Refining the molecular blueprint of in vivo CAR-T

Following the first commercial approvals in 2017, the chimeric antigen receptor T-cell (CAR-T) therapy market has expanded rapidly thanks to the significant impact these treatments have had on patient outcomes in hematologic cancers. There are now more than 10 CAR-T cell therapeutics approved globally, and the market is valued at approximately $7 billion¹ and growing. Projected market growth is anticipated to follow the administration of CAR-T cell therapy earlier in the patient’s treatment journey, as well as expansion of target indications from oncology into autoimmune conditions.

Reprogramming a patient’s T-cells in vivo is a promising approach to transforming the accessibility of these life-saving therapies to a larger patient population. This offers several potential advantages: patients could be treated in a single hospital visit; treatments would be available ‘off-the-shelf’; and manufacturing processes would be significantly less complicated and more cost effective without the need to culture, transduce, purify, test and re-infuse the patient’s cells.

Manipulating tropism for precision targeting

The most commonly pseudotyped lentiviral envelope protein, Vesicular Stomatitis Virus Glycoprotein (VSV-G) targets the low-density lipoprotein receptor (LDLR), which is found on the surface of most mammalian cell types. This pantropism is of little concern ex vivo, when the target cells can be brought into close contact with the viral vector, but it becomes a serious concern in vivo, where precise tissue targeting is critical for safety and efficacy. Indeed, it is essential that the viral vector does not transduce every cell it encounters, and instead reserves its cancer-fighting payload for the T-cells it’s intended to reprogram. Ironically, since resting T-cells express very low levels of LDLR, VSV-G is naturally inefficient at transducing these cells.

To re-engineer tropism, lentiviral vectors can be ‘dressed’ in a new envelope protein that has a different mechanism of viral entry and naturally targets a different cell type. Measles virus, for example, can infect lymphocyte cell lines via the SLAM receptor.² However, while this is a narrower, and more relevant, tropism for CAR-T cell therapies, it isn’t precise enough; viral vectors re-engineered to target SLAM would still transduce B cells, dendritic cells, macrophages and NK cells; potentially also crossing into the bone marrow to transduce hematopoietic stem cells.

In fact, while T-cells can be targeted by wild-type viruses, for example HIV, there isn’t a known virus with a natural tropism specific enough that it can be used for in vivo CAR-T cell modification without further engineering to blind the envelope protein to its natural receptor target. Engineering of an envelope protein is a common way to drive cell specific targeting. For example, the measles viral envelope can be blinded by truncating the targeting protein (H; Hemagglutinin), without affecting the ability of the fusion (F) protein to enter the target cell. Adding a single chain antibody fragments (scFv) that binds a T-cell specific receptor can target the envelope protein to the desired T-cell population.³ A similar approach to lentiviral pseudotyping and targeting has been demonstrated with engineered Nipah viral protein^4,5 and VSV-G.⁶

However, as with many things in biology, further investigation uncovers additional complications. For example, biophysical properties of scFv can cause problems with lentiviral vector purification, dosing and immune reactivity in the patient.7 On the other hand, using a CD8 targeting DARPin (Designed Ankyrin Repeat Protein) allows higher titre production than engineering with the equivalent CD8-targeting scFv, as well as superior transduction of cytotoxic T-cells in vitro and in vivo.⁸

Engineering biological safety mechanisms

Once the lentiviral envelope is engineered to specifically target T-cells for transduction, the question becomes whether this is enough to reduce the risk of off-target CAR expression in an in vivo therapy. So far, the general scientific consensus seems to be ‘why risk finding out?’

The addition of tissue specific promoters to the CAR construct represents a ‘belt and braces’ approach to biological safety, exemplified by EsoBiotec’s first-in-human trial of their ESO-TO1 in vivo CAR-T cell therapy early last year. Their anti-B-cell maturation antigen (BCMA) CAR construct includes an anti-CD3 T-cell receptor nanobody for T-cell targeting, and a T-cell specific synthetic promoter driving expression of the CAR payload.^9,10 As well as adding an extra layer of biological safety, ongoing research is exploring whether the addition of tissue specific promoters can also enhance efficacy. CAR-T cell therapies have been extremely effective in treating blood-based malignancies, but have so far shown little promise in solid tumours. A recent paper investigates whether the addition of a lactic-acid responsive promoter can be used to overcome the suppression of CAR-T cell activity caused by the acidity of the tumour microenvironment, while also reducing on-target, off-tumor effects by only activating CAR transcription in highly acidic conditions. Their LAR CAR T-cells demonstrated comparable tumor eradication and superior safety profiles compared to conventional CAR T-cells in in vivo mouse models.¹¹

Whether administered ex vivo or in vivo, CAR-T cell treatment still confers unavoidable, and potentially fatal, side effects including cytokine release syndrome, tumor lysis syndrome and on-target off-tumor toxicity.¹⁰ While it may not be possible to avoid these side effects, it is possible to engineer a molecular ‘emergency brake’ into the CAR construct, so that should side effects become intolerable or life-threatening, or unexpected toxicities occur, treatment can be stopped. This may be especially important for in vivo CAR-T cell therapies, where the risk of uncontrolled T-cell production is somewhat higher than for ex vivo approaches. Examples of such molecular brakes include:

