Beating the abundance problem

Sri Bandhakavi discusses how a novel approach to low abundance protein enrichment can enhance studies aimed at the identification of proteins and their differential analyses

Changes in the abundance levels of proteins and their post-translational modifications (PTMs) are linked to various disease states and detecting these disease-specific changes is important for effective health and diagnostic monitoring. However, several of the proteins underlying health and disease transitions are typically lower in abundance and their detection is often a challenge. This challenge is further magnified in bodily fluids, where disease/health ‘biomarkers’ are typically diluted (compared to their levels in affected cells/tissues). Consequently, enhancing ability to detect changes in low abundance proteins and their PTMs can improve health diagnostics.

Even when highly sensitive mass spectrometers are used to analyse complex biological samples (e.g. bodily fluids), owing to the wide dynamic range of protein abundance, high-abundance proteins obscure the detection of lower-abundance proteins and their PTMs. Thus, the low-abundance proteins that are potentially associated with health or disease status remain undetected. Protein/PTM dynamic range is particularly wide in bodily fluids (plasma, saliva, urine etc), thus hampering the sensitivity of biomarker and health diagnostic studies.

Hexapeptide libraries (Bio-Rad, ProteoMiner) effectively compress protein dynamic range. This significantly enhances mass spectrometric detection of low-abundance proteins. But what about the differential analyses of proteins or identification of low-abundance PTMs? Recent research demonstrates that Dynamic Range Compression (DRC)/ProteoMiner technology boosts differential analysis studies wherein researchers might be interested in identifying disease-specific changes to low-abundance proteins/PTMs.

In the laboratory of Timothy J Griffin at the University of Minnesota, we were interested in exploring the diagnostic potential of human saliva, as a non-invasive alternative to human plasma and serum. Towards this goal, we initially combined hexapeptide libraries/DRC treatment of human saliva with mass spectrometry, and significantly expanded the known proteome of human saliva¹. Bioinformatic analysis of this dataset revealed that saliva had several proteins with system-wide diagnostic potential meaning that it was implicated in systemic diseases and not just those restricted to the oral cavity. As a test of this bioinformatic prediction, we sought to identify breast cancer associated changes to salivary proteins. Of particular interest were changes to low-abundance proteins in saliva that have been previously implicated in breast cancer from plasma/serum and breast cancer tissue profiling studies (where they are present at higher abundance levels and hence more easily detectable).

From our 2009 study, these low-abundance proteins could be detected in saliva only by the addition of DRC to existing peptide fractionation/mass spectrometric workflows. When tryptic digests of untreated human saliva were fractionated and analysed by mass spectrometry, these proteins are not detected. Thus, we incorporated DRC/ProteoMiner technology via a novel workflow to identify potentially breast cancer associated changes to these proteins in human saliva².

[caption id="attachment_24352" align="alignright" width="200" caption="Each bead features a different hexapeptide ligand with affinity for specific proteins in a sample. Samples are applied to the beads, allowing proteins to bind their specific ligands. Proteins in excess are washed away, and those proteins bound to the beads are eventually eluted, allowing further downstream analysis"][/caption]

ProteoMiner technology consists of millions of unique hexapeptides immobilised on beads; each bead has potentially unique protein-binding properties. It causes partial depletion of high-abundance proteins and simultaneous concentration of low-abundance proteins, resulting in DRC of complex samples. Consequently, high-abundance proteins do not retain quantitative accuracy, while lower-abundance proteins remain quantifiable post-DRC. Distinguishing between these classes of proteins is critical for successful differential analyses when employing DRC.

We compared saliva samples from healthy women with those with metastatic breast cancer, and processed the samples with DRC (ProteoMiner treated saliva) and without DRC (untreated saliva). Tryptic peptides from untreated and treated saliva samples were stable isotope labelled and processed via multidimensional peptide fractionation followed by analysis on an Orbitrap XL mass spectrometer. All proteins identified/quantified without treatment (i.e. in untreated saliva) were designated as ‘high-abundance proteins’. Those proteins identified/quantified only after DRC were designated as ‘low-abundance proteins.’

The workflow we developed categorised salivary proteins with higher absolute abundance (i.e. identified without DRC and whose quantitative ratios are altered by ProteoMiner treatment) from proteins of lower absolute abundance (identified only after DRC/ProteoMiner treatment). This enabled us to identify low-abundance proteins that were affected in the breast cancer samples and retain quantitative accuracy following treatment with the ProteoMiner kit. Consistent with the mass spectrometric results, putative breast cancer associated changes to proteins from each of these categories were identified and validated independently.

A variety of diseases are also characterised by specific post-translational modifications (PTMs) caused by altered signal transduction networks. However, detecting PTMs is also challenged by the wide dynamic range of protein abundance and their sub-stoichiometric nature. Hence, we tested the potential for DRC/hexapeptide libraries in enhancing detection of PTMs in human saliva.

Coupling DRC of saliva with covalent glycopeptide enrichment and tandem mass spectrometry enabled us to significantly expand the known salivary N-glycoproteome. With DRC, we identified two times more N-linked glycoproteins and their glycosylation sites than without DRC. In a single study, 193 N-glycoproteins were identified in saliva, compared to the previously known ~ 60 salivary N-glycoproteins.  Based on this result, we postulated that ProteoMiner technology should also boost detectability of other biologically interesting PTMs such as phosphorylation.  This was shown to be true by work done by Matthew Stone and colleagues at the University of Minnesota³.

The workflow is readily transferable to diverse samples, signalling the appropriateness of including ProteoMiner technology in differential analyses of clinical samples or cellular lysates. Given our ability to identify proteins putatively associated with breast cancer, the study further highlights the systemic diagnostic potential of saliva as a non-invasive alternative to serum, long the gold standard for biomarker discovery. Our proof-of-concept also demonstrates wider applications of ProteoMiner/DRC technology for PTM detection as part of disease diagnostic efforts, particularly in bodily fluids.

Unlike immunodepletion methods that are sample-specific, ProteoMiner can be used with a variety of samples and across different species – it is, for example, also applicable in situations where antibody depletion tools might not exist. Its ability to work with a variety of samples makes it much more suitable for studies that go across different models. For example, a biological researcher who needs to work with different models through the different stages of the research, for example mouse, dog, monkey, human, can use ProteoMiner throughout the research study. The technology is an essentially universal approach for protein dynamic range compression across biological samples.

In spite of the versatility of the ProteoMiner technology, there is still much to be explored. We are now focusing our efforts on developing additional applications for a variety of samples. Currently we are exploring its ability to work in the presence of reagents such as SDS and fluoroalcohols for solubilising intractable proteins such as membrane proteins. We have proven the basic concept that it functions in the presence of certain concentrations of these reagents, and are investigating further applications along these lines.

The wide dynamic range of protein expression/abundance in various samples has been a serious problem for biological and proteomics research. Our results should expand the use of DRC with hexapeptide libraries as a general tool for increasing protein detection, PTM characterisation, and relative protein/PTM abundance pro?ling using mass spectrometry-based proteomics.

References

1. Bandhakavi et al., 2009 J Proteome Res. 2. Bandhakavi et al., 2011 J Proteome Res. 3. Stone MD et al., 2011 J Proteome Res.

Author: Sri Bandhakavi, Senior Scientist, Bio-Rad Laboratories