If there’s one thing the weight of the scientific evidence has demonstrated, it’s that cancer cells are, to a very good first approximation, little bastards. They try in various ways to pose as normal cells to evade the immune system while they grow out of control and wreck everything around them.
But as we continue to learn more about them, and about our own immune system, the tricks we are able to pull on them get more and more sophisticated and clever. When Joe Biden talks about a “cancer moonshot”, he’s alluding to the right level of complexity, and the right level of ingenuity.
This week, a team at UCSF shows how to get a cancer cell to score an own goal on itself by unwittingly announcing its presence to the immune system and even actively helping to attract T cells to come over and kill it. This new approach is explained in the September 12 issue of Cancer Cell.
While this particular example is aimed at cancers of a type called KRAS-G12C, it needn’t end there; the same kind of approach could be used against several other cancers.
Human cells use a protein called KRAS to tune their growth and activity based on what’s happening in their environment. But there’s a defective form of KRAS called G12C that commonly shows up as a cause of lung, colon, and pancreatic cancer. It’s called “G12C” because its 12th amino acid is cysteine (C) instead of glycine (G). That’s because of a single-letter mistake that can develop in DNA, one particular thymine (T) that is supposed to be a guanine (G). And that’s a really important little mistake, because the G12C mutation leaves KRAS stuck in the “on” form, so cells with KRAS G12C basically have an anvil on the gas pedal, which is the hallmark of cancer.
There’s actually some good news right off the top on this, though. It was long thought that KRAS G12C was “undruggable” because the mutation is hidden inside the folded-up protein, and a lot of drugs were thrown at it and failed. But as we got more and more precise information about KRAS’s structure, and we got better at designing molecules to fit in there, we started finding some molecules that could indeed access that 12th amino acid after all. Here’s one called ARS-1620, found in 2018, and look how nicely it fits right in there:
Of course, we can’t take drugs that indiscriminately shut down all our KRAS molecules, including normal ones, because then all our cells would die. So we need a way to go after only the G12C form, to specifically land a punch on that cysteine-12. You can see at the left edge of the ARS-1620 molecule — the part that gets close to cysteine-12 — that there’s a double bond (depicted as =) sticking out. Keep your eye on that.
Cysteine has a group called a sulfyhdryl (—SH) that glycine doesn’t have, and ARS-1620’s double bond (still got your eye on that?) can attack that —SH. Once that happens, it’s irreversible, and the KRAS protein is shot. The great thing is that only mutated KRAS proteins get the kaibosh, while normal ones are left alone:
This approach by itself has worked well enough that just last year, the FDA accelerated the approval of a drug called sotorasib (very similar to ARS-1620) for non-small-cell lung cancer. Since then, newer drugs that are even more specific to KRAS G12C have been developed.
That’s awesome, but our story doesn’t end there. Not even close!
Sotorasib did stop the progression of cancer in 81% of patients, and that typically lasted on the order of six months to a year, but then often the cancer cells started to become resistant to the treatment. Resistance to KRAS G12C inhibitors like ARS-1620 often is caused by other mutations cropping up to compensate somehow in ways that aren’t always well-understood.
But, undaunted, this is where the UCSF researchers really pulled a nasty double move on the cancer cells.
First we have to step back here and absorb one amazing thing that human cells do. They’re constantly making proteins, but believe it or not, somewhere between 30-70% of those proteins don’t quite get made properly. Plus lots of other proteins just outlive their usefulness. So the cell chops up all those past-their-prime proteins, sticks the resulting snippets onto something called major histocompatibility complex I (MHC I), and sends those out to the surface of the cell, just to show the outside world what the cell has been up to lately. This is called “antigen processing and presentation”. Here is a good simple cartoon of that:
Every cell has about 20,000 copies of MHC I at its surface at any time (!), meaning that every cell displays a pretty wide snapshot of its own recent protein history out on its surface. You can imagine that if a cell gets infected by a virus, before long a lot of its MHC I-associated surface proteins are going to be viral proteins, and the immune system is going to spot that and go attack those infected cells.
