This article is mentioned in:

Introduction

Each cell follows a distinct developmental trajectory and history. Mutations occur during cellular development due to factors such as the cell cycle, fluctuations in the cellular microenvironment, developmental processes, senescence and infection (1). Multicellular organisms integrate trillions of highly specialized cells derived from a single zygote, continually developing specific genomic landscapes during differentiation to form complex functional entities (2). Extensive research has been conducted on cells [including cell lineage tracing (3) clonal behavior and dynamics (4) and single-cell transcriptomics (5). The introduction of nucleic acid sequence barcoding technology has provided a powerful tool for the aforementioned studies. Prior to this, various methods were employed to label cells, such as time-lapse microscopy for direct observation in Caenorhabditis elegans (6), dye injection, transplantation, viral transduction (7), or genetic recombination of fluorescent proteins to mark (8–13) and track cells of interest. Despite the power of these methods, they have limitations, including a limited number of generations still marked after dye injection, non-physiological settings of cell transplantation, low frequency of viral barcode insertion, or a limited number of fluorescent proteins available to mark complex tissues and these methods are labor-intensive. There is a need for higher throughput, improved quantification and the ability to track large numbers of single-cell systems simultaneously to further advance the field.

Therefore, the concept of nucleic acid sequence barcoding has been introduced to effectively address the aforementioned issues in cell research (14,15). Cell barcoding technology marks each cell in highly heterogeneous cell populations with a unique barcode sequence. Nucleic acid sequence barcodes are inherited from parent cells to progeny along developmental trajectories, thus allowing the reconstruction of their genetic relationships by deciphering the nucleotide sequence information within the barcodes (16). This technology can track cells or molecules across temporal and spatial boundaries. In theory, the number of possible barcodes is limitless, allowing the parallel tracking of millions of cells. These barcodes used for marking are akin to a cell's ‘identification card’, enabling the rapid and accurate identification of different cell types (17). In recent years, the application of nucleic acid sequence barcoding has expanded across various fields. For instance, deciphering clonal dynamics in disease (18), further improving the detection of natural or synthetic barcodes and performing multimodal cell state measurements at the single-cell resolution can improve understanding of the origins of human diseases, explore cancer heterogeneity (19) and investigate cancer metastasis mechanisms (20). Using mice as a model, clustered regularly interspaced short palindromic repeats (CRISPR) barcoding has constructed lineage trees throughout the developmental process, providing a viable multifunctional platform for in vivo barcoding and lineage tracing in animal model systems (16). Additionally, in the field of cancer liquid biopsies, cell barcoding has been applied to reveal the immunological characteristics of cancer and the heterogeneity of extracellular vesicles (21,22). Furthermore, ‘RNA barcodes’ have been developed to elucidate the combinatorial signaling pathways driving embryonic stem cell (ESC) differentiation (23). However, cell barcoding technology still faces challenges and risks. The first is the issue of incorrect assignment due to barcode similarity, especially in genetic barcoding, where highly similar DNA tags may be erroneously allocated. Other inherent risks include mutations in genetic markers during PCR and conventional library sequencing steps, or partial loss of barcode reads in single-cell data when using transcriptomic libraries (7). Moreover, the dynamic changes of barcodes present limitations in tracking clonal evolution, especially in scenarios requiring the tracking of clones over time (18). Lastly, the data analysis and computational complexity of cell barcoding are high and the development and application of these methods remain a challenge.

With the rapid development of microfluidic technology and advancements in RNA sequencing, precise analysis at the single-cell level has been achieved (24). By integrating cellular barcoding techniques, high-throughput transcriptomic analyses can be conducted on a vast number of cells within heterogeneous populations, thereby elucidating population structures, the interplay of gene expression and the heterogeneity during differentiation processes (25). Previously, DNA barcodes for spatial omics have been described and the integration of these barcodes with fluorescence microscopy has been explored, with a focus on cellular localization and tissue structure studies (26). Although the present review advanced the application of spatial omics, it did not delve into the challenges and diverse applications of nucleic acid barcodes in lineage tracing and disease modeling. By contrast, the present review focused on the complex challenges and broad applications of nucleic acid barcodes in these fields. Additionally, previous studies have mainly concentrated on the application of barcodes in lineage tracing, particularly in stem cell biology and cancer research (17,27). However, these studies did not comprehensively address the emerging complexities of barcode technology, such as high-throughput sequencing and the integration of multidimensional data. More importantly, while the literature reviews the historical development of lineage tracing methods and their applications in developmental biology and cancer research (3), it does not cover the latest advances and breakthroughs in barcode technology. The present review examined nucleic acid sequence-based barcoding strategies, which are primarily categorized into two main types: Natural barcodes and synthetic barcodes. It further discussed the advantages, challenges and potential solutions of nucleic acid sequence barcoding in practical applications; despite significant progress in nucleic acid sequence barcoding technology, there is still room for improvement in sensitivity, specificity, cost-effectiveness and multiplexing capabilities. Lastly, the present review projected the future development of nucleic acid sequence barcoding technology. With the progress in synthetic biology and gene editing technologies, this field is expected to witness further innovations and breakthroughs. In the future, we may see the emergence of more efficient and cost-effective barcoding systems, which will markedly advance biomedical research and clinical applications. It is expected that nucleic acid sequence barcoding technology will make a more substantial contribution to understanding complex biological processes and improving human health (Fig. 1).

Figure 1. Schematic diagram of the architecture and application of nucleic acid. Image created using Adobe Illustrator 2023 (Adobe Inc.).

Principles and types

Since the inception of nucleic acid sequence barcodes, a variety of distinct barcode types have been developed, including Polylox barcodes, CRISPR barcodes, integration barcodes, droplet barcodes and native barcodes (3). Nucleic acid sequence barcodes distinctly label each cell within a highly heterogeneous cell population through unique barcode sequences. As the developmental trajectory unfolds, barcode sequence information is inherited from progenitor cells to their progeny, thereby elucidating lineage history (28). The principles, advantages, limitations, applications and challenges associated with various barcode strategies is comprehensively discussed later and summarized in Table I.

Table I. Barcoding technology.

Table I.

Barcoding technology.

