|Home | About | Journals | Submit | Contact Us | Français|
The advantage of using proteins and peptides as biomarkers is that they can be found readily in blood, urine, and other biological fluids. Such sample types are easily obtained and represent a potentially rich palette of biologically informative molecules. Matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF-MS) represents a key tool for rapidly interrogating such sample types. The goal of clinical proteomics is to harness the power of this tool for identifying novel, condition-specific protein fingerprints that may, in turn, lead to the elucidation and use of diseasespecific biomarkers that may be used to diagnose disease as well as to evaluate disease severity, disease progression, and intervention efficacy. Here we have evaluated a simple, affordable bench-top MALDI-TOF mass spectrometer to generate protein profiles from human plasma samples of asthma patients and healthy individuals. We achieve this profiling by using C8-functionalized magnetic beads that enrich a specific subset of plasma proteins based on their absorption by this resin. This step is followed by elution, transfer onto a prestructured sample support (AnchorChip technology), and analysis in a bench-top MALDI-TOF mass spectrometer (OmniFLEX) with AutoXecute acquisition control which enables automated operation with reproducible results. Resulting spectra are compiled and analyzed through the pattern recognition component of ClinProTools software. This approach in combination with ClinProTools software permits the investigator to rapidly scan for potential biomarker peptides/proteins in human plasma. The reproducibility of plasma profiles within and between days has been evaluated. The results show that the novel and facile approach with manual magnetic-bead sample preparation and a low-cost bench-top MALDI-TOF mass spectrometer is suitable for preliminary biomarker discovery studies.
Asthma is an inflammatory disease of the airways that is characterized by intermittent and partially reversible bronchoconstriction, as well as by airway hyperresponsiveness to a wide variety of stimuli.1 The symptoms associated with this disease state and the response to specific pharmacological interventions vary widely among affected individuals, indicating potential differences in disease mechanism. This has led to the suggestion that asthma represents many different types of respiratory diseases. Understanding these differences may allow clinicians to distinguish asthma types and may lead to the development of more effective therapeutic approaches.
One approach to delineating these differences in clinical settings is to identify biomarkers that are unique to specific asthma subtypes—an approach amenable to mass spectrometry (MS)-based proteomic investigations. The advantage of using proteins and peptides as biomarkers is that they can be found readily in blood, urine, and other biological fluids. Such sample types represent a dynamically changing biological milieu. Many of the changes that occur within such sample matrices reflect the underlying physiological/pathological state of the individual. It is the goal of proteomic investigations to identify those proteins or peptides that are specific reporters or biomarkers of the underlying state. Biomarkers that are found to be specific and sensitive reporters of disease offer enormous power: They can potentially be used to (1) indicate the presence of disease; (2) diagnose a specific disease that may otherwise be clinically indistinguishable from related disease processes, for example, asthma variants; (3) evaluate disease severity; (4) monitor disease progression; and (5) evaluate the efficacy of experimental or therapeutic interventions. Matrix-assisted laser desorption/ionization time-of-flight MS (MALDI-TOF-MS) has become a powerful tool for surveying such sample types.2 The goal of clinical proteomics is to harness the power of this tool for identifying novel, condition-specific protein fingerprints that may, in turn, lead to the elucidation and use of disease-specific biomarkers.3 Here we have evaluated a simple affordable bench-top MALDI-TOF mass spectrometer in combination with magnetic bead-based sample cleanup and ClinProTools (CPT) software to generate and analyze protein profiles from human plasma samples of asthma patients and healthy individuals.
In this preliminary investigation, we evaluated a software–hardware combination designed to allow the automated processing and interrogation of multiple samples from a group of control and asthmatic individuals. The overall goals of this study were to assess in a small sample population the utility of protein/peptide profiling to identify potential biomarkers of asthma, to evaluate the intraday and interday reproducibility of selected plasma samples, and to determine the usefulness of the algorithms for identifying differentially represented proteins/peptides in typically complex biological fingerprint patterns.
Human plasma samples were collected from patients with a clinical diagnosis of asthma during an acute exacerbation of symptoms and from healthy individuals at Children’s Hospital and Research Center at Oakland. The study protocol was approved by the Institutional Review Board at Children’s Hospital and Research Center at Oakland, and informed consent was obtained for all patients. Heparin-anticoagulated blood was collected and centrifuged at 3600 rpm at room temperature for 10 min. The separated plasma samples were removed and frozen at −80°C for future analysis.
