Christopher Shuford (Presenter)
Laboratory Corporation of America
Authorship: Christopher M. Shuford(1), Patricia M. Holland(1), Kevin Ray(2), James J. Walters(2), Uma Sreenivasan(3), Sarah Aijaz(3), Russell P. Grant(1)
(1) Laboratory Corporation of America Holdings, 1447 York Ct., Burlington, NC, 27215 (2) MilliporeSigma, 2909 Laclede Ave., Saint Louis, MO, 63103 (3) MilliporeSigma, 811 Paloma Dr., Round Rock, Texas, 78665
Fully-tryptic stable-isotope labeled (SIL) peptides, cleavable SIL peptides, and a full-length SIL protein were compared as internal calibrators for quantifying 3 forms of unlabeled thyroglobulin (Tg) by protein cleavage-isotope dilution mass spectrometry using multiple digestion conditions. All SIL calibrators and unlabeled proteins were standardized by amino acid analysis to allow confident assignment of accuracy. Collectively, the results demonstrate lack of commutability for peptide calibrators across digestion conditions and potential for different proteoforms to provide disparate concentration assignments due to variations in the peptide formation despite altering denaturation/digestion stringency. Nonetheless, these results still support the use of recombinant, full-length proteins as calibrators and internal standards, in favor of pseudo-protein calibrators.
A logical calibration hierarchy for Protein Cleavage-Isotope Dilution Mass Spectrometry (PC-IDMS) assays is as follows: fully cleaved peptides (lowest), cleavable peptides/concatamers, partial proteins, full-length proteins (highest). This ranking stems from a general understanding that during protein digestion, typically with trypsin, stoichiometric conversion of the protein into its signature peptide(s) is not guaranteed – more likely the opposite. Thus, quantifying the peptide resulting from digestion does not result in an estimate of the protein quantity, but rather an estimate of the combined protein quantity and the relative digestion efficiency of the signature peptide being derived and measured. Conversely, it has been proposed that full-length protein calibrators enable true “absolute” quantification of protein concentrations as demonstrated by the agreement in the protein quantities derived from multiple signature peptides. To exemplify this hypothesis, we compared the use of 8 fully-tryptic stable-isotope labeled (SIL) peptides, 8 cleavable SIL (cSIL) peptides, and a full-length SIL protein as internal calibrators for quantifying 3 forms of naturally-labeled (NAT) thyroglobulin (Tg). The 3 forms of NAT thyroglobulin quantified were 1) BCR457 (thyroglobulin international reference material), 2) a commercially available purified thyroglobulin, 3) a full-length recombinant protein expressed and purified in identical fashion to the full-length SIL protein, but without labeled amino acids.
Independent, neat stock solutions of each SIL peptide, cSIL peptide, SIL Tg, and NAT Tg were initially standardized by replicate amino acid analysis (AAA) using NIST amino acid calibration materials. A NIST-traceable BSA control was run in parallel to qualify the accuracy of the AAA. Subsequently, all working solutions were gravimetrically prepared to enable traceability to the original AAA-assigned values. All SIL materials were labeled at different amino acid positions within the final signature peptide sequences to ensure selective detection of each peptide-form in the same sample, which was confirmed through negative control experiments.
For quantification by PC-IDMS, 3 pmol of each NAT-Tg was combined gravimetrically with 3 pmol of SIL-Tg in a matrix of 0.1% HSA, 1X PBS. Samples were then mixed 1:1 with denaturant containing 10 mM DTT, then heated for 30 minutes at 56 °C. Denatured samples were further diluted 1:1 with 125 mM Tris-Buffer (pH 8) containing 1.5 pmol of each cSIL peptide prior to adding 5 µg of trypsin and incubating for 30 minutes at 37 °C. Finally, digestion was terminated in each sample by addition of formic acid to a final concentration of 1% (v/v) and addition of 3 pmol of each SIL peptide. This procedure was tested with multiple denaturants including 4 M urea (1M during digestion), 1% sodium deoxycholate (DOC, 0.25% during digestion), and 10% trifluoroethanol (TFE, 2.5% during digestion). Quantities of NAT-Tg were calculated by taking the quotient of the NAT:SIL peptide peak areas measured by LC-SRM (with matching transitions) and multiplying by the nominal amount of the SIL calibrator added.
