Inhibition of repulsive guidance molecule A ameliorates diabetes-induced cognitive decline and hippocampal neurogenesis impairment in mice

Abstract

Although diabetes mellitus is strongly associated with dementia, the mechanism underlying diabetes-induced cognitive dysfunction has not been clarified. Here, we demonstrate the vital role of repulsive guidance molecule A (RGMa) in the regulation of adult hippocampal neurogenesis and cognitive impairment under diabetic conditions. In type 2 diabetic db/db mice and streptozotocin-mediated type 1 diabetic mice, RGMa is upregulated in the granular cell layer of the dentate gyrus. Additionally, both neural stem cells (NSCs) and immature neurons express its receptor, neogenin. In vitro experiments revealed that high glucose-conditioned hippocampal neurons inhibited the differentiation of NSCs, and the application of an anti-RGMa antibody restored it. The treatment with an anti-RGMa antibody ameliorated diabetes-induced cognitive decline and impairment of hippocampal neurogenesis. These findings suggest that the RGMa negatively regulates hippocampal neurogenesis and is involved in diabetes mellitus-induced cognitive decline.

Repulsive Guidance Molecule A (RGMa) plays pivotal roles in disrupting adult hippocampal neurogenesis and cognitive impairment under diabetes.

Introduction

The prevalence of diabetes, the most common metabolic disorder, has been on the rise worldwide¹. Diabetes causes cardiovascular damage, inflammation, and metabolic imbalance throughout the body, resulting in chronic organ malfunction. In the central nervous system (CNS), diabetes negatively affects neuronal function, inducing cognitive dysfunction and mood impairments^2,3. Although the mechanism underlying diabetes-associated cognitive impairment has not been elucidated, an MRI study revealed that patients with type 2 diabetes show decreased gray matter volume in several regions involved in cognition and depression, including the prefrontal cortex, hippocampus, para-hippocampus, and amygdala. This suggests that brain pathology underlies neuropsychological dysfunction in patients with diabetes⁴. Multiple mechanisms, such as neuroinflammation, insulin deficiency, or glycation end-product (AGEs)-induced oxidative stress have been suggested to play a role in cognitive dysfunctionl^2,5–9; however, the detailed molecular mechanisms of diabetes-related dementia are poorly understood, which results in the unavailability of molecular-targeted therapy for this condition. The development of a new treatment strategy for diabetes-related dementia may break the cycle of deteriorating diabetes management and worsening cognitive impairment caused by low medication adherence^10,11.

Recently, the possible involvement of adult hippocampal neurogenesis in the pathology of diabetes-related cognitive dysfunction was reported^7,9,12,13. Using animal models of diabetes, Stranahan et al. revealed that elevated corticosterone levels negatively regulate hippocampal neurogenesis, synaptic plasticity, and hippocampus-dependent memory⁷. These results clearly demonstrate the importance of hippocampal neurogenesis as a therapeutic target for diabetes-related cognitive dysfunction. Nevertheless, no further studies have succeeded in elucidating the molecular mechanisms of impaired adult neurogenesis under diabetic conditions.

Repulsive guidance molecule A (RGMa) is a glycosylphosphatidylinositol (GPI)-anchored glycoprotein that is known as an axonal guidance molecule during the developmental phase^14–18. Notably, RGMa is negatively involved in the pathology of various neurological disorders, including spinal cord injury, multiple sclerosis, stroke, neuromyelitis optica, and Parkinson’s disease^19–26. These findings lead us to investigate the involvement of RGMa in neurological diseases that induce cognitive impairment. In the present study, we elucidated the contribution of RGMa to diabetes-related disruption of hippocampal neurogenesis and cognitive impairment. We found that RGMa is increased in the hippocampal dentate gyrus of insulin-resistant type 2 diabetic db/db mice. In vitro experiments revealed that high glucose directly induces RGMa expression in hippocampal neurons and inhibits neural development in differentiating neural stem cells (NSCs) via RGMa. We also found that the anti-RGMa antibody improved the reduction in doublecortin (DCX)-positive cells in the hippocampal dentate gyrus and mitigated memory deficits in both db/db mice and streptozotocin (STZ)-mediated insulin-deficient type 1 diabetic mice. These findings suggest that RGMa plays a crucial role in disrupting adult hippocampal neurogenesis and causing cognitive impairment in diabetes.

Results

RGMa expression is upregulated in the granular cell layer of the dentate gyrus in db/db mice

The db/db mouse, homozygous for spontaneous leptin receptor mutations, has been used as a model of type 2 diabetes mellitus. Db/db mice exhibit obesity, hyperglycemia, and diabetes-related complications, including cognitive decline^7,27. First, we monitored the body weight and blood glucose levels in db/db and control mice (non-diabetic +/db mice) and confirmed significantly higher blood glucose levels in db/db mice than in age-matched +/db mice (Supplementary Fig. 1A, B). Next, we performed immunohistochemical analysis to investigate the number and morphogenesis of DCX-positive immature neurons in the dentate gyrus, aiming to evaluate hippocampal neurogenesis. This evaluation is based on the established understanding that DCX-positive immature neurons are generated from NSCs within the subgranular zone (SGZ) and play important roles in learning and memory formation^28,29. Additionally, morphological complexity is recognized as a characteristic feature in differentiating NSCs²⁹. The number of DCX-positive immature neurons was significantly reduced in db/db mice compared to in non-diabetic controls (Fig. 1A–C). These DCX-positive immature neurons exhibited short-distance neurites (Fig. 1D). To quantitatively evaluate this morphological alteration, we measured both the DCX-positive area in the granular cell layer (GCL) and the area per cell, both of which were markedly reduced in db/db mice (Fig. 1E, F). These results suggest that the morphogenesis of DCX-positive immature neurons is impaired in the context of diabetes, and that hippocampal neurogenesis was disrupted in db/db mice.

Fig. 1. Repulsive guidance molecule A (RGMa) expression is increased in the granular cell layer (GCL) of type 2 diabetes db/db mice.

