SLIDES & TRANSCRIPTS
Tuesday, February 1, 2000

Immunotherapy
Stanley Riddell, MD

DR. LARSON: I think we must move directly on in order to give the final two speakers adequate time for their presentations.

Thank you, Irv.

As you will recognize the titles of these talks are also the titles of the discussion sections this afternoon. There should be adequate time to question the speakers within those sessions.

Next I will ask Dr. Stanley Riddell from the Fred Hutchinson Cancer Research Center to give us an overview on immunotherapy.

DR. RIDDELL: Actually talking about cellular immunotherapy and malignancy, in fact, working in the field is a little bit like wandering in the wilderness. So in order to keep myself grounded, I have actually spent a fair amount of my time trying to understand the immunobiology of viruses. I actually took this slide from that group of slides to illustrate some points about how T cell recognition of virus-infected cells can provide us with some insight into principles as to how tumor cells might be recognized.

Essentially, we really have two subsets of T cells, CD8 cells that recognize target cells in the context of Class I MHC molecules and the recognition structure here is really a peptide that is presented in the context of Class I. These peptides are displayed in the cell surface after processing and presentation of proteins from within the cell.

The reason the cells display MHC molecules is actually to provide the immune system with some idea of what is going on inside that cell. So when a cell becomes infected with a virus, the novel peptides are now shown at the cell surface, and these can be recognized by the immune system.

Many of the first talks you heard this morning discussed chromosomal translocations, fusion proteins, mutations in various signaling molecules that are expressed in leukemic cells, and obviously these mutant proteins could give rise to novel epitopes that are presented in the context of Class I in this situation or in the context of Class II MHC on cells that express those molecules.

There are some constraints in that conceptual framework in that the proteins have to be processed and presented, and the peptides that are actually bound to the MHC molecules have to conform to a peptide binding motif for particular individual HLA alleles.

There are anchor residues in the peptides that bind MHC molecules. Not all mutant proteins could give rise to epitopes that are capable of binding, but having said that, this understanding of how T cells recognize target cells has at least provided the opportunity to suggest that there are many proteins inside tumor cells including leukemic cells that could essentially be immunologic targets. In fact, several people in this room have demonstrated that you can isolate T cells that recognize epitopes derived from the BCR-ABL fusion protein, mutant ras and other overexpressed proteins in myeloid lineage cells such as proteinase 3 and proteins that may be important for the malignant phenotypes such as telomerase,

TOP

but having said that, those descriptions are relatively rare. I must say that there is very little evidence that in the tumor-bearing host that T cell immunity plays very much of a role in controlling the proliferation and growth of AML, but that is not the case after allogeneic marrow transplantation.

In fact, acute leukemia is one of the malignancies for which we can actually confidently say that the immune system plays a major role in eliminating the tumor.

TOP

This is old data now from the International Bone Marrow Transplant Registry that looks at relapse of acute myeloid leukemia after marrow transplantation depending on the type of transplant. What I am showing here is that the relapse rates if you receive an allogeneic transplant are dramatically lower than if you receive a syngeneic transplant or if you receive a T cell depleted allogeneic transplant, and that even allogeneic recipients who do not develop graft-versus-host disease have a much lower risk of relapse.

This reduction in relapse has been termed the graft versus leukemia effect. It is only seen in allogeneic transplants that are T cell replete, the implication being of course that what is being recognized here are minor histocompatibility determinants that are expressed on the recipient cells that can serve as foreign antigens for the donor immune system. Using donor lymphocyte infusions, this is some data actually from Mary Flowers at our center where we have actually treated patients who have had relapse of CML after allogeneic transplant with normal lymphocyte infusions either taken directly from the donor or after G-CSF mobilization, demonstrating that a substantial fraction of these patients can actually develop a complete remission. What this data really demonstrates is in fact the immune system is very important in eliminating leukemia. This has sent some of the high-dose therapists that are typically resident in transplant centers scurrying for the hills and then coming back out newly rejuvenated as non-myeloablative therapists where in fact now what we are trying to do is do marrow transplants without giving all this high-dose chemoradiotherapy but in fact giving very low doses of therapy that allow us to create mixed hematopoietic chimerism where we actually have resident in the recipient both a donor immune system and a recipient immune system and we allow the donor immune system to eliminate the leukemia.

