A growing number of biotechnology companies are racing to eliminate the most cumbersome step in CAR-T cell therapy: the weeks-long, patient-specific manufacturing process that happens outside the body. Instead, these companies aim to reprogram a patient’s own immune cells directly inside the body using a single injectable product, turning what is currently a bespoke, expensive hospital procedure into something closer to an off-the-shelf drug. The approaches range from engineered lentiviral vectors to lipid nanoparticles carrying mRNA or circular RNA, and the therapeutic targets now extend well beyond cancer into autoimmune disease and even organ fibrosis.
Why the Field Is Moving Toward In Vivo Engineering
Conventional CAR-T therapy works remarkably well for certain blood cancers, but the process to make it is punishing. A patient’s T cells are collected, shipped to a specialized facility, genetically modified, expanded over days to weeks, and then shipped back and infused. The whole cycle can take three to five weeks, and the cost regularly exceeds $400,000 per patient before hospitalization. For patients with fast-moving cancers, that timeline can be a death sentence. And for healthcare systems in lower-income countries, the price alone makes the therapy inaccessible.1PubMed Central. From ex vivo to in vivo chimeric antigen T cells manufacturing: new horizons for CAR T-cell based therapy
In vivo CAR-T engineering flips this model. Rather than modifying cells in a lab, you inject a delivery vehicle, whether a virus or a nanoparticle, directly into the patient. That vehicle finds T cells in the bloodstream, enters them, and delivers the genetic instructions to build a chimeric antigen receptor on their surface. The patient’s own body becomes the manufacturing facility. Because the drug product itself has defined physical and chemical properties and is not unique to each patient, a single production run can serve many patients, which dramatically simplifies manufacturing, shortens supply chains, and could eventually bring costs down to a fraction of what they are today.2Trends in Pharmacological Sciences. In Vivo CAR-T Companies: A Deep Dive Into Current Advances
Lentiviral Vector Platforms
Lentiviral vectors have been the workhorse of conventional ex vivo CAR-T manufacturing for years, so it is not surprising that several companies have adapted them for in vivo use. The key engineering challenge is specificity: a lentiviral vector injected into the bloodstream needs to find and enter T cells while ignoring other cell types. Companies solve this by displaying targeting molecules on the viral surface, essentially molecular zip codes that direct the vector to T cells.
Umoja Biopharma built its VivoVec platform around this idea. Their lead candidate, UB-VV100, displays an anti-CD3 targeting molecule on the lentiviral surface so it binds T cells selectively. Once inside, the vector delivers a genetic payload encoding an anti-CD19 CAR for targeting B-cell cancers, along with a clever addition: a rapamycin-activated cytokine receptor system, abbreviated RACR. The RACR system is designed to let clinicians support the expansion and survival of the new CAR-T cells by administering rapamycin, without needing the harsh lymphodepleting chemotherapy that standard CAR-T regimens require.3PubMed Central. Preclinical proof of concept for VivoVec, a lentiviral-based platform for in vivo CAR T-cell engineering Umoja has also advanced UB-VV111, a related anti-CD19 VivoVec product that has been through nonclinical toxicology, biodistribution, and pharmacokinetic studies, the standard preclinical package needed to support entry into human trials.4Blood. Nonclinical Toxicology, Biodistribution, and Pharmacokinetics of UB-VV111, an In Vivo Anti-CD19 CAR T Cell Therapy
Interius BioTherapeutics (now part of the IN8bio pipeline) has taken a different angle with its lentiviral platform. Their INT2104 candidate uses a novel detargeted viral fusogen combined with a CD7-binding molecule. CD7 sits on both T cells and natural killer (NK) cells, which means a single intravenous dose can generate both CAR-T and CAR-NK cells simultaneously. In preclinical studies in mice and cynomolgus monkeys, INT2104 produced both cell types and depleted CD20-positive B cells after a single injection.5PubMed Central. Targeted in vivo delivery of genetic medicines utilizing an engineered lentiviral vector platform results in CAR T and NK cell generation Recruiting NK cells alongside T cells is potentially significant because NK cells are better at infiltrating solid tumors, a setting where conventional CAR-T therapy has mostly struggled.
A separate academic effort used a Sindbis virus envelope engineered onto a lentiviral backbone to achieve high specificity for human T cells while avoiding uptake by other immune cell types.6Scientific Reports. A targeting lentiviral vector for generation of CAR-T cells in vivo This kind of basic research feeds the commercial pipeline, even when the specific vector has not yet been picked up by a named company.