1. Inducible Caspase 9 (iCasp9): A fast-acting molecular safety switch made by fusing human caspase 9, modified to remove its natural dimerization domain, to the FKBP molecule, a receptor for the immunosuppressant drug FK506 engineered to bind a small molecule chemical inducer of dimerization (CID). When a CID is administered, the drug-binding domains of the chimeric ‘switch’ protein crosslink, dimerizing caspase 9 and activating downstream expression of caspase3, triggering T-cell apoptosis.
2. Herpes simplex virus 1-derived thymidine kinase (HSV-TK): Engineering HSV-TK into the CAR construct renders modified T-cells susceptible to the antiviral drug ganciclovir. Upon administration of ganciclovir, HSV-TK catalyses its phosphorylation into a potent active metabolite that acts as a structural analogue to natural nucleotides, interfering with DNA synthesis and eventually killing the cells.
3. Co-expression with truncated epidermal growth factor receptor (EGFRt): Including the surface antigen EGFRt within the CAR construct renders the CAR-T cells susceptible to antibody dependent cell-mediated toxicity (ADCC) upon administration of the monoclonal antibody cetuximab.^12,13

While the risk of insertional mutagenesis from lentiviral vectors is typically low, and the use of self-inactivating viral backbones reduces this further, the risk-reward ratio changes when CAR-T cell therapies stop being considered an ‘end of line’ treatment or are used to treat non-fatal conditions. At this point any risk of insertional mutagenesis is unacceptable.

Non-integrating lentiviruses may offer a solution. However, while the risk of insertional mutagenesis is eliminated, so is the potential to maintain expression of the transgene through cell division. The holy grail of safe lentiviral vector design is a non-integrating lentivirus whose transgene expression can still be maintained throughout multiple cell divisions as the activated CAR-T cells divide as they fight the cancer.

Several viruses, including Epstein Barr Virus, Kaposi’s sarcoma associated herpesvirus and human papilloma virus achieve this naturally, persisting as multicopy episomes; circular genomes that are assembled into chromatin with histone and DNA modifications like host genomes.¹⁴ Epstein Barr virus carefully maintains episome copy number through cellular generations using a viral protein to tether the viral origin of plasmid replication (OriP) to the host cell’s metaphase chromosomes. It then recruits host cell DNA replication machinery to synchronize replication of the episome exactly with the host cell cycle.¹⁵ Researchers at ViroCell Holdings are currently working on the identification of novel regulatory elements and sequence motifs to mimic this persistence with non-integrating lentiviruses, an endeavor in which they would welcome additional scientific collaboration.¹⁵

Purity and potency: The backbone of manufacturability

Residual host cell proteins or DNA can be ‘washed off’ reprogrammed T-cells prior to re-infusion with an ex vivo therapy, but there’s no such opportunity for clean-up of the vector prior to injection of an in vivo therapeutic. This means the manufacturing process must be optimized for production of mature, functional lentiviral vectors and to minimize production of non-functional byproducts.

The starting point for vector purity once again begins at the molecular level. Assembly of mature lentiviral particles requires the Gag protein to recognize and interact with a specific sequence on the RNA genome, carrying it to the cell membrane for budding. This process is not always executed perfectly; lentiviral particles may form that are only partially packaged, completely empty or contain extraneous cellular RNA. Simultaneously, extracellular vesicles are also likely to bud from the cell, with or without viral RNA inside, and most likely with viral envelope proteins on the EV surface.¹⁶

Some of this incorrect packaging is caused by aberrant splicing of the lentiviral genome, driven by interactions between a strong splice donor site adjacent to the 5’ LTR in the psi packaging signal; and a weak or cryptic splice acceptor site, commonly found in standard promoters that drive expression of the CAR in transduced cells. This can cause unwanted breakthrough expression of the CAR, which can end up on the surface of the manufacturing cell. From there the lentivirus can become contaminated with the CAR. This has the potential to negatively affect vector potency, as the CAR-‘pseudotyped’ lentivirus can bind to the CAR-T target cells (e.g. CD19 +ve B-cells) as well as the intended T-cell target. Such CAR contamination may also impact vector safety, especially in vivo. Lentitek are tackling this problem using a proprietary promoter that sits upstream of the 5’ LTR and produces a viral genome that has been demonstrated as being resistant to splicing, and therefore the unwanted expression of the CAR payload during lentivirus production.¹⁷

To add an additional layer of complication, genetic modifications that improve lentiviral vector safety and precision may simultaneously reduce the number of functional lentiviral particles produced. Both lentiviral pseudotyping and increased transgene sizes have been associated with reduced functional titer.¹⁶ Therefore, additional engineering of the Gag gene may be necessary to optimize production of properly packaged lentiviral vectors.

