Single-Cell Transcriptomics in Spinal Cord Studies: Progress and Perspectives

Abstract

Single-cell RNA sequencing (scRNA-seq) has revolutionized neuroscience by enabling the analysis of cellular heterogeneity and dynamic molecular processes at the single-cell resolution. In spinal cord research, scRNA-seq provides critical insights into cell type diversity, developmental trajectories, and pathological mechanisms. This review summarizes recent progress in the application of scRNA-seq to spinal cord development, injury, and neurodegenerative diseases and discusses the current challenges and future directions. Relevant studies focusing on the key applications of scRNA-seq, including advances in spatial transcriptomics and multi-omics integration, were retrieved from PubMed and the Web of Science. scRNA-seq has enabled the identification of distinct spinal cord cell populations and revealed the gene regulatory networks driving development. Injury models have revealed the temporal dynamics of immune and glial responses, alongside potential regenerative processes. In neurodegenerative conditions, scRNA-seq highlights cell-specific vulnerabilities and molecular changes. The integration of spatial transcriptomics and computational tools, such as machine learning, has further improved the resolution of spinal cord biology. However, challenges remain in terms of data complexity, sample acquisition, and clinical translation. Single-cell transcriptomics is a powerful approach for understanding spinal cord biology. Its integration with emerging technologies will advance both basic research and clinical applications, supporting personalized and regenerative therapy. Addressing these technical and analytical barriers is essential to fully realize the potential of scRNA-seq in spinal cord science.

1. Introduction

The spinal cord, a pivotal component of the central nervous system (CNS), transmits sensory and motor signals between the brain and peripheral tissues and coordinates reflex activities independently of the brain. The structure of the spinal cord is intricately connected to its function; the outer white matter comprises myelinated ascending and descending nerve tracts, whereas the central gray matter contains neuronal cell bodies and local neural circuits [1].

The complexity of the spinal cord arises from its diverse cellular composition, including neurons, glial cells, and other specialized cell types, as well as the precise functional specialization and intricate interactions among these cells [2]. Advances in single-cell transcriptomics have revealed the heterogeneity of neuronal and non-neuronal cell populations, offering new insights into their roles in health and disease [3]. For example, motor neurons in the spinal cord have been found to consist of multiple subtypes, each with distinct gene expression profiles and functional roles [4].

Moreover, the response of glial cells to spinal cord injury (SCI) has attracted considerable attention. Research has shown significant alterations in astrocytes, microglia, and oligodendrocytes post-injury, which critically influence neuronal survival and regeneration [5,6,7]. These findings highlight the importance of exploring the cellular composition and function of the spinal cord, contributing to our understanding of its roles in health and disease and opening new avenues for therapeutic intervention.

Single-cell RNA sequencing (scRNA-seq) has emerged as a powerful tool for elucidating the intricate cellular and molecular mechanisms underlying spinal cord function and disease. This technology allows for a detailed analysis of gene expression at the single-cell level, providing unprecedented insights into cellular heterogeneity and function [8]. Specifically, methods such as SPLiT-seq and Smart-seq3 have been shown to be particularly suitable for spinal cord tissues due to their ability to handle complex and heterogeneous samples with high sensitivity and accuracy. For instance, SPLiT-seq has been successfully applied to study the developing mouse spinal cord, revealing distinct cell types and lineages [9]. Similarly, Smart-seq3 has been highlighted for its ability to generate high-quality, full-length transcriptomes [10], which is crucial for understanding the nuanced gene expression profiles in spinal cord cells. These advancements not only enhance our understanding of spinal cord biology but also pave the way for future therapeutic interventions.

The groundbreaking development of single-cell transcriptomics dates back to 2009, when Tang et al. first successfully sequenced the transcriptome of individual mouse embryonic cells, overcoming the challenge posed by low RNA content in single cells [11]. Subsequently, advancements in microfluidics and laboratory automation have dramatically increased the capacity of single-cell transcriptomic analyses, facilitating the commercialization of platforms capable of processing thousands to tens of thousands of cells per experiment [12]. The rapid evolution of scRNA-seq has emerged as a powerful tool for unraveling cellular heterogeneity within spinal cord tissues.

Spatial transcriptomics has further advanced this field by integrating gene expression profiles with spatial tissue context. Its origins can be traced back to the application of in situ hybridization by Walter Gehring’s group in 1986, which localized antennapedia mRNA in Drosophila embryos and imaginal discs [13]. The term “spatial transcriptomics” was coined by Ståhl et al. in 2016, describing a novel method that captures mRNA molecules directly from tissue sections on glass slides, enabling the high-resolution mapping of gene expression integrated with tissue morphology [14]. While single-cell transcriptomics provides genome-wide expression profiles at single-cell resolution, it lacks contextual tissue information, a gap effectively addressed by spatial transcriptomics, which elucidates intercellular interactions and tissue heterogeneity [12,15]. However, spatial transcriptomics still faces challenges, particularly regarding resolution and the detection of low-abundance transcripts [15].

The integration of single-cell and spatial transcriptomics is driving advances in precision medicine, offering robust tools for disease research, drug development, and regenerative medicine [16]. By sequencing transcripts at the single-cell level, researchers can identify distinct cellular subpopulations and clarify their specific roles in spinal cord development, homeostasis, and injury repair processes. For instance, Delile et al. (2019) employed scRNA-seq to profile the developing mouse spinal cord, uncovering spatial and temporal gene expression dynamics essential for neural progenitor differentiation and neuronal maturation, significantly enriching our understanding of spinal cord developmental processes [17]. Blum et al. (2021) utilized single-nucleus transcriptomics to reveal substantial heterogeneity among motor neurons in adult mouse spinal cords, identifying subtype-specific transcriptional signatures closely associated with motor functions and disease mechanisms [4]. Recently, Zhang et al. (2024) [18] combined spatial transcriptomics with single-nucleus RNA sequencing to create a detailed spatial atlas of adult human spinal cord cells and identified and localized 21 neuronal subclusters. This comprehensive mapping linked molecular identities directly to anatomical structures, greatly expanding our knowledge of the cellular composition of the human spinal cord [18]. Collectively, advances in single-cell and spatial transcriptomics offer unparalleled opportunities to elucidate the complex cellular architecture of the spinal cord and open innovative pathways for precise therapeutic interventions in spinal cord injury and associated diseases.

With the widespread application of single-cell and spatial transcriptomics in spinal cord research, a growing number of studies have uncovered the cellular heterogeneity and dynamic changes that occur in the spinal cord during development, physiological function, and pathological conditions [9,19]. Despite these advances, a comprehensive review that systematically integrates these findings and explores their implications for neuroscience and clinical translation is lacking. This review aims to fill this gap by providing a thorough overview of the recent progress in the application of single-cell and spatial transcriptomic technologies in spinal cord research. By synthesizing the current findings, we aim to offer a conceptual framework that deepens the understanding of complex cellular behaviors within the spinal cord and guides future developments in related fields.

This review covers several key aspects, including technological advancements, the construction of comprehensive cellular atlases, an analysis of microenvironmental dynamics, and challenges associated with translational medicine. By tracing the evolution of these technologies and highlighting representative clinical applications, we aim to bridge basic and translational research, offering insights into both the fundamental biology of the spinal cord and its relevance to regenerative medicine and therapeutic development for neurological disorders. Ultimately, this review aspires to serve as a valuable resource for neuroscientists and clinicians alike, providing a systematic reference that promotes a deeper understanding of spinal cord complexity and informs future research directions and treatment strategies.

2. Single-Cell Transcriptomics: Technologies and Methodologies

2.1. Principles and Evolution of Single-Cell RNA Sequencing Technologies

scRNA-seq is a powerful technique that enables transcriptomic analysis at the resolution of individual cells, allowing for an exploration of cellular heterogeneity and complex gene expression patterns. The core workflow of scRNA-seq typically involves the isolation of single cells, cell lysis, the reverse transcription of mRNA into cDNA, library construction, and high-throughput sequencing. These steps allow researchers to capture a comprehensive transcriptional profile of individual cells [20].

Since its initial introduction in 2009 [11], scRNA-seq has undergone rapid technological advancements. In 2012, the development of SMART-seq significantly improved full-length cDNA amplification strategies, resulting in enhanced transcript coverage [21]. One of the most notable breakthroughs in scRNA-seq is the exponential increase in throughput. While early methods such as STRT-seq could only process hundreds of cells [22], the advent of droplet-based microfluidics in 2015—exemplified by Drop-seq and the 10x Genomics Chromium system—enabled the profiling of tens of thousands to millions of cells in a single experiment [23,24,25,26]. This high-throughput capacity has made it feasible to investigate cellular heterogeneity in complex tissues, such as the spinal cord.

