Discussion

The goal ofthe SCHC is to develop a standard operating procedure (SOP) for blood collection and handling for proteomic studies. This is a daunting task considering the large number of variables that must be thoroughly studied and the complica

Fig. 5 Examination of the effect of a protease inhibitor cocktail and individual protease inhibitors on the 2-DE profile of human plasma. Eight samples (50 mL) of heparin plasma (HUPO PPP BDAA01-Hep) were subjected to the addition of the following reagents: Protease inhibitor cocktail (Panel A), water (Panel B), AEBSF (Panel C), Aprotinin (Panel D),

Fig. 5 Examination of the effect of a protease inhibitor cocktail and individual protease inhibitors on the 2-DE profile of human plasma. Eight samples (50 mL) of heparin plasma (HUPO PPP BDAA01-Hep) were subjected to the addition of the following reagents: Protease inhibitor cocktail (Panel A), water (Panel B), AEBSF (Panel C), Aprotinin (Panel D),

Leupeptin (Panel E), Bestatin (Panel F), Pep-statin A (Panel G), and E-64 (Panel H). All samples were depleted of albumin and IgG. A portion of each of the depleted samples (Panels A and B - 500 mg, Panels C-H -200 mg) was separated on 2-DE gels as described in Section 2.

tion and difficulties in standardization, as evidenced within the HUPO PPP efforts to date. A microcosm of this complexity is captured in the data presented above: a wide variety of analytical techniques addressing specific but varied parameters, from which to draw concrete and broadly applicable conclusions. As such, we do not put forward

Fig. 6 Representative antibody microarray images from four experimental conditions: (A) No PI added; (B) PI added after the labeling; (C) PI added before the labeling; (D) PI added both before and after the labeling. The composite images from the 543 nm (green) and the 633 nm (red) channels are shown.

Fig. 6 Representative antibody microarray images from four experimental conditions: (A) No PI added; (B) PI added after the labeling; (C) PI added before the labeling; (D) PI added both before and after the labeling. The composite images from the 543 nm (green) and the 633 nm (red) channels are shown.

broad, sweeping recommendations based on the individual data sets. However, we believe that our data represent important lessons for the field of plasma proteomics, so we present here general recommendations and cautions that can help guide researchers toward a more thoughtful experimental design, hopefully leading to even more robust plasma proteome studies in the future. As the collective knowledge ofthe field grows, eventually we may achieve a definition for a handling and collection SOP.

The data as presented, although derived from independent experiments, provides insight into pre-analytical variables that could prove detrimental to proteomics experimental design and outcome. The following discussion ofthe data illustrates particular pre-analytical variables, and how the proteomic analyses in question were altered. Topics touched by this data include platelet contamination, storage, and the use of protease inhibitors as a protective mechanism. This is the first step towards developing a sample acquisition and handling SOP for plasma proteomics.

A.

18000 --

■ Patient 21 □ Patient 31_

16000 -14000 -12000 -10000 -8000 -6000 -4000 -2000 --

0

f

No PI Labeling only Column only Both

Fig. 7 Average backgrounds and signal-to-background ratios for the four conditions. (A & B) For each of the three serum samples, the local backgrounds from all the antibodies on the array were averaged and plotted with respect to experimental condition for both the 543 nm (A) and the 633 nm (B) channels.

Fig. 7 Average backgrounds and signal-to-background ratios for the four conditions. (A & B) For each of the three serum samples, the local backgrounds from all the antibodies on the array were averaged and plotted with respect to experimental condition for both the 543 nm (A) and the 633 nm (B) channels.

D.

(C & D) The signal-to-background ratios from all the antibodies on the array were averaged and plotted with respect to experimental condi tion for both the 543 nm (C) and the 633 nm (D) channels. The signal-to-background ratio was calculated as (S-B)/B, where S = fluorescence in the spot and B = local background.

(C & D) The signal-to-background ratios from all the antibodies on the array were averaged and plotted with respect to experimental condi tion for both the 543 nm (C) and the 633 nm (D) channels. The signal-to-background ratio was calculated as (S-B)/B, where S = fluorescence in the spot and B = local background.