Although this is I think a very encouraging development in the transplant field, there are some problems with this approach. First of all, donor lymphocyte therapy and in fact, non-myeloablative therapy is much less effective for blast phase CML and for acute leukemias. I think there are lots of potential reasons for this that I will be happy to discuss later, but clearly this is a problem in terms of the clinical efficacy of this approach. Secondly, both donor lymphocyte infusion and non-myeloablative therapy that rely on the immune system to eliminate the leukemia are non-selective and so what you often end up with is significant graft versus host disease.

TOP

One of the efforts in my lab has been really to try to begin to separate, to really understand the molecular nature of what is being recognized in graft versus leukemia responses, and to determine if in fact one could separate the GVL effect from the graft versus host effect. You might remember in that original slide I showed you from Mary Horowitz that there is a GVL effect even in patients who do not develop graft versus host disease. So the concept here is that minor histocompatibility antigens really represent polymorphic genes that differ between donor and recipient, and of course many of these polymorphic genes will be selectively expressed in differentiated tissues.

What I have depicted here are some polymorphisms that are ubiquitously expressed. You could have CTL that recognized those antigens and they would recognize all tissues and would presumably be cells that mediated graft versus host disease effect.

However, there may be T cells against minor antigens that are more selectively expressed, for example, those that might be involved in hematopoiesis or expressed only in hematopoietic lineage cells. If these antigens were expressed on leukemic cells, including leukemic progenitors, then CTL that would target these antigens could potentially mediate a GVL effect without causing graft versus host disease. This is the conceptual framework that we have worked in.

TOP

Houston Warren, in the lab, has developed culture techniques essentially to begin to identify T cell clones specific for minor antigens after allogeneic transplant so that we can characterize the molecular nature of these antigens and their distribution on leukemic cells. He uses a very simple technique in that he takes cells from the recipient after transplant. These are donor lymphocytes now developing in the recipient, and he stimulates them in vitro with gamma irradiated recipient lymphocytes that were stored pretransplant. After stimulation these cells are cloned by limiting dilution, and we characterize the clones that are reactive with recipient cells but not donor cells.

TOP

This works very reproducibly. You can generate monohistocompatibility antigen specific T cells from the vast majority of allogeneic HLA-matched transplant recipients. This just shows the lytic activity against recipient target cells and the absence of activity against donor cells.

TOP

. When you clone these cell lines, you typically get out two types of clones. You get out T cell clones that recognize recipient cells derived from any lineage, and here I am just showing B cells, T cells and fibroblasts derived from the skin. These two clones will recognize any of these recipient cells but not donor cells.

You can extend this panel. We have done this on occasion where we can get other tissues. They will also recognize keratinocytes, for example, and epithelial cells from other tissues.

These appear to represent ubiquitously expressed antigens and obviously would not be the kind of clone that you might think would mediate a selective GVL effect.

TOP

However, you also get out other types of clones which actually are somewhat more frequent in our experience. These clones will recognize patient B cells, T cells and dendritic cells, but do not kill fibroblasts or cells from other epithelial sites such as keratinocytes.

These minor antigens exhibit at least some preferential expression on hematopoietic cells and they could be potential targets to exert a GVL effect if they are expressed on leukemic cells.

TOP

One of the issues in terms of the

TOP

kind of data that I have previously shown you is that of course what one really needs to know is the genes that encode these antigens so you can determine their expression on other tissues. It is very difficult to look with in vitro cytotoxicity assays at

TOP

brain or heart or liver or other tissues that are more difficult to get.

In order to try to understand where these antigens are expressed, we have turned to strategies to identify the genes that encode them.

TOP

The approach that we have taken is a cDNA expression cloning strategy where essentially what one does is make cDNA library from a recipient cell that expresses the antigen and then transfects pools of this library with the class I MHC restricting allele into COS cells. These COS cells can then be screened with the CTL, and if the pool of cDNA that was transfected contains a gene encoding the antigen, you will get cytokine release from the co-culture of the T cells with the COS cell.