Moreover, these technical advancements have not only increased scalability but also reduced the cost per cell, facilitating the broad application of scRNA-seq across various disciplines, including developmental biology, oncology, and neuroscience. In summary, the development of scRNA-seq has provided a transformative tool for dissecting cellular functions and tissue architecture, particularly in elucidating the diverse cellular landscape of complex systems, such as the spinal cord.

2.2. Comparison of Major scRNA-Seq Platforms and Their Features

The selection of scRNA-seq technologies should be guided by specific research objectives and experimental conditions, as different platforms vary significantly in terms of throughput, sensitivity, and cost.

Among the currently available scRNA-seq platforms, the 10x Genomics Chromium system is one of the most widely adopted high-throughput technology. The 10x Genomics Chromium platform has significantly advanced scRNA-seq with its high-throughput capabilities and ability to detect rare cell types [24]. Based on droplet microfluidics, it encapsulates thousands of individual cells into nanoliter droplets containing unique barcodes, enabling efficient and parallel transcriptomic profiling [25]. Its high throughput allows for the detection of rare cell populations, and its relatively low cost and scalability make it particularly suitable for complex experimental designs involving multiple conditions or time points [27]. However, it is important to acknowledge the limitations of the platform, particularly in the context of specific applications. For instance, the 10x Genomics Chromium system is known for its 3’ bias, which can limit the detection of full-length transcripts and affect the identification of transcript isoforms [28]. This bias may be particularly problematic when studying alternative splicing events or low-abundance transcripts, as these require comprehensive transcript coverage [27]. Additionally, the platform experiences higher dropout rates for low-expression genes, which can result in the loss of important biological information [27]. These limitations highlight the need for researchers to carefully consider the specific requirements of their studies when selecting an scRNA-seq platform.

In contrast, SMART-seq2 and SMART-seq3 are well-established plate-based scRNA-seq methods that enable full-length mRNA amplification and sequencing. These technologies are particularly advantageous for detecting low-abundance transcripts and transcript isoforms [29,30]. SMART-seq3 further improves the sensitivity over its predecessor and supports quantitative transcript detection at the single-molecule level [30]. However, these methods are generally limited by lower throughput, longer processing times, and higher per-cell costs, making them more suitable for studies focusing on rare cell populations and functional subtypes.

Drop-seq and inDrops, both droplet-based microfluidic approaches, offer cost-effective and scalable solutions for high-throughput scRNA-seq. Drop-seq, developed by Macosko et al., co-encapsulates single cells with barcoded beads in droplets, providing a robust method with high efficiency and relatively low cost [23]. inDrops introduces improvements in barcode design and reaction chemistry, enhancing cell capture rates and data consistency while also exhibiting remarkably low background noise [31]. Furthermore, its flexible architecture facilitates adaptation to other sequencing-based assays [31]. While both methods maintain high throughput, their sensitivity is generally lower than that of SMART-seq-based techniques [32].

SPLiT-seq is a unique combinatorial barcoding approach that eliminates the need for physical cell isolation. By sequentially tagging fixed cells through multiple rounds of barcode delivery, SPLiT-seq significantly reduces the reliance on specialized instrumentation and greatly increases throughput [9]. Using this method, Rosenberg et al. mapped the spatiotemporal dynamics of 30 neuronal subtypes in the developing mouse spinal cord, demonstrating the utility of SPLiT-seq for comprehensive transcriptomic profiling in complex multicellular systems [9].

In recent years, spatial transcriptomics technologies, such as 10x Genomics Visium and BGI’s Stereo-seq, have enabled the in situ mapping of gene expression across complex tissues, such as the spinal cord. For instance, Zhang et al. utilized single-nucleus RNA sequencing in combination with Visium spatial transcriptomics to delineate the spatial distribution of spinal cord neurons in nine human donors [18]. Stereo-seq further advances the field by improving spatial resolution and expanding the capture area, offering the potential to generate spatial gene expression maps at subcellular resolution [33]. To aid comparison,

Table 1. Summary of Single-cell RNA Sequencing Platforms Used in Spinal Cord Research.

Sequencing Method	Key Features	Advantages	Limitations	Applications in Spinal Cord Research	Temporal Evolution
SMART-seq2/SMART-seq3	Plate-based; full-length transcript coverage (SMART-seq3 improves sensitivity)	High sensitivity; isoform detection; full-length transcripts	Low throughput; expensive; time-consuming	Detecting rare cell types or novel isoforms	SMART-seq2 (~2013); SMART-seq3 (~2019)
10x Genomics Chromium	Droplet-based; barcode encapsulation of cells	High throughput; scalable; relatively low cost	Partial transcript coverage; dropout effect	Developmental mapping; SCI response; human/fetal profiling	Mainstream since ~2016
Drop-seq	Droplet-based; barcoded beads in droplets	Cost-effective; scalable; high efficiency	Lower sensitivity than SMART-seq	Identifying glial/inflammatory responses	Introduced ~2015
inDrops	Improved droplet-based; enhanced barcoding and chemistry	Low background noise; flexible; improved capture	Still limited in sensitivity; complex protocol	Flexible transcriptomic profiling with high consistency	Developed post-2015
SPLiT-seq	Combinatorial barcoding; no physical isolation required	Ultra-high throughput; low equipment requirement	Sparse gene detection; complex processing	Mapping neuron subtypes in mouse spinal cord development	Introduced ~2018
10x Visium/BGI Stereo-seq	Spatial transcriptomics; spatial gene expression in situ	High spatial resolution; large capture area	Requires tissue sectioning; limited to spatial resolution	Mapping spatial gene expression in human spinal cord	Emerging since ~2020

summarizes key single-cell RNA sequencing platforms, highlighting their technical features, advantages, limitations, and representative applications in spinal cord research.

2.3. Technology Selection Strategies and Application Cases in Spinal Cord Research

In spinal cord research, the selection of an appropriate scRNA-seq technology must be carefully tailored to specific research objectives and the nature of the biological samples. For studies aiming to identify novel transcript isoforms or rare cell populations, SMART-seq is often preferred because of its high sensitivity and ability to capture full-length transcripts. In contrast, high-throughput platforms, such as 10x Genomics Chromium or Drop-seq, are more suitable for constructing large-scale single-cell atlases of the spinal cord.

In the context of spinal cord development, Delile et al. employed the 10x Genomics Chromium platform to perform scRNA-seq on the cervical and thoracic regions of the embryonic mouse spinal cord, covering multiple developmental stages from embryonic day 9.5 to 13.5. This study generated a high-resolution transcriptional atlas, revealing the spatial and temporal dynamics of neural progenitors and neuronal populations. Moreover, this study identified novel marker genes and regulatory sequences involved in neuronal differentiation, providing a valuable resource for elucidating the cellular diversity and molecular mechanisms underlying spinal cord development [17].

To construct a comprehensive cellular map of the human spinal cord, Andersen et al. applied the 10x Genomics Chromium system to systematically profile the human fetal spinal cord. Their study uncovered substantial cellular heterogeneity and regional specialization during the development process. By integrating data across multiple developmental time points, the authors tracked temporal changes in cell types and mapped disease-associated genes to specific cellular populations, offering novel insights into the cellular architecture of motor control and the pathogenesis of neurological disorders [34]. This study represents a critical step toward building a detailed single-cell atlas of human spinal cord development.

In studies of SCI, researchers have utilized single-nucleus RNA sequencing (snRNA-seq) to profile transcriptional changes in the injured spinal cord at high resolution. For example, Matson et al. used the 10x Genomics Chromium platform to generate a cellular map of the lumbar spinal cord following injury in a mouse model. Their findings revealed injury-induced alterations in distinct neuronal subpopulations and highlighted specific neuronal subsets with regenerative potential [35]. These results provide a theoretical framework for future therapeutic strategies targeting specific neuronal groups to promote spinal cord repair.

In summary, the 10x Genomics Chromium system has emerged as the predominant platform used in most spinal cord studies because of its balance of throughput and resolution. While it is crucial to maintain a balanced perspective and recognize the platform’s limitations, it is equally important to acknowledge the significant advancements that 10x Genomics has brought to this field. The thoughtful selection and application of scRNA-seq technologies, aligned with the specific aims and sample characteristics of a study, are essential for advancing our understanding of spinal cord development, function, and regeneration.