The defining feature of lentiviral delivery is permanence. When a lentivirus inserts its genetic cargo into a T cell’s DNA, that cell and all its descendants carry the CAR gene. This integration-based expression supports sustained tumor control, which matters in cancers where relapses can happen months or years later.7Translational Insights. In Vivo CAR-T Cell Engineering: Translating Vector Pharmacology into Therapeutic Performance But permanence also introduces risks, as we will see in the safety section below.
Lipid Nanoparticle Platforms
The success of mRNA COVID vaccines put lipid nanoparticles on the map, and several companies have repurposed the technology for in vivo CAR-T engineering. The basic idea: wrap CAR-encoding mRNA inside a tiny fat bubble decorated with a targeting antibody, inject it intravenously, and let the nanoparticle find and transfect T cells. Because mRNA degrades within days, the CAR expression is transient. That is a limitation for cancer, where you want long-lived killer cells, but it can be an advantage for diseases where you only need a short burst of immune activity.
Capstan Therapeutics (acquired by Roche in 2024) developed targeted LNPs, or tLNPs, specifically engineered to avoid the liver, which normally soaks up conventional nanoparticles. By functionalizing their LNPs with a T cell-specific antibody, Capstan demonstrated specific delivery of CAR mRNA to T cells and generation of functional anti-tumor CAR-T cells in vivo.8Journal for ImmunoTherapy of Cancer. In vivo engineering of CAR T cells using a novel targeted LNP-mRNA technology A peer-reviewed study from the same research group showed that these tLNPs could reprogram CD8-positive T cells in both healthy donors and autoimmune patient samples, and that in vivo dosing controlled tumors in humanized mice and depleted B cells in cynomolgus monkeys.9PubMed. In vivo CAR T cell generation to treat cancer and autoimmune disease
Which surface receptor makes the best target for getting nanoparticles into T cells? A systematic comparison tested antibodies against CD2, CD4, CD5, CD7, CD8, and a CD4-plus-CD8 combination under identical conditions. CD7-targeted LNPs came out on top for mRNA delivery efficiency and were the best at generating functional anti-CD20 CAR-T cells in vivo in humanized mice.10PubMed. Rapid receptor internalization potentiates CD7-targeted lipid nanoparticles for efficient mRNA delivery to T cells and in vivo CAR T-cell engineering The reason seems to involve how quickly the targeted receptor gets pulled into the cell after the nanoparticle binds it, a process called receptor internalization. CD7 internalizes rapidly, which drags the nanoparticle inside before it can detach.
Orna Therapeutics has taken a distinctive twist on the LNP approach by replacing standard mRNA with circular RNA, which the company brands as oRNA. Circular RNA lacks the exposed ends that cellular enzymes use to degrade linear mRNA, so it persists longer inside cells and can drive protein expression for an extended window. Orna’s panCAR platform combines this circular RNA with a proprietary immunotropic LNP formulated to deliver cargo to immune effector cells including both T cells and NK cells. In non-human primates, an anti-CD20 panCAR achieved B cell depletion, and the company positions the platform as transient and re-dosable, meaning patients could receive repeat injections as needed rather than relying on a single infusion to produce lifelong CAR-T cells.11Blood. In Vivo panCAR-Mediated Depletion of B Cells in Non-Human Primates Using a Circular (oRNA®) Anti-CD20 CAR Orna has also reported preclinical work on an anti-BCMA panCAR for multiple myeloma, showing dose-dependent CAR expression on immune cells that persisted for at least 72 hours in vitro.12Blood. In Vivo pan CAR therapy utilizing circular RNA for treatment of multiple myeloma
Polymeric Nanoparticles and Engineering Beyond T Cells
Not all non-viral delivery relies on lipid nanoparticles. Several research groups have built polymeric nanoparticle platforms using biodegradable polymers that self-assemble with mRNA through electrostatic interactions. One well-studied polymer, poly(beta-amino ester) or PBAE, was originally developed for brain cancer treatment and has since been adapted for T cell delivery. By coupling anti-CD8 antibodies to the nanoparticle surface, researchers achieved targeted mRNA delivery to circulating T cells in vivo, producing transient CAR expression.13Nature Communications. In vitro-transcribed antigen receptor mRNA nanocarriers for transient expression in circulating T cells in vivo
A more recent polymeric approach conjugated T cell-activating ligands directly to biodegradable mRNA nanoparticles, so the particles not only delivered the CAR gene but also stimulated the T cells as they transfected them. In healthy mice, these particles delivered anti-CD19 CAR mRNA in vivo and depleted roughly 95% of B cells in peripheral blood and about half of splenic B cells.14PubMed Central. Biodegradable targeted polymeric mRNA nanoparticles enable in vivo CD19 CAR T cell generation and lead to B cell depletion Those are striking numbers for a non-viral platform, and the built-in activation step addresses a real bottleneck: T cells that receive a CAR gene but do not get activated tend to sit idle rather than attacking their targets.