Barcode type	Principle	Advantages	Disadvantages	Applications	Challenges
CRISPR-Cas9	Uses the CRISPR-Cas9 system to insert barcode	1. High specificity and accuracy	1. Barcode insertion may interfere with Gene expression	1. Used for cell tracking	1. Optimizing editing efficiency and accuracy
barcodes	sequences at specific genomic locations guided by gRNA. Each	2. Enables quantification and tracking of individual cells or EVs	2. Requires precise gene editing techniques	2. Post-genome editing tracking	2. Challenges in cross-species applications
	barcode corresponds to a different cell or EV	3. Strong scalability	3. Low editing efficiency possible	3. Single-cell analysis	3. Biosafety concerns
Polylox barcodes	Based on the Polylox system, which generates	1. Multiple barcode system	1. System complexity may lead to instability in barcode insertion	1. Used for cell population tracking	1. Barcode variability
	multiple unique barcode genes randomly inserted into the	2. High-throughput tagging and tracking in cell populations	2. Barcode mutations may reduce tracking accuracy	2. Cancer research	2. Long-term stability of the system
	genome for high-throughput analysis in cell populations	3. Suitable for long-term cell fate tracking		3. Long-term cell fate analysis, etc.	3. Challenges in multi-cell applications
Integration barcodes	Uses insertion barcode systems where	1. Long-term stability	1. Barcode insertion may be biased by genomic location	1. Used for long-term cell line tracking	1. Selection bias in insertion sites
	viral or other vectors directly insert barcode sequences into the target cell genome,	2. No external intervention required; barcodes are stably passed on during cell division	2. Potential interference with target cell function	2. Stem cell research	2. Potential conflict with cell function
	which are inherited during cell division	3. Precise single-cell tracking		3. Genome editing tracking	3. Requires efficient viral or vector systems
Droplet barcodes	Microfluidic droplets are used to partition and	1. High throughput and sensitivity	1. Droplet stability can affect analysis	1. Used for single-cell analysis	1. Stability and consistency of microfluidic technology
	tag individual cells or EVs, enabling high-	2. Precise single-cell or single-EV analysis	2. Requires complex microfluidic devices	2. High-throughput single-molecule sequencing	2. Equipment and data analysis costs
	throughput parallel analysis	3. Can perform multiparametric analysis for large-scale data generation	3. High complexity in data analysis	3. EV analysis	3. Droplet marking errors
Native barcodes	Relies on natural barcodes, such as RNA or DNA	1. No external insertion required, naturally occurring system	1. Dependent on the presence and stability of natural markers	1. Used for cell tracking, RNA studies	1. Limitations of natural markers
	barcodes, which are directly recognized by	2. Suitable for molecular-level tagging	2. Potential competition with target molecules or systems for binding	2. Species separation based on natural markers	2. Stability of natural markers in the system
	biological markers without external insertion	3. Relatively stable markers for long-term monitoring			3. Sensitivity issues in localization and recognition

Principles of cellular barcoding

DNA cellular barcoding is a high-throughput method widely used to track cell lineages across various fields (18,29), including hematopoiesis (7,30,31), development, cancer (20,32–34) and infectious disease dynamics (35). It employs a unique, inheritable DNA sequence integrated into the genome of an ancestral cell, with descendants detected through sequencing. In theory, the number of cellular barcodes is limitless, with the number of possible barcodes being 4n, where n is the length of the sequence (since each position can encode for one of the four bases). Thus, a random 10-base pair (bp) barcode can assume any of 410 (~106) different sequences, while a random 30-bp barcode can assume any of 430 (~1018) different sequences, each serving as a unique identifier. Regarding the design principles, barcodes are typically categorized into random and semi-random types. Random barcodes, such as those used by Macosko et al in 2015 (36) who combined cellular barcoding with microfluidic technology for high-throughput cell labeling, consist of completely randomized nucleotide sequences. Semi-random barcodes, on the other hand, leverage recombinase-based or gene-editing technologies such as CRISPR/Cas9 and are integrated into cellular DNA via viral vectors or other methods for cell marking (37,38). As cellular barcoding technology continues to evolve, researchers have designed various types of DNA cellular barcodes to suit different research needs and experimental designs. The design and application of these barcodes provide a powerful tool for gaining deep insights into cellular behaviors and biological processes (17).

Types of barcode design and challenges

Design

The Polylox system facilitates high-resolution lineage tracing and multiplexing, enabling concurrent tracking of multiple cell populations. This system offers valuable insights into tissue regeneration, developmental biology and cancer progression. Understanding the cellular origin of various cell types or tumors can uncover critical biological mechanisms (39). In the seminal work by Cai et al a recombinase-based nucleic acid sequence barcode technology was first introduced. This technology employs Cre recombinase to excise or invert specific DNA sequences, which interact with a series of open reading frames of fluorescent proteins, followed by analysis through fluorescence imaging (40). However, the repetitive action of Cre may result in random recombination and collapse of the target array. In 2017, Pei et al (41,42) used the Polylox barcode system to elucidate the in vivo fate of hematopoietic stem cells and it was further applied it in 2019 for lineage tracing in mice (41,42). The Polylox system is an advanced in vivo cellular lineage tracing technology that introduces artificially synthesized recombination substrates into specific loci in mice (such as the Rosa26 locus) and employs the transient activity of Cre recombinase to induce random DNA recombination, generating unique DNA barcodes (Fig. 2). The Polylox system comprises 10 loxP sites, spaced 11 base pairs apart, with the nine intermediate DNA fragments carrying unique sequences from the Arabidopsis AT2G21770 gene, thereby forming the ‘alphabet’ of the barcode (3,43,44). This innovation confers the Polylox barcode with enhanced functionality, offering a more robust and scalable fate mapping method compared to traditional techniques, with high precision, as well as superior multiplexing capabilities and long-term tracking potential. The versatile application of the system renders it suitable for a broad range of experimental models. It can be employed in studies on stem cell behavior, tumor progression, or immune cell tracking, providing a versatile approach for diverse biological investigations, all without the necessity of exogenous markers. The Polylox system employs an endogenous genetic barcode to track cells, offering a more natural and generalizable method of tracking (41). However, the complexity of system setup, such as the insertion of multiple recombinase target sites (loxP or other variants), may present technical challenges and restrict its use in certain laboratories (41). Additionally, the system may result in mosaic expression in certain cells and not all cells may undergo recombination events, potentially introducing bias when analyzing cellular fate. Another consideration is tissue-specific recombination efficiency, as different tissues and developmental stages exhibit variable recombination efficiencies. This could lead to discrepancies between barcode outcomes and expected results. During this process, barcode drift or loss may occur, resulting in insufficient representation of certain barcodes over time, particularly in rapidly dividing or regenerating tissues (45). It is also important to note that, during the tracking of large cell populations, barcode overlap may occur, complicating the differentiation of closely related cells or their progeny (44).

Figure 2. Polylox Barcodes. Designed as an endogenous DNA cassette that generates barcodes based on the Cre-lox recombination system and is capable of expression driven by the Rosa26 promote. Image created using Adobe Illustrator 2023 (Adobe Inc.); image modified from (41).

Challenge

The Polylox barcode system generates barcodes by chemically inducing alterations to specific sequences. However, the insertion of these barcodes is inherently stochastic, resulting in uneven distribution and expression across cell populations (46,47). This issue can be addressed by optimizing the chemical induction conditions and reaction system to enhance the uniformity and stability of barcode insertion. Barcodes are frequently employed in cellular contexts; however, their introduction may alter cellular states, thereby confounding the interpretation of biological outcomes. Moreover, the recovery rates of both cells and barcodes pose a potential risk. Given the stochastic nature of barcode insertion, some cells may fail to successfully generate a barcode, resulting in low recovery rates. Employing multiple rounds of selection can ensure that a sufficient number of cells receive functional barcodes, while the development of optimized capture techniques can further improve recovery rates. It is well-established that the diversity of barcodes facilitates the generation of millions, or even tens of millions, of distinct barcodes. However, the intrinsic randomness of the Polylox barcode generation process further complicates data analysis. This challenge is particularly pronounced in large-scale cell populations, where the accurate separation and identification of barcode sequences constitutes a significant obstacle. To address this challenge, statistical model algorithms, such as Bayesian analysis, have been developed to handle the diverse barcode data, thereby enhancing the accuracy of data analysis (47–49).