3. Applications of Single-Cell Transcriptomics in Spinal Cord Research

Single-cell transcriptomics has rapidly evolved over the past decade, giving rise to a diverse set of technologies that have been instrumental in advancing spinal cord research. These platforms vary in resolution, throughput, and scalability, enabling applications ranging from transcript isoform discovery to the construction of comprehensive cellular atlas. In spinal cord studies, scRNA-seq technologies have been applied across multiple major domains, including the characterization of developmental processes, the elucidation of injury-induced molecular changes, the investigation of neurodegenerative disease mechanisms, and integration with spatial transcriptomics and multi-omics approaches for translational insights. Figure 1 illustrates the chronological development of key scRNA-seq platforms and their representative applications in spinal cord research.

3.1. Decoding Spinal Cord Developmental Biology

Single-cell transcriptomic technologies have emerged as pivotal tools for deciphering cellular diversity and lineage specification during spinal cord development. By generating cell type-specific gene expression maps, researchers can precisely track processes such as neural tube formation, fate specification, and circuit assembly. This section highlights key discoveries enabled by single-cell approaches, focusing on the temporal regulation of neurogenesis, mechanisms underlying cell type diversification, and the dissection of critical signaling pathways at single-cell resolution (

Table 2. Representative Single-Cell Transcriptomic Studies of Spinal Cord Development.

Study	Species	Developmental Stage	Platform	Key Findings
Delile et al. [17]	Mouse	E9.5–E13.5	10x Genomics Chromium	Spatial/temporal dynamics of neural progenitors; novel markers in dorsoventral domains
Andersen et al. [34]	Human	~22 weeks gestation	10x Genomics Chromium	Glial heterogeneity, astrocyte regionalization, disease gene mapping to specific cell types
Sathyamurthy et al. [19]	Mouse	Adult	Drop-Seq	43 neuronal subtypes, region-specific distribution, spinal neuron molecular map
Blum et al. [4]	Mouse	Adult	10x Genomics Chromium	Motor neuron heterogeneity, transcriptional profiles linked to axonal targeting and function
Cao et al. [36]	Mouse	Organogenesis (multi-stage)	SPLiT-seq	Spinal progenitor transcriptional transitions, Hox genes, Shh pathway in developmental regulation
Zhang et al. [18]	Human	Adult	10x Genomics Chromium	21 neuronal subtypes, spatial distribution, human-mouse comparison, sex-specific transcription

).

Delile et al. employed the 10x Genomics platform to perform scRNA-seq on cervical and thoracic spinal cord regions from embryonic day (E) 9.5 to 13.5 in mice, generating a high-resolution developmental atlas that revealed the spatial and temporal dynamics of neural progenitor and neuronal populations [17]. The study identified novel marker genes expressed in distinct dorsoventral domains of progenitors and neurons, providing insights into the transcriptional programs that guide cell fate decisions during spinal cord development [17].

In human spinal cord development, Andersen et al. conducted scRNA-seq analysis on mid-gestation (~22 weeks) fetal samples and uncovered pronounced heterogeneity between and within major cell types [34]. For example, glial cells show spatially dependent transcriptional programs along the dorsoventral and rostrocaudal axes, and astrocytes display differential gene expression based on their localization in the white versus gray matter. Moreover, the study integrated multiple existing datasets covering the same developmental window to trace temporal changes in cellular composition and mapped disease-related genes to specific cell types, offering novel insights into the cellular basis of motor control and opportunities for stem cell-based disease modeling [34].

A comparative analysis of these studies revealed both similarities and differences in spinal cord development between mice and humans. While Delile et al. [17] identified distinct spatial and temporal dynamics of neural progenitor and neuronal populations in the mouse spinal cord, Andersen et al. [34] demonstrated significant heterogeneity within and between the major cell types in the human spinal cord. Notably, the identification of novel marker genes in mouse models by Delile et al. [17] aligns with the discovery of distinct transcriptional programs in human glial cells by Andersen et al. [34]. However, there are notable differences. For instance, the spatial dependency of transcriptional programs in human glial cells along the dorsoventral and rostrocaudal axes [34] contrasts with the more generalized spatial dynamics observed in mouse models [17]. These differences highlight the importance of considering species-specific variations when interpreting findings from mouse models in the context of human spinal cord development. A critical reflection on these variations suggests that while mouse models provide valuable insights into the fundamental mechanisms of spinal cord development, they may not fully capture the complexity and heterogeneity observed in human tissue. This underscores the need for further comparative studies to bridge the gap between mouse and human data, particularly in understanding neuronal subtypes and glial cell heterogeneity.

Significant progress has been made in resolving motor neuron diversity at the single-cell level. Sathyamurthy et al. utilized massively parallel snRNA-seq to profile spinal cord neurons in adult mice and identified 43 transcriptionally distinct neuronal subtypes with region-specific distributions [19]. Their work established a comprehensive molecular map of spinal neurons and linked transcriptional identities to the anatomical and functional domains. These findings have led to new hypotheses regarding the functional roles of specific neuronal subtypes in motor control and disease development. For example, the identification of distinct transcriptional programs in motor neurons suggests that targeted interventions can be developed to modulate the activity of specific neuronal populations, potentially offering new therapeutic directions for motor neuron diseases [19].

Building on these findings, Blum et al. focused specifically on adult spinal motor neurons and used scRNA-seq to investigate transcriptional heterogeneity within this population [4]. Despite similar morphological and positional features, individual motor neurons exhibit distinct transcriptional programs related to their electrophysiological properties, spatial organization, and axonal targeting [4]. This study provides a detailed molecular atlas of motor neurons, laying the groundwork for future investigations into motor system function and dysfunction. The detailed molecular atlas generated by Blum et al. [4] has opened new avenues for exploring the mechanisms underlying motor neuron diversity and has led to new hypotheses regarding the role of transcriptional heterogeneity in motor neuron function. For instance, this study suggests that specific transcriptional profiles may be associated with different motor neuron subtypes, which may have implications for understanding the pathophysiology of motor neuron diseases [4].

In a broader developmental context, Cao et al. performed large-scale scRNA-seq during mouse organogenesis and constructed a multi-organ cell atlas, identifying developmental trajectories across multiple germ layers [36]. Their analysis showed that spinal cord progenitors undergo clear transcriptional transitions over time, driven by key regulatory factors such as Hox genes and the Sonic Hedgehog (Shh) signaling pathway [36]. This study offers a valuable resource for studying neural development and the emergence of cellular heterogeneity in the spinal cord. The identification of key regulatory factors and developmental trajectories in the spinal cord has led to new hypotheses regarding the mechanisms driving cellular differentiation and tissue organization. For example, the role of Hox genes and the Shh signaling pathway in spinal cord development suggests potential targets for therapeutic interventions aiming to promote neural regeneration and tissue repair [36].

Advances in computational methods have further enhanced our ability to analyze the developmental dynamics. The reversed graph embedding (RGE) algorithm proposed by Qiu et al. enables the reconstruction of pseudotemporal trajectories from scRNA-seq data without time labels [37]. Implemented in Monocle 2, this method has been widely used to model nonlinear developmental progressions in complex systems, including the differentiation of spinal cord progenitors into diverse neuronal lineages. Computational tools, such as pseudotime analysis, provide a more detailed understanding of cellular transitions and developmental trajectories in the spinal cord [38]. Pseudotime analysis allows researchers to infer the temporal sequence of cellular states from static scRNA-seq data, offering insights into the dynamic processes underlying spinal cord development and regeneration [37]. This is particularly valuable for studying the differentiation of neural progenitors and the maturation of neuronal subtypes, which are critical for understanding the functional organization of the spinal cord [17].

Recently, Zhang et al. combined snRNA-seq with spatial transcriptomics to profile spinal cord neurons from nine human donors, identifying 21 neuronal subtypes and defining their spatial distribution [18]. Comparative analysis between human and mouse spinal cords revealed broad similarities in neuronal composition and species-specific transcriptional signatures. Notably, sex-specific differences in gene expression were also observed in motor neurons, including SCN10A and HCN1 [18]. Spatial transcriptomics has revolutionized our understanding of the spatial organization of cells within the spinal cord. Unlike traditional scRNA-seq, spatial transcriptomics provides a snapshot of gene expression without spatial context and allows for the simultaneous measurement of gene expression and cellular localization [39].

The studies selected for this review were chosen to represent the latest advancements in single-cell transcriptomics applied to spinal cord research. Collectively, these studies illustrate the power of single-cell and spatial transcriptomics in resolving the cellular diversity and complex regulatory networks underpinning spinal cord development. These findings provide foundational insights into neurodevelopmental biology and inform our understanding of the molecular basis of spinal disorders.

