At A Glance:
Name: Jan Schnitzer
Position: Scientific director, Sidney Kimmel Cancer Center, since 1999.
Published paper in June 10 issue of Nature: “Subtractive Proteomic Mapping of the Endothelial Surface in Lung and Solid Tumors for Tissue-Specific Therapy.”
Background: Scientific founder and director, Vascular Genomics, 1995-2001.
Assistant/Associate professor of pathology, Harvard Medical School, 1994-98.
Assistant professor of medicine and pathology, University of California San Diego, 1990-93.
Post-doc in cell biology, Yale University School of Medicine, 1986-88.
MD, University of Pittsburgh Medical School, 1985.
B.S.E. in chemical engineering, Princeton University, 1980.
Tell me what you did in the Nature paper.
There are several different perspectives that we were trying to get across in this paper, but the most relevant one here has to do with the fact that when you apply a global analytical technique on tissues, there’s such a morass of data. What we have done is designed and developed a variety of different techniques to allow us to hone in on what we believe to be a critical interface to find targets that are inherently accessible to agents that are injected in the blood. What we deem that to be is the tissue-blood interface. The tissue blood interface is the lumenal cell surface of the endothelium. We’ve developed technology to allow us to isolate that membrane from the tissues. It’s sort of the proverbial magnet that allows you to pull the pin out of the haystack, because the endothelium is a very minor component of the tissue. It can be much less than 1 percent of the tissue. And then this membrane is about 1 percent of the total cellular membrane. So .01 percent of the total tissue is what you’re actually looking at. But by [using our technology], we get rid of the other 99.99 percent of the tissue, which then allows us to unmask targets. We find tissue-specific candidate molecules in these membranes by doing the in silico subtraction within our proteomic database, then we make antibodies to them for expression profiling and in vivo molecular imaging, that tell us which one is specific and exposed to the antibody in the circulation.
What we did in [the] part of the paper which focuses on lung, [is] we said, ‘OK, we’ve isolated this from lung tissue and we isolated it from endothelial cells grown in culture. What are the differences between the two?’ We had hypothesized that if a molecule is tissue-specific, then one of the reasons is because of the environment in which the endothelial cell is growing. In other words, each tissue is unique in terms of the environment that the endothelial cell is experiencing. So obviously, when you take endothelial cells out of their normal environment in the lung, and you throw them onto plastic in a tissue culture dish, all the stroma, the blood vessel, the vascular flow — all of these things aren’t present. So we said, ‘Let’s subtract what we see in vitro — in cell culture — from what we see in vivo, and we should, by doing that, be able to hone in on molecules that tend to be expressed because of the tissue microenvironment.’ That would mean that in that subset we might find tissue-specific molecules. And that’s exactly what we did. Every molecule that we identified showed some level of tissue modulation when you looked at the expression pattern in different tissues of the body on the endothelium. When we looked at all [of them], we found two where the only place we could see them expressed on endothelium was in the lung. One of them ended up being the best one.
To what else have you applied the technique?
We applied it in the paper to tumors, too. In this case, we didn’t get cultured cells. We used the model of mammary adenocarcinoma: tumors that were injected, seeded the lung, and grew into the lung, and we isolated our membranes from the tumor-bearing lungs. After doing the mass spec analysis of those membranes, we compared those proteins that were identified in the tumor endothelial cell surface membranes to those that were identified in normal lung. So, in this case, you’re taking membranes in a normal tissue and you’re subtracting [their measurements] away from membranes in a diseased tissue. And when we did that, we focused in on about 20 proteins that came up in that in silico subtraction. Some of them were known angiogenic markers — the VEGF receptors, neuropilin, things like that that were already known. That told us we were looking at the right membrane. And then on top of that we found another subset of about 10 proteins, which we then did expression profiling on using antibodies that we made or bought, and we were able then to find one molecule — the annexin A1 molecule — that was a 34 kDa protein that was expressed and exposed on the tumor endothelial cell surface, only in the tumors. So when we made a monoclonal antibody to it and injected it into the circulating blood, it traveled via the blood vessels to the brain, and didn’t bind to anything because the antigen is not there. It went through the normal lung, [and] didn’t bind anything, because the molecule’s not there. When it went through blood vessels of the tumor, it saw the molecule, and thus bound to the endothelial cell surface. Because we had added a radionuclide so that we could follow it with whole body imaging of the animal. And when we did that, sure enough, we saw this stuff light up the tumors. In the process of doing whole body imaging, we started to notice that the tumors were disappearing. So we did a survival study and showed that when you injected this antibody into the rats that had lung tumors, the control rats would all die in five to seven days. And when we did it with the annexin A1 antibody that we created, it targeted the tumor, caused the tumor to disappear, and the animal survived. We had complete remissions. The very first [animal] that survived has now survived [for] over a year and a half.
