The Paradigm Shift
Current CAR-T therapy requires extracting a patient's T cells (leukapheresis), shipping them to a centralized facility, genetically engineering them ex vivo over 2–4 weeks, then infusing them back. This costs $300K–$500K per patient and is limited by manufacturing variability, production time, and logistical complexity.
This Nature 2026 study demonstrates the first site-specific integration of a large DNA payload in primary human T cells directly in the body. Using a dual-vector system — anti-CD3 enveloped delivery vehicles (EDVs) carrying Cas9-RNP + evolved AAV-hT7 delivering homology-directed repair templates — they achieve promoterless CAR integration at the endogenous TRAC locus, producing T cell-specific, physiologically regulated CAR expression without any ex vivo manufacturing.
Three Engineering Barriers Solved
Neutralizing Antibodies
AAV6 is inactivated by pre-existing human serum antibodies. Solution: Directed evolution of AAV6 capsid on human T cells in serum → AAV-hT7 with HAPRVEE motif, 20,000× enrichment, CD7-dependent entry. Maintains full transduction even in serum.
T Cell Selectivity
VSVG-EDV has broad tropism — can transduce HSCs (oncogenic risk) and tumor cells (antigen-negative relapse). Solution: Anti-CD3 scFv on EDV + CD7-targeting AAV-hT7 + promoterless TRAC. Triple-layer specificity eliminates off-target.
Cycling T Cells
HDR requires actively dividing cells, but most circulating T cells are quiescent. Solution: Anti-CD3 scFv on EDV surface activates naive T cells (CD25+ CD69+ comparable to Dynabeads), driving entry into cell cycle for efficient HDR.
Key Results Summary
Technology Stack
-
1
Anti-CD3 EDV — Enveloped delivery vehicle with mutated VSVG (no LDLR binding) + anti-CD3 scFv for T cell targeting and activation. Delivers Cas9-RNP transiently.
-
2
AAV-hT7 — Evolved AAV6 capsid resistant to neutralizing antibodies. Targets CD7 (pan-T/NK receptor). Delivers HDRT encoding promoterless CAR.
-
3
TRAC Locus — Endogenous TCR alpha promoter drives physiological CAR expression. Disrupts TCR (reduces GvHD risk). Uniform expression across all edited cells.
-
4
1XX CAR Architecture — CD28ζ-1XX with mutated ITAMs 2&3. Delays exhaustion, improves persistence. Consistent across CD19/BCMA/B7H3 targets.
-
5
Triple Specificity — EDV (anti-CD3) + AAV (CD7) + Promoter (TRAC) = zero integration in HSCs, B cell tumor lines, and macrophages.
EDV + AAV Dual-Vector Mechanism
EDV Specifications
| Envelope | Mutated VSVG (ablated LDLR binding) + anti-CD3 scFv |
| Cargo | Cas9 protein with 4× NLS + sgRNA (TRAC-targeting) |
| Dose | 2.5–5 × 10¹¹ sgRNAs per mouse |
| Expression | Transient (protein delivery, no integration) |
| Activation | CD25⁺/CD69⁺ induction comparable to Dynabeads |
| TCR KO | 50% at lowest MOI; 67% at standard MOI |
| Specificity | CD3⁺ cells only (T cells). No HSC/macrophage transduction |
AAV-hT7 Specifications
| Capsid | Evolved AAV6 variant (HAPRVEE motif at pos 454–456) |
| Cargo | ssDNA HDRT: LHA–P2A-CAR-P2A-EGFRt–RHA |
| Dose | 1 × 10¹² viral genomes per mouse |
| Receptor | CD7 (pan-T/NK, KIAA0319L co-required) |
| Enrichment | 20,000× from parental library (3 evolution cycles) |
| Serum resistance | Full transduction in human serum (vs ~0 for AAV6) |
| Off-target | Abolished HSC transduction; ↓ B cell tumor integration |
Specificity by Cell Type
| Cell Type | VSVG + AAV6 | VSVG + AAV-hT7 | αCD3 + AAV6 | αCD3 + AAV-hT7 |
|---|---|---|---|---|
| CD4⁺ T cells | 100% | ~80% | ~60% | ~50% |
| CD8⁺ T cells | 100% | ~85% | ~55% | ~45% |
| NK cells | Low | Low | 0% | 0% |
| CD34⁺ HSCs | Positive | 0% | 0% | 0% |
| Macrophages | Positive | Low | 0% | 0% |
| B cell lines (NALM6, SupB15, JeKo1, Raji) | Positive | Reduced | Reduced | 0% |
Directed Evolution of AAV-hT7
Serum Resistance Comparison
Host Factor Dependencies
CD7 Biology — The Key Receptor
Expression Pattern
CD7 is broadly expressed across T cell populations (CD4⁺, CD8⁺, γδ) and mature NK cells. Not expressed on B cells, HSCs, macrophages, or tumor cells — providing inherent selectivity.
