(Click on image for a larger version)
I want to talk about some work we have been doing over the last couple of years on how the biological activity of IL-2 and certain other gamma(c) cytokines, in vivo, can be considerably enhanced through binding with antibodies to that cytokine. And I am going to finish by showing how, in the case of IL-2, and one particular anti-IL-2 antibody, this can be used to selectively expand T-regulatory cells in vivo, and how this might be useful for inducing transplantation tolerance.
(Click on image for a larger version)
How did we begin with this story? We began when Onur Boyman, in my lab at Scripps, was interested in what it is that keeps memory T cells, CD8 cells, alive. So what do I mean by memory CD8 cells?
Let’s say you stain a spleen or a lymph node for CD8. The left-hand chart here refers to a CD8 cell which we have stained for a marker, CD44, that distinguishes between naïve and memory cells. In a young B6 mouse, most CD8 cells are naïve, so they express a low density of CD44. But about 20 per cent of the cells have a memory phenotype – they are CD44 hi. And this is the same phenotype you see on antigen-specific memory cells. We just use these naturally occurring cells for convenience.
Now, it just so happens that most of the CD44 hi cells in the right-hand part of this chart co-express the molecule CD122. So this is IL-2 receptor beta, and it is, as you would know, a receptor for both IL-2 and IL-15.
If you double-stain, as in the image at the right of this slide, you see that roughly two-thirds of these memory cells have a high density of CD122.
So we were interested in what it is that keeps these cells alive. I can’t talk about the naïve cells or the subset of memory cells in the areas here marked 6 and 82, just the ones in the area marked 11.
(Click on image for a larger version)
It turns out that it is contact with one cytokine that keeps them alive, and that is IL-15. We know that because, if you deprive these cells of contact with IL-15, by looking in an IL-15 knockout mouse, they totally disappear. So here they are, 10 per cent or so, in a B6 mouse, but totally gone in an IL-15 knockout mouse.
That to me was a surprise, because CD122, IL-2 receptor beta, binds IL-2 as well as IL-15. So why are they totally dependent on IL-15? Why doesn’t IL-2 substitute?
Onur Boyman, in the lab, was intrigued by this, and so he decided to do some experiments. The simple one was to take memory CD8 cells, transfer them and ask whether, under in vivo conditions, they can respond to IL-2.
And so he did these experiments. (These data were published a couple of years ago, so I am just going to go over them very quickly.)
(Click on image for a larger version)
He took purified memory CD8s, labelled them with CFSE, transferred them into B6 mice and gave them IL-2. I am not showing you that data; the result was not very interesting. They did indeed proliferate in response to IL-2 – exogenous IL-2. They just can’t respond to endogenous IL-2.
But as a control he injected some of the mice with an antibody to IL-2. Why would he do that? Because a few years before, Marić and Čaplar had published a highly influential paper that showed an antibody to IL-2 makes memory CD8 cells proliferate. And here indeed it did, Onur Boyman confirming what the Marić-Čaplar group had found.
(Click on image for a larger version)
So in the first instance you transfer the cells into a normal mouse. In the second instance you transfer them in and you give them antibody to IL-2 and it makes the cells proliferate like crazy. That makes no sense, because we know that if you inject a mouse with anti-IL-2 antibody it deprives T-regulatory cells of IL-2 – and Tregs are dependent on IL-2. So it neutralises the IL-2, and yet what we are seeing here is that if you inject the antibody it makes the memory cells proliferate. So what’s going on?
(Click on image for a larger version)
So this is the paradox, then. For Tregs it blocks IL-2 activity, and yet for these memory cells and also for NK cells it has the opposite effect: it enhances.
These data have been around for, as I say, five years or so, and Onur Boyman was very intrigued by this.
(Click on image for a larger version)
He decided to do an experiment in which he showed that if you inject these memory cells into a normal mouse with an antibody to IL-2 they proliferate. But he did a ridiculous control – by which I mean a control such that, if he had asked me, ‘Should I do the experiment?’ I would have said, ‘No way. It’s a ridiculous experiment.’ He transferred the cells into an IL-2 knockout mouse, and under those conditions there was no proliferation. So he showed that the proliferation induced by the antibody to IL-2 was dependent on IL-2. And the only explanation was that the antibody to IL-2 was boosting – not blocking, but boosting – the biological activity of endogenous IL-2.
So how can one explain this?
Well, it turns out that the explanation is that these two types of cells, Tregs and memory CD8s and NK cells, have different IL-2 receptors. Let me just go through this.
