High-throughput editing boosts the construction of next-generation microbial cell factories

Abstract

Traditional metabolic engineering faces numerous challenges in constructing microbial cell factories, such as inefficient gene editing, time-consuming and labor-intensive screening processes, and the complexity of multi-gene optimization. High-throughput (HTP) genome editing technology accelerates the optimization of microbial metabolic pathways by precisely and efficiently modifying multiple genes. Furthermore, HTP genome editing technology enables the rapid screening and modification of key enzymes or regulatory factors across multiple metabolic pathways, facilitating the analysis of complex regulatory mechanisms. These advantages make it a key enabling tool for both top-down analysis and bottom-up assembly of metabolic pathways. This review summarizes the mechanisms and applications of various HTP genome editing and discusses the development of HTP strategies that accelerate the design-build phase for microbial cell factories (MCFs). These advancements promised to significantly enhance the performance of MCFs and drive the next generation of sustainable, bio-based production technologies.

1. Introduction

As a core component of green biomanufacturing, microbial cell factories (MCFs) represent advanced biotechnological platforms that leverage genetically engineered microorganisms to efficiently convert simple substrates into complex target molecules, such as biofuels, pharmaceuticals, enzymes, and specialty chemicals, through engineered metabolic pathways [1,2]. Achieving commercially viable fermentation processes with MCFs necessitates extensive reprogramming of cellular metabolism. Traditionally, metabolic engineering has relied on trial-and-error approaches predicated on understanding microbial metabolic networks and strain physiology [3,4]. However, developing workable MCFs from concept design demands 50–300 person-years and hundreds of millions of dollars in investment [5], rendering this approach cost-prohibitive for many industrial applications. Fortunately, metabolic engineering has greatly benefited from synthetic biology tools since the late 1990s and early 21st century, enabling a decisive shift from traditional methodologies. Advanced tools, including genome editing systems, whole-genome assembly techniques, genome-scale metabolic modeling, high-throughput phenotypic screening, and other novel technologies, have driven this shift.

HTP genome editing enables the precise and combinatorial modification of multiple genomic loci, allowing for the large-scale manipulation of microbial genomes and the optimization of microbial strains for industrial-scale production [[6], [7], [8], [9]]. The primary aim of applying HTP genome editing in MCF research is to enhance the performance and capabilities of these cell factories. By systematically editing and optimizing genes involved in key metabolic pathways, it is possible to accelerate the production of biofuels, pharmaceuticals, and other essential chemicals [10,11]. These genetic alterations may include the insertion or deletion of genes, as well as the fine-tuning of gene expression, which can lead to improved metabolic flux, higher product yields, and better resistance to environmental stressors such as high temperatures or toxic by-products [12,13].

This review systematically classified HTP genome editing technologies into three categories based on their core mechanisms and provided a comprehensive overview of HTP genome editing in MCFs, highlighting the potential of HTP genome editing to accelerate strain development and improve product yields. Furthermore, this review outlines key strategies for enhancing genome editing efficiency and scalability. It would guide the development of next-generation cell factories and drive significant advancements in industrial biotechnology, making bio-based production more efficient and sustainable.

2. HTP genome editing

Genome editing has evolved from early random mutagenesis techniques to highly precise methods for gene modification. Initially, random mutations were induced using chemical mutagenesis and UV radiation [14,15]. In the 1980s, the discovery of restriction enzymes and DNA ligases paved the way for gene cloning and early genetic manipulation [16]. The early 2000s saw the development of Zinc Finger Nucleases (ZFNs) [17] and Transcription Activator-Like Effector Nucleases (TALENs) [18], which offered greater precision in targeting DNA but were complex to design and less efficient. In 2012, the breakthrough of the CRISPR/Cas9 system revolutionized genome editing by enabling precise DNA cutting through RNA-guided targeting, significantly simplifying the process [19]. Additionally, the rise of high-throughput genome editing enabled multiplexed editing, driving applications in metabolic engineering. Nevertheless, HTP genome editing remains a nascent technology, dominated by exploratory research despite its transformative potential (Fig. 1). Genome editing relies on three main core mechanisms: transposition systems, DNA recombination, and programmable endonucleases for precise gene insertions, deletions, replacements, or functional alterations of regulatory elements. Here, the prime genome editing technologies are systematically summarized, with comparative analysis of their characteristics and applications (Table 1).