Despite these advancements, current single-cell approaches in spinal cord studies face several limitations. For instance, spatial resolution remains a challenge, as many single-cell techniques do not provide detailed spatial information regarding cell locations within tissues [14]. This limitation is particularly significant in the spinal cord, where spatial organization is crucial for understanding cellular interactions and tissue function. Sampling bias can also affect the representativeness of the data, particularly when dealing with rare cell types or small sample sizes [40]. Additionally, the technical noise inherent in single-cell sequencing can introduce variability and limit the accuracy of gene expression measurements [41]. These limitations highlight the need for continued technological improvements and methodological refinements to enhance the reliability and applicability of single-cell studies in spinal cord research. Future studies should address these challenges by integrating spatial transcriptomics with single-cell RNA sequencing and developing new computational tools to mitigate sampling biases and technical noise.

3.2. Investigating Spinal Cord Injury and Neurodegeneration

scRNA-seq has provided unprecedented resolution for dissecting the cellular heterogeneity and molecular mechanisms underlying SCI and neurodegenerative diseases. By enabling the high-throughput profiling of cell populations at various time points and pathological states, researchers can identify critical cellular subtypes, signaling pathways, and transcription factors, thereby advancing the understanding of disease mechanisms and uncovering potential therapeutic targets.

SCI is a severe CNS disorder characterized by disruption of spinal cord structure and function, leading to impairments in motor, sensory, and autonomic functions [42]. The application of single-cell transcriptomic technologies offers unique opportunities to explore cellular responses following SCI at single-cell resolution, allowing for the characterization of cellular composition, state transitions, and gene expression dynamics in injured spinal tissues [43]. Following SCI, the spinal cord undergoes complex cellular and molecular alterations, with glial cell responses being particularly well delineated by scRNA-seq studies.

Research has demonstrated that microglia and astrocytes exhibit pronounced dynamic changes after injury, forming multiple functional subtypes. Notably, microglia undergo a second wave of activation approximately 14 days post-injury, coinciding with secondary neuronal damage [44]. While other cell types tend to revert to their pre-injury states, microglia remain persistently activated, potentially altering the immune microenvironment in the long term [44]. For instance, certain microglial subtypes with high Cd68 expression may suppress regenerative capacities [44]. Zhang et al. (2025) conducted a cross-species study utilizing single-nucleus RNA sequencing and lineage tracing to comprehensively characterize neural cells in the spinal cord from development to injury [45]. This study revealed, for the first time, that astrocytes in the adult spinal cord possess significant multipotency and can transdifferentiate into mature oligodendrocytes following injury [45].

Oligodendrocytes and their progenitors are crucial for maintaining myelin integrity and promoting axonal regeneration [46]. scRNA-seq has enabled an in-depth analysis of the cellular heterogeneity within the oligodendrocyte lineage, elucidating their functional changes and intercellular interactions at various post-injury stages [47]. Notably, the Psap (prosaposin)/Gpr37l1 and Psap/Gpr37 ligand-receptor pairs have been implicated as key regulators within the oligodendrocyte lineage, and several transcription factors with potential regulatory roles have been predicted [47].

Moreover, snRNA-seq has revealed significant alterations in cells distal to the injury sites. In a mouse model, researchers analyzed lumbar spinal cord cells following thoracic SCI and observed dynamic responses across multiple cell types during the acute, subacute, and chronic phases [35]. Particularly, a rare neuronal population, including major spinocerebellar projection neurons, was found to express regeneration-associated genes post-injury [35]. These neurons demonstrated axonal preservation, extension, and remodeling, suggesting a potential role in functional recovery [35].

snRNA-seq has also shown that neurons exhibit distinct gene expression changes and structural and functional plasticity after SCI. Sarathwath et al. (2024) investigated regenerative processes in adult zebrafish six weeks post-severe SCI and discovered that adult neurogenesis and neural plasticity synergistically promote spinal repair [48]. They identified a population of injury-responsive neurons (iNeurons), which displayed significant plasticity and adopted neuroblast-like transcriptional features one week post-injury, thereby establishing zebrafish as a model for plasticity-driven neural repair [48].

Single-cell transcriptomic analyses have elucidated the complex immune responses following SCI. Gillespie et al. (2024), using humanized NSG-SGM3 mice, systematically analyzed human immune responses post-SCI via scRNA-seq [49]. The study identified diverse immune cell subpopulations and revealed the downregulation of T cell receptor signaling and antigen presentation, confirming the presence of spinal cord injury-induced immune deficiency syndrome (SCI-IDS) [49]. Concurrently, immune cells in the local lesion microenvironment are activated, with an upregulation of genes associated with proliferation and oxidative phosphorylation [49]. This model provides a valuable platform for translational studies on immune responses in SCI.

The disruption of the spinal vascular system is a critical pathological factor post-SCI, contributing to a deteriorating microenvironment, exacerbated inflammation, and nutrient deprivation at the injury site. Yao et al. (2022) conducted a comprehensive scRNA-seq analysis of vascular endothelial cells (ECs) in a rat SCI model and discovered that microglia and macrophages regulate specific EC subtypes via the SPP1 and IGF signaling pathways, promoting endogenous angiogenesis [50]. This study highlights the importance of immune–endothelial interactions in post-SCI vascular remodeling, offering new avenues for therapies aiming to enhance angiogenesis after SCI. Recent studies have also highlighted the role of inflammatory signaling pathways, such as IL-33/ST2, in modulating neuroinflammation and neuroprotection after SCI, offering potential therapeutic targets for intervention [51].

Spinal muscular atrophy (SMA) is a neurodegenerative disorder caused by the loss of SMN protein function [52]. While SMA primarily affects anterior horn motor neurons, non-neuronal cells also play significant roles in its pathogenesis [53]. The single-cell RNA sequencing (scRNA-seq) of spinal cord cells from severe SMA mouse models identified ten distinct cell types and their differentially expressed genes. CellChat analysis has revealed markedly reduced intercellular communication in SMA spinal cords [54]. Furthermore, a specific vascular fibroblast subpopulation is significantly diminished, potentially leading to vascular defects and widespread impairments in protein synthesis and energy metabolism [54].

During the secondary phase of SCI, the immune microenvironment at the injury site plays a critical role in spinal regeneration [55]. Previous studies have shown that macrophages/microglia exhibit both pro-inflammatory and anti-inflammatory functions during the subacute phase of SCI [56]. Zhang et al. (2023) integrated scRNA-seq and bulk RNA-seq data to delineate the immune infiltration landscape and identify potential therapeutic agents for SCI [57]. During the subacute phase, B2m, Itgb5, and Vav1 were identified as key molecular targets in macrophages/microglia, potentially promoting neural regeneration [57]. Additionally, low-dose decitabine modulates immune cell polarization and enhances the regenerative capacity of spinal tissues [57]. Identifying hub genes and targeted therapies for macrophage/microglial responses post-SCI is crucial for advancing neural regeneration.

In conclusion, single-cell transcriptomic technologies have been instrumental in unveiling cellular heterogeneity, disrupted cell-cell communication, and immune microenvironmental changes associated with SCI and neurodegenerative diseases. These tools have enabled researchers to dissect complex cellular responses and molecular mechanisms with unprecedented resolution. These studies have elucidated critical processes such as neuronal plasticity, glial trans-differentiation, immune modulation, and vascular remodeling, providing a robust theoretical foundation for the development of novel therapeutic strategies.

3.3. Spatial Transcriptomics and Multi-Omics Integration

scRNA-seq has significantly advanced our understanding of the cellular heterogeneity within the spinal cord. Proteomic insights, such as UBL3-mediated EV protein secretion in cancer cells [58], highlight the value of integrating transcriptomics with extracellular vesicle studies in spinal cord research. However, the lack of spatial context limits comprehensive insights into the localization and interactions of cells within the tissue microenvironment. The emergence of spatial transcriptomics addresses this limitation by enabling transcriptome-wide gene expression profiling while preserving the spatial organization of the tissues. This allows researchers to capture the spatial distribution of gene expression at a high resolution within tissue sections, facilitating a deeper understanding of spinal cord structure and function. Compared to conventional scRNA-seq, spatial transcriptomics can precisely map the spatial distribution patterns of distinct cell types within the spinal cord, retaining structural information and thereby offering a more holistic view of the spatiotemporal characteristics of neurons and glial cells.

Delile et al. (2019) utilized single-cell transcriptomic analysis to construct a spatiotemporal atlas of gene expression during mouse spinal cord development, identifying various neuronal subtypes and their spatial localization, thereby enhancing our understanding of spinal cord developmental mechanisms [17]. Russ et al. (2021) integrated six independent scRNA-seq datasets to generate a comprehensive cell-type atlas of the mouse spinal cord [59]. This study identified 84 distinct cell types and validated their spatial distribution using high-throughput in situ hybridization, revealing cell-type-specific distributions along the dorsoventral and rostrocaudal axes [59]. Additionally, they developed an open-source cell type classification tool, SeqSeek, to facilitate standardized cell type identification, providing a valuable resource for exploring spinal cord cellular diversity and spatial organization [59].

