Neuronal guidance signaling in neurodegenerative diseases: Key regulators that function at neuron-glia and neuroimmune interfaces

Abstract

The nervous system processes a vast amount of information, performing computations that underlie perception, cognition, and behavior. During development, neuronal guidance genes, which encode extracellular cues, their receptors, and downstream signal transducers, organize neural wiring to generate the complex architecture of the nervous system. It is now evident that many of these neuroguidance cues and their receptors are active during development and are also expressed in the adult nervous system. This suggests that neuronal guidance pathways are critical not only for neural wiring but also for ongoing function and maintenance of the mature nervous system. Supporting this view, these pathways continue to regulate synaptic connectivity, plasticity, and remodeling, and overall brain homeostasis throughout adulthood. Genetic and transcriptomic analyses have further revealed many neuronal guidance genes to be associated with a wide range of neurodegenerative and neuropsychiatric disorders. Although the precise mechanisms by which aberrant neuronal guidance signaling drives the pathogenesis of these diseases remain to be clarified, emerging evidence points to several common themes, including dysfunction in neurons, microglia, astrocytes, and endothelial cells, along with dysregulation of neuron-microglia-astrocyte, neuroimmune, and neurovascular interactions. In this review, we explore recent advances in understanding the molecular and cellular mechanisms by which aberrant neuronal guidance signaling contributes to disease pathogenesis through altered cell–cell interactions. For instance, recent studies have unveiled two distinct semaphorin-plexin signaling pathways that affect microglial activation and neuroinflammation. We discuss the challenges ahead, along with the therapeutic potentials of targeting neuronal guidance pathways for treating neurodegenerative diseases. Particular focus is placed on how neuronal guidance mechanisms control neuron-glia and neuroimmune interactions and modulate microglial function under physiological and pathological conditions. Specifically, we examine the crosstalk between neuronal guidance signaling and TREM2, a master regulator of microglial function, in the context of pathogenic protein aggregates. It is well-established that age is a major risk factor for neurodegeneration. Future research should address how aging and neuronal guidance signaling interact to influence an individual’s susceptibility to various late-onset neurological diseases and how the progression of these diseases could be therapeutically blocked by targeting neuronal guidance pathways.

Introduction

During development, neuronal guidance genes regulate cell migration, axon pathfinding, cell-cell communications, and synapse formation, thereby generating initial patterns of neural connectivity, and subsequently fine-tuning the neural architecture in combination with neural activity (Dickson, 2002; Guan and Rao, 2003; Roig-Puiggros et al., 2020; Sanes and Zipursky, 2020; Dorskind and Kolodkin, 2021; Luo, 2021; Südhof, 2021; Zang et al., 2021; Yuasa-Kawada et al., 2023). Precise neural wiring underlies brain function, whereas miswiring may lead to various neurological disorders, exemplified by mirror movement diseases and horizontal gaze palsy with progressive scoliosis (Jen et al., 2004; Srour et al., 2010). In parallel, neuronal guidance pathways coordinate the morphogenesis of various organs outside the nervous system. During adulthood, neuronal guidance genes modulate synaptic connectivity and immune cell trafficking in response to environmental stimuli, to maintain nervous system homeostasis. Increasing evidence supports the notion that dysregulation of neuronal guidance signaling causes defects in synaptic maintenance and disruption of brain homeostasis, contributing to a wide range of neurodegenerative and neuropsychiatric diseases (Schmidt et al., 2009; Van Battum et al., 2015; Gan et al., 2018). However, there is a lack of understanding regarding how miswiring and/or functional defects in neuronal guidance signaling lead to neurodegeneration.

Elucidating the molecular pathogenic mechanisms by which mutations or dysfunction in neuronal guidance genes affect susceptibility to– and initiation and progression of– neurological diseases remains a major challenge. Many pathogenic mechanisms, including abnormal protein aggregation and disruption of proteostasis, have been extensively investigated. Furthermore, several therapeutic approaches to neurodegenerative diseases are under development. Nevertheless, we still have no effective treatment for these diseases or strategies that can block disease progression (reviewed in Gan et al., 2018; Wilson et al., 2023).

In this review, we consider how dysregulated neuronal guidance signaling affects disease pathogenesis. Single nucleotide polymorphisms (SNPs) in neuronal guidance genes have been found in patients with various neurodegenerative diseases, including Alzheimer’s disease (AD), Parkinson’s disease (PD), and amyotrophic lateral sclerosis (ALS) (Lesnick et al., 2007, 2008; Lin et al., 2009; Van Battum et al., 2015). Because several neurodevelopmental and neuropsychiatric diseases show signs of neurodegeneration, together with neural circuit-level alterations, we have also included these diseases, namely autism spectrum disorders (ASD), major depressive disorder/bipolar disorder, and schizophrenia (SCZ). We discuss the challenges in deepening our understanding of the pathogenic mechanisms underlying these diseases. Finally, we present translational approaches for developing novel therapeutic strategies against neurodegenerative diseases, by targeting neuronal guidance pathways. Specifically, we consider how modulation of these pathways might help control the apparently detrimental, pro-inflammatory behavior and function of glia and immune cells, and their interactions with other cells in the nervous system, offering novel therapeutic avenues for addressing neuroinflammation and neurodegeneration.

Search Strategy

In this narrative review, we focused on studies that examined the roles of neuronal guidance signaling in neurodegenerative and neuropsychiatric diseases. Articles were selected based on relevance to the review’s scope, emphasizing studies published in the last 20 years.

An Overview: Neuronal Guidance Signaling in Neurodevelopment and Neurodegenerative Diseases

Following a rapid advancement of molecular studies in the 1990s, neurodevelopmental research has been directed toward identifying various guidance cues and receptors (Additional Figure 1 ^{(2.2MB, tif)} ) and uncovering the neural wiring mechanisms that are critical in the formation of neural architecture (Guan and Rao, 2003; Dickson and Zou, 2010; Dorskind and Kolodkin, 2021; Luo, 2021; Zang et al., 2021; Yuasa-Kawada et al., 2023). Since then, a few basic concepts have emerged. (1) The expression and function of genes that regulate neuronal guidance, encoding guidance cues, receptors, and downstream effectors, are now known to be subject to spatiotemporal control at multiple levels. (2) Neuronal guidance genes regulate various cellular responses, including attraction, repulsion, and adhesion. Notably, axonal responses to different cues can be switched in spatiotemporally regulated manners (Figure 1A). Additionally, the same cues and/or receptors can elicit different, even opposing, responses at different locations and times in context-dependent manners (for example, affected by different intracellular states of responding cells or the presence of co-presented guidance cues or co-receptors). (3) Neuronal guidance genes regulate both short- and long-range cell-cell communications. (4) There are complex crosstalk combinations at the levels of guidance cues, receptors, and downstream signal transducers. (5) Neuronal guidance genes regulate synaptic formation and plasticity. (6) The neuronal guidance system acts together with neural activity to form and remodel neural architecture. (7) Some neuroguidance cues, including NOGO and repulsive guidance molecule-a (RGMA), can act as myelin-associated inhibitors of axon regrowth after injury of the brain or spinal cord (Harel and Strittmatter, 2006; Mueller et al., 2009; Varadarajan et al., 2022; Winter et al., 2022); these inhibitory neuroguidance cues prevent axon rewiring and neuroregeneration. A recent study also revealed that silencing SLIT-ROBO signaling is required for long-distance nervous system repair and behavioral recovery (Delpech et al., 2024). (8) Neuronal guidance genes regulate cell migration, cell-cell communications, and stem cell production during organogenesis and tissue homeostasis inside and outside the nervous system. Thus, the effects of neuronal guidance signaling extend well beyond neural wiring, through modulation of the function of various cell types, such as glia, immune cells, and vascular cells.

Figure 1.

Roles of the neuronal guidance system under physiological and pathological conditions.

The functionality of neuronal guidance signaling changes during developmental, adult, and neurodegenerative stages. (A) Axon midline crossing. Growing axons sense guidance cues and exhibit specific responses, such as attraction, repulsion, and cell adhesion. Their responsiveness can be switched in spatiotemporally regulated manners. For example, in the vertebrate neural tube, commissural axons initially extend toward the ventral midline by sensing netrin-1 (NTN1) derived from the floor plate or ventricular zone, which acts as an attractive or adhesion-promoting cue, respectively (Dominici et al., 2017; Moreno-Bravo et al., 2019; Wu et al., 2019). Upon reaching the midline, the axons lose their responsiveness to NTN1 and other attractants, including sonic hedgehog, and acquire responsiveness to repellents, such as SLIT, so that they smoothly cross the midline (Dickson and Zou, 2010; Yuasa-Kawada et al., 2023). (B) During prenatal and adult stages, synapses are remodeled to generate the mature neural architecture. Microglia-mediated synaptic pruning contributes to the refinement of the neural architecture. Locally externalized PtdSer (ePtdSer) is sensed as an “eat-me” signal by its receptors, such as TREM2 and GPR56, in microglia, and activates microglia to selectively engulf the “marked” presynaptic endings. Semaphorins (e.g., SEMA7A) may act as retrograde signals, to transduce neural activity and eliminate inappropriate presynaptic endings. (C) Aβ-exposed hyperactive presynaptic and postsynaptic structures are removed by activated microglia, via ePtdSer signaling. This mechanism exhibits beneficial effects during the early stages of neurodegeneration and possibly detrimental effects in the late stages (See Rueda-Carrasco et al., 2023 for details). (D and E) Cell contact-dependent SEMA-PLXN pathways mediate microglia-astrocyte and neuron-glia interactions and regulate glial activities under pathological conditions. (D) SEMA4D-PLXNB1 signaling induces the formation of peri-plaque glial nets by regulating glial cell spacing around Aβ plaques, promotes neuroinflammation, and affects microglial phagocytic behavior (see Huang et al., 2024 for details). Note that there is no verification to date whether the corresponding ligand is SEMA4D, although it is likely. (E) SEMA6D-PLXNA1-TREM2 signaling, which mediates interactions between Aβ-exposed excitatory neurons and microglia, induces microglial activation and promotes neuroinflammation around Aβ plaques (see Albanus et al., 2023 for details). Aβ: Amyloid-β; C: caudal; DAM: disease-associated microglia; FP: floor plate; PLXN: plexin; R: rostral; RP: roof plate; SEMA: semaphorin.

Accumulation of amyloid and amyloid-like protein aggregates has been recognized as a major pathological hallmark for various neurodegenerative diseases (Alberti and Hyman, 2021; Sawaya et al., 2021). Since the identification of key disease-associated aggregation-prone proteins, such as amyloid-β (Aβ), tau, α-synuclein, and transactive response DNA-binding protein 43 kDa (TDP-43), studies have addressed the formation, accumulation, and propagation of pathogenic protein aggregates, leading to synaptic and neuronal loss and to neurodegeneration. In their search for effective therapies, researchers have explored how such protein aggregation and propagation processes could be blocked. While evidence suggests that neuronal guidance genes promote or suppress the progression of neurodegenerative diseases (e.g., Van Hoecke et al., 2012; Karch and Goate, 2015; Albanus et al., 2023; Huang et al., 2024), two critical and interrelated questions need to be addressed: (1) how neuronal guidance signaling modulates protein aggregation; and (2) how neuronal guidance signaling affects other crucial aspects of neurodegeneration, especially aggregate propagation and neuroinflammation. Addressing these questions will be important for the development of therapeutic strategies aimed at mitigating debilitating neurodegenerative diseases. It is hypothesized that targeting neuronal guidance signaling may be effective for controlling microglia-mediated neuroinflammation and neurodegeneration.

Microglia are central nervous system (CNS)-resident macrophages (Ginhoux et al., 2010; Prinz et al., 2021). Under physiological conditions, microglia maintain brain homeostasis through phagocytosis and regulate synaptic pruning in the developing and adult brain (Sierra et al., 2010; Faust et al., 2021). Microglia suppress neural activity by sensing extracellular ATP released from excited neurons, to protect neural circuitry from excessive neuronal excitation (Badimon et al., 2020). During synaptic remodeling, microglia selectively engulf inappropriate synapses, which are marked by externalized phosphatidylserine (ePtdSer), a guidance cue that directs microglial process extension (Figure 1B; details will be discussed later) (Li et al., 2020; Scott-Hewitt et al., 2020). Early in disease pathology, microglia remove ePtdSer-marked hyperactive synapses that are exposed by abnormal protein aggregates (Figure 1C; Badimon et al., 2020; Bartels et al., 2020; Rueda-Carrasco et al., 2023). Under sustained pathological conditions, however, microglia act as carriers of the engulfed protein aggregates and propagate aggregation-mediated pathology into unaffected brain regions; moreover, they engulf synapses in dysregulated manners (Eroglu and Barres, 2010; Hansen et al., 2018; d’Errico et al., 2022). Importantly, neuronal guidance signaling modulates states, dynamics, and functions of microglia and astrocytes, through neuron-glia interactions (e.g., Stogsdill et al., 2022). Neuronal guidance signaling can affect the activity and behavior of microglia and astrocytes, contributing to neuroinflammation and neurodegeneration, both positively and negatively (Figure 1D and E; discussed below in detail; Clark et al., 2021; Albanus et al., 2023; Huang et al., 2024).

Genetic and Transcriptomic Evidence for the Association Between Neuronal Guidance Genes and Neurodegeneration

Genome-wide association studies (GWAS) have identified causal genes and risk factors for various neurodegenerative and neuropsychiatric diseases. Single-cell RNA-sequencing analyses have revealed differential expression of neuronal guidance genes, which include risk factors of neurological diseases, in various cell types in patients, suggesting that dysregulated expression of neuronal guidance genes affects disease susceptibility. Individual neuronal guidance genes may act as a weak or moderate risk factor/modifier in one population, but not in others. Combinatorial effects of such factors may affect disease susceptibility. There is a need to cross-validate genetic, genomic, and transcriptomic findings in multiple datasets across different human populations. Another key point is that many risk factors of neurodegenerative diseases, which include components of neuronal guidance signaling, function in immune cells, including microglia (Table 1), as exemplified by TREM2 (Deczkowska et al., 2020; Colonna, 2023), EPHA1 (Karch and Goate, 2015), and PLXNA1 (Kumanogoh and Kikutani, 2013; Worzfeld and Offermanns, 2014).

Table 1.