Fig. 1.

A historical timeline of high-throughput gene editing technologies. The recombinant DNA and transposons were discovered successively in the 1980s. In 2002, Zhang developed genome shuffling technology, marking a breakthrough that advanced genomic library construction. In 2012, CRISPR technology revolutionized the field of gene editing. In 2016, base editing expanded the boundaries of precision genome editing. Then, various genomic editing technologies had been developed to realize more precise and effective editing applied in the construction of MCFs.

Table 1.

Comparison of HTP genome editing technologies.

Technology	Editing Range	Editing Types	Editing Efficiency	Editing Precision	Use Cases
Transposase-mediated genome editing	Random/semi-random insertions	Insertion, large insertion libraries	High for insertion	Random, possible insertion at unknown loci	Construction of genome-scale insert libraries and large-scale screening
HR mediated genome editing	Any locus	Insertion, deletion, replacement	Medium-high (depends on host repair mechanisms, donor DNA design)	High precision, minimal random effects	Precision gene replacement and large-fragment integration
SSR mediated genome editing	Requires pre-installed "landing pad/recombination site" in genome	Insertion, replacement, cassette exchange, deletion, inversion, rearrangement	Medium-high (if landing pad is installed)	Precise, low off-target effects	Large-scale path integration and multi-round insertion/replacement
CRISPR/Cas mediated genome editing	Any locus with PAM	Insertion, Deletion, Replacement, Knock-out, large deletions/insertion, rearrangement	High (especially for knockouts/indels via NHEJ)	Medium-high precision; high off-target effects	Extremely versatile, large-scale knockout/knockin
BE mediated genome editing	Limited to loci with suitable PAM and editing window	Single-base substitution	High (especially for point mutations)	High precision; Medium-high off-target effects	Fine-tuning: amino acid changes, gene emxpression tuning, point mutation repairs, protein engineering
PE mediated genome editing	Any locus	All 12 base conversions, small insertions/deletions	Medium-high (efficiency depends on pegRNA design, cell repair mechanisms, delivery)	Very high precision	Simultaneously requiring high precision and multiple mutations

2.1. Transposition-mediated HTP genome editing

Transpositional recombination describes the mobilization and genomic integration of DNA segments, facilitated by transposable elements or transposons [20]. Transposase identifies two specific reverse repeats and transfers both the repeats and intervening DNA to random locations in the genome. Transposition enables random mutations for genome function studies, like identifying phenotype-related genes and determining the minimum genome. It also facilitates the stable integration of heterologous cassettes into host chromosomes for biosynthetic gene expression.

Transposon insertion sequencing (Tn-Seq) generates high-quality genome-wide knockout libraries by randomly integrating transposons into the host genome (Fig. 2) [21,22]. This method was often used to dissect essential genes in many organisms [23,24]. An elegant study by Goodall et al. [25] to dissect crucial genes in E. coli, employing a transposon library of 901,383 unique transposon insertions. This approach uncovered novel features missed, such as short essential gene regions, orientation effects, and precise genomic identification. It also identified new metabolic pathways for bioprocessing, including assessing aromatic catabolism genes using RB-TnSeq. [26]. In the RB-TnSeq library, a unique 20-nucleotide barcode tracks transposon locations across various growth conditions. [27]. Esteban et al. [28] utilized a series of broad-host-range mini-Tn5 vectors (pBAMD) to integrate exogenous genes into the chromosomes of various Gram-negative bacteria, achieving complete integration of the heterologous poly(3-hydroxybutyrate) metabolic pathway.

Fig. 2.

The mechanisms of Tn-Seq. A gene disruption library is created by transposing the mini-transposon into bacterial genomic DNA in vitro and transforming it into a bacterial pool. DNA is isolated and digested with MmeI. PCR is then performed to amplify a 160-bp sequence containing bacterial-specific DNA, flanked by Illumina-specific sequences for sequencing. The 20-bp reads are mapped to the genome. Then, insertion counts are used to calculate fitness for each sample.