Across the 3 different denaturation/digestion conditions, the 8 SIL peptides resulted in quantitative accuracy ranging from 1.4 to 124% - spanning 2-orders of magnitude - with only 16 of 72 possible quantities being within +/- 20% of the expected quantity for the 3 forms of NAT-Tg tested. Indeed, 54 of 72 results demonstrated accuracy <80% due to incomplete conversion of the protein targets into the corresponding signature peptides. There were also noticeable differences in quantities obtained with each signature peptide across the 3 digestion conditions. For example, one signature peptide (VIFDANAPVAVR) resulted in quantitative accuracy between 80 and 105% across all 3 forms of NAT-Tg and 3 digestion conditions, while another signature peptide (SQAIQVGTSWK) yielded between 104 and 124% accuracy when denaturing with DOC or urea, but only 50 to 58% accuracy when denaturing with TFE. Similarly with cleavable SIL peptides, quantitative accuracy ranged from 20 to 165% across the 3 digestion conditions and 3 NAT-Tg forms, with only 13 of 72 quantities being within +/- 20% of the expected quantity. Time course analysis of the trypsin digestion demonstrated these results were not confounded by insufficient time for peptide formation/generation, but rather confirmed the peptide formation from all substrates had plateaued well before the 30 minute time point used for quantification.
When quantifying using the SIL-Tg as the internal calibrator, all 8 signature peptides across the 3 digestion conditions were systematically biased by ~20% for all 3 NAT-Tg forms. Given these results were conserved across all 3 digestion conditions and by repeat analysis, this result suggests a potential error in the concentration assignment of the SIL-Tg stock material by AAA or loss (e.g., due to instability or adsorption) of the SIL-Tg during manufacturing or storage of the stock material. When comparing to the overall mean of all conditions in order to adjust for this apparent systematic bias, 57 of 72 possible quantities were within 80 to 120% accuracy when using SIL-Tg as the internal calibrator.
Most interestingly, was the fact that all 8 peptides demonstrated good agreement (+/- 20%) in the resulting protein quantity calculated for the recombinant NAT-Tg, but there was marked disagreement between 2 of 8 signature peptides when quantifying the BCR-457 and the commercial Tg. When quantifying BCR-457 using SIL-Tg, the signature peptide TFPAETIR demonstrated 67% accuracy, while the peptide LEDIPVASLPDLHDIER demonstrated 124% accuracy relative to the mean quantity derived from all 8 peptides. Likewise, these two peptides demonstrated 66 and 120% relative accuracy, respectively, for the commercial Tg, indicating that the digestion recovery of these two peptides is substantively different between the recombinant SIL-Tg and the native NAT-Tg forms, but not different between the recombinant SIL-Tg and recombinant NAT-Tg.
Further demonstrating the difference in trypsin digestion efficiency for native and recombinant proteins, 39 signature peptides (including 7, single missed-cleavage peptides) were subsequently monitored and compared between the SIL-Tg and 3 forms of NAT-Tg across all 3 digestion conditions. In quantifying the recombinant NAT-Tg, all 39 peptides demonstrated 80 – 120% accuracy across all 3 conditions. In contrast, only 34, 33, and 30 peptides demonstrated 80 – 120% accuracy in concentration assignments for BCR457 when digesting with DOC, urea, and TFE, respectively. Likewise, only 24, 27, and 21 peptides demonstrated equivalent accuracy in concentration assignments for the commercial Tg under the same digestion conditions, respectively.
Collectively, these results unequivocally demonstrate 1) the lack of commutability for peptide calibrators, fully tryptic or cleavable, across various assays (i.e., digestion conditions) and 2) the potential for different proteoforms (i.e., recombinant versus native) to result in disparate concentration assignments by PC-IDMS due to variations in the peptide formation. Nonetheless, these results still support the use of recombinant, full-length proteins as calibrators and internal standards, in favor of pseudo-protein calibrators.
References & Acknowledgements:
|Salary||yes||Laboratory Corporation of America|
|Stock||yes||Laboratory Corporation of America|
IP Royalty: no
|Planning to mention or discuss specific products or technology of the company(ies) listed above:||