CRISPR Cas9-based barcodes

The CRISPR-Cas9-based barcoding technology exploits the potential of the CRISPR-Cas9 system to introduce specific DNA modifications (that is, barcodes) into the genomes of individual cells (50). These barcodes, comprising unique synthetic DNA sequences, function as molecular tags, facilitating the tracing of cell lineages and associating genetic information with various cellular attributes, including gene expression and therapeutic responses. A variety of CRISPR-Cas9 barcode systems have been developed, including GESTALT, scGESTALT, macsGESTALT, synthetic barcodes delivered via piggyBac transposase, CREST and ScarTrace (51–56) (Fig. 3). In zebrafish models, a CRISPR/Cas9-based synthetic barcode system was evaluated, capable of accumulating informational mutations during cellular division events and throughout the organism's developmental process (47). This system is applicable to a wide range of organisms. However, due to saturation issues, the GESTALT system is restricted to early embryonic stages and exhibits reduced precision in reconstructing complex lineage trees. To overcome this limitation, single-cell RNA sequencing (scRNA-seq) technology is incorporated, as demonstrated by scGESTALT and its applications in zebrafish and macsGESTALT mouse models (51,53,54). This technology facilitates barcode editing at multiple time points, capturing lineage information from later developmental stages. The scGESTALT method is applicable to various multicellular organisms, enabling the characterization of molecular identities and lineage histories of thousands of cells throughout development and disease progression (53). As an inducible CRISPR-Cas9-based lineage recorder, macsGESTALT efficiently captures single-cell transcription and phylogenetic data (54). When applied to an in vivo model of pancreatic cancer metastasis, macsGESTALT revealed that metastatic activity peaked at a specific late-stage hybrid epithelial-mesenchymal transition (EMT) state (51,53,54). Additionally, CARLIN technology has been employed to identify inherent biases in the clonal activity of fetal liver hematopoietic stem cells (HSCs), enabling an unbiased, global analysis of both lineage history and gene expression in single mouse cells (55). Another variant of the CRISPR/Cas9-based synthetic barcode approach uses the piggyBac transposon, which efficiently integrates synthetic barcodes and active specific guide RNAs (sgRNAs) into target cells. In an initial model, engineered K562 cells containing a piggyBac library with sgRNAs and their respective target site boxes (including an inducible Cas9 system) were introduced into fertilized eggs to reconstruct lineage development during mouse embryogenesis (56). This model was subsequently optimized to enable the direct implantation of Kras-mutant A549 cells into mouse lung tissue or the generation of a genetically engineered mouse model of induced lung adenocarcinoma (KP-Tracer), facilitating the tracking of tumor evolution via barcode information (57,58). Moreover, a CRISPR-based lineage-specific tracing method, CREST, was developed for clonal tracing in Cre mice. Using two complementary strategies based on CREST, the single-cell lineage of the developing mouse ventral midbrain was mapped. The CREST method and its strategies allow for comprehensive single-cell lineage analysis, offering new insights into the molecular programming of neurodevelopment (59). Fluorescent proteins, in addition to serving as standalone color barcodes, were integrated with CRISPR-Cas9-mediated nucleic acid sequence barcodes to create fluorescent protein genetic markers. Most CRISPR/Cas9-based lineage tracing methods employ an inducible editing system, enabling temporal and/or spatial regulation within specific developmental timeframes to track diverse cell lineages. These methods have also been widely used to study lineage relationships across various cell types (38,60). However, CRISPR/Cas9 editing may induce off-target effects, necessitating highly specific sgRNA sequences to precisely target synthetic barcodes and avoid unintended mutations (61,62). Furthermore, non-homologous end joining repair mechanisms inevitably generate insertion-deletion mutations of varying lengths, presenting challenges for clonal analysis. Lastly, multiple double-strand breaks in synthetic barcodes can lead to complete loss of barcode information, thus reducing the accuracy of lineage tracing analyses.

Figure 3. CRISPR-Cas 9 barcodes. Combination of InDels generated by using RNA molecules transcribed by CRISPR sequences to guide the cleavage of Cas proteins at the corresponding sites. Image created using Adobe Illustrator 2023 (Adobe Inc.); image modified from (169). CRISPR, clustered regularly interspaced short palindromic repeats; InDel, insertion-deletion mutations.

Challenge

The CRISPR-Cas9 barcode technology typically inserts barcode sequences into the target genome through precise gene editing. However, despite the high targeting fidelity of the CRISPR system, off-target effects remain a concern, potentially leading to barcode insertion into non-target genomes and thereby increasing the data complexity (63). To minimize off-target effects, researchers have developed improved Cas9 variants, such as high-fidelity Cas9, to enhance specificity and have designed sgRNAs to further reduce off-target activity (64). Several barcode technologies utilizing CRISPR-Cas9 gene editing have been developed, addressing various challenges. However, issues such as insufficient recovery rates of both cells and barcodes persist. In certain cells, CRISPR-mediated barcode insertion may be incomplete or may fail to occur successfully due to variations in DNA repair mechanisms, thereby affecting subsequent lineage tracing and sequencing data (65). The use of more robust viral vectors or optimized CRISPR-Cas9 systems, in conjunction with multiple rounds of selection, can help ensure a higher barcode insertion efficiency. Furthermore, the insertion of barcodes into cells may interfere with the original gene expression, potentially leading to erroneous data in clonal and genetic analyses (65,66). Finally, the integration of barcode technologies with sequencing methods enables the reading of barcode sequences; however, due to off-target effects, precise localization of barcode insertion sites and robust computational support are essential for identifying and quantifying barcode data in complex cell populations (67). Researchers have exploited high-throughput sequencing and specialized bioinformatics tools, such as CRISPResso, to align and correct potential off-target insertion points, optimizing barcode recovery and analysis (68–71).

Integration barcodes

Integration barcodes, also referred to as genetic barcodes, are short sequences positioned at the ends of transposons or viral integration sites, typically at the 3′ end of a reporter gene (such as GFP) and the 5′ end of the adjacent 30-LTR. These barcodes facilitate the unique labeling of individual cells by integrating into the host genome, thereby enabling their differentiation and tracking in subsequent experimental analyses (72). Due to the high transduction efficiency of retroviral and lentiviral vectors, along with the availability of standardized experimental protocols, transposons or lentiviruses are commonly employed to integrate barcodes into the host genome. During barcode generation, the original marked cells pass genetic markers to progeny, leading to the formation of unique barcode sequences that are randomly generated (19). Although unique labeling of cells has been accomplished, current methodologies encounter challenges in efficiently reading barcode sequences during single-cell RNA sequencing (scRNA-seq) (3). This limitation poses challenges for the precise tracking and analysis of cells. In response to this challenge, integration barcode technology has undergone several improvements. Currently, barcodes are classified into two primary types based on their design: Random barcodes and semi-random barcodes. Random barcodes typically consist of four nucleotides (A, G, C, T) randomly arranged and denoted as (n) (72,73), as illustrated in Fig. 4. While these barcodes exhibit considerable diversity, theoretically accommodating a vast number of unique sequences, they also present several drawbacks. First, the amplification of these barcodes necessitates the PCR process, which may induce sequence biases or small insertions and deletions. Second, the complete randomness of the sequences, devoid of structural information, complicates subsequent decoding efforts. In contrast, semi-random barcodes feature a more structured design, often incorporating a combination of strong and weak base sequences. For example, a previous design used strong bases (such as S-C, G) and weak bases (such as w-A, T) (72,74). This approach presents significant advantages over fully random barcodes. Notably, semi-random barcodes exhibit a discernible sequence pattern, facilitating easier identification and retrieval during sequencing. Moreover, these barcodes can be directly integrated into vectors without the need for PCR amplification, thus circumventing the biases and insertions associated with PCR (7). The inherent structural regularity of semi-random barcodes not only maintains their efficiency but also enhances their stability and reliability. The primary advantage of integration barcode technology lies in its capacity to efficiently label and track cells, enabling researchers to precisely identify the genetic characteristics of individual cells within complex cellular populations (75). With ongoing advances in barcode design (from random to semi-random barcodes and even the combination of distinct barcode strategies) this technology holds considerable promise across various applications in genomics, drug screening and gene editing. However, challenges persist, including amplification errors and issues related to sequence read accuracy, which necessitate further optimization in future research endeavors.

