Direct Tissue Proteomics in Human Diseases

Proteomic Technologies in Biomarker Discovery

SELDI-TOF-MS technology is a combination of conventional choromatography and MS (Ciphergen Biosystems, CA, USA). The SELDI-TOF MS system, a modified MALDI-TOF-MS system, is an affinity-based MS method using protein chip modified with a specific chromatographic affinity surface. The sample is applied to chips with a variety of chromatographic affinity surfaces and captured proteins are ionized using matrix-assisted ionization. SELDI-TOF-MS generates spectra of ion mass-to-charge ratios from a sample such as blood serum. These spectra do not identify proteins in the sample, but instead demonstrate the relative abundance proteins at each mass-to-charge ratio value. Several studies have shown the feasibility of SELDI-TOF-MS to directly analyze patient serum for protein expression patterns that have the capability to distinguish cancer patients.^[16,17] Mian et al. reported the potential prediction of biological resistance of breast cancer cell lines to specific antitumor reagents with SELDI-TOF-MS protein profiling.^[18] In lung cancer proteomics, several reports have been published with SELDI-TOF-MS proteomic profiling analysis with patient serum and tumor specimens.^[19,20] With MALDI-TOF-MS profiling, samples are mixed with a laser-reactive matrix solution and deposited on a metal MALDI plate; the matrix/sample solution is applied to the surface and the proteins bound to the substrate are ionized and analyzed by TOF-MS. Both instruments are compatible with automated high-throughput robotic sample preparation, improving speed and reducing labor.

In both MALDI-TOF-MS and SELDI-TOF-MS, high resolution and mass accuracy are needed to reliably identify proteins, and thus, analyses are greatly dependent on the quality of the sample. Uncertain procurement protocols and handling of fresh and frozen specimens casts doubt on the quality of samples collected prior to understanding the specific needs for novel proteomic technology analyses. In SELDI-TOF protein profiling analysis, demonstration of interinstitutional reproducibility is crucial for further clinical use and identification of signature peptides or their parent proteins are indirect. Identification would require a second dimension of MS and increase the requirements of sample size. Currently, their robustness and reliability in the clinical setting is being challenged, owing to the lack of definitive protein identification as well as multiple erroneous predictions associated with the aforementioned techniques.

Current proteome separation platforms, including 2D-PAGE and shotgun-based multidimensional liquid chromatography (LC) separations, require large cellular samples that are generally incompatible with the protein quantities obtained from laser-capture microdissection (LCM) samples.^[21,22] While limited 2D-PAGE analyses of LCM-derived tissue samples have been attempted, these studies involve significant time effort (from 3 h to 4 days, depending on tissue type) to extract sufficient amounts of protein for obtaining good visual quality of gel patterns, while providing little proteome information beyond a relatively small number of high-abundance protein identifications. In addition, the 2D-PAGE-MS approach suffers from low throughput and lack, of proteome coverage, dynamic range and sensitivity.^[23–25] Peak capacity improvements in multidimensional LC separations have increased the number of detectable peptides and, therefore, proteins. However, the quantity needed for samples from an LCM-captured cell sample has restricted analysis to the use of only a single chromatography separation prior to tandem MS (MS/MS) analysis in recent studies, limiting the ability to mine more deeply into the tissue proteome. Since the sizes of human tissue biopsies are becoming significantly smaller due to earlier detection and diagnosis, a more effective discovery-based proteome technology is critically needed to enable sensitive studies of protein profiles within tissue specimens procured by LCM and other microdissection techniques.

