Proteomics, morphoproteomics, saliva and breast cancer: An emerging approach to guide the delivery of individualised thermal therapy, thermochemotherapy and monitor therapy response

Abstract

The field of proteomics is in its infancy; however the discipline, its technology, and our abilities to translate the proteomic data are rapidly evolving. In the near future proteomics should significantly improve our ability to make early cancer diagnoses, direct appropriate personalised therapy, and monitor response to therapy, including thermal therapy. The potential role of proteomics in breast cancer early diagnosis, prediction of aggressiveness is clear. Its potential importance in guiding treatment choice and prediction of treatment response is especially intriguing. This paper reviews the varied methodologies used in the field of proteomics, including gel-free, label-free proteomics, quantitative proteomics, phosphoproteomics, protein extraction from formalin-fixed, paraffin-embedded tissue sections (FFPE) proteomics, laser capture microdissection proteomics, and targeted proteomics. It also discusses two new areas, morphoproteomics and salivary proteomics cancer diagnostics, as well as selected pre-clinical and clinical analyses using the described methodologies. Morphoproteomics defines which signal transduction pathways exist within the tumour cells and the surrounding tissue comprising a patient's cancer biopsy specimen. Morphoproteomics, and the other histology-based proteomic techniques are actually beginning to clinically make possible individualised treatment of breast cancer.

Salivary proteomics, in part because it is non-invasive, is a new area of breast cancer diagnostics that can be used to non-invasively monitor an individual patient's response to treatment with every treatment cycle. The current literature demonstrates that a diagnosis of breast cancer can be readily made using proteomic methodologies, and that proteomics can also define cancers with a poor prognosis at the time of diagnosis. With such early prognostic information we expect proteomics will soon be a science that on the basis of prognosis, guides individualised therapy and as well, have the ability to monitor the results of thermal therapy, radiation, and chemotherapy treatment during therapy.

• morphoproteomics
• proteomics
• saliva
• signal transduction pathways

Introduction

Breast cancer is the most common cancer and the second leading cause of cancer death in American women [11]. Despite advances in early detection and the understanding of the molecular bases of breast cancer biology, approximately 30% of patients with early-stage breast cancer progress to recurrent disease after primary therapy [12]. We increasingly use information related to the specific molecular characteristics of the tumour to distinguish patients with aggressive tumours and bad prognoses. Information about prognosis allows the oncologist to select the most effective therapies for an individual patient [12], [13]. While we have recognised the important clinical prognostic factors associated with therapy outcome such as the size of the primary tumour, and the presence of positive axillary lymph nodes for more than 30 years; our understanding of the molecular markers of a tumour is much more recent. It is becoming important to recognise entire signal transduction pathways to predict the aggressive behaviour of the cancer as well as guide the choice of a particular therapy [14]. These pathways can be analysed by proteomic techniques[13].

The proteomic analysis of blood, urine, and tissue continues to be a rapidly evolving field [15], [16] and is expected to become an exceptional diagnostic tool for cancer and other disease. This has engendered a great deal of interest in the field [16], [17]. In addition to its use as a diagnostic tool, proteomics will also be used to guide the choice of individualised treatment for a patient based on identification of proteins and entire signal transduction pathways in that primary or metastatic tumour [18–20]. We expect proteomics, as well, allow us to monitor an individual's response to therapy, whether the particular treatment is thermal therapy, thermochemotherapy, chemotherapy, targeted therapy or radiation [21], [22].

Proteomic methodology is capable of demonstrating that radiation has been administered, and can also quantify the radiation dose [23]. Proteomics can also predict resistance to radiation [24]. Since heat is a form of radiation, we anticipate proteomics to have a role in ascertaining what dose of heat has been administered to a primary breast cancer, or to metastases. We also expect that proteomics will allow us to understand why thermal therapy increases the cytotoxicity of chemotherapy and radiation.

The methods used for proteomic analyses include gel-free, label-free proteomics [1–3]; quantitative proteomics; phosphoproteomics [4], [5]; protein extraction from formalin-fixed, paraffin-embedded tissue sections (FFPE) proteomics [6], [7]; laser capture microdissection proteomics [7], [8]; and targeted proteomics [9], [10]. Additionally, two new research areas of proteomics are morphoproteomics and salivary proteomics cancer diagnostics [14], [25], [26].

Morphoproteomics

Morphoproteomics and its method variations is a developing new tool in pathology. A major value of morphoproteomics is its ability to analyse signal transduction pathways within tumour cells. As the name ‘morphology’ suggests, signal transduction pathways can be identified within cells within an anatomic location within a tumour. The cellular anatomy allows the determination where the cellular pathways are located within, and proximal to the tumour. Identifying signal transduction pathways can potentially determine an effective therapy for an individual patient before beginning treatment [25], [27], [28].

The pathological determination of oestrogen and progesterone receptor positivity is one of the most critical analyses in both primary breast cancer and in breast cancer metastases. The immunohistochemical analysis of tumour hormone receptor status has been routine for more than 20 years [29], [30], and determines whether an anti-oestrogen drug (tamoxifen/fulvestrant), or an aromatase inhibitor will be used as either primary or as adjuvant therapy.