One then could take the pool that was positive, subclone that and eventually identify the gene that encodes the antigen.

TOP

I am just going to demonstrate how we have used this to identify a couple of genes. This was the first one that we identified. The clone was called DRN7. It recognized a minor antigen presented by HLA-3, and we were interested in this because A-3 is a fairly common HLA allele, and this antigen was roughly evenly distributed in the population.

If you look at HLA-3 positive people, about half of them expressed the antigen and half did not. Moreover this antigen was expressed on hematopoietic cells, and we could kill fresh leukemic blasts from HLA-3 positive donors with these T cell clones.

TOP

We were interested in identifying the gene and using the cDNA cloning technique. We identified the gene as a human nuclear phosphoprotein. This was in GenBank. The cDNA was an allelic 1113 base pair gene. The epitope results from a single base pair change that results in a glycine to arginine substitution. So the recipient allele has this arginine. The donor allele has the glycine and this creates an epitope now that is recognized by the immune system. These are other individuals in the population, and some of them have the G allele and some of them have the arginine allele.

TOP

It turned out that when we looked at what was known about this gene, it actually was originally cloned from a gamma interferon-inducible library. What we found when we treated fibroblasts with gamma interferon, this gene was actually up regulated and now the fibroblasts were recognized, suggesting that in fact this gene may be expressed in other tissues under certain situations and may not be a suitable target to mediate a selective GVL effect.

TOP

I want to just use as an example one other gene that we are very interested in that we have recently identified. This came from some clones that we isolated from a transplant with a female donor into a male recipient.

We isolated a panel of clones, and the clones appeared to recognize an antigen that was encoded or regulated by the Y chromosome because they recognized target cells from HLA-B8 positive men but not from women. So this was an epitope that was presented by HLA-B8 and was only expressed in male cells.

TOP

We were interested in this because this antigen was also not expressed in fibroblasts even if you pretreated them with gamma interferon. It was only in hematopoietic cells including acute leukemia cells.

This exhibited tissue restricted expression, and so we were very interested in identify the gene.

TOP

We took a somewhat different approach initially to identify this gene. David Page at the Whitehead Institute had a series of cell lines that were deleted in portions of the Y chromosome. Houdi Warren in my lab obtained these cell lines and used them actually to map the region of the Y chromosome encoding the epitope. So this cell line MRCY1 contains the full length of the Y chromosome and then these two other cell lines, MRCY10 and WHY14, contain deletions in the terminal portion of the Y chromosome. Once you delete this section of the Y chromosome, you lose recognition by the T cell clone, and so that suggested that the genes encoding this antigen were in this region.

As you know the Y chromosome contains both testes specific genes and X homologues. The testes specific genes obviously could not encode this antigen because it was expressed in hematopoietic cells. So we focused our attention on these X homologues.

TOP

These were the candidate genes in the area and essentially what we did was use PCR to make cDNA from male cells for each of these genes and transfected them with HLA-B8 into COS cells and identified the gene to be UTY and the epitope is in fact this region here. Of UTY it has two amino acid differences from the X homologue in female cells, and these two amino acid differences are enough to provide this epitope that is recognized by the immune system.

When you look at the transcription of UTY, the U stands for ubiquitous. It was thought to be ubiquitously expressed, but actually what we found is that this gene is expressed at variable levels in different tissues. It is highly expressed in hematopoietic cells but minimally expressed actually in other tissues, even though you can detect a transcript. In fact this illustrates another principle that you don't have to have a gene that is not completely expressed. If it is expressed at low levels, it may not be processed sufficiently enough to generate enough

TOP

In fact, when we got this data I had Reg Cliff just look at our database for patients who were transplanted for CML, female into male transplants where HLA-B8 was involved, and we have had no relapses in that subgroup of patients. So again, although that wasn't statistically significant when you compared it to the other groups, as Reg put it, and for Reg this is actually a concession, he said, "It is very tantalizing."

