Gene therapy achieved through genetic manipulation and transfer of hematopoietic stem and progenitor cells (HSPC) is becoming an increasingly attractive option for personalized medicine. In the current clinical scenarios, use is restricted to overcoming genetic disorders, but it is envisioned that more diverse applications will be developed in the future. Immunotherapy of malignant disease already employs non-HSPC-based gene therapies, and there is a body of evidence that HSPC-mediated gene therapy has potential as a powerful immunotherapeutic for some diseases of immune dysregulation such as autoimmune disease and allergies. Administration of HSPC gene-engineered to encode antigen(s) induces antigen-specific immune tolerance in animal models. To translate this technology, we sought to identify whether human dendritic cells (DC) and other antigen-presenting cell populations derived from lentivirally transduced HSPC would express, process, and present antigens of interest. Lentiviral vectors were generated that encoded preproinsulin, a relevant autoantigen implicated in type 1 diabetes, or a control protein carrying a viral epitope along with reporter genes under transcriptional control of ubiquitous or antigen-presenting cell-directed promoters. In vitro, DC were differentiated from transduced HSPC and tested in antigen presentation assays using Jurkat lines carrying relevant type 1 diabetes-associated T cell receptors. In vivo, transduced HSPC were used to generate humanized mice, and the development of a range of antigen-presenting cells and their expression of lentivirus-encoded genes were determined. Antigen-presenting cell progeny of transduced HSPC expressed antigens and reporter genes both in vitro and in vivo, and in vitro-generated DC presented lentivirally encoded antigens. The studies characterize human chimerism and therapeutic gene expression in a new model and provide support for the use of HSPC-based gene therapies for immune dysregulation in humans.

Total elimination of replication-competent human immunodeficiency virus type 1 (HIV-1) remains a major clinical challenge, in part due to random integration of the proviral DNA into host cell chromosomes, which enables lifelong persistence and production of progeny. Although antiretroviral therapies (ARTs) suppress viral replication, they cannot eliminate integrated proviral DNA, which remains a fundamental obstacle to achieving a cure. To overcome this problem, we developed a combinatorial clustered regularly interspaced short palindromic repeats-Cas9 gene editing strategy to disrupt viral replication and inactivate host factors essential for HIV-1 entry and spread. This approach targets C-C chemokine receptor type 5 (CCR5), a chemokine receptor central to HIV-1 host cell entry, and mannosyl-oligosaccharide glucosidase (MOGS), a key enzyme in glycoprotein processing that modifies the HIV-1 envelope glycoprotein gp120, facilitating receptor engagement, viral entry, and morphogenesis of infectious virion. We demonstrate that our strategy, which includes editing of the integrated proviral DNA, in concert with two cellular genes whose products facilitate viral entry, results in robust suppression of viral replication in vitro and in ex vivo-infected cells. Using transmission electron microscopy, HIV-1 p24 ELISA, and GFP-based viral infection assays, we show that the combination knockout of CCR5, MOGS, and viral sequences profoundly reduces HIV-1 replication in an ex vivo cellular model, that is, HIV-1-infected peripheral blood mononuclear human cells, thus offering a pathway to launch further preclinical studies.

This retrospective follow-on study reviewed 25 consecutive patients with genetically confirmed choroideremia who were treated with subretinal AAV2-CAG-REP1 in dose-escalation gene therapy clinical trials conducted at Oxford University Hospitals NHS Foundation Trust and the University Eye Hospital Tübingen. Six patients in Oxford received 10[10] vector genomes in 100 µL, while a further 19 patients across both centers received 10[11] vector genomes in 100 µL. Although the primary safety outcomes of both investigator-led academic studies have previously been reported, retinal imaging changes were observed during follow-up, prompting a more detailed characterization of these findings. Twelve of 25 treated eyes developed imaging features consistent with perivascular astrocytic gliosis during follow-up, representing 48% of the cohort. All affected eyes had received the higher dose, accounting for 12 of 19 eyes treated with 10[11] vector genomes. No patients receiving the lower dose (1.0 × 10[10] vector genomes) developed this phenotype. The lesions followed the major vascular arcades toward the optic disc, and optical coherence tomography demonstrated inner retinal thickening with a fine, granular hyperreflective texture, without intraretinal cysts. Longitudinal imaging showed no progressive expansion of the involved regions. Clinically, the findings were silent: no patient reported uveitic symptoms, ocular examinations were largely quiet, and no peripheral cellular immune responses were detected in patients with available assays. Perivascular astrocytic gliosis therefore appears to be a common, clinically silent, nonprogressive, and noninflammatory retinal response after subretinal AAV2-CAG-REP1 gene therapy for choroideremia at a dose of 10[11] vector genomes, which is generally accepted as the therapeutic human dose for targeting photoreceptors. We hypothesize that local REP1 overexpression in perivascular astrocytes suppresses FOXO3-mediated apoptosis and engages EGFR/STAT3 signaling, driving benign perivascular astrocytic accumulation analogous to retinal astrocytic hamartomas seen in phacomatoses. Gene therapy involving ubiquitous promoters may have unexpected effects, and while benign, may not be predicted.