A Immunostaining of doublecortin (DCX) in the hippocampal dentate gyrus of 12-week-old +/db and db/db mice. Scale bars: 200 µm. B, C Quantification of the total number (B) and density (number per GCL area) (C) of DCX-positive immature neurons in the GCL of 12-week-old +/db and db/db mice. P values were determined by an unpaired t test with Welch’s correction. ***P < 0.001. N = 5 per group. D Magnified images of DCX-positive cells in the GCL. Scale bars: 20 µm. E, F Quantification of the total area (E) and area per cell (F) of DCX-positive immature neurons in the GCL of 12-week-old +/db and db/db mice. P value was determined by unpaired t test with Welch’s correction. **P < 0.01. ***P < 0.001. N = 5 in each group. G The images of quantitative in situ hybridization, RNAscope for RGMa of 12-week-old +/db and db/db mice. GCL was indicated by the yellow broken line. Scale bars: 20 µm. H, I Quantification of RNAscope in the GCL (H) and hilus (I) of the hippocampal dentate gyrus of 12-week-old +/db and db/db mice. P value was determined by unpaired t test with Welch’s correction. *P < 0.05. N = 4 in each group. J Whole images of the immunohistochemical staining of RGMa in 12-week-old +/db and db/db mice. Scale bars: 200 µm. K Magnified images of the immunohistochemical staining of DCX, RGMa, and Nissl staining in the hippocampal dentate gyrus of 12-week-old +/db and db/db mice. GCL was indicated by the white broken line. Scale bars: 50 µm. L Quantification of the fluorescent intensity of RGMa in the granular cell layer of the hippocampal dentate gyrus of 12-week-old +/db and db/db mice. P value was determined by unpaired t test with Welch’s correction. ***P < 0.001. N = 4 in each group. Data are shown as mean ± SEM.

To date, whether RGMa is associated with the impairment of adult neurogenesis and cognition under pathological conditions remains unknown. First, we evaluated total RGMa expression in the hippocampus of db/db mice, finding no significant difference from that in controls (Supplementary Fig. 2). Given prior evidence suggesting that the local environment surrounding brain stem cell niches regulates the state of NSCs and immature neurons³⁰, we performed quantitative in situ hybridization, RNAscope (Fig. 1G–I), and immunofluorescent staining (Fig. 1J–L) to evaluate RGMa expression specifically in the hippocampal neurogenic niche. RNAscope, which allows for quantitative RNA expression analysis without histological disturbance, revealed a significant increase in RGMa mRNA within the GCL of db/db mice (Fig. 1G, H), whereas no change was observed in the hilus (Fig. 1I). Immunohistochemical staining confirmed these findings, showing elevated RGMa expression in Nissl-positive GCL neurons, with significantly increased RGMa signal intensity in db/db mice (Fig. 1J–L).

To identify the specific cell types expressing RGMa within the hippocampal dentate gyrus, we examined various markers. Although RGMa has been previously detected in glial cells within the cerebral cortex, spinal cord^22,26, and in hilar GABAergic interneurons³¹, we found no RGMa signal in glial fibrillary acidic protein (GFAP)-positive astrocytes, ionized calcium-binding adapter molecule 1 (Iba-1)-positive microglia, and Sox-2 positive NSCs in the dentate gyrus (Supplementary Fig. 3A, B). On the other hand, in the Nissl-positive GCL, we found that DCX-positive immature neurons expressed RGMa (Fig. 1K, white arrows).

These results indicate that both immature and mature neurons in the hippocampal dentate gyrus express RGMa, with expression specifically upregulated in the GCL of db/db mice.

NSCs and immature neurons express the RGMa receptor, neogenin

Based on a previous study reporting that neural stem and progenitor cells express neogenin^32,33, we investigated cell types expressing neogenin in the hippocampal dentate gyrus of +/db and db/db mice by immunohistochemical staining. We found that neogenin was expressed in Sox2-positive NSCs and DCX-positive immature neurons, but not in glial cells, in both +/db and db/db mice (Supplementary Fig. 4A–C). Notably, we found that Sox2-positive NSCs did not always express neogenin (Supplementary Fig. 4D: white arrows and arrowhead), and Sox2, DCX, and neogenin triple-positive cells were observed (Supplementary Fig. 4D: yellow arrowhead). These observations suggest that NSCs begin to express neogenin when they differentiate into immature neurons (Supplementary Fig. 4E).

Moreover, to determine whether RGMa-expressing GCL neurons directly interact with neogenin-expressing cells, we performed co-staining for neogenin and RGMa. Our results revealed that RGMa-positive cells were in close contact with neogenin-expressing cells within the GCL of db/db mice (Supplementary Fig. 5).

These findings suggest that, under diabetic conditions, RGMa-expressing neurons in the GCL directly interact with neogenin-expressing NSCs and immature neurons.

The high glucose condition directly induces RGMa expression in hippocampus-derived neurons in vitro

Mature neurons are the primary components of GCL. To identify diabetes-related factors that increase RGMa expression in the GCL, we treated the HT-22 neuronal cell line derived from the mouse hippocampus with high glucose (Fig. 2A). Hyperglycemia is a major and common feature of type 1 and type 2 diabetes, with high concentrations of glucose directly and negatively affecting brain function via AGEs and reactive oxygen species (ROS) and promoting abnormal DNA modification³⁴. The immunoreactivity of RGMa in HT-22 cells treated with high glucose increased 24 h after the treatment (Fig. 2B, C). To elucidate the involvement of RGMa in the neural differentiation of NSCs, we cultured neurospheres, which are clusters composed of NSCs from neurogenic niches, as described previously^35,36 (Fig. 2D). We confirmed that nestin-positive NSCs and DCX- and MAP2-positive immature neurons differentiated from NSCs expressed neogenin (Fig. 2E, F).

Fig. 2. Upregulated repulsive guidance molecule A (RGMa) in glucose-conditioned HT-22 directly inhibits neural development of differentiating neural stem cells (NSCs) in vitro.

A Phase-contrast images of HT-22 cells maintained in medium that contained basal (25 mM) and high (100 mM) glucose for 24 h. Scale bars: 100 µm. B Immunocytochemical RGMa staining of HT-22 maintained in medium that contained basal (25 mM) and high (100 mM) glucose for 24 h. Scale bars: 50 µm. Magnified images were shown on the right. Scale bars: 20 µm. C Quantification of fluorescent intensity of RGMa in HT-22 maintained in medium that contained basal (25 mM) and high (100 mM) glucose for 24 h. P value was determined by unpaired t test with Welch’s correction. **P < 0.01. N = 3 in each group. D Phase-contrast image of artificial NSCs (aNSCs) cultured as neurosphere. Scale bar: 100 µm. E, F Immunocytochemical staining of aNSCs and differentiating NSCs. Nestin (E), and doublecortin (DCX) and MAP2 (F) were co-stained with neogenin. Scale bars: 20 µm. G Schematic illustration of neurosphere co-culture assay with glucose-conditioned HT-22 cells. H Immunocytochemical MAP2 staining of neural precursor-like cells differentiating from neurospheres co-cultured with HT-22 cells conditioned with normal or 100 mM glucose and incubated with Control or anti-RGMa Ab. Scale bars: 20 µm. I–K Quantification of the total number (I), neurite length (J), and MAP2-positive area (K) of differentiating NSCs. Neurite length and MAP2-positive area were normalized with Hoechst counts. Each P value was determined by one-way ANOVA with Turkey’s multiple comparisons test, respectively. *P < 0.05. ***P < 0.001. N = 5 in each group. Data are shown as mean ± SEM.