Neuronal guidance cues, receptors, and their function in neurological diseases

Cues (and expressing cells)	Receptors	Co-receptor and binding partners	Receptor-expressing cells	Diseases involved	Mechanisms	References
Canonical cues
NTN1	UNC5	DCC	DA neurons (SN)	AD, PD, ALS	Neuronal survival/apoptosis	Williams et al., 2006; Tang et al., 2008; Ahn et al., 2020; Jasmin et al., 2021
RGMA (DA neurons [SN])	NEO1	UNC5	DA neurons (SN) and possibly immune cells	PD	Induce neuronal death	Korecka et al., 2017
SLIT2	ROBO1/2	APP?	Purkinje cells, Microglia	SCA, SCZ, glioblastoma	Dendrite self-avoidance; microglia and macrophage chemotaxis and invasion	Gibson et al., 2014
SLIT2	ROBO4		Endothelial cells	Retinal vascular diseases	Vascular stabilization	Jones et al., 2009
SEMA3A (Schwanm cells)	PLXNA1	NRP1/2, CRMP1/2	Motor neurons	ALS	Axon retraction and degeneration	Numata-Uematsu et al., 2019; Asano, 2022; Kawamoto et al., 2022
SEMA3A	PLXNA2			SCZ		Mah et al., 2006
SEMA3A (neurons)	PLXNA4	NRP1/2, CRMP1/2	Microglia	PD, ALS	Link Aβ and tau pathologies; organize microglial states	Kang et al., 2016; Chung et al., 2021; Stogsdill et al., 2022
SEMA4D (microglia)	PLXNB1		Astrocytes	AD, PD, ALS, MS, SCZ	Promote or suppress microglia activation; promote neuroinflammation by controlling glial net formation	Clark et al., 2021; Huang et al., 2024
SEMA4C/4D/4G	PLXNB2		Microglia	ALS, SCZ, SCI	Enhance microglial motility; required for peripheral nerve regeneration	Saez-Atienzar et al., 2021; Li et al., 2022; Humphrey et al., 2023
SEMA5A	PLXNA2/C1			PD, ALS, SCZ		Ho et al., 2019
SEMA6B	PLXNA2			SCZ, intellectual disability		Mah et al., 2006; Allen et al., 2008
SEMA6D (damaged excitatory neurons)	PLXNA1	TREM2, DAP12/10	Microglia	AD	Microglial activation	Albanus et al., 2023
SEMA6D (macrophage)	PLXNA4		Macrophage		Anti-inflammatory macrophage polarization (reverse signaling)	Kang et al., 2018
SEMA6D (tumor and nonhematopoietic cells)	PLXNA4		CD8+ T cells	Head and neck cancer	Suppress anti-tumor CD8+ T cell activation and infiltration into tumor microenvironments (forward signaling)	Hirai et al., 2024
SEMA7A	PLXNC1			PD, SCZ		Lesnick et al., 2007
EFNA1	EPHA1		Epithelial cells, T cells, microglia	AD	AD risk factor (may stimulate neuroinflammation)	Karch and Goate, 2015; Andrews et al., 2023
EFNA3 (astrocytes)/A1	EPHA4		Neurons	AD, ALS	Dendritic spine retraction; regulate glutamate transporter trafficking; ALS modifier (LOF variant is protective for ALS)	Murai et al., 2003; Filosa et al., 2009; Van Hoecke et al., 2012; Fu et al., 2014
EFNB1 (astrocytes)	EPHB1		Injured neurons	ALS, PD, MS	Activate neuroprotective responses of astrocytes (reverse signaling)	Tyzack et al., 2017
EFNB1/B2	EPHB2	NMDAR	Neurons	AD, MS, anti-NMDAR encephalitis	Regulate dendritic spine growth, NMDAR trafficking, and NMDAR-dependent LTP	Cissé et al., 2011; Mikasova et al., 2012
EFNB3 (microglia)	EPHB3		Astrocytes	MS	Promote neuroinflammation	Clark et al., 2021
Non-canonical cues
RELN (neurons)	VLDLR/APOER2	NRP1, DAB1	Neurons, glia	AD	Reduce tau phosphorylation in the hippocampus	Lopera et al., 2023
VAPB	EPHA4			ALS (ALS8)		Tsuda et al., 2008
CLU (astrocytes)	PLXNA4		Neurons	AD	Maintain dendritic spine density	Kang et al., 2016
CLU (astrocytes)	TREM2		Neurons	AD	Maintain dendritic spine density	Deczkowska et al., 2020; Chen et al., 2021a; Colonna, 2023
ePtdSer	TREM2		Microglia	AD	Activate microglial phagocytosis of synapses	Scott-Hewitt et al., 2020; Rueda-Carrasco et al., 2023
ePtdSer	GPR56		Microglia	AD	Activate microglial phagocytosis of synapses	Li et al., 2020
CXCL12 (SDF-1)	CXCR4		T cells	PD-DLB	Immune cell trafficking	Gate et al., 2021
CCL3/4/5 (microglia)	CCR5		Neurons		Inhibit autophagy and promote neurodegeneration	Shen et al., 2022; Festa et al., 2023
Disease-associated ligands
Aβ	PLXNA4	NRP2	Neurons	AD	Promote tau phosphorylation	Chung et al, 2021
Aβ	EPHA4		Neurons	AD	Induce synaptic dysfunction	Fu et al., 2014
Aβ	TREM2		Microglia	AD	Regulate microglial responses to Aβ	Zhao et al., 2018b; Wang et al., 2022
APOE	TREM2		Microglia	AD		Yeh et al., 2016
TDP-43	TREM2		Microglia	ALS	Protect against TDP-43 neurotoxicity	Xie et al., 2022

Aβ: Amyloid-β; AD: Alzheimer’s disease; ALS: amyotrophic lateral sclerosis; CXCL12: CXC-chemokine ligand 12; CXCR4: CXC-chemokine receptor 4; DA: dopamine; DCC: deleted in colorectal cancer; DLB: dementia with Lewy bodies; EFN: ephrin; EPH: erythropoietin-producing human hepatocellular; ePtdSer: externalized phosphatidylserine; GPR56: G protein-coupled receptor 56; LOF: loss-of-function; LTP: long-term potentiation; MS: multiple sclerosis; NEO1: neogenin1; PD: Parkinson's disease; ROBO: Roundabout; SN: substntia nigra; SCZ: schizophrenia; TREM2: triggering receptor expressed on myeloid cells 2; UNC5: uncoordinated-5.

Below, we refer to several representative neurodegenerative diseases, in which neuronal guidance pathways contribute to pathogenesis. Experienced clinicians and researchers may prefer to omit the discussion regarding genetic and transcriptomic analyses on each disease and proceed to the next section, “Toward understanding molecular mechanisms underlying aberrant neuronal guidance signaling in neurodegenerative diseases.”

Alzheimer’s disease

AD is the most common form of dementia (Scheltens et al., 2021). During the early stages of AD, hippocampal atrophy can be detected, and cortical atrophy further spreads as the disease progresses. AD is pathologically characterized by extracellular senile plaques composed of Aβ species and intracellular neurofibrillary tangles (NFTs) containing hyperphosphorylated and misfolded tau, which lead to neuronal loss and eventual cognitive decline, memory loss, and functional impairment (Long and Holtzman, 2019; Andrews et al., 2023). In the initial process, amyloid precursor protein (APP) is cleaved to form Aβ (ranging from 38 to 43 amino acid residues). Aβ monomers can form oligomers (protofibrils), polymers (fibrils), and insoluble amyloid plaques through β-sheet stacking and lateral associations (Shankar et al., 2007; Benilova et al., 2012; Sawaya et al., 2021). In the amyloid-cascade hypothesis, Aβ accumulation has been recognized as the primary event that drives NFT formation and disease progression (Hardy and Higgins, 1992; McGeer and McGeer, 2013).

GWAS have identified more than 70 susceptibility loci for AD (Karch and Goate, 2015; Andrews et al., 2023), including several neuronal guidance genes. For example, EPHA1, encoding an ephrin receptor of the EPH tyrosine kinase family (A-class), was identified as a major genetic risk for late-onset AD (LOAD), which accounts for most AD cases (Table 1; Carrasquillo et al., 2011; Hollingworth et al., 2011; Naj et al., 2011; Kania and Klein, 2016; Schwartzentruber et al., 2021; Andrews et al., 2023). EPHA1 is expressed in epithelial cells, T cells, and microglia (Karch and Goate, 2015; Villegas-Llerena et al., 2016; Hampel et al., 2020). A SNP near EPHA1 (located in its promoter region), rs11767557, has been strongly associated with reduced risk for LOAD, although EPHA1 expression was not significantly changed in most AD brain samples (Karch and Goate, 2015), suggesting gene regulation mechanisms distinct from expression-level control. A meta-analysis of GWAS data revealed another EPHA1 variant, rs11771145, to be associated with reduced risk for LOAD (Lambert et al., 2013). Patients with AD (PwAD) having the minor allele of EPHA1 (rs11771145) exhibited less brain atrophy and a greater cerebral metabolic rate for glucose in the lateral occipitotemporal gyrus and inferior temporal gyrus (Wang et al., 2015a). Meanwhile, a significant association of EPHA1 (rs11771145) with Aβ or tau aggregation was not detected in this study. Furthermore, EPHA1 promotes microglia-mediated neuroinflammation via CXCL12-CXCR4 signaling (Ma et al., 2021). Thus, the major allele of EPHA1 may be involved in the propagation of Aβ/tau pathology via neuroinflammation.

PlexinA4 (PLXNA4), encoding a semaphorin (SEMA3 and SEMA6) receptor (Additional Figure 2 ^{(2MB, tif)} ), is also linked to AD risk, and expression of full-length PLXNA4 has been correlated with the Clinical Dementia Rating score (Jun et al., 2014). In cultured neurons, SEMA3A stimulation of full-length PLXNA4 increased tau phosphorylation by GSK3B (a major tau kinase), without affecting Aβ production (Figure 2A). This signaling event occurs in a CRMP2 (DPYSL2) phosphorylation-dependent manner. Furthermore, short, secreted isoforms of PLXNA4 suppressed tau phosphorylation, suggesting that PLXNA4 variants may differentially modulate tau phosphorylation and aggregation (Jun et al., 2014). PLXNA4 acts as a receptor not only for semaphorins but also for clusterin (CLU; a.k.a. apolipoprotein J [APOJ]) (Figure 2A; Kang et al., 2016; Foster et al., 2019). CLU is an extracellular chaperone and AD risk factor; it binds to Aβ and inhibits Aβ aggregation (Oda et al., 1994; Nuutinen et al., 2009; Spatharas et al., 2022). Mice with heterozygous or homozygous genetic deficiency in Plxna4 showed hyperactivity in an open-field test and impairments in learning and memory (Kang et al., 2016). Thus, CLU and PLXNA4 are biochemically and genetically associated with AD. Several other candidates for CLU receptors have been identified, such as LRP2/8 (lipoprotein receptor-related protein 2/8) and TREM2 (triggering receptor expressed on myeloid cells 2) (Figure 2A and B; Deczkowska et al., 2020; Colonna, 2023).

Figure 2.

Networks of SEMA-PLXN and TREM2 signaling that mediate neuron-glia and neuroimmune interactions around disease-associated protein aggregates.

Representative SEMAs, PLXNs, TREM2, and related signaling pathways in neurological diseases, including Alzheimer’s disease (AD) and amyotrophic lateral sclerosis (ALS), are depicted (see also Yuasa-Kawada et al., 2023, Figure 4). (A) Interactions of TREM2 and clusterin (CLU) with components of neuronal guidance signaling and abnormal protein aggregates, such as Aβ and TDP-43. TREM2 is involved in multiple diseases and may interact with multiple co-receptors. In (A), RND1 and RHOD-mediated control of the RAS/RAP-GAP activity of PLXNA4 is omitted. Risk factors for neurodegenerative diseases are shown in red. (B) SEMA6D-PLXNA1-TREM2 signaling. PLXNA1 and TREM2 form a receptor complex for SEMA6D and regulate microglial activation and function, including survival, proliferation, phagocytosis, and secretion of various cytokines. Additionally, the SEMA6D-PLXNA1 pathway utilizes both forward and reverse signaling. As shown in (A and B), TREM2 receives signals from a wide range of cues, such as Aβ, CLU, APOE, ePtdSer, and indirectly SEMA6D. (C) SEMA4D-PLXNB1/B2 signaling. PLXNB1 and PLXNB2 pathways seem to share most of their ligands and downstream signaling components. The RAS/RAP-GAP activity of PLXNB1 is regulated by RND1 and opposingly by RHOD (Liu et al., 2021). The SEMA6D-PLXNA1 pathway also utilizes both forward and reverse signaling. Such forward and reverse signaling pathways mediate bidirectional cell-cell communications in health and diseases, which can be general principles in neuronal guidance. APOE: Apolipoprotein E; APP: amyloid precursor protein; Aβ: amyloid-β; CLU: clusterin; DAP12: DNAX-activating protein of 12 kDa; GSK: glycogen synthase kinase; mTOR: mammalian target of rapamycin; NMDAR: N-methyl-D-aspartate-type glutamate receptor; NTN1: netrin; PI3K: phosphatidylinositol 3-kinase; PLXN: plexin; SEMA: semaphorin; TDP-43: transactive response DNA-binding protein 43 kDa; TLR: Toll-like receptor.

TREM2 is a risk factor for AD and ALS (Guerreiro et al., 2013; Jonsson et al., 2013; Cady et al., 2014), acting as a cell-surface receptor of the immunoglobulin superfamily and the TREM subfamily expressed in tissue macrophages, including microglia. Importantly, TREM2 senses a broad array of anionic and zwitterionic lipids that are associated with Aβ fibrils and become exposed during cell death (Wang et al., 2015b; Colonna, 2023). Lipidated CLU and APOE interact with Aβ to promote Aβ clearance through binding to microglial TREM2 (Yeh et al., 2016). TREM2 also binds to Aβ oligomers proper, regulates microglial responses to Aβ, and suppresses microglia-dependent Aβ-associated tau seeding and spreading (Asai et al., 2015; Ulland et al., 2017; Zhao et al., 2018b; Wang et al., 2022; Jain et al., 2023). TREM2 variants with reduced activity for phagocytic clearance of amyloid aggregates, such as the R47H and R62H rare variants, have been associated with AD risk (Zhou et al., 2020b). TREM2^R47H shows reduced binding and uptake of lipidated APOE and CLU and impaired ability to induce microgliosis, thus facilitating the seeding and spreading of tau aggregates (Yeh et al., 2016; Leyns et al., 2019).

CLU and TREM2 appear to act as a versatile ligand-receptor axis that prevents the progression of several diseases, besides AD. Both CLU and TREM2 bind to TDP-43 and protect against TDP-43 neurotoxicity in ALS (Figure 2A; Gregory et al., 2017; Xie et al., 2022), suggesting that dysregulation of CLU-TREM2 interactions with diverse protein aggregates may be linked to various types of neurodegeneration.

A recent GWAS analysis revealed a new member of the netrin family, netrin-5 (NTN5; rs2452170), as a risk factor for AD (Wightman et al., 2021). However, the mechanistic link between netrin-5 and AD remains to be elucidated.

Transcriptomic analyses have identified many genes and modules whose expression and/or interactions are altered in the brains of PwAD. These AD-associated molecular networks include neuronal guidance genes, such as PLXNB1, which is expressed in astrocytes in the adult brain (Mostafavi et al., 2018). PLXNB1 acts as a signaling hub in forming glial networks that responds to Aβ, as discussed later (Figures 1D and 2C; Huang et al., 2024). PLXNB1 deletion resulted in reduced neuroinflammation and improved memory performance in AD mice (carrying APP/PSEN1), suggesting that PLXNB1 may be a therapeutic target for suppressing neuroinflammation in AD. Together, genetic and transcriptomic studies have begun to reveal the contributions made by multiple components of neuronal guidance pathways, such as EPHA1, PLXNA4, and PLXNB1, and their partners, including CLU and TREM2, to AD pathogenesis.

Parkinson’s disease

PD is one of the most common neurodegenerative diseases, second only to AD. It progressively affects movement, with patients exhibiting motor symptoms, such as resting tremor, bradykinesia, rigidity, and postural instability (Bloem et al., 2021). Patients (people) with PD (PwPD) also show troublesome non-motor symptoms, including apathy, depression, and insomnia. The major neuropathological hallmarks of PD are morphologically characteristic aggregates of α-synuclein, designated Lewy bodies, and massive dopamine (DA) neuron loss in the substantia nigra (SN), which cause deficient dopaminergic neurotransmission in the striatum.

Mutations in 26 PARK genes are known to cause familial PD, including rare mutations in α-synuclein (SNCA/PARK1 or PARK4) and parkin (PRKN/PARK2), and common mutations in leucine-rich repeat kinase 2 (LRRK2/PARK8) (Hattori and Mizuno, 2017; Blauwendraat et al., 2020; Vázquez-Vélez and Zoghbi, 2021). Furthermore, more than 90 PD risk-associated variants have been found (Bloem et al., 2021). Among them, several neuronal guidance genes have been associated with PD (Table 1). In a high-resolution GWAS, two SNPs within SEMA5A (rs7702187 and rs3798097) were identified as risk factors (Maraganore et al., 2005; Clarimon et al., 2006; Fung et al., 2006). By utilizing a genomic-pathway approach, various neuronal guidance genes, such as DCC, UNC5C, EPHA4, EPHB1, PLXNA2, PLXNC1, SEMA5A, SLIT3, NTNG1, CDC42, FYN, and GSK3B (Lesnick et al., 2007), were associated with PD susceptibility, suggesting that combinations of polymorphisms in neuronal guidance genes and other categories of risk genes may contribute to PD pathogenesis.