Adeno-associated virus (AAV)-mediated gene transfer remains a promising strategy for cystic fibrosis (CF), but durability of expression and optimal dosing intervals are unresolved challenges. Here, we evaluated long-term gene transfer, transduction, and immunological responses following a single pulmonary administration of an AAV1 vector encoding Δ27-264 CFTR to juvenile rhesus macaques. Four animals received 1 × 10[1][3] vector genomes via endotracheal spray. Two were untreated. All were analyzed 180 days post-delivery. Vector genomes were detected throughout the conducting airways and distal lung at levels ranging from ∼5 × 10[6] to 9 × 10[7] vg/µg genomic DNA. Vector genomes were also detected in extrapulmonary tissues, including liver, pancreas, spleen, heart, and kidney. Cystic fibrosis transmembrane conductance regulator (CFTR) messenger RNA expression was detected in all tissues examined and in several respiratory and extrapulmonary tissues, accompanied by increased CFTR protein expression in multiple tissues. Transgene expression was observed in airway basal cells and FOXI1-positive ionocytes, indicating targeting of key epithelial and progenitor cell populations. Despite persistence of vector genomes, immune activation was limited. Cytokine and chemokine expression profiles showed only sporadic, tissue-specific changes without evidence of coordinated inflammation. Neutralizing antibodies to AAV1 developed in all animals but did not correlate with capsid-specific T cell responses, which remained low and infrequent. These findings indicate sustained gene transfer and transduction for at least 180 days following a single AAV1 administration to the primate lung, with minimal immunopathology. The durability of expression supports the potential feasibility of extended dosing intervals and informs the development of repeatable AAV-based gene therapies for CF.

To date, prediction of manufacturing failure for CAR T cells is still lacking. In a retrospective multivariate analysis of our 368 manufactured CAR-T cell batches (tisagenlecleucel), we investigated potential factors that might be associated with manufacturing failure. In this letter to the editor, we provide a summary of our findings and outline recommendations for future advanced quality control strategies.

Accurate classification of Genetically Modified Organism (GMO) gene therapies remains a significant challenge to initiating clinical trials in the United Kingdom, often delaying patient access to innovative treatments. Current regulatory processes require separate GMO risk assessments at each National Health Service (NHS) trial location, leading to inconsistent approaches and duplication of effort. This article presents the development of a decision tree designed to streamline GMO classification for Gene Therapy Investigational Medicinal Products (GTIMPs). The proposed tool aims to address key challenges in trial setup by providing a structured and evidence-based approach to GMO classification, clarifying international misalignment in hazard classification and offering practical guidance for the in vivo and ex vivo GMO GTIMPs most encountered in clinical practice. By harmonizing decision-making and clarifying containment requirements, the adoption of the decision tree across NHS organizations could reduce variability, accelerate trial setup, and strengthen the United Kingdom's position as a competitive destination for GTIMP clinical trials.

Friedreich ataxia (FA) is a progressive neurodegenerative disorder caused by reduced expression of frataxin (FXN), a mitochondrial protein essential for iron-sulfur (Fe-S) cluster biogenesis. Although gene therapy strategies aimed at restoring FXN have shown promise, excessive expression can lead to mitochondrial dysfunction, emphasizing the importance of maintaining FXN within a physiological range. Here, we evaluated a gene therapy approach based on a human mini-frataxin construct (miniFXN7) incorporating an endogenous regulatory element to enable controlled FXN expression. The construct was delivered systemically using an AAV-PHP.eB vector in the Pvalb-cKO mouse model of FA. MiniFXN7 treatment resulted in widespread neuronal transduction and restoration of FXN expression toward a near-physiological range in the neuronal populations examined. Treated mice exhibited sustained improvements in motor coordination and proprioceptive function, including normalization of H-reflex responses. At the cellular level, miniFXN7 restored succinate dehydrogenase activity, a mitochondrial Fe-S enzyme, and was associated with partial normalization of mitochondrial morphology. In parallel, neuronal integrity was preserved and astrogliosis reduced across the cerebellum. These findings demonstrate that physiologically regulated FXN replacement is sufficient to achieve substantial functional rescue in FA, supporting a gene therapy strategy based on a transgene expression driven by endogenous regulatory elements.

