Base Editors Correct Artemis Mutations, Offering New Hope for SCID Gene Therapy

A rare and devastating inherited immune disorder may have a new path toward treatment, thanks to a proof-of-concept study demonstrating that CRISPR base editors can correct disease-causing mutations in the Artemis gene outside the body. Artemis-deficient severe combined immunodeficiency, known as ART-SCID, leaves infants with almost no functional T and B cells, and current treatments fall short for many patients. The new research, published in Advanced Biotechnology, shows that cytidine and adenine base editors can repair three pathogenic Artemis variants in cell models with clinically meaningful efficiency and no detectable off-target editing at predicted sites, partially restoring the enzyme’s nuclease activity in the process.

Severe combined immunodeficiency is a group of inherited disorders characterized by impaired lymphocyte development, and mutations in more than twenty genes have been implicated in the condition, including CD3D, IL7R, IL2RG, RAG1, RAG2, PRKDC, LIG4, and DCLRE1C, the gene that encodes Artemis. Affected infants typically present with chronic diarrhea, oral thrush, skin rashes, and recurrent, refractory infections. Without early diagnosis and intervention, the disease can lead to life-threatening complications and a high risk of mortality. Since the 1960s, hematopoietic cell transplantation has been the primary treatment, and thousands of infants have been treated worldwide, but the Artemis-deficient form of the disease poses particular challenges that have driven researchers to seek alternatives.

Artemis is an endonuclease that resolves DNA hairpin intermediates during V(D)J recombination, the genome-scrambling process that generates the diverse repertoire of antigen receptors on lymphocytes. After the RAG1/2 recombinase complex cleaves recombination signal sequences, Artemis, activated by DNA-PKcs, nicks the closed DNA hairpin coding ends, creating a 3-prime overhang that primes subsequent junctional diversification. The enzyme also acts as a caretaker of genomic integrity through its participation in the non-homologous end-joining DNA repair pathway. When Artemis is absent, T and B cell development is arrested, producing a radiation-sensitive form of SCID. The clinical spectrum is highly heterogeneous: null mutations generally cause classical SCID with complete arrest of lymphocyte development, while hypomorphic alleles retaining partial nuclease activity can give rise to atypical phenotypes, including Omenn syndrome and leaky SCID.

The transplantation route itself is complicated by the very biology of the disease. Patients with radiosensitive SCID are systemically sensitive to alkylating agents, the standard myeloablative drugs used before transplantation to eradicate recipient hematopoietic stem cells. Clinical studies have also indicated that ART-SCID patients remain hypersensitive to ionizing radiation and suffer more infections long after receiving an allogeneic transplant. Even with an HLA-matched sibling donor, T and B cell reconstitution after transplantation is not as efficient in ART-SCID patients as it is in other SCID genotypes. Gene delivery strategies using lentiviral vectors carrying wild-type Artemis cDNA have been explored, and replacing a strong viral promoter with the endogenous human Artemis promoter achieved robust immune recovery in affected infants after in vivo transduction of hematopoietic stem cells. However, the lingering safety concern of insertional mutagenesis, in which semi-randomly integrating vectors may activate oncogenes, persists. Meanwhile, CRISPR-Cas9 mediated homology directed repair, another gene correction approach, remains difficult to apply clinically because quiescent hematopoietic stem cells are intrinsically resistant to HDR and the error-prone NHEJ pathway competes for repair, resulting in low efficiency.

Against this backdrop, base editors have emerged as a compelling alternative. Cytosine base editors employ cytidine deaminases such as rat APOBEC1 to convert C•G base pairs to T•A via a C-to-U intermediate, while adenine base editors use evolved adenine deaminases to convert A•T to G•C through A-to-I deamination. Both consist of an engineered deaminase fused to a Cas9-derived module for guide RNA-directed DNA targeting, and neither requires a donor DNA template or creates a double-strand break. Base editors have already entered clinical trials for sickle cell disease, beta-thalassemia, and familial hypercholesterolemia, and have been used to develop universal CAR T cells. In a striking precedent, researchers delivered a cytidine base editor by electroporation into human hematopoietic stem cells to correct the CD3D mutation that causes another form of SCID, achieving approximately 71.2 percent pathogenic-allele repair.

To systematically identify suitable editing targets for ART-SCID, the research team first compiled previously reported pathogenic point mutations in Artemis and mined the ClinVar database for additional disease-associated variants, filtering out 41 mutations documented as pathogenic and highly conserved throughout Artemis evolution. Using an established Artemis-knockout 293T cell line and a coding-joint reporter assay, an extrachromosomal V(D)J recombination system in which functional Artemis enables GFP expression quantified by flow cytometry, the team screened candidate variants. They newly identified four mutations, G6E, E10G, F19L, and L110R, that severely impair Artemis activity, reducing both its exonuclease activity toward single-stranded DNA and its endonuclease activity toward 3-prime overhang and DNA hairpin substrates. Structural modeling based on the wild-type Artemis crystal structure further revealed that mutations such as G118V, G135E, and L110R substantially reduce protein stability, and that G118 coordinates a conserved water molecule bridging residues critical for activity across the related SNM1 family members SNM1A and SNM1B.