These results indicate that NSCs and immature neurons have a potential to receive RGMa signaling via neogenin.

High glucose-treated HT-22 cells inhibit the differentiation of NSCs through RGMa

Our findings revealed increased RGMa expression within the GCL of db/db mice and close contact between RGMa-positive cells and neogenin-positive cells (Fig. 1G–L and Supplementary Fig. 5). To investigate whether the direct interaction with RGMa-expressing hippocampal neurons alter the neurogenic properties of neogenin-expressing NSCs, we established a co-culture assay of NSC-derived neurospheres and glucose-conditioned HT-22 cells (Fig. 2G and Supplementary Fig. 6). After maintaining HT-22 cells in high glucose condition medium (100 mM), the cells were treated with anti-RGMa antibody for 3 h to inhibit the function of RGMa expressed on the cells. Subsequently, HT-22 cells were co-cultured with the neurospheres for 3 h. Neurospheres were then collected, seeded, and differentiated into MAP2-positive immature neurons. In the control condition, neurospheres were co-cultured with HT-22 cells, which were maintained in normal glucose medium, and were then differentiated into immature neurons. They showed extended neurites with a complex morphology, which is a main aspect of differentiating NSCs. In contrast, when neurospheres were co-cultured with HT-22 cells pretreated with high glucose, the differentiated immature neurons showed shortened neurites with reduced MAP2-positive areas, although the number of precursor cells remained unchanged. However, treating high glucose-conditioned HT-22 cells with an anti-RGMa antibody prior to co-culture with neurospheres significantly alleviated both the impairment in neurite morphogenesis and the reduction in the MAP2-positive area in neurosphere-derived immature neurons (Fig. 2H–K).

To investigate whether neogenin expression in NSCs is altered by hyperglycemia, we cultured neurospheres under varying glucose concentrations following established protocols³⁷. Neogenin expression levels in NSCs were unchanged across glucose conditions (Supplementary Fig. 7), suggesting that NSCs retain the capacity to receive RGMa signaling via neogenin under diabetic conditions.

These results indicate that glucose-conditioned hippocampal neurons directly interact with NSCs via RGMa, altering the neurogenic properties of differentiating NSCs.

Anti-RGMa antibody treatment improves hippocampal neurogenesis impairment in db/db mice

Given that RGMa negatively affects the neural development of NSCs in vitro, we investigated whether the inhibition of RGMa signaling improves the reduction in DCX-positive cells and cognitive decline in diabetic mice. To address this, we intravenously administered humanized anti-RGMa antibody to non-diabetes control (+/db) and db/db mice over the time course shown in Fig. 3A.

Fig. 3. Inhibition of repulsive guidance molecule A (RGMa) by the humanized anti-RGMa neutralizing antibody improves hippocampal neurogenesis impairment in db/db mice.

A Schematic of the administration of humanized anti-RGMa antibody, behavioral tests, and tissue collection. B, C The transitions of the body weight (B) and blood glucose (C) in control or anti-RGMa antibody treated +/db and db/db mice. Each P value was determined using a two-way ANOVA with Sidak’s post-hoc test. +/db+Control Ab: n = 6, +/db+RGMa Ab: n = 6, db/db+Control Ab: n = 11, db/db+RGMa Ab: n = 12. D Immunostaining of doublecortin (DCX) in the hippocampal dentate gyrus of control or anti-RGMa antibody treated +/db and db/db mice. Scale bars: 200 µm. Magnified images were also shown. Scale bars: 20 µm. E–H Quantification of the total number (E), density (number per granular cell layer (GCL) area) (F), area (G), and area per cell (H) of DCX-positive immature neurons in the GCL of control or anti-RGMa antibody treated +/db and db/db mice. Each P value was determined by one-way ANOVA with Turkey’s multiple comparisons test, respectively. *P < 0.05. ***P < 0.001. ****P < 0.0001. +/db+Control Ab: n = 3, +/db+RGMa Ab: n = 3, db/db+Control Ab: n = 8, db/db+RGMa Ab: n = 8. Data are shown as mean ± SEM.

First, we confirmed that the administration of the anti-RGMa antibody did not affect body weight or blood glucose both in +/db and db/db mice (Fig. 3B, C), suggesting that the administration of the anti-RGMa antibody does not interfere with diabetes itself.

Next, we evaluated the effects of anti-RGMa antibody on hippocampal neurogenesis. Immunohistochemical staining revealed that RGMa antibody mitigated the reduction of the number and morphogenesis of DCX-positive cells in the dentate gyrus (Fig. 3D–H).

Since previous studies have reported a decrease in NSC numbers within the GCL under diabetic conditions^7,38, we further investigated whether the anti-RGMa antibody influenced NSC proliferation using immunohistochemical staining targeting Ki-67, a marker of proliferative cells. However, antibody treatment did not affect the reduced number of Ki-67-positive cells in the GCL (Supplementary Fig. 8A–C).

These results suggest that the anti-RGMa antibody specifically improves the development and the survival of DCX-positive immature neurons in the context of insulin-resistant type 2 diabetes, without altering NSC proliferation.

Anti-RGMa antibody treatment ameliorates the cognitive decline in db/db mice

To assess the effect of anti-RGMa antibody treatment on hippocampus-dependent memory, the novel object recognition test (NORT) was performed (Fig. 4A). Mice with normal cognition instinctively explore a new object more than a familiar object; however, mice with cognitive impairment lose the bias to explore new objects³⁹. Consequently, db/db mice explored new and familiar objects almost equally in the test phase, resulting in a lower Discrimination Index (DI) than non-diabetic controls (Fig. 4B). These results indicate that db/db mice exhibited cognitive impairment. After confirming that the total distance of exploration in the test phase was not altered by anti-RGMa antibody administration (Fig. 4C, D), we calculated the DI and found that anti-RGMa antibody treatment significantly improved the reduction of the DI in db/db mice (Fig. 4C, E).

Fig. 4. Administration of humanized anti-repulsive guidance molecule A (RGMa) antibody improves hippocampus-dependent memory in db/db mice.

