Pioneers of Proteomics

In part six of the Pioneers of Proteomics video series, Dr. Catherine Fenselau discusses the application of proteomics to the study of drug resistance and the requirements for continued advancement in proteomics.

Dr. Fenselau is Professor of chemistry and biochemistry at the University of Maryland. Dr. Fenselau is past president of the American Society for Mass Spectrometry, a member of the executive committee of the U.S. Human Proteome Organization and associate editor of the ACS journal Analytical Chemistry. Dr. Fenselau’s research interests include biomolecular studies using mass spectrometry, interactions of drugs and proteins, rapid characterization of microorganisms by mass spectrometry, and mechanisms of acquired drug resistance. Dr. Fenselau was one of the first trained mass spectroscopists to take a faculty position in a U.S. medical school.

On the application of proteomics to the study of drug resistance

“…the work of the cell is ultimately done by proteins”

Some people say that most cancer patients die of drug resistance rather than failure of their chemotherapy treatment, so it is quite a severe problem clinically. And actually not just for cancer, but for any condition in which the patient receives drugs for a prolonged period of time like leprosy or AIDS obviously, pneumonia. The body just does what it’s supposed to do and figures out a way to resist or get rid of this foreign substance that’s being introduced. The mechanism of drug resistance is certainly multiple mechanisms, and that makes it all the more challenging to understand and define.

I think in all chemotherapy there’s a keen awareness on the part of the clinician that sooner or later the patient will become resistant to the drug. This is why these multi-drug regimens have been developed, to try to slow the resistance to a particular drug. So I think that resistance is second only to toxic side affects in the minds of clinicians treating cancer right now.

This is a problem where there are many proteins involved. We have no idea really how many proteins are up regulated or down regulated or modified structurally in response to the challenge of therapeutic regimen. So if we want to look at many proteins at one time proteomics technologies are certainly the way to go.

…the work of the cell is ultimately done by proteins. And so if we want to monitor what’s changed in drug resistance or what’s changed during successful therapy, it’s most direct to look at the proteins. I guess it’s also worth pointing out that abundances of RNA have been found in many situations not to equate to the abundances of proteins.

On the protein changes that occur in drug resistance

“Is there some overlying theme?”

We’ve actually taken the cell apart in our studies and looked at organelles. And in every organelle we’ve looked at - cytosol, plasma membrane, nucleus, mitochondria, we’ve found proteins whose abundances and/or structures have changed in the drug resistant lines that we’re studying. And in many cases some things are already known about these proteins. They’re known to control apoptosis, so that’s a sensible response to drug therapy. In some cases we’ve have functional confirmation usually by collaborating with other cell biologists, that the function of the cell has changed. In some cases as more people begin to be interested in the drug resistance problem or in taking a proteomics approach to the drug resistance problem, we’ve had our findings confirmed by other laboratories. Now we have also been interested in putting the big picture together, if we’re going to look at individual parts of the cell. Is there some overlying theme? In some cases we have found changes that would seem to be part of a transcellular response. For example some of the pumps in the plasma membrane that remove the drug from the cell are more abundant. And these are pumps that rely on ATP, so we’ve also found in the mitochondria that the ATP syntheses are more abundant. So that’s a relationship between the inside and the outside of the cell, if you like.

In our studies we’ve used a kind of general guideline that we’re only interested in changes of three fold or more, because we do have uncertainty in our experimental methods. Also, because there is biological variability even between the same cell line grown several times. But one of the things that has impressed me the most is that in some cases we see a protein which is not detectable in the normal cancer cell and is very abundant in the drug resistant cell. People call those on/off proteins. They may not be totally off, but by our methods they’re not detectable. So some of those proteins are striking.

On monitoring responses to therapy

“…we will be in the midst of developing biomarkers for drug effectiveness…to guide therapy.”

I’m not focusing on biomarkers for early detection but rather I’m focusing on biomarkers for monitoring drug activity or drug resistance. And I think that in the ten-year timeframe we will have developed biomarkers for drug resistance which can guide therapy. And also we will be in the midst of developing biomarkers for drug effectiveness which will also guide therapy. I think we’re going to have a whole different set of cancer drugs in a decade partly as a result of the genetics emphasis now being given to cancer and drug treatment. Also as a result of the proteomics work that’s being done. Finding out what the changes are in the cancer cell as well as what the changes are in the drug resistant cancer cell.

If we can show that treatment of a tumor with a particular drug changes the abundances of a set of proteins, they can be used as biomarkers for the effectiveness of that drug. And again, some of this will be tied to genetics and some of it will be tied to availability of the drug because of the rest of the body function. But we can monitor the drug response biomarkers in real time, means as the patient undergoes therapy.

On the evolution of protein quantitation

“…of course we’re not satisfied with the methodology right now.”

