CRISPR Builds More Powerful CAR T Cells

Michel Sadelain, MD, PhD

Michel Sadelain, MD, PhD

Chimeric antigen receptors (CARs) are now being targeted with CRISPR/Cas9, the latest in a series of genome editing tools. In trials done on mice, CRISPR has been effective in creating more potent CAR T cells that enhance tumor rejection.

Senior author Michel Sadelain, MD, PhD, and fellow researchers at his lab at Memorial Sloan Kettering Cancer Center, have recently published evidence to show that this CRISPR technology can deliver the synthetic receptor CAR to the TRAC locus in the genome of the T cell.

Although CRISPR is not yet used in humans, the technology is developing swiftly and showing potential, says Sadelain.

“It is really quite mind-boggling what we can do now with efficiency and precision. It is a completely new way of thinking about a medicine that starts with a human cell, which is modified to acquire not one, but multiple features that increase its therapeutic potency.”

OncLive: Could you provide some background on CRISPR?

In an interview with OncLive, Sadelain, director of the Center for Cell Engineering and the Gene Transfer and Gene Expression Laboratory at Memorial Sloan Kettering Cancer Center, discussed CRISPR and its effect on CAR T cells.Sadelain: CRISPR is a very hot topic—it is a revolutionary technology that has gone through a number of improvements over the years. Genome editing is based on enzymes and nucleases that are targeted to a specific DNA sequence, and when the nuclease is introduced into a cell and targeted, it will cut just the gene that is chosen. There have been nucleases before this—and this one is just as effective—but is easier to manipulate. And so more people have started to use it, which includes us, for the purpose of this particular study.

Another hot topic is CAR. You can then introduce CARs to T cells, and we at Sloan Kettering and a few others have shown that if you engineer T cells with CARs, you can turn T cells into pretty effective antitumor agents. By the time it got to clinical trials, 2 other groups who had read our papers tried it as well, NCI and UPENN. So, our center, and these 2 others, provided clinical proof of activity in different forms of lymphoma and leukemia, showing that T cells engineered with CARs can be very effective in some pretty nasty cancers.

How does your study add to the development of CRISPR in the cancer space?

Now, the way that T cells are engineered to do that makes use of what is called retroviral vectors. That is what is used in research all the time, and that is also how these clinical applications were developed. Once you have designed the CAR and the gene for the CAR, you use a retroviral vector to carry the gene inside the T cell and the gene then inserts itself on a chromosome and it becomes permanent property of that T cell. That T cell now makes the CAR protein, so that is why it recognizes and kills those tumor cells. There are also a few people who use another technique, which is called a transposon, to again, shuttle the gene coding for the CAR inside the T cell.CAR T cells that are engineered with these vectors work, there are these clinical results that I just mentioned. But, what we sought to do in this study was answer the following question: could we make the CRISPR work in human T cells? And once we have that, can we pinpoint a location on the chromosome where the CAR gene will go and insert itself?

So, instead of being integrated at all of these possible locations, resulting in variegated expression, now every T cell that you make in the lab would have the gene always at that same chosen location and predictably, they would all express the CAR in the same way. We did that and we picked the location, and the location is a gene called TRAC, and we are going to put the CAR gene in that TRAC locus. We succeeded in that and it was highly efficient. And indeed—as you would expect now—all T cells express the CAR at the exact same level. That is clearly in contrast to what you get with the retroviral vectors, which, as expected, their expression is much more variable because the gene is not inserted in the same place in every T cell. Then the surprises came, one of which was when we evaluated these cells in animal models of leukemia in vivo, we found that the cells that had the CAR gene inserted into the TRAC locus worked far better than the T cells that had the randomly integrated CARs. Again, those cells can work in the clinic, there are clinical results to prove that, but TRAC CAR T cells work far, far better.

And then the question became “Why?” There are 2 reasons. The first is that when you use the usual ways of inserting the CAR gene into the T cells, there is a phenomenon that occurs which is called tonic signaling. Tonic signaling means the CAR, which is not expressed in the T cell, is sending in weak signals all the time, even if there are no antigens around—that is where the word tonic comes from. This is detrimental to the T cell, it accelerates differentiation and the T cell rapidly loses function. But, when the gene was in the TRAC locus there was no more tonic signaling, so the expression is a bit lower than what you get on average when you randomly integrate vectors. That level is still good enough to have excellent antitumor activity—but it is not too high, because if the level is too high, you run into this problem of tonic signaling.