Zhang et al. (2024) applied the 10x Genomics Visium platform to generate spatial transcriptomes of lumbar spinal cord sections from six adult human donors (three males and three females, aged 47–59 years) [18]. They identified 17 cell types with distinct spatial distribution patterns and further integrated spatial transcriptomics with single-nucleus RNA sequencing to systematically classify neuronal subtypes and map their localization [18]. Notably, they uncovered sex-specific differences in gene expression within motor neurons, including the differential expression of SCN10A and HCN1, providing a molecular basis for sex-specific spinal cord functions [18].

Han et al. (2024) employed spatial transcriptomics and scRNA-seq to analyze the molecular characteristics of ischemic hemispheres in a mouse stroke model [60]. They found that the interaction between galectin-9 (LGALS9) and CD44 plays a crucial role in post-stroke neuroinflammation and repair [60]. Lgals9, delivered via extracellular vesicles, promoted functional recovery, highlighting its potential as a therapeutic target and revealing cell-type-specific responses in injured regions [60].

Recent studies have underscored the value of integrating mass spectrometry-based omics with spatial and transcriptomic approaches to enhance the molecular resolution of spinal cord research. For instance, matrix-assisted laser desorption/ionization (MALDI)-based mass spectrometry imaging (MSI) has provided spatially resolved metabolic profiles of the spinal cord, enabling the detection of molecular alterations following injury or disease progression [61]. In vascular and neuroinflammatory contexts, mass spectrometry proteomics has facilitated the identification of lipid and protein regulators involved in immune and endothelial cell signaling [62,63]. Complementing transcriptomic insights, these mass spectrometry-based approaches offer orthogonal validation of cell type-specific functional states and metabolic reprogramming processes. Notably, reviews and protocols emphasize the synergy between MSI and omics, promoting the high-content spatial mapping of nervous tissue [64]. Therefore, the convergence of MSI with single-cell and spatial transcriptomics is poised to enrich multi-omics analyses of spinal cord tissues, providing a deeper understanding of cell metabolism, signaling dynamics, and therapeutic targets [65,66].

Spatial multi-omics has demonstrated unique value in SCI research. By integrating immunohistochemistry with multiparametric analyses, spatial multi-omics can delineate microenvironmental changes during the secondary injury phase post-SCI [67]. Researchers have systematically reviewed advancements in spatial multi-omics for studying the post-injury spinal microenvironment, including alterations in the immune milieu, and have discussed potential future therapeutic strategies [67].

Spatial multi-omics technologies offer novel perspectives for understanding spinal cord disease. In amyotrophic lateral sclerosis (ALS) research, spatial transcriptomic analyses have shown that ALS risk genes are enriched in motor neurons and microglia [68]. This pioneering study introduced TF-seqFISH, an image-based single-cell transcription factor spatial decoding method, to investigate the spatial expression and regulation of transcription factors during human spinal cord development [68]. By integrating TF-seqFISH data with scRNA-seq, researchers have elucidated the spatial distribution of neural progenitors along the dorsoventral axis and identified the molecular and spatial characteristics governing neuronal generation, migration, and differentiation along the mediolateral axis [68]. These findings provide a valuable spatiotemporal transcriptomic resource for the developing human spinal cord and offer potential intervention strategies for SCI repair and ALS treatment.

These studies, despite their diverse methodologies, collectively illustrate how spatial transcriptomics can unravel the spatiotemporal features unique to both development and injury. For instance, developmental studies in mice (e.g., Delile et al. [17]) focus on progenitor domain patterning, whereas injury models (e.g., Zhang et al. [45]) explore dynamic cell fate changes, such as astrocyte-to-oligodendrocyte transitions. Moreover, human-focused datasets (Zhang et al. [18]) reveal conserved and species-specific molecular features that are not captured in rodent models.

Spatial transcriptomics complements single-cell and single-nucleus RNA-seq by preserving anatomical architecture and enabling the localization of transcriptomic signatures identified earlier. Together, these methods offer a layered understanding of spinal cord biology; scRNA-seq identifies cell types, whereas spatial tools anchor these identities in the tissue space. Recent advances in omics technologies have revolutionized the understanding of spinal cord biology. As illustrated in Figure 2, scRNA-seq and snRNA-seq have enabled the comprehensive characterization of cellular diversity, whereas spatial transcriptomics provides crucial spatial context by preserving tissue architecture. Proteomics and multi-omics approaches further expand this toolkit by integrating protein-level data and cross-omics correlations, offering unprecedented insights into the function and disease mechanisms of the spinal cord.

Despite its promise, spatial transcriptomics faces key limitations, including low resolution, limited gene throughput, and spatial misregistration errors [69,70]. Innovations such as Slide-seq V2, Stereo-seq, and spatial deconvolution algorithms (e.g., Tangram and SIMO) aim to enhance accuracy and scalability [71,72]. These improvements are critical for mapping fine-grained structures, such as the laminar spinal cord architecture.

The integration of single-cell/nucleus transcriptomics with spatial transcriptomics offers complementary insights: while scRNA-seq provides high-resolution cell type definitions, spatial transcriptomics preserves the spatial context within the tissues [73]. Effective multi-omics data integration requires robust computational methods to handle the heterogeneity and high dimensionality of various data types. Gao et al. (2023) proposed a graph convolutional network-based framework (GCN-SC) for single-cell multi-omics integration, effectively fusing diverse modalities and enhancing the accuracy of cell type classification [74]. Wang et al. developed contrastive cycle adversarial autoencoders to align and integrate single-cell multi-omics data, maintaining structural integrity while minimizing cross-modality discrepancies, thereby improving integration performance [75].

Cao et al. (2022) introduced GLUE (graph-linked unified embedding), a modular framework for integrating unpaired single-cell multi-omics data and inferring regulatory interactions [76]. By explicitly modeling cross-omics regulatory interactions, GLUE bridges the gap between data types, outperforming existing tools in terms of accuracy, robustness, and scalability [76]. Recently, Yang et al. (2025) proposed SIMO, a probabilistic alignment-based computational method for multi-omics spatial integration, which is capable of efficiently addressing modality differences and uncovering multi-level cellular topologies and regulatory patterns [71]. GLUE integrates unpaired omics data by modeling the regulatory interactions. Although it has not yet been specifically applied to spinal cord datasets, it has the potential to infer gene regulation in injury models. SIMO, a spatial multi-omics integrator, has demonstrated efficacy in spatial heterogeneity detection, which could enhance the mapping of regenerating regions post-SCI. GCN-SC employs graph convolution to improve cell type classification, which is crucial in tissues such as the spinal cord, where rare cell types are of interest [71,74,76]. Applications in both simulated and real biological datasets have demonstrated its high accuracy and robustness in capturing multimodal spatial heterogeneity, providing a powerful computational tool for dissecting molecular spatial organization [71].

Deep learning approaches have also shown tremendous potential for the integration of spatial transcriptomics with other modalities. Researchers have systematically reviewed deep learning applications for integrating spatial transcriptomics with histological images, chromatin imaging, and scRNA-seq data [77]. These methods can enhance spatial transcriptomic data interpretation, while spatial data provide essential contextual information for single-cell analyses.

In summary, spatial transcriptomics and multi-omics integration offer unprecedented opportunities for spinal cord research, enabling the comprehensive molecular and functional characterization of spinal cells within their native spatial contexts. With continuous advancements in technology and computational methodologies, this field is poised to make significant breakthroughs in spinal cord development, disease, and injury research.

4. Current Challenges and Limitations

Single-cell transcriptomics in spinal cord research faces several key limitations that hinder both experimental and clinical advancements. These challenges can be broadly categorized into technical, computational, and biological aspects (Figure 3).

4.1. Technical Challenges

scRNA-seq has demonstrated tremendous potential in spinal cord research; however, it still faces numerous technical challenges and limitations in its application. This section systematically discusses these obstacles to provide insights into future studies. Key technical issues include sample preparation and cell dissociation, data sparsity and technical noise, difficulty in detecting rare cell types, a loss of spatial information, batch effects, and standardization concerns.

The preparation of spinal cord tissue samples presents unique challenges. The spinal cord is characterized by high structural complexity and cellular heterogeneity, comprising tightly packed neurons, glial cells, and vascular components [78]. Moreover, distal neuronal structures, such as axons and dendrites, are rich in RNA [79]. Conventional tissue dissociation methods used in scRNA-seq can disrupt these structures, leading to the loss of specific RNA species and compromising data integrity and completeness. Ament et al. reported that scRNA-seq often fails to capture transcripts localized in distal compartments of neurons, including dendrites, axons, growth cones, synapses, and endfeet—collectively referred to as the “dark transcriptome”—which play crucial roles in neural development and function [80]. In injured spinal cord tissue, additional challenges arise: many cell types do not survive the harsh dissociation protocols, while surviving cells often exhibit stress-induced transcriptional artifacts [81].