Transposase-mediated HTP editing offers advantages, including high insertion efficiency and ease of operation, which is especially useful in gene function studies and large-scale screening. Simple and efficient transposons enable large-scale gene editing in prokaryotes, such as E. coli. In contrast, complex genomes require optimized transposon tools and techniques in S. cerevisiae [24]. While non-model organisms present greater challenges, potentially necessitating the development of specific transposon systems and strategies tailored to their unique genomic characteristics. However, the random nature of transposon insertion can lead to unpredictable outcomes, including potential disruption of important genes or regulatory regions, making it less suitable than recombination-mediated gene editing technologies in applications requiring precise editing.

2.2. Recombination-mediated HTP genome editing

DNA recombination refers to the exchange, rearrangement, and transfer of nucleotide sequences within a single DNA molecule or between different DNA molecules, resulting in the reassembly of existing genetic information. Recombination can be broadly divided into two major categories: homologous recombination (HR) and site-specific recombination (SSR). These distinct forms of DNA recombination play essential roles in living systems by generating genetic diversity while simultaneously contributing to the maintenance of genomic stability within populations [29,30]. They are highly suitable for tasks requiring precise gene editing, fragment insertions, or directed mutations.

2.2.1. HTP genome editing based on HR

HR occurs between genes or DNA sequences within the same DNA molecule, using homologous DNA as a template for repair or recombination [31]. With continuous research on the mechanisms of HR in microbial cells, the field has continuously evolved and improved, offering increasingly powerful tools for probing and manipulating genome function [32,33]. Several recombination systems had been extensively studied, including the λ-Red system and the Rac prophage-derived RecET system [34,35].

Currently, most studies creating genomic mutants using recombineering still rely on the Red-based system [36,37]. Farzadfard et al. [38] developed the high-efficiency synthetic cellular recorders integrating biological events (HiSCRIBE) (Fig. 3A). HiSCRIBE enables high-efficiency, scarless bacterial genome editing via retron-generated ssDNA and Redβ recombination. It allows genome-wide, dynamic modifications without cis-elements or DNA breaks, achieving over 99 % mutation uniformity while reducing cytotoxicity and chromosomal instability [39]. Recombineering advancements, especially multiplex automated genome engineering (MAGE), had greatly enhanced genome editing (Fig. 3B). MAGE uses λ Red-mediated ssDNA annealing to E. coli replication forks, achieving up to 25 % efficiency and generating billions of variants via automated cycles [40]. Co-selection (CoS) MAGE (Fig. 3C) was proposed to efficiently introduce larger ssDNA fragments into target regions on the chromosome to improve the insertion efficiency [41]. Furthermore, for simultaneous perturbation of thousands of genomic sites, microarray-oligonucleotide MO-MAGE (Fig. 3D) amplified oligos from microarray chips for direct use in MAGE [42]. Efficient mutation incorporation requires Red enzyme expression and mismatch repair inactivation. To overcome this, Nyerges et al. [43] developed a generalized method for bacterial genome editing, named pORTMAGE (Fig. 3E). It provides genome-wide editing with high efficiency and low off-target rates, producing diverse mutants after only 24 cycles. Moreover, yeast oligo-mediated genome engineering (YOGE) [44] and eukaryotic MAGE (eMAGE) [45] were developed in the model yeast S. cerevisiae (Fig. 3F and G). eMAGE in yeast enabled simultaneous transformation of up to 12 oligos, introducing ∼60 precise mutations. Iterative modifications generated >10⁵ combinatorial variants, allowing targeted diversification of genes, promoters, and terminators, effectively optimizing pathways such as heterologous carotenoid production.

Fig. 3.

Recombination-mediated HTP genetic engineering tools. A. High synthetic cellular recorders integrating biological events (HiSCRIBE). B. Multiplex automated genome engineering (MAGE) automation. C. Co-selection (CoS) MAGE strategy. D. Microarray Oligonucleotide (MO)-MAGE. E. pORTMAGE: a set of plasmids expressing the λ Red recombinase enzymes, as well as the dominant-negative mutator allele of MutL. F. Yeast oligo-mediated genome engineering (YOGE). G. Eukaryotic MAGE (eMAGE).

HR allows for the insertion, replacement, or deletion of target fragments at specific genomic locations with high precision, which has been widely applied to various microorganisms, including yeast, E. coli, and actinomycetes. However, HR requires an exogenous template, which may increase complexity during large-scale genome engineering. SSR is not dependent on homologous sequences, enabling rapid and efficient gene insertion or deletion without relying on homologous fragments.