It has been found that expression levels of RGMA are increased in the SN of PwPD (Bossers et al., 2009). The repellent RGMA binds to the netrin receptor NEO1 and receptor complex composed of NEO1 and UNC5 (Figure 3A, inset; Hata et al., 2009; Siebold et al., 2017; Yamagishi et al., 2021). Targeted overexpression of Rgma in midbrain DA neurons in adult mice led to selective DA neuron loss and gliosis with activation of microglia and astrocytes, causing progressive motor dysfunction (Korecka et al., 2017). Consistently, in an MPTP (1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine)-induced PD mouse model, Rgma expression was increased in the SN. Suppressing Rgma signaling with a specific anti-RGMA antibody prevented DA neuron loss in mice (Oda et al., 2021). Collectively, RGMA is neurotoxic for DA neurons, and dysregulated RGMA signaling may contribute to PD progression.

Figure 3.

Neuroprotective and neuromodulatory roles of neuronal guidance signaling.

(A) NTN1-UNC5 signaling affects Parkinson’s disease progression. NTN1-UNC5B/C signaling regulates the survival of dopamine neurons in the substantia nigra (SN), whereas NTN1 deficiency triggers α-synuclein (αSyn) aggregation and dopamine neuron death in Parkinson’s disease. The inset shows that NTN1 and RGMA share multiple receptors. A schematic illustration of guidance cues and receptors is based on domain structures annotated by EMBL-SMART (http://smart.embl-heidelberg.de/). (B) RELN signaling ameliorates AD progression. RELN upregulates GSK3B phosphorylation and suppresses GSK3B activity, which blocks tau phosphorylation. The rare variant RELN-H3447R acts as a gain-of-function (GOF) mutant to suppress tau phosphorylation more effectively than the major variants, even in AD families. Interestingly, RELN and APOE seem to competitively bind to VLDLR and APOER2. (C) Astrocytic EFNA3-neuronal EPHA4 forward and reverse signaling pathways control synaptic efficacy and remodeling. EFNA3-EPHA4 forward signaling triggers dendritic spine retraction, whereas EPHA4-EFNA3 reverse signaling reduces glutamate transporter levels in astrocytes, increasing intrasynaptic glutamate levels and synaptic efficacy. EFNA3 and NEO1 in astrocytes opposingly regulate the glutamate levels at synaptic spaces. AD: Alzheimer’s disease; AMPK: Adenosine 5′-monophosphate-activated protein kinase; APOE: apolipoprotein E; DD: death domain; GAP: GTPase-activating protein; GLAST: glial glutamate/aspartate transporter; GLT1: glutamate transporter subtype-1; Ig: immunoglobulin-like; LTP: long-term potentiation; NTN: netrin; PI3K: phosphatidylinositol 3-Kinase; PM: plasma membrane; RGMA: repulsive guidance molecule-a; TSP: thrombospondin type 1; UPA: UNC5-PIDD-ankirin; ZU5: present in ZO-1 and UNC5.

In summary, several neuronal guidance genes, such as SEMA5A and RGMA, have been associated with PD. Evidence supports the notion that RGMA and netrin-1 (discussed later), and their receptors, NEO1, UNC5, and DCC, contribute to PD pathogenesis. Promisingly, a candidate for anti-RGMA antibody medicine has entered clinical trials (Huang et al., 2021; Kalluri et al., 2023).

Amyotrophic lateral sclerosis and frontotemporal lobar degeneration disease spectrum

ALS is a progressive fatal neurodegenerative disease that primarily affects upper and/or lower motor neurons in the motor cortex and spinal cord. It is pathologically characterized by the presence of TDP-43-containing proteinaceous inclusions in the motor neuron cytoplasm, which are found in more than 90% of cases (Feldman et al., 2022). Frontotemporal lobar degeneration (FTLD) is a common neurodegenerative disease causing early-onset dementia. Most FTLD cases are associated with tau or TDP-43 (Grossman et al., 2023). ALS and FTLD with TDP-43 proteinopathy (FTLD-TDP) can be concurrent as a continuum of the disease spectrum and share clinicopathological and genetic features.

More than 40 genes have been implicated in the pathogenesis of familial ALS (reviewed in Wang et al., 2023); for example, EPHA4 was proposed as a disease modifier of ALS (Van Hoecke et al., 2012). EPHA4 may affect disease onset, progression, and survival of patients with ALS (PwALS), while EPHA4 levels are correlated inversely with symptom severity; loss-of-function EPHA4 variants are associated with longer survival of PwALS (Van Hoecke et al., 2012). Pharmacological inhibition of Epha4 improved motor function and prolonged motor neuron survival in superoxide dismutase 1 (SOD1^G93A) transgenic mice (Van Hoecke et al., 2012; Zhao et al., 2018a). However, there are discrepancies among different studies regarding Epha4 in ALS mouse models. For example, one study reported that treatment with an Epha4-activating peptide increased the life span of SOD1^G93A mice (Wu et al., 2017). In another study, reducing Epha4 expression using antisense oligonucleotides did not affect motor function or life span in SOD1^G93A mice or another ALS model expressing profilin 1 mutant (PFN1^G118V) (Ling et al., 2018). Thus, the precise roles of EPHA4 in ALS pathogenesis remain to be confirmed. SNP analyses of PwALS identified other EFNs and EPHs, such as EFNA5 and EPHB1, as candidate mediators of cell-cell interactions and neuroprotective signals to block ALS progression (Lesnick et al., 2008; Lin et al., 2009; Tyzack et al., 2017; Table 1), suggesting the involvement of multiple EFN-EPH pathways in ALS.

Single-cell transcriptomic studies identified gene modules whose expression profiles are affected in patients with ALS-FTLD spectrum, including several neuronal guidance genes. For example, expression of components in the chemokine CXCL, CCL, and SEMA3/4/7 pathways, associated with neuroinflammation, vascularization, and blood–brain barrier (BBB) dysfunction, was altered in endothelial cells and astrocytes in patients with ALS and/or FTLD, suggesting defects in the neurovascular unit (NVU) in ALS-FTLD (Gerrits et al., 2022; Hasan et al., 2022; Humphrey et al., 2023). Transcriptomic analyses of spinal cord samples from PwALS identified microglial gene modules, whose expression levels were negatively correlated with disease duration, and novel risk loci for ALS-FTLD, including PLXNB2 (Figure 2C; Saez-Atienzar et al., 2021; Humphrey et al., 2023). A modified RNA-seq technique in axons enabled a comparison of the transcriptome in motor axons between healthy controls and PwALS (Nijssen et al., 2018). In mouse embryonic stem cell-derived motor neurons expressing human SOD1^G93A, Neuropilin 1 (Nrp1, a semaphorin receptor) and Drebrin1 (Dbn1, an actin-binding protein that regulates axon initiation and growth) were identified among downregulated transcripts, suggesting that neuronal guidance signaling may be affected in patients with SOD1-ALS (Nijssen et al., 2018).

Additionally, RGMA signaling promoted the uptake of SOD1 mutant protein into neurons, by compromising actin barrier function. This process may constitute previously unknown propagation mechanisms in SOD1-ALS pathogenesis (Shimizu et al, 2023). RGMA levels in the cerebrospinal fluid (CSF) were positively correlated with ALS severity and may be clinically useful as a biomarker for ALS.

Mutations in ubiquilin 2 (UBQLN2), encoding an aggregation-prone ubiquitin-like chaperone, lead to X-linked ALS (designated ALS15) and FTLD (Deng et al., 2011; Renaud et al., 2019; Todd et al., 2023). In ALS15, defects in autophagy and ubiquitin-proteasome systems have been proposed to contribute to neurodegeneration. Consistent with rodent studies, the expression of ALS-associated mutants of UBQLN2 (UBQLN2^ALS) caused neurodegeneration in Drosophila (Kim et al., 2018). A subsequent genetic screen identified two neuronal guidance genes, UNC5 and BEAT-1B/1C (the latter being members of the beaten path [BEAT] family, which are secreted by motor neurons and regulate motor axon guidance), as modifiers of ALS. Mutations in or knockdown of UNC5 or its binding partner, DCC/Frazzled, reduced neurodegenerative phenotypes (defects of neuromuscular junctions and locomotion) and extended life span of UBQLN2^ALS flies (Kim et al., 2023). These results suggest that UNC5 and DCC, two netrin receptors, mediate neurotoxicity downstream of UBQLN2 aggregation, further pointing to the convergence of netrin and UBQLN2 pathways in ALS pathogenesis.

In PwALS, expression of the repellent SEMA3A is upregulated in motor neurons and Schwann cells, which initially suggested that SEMA3A may cause axon degeneration (De Winter et al., 2006; Körner et al., 2016). Blocking Sema3a-Nrp1 signaling with an anti-Nrp1 antibody suppressed defects in neuromuscular junctions and the decline of locomotion and prolonged life span in SOD1^G93A mice (Venkova et al., 2014). However, contradictory evidence has been presented; astrocyte-specific Sema3a deletion caused the death of a subset of motor neurons in the mouse spinal cord (Molofsky et al., 2014). A subsequent study showed that SEMA3A acts as a trophic factor for motor neurons, but as a neurodegenerative cue for cortical neurons (Birger et al., 2018). SEMA3A may exert diverse effects on multiple neuron types in cellular context-dependent manners. The mechanisms by which SEMA3A contributes to ALS pathogenesis need to be further addressed.

A GWAS, using a polygenic risk score approach, in combination with single-nuclear RNA-seq (snRNA-seq) data, identified differentially expressed genes that are associated with ALS, including PLXNB2 (Saez-Atienzar et al., 2021). PLXNB2 expression was altered in PwALS and significantly associated with ALS risk (Saez-Atienzar et al., 2021; Humphrey et al., 2023). However, it remains uncertain whether and how PLXNB2 modulates the functionality of the cell populations affected by ALS, such as interneurons and glia. Because Plxnb2 regulates the proliferation and migration of neuroblasts in the subventricular zone of adult mice (Saha et al., 2012), PLXNB2 dysfunction in neurogenesis may play a role in ALS pathogenesis.

Another risk factor for ALS-FTLD is fused in sarcoma (FUS), an RNA-binding protein that forms cytoplasmic inclusions and causes neurodegeneration in the ALS-FTLD spectrum (Vance et al., 2009; Deng et al., 2014). Overexpression of human wild-type FUS in mice led to cognitive deficits, an age-dependent reduction in dendritic spine density and long-term potentiation (LTP) in the hippocampus (Ho et al., 2019). Transcriptomic analyses of the FUS mice revealed changes in expression of genes, such as Sema5a, in neurons and oligodendrocytes. FUS binds to Sema5a mRNA and regulates its expression (Ho et al., 2019), suggesting that FUS dysfunction coupled with aberrant semaphorin signaling may contribute to ALS pathogenesis.

Together, dysregulation and/or altered expression of several neuronal guidance genes are linked to the pathogenesis of the ALD-FTLD spectrum. As discussed later, the roles of several genes, such as PLXNB2 and a cognate, PLXNB1, in neuroinflammation have been unveiled.

Multiple sclerosis

Multiple sclerosis (MS) is the most common autoimmune, neuroinflammatory, and neurodegenerative disease of the CNS (Attfield et al., 2022; Garton et al., 2024; Jakimovski et al., 2024). MS is also a model disease of neuroinflammation, whose aspects are shared by several neurodegenerative diseases, including AD and PD. MS is pathologically characterized by BBB breakdown, chronic inflammation, and progressive CNS demyelination, which is caused by monocytes and autoreactive lymphocytes infiltrating into the brain parenchyma. Recent studies showed a significant link between Epstein-Barr virus infection and initiation of MS pathogenesis, as well as proinflammatory roles of infiltrating CD8⁺ T cells (Bjornevik et al., 2022; Lanz et al., 2022; Kavaka et al., 2024). GWAS and the network-based analyses have identified multiple susceptibility loci, including EFN-EPH signaling genes, as well as components in immunological pathways (Baranzini et al., 2009; Clark et al., 2021; Attfield et al., 2022; Goris et al., 2022; Jacobs et al., 2022).

Spinocerebellar ataxias

Spinocerebellar ataxias (SCAs) are a group of autosomal dominant neurodegenerative diseases characterized clinically by loss of balance and movement coordination, and pathologically by Purkinje cell (PC) loss and cerebellar atrophy. Forty-eight SCA subtypes have been reported (Klockgether et al., 2019). Among them, spinocerebellar ataxia 14 is caused by missense mutations or deletions in PRKCG, encoding a PC-specific protein kinase C (PKCγ) (Mezey et al., 2022). A transcriptomic study in heterozygous and homozygous mice carrying PKCγ (p.A24E) revealed changes in neuronal guidance pathways, including Slit1, Ntng1, Sema5a, and Plxnb3. SLIT1 and SLIT3 are expressed in cerebellar granule cells, whereas SLIT2 and ROBO2 are highly expressed in PCs. PC-specific Robo2 knockout mice exhibit gait alterations and motor deficits, showing that SLIT-ROBO signaling is involved in self-avoidance in PC dendrites (Gibson et al., 2014). Although the cerebellum has long been used as a platform in neurodevelopmental research, the pathogenic roles of neuronal guidance signaling in cerebellar diseases remain poorly defined.

Neuropsychiatric diseases, including schizophrenia, attention deficit hyperactivity disorder, bipolar disorder, depression, and autism

SCZ is a neuropsychiatric disease characterized by abnormal cognition and behavior. Despite decades of genetic and transcriptomic studies, there is still no consensus about gene pathways critical for SCZ pathogenesis (reviewed in Merikangas et al., 2022; commented by Hoffman et al., 2023; replied by Merikangas et al., 2023 and references within). Nonetheless, defects in prenatal cortical neurogenesis and impairment of synaptic connectivity have been implicated in SCZ. GWAS and meta-analyses identified PLXNA2 as a SCZ susceptibility gene (Table 1; Mah et al., 2006; Allen et al., 2008). Possibly in a related manner, expression of SEMA3A, a PLXNA2 ligand, is elevated in the cerebellum of patients with SCZ (Eastwood et al., 2003).

Rare heterozygous loss-of-function mutations in GRIN2A, encoding an N-methyl-D-aspartate (NMDA) receptor subunit, were found as a SCZ risk factor (Singh et al., 2022). In heterozygous Grin2a mutant mice, a new SCZ model, pyramidal neurons in the prefrontal cortex (PFC) showed drastic upregulation of components in semaphorin-plexin signaling, such as Plxnb1, Plxnb2, and Sema3g (Farsi et al., 2023). Thus, defective semaphorin signaling may contribute to SCZ pathogenesis.

Whole-exome/whole-genome sequencing and transcriptomic analyses revealed many genetic aberrations in SCZ and other mental illnesses. De novo single nucleotide variations in neuronal guidance genes, including SLIT1 and DCC, were identified in patients with SCZ (Gilman et al., 2012). In another study, de novo mutations in genes regulating neuronal guidance, synaptic transmission, transcription, and axon transport, such as SLIT2, SLIT3, ITGA3 (integrin), LAMA2 (laminin), CELSR2 (cadherin), ASAP2 (ArfGAP), and MYH9 (myosin), were identified in SCZ-affected gene co-expression networks in the PFC. Gene networks in neurogenesis were also affected in SCZ patient brains (Gulsuner et al., 2013). Thus, dysregulation of neuronal guidance and neurogenesis pathways may contribute to SCZ pathogenesis.

