Illustration of a pancreas gland with islets ADOBE (STOCK.ADOBE.COM)

During the January 2023 Advanced Therapies Week in Miami, FL, Leonardo Ferreira, an assistant professor from the Medical University of South Carolina, spoke about his team’s work in developing chimeric antigen receptor (CAR) regulatory T cells (Tregs) as living therapeutics for type 1 diabetes. Ferreira has focused on type 1 diabetes since 2016, when he undertook postdoctoral work at the University of California at San Francisco under Qizhi Tang and Jeff Bluestone before continuing his research at his own laboratory in Charleston, SC. He spoke to BPI about his lab’s progress toward developing a therapy for type 1 diabetes in a two-part interview. In next month’s Part 2, Ferreira will discuss the characteristics that make CAR Tregs a good choice for treating Type 1 diabetes.

The Biology of Type 1 Diabetes
Can you tell us about type 1 diabetes and how it affects those who have the disease? Type 1 diabetes is a devastating autoimmune disease. It’s a process in which your body’s own T lymphocytes are recognizing and destroying your pancreatic beta cells. Those cells make the insulin that regulates your blood sugar. If your body doesn’t make insulin, the sugar cannot enter your cells, so instead it stays in the blood and creates problems. When a child is afflicted with the disease, it’s every parent’s nightmare. Children will experience hypoglycemia during their sleep, which can lead to complications such as nerve damage, blindness, general health problems, and even the need for limb amputation.

What are the currently best available treatments for type 1 diabetes, and how effective are those approaches? People live for many decades with type 1 diabetes despite its complications, but treatment for type 1 diabetes helps only to control the symptoms. Patients depend on insulin injections several times per day. They have to monitor their blood glucose levels closely at all times. If the disease goes unattended, such as when patients are sleeping, they can’t inject themselves with insulin. Current technology has helped with continuous glucose monitoring, which connects to a patient’s or caregiver’s phone. So there’s a lot of monitoring, but no cure.

As for the insulin molecule itself, companies such as Eli Lilly make special versions of this protein to control potency and limit side effects by minimizing the amount needed. Before the discovery of insulin (in 1921), people would simply die from having type 1 diabetes. Scientists discovered insulin first in dogs and then in pigs. People would receive injections of pig insulin, which is not the same thing as human insulin, but it worked to a point. Over time, though, a person’s body would create resistance and antibodies to pig insulin, but it was the only available treatment. The field has come a long way in terms of managing the disease, which no longer means a death sentence. Now it is time for a cure.

Today, the curative process involves transplanting a new pancreas, which is composed of 2% endocrine tissue. That is the insulin-producing part that gets destroyed by type 1 diabetes. You don’t want to transplant the whole pancreas, but rather the islets, which contain beta cells. A destroyed pancreas can be replaced by one from a cadaveric donor, but more recently, there is the hope of using stem cells. You can differentiate pluripotent stem cells into functional beta cells and then transplant those.

Although transplantation is possible, doing that alone would be like building a beautiful sandcastle right next to the ocean. The waves will come ashore and just wash it away. You need to make more and better beta cells and protect them from an immune standpoint. Insulin-derived peptides are recognized by T cells and attacked. That’s how the beta cells get killed. When type 1 diabetes patients get a transplant, there are two types of immunity to thwart. Aside from an autoimmune reaction, there is also an alloimmune reaction. If you take a heart, a kidney, a lung, or a pancreas from someone else, your body will recognize it as foreign, resulting in transplant rejection.

Like other organ transplant patients, pancreas recipients have to take immunosuppressive drugs for the rest of their lives. So they have a hard time fighting infections and some types of blood cancers. Vaccines don’t work as well for them. With the currently available COVID-19 vaccines, for example, immunocompromised patients need to take three or four boosters to get some level of response and protection. Your immune system cannot be fully effective because if it were, then yes, it would kill microbes, but it would also kill the beta cells that you just received.

There’s really only one therapy so far that is shown to delay the onset of stage 3 in people newly diagnosed with type 1 diabetes. That’s a monoclonal antibody (MAb) called teplizumab, which binds to T cells. Science-wise, that demonstrates that T cells are involved in initiating the disease because if you disrupt them, the disease takes longer to develop. But you need to take care of the T-cell problem permanently if you want to cure type 1 diabetes.

Some of us are working to make a new class of treatment that will be more specific. We know what we want to shut down, and the first step to solving a problem is to identify it. In this case, the problem is that beta cells are getting killed by immune cells. Can we stop that killing without stopping the immune cells from killing other harmful cells? You still want to be able to kill cancer cells and virus-infected cells. That’s where the work for some of us comes in, and it’s why we choose to use regulatory T cells.

Tregs are a subset of T cells that are not as interested in killing as they are into preventing other T cells from killing, and they represent a new class of therapy. They show promise in stopping the immune reaction against beta cells without stopping reactions against cancer and virus-infected cells.