A Schematic of novel object recognition test, NORT. B Discrimination index of NORT-test phase in 16-week-old +/db (n = 6) and db/db (n = 7) mice. P value was determined by unpaired t test with Welch’s correction. **P < 0.01. C Tracking and heat map images of control or RGMa Ab treated db/db mice in the NORT-test phase. D, E Quantification of traveled total distance (D) and discrimination index (E) of +/db or db/db mice treated with control Ab or RGMa Ab in the NORT-test phase. Each P value was determined by one-way ANOVA with Turkey’s multiple comparisons test, respectively. **P < 0.01. ***P < 0.001. ****P < 0.0001. +/db+Control Ab: n = 6, +/db+RGMa Ab: n = 6, db/db+Control Ab: n = 11, db/db+RGMa Ab: n = 11. F Schematic image of object placement test, OPT. G Tracking and heat map images of control or RGMa Ab treated db/db mice in the OPT-test phase. H, I Quantification of traveled total distance (H) and discrimination index (I) of +/db or db/db mice treated with control Ab or RGMa Ab in the OPT-test phase. Each P value was determined by one-way ANOVA with Turkey’s multiple comparisons test, respectively. +/db+Control Ab vs dbdb+Control Ab; P = 0.0512. *P < 0.05. **P < 0.01. +/db+Control Ab: n = 6, +/db+RGMa Ab: n = 6, db/db+Control Ab: n = 11, db/db+RGMa Ab: n = 10. Data are shown as mean ± SEM.

We also performed an object placement test (OPT), another paradigm for evaluating hippocampus-dependent memory^40,41. The experimental time-course is shown in Fig. 4F. As previously reported⁷, spatial memory was disrupted in db/db mice; in the test phase, db/db mice treated with the control antibody explored duplicated objects randomly, that is, regardless of their locations, and thus showed low DI values. In contrast, the DI value of db/db mice treated with the anti-RGMa antibody significantly increased (Fig. 4G–I).

These results indicate that anti-RGMa antibody treatment effectively improves impaired hippocampal neurogenesis and cognitive decline in insulin-resistant type 2 diabetes.

Anti-RGMa antibody prevents hippocampal neurogenesis reduction and cognitive impairment in STZ-treated type 1 diabetic mice

Finally, we aimed to address whether humanized anti-RGMa antibody treatment is effective against memory deficits in STZ-mediated type 1 diabetes model mouse. First, we monitored the body weight and blood glucose levels and confirmed that STZ-treated mice showed a significant loss of body weight and rapid elevation of blood glucose level (Supplementary Fig. 9A, B). We then conducted immunohistochemical staining and analyzed the number of DCX-positive cells in the dentate gyrus to evaluate hippocampal neurogenesis. The number of DCX-positive cells was significantly lower in STZ-treated mice than in non-diabetic mice (Fig. 5A, B). We also assessed hippocampus-dependent memory by NORT, and STZ-treated mice showed a lower DI values than wild-type (WT) mice (Figs. 4A and 5C). Thus, STZ-treated type 1 diabetic mice show cognitive decline and impaired hippocampal neurogenesis, as observed in type 2 diabetic db/db mice.

Fig. 5. Humanized anti-repulsive guidance molecule A (RGMa) antibody also improves hippocampal neurogenesis reduction and cognitive decline in streptozotocin (STZ)-mediated type 1 diabetic mice.

A Immunohistochemical staining of DCX in the hippocampal dentate gyrus of age-matched wild-type (WT) and STZ-treated mice 4 weeks after the diabetes induction. Scale bars: 200 µm. B Quantification of the number of doublecortin (DCX)-positive immature neurons in the granular cell layer (GCL) of age-matched WT and STZ-treated mice 4 weeks after the diabetes induction. P value was determined by unpaired t test with Welch’s correction. *P < 0.05. N = 4 in each group. C Discrimination index of age-matched WT (n = 25) and STZ-treated (n = 12) mice 4 weeks after the diabetes induction. P value was determined by unpaired t test with Welch’s correction. *P < 0.05. See also Fig. 4A. D RNAscope images for RGMa in the hippocampal dentate gyrus of age-matched WT and STZ-treated mice, 2 weeks after diabetes induction. The GCL is indicated by the yellow broken line. Scale bars: 20 µm. E, F Quantification of RNAscope in the GCL (E) and hilus (F) of the hippocampal dentate gyrus of age-matched WT and STZ-treated mice, 2 weeks after diabetes induction. P values were determined by an unpaired t test with Welch’s correction. ***P < 0.001. WT: n = 7, STZ: n = 9. G Immunostaining of DCX in the hippocampal dentate gyrus of control or anti-RGMa antibody administrated STZ mice 6 weeks after the diabetes induction and age-matched WT mice. Scale bars: 200 µm. H–K Quantification of the total number (H), density (number per GCL area) (I), area (J), and area per cell (K) of DCX-positive immature neurons in the GCL of control or anti-RGMa antibody administrated age-matched WT and STZ-treated mice 6 weeks after the diabetes induction. Each P value was determined by one-way ANOVA with Tukey’s multiple comparisons test. *P < 0.05. **P < 0.01. ***P < 0.001. ****P < 0.0001. N = 6 in each group. L Discrimination index of control or anti-RGMa antibody administrated age-matched control and STZ-treated mice 4 weeks after the diabetes induction. Each P value was determined by one-way ANOVA with Turkey’s multiple comparisons test. *P < 0.05. WT: n = 8, STZ+Control Ab: n = 9, STZ+RGMa Ab: n = 10. See also Fig. 4A. Data are shown as mean ± SEM.

Next, we performed quantitative in situ hybridization (RNAscope) and immunohistochemistry to identify RGMa expression in the hippocampal dentate gyrus of STZ-mice. RNAscope revealed that RGMa expression was significantly and specifically increased within the GCL of both STZ-treated and db/db mice (Fig. 5D–F). Immunohistochemical staining supported this result by showing RGMa fluorescent signal within the GCL of STZ-treated mice (Supplementary Fig. 10). Moreover, we confirmed that Sox2-positive NSCs and DCX-positive immature neurons in STZ-treated mice expressed neogenin, consistent with observations in db/db mice (Supplementary Fig. 11).

We then investigated the effects of the anti-RGMa antibody on hippocampal neurogenesis and cognitive function in STZ-treated mice. We intravenously administered the anti-RGMa antibody once a week for 6 weeks. First, we confirmed that anti-RGMa antibody administration did not affect the reduction in body weight or increase in blood glucose observed in STZ mice (Supplementary Fig. 12A, B). Subsequently, we performed immunohistochemical DCX-staining (Fig. 5G) and found that anti-RGMa antibody treatment significantly improved the reduction in number and morphogenesis of DCX-positive immature neurons within the GCL (Fig. 5H–K).

Finally, we assessed the effect of the anti-RGMa antibody on the memory deficits induced by NORT (Fig. 4A). The DI of STZ mice treated with anti-RGMa antibody was significantly higher than that of STZ mice treated with the control antibody (Fig. 5L).

These results indicate that the anti-RGMa antibody ameliorates the impairment of hippocampal neurogenesis and cognitive decline not only in insulin-resistant type 2 diabetic mice but also in insulin-deficient type 1 diabetic mice.

