Scientist Stories: Klaus Ley, Translating an Atherosclerosis Vaccine to Clinical Trials

Our next speaker will be Dr. Klaus Ley. He is going to be talking to us on translating an atherosclerosis vaccine to clinical trials. He is the professor and head of the Division of Inflammation Biology at La Jolla Institute for Allergy and Immunology. Most of his work has centered on the inflammatory component of atherosclerosis and is actively working on an anti-inflammatory heart vaccine aimed at dialing down inflammation in affected arteries. This is something that many, many people would look forward to your being successful at, and that's what he's going to focus on this, this morning. He comes to us from Germany, where he did his medical studies and, postdoctoral research. He visited, uh, in San Diego, um, during that period, um, and then came back to the U.S. at the University of Virginia, where he served as director of the Robert M. Byrne Cardiovascular Research Center from 2001 to 2007, um, and then joined the group at La Jolla. He is distinguished, as have been our other speakers, in terms of honors and awards, including most recently winning the Eugene M. Landis Award which is the Microcirculatory Society's top honor in recognition of his pioneering work in vascular biology and microcirculation. So please proceed.

Well, thank you for your kind introduction, and thank you for inviting me here. This is very exciting to be at a real vaccine meeting. And I couldn't agree more with Wayne and Mark and others. In order to do this, we need to understand immunology and especially human immunology much better. So my lab is a vascular biology lab and we do a lot of imaging. And I just wanted to show you a handful of pretty movies because one part of the immune system that we haven't heard about today is the innate immune system. And that's very important in many diseases. And let me just speak for atherosclerosis. So this is a cleared whole mount of a carotid artery in which Different macrophages were labeled by different types of GFP and YFP. And there are 4 subsets of macrophages, and nobody has a clue what they do. We can also do in vivo imaging, which is technically challenging because the mouse has a heartbeat of 300 to 400. And this is a live mouse carotid artery, and you can see these subsets of macrophages doing their thing. The workflow in my lab is digesting either mouse or human atherosclerotic plaques, then do standard flow cytometry and CyTOF, which is mass cytometry where you get better resolution with about 40 markers. Then we do, in the beginning when we have no idea what we're looking for, we do DropSeq where you get single-cell RNA-seq for all the cells in the digest. But once we know which cells we want, From either the flow cytometry or the CyTOF, we can sort for these cells and sequence them and get much deeper transcriptomes, analyze this by various tools that are basically all machine learning based. This is not an atherosclerosis crowd, so I have to introduce the residual inflammatory risk. In atherosclerosis, we have lipid risk and inflammatory risk. Basically, we can say that in the age of PCSK9 antibodies, that the lipid risk can be taken care of. If your LDL is not controlled by statins, it can be controlled by PCSK9, but your risk is still about half even if you control to target. And that is true whether you control with statin or whether you control with PCSK9. In other words, there is not more to be had by controlling lipid risk. But earlier this year, the CANTOS trial came out and showed that you can also do something about the inflammatory risk because when you block interleukin-1 beta, and this is a large clinical trial, you get an improved hazard ratio, and this is significant. However, you also get a higher fatal infection rate because, of course, if you mess with a major cytokine like IL-1 beta, this is not good for host defense. So the rationale for immunotherapy is that, yes, IL-1 blockade works, but it it broadly impairs host defense, so overall mortality is actually not changed much in spite of significantly reducing major adverse cardiovascular events. But immunotherapy is narrowly antigen-specific and therefore spares host defense. I don't have to tell you about CD4 T cell subsets. This is an immunology crowd. But in atherosclerosis, mostly the Tregs are atheroprotective. That has the best evidence. TR1s, the IL-10 producers, may be atheroprotective. And Th1s and Th17s are bad. And there are not many Th2s in the plaque. So, okay. Can you— yeah. This movie is not working but simply shows MHC Class II restriction. So when we first started this, we had to show that in atherosclerosis we actually had Class II-restricted CD4 T cell responses. And this was initially done. With OTII or SMARTA transgenic T cell receptor cells that were incubated with an aorta from an atherosclerotic mouse in which the APCs were labeled with GFP. And then you get movies like this where the blue and the red cells are T cells and the green cells are the antigen-presenting cells. And then you can study how well they interact with each other and you can get numbers. And so you see two things. You see that when you have antigen present, you get long interactions and the T cell velocity goes down. And you can also count the number of cells that interact. In a positive case, that can be 15%. And here are the specificity