Recent Advances in Treg Therapy
Can you describe the success that your team’s therapy has had in preclinical models and how those successes might translate to treating disease in humans? What we’ve seen in vitro is exciting because it shows us that our CAR works. We based a whole project on one particular CAR because it is a synthetic molecule. Nature didn’t come up with it, so there was a chance that it might not work at all. And sure enough, some combinations of binding and signaling domains just don’t work together. The protein aggregates don’t get expressed, and the cells die.

The CAR-making process requires protein engineering. We failed to get cell-surface expression for quite a while with some of the CARs we made. We had to go back to the antibody against human leukocyte antigen A2 (HLA-A2) that it came from and take certain sequences from it called complementarity-determining regions (CDRs). Then we grafted those molecularly into a different single chain fragment variable (scFv) backbone in what is basically a transplant of specificity. You have this new DNA sequence that used to recognize x and now it’s going to see y. The y in this case is HLA-A2.

That scFv molecule would bind previously to human epidermal growth factor receptor 2 (HER2). With this development program we wanted to see whether it would bind to HLA-A2 instead. Because it maintains the scaffolding of the anti-HER2 scFv, we hoped that the CAR would work, and in this case it did. That process took a while to figure out, but it could be a useful discovery for the field. If your specific antibody-based CAR is not working, then try to use an existing antibody that you know works, and then try to change the specificity of that antibody to see if that makes it work. We were successful in making a new CAR, and it’s not always easy to do because CARs are designer molecules with many moving parts.

We used clustered regularly interspaced palindromic repeats and associated protein 9 (CRISPR-Cas9) to knock in sequences for a CAR that works in Tregs. We inserted a T-cell receptor (TCR) locus, then put in a grafted anti-HLA-A2 CAR that is suppressed in vitro. In the mouse model, we saw that HLA-A2 CAR Tregs suppress immune responses in vivo in an HLA-A2–dependent way. That uses a graft-versus-host disease model with HLA-A2–expressing cells. We can use disease-causing white blood cells with HLA-A2, the recipient transgenic mouse that has HLA-A2, or both.

The anti-HLA-A2 CAR Tregs suppressed human immune responses when they recognized HLA-A2 in the injected peripheral blood mononuclear cells (PBMCs), the mouse tissue, or both. We also have shown that they work in vivo and that they are trafficking. We induced diabetes in the mice by destroying their beta cells so that the mice depend on an islet transplant for normal glycemic levels. When we inject our HLA-A2 CAR Tregs into a mouse to transplant, the Tregs circulate, find the transplant, and stay there for weeks without damaging tissue. During that time, mice maintain normal glycemic levels.

Blood glucose measurements in those mice remain normal, demonstrating that their islets are functional. We tag CAR Tregs with luciferase, which glows when you add a substrate, allowing you to image the cells inside a living mouse. We saw that the living mice have normal glycemic activity, their islet transplants are healthy and functional, and the CAR Tregs have accumulated. That’s exactly what we want. For a patient with type 1 diabetes, we want CAR Tregs inside a pancreas to address the destruction that is happening there without causing further damage. We have confirmed that in preclinical work. Now in the clinical phase, the field as a whole has shown several times that Tregs are safe. You can inject up to three billion Tregs into a person with type 1 diabetes and get no adverse effects.

So that’s phase 1. For phase 2 trials, some of us think that antigen-specific Tregs will be required rather than just polyclonal, total Tregs. Otherwise, how will they get into only the pancreas without spreading elsewhere? That’s where CAR Tregs come in and also why companies such as Quell Therapeutics, Sonoma Biotherapeutics, GentiBio, and Kyverna Therapeutics are showing interest. As an academic, I think it’s a good sign to see company involvement, because that represents a real belief that this technology can progress beyond laboratory scale.

I don’t have any shares in any of those companies, but I am excited that they are taking the next steps. In the meantime, I’m working on how to make better CAR Tregs and how to understand CAR signaling better. Some basic biology questions must be answered because the CAR Treg is a Frankenstein cell. CAR Tregs do not look exactly like TCR natural Tregs. We want to foresee possible problems with inflammation, instability, and lack of functionality. Companies are recruiting patients, so I think that CAR Tregs are going to see some results from upcoming trials. The next few years of this work will be exciting.

What the Field Needs Next?
What does your field need most in terms of biological understanding, technological support, and the like to ensure the success of CAR Tregs? For CAR Tregs to succeed, you need both biological understanding and technological support. You get to go from basic science to translational science, then back to basic. On one hand, we need more biological understanding. For example, Tregs do not come in only one cell type. With the advent of single-cell sequencing, we see marked type differences among Tregs. There are thymic, peripheral, tissue, and induced Tregs.