Internalization
CD7 undergoes rapid ligand-induced internalization, likely facilitating AAV-hT7 uptake. High CD7 surface expression on T cells enables efficient vector entry even at low MOI in vivo.
Pan-T Advantage
Unlike CD4/CD8-targeting AAVs that engineer only one subset, CD7 enables pan-T cell access — engineering both CD4⁺ and CD8⁺ subsets simultaneously for balanced CAR T pools.
In Vivo CAR T Generation by Vector Combination
Tumor Model Results
| Model | Target | CR Rate | Details |
|---|---|---|---|
| B-ALL (NALM6) | CD19 | 18/20 (90%) | 4 PBMC donors, single injection, B cell aplasia confirmed |
| B-ALL Rechallenge | CD19 | Controlled | 5×10⁶ NALM6 rechallenge at d39 → no tumor increase over 2 weeks |
| Myeloma (OPM2) | BCMA | 8/8 (100%) | Complete response, 3/4 durable after rechallenge |
| Sarcoma (MES-SA) | B7H3 | 5/6 donor 1 | First solid tumor demo; 3/8 donor 2 |
CAR T Cell Phenotype
Head-to-Head: TRAC vs Lentiviral (In Vivo)
Key Advantages of TRAC Integration
-
▲
21–50× more CAR T cells at week 2 vs lentiviral in vivo
-
▲
Uniform CAR expression — high/homogeneous MFI vs variable lentiviral
-
▲
Earlier peak expansion — week 2 (TRAC) vs week 3 (lentiviral)
-
▲
6/6 complete response vs 1/6 CR for high-dose lentiviral
-
▲
Higher stem cell memory — enriched CD45RA⁺CD62L⁺ CD8⁺ subset
-
▲
Normal effector contraction after tumor clearance
CAR-T Manufacturing Paradigm Arena
| Feature | Autologous (Standard) | Allogeneic (Off-shelf) | In Vivo LNP | In Vivo Lentiviral | In Vivo TRAC (This Work) |
|---|---|---|---|---|---|
| Manufacturing | Patient-specific ex vivo | Donor-derived ex vivo | None (injectable) | None (injectable) | None (injectable) |
| Integration | Random (retroviral) | Site-specific (TRAC) | None (transient mRNA) | Random (lentiviral) | Site-specific (TRAC) |
| CAR Expression | Variable, constitutive | Uniform, physiological | Transient (~days) | Variable, constitutive | Uniform, physiological |
| T Cell Specificity | High (ex vivo sorted) | High (ex vivo sorted) | Medium (LNP targeting) | Medium (envelope eng.) | Very high (triple layer) |
| HSC Risk | None | None | Low (transient) | Moderate | Abolished |
| Tumor Transduction | Reported (rare) | None | Low (transient) | Risk (Ag⁻ relapse) | Abolished |
| GvHD Risk | None (autologous) | Present (TCR retained) | None | None | None (TCR disrupted) |
| Persistence | Good | Limited (rejection) | Poor (re-dosing needed) | Variable | Durable (rechallenge shown) |
| Cost (est.) | $300–500K | $100–200K | $10–50K | $10–50K | $10–50K (est.) |
| Time to Treatment | 4–6 weeks | Days (off-shelf) | Same day | Same day | Same day |
| Stage | 7 FDA approved | Phase 1/2 | Phase 1 | Phase 1 | Preclinical |
Company Landscape
| Company | Approach | Stage | Vector |
|---|---|---|---|
| Novartis (Kymriah) | Autologous | Approved | Lentiviral ex vivo |
| Gilead/Kite (Yescarta) | Autologous | Approved | Retroviral ex vivo |
| BMS (Abecma, Breyanzi) | Autologous | Approved | Lentiviral ex vivo |
| Umoja Biopharma | In vivo LVV | Phase 1 | Fusosome (anti-CD3 LVV) |
| Capstan Therapeutics | In vivo LNP | Phase 1 | Targeted LNP (mRNA) |
| CRISPR Therapeutics | Allogeneic | Phase 1/2 | Cas9 ex vivo |
| Allogene | Allogeneic | Phase 1/2 | TALEN ex vivo |
| This Work (MSK/UCSF) | In vivo TRAC | Preclinical | EDV + AAV-hT7 |
Approach Capability Radar
Market Impact Analysis
Roadmap to Clinical Translation
Translational Challenges
Technical Challenges
-
!
Redosing limitation: Anti-AAV-hT7 antibodies generated after first injection. May require capsid switching or immune modulation for repeat dosing.
-
!
TCR clonality: More limited T cell repertoire compared to ex vivo products. Clinical significance unknown but consistent with other in vivo approaches.
-
!
Dosing in humans: Mouse-to-human dose scaling unclear. Will require careful dose-finding with real-time CAR T cell monitoring.