Discussion

In the present study, we provide direct evidence that RGMa is involved in the pathophysiology of diabetes-related cognitive impairment. We also revealed that anti-RGMa antibody treatment effectively ameliorated cognitive decline and impairment of hippocampal neurogenesis in db/db and STZ-mediated diabetic mice.

Adult neurogenesis in the hippocampus has been suggested to play a pivotal role in maintaining hippocampal function^28,29. NSCs (radial glia-like cells) with a proliferative capacity differentiate into intermediate neural progenitor cells, which can give rise to neuroblasts and then differentiate into mature neurons²⁹. Newborn neurons are integrated into the dentate circuitry within 6 weeks and involved in hippocampus-dependent memory formation. X-ray-mediated or pharmacogenetic ablation of hippocampal neurogenesis results in increased spontaneous neuronal firing in the dentate gyrus, indicating that immature neurons inhibit neural hyperactivation and modulate recurrent network activity⁴². Immature neurons also contribute to memory formation directly. Because of the lack of inhibitory inputs from interneurons in the hilus, immature neurons have a lower threshold for activation than mature neurons and respond to various afferent stimuli from multiple brain regions, including the lateral and medial entorhinal cortices, resulting in pattern separation and memory flexibility, which is essential for hippocampus-dependent spatial and episodic memory formation^28,29,43,44. Importantly, the relationship between hippocampal neurogenesis and cognitive impairment in diabetes has recently been investigated both in human patients and animal models^3,13. However, to date, strict glycemic control is the only choice to ameliorate diabetes-related cognitive decline, which emphasizes the necessity of identifying draggable molecular targets.

We demonstrated that NSCs and immature neurons express the RGMa receptor, neogenin, and that the expression of RGMa is upregulated in the granular cell layer of the dentate gyrus of db/db and STZ-treated mice. A recent study by Kandilya et al. reported that high glucose treatment of human fetal neural progenitor cells (ReNcell VM) induces epigenetic modification to alter the DNA methylation pattern of neurodevelopment-associated genes³⁴. Based on whole genome DNA methylation analysis in this previous report, the promoter region of RGMa was hypomethylated, which may result in upregulation of target gene expression. Because the concentration of glucose is increased in the cerebrospinal fluid of patients with diabetes⁴⁵, it is possible that hyperglycemia induces RGMa upregulation in the GCL. Further research is needed to clarify the reason behind the specific increase in RGMa expression in GCL neurons under these conditions.

In this study, we identified that RGMa is upregulated in the GCL surrounding the neurogenic niche in diabetes. We also showed that high glucose-treated hippocampal neurons directly interact with neogenin-expressing NSCs, influencing their morphogenesis during neurogenesis via RGMa. Given previous findings showing that the microenvironment around NSCs and immature neurons play a critical role in regulating hippocampal neurogenesis⁴⁶, it is reasonable to speculate that upregulated RGMa modulates the neurogenic capacity of NSCs through trans-signaling. Considering the downstream signals of RGMa stimulation in NSCs, the RhoA/ROCK pathway has a potential to play a key role in suppressing neurogenesis. It is well established that RGMa stimulation of neurons leads to activation of RhoA and Rho kinase via neogenin and inhibition of neurite growth. When RGMa binds to neogenin, RhoA is activated through LARG, a guanine nucleotide exchange facto, and subsequent activation of its downstream target ROCK inhibits neurite outgrowth via cytoskeletal rearrangement^19,47. Importantly, RhoA/ROCK activity plays a significant role in adult neurogenesis; the inhibition of Rho kinase is essential during differentiation stage of neurogenesis and ROCK inhibitor rescues the neurogenesis defects^48,49. Therefore, it is possible that RGMa expressed in the GCLs induces RhoA/ROCK activation of neogenin-expressing NSCs and inhibits neural development.

We also confirmed that neogenin is expressed in immature neurons, indicating that its presence may directly contribute to impaired neural maturation. Given recent findings suggesting that recombinant RGMa induces cleaved caspase 3-mediated apoptosis in differentiating neurons³¹, RGMa-mediated apoptosis could accelerate the reduction of DCX-positive immature neurons under diabetic conditions. Additionally, we observed that immature neurons themselves also expressed RGMa, raising the possibility that cis-signaling by RGMa-neogenin may also affect the differentiation capacity of immature neurons. Conversely, mice with neogenin depletion in NSCs and immature neurons show hippocampal neurogenesis impairment and depressive-like behavior³³. Therefore, the function of neogenin in adult neurogenesis is context-dependent and may rely on its ligands. These results emphasize the suitability of RGMa, instead of neogenin, as a therapeutic target for treating diabetes-associated cognitive decline.

Insulin has important roles in maintaining hippocampus function, and insulin resistance is linked to cognitive deficits due to impaired synaptic plasticity and neurogenesis^50–52. Our study demonstrates that RGMa expression is upregulated in the hippocampal GCL and that anti-RGMa antibody treatment ameliorates deficits in DCX-positive immature neurons development and cognitive decline in both insulin-resistant and insulin-deficit diabetic mice. In addition, our neurosphere co-culture assay revealed that high glucose-treated hippocampal neurons directly interact with NSCs and inhibit neurite elongation via RGMa. These results indicate that RGMa contributes to cognitive decline in diabetes through an insulin-independent mechanism, underscoring the potential of RGMa inhibition as a complementary therapy to insulin replacement in diabetes management. On the other hand, therapeutic potential of Metformin and SGLT2 inhibitors, which are known to alter blood glucose levels independently of insulin, on diabetic cognitive impairment has recently been discussed^53–60. Metformin and SGLT2 inhibitors are reported to attenuate cognitive function of rodent models and patients with diabetes^55,60–62. Especially in diabetes animal models, it is revealed that these drugs contribute to the improvement of hippocampal neurogenesis^55,62. Following alleviation of diabetic symptoms, Metformin leads to the reduction of AGEs and inflammatory mediators such as phosphorylated NF-κB⁶³. Also, SGLT2 inhibitors are suggested to modulate ROS- and TNFα-mediated neuroinflammation and promote the production of neurotrophic factors including BDNF^54,64,65. These multiple benefits of these drugs may contribute to the enhancement of the hippocampal neurogenesis and improvement of diabetes-related cognitive dysfunction comprehensively. Our current results revealed that anti-RGMa antibody treatment improves the disruption of adult neurogenesis and hippocampus-dependent memory without affecting blood glucose levels, and the increase in RGMa signaling can be induced by hyperglycemia. Therefore, it is possible that the mechanism of RGMa-dependent inhibition of hippocampal neurogenesis partially underlies the therapeutic effect of Metformin and SGLT2 inhibitors on diabetes-related cognitive impairment.