Figure 4. Integrated barcodes. One or more integrated barcodes may serve as cellular identifiers to record clonal information. Image created using Adobe Illustrator 2023 (Adobe Inc.); Image modified from (3).

Challenge

Integration barcodes are typically introduced into the genomes of target cells through viral vectors or other delivery systems. However, due to the inherent randomness of insertion, the distribution of barcodes within cell populations is often uneven, which may affect the reliability of experimental results (76). The integration of barcodes is influenced by DNA repair mechanisms, potentially leading to the insertion of barcodes at random genomic loci. In single-cell and lineage tracking applications, this randomness may introduce biases or inaccuracies, as barcodes may integrate into non-representative genomic regions. For instance, insertions into non-coding regions or functionally irrelevant sites could affect gene expression or cellular pathways, thereby confounding the analysis (77,78). Furthermore, the insertion of barcodes may present a potential conflict with cellular function. Insertion at critical genomic loci could disrupt essential genes or regulatory elements, potentially interfering with cell survival, division, or differentiation, leading to unintended cellular changes (79,80). In cancer research, such insertions could alter tumor biology, obscure tumor heterogeneity and hinder accurate understanding of distinct tumor subtypes. Finally, the integration of barcodes relies on viral vectors to deliver the barcode sequences. However, the efficiency of these vectors can be unstable, leading to low integration rates or ineffective barcode delivery, thereby compromising experimental outcomes. To address this challenge, it is crucial to develop efficient and stable viral vector systems that ensure the effective and reproducible integration of barcodes into target cells (81,82). Additionally, the cost and complexity associated with the development and optimization of such systems may pose significant challenges, particularly in high-throughput experiments.

Droplet barcoding

Nucleic acid sequence barcodes are comprised of random, semi-random and non-random components. Random barcodes are generated by permuting various nucleotide elements and adjusting base positions and sequence lengths to achieve both diversity and uniqueness. The design of these random barcodes includes: i) PCR handles for reverse transcription and amplification; ii) cellular barcodes; iii) unique molecular identifiers (UMIs) for distinguishing PCR duplicates; and iv) a 30 bp oligo-dT sequence for capturing polyadenylated mRNA and initiating reverse transcription (36,83)(Fig. 5). Several barcode platforms, including Drop-Seq, InDrop and 10X Genomics, have been developed based on the principle of random barcodes (5,24,36,84). Prior to these, similar approaches, such as CEL-Seq and Smart-seq2 (85,86) employed random sequence barcodes for cell labeling. While CEL-Seq is highly effective in gene expression analysis, it has limited resolution when detecting complex alternative splicing events or analyzing full-length transcript isoforms (85). By contrast, methods such as Smart-seq offer enhanced efficiency, resulting in the development of two versions of SMART-seq. These versions address some of the limitations of other single-cell RNA sequencing technologies by capturing and analyzing complete transcript sequences, including both coding and non-coding regions. Despite the high sensitivity of SMART-Seq2, the method is still susceptible to amplification biases, which may compromise the accuracy of gene expression quantification, particularly for low-abundance genes (86). The random barcode design used by the Drop-Seq, InDrop and 10X Genomics platforms is largely similar, incorporating PCR handles, cellular barcodes, UMIs, poly-T cleavage regions and T7 promoters (24,36,87). However, there are subtle differences in materials, barcode capacity and post-demulsification reactions. Drop-Seq, developed by Macosko et al, enables the tagging of hundreds of cells for single-cell genomic analysis. Nucleic acid barcodes are integrated with polyacrylamide microspheres or particles, which are then linked to target cells and encapsulated into nanoliter-sized droplets via a microfluidic platform (36). Nevertheless, limitations remain, such as issues with cell capture efficiency, dropout events and limited resolution for rare cell types. Like Drop-Seq, InDrop employs microfluidic droplets to isolate single cells, but with distinct bead chemistry and barcode designs that provide higher sensitivity (24). InDrop can capture full-length transcripts, offering more detailed insights into gene expression, alternative splicing and transcript isoforms, making it particularly suitable for studying gene isoform diversity at the single-cell level (24). 10X Genomics offers high sensitivity, particularly in detecting low-abundance transcripts, thus enabling researchers to gain deeper insights into the gene expression dynamics of rare or low-expressed genes. This system uses unique molecular identifiers (UMIs) to improve quantification accuracy and minimize amplification biases. However, it captures only a portion of each transcript (typically the 3′ end), which may limit its ability to assess full-length transcript diversity or alternative splicing events comprehensively (88).

Figure 5. Droplet barcoding. The droplet barcode consists of four sections, including a ‘cell barcode’ section for labeling individual cells, which is looped through the pool to form a non-stop base sequence barcode; a UMI section for digital counting of mRNA transcripts; and a polyT sequence for capturing cellular mRNA. Image created using Adobe Illustrator 2023 (Adobe Inc.); image modified from (36). UMI, unique molecular identifier.

Challenge

In droplet-based barcode systems, individual cells are encapsulated within microdroplets and tagged with unique barcode sequences. However, in complex systems, cross-droplet contamination and barcode swapping are common issues, which may result in cross-contamination of barcodes between cells (89,90). To mitigate such concerns, single-cell separation technologies, such as microfluidic chips, should be employed during droplet generation to reduce barcode cross-contamination (91,92). Currently, the mRNA capture efficiency within droplets is low, at only 7% (93). Despite robust quantification through UMI filtering, this low efficiency hampers the reliable detection of genes with fewer than 20–50 transcripts per cell (90). This issue, which affects single-cell RNA sequencing, necessitates the implementation of improved cell lysis methods or optimized enzymatic reactions during library preparation (94). Additionally, as the barcodes in this system are random, they do not permit the association of a given barcode with a specific cell identification. Background noise, a critical factor contributing to the low signal-to-noise ratio in this technology, originates primarily from non-specific binding molecules, residual substances in solutions and cross-contamination between cells (95). This noise compromises data accuracy, particularly in high-throughput screening and cancer research, where it may obscure critical biological signals. Therefore, reducing non-specific binding agents and enhancing the purity of single-cell analysis is essential. The use of refined reagents, optimized culture conditions and more sensitive sequencing platforms can help mitigate background noise (95,96). To extract barcode sequence information, sequencing technologies are indispensable; however, current sequencing methods face limitations in detecting genes of extremely low abundance, especially at the single-cell level, where sequencing depth is often insufficient. Moreover, high-throughput sequencing technologies themselves are prone to errors. Consequently, sequencing workflows must be optimized, increased sequencing depth must be implemented and multiplex amplification strategies employed to improve the detection of low-abundance genes, reduce sequencing error rates and integrate error-correction algorithms to enhance data quality (69). The vast volume of data generated by barcodes necessitates the use of efficient algorithms and computational platforms. The integration of various computational models is essential for data consolidation, enabling a deeper understanding of the relationship between barcodes and gene expression, thereby enhancing data interpretation capabilities.