-
!
Pre-existing AAV immunity: ~30–60% of humans have anti-AAV6 antibodies. AAV-hT7 shows resistance, but prevalence of cross-reactive antibodies against evolved capsid needs testing.
Safety Considerations
-
✓
No systemic inflammation: No cytokine elevations at day 1 or day 7 post-injection in mice. Favorable safety signal.
-
✓
Treg depletion: Reduced FOXP3⁺CD25⁺ Tregs in CAR⁺ fraction — beneficial for anti-tumor response.
-
✓
Triple specificity: Zero off-target integration in HSCs, macrophages, and tumor cell lines.
-
?
Long-term genotoxicity: Site-specific integration is safer than random, but large-scale insertional analysis in human cells is needed.
Beyond CAR-T: Platform Applications
TCR Replacement
Same EDV/AAV platform could integrate tumor-specific TCR sequences at TRAC — in vivo TCR-T cell therapy for solid tumors with known neoantigen peptides.
Synthetic Receptors
SynNotch, SUPRA-CAR, or logic-gated receptors could be integrated at TRAC for programmable T cell responses — smart T cells generated directly in the patient.
Other Cell Types
The dual-vector concept (cell-specific EDV + evolved AAV + lineage promoter) is generalizable to NK cells, macrophages, B cells, or any cell type with unique surface markers.
Autoimmune Disease
In vivo CAR-T targeting autoreactive B cells (anti-CD19/BCMA) for lupus, MS, myasthenia gravis — same-day treatment for autoimmune conditions.
Infectious Disease
HIV-specific CAR integration at TRAC for functional cure strategies — AAV9 already in clinical trial (NCT05144386). This approach offers better specificity.
Integration Methods
PASTE, PASSIGE, CAST technologies could replace HDR for even larger payloads. The modular dual-vector architecture accommodates any integration enzyme.
In Vivo Efficacy Estimator
References
- Eyquem, J. et al. Targeting a CAR to the TRAC locus with CRISPR/Cas9 enhances tumour rejection. Nature 543, 113–117 (2017). DOI
- Eyquem, J., Sadelain, M. et al. In vivo site-specific engineering to reprogram T cells. Nature s41586-026-10235-x (2026). DOI
- Sadelain, M. et al. Therapeutic T cell engineering. Nature 545, 423–431 (2017). DOI
- Michels, K. R. et al. Evolved AAV achieves site-specific editing in murine T cells in vivo. Sci. Transl. Med. (2024). DOI
- Ruella, M. et al. Induction of resistance to chimeric antigen receptor T cell therapy by transduction of a single leukemic B cell. Nat. Med. 24, 1499–1503 (2018). DOI
- Hamilton, J. R. et al. In vivo human T cell engineering with enveloped delivery vehicles. Nat. Biotechnol. 42, 1684–1692 (2024). DOI
- Urnov, F. D. et al. Genome editing with engineered zinc finger nucleases. Nat. Rev. Genet. 11, 636–646 (2010). DOI
- Umoja Biopharma. VivoVec in vivo CAR-T platform — Phase 1 clinical update. ClinicalTrials.gov (2025–2026).
- Capstan Therapeutics. Targeted LNP platform for in vivo T cell reprogramming — Phase 1 first patient dosed. Press Release (2025).
- Frangoul, H. et al. CRISPR-Cas9 gene editing for sickle cell disease and β-thalassemia. NEJM 384, 252–260 (2021). DOI
- Foy, S. P. et al. Non-viral precision T cell receptor replacement for personalized cell therapy. Nature 615, 687–696 (2023). DOI
- June, C. H. & Sadelain, M. Chimeric antigen receptor therapy. NEJM 379, 64–73 (2018). DOI
- Labanieh, L. et al. Enhanced safety and efficacy of protease-regulated CAR-T cell receptors. Cell 185, 1745–1763 (2022). DOI
- Yousefpour, P. et al. Targeted modulation of the human immune system using LNPs. Nat. Rev. Drug Discov. (2024). DOI
- Verma, M. et al. Clinical application of in vivo gene editing for CAR-T generation. Blood Adv. (2025).
- Palanki, R. et al. Ionizable LNPs for in vivo mRNA delivery to the hematopoietic system. Sci. Adv. (2024).
- Mingozzi, F. & High, K. A. Overcoming the host immune response to AAV gene therapy vectors. Blood 122, 23–36 (2013). DOI
- Wang, D. et al. AAV gene therapy: the first approved gene therapy product. Mol. Ther. 27, 1195–1199 (2019). DOI
- Anzalone, A. V. et al. Programmable large DNA insertions at predetermined sites (PASTE). Nat. Biotechnol. 40, 968–979 (2022). DOI
- FDA. Approved Cellular and Gene Therapy Products. FDA.gov (2026).