TOP

One of the other critical issues now that we are beginning to identify some of the genes is really to understand whether these genes are in fact expressed in the early leukemic progenitor cells that you ultimately have to eliminate. John Dick has very elegantly shown that in acute myeloid leukemia, there is a leukemic progenitor cell which he has termed the SCID leukemia initiating cell based on engraftment studies in NOD/SCID mice. This cell is actually a rare cell in the blast population, much rarer than the clonogenic leukemic progenitors and which represent a small fraction of what you actually see in the peripheral blood.

This really tells us there is some differentiation going on in the leukemic population, and what we really need to know is are these minor antigens actually expressed on this very rare stem cell because that is the cell you ultimately have to target to have an effective therapy.

TOP

TOP

In collaboration with John Dick and Dominique Benet in John's lab, we have done experiments where we have looked at the ability of these CTL clones to eliminate SCID leukemia initiating cells from fresh acute myeloid leukemia samples, by coculturing the clone with the leukemia and then injecting the mixture into NOD/SCID mice and analyzing the ability of these cells to engraft. This is the DRN-7 clone that recognizes the human nuclear phosphoprotein. This clone actually has a very potent effect in this model. This is actually one of the worst mice. Usually, you know, you pick the best slide. I am actually picking, I think, the worst slide.

We did get a reduction from about 50 percent engraftment to about 1-1/2 percent engraftment on this experiment, and a control clone had no effect.

TOP

When we looked at several mice by Southern blot now to detect human DNA in these mice, you can see the DRN-7 clone completely eliminated the leukemia in the majority of mice whereas mice that got the leukemia with no clone or a non-specific clone essentially engrafted to very high levels.

TOP

Now, this effect on the leukemic stem cell is actually a specific effect that requires cell-cell contact. If you do mixing experiments where you inject two leukemia samples that are differentially recognized by HLA antigen specific T cell clones and a single clone, you can show that only with the clone that recognizes the leukemia will one of the leukemias will be eliminated.

This just shows that this clone MRR-24 which actually happens to recognize the UTY gene will kill this leukemia cell in vitro but not this other one.

If we mix these two leukemias and then treat the mice with the MRR-24 clone, as I will show you, we will selectively eliminate this leukemia and not this one, and the converse experiment with ATT-7.

TOP

This is just the data for that experiment. Each of these dots represents a single mouse, and as you can see, in mice that are inoculated with MRR-24 this selectively kills the HLA-A1 positive leukemia cell but not the other one, and in the converse experiment, it has no effect on the other leukemia. So this says that it is not an indirect effect mediated by some cytokines that these T cells produced when they recognized the target cell. This is a direct effect on the leukemia and probably requires cell-cell contact and perforin and granzyme mediated killing.

TOP

Where do you go with this data? We have demonstrated that there are lots of minor histocompatibility antigens. We now have T cell clones that recognize at least 35 distinct specificities. About half of those are tissue restricted and we are in the process of mapping the genes that encode those antigens, but until we have a very large catalog, it is going to be very difficult prospectively to design studies that would actually allow you to investigate T cell therapy targeting these antigens. What we have decided to do, since the patients with AML can be stratified based on the risk for relapse after transplant, is to take patients that are at high risk for relapse and prospectively isolate T cell clones from these patients. Then we will select clones that recognize minor antigens that are tissue restricted even though we may not know the gene in all cases. We will modify these clones with the thymidine kinase gene, and then if the patients relapse, we will actually administer the clones in the dose escalation.

From this kind of study we are likely to learn several things. First of all we are likely to identify antigens that are going to be targets of graft versus host disease and would not be suitable for this. In that setting we could use the thymidine kinase gene as a suicide gene to eliminate the clones and eliminate the toxicity, but what we hope to identify are antigens that are selectively expressed where we would not see graft versus host disease and we would see an anti-leukemic effect.

Those would provide antigens for which we would pursue gene identification and then begin to prospectively genetically type donors and recipients so that you can actually select settings where you can manipulate this immunologic effect in advance and prevent relapse.

I think the other thing that this kind of study will have implications for is going back to the non-transplant setting where we actually start to look at targeting other antigens that we know can be potentially presented. We really need to begin to characterize what the best targets are, begin to understand issues related to tolerance in individuals to tumor antigens so that we can begin to manipulate those effects,

TOP

and I think with that I will stop there.

TOP