Native barcodes

Somatic mutation barcoding technology represents an advanced and increasingly indispensable method for tracking and analyzing somatic mutations at the single-cell level. This method is key for investigating genetic alterations linked to disease progression, aging and cellular evolution. The primary types of mutations detectable through somatic mutation barcoding include microsatellite instability (MSI), copy number variations (CNVs) (97), single nucleotide variants (SNVs) (98) and mitochondrial DNA mutations (Fig. 6). MSI involves alterations in the length of short, repetitive DNA sequences, typically indicative of defective DNA repair mechanisms and is commonly associated with cancer and various genetic disorders (99,100). CNVs refer to changes in the copy number of specific genomic regions, frequently linked to neurological disease, cancer and developmental disorders. SNVs represent single base-pair changes in the DNA sequence, which are pivotal for understanding disease susceptibility and genetic diversity. Although mitochondrial DNA mutations remain less well understood, they play a crucial role in various metabolic and neurodegenerative diseases, influencing cellular energy production and function. The application of somatic mutation barcoding technology in biomedical research and clinical diagnostics holds revolutionary potential (101–103). It enables researchers to capture the full complexity of somatic mutations at the single-cell level, offering an unprecedented approach for analyzing genetic diversity within tissue. In oncology, this technology is particularly valuable, as clonal evolution within tumors frequently dictates therapeutic resistance and metastasis. It also provides new opportunities for studying neurological diseases, as analyzing somatic mutations in neuronal populations may elucidate the mechanisms underlying diseases such as Alzheimer's disease, autism and schizophrenia. Furthermore, as somatic mutations accumulate in various tissues throughout aging, this technology can deepen our understanding of the aging process and age-related diseases (104,105). The advantages of somatic mutation barcoding technology are clear. Its high sensitivity facilitates the detection of rare mutations in heterogeneous cell populations and allows for longitudinal monitoring of mutations in response to treatment or environmental factors. By using unique molecular barcodes, researchers can distinguish mutations that arise in different cells, thereby facilitating the analysis of clonal evolution and cellular lineage (106). Additionally, the application of this technology provides novel insights into the relationship between somatic mutations and disease mechanisms, potentially leading to the development of personalized therapeutic strategies. Despite its numerous advantages, somatic mutation barcoding technology faces several challenges (107). Sequencing errors, amplification biases and cross-contamination can result in inaccuracies in mutation identification and quantification, particularly when addressing low-abundance mutations, which are prevalent in complex diseases such as cancer and neurological disorders. Moreover, distinguishing pathogenic mutations from benign ones remains a significant challenge, as not all mutations contribute to disease development. To overcome these limitations, advances in sequencing technologies, error-correction algorithms and bioinformatics tools are essential. Innovations, such as enhanced error-correction methods, longer read lengths and more sensitive sequencing platforms, are anticipated to improve (108–110).

Figure 6. Native barcodes. Image created using Adobe Illustrator 2023 (Adobe Inc.); image modified from (169). SNVs, single nucleotide variants; CNVs, copy number variations; InDel, insertion-deletion mutation.

Challenge

Native barcodes are derived from natural variations or mutations, rather than exogenous sequences, which inherently restrict their insertion sites and diversity. This limitation may lead to an insufficient number and diversity of barcodes, rendering them inadequate for large-scale tracking purposes. To mitigate this limitation, a combination of diverse natural variations and mutations, coupled with specific cell-labeling techniques, may be employed to augment barcode diversity and resolution. These barcodes generally rely on inherent cellular markers, such as specific gene expressions or mutations, which often exhibit uneven expression levels across different cell types and lack universal applicability. Moreover, certain cell types may lack sufficient natural markers, or the markers may be difficult to detect with adequate sensitivity. The range of natural markers is restricted and their levels are susceptible to fluctuations in cell state and environmental factors, resulting in inconsistent marker expression. To overcome this limitation, a combination of various natural markers or alternative exogenous labels is required to increase marker diversity, while high-sensitivity techniques, such as single-molecule RNA sequencing, must be employed to enhance detection sensitivity. Stability is also a critical factor to consider when utilizing natural barcodes, as they are often influenced by external factors such as cell division, environmental changes and epigenetic modifications. Throughout research, barcodes may undergo alterations during cell development, compromising their stability and impeding accurate lineage tracing. Furthermore, long-term stability may deteriorate over extended periods, rendering sustained tracking difficult. Barcode stability can be enhanced through the optimization of probe design and the use of advanced imaging techniques, such as fluorescence confocal or super-resolution microscopy, to improve spatial localization. Additionally, sensitivity remains a critical concern, especially in low-abundance or rare cell populations, where natural barcode concentrations may be too low for effective signal detection. Finally, the intrinsic nature of native barcodes leads to heightened background noise interference, complicating data interpretation. The application of data-cleaning techniques to reduce background noise and multi-channel data fusion methods can improve the accuracy of barcode detection.

Applications of barcodes

Single cell lineage tracing and fate mapping

Lineage tracing has emerged as a powerful tool for studying tissue development, homeostasis and disease, particularly when combined with experimental manipulation of signaling pathways that regulate cell fate decisions. The evolution of lineage tracing methods has advanced from traditional techniques, such as optical microscopy and dye-based cell labeling, to more sophisticated approaches, including recombinase-based systems, barcode technologies and natural gene mutations (111–113). This evolution has enabled the tracking of thousands of cells simultaneously, facilitated by advances in barcode analysis technologies, from PCR (114) to sequencing methods, microarrays (15) Sanger sequencing (115,116) high-throughput sequencing (7,31) and single-cell RNA sequencing (scRNA-seq) (117). Since these foundational studies, barcoding has been widely adopted for fate mapping in both multicellular organisms and single-cell microbial communities. In cell biology, this includes investigating clonal relationships among cell types (18), analyzing differentiation trajectories (118), identifying molecular signatures tied to clonal identity, resolving lineage relationships and dissecting cellular dynamics (119). At present, fate mapping extends to cancer research, including studies on cancer heterogeneity, clonal evolution, drug resistance and metastasis mechanisms. Barcodes in this context serve as both static and cumulative markers (118). Polylox barcodes (44), transposon-based barcodes (120) and CRISPR-based barcodes (38) are commonly used in lineage tracing, with applications across various systems. Integrated barcodes are often employed to mark progenitor cells, with this approach being widely used in hematopoietic stem cell research (44,121). By tracking individual cells and their progeny, these studies have linked specific transcriptional programs to various differentiation states, highlighting the dynamic relationship between transcriptional regulation and lineage commitment. Tracking the lineage history of cells is essential for addressing fundamental biological questions. For instance, in 2020, Bowling et al (55) used a CRISPR-based array system to track lineage in mice, uncovering intrinsic biases in fetal liver hematopoietic stem cell clonal activity and providing insights into stem cell responses to injury (Fig. 7A). In the same year, Weinreb et al (118) employed expressed DNA barcodes to perform clonal tracing of transcriptomes over time, applying this approach to investigate fate determination during hematopoiesis. Unlike CRISPR-based methods, this technique does not require lineage tree inference to establish sister-cell relationships; it exhibits remarkably low single-cell barcode dropout rates and eliminates the need for delivering multiple components. The study identified states that initiate fate potential and mapped them onto a continuous transcriptional landscape. (Fig. 7B). Lineage tracing has broader applications, including fate mapping in entire organisms (52), vertebrate brains (51), EMT in cancer and cancer xenograft metastasis (122). The application of barcode technologies in lineage tracing has also addressed several key challenges. For instance, barcodes in dynamic systems, such as cancer stem cells or regenerative tissues, offer valuable insights into cellular identity plasticity and lineage decisions. These methods enable the exploration of tissue heterogeneity, revealing how different cell populations within the same tissue can behave based on their lineage and gene expression profiles. Despite the advances, several challenges remain. High-throughput barcode generation and integration of multi-dimensional data (e.g., genomic, transcriptomic and epigenomic profiles) from lineage-traced cells still pose technical difficulties (7,123,124). Moreover, distinguishing between genetic mutations and transcriptional shifts in lineage tracing experiments requires careful control of experimental conditions and sophisticated computational methods. These challenges underscore the need for ongoing refinement of lineage tracing techniques and the development of more scalable and precise methods for high-dimensional data integration. lineage tracing, particularly when integrated with barcode technology, has transformed our understanding of cellular fate, lineage commitment and plasticity. Its applications in basic research and disease modeling, especially in cancer, provide a valuable framework for investigating the complexities of cellular behavior during development and disease (123,125).