The topic of platelet contamination and also that of choice of sample type, i.e., serum versus plasma, are addressed by an analysis of peptide contents of various samples. During processing of venous blood into serum, various ex vivo processes occur which lead to neo-generation of many peptides. Since serum generation relies on a biochemical process, it is reasonable to expect that various parameters like temperature after sample collection, time for sample processing/clot formation or medication of patients, can alter the peptide content of serum. These issues are difficult to standardize in routine clinical practice, in cohorts, and among different centers. Furthermore, certain proteins may bind to the clot in an uncontrolled fashion (causing a concominant decrease in free protein concentration during clot formation). Finally, serum shows many highly concentrated and intense peptide signals, which impede the detection of endogenous peptides. Therefore the use of serum samples for peptidomic mono/oligo biomarker discovery should be avoided with these caveats in mind. Different research groups [13, 14] use serum peptide patterns for prediction of early stage cancers and the controversial debate [15, 16] about this approach is still ongoing. At this time it is not clear whether the proteomic patterns reflect directly disease related peptides or peptides which are generated due to secondary effects during ex vivo coagulation and may only be loosely connected to the disease. Nevertheless, serum is the most commonly archived sample and may be helpful to validate results from biomarker studies, so care must be exercised when using archived samples, given the potential pitfalls described.

With regard to selection of the optimal sample type, we suggest that this will be dependent on the downstream analysis that is performed. Each of the individual sample types: serum, EDTA-plasma, heparin-plasma, and citrate-plasma, all exhibit shortcomings and should not be used under particular circumstances. We have discussed above the issues relating to serum ([12] and Fig. 1). EDTA is an aminopoly-carboxylic acid and is negatively charged. It forms soluble complexes with metal ions and prevents them from further reaction [17]. If the endpoint measurement of interest involves assays wherein divalent cations, such as Mg21 or Ca21, are necessary, EDTA-plasma is not an ideal sample choice. This can occur in cases where the assay is used to measure the free ion, the metal ion is required as an enzyme cofactor, or when the metal ion is an intermediate in the assay reaction. Heparin can result in interference in some affinity processes, such as SELDI-TOF analysis [6]. Heparin is a sulfated glycosaminoglycan that prolongs the clotting time of blood [17]. It is a highly charged molecule, and its presence in solution can compete for or prevent binding of molecules to charged surfaces. This step is important for the surface adsorption of proteins to protein chips; in the SELDI process, these bound proteins are laser-de-sorbed, ionized, and then detected after traversing the length of the flight tube. Finally, citrate can bind calcium [17] and has been shown to cause falsely lowered readings of immunoassay measurements for multiple analytes [18]. The citrate anticoagulant is present in collection tubes in liquid form, and thus, exerts a dilutional effect when blood is added during sample collection. Thus, an educated choice for the selection of optimal sample type requires one to be cognizant of each of these caveats and to take the necessary precautions in choosing the most appropriate specimen.

The HUPO plasma samples show significant differences compared to reference plasma collected according to the BioVisioN sample protocols. In our study, different protocols for EDTA/citrate specimen collection were investigated. A major finding is the detection of platelet-derived peptides in HUPO specimens. After correlation analysis (detailed in [12]), we estimate that at least 14% of peptides are platelet related. This was confirmed by sequencing a subset of 20 peptides. The appearance of platelet related peptides is either because residual platelets were present in HUPO samples, or platelets were activated prior to their removal. This may have led to release of platelet derived peptides or enzymes causing further proteolysis of proteins. The elimination of platelets prior to freezing is recommended; in addition, activation ofplatelets prior to their removal has to be kept to a minimum. We established protocols to remove platelets (total residual platelets <10/nL, see [12]) to obtain a platelet-poor plasma by combining centrifugation with a filtration step. Alternatively, sequential centrifugation at room temperature may be useful. We suggest the use of platelet-poor plasma as the preferred specimen for peptidomic analysis. Further, we suggest that protein sequence identifications corresponding to intracellular proteins should be carefully characterized to distinguish between possible biomarker candidates from distant organ systems and platelet derived, contaminating proteins.

Regarding storage conditions, the purpose of our studies was to determine what changes occur to a reference serum sample upon storage under different conditions of temperature and length of time (Tab. 3). Interestingly, in contrast to degradation in cell lysates where incubations of minutes to hours is often sufficient to completely destroy most protein bands and convert gel images to smears, it was observed that even very long term degradation at room temperature in serum only marginally or moderately degrades most high abundance proteins. Thus, the general pattern displayed by serum proteins upon analysis by SDS-PAGE or 2-D gels remains relatively similar, because such read-out is dominated by the abundant proteins. However, as the above data shows, while serum is more stable to proteolysis than cell lysates, substantial proteo-lysis and other modifications to a subset of proteins is clearly occurring, even at 4°C or —20°C. In addition, the changes in enzyme activity are similar to those described for other analytes in the literature [19], and are not unexpected. Although different enzymes are susceptible to varying degrees, a survey such as this, which comprises a panel of proteins, is likely to reveal changes to a subset of the analyzed proteins.

