Transduction efficiency is a critical determinant of CAR T-cell manufacturing quality. Analysis of 204 clinical CAR T-cell products revealed that transduction efficiency is shaped primarily by manufacturing workflows and protocol-dependent starting material composition. Higher transduction efficiency was associated with early memory-like cellular states, providing insights into optimizing CAR T-cell.

He and colleagues developed a transient, non-gene-editing platform for off-the-shelf allogeneic TCR-T therapy. By conferring tacrolimus resistance characteristics to IL-2/4/7 expanded TCR-T cells that display reduced alloreactivity, He et al. provided a proof-of-concept for an allogeneic TCR-T therapy approach that addresses both manufacturability, and unwanted graft and host alloreactivity concerns.

CRISPR hybrid guides containing RNA and DNA nucleotides (chRDNA) enhance Cas nuclease specificity by reducing off-target editing. Intravenous administration of LNP-encapsulated Cas12a mRNA with a chRDNA guide in mice led to dose-responsive hepatic genome editing without detectable off-target editing and near elimination of the targeted plasma protein for one year.

Utilizing medicinal chemistry approach, Datta and colleagues demonstrate that EZH2-EGFR dual targeting halts protein translation in TNBC and inhibits tumor growth and metastasis in preclinical animal models.

LMNA-related congenital muscular dystrophy (L-CMD) is a devastating early-onset muscle disease caused by dysfunctional nuclear lamina. Current models fail to capture the complexity of human muscle pathology, limiting translational progress. This study overcomes this limitation by establishing a robust, human iPSC-based platform for modelling L-CMD and testing gene editing strategies.

Obesity expands a pathogenic ITGA7high adipose stromal cell population that promotes fibrosis. We engineered ITGA7-targeted lipid-coated mesoporous silica nanoparticles to selectively deliver plasminogen activator inhibitor-1 small interfering RNA, suppress profibrotic signaling, and restore a healthier adipose microenvironment.

In preclinical studies, UB-VV400, an off-the-shelf, investigational lentiviral drug product, generates fully human anti-CD22 CAR T cells in vivo without the need for lymphodepletion. Activation of the synthetic rapamycin-activated cytokine receptor drives selective CAR T cell expansion and enrichment, resulting in complete tumor clearance and B cell depletion.

Cellular antiviral defenses limit lentivector gene delivery. The authors describe lentivector transduction enhancement via inhibition of antiviral IFITM3 by novel inhibitor BG147. BG147 has activity ex vivo in human stem cells and in vivo in mouse subretinal cells, driving IFITM3 degradation via lysosomes. BG147 can expand lentivector utility and reduce treatment costs.

Vishwakarma and colleagues demonstrate that small extracellular vesicles derived from lineage-tailored human retinal ganglion-like cells restore mitochondrial and metabolic homeostasis in glaucomatous neurodegeneration. These vesicles reprogram retinal bioenergetics and promote mitochondrial homeostasis to enhance neuroprotection, offering a promising translational strategy for glaucoma and other optic neurodegenerative disorders.

To overcome the strict packaging limits of AAV vectors, this study utilizes a triple-AAV system paired with orthogonal split inteins to reconstitute the exceptionally large LAMA2 protein. This scarless, multi-vector approach successfully rescues the dystrophic phenotype in vivo, offering a scalable platform for large-gene therapies.

CAR T cells engineered to secrete tumor-localized anti-EpCAM bispecific T cell engagers (BTCEs) overcome antigen escape and heterogeneity across multiple epithelial carcinomas in preclinical models, achieving complete tumor eradication where conventional single-target CAR T cell therapies often failed, supporting broad translational potential for solid tumor immunotherapy.

Injectable methylcellulose microfoam enables rapid bedside/in vivo generation of CAR-T cells by colocalizing patient PBMCs, gene therapy vectors, and stimulatory cues. The platform enhances gene transfer, minimizes systemic vector exposure, and produces therapeutically effective CAR-T cells without conventional ex vivo manufacturing, potentially expanding patient access to cellular immunotherapy.

In vivo gene therapy using lentiviral vectors (LVs) has been moving into focus. This review discusses strategies to redirect LV particles to specific cell types through transductional targeting, with special emphasis on vesicular stomatitis virus glycoprotein, its advantages and limitations, current challenges, and potential fields of application, in addition to an overview of ongoing clinical trials.

Multivalent DNA plasmids, each encoding a diverse set of Mtb antigens, elicited robust polyfunctional T cell responses in naive and BCG-primed mice. Boosting with the lead construct, pESX, resulted in superior immunogenicity that correlated with prolonged pulmonary bacterial control in BCG-primed mice.