2.2.2. HTP genome editing based on SSR

SSR happens at specific DNA sequences, often mediated by specific enzymes or proteins. Site-specific recombinases mediate the excision or integration of DNA fragments flanked by specific sites [46]. Site-specific recombinases include two families, the tyrosine recombinase family (such as Cre, FLP, or Dre) and the serine recombinase family (such as φC31) [47]. SSR-based HTP genome editing has been applied sparingly in prokaryotes. Recombinase-assisted genome engineering (RAGE) based on Cre recombinase and mutually exclusive lox sites was successfully used to introduce a 34 kb alginate metabolism pathway into the E. coli [48].

In eukaryotes, synthetic chromosome rearrangement and modification by loxP-mediated evolution (SCRaMbLE) showed good HTP editing prospects [49] (Fig. 4A and B). Using this technique, Shen et al. [50] generated a variety of genetic changes in yeast synthetic chromosome arm synlXR. Xiong et al. [51] created a library of haploid and diploid synthetic strains with chromosome end variations using hundreds of loxPsym sites (Fig. 4C). Diploid strains generally outperform adapted haploid strains in growth fitness and exhibit a lower rate of SCRaMbLE-induced mortality [52]. L-SCRaMbLE, a light-controlled Cre recombinase for S. cerevisiae, offered tight modulation of recombinase activity, with up to a 179-fold induction upon exposure to red light [53]. Various applications of SCRaMbLE, such as in the ring synthetic yeast chromosome V [54] and the in vitro DNA library-building technique [55] (Fig. 4D), had demonstrated the generation of complex genomic variations, enhanced production of compounds like prodeoxyviolacein and β-carotene, and improved tolerance to environmental stresses, highlighting the potential of recombinase-based systems in genetic engineering and evolution. [56,57]. Furthermore, site-specific recombinase (SSR) tools, such as Cre/loxP, Flp/FRT, Dre/rox, and PhiC31 integrase systems, were developed to meet the need for precise and efficient genomic modifications. Another key application was recombinase-mediated cassette exchange (RMCE), which allowed for the rapid swapping of expression cassettes, streamlining strain development and optimization [58,59].

Fig. 4.

Illustrates the process of synthetic chromosome rearrangement and modification by SCRaMbLE. A. The mechanisms of SCRaMbLE involve the induction of recombination between two loxP sites by site-specific recombinase systems. B. illustrates the Multiplex SCRaMbLE Iterative Cycling (MuSIC) method. C. The objectives of SCRaMbLE encompass different numbers and forms of synthetic chromosomes. D. The methods of SCRaMbLE include haploid or diploid SCRaMbLE and in vivo or in vitro SCRaMbLE.

Relying on the precise homology between cargo and desired genomic segments, HR-induced genomic editing is suitable for applications requiring scarless editing, site-specific insertion, or precise sequence replacement. However, it is more widely used in model organisms, such as B. subtilis and S. cerevisiae, with high HR efficiency. Its application in non-model organisms with low HR efficiency remains challenging. SSR systems exhibit high specificity, controllability, and precision, supporting complex genomic operations. But in bacteria, SSR systems like Cre/loxP/φC31 are suitable for fragment insertion, while not commonly used for constructing large editing libraries. Specific sites must be introduced into the bacterial chromosome, and the natural repair mechanisms of prokaryotes do not couple naturally with SSR. The development and application of CRISPR technology have overcome the limitations of recombinant-mediated genome editing, such as low efficiency, complex operations, and insufficient precision.

2.3. Nucleic acid endonuclease-mediated HTP genome editing

Nucleic acid endonuclease-mediated genome editing involves the use of nucleases (e.g., CRISPR/Cas9, ZFNs, TALENs) to introduce double-strand breaks (DSBs) at specific genomic loci [60]. These breaks are then repaired by the cell using either non-homologous end joining (NHEJ) or homology-directed repair (HDR) pathways, enabling precise modifications, such as insertions, deletions, or replacements. Multiple genomic loci can be targeted simultaneously using multiplexed nucleases, enabling efficient and scalable genetic modifications across various species. It is particularly well-suited for tasks requiring rapid construction of variant libraries, multi-site editing, and exploratory gene function screening.