We have recently introduced a modified shotgun proteomic strategy, termed direct tissue proteomics (DTP), which is able to identify proteins directly from tissues by micro-reverse-phase (µRP) LC-MS/MS. As a proof-of-principle demonstration, commercially available prostate tissue arrays were used for DTP method development. All of the prostate tissue arrays were core biopsy samples with identical dimension (2-mm rings with 2-µm thickness). Manual counting of defined microscopic fields revealed an estimation of approximately 40,000–60,000 cells/sample. In order to assess the usefulness of this proteomic study, all tissue samples were carefully examined and categorized for prostate cancer according to the Gleason score, such as low, medium and high grade. To initiate the method, samples were then heated at 94°C in a buffer compatible with MS (30% acetonitrile and 100 mM ammonium bicarbonate) for formaldehyde de-crosslinking and then digested with trypsin. This MS-compatible buffer is critical especially for minute quantities of clinically relevant analysis, such as thin sections of needle biopsy tissue. Once the peptides are extracted from tissue samples, they are directly analyzed on a Finnigan LTQ-linear ion-trap mass spectrometer coupled to the nano-electrospray source. Using this strategy, we have been able to identify over 400 proteins from prostate cancer biopsies, including known prostate-specific antigen.^[26]

The most important issue in shotgun proteome analysis of formalin-fixed tissue is to efficiently extract proteins or peptides from the fixed tissues. In order to improve extraction efficiency for the DTP method, protein yield was compared with five different buffer conditions before and after the formaldehyde fixation and it was found that our initial method using acetonitrile buffer (compatible with trypsin digestion and direct LC-MS/MS analysis) can extract from 13% (after crosslinking) to 42% (before crosslinking) of the total extractable proteins. This extraction procedure was replicated five times and also compared with radioimmunoprecipitation assay buffer plus 2% SDS extraction. It was concluded that a radioimmunoprecipitation assay containing 2% SDS at high temperature was a better protocol to extract proteins.

Several proteomics studies using archival FFPE tissues have been reported in recent years.^[27–31] The majority of these studies employ protein extraction methods to the use of both heating and detergent (e.g., SDS) as supported by our studies. In fact, Fowler et al. have compared six different protocols for protein extraction from FFPE tissues, and concluded that the most effective protein extraction buffer tested was a 20-mM Tris solution containing 2% SDS and 0.2 M glycine at high temperature.^[32] The addition of a reducing agent did not improve protein recovery; however, recovery varied significantly with pH and optimal protein extraction was obtained with pH 4. Recently, Fowler et al. also demonstrated that elevated hydrostatic pressure (at 45,000 psi and 80–100°C in Tris buffer containing 2% SDS and 0.2 M glycine at pH 4.0) is a promising approach for improving the recovery of proteins from FFPE tissues for proteomic analysis.^[33] We have assessed extraction of the proteins from FFPE tissues using 20-mM Tris buffer containing 2% SDS and 0.2 M glycine at pH 4 and found this method to be most efficient when we assess the sharpness of protein bands in SDS-PAGE.

However, as far as we are aware, the extent of de-crosslinking from the formalin-fixed tissues has not been accurately measured. In order to accomplish this, one must first identify the crosslinked site on proteins, synthesize two isotope-labeled tryptic peptides encompassing the site of interest and perform absolute quantification (AQUA) of native and crosslinked peptide before and after the de-crosslinking procedure.^[34] Once this method is established, the efficiency of de-crosslinking can be assessed for different fixation conditions. Although only a few papers published the shotgun method used to analyze the proteome of formalin-fixed tissue, DTP and other studies indicated the possibility of using the archival collected formalin-fixed tissues for discovery-driven biomarker research.