Cancer cells with the KRAS G12C mutation thus end up displaying snippets containing that very mistake on their surfaces, but the problem is that one teeny mutation like G12C is going to be pretty hard for the immune system to recognize as abnormal through all the noise.
So the big question our UCSF researchers asked is: If we treat cancer cells with ARS-1620, so that the KRAS G12C protein now has this big honking pendant dangling from it, will that part actually make it through antigen processing and get displayed on the surface? Because if it does, now we’d have cancer cells that not only are slowed down by KRAS being inhibited by the drug, but those same cancer cells would also be displaying a huge “Look at me! I’m a cancer cell!” flag.
It turns out that does indeed happen. Cancer cells, in their quest to look like normal cells, will keep running the MHC I system (Dum-dee-dummm, nothing to see here!), but too bad for them that the G12C snippet makes it out to the surface with that big ARS-1620 clown nose attached to it. Busted!
Based on the halting success of sotorasib, we know that the immune system does recognize that to some extent, but the UCSF crew went a couple steps further, juuust to make it nice and clear.
We want T cells to recognize and attack these cancer cells, and we can guide those T cells right to the target by using something called a BiTE, or bispecific T cell engager. This BiTE is basically a double antibody. One end of it sticks to the T cell, and the other end sticks to the unique feature of the cancer cell we’re trying to attack. Cancer cell, meet Mr. T.
We start with two regular antibodies. One of these sticks to CD3 (a protein all T cells have on their surfaces), and that’s a well-known antibody that we can easily buy. But we also need to find or make another custom antibody that will stick to (in our case) the G12C snippet with ARS-1620 hanging off of it, because that’s on the surface of the cancer cells. We’ll make the BiTE out of the sticky ends of these two antibodies, and that will pull a T cell right up next to the cancer cell. The T cell will then secrete Nasty Juice at the cancer cell and put it out of its misery.
If you’re wondering where they got the custom antibody to the G12C/ARS-1620 snippet, they made and found one they call P1A4 using a Fab-phage library, the same technology for which the 2018 Nobel Prize in Chemistry was awarded to George P. Smith and Sir Gregory P. Winter. In a Fab-phage library, you make a huge number of viruses so that each one has a random antibody hanging off its outer coat. The virus particles that stick to your target can be fished out of the gigantic library pretty easily because of this selective stickiness, sort of like collecting tiny meteorites with a magnet. Then you can infect bacteria with your favorite virus and make zillions more of it, and thus buttloads of your favorite antibody. Neato.
Our UCSF team showed that the BiTE strategy works really nicely on human KRAS G12C cells that have been treated with ARS-1620. In the chart below we have a particularly fast-growing line of KRAS G12C pancreatic cancer cells called Miapaca-2, cultured along with a mix of garden-variety immune cells called peripheral mononuclear blood cells that haven’t been primed for anything in particular. The cancer cells don’t respond much to ARS-1620 alone, but when the BiTE is added, suddenly they find themselves in a lot of trouble — watch the red bars go downward — ouch:
The approach also worked well with a resistant cancer-cell line called SW1573 that ARS-1620 by itself is not very effective on. That’s really important because now ARS-1620 has two different punches to throw at the same cell: inhibiting the bonkers version of KRAS, and, even if that stops working, recruiting T cells to come over and finish the job. There are many other mutations besides KRAS G12C that this approach could be applied to. Different molecules, but the same line of reasoning. So let’s call this a “platform technology”.
As always, we have to understand that no one treatment is going to singlehandedly wipe out all cancers, and there’s still work to be done here to test and refine this treatment in real human beings. But what we’ve got here is a totally new and effective approach that creatively combines a whole lot of prior knowledge and techniques with some brand-new insights. You see how much collective effort, over generations, ultimately has to go into fighting cancer successfully. Moonshot indeed.
So cancer cells may be able to get around our drugs for awhile, but our tricks will keep getting wilier. I mean, when you have a T cell attached to your face, you can’t front on that.