The second point is that we discovered when the CAR binds to antigen, the CAR molecules, the receptor, now it is the protein made from that gene—it is on the surface of the T cell. When it binds to the antigen, which in this case was CD19, the CAR disappears from the cell surface—it goes inside the cell and is degraded. That means that for the next 12 hours or so, there is no more CAR in the T cell, which means the T cell cannot attack the leukemia, and in fact, it gives a break to the T cell. The CAR protein disappears so the T cell gets a little break, but then from the CAR gene that is centered in the TRAC locus, it starts making new CAR proteins, and after 24 hours, the CAR is back on the cell surface.

What were the findings of this study?

As we saw this, we started inserting the CAR at different locations, and in the paper there are 9 locations in all that have been investigated. And what we found is the 8 other ways to engineer the CAR using CRISPR. We’re not nearly as good as we were when it was in the TRAC, which means the TRAC was a really good place to insert the gene. And what we saw when we analyzed the other 8, were different patterns—some expressed much higher CAR than the TRAC, resulting in tonic signaling and rapid dysfunction of the cell, also called exhaustion. Some expressed very low CAR, but then they were not good killers. The initial expression of the CAR is quite comparable than what you get from the TRAC. However, when the CAR is degraded inside the cell—which takes 24 to 36 hours to be fully replenished—some of the locations did not work, the CAR came back too fast. What we learned are 2 important things—if you express the CAR at the right level, not too much, not too little, you can avoid tonic signaling, which is a bad thing for the cells. The CAR does all of these very sophisticated things after binding to the antigen—it disappears and it comes back. But the tempo is very critical, coming back too late or too early isn’t great for the T cell.

Now, this has a bunch of practical ramifications, one is that if you make much better cells, then you do not need to make as many. So, this is positive for the manufacturing of cells.

Another important feature of this is that when the CAR goes into that TRAC locus, it destroys the natural receptor for antigen that the T cell has because TRAC is actually one of the locations that codes for one of the two chains of the normal receptors that all T cells have. When we put the CAR in that location, we destroy the normal genes. What this means is that the T cell no longer has its original natural receptor because we just zapped it, but it has the CAR.

There are clinical consequences of this that remain to be proven, but there are logical expectations from this. One is, in clinical trials of CARs, they usually exclude patients with autoimmunity because people are worried that if you take their T cells and give them back that maybe some of these T cells could attack the patient because they have an autoimmune disease. But, in this approach, since you've destroyed their natural T cell receptor, they could not attack normal cells, all they will do to those normal cells is to attack what the CAR tells them to attack.

Lastly—and this has the broadest implications—you could use the cells from someone else, put cells from "Person A" into "Recipient B." Normally, people do not do that—nobody does that because the concern is that the cells of Person A will attack Person B, like graft-versus-host disease in bone marrow transplants. Therefore, all of these therapies are autologous—your own cells will not attack you—unless there is an autoimmune condition. But with CARs, if you take the cells of Person A and you take away their T-cell receptors and give them to Person B it will not cause graft-versus-host disease, because they do not have a receptor anymore, and the cells will just do what the CAR tells them to do.

This has been done before by others, in part. They destroyed the T-cell receptor, and showed that you could then avoid this graft-versus-host disease. So, we are not the first to do just that, however, we both destroyed the T cell receptor and put the CAR in the TRAC locus. And that resulted in the superior function in the CAR, which hasn’t been done before.

What are the next steps?

So, you can make these cells now that do not cause graft-versus-host disease, and you have a CAR that is expressed optimally. In general, CRISPR is thought to bring a lot of good things to a variety of medicines by engineering stem cells for treatment, and there is an expectation that it will do wonders. The precision of this editing tool, as you could call it, will be able to find many medical applications. It is already used a lot as a research tools when people in the lab want to change cells or genes. It is also used in animals, plants, rice—people are using it to change the genes for a lot of organisms.

Now, as much as there is the anticipation that it will be useful in medicine, it has not been done yet. CRISPRs have never been used in humans—yet. And it is a question of what will be the first applications. It is possible that these CAR-engineering strategies may turn out to be the first application of CARs in humans. Maybe there will be other ones, but it is likely that these CAR T-cell therapies that are already based on genetic engineering, which are already showing great results in patients, will logically be one of the first approaches incorporating CRISPR technologies.

Importantly, we are on the path—we are in the process of making it clinically acceptable. I cannot speak for everything happening on our planet, but I can tell you that at Memorial Sloan Kettering, we are planning to bring this to the clinic because based on these results and this manuscript, we think that CAR T cells engineered in this way will have many benefits and will be better than the CAR T cells that are engineered today—and we want to test that in clinical trials as soon as we can.

Eyquem J, Mansilla-Soto J, Giavridis T, et al. Targeting a CAR to the TRAC locus with CRISPR/Cas9 enhances tumour rejection. Nature. 2017;543(7643):113-117.