Plasma samples (5 μL) were processed using a magnetic bead-based C8-hydrophobic interaction chromatography resin (MB-C8, Bruker Daltonics, Leipzig, Germany) according to the manufacturer’s protocols. The MB-C8 was supplied as part of a kit with standard protocol and binding and washing buffer provided. In brief, 5 μL plasma sample was mixed with 10 μL binding solution in a standard, thin-wall PCR-tube. Then, 5 μL MB-C8 was added and the solution was mixed carefully by pipetting up and down several times. To separate the unbound solution, the tube was placed in a magnetic bead separator and the supernatant was removed carefully using a pipette. The magnetic beads were then washed three times with 100 μL wash buffer. Following binding and washing, the bound proteins/peptides were eluted from the magnetic beads using 5 μL of 50% acetonitrile. A portion of the eluted sample was diluted 1:10 in matrix solution comprised of α-cyano-4-hydroxycinnamic acid (0.6 g/L in 2:1 ethanol:acetone). Then 0.5 μL of the resulting mixture was spotted onto a SCOUT 600 μm prestructured sample support (AnchorChip target, Bruker Daltonics) and allowed to air dry for approximately 5 min at room temperature.
The processed samples were analyzed using a bench-top MALDI-TOF mass spectrometer (OmniFLEX, Bruker Daltonics) equipped with a pulsed ion extraction ion source. Ionization was achieved by irradiation with a nitrogen laser (λ = 337 nm) operating at 5 Hz. Ions were accelerated at +19 kV with 200 nsec of pulsed ion extraction delay. Each spectrum was detected in linear positive mode and was externally calibrated using a mixture of peptide/protein standards between 1000 and 12,000 Da. In order to increase detection sensitivity, excess matrix was removed with 10 shots at a laser power of 83% prior to acquisition of spectra with 100 shots at a fixed laser power of 67%. AutoXecute acquisition control, a software tool, was applied for automated data acquisition.
The spectra from all samples (training data) were imported into CPT software for spectrum postprocessing and generation of proteomic fingerprint. ClinProTools is designed to facilitate the processing and comparison of multiple spectra by automatically normalizing, baseline subtracting, peak defining, and recalibrating. The spectra were first processed using the following workflow:
The result of this data preparation was a collection of normalized peak areas for each spectrum.
Then these processed spectra were used, in a supervised training, to build a model describing a proteomic fingerprint that can discriminate asthma from normal. Model building was performed by selecting a small subset of relevant peaks and establishing clusters using the areas of these peaks. Most of the parameters of this Bruker algorithm were automatically determined to optimize the performance of the peak selection and to make it easy to use. The selected peaks were used to classify the spectra. With CPT software this can be done by either using a centroid clustering or a k nearest neighborhood criterion. The result of the peak selection algorithm is the peak combination (proteomic fingerprint) that proved to separate best between the different classes. After the proteomic fingerprint model was generated, a validation using training data was performed to estimate the error rates of this proteomic fingerprint model. The goal of model building in supervised training was to describe the training data in such a way that new data (test data) could be classified afterwards. The percentage of correctly classified training and test data is known as recognition and prediction capability, respectively. The visualization features of the CPT identify, highlight, and locate those ions of the spectra that have peak areas significantly different from those of the reference group. These identified ions represent candidate protein/peptide biomarkers of disease.
The experimental flow chart of the integrated approach is shown in Figure 11.. The human plasma profiling was achieved by using C8-functionalized magnetic beads (MB-C8) that enrich a specific subset of plasma proteins based on their absorption by this hydrophobic resin. Sample spots used in MALDI-TOF- MS analysis often suffer from heterogeneity. This leads to high variability in signal intensities and coefficients of variation that make quantitative analysis by this ionization technique difficult. We overcame many of these limitations by employing prestructured sample supports (AnchorChip technology), a sample focusing surface chemistry that greatly improves the homogeneity and increases the effective concentration of the spotted analyte.4–6 This allows more consistent data signals to be acquired at fixed laser power, thus enabling the user to acquire MALDI-TOF-MS data that are quantitatively more consistent than is typical for this type of ionization technique.
The intraday reproducibility of plasma spectral fingerprints has been evaluated. By improving the homogeneity of the sample spot, a fixed laser power can be used, thereby improving the consistency of ionization and protein profiling. Figure 22 shows the reproducibility for MALDI-TOF-MS measurement with AnchorChip technology. The assessment of sample preparation consistency with MB-C8 is shown in Figure 33.. To increase detection sensitivity, the spotted sample was initially exposed to a laser power of 83%, a technique used to facilitate desorption of excess matrix that appears to otherwise significantly inhibit the ionization of the analyte. Acquisition of spectra was then performed at a fixed laser power of 67%, an intensity optimized for desorption and mass resolution of analyte. The data were automatically collected in positive linear mode.