Here you see a Treg, or an activated T cell, and it has a high affinity receptor for IL-2: alpha chain, and the beta chain, and the gamma chain. This trimeric receptor has a very strong affinity for IL-2, so these T cells see the small amount of IL-2 that is present in vivo and they live on it. If you deprive them of IL-2, they disappear.
Memory CD8 cells are different. They lack the alpha chain, so they have a low affinity receptor for IL-2. And so the amount of IL-2 present in vivo is not enough; they can’t survive on that. Instead they have to make use of a related cytokine, IL-15. That is why they are totally dependent on IL-15 and not on IL-2.
Now, what happens if you inject the antibody?
It turns out that in the case of Tregs, the antibody blocks – I have already mentioned this – and it blocks probably because the antibody binds an epitope on the IL-2 molecule that is crucial for interacting with the alpha chain, whereas for the memory cells that lack the alpha chain, covering up this epitope does not matter.
If we do the experiment in vitro, then the binding of the antibody to IL-2 has no effect on the activity of IL-2. It doesn’t enhance, it doesn’t block, it has no effect. But in vivo – and this is the curious thing – the association of the antibody with IL-2 enormously increases its biological activity. Why?
This we have been trying to figure out for the last year and a half, and we still don’t really understand it. I have mentioned that it only works in vivo, not in vitro; it is Fc-dependent, if you use F(ab 1)2 fragments in vivo it doesn’t work; it requires neonatal Fc receptors but not the conventional Fc receptors; it is partly because binding to the antibody increases the half-life of the cytokine but that is not the only explanation, because it doesn’t matter how much IL-2 we give over a period of several days, we can’t reproduce what we see with the complex.
So we think it is something to do with the positioning of the IL-2 in a particular microenvironment, but precisely what is going on we still don’t know.
(Click on image for a larger version)
So this model would suggest that if we pre-associate IL-2 with an antibody to IL-2 before injection, it ought to boost the activity of the IL-2. And this slide shows that that is indeed the case.
Here in the left-hand chart column we are injecting normal B6 mice with purified memory CD8 cells, so they are CD44 hi and CD122 hi, transferring them together with daily injections of the stimuli, and looking in the host spleen a week later. So you give the memory cells IL-2, they proliferate; you give them anti-IL-2, they proliferate even more; and if you pre-mix them as a 2:1 ratio of IL-2 to anti-IL-2, they proliferate like crazy. The number of cells you get in the bottom chart versus the ones you get in the one above after about one week is about 100-fold higher. So there is enormous proliferation.
And it doesn’t work if you inject naïve cells – in other words, cells that lack CD122, the receptor for the cytokine.
(Click on image for a larger version)
This is not unique to the one antibody I have been discussing, S4B6, which is what people use to get rid of Tregs. You see here a column for S4B6; one for another antibody, JES6-5; and, at the right of the slide, human IL-2 with an antibody to human IL-2. (This is with mouse T cells.) And you see that we still find the same effect.
I want to draw your attention to one antibody that is quite different, called JES6-1. It is a true blocking antibody, unlike those other ones. So here we inject the mouse with IL-2, the memory cells proliferate, and we inject it with IL-2 together with this JES6-1 antibody: total blockade.
(Click on image for a larger version)
I am going to show you data on JES6-1 for transportation immunity in a moment, but before I do, let me talk about not just CD8 cells and memory cells, which is what I have been talking about, but CD4 cells.
It doesn’t work on CD4 cells, largely because the expression of CD122 is much lower on CD4 cells than it is on CD8 cells. To look at the data at the top left: this is injecting the memory CD8 cells into normal B6 mice and then looking at the donor cells and then the host cells in the spleen a week later. When we look at the ratio of the memory CD8 cells to Tregs, CD4 + CD25 + cells, it just so happens that the ratio is about 1:1.
If we inject the stimulatory antibody together with IL-2, the donor cells proliferate and in the host cells we see the selective expansion of the memory CD8 cells, but we do not see much effect on CD4 cells. There is some slight increase in Foxp3 + CD25 + cells, but if you look at in terms of a ratio, then it is 15:1 for the CD8 cells.
That is for S4B6, but for JES6-1 you see there is no effect on either the donor or the host CD8 cells, or NK cells – I am not showing you those – but for the CD4 cells there is selective expansion of Tregs. The ratio goes from 1:1 to 1:4. So we have this mechanism of being able to selectively expand Tregs in vivo.
(Click on image for a larger version)
Let me make the point that what we see with IL-2 and anti-IL-2 in terms of stimulation also applies to certain other cytokines. Here you see some data on IL-4, and on the right you can see that IL-4 plus anti-IL-4 gives much more proliferation than either reagent by itself.