Some groups are looking beyond T cells entirely. Macrophages are immune cells that naturally infiltrate solid tumors, something T cells often struggle to do. Researchers have created nanocomplexes using macrophage-targeting nanocarriers loaded with plasmid DNA encoding both a CAR and the cytokine interferon-gamma. Injected intravenously, these nanocomplexes programmed macrophages in vivo into what the team called CAR-M1 macrophages, cells capable of directly eating cancer cells through CAR-mediated phagocytosis while also reshaping the tumor’s immune environment. In mouse models, this approach inhibited solid tumor growth.15PubMed. Nanocomplex-Mediated In Vivo Programming to Chimeric Antigen Receptor-M1 Macrophages for Cancer Therapy If this translates clinically, it could open a path to treating solid tumors that have resisted every CAR-T approach tried so far.
Autoimmune Disease, Fibrosis, and Other Non-Cancer Uses
One of the most exciting implications of in vivo CAR-T engineering is how naturally it fits diseases where you want temporary, repeatable immune interventions rather than permanent tumor-killing machinery. Autoimmune diseases are the obvious frontier.
In a mouse model of systemic lupus erythematosus, researchers used CD5-targeted lipid nanoparticles carrying anti-CD19 CAR mRNA to generate CAR-T cells in vivo. These cells depleted pathogenic B cell subsets in the spleen, reduced autoantibodies and inflammatory cytokines, and reversed kidney and skin damage, all without lymphodepleting preconditioning or any ex vivo cell work. The mice achieved sustained remission.16PubMed. In vivo generation of CD19 CAR T cells for the treatment of mouse systemic lupus erythematosus This result matters because lupus patients are often immunocompromised to begin with, making the standard lymphodepletion step of conventional CAR-T particularly risky for them.
Cardiac fibrosis represents another frontier. Separate research teams have shown that LNP-delivered mRNA can generate short-lived anti-fibrotic CAR-T cells in vivo, cells that home to injured heart tissue and reduce fibrotic burden in mouse models.17Genes & Diseases. In vivo CAR-T for the treatment of autoimmune diseases: Current progress and future directions This is an area where transient expression is actually ideal: you want to clear the fibrotic tissue, not maintain a standing army of killer cells in the heart indefinitely.
A particularly creative application targets senescent cells, the “zombie cells” that accumulate with age and drive chronic inflammation. Using a novel cardiolipin-mimic LNP, one group delivered mRNA encoding a CAR targeting uPAR, a protein found on senescent and inflammatory cells. In mouse models, the resulting CAR-T cells alleviated liver fibrosis and rheumatoid arthritis.18Cell Reports Medicine. In vivo T cell engineering and senolytic CAR-T therapy using cardiolipin-mimic lipid nanoparticles The senolytic angle is worth watching because the target population, people with age-related fibrosis or arthritis, is vastly larger than the cancer populations where CAR-T got its start.
Transient Versus Permanent Engineering and the Persistence Question
The choice of delivery platform shapes more than manufacturing logistics. It determines whether the engineered cells last days or years. Lentiviral vectors integrate the CAR gene into the T cell’s DNA, so the modification is inherited through every cell division and can persist for over a decade after infusion.19Blood Immunology & Cellular Therapy. In vivo generation of CAR T cells: biology, delivery platforms, clinical promise, and translational challenges LNP-delivered mRNA, by contrast, produces CAR protein only until the mRNA degrades, typically a matter of days. Circular RNA from platforms like Orna’s lasts somewhat longer than linear mRNA but is still transient.
For cancer, long persistence is generally what you want. Relapses can occur months or years after initial treatment, and having CAR-T cells on patrol provides ongoing surveillance. For autoimmune diseases, temporary CAR expression may be preferable because it allows a self-limiting intervention: the CAR-T cells deplete the problematic B cells, the mRNA fades, and normal immune reconstitution follows. If the disease returns, the patient can simply receive another dose. This re-dosability is a major selling point for nanoparticle-based approaches in non-oncology settings.
Some researchers are trying to bridge the gap. One strategy uses the piggyBac transposon system, where the CAR gene is flanked by specific DNA sequences and co-delivered with a transposase enzyme. The transposase cuts the CAR gene out of the delivered DNA and pastes it into the T cell’s genome, achieving stable integration without a viral vector.20Trends in Pharmacological Sciences. In Vivo CAR-T Companies: A Deep Dive Into Current Advances This hybrid approach could, in theory, combine the manufacturing simplicity of nanoparticles with the durability of viral integration.