controls. So when you add the OVA peptide, the OT-II cells will react. And SMARTA peptide, the SMARTA T cells will react. This is simply showing that this is class II restricted. And when you block class II, you don't get any interaction. At that time, 2012, nobody believed that there were CD4 T cell responses in atherosclerosis that were class II restricted. This movie doesn't seem to work because the Apple computer doesn't have this codec, but this is the same thing in atherosclerotic mouse. The movie would look similar. And when I saw this data in 2011 or '12, I almost fell out of my chair because this is now a polyclonal response. This is not a transgenic. This is T cells harvested from an ApoE knockout mouse incubated with an ApoE knockout, which is standard model of atherosclerosis. And the number of interacting cells is the same almost as in the transgenic model. So this works extremely well, much better than I would have thought. So I'll summarize this work. We know the antigen presentation is by some sort of macrophage that is CD11c positive to CD4 T cells. It's class 2 restricted peptide specific and triggers various cytokines. Now, one of the known atherosclerosis antigens is ApoB. And ApoB is the core protein LDL, but also in chylomicrons. And it winds its way around and through the various lipids, the phospholipid monolayer and the cholesterol and the cholesterol esters. And so we wanted to immunize with class 2 restricted ApoB peptides, and to do that, We first used a computer-based prediction of epitopes similar to the last talk. We used the NN algorithm for class II binding. Measured the affinity of candidate peptides and then the good ones we tested in mice. We used the ApoE knockout model. And the vaccination scheme we use is basically 1 prime and 4 boosts right now. And that clearly needs to be optimized. This is not translatable yet. So I will summarize all these data in one slide because they're all published. So we tested so far 6 of the 27 epitopes that we found, and 6 out of 6 worked. In all cases, you get 40 to 50% reduction in lesions. I will come back to P18, which is a special peptide and for which we have some human data. And the controls were negative, so adjuvant only or peptide only or irrelevant peptide like MOG or ovalbumin had no, protective effect. And recently we screened some adjuvants because CFA-IFA is not translatable. And we screened about 8 adjuvants. And it turns out that Adavax actually works. So Adavax is similar to MF59, which is clinically used. And you see the lesion size in the aortic arch here on the left and in the whole aorta on the right. Maybe the images are more impressive. But what was very interesting, unlike CFA-IFA, the Adavax did not induce antibodies. There was no IgG1, no IgG2c. These are the two types that you measure in black 6 mice. Whereas, so the same protection was afforded with CFA-IFA and with Adavax, but Adavax did not induce antibodies. And I've for a long time believed that this vaccine is mostly a CD4 T-cell vaccine, not an antibody-based vaccine. Vaccine. So this kind of gives additional support to this hypothesis. In order to see the class II restricted CD4 T cells, we made MHC class II tetramers. For the mouse, were made by Mark Jenkins, and Bill Kwok made the human tetramers. Basically, I think you all know what tetramers are. This is an immunology crowd. And so, so this is the, the p18 peptide And you can see here that when you immunize with P18, in this case in IFA, you get an expansion and you get like 0.05% of your CD4 T cells to be specific. So this is, of course, more than the natural repertoire, but these numbers are starting to make sense. And of these, many, many, 50% are CD25 positive, FOXP3 positive Tregs. We think the induced Tregs are what are atheroprotective. Some evidence for that. So next we took a look by single-cell RNA-seq. And as you know, single-cell RNA-seq, as elegant as it is, the transcriptomes are not very deep. You get about 1,200 genes per cell. And so we did this t-SNE clustering, and I'll just highlight the ones that have transcript for FOXP3. So these are clusters 1 through 6. They had FOXP3 in them. And when you look at the transcriptome, you see very interesting things. So these have the FOXP3, but they also have— some have ROR-alpha, ROR-gamma, some have ICOS, TBX21 is T-bet. So you get a mixed phenotype. These are not your normal garden variety T-regs, and we're working on this to figure out to figure out what exactly they are. So what do these cells do? Well, if you do an adoptive transfer, you sort these by tetramer and put them in a recipient mouse. They go to the order, so that's good. This is an atherosclerotic recipient, so they can go there. And you can show this by imaging. So here, the control cells are green, and you can see that antigen-specific T cells in red go into where the antigen-presenting cells in blue are. And are they good or bad? So this was a system in which we went into CD4 knockout with 60,000 antigen-specific regulatory T cells from a FOXP3-GFP mouse sorted again by tetramer. So this explains why we have so few cells. And when you give these cells, you see a decrease in the on-fuzz lesion size with these ApoB-specific Tregs, and also root lesion size. I'm not showing pictures, but there's another lesion near the aortic valves that you can analyze. So what do we learn from the mouse so far in