Heterozygous variants of the transmembrane semaphorin SEMA6B have been found in individuals with intellectual disability, a mostly monogenic neurodevelopmental disorder presenting limited intelligence and adaption (Cordovado et al., 2022). In cultured neurons, overexpression of variants or knockdown of SEMA6B reduced spine density. In midline-crossing commissural axons, axonal SEMA6B acts as a receptor for midline-derived PLXNA2 (reverse signaling) (Andermatt et al., 2014). In the experimental setting with endogenous SEMA6B knocked down, SEMA6B pathogenic variants failed to rescue commissural axon midline crossing, suggesting that functionally deficient SEMA6B variants may contribute to intellectual disability.

Mutations in genes encoding teneurin-4 and its binding partners, latrophilins (LPHNs), have been associated with bipolar disorder and SCZ (Croarkin et al., 2017; Yi et al., 2021). LPHN3 variants have also been implicated in attention deficit hyperactivity disorder (Franke et al., 2012), suggesting that dysfunctional teneurin-latrophilin pathways, which generate topographic hippocampal networks during development (Pederick et al., 2021), may contribute to the pathogenesis of neuropsychiatric diseases.

Several neuropsychiatric disorders, including depression, are sexually diverse in susceptibility. Female-specific decrease in SLIT1 expression was reported in the ventromedial PFC of patients with major depressive disorder. Depleting Slit1 reduced dendrite morphogenesis in ventromedial PFC neurons and neuronal excitability in female, but not male, mice. RNA-seq analyses detected stress-response signatures caused by Slit1 depletion in females (van der Zee et al., 2022). Thus, neuronal guidance genes may influence sex differences in susceptibility to depression.

ASD is a group of heterogeneous neurodevelopmental and psychiatric disorders characterized by impairments in social interactions, communications, sensory anomalies, repetitive behaviors, and intellectual disability (Lord et al., 2020). A machine learning-based approach revealed that axonal guidance and cell migration are among highly enriched biological processes associated with ASD (Di Giovanni et al., 2023). In a magnetic resonance imaging-based study, the cortical thickness was evaluated, and gene co-expression analyses were employed to identify disease-associated genes in patients with one of six different neuropsychiatric disorders, including ASD, bipolar disorder, major depressive disorder, obsessive-compulsive disorder, attention deficit hyperactivity disorder, and SCZ. Axon guidance in the prenatal co-expression gene cluster (e.g., SLIT1, ROBO2, SEMA7A, PLXNC1, SYNGAP1, NRXN1, PIK3CA, DOK4/6, and DOCK7) and synaptic function in the postnatal co-expression cluster (SHANK2) were identified as key gene ontology terms common in all six disorders (Writing Committee for the Attention-Deficit/Hyperactivity Disorder; Autism Spectrum Disorder; Bipolar Disorder et al., 2021). Differences in cortical thickness are associated with these disorders and related to axon guidance, brain lamination, and postnatal synaptic function.

Together, the roles of neuronal guidance signaling in the pathogenesis of various neurodevelopmental and neuropsychiatric disorders, in addition to neurodegenerative diseases, have begun to be unveiled. It may be noted that relatively similar sets of neuronal guidance genes, such as SEMAs and PLXNs, have been listed in both neurodegenerative and neuropsychiatric diseases, probably reflecting microglial dysregulation of synaptic function and neuroinflammation. Ongoing and future research will focus on how dysregulated neuronal guidance genes and their transcript-isoform diversity contribute to various neurological diseases (Patowary et al., 2024).

Toward Understanding Molecular Mechanisms Underlying Aberrant Neuronal Guidance Signaling in Neurodegenerative Diseases

Recent studies have revealed that neuronal guidance signaling regulates synaptic plasticity and neural remodeling (reviewed by Sanes and Zipursky, 2020; Südhof, 2021). In the following sections, we discuss molecular pathological mechanisms by which defects in neuronal guidance signaling contribute to various aspects of the pathogenesis of neurodegenerative diseases. We then review recent advances in therapeutic approaches to these diseases.

Microglia are executors of synaptic pruning and neural remodeling in physiological contexts, which are essential steps for establishing and maintaining the neural architecture (Paolicelli et al., 2011; Faust et al., 2021). Recent data suggest that microglia and immune cells infiltrating into the brain parenchyma may dramatically alter the symptomatic course of neurodegenerative diseases (e.g., Sulzer et al., 2017; Gate et al., 2021; Chen et al., 2023; Dejanovic et al., 2024). Importantly, activation of Th17 cells (helper memory CD4⁺ T cells expressing the proinflammatory cytokine interleukin (IL)-17) by α-synuclein peptides was detected in patients with the disease spectrum of PD-dementia with Lewy bodies (Gate et al., 2021). Such T-cell populations upregulated CXCR4. Its ligand, CXCL12/SDF-1, was also associated with PD-dementia with Lewy bodies. Thus, a renewed theory has been emerging that dysregulated mechanisms of chemokine and neuronal guidance signaling affect neuron-glia and neuroimmune interactions, which are also coupled with immune cell trafficking and their infiltration into the brain parenchyma. These processes promote neuroinflammation, eventually directing microglia and immune cells to attack functional synapses, neurites, and even cell bodies. We will discuss ongoing research efforts that have been addressing how the brain is protected and pathologically destroyed by the immune system, and how targeting neuronal guidance signaling may modulate the detrimental behavior and upregulate beneficial effects of microglia and immune cells.

Synaptic Dysfunction and Neuronal Death Caused by Protein Aggregation

Synaptic dysfunction and loss are pathological features of many neurodegenerative diseases (Tzioras et al., 2023; Dejanovic et al., 2024). Evidence supports the notion that dysregulated neuronal guidance signaling causes defects in synaptic connectivity, which may lead to miswiring and neurodegeneration. We first discuss that Aβ-stimulated Epha4 signaling triggers synaptic dysfunction in the hippocampus of AD mice (APP/PSEN1) (Figure 2A; Fu et al., 2014). EPHA4 is also a risk factor for ALS (Van Hoecke et al., 2012). Consistently, Epha4 activation led to dendritic spine retraction (Fu et al., 2007; Kania and Klein, 2016). Soluble Aβ oligomers, which cause synaptic loss, activate Epha4 in mouse hippocampal neurons, indicating that Aβ oligomers are pathogenic ligands for EPHA4. Blocking postsynaptic Epha4 by its small-molecule inhibitor, rhynchophylline, ameliorated synaptic dysfunction in the hippocampus of AD mice (Fu et al., 2014), suggesting a new therapeutic approach to AD.

In contrast to EPHA4 function in promoting AD progression, EPHB2, belonging to the EPH receptor B-class family, plays an opposite role (Figure 2A; Cissé et al., 2011). EPHB2 activation promotes dendritic spine growth and maturation. Aβ-EPHB2 binding triggers proteasomal degradation of EPHB2. Consistently, EPHB2 is seen to be depleted in AD brain samples and mouse models (Cissé et al., 2011). EPHB2 downregulation reduces NMDA-type glutamate receptor (NMDAR)-mediated synaptic transmission and LTP in the dentate gyrus. In AD mice (human APP transgenic), overexpression of Ephb2 in the dentate gyrus improved NMDAR-dependent LTP and cognitive behavior (Cissé et al., 2011).

Relatedly, in anti-NMDAR encephalitis, anti-NMDAR autoantibodies perturb NMDAR-EPHB2 interactions, diverting NMDAR trafficking from the cell surface to the degradation pathway (Mikasova et al., 2012). EPHB2 activation by its ligand, EFNB2, suppresses aberrant NMDAR trafficking, thereby upregulating surface levels of NMDAR. The imbalance among multiple EPH pathways, including EPHA1 (AD risk factor), EPHA4 (ALS modifier), and EPHB2, may affect the progression of various types of neurodegenerative and neuroinflammatory diseases.

Homeostatic synaptic scaling is a key mechanism that enables neurons to respond to changes in their firing rates, by regulating neurotransmitter receptor trafficking, and to optimize neuronal activity at the synapses (Turrigiano, 2017). Semaphorin signaling contributes to the balance between excitatory and inhibitory synaptic transmission. At excitatory synapses of cortical neurons in response to SEMA3F, NRP2-PLXNA3 complexes interact with postsynaptic AMPA-type glutamate receptors and downregulate their surface levels (Wang et al., 2017b). Microglia-derived tumor necrosis factor alpha (TNFα) also modulates homeostatic scaling (Renno et al., 1995; Pandey et al., 2022). Notably, NTN-DCC signaling is required for NMDAR-dependent LTP in the adult hippocampus (Glasgow et al., 2018, 2020). In an autocrine manner, activity-dependent secretion of netrin-1 from postsynaptic dendritic spines elevates AMPA-type glutamate receptor surface insertion at the CA3-CA1 excitatory synapses, thereby enhancing synaptic transmission. Both postsynaptic and presynaptic contributions of DCC to LTP were revealed in brain region-specific conditional knockout mice. Therefore, an imbalance of cytokines, chemokines, and neuroguidance cues in the homeostatic synaptic transmission machineries may contribute to various neurological diseases, including epilepsy, learning disability, and cognitive disorders (Van Battum et al., 2015; Yuasa-Kawada et al., 2023).

In addition to the above-mentioned function of chemotropic and adhesive neuronal guidance activities, netrins can act as trophic factors. UNC5s and DCC may act as dependence receptors for netrins (Figure 3A). Netrin signaling has been linked to neurodegenerative diseases, including PD and AD. Netrin-1 is highly expressed in DA neurons of the SN of human and rodent brains. In PwPD, netrin-1 is depleted in the SN (Ahn et al., 2020). Consistently, genetic deletion in netrin-1 in the SN results in DA neuron death in mice (Jasmin et al., 2021).

Upon netrin-UNC5B binding, the GTPase domain of AGAP2 (Arf GTPase-activating protein with GTPase, Ankyrin repeat, and Pleckstrin homology domain 2; a.k.a. phosphoinositide-3 kinase enhancer L [PIKE-L]), a brain-specifically expressed GTPase, interacts with UNC5B, to enhance neuronal survival via the PI3K/AKT pathway (Figure 3A, left; Tang et al., 2008). AGAP2 binds to α-synuclein to suppress its aggregation under physiological conditions; however, upon netrin-1 depletion, α-synuclein aggregates sequester AGAP2 into Lewy bodies to induce neuronal death (Figure 3A, right; Kang et al., 2017); this explains the pathogenic mechanism by which α-synuclein aggregates induce DA neuron loss in PwPD. In parallel, netrin-1 deficiency activates MST1, a Hippo homolog, and proapoptotic Ser/Thr kinase, which phosphorylates UNC5B at T428, promoting its pro-apoptotic activity (Figure 3A, right). MST1 activation is accompanied by phosphorylation of LATS1 (Ser/Thr kinase) and YAP1 (nuclear effector), leading to a reduction of YAP levels and DA neuron apoptosis (Ahn et al., 2020).

Thus, in contrast to RGMA, netrin-1 exerts neuroprotective effects on DA neurons in the SN, suggesting the involvement of netrin-1 depletion in PD pathogenesis. Consistently, in vivo, administration of recombinant netrin-1 restored axonal projections in DA neurons of PD mice (Jasmin et al., 2021).

The cognate receptor, UNC5C, is expressed in the hippocampus and cerebellum, as well as in midbrain DA neurons (Wetzel-Smith et al., 2014; Chen et al., 2022). An autosomal dominant rare mutation in UNC5C, T835M, has been associated with LOAD (Wetzel-Smith et al., 2014). When expressed in cultured cells, UNC5C^T835M increases apoptosis, compared with wild-type UNC5C. Upon netrin-1 depletion, cleavage of wild-type UNC5C by δ-secretase increases its proapoptotic activity, exacerbating hippocampal pathology in AD mice and DA neuron degeneration in PD mice (Figure 3A, right; Chen et al., 2021b, 2022). Cleaved UNC5C^T835M further elevates its cytotoxic activity. Thus, strategies enhancing netrin signals might be possible for treating AD and/or PD. However, another study reported that UNC5 induces neuronal apoptosis, irrespective of netrin-1 (Williams et al., 2006). The significance of netrin-UNC5 signaling in regulating neuronal survival and loss remains to be clarified.

Roles of APP and Aβ in AD pathogenesis have been extensively studied. APP is a transmembrane protein whose proteolytic cleavage by secretases generates pathogenic Aβ species (Figure 2A). Studies in rodents have revealed potential roles of APP in regulating neuronal migration and cortical lamination during development (Herms et al., 2004; Young-Pearse et al., 2007; Callahan et al., 2017). The mass spectrometric analysis identified amyloid precursor-like protein 1 as an interactor of the netrin-DCC complex (Lourenço et al., 2009), suggesting that APP family members act as co-receptors for netrin and/or modulators of netrin-DCC signaling. Additionally, netrin-APP binding suppressed the production of Aβ (1–40 and 1–42) in AD mice (Lourenço et al., 2009). When netrin-1 was delivered intracerebroventricularly in AD mice, Aβ levels were decreased and behavioral deficits were ameliorated (Lourenço et al., 2009). Thus, therapeutic approaches targeting netrin-1 function appear promising for AD and PD.

Furthermore, APP acts as a receptor for SLIT. Dual deficiencies of APP and a cognate protein, amyloid precursor-like protein 2, abolished SLIT repulsion of neurons migrating from olfactory bulb explant cultures (Wang et al., 2017a), suggesting that APP forms a SLIT receptor complex with ROBO. These data demonstrate the physiological roles of APP in neuronal guidance, further implicating SLIT-APP signaling in AD pathogenesis.

CLU is a ligand for PLXNA4 and TREM2 (Figure 2A); importantly, these molecules are genetic risk factors for LOAD (Harold et al., 2009; Lambert et al., 2009; Jun et al., 2014; Kang et al., 2016; Yeh et al., 2016; Foster et al., 2019). In the CNS, CLU is secreted by astrocytes and localizes to the presynaptic puncta of excitatory neurons. Genetic deletion of CLU reduced dendritic spine density in hippocampal neurons and impaired excitatory synaptic transmission (Chen et al., 2021a). In wild-type mice, astrocytic CLU overexpression promoted excitatory synaptic transmission, while, in AD mice (5xFAD), astrocytic CLU overexpression reduced Aβ pathology and rescued Aβ-induced presynaptic impairment (Chen et al., 2021a). CLU may act as an anti-amyloid synaptogenic ligand, possibly by binding to TREM2 and/or PLXNA4. Interestingly, PLXNA4, but not PLXNA1, A2, or A3, binds to Aβ (Chung et al., 2021). PLXNA4 and NRP2 appear to form a receptor complex for Aβ (Figure 2A). In AD mice (5xFAD, tau P301L-JNPL3), knocking down Plxna4 blocked CDK5 activation and reduced Aβ-induced phosphorylation and aggregation of tau in the hippocampus, ameliorating spatial memory impairment (Chung et al., 2021). However, it should be noted that heterozygous and homozygous deletion of Plxna4 in wild-type mice resulted in impairments in learning and memory (Kang et al., 2016). The exact roles of PLXNA4 function need to be further clarified. Collectively, SEMA-PLXN signaling can modulate Aβ and tau pathologies. PLXNA4 may also crosstalk with TREM2, similar to PLXNA1 (Figure 2A and B). The detailed mechanisms of SEMA-PLXN and TREM2 signaling in neuroinflammation will be discussed later.

Roles of non-canonical neuroguidance cues and signaling proteins, such as WNT pathways, in modulating AD, PD, and other neurological disorders have been studied (Freitas et al., 2023). This pathway physiologically regulates planar cell polarity (Zou, 2020) and pathologically modulates Aβ production. Aβ oligomers induce expression of Dickkopf-1 (DKK1), a secreted antagonist that blocks CTNNB1 (β-catenin)-dependent canonical WNT signaling. Consistently, DKK1 expression is upregulated in AD brains (Caricasole et al., 2004). Non-canonical WNT signaling enhances Aβ production, and DKK1 downregulates the canonical WNT signaling that suppresses Aβ production (Elliott et al., 2018). In a pathogenic positive-feedback loop, Aβ-induced DKK1 may switch WNT signaling from the canonical to non-canonical mode, and may further enhance Aβ production and synaptic loss. Thus, DKK1-neutralizing antibodies may have therapeutic potential against AD through suppressing Aβ-induced synapse disassembly (Purro et al., 2012).