Tissue proteomics seeks to identify differential protein expression as a function of a disease state or exposure to stimuli. Additional applications include determining how proteins interact within a system, which allows mapping of components to assess complex functional networks. Tissue proteomics can be particularly challenging because samples are highly heterogeneous with respect to cellular composition, secretions and stage of disease.^[22] LCM is a technique that enables procurement of pure cell populations from heterogeneous tissue sections under direct microscopic visualization. LCM can be used to isolate tumor and stroma from a single core of tissue, providing the opportunity for independent analysis of the tumor and its local microenvironment.^[35] The combination of DTP with LCM will facilitate the systematic analysis changes in protein expression that occur with disease development and progression and the effects of treatment. Application of DTP in combination with LCM to map the proteome of paraformaldehyde-fixed archival coronary arteries shows great promise.^[36] Analysis of 35 human coronary atherosclerotic samples using the DTP method allowed identification of a total of 806 proteins, which provided the first large-scale proteomic map of human coronary atherosclerotic plaques. Four specific proteins that are expressed in atherosclerosis lesion, PEDF, periostin, MFG-F8 and annexin-1, which were all validated using immunohistochemistry staining techniques to confirm their presence in human plaques.^[36] This study also applied AQUA technology to quantify low abundance cytokine/growth factor in human coronary arteries. This information will provide a basic understanding of the disease process as well as clinical applications, such as diagnosis and early detection of pathological conditions.

A critical step in protein biomarker discovery is the ability to contrast proteomes, a process generally referred to as quantitative proteomics. Stable-isotope labeling techniques (e.g., ICAT, ¹⁸O- or ¹⁵N-labeling, or AQUA) have been integrated with LC-MS/MS for relative protein quantification.^[7,8] To do this, protein samples are first labeled separately with stable isotopes, either by metabolic incorporation or chemical reaction. The differentially labeled samples are then mixed and subjected to MS analysis. This approach depends on the efficiency and completion of labeling reactions and the ability of MS to differentiate the mass-to-charge ratios of paired peptides different by a few mass units.

Despite the progress made in the past few years, practical difficulties associated with gel-based and isotope-labeling methods exist; therefore, other alternative approaches have been attempted. A semiquantitative method based on the number of peptides identified for each protein was recently described.^[37] Wang et al. and others reported a quantification method by direct comparison of peptide peak areas between LC-MS runs without any isotopic labeling.^[38–43] By varying the amount of a single protein or a few standard proteins, it was shown that the intensities of peptide peak signals correspond nearly linearly to their concentrations in the sample, and that the ratios of peak areas between different LC-MS runs reliably reflect their relative quantities in the sample. However, the linearity of applying such a method to compare whole proteomes remains to be demonstrated. Recently, using MudPIT technology, a simple and fast method for rough estimation of relative protein abundance has been described by Liu et al..^[44] They found that the number of tandem mass spectra (spectral count) collected from peptide mixtures displayed perfect linearity with respect to concentration. By contrast, percentage sequence coverage and number of peptides per protein did not demonstrate as good a linear correlation as a spectral count. However, it is unclear how these label-free (spectral count) methods will perform on complex human gel-based LC (GeLC)-MS/MS datasets. Laudgren et al. demonstrated that spectral count can predict differential protein expression with confidence using GeLC-MS/MS and a minimum of two replicates per sample (Lundgren DH, Wu L, Hwang S-I et al. Evaluating sensitivity and specificity of spectral count method to detect regulated proteins in complex mixtures. Manuscript in preparation). These label-free indices can add a new dimension to tissue proteomic studies for differential protein expression with a hope of candidate biomarker discovery. Besides the relative quantification of proteins, there is a strong need for the analysis of low-abundance proteins and the determination of absolute quantities of proteins in ultra-complex mixtures. The highly dynamic proteome is a great challenge. As a promising approach, stable-isotope dilution techniques in combination with shotgun proteomics are emerging. Proteins of interest are tryptically digested in the presence of synthetic peptide standards of known concentration with an incorporated stable isotope (¹³C or ¹⁵N). These standards are identical to the analyte peptides of interest, but are distinguished by mass difference. Stable isotope-labeled and unlabeled peptides co-migrate during chromatography and absolute levels of proteins are achieved by comparison of the peak abundances of the internal standard peptide with corresponding native counterpart due to, for example, multiple reaction monitoring via MS/MS.^[45–48]

Comments

Commenting is limited to medical professionals. To comment please Log-in.

Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.