Figure 7. Lineage tracing and fate mapping based on nucleic acid sequence barcodes. (A) The CRISPR-Cas9 mouse engineered cell lines are employed to simultaneously decode the lineage history and gene expression profiles of individual cells. Reprinted with permission from reference (55) (B) Polylox barcodes used for lineage tracing. Reprinted with permission from reference (118). CRISPR, clustered regularly interspaced short palindromic repeats.

Barcoding for single cell transcriptome

Single-cell transcriptomics is used to investigate the gene expression profiles of individual cells. This technology is pivotal in uncovering cellular heterogeneity, identifying novel cellular subpopulations and is indispensable for understanding the functional roles of diverse cell types within tissues. Among the primary methodologies used in single-cell transcriptomics is the droplet barcode technology (24). Prior to this, several methodologies, including SMART-seq, CEL-seq and Cyt-seq, were developed to study embryonic samples (126,127), the early developmental stages of the nematode Caenorhabditis elegans and the transcriptomes of rare cells within the hematopoietic system (128). Nevertheless, these approaches remain constrained by limitations such as low throughput (129), reduced specificity and sensitivity and PCR bias (130). The emergence of microfluidic technologies has substantially mitigated these challenges. Klein et al (24) integrated microfluidic technology with barcode beads to devise a calibration technique suitable for RNA sequencing of thousands of individual cells, widely applied in sequencing analyses of embryonic stem cells and their differentiated forms. This technology offers boundless scalability in cell processing, facilitating highly efficient cell capture, rapid sample collection and minimal technical noise (Fig. 8A) (24).

Figure 8. Nucleic acid sequence barcoding in single-cell transcriptomics. (A) Droplet barcoding combined with microfluidic technology is employed for single-cell transcriptomics. (B) Single-cell transcriptomes are extracted and processed using Drop-Seq technology. (C) The barcode partitioning process of soluble hydrogel beads for scRNA-seq and transposase-based scATAC-seq. (D) By encapsulating drugs with DNA barcode fragments, multiplexed drug screening is conducted within single-cell droplets. (E) The principles and performance of OAK in single-cell atlas analysis across various molecular modalities. Reprinted with permission from references (5,24,36,133,134). scRNA-seq, single-cell RNA sequencing; scATAC-seq, single-cell accessible chromatin analysis; OAK, overloading and dissociation.

In the same year, the development of the Drop-seq platform introduced a rapid strategy for analyzing thousands of individual cells. This platform analyzed the transcriptomes of 44,808 retinal cells from mice, constructing a molecular gene expression map encompassing both established retinal cell types and novel candidate subtypes (Fig. 8B) (36). Single-cell sequencing offers several advantages, including the precise identification of genetic differences between healthy and aberrant cells, as well as the construction of cellular maps for entire organisms, tissues and microbiomes. Nevertheless, existing chromatin extraction methods remain prohibitively expensive and there are currently no publicly available tools for the efficient processing of tens of thousands of cells. Moreover, although certain single-cell RNA sequencing tools are available at no cost, they often exhibit limitations in sensitivity and practicality. To address these issues, Florian et al developed HyDrop, an innovative open-source platform that surmounts these obstacles (5). This platform employs a novel type of barcode bead and refines existing microfluidic protocols using open-source reagents. These advancements offer users a more intuitive workflow while enhancing sequencing sensitivity without incurring additional costs. Leveraging this platform, thousands of single-cell RNA and open chromatin profiles were successfully extracted from the brains of mice and fruit flies (Fig. 8C) (5).

Traditional droplet-based microfluidic methods often generate a significant number of droplet-only cells, failing to fully exploit barcode beads and reagents. Although combinatorial indexing on microplates proves more effective for barcode usage, it is labor-intensive (84,131,132). To address this, researchers have developed a combinatorial indexing system based on overloading and dissociation (OAK) for ultra-high-throughput single-cell multi-omics analysis. Initial partitioning is performed using a droplet barcode system, followed by a second round of identity-tagged references to achieve combinatorial indexing. The OAK system boasts ultra-high throughput, broad compatibility, high sensitivity and a simplified workflow, making it a powerful tool for large-scale molecular analysis, particularly for rare cell populations (Fig. 8E) (133). In recent years, this technology has also been applied to drug screening. Mathur et al (134) proposed an expandable microfluidic workflow, named Combi-Seq, which uses transcriptomic changes as a readout for drug effects, screening hundreds of drug combinations within microliter-sized droplets (Fig. 8D). Applying Combi-Seq, they screened 420 drug combinations for their effects on the K562 cell transcriptome, using ~250 single-cell droplets per condition, successfully predicting synergistic and antagonistic drug interactions and their pathway activities.