A recent clinical study identified a rare variant in RELN, encoding another non-canonical neuroguidance cue that confers resilience to familial AD (Lopera et al., 2023). A male heterozygous for the RELN^H3447R variant was resilient to autosomal dominant AD caused by a PSEN1-E280A mutation (Figure 3B). This individual presented severe Aβ burden but relatively low tau pathology in the entorhinal cortex. The RELN^H3447R variant may suppress the progression of the Aβ-tau cascade. The huge extracellular matrix ligand RELN (3460 amino acid residues) binds to two APOE receptors, VLDLR and APOER2 (D’Arcangelo et al., 1999), and to EFNs (Sentürk et al., 2011), leading to tyrosine phosphorylation of DAB1 and suppressing tau phosphorylation by inhibiting GSK3B activity (Figure 3B; Hiesberger et al., 1999; Beffert et al., 2002; Alexander et al., 2023). RELN-H3447R turned out to be a gain-of-function variant that enhances DAB1 phosphorylation and reduces tau phosphorylation in the hippocampus of AD mice (Lopera et al., 2023). Because the C-terminal basic region of RELN interacts with NRP1 (Kohno et al., 2020), the H3447R mutation at the C-terminal basic region may affect RELN-NRP1 interactions. Although the mechanistic details by which this RELN variant exerts stronger activities than the wild-type to protect against AD remain unclear, a new therapeutic potential of targeting RELN signaling is presaged.

Taken together, dysregulation of multiple neuronal guidance pathways may lead to synaptic loss and neuronal death. A deeper understanding of the signaling pathways of netrins, semaphorins, ephrins, and RELN will provide new insights into drug development for treating AD and other types of dementia.

Axonal and Dendritic Retraction, Pruning, and Degeneration

In addition to synaptic loss, defects in neuritic maintenance cause brain dysfunction in patients with neurodegenerative diseases. For example, clinical symptoms of ALS and peripheral neuropathies initially appear upon the degeneration of distal axons, which is followed by distal-to-proximal progression (known as “dying-back” pathology), leading to axon fragmentation and removal of the axon debris by glia (Raff et al., 2002; Neukomm et al., 2014; Coleman and Höke, 2020; Varadarajan et al., 2022).

To fine-tune the neural circuitry during development, excessive axons and dendrites are eliminated in three modes: neuritic retraction (small-scale pruning), shedding of axosomes (organelle-containing axonal compartments), and degeneration of neuritic branches (large-scale elimination) (Neukomm et al., 2014). The refinement processes are highly regulated in which appropriately connected axon branches are maintained and stabilized, whereas inappropriately targeted axon branches are selectively removed without causing neuronal cell body death (Luo and O’Leary, 2005; Neukomm et al., 2014). Several neuronal guidance genes have been implicated in neural remodeling, as exemplified by Sema3F-dependent retraction of transiently formed infrapyramidal tracts in the hippocampus (Riccomagno et al., 2012; Ziak et al., 2020) and EFNB3-dependent pruning of hippocampal mossy fibers (Xu et al., 2009). Injury-triggered axon degeneration proceeds in a programmed manner, called Wallerian degeneration (WD) in which axons degenerate distal to the injury site through disintegration of the cytoskeleton and axonal fragmentation (Coleman and Höke, 2020). Importantly, morphological alterations and degeneration of axons, similar to those in WD but without physical axon damage, have been found in neurodegenerative diseases, including PD (Chu et al., 2012). Understanding molecular mechanisms that trigger axon degeneration will be critical for identifying novel therapeutic targets for neurological diseases.

There is limited evidence on whether abnormal neuronal guidance signaling directly triggers axonal degeneration (evidence regarding SEMA3A will be discussed below). Furthermore, it is known that axon regrowth and regeneration, including axon branching, is regulated by neuronal guidance signaling (e.g., Harris et al., 2020; Zou, 2021; Li et al., 2022; Smith et al., 2023). Until recently, it remained unclear to what extent pathological and developmental axon degeneration machineries are shared at molecular levels (Neukomm et al., 2014; Raff et al., 2002). However, recent research has revealed that axon degeneration, stabilization, and branching during both developmental and adult stages are controlled by the same evolutionarily conserved signaling network that regulates WD, consisting of a nicotinamide adenine dinucleotide (NAD⁺)-ase, SARM1 (sterile α- and Toll/interleukin-1 receptor motif-containing protein 1), and its regulators, such as with-no-lysine kinases (WNKs) (Figure 4; Osterloh et al., 2012; Gerdts et al., 2015; Izadifar et al., 2021; Ketschek et al., 2022). GWAS also linked variants in SARM1 to sporadic ALS (Fogh et al., 2014; Akçimen et al., 2023), suggesting that these machineries may be at work in neurodegenerative diseases.

Figure 4.

Evolutionarily conserved SARM-NMNAT-WNK machinery regulates axon degeneration, stabilization, and branching.

(A) Signaling pathways of Wallerian degeneration in response to axon damage, driving axon destruction. Two enzymes, nicotinamide mononucleotide adenylyl-transferase 2 (NMNAT2), which synthesizes NAD⁺, and sterile α- and Toll/Interleukin-1 receptor motif-containing protein 1 (SARM1), which hydrolyzes NAD⁺, control Wallerian degeneration. The inset shows cycles of the metabolites and enzymes involved. NMN and NAD⁺ opposingly regulate SARM1 activity by binding to its armadillo repeat (ARM) domain. (B) Axonal fates, such as branching, stabilization, and degeneration, are controlled by the SARM-NMNAT-WNK mechanisms, in combination with AXED and MYCBP2, and possibly regulated by neuroguidance cues, including SEMA3A, RGMA, and SLIT. ADPR: ADP-ribose; NAD: nicotinamide adenine dinucleotide; NAM: nicotinamide; NMN: nicotinamide mononucleotide; NMNAT: nicotinamide mononucleotide adenylyltransferase; SEMA: semaphorin; WNK: with-no-lysine kinase.

NAD⁺ and its precursor, nicotinamide mononucleotide (NMN), are two key metabolites that regulate axon degeneration and survival (Figure 4A). In response to axonal damage, SARM1 breaks down NAD⁺ into nicotinamide and ADP-ribose (a potent calcium mobilizer), leading to neurofilament degradation and axon fragmentation by activating calcium-activated proteases of the calpain family. Preceding this, NMNAT2 (nicotinamide mononucleotide adenylyltransferase 2), a biosynthetic enzyme that catalyzes the conversion of NMN to NAD⁺, is dramatically reduced in damaged axons (Coleman and Höke, 2020). Due to axonal NMNAT2 depletion, NAD⁺ production rates are also decreased, while NMN levels are increased in axons. SARM1 is activated in response to an increase in the ratio of NMN to NAD⁺ (Figley et al., 2021; Shi et al, 2022); namely, SARM1 activation relies on both an increase in NMN and a decrease in NAD⁺ (Figure 4A). Consistently, supplying NAD⁺ was shown to block SARM1-mediated axon destruction (Gerdts et al., 2015). Furthermore, Axundead, a putative member of the BTS-Kelch family of E3 ubiquitin ligase, acts downstream of SARM1 to mediate NAD⁺-dependent axon degeneration in Drosophila (Neukomm et al., 2017).

Furthermore, both WNK1/2 support NMNAT2 function and suppress SARM1 and Axundead, to maintain axonal integrity and promote axon branching and growth (Izadifar et al., 2021; Ketschek et al., 2022). Thus, WNK1/2 and SARM1 regulate not only axon degeneration but also branching for neural remodeling. E3 ubiquitin ligase MYCBP2 (also termed Highwire, Hiw in Drosophila) promotes WD by targeting NMNAT(s) for ubiquitin-mediated proteolysis (Figure 4B; Xiong et al., 2012). Together, axon morphogenesis, survival, and degeneration are controlled by the SARM-NMNAT-WNK network and regulators. Further studies are needed to delineate how canonical neuronal guidance pathways are merged into this SARM1- and WNK-dependent mechanism to determine the axonal fate.

Additionally, it has been postulated that various neuronal guidance pathways control mitochondrial mobilization and immobilization along axons, to meet huge metabolic demands at active synapses. The metabolic failure caused by mitochondrial dysfunction has been implicated in programmed axon death during neurodegeneration (Merlini et al., 2022). Consistently, mitochondrial toxins, including CCCP (carbonyl cyanide m-chlorophenyl hydrazone), can induce SARM1-dependent axon degeneration, following mitochondrial depolarization.

Although there is still limited evidence, several studies have addressed the mechanistic links between aberrant neuronal guidance signaling and axonal destruction. As discussed above, in ALS-SOD1^G93A mice, Sema3a is upregulated in Schwann cells at the distal end of motor axons near neuromuscular junctions that are vulnerable to denervation in ALS (De Winter et al., 2006). The administration of either anti-Nrp1 antibodies (Venkova et al., 2014) or small-molecule drugs to inhibit phosphorylation of CRMP2 (Figure 2A), a downstream signal transducer in SEMA3A signaling, or introduction of phospho-null CRMP1, a cognate CRMP (Numata-Uematsu et al., 2019; Asano, 2022; Kawamoto et al., 2022), improved motor function and survival in SOD1^G93A mice. Thus, SEMA3A acts as an axon degeneration signal released from Schwann cells in pathological contexts. By contrast, a phospho-mimicking mutant of CRMP1 suppressed neurite growth in Neuro2A cells (Kawamoto et al., 2022). In spinal cord samples of PwALS, phosphorylated CRMP1 accumulated and was colocalized with phosphorylated neurofilaments at sites of axon damage and swelling (spheroids) (Kawamoto et al., 2022), suggesting that SEMA3A-induced CRMP phosphorylation may trigger axon degeneration in ALS.

ALS8/vesicle-associated membrane protein-associated protein B (VAPB) is cleaved, and its N-terminal fragment is secreted, acting as a ligand for EPH receptors (Tsuda et al., 2008). In humans, the P56S mutation in VAPB causes familial ALS (Nishimura et al., 2004; Tripathi et al., 2021). In Drosophila, the corresponding mutant, dVAP^P58S, is defective in secretion, leading to accumulation of the dVAP mutant protein, triggering an unfolded protein response and inducing neuronal death in a cell-autonomous manner; reduced secretion of VAP may also perturb EPH signaling (Tsuda et al., 2008). Thus, both cell-autonomous and non-cell-autonomous effects of VAPB variants may contribute to neurodegeneration in ALS. Interestingly, VAPB acts as a tether at endoplasmic reticulum-mitochondria contact sites (ERMCS) by binding to the outer mitochondria-localized protein tyrosine phosphatase interacting protein 51, where it regulates ERMCS dynamics and calcium homeostasis (Obara et al., 2024). VAPB^P56S reduces the ERMCS dynamics and increases mitochondrial Ca²⁺ uptake (De Vos et al., 2014). ERMCS component dysfunction has been implicated in ALS, PD, and neuropathies, such as hereditary spastic paraplegia and Charcot-Marie-Tooth disease. ERMCS dynamics may affect endoplasmic reticulum and mitochondrial homeostasis, suggesting a new notion that ERMCS dysregulation may contribute to axon degeneration (Kuijpers et al., 2013; Krols et al., 2016; Paillusson et al., 2017). Several guidance cues, including NOGO, may regulate endoplasmic reticulum-mitochondria interactions (Sutendra et al., 2011). Issues of how neuronal guidance signaling regulates the ERMCS dynamics and how their failures contribute to neurodegeneration warrant future investigation.

In a dog model of X-linked progressive retinal atrophy 1 (XLPRA1), rod photoreceptor axon terminals were retracted into the outer nuclear layer before rod photoreceptor loss (Appelbaum et al., 2020a). Although, in this model, ROBO1/2 expression was normal during retinal development, ROBO1 levels were drastically reduced upon disease onset, suggesting that axon retraction may correlate with ROBO1 expression. Furthermore, a GWAS revealed ROBO1/2 as a candidate genetic modifier of XLPRA1 (Appelbaum et al., 2020b). The significance of dysregulated SLIT-ROBO signaling in retinal axonopathies remains to be clarified.

Collectively, axon degeneration is critical in the pathogenesis of many neurological diseases. It has become clear that neuroguidance cues, receptors, downstream signaling components, and effectors control axon branching and/or degeneration. Further efforts will be necessary to elucidate the entire mechanisms regulating axon morphogenesis and degeneration and to therapeutically target axonal degenerative processes, which are common among many neurodegenerative diseases.

Regulation of Neuron-Microglia-Astrocyte Triangular Interactions by Neuronal Guidance Genes

It has been postulated that aberrant protein aggregates trigger axonal and synaptic loss and neuronal death. Peri-aggregate glia are also involved in aggregation-dependent neurodegeneration (Heneka et al., 2015, 2018; Ransohoff, 2016; Hammond et al., 2019; Bettcher et al., 2021; Leng and Edison, 2021; Patani et al., 2023).

The major types of glia are astrocytes, microglia, oligodendrocytes, and Schwann cells (Eroglu and Barres, 2010). Physiologically, glia regulate the homeostasis of the nervous system, providing microenvironments suitable for neuronal function (Greenhalgh et al., 2020). Glia constantly surveil the nervous system for abnormal protein aggregates, cell debris, and pathogens to maintain neuronal health. It has also become clear that glia participate in neural circuit development and function through the modulation of synaptic formation, connectivity, elimination, and plasticity (Eroglu and Barres, 2010). Proper communications between neurons and glia maintain optimal neural circuit function. In this and the following sections, we focus on the pathophysiological roles of astrocytes and microglia, especially regarding how aberrant neuronal guidance signaling affects glial cell behavior in neuroinflammation and neurodegeneration.

Through bidirectional interactions with neurons and their synapses, astrocytes execute a variety of housekeeping functions, including metabolic and trophic support for neurons, coordination of synaptic transmission, and regulation of cerebral blood flow (Khakh and Sofroniew, 2015; Allen and Eroglu, 2017). Microglia are key components of the innate immune system in the brain. Astrocytes and microglia collaboratively regulate synaptic connectivity through neuritic pruning and synaptic remodeling (Vainchtein et al., 2018; Li T et al., 2020; Vainchtein and Molofsky, 2020). Furthermore, even under pathological conditions, astrocytes and microglia form functional units to exert beneficial or detrimental effects on neural circuits.

Upon stimulation with IL-1α, TNF, and C1q, released by disease-associated microglia (DAM), astrocytes lose their normal ability to support neuronal survival and are differentiated into inflammatory, neurotoxic astrocytes, termed A1 astrocytes (Deczkowska et al., 2018; Vainchtein and Molofsky, 2020). Strikingly, in the absence of A1 astrocytes, the death of axotomized neurons is prevented (Liddelow et al., 2017), suggesting that neuronal death pathways are activated by non-cell-autonomous mechanisms. A1 astrocytes secrete toxic cytokines, including IL-1α and TNF, to induce neuronal death (Liddelow et al., 2017). In later stages, interactions between astrocytes and microglia, and their communications with neurons, via cytokines, chemokines, and secreted or membrane-associated neuroguidance cues, generate inflammatory positive-feedback loops that further damage neurons beyond the original injury site. Thus, controlling microglia-astrocyte crosstalk may be therapeutically critical in preventing excessive neuronal loss.