To address this, researchers have adopted snRNA-seq, which does not require intact cells and better preserves cellular diversity, particularly in tissues that are difficult to dissociate. For instance, Sathyamurthy et al. utilized snRNA-seq to construct a cellular atlas of the adult mouse spinal cord, identifying 43 neuronal populations, thus demonstrating the utility of this method in complex tissues [19].

The detection of rare cell types remains a critical limitation. The spinal cord contains rare but functionally important cell types, such as specific neuronal and glial subpopulations [35]. Due to their low abundance, conventional scRNA-seq often fails to efficiently capture these cells, limiting our understanding of their biological functions. Although high-throughput platforms such as 10x Genomics have improved detection sensitivity, enrichment strategies or specific markers are still required to reliably identify rare populations [35]. For example, Milich et al. investigated the origin and fate of macrophages after SCI and highlighted their critical role in injury pathology, although their heterogeneity and dynamic changes remain insufficiently characterized [82].

Data sparsity and technical noise present significant analytical challenges. scRNA-seq datasets are inherently sparse, with many genes exhibiting zero expression in individual cells. These zero inflations result from low mRNA content, inefficient reverse transcription, and amplification bias [83]. High technical noise and biological variability further complicate data interpretation. Standard RNA-seq analytical approaches may be biased by cell-to-cell variability. Kharchenko et al. modeled single-cell measurements as a mixture of success and failure in transcript detection and proposed a Bayesian framework to mitigate expression distortions and enhance the accuracy of differential expression analysis [84]. Additionally, Liu et al. noted that in small-scale organisms, single-cell transcriptomics is severely affected by high variability, which affects data reliability and interpretation [85]. The heterogeneity of spinal cord cell populations further exacerbates these issues. Studies on SCI and neurodegenerative diseases have indicated that different cell types exhibit distinct responses to SMN protein deficiency, necessitating highly sensitive techniques to discern subtle transcriptional differences [54].

The growing volume of scRNA-seq datasets enables researchers to systematically map cellular transcriptomes in diverse biological contexts. However, integrating multiple datasets remains challenging because of the biological and technical heterogeneity across platforms [86].

Another critical limitation is the loss of spatial context of the data. Traditional scRNA-seq dissociates cells from their native tissue architecture, thus losing spatial information essential for understanding cell-cell interactions and microenvironments within the spinal cord. To overcome this, spatial transcriptomics techniques, such as Visium and LCM-seq, have been developed to capture gene expression data while preserving tissue structure [14,87]. Despite advancements in spatial methods such as Slide-seq and MERFISH, challenges remain in terms of resolution, throughput, and cost, preventing them from fully replacing scRNA-seq [39,69]. As Marx (2021) noted, while spatial transcriptomics allows transcriptome profiling within the tissue context, further technological improvements are needed for broader applications [70].

Batch effects also threaten the reliability of scRNA-seq data. Technical variability between experimental batches can introduce significant batch effects that compromise data comparability and integration. Although several algorithms (e.g., MNN, Harmony, and Seurat) exist to correct batch effects, challenges persist, especially when integrating data across conditions, platforms, or species [88]. Furthermore, the lack of standardized workflows limits the comparability of datasets between studies. Luecken and Theis outlined the current best practices in scRNA-seq analysis, including preprocessing, normalization, and batch correction strategies [89].

Despite these challenges, ongoing methodological advancements and computational innovations are gradually overcoming these limitations, offering new opportunities for mechanistic insights and therapeutic development for spinal cord diseases.

4.2. Computational and Interpretive Bottlenecks

The application of scRNA-seq in spinal cord research has expanded significantly; however, it faces numerous computational and interpretive challenges. Compared to traditional bulk RNA sequencing, scRNA-seq data are inherently noisier and more complex, presenting substantial analytical difficulties [90]. Due to the limited starting material, scRNA-seq data face various computational challenges, including normalization, differential gene expression analysis, and dimensionality reduction [91]. These bottlenecks not only compromise the accuracy and reproducibility of the data but also become critical limiting factors for the broader application of single-cell transcriptomics in spinal cord studies.

A major issue in scRNA-seq data is their high sparsity and technical noise, which complicate data preprocessing and quality control [92]. For instance, the dropout phenomenon of lowly expressed genes and batch effects significantly impact the accuracy of downstream analyses [92,93]. Therefore, quality control (QC) is essential for identifying and removing low-quality cells to ensure the reliability and reproducibility of the results. Although various normalization and batch correction methods, such as Seurat and Harmony, have been developed, their efficacy remains limited in complex tissues, such as the spinal cord [88].

The high sparsity of scRNA-seq data is one of the principal barriers to analysis. The imputation of missing values is a crucial strategy for addressing this issue. Several imputation methods tailored for scRNA-seq data, including ScImpute, SAVER, and MAGIC, have been proposed [94,95,96]. These methods vary in their assumptions and scalability, which influence their performance. However, some researchers argue that dropout patterns might carry biologically meaningful information useful for cell type identification, offering new directions for computational method development [97].

In the context of spinal cord research, these computational and interpretive bottlenecks are particularly pronounced. A study on the single-cell transcriptomics of the developing human spinal cord revealed a significantly higher number of cells detected in the CS17 sample than in the others, potentially due to environmental RNA accumulation [98]. This phenomenon was closely linked to a 24 h delay in sample preparation, underscoring the critical impact of sample preservation conditions and processing timeliness on data quality [98]. Moreover, the complexity and high heterogeneity of spinal cord tissues impose greater demands for the interpretation of single-cell sequencing data.

Integrating scRNA-seq with other omics data, such as ATAC-seq and proteomics, can provide a comprehensive understanding of cellular function and regulatory mechanisms. However, differences in scale, noise, and missing data across these data types pose significant challenges for multi-omics integration. Although methods such as MOFA and Seurat v3 have attempted to address these issues, limitations persist in their practical applications [99,100].

The biological interpretation and validation of scRNA-seq results are crucial, yet challenging steps. The potential for false positives, technical artifacts, and dependency on extensive biological knowledge complicates the interpretation process. Furthermore, the lack of standardized validation protocols and reference databases limits the reproducibility and credibility of the findings.

In summary, the computational and interpretive bottlenecks in spinal cord scRNA-seq research include complex data preprocessing, difficulties in cell type identification, limitations in trajectory inference, challenges in multi-omics integration, computational resource constraints, and obstacles in result interpretation. Addressing these issues requires the development of more efficient, accurate, and interpretable analytical methods to advance the application of scRNA-seq in spinal cord research.

4.3. Biological Complexity and Clinical Translation

scRNA-seq has significantly advanced our understanding of the cellular heterogeneity of the spinal cord. However, the clinical translation of these findings is impeded by several factors, including the intricacies of biological complexity, interspecies differences, and barriers to transitioning from experimental research to clinical applications.

Cellular heterogeneity and dynamic changes within the spinal cord represent major biological challenges. The spinal cord consists of various cell types, including neurons, astrocytes, oligodendrocytes, microglia, and vascular-associated cells [101,102]. These cells exhibit substantial heterogeneity and undergo dynamic shifts in different physiological and pathological conditions. For example, in SCI models, microglia and astrocytes have been shown to display distinct subtypes and functional states at various stages post-injury, indicating the presence of complex regulatory mechanisms in response to injury and repair [44]. Moreover, Fan et al. (2023) applied single-cell molecular profiling to rhesus macaque SCI models and identified unique molecular heterogeneity, spatiotemporal cellular dynamics, and intricate intercellular interactions, further highlighting the dynamic complexity of spinal cord tissue during injury [103].

Interspecies differences and their impact on clinical translation are critical challenges. Although animal models, particularly mice, are widely employed in spinal cord studies, substantial discrepancies exist in cell composition and gene expression profiles between species. Zhang et al. (2024) reported that certain genes highly expressed in the mouse spinal cord are minimally expressed or entirely absent in humans and vice versa [18]. A comprehensive transcriptomic comparison of canonical neuronal markers between human and mouse spinal cords revealed notable divergences in genes such as CALCA and GRP, underscoring the need for caution regarding species-specific gene expression in clinical translation [18]. These interspecies variations may result in the failure of therapeutic strategies validated in animal models during human clinical trials. Additionally, sex differences further complicate translational efforts, as Zhang et al. (2024) demonstrated sex-specific gene expression patterns in the human spinal cord that could influence disease susceptibility and treatment efficacy [18].