Stanek and colleagues uncover a novel PRPF31 intronic mutation that disrupts splicing and lowers protein levels. Targeted antisense oligonucleotides restore normal splicing and boost PRPF31 expression in patient-derived RPE, highlighting a potential therapeutic strategy for retinitis pigmentosa.

Nature Biomedical Engineering, Published online: 26 August 2026; doi:10.1038/s41551-026-01774-9

Autophagy receptors are engineered fusing LC3 to cytoplasm-stable antibody variants and are delivered via small extracellular vesicles or adeno-associated virus to eliminate pathological Tau and TDP-43 in animal disease models.

arXiv:2609.09306v1 Announce Type: new Abstract: This paper investigates the hypothesis that the first-order structure of physical interactions, i.e. gradients or Jacobians, characterizes the structure of phenomenal experience. It does so in an idealized world inhabited by neural networks, Gradland, where the physics are known and the functions are (mostly) differentiable. The paper introduces two measures of Jacobian structure: effective rank and cohesion, based on Kirchhoff complexity. Applying the measures to a series of worked examples shows the hypothesis accounts for: (1) the duration of experience, that it can prolong over hundreds of milliseconds; (2) the difference between what is experienced vividly and obscurely; (3) the experience of texture; (4) the blooming buzzing confusion presumably experienced by newborns; (5) the difference between ideas that are held distinctly in mind and ideas that are confused; (6) what learning is like; and finally (7) the paper explains the function of rich, dense experience.

arXiv:2609.09647v1 Announce Type: new Abstract: Agentic systems are rapidly moving to production, where they read untrusted inputs, call tools with real permissions, and act autonomously, expanding the security surface beyond chat-only models. Yet standard evaluations remain single-turn and fail to capture multi-step agent vulnerabilities. We present a systematic black-box framework for risk-aware agent evaluation requiring only basic system descriptions. Our approach introduces: (1) a seven-domain taxonomy mapping observable behaviors to risk categories, (2) fully automated SAGE-RT red teaming producing 120 adversarial scenarios per domain, and (3) human-validated evaluation using LLM judges. Empirical validation across two agent architectures (CrewAI and AutoGen) with four base models reveals alarming patterns: 56.25\% average governance risk, 65\% privacy risk in multi-agent configurations, and agent behavior vulnerabilities reaching 85\%. Our black-box approach effectively identifies critical architectural vulnerabilities without privileged access, providing a scalable path toward safer agent deployments.

arXiv:2609.09776v1 Announce Type: new Abstract: Frontier gains in language-model reasoning come from reinforcement learning on reasoning traces and are concentrated in domains with a cheap, sound verifier. We argue the field's binding constraint is the verification gap: no scalable, incorruptible reward for reasoning outside formal domains. We make four contributions. (1) Theory: in a joint-Gaussian model of best-of-N selection, verifier-gold correlation rho is the exact exchange rate between test-time compute and capability, and an unsound verifier pays a polynomial penalty N^(1/rho^2); a margin-free copula form predicts realized soundness of real LLM judges to 4% median error. (2) Demonstration: in program-synthesis testbeds with executable ground truth, including a pre-registered scaled replication, unsound verifiers lose Soundness-under-Pressure as optimization grows (0.94 to 0.32 at N=4096) while a sound verifier improves monotonically; reality-anchored settlement beats a frozen verifier under i.i.d. and adversarial pressure, driving the hacking gap from ~0.27 to ~0; soundness scales log-linearly with settled labels, with on-policy settlement ~10x more label-efficient than random labeling. With real LLM judges and unit-test execution as gold, a weak judge loses soundness under best-of-N (p

arXiv:2609.09853v1 Announce Type: new Abstract: LLM agents for enterprise systems of record cannot be evaluated on customer production data, and no existing substitute provides ground truth. We present the Era by Eon Benchmark for evaluating LLM agents that use enterprise tools. The benchmark is built around a complete fictional company. It includes product simulators, company-specific internal databases, benchmark questions, and computed answer keys. Industry, company size, business model, application portfolio, and a seed define each company. One seeded entity graph supplies shared company data to simulators of Salesforce, Zendesk, Slack, Gong, and other products. A questionconditioned generator creates the schemas and records for internal databases. It takes shared entities, keys, and values from the same graph before generating database-specific facts. Both mechanisms therefore describe one consistent enterprise estate. Every expected answer is computed from the final records, so grading is exact. Design and answer-key checks validate the internal databases. A realism scorecard and adversarial detector validate the entity graph. Across 23 generated companies, the mean realism score rose from 61.8 to 97.0, with zero records flagged as synthetic. In the reported simulator-track comparison, nine models answered the same 33 questions three times each. Accuracy estimates ranged from 42.4% to 76.8%, and three of 36 pairwise differences remained supported after correction.