terms of mechanism? CD4 T cells interact with APCs in the aortic wall. This interaction is epitope-specific and class II restricted. Immunotherapy, either by vaccination or cell-based with class II restricted ApoB peptides, is atheroprotective and transferred to Tregs home to the aorta and they're also atheroprotective. So is this just a mouse phenomenon? So let's look at patients. So in patients, of course, like in all of us, we have a very diverse array of class II molecules, DRB, DP, and DQ, and these are the 30 epitopes that we have found in human ApoB. And you can see that some alleles bind many of the peptides and some peptides bind many of the alleles. So there's hope that there might be a universal vaccine possible. Here I'm coming back to this peptide 18, which is unique because it's the only peptide that is sequence identical between mouse and human and also binds both IAFB, the mouse class II, and also binds two alleles of human class II. We got the human 0701 tetramer to work so that we could interrogate people. Using this tetramer to this endogenous ApoB epitope. So, this— the samples come from the Women's Interagency HIV Study. Subclinical atherosclerosis was shown by carotid ultrasound. Actually, here at USC, Hodges did that work. And frozen PBMCs from over 100 participants, we had them MHC typed, and we found some that were addressable by our 701p18 tetramer.. And they were tested for surface and intercellular markers. So this is what the ultrasound looks like, and it's basically a lesion count. So this is carotid artery wall, and here you have a lesion. So we had this for all the participants. And we have this, the human tetramer, as I said, it's 701. And so we can now see how many cells we have. We actually use the tetramer always in two colors, in APC and PE.. So that reduces the noise, only the double positives are counted. And so you can see here's a signal-to-noise, so there is specificity. In humans, by the way, the number of cells is much higher than in mice because humans are not naive, they've already seen antigen. So the numbers are roughly above 0.1% of all CD4 T cells. So what changes most in the patients, and that came somewhat of a surprise, not so much in numbers but the flavor. In healthy people, most of these, two-thirds of these, are FOXP3-positive Tregs, and some have FOXP3 and some other transcription factors. But in the cases, you see that FOXP3 part of the pie is much smaller, and many acquire ROR gamma T and also T-bet and some other combinations. And this is highly significant. So this ROR gamma T-positive, FOXP3-double-positive cell is really only seen in the people with cardiovascular disease and not in the healthies. So the second study, we assessed the cardiovascular disease by coronary angiography. We did PBMCs and then we incubated, we did a restimulation assay with the ApoB peptides and did flow cytometry. And what we saw, a few things. One, in healthy samples we saw about 1% of CD4 T cells makes IL-10. By intracellular staining spontaneously. And this is much lower in the cases and not responsive to the peptides. But maybe more excitingly, when you look at CD40 ligands, so that's CD154, this is only, this comes only up when you restimulate cases, so people with atherosclerosis and not in controls. And here are some other cytokines, IL-17, a very large number. 1.5% of all CD4 T cells, when we stimulated with ApoB peptide pool, make IL-17, and a smaller number, about 0.3%, make TNF. So what do we know about humans so far? Of course, I have no vaccination data yet. From the WIHS study, we know that most ApoB-specific CD4 T cells in healthy or also in HIV-positive women are Tregs. In the Freiburg study, which is the coronary angiography study, we see that 1% of CD4 T cells from healthy subjects, but not from CAD patients, make IL-10. And this is not boosted by the ApoB peptides, it's just spontaneous. Again, in the WIHS study, we see that many ApoB-specific CD4 T cells in women with atherosclerosis now express ROR gamma T, some also T-bet, a few GATA3, and a few postdocs in my lab work on this, what we call the switch. Project figuring out how this works mechanistically. And in the Freiburg study, we see that 0.3 to 1.5% of CD4 T cells from subjects with CAD but not from healthy controls make TNF IL-17 and CD40 ligand when restimulated with the APOB peptide pool. So I would like to map out how we can progress. We have— I've shown you data in mice, in transgenic mice. We have done transcriptomes, we've done FACS, CyTOF, TCR sequencing, which I didn't have time to show, single-cell RNA-seq. We have humanized mice in which we have the human ApoB, which is useful. We have an adjuvant that might be translatable and we're working more on this. We have reagents and we have the human peptides. We have patients at high risk for cardiovascular disease you can do Patient exomes, you can find people that have 8-fold elevated over general population. What we don't have, we don't have primate data yet, and of course we don't have clinical data yet. And I would like to close by thanking the people who did the work. Dennis Wolff, a cardiologist who did the second clinical study, and Holger Winkels who did the single-cell RNA-seq, and Alex Sette who is a PI at La Jolla Institute who helped me defined the epitopes, Mark Jenkins and Bill Kwok, who made the tetramers. And I thank you for your attention.