We demonstrated that the antibody did not affect neural differentiation in control mice (non-diabetic WT and +/db mice). Considering the results of our in vitro experiments, the anti-RGMa antibody is likely to block the direct and disease-specific interaction between RGM-expressing mature neurons in the GCL and neogenin-expressing neurogenic cells, including NSCs and immature neurons. Nonetheless, there is a possibility that anti-RGMa antibody improves the diabetes-related pathology through multiple targets. Disruption of the blood brain barrier (BBB) has been documented in diabetes, and there is an ongoing debated regarding the correlation between BBB disruption and cognitive decline^66,67. Hyperglycemia is thought to affect endothelial cells by altering tight junctions and increasing BBB permeability⁶⁸. BBB disruption may enhance the unwanted effect of peripheral-derived factors, and Erion et al. proposed that Il-1b, which is secreted from peripheral adipocytes, contributes to cognitive decline in db/db mice by activating microglia in the hippocampus⁶⁹. In this context, we recently reported that under experimental autoimmune encephalomyelitis conditions, anti-RGMa antibody treatment promotes the restoration of the blood-spinal cord barrier⁷⁰. Therefore, the therapeutic effects of the anti-RGMa antibody on diabetes-related cognitive decline can be enhanced by its ameliorating effects on BBB disruption and inflammation in the brain. Further investigation is needed to uncover the cell type-specific mechanism of RGMa-neogenin signaling under diabetic conditions using RGMa- or neogenin-conditional knockout mice.

In our study, we investigated the alterations in both the number and morphology of immature neurons, evaluating hippocampal neurogenesis both in vivo and in vitro. Morphologically, we found that anti-RGMa antibody treatment effectively ameliorated the structural changes in DCX-positive immature neurons (Figs. 3G, H and 5J, K). These findings are consistent with the results from our co-culture assay (Fig. 2); high glucose-treated hippocampal neurons interact with NSCs via RGMa and suppress their neurite growth in differentiating process. However, the in vitro co-culture assay did not detect a reduction in the number of neural progenitors (Fig. 2I), despite the alteration in both the number and morphogenesis of DCX-positive immature neurons observed under diabetic conditions. Given that HT-22 cells were treated with high-glucose prior to co-culture with NSCs, the neurospheres themselves were not directly exposed to high glucose, and similarly, differentiating immature neurons were not directly exposed to either high glucose or RGMa. Therefore, this limited experimental setup may not fully reflect the reduction in differentiating cell numbers seen in diabetic conditions.

Considering that RGMa has a potential to induce cleaved caspase 3-mediated cell death in immature neurons³¹, RGMa upregulation likely impairs the survival of DCX-positive immature neurons, leading to a reduction in their number under diabetic condition.

In this study, we provided a new insight that RGMa negatively regulates hippocampal neurogenesis and is involved in diabetes mellitus-induced cognitive decline, offering a new approach for the treatment of cognitive decline following an imbalance in adult neurogenesis. Impairment in adult neurogenesis has also been observed in neurodegenerative diseases presenting with dementia, including AD and dementia with Lewy bodies (DLB)^28,71,72. Therefore, it would be interesting to explore whether the RGMa antibody also exerts therapeutic effects on other types of dementia in a future study.

Methods

Animals and induction of type 1 diabetes

We have complied with all relevant ethical regulations for animal use. Animal care and experimental procedures were approved by the institutional regulations of Osaka University. 8 weeks-old C57BL/6JJmsSlc mice were purchased from Japan SLC, Inc. (Shizuoka, Japan). To induce type 1 diabetes, we administered 200 mg/kg STZ (S0130, Merck, Darmstadt, Germany) intraperitoneally as previously described⁷³. STZ was dissolved in citrate buffer (pH 4.5). Age-matched mice injected with the vehicle alone were used as controls. In our study, STZ-treated mice were required to have serum glucose ≧ 300 mg/dL 3 days after the STZ injection. Blood samples were obtained from the tail vein and blood glucose levels were measured using LAB Gluco (Bio Medical Sciences, Tokyo, Japan).

Six-week-old male and female mice with mutant leptin receptors (BKS.Cg-+Leprdb/+Leprdb/Jcl: db/db mice) and non-diabetic control mice (BKS.Cg-m + /+Leprdb/db: +/db mice) were purchased from CLEA Japan, Inc. (Shizuoka, Japan).

The mice were housed under a 12 h light/dark cycle. Water and food were provided ad libitum.

Administration of the anti-RGMa antibody

The humanized anti-RGMa antibody (unasnemab/MT-3921; Mitsubishi Tanabe Pharma Corp., Osaka, Japan) and an isotype control antibody (palivizumab; AbbVie GK, Tokyo, Japan) were administered intravenously at a dose of 30 mg/kg once a week from 3 days after STZ injection or at 6 weeks of age in +/db and db/db mice.

Behavioral testing

Handling

Animals were handled for 5 min by experimenters 3 consecutive days before the beginning of behavioral tests to reduce their stress because human contact affects the results of behavioral tests^74,75. The handling processes were conducted between 5 p.m. to 8 p.m. during the light cycle.

NORT

The previously described NORT procedure was modified as shown in Fig. 4A^40,41. The mice were allowed to explore an open field (40 × 40 cm) for 15 min (Habituation phase 1). After 24 h, the mice were placed in the field again for 5 min (Habituation phase 2) and then allowed to freely explore the field with duplicate objects described previously for 5 min (Acquisition phase)⁷⁶. Approximately 12 h later, mice freely explored the field with the familiar object used in the acquisition phase and the new object for 10 min (Test phase). Each object was placed 100 mm diagonal to the wall. The field and objects were cleaned with 70% ethanol in every session except between the Habituation phase 2 and Acquisition phase. The time spent exploring each object in each session was manually measured. In our study, exploration of an object was defined as sniffing and touching, but not climbing, as mentioned previously⁷⁶. The DI was calculated as follows: {(duration of exploration for a new object in the test phase) – (duration of exploration for familiar objects in the test phase)} / (total exploration time for both objects in the test phase). The total distance traveled in the field was calculated automatically using a smart video tracking software (Panlab, Barcelona, Spain).

OPT

The previously described OPT procedure was optimized, as shown in Fig. 4F⁴⁰. The same field as that in the NORT was used in each session, but vertical stripes were placed on a wall inside the box as visual cues, and the objects were completely different from those in the NORT. The mice were allowed to explore the open field for 5 min (Habituation phase 1). Immediately after the habituation phase 1, the mice freely explored the field with duplicate objects placed 80 mm parallel to the corners of the striped wall for 5 min (Acquisition phase). After 3 h, the mice explored the field with duplicate objects, but one was relocated to a novel location for 10 min (Test phase). The novel location was at the right corner, opposite the stripe wall and 80 mm away from the walls. The DI was defined as follows: {(duration of exploration for a newly located object in the test phase) – (duration of exploration for an old-located object in the test phase)}/(total time of exploration for both objects in the test phase). The total distance traveled in the field was calculated automatically using smart video tracking software.

