In vivo CAR NK cells created with targeted mRNA nanoparticles

In vivo CAR NK cells normally sound like a manufacturing shortcut that has not yet happened: instead of removing immune cells, engineering them in a facility and infusing them back, a patient would receive a formulation that reprograms the right cells directly in the bloodstream. Researchers at MIT have now demonstrated that concept in mice using targeted lipid nanoparticles carrying messenger RNA.

Their best formulations delivered mRNA to as many as roughly 80% of targeted NK cells in mice while producing relatively little off-target transfection.

The more important result, however, was what those cells could be instructed to do. By changing the mRNA cargo, the same nanoparticle platform redirected NK cells toward different immune-cell targets and tumors.

Turning an NK cell into a temporary CAR cell

Natural killer cells are part of the innate immune system. Unlike conventional T cells, they can recognize and destroy stressed, infected or malignant cells without first being trained against one specific antigen.

A chimeric antigen receptor adds another layer of recognition. CARs are synthetic receptors designed to bind a chosen target on another cell and trigger an immune attack.

The best-known version of this idea is CAR T therapy. Producing those treatments is logistically demanding: cells must be collected from a patient, genetically modified, expanded, tested and then infused back.

NK cells may offer a particular advantage when the unwanted cells are themselves T cells. A CAR T cell directed against a broadly expressed T-cell antigen can attack other therapeutic T cells, creating the problem of fratricide. NK cells can target those same T-cell antigens without belonging to the population they are eliminating.

The challenge has been getting genetic instructions specifically into NK cells in vivo.

Yuang Chen and colleagues approached that problem with lipid nanoparticles carrying mRNA. The particles were decorated with engineered binding molecules that engage receptors on NK cells and promote receptor-specific uptake. Once inside, the mRNA serves as a temporary template for production of the desired protein.

Because the cargo is mRNA, the approach provides transient programming rather than permanent insertion of a CAR gene into the genome.

One delivery system, several immune programs

The researchers first tested whether NK cells could be redirected toward B cells. Nanoparticles carrying an mRNA encoding a CD19-specific CAR generated CAR NK cells in vivo and led to complete depletion of circulating B cells in healthy mice.

They then turned to dysfunctional T cells.

In a mouse model of autoimmune hepatitis, autoreactive T cells attack liver tissue. The team programmed NK cells with a CAR recognizing CD3, a T-cell surface molecule. After a single nanoparticle treatment, nearly all autoreactive T cells were eliminated and liver injury improved.

That experiment illustrates an unusual potential use for engineered immune cells: not attacking cancer, but selectively removing another part of the immune system that is driving disease.

The platform was also tested against cancer. In a humanized mouse model of T-cell lymphoma, five nanoparticle administrations eradicated tumors. The outcome was similar to treatment with CAR NK cells manufactured by the conventional ex vivo route.

In another set of experiments, the nanoparticles were used not to install a CAR but to target NKG2A, an inhibitory checkpoint on NK cells. This relieved tumor-mediated suppression and improved NK-cell antitumor activity in a solid-tumor model.

The platform was therefore acting less like a single drug and more like a delivery system for different immune-cell programs.

The manufacturing step moves inside the body

Every therapeutic result in this study came from mice, including humanized mouse models. The work does not establish efficacy or safety in people. Human translation will require much more information about targeting accuracy, dose, durability, repeat administration and immune reactions to the nanoparticle components.

The proof of principle is nevertheless distinct from simply finding another useful CAR target. The researchers showed that NK cells can be selectively programmed with mRNA after systemic administration, without first being isolated from the body.

That changes where the “manufacturing” happens.

In one experiment, the particles instructed NK cells to attack B cells. In another, they removed pathogenic T cells. Elsewhere, they modified an inhibitory NK-cell pathway. The carrier remained modular while the molecular instruction changed.

If this approach ultimately translates to people, one of the most complicated steps in cellular immunotherapy could shift from manufacturing millions of engineered cells in specialized facilities to delivering a short-lived genetic instruction to the right immune cells inside the patient's bloodstream.