Equally important is the analysis of the presence or absence of amplification of the human epidermal growth factor receptor-2 (HER2/neu, or c-erb-2B). While this analysis can be immunohistochemical, the more quantifiable assay uses a molecular technique of fluorescent in situ hybridisation (FISH) to assess the level of the plasmalemmal (cell membrane) expression of HER2/neu total protein to determine HER2/neu gene amplification. Amplification of the gene dictates using with trastuzumab, a targeted therapy alone, or most commonly with a neoadjuvant, or adjuvant chemotherapy regimen, or as first, second, or third line chemotherapy regimens for metastatic breast cancer [31], [32]. Concomitantly, if the HER2/neu gene is amplified, anti-estrogens (tamoxifen/fulvestrant) will not as readily be used as single agents because of decreased efficacy of hormone therapy in the presence of HER2/neu amplification [27]. Without trastuzumab the antiestrogen drugs are often ineffective even if the tumour is ER+/PR+ [33], however, recently fulvestrant has been combined with trastuzumab in tumours that are ER+ and HER2/neu positive with improved efficacy compared to fulvestrant alone [34].

Yet, the predictive value of these latter molecular assessments of HER2/neu, they must be viewed in perspective. Despite the finding of a grade 3 plasmalemmal expression of HER2/neu amplification, trastuzumab, as a single agent has only a 15% response rate. It is true, however, that the response rate to chemotherapy combined with trastuzumab of HER2/neu amplified tumours is much higher [32]. Nevertheless, the results of trastuzumab administered as a single agent in an individual patient after routine pathological testing of HER2/neu amplification suggest that the test does not have good utility in predicting response to the monoclonal antibody targeted drug. An analysis of the full signalling pathway of HER2/neu would increase the ability of the test to predict response to trastuzumab.

Another important clinical area that points out the importance of understanding the entire signalling pathway to effectively treat the cancer is colon cancer treated with targeted monoclonal antibodies. Cetuximab and panitumumab are monoclonal antibodies that target the epidermal growth factor receptor (EGFR). These new agents have increased treatment options for metastatic colorectal cancer. Yet, the initial evaluation of these antibodies as single agents in patients with EGFR-expressing chemotherapy-refractory tumours elicited only a 10% response rate. The realisation that detection of positive EGFR expression does not predict clinical outcome of EGFR-targeted treatment dependably led to an intense search for alternative predictive biomarkers. Kirsten Ras sarcoma viral oncogene protein (KRAS) mutations, are present in 45% of patients with colorectal cancer. Malignant activation of signalling pathways downstream of the epidermal growth factor receptor (EGFR), such as the mutation in the KRAS oncogene is now recognised as critical to colorectal cancer growth. Documenting KRAS mutations have become an important predictive marker of resistance to panitumumab or cetuximab therapy. There are additional mutations that appear to be related to resistance to EGFR-targeted monoclonal antibody treatment. This entire signalling pathway can be assayed by morphoproteomic analysis for an individual patient. A more complete knowledge of the molecular basis for sensitivity or resistance to EGFR targeted monoclonal antibodies should allow the development of new treatments and may also provide rationale for combining therapies to overcome primary resistance [35].

The application of morphoproteomics involves the immunohistochemical assessment of the activation of signalling as well as metabolic pathways in cancer cells, in order to predict susceptibility to monoclonal antibodies, small-molecule inhibitors, and specific chemotherapeutic agents. It must be recognised that the morphological component of protein analysis is essential. By using morphologic analysis, the translocation of a protein from one sub-cellular compartment to another can be assessed. For example, if the protein is translocated from the cytosolic compartment to the plasmalemmal compartment it signifies the activation of certain proteins. Moreover, morphoproteomics examines the phosphorylation of a protein analyte using phosphospecific immunohistochemical probes directed against sites of putative activation of a given molecule. As an example, in the Akt protein, a serine/threonine kinase, serine 473, is one of the activation sites. In p70S6-kinase, threonine 389 is an activation site. In nuclear factor (NF)- κBp65, the activation site is serine 536. These activation sites of signalling molecules can be detected using bright-field microscopy and phosphospecific immunohistochemical probes [36]. Morphoproteomics details the correlative expression of the proteins, upstream and downstream along signal transduction pathways, to determine expression in the same tumour [25].

Morphoproteomics can help explain why certain drugs might be applicable in an individual patient's tumour. And notably, morphoproteomics can guide specific treatment combinations to optimally treat a particular patient's tumour by suggesting combinations of cytotoxic agents, thermal therapy, radiation therapy, chemotherapy agents, small molecules, and ‘targeted agent’ signal transduction inhibitors on the basis of signal transduction pathways [14], [37].

Of emerging importance, morphoproteomics can also analyse the context of what is happening in the tumour cell with what is happening in the surrounding cell population [25]. Morphoproteomic analysis demonstrated the interaction of tumour cells with surrounding non-tumour stromal tissue. It can determine what is occurring in the endothelial cell, and what is happening in the stromal cell. These analyses potentially can give us important information about how we might direct therapies.

Pavlides et al. described a highly sophisticated example of the use of stromal cell-based proteomics technology [23]. They analysed stromal cells, specifically myo-fibroblasts and cancer-associated fibroblasts, to determine the effect of the caveolin (CAV) gene of these stromal cells on breast cancer progression in mice, and in human breast cancer as well as the gene's effect on response to therapy [23]. They used stromal cells derived from Cav-1 (-/-) null mice to identify which proteins are selectively up-regulated by an absence of Cav-1.