Barcodes for high-throughput screening

High-throughput screening technology has become a crucial tool in the study of genetic variations and perturbations, driving advancements in our understanding of biological systems and enhancing the application of biotechnologies. The development of large-scale screenings and genetically diverse variant libraries has played a central role in advancing this field. At the genomic scale, RNA interference (135,136) and CRISPR-based technologies have been widely employed to screen pooled libraries (137), investigating cellular phenotypes such as cell viability and transcriptomes at single-cell resolution. Traditional screening methods are often constrained by low throughput, extended durations and high costs. Although progress in genome-scale screening has enabled the assessment of mammalian gene functions, these methods still rely on simple phenotype readouts that cannot differentiate between similar responses induced by distinct mechanisms. The advent of droplet-based single-cell RNA sequencing technology has introduced a powerful platform for high-throughput screening (138). When paired with CRISPR-based transcriptional interference platforms, this has facilitated the development of the ‘perturb-seq’ method. (Fig. 9A) (139). This approach is not only easily implementable but also highly scalable, allowing parallel screening at the single-cell level to generate rich phenotype data. Using this technology, studies on the unfolded protein response in mammals have elucidated how three sensors within the endoplasmic reticulum monitor various types of stress. Furthermore, CRISPR-based transcriptional interference has been combined with scRNA-seq to dissect immune circuits. Nucleic acid barcodes also play a pivotal role in phenotype analysis (140), particularly in situations requiring high-resolution imaging to localize cellular morphology, proteins and RNA. Emanuel et al (141) introduced an innovative method that combines genetic variant barcodes with multiplexed fluorescence in situ hybridization (FISH) technology (141). Each genetic variant is associated with a unique nucleic acid barcode, enabling identification via multiplexed FISH imaging. This method identifies brighter and more photostable YFAST fluorescent protein variants from a pool of 60,000 variants (141) (Fig. 9B). By integrating genome engineering, nucleic acid barcode technologies, high-resolution microscopic imaging and microfluidics, researchers can now conduct high temporal and spatial resolution studies of complex phenotypes in live cells, enabling precise control of single strains through single-cell observation. However, effective studies on the localization of phenotypes related to intracellular features or their correlation with specific genotypes are still limited (142). Xie et al (143) developed the Mosaic-seq platform, which integrates CRISPR barcodes with transcriptomic data and sgRNA regulators, enabling the engineering and analysis of multiple enhancer activities at the single-cell level (Fig. 9C). Lawson et al (144) expanded advanced live-cell microscopy techniques to bacterial strain libraries generated through pooling (Fig. 9D). They successfully confirmed the presence of plasmid-encoded strains using single-molecule fluorescence lifetime imaging, where the plasmid expressed sgRNA and employed dCas9 interference to suppress distinct genes within the Escherichia coli genome. This method effectively addresses the challenge of characterizing complex dynamic phenotypes within genetic libraries of various bacterial strains. For instance, it enables the screening of how changes in regulatory or coding sequences affect the timing, localization, or functionality of gene products, or how shifts in the expression of a set of genes influence the intracellular dynamics of a reporter gene.

Figure 9. High-throughput screening with nucleic acid sequence barcodes. (A) A multiplexed single-cell CRISPR screening platform is capable of systematically dissecting unfolded protein responses. (B) A high-throughput, image-based screening method utilizing large-scale multiplexed fluorescence in situ hybridization technology. (C) ‘Mosaic-seq,’ an index-based CRISPR sequencing approach for mosaic single-cell analysis, is a method that directly measures transcriptional changes as a phenotype resulting from enhancer repression at single-cell resolution. (D) In situ genotyping of pooled strains following the characterization of complex phenotypes. Reprinted with permission from references (139,141,143,169). CRISPR, clustered regularly interspaced short palindromic repeats.

Exploring cancer heterogeneity

Human cell lines have been pivotal in cancer research, driving discoveries of oncogenic mechanisms and therapeutic targets. Tumor heterogeneity, characterized by differences in tumor cell growth rate, invasive ability and regulation by drug-sensitive genes or molecules, results from genetic, epigenetic and environmental influences (145). Additionally, cancer heterogeneity depends on extrinsic factors such as time, the tumor microenvironment and interventions. Linking the cellular behavior of cancer cells with their molecular profiles can aid in proposing more effective methods to predict clinical outcomes and treat cancer patients. Cancer cells originate from individual malignant cells in the body and accumulate mutations in their DNA as they proliferate (146). Thus, tumors evolve into highly distinct cancer clones over time. With the advent of next-generation sequencing technologies (such as sc-RNA-seq), tumor heterogeneity and clonal frequency can be further investigated before and after treatment. However, carcinogenesis involves millions of cells and current technologies cannot fully cover such a large, heterogeneous cell population. The introduction of nucleic acid sequence barcoding provides a solution by labeling individual cells with a ‘unique’ barcode. This barcode can be used alongside sequencing to track progenitor cells, as it is heritably replicated and inherited by daughter cells during cell division (17,37,72,147) In recent years, The development of cancer is primarily driven by the processes of clonal selection and clonal expansion. However, the non-genetic mechanisms underlying intratumoral heterogeneity and malignant clonal heterogeneity remain largely poorly understood. Fennell et al (148) developed an innovative expression barcoding strategy, termed Single-cell Profiling and Lineage Tracing (SPLINTR). Using a mouse model of acute myeloid leukemia, the study employed isogenic clonal tracking to demonstrate, for the first time, that the clonal output characteristics of leukemia stem cells are critical determinants of their sensitivity to chemotherapy. By using SPLINTR technology, the team was able to simultaneously track the dynamic changes of thousands of malignant clones in a time-resolved manner, thereby uncovering principles of clonal adaptation that extend beyond the cancer genome. Notably, the study identified Slpi, a pivotal mediator of malignant clonal dominance that is exclusively detectable in vivo. This discovery offers new theoretical insights and research directions toward a deeper understanding of the mechanisms underlying cancer clonal evolution (Fig. 10A). In addition, Merino et al (20) aimed to uncover the subclonal relationship between primary triple-negative breast cancer and its metastasis. They employed cellular barcoding of two untreated triple-negative breast cancer patient-derived xenografts to track thousands of barcoded clones across primary tumors and metastases. Their findings suggest that cancer dissemination might relate to persistent shedding of material (Fig. 10C). Furthermore, Jin et al (149) employed an in vivo barcoding strategy to assess the metastatic potential of human cancer cell lines in mouse xenografts. They developed a first-generation metastasis map (MetMap) that elucidated organ-specific metastasis patterns (Fig. 10B). Further research into barcoding technology enables precise mapping of individual cell behavior at the cellular level and comprehensive analysis of intra-tumor heterogeneity mechanisms. Additionally, it facilitates molecular characterization of cancer clones, enhancing understanding of their characteristics. Barcoding technology holds substantial potential in advancing cancer research across cloning, metastasis, origin, development, treatment and drug selection.

Figure 10. Nucleic acid barcoding in cancer heterogeneity studies. (A) Splinter, an expression barcode strategy to trace isogenic clones in three clinically relevant mouse models of acute myeloid leukemia. (B) An in vivo barcode strategy capable of large-scale identification of the metastatic potential of human cancer cell lines in mouse xenografts. (C) Using cell barcode technology in two untreated PDXs of TNBC to track the spatiotemporal fate of thousands of barcode clones from the primary tumor and their metastases. Reprinted with permission from references (20,147,148). Splinter, single-cell analysis and lineage tracing; PDX, patient-derived xenografts; TNBC, triple-negative breast cancer.

