Feature: Going rogue

This feature appeared in the July/August 2011 issue of Australian Life Scientist. To subscribe to the magazine, go here.

Mission director is Associate Professor Tony Purcell, an NHMRC Senior Research Fellow at the Bio21 Molecular Science and Biotechnology Institute at the University of Melbourne. His group’s brief (should they choose to accept it) is to understand exactly what it is that the immune system recognises in the context of autoimmune disease.

Type I diabetes, multiple sclerosis and rheumatoid arthritis are the three biggies on their hit list. What all such diseases have in common is an over-eager army of T cells floating around the body programmed to seek and destroy ‘self’ antigens , leading to damage and destruction of the body’s own tissues.

One of the big issues in tackling autoimmune disease is working out what went wrong in the first place, and Purcell’s part in that quest is working out what it is on the surface of the body’s own cells that these ‘rogue’ T cells actually see.

In most cells, any sort of antigen encountered is degraded by a series of proteolytic enzymes into constituent protein fragments or peptides. These are then bound by specific HLA (human leukocyte antigens) molecules in the cell and ferried to the surface for presentation to T cells as part of the immune response’s normal ‘screening’ program.

According to Purcell, it is like the immune system performing its very own proteomics experiment – an intracellular multiprotease digest followed by a shotgun proteomics experiment to sort out all of the peptides from the screening round for the ones it needs to worry about. It then differentiates the good guys (‘self’) from the bad guys (‘non-self’), which include pathogens like viruses and bacteria, cancer cells and other abnormal proteins.

Ideally, the good ones are left alone and the bad ones are shunted along to the next set of protective immune responses to be eliminated. According to Purcell, all protein turnover yields a complex display of peptides on the cell surface providing a snapshot of the cellular proteome and alerting the relevant T cells of the immune system to potential dangers such as infection or malignancy.

When immune cells go bad

To do the job it is designed for, our immune system is keenly attuned to subtle differences in antigen presentation and to remember those differences. “If a virus or tumour-derived peptide cell is thus presented, there are generally T cells in the body that recognise those peptides as foreign and order destruction of the affected cell before the infection or malignancy spreads,” he says.

“Meanwhile, a normal healthy cell will just show off pieces of constitutively expressed proteins that the immune system is usually well-educated enough not to recognise as self and therefore leave alone.”

---PB---

However, in autoimmune disease the normal situation breaks down. What interests Purcell are the sorts of peptide changes that can precipitate such a breakdown and what is the trigger: is it some sort of stress on the cell or some other factor that makes the cell start to present a different and less friendly face?

“We are particularly interested in defining what biochemical changes go on in the insulin-producing beta cells of the pancreatic islets during diabetes,” says Purcell. “The problems in diabetes really start when enough T cells accumulate around the islets to cause an inflammatory infiltrate – then comes the invasion of T cells into the islets where they start killing the beta cells and eventually causing symptoms.

“So, we have spent the last 15 years or so developing some quite high-end skills in mass spectrometry (MS) so that we can directly sequence and analyse the peptides involved in immune recognition.

“For example, we take beta cells from pancreas as well as professional antigen-presenting cells (APCs) such as dendritic cells from nearby lymph nodes and we grind them all up. We then affinity purify the HLA molecules on the cell surface to separate their bound peptide cargo, which we can then identify using MS.”

Over the last couple of years, Purcell and his team have really started to drill very deeply into this so-called ‘immunopeptidome’ – that is, those peptides bound to the HLA molecules. “When we started, we only got back a few hundred peptides at the end of the process, but new technology and instrumentation are now revealing tens of thousands of different peptides in a preparation. We are now really starting to get a good insight into these changes.”

Looking through T cell eyes

So far, Purcell’s work has revealed quite large quantitative changes in the types of peptides that are presented during diabetes. “For instance, there is a particular peptide in our mouse model of diabetes that is not detected until there is some sort of immune stimulation by cytokines, which are chemical immunoregulators released by activated cells. So clearly an initial (and perhaps chronic) inflammatory event can quite significantly change the sorts of peptides presented,” he says.

“T cells that recognise some of these peptide antigens, like the glucose-6-phosphatase-like molecule called IGRP from pancreatic islets, can be detected and measured in the blood of these mice and increase in frequency as disease progresses. So, this molecule becomes a potential prognostic marker to perhaps assess the disease status or progression.”

