Feature: Protein treasure hunt

Tissue banks in hospitals and research centres round the world hold a resource of profound importance to cancer researchers: a vast archive of tumour samples, stretching back decades.

Fixed in formalin and stored in paraffin wax, they were preserved for purposes of studying pathological changes in cell morphology and organisation, long before the word “proteomics” was coined in 1997.

However, the advent of the new science of proteomics has made them a potential treasure trove of information concerning the macromolecular changes involved cancers, such as abnormal expression and distribution of proteins within cells and the extracellular matrix during tumour development and metastasis.

But how to identify proteins in situ? There’s a chicken-and-egg problem – a double-yolker, at that. A huge selection of off-the-shelf antibodies are available to highlight the distribution individual proteins, but the molecular oncologist’s choice of antibodies relies on knowing in advance what proteins are likely to be present in the tissue sample.

Dr Peter Hoffman is Director of the Adelaide Proteomics Centre, a joint venture involving the University of Adelaide and Adelaide’s Hanson Institute. His research team is developing new methods for identifying, quantifying and mapping proteins in tissue samples, using tandem mass spectrometry.

The complexity of the human genome, particularly the unique diversity of splice variants of the proteins encoded by the 20,000-odd human genes, means that human tissue samples are usually more heterogeneous than those of laboratory animals. Some commercial antibodies for some proteins have poor specificity and may provide negative or false results.

Moreover, antibodies to unique, mutant or alternatively spliced proteins simply may be unavailable commercially, yet these are the very proteins that are of greatest interest because of their potential to be used as diagnostic and prognostic markers for different forms of the same cancer.

Dr Peter Hoffmann, Director of the Adelaide Proteomics Centre, at the University of Adelaide’s School of Molecular and Biomedical Science, heads a team that has made enticing progress in imaging proteins and peptides in formalin-fixed tumour samples, using MALDI-IMS, a hefty acronym that stands for matrix-assisted laser desorption/ionisation (MALDI) imaging mass spectrometry (IMS).

According to Hoffmann, MALDI-IMS can also image drug molecules in situ in tumour samples, highlighting sites of drug activity, and illuminating mechanisms of resistance. “Basically, we use thin tissue sections mounted on glass slides, and use the MALDI mass spectrometer to perform a raster screen, giving us the molecular signatures of all proteins,” he says. “We then create a 2-D image of each protein’s distribution from the raster data.”

One potential application is in analysing biopsied tissue samples from patients who have undergone surgical removal of tumours. MALDI-IMS could provide some surety that surgery has been successful by providing high-resolution images of the borders between cancerous and normal tissues, to ensure that a tumour has been completely excised.

---PB---

If the surgeon has excised a tumour with a 2 cm border as a safety margin, and MALDI imaging detects a mobile marker protein associated with the tumour invasion in adjacent, apparently normal tissue, this might indicate undetected tumour cells. The surgeon might then opt for a larger safety margin next time.

“It goes even further,” says Hoffmann. “Once you’ve done the imaging, you can see how some patients have responded to chemotherapy, while others show no response.

Tissue microarrays

MALDI-IMS has other applications too. “You can take tissue samples from 100 cancer patients, punch out sample cores from the tumour area and create a tissue microarray,” says Hoffman.

“A single microscope slide can contain more than hundred different patient samples. You might include samples from 50 responders and 50 non-responders, and look for consistent proteomic differences between them.”

For biomarker discovery, MALDI-IMS can detect candidate protein markers in the cancer tissue and its immediate surroundings. Researchers can then look for the same proteins among the plethora of familiar and anonymous proteins in the bloodstream.

An HE-stained thin section of an ovarian tumour, with two MALDI-INS images of the same sample. The first, based on the signal-to-noise ratio, highlights cancerous tissues surrounded by normal tissue. At right, a sample from the same ovarian tissue shows the absolute intensity of the signal, clearly delineating malignant tissues.

By comparing sera from healthy subjects with sera from patients at different stages of tumour progression, they can also evaluate their potential as cancer biomarkers. A French research group has recently identified a potential biomarker for ovarian cancer using just this approach.

