Skip to : [Content] [Navigation]
 

IVD Technology

 
 
Email article link Print article

Originally Published IVD Technology November/December 2005

ASSAY DEVELOPMENT

Array analyses of human whole blood changes in protein fractions

Protease inhibitors may provide for more-sensitive detection of biomarkers in blood.

James L. Schram and Robert E. Pearson

Figure 1. Serum (S) and plasma (P) proteins labeled with cyanine 3 (Cy3) and cyanine 5 (Cy5). After clotting at room temperature for 1 hour (T = 0), sample tubes were centrifuged and stored for 24 hours (T = 24).

Proteomic analysis of blood promises to revolutionize diagnostics through the discovery of novel markers and new approaches to understanding disease.1 Investigations using two-dimensional gel electrophoresis and mass spectrometry have had an enormous impact on the identification of proteins and diagnostic paradigms.2–4 Once these proteins are identified, antibody arrays can then be used to detect them at low concentrations, simultaneously monitor a large variety of proteins, and evaluate expression changes. Although clinical technologies will continue to evolve, the impact of proteomic technologies on health may well depend on the in vitro blood protein changes that occur after the sample is collected.

Protein changes that occur in serum, plasma, and other blood fractions can have a direct effect on the detection of biomarkers and clinical targets. For example, in serum, a number of hormones degrade rapidly, and an inaccurate analysis could affect therapy.5 Often, the stability of hormones in blood serum is less than in ethylene diamine tetraacetic acid (EDTA)–anticoagulated plasma, suggesting that the coagulation proteases affect blood samples.6 In such blood samples, EDTA acts as a protease inhibitor. Protein degradation in serum is an important process that can create sample artifacts, and serum has been used in several biomarker studies.7

Aside from coagulation, proteases are often found in blood as a consequence of physiological disorders like tumor invasion, pancreatic dysfunction, and inflammation.8–10 In some cases, disease states may cause sample degradation prior to diagnostic testing.

Recently, several investigations have demonstrated that protease inhibitors slow protein degradation, but their usefulness in the global protein stabilization of a wide variety of targets remains to be demonstrated.11 To further understand serum and plasma changes, researchers at BD Technologies (Research Triangle Park, NC) investigated the blood protein profiles of matched serum and plasma samples and used array technology to follow the changes induced by an overnight incubation.

Antibody array technology is a multiplex platform for analyzing large numbers of blood proteins, developing disease-specific protein patterns, and detecting sensitive biomarkers. Arrays can detect changes in protein expression and are often as sensitive as enzyme immunoassays.12 Recently, antibody arrays have been used to identify novel protein expression associated with neurological disease.13

Since blood proteins are complex and may contain proteins at very low levels, antibody arrays were used in the study to look for the presence of intercellular proteins in blood protein. Many of these types of proteins could become critically important for the diagnosis of future diseases. In this paper, it is shown that more protein changes occur in serum than in plasma after incubation. Many of these changes correspond to decreases in proteins levels over the same period. Further, it is shown that more proteins were detectable in plasma than in serum. Finally, it is demonstrated that serum proteases degrade the clinically important target B-type natriuretic peptide (BNP). In the future, clinical identification of disease markers may depend on EDTA and other protease inhibitors for sensitive detection of biomarkers in blood samples.

Figure 2. Dye-to-protein ratios for the tested serum and plasma samples.
Figure 3. A comparison of serum and plasma array signals at the initial time and after 24 hours. The percentage of detectable antibody signals is provided in parentheses. Antibody signals at the background are identified as “No Signal.”

Background and Procedure

In vitro whole-blood protein changes were studied by comparing samples before and after an overnight incubation at 37°C. While many sample iterations could be and have been compared, we arbitrarily selected these conditions as they might be expected to generate the greatest amount of blood alteration from cell changes and proteases.

Figure 4. Proteins ranked by total array signal. Labeled proteins for each sample type are presented for T = 0 and T = 24. The averaged total signals consist of the replica means for the Cy3 and Cy5 signals for each protein target.

To evaluate the most abundant proteins, proteins labeled for array detection were investigated. One-dimensional denaturing gels were employed to evaluate cyanine 3 (Cy3) and cyanine (Cy5) labeling profiles and to detect incubation changes in proteins such as albumin and immunoglobulins (see Figure 1).