They performed unbiased proteomic analysis and genome-wide transcriptional profiling of Cav-1 (-/-) stromal cells. The immuno-staining of human breast cancer samples that lack stromal Cav-1 expression were then used to validate results from this screening approach. Using their proteomic method, they identified >25 candidate stromal biomarkers that are up-regulated by a loss of Cav-1 in stromal cells. Interestingly, these proteins include five myofibroblast markers, three signalling molecules, one oncogene, eight metabolic and glycolytic enzymes, as well as three extra-cellular matrix proteins–known to be associated with fibrosis and tumourigenesis. These studies elucidate the close relationship of myofibroblasts, as well as cancer-associated fibroblasts to tumour initiation, progression, and metastasis. The active communication between tumour and stroma is detrimental to the host. Their studies suggest new therapeutic strategies to target the tumour microenvironment [23], as well as another potential use of proteomic analysis to follow similar protein and gene responses to therapy, including thermal therapy combined with radiation and chemotherapy.

Using another technique to combine histology with proteomics, Amann et al. used a matrix assisted laser desorption/ionisation time-of-flight (MALDI-TOF) MS compatible fixation and staining techniques for high-resolution cellular morphology that allowed selective identification of tumour cells [6] in tumour biopsy samples. They found protein profiles of fine-needle aspirates to be highly reproducible and similar to the pathological profiles of the tissue from which they were obtained. They were able to differentiate xenograft tumours derived from two different human cell lines (A549 and H460). They concluded that the procedure resulted in high-quality, cancer cell-specific protein profiles. The authors suggested that the highly reproducible technique could be applied to other types of clinical samples and had potential to be used in clinical diagnosis, classification, and potentially to individualise cancer treatment [6].

Imaging proteomics

While molecular histology is acknowledged to have important value in the field of pathology, imaging mass spectrometry brings a new dimension of molecular data to molecular histology. Cellular histology is also critically important in morphoproteomics [14], [25].

Profiling and imaging mass spectrometry is a technology that provides new insights to molecular processes ongoing in living systems [38]. The classification of cancerous tissue based on the Matrix-assisted laser desorption/ionization mass spectrometry (MALDI/MS) protein profiles is one of the growing diagnostic applications of the field of proteomics [38], [39]. Tumours are typically categorised by non-invasive techniques before surgery. However, one of the newer techniques is a combination of magnetic resonance imaging (MRI) imaging with magnetic resonance (MR) spectroscopy, and brain metabolite information [40].

Chaurand et al. described the technique and usefulness of direct tissue profiling combined with imaging mass spectrometry to provide a comprehensive assessment of the complex protein pattern in a tissue sample [41]. Using MALDI mass spectrometry (S), the authors profiled and imaged thin tissue sections and found more than 500 individual protein signals in the mass range of 2 to 70 kDa that directly correlated with protein composition within a specific anatomic region of the tissue sample. MS was applied to multiple diseased tissues, including breast cancer, human gliomas and non-small-cell lung cancer [41]. Chaurand et al. combined modern biocomputational tools to analyse the complex MS data sets and this allowed the investigators to identify both disease-state and patient-prognosis-specific protein patterns. The potential of this type of molecular imaging technology is considerable. Their results suggest that this type of proteomic information will become more and more important in assessing disease progression, prognosis, and drug efficacy [42]. The fundamental contributions of the technology to rapidly provide molecular weight specific profiles and images at high resolution and sensitivity can provide important data for the investigation of cellular processes in both health and disease. Their studies suggest the usefulness of proteomic information in assessing disease progression, prognosis, as well as drug efficacy.

Crecelius et al. used a somewhat different proteomic methodology to visualise the corpus callosum of the mouse brain. They used matrix-assisted laser desorption ionisation imaging mass spectrometry (MALDI IMS) data combined with optically determined tissue structures in a 3-D reconstruction of myelin basic protein (MBP) [39]. Their technique consists of obtaining optical images from serial coronal sections. The optical images are first aligned to each other to reconstruct a surface of the corpus callosum from segmented contours of aligned images. The MALDI IMS data are then co-registered to the optical images and superimposed into the surface to create the final 3-D visualisation. The researchers found that correlating proteomic data with anatomical structures provided a comprehensive understanding of healthy and pathological brain functions [39]. They correlated changes to local protein distributions in brain caused by diseases such as cancer, Alzheimer's dementia, and Parkinson's disease [39]. Cross-correlating this approach with a classification based on differentially expressed proteins employing MALDI MS could improve the diagnosis of tumours. Changes in protein expression induced by a drug, the spatial localisation of drugs and the resulting changes in protein expression can give important in vivo biological information. This technique would also permit the visualisation of tumour heterogeneity on a proteomic level in addition to commonly used histological characterisation [39].

Neurological diseases, including leptomeningeal carcinomatosis and multiple sclerosis (MS), cause changes in the functioning of the endothelial and epithelial brain barriers giving rise to disease-associated alterations of the cerebrospinal fluid (CSF) proteome [43]. Examining the cerebral spinal fluid, Noben et al. pooled and ultrafiltered CSF of patients with MS and non-MS patients. The CSF fluid was enzymatically digested and analysed by off-line strong cation-exchange chromatography (SCX) coupled to on-line reversed-phase LC-ESI-MS/MS. In an alternative approach, the trypsin-treated sub-proteomes were analysed directly by Liquid chromatography electrospray ionisation tandem mass spectrometry (LC-ESI-MS-MS) and gas-phase fractionation in a mass spectrometer. The proteomic approach in combination with a three-step evaluation process identified 148 proteins in CSF. Sixty proteins were identified in CSF for the first time [43].

