With the rapid rise in gene-editing technology, pluripotent stem cells (PSCs) and their derived organoids have increasingly broader and practical applications in regenerative medicine. Gene-editing technologies, from large-scale nucleic acid endonucleases to CRISPR, have ignited a global research and development boom with significant implications in regenerative medicine. The development of regenerative medicine technologies, regardless of whether it is PSCs or gene editing, is consistently met with controversy. Are the tools for rewriting the code of life a boon to humanity or a Pandora’s box? These technologies raise concerns regarding ethical issues, unexpected mutations, viral infection, etc. These concerns remain even as new treatments emerge. However, the potential negatives cannot obscure the virtues of PSC gene editing, which have, and will continue to, benefit mankind at an unprecedented rate. Here, we briefly introduce current gene-editing technology and its application in PSCs and their derived organoids, while addressing ethical concerns and safety risks and discussing the latest progress in PSC gene editing. Gene editing in PSCs creates visualized
Pluripotent stem cells (PSCs), such as embryonic stem cells (ESCs) and induced pluripotent stem cells (iPSCs), are extensively used and considered to be viable cellular therapies against complicated and malignant diseases, like leukemia [
Gene editing is broadly applied in disease modeling [
The applications and potential expansion of gene editing of PSCs and their derived organoids are endless. Here, we systematically analyze and compare several gene-editing methodologies and provide examples of how gene editing has been used in the treatment of diseases, construction of disease models, and exploration of disease mechanisms. Combined with the experiences and ongoing work in our lab, we have expounded the perspectives as well as opportunities associated with gene editing in PSCs and their derived organoids.
PSCs are self-renewing with infinite proliferation and multipotency. In 2006, Shinya Yamanaka was the first Japanese scholar to use a viral vector to introduce four transcription factors (Oct4, Sox2, Klf4, and c-Myc) into somatic cells to obtain iPSCs, which revolutionized the field of regenerative medicine [
Genome editing tools can be divided into four types that are described here according to the timeline of their discovery from the earliest to the most recent: meganucleases (MegNs, also termed homing endonucleases), zinc finger nucleases (ZFNs), transcription activator-like effector nucleases (TALENs), and CRISPR RNA-guided nucleases. The characteristics of each of these four editing tools relative to existing genetic technologies, as well as their advantages and disadvantages, are summarized in Table
Characteristics of current gene editing technologies and their advantages and limitation.
Identifying patterns | Cleavage domain | Recognition length | Identification conditions | Minimum identification unit | Accuracy | Molecular weight size of editing tools | Operational difficulty | Off-target level | Cytotoxicity | Advantages | Limitations | |
---|---|---|---|---|---|---|---|---|---|---|---|---|
MegNs | Binds specific DNA through protein-DNA interactions | 4 bp | Double-stranded DNA sequences of 12 to 40 base pairs | Monomer, target DNA | Indeterminate | +++ | 200-400 aa | +++ | + | + | Higher specificity | Limited variety; difficult to retrofit |
ZFN | Binds specific DNA through protein-DNA interactions | 5-7 bp | 9-18 bp per ZFN | Dimers, 3 bp units of target DNA | 3 bp | ++ | 500-1300 aa | ++ | ++ | ++ | Mature platform; more efficient than homologous recombination | High off-target rate; low specificity; design dependent on upstream and downstream sequences; only for in vitro operations |
TALEN | Binds specific DNA through protein-DNA interactions | 5-7 bp | 14-20 bp per TALEN | Dimer, transcription activator-like effector or transcription activating effector nuclease 5 | 1 bp | ++ | 900-1100 aa | ++ | ++ | +++ | Unrestricted target sites; easier design than ZFN; higher specificity | Cumbersome module assembly; requires large sequencing effort; high cost |
CRISPR | Binding of specific DNA through base complementary pairing and protein-DNA interactions | 0 bp | 20 bp | Monomer, 3 | 1 bp | ++ | 1300-1500 aa | + | + | + | High rate of gene modification; diverse gene regulation; allows simultaneous knockout of multiple target loci; precise targeting inexpensive | No PAM in the pretarget region cannot be cut; transfection difficulties |
PAM: protospacer adjacent motif.