Cell culture

The HT-22 cells were purchased from Merck (SCC129). The cells were maintained at 37 °C with 5% CO₂ in Dulbecco’s modified Eagle’s medium (DMEM) with 10% heat-inactivated fetal bovine serum (FBS) and antibiotics (100 IU/mL penicillin and 100 μg/mL streptomycin).

Isolation and passaging of artificial NSCs

The isolation and passaging of artificial NSCs (aNSCs) were previously described with some modifications^35,36. The brains of 8- to 12-week-old WT C57BL/6 J male mice were dissected and sliced into 1 mm coronal sections. Subventricular zone and SGZ were dissected and collected into DMEM/F12 with antibiotics (100 IU/mL penicillin and 100 μg/mL streptomycin). After the centrifuge, digestion was performed in 0.05% trypsin-EDTA (T4174, Sigma-Aldrich, MO, USA) for 5 min at 37 °C and stopped by adding an equal volume of trypsin inhibitor (T6522, Sigma-Aldrich) in phosphate-buffered saline (PBS). Cells were triturated to a single cell using a 1 mL syringe with a 25 G needle. After the centrifuge, cells were resuspended in 20 mL proliferation medium containing Neural Stem Cell Basal Medium (SCM003, Merck Millipore), B27 without vitamin A (12587010, Thermo Fisher Scientific, Massachusetts, USA), GlutaMAX (35050061, Thermo Fisher Scientific), antibiotics (100 IU/mL penicillin and 100 μg/mL streptomycin), 1 mg/mL epidermal growth factor (E9644, Sigma-Aldrich), and 1 mg/mL fibroblast growth factor 2 (100-18B, PeproTech, NJ, USA), and incubated at 37 °C with 5% CO₂. Approximately 48 h later, 40 µL of 10 µg/mL epidermal growth factor and 20 µL of 10 µg/mL fibroblast growth factor were added. Clusters of neural stem cells (neurospheres) were visible after 7 days.

For passaging, the proliferation medium containing the neurospheres was collected and centrifuged. Neurospheres were digested in 0.05% trypsin-EDTA and an equal volume of trypsin inhibitor and triturated using a 1 mL syringe with a 26 G needle. After centrifugation, the cells were split according to the stage of passage. The ratio of primary neurospheres to secondary neurospheres was 1:1. The ratio of secondary neurospheres was 1:3. After 48 h, 40 µL of 10 µg/mL epidermal growth factor and 20 µL of 10 µg/mL fibroblast growth factor were added. All experiments were performed using aNSCs between 3 and 10 passages.

aNSCs co-culture assay with high glucose conditioned HT-22

This procedure is illustrated in Fig. 2G. HT-22 cells were seeded into 24 well plates at a density of 1.0 ×  10⁴ cells/well. After 24 h, the medium was changed to the glucose condition medium containing DMEM with 10% FBS, antibiotics (100 IU/mL penicillin and 100 μg/mL streptomycin), and glucose (25 mM (basal) or 100 mM). After incubation at 37 °C with 5% CO₂ for 24 h, control or anti-RGMa antibody was added at a concentration of 100 µg/mL and incubated at 37 °C with 5% CO₂. The medium containing the control or anti-RGMa antibody was removed 3 h later, and the cells were washed in PBS. A 1 mL proliferation medium containing neurospheres 7 days after the passage was added and cultured for 3 h at 37 °C with shake (120 rotations/min). The medium containing neurospheres was collected gently to avoid disturbing the adhered HT-22 cells. After the centrifuge, digestion was performed in 0.05% trypsin-EDTA for 5 min at 37 °C and stopped by adding an equal volume of trypsin inhibitor. Cells were triturated to a single cell using a 1 mL syringe with a 25 G needle. After centrifugation, cells were resuspended in 500 µL proliferation medium and seeded into poly-L-ornithine and laminin coated 4 wells slide at a density of 1.0 × 10⁴ cells/well and incubated at 37 °C with 5% CO₂.

aNSCs differentiation assay

After 24 h of incubation in the proliferation medium, half of the medium was replaced with differentiation medium containing Neural Stem Cell Basal Medium, B27 without vitamin A, GlutaMAX, antibiotics (100 IU/mL penicillin and 100 μg/mL streptomycin), 1 mg/mL epidermal growth factor, 1 mg/mL fibroblast growth factor 2, 1 μM retinoic acid (R2625, Sigma-Aldrich), and 1 μM forskolin (F6886, Sigma-Aldrich). The cells were incubated continuously, with half of the medium replaced with differentiation medium every day for 3 days.

Cells preparation for the immunocytochemistry of non-differentiated neural stem cells from neurospheres

Medium containing neurospheres was collected and digestion was performed in 0.05% trypsin-EDTA for 5 min at 37 °C and stopped by adding an equal volume of trypsin inhibitor. Cells were triturated to a single cell using a 1 mL syringe with a 25 G needle. After centrifugation, cells were resuspended in 500 µL proliferation medium and seeded into poly-L-ornithine and laminin coated 4 wells slide at a density of 1.0 × 10⁴ cells/well and incubated at 37 °C with 5% CO₂.

Immunocytochemistry

Cells were fixed in 4% PFA for 20 min at room temperature. After washing with PBS, the cells were permeabilized and blocked in PBS containing 5% donkey serum and 0.1% Tween-20 (blocking buffer) for 1 h. Samples were incubated overnight at 4 °C with following primary antibodies dissolved in blocking buffer: anti-MAP2 (1:200; 4542S, Cell Signaling Technology, MA, USA), anti-DCX (1:500; AF100-25, R&D systems, MN, USA), anti-nestin (1:500; ab81462, Abcam, Cambridge, UK), goat anti-RGMa (1:100; AF2458, R&D systems), and neogenin (1:100; NBP1-89651, Novus biologicals, Abingdon, UK). After washing with PBS, secondary antibodies were applied, and the samples were incubated for 2 h at room temperature. After washing with PBS, the nuclei were counterstained with Hoechst 33342 (1:1000), and coverslips were mounted with Dako fluorescent mounting medium (S302380-2, Agilent Technologies). Images were acquired using a confocal laser-scanning microscope (FV3000; Olympus Corporation, Tokyo, Japan). Thirty to sixty cells from each group were analyzed. All analyses were conducted using ImageJ software. For the quantification of RGMa fluorescence in HT-22 cells, images were acquired under the same microscopic conditions. The fluorescence intensity was measured in 60 cells per well. The MAP2-positive area in each cell was calculated automatically and divided by the Hoechst count. Neurite length was determined using the Simple Neurite Tracer (SNT) plugin in the ImageJ software. Ten cells were analyzed in each independent experiment.