Saliva

Until recently blood, urine and tissue have been the only substances analysed by proteomics. However, saliva has now been shown to be a useful diagnostic fluid to analyse proteomically in breast cancer [17], [44]. Saliva has several important advantages compared to blood [45].

With recent developments in proteomic technology, saliva has become a ‘diagnostic fluid’. With the use of two dimensional gel electrophoresis and mass spectrometry technology, the protein catalogue of saliva has expanded several hundred-fold [45], revealing proteins that are altered in presence of cancer [46].

The need to overcome invasive methods of specimen collection and labour-intensive analysis is important. The logistical advantages of salivary diagnostics are that the collection of saliva is safe, relatively simple, and non-invasive (e.g. no needle punctures or even more invasive biopsies). Additionally, saliva may be collected repeatedly without discomfort to the patient [47].

It is of practical importance that saliva is not as complex a media as blood [47]. Since blood possesses more proteins than saliva, assaying trace amounts of ‘factors’ (e.g. oncogenes, etc.), may result in a greater risk of non-specific interference and a greater chance for hydrostatic (and other) interactions between the factors and the abundant blood proteins. For example blood possesses numerous carrier proteins, such as albumin, which must usually be either removed or treated prior to being assayed for protein content.

Secondly, the concentration of proteins in blood can vary over nine orders of magnitude. This complicates the detection of low abundant proteins.

A third advantage to salivary proteomics is temporally related. An analysis of saliva yields a protein analysis at that precise instant in time in contrast to blood. Blood is composed of peptides, proteins and cells that have half-lives ranging from seconds to weeks or even a month or more. As a consequence, the presence of a given substance might not accurately reflect the current state of the patient. In contrast, the physiology of saliva is such that the flow of secreted fluid is continually flushing and refreshing the fluid content of the mouth. Therefore, the composition of the fluid at any moment temporally reflects the metabolic activity of the secretory elements generating that fluid.

Finally, because of blood's direct connection with multiple organs of the body, the co-morbidity effects of one disease can influence the protein profile of another, and obscure cancer related biomarkers. For example, will a cancer-related biomarker be the same in a patient with only cancer as a pathological diagnosis if compared to another patient with cancer being treated for HIV? Although there is little information regarding co-morbidity effects on saliva, there is reason to predict that the variable of co-morbidity may not be as influential with salivary diagnostics as it can potentially be with blood [17]. More research is needed in this area.

Another advancement will be to understand the underlying mechanisms associated with how the biomarkers enter the saliva. For example, how does a 185 kDa protein enter into saliva? Also, why are there so many protein fragments in saliva? Most importantly, how are cancerous tissues communicating with the oral cavity and saliva? To date, this remains poorly understood, yet an understanding of these processes would shed light on cancer progression as well as treatment intervention.

Additionally, there are serious limitations associated with many current diagnostic methodologies in the field of cancer diagnosis and follow-up treatment. Our current diagnostic methodologies involve physical exam, laboratory blood counts and chemistries, and most importantly radiological monitoring of tumour size. Radiological measurements of change require several months to determine whether a therapy is efficacious. Using the advancement of proteomic research, saliva is being investigated as a potential diagnostic medium to non-invasively monitor tumour growth. Saliva can be queried early, and repeatedly after a therapy is administered rather than waiting months to follow tumour size.

To monitor therapeutic response there will need to be many changes in the field of salivary proteomics for cancer detection. The first advance will be a standardisation of salivary collection and proteomic techniques. This is an important step, as at present the salivary literature is a mixture of techniques. A consensus is being developed to determine the proper procedures for salivary biomarker discovery [48–50]. Another advance will be the cataloguing of both ‘high and low’ abundant proteins in saliva.

Recently a consortium of three research teams catalogued the proteins in the saliva of normal individuals [50]. Collectively they identified 1166 different proteins; 914 in parotid saliva and 917 in submandibular/sublingual saliva. They found that a high proportion of proteins found in blood and tears are also present in saliva along with other unique proteins. The proteins they identified in saliva were involved in molecular processes ranging from cellular structural functions to enzymatic/catalytic activities. As one would expect, the majority mapped to the extracellular and secretory compartments. Data derived from their work can be used to construct clinical laboratory tests that use saliva rather than blood [50]. Additionally, a library of the salivary proteome of healthy individuals can form the background necessary for future analyses of salivary samples from individuals with cancer. Such studies are expected to identify many more salivary proteins with diagnostic and/or prognostic value.

Using proteomic techniques to analyse saliva, Streckfus et al. successfully predicted whether women with primary breast cancer would have positive or negative ipsilateral lymph nodes at surgery [17]. Their non-invasive salivary proteomic test can potentially be used to distinguish those patients with small breast cancers (T1 tumours) who may need neoadjuvant chemotherapy from those patients who will do well without chemotherapy administration prior to surgery [17], [26].

General proteomics

The development of proteomics dates back several decades to the development of two-dimensional gel electrophoresis and the cataloguing of individual gel spots to create early protein databases [42]. The rapid growth of information in the genomic era quickly led to new, robust methods for the analysis of protein targets [51]. These include protein identification [52], recognition of post-translational modifications such as glycosylation and phosphorylation and sequence variations [53], as well as assessment of protein function and characterisation of protein-protein interactions [28]. Many of the methods are currently being used for quantification and screening of proteins that differ between health and disease [25].