Proteins labeled with Cy3 and Cy5 were separated by 12.5% sodium decyl sulphate polyacrylamide gel electrophoresis. Blood from three healthy males was collected into serum (red top) and EDTA plasma tubes. The serum tubes were allowed to clot for 1 hour at room temperature and became the zero-time point for serum (T = 0). The plasma tubes were kept at 4°C during the serum clotting. All tubes were centrifuged at 1500 rpm for 5 minutes at 4°C before the protein fractions were removed, pooled, and stored at –70°C. Another set of tubes was placed at 37°C and processed in a similar fashion and is named T = 24. The T = 24 samples were incubated for 24 hours at 37°C. The proteins were labeled with both Cy3 and Cy5 from GE Healthcare (Piscataway, NJ), as described by Clontech (Palo Alto, CA). Approximately 0.5 mg of protein was separated in each well of a Criterion precast gel system by Bio-Rad Laboratories Inc. (Hercules, CA) (see Figure 2).

The resulting gels demonstrate that most of the major blood components are labeled in a similar fashion within a dye group and exhibit few changes during the incubation period. However, there are subtle differences between a few of the abundant proteins. In the case of serum, for example, the modifications at 55,000 molecular weight most likely correspond to fibrinogen degradation. Denaturing gels of unlabeled proteins demonstrated few detectable differences in the most abundant blood proteins after incubation.

Because arrays can monitor multiple targets at low concentrations, biomarkers in serum and plasma were evaluated by antibody array comparisons. The arrays contain a wide collection of antibody specificities, including cytokines, apoptosis proteins, transcription factors, and cell-signaling proteins.

Initially, the number of protein signals twice as strong as the average local background signal and the number of proteins with no signal above the background were counted for each sample type. Total averaged signals were calculated for all of the proteins by averaging the replicate signals for the dye-labeled targets at either T = 0 or T = 24, and adding the Cy3 and Cy5 signals for a sample type. Any Cy3 and Cy5 labeling differences in the protein samples were averaged over the complement arrays with flipped dyes, in which Cy3 from

Figure 5. Internal normalized ratios (INRs) for serum and plasma antibody signals. The bulk total averaged signals are included for both time points. Only protein ratios of the T = 0 spots more than twice as great as the local background are presented.

T = 0 with Cy5 from T = 24 were in one array assay and Cy3 from T = 24 and Cy5 from T = 0 were in the other.

Because all of the arrays were from the same manufacturing lot, cross-reactions between blood proteins and the array antibodies should be consistent.

Overall, none of the total signals varied by more than 35%, and the highest coefficients of variance were less than 1.3. The data in Figure 3 demonstrate that the number of serum protein signals above the arbitrary cutoff value decreases with whole-blood incubation at 37°C.

Dual antibody array sets with 507 antibodies were purchased from the same manufactured lot. The arrays were probed with 100 mg of total protein containing 50 mg of each dye-labeled protein sample. One array contained the Cy3-labeled protein from one processing time and the Cy5-labeled protein from the other processing time. The complement array had the dye-labeled proteins in the opposite combination. The dried arrays were scanned with a Typhoon 9410 imager by GE Healthcare and the spot signals evaluated with ImaGene image-analysis software by BioDiscovery Inc. (El Segundo, CA). The local backgrounds were determined as the average local background for the Cy3 and Cy5 signals for each sample array. The number of spots with signals twice as great as the background was determined within each blood-sample type and incubation time. All of the arrays had a local signal background of less than 135 with an average signal/ average background of greater than 2:1 for both the Cy3 and Cy5 channels.

After the 24-hour incubation, 22% fewer proteins were detected in the healthy sample. Also, the number of antibody spots with no detectable signal was higher in the serum fraction. In contrast to serum, the plasma proteins remained consistent during the incubation and more spots were detected.

The composition of and changes to blood samples like serum and plasma can be evaluated by comparing bulk signals from the two time points. Figure 4 shows the top 20 proteins ranked by total signals for biomarkers from serum and plasma. The data demonstrate that many of the detectable serum signals decrease after incubation at 37°C. At both sample time points, 55% of the serum proteins were present, while at T = 0, only 35% were found. Unlike serum, many of the plasma proteins displayed higher signals during the incubation period, suggesting that they were accumulating. The overall consistency of the plasma sample suggests relative protein stability during incubation.

The relative ranks of the blood proteins suggest that cell processes are activated. In serum, for example, the higher levels of caspase-4 and caspase-8 present after incubation could be related to apoptosis activation in lymphocytes. The elevated p53 signals in the incubated samples suggest a cellular response to DNA damage, which may also be related to apoptosis. Finally, a number of proteins are unique to either serum or plasma within an incubation sample. More experiments with cell fractions will be necessary to better understand the protein changes related to pathway activation, apoptosis, and nucleic acid damage.

Figure 6. The change in cyclin D3 Western blot band-intensity levels after overnight incubation at 37ÞC.

