MSACL 2016 US Abstract

Is It Better to Be Lucky or Absolute in Protein Quantification?

Christopher Shuford (Presenter)
Laboratory Corporation of America

Authorship: Christopher M. Shuford, Patricia M. Holland, Russell P. Grant
Laboratory Corporation of America Holdings, Burlington, NC

Short Abstract In order to overcome the shortcomings of peptide calibrators, many efforts have been made to use full-length proteins as calibration materials with the premise that conversion of a full-length protein calibrator into its signature peptide(s) will occur with the same efficacy as conversion of the endogenous protein to be measured. If this is universally true, then the accuracy of protein-calibrated assays would be independent of 1) the digestion conditions employed and 2) the signature peptide selected. In our efforts to quantify the endogenous protein biomarker, thyroglobulin, using full-length protein calibrators and internal standards we have observed this not to be the case – indicating full-length proteins calibrators do not universally provide absolute quantification.

Long Abstract

The notion of absolute quantification of proteins when employing enzymatic digestion is tenuous. It is now generally accepted that digestion-based methods employing synthetic proteolytic peptides as calibration material cannot reliably provide absolute protein measurements because this requires stoichiometric conversion of the parent protein(s) into its signature peptide(s), the success of which varies based on the digestion conditions employed, the signature peptide(s) selected, and even the laboratory performing the measurement. Consequently, there are significant implications on the translatability of peptide-standardized assays between laboratories because the measurand in each laboratory is a composite of the absolute protein concentration in the sample as well as the inherent biases imparted by the assay methodology on that sample (which are potentially un-controlled/un-corrected with peptide standards). To that end, additional efforts have been placed on using full-length proteins as calibration materials, with the premise that conversion of a full-length protein calibrator into its signature peptide(s) will occur with the same efficacy as conversion of the endogenous protein to be measured. If this is universally true, then the accuracy of protein-calibrated assays would be independent of 1) the digestion conditions employed, 2) the signature peptide selected, and 3) the sample to be measured. To varying degrees, this has been suggested as common place when using of full-length, recombinant proteins as internal (stable isotope-labeled, SIL) and/or external (un-labeled) calibrators for quantification. Nonetheless, in our efforts to quantify the serum protein biomarker, thyroglobulin (Tg), using full-length protein calibrators and internal standards we have observed numerous discrepancies between digestion conditions, signature peptides, and even protein calibrators – indicating full-length protein calibrators do not universally provide absolute quantification.

Firstly, in a study of 95 serum specimens, two signature peptides for Tg, VIFDANAPVAVR and FSPDDASAVLLR, were compared using an external protein calibration curve prepared in a surrogate matrix and cleavable peptide internal standards. In the 78 specimens negative for Tg autoantibody (TgAb), protein measurement by the two peptides showed good agreement, yielding a deming slope of 1.077 and correlation coefficient of 0.99. Moreover, Tg concentrations obtained with each signature peptide were within ±20% agreement for 97% of the TgAb-negative samples indicating negligible impact for inter-patient differences on quantitative accuracy. By contrast, in the remaining 18 samples that were positive for TgAb, 5 of 15 with quantifiable amounts of Tg showed a significant negative bias (<-20%) by the VIF peptide relative to the FSP peptide. Similarly, biases as great as -80% were observed in serum pools comprised of autoantibody positive specimens, suggesting the VIF signature peptide digestion recovery was reduced by the presence of autoantibodies. To confirm under recovery by the VIF peptide was linked to autoantibody interference, a pool of antibody-negative sera was assayed for both signature peptides and assayed again following spiking with two exogenous anti-thyroglobulin antibodies at increasing concentrations. At lower concentrations of antibody, Tg recovery by both peptides was identical to that observed in the absence of antibody (+/- 15%). However, at higher concentrations of one antibody (2.4 IU/mL), the VIF peptide showed recovery less than 50%, confirming that protein-antibody interactions can affect the digestion process. From these data, it can be concluded that bias in the digestion process between samples is not accounted for by the use of 1) external protein calibrators because they inherently lack the ability to mimic heterogeneous protein-protein interactions and 2) cleavable peptide internal standards because they lack the higher order structure to experience and be affected by protein-protein interactions. This has implications on all externally calibrated assays, as well as assays employing peptide internal standards, as the bias imparted by such protein-protein interactions will likely be fundamental to 1) the digestion conditions employed and 2) the signature peptide selected.

To that end, we have also explored the use of full length SIL-Tg as both an internal standard to correct for sample-sample biases imparted by digestion, as well as sample-calibrator biases imparted by matrix-specific digestion. Initially, we compared results obtained with the VIF and FSP peptides in TgAb-negative patients again using external protein calibrators in combination with the SIL-Tg internal standard. In this study, the VIF peptide produced systematically higher Tg concentration assignments than the FSP peptide (Deming Slope = 1.164), while maintaining good correlation (R = 0.99). Nonetheless, these results indicate a systematic bias between peptides produced in the calibration and sample systems, which is not wholly corrected even by the SIL-Tg internal standard. Comparing the light:heavy response ratios measured in serum samples using the SIL-Tg revealed almost no bias between the two peptides in serum specimens (Deming Slope = 0.987), but a systematic bias in calibrators (Deming Slope = 0.863). Indeed, identical light:heavy ratios should be observed for all peptides produced from the light and heavy protein forms given the molar equivalency is ensured for all peptides within a given protein (in-vitro and in-vivo modifications notwithstanding). Consequently, this data suggests yield of the two peptides from SIL-Tg and native Tg in serum samples is identical under the digestion conditions used, but different from the exogenous Tg used in external calibrators.

As such, we next tested using recombinant SIL-Tg as the internal standard with 4 sources of full-length sources of Tg in combination with 3 calibration matrices to create 12 external calibration systems (i.e., 12 protein-matrix combinations total) to determine which combination would provide inter-peptide agreement within calibrators (and, thus, samples). Within a given external calibration system, variable (dis)agreement was observed between peptides, with mean biases spanning -2 and +31% and only 4 of 12 external calibration systems showing inter-peptide agreement with +/- 10%. However, if all combinations of peptide and external calibration systems (24 combinations total) were compared with one another, only 80 of 276 possible non-redundant comparisons (29%) showed agreement within +/- 10%, with mean biases spanning -40 to +84% . The disagreement observed in this study between peptide and between external calibration systems exemplify the difficulty in obtaining inter-assay agreement between laboratories as each of these 24 peptide-external calibration combinations represent rationale assay systems for quantifying intact proteins. Indeed, given the disagreement was observed despite the use of full-length proteins as both calibrator and internal standard (presumably “best practice”), this data suggests different forms of a common protein may not digest with the same efficacy, which has implications on the translatability of biomarkers from discovery to clinically use.

Details of these studies will be presented together with the following studies: 1) incorporation of additional peptides for comparison, 2) exploration digestion efficiency of additional sources of purified Tg, and 3) determination if recombinant SIL-Tg can mitigate inter-peptide differences in the presence of Tg autoantibodies.

References & Acknowledgements:

Financial Disclosure

IP Royalty: no

Planning to mention or discuss specific products or technology of the company(ies) listed above:

yes