LC-MS/MS mass spectroscopy with isotopic labelling

Progress in mass spectrometry (MS), liquid chromatography, analytical software, and bioinformatics have allowed the analysis of complex peptide mixtures. The main approach for querying any tissue or body fluid is a mass spectroscopy-based method that uses isotope coding of complex protein mixtures such as tissue extracts, blood, urine, or saliva to identify variably expressed proteins [15], [54]. The technology allows the detection of proteins that vary by more than eight orders of magnitude [55] by using isotope labelling coupled with liquid chromatography tandem mass spectrometry (IL-LC-MS/MS) to characterise the proteome [56]. The method identifies changes in the level of expression and permits the analysis of regulatory pathways to show significant changes in response to therapy in a breast cancer, or of changes in host immune markers before, during and after therapy. The analysis is performed on a tandem mass spectrometer using high-performance liquid chromatography (HPLC) for capillary chromatography with the HPLC coupled to the mass spectrometer. The advantage of tandem mass spectrometry combined with liquid chromatography (LC) is improved sensitivity combined with the peptide separations afforded by chromatography. Thus even in complex protein mixtures MS/MS data can be used to sequence and identify peptides by sequence analysis with a high degree of confidence [26], [54], [56–58].

Isotopic labelling of protein mixtures is a useful technique for the analysis of relative expression levels of proteins in complex protein mixtures such as plasma, saliva, urine, or cell extracts. There are many methods that are based on isotopically labelled protein modifying reagents to label or tag proteins to determine relative or absolute concentrations in complex mixtures. The higher resolution offered by the time of flight (TOF) tandem mass spectrometer is ideal to analyse isotopically labelled applications [26], [56], [59], [60].

The use of Isobaric tags for relative and absolute quantification (ITRAQ) reagents is a relatively new innovation [56], [59], [60]. The iTRAQ reagents are a set of isobaric reagents which are amine specific and allow for the identification and quantification of up to four different samples simultaneously. These reagents are amino reactive compounds that label peptides in a total protein digest of a complex fluid. The advantage of using iTRAQ reagents is that the tag remains intact through TOF-MS analysis. During collision-induced dissociation by MS/MS analysis the MS/MS spectrum for each peptide has a fingerprint allowing the quantification of that peptide from each of the different protein pools. Because effectively all of the peptides in a mixture are labelled by the reaction, numerous proteins in complex mixtures are identified and can be compared for their relative concentrations in each mixture. Thus even in complex mixtures there is a high degree of confidence in the identification of the individual peptide.

Quantitative protein analysis

Liquid chromatography/mass spectroscopy (LCMS) has been used to find disease pathways, for biomarker discovery, and for new insights into biological processes for drug discovery, yet its methodology yields results that are only relatively quantifiable. Systems biology depends on data sets in which the same group of proteins is consistently identified and precisely quantified across multiple samples, often over serial time periods [9]. For studies involving longitudinal studies over time, such as monitoring changes in patients with cancer before and after therapy, it becomes critical to perform an absolute quantification of proteins whether in blood, urine, saliva, or in tissue biopsies.

A quantifiable mass and time tag approach for proteomic analyses was described by Pasa-Tolic et al. [1]. The researchers developed an accurate mass time tag data base for an organism, tissue, or cell line. The investigators first performed standard proteomic analyses, and then validated peptide identifications using the mass measurement accuracy of Fourier transform ion cyclotron resonance mass spectrometry and liquid chromatography elution time constraint. Pasa-Tolic et al.'s method has the ability to detect >10⁶ differences in protein abundance and identify more abundant proteins from sub-picogram amounts of samples [1].

Silva et al. developed two different methods to derive absolute values from LC/MS [2], [3]. The first method quantifies proteins by using a simple LC/MS-based methodology to measure relative changes in abundance of proteins in highly complex mixtures to be determined [2]. Using chromatographic separations with the high mass resolution and mass accuracy of an orthogonal time-of-flight mass spectrometer, they were able to quantitatively compare tens of thousands of ions from identically prepared control and experimental samples [2].

Later, using a second method Silva et al. described an absolute quantification of proteins by LC/MS [3]. The second method is based on the relationship the investigators found between mass spectroscopy signal response and protein concentration. They established that an average MS signal response is constant with a coefficient of variation of less than ±10% for three of the most intense tryptic peptides per mole of protein in their material. Applying an internal standard, they used the relationship to calculate a universal signal response factor. They demonstrated that the universal signal response factor (counts/mol) was the same for all proteins tested in the studies. They then determined the absolute concentration of 11 common serum proteins and compared them with the literature. The absolute concentrations were close to the expected concentration values from Specialty Laboratories (Valencia, CA) [3].

Using an unfractionated Escherichia coli lysate, Silva et al. investigated a sub-set of identified proteins that exist as functional complexes. They used the absolute quantities calculated to accurately determine the stoichiometry of those functional complexes [3].

Malstrom et al. developed yet another new quantitative mass spectrometry strategy [61], [62]. Using their method Malstrom et al. were able to determine the absolute quantity of protein copies per cell for a major fraction of the proteome in a genetically stable bacterial cell population. They determined an absolute protein abundance scale for greater than 80% of the proteome of the human infectious pathogen Leptospira interogans. They calculated the average absolute value of selected proteins determined by mass spectrometry at a single cell level in the Leptospira cells. With the same cells they independently verified the quantified proteomic results using cryo-electron tomography measurements [61], [62].