Adeno-associated virus (AAV) gene transfer vectors mediate long-term expression in nondividing cells, an advantage for treating chronic disorders. However, current platforms lack a way to selectively shut down transgene expression if adverse effects arise. To create an "off switch," we hypothesized that incorporating unique artificial microRNA (amiRNA) target sequences into an AAV expression cassette would allow subsequent suppression of transgene expression using a second AAV vector encoding the cognate amiRNA. We introduced 22-nt sequences absent from human and mouse transcriptomes into the 3' untranslated region (UTR) of a therapeutic AAV cassette. To identify optimal amiRNAs, two tandem copies of each amiRNA were cloned into the 3'UTR of an mCherry reporter gene. In vitro assessment of six amiRNA/target pairs using a dual luciferase assay identified four amiRNAs that efficiently suppressed reporter expression. Cells cotransfected with target site 3 (TS3) and amiRNA-T3B showed the greatest reduction in luciferase activity (80%, p < 0.0001) and were selected for further study. The "off-switch" system was then evaluated using an AAV5 therapeutic vector expressing a recombinant humanized anti-IgE monoclonal antibody (AAV5-TBG-anti-IgE-TS3), designed for long-term suppression of allergen-induced reactions. Co-transfection of HEK293T cells with anti-IgE-TS3 and amiRNA-T3B significantly reduced anti-IgE mRNA and protein levels relative to a control amiRNA (p < 0.0001). In vivo testing in Balb/c mice (n = 5) involved intravenous administration of AAV5-anti-IgE-TS3 (3.2 × 10[10] gc), followed 4 weeks later by an AAVrh.10 amiRNA vector (AAVrh.10-TBG-amiRNA-T3B; 1 × 10[11] gc). Control mice receiving only the therapeutic vector expressed 18.4 ± 13.8 µg/mL serum anti-IgE at 10 weeks. In contrast, mice receiving the amiRNA "off" vector showed marked suppression of anti-IgE (0.3 ± 0.15 µg/mL, p < 0.0001). These findings provide proof-of-concept that AAV-delivered amiRNAs can selectively switch off transgene expression, offering a strategy to improve the safety of AAV-mediated gene therapies.

Delandistrogene moxeparvovec-rokl (DMR) is the only U.S. Food and Drug Administration-approved gene therapy for patients with Duchenne muscular dystrophy (DMD). After reports of two deaths due to acute liver injury (ALI)/acute liver failure, our center began prophylaxis with sirolimus to help prevent ALI. The objective of this study is to report the initial safety, tolerability, and efficacy of sirolimus prophylaxis. The clinical course of patients receiving sirolimus prophylaxis with a target trough of 4-6 ng/mL was retrospectively evaluated and compared with those who did not receive sirolimus. Laboratory testing, particularly gamma glutamyl transferase, aspartate aminotransferase, alanine aminotransferase, and total bilirubin, was collected. A Fisher's exact test was used to assess the difference between groups. Twelve patients were started on sirolimus prophylaxis, 10 prior to DMR infusion and 2 within 2 weeks of DMR. The median age at DMR dosing was 11 years old (interquartile range = 8.5-14); 11 patients were ambulatory. No patients experienced a serious adverse event related to sirolimus administration. Seven patients developed elevated triglyceride or cholesterol levels and began omega-3 supplementation. None of the patients receiving sirolimus prophylaxis experienced ALI compared with 3 of the 14 patients (21%) infused prior to implementation of the sirolimus protocol. Sirolimus prophylaxis appears to be safe and well-tolerated in DMD patients receiving DMR. While not statistically significant, these results suggest that further study of sirolimus prophylaxis is warranted.

Adoptive immunotherapies have emerged as a promising strategy in the treatment of hematological malignancies. To date, seven chimeric antigen receptor (CAR)-T cell products have obtained market authorization in Europe for B-cell malignancies, where they have revolutionized treatment for eligible patients. In addition, natural killer (NK) cells and γδ T cells are of particular interest for cell-based therapies due to their strong intrinsic cytotoxicity and favorable safety profile. Addressing challenges facing the clinical translation of NK cell therapies was one of the issues discussed in the NK & ILC Symposium that took place in Freiburg, Germany from March 11 to 13, 2026, as the annual meeting of the NK & ILC study group of the German Society for Immunology (DGfI). Topics ranging from basic immunological research to technological innovations and clinical trial results were discussed during the 3-day conference, spanning over 40 presentations and 100 posters. A highlight of the conference was a workshop featuring short presentations and a panel discussion that focused specifically on the current challenges and future prospects of the clinical implementation of NK cell therapies; the findings of this workshop are summarized in this opinion paper.