The raw mean signals can be converted into internal normalized ratios (INRs) to evaluate protein expression levels. The ratio was developed by Clontech as a way to compare changes in protein expression levels between two experimental conditions. The INR value for each protein is the average square root of array ratios of the Cy5 mean signal divided by the Cy3 mean signal for each array. Thirty-five serum signals had INR values above 2 and were twice as strong as the local background. No plasma protein had an INR value greater than 2. Figure 5 is a summary of the INRs for serum and plasma along with their total array signals. Cyclin D3, a cell-signaling protein that displayed one of the top 20 signals, has an INR of 2.12, and the combined averaged array signal changed from 4027 to 798. The INR values of serum proteins demonstrate that array detection of some serum biomarkers may be problematic.

A Western blot was performed to more directly evaluate the level of cyclin D3 in blood samples (see Figure 6). Proteins were separated on a 10–14% gradient gel, then transferred to a nitrocellulose membrane and probed with clone DCS-22 from Biosource International (Camarillo, CA), a monoclonal antibody specific to cyclin D3.14 Bound antibody was detected with the Immun-Star chemiluminescent kit from Bio-Rad Laboratories. Bands were quantified using ImageQuant software by GE Healthcare by highlighting the areas with rectangles and then determining volume intensities with local background subtraction.

The band intensities demonstrate that the cyclin D3 signal decreased during the overnight incubation of the serum sample, but not in plasma. The signal at 4°C was 2.18 times greater than the signal after the incubation. In contrast, the total pixel intensities for plasma cyclin D3 were similar at both times and were consistent with bulk array signals for cyclin D3—1070 at T = 0 and 1150 at T = 24. The bulk signals for the Western blot confirmed the decreasing serum cyclin D3 levels detected by the arrays. These changes may be related to active proteases present in serum.

Figure 7. B-type natriuretic peptide (BNP) band changes in serum and plasma during incubation at 37ÞC. Bands decrease in intensity during the incubation.

Since none of the array antibodies is used for human diagnosis, protease activity in serum and plasma was tested by monitoring the turnover of BNP, a critical marker of cardiac function. BNP concentrations greater than 100 pg/ml are highly correlated with heart failure.15 An accurate determination of BNP levels is important for appropriate diagnosis and therapy. Proteolysis was sought by monitoring dye-labeled BNP band intensity decreases after protein spikes in serum and plasma by denaturing gels (see Figure 7).

Amino terminal fluorescein–labeled BNP was obtained from Phoenix Pharmaceuticals Inc. (Belmont, CA) and spiked into blood fractions as follows: 95 ml of serum or plasma was spiked with 2.5 ml (1 nM BNP), and 5-ml aliquots were removed after each hour for the first four hours at 37°C.

The aliquots were transferred to 95 ml Laemmli buffer and boiled for 3 minutes. A sample of 7.5 ml was separated on Criterion 18% gels from Bio-Rad Laboratories and imaged with the Typhoon 9410. Bands were quantified with ImageQuant.

Measurements of the band signals demonstrate that 18% of the serum BNP protein is degraded within the first hour of incubation, and 45% after four hours. In contrast to serum, little BNP was degraded in the plasma from healthy blood as indicated by the bands and measurements of the BNP band intensities.

Conclusion

Because of the importance of novel biomarkers to clinical diagnostics and medicine, blood proteins remain at the forefront of human proteomics. The experiments described here demonstrate that a high number of low-abundance cellular proteins can be detected in whole-blood fractions despite the presence of albumins and immunoglobulins that often complicate analytical techniques such as gel electrophoresis and mass spectrometry. Typically, the Clontech array–detectable proteins should not be present at high concentrations in blood because the antibody specificities include cell surface markers, transcription factors, DNA modification proteins, and oncogene expression products. The data demonstrate that arrays offer an approach to detecting multiple targets at low concentrations. Furthermore, array detection of many serum proteins was shown to decrease during the incubation period while plasma protein detection appeared more consistent. These serum changes could be related to the cell components in blood or active proteases involved in coagulation.

In some cases, the signals and patterns of detectable proteins can be employed to understand the sample biology. For example, the detection of proteins such as caspase-4 and caspase-8 suggests apoptosis induction, while the presence of CDC 27, Rag-1, and IL1b imply T-cell activation, B-cell differentiation, and a proinflammatory response, respectively. On the other hand, plasma proteins such as Max, eIF-4e, Xpa, and Rag-1 are involved in elongation, DNA repair, and DNA recombination, respectively. Array analyses suggest that the cell components of collected samples are different and could affect the in vitro sample.

James L. Schram (left) is a scientist, clinical diagnostic research, and Robert E. Pearson, PhD, is a senior scientist, clinical diagnostic research, at BD Technologies (Research Triangle Park, NC). The authors can be reached at jim_l_schram@bd.com and bob_e_pearson @bd.com, respectively.