They found that a major component of protein synthesis of the bacteria was devoted to protein synthesis and folding, electron transport, cell motility, and to maintain the external encapsulating structure.

Of considerable interest, they were able to study the effects of an antibiotic on the proteome of the bacteria using their absolute quantification method. They found that after treatment with an antibiotic, ciprofloxacin, the bacterial cell reacted by synthesising massive amounts of a small number of proteins normally not expressed in a state of unperturbed growth. However, while these new proteins comprised 20% of the proteome, the normally expressed total cellular protein concentration did not significantly change. Instead, the redistribution of proteins required only a small reduction of the highly abundant proteins that were important to cell maintenance. The decrease necessary to maintain protein concentration homeostasis occurred through a loss of the less abundant proteins. It appeared the proteins that decreased in concentration were not essential proteins to the organism. The data suggest that those proteins the organism required to maintain critical functions of the cell were highly conserved despite antibiotic perturbation [62].

Selected reaction monitoring (SRM) targeted proteomics

As stated earlier, systems biology depends on data sets in which the same group of proteins is consistently identified and precisely quantified across multiple samples, often over serial time periods [9]. This is not achieved by current standard mass spectroscopy-based proteomics. Selected reaction monitoring (SRM), also called multiple reaction monitoring, is an even newer technology that enhances the discovery capabilities of ‘shotgun strategies’ by a reliable quantification of analytes of low abundance in complex mixtures. Lange et al. used a triple quadrupole (QQQ) MS for quantitative analysis named SRM [9]. In SRM, the first and the third quadrupoles act as filters to specifically select predefined m/z values corresponding to the peptide ion and a specific fragment ion of the peptide, whereas the second quadrupole serves as a collision cell. Several transitions (precursor/fragment ion pairs) are monitored over time, yielding a set of chromatographic traces with the retention time and signal intensity for a specific transition as coordinates. The two levels of mass selection with narrow mass windows result in a high selectivity. Unlike standard MS-based proteomic techniques, no full mass spectra are recorded in QQQ-based SRM analysis. The non-scanning nature of this mode of operation translates into an increased sensitivity by one or two orders of magnitude compared with conventional ‘full scan’ techniques. Typically, a large number of peptides are quantified during a single LC-MS experiment [9].

Whiteaker et al. developed an alternative method to validate large numbers of protein biomarker analytes produced in proteomic assays [10]. They used an automated multiplexed stable isotope standards with capture by anti-peptide antibodies (SISCAPA) assay (nine targets in one assay). Using this automated, multiplexed platform, they were able to detect analytes in a physiologically relevant ng/mL range (from 10 µL of plasma) with sufficient precision (median coefficient of variation, 12.6%) to quantify biomarkers. They also demonstrated that enrichment of peptides from larger volumes of plasma (1 mL) could extend the limits of detection to the low pg/mL range of protein concentration. Whiteaker's method is generally applicable to any protein or biological specimen of interest and is expected to be capable of analysing large numbers of biomarker candidates [10].

Quantitative phosphoproteomics

Protein phosphorylation is a complex network of signalling and regulatory events that affects virtually every cellular process. The understanding of the nature of this network as a whole remains limited, largely because of an array of technical challenges in the isolation and high-throughput sequencing of phosphorylated species [5]. Villien et al. demonstrated that a combination of tandem phosphopeptide enrichment methods, high performance MS, and optimised database search/data filtering strategies could be used to survey the phosphoproteome. Villien et al. used an integrated analytical platform and identified 5,635 non-redundant phosphorylation sites from 2,328 proteins from mouse liver. From this list of phosphorylated sites they found both novel and known motifs for specific serine/threonine kinases including a ‘dipolar’ motif. They also found that C-terminal phosphorylation was more frequent than at any other location and that the distribution of potential kinases for these sites was unique. Finally, they identified double phosphorylation motifs that may be involved in ordered phosphorylation [5].

While proteomics has previously been used to identify the direct targets of kinase inhibitors, Pan et al. introduced an approach to evaluate the effects of kinase inhibitors on the entire cell-signalling network [4]. They developed triple labelling stable isotope labelling by amino acids in cell culture (SILAC) to compare cellular phosphorylation levels for control, epidermal growth factor stimulus, and growth factors combined with two kinase inhibitors. They found that either or both inhibitors affected 83% of the growth factor-induced phosphorylation events. Their data suggested that early signalling processes are predominantly transmitted through the mitogen-activated protein kinase (MAPK) cascades. In addition they found that dasatinib affected nearly 1,000 phosphopeptides. Dasatinib is a potent clinical drug directed against the breakpoint cluster region gene mutation (BCR-ABL), the causal mutation of chronic myelogenous leukaemia. In addition to the proximal effects on BCR-ABL and its immediate targets, dasatinib broadly affected the downstream MAPK pathways [4]. Pathway mapping of regulated sites implicated a variety of cellular functions, such as chromosome remodelling, RNA splicing, and cytoskeletal organisation, some of which have been described in the literature before. Pan et al.'s assay is streamlined and can become a useful tool in kinase drug development [4].