How digital lab management enables complex workflows

There is no question that designing a lentiviral vector encoding a CAR construct for in vivo use is a major feat of genetic engineering. It’s also clear that there is going to be no single ‘right’ way to go about it; every approach will have to balance improvements in precision and/or safety, with potential decreases in functional titer or other complications. Research teams involved in this endeavor are going to need to collaborate closely, and work through multiple iterative rounds of trial and error to identify the optimal vector design. Plasmids are likely to be sourced and constructed from institutions around the world, making careful annotation necessary to keep track of provenance, usage and outcomes.

This is where digital lab management technology can help by providing a robust digital infrastructure for scientific discovery, project management and collaboration. This is also an area where new IP is constantly being generated, and here again, digital records and timestamped records of relevant discoveries can be fundamental in proving novelty and priority.

Advanced molecular biology software facilitates complex genetic engineering. Tools like Lab Thread’s ‘in sequence’ commenting functionality also allows detailed annotation, as well as sequence sharing to streamline collaboration. Using a software that’s designed to connect each design to its experimental outcome and sample location can help manage the assessment of each genetic variable involved in complex plasmid engineering, streamlining the process of iterative validation to reach an optimized construct design. Meanwhile, if you can link each molecular design to a complete digital record of where it has been used — which experiments, cell lines, bioreactor runs etc. — it makes it infinitely easier to assemble a fully traceable regulatory package or patent specification when the time comes. Adding in commenting, feedback and collaboration tools at sequence, design, task and project levels facilitates ready communication within and between teams, while big picture project management tools support the entire program.

Laboratory management software can’t build a perfectly optimized construct for an in vivo CAR-T cell therapy, but it can make the lives of the scientists doing so considerably easier through detailed molecular design, collaborative project management, easily accessible data, and comprehensive sample tracking. 

References

1. MarketsandMarkets. (2026, Feb). CAR T-Cell Therapy Market.
2. Duvergé, A. and Negroni, M. (2020 Nov). Pseudotyping Lentiviral Vectors: When the Clothes Make the Virus. Viruses. 12(11): 1311.
3. Funke, S., et. al. (2008 Aug). Targeted Cell Entry of Lentiviral Vectors. Mol Ther. 16(8): 1427-36.
4. Mhaidly, R. and Verhoeyen, E. (2019 Apr). The Future: In Vivo CAR T Cell Gene Therapy. Mol Ther. 27(4): 707–709.
5. Pfeiffer, A., et. al. (2018 Nov). In vivo generation of human CD19-CAR T cells results in B-cell depletion and signs of cytokine release syndrome. EMBO Mol Med. 10(11): e9158.
6. Andorko, JI., et. al. (2025 Oct). Targeted in vivo delivery of genetic medicines utilizing an engineered lentiviral vector platform results in CAR T and NK cell generation. Mol Ther. 33(10): 4937–4952.
7. Friedel, T., et. al. (2015 Apr). Receptor-targeted lentiviral vectors are exceptionally sensitive toward the biophysical properties of the displayed single-chain Fv. Protein Eng Des Sel. 28(4): 93–106.
8. Frank, AM., et. al. (2020). CD8-Specific Designed Ankyrin Repeat Proteins Improve Selective Gene Delivery into Human and Primate T Lymphocytes. Hum Gene Ther. 31(11-12): 679–691.
9. Xu, J., et. al. (2025 Jul). In-vivo B-cell maturation antigen CAR T-cell therapy for relapsed or refractory multiple myeloma. The Lancet. 406: 228–231.
10. An, N., et. al. (2026). In vivo generation of anti-BCMA CAR-T cells in relapsed or refractory multiple myeloma: a phase 1 study. Nat Med.
11. Feng, M., et. al. (2026 Feb). Harnessing tumor acidity: innovative lactic acid-responsive promoter enables precision control of CAR-T cell activity in solid tumors. J Immunother Cancer. 14(2): e013672.
12. Moghanloo, E., et. al. (2021 Jun). Remote controlling of CAR-T cells and toxicity management: Molecular switches and next generation CARs. Transl Oncol. 14(6): 101070.
13. Głowacki, P. and Rieske, P. (2022 Jun). Application and Design of Switches Used in CAR. Cells. 11(12): 1910.
14. De Leo, A., et. al. (2020 Feb). Control of Viral Latency by Episome Maintenance Proteins. Trends Microbiol. 28(2): 150–162.
15. Faraneh, F. and Ostrout, N. (2025 Jul). De-Risking Cell and Gene Therapies with Innovative Solutions: A Review for Leveraging a Proven Workhorse Technology in New Ways. Pharma's Almanac.
16. Barbieri, E. and Heldt, CL. (2026 Feb). Challenges and Opportunities in Lentivirus Viral Vector Manufacturing for In Vivo Applications. Biomedicines. 14(2): 369.
17. Science: The Lentitek Platform. Lentitek.com.