Clotted whole blood contains both a fibrin mass and active proteases that can degrade study-relevant proteins. The study data demonstrate that detection of proteins in serum samples is different from results obtained by analyzing the same proteins in plasma samples. This is true both in terms of detectable markers, bulk signals, and normalized ratios. The results indicate that the most appropriate blood-protein sample for biomarker discovery and clinical diagnostics of proteins is anticoagulated blood plasma.

Protease cascades are very common in a number of biological systems, such as clotting and complement fixation. While the cleavage specificities of the cascade proteases are generally good, some proteases lack target-sequence specificity. Therefore, they are most likely responsible for biomarker degradation. Besides protease cascades normally found in healthy individuals, other proteases could be present in blood as a consequence of a disease process. For example, prostate-specific antigen is a protease that is released into blood and serves as an indicator of prostate cancer. Dipeptidyl peptidases are often activated and released into blood during lymphocyte activation.16 Similarly, apoptosis enzymes can also be released into blood during inflammation related to a disease like septic shock.17 Proteases, as a consequence of disease, might degrade clinically important targets during collection and processing. Including protease inhibitors as part of the blood-collection process might stabilize diagnostically relevant proteins and lead to improved biomarker detection.

References

  1. S Hanash, “Disease Proteomics,” Nature 422 (2003): 226–232.

  2. T Rabilloud, “Two-Dimensional Gel Electrophoresis in Proteomics: Old, Old Fashioned, but It Still Climbs Up the Mountains,” Proteomics 2 (2002): 3–10.

  3. JN Adkins et al., “Toward a Human Blood Serum Proteome: Analysis by Multidimensional Separation Coupled with Mass Spectrometry,” Molecular and Cellular Proteomics 1 (2002): 947–955.

  4. EF Petricoin et al., “Use of Proteomic Patterns in Serum to Identify Ovarian Cancer,” The Lancet 359, no. 9306 (2002): 572–577.

  5. MJ Evans et al., “Effect of Anticoagulants and Storage Temperatures on Stability of Plasma and Serum Hormones,” Clinical Biochemistry 34, no. 2 (2001): 107–112.

  6. JH Livesey et al., “Effect of Time, Temperature, and Freezing on the Stability of Immunoreactive LH, FSH, TSH, Growth Hormone, Prolactin, and Insulin in Plasma,” Clinical Biochemistry 13, no. 4 (1980): 151–155.

  7. JY Kwak et al., “The Comparative Analysis of Serum Proteomes for the Discovery of Biomarkers for Acute Myeloid Leukemia,” Experimental Hematology 32, no. 9 (2004): 836–842.

  8. P Friedl and K Wolf, “Proteolytic and Non-Proteolytic Migration of Tumour Cell and Leucocytes,” Biochemical Society Symposia 70 (2003): 277–285.

  9. DM Goldberg, “Proteases in the Evaluation of Pancreatic Function and Pancreatic Disease,” Clinica Chimica Acta 291, no. 2 (2000): 201–221.

  10. M Levi, E De Jonge, and T Van Der Poll, “New Treatment Strategies for Disseminated Intravascular Coagulation Based on Current Understanding of the Pathophysiology,” Annals of Medicine 36, no. 1 (2004): 41–49.

  11. JD Hulmes, “An Investigation of Plasma Collection, Stabilization, and Storage Procedures for Proteomic Analysis of Clinical Samples,” Clinical Proteomics 1, no. 1 (2004): 17–32.

  12. R Ekins, F Chu, and E Biggart, “Multispot, Multianalyte, Immunoassay,” Annales de Biologie Clinique 48, no. 9 (1990): 655–666.

  13. K Anderson et al., “Protein Expression Changes in Spinal Muscular Atrophy Revealed with a Novel Antibody Array Technology,” Brain 126, no. 9 (2003): 2052–2064.

  14. J Sambrook, EF Fritch, and T Maniatis, Molecular Cloning: A Laboratory Manual, 3 vols. (Cold Spring Harbor, NY: Cold Spring Harbor Laboratory Press, 1989).

  15. C Mueller et al., “Use of B-Type Natriuretic Peptide in the Evaluation and Management of Acute Dyspnea,” New England Journal of Medicine 350, no. 7 (2004): 647–654.

  16. S Iwaki-Egawa et al., “Dipeptidyl Peptidase IV from Human Serum: Purification, Characterization and N-Terminal Amino Acid Sequence,” Journal of Biochemistry 124, no. 2 (1998): 428–433.

  17. VD Joshi, DV Kalvakolanu, and AS Cross, “Simultaneous Activation of Apoptosis and Inflammation in Pathogenesis of Septic Shock: A Hypothesis,” FEBS Letters 555, no. 2 (2003): 180–184.

Copyright ©2005 IVD Technology