So the next step is to take this to humans?
Are you collaborating with any companies to do that?
We’re talking with several institutions across the country to identify groups that would be interested in doing this and have the necessary antibody experience. We did tissue staining, and, sure enough, this antigen is indeed expressed in human solid tumors. We saw it in a variety of human tumors — breast, prostate, colon, lung tumors, kidney tumors, brain, liver, even metastatic tumors. Now we want to find the right group that can make the antibody and then do the imaging as well as the therapeutic studies with us.
So the antibody could be used in a variety of tumors?
Absolutely — about 80 to 85 percent of cancers are solid tumors, so the question will be, what percentage of any given tumor type will express this protein and have it exposed so that an antibody can target it and we can achieve the same results. We also need to determine whether the profile is going to be identical to what we saw in a rat. Rats have a cleaner lifestyle than human beings — at least research rats do. They don’t smoke, they don’t abuse alcohol, and so when we’re doing the experiment, we’re giving them one disease. So far, when we’ve done stains of normal tissue in humans, we’ve seen the same profile in the human that we saw in the rat. But that’s going to require us to broaden our investigation, and that’s why we want to do these imaging studies. Does emphysema, for instance, cause this protein to be expressed in the blood vessel of the lung?
So you need to do diseased controls?
Yes. One can look at it as being a bad thing if it’s found in another disease, but one can also look at it as being a good thing, because it may allow us to then target patients who have emphysema and do something to help their disease. On the other hand, it would complicate the way in which we would have to create a therapy. We would have to be a bit more sophisticated than injecting a radioactive antibody into the body with the idea of destroying the vasculature. Because obviously you wouldn’t want to destroy normal tissue too.
If all conditions were ideal, how far out do you think you would be from putting out a therapy?
Our goal is within the year to get it into patients for clinical trials. If it gives us images that were the same as what we were getting in the rats, then I can see this thing getting fast-tracked. If you can truly achieve tumor-specific targeting in a human being, you’ve overcome 50 percent of the battle.
Are you working on markers for other diseases?
Yes. We have a huge proteomics effort, and we get a lot of money from the NIH. I have several proteomic mapping grants from NIH, and we look like we’re going to be receiving money from the [National Heart, Lung, and Blood Institute] also to do a global proteomic mapping of the endothelium in normal organs. We also have a paper coming out in Nature Biotechnology next month that is the first comprehensive proteomic map of the endothelial cell surface in an organ of the body, where we identify about 450 proteins at this cell surface. That’s with a very stringent three-peptide requirement.
We’re also beginning collaborations with gene therapists around the country to be able to do [targeting] for gene therapy. One of the two major problems with gene therapy is delivery — how do you get the vector, whether it be viral or nonviral, to the site where you want the gene expressed? Because you don’t want the gene expressed in aberrant sites, because it will have effects. That is a major problem in gene delivery. And that’s what we are pretty good at moving forward with.
So you would use the maps to target gene therapy?
Right. What you do is map the endothelium in one organ, versus another organ, versus another, and you do in silico subtraction to figure out what proteins are expressed in the organ of interest. Now we’re trying to figure out how [to] deliver adenoviruses or nanoparticles that carry genes or drugs only to [that organ].
Once you’ve targeted a particular tissue, how could you get the antibody or virus into the cell?
That’s the third part of what we’re doing, and we’re in the process of putting together some pretty amazing papers showing that we can target a particular vesicle, called a caveola. What these vesicles do is they’re involved in a process called transcytosis. That means movement from one side of the cell to the other side of the cell. We have shown that antibodies can enter these caveolae and be ferried across. And so that’s how we can achieve penetration into the tissue. We’re going to have a series of papers that show this process can be done fairly rapidly both in normal tissues as well as in tumors.