2.3.1. HTP genome editing based on CRISPR/cas

Compared to ZFNs and TALENs, which require altering the protein sequences to achieve recognition of specific sites, CRISPR/Cas9 offers capabilities with simplified design and operation, lower costs, high editing efficiency, and flexible targeting. Recently, CRISPR/Cas emerged as a key enabling technology for genome editing in prokaryotes [61] and eukaryotes [62]. Genome editing is achieved by Cas proteins, including cleavage guided by a guide RNA (gRNA) to cleave the targeted genomic locus (Fig. 5A) [63]. Compared to traditional genome editing methods such as recombination or random insertion, CRISPR/Cas significantly enhances efficiency and shortens construction cycles, particularly in the development of cell factories requiring multi-gene, multi-module, and multi-round iterative editing. CRISPR/Cas systems are continually being discovered in large numbers of species with wide phylogenetic diversity, and among which, Cas9 from Streptococcus pyogenes [64] and Cas12a from Acidaminococcus sp. [65] remain the two most widely used Cas proteins.

Fig. 5.

Nucleic acid endonuclease-mediated HTP genome editing. A, CRISPR/Cas9 genome editing system, in which the Cas9 protein induces a double-stranded DNA break at the target loci guided by gRNA. This cleavage can result in base substitution, insertion, or deletion of new DNA through HDR or NHEJ mechanisms. B, base editing, where the nCas9 protein fused with cytidine deaminase (CBE) or adenine deaminase (ABE) is used to perform targeted base substitutions on the target DNA sequence. C, prime editing consists of nCas9 and the MMLV reverse transcriptase (RT), which allows precise base substitutions, insertions, or deletions editing.

CRISPR-enabled trackable genome engineering (CREATE) connected each gRNA to homologous repair cassettes that both edit genome loci and serve as barcodes for tracking genotype-phenotype relationships in E. coli [66]. Each CREATE cassette contained a targeting gRNA and homology arm for mutation introduction, enabling editing of 10⁴–10⁵ loci with a 98 % identity threshold in evolution experiments. The CREATE approach enableed design-driven genome engineering, identifying novel mutations conferring resistance to stressors and antibiotics in E. coli [[67], [68], [69]]. Similarly, CRISPR/Cas9-assisted genome-scale genetic modification methods that enable rapid engineering of S. cerevisiae with precise and traceable edits were reported [70,71]. A mutant library of 43,020 guide sequences targeting 25 regulatory genes was created in S. cerevisiae, achieving 98 % editing efficiency. [71]. Specific mutants enhanced tolerance to growth inhibitors and doubled ethanol production, with iCREATE enabling multiplex mutation library creation [72].

CRISPR-mediated genome editing technology has emerged as the most powerful gene editing tool due to its high efficiency, precision, simplicity, and broad applicability. It enables efficient and precise gene knockout, insertion, and regulation, while also facilitating complex metabolic engineering designs, thereby advancing the efficient construction and application of microbial cell factories. With the advancement of DBTL (Design-Build-Test-Learn), CRISPR/Cas editing technology has become the core tool in the “build” phase. The CRISPR/Cas system relies on the characteristics of the host cell for its efficiency and precision in genome editing, including factors such as genome structure, repair mechanisms, transformation efficiency, and cell cycle. Bacteria such as E. coli possess a simple-structured genome, high transformation efficiency, and relatively high editing efficiency, making them suitable for large-scale genome editing and screening using the CRISPR/Cas system. Although S. cerevisiae has a complex genome, it can also efficiently conduct gene function studies and genetic modification by optimizing the CRISPR/Cas system. For non-model microorganisms with complex genomic structures, transformation efficiency may be low, and they may possess additional defense mechanisms, resulting in suboptimal high-throughput gene editing efficiency. Nevertheless, the CRISPR/Cas9 system performs editing by DSBs, which may result in insertions or deletions during repair or cause off-target effects at other similar sequence locations.