Furthermore, differences in the timing of glial cell development between humans and mice add another layer of complexity. Zhang et al. (2021) found that human astrocytes and oligodendrocytes begin differentiating as early as the 8th gestational week, whereas in mice, GFAP expression—an astrocytic marker—appears only at embryonic day 16 (E16) [104]. Such species-specific developmental timelines complicate the translation of findings from animal models to human contexts, as emphasized by Kushnarev et al. [105].

Barriers to translating experimental research into clinical applications further impede progress. One major limitation is the difficulty in acquiring human spinal cord samples, which restricts large-scale single-cell studies in clinical settings. Although Zhang et al. (2024) successfully sequenced lumbar spinal cord samples from nine human donors, such studies are rare [18]. The acquisition of paired pre- and post-treatment clinical samples is particularly challenging, limiting the exploration of therapeutic mechanisms at the molecular level in humans.

Although scRNA-seq offers new avenues for precision medicine in spinal cord disorders, translation from bench to bedside remains a formidable task. Awuah et al. (2023) highlighted the potential of scRNA-seq to revolutionize clinical practice in neurological diseases by providing personalized insights and improving prognoses [106]. However, most current research remains descriptive, and substantial work is required to translate these data into diagnostic tools, prognostic biomarkers, or therapeutic targets.

Finally, technical and operational hurdles in clinical settings remain to be significant. There is no universally accepted standard for sample collection and processing, resulting in difficulties in comparing or integrating data across studies [107]. Additionally, the analysis and interpretation of scRNA-seq data require sophisticated bioinformatics expertise, which currently limits its widespread clinical implementation [107]. The high cost and technical complexity of scRNA-seq also pose barriers, although technological advancements have somewhat mitigated these issues. Further efforts are required to decrease costs and simplify the workflows for large-scale clinical research.

Although scRNA-seq excels at identifying rare cell types and transient states, offering a resolution that bulk RNA sequencing cannot match, it also suffers from significant limitations. The high cost and technical complexity of scRNA-seq render it inaccessible to many laboratories lacking specialized equipment or expertise [92]. Moreover, the process of dissociating spinal cord tissue into single cells can lead to cellular stress and changes in gene expression, introducing artifacts that may compromise data accuracy [108]. Additionally, sampling biases can result in the underrepresentation of rare cell types or specific developmental stages, while technical noise inherent to single-cell sequencing can obscure low-abundance transcripts and subtle gene expression changes.

The integration of spatial transcriptomics with scRNA-seq has begun to address some of these issues by providing a spatial context for gene expression data. This integration enhances our understanding of cellular interactions and tissue architecture. However, spatial resolution remains a significant challenge. Current techniques often fail to capture the intricate spatial organization of cells within the spinal cord, which is critical for understanding cellular functions and disease mechanisms [109]. Furthermore, the integration of multi-omics data, including proteomics and epigenomics, is essential for a comprehensive understanding of cellular regulation and function. This integration could potentially reveal new layers of complexity and provide a more comprehensive picture of spinal cord biology.

Despite the promise of scRNA-seq in identifying novel therapeutic targets, significant challenges remain in translating these findings into clinical applications. Bridging the gap between basic research and clinical practice remains a major hurdle. While scRNA-seq can identify distinct transcriptional programs in motor neurons, suggesting potential targets for modulating neuronal activity, the path from discovery to effective treatment is fraught with difficulties [19]. The development of advanced computational tools is crucial for handling the large-scale data generated by scRNA-seq and mitigating biases and noise. These tools must be refined to improve the reliability and applicability of single-cell studies in spinal cord research in the future. As the field continues to evolve, addressing these challenges is essential to fully realize the potential of scRNA-seq in advancing our understanding of spinal cord biology and disease.

In summary, despite these challenges, continued technological innovation and interdisciplinary collaboration are expected to enable scRNA-seq to drive transformative progress in spinal cord research, ultimately benefiting patients with spinal cord disorders.

5. Future Directions and Perspectives

5.1. Toward a Comprehensive Spinal Cord Cell Atlas

The construction of a comprehensive spinal cord cell atlas is one of the primary goals of current single-cell transcriptomic research. Although several studies have classified the cell types of the mouse spinal cord, these datasets have not yet been integrated into a unified reference framework. For instance, Russ et al. compiled existing single-cell transcriptomic data to construct a mouse spinal cord cell atlas, revealing the hierarchical organization and spatial distribution of diverse cell types [59].

In human spinal cord studies, Yadav et al. employed single-nucleus RNA sequencing combined with spatial transcriptomics to classify adult spinal cord cell types, enriching our understanding of human spinal cord cellular diversity [78]. Zhang et al. utilized single-nucleus RNA sequencing and spatial transcriptomics to create a detailed cell atlas of the adult human spinal cord, identifying 21 neuronal subtypes and elucidating their spatial distributions and sex-specific differences [18]. Sex-specific transcription in motor neurons, such as the differential expression of SCN10A and HCN1 [18], may underlie the sex-biased prevalence of disorders such as ALS and neuropathic pain. While initial findings come from human datasets, comparative analyses with mouse models suggest a partial conservation of these differences, although further investigation is required to determine their developmental regulation and functional relevance.

To achieve a fully comprehensive spinal cord cell atlas, future research must integrate single-cell data from diverse studies and establish standardized criteria for cell type classification. It is also essential to consider interindividual variability, including factors such as sex, age, and disease states, which may influence cellular composition. For example, the MOp census and atlas provide a robust framework for integrating cross-species cell-type data, offering a model that spinal cord research can adopt to standardize annotation and enhance cross-study compatibility [110].

Moreover, future spinal cord atlases should integrate multi-omics data that extend beyond transcriptomics alone. A recent example, Tabulae Paralytica, exemplifies this trend by incorporating four SCI datasets, including a single-nucleus transcriptomic atlas of 500,000 cells, multi-omics data combining transcriptomic and epigenomic measurements, and two spatial transcriptomic maps covering four spatial and temporal dimensions [81]. This platform facilitates the integration of four-dimensional, multimodal, and genome-scale datasets for biological and medical research [81].

With the growing availability of spinal cord single-cell datasets, establishing a unified reference framework and standardized cell-type identification methodologies has become increasingly critical. Additionally, future research should focus on the development of more advanced computational tools, such as SeqSeek, an open-source cell type classifier, to promote standardized cell type annotation [59].

These efforts will collectively contribute to a deeper understanding of the cellular composition and functional organization of the spinal cord, laying the groundwork for advancing research and therapies targeting neurological diseases.

5.2. AI and Machine Learning in Single-Cell Analysis

With the exponential growth of single-cell transcriptomic data, artificial intelligence (AI) and machine learning (ML) techniques have emerged as critical tools for managing and analyzing complex datasets, offering vast potential for discovery. Deep learning, in particular, has demonstrated strong promise in single-cell data analysis and is expected to unlock further opportunities in the future [111]. In spinal cord research, these computational methods have revolutionized research paradigms by optimizing data preprocessing, enhancing cell type identification and marker gene prediction, improving annotation accuracy, elucidating dynamic trajectories, and integrating multimodal data.

Recent advances in deep learning, especially graph neural networks (GNNs) and autoencoders, have shown great efficacy in preprocessing and denoising scRNA-seq data. For example, Wang et al. introduced the scGNN framework, which constructs graph-based representations of cellular relationships and applies multimodal autoencoders to mitigate data sparsity, improving both cell clustering and gene expression imputation, with broad applicability in single-cell neural studies [112]. Furthermore, Xi et al. systematically evaluated various autoencoder architectures for single-cell data imputation, and their optimized designs significantly enhanced downstream analysis accuracy, providing a solid foundation for high-quality data mining [113].

AI/ML methods have also excelled in cell-type identification and marker gene prediction. Cao et al. systematically assessed 13 supervised learning algorithms for phenotypic classification in scRNA-seq data and found that ElasticNet and XGBoost consistently outperformed the others across diverse dataset scales, demonstrating their potential for robust cell-type classification [114].

For cell type annotation, tools such as scDeepSort, ACTINN, and scBERT enable highly accurate predictions across platforms. Meanwhile, preprocessing and noise reduction are addressed by models such as scGNN and autoencoder-based frameworks. scDeepSort, a weighted GNN-based pretrained annotation tool, enables high-accuracy cell type labeling without relying on predefined marker genes or RNA-seq references, making it broadly applicable across tissues and platforms [115]. Similarly, Automated Cell Type Identification using Neural Networks (ACTINN) offers the rapid and accurate identification of cell types, validated across murine leukocytes, human PBMCs, and human T-cell subtypes, suggesting its utility as an adjunct in scRNA-seq pipelines [116]. Yang et al. developed scBERT, a BERT-inspired model for gene expression pattern recognition [117]. Separately, Wang and Du introduced WCSGNet, a GNN-based approach that focuses on cell-specific gene networks, enabling precise annotation in diverse cellular contexts [118].