Microglia comprise highly heterogeneous and dynamic cell populations. Microglial states have been classified by epigenomic, transcriptomic, and proteomic analyses (Lopes et al., 2022; Paolicelli et al., 2022). Single-cell and multiplex analytic strategies have begun to uncover microglia subtype-specific changes of gene expression in different ages, brain regions, and under various pathological conditions. According to single-cell and spatial transcriptomic studies based on the analyses of expression and splicing quantitative trait loci, gene expression/splicing profiles in microglia are altered at many risk loci associated with neurological diseases (Lopes et al., 2022). CNS insults, including abnormal protein aggregates released from neurons, trigger drastic responses in astrocytes and microglia, termed reactive gliosis, which constitutes a major part of neuroinflammation (Pekny and Pekna, 2016). As the disease progresses, their combinatorial responses generate further heterogeneity of microglia (much more than the originally proposed DAM) and possibly of astrocytes (Escartin et al., 2021; Leng and Edison, 2021; Paolicelli et al., 2022).

Uptake and processing of protein aggregates by microglia may generate disease-specific toxic protein strains through autophagy and the lysosomal degradation system and propagate such protein aggregates into unaffected brain regions (Bartels et al., 2020; d’Errico et al., 2022). Dysregulation of the multi-layered and interconnected cell-cell interactions and dysfunction of the waste-disposal system in microglia may promote the propagation of neurotoxic aggregates, exacerbating neuroinflammation, and thereby driving disease progression.

A pioneer study on neuron-glia interactions revealed that cell contact-dependent astrocytic EFNA3-neuronal (CA1 pyramidal) EPHA4 forward signaling induces dendritic spine retraction and morphogenesis at excitatory synapses (Murai et al., 2003). Interestingly, CA1 pyramidal neuronal EPHA4-astrocytic EFNA3 reverse signaling reduces levels of glial glutamate transporters (GLAST and GLT1) to suppress excessive neuronal excitation and to modulate LTP (Figure 3C) (Filosa et al., 2009). By contrast, astrocytic NEO1 signaling upregulates surface levels of glutamate transporters in astrocytes, to reduce glutamate levels at synaptic spaces and to prevent epileptic responses (Figure 3C; Sun et al., 2021). Thus, neuroguidance cue-mediated neuron-glia interactions physiologically contribute to the fine control of synaptic transmission.

A genome-wide ligand-receptor atlas revealed correlations between pyramidal neuron subtypes and microglial states in the neocortex (Stogsdill et al., 2022). Families of neuronal guidance genes, including SEMA3, PLXNA4, and NRP1/2, were identified in the neuron-microglia interactome. Thus, pyramidal neurons may organize microglial states, in part via neuronal guidance signaling.

TREM2, an innate immune receptor expressed in microglia and a genetic risk factor in AD, regulates the capacity of microglia to respond to Aβ or other protein aggregates, such as TDP-43, to form physical barriers, and to prevent the spreading of protein aggregates, at least during early disease stages. TREM2 interacts with dimers of transmembrane adaptors, including DNAX-activating protein of 12 kDa (DAP12), encoded by TYROBP (Additional Figure 2 ^{(2MB, tif)} ), and DAP10, encoded by HCST (Takegahara et al., 2006; Ulland et al., 2017; Hou et al., 2022; Colonna, 2023). Microglia respond to Aβ via TREM2-DAP12 and/or TREM2-DAP10 pathways (Wang et al., 2022).

Although TREM2 binds to the protein aggregates and CLU (Figure 5A; e.g., Wang et al., 2022), it remains to be clarified how diverse signals from multiple ligands and protein aggregates are processed and possibly integrated at microglial TREM2 in degenerating brain regions. SYK (spleen tyrosine kinase, initially identified from the porcine spleen) is a versatile signal transducer in both immune signaling and neuronal guidance pathways (Taniguchi et al., 1991; Mócsai et al., 2010; Noraz et al., 2016). Recent studies revealed that SYK regulates TREM2 signaling and that SYK is required for neurodevelopment and for full activation and phagocytosis of microglia (Figure 5A; Ennerfelt et al., 2022). Consistently, genetic deletion in Syk increases Aβ burden and worsens AD pathology in mice (Wang et al., 2022).

Figure 5.

Microglia-mediated neuroinflammatory responses and endothelial-to-mesenchymal transition (EndoMT) are regulated by neuronal guidance signaling.

(A) Microglia-mediated neuroinflammation is triggered and maintained by SEMA6D-PLXNA1-TREM2 signaling in AD. Broad arrays of signals, including Aβ, SEMA6D, ePtdSer, CLU, and APOE, may be integrated at TREM2 to coordinate microglial responses, as shown on the right. SEMA3A-PLXNA4 signaling may promote Aβ-tau cross-seeding. It is uncertain whether PLXNA4 interacts with TREM2. (B) To prevent chronic neuroinflammation, BBB integrity is regulated by ARF6 signaling. BBB breakdown is driven by EndoMT, which is regulated by opposing actions of inflammatory cytokines (such as IL-1β and VEGFA) and neuroguidance cues (such as SLIT2). BBB breakdown leads to infiltration of immune cells into the brain parenchyma, and eventually to CNS demyelination, which is a clinical feature of multiple sclerosis. APOE: Apolipoprotein E; ARF: ADP ribosylation factor; Aβ: amyloid-β; BBB: blood–brain barrier; CLU: clusterin; DAP: DNAX-activating protein; GSK: glycogen synthase kinase; IL: interleukin; mTOR: mammalian target of rapamycin; NTN: netrin; PI3K: phosphoinositide 3-kinase; PLXN: plexin; PRR: pattern recognition receptor; SEMA: semaphorin; TNF: tumor necrosis factor; VEGFA: vascular endothelial growth factor A; VEGFR: vascular endothelial growth factor receptor.

In wound-healing responses after spinal cord injury (SCI), glia and immune cells are recruited to form protective barriers that physically seal the wound, remove cell debris, and facilitate neuroinflammation. SEMA4C, 4D, 4G, and PLXNB2 are upregulated in injury-associated microglia and infiltrating macrophages. Furthermore, PLXNB2 is required for motosensory recovery and peripheral nerve regeneration (Zhou et al., 2020a; Li et al., 2022). PLXNB2 enhances microglial motility and regulates their spatial segregation from surrounding cells, such as reactive astrocytes, to delineate wound compaction and limit inflammatory spread. Depletion of PLXNB2 leads to impaired recovery following SCI. On the other hand, Sema4C knockouts showed no obvious phenotypes in SCI, suggesting that multiple SEMA4 ligands may activate PLXNB2.

In response to ischemic stroke, microglia in brain samples from patients express higher levels of netrin-1 and UNC5A compared with healthy controls. Autocrine- and paracrine-acting netrin-1 suppresses inflammatory responses of microglia and microglial apoptosis in hypoxic conditions in vitro and reduces neuronal loss in a mouse stroke model (Yang et al., 2023). Following ischemic stroke, the migration of neural stem cell-derived neural progenitors from the ventricular and subventricular zones toward the lesion site is hampered by reactive astrocytes. In a recent study using a mouse poststroke model, transplantation of SLIT2-overexpressing neural progenitors disrupted actin cytoskeleton in reactive astrocytes at their contact sites, increased the efficacy of neuronal migration to injury sites, and promoted functional recovery (Kaneko et al., 2018). Thus, functionally distinct neuroguidance cues may regulate functional recovery of the poststroke brain.

During postnatal development, microglia accumulate along projection neuron axons in the subcerebral region and support neuronal survival. Netrin-G1 is a netrin family member with a glycosylphosphatidylinositol anchor and is expressed in layer V neurons, while its receptor NGL1 is expressed in microglia. Netrin-G1 and NGL1 are required for microglial accumulation along axons and neuronal survival (Fujita et al., 2020). SNPs and abnormal expression of netrin-G1 and -G2 have been implicated in several types of neuropsychiatric diseases (Yamagishi et al., 2021).

Microglial activities are essential for proper axon innervation of DA neurons and laminal positioning of interneurons (Squarzoni et al., 2014). They participate in the refinement of neural connectivity, where they sense neural activity and metabolic states of neurons, and execute synaptic pruning (Hammond et al., 2018; Badimon et al., 2020; Bartels et al., 2020; Scott-Hewitt et al., 2020).

Since the discovery by Hubel and Wiesel (1959), neuroscientists have endeavored to uncover molecular mechanisms by which neural activity directs synaptic remodeling to form mature neural architecture. According to Hebb’s rules, (1) cells that fire together, wire together, whereas “out of sync” cells lose their connections and (2) “stronger” synapses are retained, whereas “weaker” synapses are eliminated (Hebb, 1949; Brown et al., 1990). More recently, it has become clear that microglia and astrocytes play roles in synaptic selection (Stevens et al., 2007; Paolicelli et al., 2011; Schafer et al., 2012; Faust et al., 2021). The complement-tagging cascade initiated by neighboring astrocytes mediates developmental synapse elimination. Consistent with the hypothesis that this mechanism may be “reactivated” in neurodegenerative diseases, complement directs microglia-mediated synaptic loss in the early stages of AD (Hong et al., 2016; Lui et al., 2016). However, detailed mechanisms by which “specific” synapses are eliminated by microglia remained unknown.

As discussed above, phosphatidylserine (PtdSer) is a lipid ligand for cell-surface receptors expressed in microglia, including TREM2; it acts as a new category of neuronal cue and triggers microglia-mediated synapse pruning (Filipello et al., 2018; Scott-Hewitt et al., 2020). Under physiological conditions, PtdSer is localized onto the inner leaflet of the plasma membrane to prevent microglial recognition. However, upon cellular damage, PtdSer becomes externalized onto the cell surface. Phagocytes then sense the externalized PtdSer (ePtdSer), which acts as an “eat-me” signal and leads to microglial engulfment. Localized PtdSer exposure can occur at synapses destined for microglial engulfment, and it tags mostly presynaptic terminals under physiological conditions (Figure 1B) (except for postsynaptic spines in cases of inhibitory neurons) (Li et al., 2020; Scott-Hewitt et al., 2020; Park et al., 2021). In C1q-knockout mice, in which neural refinement is impaired, ePtdSer levels are elevated, but microglial engulfment is reduced (Scott-Hewitt et al., 2020), suggesting that microglial engulfment of ePtdSer⁺ synapses requires the complement system. Many receptors for ePtdSer, including TREM2 and GPR56 (adhesion G protein-coupled receptor ADGRG1), are expressed in microglia and regulate their engulfment of ePtdSer⁺ synaptic structures. It is currently unclear how various pathways of ePtdSer signaling crosstalk with each other or work in parallel to coordinate microglial phagocytosis for synaptic remodeling.

Aβ oligomers and aggregates stimulate PtdSer externalization at synapses, which results from caspase 3-mediated cleavage of phospholipid flippase in apoptotic and non-apoptotic manners (Rueda-Carrasco et al., 2023). At Aβ-exposed, abnormally hyperactive synapses, TREM2 recognizes ePtdSer, causing microglia-dependent removal of ePtdSer⁺ presynaptic and postsynaptic structures (Figure 1C; Segawa et al., 2014; Wang et al., 2015b; Rueda-Carrasco et al., 2023; Tzioras et al., 2023). In TREM2^R47H and human APP^NL-F knock-in AD mice, microglial engulfment activity against ePtdSer⁺ synapses is markedly impaired, suggesting that TREM2-mediated microglial engulfment of aberrant synapses may be a potentially beneficial mechanism, at least in early AD (Rueda-Carrasco et al., 2023). Dysregulation of these microglial engulfment systems may lead to drastic synaptic losses.

It remains unclear what types of neural signals, downstream of neural activity in postsynaptic neurons, selectively label elimination-destined synapses with ePtdSer under physiological and pathological conditions (Figure 1B). Retrograde SEMA-PLXN signaling may mediate such transsynaptic communications for the elimination of weaker synapses that undergo long-term depression (Uesaka et al., 2014; Faust et al., 2021). How are synapses that undergo long-term depression labeled with ePtdSer for elimination? Notably, activity-dependent expression of MHCI (major histocompatibility complex class I) proteins in postsynaptic spines has been linked to synaptic plasticity, long-term depression, and LTP (Huh et al. 2000; Shatz, 2009). Thus, MHCI may direct synaptic elimination by microglia and immune cells, in combination with ePtdSer. Interestingly, MHCI is capable of binding Aβ (Spires-Jones and Hyman, 2014), suggesting that this system may also be aberrantly utilized in neurodegeneration.

Future research should address the long-standing question of how neural activity directs microglia-mediated synaptic sculpting. Under pathological conditions, this mechanism may become dysregulated, and synaptic pruning may be exacerbated, e.g., dysregulation of ePtdSer tagging and/or microglial phagocytosis of synapses.

Neuronal Guidance Genes in Neuroinflammation

As discussed, amyloid and amyloid-like aggregates of disease-associated proteins, such as Aβ, tau, α-synuclein, and TDP-43, in addition to neuronal loss, are major hallmarks of neurodegenerative diseases and are considered to drive disease pathogenesis. However, paradoxically, neurodegeneration can proceed in the absence of detectable protein aggregates. For example, the presence and extent of Aβ plaques are often not correlated with brain atrophy or severity of clinical symptoms in AD (Herrup, 2021; Leng and Edison, 2021). On the other hand, neuroinflammation, including microgliosis and astrogliosis, can be found in almost all cases with neurodegenerative diseases, although initially suspected to be merely secondary responses to abnormal protein aggregation (Heneka et al., 2013, 2015, 2018; Hammond et al., 2019; Leng and Edison, 2021; Wilson et al., 2023). As introduced by Alzheimer (1907), DAM are localized in proximity to Aβ plaques and NFTs in AD brains (Figure 1D; Keren-Shaul et al., 2017; Hammond et al., 2019). Analyses of the pathological temporal sequences of AD led to the “amyloid cascade-inflammation” hypothesis, in which, in response to Aβ plaque deposition, microglia-mediated inflammation may promote aggregation of hyperphosphorylated tau, driving disease progression (McGeer and McGeer, 2013).

Genetic and single-cell transcriptomic studies have revealed that many risk genes for neurodegenerative diseases encode components in the protein/lipid clearance or inflammation systems that function in immune cells, including microglia (Bartels et al., 2020). Although microglia can exert neuroprotective effects (Spiller et al., 2018), their pathogenic roles of microglia in driving neurodegenerative diseases have also begun to be elucidated (Wilson et al., 2023). Aggregation-prone oligomers and aggregates of amyloids and amyloid-like proteins are phagocytosed by microglia and often modified into more toxic species. Activated microglia release neurotoxic ligands and promote neuroinflammation, contributing to neurodegeneration (Hammond et al., 2019; Leng and Edison, 2021). It is undeniably important to study how pathogenic microglial activation can be therapeutically controlled, because activated microglia drastically affect the extent of neuroinflammation and the propagation of abnormal protein aggregates (Asai et al., 2015).

Functionally distinct populations of microglia participate in both anti- and pro-inflammatory responses, together with recruited monocyte-derived macrophages, dendritic cells, and B/T cells. These non-CNS-resident immune cells can infiltrate into the brain parenchyma, especially when BBB integrity is broken down. At sites of abnormal protein aggregation and/or neuroinflammation, immune cells secrete small molecules, including nitric oxide and reactive oxygen species, cytokines, chemokines, growth factors, and interferons that cause synaptic dysfunction and loss, and neuronal death. Furthermore, immune cell behavior can be modulated by neuronal signaling in that immune cells express receptors for neuropeptides, neurotransmitters, and neuroguidance cues (Rustenhoven and Kipnis, 2022). Thus, bidirectional immune and neuroimmune interactions regulate neuroinflammation.