2.3.2. HTP genome editing based on base editing and prime editing

Base editing (BE) is an emerging genome editing technique that allows direct nucleobase transitions in genomic DNA without the need for DSBs, DNA donor templates, or HR [73]. It involves the use of catalytically impaired Cas nucleases fused with nucleobase deaminases [74]. This technology is beneficial for modifying target codons into desired outcomes, such as replacing codons with premature stop codons to deactivate target genes (Fig. 5B) [75,76]. Researchers have developed various CRISPR-based base editing systems for Streptomyces [77]. CRISPR/Cas9 cytidine deaminase fusion system (CRISPR/Cas9-Rapobec1) for A. niger [78], and Target activation-induced cytidine deaminase (Target-AID) for Y. lipolytica [79] also expand the application of base editors in microbial studies. Furthermore, Liang et al. [80] developed AGBE comprises dual deaminases to introduce indels as well as four different base conversions (C-to-G, C-to-T, C-to-A, and A-to-G) simultaneously. AGBE enhances gene-editing diversity, creating saturated mutant populations for functional gene verification.

Wang et al. [81] developed MACBETH, a multiplex base editing technique using dCas9 and AID in C. glutamicum. Automated gene silencing of 94 transcription factors using MACBETH achieved >50 % editing efficiency, contributing to the identification of novel genes conferring resistance to 5-fluoroorotate, oxidative stress, or furfural. CRISPR-guided base editors can also serve as a replacement tool for streamlining the laborious process of RBS/promoter engineering in bacteria, offering means to fine-tune production pathway activity [82]. Base Editor-Targeted and Template-free Expression Regulation (BETTER) was a technique that allowed simultaneous diversification of multigene expression [74]. BETTER eliminated library generation, transformation, and donor DNA integration, enabling in-situ creation of diverse genetic combinations, including ribosome binding sites and promoters. The library complexity for glycerol catabolism in B. subtilis and xylose catabolism in C. glutamicum reached 1 × 10¹¹ and 6 × 10¹⁰, respectively.

Base editors have shown tremendous potential in cell factories and systems biology, offering precise, efficient, and less error-prone genome editing tools. However, bystander and off-target editing constrain the development of base editors. Prime editing (PE) is another CRISPR-based genome editing technology that would not induce DSB and could realize 12 base substitutions and small insertions and deletions, giving much greater advantages for bacteria activity manipulation and cell factory construction (Fig. 5C) [83]. PE has been used in E. coli to introduce precise point mutations that improve its tolerance to high concentrations of organic acids, enhancing the yield of biofuels during fermentation [84]. PE has also been applied in S. cerevisiae to efficiently modify the genome, optimizing the production of bio-based chemicals by correcting key metabolic genes involved in sugar consumption and fermentation pathways [85,86]. Several recent studies have demonstrated its potential in high-throughput applications, particularly for microbial cell factories and metabolic engineering. Tong et al. [87] offered a CRISPR-Prime Editing-based toolkit for genome editing in E. coli, including base substitution, insertion, and deletion. With a second guide RNA, the toolkit could achieve multiplexed editing, albeit with low efficiency. It has been indicated that while PE has shown significant promise in high-throughput genetic screening, the transition to large-scale microbial applications requires further optimization in terms of editing efficiency, functional screening, and long-term stability in industrial fermentation systems.

2.4. HTP genome editing by the combination of CRISPR and HR

Although the CRISPR/Cas system can efficiently achieve extensive genome editing, it may induce off-target effects and random insertions. Homologous recombination enables precise gene replacement, but its efficiency is low in certain microorganisms. The combination of these two mechanisms compensates for the shortcomings of either technology alone, enabling precise, efficient, and multifunctional genome editing. The CRISPR/Cas system precisely targets specific gene loci. At the same time, HR provides homologous DNA templates to accurately replace target genes or insert specific sequences, which enables precise gene replacement, point mutations, insertions, or large-scale segmental replacements, minimizing unwanted indels. Various genome editing platforms combining CRISPR with HR have been developed, offering high efficiency, diversity, and scalability, which have been applied to multiple model microorganisms and non-model strains, providing versatile tools for reprogramming complex metabolic networks [[88], [89], [90]].

The EASY-edit (Engineered assembly of synthetic operons for targeted editing) enabled efficient editing of different components by utilizing a set of optimized guide RNAs and single- or double-stranded DNA repair templates carrying relatively short homologous arms. It allowed replacement of open reading frames and regulatory elements with low constraints and at high-throughput by targeting a specific locus on the chromosome [90]. In eukaryotic organisms, CHASE (CRISPR/Cas9-and homology-directed repair-assisted saturation editing) was a saturation gene editing method capable of achieving high-saturation codon exchange across extended genomic regions. By applying CHASE to perform large-scale editing of global transcription factor genes in S. cerevisiae, previously unreported genetic variations affecting industrially relevant microbial traits were discovered [91]. These approaches are increasingly being integrated with deep learning and structure prediction to finely tune enzyme activity or tolerance, holding significant promise for the engineering of key enzymes in synthetic pathways [92].