Clinical examples of various proteomic analyses

The functional machinery of protein signal transduction is embodied in the post-translational modifications and binding partner complexes that assemble and disassemble over time following a molecular stimulus [63]. Wu et al. described the profiling of proteins in a sample from a breast cancer cell line (SKBR-3) prepared by laser capture microdissection (LCM) [64]. An important component of this study was a combination of the LCM process with an extraction/digestion procedure that allowed effective solubilisation of a significant part of the cellular sample in a single step. The identity of the peptides was determined by tandem mass spectrometry measurements in which the resulting spectra were compared with genomic and proteomic databases. While Wu et al. only used peptides with a high probability assignment; the investigators also confirmed their interpretation of mass spectral fragmentation patterns by a manual interpretation of the spectra. Also, for the more abundant proteins they strengthened the initial protein assignment from the best match peptide by identifying additional confirmatory peptides. The authors correlated the mass spectrometric studies of the more abundant proteins with clinical and genomic studies of cancer markers in tumour samples. Their proteomic study allowed identification of HER-2/neu and the related kinases HER-3 and HER-4, gene products from breast cancer type I and II susceptibility genes and cytoskeletal components such as cytokeratins 8, 18 and 19. Their experimental approach can be used for proteomic studies on selected tissue samples and for studies of specific cell types.

In another report, Wu et al. described extended range proteomic analysis (ERPA), an intermediate approach between top-down and bottom-up proteomics, for the comprehensive characterisation at the trace level (fmol level) of large and complex proteins [63]. They used ERPA in cultured cells to determine quantitatively the temporal changes that occur upon factor stimulation of the epidermal growth factor receptor (EGFR). Specifically, A431 cells were stimulated with epidermal growth factor after which EGFR was immunoprecipitated at stimulation times of 0, 0.5, 2, and 10 min as well as 4 h. The investigators obtained high sequence coverage (96%), and developed methods for label-free quantification of phosphorylation and glycosylation. They identified a total of 13 phosphorylation sites on EGFR, and they also estimated stoichiometry changes in the receptor over the stimulation time points. Additionally, they identified a total of 10 extracellular domain N-glycan sites, and quantified major glycoforms at each site. As expected they did not observe a change in the extent of glycosylation with stimulation. They finally identified potential binding partners to EGFR based on changes in the amount of protein pulled down with EGFR as a function of time of stimulation. Many of the 19 proteins identified are known binding partners of EGFR [63]. Wu et al.'s work demonstrates that comprehensive characterisation provides a powerful tool to aid in the study of important therapeutic targets. Their technique is a tool to study important therapeutic targets [63].

Clinical breast cancer

In clinical breast cancer, metaplastic breast cancers (MBC) are aggressive, chemo resistant tumours associated with poor outcomes [65]. They are triple-negative breast cancers (oestrogen receptor-negative, progesterone receptor-negative, HER-2-negative) characterised by mesenchymal/sarcomatoid and/or squamous metaplasia (lineage plasticity) of malignant breast epithelium [65], [66]. MBCs are typically treated like basal-like or triple receptor-negative ductal cancers. However, whereas basal-like carcinomas show a high pathologic complete response rate to neoadjuvant chemotherapy, MBCs are most frequently chemo resistant [60].

Hennessey et al. compared 28 MBCs to common breast cancers using functional proteomic reverse-phase protein arrays combined with transcriptional profiling, comparative genomic hybridisation, and gene sequencing [67]. The MBCs showed unique DNA copy number aberrations compared with common breast cancers. PIK3CA mutations were detected in 9 of 19 MBCs (47.4%) compared to 80 of 232 hormone receptor-positive cancers (34.5%; P = 0.32), 17 of 75 HER-2-positive samples (22.7%; P = 0.04), 20 of 240 basal-like cancers (8.3%; P < 0.0001), and 0 of 14 claudin-low tumours (P = 0.004). Of seven phosphatidylinositol 3-kinase/AKT-pathway phosphorylation sites, six were more highly phosphorylated in MBCs than in other breast tumour sub-types. By transcriptional profiling, MBCs and the claudin-low breast cancer sub-set constitute related receptor-negative sub-groups characterised by low expression of GATA3-regulated genes and of genes responsible for cell–cell adhesion with enrichment for markers linked to stem cell function and epithelial to mesenchymal transition (EMT). In contrast to other breast cancers, most MBCs and claudin-low tumours and showed a significant similarity to a ‘tumourigenic’ signature defined using CD44 + /CD24 breast tumour initiating stem cell-like cells. MBCs and claudin-low tumours are thus enriched in EMT as well as stem cell-like features. Hennessey suggested that the MBC cancers may arise from an earlier, more chemo-resistant breast epithelial precursor than either basal-like or luminal cancers. The PIK3CA mutations, EMT, and stem cell-like characteristics likely contribute to the poor outcomes of MBC [67].

Bauer et al. examined a group of patients with high-risk, operable breast cancer who were treated with three cycles of paclitaxel followed by concurrent paclitaxel/radiation [68]. Tumour tissue from pretreatment biopsies was obtained from 19 of 38 patients enrolled in the study. Bauer et al. performed protein and gene expression profiling on serial sections of the biopsies from patients that achieved a pathologic complete response (pCR) and compared the protein and gene expression to those patients with residual disease, (non-pCR; NR). Proteomic and validation immunohistochemical analyses revealed that α-defensins (DEFA) was overexpressed in tumours from patients with a pathological complete response (pCR). Gene expression analysis revealed that MAP2, a microtubule-associated protein, was expressed at significantly higher levels in patients that achieved a pCR. They found that elevation of MAP2 in breast cancer cell lines was associated with increased paclitaxel sensitivity. Furthermore, expressions of genes associated with the basal-like, triple-negative phenotype were enriched in tumours from patients that achieved a pCR. Analysis of a larger panel of tumours from patients receiving pre-surgical taxanes-based treatment showed that DEFA and MAP2 expression, as well as histological features of inflammation, were all statistically associated with response to therapy at the time of surgery.