Meganucleases (MegNs) rely on the length of the target sequence and the structure of the DNA contact surface to specifically, accurately, and effectively identify the target. The mechanism of DNA recognition by MegNs involves binding patterns of protein side chains and nucleotide bases [
Zinc finger nucleases are constructed by fusing a DNA cleavage domain, like the Type II restriction enzyme FokI, to a zinc finger protein (ZFP) [
Transcription activator-like effector nucleases (TALENs) evolved from transcription activator-like (TAL) effectors, which are transcription activators that have peculiar properties of DNA recognition. The monomeric protein chains of TALENs bind DNA in a right-handed spiral manner, without inducing any bend or other substantial structural distortion. Each base is recognized by a highly conserved sequence of typically 33–35 amino acids. Based on the one-to-one corresponding relationship [
CRISPR, clustered regularly interspaced short palindromic repeats, is named for the conserved primitive sequence structure of the bacteria and archaea immune defense system [
Among gene-targeting nucleases, MegNs are the most difficult to synthesize. However, they exhibit small sizes, single-chain structures, and high specificity. TALENs are good at targeting specific individual DNA base pairs without affecting the activity or binding force of the nucleases. Only a pair of TALENs can accurately bind to a double-strand break, which may result in a low probability of off-target effects. Engineering and redesigning specific recognition of DNA-binding proteins are a challenging area of research and development. Proteins and DNA have different molecular interface compositions, and their complex relationships include directional hydrogen bonds, electrostatic contacts, ordered solvent molecules, and bound counterions, making protein–DNA interactions elusive and unpredictable. The CRISPR/Cas9 system is the most operable tool because of its RNA-DNA recognition characteristics, which avoids complex protein engineering.
The combination of stem cell and gene-editing technologies has led to new innovations in the field of medicine, opening up a new wave of personalized and precision medicine. The creation of organoid disease models through genetic engineering and gene-editing technologies has led to the elucidation of underlying mechanisms of major diseases, with clinically translatable applications. Table
Landmarks and trends of gene editing in life and medical sciences.
Editing methods | Target cells | Targeted genes | Virus transfection | Animal models | Points | Year | |
---|---|---|---|---|---|---|---|
MegN | KI | mESCs | Villin locus | Yes | / | Induction of gene-targeting and homologous recombination events | 1998 [ |
M | 293T | RAG1 locus | No | / | Targeting endogenous genes; low targeting efficiency; with cytotoxicity | 2009 [ | |
KI | 293T | Yes | / | Delivering meganucleases into cells in a transient and dose-controlled manner; low targeting efficiency; with cytotoxicity | 2011 [ | ||
ZFN | KI/GFP | hESCs | OCT4 locus、AAVS1 locus | Yes | m | Gene targeting in hESCs | 2009 [ |
M | hESCs | Genomic | No | / | Genome editing in hESCs; off-target detection needs to be improved; targeting efficiency needs to be enhanced | 2011 [ | |
KO | hiPSCs | LRRK2 (sigma) | No | m | Parkinson’s pathogenesis; patient-derived iPSCs; low targeting efficiency; with cytotoxicity | 2013 [ | |
M | hiPSCs | MAPT | Yes | / | Designed mutation iPSCs; FTD pathogenesis; targeting efficiency needs to be enhanced | 2018 [ | |
KO | FRT cells | CFTR | No | r | Disease targets; designed KO model | 2020 [ | |
TALEN | KI/GFP | hPSCs | OCT4 locus | No | / | Genetic engineering for hPSCs; targeting efficiency like ZFN | 2011 [ |
KO | hiPSCs | TNNT2, LMNA/C, TBX5, MYH7, ANKRD1NKX2.5 | Yes | / | Human-based KO cell model | 2017 [ | |
KI/GFP | hiPSCs | AAVS1 locus | No | / | 3D organoid models; GFP; mechanistic studies | 2019 [ | |
CRISPR | KO | hiPSCs | AAVS1 safe harbor locus | No | / | TetO inducible system; feasibility and reversibility of CRISPRi; high off-target efficiency | 2016 [ |
M | hiPSC | FBN1 | No | / | Vascular models, human iPSCs; pathogenesis of MFS | 2017 [ | |
KI/GFP | hESCs | gRNAs made from the lentiGuide-puro construct | Yes | / | A genome-scale screening; hESCs; impaired differentiation | 2019 [ | |
KO | hPSCs | NRL | No | / | A 3D organoid model; disease pathogenesis; high targeting efficiency | 2021 [ |
ESCs: embryonic stem cells; FTD: frontotemporal dementia; FRT: Fischer rat thyroid; h: human; iPSCs: induced pluripotent stem cells; KI: knockin; KO: knockout; m: mouse; M: mutation; MFS: Marfan syndrome; PSCs: ESCs and iPSCs; r: rat.