Think about how people used to quantitate proteins. Perhaps it required completely isolating the protein, purifying it. And then western blots with antibodies is another older technique which required a lot of, well it’s required knowing what you were looking at and developing the antibody to it. Now if we’re using isotope labels, stabilized isotope labels in mass spectrometry to measure ratios, that’s a much easier method, faster, which does not require deciding what you’re going to look at before you get into the cell. The newest method which people are talking about now is to try to define the peak of each peptide as it comes out of a HPLC, a chromatographic separation resolving that peptide from the many others that co-elude in these large scale experiments using lots of sophisticated computer programs. That would be very exciting as well because then we wouldn’t have to use the isotopes to do chemical reactions. But we’re not quite sure yet how complex the mixtures can be that we apply that method. So people are now calling it isotope free quantitation, but twenty-five years ago it was developed for small compound quantitation by LCMS or GCMS using the pharmaceutical industry. So it’s not an entirely new idea, so we’ll see how far we can go with that method. That’s a comparison of how things have evolved. So the modern methods allow us to quantitate more peptides and thus proteins more rapidly and many of them at once with more sensitivity. So that’s real advances. But of course we’re not satisfied with the methodology right now. I think most isotope labeling methods have a very limited dynamic range. So we can see maybe a ten to one ratio, but not a hundred to one change. Part of mass spectrometry limitations are one of the areas where people are trying to improve mass spectrometry is to improve the dynamic range of our isotope ratio measurements. So we’re never satisfied!

On the need for bioinformatics

“…the database can make a difference…”

I’ve really learned a lot about bioinformatics in the last year or two since I have begun to work with a really good bioinformatics collaborator here on campus. We teach a course together so I learned that the database can make a difference, the search program can make a difference in how well one identifies the peptide or protein one is interested in. And not all databases are equally reliable! I think that bioinformatics is one of the three legs of the proteomic stool. And the other two legs of the proteomic stool are mass spectrometry and sample preparation. And sample preparation includes all these separation we’re talking about.

I think that the data that’s generated, not just across our country but around the world, is being collected together. Protein databases are being expanded continuously, and of course the search programs, the new search program is usually developed rapidly or evaluated rapidly around the world. And there are various other kinds of databases that are being developed for proteins that have been found to be over expressed in association with certain diseases. So I think the data, we will benefit from coordinating our data.

On detecting low abundance proteins

“I think it’s still a very big challenge…”

I think it’s still a very big challenge that we’ve made almost no progress on it. I think that probably no one has detected an entire proteome. And every time a lab uses two different methods they obviously have a lot of the same proteins in both methods, but also each method turns up unique proteins. And this is evidence that we never see all our proteins, characterize and identify all our proteins and peptides. One can think about the minor abundance proteins in two directions. One is to try to pull them off preferentially, which can be done with immunoprecipitation or some kind of affinity methods, but only if we know what we’re looking for. And then the other approach is to try to remove the most abundant proteins, which of course is what people are doing with plasma. But we find other laboratories have reported that when they remove albumin from the plasma there’s a reproducible cohort of proteins, biomarkers that are associated with the albumin. And they think it can be called the albuminome. So I’m not sure that we want to even remove the top ten most abundant proteins in plasma and discard that entire fraction. The answer right now to getting at low abundance proteins is to use many multiple separation techniques; fractionation, fractionation, fractionation. And that of course is more work, may require more tissue or blood material, but does seem to give us access to a broader selection of proteins.

Well I think that our biggest problem is one we’ve already talked a little bit about and that’s the dynamic range of our analytical methods. Can we see the low abundance proteins in the presence of higher abundance proteins? Some believe or suggest that this is a limitation of mass spectrometry, one of the three legs of the stool, and that we need to work on that dynamic range as well, a sensitivity of our mass spectrometry detectors. That’s a big problem.

On the questions that can be answered with proteomics

“There are a lot of different questions we can ask…”

There are a lot of different questions we can ask with proteomics technologies and a lot of questions, which can really be answered much better with proteomics technologies. And those would include the search for biomarkers; include the questions about cell biology, such as what are the mechanisms by which cells develop to resist drugs? You can also ask questions in the more biochemical realm; what are the substrates of a particular protein, or what are the proteins which are altered by a particular mechanism, or whose substrates are altered by a particular mechanism? All these are in fact areas of research that are being followed up in labs around this country.

I think there’s some very important work to be done with proteomics that addresses fundamental biochemistry. What is the biochemical basis of drug resistance? That does lead to biomarkers if we understand what changes. We can then look for the change in that protein. And a lot of the folks who are interested in biomarkers recognize that when they find a candidate they have to go back and look at the biochemistry to make sure that it is a real biomarker.