For long fragments integration, ReaL-MGE (Recombineering and Linear CRISPR/Cas9 assisted Multiplex Genome Engineering) enabled precise manipulation of numerous large DNA sequences, as demonstrated by the simultaneous insertion of multiple kilobase-scale sequences into E. coli, Schlegelella brevitalea and Pseudomonas putida genomes without any off-target errors [93]. With ReaL-MGE, intracellular malonyl-CoA production was increased by 13–26-fold or more in three bacterial strains, demonstrating its potential for designing complex product pathways. In the non-model strain Ogataea (Hansenula) polymorpha, rME (recombination machinery engineering) was developed, which increases HR efficiency from 20 %–30 % to 60 %–70 %. The rME system has been applied for homologous integration of large fragments and in vivo assembly of multiple fragments, which enables the production of fatty alcohols in O. polymorpha [94].

The CRISPR-HR genome editing technology significantly reduces the random insertion or deletion mutations commonly observed when relying on the NHEJ repair mechanism. It represents a highly promising method for precise gene repair, offering substantial advantages in applications requiring high-fidelity editing, such as gene knock-in and gene replacement. However, this technology has advanced rapidly in model organisms like E. coli and S. cerevisiae due to its reliance on exogenous recombination systems, yet its application in non-model organisms remains limited.

3. Conclusions and future perspectives

Metabolic engineering has evolved from the modification of a few genes in a close metabolic network to complex designs involving dozens of genes. HTP genome editing technologies provide more efficient and precise tools for constructing cellular factories, accelerating the optimization of microbial metabolic pathways. This review highlights the recent advances in HTP genome editing technologies, systematically summarizing the high-throughput genome editing techniques based on transposons, recombination, and nucleic acid endonuclease, along with their applications in elucidating metabolic mechanisms, constructing complex metabolic pathways, and enhancing tolerance to stress conditions. Although genome editing technologies offer significant advantages in the construction of cellular factories, unclear editing objectives, inherent limitations of editing tools, editing sequence dependencies, and mismatched screening methods are pressing challenges that restrict the development of highly efficient MCFs (Fig. 6).

Fig. 6.

A comprehensive strategy demonstrated applications of deep learning in enzyme modification and metabolic pathway design, multi-mechanism combination genome editing technologies, PAM-free genome editing techniques, and robotic platforms with high-throughput microbial reactor systems offer insights into how various stages of design, build, learn, and test can be leveraged in the context of metabolic engineering and synthetic biology.

3.1. Intelligent target design significantly enhances editing efficiency

Genome-wide HTP editing requires the construction of large-scale libraries when targets are undefined. While this approach enables researchers to explore gene functions and metabolic pathways comprehensively, it also presents significant challenges in terms of high costs, substantial workload, and data analysis [95,96]. The rational or semi-rational design of editing targets offers the potential to overcome the challenges associated with efficiently engineering complex phenotypes, which often require the complete reprogramming of a wide range of genes and traits. With the advancement of data science and artificial intelligence (AI), machine learning can automate target design by analyzing massive genomic datasets to predict optimal target sites and editing strategies. Segler et al. trained three deep neural networks based on Monte Carlo tree search using 12.4 million reaction rules from the Reaxys chemistry database to discover retrosynthetic pathways for small molecules [97]. Similarly, a Monte Carlo Tree Search method has been extended to predict synthetic metabolic pathways within biological systems, enabling systematic pathway design for metabolic engineering [98]. Experimental data obtained can then be analyzed and fed back into predictive models for iterative updates, optimizing design outcomes to guide subsequent experiments [99,100]. This closed-loop optimization enables the establishment of more precise gene-phenotype association models and provides more efficient design solutions for multi-gene, multi-pathway optimization.