(Click on image for a larger version)
We have the same thing for IL-7 here, and anti-IL-7. Curiously, it works much better for CD8s than for CD4 cells, even though the density of the IL-7 receptor is as high on CD4s as on CD8 cells.
(Click on image for a larger version)
It also works for IL-15, although here let me point out that IL-15, unlike the other cytokines, is not a secreted cytokine. It is a cell-associated cytokine, and it is normally presented bound to the IL-15 receptor alpha chain on various cells. And the affinity of IL-15 for the IL-15 receptor alpha is extremely high. What we are showing in this experiment is that if you make a soluble recombinant form of the IL-15 receptor alpha, and you add it to IL-15, it greatly potentiates the activity of IL-15, not just in vivo but also in vitro.
So here at the top is a small amount of IL-15 by itself, the memory cells don’t proliferate. When you add IL-15 receptor alpha-Fc, they proliferate. It is not just cross-linking, because if you use monomers they proliferate even more. This is unique to IL-15.
(Click on image for a larger version)
If you go to IL-2 and soluble IL-2 receptor alpha CD25, it doesn’t work, and that is because the affinity of IL-2 for soluble IL-2 receptor alpha is 3 logs less.
(Click on image for a larger version)
We are, obviously, interested in the possible clinical applications of the capacity of these complexes to stimulate T cells in vivo. So although I am not going to show these data we are interested in the issue of whether it can be used to amplify T cells for immunotherapy, for priming CD8 cells and for expanding T cells after bone marrow transplantation. And we have data for each of these situations that is rather promising.
(Click on image for a larger version)
Let me just show you one brief experiment for T cell priming. This is some data from Mark Rubinstein, who was in my lab at Scripps. In this experiment he transferred a small dose of TCR transgenic cells, 2 x 10 4 OT-1 cells, which you would know are reactive to OVA, presented here by a virus, VSV, transferred then with IL-15 bound to soluble IL-15 receptor alpha every two days, and measured the donor cells in the blood.
(Click on image for a larger version)
This graph is for the virus presenting ovalbumin, together with IL-15. So it is really quite a powerful accentuator of the response and generation of early memory cells. And we are, obviously, interested in pursuing this for long-term memory.
But now I want to finish with some recent unpublished data on this other antibody, JES6-1. This work was begun by Onur Boyman when I was at Scripps and it has been carried on by an excellent postdoc with me at the Garvan Institute, Kylie Webster.
(Click on image for a larger version)
Just to remind you: this is the anti-IL-2 antibody that blocks IL-2 presentation for cells that have the high-affinity receptor – Tregs and activated T cells – whereas for cells that have the low-affinity receptor there is really no effect.
(Click on image for a larger version)
In preliminary experiments, Kylie Webster looked at how much of the complexes you have to inject and what was the molar ratio. She found that a 2:1 molar ratio of IL-2 to anti-IL-2 worked best.
Unless I mention otherwise, for each injection that the mice received it was a total of 6 micrograms per injection – 1 μg of IL-2 and 5 μg of the antibody.
(Click on image for a larger version)
So here we are injecting, giving the three doses one day apart, and we are looking at the appearance of Tregs. This is in the spleen, so these are CD25 + cells. You see that it goes up from a background of about 8 per cent to above 50 per cent.
At the bottom of the slide you can see that the vast majority of the CD25 + cells are indeed Foxp3 +, so they have the phenotype of Tregs.
(Click on image for a larger version)
So how long do these cells stick around? The answer is: not very long. Again these mice are receiving three doses, 6 μg per injection – this was a better quality antibody in this experiment, so the results are better than in the previous slide. You can see that we get a peak of 50–60 per cent of the CD4 cells that are Foxp3 +, CD4/25 +. And if you pre-treat the mice with this PC61 antibody, which wipes out CD25 + cells, you don’t see this.
At the top right you see the number of cells in the spleen, which peaks after about a week. And then it is gone, by another week. And again, at the lower right, you can see that these are nearly all typical Foxp3 + Tregs.
(Click on image for a larger version)
What is the phenotype of these expanded Tregs? It is similar to that of naturally occurring Tregs but the expression of the typical markers is accentuated. So they have a higher density of CD25, Foxp3 is a bit higher, CTLA-4, GITR, TGFβ and the various other markers here.
(Click on image for a larger version)
What about functionally? If you take the cells immediately after you have finished giving the injections, they are better suppressors than normal Tregs. That is shown here in the red square. So this is a typical in vitro Treg assay. (I don’t have a great deal of confidence in these assays, but everybody does them.) You just take your effector cells and you stimulate them with anti-CD3, and then you put in your Tregs at the sort of ratio shown here, and the Tregs suppress.
