The Immune Cell That Could Make CAR Macrophage Therapy Renewable

For decades the blood system has been read like a one-way street. At the top sits the hematopoietic stem cell, the master cell that can renew itself indefinitely and spin off any blood or immune cell the body needs. Below it, a cascade of progenitors that have already chosen a path, dividing a few times and then committing fully to becoming whatever they were destined to be. A granulocyte-monocyte progenitor, or GMP, is one of these committed cells. It was supposed to be a transient thing, a waystation with a short shelf life.

A team led by USC Stem Cell and Stanford has now grown these supposedly short-lived cells in the laboratory for more than fifty days, expanding a starting pool by roughly eight orders of magnitude, to something on the order of ten trillion cells. The GMPs kept dividing without losing their identity, and they kept their ability to churn out functional immune cells.

That is not how the textbook says committed progenitors behave. “The prevailing view has been that long-term self-renewal in the blood system is primarily a property of the hematopoietic stem cells that can generate any type of blood or immune cell,” says Qi-Long Ying at the Keck School of Medicine of USC, the study’s corresponding author. His team, reporting in Cell, found something the hierarchy was not supposed to allow. “We found that, under the right conditions, GMPs can also self-renew, dividing extensively while keeping their identity and ability to produce functional immune cells,” Ying says.

The point of all this is not really to settle a debate in stem cell biology, though it does poke a hole in one. The point is macrophages.

Macrophages, the immune system’s big eaters, are appealing for cancer therapy because of what they naturally do. They crawl into tumors, engulf cells, and reshape the local immune response, and unlike the T-cell therapies that have transformed treatment for blood cancers, they might actually work against solid tumors. The trouble is that mature macrophages make terrible drugs. They barely multiply outside the body, they resist genetic engineering, freezing damages them, and once infused they pile up in the lungs and liver instead of going where they’re needed.

So first author Shi Yue and colleagues did something slightly counterintuitive. Rather than fight the mature cell, they stepped one rung back up the ladder, to the progenitor that makes it.

A CAR trick, and an enzyme nobody was watching

The cells expand in a defined chemical cocktail the team calls SCF/2i, built around stem cell factor and two small molecules, and the key, it turned out, was an old enzyme caught in a new act. One of the compounds, SKL2001, had been filed away as a Wnt-pathway tweaker. It is not, at least not here. When the researchers went hunting for what the molecule actually grabbed onto inside the cell, the dominant catch was myeloperoxidase, or MPO, the enzyme granulocytes use to make bleach to kill bacteria. Inhibit MPO, and the proliferative brake on the progenitor comes off. Two unrelated MPO inhibitors, drugs already developed for other purposes, reproduced the same sustained growth, which is the sort of independent confirmation that makes a mechanism credible. Getting human cells to behave took extra work and a custom-built analog, but cord blood, bone marrow, and mobilized blood all eventually expanded, and whole-genome sequencing after 28 days turned up no detectable jump in mutation burden, an encouraging if preliminary sign for anything aimed at the clinic.

And the expanded cells, unlike their grown-up descendants, take genetic engineering beautifully. They accept a chimeric antigen receptor, the CAR that lets an immune cell recognize a specific tumor marker, and they tolerate being engineered more than once, which matters if you want to stack features. The researchers think this is a property of the developmental stage rather than the species: progenitor cells are simply more pliable than the finished article, in mouse and human alike.

Injected into mice, the engineered GMPs did what mature macrophages never manage. They settled into the bone marrow and other blood-forming niches and set up shop, producing a rolling supply of engineered macrophages and granulocytes that could be detected months later. Where infused macrophages lodged in the lungs within a day and vanished within a week, the progenitors persisted for 150 days and more. In mice with CD19-positive leukemia, anti-CD19 CAR-GMPs kept six of ten treated animals nearly disease-free out to day 80; all ten untreated controls were dead by day 38. The same trick worked on HER2-positive solid tumors.

There is an obvious catch, and the team has not hidden it. Macrophages can kill what they eat, but durable cancer control usually needs the host’s own immune system pulled into the fight, especially the dendritic cells that carry tumor fragments to lymph nodes and rouse T cells. A donor macrophage acting alone is a closed shop.

Teaching one cell to recruit a crowd

To pry the shop open, the researchers bolted an antibody tail, an IgG Fc domain, onto the CAR. The modified receptor, CAR-Fc, still lets the engineered cell eat its target directly, but it also flags that target for the host’s own scavenger cells, which pick it up, process it, and route the tumor’s identity into the body’s broader immune machinery. Crucially this worked even when donor and recipient were immunologically mismatched, the situation that would arise with an off-the-shelf product made in advance from donor cells rather than built one patient at a time. In immunocompetent mice with mismatched tumors, CAR-Fc outperformed the conventional version: against a disseminated lymphoma, five of nine treated animals lived past day 60, and in an aggressive breast cancer model median survival roughly doubled, with more activated T cells crowding into the tumors. Ravi Majeti at Stanford, whose lab independently reproduced the expansion and engineering, reckons the approach “opens the door to numerous translational applications, much like T cell expansion and engineering.”

The therapy beyond cancer is worth a mention too. In mice with chronic granulomatous disease, an inherited disorder that cripples the immune system’s ability to fight bacteria, infusing the progenitors restored the missing defense, clearing infections that the animals could not otherwise survive.

None of this is a clinic-ready treatment, and the limitations are real ones. The human cells were tested only in mice without a working adaptive immune system, so the human data cannot yet speak to how memory or rejection would play out; the work covers just two target antigens; and in mismatched hosts the donor cells were cleared faster, though the team argues that a therapy that fades may actually be safer for risky targets, where a cell engineered to persist for years is a liability you cannot recall. Several of the researchers, Ying and Majeti among them, have founded a company around the platform, the usual sign someone intends to test whether the mouse results survive contact with people. What the study mostly does is reframe a design question. “Our study suggests that the future of immunotherapy may depend not only on designing better CAR receptors, but also on choosing the right developmental stage of the cell,” says Ying.

Which leaves the field with a slightly different map than it had a week ago. The dead-end cell, it seems, was only ever waiting for someone to release the brake.

Frequently Asked Questions