Microglia sense Aβ species and various protein aggregates via pattern recognition receptors, including Toll-like receptors (TLR1, 2, 4, and 6), TREM2, and other receptors and associated co-receptors (Liu et al., 2012; Zhao et al., 2018b). TREM2 modulates TLR signaling; for example, TREM2 is capable of suppressing TLRs by sequestering the downstream components from TLRs (Ito and Hamerman, 2012; Peng et al., 2013; Colonna, 2023). Upon TLR-mediated priming in response to Aβ, microglia are activated to assemble the NLRP3 (NOD-, LRR- and pyrin domain-containing protein 3) inflammasome, a major intracellular signaling center of inflammation, in response to diverse microbial and danger signals in their cytoplasm. As a result, microglia secrete a variety of pro-inflammatory cytokines and chemokines, such as IL-1β, IL-8/CXCL8, and IL-18 (Swanson et al., 2019; Leng and Edison, 2021). Upon activation, NLRP3 recruits a prion-like adaptor protein, ASC (apoptosis-associated speck-like protein containing a C-terminal caspase recruitment domain; a.k.a. PYCARD), and caspase-1 to induce ASC polymerization (Figure 5A; Cai et al., 2014; Broz and Dixit, 2016). The assembled ASC specks are released from microglia and taken up by neighboring cells, driving inflammatory propagation. Extracellular ASC specks bind to Aβ, serving as the core for Aβ seeding/cross-seeding, thereby enhancing Aβ pathology (Heneka et al., 2013, 2018; Venegas et al., 2017; Swanson et al., 2019). Aβ-ASC composite fibrils boost Aβ toxicity in microglia, inducing their death (Friker et al., 2020). Similar schemes may be used for microglial responses to TDP-43 (Zhao et al., 2015; Xie et al., 2022). Furthermore, Aβ-activated NLRP3 inflammasomes and ASC specks upregulate tau kinases in microglia and drive aggregation of hyperphosphorylated tau in neurons (Ising et al., 2019), by as yet unknown mechanisms (Figure 5A). This effect is augmented by tau monomers/oligomers released from neurons. Additionally, complement receptors and Fc receptors expressed in microglia activate both innate and adaptive immune systems and are linked to microglial engulfment-mediated synaptic loss in AD (Hong et al, 2016; Leng and Edison, 2021). Collectively, activation of NLRP3 and ASC transduces Aβ-triggered tau pathology. This mechanism can be modulated by SEMA-PLXN signaling (Figure 5A).

It is currently unclear how tau aggregation proceeds within the neuronal cytoplasm in microglial NLRP3-dependent and presumably Aβ-bound ASC speck-dependent manners. Microglial phagocytosis, exosome secretion and uptake may mediate tau spreading into neurons. Additionally, it is tempting to ask whether and how SEMA3A-PLXNA4 signaling, together with NLRP3 inflammasomes, regulates NFT formation in response to Aβ species (Figure 5A). Furthermore, the whereabouts of NLRP3 inflammasome activation (for example, in microglia or neurons/probably both) is an important issue for identifying novel therapeutic targets. Relatedly, NLRP3 is a substrate for the E3 ubiquitin ligase parkin (PRKN, a.k.a. PARK2). Upon parkin inactivation or depletion, the assembly of neuronal NLRP3 inflammasomes drives neuroinflammation and death of DA neurons in PD (Panicker et al., 2022). PRKN activation may be a key step to control neuroinflammation and neurodegeneration. Thus, microglial phagocytosis and aggregate-(cross-)seeding, coupled with neuroinflammation, can drive neurodegeneration.

Accumulating data indicate that neuronal guidance pathways modulate microglial activation and subsequent inflammatory responses of various types of immune cells, including T cells that infiltrate into the brain parenchyma (Mirakaj and Rosenberger, 2017; Lee et al., 2019). Microglia-mediated and chemokine-dependent T cell infiltration has been shown to drive tauopathy (Chen et al., 2023). The first report of neuronal guidance genes regulating the function of immune cells involved SLIT inhibition of leukocyte chemotaxis in response to chemokines (Wu et al., 2001; reviewed in SenGupta et al., 2021). Remarkably, SLIT-ROBO signaling protects against excessive neuroinflammation by reducing immune cell infiltration in animal models of brain injuries (Altay et al., 2007; Sherchan et al., 2016). Recombinant SLIT2 attenuates neuroinflammation by inhibiting immune cell infiltration via the ROBO1-SRGAP1 pathway in rat models of brain injuries or geminal matrix hemorrhage (the latter disease causes early neonatal death) (Wong et al., 2001; Sherchan et al., 2016; Li et al., 2023). These studies suggest that targeting chemokine signaling with neuroguidance cues for therapeutical control of neuroinflammation warrants future investigation.

Members of the semaphorin family, known as immune semaphorins, and their receptors have been implicated in regulating immune cell communications and function (Suzuki et al., 2007; reviewed in Kumanogo and Kikutani, 2013; Nakanishi et al., 2021). In dendritic cells, PLXNA1 interacts with TREM2 and DAP12 (TYROBP) adaptor, forming a receptor complex for SEMA6D, a transmembrane semaphorin expressed in T cells (Takegahara et al., 2006). SEMA6D-PLXNA1-TREM2 costimulatory signaling confers dendritic cells to activate antigen-specific T cells during the early phases of immune responses. This signaling pathway is also used in microglia (Additional Figure 2 ^{(2MB, tif)} ; Albanus et al., 2023).

As discussed above, TREM2 is essential for microglial activation and the full differentiation into DAM, which also undergo metabolic reprogramming (Ulland et al., 2017; Colonna, 2023). In AD and possibly ALS, DAM activation from homeostatic states proceeds in two steps: a TREM2-independent first step, in which microglial checkpoint pathways are downregulated, in part via suppressing CX3CL1-CX3CR1 signaling, and a TREM2-dependent second step (Keren-Shaul et al., 2017). In the TREM2-dependent process, diverse microglial responses, such as proliferation, clustering around plaques, and phagocytosis, are triggered.

However, a missing ligand responsible for TREM2 activation was recently found. A snRNA-seq meta-analysis, which was combined with network biology of cell-cell interactions, revealed that multiple ligands, receptors, and signal transducers, including AD risk factors, regulate neuron-microglia interactions. Among them, microglial TREM2 and PLXNA1 form a SEMA6D receptor complex that upregulates phagocytosis and cytokine secretion in response to SEMA6D, which is presented on excitatory neurons around Aβ plaques (Figure 5A; Albanus et al., 2023). APOE and CLU may act as co-ligands for TREM2, which work with SEMA6D (Yeh et al., 2016; Albanus et al., 2023; Colonna, 2023). However, it should be noted that APOE expression is upregulated in DAM in AD mice (Keren-Shaul et al., 2017), suggesting an autocrine-signaling mode. The gene network mediating SEMA6D-TREM2 interactions is negatively associated with the Braak stage and is active during the early stages of AD. At the later stages, expressions of TREM2 and SEMA6D are downregulated, leading to the destruction of this signaling in microglia. Again, SEMA6D-PLXNA1-TREM2 signaling may exert beneficial effects in early AD, while it is unclear how SEMA6D expression is dysregulated in Aβ-exposed neurons in late AD. Further investigation will be necessary to fully understand the significance of SEMA6D-PLXNA1-TREM2 signaling in microglial activation. Additionally, further research is needed of whether and how SEMA6D-PLXNA1-TREM2 signaling, which potentially suppresses AD progression, may communicate with SEMA3A-PLXNA4 signaling that links Aβ and tau pathologies (Kang et al., 2016; Chung et al., 2021). Furthermore, TREM2 may interact with several plexins, such as PLXNA4, B1, or B2, in addition to PLXNA1, and form receptor complexes that simultaneously sense ePtdSer, semaphorins, and abnormal protein aggregates to integrate glial responses.

Intriguingly, SEMA6D can also bind to PLXNA4. However, the situation seems to be complicated, because PLXNA4-SEMA6D inter-macrophage reverse signaling mediates anti-inflammatory macrophage polarization, eventually suppressing inflammation (Table 1; Kang et al., 2018). Furthermore, a recent study showed that SEMA6D-PLXNA4 forward signaling prevents anti-tumor CD8⁺ T cell activation and infiltration into tumor microenvironments (Hirai et al., 2024). Immune checkpoint inhibitors were effective in SEMA6D-KO mice; additionally, SEMA6D expression is negatively correlated with that of PD-L1 (programmed cell death 1-ligand 1). Thus, it is tempting to speculate that suppressing SEMA6D-PLXNA4 signaling may enhance the efficacy of immune checkpoint inhibitors and be applicable to treating neurodegeneration. Which cells, besides damaged excitatory neurons, present SEMA6D or related cues, and which cells express PLXNA1 or PLXNA4, besides microglia, may be key issues in addressing how to control neurodegeneration via manipulating SEMA6-PLXNA1/A4 signaling in neurological diseases. It is also unclear whether PLXNA4, in combination with TREM2, exerts detrimental and/or beneficial effects. Furthermore, it will be of interest to examine whether SEMA6D-PLXNA1-TREM2 signaling regulates neuronal and glial responses to TDP-43 and other protein aggregates.

In similar contexts, SLIT2-ROBO signaling plays roles in tumor-associated microglia that control glioblastoma microenvironments (Geraldo et al., 2021). However, it is currently unclear how SLIT2-mediated mechanisms regulate neuron-microglia interactions.

It is pertinent to discuss now the relationship between neuronal guidance and MS. Differential expression of Sema3a and Sema7a has been reported in encephalomyelitis (EAE) mice, an MS animal model (Gutiérrez-Franco et al., 2016). SEMA4A was also characterized as a cytotoxic cue to oligodendrocytes and its expression is elevated in activated microglia in patients with MS (Leitner et al, 2015). Furthermore, intracranial injection of recombinant SEMA4A led to microglial activation in mice (Chiou et al, 2019). Another cognate immune semaphorin, SEMA4D (also known as CD100), which is presented on or secreted as soluble CD100 by activated T cells, can induce the collapse of oligodendrocyte processes and ultimately lead to neuronal cell death (Giraudon et al., 2004). Although it was initially reported that SEMA4D suppressed microglial activation (Toguchi et al., 2009), in a following study, infiltrating mononuclear cell-derived Sema4d promoted microglial activation and neuroinflammation through Plxnb1 in EAE mice (Okuno et al., 2010). A third study in EAE mice showed that microglial Sema4d-astrocytic Plxnb1/b2 signaling promoted neuroinflammation (Clark et al., 2021).

Furthermore, bidirectional SEMA4D-PLXNB1 signaling seems to play key roles in astrocyte-microglia interactions to regulate neuroinflammation upon the exposure of protein aggregates (Figures 1D and 2C). In AD mice and most clinical cases of AD, PLXNB1 expression is upregulated in peri-plaque reactive astrocytes and is positively correlated with disease severity and Aβ burden. Although the responsible ligand on microglia, which signals to PLXNB1, has not yet been characterized, SEMA4D is a strong candidate, as in EAE (Figure 1D; Huang et al., 2024). Astrocytic PLXNB1 signaling may cause contact inhibition of cell locomotion and enlargement of amyloid plaques. In AD mice, Plxnb1 deletion increased microglial coverage, which forms a physical barrier that reduces the exposure of neurons to Aβ. This promotes the compaction of amyloid plaques, which results from enhanced microglial phagocytosis and brings microglia closer to amyloids. As a result, Aβ burden and neuroinflammation are reduced, leading to improved memory performance (Huang et al., 2024). Thus, physiologically, PLXNB1 signaling decreases glial interactions with amyloid plaques and increases inter-astrocytic and inter-microglial distance, promoting neuroinflammation and neurodegeneration (Figure 1D).

In this setting at the astrocyte-microglia interface, microglial activation is presumably suppressed via astrocytic PLXNB1-microglial SEMA4D reverse signaling (Huang et al., 2024), supporting the initial finding by Toguchi et al. (2009). Thus, relaxing the peri-plaque glial net, through inhibiting SEMA4D-PLXNB1 signaling, may offer a new therapeutic strategy for controlling microglial and astrocytic activation and neuroinflammation. Manipulating astrocytic and microglial dynamics in peri-aggregate spaces may be critical in controlling the phagocytic activity of microglia and for reducing amyloid burden and neuroinflammation. Thus, SEMA4D-PLXNB1 signaling may be capable of coordinating glial cell spacing and the microenvironments around Aβ plaques. Further research will be necessary to better understand the molecular mechanisms controlling microglial function and neuroinflammation.

Collectively, therapeutic targeting of plexins and TREM2 may provide effective candidate strategies for controlling neuron-glia and neuroimmune interfaces, for the treatment of various neurodegenerative and neuroinflammatory diseases. Promisingly, Pepinemab, a humanized monoclonal antibody against SEMA4D, has entered a phase 1/2 study of early AD (Evans et al., 2020).

Defects of Neurovascular Interactions in Neurological Diseases

Brain function strictly depends on the moment-to-moment control of the vascular system. The nervous and vascular systems constantly communicate with each other and collaborate to promptly execute information processing and computation in response to external and internal stimuli. Vascular contributions to neurodegenerative diseases are now being dissected (Zlokovic, 2011; Sweeney et al., 2018; Schaeffer and Iadecola, 2021).

Neurons, glia, their processes, and vascular cells (tight junction-forming endothelial cells, pericytes, and vascular smooth muscle cells) are in close contact throughout neurovascular networks, forming the NVU (Kaplan et al., 2020; Schaeffer and Iadecola, 2021). Endothelial cells and other NVU components generate the BBB, also known as blood–central nervous system barrier. The blood-spinal cord barrier is a functional equivalent of the blood–central nervous system barrier that maintains spinal cord integrity, but these barriers are physiologically independent. Disruption of the blood–spinal cord barrier has been implicated in ALS pathogenesis (Bartanusz et al., 2011). At the NVU level, oxygen, energy substrates, and nutrients are supplied to neural cells, while carbon dioxide, toxic metabolites, aggregates, and other waste products are removed. NVUs coordinate the permeability of the blood–central nervous system barrier/blood–spinal cord barrier (hereafter designated BBB) and the rate of cerebral blood flow. Meanwhile, the BBB prevents the entry of various toxic biomaterials and pathogens into the brain parenchyma and maintains intracerebral environments.

The establishment of the spatial organization of neurovascular networks during development is an important issue in neurology. In the body and brain, nerves align with vessels, forming branched, looped, and interdependent networks. Developmental angiogenesis and neural wiring are regulated by shared neuroguidance cues, growth factors, and chemokines, including vascular endothelial growth factor A (VEGFA) and CXCL12/SDF-1, and their receptors, VEGFRs and CXCR4, respectively (reviewed by Adams and Eichmann, 2010; Wälchli et al., 2023). Although neuronal and vascular morphogenesis occurs in monocellular and multicellular manners, respectively, neurite branching and vascular sprouting are choreographed by similar sets of attractive and repulsive signals, including netrins, SLITs, EFNs, and SEMAs (Adams and Eichmann, 2010; Yuasa-Kawada et al., 2023). Representative neuronal guidance pathways in angiogenesis involve EFN-EPH signaling (Wang et al., 1998), which determines atrial-veinous vessel fates, and SLIT-ROBO signaling (Jones et al., 2009; Rama et al., 2015). The role of SLIT as pro-angiogenic or anti-angiogenic cues may be context-dependent. Additionally, the role of netrins in angiogenesis remains controversial, as they have been reported as both inhibitors and promoters of angiogenesis (Lu et al., 2004; Park et al., 2004; Wilson et al., 2006). Interestingly, in EAE mice, netrin-1 expression is induced in endothelial cells in response to astrocyte-derived sonic hedgehog. Autocrine and microglia-derived netrin-1 signaling enhances BBB stability by upregulating endothelial junction protein expression, via WNT/CTNNB1 (β-catenin) signaling (Figure 5B; Podjaski et al., 2015; Boyé et al., 2022).

Since the formulation of the NVU concept, evidence suggests that NVU dysfunction and BBB breakdown lead to the penetration and accumulation of toxic molecules and protein aggregates, and even to the infiltration of immune cells into the brain parenchyma. Furthermore, the infiltration of dendritic cells, macrophages, and T cells into the brain results in an enhancement of neuroinflammation, cerebral blood flow reduction, and hypoxia (Iadecola, 2017). Thus, vascular-derived insults and NVU dysfunction contribute to neuroinflammation, demyelination, and eventually neurodegeneration.