Safety Challenges Specific to In Vivo Delivery
In vivo CAR-T engineering inherits some safety concerns from conventional CAR-T therapy, including cytokine release syndrome and neurotoxicity, while introducing new ones that are unique to the delivery vehicles themselves.21PubMed Central. In vivo engineering of CAR-T cells: delivery strategies and clinical translation
For lentiviral vectors, the biggest theoretical worry is off-target transduction. If the vector accidentally integrates its genetic cargo into a non-T cell, or into a germ cell, the consequences could include unintended heritable genetic changes or insertional mutagenesis, where the inserted gene disrupts a tumor-suppressor gene or activates an oncogene. These risks are theoretical rather than observed in current preclinical models, but regulators take them seriously, and they are a major reason why targeting specificity gets so much engineering attention.19Blood Immunology & Cellular Therapy. In vivo generation of CAR T cells: biology, delivery platforms, clinical promise, and translational challenges
LNPs bring a different set of issues. The innate immune system can recognize the nanoparticle itself as foreign material, triggering inflammatory responses independent of the CAR payload. Repeated dosing, which is the whole advantage of transient platforms, raises questions about whether the immune system will mount stronger reactions with each injection. Vector immunogenicity is a known challenge in gene therapy more broadly, and in vivo CAR-T platforms will need to demonstrate that repeat dosing remains safe and effective over time.
Both approaches also face the fundamental question of dose control. In ex vivo manufacturing, clinicians know exactly how many CAR-T cells they are infusing. With in vivo engineering, the number of T cells that actually get transduced depends on variables that are harder to control: the patient’s T cell count, their immune status, how much of the vector reaches the bloodstream versus getting trapped in the liver or spleen. This unpredictability makes dosing more complex and could lead to under- or over-engineering of the immune response.
Built-In Safety Switches
Given that in vivo platforms introduce genetic instructions directly into a patient’s body, there is intense interest in engineering “off switches” that let clinicians shut down CAR-T cells if something goes wrong. Several strategies are under active investigation.
Small-molecule safety switches are the most advanced concept. In one design, the CAR itself is split into two inactive subunits that only come together in the presence of a specific drug. Without the drug, the T cell carries the CAR components but they cannot signal. When the drug is administered, the subunits dimerize and the CAR becomes functional. Remove the drug and the CAR goes quiet again. This gives clinicians a dial they can turn, not just an on/off switch but a way to titrate activity.22PubMed Central. Enhancing the safety of CAR-T cell therapy: Synthetic genetic switch for spatiotemporal control
Multiple drug systems are being explored for this purpose, including rapamycin-based dimerization (which Umoja’s RACR system also leverages for cell expansion), lenalidomide-based switches, and even dasatinib, a cancer drug that happens to reversibly suppress T cell signaling. Other approaches use molecules like FITC or folate as the key that activates a split CAR system, and proteolysis-targeting chimeras (PROTACs) have been investigated to degrade the CAR protein on demand.23Journal of Medicinal Chemistry. Optimization of CAR‑T Cell-Based Therapies Using Small-Molecule-Based Safety Switches Safety switches matter more for in vivo platforms than for ex vivo therapy because the manufacturing happens inside the patient. If something goes wrong during ex vivo production, you can catch it before infusion. With in vivo engineering, the cells are already in the patient’s body when they start expressing the CAR, so the ability to shut them down quickly becomes a genuine clinical necessity rather than a nice-to-have feature.
How CAR-Macrophages Could Change the Solid Tumor Problem
Nearly all the in vivo CAR-T platforms described so far target blood cancers or autoimmune diseases, conditions where the target cells circulate freely or reside in accessible lymphoid tissues. Solid tumors remain the hard problem. T cells have trouble physically entering many solid tumors, and even when they do, the tumor microenvironment suppresses their activity through a web of immunosuppressive signals.
In vivo CAR-macrophage engineering offers a conceptually different attack. Macrophages are among the most abundant immune cells within solid tumors. They get in easily, but in most cancers they are co-opted by the tumor into supporting its growth rather than fighting it. The nanocomplex approach that programs macrophages into pro-inflammatory CAR-M1 cells tries to flip this dynamic: the macrophages keep their natural tumor-infiltrating ability but gain a CAR that directs them to eat cancer cells, plus interferon-gamma expression that shifts them into an anti-tumor mode.15PubMed. Nanocomplex-Mediated In Vivo Programming to Chimeric Antigen Receptor-M1 Macrophages for Cancer Therapy Whether this approach will work in human solid tumors, where the immunosuppressive environment is far more complex than in mouse models, remains an open question. But the in vivo delivery format makes it feasible to try: you do not need to collect a patient’s macrophages, engineer them in a lab, and hope they find their way back to the tumor. The nanoparticles do the targeting and the engineering in one step, inside the body where the macrophages already live.