Further, the use of glycerol, or other similar additives, to stabilize protein structure and/or function during storage, crystallization, lyophilization, and other harsh conditions, has been previously described [20-25]. Results of these studies demonstrate that such additives can help mitigate damage to protein structure and/or stability, demonstrating the sensitivity that certain proteins can have towards their physical environment.

Extending beyond this generic use of alcohol additives, several approaches have been used to measure the specific benefits ofprotease inhibitors to protect plasma proteins. Interestingly, on 2-D gels, there was no direct evidence that PIs protected against a decrease in molecular weight. Instead, findings suggest that inclusion of a particular protease inhibitor cocktail causes isoforms of certain human plasma proteins (depleted of albumin and IgG) to be shifted to higher pi isoforms. These phenomena are fully discussed at a high level of detail in a separate manuscript [26]. Systematic analysis of the PI cocktail leads to the conclusion that AEBSF was the cause of this significant distortion of the 2-DE gel profile using citrated human plasma. Similar results were observed using heparinized and potassium-EDTA plasma (results not shown). AEBSF (also called Pefabloc) is a serine protease inhibitor, which inhibits proteases such as trypsin, chymotrypsin, plasmin, kallikrein and thrombin [27]. Inhibition of these proteases takes place by covalent (irreversible) modification of serine residues in the active site through the formation of sulfonate esters. AEBSF has also been shown to deriva-tize other proteins by serine modification [28]. Such a covalent modification would add an amine functional group, thereby shifting the proteins to higher pi forms.

In contrast, the SELDI study presented suggests that a PI cocktail has immediate and detectable benefits to stabilize proteins at the time ofphlebotomy. For processed blood samples, either serum or plasma, the concept of "time zero" is such that at least several minutes pass during processing before any newly acquired blood sample is available for further analysis. Even under optimal collection conditions, blood processing requires up to 10 min post-phlebotomy for plasma, and at least 30 min or more for serum. During this processing time, the intrinsic biochemical processes ofblood continue to operate, including enzymatic proteolysis that could possibly affect prote-

omic analyses [29]. The potential beneficial effects of protease inhibitors have been analyzed. The presence of protease inhibitors at the moment of blood draw, could reduce proteolytic damage, and thus prevent any time dependence of sample processing to overall proteomic and diagnostic outcomes. A representative result (Fig. 4) is included here, and a more thorough description of the experiment and results will be described in a subsequent manuscript (Yi and Gelfand, in preparation).

Interestingly, these observed differences are present at "time zero," suggesting the requirement for and benefit of the immediate mixing of PIs with blood during phlebotomy. These data are consistent with Pi-mediated preservation of the species corresponding to this peak of interest. There are a number of other peaks (Yi and Gelfand, unpublished observations) that appear "preserved" by PI presence, and also a number of peaks, generally in the peptide range, that are more intense in the EDTA samples, so the figure presented does not represent an isolated incident. All data thus far are consistent with the benefits of blocking protease activity, and, perhaps more importantly, of blocking this activity immediately, during sample acquisition.

Another observed benefit of PI use has been seen with an antibody microarray approach. The decrease in background signal with the addition of PI could be the result of cleavage products being more "sticky," or likely to bind nonspecifically, as compared to their whole-protein counterparts. Thus, when the production of cleavage products is suppressed, nonspecific binding to the nitrocellulose is also suppressed. The cumulative effect from addition of PI both before and after labeling likely reflects the amount of time the samples were incubated with or without inhibitors. Further research will focus on the effects of PI on data quality when using other surfaces and on the quantification of specific proteins.

In summary, with regard to the use of PIs, we conclude that their use is recommended for top-down analyses. However, the researcher must be careful to avoid some of the potential drawbacks that might arise with use of PIs, with particular regard to the analytical method and targets being evaluated. Care should be exercised in the selection of the specific inhibitor. Those which act through a non-covalent mechanism should not cause any unexpected modifications. The observations of alterations in isoelectric forms in the presence of AEBSF provide an example of artifactual modification of proteins that are possible when using covalently modifying PIs, potentially adding analytical complexity.

Beyond potential artifactual modifications, there are process considerations as well. Obviously, inclusion of PIs immediately prior to a complete proteolytic sample digestion (e.g., for MS/MS analysis in a "bottom-up" approach) is counterproductive. Further, the use oflow molecular mass PIs, or other chemical additives, can mask the presence of species at similar mass ranges, such as for peptide analyses.

Was this article helpful?

0 0

Post a comment