MS is characterized by BBB breakdown. Molecular mechanisms underlying BBB disruption remain unsolved; however, several mechanisms involve a small GTPase, ADP ribosylation factor 6 (ARF6). ARF6 is a key hub of the intracellular signaling network and of SLIT-ROBO signaling in the nervous and vascular systems (Figure 5B; Jones et al., 2009; Kinoshita-Kawada et al., 2019). In the vasculature, IL-1β and VEGFA signaling pathways reduce endothelial stability by activating ARF6, whereas SLIT2-ROBO4 and ANXA2-ROBO4 signaling pathways promote endothelial stability by suppressing ARF6 activity (Jones et al., 2009; Zhu et al., 2012, 2017; Li et al., 2019). The ARF6-dependent endothelial-to-mesenchymal transition drives BBB breakdown (Figure 5B). Consistently, pharmacological inhibition of ARF6 stabilizes the BBB and suppresses CNS demyelination (Sun et al., 2022). Thus, ARF6 has emerged as a new therapeutic target for treating MS and other neuroinflammatory and neurodegenerative diseases.

Several neuronal guidance pathways are candidates for therapeutic targets to control MS progression. When EAE mice were treated with recombinant netrin-1, the severity of BBB disruption was reduced (Figure 5B; Podjaski et al., 2015). However, distinct cues and receptors seem to promote neuroinflammation in MS. In peak stages of EAE and MS, microglia strongly express SEMA4D and EFNB3, and astrocytes express PLXNB1/B2 and EPHB3. Both forward and reverse signaling of the two pathways boost proinflammatory responses of both astrocytes and microglia in EAE and MS (Clark et al., 2021). Pharmacological inhibition of EPHB3 suppressed proinflammatory responses of microglia and astrocytes in EAE, suggesting a new therapeutic strategy against MS that prevents from exacerbating neuroinflammation.

Neuronal guidance signaling and aging affect neurocardiovascular dynamics. Sema3a is expressed in endothelial cells of aged mice (Wagner et al., 2023). Its endothelial expression is physiologically repressed by microRNA-145. However, microRNA-145 is downregulated with aging, upregulating Sema3a, which leads to a reduction in sympathetic, parasympathetic, and sensory axon fibers in the heart. Thus, aging can induce cardiac denervation. Senolytic drug treatment reduced Sema3a expression in endothelial cells, recovering nerve fiber density in the heart and improving cardiac function. This study provided evidence for the impact of aging, in combination with neuronal guidance signaling, on neurocardiovascular interactions.

Impaired lymphatic drainage and dysregulated immune cell trafficking have also been candidates for therapeutic targets in neurodegenerative diseases (Rustenhoven and Kipnis, 2022; Kipnis, 2024). Interestingly, netrin-4 stimulates the formation of the lymphatic system (Larrieu-Lahargue et al., 2010). Additionally, EPHB4 signaling is required for lymphatic valve development (Zhang et al., 2015). Transcriptomic profiling of meningeal lymphatic endothelial cells, microglia, and brain endothelial cells (BECs) revealed that meningeal lymphatic function is linked to those of microglia and the neurovascular system. As expected, impairment of meningeal lymphatic drainage decreased the efficacy of anti-Aβ immunotherapy to clear Aβ aggregates (Da Mesquita et al., 2021a).

The glymphatic system, which transports CSF, is connected to the meningeal lymphatic system; these systems are physically separated from the vascular system by the blood–CSF barrier. The blood–CSF barrier is formed by epithelial cells of the choroid plexus, while the BBB is formed by endothelial cells of CNS microvessels (Kratzer et al., 2020). Astrocytes also constitute the glymphatic system, which acts together with the lymphatic system to remove neurotoxic materials, such as amyloids (He et al., 2020; Nedergaard and Goldman, 2020; Da Mesquita et al., 2021b; Jiang-Xie et al., 2024). It remains to be understood how the formation and function of the lymphatic and glymphatic systems are coordinated by neuronal guidance signaling.

snRNA-seq studies revealed that the molecular signature of BECs is distinct from that of endothelial cells in other organs and that BECs from various brain regions have differential expression profiles (Garcia et al., 2022; Yang et al., 2022). Heterozygous mutations in the progranulin (GRN) gene, caused haploinsufficiency and thus autosomal dominant FTLD (designated FTLD-GRN), which is associated with TDP-43 pathology and neuroinflammation. snRNA-seq analyses of brain samples from patients with FTLD-GRN revealed disease-associated subtypes of astrocytes and endothelial cells (Gerrits et al., 2022). In FTLD-GRN brains, vasculature-supportive function and capillary coverage by pericytes were reduced, leading to vascular hypertrophy. Notably, SEMA3, chemokine, and NGF pathways were altered in patients with FTLD-GRN. Remarkably, almost all cell types in the NVU, including astrocytes, endothelial cells, pericytes, fibroblasts, and smooth muscle cells, as well as the NVU interactome, were affected in FTLD-GRN samples (Gerrits et al., 2022). Alterations in cell–cell interactions at the NVU may contribute to BBB breakdown in FTLD-GRN.

Furthermore, how regional and molecular heterogeneity affects functional diversity and stress resistance in healthy and diseased brain vasculature remains to be understood. The hippocampus is known to be susceptible to vascular inflammation. A molecular atlas generated from healthy individuals and PwAD suggests that the vulnerability of BECs in the hippocampus, compared with the cortex, may be attributed to inflammatory interferon-γ signaling, which is obvious even in healthy individuals (Yang et al., 2022). A further note was that the loss of vascular cells (BECs and pericytes) was significantly widespread in PwAD. Together, selective anti-inflammatory therapeutic strategies will be necessary to treat neurodegenerative diseases. Components in neuronal guidance signaling are among the strong candidates for such molecular targets.

Discovering Biomarkers for Neurological Diseases and Developing Therapeutic Approaches by Targeting Neuronal Guidance Signaling

Over the past two decades, significant progress has been made in discovering biomarkers, including prions, amyloids, and amyloid-like fibrils from serum and CSF for neurodegenerative diseases (for example, see Atarashi et al., 2011; Okuzumi et al., 2023, for detecting protein aggregation using real-time quaking-induced conversion methods). Additionally, soluble forms of TREM2 (sTREM2), which are released by activated microglia, through shedding by proteases such as ADAM10 (a disintegrin and metalloproteinase domain-containing protein 10) and γ-secretase, can be detected in CSF (Additional Figure 2 ^{(2MB, tif)} ; Piccio et al., 2008). sTREM2 levels were correlated with tau levels in CSF, suggesting its potential as a biomarker of neuroinflammation and microglial activation in several neurodegenerative diseases. Genetic and proteomic studies have validated canonical Aβ, tau, and α-synuclein pathways for AD and PD, and unveiled novel components in extensive gene and protein networks for neurological diseases, which regulate RNA metabolism, nuclear pore function, membrane transport, lipid metabolism, synaptic transmission, and mitochondrial damage control (Nussbacher et al., 2019; Bai et al., 2021; Fare and Rothstein, 2024). Multiple components of neuronal guidance signaling have also been identified. For example, netrin-1 and SLIT2 were detected among proteins that are significantly increased in AD brain samples, as compared with brain samples from control subjects or those with mild cognitive impairment (Bai et al., 2020, 2021). Together with neuroimaging, further expansion of proteomic coverage of biofluids will accelerate biomarker discovery and treatment development for diverse neurological diseases (Dhindsa et al., 2023).

Remarkable progress has also been made in therapeutic approaches to neurodegenerative diseases, although effective treatments that block neuronal loss or disease progression are still unavailable. Several anti-Aβ therapies have been raised, including aducanumab and lecanemab, which are humanized monoclonal antibodies that recognize insoluble and soluble forms of Aβ aggregates, respectively, and suppress Aβ accumulation. Although lecanemab has been approved by the US FDA and the EMA CHMP, the clinical efficacy of such anti-Aβ antibodies for improving patient cognition is modest with significant side effects, including brain edema (van Dyck et al., 2023).

One major issue in neural repair is that the adult central nervous system has only a limited capacity to regenerate upon injury or neurodegeneration, due to the inhibitory environment (Harel and Strittmatter, 2006; Mueller et al., 2009; Varadarajan et al., 2022; Delpech et al., 2024). Elezanumab, a human monoclonal antibody against RGMA (a potent inhibitor of axon growth) and a promising drug candidate, has entered phase 1/2 clinical trials for SCI and MS (Huang et al., 2021; Kalluri et al., 2023).

Additionally, chemokine signaling can modulate neuronal function. While neuronal chemokine receptor CCR5 suppresses memory linking (Shen et al., 2022), CCR5 and microglia-derived CCL3/4/5 inhibit autophagy in neurons, promoting neurodegeneration (Festa et al., 2023). CCR5 is also a co-receptor of CD4 for the HIV entry. Maraviroc, an FDA-approved CCR5 inhibitor used to treat HIV patients, reversed CCR5-mediated inhibition of autophagy and mitigated neurodegeneration in mouse models of HD and tauopathy (Festa et al., 2023).

A key to whether neuroinflammation is beneficial or detrimental depends on pathophysiological contexts (Bartels et al., 2020; Leng and Edison, 2021); thus, they need to be taken into consideration for the development of therapeutic approaches to neuroinflammation.

In AD and Lewy body dementia, immune cells, including T cells, were recruited to the aggregation sites of hyperphosphorylated tau and synuclein, respectively (Gate et al., 2020, 2021; Chen et al., 2023). In both diseases, upregulation of components of multiple chemokine pathways is associated with the severity of symptoms. Antibody-mediated depletion of microglia or T cells blocked tau-mediated neurodegeneration in mice (Chen et al., 2023). Furthermore, inhibiting interferon-γ or immune checkpoint, PD-1 (programmed cell death-1) signaling, reduced brain atrophy in AD mouse models with tauopathy and APOE4 (Chen et al., 2023). The NFT-triggered immune communications between activated microglia and cytotoxic T cells, via chemokine or SEMA-PLXN-TREM2 signaling, may serve as therapeutic targets for neurodegeneration in AD and other tauopathies and proteinopathies, although mechanistic details need to be further uncovered.

There is also a new direction to MS treatment. Currently used therapeutic treatments against MS using general immunosuppressants inevitably affect systemic immune responses and create severe side effects. A pharmacological inhibitor of ARF6, NAV-2729, was examined in EAE mice, where it prevented the infiltration of CD4⁺ or CD8⁺ T cells and CNS demyelination. As expected, host immune function was not affected, unlike the currently used natalizumab, a humanized monoclonal antibody against α4-intregrin (Sun et al., 2022).

Various therapeutic approaches have been under development for stroke and several neurovascular diseases. The manipulation of neuronal guidance signaling, such as netrin pathways, may be used for vascular repair and NVU reconstruction in the ischemic CNS (Podjaski et al., 2015; Rust et al., 2019). Meanwhile, intrathecal injection of EFNA1 fragment enhanced DA neurogenesis and angiogenesis in PD rodent models (Jing et al., 2012). EFNA1 regulates the dynamics of the subventricular niche for stem cell production and angiogenesis. Consistently, inhibition of EPHA2, an EFNA1 receptor, suppressed pathological angiogenesis (Rust et al., 2019). Function-blocking anti-EPHA2 antibodies may reduce BBB permeability and these have been explored in cancer treatments. Furthermore, EFNB2-EPHB4 signaling increased pericyte recruitment and endothelial cell–pericyte interactions, promoting post-stroke neurovascular repair (Ghori et al., 2017). Obviously, further efforts are necessary to develop effective treatments for various types of neurological diseases, along with mechanistic elucidation of neuronal guidance signaling.

Concluding Remarks and Perspectives

How the nervous system is formed and affected by various diseases has been a long-standing question in neurobiology and neurology (Luo, 2020). Accumulating data now support the crucial roles of neuronal guidance genes in the formation, function, maintenance, and repair of the nervous system. Strikingly, neuronal guidance signaling involves a versatile machinery that can mediate cell–cell communications and can also modulate various activities of immune cells, including microglia and T cells, in the nervous system. Despite remarkable progress, significant gaps remain in our knowledge regarding the pathophysiological roles of neuronal guidance signaling. We have just begun to understand the molecular principles by which multiple signaling pathways, including those mediated by neuronal guidance genes, exert both short-range and long-range effects to enable cells to adapt to a constantly changing environment. We are also confronted by an important problem: how highly sophisticated mechanisms, for instance, those used for cell–cell communications at neuron–glia and neuroimmune interfaces, go awry with aging (Wyss-Coray, 2016). Further advances in epigenetics, genetics, genomics, single-cell and spatial transcriptomics, and proteomics, together with the development of new technologies, will provide more comprehensive information about extensive interactions among different cell types in the nervous system under physiological and pathological conditions.

Patterns of neuron–glia and neuroimmune interactions often appear complicated. It will be important to further decipher the precise location(s) of points of action, namely which cells express the corresponding ligands and receptors, as well as the directionality of the signaling, in addition to characterizing their effects. Neuroinflammation, which eventually leads to BBB breakdown and infiltration of immune cells into the brain parenchyma, is a common pathomechanism that worsens various neurodegenerative diseases during later stages. It has become clear that networks of neuronal guidance pathways regulate the activities of microglia and immune cells in neuroinflammation and neurodegeneration. A deeper understanding of the complex interactions among neurons, glia, and immune and vascular cells that contribute to neuroinflammation will further accelerate the development of therapeutic approaches to neurodegenerative and neuropsychiatric diseases. Neuronal guidance research will find novel molecular targets that enable the development of truly effective drugs for treating neurodegenerative diseases.

Additional files:

Additional Figure 1 ^{(2.2MB, tif)} : Neuronal guidance cues and receptors.

Additional Figure 1

Neuronal guidance cues and receptors.

Schematic illustrations of guidance cues and receptors based on domain structures annotated by EMBL-SMART (http://smart.embl-heidelberg.de/). Reprinted with permission from Yuasa-Kawada et al. (2023). Each LRR domain of SLIT2 is composed of multiple LRR assemblies. EFNBs and EPH receptors C-terminally bear PDZ domain (the scaffold domain shared by Postsynaptic density-95)-binding motifs. Reticulon-4 (RTN4)/NOGO interacts with RTN4 receptors (RTN4Rs). CC: Conserved cytoplasmic; CT: C-terminal cysteine knot; CUB: found in C1r, C1s, uEGF, and BMP; DD: death domain; FN3: fibronectin type 3; F V/VIII: factors V and VIII (coagulation factors V and VIII); GAP: GTPase-activating protein; Ig: immunoglobulin-like; IPT: immunoglobulin-plexin-transcription; Lam: laminin-type; LBD: ligand-binding domain; LRR: leucine-rich repeat; MAM: present in meprin, A5, receptor protein tyrosine phosphatase mu; PSI: plexin-semaphorin-integrin; SAM: sterile alpha motif; TSP: thrombospondin type 1; UPA: UNC5-PIDD-ankirin; ZU5: present in ZO-1 and UNC5.

Additional Figure 2 ^{(2MB, tif)} : SEMA pathways in neurodevelopment and neurodegenerative diseases.

Additional Figure 2

SEMA pathways in neurodevelopment and neurodegenerative diseases.

Schematic illustrations of representative semaphorins and their receptors discussed in this review. On the left, PLXNA1/A4 and NRP1/2 form a receptor complex for SEMA3A. The solved structure of the SEMA3A-PLXNA4- NRP1 complex shows a symmetric 2:2:2 assembly (Lu et al., 2021). In the middle, PLXNA1, TREM2, and DAP12 form a receptor complex for SEMA6D. TREM2 is cleaved, upon microglial activation. Released sTREM2 may allow diagnostic detection of neuroinflammation. The SEMA4D and PLXNB1 pair, which upregulate neuroinflammation, is shown on the right.

Funding Statement

Funding: This work was supported by JSPS (KAKENHI: 21K06205; 23K06937; 24K23419) and AMED (to JYK, SaY, TM, SiY, YT, and NH); JYW had long been supported by the NIH.

Data availability statement:

All relevant data are within the paper and its Additional files.