That diversity leads us to wonder about the plasticity or stability of those Treg subsets. Is a Treg always the same, or can a Treg type-1 become a Treg type-2 easily? We know about types of T-helper (Th)–like Tregs. We have seen and read about Th1 Tregs, Th2 Tregs, and Th17 Tregs. Those all share certain genes with helper cells. For example, Th1 Tregs share some chemokine receptors with Th1 cells.

That makes sense because if you want to suppress Th1 cells, then you need to have Tregs alongside them. Th1 Tregs chase Th1 cells, so wherever Th1 cells migrate, so do the Th1 Tregs. The same principle is true for Th2 Tregs and Th17 Tregs. In the gut and intestine, many Tregs coexpress a factor called RAR-related orphan receptor gamma (RORγ) T, which is associated with inflammatory Th17 cells.

Transcription programs of different Tregs and expression issues related to tissue residents introduce a great deal of complexity. Again, the intestine is a good example in which a Treg has some genes that usually are found in other types of T cells. Perhaps even more baffling, at least to me, are tissue-resident Tregs that reside in fat tissue, for example. Adipose-resident Tregs express Treg genes of course, and they are suppressive, but they also express peroxisome proliferator-activated receptor gamma (PPARγ), which is a fat-transcription factor.

It’s as though Tregs learn by osmosis. If a Treg resides in adipose tissue for long enough, then it becomes a fat-resident Treg and will express genes that are associated with fat. Also, muscle Tregs and other different tissues contain their own specialized populations of Tregs. This matters because it relates to translation. Maybe I want a Th1 Treg for a certain disease indication. Maybe I want a pancreas-resident Treg cell for type 1 diabetes. Understanding the differences among Tregs is step one. Step two is using that knowledge to select a specific Treg. If I want a specific “flavor” of Treg, do I isolate and expand the cells, or do I introduce genes to a “vanilla” Treg to transform it into a Th1 Treg or an islet-resident Treg?

In terms of technological support, we are working toward how best to modify, purify, and expand Tregs. Regarding modification, Tregs are more fragile, sensitive, and delicate than conventional T cells, so the methods that we’ve used to modify T cells to make CAR T cells for cancer therapies might need further adjustment to optimize the gains for CAR Tregs. For purification, we are using sterile fluorescent-activated cell sorting (FACS) (Miltenyi Biotec). I think there’s going to be a big push in that direction, because then you can have rigorous good manufacturing practice (GMP)–compatible sorting. And for expansion, automation can help you decrease variability from donor to donor. Finally, gene-engineering technology is an important technological consideration when treating a disease with Tregs that might have been caused by Treg defects from the start. What technique is going to work best in that case? Are we going to take the defective Tregs out of a patient, gene-correct them, and then put them back in?

That idea comes from a treatment for sickle-cell disease by which faulty cells are removed, the abnormal gene is corrected, and the cells are returned to the patient. But is it better instead to make healthy Tregs compatible because they are more abundant and easier to grow? Imagine knocking out a TCR or replacing it with a CAR. Let’s say you have an HLA-A2 CAR knock-in Treg, and it’s an HLA-A2 knock-out Treg. Such a Treg could go into a donor and recognize only HLA-A2. It would not be recognized and destroyed by the recipient’s immune system before it gets through its job.

Now, the caveat here is that immune surveillance is good. We might want to include a suicide gene so that if those CAR Tregs become dangerous, we can shut them off. I think that’s probably going to be a parallel process. Do you modify a patient’s own Tregs, correct them, and put them back in, or do you modify healthy control Tregs for compatibility with a patient and put those in?

It also is important to note that assaying Tregs in vitro is difficult. A couple of suppression assays can be used in vitro, but in vivo the Tregs are talking to 15 to 20 cell types. In vitro observations allow me to put Tregs only with T cells, after which those cells proliferate less. But that needs to be confirmed by an assay before the cells are injected into a patient. Sometimes you don’t see differences in vitro, though, even when there’s a gigantic difference in vivo. That is an important part of biology and assay development. How do you understand whether those cells are working and how well they are working? And which mechanism are they using at each time point and location?

Look for Part 2 of our interview in next month’s issue of BPI.

Further Reading
Alberto SF, et al. Applicability, Safety, and Biological Activity of Regulatory T Cell Therapy in Liver Transplantation. Am. J. Transplant. 20(4) April 2020; https://doi.org/10.1111/ajt.15700.

Ferreira, et al. Next-Generation Regulatory T Cell Therapy. Nat. Rev. Drug. Discov. 18, 749-769 2019; https://doi.org/10.1038/s41573-019-0041-4.

Muller, et al. Precision Engineering of an Anti-HLA-A2 Chimeric Antigen Receptor in Regulatory T Cells for Transplant Immune Tolerance. Front. Immunol. 12, 2021; https://doi.org/10.3389/fimmu.2021.686439.

Josh Abbott is associate editor of BioProcess International; [email protected]. Leonardo Ferreira, PhD, is an assistant professor of microbiology and immunology at the Medical University of South Carolina; [email protected].