“We’ve been working a lot on ovarian cancer, which is notorious for being asymptomatic until it begins to produce secondary tumours in the abdominal cavity – and by then the chances for full recovery are slim,” says Hoffman. “But we don’t yet have any candidate biomarkers for ovarian cancer.”

Hoffmann and his group have developed MALDI-IMS to the point where it can be used for research on formalin-fixed, paraffin embedded tumour samples, which was no trivial task. Formalin cross-links peptide sequences within proteins, and the links must removed to facilitate analysis.

The protein-digesting enzyme trypsin is then introduced into the sample to digest the freed-up proteins and dice them into peptides of a length suitable for identification by mass spectrometry. The matrix material required to absorb and transfer laser energy to desorb and ionise the peptide fragments is added at the same time.

The so-called tryptic peptides can be analysed using time-of-flight (TOF) mass spectrometry to detect their distinctive mass-charge signatures. The identity of the parent protein can then be reconstructed from the peptide sequence. Many proteins can be readily identified from unique peptide sequences.

---PB---

Treasure hunt

But that still leaves the problem of knowing which proteins to look for. These must be selected from a dauntingly large field of at least 20,000 primary proteins, with up to 10 splice variants each, and a host of possible protein modifications, such as adding sugars, lipids and phosphates to certain amino acids, or enzymatic cleavage of the protein to modify its function.

Researchers need a “ground truth” list of the actual sub-set of proteins and peptides expressed by normal and cancerous cells in the tissue sample, as a basis for comparing and mapping the proteins mapped by MALDI-IMS in the tissue sections.

A colon cancer tissue sample after conventional haematoxylin-eosin staining, showing cancerous epithelial tissue of the bowel lining, overlying normal muscle tissue. The three MALDI-INS images highlight the distribution of cancerous cells in the tumour, and associated disordered connective tissues and in the normal muscle tissue.

Hoffmann’s team employs liquid chromatography coupled to tandem mass spectrometry (LC-MS/MS) for this task. The tryptic peptides from the protein digest are extracted, separated with a nano-high-performance liquid chromatography, collected as droplets containing nearly pure fragments, and dried on a sample plate for mass spectrometric analysis.

The dried samples are then ionised and identified by tandem mass spectrometry, and with the help of in silico data processing, the parent proteins can be reconstructed from the peptide data.

“Many hospitals have huge archives of formalin-fixed paraffin embedded samples covering several decades,” says Hoffmann. “We now have methods for digesting the peptides and extracting them from formalin-fixed tissues and using microarrays to image multiple tissues at the same.

“We have a much larger number of patients’ tissue samples available today, so we should be able to find diagnostic and prognostic markers and validate them in a relatively short time frame. Time-series samples are available for many patients’ cancers, giving have the potential to image and quantify proteins in the primary tumour, secondary tumours, and tumours that have become resistant to chemotherapy.

“People are dying because their chemotherapy doesn’t work any more. We now have the means to answer questions about why a cancer comes back for a second or even a third time in some patients, while another patients experiences prolonged or permanent remission.

“If the oncologist detects markers for resistance to particular chemotherapy agents early in a patient’s treatment, they can switch to another frontline therapy. In families with a history of colorectal cancer, there’s a progression from normal tissues to abnormal but benign polyps, that may or may not become cancerous.

“With MALDI-IMS, we hope to determine why some family member’s abnormal polyps become cancerous, while those of other family members don’t. If we can understand the changes involved in progression to cancer, and identify the risk factors involved, it should be possible to predict whether the patient’s polyps are at high or low risk of becoming colonic tumours.”

MALDI-IMS

MALDI (matrix-assisted laser desorption/ionisation), is laser-based, soft-ionisation technique that has become one of the most successful ionisation methods for mass spectrometric analysis and investigation of large molecules.

It is used to analyse tissue samples embedded in a chemical matrix that releases intact, gas-phase ions from large, non-volatile and thermally labile compounds such as proteins, peptides, nucleotides, synthetic polymers and organic molecules, such as metabolites and drugs.

A pulsed ultra-violet or infra-red laser doubles as the desorption and ionisation source. The matrix material absorbs the laser beam’s energy, causing some of the substrate to vaporise. The ionised molecules are then fed into an mass analyser that separates them for identification by time-of-flight (TOF) mass spectrometry.