From this catalog, the team selected three loss-of-function mutations amenable to correction by established base editors: c.49G>A (D17N) and c.404G>A (G135E), both targetable by adenine base editors, and c.181T>C (C61R), a representative cytidine-base-editor target. The clinical stakes of these variants are well documented. The c.49G>A mutation was identified in a Japanese SCID study in which two of eight patients harbored the homozygous variant, presenting before six months of age with severe infections and profound T cell deficiency. The c.404G>A mutation was reported in a study of four ART-SCID patients, two of whom carried the homozygous variant with a T-negative, B-negative, NK-positive immunophenotype; one presented with persistent varicella-zoster virus infection, and both ultimately succumbed to severe complications including multiorgan failure. Before editing, the team also assessed predicted bystander substitutions within the editing windows, finding that S62L moderately attenuates activity, I16M and Y133C cause minor reductions, T134A has negligible impact, and I16V has no detectable effect, confirming the three sites as suitable and comparatively safe targets.

The ex vivo editing results were encouraging. For the c.181T>C mutation, the team tested several cytidine deaminases incorporated into the BE4max backbone, including rat APOBEC1, APOBEC3A, and evoFERNY, alongside narrow-window variants and a combination of rat APOBEC1 with SpRY-HF1, a PAM-relaxed, high-fidelity Cas9 variant. Deep sequencing showed that rAPOBEC1-SpRY-HF1-BE4max paired with a suitable guide RNA achieved approximately 50 percent editing efficiency at the target site, with no significant off-target activity at any of five computationally predicted genomic loci per guide, as assessed by targeted deep sequencing and CRISPResso2 analysis. Flow cytometry confirmed that editing partially restored Artemis endonuclease activity in the mutant cells. Haplotype-level analysis of bystander mutations revealed that most co-occurring substitutions were silent due to codon degeneracy, with the predominant functional bystander being S62L, which only moderately reduces activity.

For the adenine base editor experiments, five editors were tested, including ABE7.10, ABEmax, ABE8e, ABE8e-SpRY-HF1, and ABE8e-NRRH. ABE8e achieved approximately 35 percent correction at the c.49G>A site, comparable to ABE8e-NRRH with a PAM-relaxed guide, and flow cytometry demonstrated significantly increased Artemis activity following correction, with no significant off-target mutations at predicted sites and a dominant bystander, I16V, that does not impair function. The c.404G>A site proved more difficult: ABE8e and ABE8e-SpRY-HF1 reached roughly 20 percent editing efficiency, but Artemis activity was not restored. The culprit appears to be a frequent bystander mutation, D136G, which completely abolishes enzyme activity. The authors suggest that prime editing, which couples Cas9 to a reverse transcriptase and an exogenous RNA template to enable precise nucleotide correction with markedly reduced bystander effects, may be better suited for mutations of this type.

The authors are candid about the limitations of the work. All editing experiments were performed in 293T cells, which provide a convenient and robust evaluation system but do not fully recapitulate the biology of human hematopoietic stem and progenitor cells, which exhibit distinct DNA repair pathway preferences and greater sensitivity to editing-associated cytotoxicity. Future studies in patient-derived CD34-positive cells will require clinically relevant delivery methods such as electroporation of ribonucleoprotein complexes, specialized culture conditions to preserve stemness, and long-term engraftment assays in immunodeficient mouse models to confirm safety and self-renewal capacity. The off-target analysis was also limited to predicted sites, and genome-wide evaluation will be needed. Moreover, base editing cannot address patients with homozygous large genomic deletions, though correcting a single allele in compound heterozygous patients is supported by the observation that parents carrying only one mutant allele remain asymptomatic. Nevertheless, by linking precise nucleotide correction to functional recovery of a critical immune gene, the study establishes a theoretical foundation for base-editing therapies for ART-SCID and related monogenic disorders, and underscores how rapidly evolving editor toolkits, including newer glycosylase-based editors capable of C-to-G and T-to-G conversions, continue to expand the range of mutations that may one day be corrected at the bedside.

Subject of Research: Base editing correction of Artemis gene mutations as a potential gene therapy approach for Artemis-deficient severe combined immunodeficiency

Article Title: Base editing of Artemis mutations ex vivo sheds light on gene therapy for Artemis-deficient SCID

Article References: Base editing of Artemis mutations ex vivo sheds light on gene therapy for Artemis-deficient SCID. (n.d.). https://doi.org/10.1007/s44307-026-00115-w

Image Credits: AI Generated

DOI: 10.1007/s44307-026-00115-w

Keywords: Artemis, ART-SCID, base editing, CRISPR, cytidine base editor, adenine base editor, V(D)J recombination, severe combined immunodeficiency, gene therapy, ex vivo editing, DCLRE1C, off-target effects