenotype that mirrors memory loss described in Alzheimer’s disease models [25,55,56,57,58,59,60]. In human post-mortem tissue, hippocampal neurogenesis levels correlate with Alzheimer’s disease [61,62]. Animal models recapitulate these findings, as hippocampal neurogenesis is downregulated in many mouse models of Alzheimer’s disease [63,64]. Further, LRP1 is a considerable target of interest in relation to Alzheimer’s disease, as LRP1 is a receptor for amyloid ꞵ [18], tau [19], and Apolipoprotein E (ApoE) [65]. It is entirely possible that conditions in Alzheimer’s disease impair LRP1 to influence hippocampal memory and function via influencing neurogenesis. With this mechanistic framework in mind, agonism of LRP1 could potentially be a viable therapeutic strategy. Presently, no therapeutically approved LRP1 agonist exists, but clinical trials are underway testing the efficacy of LRP1 agonism in unrelated conditions [66], which emphasizes the multimodal therapeutic potential of targeting LRP1.

The hippocampus is also critical in psychiatric disorders, and our results suggest anxiogenic effects. Indeed, the hippocampus is important in regulating anxiety, with evidence from Weeden and colleagues (2015) suggesting specific involvement of the ventral dentate gyrus [67]. The hippocampus also influences depression through changes in plasticity [68]. While there is controversy about whether patients with depression experience decreased hippocampal volume and decreased adult neurogenesis [41,69,70], increasing adult neurogenesis can alleviate depressive symptoms [71,72]. The dentate gyrus also plays a key role in post-traumatic stress disorder (PTSD) through influencing pattern separation. Patients with PTSD often struggle to discriminate between similar stimuli, leading to exaggerated startle responses to non-threatening stimuli [73]. Stress is known to alter adult neurogenesis by both decreasing proliferation [74] and increasing cell death of hippocampal adult-born neurons [47,48]. Stress can also exacerbate psychiatric symptoms [75]. Given that our data suggests LRP1KO mice may be resistant to stress-induced apoptosis of newborn neurons, further exploration might lead to the identification of new targets for the treatment of stress-related hippocampal dysfunction.

Altogether, our results further emphasize that dysfunction within a very small population of cells could be a significant driver of hippocampal pathologies. Exposing the forces that are driving these behaviors within our model may provide undiscovered insight into human disease conditions and underscores the profound effects both LRP1 and NSCs can have on basic neurobiology.