---PB---

Purcell adds that this sort of finding could potentially meet the strong need in humans for some sort of biomarker or assay. “For instance it could help in monitoring young kids at risk of developing type I diabetes. We use antibodies at the moment for detection of risk, which is helpful, but you often see antibody-positive individuals without disease symptoms and have no real idea of when disease may precipitate.

“Having a marker for the sort of immune responses going on could allow more effective immunotherapy to begin quite early, and those specific T cells causing the problem would be eliminated from the T cell repertoire from an early age in at-risk individuals. This would really be the holy grail of all these studies.

“One of the nice things about MS is that you can start to do some really good and meaningful quantitation,” says Purcell. “And in the last few years we have invested quite a bit of time in developing multiple reaction monitoring (MRM)-based assays, which are basically a very sensitive way of quantifying our data.”

This is a particularly important feature in the immunoproteomics field where some of the peptides they are hunting may be initially presented at very low levels before being ramped up.

“So, potentially there may be a quantitative threshold at which the body goes from ignoring a peptide presented on the beta cell surface to noticing a bit of a grumble, to finally triggering a bigger reaction to start these cascades of immune-mediated destruction. Having that quantitative ability is definitely a big bonus in detecting such thresholds.”

Diabetes focus in more detail

There are two main arms to the group’s current diabetes focus in proteomics and, according to Purcell, both are progressing well. “In the first we are looking at the effect of inflammation on protein expression in the beta cells of our mouse model.

“Holistically we are interested in how these changes in expression may lead to changes in peptides on the cell surface of antigen-presenting cells. In doing this, we are really moving towards more of a systematic understanding of what goes on, and this something that the MS approach is really good at addressing.

“For instance, we are finding that compared to the self peptides, these disease-related peptides are in very low abundance, and at the assay level are usually swamped by everything else going on in the cell. However, in the same way that the immune system is amazingly good at picking out these low-abundance or relatively subtle peptide changes – needle in a haystack territory – we need to be also, and we think MS is the way to go.”

---PB---

The group is also getting mathematical with their immunity questions – taking a systems biology approach – and Purcell is very excited about taking this forward over the next 12-18 months.

“Asking questions like: ‘how many different presented peptide complexes are there’; ‘how many T cells are generated’; and ‘what is it about these changes in antigen presentation that leads to the immune outcomes that we see in disease’? For immunotherapy to be feasible, we want to be able to monitor relatively simple things non-invasively, such as taking peripheral blood from patients to enumerate T cells or to monitor the treatment efficacy.”

The other focus lies more on the human immunogenetics and antigen presentation side of things, involving massive sequencing runs of the isolated peptides. “Indeed, we are now getting around 20,000 different peptides from these cell lines and this is a lot more than anyone else is able to do at the moment.

For example, such experiments are helping us understand how different disease-susceptibility HLA molecules for type I diabetes might be selecting different peptides to bind and present, with one peptide menu definitely more popular than another in terms of immune response.

“To work out which are the really important genetic differences at the allele level for disease susceptibility, we are also feeding known autoantigens, such as pro-insulin to our cells to see the scope of different peptides and different T cell responses that correspond to different alleles. We are seeing definite differences, which might ultimately reflect a functional difference in risk of disease.”

Purcell’s group is one of only a handful of groups in Australia and a handful worldwide looking at the proteomics of HLA-associated peptides.

“This is mainly because it is hard stuff to get right and to keep going. You need dedicated equipment and you constantly have to push the envelope in terms of specificity and sensitivity to get more out of it. We have really spent the last 15 years building up the expertise and the bioinformatics know-how, as well as acquiring the latest and greatest instrumentation.”

This build-up has certainly paid off: every new equipment upgrade has meant tremendous leaps in the numbers of peptides detected. For example, in the latest upgrade Purcell went from seeing about 500 on average to about 4000 per run, and each time reveals a whole new subset of molecule types.

“We are now starting to see really low-abundance peptides that we need to see to identify the disease-relevant and tissue-specific sets. There is obviously a whole lot of validation that has to go on with each new finding, but one mantra we have in our lab is that the mass spec never lies – it gives you a mass, it gives you a fragmentation pattern that you derive sequences from, and then it comes down to the interpretation. But the answer is always there… somewhere.”