With established cell atlases, the full potential of gene expression analysis can be leveraged to define and decode spinal cord cell types [119]. The ScnML model, specifically designed for spinal cord neuron subpopulation recognition, enhances the classification accuracy in high-dimensional feature spaces and addresses the challenges of computational efficiency and overfitting seen in conventional models [120].

Developmental trajectory inference and dynamic process modeling are crucial for uncovering cell-state transitions, fate decisions, and tissue developmental mechanisms. Fang et al. introduced Chronocell, a trajectory inference approach based on “process time,” which improves the biological interpretability of developmental pathways and is particularly suited to complex dynamic reconstructions [121]. In spinal cord development, Shi et al. integrated scRNA-seq, spatial transcriptomics, and TF-seqFISH to delineate the migration and differentiation trajectories of neural progenitors along the dorsoventral and mediolateral axes, highlighting the finely tuned spatiotemporal regulation by transcription factors [68]. This study deepens the understanding of spinal cord developmental dynamics and proposes novel time-space-integrated modeling approaches [68].

Deep learning also shows great potential for multimodal integration and spatial transcriptomics. Biancalani et al. developed Tangram, a deep learning-based framework for the precise alignment of scRNA-seq data with various spatial transcriptomics technologies (e.g., MERFISH, Visium), enabling genome-wide spatial expression reconstruction at single-cell resolution [72]. Recently, researchers introduced scNET, which combines scRNA-seq data with protein–protein interaction networks, leveraging GNNs to enhance gene function representation and pathway identification, providing a novel tool for multi-omics integration [122].

In the context of SCI, researchers have pursued multidimensional explorations, from molecular characterization to clinical trajectory analysis. Liu et al. applied the ScnML model to analyze scRNA-seq data, achieving a high-precision prediction of neuronal subpopulations and identifying functional state-related marker genes, thus offering molecular insights into post-injury cellular dynamics [120]. Concurrently, Jaja et al. developed a classification system for recovery trajectories in patients with cervical SCI based on longitudinal clinical data, revealing dynamic heterogeneity in functional recovery and underscoring the prognostic significance of recovery patterns over initial injury severity [123]. These studies jointly establish a comprehensive framework linking single-cell molecular features with clinical outcomes, thereby advancing precision diagnostics and therapy for SCI.

The combined application of ML, statistical methods, and AI in scRNA-seq analysis significantly enhances cell type identification and an understanding of cellular heterogeneity and provides insights into biological processes [124]. These approaches are anticipated to improve disease stratification, foster innovative therapeutic strategies, and propel precision medicine [125].

Despite these advancements, AI/ML techniques still face challenges in terms of data quality control, model interpretability, and cross-platform data integration. Future research should focus on developing more robust algorithms, enhancing interpretability, and standardizing data processing protocols to facilitate the widespread application of AI/ML in single-cell spinal cord research.

In conclusion, AI and machine learning hold great promise for single-cell transcriptomic analysis, particularly in spinal cord studies. Through continuous algorithm refinement and multisource data integration, these technologies are poised to drive spinal cord disease diagnosis and treatment into a new era of precision medicine.

5.3. Personalized Medicine and Regenerative Therapies

While AI/ML tools significantly enhance the analysis of scRNA-seq data, their full potential is realized when applied in translational contexts, such as regenerative medicine, where single-cell insights can guide therapeutic innovations. With the rapid advancement of scRNA-seq technologies, spinal cord research has entered a new era characterized by personalized medicine and regenerative therapies. By enabling a detailed analysis of cellular heterogeneity and molecular signatures, scRNA-seq provides a robust foundation for precise diagnosis and individualized therapeutic strategies in cancer.

The application of scRNA-seq in personalized medicine lies in its capacity to delineate the transcriptional profiles of diverse cell types within spinal cord tissues with high resolution. In particular, following SCI, scRNA-seq facilitates the identification of specific cellular subpopulations and their dynamic responses. For instance, during the subacute phase of SCI, immune-related genes such as B2m, Itgb5, and Vav1 are upregulated in microglia and macrophages, suggesting their potential as key targets for modulating immune responses and promoting neural regeneration [57]. Furthermore, machine learning models, such as ScnML, integrated with scRNA-seq data, have demonstrated high efficiency in predicting spinal neural cell subtypes and identifying novel marker genes, thereby offering powerful tools for personalized interventions [120]. Moreover, scRNA-seq has revealed significant heterogeneity and dynamic changes among various cell types in the post-injury microenvironment, underscoring its pivotal role in elucidating the mechanisms of injury and guiding regenerative strategies. [126].

In regenerative medicine, scRNA-seq is a powerful tool for understanding cellular diversity and developing novel therapeutic approaches. In neural stem cell studies, scRNA-seq has shown that Nestin-positive neural stem cells are activated post-SCI, exhibiting regenerative potential, implying that these endogenous cells may contribute to neural repair [127]. Additionally, astrocytes have been found to transdifferentiate into oligodendrocytes following injury, participating in remyelination processes and offering new strategies for regeneration [104]. Notably, adult zebrafish exhibit an intrinsic capacity for spinal cord regeneration, with the neurogenesis of glutamatergic and GABAergic neurons helping to restore excitatory/inhibitory balance, making them an ideal model for plasticity-driven repair [48]. Furthermore, Li et al. utilized human pluripotent stem cell-derived spinal cord organoid (SSCO) models in conjunction with scRNA-seq to map the developmental trajectory of the spine and spinal cord, identifying HMMR-positive bipotent neuro-mesodermal progenitors, thereby offering novel tools for regenerative therapies and disease modeling [128].

Looking ahead, the integration of scRNA-seq with spatial transcriptomics, epigenomics, and proteomics promises a more comprehensive understanding of the molecular mechanisms underlying SCI, thereby accelerating the development of personalized medicine and regenerative treatment. For example, the integration of single-cell and spatial transcriptomics has enabled the construction of a cellular atlas of the adult human spinal cord, providing new insights into its complex architecture [78]. Moreover, the application of artificial intelligence and machine learning will further expedite the extraction of biologically meaningful information from large-scale single-cell datasets, driving the realization of precision medicine applications.

In summary, single-cell transcriptomics has brought about transformative changes in spinal cord research, deepening our understanding of spinal development and pathology, and laying a solid foundation for the advancement of personalized medicine and regenerative therapies. As technologies continue to evolve and interdisciplinary collaborations intensify, scRNA-seq will undoubtedly drive further progress in both basic research and clinical applications related to spinal cord injury.

6. Conclusions

In this review, we systematically summarize recent advancements in scRNA-seq in the field of spinal cord research. By integrating findings from recent studies, it is evident that scRNA-seq has led to groundbreaking progress in various areas, including spinal cord developmental biology [17,68], mechanisms of injury repair [126,129], and neurodegenerative disease research [54]. scRNA-seq has revealed cellular heterogeneity, developmental trajectories, and intercellular interactions within spinal cord tissues through high-throughput sequencing of transcripts at the single-cell level, offering unprecedented resolution for deciphering the complexity of spinal cord biology. Clinically, single-cell technologies are reshaping the paradigms of diagnosis and treatment of spinal cord disorders.

Despite these notable advantages, the application of single-cell transcriptomics in spinal cord research faces several challenges. These include difficulties in sample acquisition, the computational complexity of data analysis, and the necessity for a biological validation of the findings. Moreover, high technical costs and a lack of standardized protocols limit its broader implementation in clinical settings.

Looking forward, the integration of scRNA-seq with emerging spatial transcriptomics and multi-omics technologies is expected to facilitate the construction of a more comprehensive cellular atlas of the spinal cord, elucidating the spatiotemporal dynamics of the cellular processes. The adoption of artificial intelligence and machine learning approaches will further enhance the efficiency and precision of data analysis, thereby advancing personalized medicine and regenerative therapies. The convergence of single-cell transcriptomics with these cutting-edge technologies is poised to propel spinal cord research into a new era of discovery.

In conclusion, single-cell transcriptomics offers a powerful tool for spinal cord research, deepening our understanding of spinal cord development, injury, and disease. With continuous technological innovations and interdisciplinary collaborations, scRNA-seq is anticipated to play an increasingly vital role in both fundamental research and clinical applications, fostering sustained progress in neuroscience.