PSC-derived organoid models can be used to visually trace the fate of cells through development or growth, by constructing knockin reporter genes for specific target genes. To study vocal dysphonia, caused by vocal fold (VF) disorder, a hiPSC-derived VF model with a GFP reporter was transfected via TALEN to simulate the development of VF epithelial cells in utero. This system consisted of a 3D
To elucidate physiological mechanisms, gene editing and PSC differentiation models may be a perfect combination. Through CRISPR/Cas9-based genome editing technology, key segmentation-clock gene expression showed phase changes in the hPSC-derived presomitic mesoderm. This provided insights into the human segmentation clock related to diseases associated with human axial skeletogenesis [
The ability to selectively modify genes is important to identify the role of genes in specific pathological changes. In one study, genetically modified hPSCs were generated by CRISPR/Cas9 editing revealing that noncoding gene variants have undeniable effects on
To clarify the mechanism of action of known mutations, gene-editing interventions were carried out on pathogenic genes in patient-derived iPSCs or organoids. Mutations in
To explore unknown mutations and their effects, comparing PSC models with and without mutations may be effective. For instance, comparing CRISPR/Cas9-based gene editing in hPSC-derived neurons and isogenic controls, it was determined that the internal mechanism of neuronal network dysregulation was due to the
To screen for unknown genes responsible for diseases, creating mutations and overexpressing or inhibiting gene expression in PSCs and organoid models could help clarify and define key genes of interest. Using CRISPR/Cas9 in hPSCs with an E50K mutation in the optineurin (
Gene editing in PSCs through knockout (KO) or knockin (KI) genes enables observation of phenotypic changes and, potentially, the identification of disease targets for clinical research and therapy. Table
Current challenges of gene editing in different diseases.
Disease names | Related genes | Editing technologies | Model types | Clinical trials | Challenge points and limitation | Years | |
---|---|---|---|---|---|---|---|
Respiratory disorders | CF | CFTR | CRISPR | Organoids | \ | Proof of concept only, gene editing off-target effects; needs further evaluation for safety | 2013 [ |
TALEN | Cells | \ | Delivery efficiency needs to be improved; targeting accuracy needs to be improved | 2019 [ | |||
CRISPR | Patient-derived cells | \ | Difficulty of in vivo delivery, genetically corrected airway stem cell transplantation and recovery of in vivo mucus cilia transport | 2021 [ | |||
NSCLC | PD-1 | CRISPR | \ | Phase I (first) | Underexpansion and low response rate of T cells after gene editing; small study sample | 2020 [ | |
Circulatory disorders | HC | Protein PCSK9 | MegNs | Macaques | \ | Off-target effects, with cytotoxicity, immunogenicity to be overcome | 2018 [ |
NS-associated HCM | RAF1 | CRISPR | Patient-derived cells | \ | RAF1 lacks a nuclear localization sequence (NLS), its translocation mechanism is unknown, and the molecular mechanism of the disease needs to be further explored | 2019 [ | |
HC | Ldlr | CRISPR | Mouse | \ | Genome editing efficiency to be improved and off-target effects to be overcome | 2020 [ | |
NS-associated HCM | LZTR1 | CRISPR | Patient-derived cells | \ | Proof of concept only, needs in vivo evaluation, patient-specific iPSC-CM model is still immature and needs to be improved | 2020 [ | |
LDS | TGFBR1 | CRISPR | Patient-derived cells | \ | Needs further proof from in vivo experiments, off-target effects | 2021 [ | |
Infectious diseases | HIV | CCR5 | ZFN | \ | Yes | A serious adverse event was associated with the infusion of ZFN-modified autologous CD4 T cells, with off-target safety issues to be overcome | 2014 [ |
ZFN | Mouse | \ | Reduced proliferation of editorial cells transplanted in vivo, delivery efficiency and targeting accuracy need to be improved | 2013 [ | |||
TALEN | Cells | \ | Delivery efficiency and targeting accuracy need to be improved | 2015 [ | |||
CRISPR | Mouse | \ | Safety issues to be further assessed | 2017 [ | |||
CRISPR | \ | Yes | Off-target efficiency needs to be improved, targeting accuracy needs to be improved, and generalizability needs to be further assessed | 2019 [ | |||
CRISPR | Patient-derived cells | \ | Off-target efficiency needs to be improved, and targeting accuracy needs to be improved | 2020 [ | |||