As you can see, if you take the cells one day after the three injections they suppress better than if you take natural Tregs or Tregs that have received – not shown here – just free IL-2, or if, in the case of the complexes, we wait five days. So this enhancement only lasted a brief period.
(Click on image for a larger version)
What could one use this phenomenon for? We decided, ‘Well, if we can induce these Tregs, maybe we could use it to prevent the induction of autoimmune disease.’ So we arbitrarily decided to look at the induction of EAE (experimental autoimmune encephalitis). I am just going to show you one experiment here, done by Kylie Webster.
Again the mice received these three injections of IL-2 and the IL-2 antibody JES6-1, and then they received MOG peptide to induce EAE, together with CFA, a bit of pertussis, and then we looked at them to look for the onset of disease.
(Click on image for a larger version)
Disease incidence is shown on the left here. Disease score I think is the more important.
As you can see, if the mice received either saline alone or IL-2 alone, they developed severe disease. (They all had to be killed after three weeks, because of the severity of the disease.) But if we had pre-treated them with the IL-2 complexes, the disease was much less and in fact most of the mice eventually recovered.
This was encouraging. And it makes you think about therapeutic significance. But, to think about multiple sclerosis, you are presented with a situation where the patient with MS already has the disease, so they already have activated T cells, we presume, that are attacking the brain. That is not really relevant here, because we are inducing the Tregs before they receive the MOG peptide. So the question is: what happens if you give the complexes after you give the antigen?
It doesn’t work. That is shown in the lower half of this slide. As you can see, if we give the treatment on days 7, 8 and 9, after the MOG and CFSE, now the immune complexes don’t provide any protection. Why? Probably because the complexes are inducing expansion not only of Tregs but of the activated CD4s that are inducing the disease.
So we lost interest in trying to cure autoimmune disease, though I must say that it is a possibility that in the future, if we used rapamycin (which is supposed to protect Tregs but eliminate effector cells), it might work. But we don’t have any direct data on that.
Instead we wanted to enlarge upon the finding in relation to JES6-1 / IL-2 that if you start off with a naïve animal then you can induce tolerance. So we decided to look at transplantation immunity. In particular, we decided to look at islet transplantation.
(Click on image for a larger version)
So again, same situation, we took normal B6 mice and we induced diabetes by giving Streptozotocin. We gave the three injections, and then we rescued the mice by giving islet allographs from BALB/c mice.
Let me emphasise two things. One is that these mice received no form of immunosuppression other than these three initial injections of antibody and IL-2. Secondly, for those of you who are not familiar with mice: B6 and BALB/c is a very strong barrier. It is a combination of MHC (major histocompatibility complex) class I, class II and multiple minors, about the strongest barrier you can come up with. So what happens?
(Click on image for a larger version)
For an answer, let’s look at the controls. If the mice received just saline or IL-2, then they all developed diabetes within three weeks. But the rather surprising finding was that, of the mice that got the complexes, about 20 per cent did become diabetic, but after about day 40 not a single mouse – and this is data from six experiments, involving a total of 22 mice – showed graft rejection. So this is a very long-term type of tolerance. No immunosuppression here at all. What’s going on?
My own hunch was that what the Tregs are doing, round about the day 1–25 stage, is preventing graft rejection, and allowing the graft to heel in and thereby become part of self. In other words, there is a state of ignorance that is created here. The T cells are not tolerant, and we know that because they respond perfectly well in MLR (mixed lymphocyte reaction) to BALB/c. But they can’t see the antigen.
That led me to predict that if we injected the mice at the relevant stage with BALB spleen cells, the grafts should all come off. Ignorance would be overcome.
(Click on image for a larger version)
Like many of my predictions, this turned out to be wrong. In fact, the grafts did not come off. Well, one of them did – and in the ‘Blood Glucose’ results we see the one mouse that developed diabetes. But the other grafts just stayed on, no change at all. And you would notice that the one graft that did come off was at day 75, but with the other ones, including on this mouse, after nearly a year there was no change. I still don’t understand this, actually.
(Click on image for a larger version)
We don’t know, really, why these grafts don’t come off. We have this very robust form of tolerance, which I still feel is probably a reflection of some type of ignorance, but we can’t break ignorance.
Now, this from a clinical point of view is what one is aiming for: long-term tolerance without immunosuppression. Whether it is really clinically significant, going from mice, of course is way in the future.Visit these websites
The Sir Mark Oliphant Conference Series is supported by the Australian Government under the International Science Linkages Program.
© 2008 Sir Mark Oliphant Conferences.
Website design acidgreen