The acquired spectra from replicate analyses (n=3) exhibit reproducibility sufficient for semiquantitative analysis. The signal intensity variability was evaluated by randomly selecting 10 ions (annotated with asterisks in Figs. 22–4) representing high-, medium-, and low-abundance signals in the m/z 1000–10,000 mass range. The variability of each peak area (Fig. 22)) has been measured by the coefficient of variance (CV) that was calculated to be between 11% and 25% with an average intraday variability of 18% (Table 11).). The peak area ratios are used to evaluate the variability of the same sample but processed with MB-C8 in two different vials in a single sample preparation (Fig. 33).). The ratios for the signal intensities of the 10 randomly selected peaks are between 0.95 and 1.38, with an average of 1.14 (Table 22).
The interday reproducibility study including sample processing and mass spectral analysis was evaluated on three different days (Fig. 44);); variability was calculated to be between 4% and 43% with an average CV of 26% (Table 33).
After profiling, the resulting spectra were compiled and analyzed through the pattern recognition component of CPT software to generate and visualize the plasma profiles. Plasma samples from 4 asthma patients and 4 healthy individuals were manually fractionated using the MB-C8 kit. Each sample was spotted in triplicate on the AnchorChip targets to generate spectra of 12 normal and 12 asthma subjects. In order to do proper pattern recognition analysis, the spectra were duplicated; thus, a total of 48 spectra were processed with CPT software to interrogate this dataset for the potential discovery of disease-specific biomarkers. The plasma profiles generated by CPT software allow the user to visualize differences in the mass fingerprints of multiple spectra using three different views. This approach can be used to visually inspect and distinguish normal and asthma protein/peptide profiles as shown in Figure 55.
ClinProTools is a user-friendly data processing software for the interactive inspection and comparison of large data sets originating from samples with different clinical diagnoses to find potential biomarkers in complex protein/peptide profiles. It supplies highly sophisticated mathematical algorithms for the discovery of complex biomarker pattern models and supports research in initial diagnosis, screening, monitoring of disease progression, and prediction of response to therapy. As illustrated in Figure 66,, CPT software calculates the peak area for each ion and compares them between subject groups (asthma vs. control). Significantly different average ion peak area values are determined. In the future study with statistically sufficient number of samples in each data set, the confirmation of the calculated differences might allow the investigator to classify the normal and asthma profiles with 100% recognition capability with the training data (marked with red circles in Fig. 66).
This study highlights our initial effort to use a software–hardware combination designed to automate the acquisition and integration of spectra, catalog the ion intensities from each spectra, calculate the averages of ion intensities from user-specified groups (treatment vs. control), and allow the software algorithm of this described software to automatically identify and locate possible disease-specific biomarkers. The work described would optimally require a greatly expanded sample set of perhaps many hundreds of samples—closer to the functional capacity of the described software–hardware combination—to allow any conclusions to be drawn about the usefulness of the identified ions as asthma-specific biomarkers. It will be important to incorporate in future comparisons plasma samples from other types of diseases that proceed through similar, but distinct, inflammatory mechanisms to determine whether the identified markers are indeed specific or instead representative of a general inflammatory state. Availability of samples from asthma variants and other related pulmonary diseases would be a possible point of continuation for this clinical and basic research effort.
The combination of MB-C8, AnchorChip technology, and MALDI-TOF-MS generates protein/peptide profiles with good quantitative reproducibility. This approach in combination with CPT software allows rapid scanning for potential peptide/protein biomarkers in human plasma. Subtle but significant differences between normal and asthma plasma profiles have been revealed by implementing the software algorithms in CPT software. And by using those potential differences, the normal and asthma plasma profiles are correctly classified with 100% recognition capability. Further study with large data sets will be needed to confirm whether these signals represent proteins specific to asthma, or whether they are general to a process (i.e., acute phase proteins increase in response to inflammatory stress). ClinProTools software enables efficient visualization and comparison of clinical proteomic data from different patient groups. The novel and facile approach with manual magnetic-bead sample preparation and a low-cost bench-top MALDI-TOF mass spectrometer is suitable for preliminary biomarker discovery studies. When using a fully automated sample preparation robot, together with higher performance automated TOF/TOF instruments, the CV values should be much lower and throughput much higher.
Supported by P60MD00222 (MKS). Supported in part by National Institute of Health grants HL-04386-03 and MO1-RR01271, Pediatric Clinical Research Center (to CRM).