Hematologic disorders | TDT & SCD | BCL11A | CRISPR | \ | Yes | No comprehensive genomic analysis of clinical samples and the generalizability of the results needs to be further determined | 2021 [ |
TDT & SCD | HPFH5 | CRISPR | Cells | \ | Off-target effects to be overcome and safety to be improved | 2016 [ | |
SCD | HBB | CRISPR | Mouse | \ | The off-target efficiency needs to be reduced, and more sensitive off-target analysis methods are needed | 2019 [ | |
SCD | HBB | CRISPR | Humanized mouse | \ | Delivery methods to be optimized and delivery efficiency to be improved | 2021 [ | |
ALL | CD52 | TALEN | \ | Yes | Immunogenicity needs to be further reduced; safety needs to be further tested; small sample size | 2017 [ | |
MM | TRAC、CD52 | TALEN | Mouse | \ | Delivery efficiency needs to be improved, and long-term safety issues need to be further studied | 2019 [ | |
Neurological disorders | FXS | FMR1 | CRISPR | Mouse | \ | The off-target efficiency needs to be reduced, more sensitive off-target analysis methods are needed, and safety issues need to be further tested | 2018 [ |
AD | TREM2 | CRISPR | Humanized SCD mouse | \ | Further analysis of the mechanism of action is needed to find effective therapeutic targets for disease treatment | 2020 [ | |
CD | ASPA | TALEN | Mouse | \ | Proof of concept only, how to achieve sustained efficacy remains to be addressed, and the issue of safety still needs to be improved | 2020 [ | |
Ophthalmology | XLRP | RP2 | CRISPR | Organoids | \ | Retinal-like organs are still immature and need further improvement | 2020 [ |
LCA10 | CEP290 | CRISPR | Mouse | Yes | Impact of individual differences on safety of off-target effect delivery, durability of efficacy to be further assessed | 2019 [ |
AD: Alzheimer’s disease; ALL: acute lymphocytic leukemia; CD: Canavan disease; CF: cystic fibrosis; FXS: fragile X syndrome; HC: hypercholesterolemia; HCM: hypertrophic cardiomyopathy; iPSC-CMs: iPSC-derived cardiomyocytes; LDS: Loeys-Dietz syndrome; LCA10: Leber congenital amaurosis type 10; MM: multiple myeloma; NS: Noonan syndrome; NSCLC: non-small-cell lung cancer; SCD: sickle cell disease; TDT: transfusion-dependent
Immunological rejection is common following organ transplantation. A study based on hiPSC gene editing found that the
Disease models were constructed by PSC differentiation and gene editing. For fragile X syndrome (FXS), an inherited intellectual disability in males, FMR1 was reactivated after the heterochromatin status switched, by targeting demethylation of the CGG expansion using dCas9-Tet1/single guide RNA (sgRNA) in FXS iPSCs. This suggested potential therapeutic strategies for FXS [
There are many refractory diseases without effective treatment, some of which are fatal. Although the best therapy cannot be confirmed at once, potential targets can be identified through gene editing carried out on hiPSC models. Both long-QT syndrome and short-QT syndrome are fatal inherited arrhythmogenic syndromes, which can cause apopsychia and death. A human ether-a-go-go-related gene-deficient CM model [
Sometimes, as the basis of a definitive etiology, gene-editing treatment methods can be manipulated in PSC-derived models to identify rescue treatments. PSC-derived alveolar epithelial type 2 cells (AEC2s) provide a platform for disease modeling, exhibit self-renewal capacity, and display additional AEC2 functional capacities. In iAEC2s generated from a child with severe lung disease carrying an
Some mutations can be rescued by gene editing, and gene-rescued PSCs can differentiate into mature cells and be transplanted into animal models, improving and possibly curing the animal. TWIK-related spinal cord K+ channel (TRESK) is implicated in nociception and pain disorders; a CRISPR/Cas9-corrected TRESK function-related mutation, F139WfsX2, showed a reversal in neuronal excitability. This suggests TRESK activators may be a promising therapeutic approach to pain and migraine [
Ethical issues have always been unavoidable in the context of gene editing [
Although viral vectors are known to have high delivery efficiency, they can be double-edged swords, with continuous expression of CRISPR/Cas9 nuclease and gRNA causing off-target mutagenesis and immunogenicity. Off-target risk has always been a major concern for genetic treatment; however, through the use of PSC culture and differentiation technology, cells that are deemed to be safe can potentially be used for clinical applications. At the same time, more studies that are committed to safe and efficient gene-editing strategies are needed, similar to those described below.