Extracellular vesicle analysis

Extracellular vehicles (EVs), membrane-enclosed structures released by cells, are typically classified into several subgroups based on size or origin (150,151). Rich in molecular cargo, EVs play a pivotal role in cellular signaling and communication through the transport of bioactive substances and are involved in the signaling pathways associated with numerous diseases, including cancer (152–154). In recent years, EVs have emerged as valuable biomarkers for disease diagnosis and therapeutic monitoring (155). Small EVs (sEVs), derived from multicellular organisms, display a highly heterogeneous vesicular repertoire. Since the surface protein composition of EVs markedly influences their biological behavior (156), there is a growing need for high-throughput, single-EV analytical methods to more precisely characterize EV subpopulations (157). In 2021, Ko et al (158) introduced an antibody-based immunosequencing method that combines droplet microfluidics with nucleic acid sequence barcodes. This technique uses droplet microfluidics to partition EVs and barcodes, followed by sequencing of the barcode-antibody-DNA constructs, enabling multiplexed protein analysis from individual nanosized EVs to delineate their protein composition (Fig. 11A). This method offers the advantages of high sensitivity and throughput, with potential for further refinement to detect low-abundance proteins, such as enhancing the signal-to-noise ratio, reducing nonspecific binding and improving sequencing depth. In 2022, a novel alternative method based on droplet barcode sequencing (DBS) technology was proposed for the detection of individual sEVs (22) (Fig. 11B). This scalable method requires no specialized equipment or barcode gel beads and demonstrated that DBS-Pro facilitates the analysis of individual sEVs with a mixing rate of <2%. The approach was applied to analyze surface proteins of over 120,000 sEVs derived from non-small-cell lung cancer (NSCLC) cell lines and malignant pleural effusion samples from patients with NSCLC. This method enables the surface protein analysis of single EVs and the characterization of expanded sEV subtypes. However, due to the limited quantity of material available, this strategy may be impractical for clinical applications. Furthermore, a multiplexed approach based on droplet barcode technology has been developed, in which lipid nanoparticles carrying single-stranded DNA barcodes are fused with control particles, enabling the simultaneous tracking of EVs from different cell lines within a single in vivo experiment (159). In addition to barcode-based analysis of EV subpopulations, CRISPR-assisted single-barcode-based screening of regulators of sEV release has been employed (Fig. 11C) to identify key players in sEV release (160). This screening uses the interaction between gRNA and a deathCas9 fusion protein linked to sEV markers, combined with CRISPR-gRNA barcode encoding of sEVs (Fig. 11D) (159). Barcode quantification enables large-scale, parallel estimations of sEV release from individual cells. The barcode-sEV and CRISPR-based screening approach allows for genome-wide exploration of sEV release regulators, facilitating the identification of previously unrecognized regulators and the discovery of exosome/exosome-like properties, such as CD63+/CD9+ sEVs and the synchronous release of CD9+ sEVs during the cell cycle. In comparison with traditional one-to-one analyses, this multiplexed approach offers markedly higher throughput for identifying sEV release regulators, independent of the effects of sEV abundance in culture media. Despite significant advances in sequencing-based single-EV protein analysis, several challenges persist that hinder broader clinical application. These challenges include the isolation of individual EVs, efficient capture of low-abundance proteins, clinical feasibility and the need for comprehensive and accurate insights into the biological functions of EVs. The integration of nucleic acid sequence barcodes with sequencing-based single-EV protein analysis has revolutionized the study of extracellular vesicle biology. With its high sensitivity, scalability and ability to provide multiparametric analyses, this method holds substantial potential for advancing our understanding of EVs as biomarkers for disease diagnosis, prognosis and treatment. However, technical obstacles remain, particularly in the isolation of single EVs, optimization of sequencing technologies and the interpretation of complex datasets. Continued innovation in these areas will pave the way for the routine clinical application of single-EV analysis, opening new frontiers for precision medicine.

Figure 11. EV profiling. (A) An antibody-based immunosequencing method that enables multiplexed measurement of protein molecules from single nanometer-sized electric vehicles. (B) The principle of droplet barcode sequencing analysis for protein analysis, used for the quantification of surface proteins on individual vesicles. (C) CIBER screening utilizes the interaction between gRNA and death Cas9 fused with sEV markers, along with barcode-encoded sEVs via CRISPR-gRNA. Barcode quantification allows for the large-scale, parallel estimation of the amount of sEVs released from each individual cell. (D) EVs from 16 cell lines were used to generate a unique barcode-hEV library, through fusion with LNPs carrying single-stranded DNA barcodes as controls. Reprinted with permission from references (22,157–159,) gRNA, guide RNA; sEV, small extracellular vehicles; CRISPR, clustered regularly interspaced short palindromic repeats; hEV, hybridized EV; LNPs, lipid nanoparticles.

Conclusions and perspectives

In the 1950s, the elucidation of the double helix structure of DNA opened a new chapter in life science. The introduction of nucleic acid barcoding has had a similarly profound effect on the biomedical field. Given the role of DNA in genetic replication, nucleic acid barcoding serves as a robust tool for cell labeling and tracking. The diversity and uniqueness of nucleic acid barcodes imbue them with high throughput, resolution and accuracy. Recent technological advances have further enabled barcoding to offer a comprehensive, multi-modal understanding of tissue cell composition across temporal and spatial dimensions (161). Looking ahead, nucleic acid barcoding will extend beyond high-throughput screening, single-cell lineage tracking and the characterization of tumor heterogeneity. It holds promise for clinical cancer diagnosis, adjuvant therapy, exosome detection and subpopulation classification based on surface markers (18,19). Additionally, nucleic acid sequence barcoding is increasingly applied in areas such as active protease sensing (162), multiplex nucleic acid detection platforms (163), bacterial single-cell RNA sequencing for toxin regulation prediction (164) and DNA sequence-based optical-free single-cell spatial proteomics (165) to infer the spatial distribution of cell surface proteins. After nearly two decades of development, nucleic acid sequence barcoding offers a comprehensive, multimodal understanding of tissues and whole organisms across temporal and spatial dimensions. However, several challenges continue to impede its broader application. First, the diversity of barcodes does not yet comprehensively cover all cell types in a number of organs and biases introduced during barcoding remain unaddressed. Second, obtaining barcode sequence readouts requires library creation and target sequencing, which do not provide direct results. Lastly, despite advances in sequencing technology, inaccuracies, information loss and high error rates persist, limiting the reliability of results.

New nucleic acid sequence barcoding libraries and applications are reported annually, facilitating increasingly detailed and comprehensive analyses of cell and cancer heterogeneity. This technology has filled gaps left by other sequencing methods, offering broader insights into biological development, cloning and disease research. As the technology continues to evolve, the integration of optical barcoding with nucleic acid sequence barcoding holds promise for spatially localizing cells and analyzing gene expression profiles in unprecedented detail (115,166–168). Addressing the limitations of barcode diversity, sequencing accuracy and data integration will require continued innovation. The development of barcode libraries that can comprehensively cover a broader spectrum of cell types, coupled with improvements in sequencing technology and error correction, will markedly enhance the power of nucleic acid barcoding. In addition, expanding the application of multi-modal data, including genomic, transcriptomic, proteomic and spatial information, will provide a deeper understanding of the mechanisms driving disease. Machine learning and AI-driven approaches will also play a key role in making sense of this complex data and enabling more accurate predictions about disease progression and treatment response. The integration of optical barcoding techniques into nucleic acid sequence barcoding holds significant potential for spatially resolving cellular dynamics within tissues, opening new avenues for research and clinical applications.

Interdisciplinary collaborations are essential to address these challenges. Bioinformaticians, computational biologists, molecular biologists and clinicians must work together to refine the technology, improve error correction and optimize data analysis tools. Engineers and physicists can contribute by developing next-generation sequencing technologies and advancing optical barcoding methods. Collaborations with clinicians and oncologists are crucial for translating these advances into clinical settings, particularly for the detection of tumor heterogeneity and monitoring therapeutic responses. Such collaborations will undoubtedly accelerate the translation of nucleic acid barcoding technology into practical, clinically relevant applications, transforming our approach to disease diagnosis, treatment and prevention.

Acknowledgements

Not applicable.

Funding

The present study was supported by the Scientific and Technological Innovation Major Base of Guangxi (grant no. 2022-36-Z05) and the Guangxi Science and Technology Major Program (grant no. AA24263028).

Availability of data and materials

Not applicable.

Authors' contributions

YW was responsible for conceptualization, investigation, visualization and writing the original draft. FL was responsible for conceptualization, funding acquisition, validation, visualization, writing, reviewing and editing. Data authentication is not applicable. All authors read and approved the final manuscript.

Ethics approval and consent to participate

Not applicable.

Patient consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.

References