The authors definitively demonstrated the value of molecular profiling of pretreatment biopsies to discover markers of response and to suggest the potential use of immune signalling molecules such as DEFA as well as MAP2, a microtubule-associated protein, as tumour markers that predict response to neoadjuvant taxanes-based therapy [68].

Clinical ovarian cancer

In patients with advanced-stage, high-grade serous ovarian cancer Carey et al. measured protein expression associated with response to primary chemotherapy using reverse phase protein array technology [69]. They obtained tumour samples from 45 patients with advanced high-grade serous cancers from the Gynecology Tumor Bank at the British Columbia Cancer Agency. The treatment for the patients consisted of platinum-based chemotherapy following debulking surgery. They prepared protein lysates from fresh frozen tumour samples. They measured 80 validated proteins by reverse phase protein array from signalling pathways implicated in ovarian carcinogenesis. As the primary outcome measure of chemotherapy response, the investigators used normalisation of serum Ca-125 by the third cycle of chemotherapy. They used logistic regression for multivariate analysis to identify protein predictors of Ca-125 normalisation and Cox regression to test for the association between protein expression and progression-free survival, and a significance level of P ≤ 0.05. They found that the epidermal growth factor receptor, YKL-40, and several transforming growth factor TGF-β pathway proteins showed significant associations with Ca-125 normalisation on univariate testing. On multivariate analysis, epidermal growth factor receptor, JNK, and Smad3 were significantly associated with normalisation of Ca-125. The investigators concluded that TGF-β pathway signalling likely plays an important role as a marker or mediator of chemo-resistance in advanced serous ovarian cancer [69]. Using their methods, future studies to develop and validate useful predictors of treatment failure in ovarian and also breast cancer are warranted.

Clinical head and neck squamous cell carcinoma (HNSCC)

Examining normal mucosa and squamous cell head and neck cancers (HNSCCs), Patel et al. used laser capture microdissection combined with protein extraction from formalin-fixed paraffin-embedded (FFPE) tissues using a novel proteomics platform [7]. They analysed approximately 20,000 cells procured from FFPE tissue sections of normal oral epithelium and well, moderately and poorly differentiated HNSCC using mass spectrometry and bioinformatic analysis. They found that a large number of proteins expressed in normal oral epithelium and HNSCC, including cytokeratins, intermediate filaments, differentiation markers, and proteins involved in stem cell maintenance, signal transduction, migration, cell cycle regulation, growth and angiogenesis, matrix degradation, and proteins with tumour suppressive and oncogenic potential, were detected. Representative proteins were further validated using immunohistochemical studies in HNSCC tissue sections and tissue microarrays. Not surprisingly, the relative expression of many of these molecules followed a distinct pattern in normal squamous epithelia and a progressive difference in the evolution of well, moderately and poorly differentiated HNSCC tumour tissues. Patel et al. also identified several molecules with a role in the transduction of proliferative signals in normal and tumour.

HNSCC cells

However, the tumour tissue also demonstrated proteins not seen in normal oral mucosa. For example, they identified epidermal growth factor receptor in tumour samples but not in normal oral epithelial tissues, reflecting the over-expression of this tyrosine kinase receptor in HNSCC [7]. Several proteins involved in cell cycle progression, particularly G2-M transition and mitosis were only detected in tumour samples, reflecting their active state of proliferation. An unusual cell cycle regulating protein, prohibitin was only detected in tumour tissue. Prohibitin has been observed to play an unexpected function in the activation of Raf/MEK/ERK pathway by the oncogene RAS, and it also plays a part in modulating epithelial cell adhesion and migration [70]. Another surprising finding was the detection of two peptides derived from the EVI-5 oncogene. This protein was first identified in experimental T-cell lymphomas by retroviral insertion strategies and has been recently shown to ensure mitotic fidelity [71]. Thus, both prohibitin and EVI-5 may represent excellent candidates that play a role in aberrant cell growth in HNSCC. The authors concluded that combining laser capture microdissection and in-depth proteomic analysis of FFPE tissues yielded a significant amount of valuable data about the proteins expressed in normal squamous epithelium and the changes that occur during HNSCC progression [7]. This knowledge can lead to the development of novel biomarkers of diagnostic and prognostic value as well as the identification of novel targets for therapeutic intervention in HNSCC.

Rodriguez et al. also described the use of microdissection proteomics to perform tissue proteomics [8].

Conclusions

The field of proteomics is a multifaceted and rapidly evolving technology. It is apparent that the various methodologies of proteomics will significantly aid our ability to make cancer diagnoses, direct appropriate personalised therapy, and monitor response to therapy. Proteomic methodologies are certain to play an important role in how we will diagnose and establish treatment for breast and other cancer in the immediate future.

Declaration of interest: The authors report no conflicts of interest. The authors alone are responsible for the content and writing of this paper.