A nanovesicle-based delivery system, NanoMEDIC, delivers large molecules, such as ribonucleoprotein; the nanovesicles are cleared within 3 days [
Cytosine and adenine base editors (CBEs and ABEs) are powerful tools for single-base modification. However, editor components, DNA repair proteins, and local sequence context interact, resulting in unpredictable editing outcomes. Researchers who focused on illuminating base editing have provided refined and novel insights, which may improve the precision of base editing [
Nonhomologous end-joining (NHEJ), microhomology-mediated end-joining (MMEJ), and homology-directed repair (HDR) are the three main types of cellular DNA repair machinery. To determine the most efficient HDR strategy, researchers introduced different forms of donor DNA and observed that editing with a 400 bp dsDNA repair template increased the efficiency of repair [
Prime editing is a genome editing technology combining Cas9-nickase and reverse transcriptase with greater precision than Cas9-mediated HDR. When performed, nearly no off-target effects are observed; thus, it has potential in future clinical applications to safely repair human monogenic diseases [
Although various gene-editing methods have emerged, their broad and direct use in clinical settings remains a long road ahead.
The rapid advancement of genome editing technologies, from MegNs to CRISPR, has improved the operability, efficiency, and safety of gene editing. The combination of gene editing and stem cell technologies has advanced the research and development of the life and medical sciences. Through knockin and knockout technologies, human genetic and pathogenic mechanisms of disease can be better explored, and gene expression and disease progression can be traced. Drug development can also be accelerated, contributing to the advancement of personalized precision gene therapy for inherited diseases.
The existing gene-editing technologies each have their particular characteristics and advantages, but all have some corresponding challenges. Although MegNs have high specificity and low cytotoxicity, they are difficult to manipulate, limited in variety, and time-consuming, and it is expensive to design sequence-specific enzymes. ZFNs, although a relatively mature platform and more efficient than homologous recombination, are highly off-target and cytotoxic, have low specificity, are sequence-dependent upstream and downstream, and are only suitable for
Although current preclinical trials have demonstrated initial safety and efficacy of gene editing, existing studies have also shown that the immunogenicity and cytotoxicity of these vectors are of concern. Improving the accuracy of detecting and then reducing off-target effects remain a challenge. Only when these problems are solved can gene-editing technology be better applied in the clinical setting.
The authors declare that there is no conflict of interest regarding the publication of this paper.
LPL and YWZ conceived and designed the study. HZ and YW drafted and revised the manuscript. YWZ, LPL, and HZ contributed to reviewing and discussing the manuscript. All authors approved the final manuscript. YML supplied resources and materials. HZ and YW contributed equally to this work as co-first authors. YML and YWZ are senior authors and contributed equally to this work as cocorresponding authors. Hang Zhou and Yun Wang contributed equally to this work.
This research was supported partly by the National Natural Science Foundation of China (82070638 and 81770621), JSPS KAKENHI (18H02866), and the Natural Science Foundation of Jiangsu Province (BK20180281).