Tissue processing

Mice were deeply anesthetized and transcardially perfused with cold PBS and 4% PFA. Brains were post-fixed in the same fixative overnight and cryoprotected in 30% sucrose. the samples were embedded in Tissue-Tek O.C.T. Compound (Sakura Finetek Japan Co., Ltd., Tokyo, Japan). Each brain was cut into 30 µm thick sagittal sections in a cryostat (CM3050S, Leica, Wetzlar, Germany) and mounted on MAS-coated glass slides (FF-901, Matsunami Glass Inc., Osaka, Japan). Samples were stored at −80 °C until further processes.

Immunohistochemistry

Brain sections were washed three times for 5 min in PBS and then retrieved in pH 6.0 citrate buffer for 10 min at 80 °C. After washing thrice for 5 min each in PBST, the samples were blocked with 5% donkey serum in PBST (blocking buffer) for 1 h at room temperature. Following rinses with PBST, sections were incubated overnight at 4 °C with following primary antibodies dissolved in blocking buffer: rabbit anti-DCX (1:500; ab18723, Abcam), sheep anti-DCX (1:100; AF100-25, R&D systems), anti-Sox2 (1:100; sc-365823, Santa Cruz Biotechnology, TX, USA), anti-Sox2 (1:100; MAB4343, Merck Millipore), anti-Ki-67 (1:100; ab16667, Abcam), mouse anti-glial fibrillary acidic protein (GFAP) (1:400; G3893, Sigma-Aldrich), guinea pig anti-Ionized calcium-binding adapter molecule 1 (Iba-1) (1:500; 234-308, Synaptic Systems, Goettingen, Germany), goat anti-RGMa (1:100; AF2458, R&D systems) and rabbit anti-neogenin (1:100; NBP1-89651, Novus biologicals). For secondary staining, the sections were washed three times for 5 min in PBS and incubated for 1 h at room temperature with secondary antibodies dissolved in blocking buffer. Nissl staining with NeuroTrace™ 530/615 Red Fluorescent Nissl Stain (1:500; N21482, Thermo Fisher Scientific) was conducted during the treatment of secondary antibodies. Sections were washed thrice for 5 min each with PBS, and nuclear staining was performed using DAPI (1:1000) dissolved in PBS. After rinsing with PBST, the samples were mounted using a fluorescence mounting medium. Images were acquired using a confocal laser-scanning microscope. The number of Ki-67-positive and DCX-positive cells was counted manually and normalized to the area of the granular cell layer (GCL). For the quantification of RGMa fluorescence in the GCL, images were acquired under the same microscopic settings. The area of DCX-positive cells and the GCL, and the fluorescence intensity of RGMa staining were calculated using ImageJ software.

Quantitative in situ hybridization, RNAscope

RNA was extracted from fixed frozen sections prepared for immunohistochemistry. RNAscope was performed using the RNAscope Multiplex Fluorescent Reagent Kit v2 (323110, ACD Bio, CA, USA) according to the manufacturer’s instructions. Briefly, sections were washed in PBS for 5 min and baked in HybEZ™ II Hybridization System With EZ-Batch Slide System (321710, ACD bio) for 30 min at 60 °C prior to the post-fixation in 4% PFA for 15 min at 4 °C. The slides were dehydrated in 50%, 70%, and 100% ethanol for 5 min each at room temperature. After drying for 5 min at room temperature, H₂O₂ was treated for 10 min at room temperature. Slides were rinsed in PBS and boiled in target retrieval buffer for 5 min at 95 °C to improve probe accessibility. After rinsing in PBS, slides were incubated in 100% ethanol for 3 min at room temperature, dried for 5 min at room temperature and baked for 30 min at 60 °C. Slides treated with Protease III for 30 min at 40 °C to improve probe accessibility. After washing with wash buffer for 2 min, Ms-RGMa C1 probe (464141, ACD bio) was treated for 2 h at 40 °C. After washing with wash buffer for 2 min, hybridization was amplified using AMP1 for 30 min, AMP2 for 30 min and AMP3 for 15 min, and detected using HRP-C1 for 15 min and Opal 690 dissolved in TSA buffer (1:1500) for 30 min. The slides were washed with wash buffer for 2 min after each treatment. After treatment with HRP blocker for 15 min and washing with wash buffer for 2 min, nuclear staining was performed with DAPI, and the slides were mounted with fluorescence mounting medium. Images were acquired using a confocal laser-scanning microscope. RGMa signals were automatically counted using the ImageJ software. Counts were normalized to the total area of the hippocampal GCL or hilus, which was calculated using DAPI images.

RNA extraction and RT-qPCR

RNA extraction from the hippocampus and neurosphere was performed using TRIzol RNA isolation reagents (15596026, Thermo Fisher Scientific) and either the RNeasy Mini or Micro kit (74104 and 74004, Thermo Fisher Scientific). cDNA was synthesized using High-Capacity cDNA Reverse Transcription Kit (4368814, Applied Biosystems). Finally, quantified real-time PCR (qPCR) analysis was conducted using TaqMan Gene Expression Assays (RGMa: Mm00624998_m1, Neo1: Gapdh: Mm99999915_g1, Thermo Fisher Scientific) as instructed by the manufacturer.

Statistics and Reproducibility

The results are presented as the mean ± standard error of mean (SEM), and statistical analyses were performed using GraphPad Prism 9. Statistical significance was defined as P < 0.05. The number of individual experiments is specified in the figure legends and figures. Error bars indicate the SEM.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

H.U., T.I., and T.Y. designed the experiments. H.U. performed the experiments and analyzed the data. H.U., T.I. and T.Y. wrote the manuscript. T.I. and T.Y. conceived the study. All the authors reviewed the manuscript.

Peer review

Peer review information

Communications Biology thanks the anonymous reviewers for their contribution to the peer review of this work. Primary Handling Editors: Christian Wozny and Joao Valente.

Data availability

The Supplementary Data file contains all source data underlying the graphs in the main and Supplementary Figs. The data that support the findings of this study are available from the corresponding author upon reasonable request.

Competing interests

A humanized anti-RGMa monoclonal antibody (unasnemab/MT-3921) was developed by Osaka University, Chiba University, and Mitsubishi Tanabe Pharma Corporation and are the property of them. The Department of Neuro-Medical Science was co-founded by Osaka University and Mitsubishi Tanabe Pharma Corporation.

Supplementary information

The online version contains supplementary material available at 10.1038/s42003-025-07696-7.