Targeted in vivo transduction, entailing direct administration of viral vector preparations to patients, is the next big step in gene therapy. To redirect lentiviral vector (LV) particles selectively to desired cell type(s), different glycoproteins have been engineered to alter their tropism, including the glycoprotein G from Vesicular Stomatitis Virus (VSV-G) as the most commonly employed tropism-defining protein for LV particles. For detargeting from its natural receptor, the low-density lipoprotein receptor (LDLR), a VSV-G variant with two blinding substitutions, K47Q and R354A, is commonly used (VSV-G.pub). We provide first evidence for insufficient blinding of VSV-G.pub in human and murine cell lines and primary cells. In silico modeling pointed toward only slightly reduced LDLR binding affinity of VSV-G.pub. To lower the affinity further, we generated three novel VSV-G variants based on charge-reversing substitutions in two, four, or six key residues. These variants achieved a more stringent blinding compared with VSV-G.pub in the tested cell lines and primary cells. Codisplay of a CD4 binder enabled lentiviral particles pseudotyped with the novel variants to selectively transduce CD4-expressing cells. In summary, we present improved VSV-G variants with a better on/off-target ratio as attractive tools for in vivo gene therapy applications.

Common genetic variants of APOE are major risk factors for sporadic late-onset Alzheimer's disease (AD). APOE has three common variants: APOE3, APOE4, and APOE2. Epidemiological, clinical, and experimental evidence demonstrate that APOE3 is associated with an average risk for AD, APOE4 is pathogenic and conveys a high risk, and APOE2 is protective and reduces risk. In prior mouse studies, we have demonstrated that (1) adeno-associated virus (AAV)-mediated central nervous system (CNS) gene transfer of APOE2 is highly protective against the toxic effects of APOE4 and (2) AAV-mediated CNS transfer of anti-APOE4 microRNA (miRNA) significantly suppresses the expression of APOE4. The ideal therapy for APOE4 homozygotes would both reduce CNS levels of APOE4 and substitute APOE4 with the addition of APOE2. We have developed a "silence-and-replace" therapy, where suppression of CNS APOE4 is achieved through AAV-mediated expression of miRNAs designed to silence the endogenous human APOE4 messenger RNA (mRNA), together with simultaneous replacement by an APOE2 coding sequence [APOE2(R)] resistant to those miRNAs. AAV vectors expressing APOE2(R) with and without the miRNAs were administered to the hippocampus of human APOE4 mice, and APOE levels at the mRNA and protein levels were the same for both vectors. Dideoxy Sanger sequencing and allele-specific real-time quantitative PCR were used to assess the CNS APOE2 to APOE4 ratio. As expected, both vectors with APOE2(R) increased the E2/E4 ratio relative to controls by 3.5 ± 0.6-fold for APOE2(R) and 6.3 ± 0.4-fold for APOE2(R) + miRNAs. We conclude that a single AAV vector can deliver a cassette with dual functionality: suppression of the toxic APOE4 variant and simultaneous delivery of the protective APOE2. This dual approach may provide a more potent gene therapy for APOE4-dependent AD than either strategy.

Clustered Regularly Interspaced Short Palindromic Repeats (CRISPR) technology has revolutionized genetic medicine by enabling precise genome editing for therapeutic benefit. CRISPR nucleases are programmed to target genomic sites with sequence complementarity to the spacer region of an associated guide RNA. However, these nucleases may target genomic loci with sequences similar to the target site, which can lead to unintended disruption of off-target genes. The risks associated with these off-target editing events are critical to assess as CRISPR-based in vivo editing systems advance to clinical development. Nonhuman primates (NHPs) are common model species for evaluating the human safety of many therapeutic modalities, but their relevance for evaluating human CRISPR off-target activity has yet to be determined. In this study, 1,220,908 Cas12a and 6,159,066 Cas9 spacer sequences targeting human genes were designed, and off-target editing sites were computationally predicted in humans and five common NHP species. Of the 7,413 Cas12a and 570,754 Cas9 spacers meeting defined on- and off-target inclusion criteria, only 14-21% of Cas12a and 7-15% of Cas9 human off-targets per spacer are recapitulated in the genomes of NHPs commonly used in preclinical studies. These results highlight the limitations of NHPs for the study of human CRISPR spacer specificity and contextualize human risk informed by these studies.