3.2. Multi-mechanism-mediated genome editing enhances editing precision and efficiency

As described above, single-mechanism-mediated HTP genome editing suffers from inherent limitations, including the inability to precisely control insertion sites, low editing efficiency, and the presence of off-target effects. With the continuous advancement of genome editing technologies, multi-mechanism-mediated genome editing combines the strengths of different gene editing tools to achieve more efficient and precise editing. Alison Fanton et al. [101] engineered large serine recombinases (LSRs) through directed evolution and fused them with dCas9, achieving integration efficiencies of 53 % and genome-wide specificity of 97 %, while enabling efficient integration of large DNA fragments up to 12 kb. The CRISPR-associated transposase (CAST) system, which comprised transposition proteins and CRISPR/dCas, could insert sequences as long as 10 kb into bacterial genomes without inducing DSBs by finding the designated insertion site via the gRNA [102]. In the future, combining CAST with multi-fragment assembly technology could enable efficient integration of multi-gene pathways and synergistic optimization of large-scale synthetic metabolic networks. Multi-fragment assembly technologies, such as Gibson Assembly, Golden Gate, and ligase cycling reaction, can efficiently splice multiple gene modules together, each of which can be designed as a standardized fragment with different RBS or promoters [103]. The multi-site integration function of CAST is responsible for inserting multiple genes or gene modules into multiple target sites within the genome.

3.3. Unconstrained editing expands editing scope and flexibility

Identification of a short sequence context next to target sites through DNA-targeting Cas enzymes is a vital step in CRISPR/Cas systems. The protospacer adjacent motif (PAM) sequence, though it restricts the broad editing scope, ensures the editing specificity for genome editing applications. Various near-PAMless variants have been obtained by modifying the PAM recognition domain through protein engineering, such as SpCas9 [104] and Cas12i^HiFi [105]. These protein variants overcome the limitations of previous methods restricted to editing at specific sites, enabling the targeting of arbitrary sites. Future efforts will require further balancing the trade-off between targeting range and editing specificity [106].

3.4. Novel HTP screening enables large-scale complex phenotypic analysis

HTP editing can rapidly generate gene diversity libraries, but efficiently screening for beneficial phenotypes remains an urgent challenge. Biosensors [[107], [108], [109]], robotic platforms [110,111], and HTP microbioreactor systems [112,113] are effective strategies for achieving HTP screening. Biosensors can convert the concentration of intracellular metabolites or environmental factors into detectable signals, which have been used for large-scale screening of organic acids, amino acids, aromatic compounds, lipids, and other substances [114]. Future work requires continued optimization of the sensitivity, specificity, and response range of biosensors based on transcription factors and responsive promoters to achieve precise detection of specific compounds or environmental factors. In addition, robotic platforms and HTP microbioreactor systems are capable of identifying potentially promising production clones for further experiments. For example, droplet microfluidic technology allows for the parallel processing of thousands to millions of experiments on a single microchip through automated equipment that integrates detection, sorting, and analysis. This enables rapid screening to identify microbial strains with optimal metabolic performance. In the future, the cost of manufacturing high-quality chips should continue to decrease. Additionally, by integrating AI and machine learning models to optimize automated droplet formation and enhance image analysis, HTP screening can be achieved at low cost, with high efficiency and stable core performance. To enable the screening of complex phenotypes in the future, it will be essential to develop rapid assessment techniques, such as those based on growth rate or fluorescence measurements. These methods will be critical for facilitating the phenotypic analysis of large strain libraries containing billions of variants.

In summary, HTP genome editing technologies have revolutionized the field of metabolic engineering. With the continuous evolution of new methods such as CRISPR-based tools, transposase-mediated systems, and recombination technologies, they offer scalable solutions for complex biological problems. Through interdisciplinary collaboration spanning engineering, bioinformatics, and experimental validation, a new generation of genome editing tools will emerge, enabling more efficient construction of next-generation MCFs.

CRediT authorship contribution statement

Xi Sun: Writing – review & editing, Writing – original draft, Investigation. Yangyang Zheng: Writing – review & editing, Writing – original draft, Investigation. Jiancheng Luo: Writing – original draft, Investigation. Zhouxiao Geng: Investigation. Ziyao Wang: Writing – original draft. Jianxiao Zhao: Investigation. Tao Chen: Writing – review & editing, Conceptualization. Zhiwen Wang: Writing – review & editing, Project administration, Funding acquisition, Conceptualization.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.