Thank you. And we may have to give you an additional award for ending early, so thank you.

Are there questions?

Yes.

Yeah, so it's really interesting that you saw CD17. We saw that same phenotype in CD4 cells infiltrating the colon carcinomas that we looked at. It is some— quite a fair fraction of colon carcinomas' T cell infiltrates are of a Th17 type, and we did see those, but they often were part of a clonal expansion where it looked like the earlier version of those, at least some members of this one clone of T start sequence, some were making IL-17 and FOXP3 and ROR gamma, and then others had dropped the FOXP3 and they became more conventional looking IL-17 cells. So there's something there in terms of cells that maybe were thinking about being regulatory T cells Sure.

Yeah, thank you for that question, Mark. So all the data I showed you in those pie charts, these are all single-cell data. So when it was orange, it means the same cell had the FOXP3 and the ROR gamma T, just like you observed. These cells were discovered in 2008 in gastrointestinal biology. And these people at that time called them, I guess, Th17 Tregs or something. They were thought to be Tregs that are specialized to regulate Th17 responses. Similarly, there's a Th1 Treg which is also seen in the patients. So that's T-bet positive and FOXP3 positive. It is not known how this works. So there are a couple, I mean the simplest hypothesis is that there's clonal expansion of a minor population that in the disease now becomes the major population. And that can be addressed by TCR sequencing. But there are other interesting hypotheses that are also possible, and we're working on trying to find out how the switch works.

Yeah. Thank you. Any other questions? I am curious. Perhaps you said and I missed it, but why did you choose HIV-positive women as a subset to study?

Well, you're an MD, so you understand where the samples come from sometimes dictates what you do. We had the HIV interagency study study. They wanted to find out why people living with HIV have higher cardiovascular risk, and we basically sequenced all their monocyte subsets and did that work. Well, and then we had these samples in the freezer and we typed them and there were enough 701s in there, so we just— and they were phenotyped for atherosclerosis, so they were a target of convenience. But the second study, the Freiburg study, is men and women and non-HIV.

So basically confirming that. Interesting. Well, thank you so much. Important work.