Multiplexed MRM concentration measurements of >200 endogenous proteins in dried blood spots for biomarker discovery and validation
Pre-analytical sample preparation
DBSs were visually examined for uniformity, and if a spot displayed visual discoloration or clusters of solidified material, it was omitted from further analysis. DBS discs (6 mm) were punched out using a pneumatic card punch (Analytical Sales & Services, Inc., Flanders, NJ). Disc punches were performed by punching each spot either west, north, east, south, or center (Supplemental Figure 1). All five discs were collected in a single 1.5-mL conical tube and 1250 µL (i.e., 250 µL per punch) of 50 mM ammonium bicarbonate – 2% sodium deoxycholate (DOC) solution was added. Solid to liquid phase extraction from the discs was achieved by incubation for 45 min at 37 ⁰C with shaking at 1400 rpm using an Eppendorf Thermomixer R (Fisher Scientific, Toronto, ON). Samples were centrifuged for 15s to pellet the discs and particulates, and positive displacement pipette was used to transfer 200 µL to a new conical tube. A 10-µL aliquot of 0.5 M Bond-Breaker TCEP solution (Fisher Scientific, Toronto, ON) was added, and the solution was incubated at 99 ⁰C for 15 min with shaking at 950 rpm on an Eppendorf Thermomixer R. Samples were centrifuged briefly at 17,000 xg, 30 µL of an aqueous 100-mM iodoacetamide solution was added, and the samples were incubated at 37 ⁰C for 15 min in the dark. Using a positive displacement pipette, 50 µL of the sample solution was transferred to a new conical tube, and 60 µL of TPCK-treated trypsin (>180 units per mg protein, 5.5 µg/ml in 50 mM ammonium bicarbonate) (Worthington-Biochemicals, Lakewood, NJ) was added. Sample was incubated in a thermomixer for 9 h at 37 ⁰C with shaking at 750 rpm. Subsequently, 12 µL of 10% formic acid (in water) was added to inactivate the trypsin and to precipitate the DOC. Sample was centrifuged at 21,000 x g for 10 min, and 70 µL of the supernatant was transferred to a 300-µL sample injection vial (PP screw vial, 12 x 32mm, 9mm thread, from Canadian Life Science, Edmonton, AB). At this step the sample was ready to be spiked with peptide standards prior to LC-MS/MS analysis.
A surrogate matrix was made for formulating the samples used to generate the calibration curves. Unused DBS extracts which had been processed up to the addition of trypsin, were pooled to make a surrogate matrix (SM). To 500 µL of pooled DBS extract, 600 µL of proteinase K (≥30 units/mg protein, 5.5 µg/ml, in 50 mM ammonium bicarbonate) (Sigma-Aldrich) was added and sample was incubated in an Eppendorf Thermomixer R at 37 ⁰C for 9 h with shaking at 750 rpm. To stop the digestion and precipitate the DOC, 120 µL of 10% aqueous formic acid was added to the reaction mixture. The sample was centrifuged for 10 min at 21,000 x g to pellet the DOC and any other debris, and the supernatant was transferred to a 5000 MW-cut-off ultra-filtration conical tube (Agilent Technologies, Cedar Creek, TX) and centrifuged 4,000 x g for 45 min to remove the proteinase K from the surrogate matrix. The flow-through was used for preparing the samples for generating the calibration curve.
Protein identification using DBS
Proteins were selected for MRM assay development based on their detectability in DBSs, and were identified by performing bottom-up proteomics in-house. After pre-analytical sample preparation (described above) of three replicate DBS samples, the samples were subjected to solid phase extraction  and lyophilisation. Samples were solubilized in 210 µL of 3% aqueous acetonitrile in water and used for LC-MS/MS on an Orbitrap Fusion Tribrid coupled to an EASYnLC 1000 HPLC system via a Nanospray Flex NG source (Thermo Fisher Scientific), as previously described [3, 4]. The only modification made was to inject 1 µL of each technical replicate, which was equivalent to ~ 1 µg of total protein. The raw data files were processed as previously described  with only one modification: the peak lists submitted to the Mascot 2.4.1 server were searched against the UniProt human database . Proteins were selected for MRM assay development by consulting PeptideTracker  for available MRM compatible stable isotope labelled standard (SIS) peptides. Using this resource, 356 SIS peptides, targeting 245 unique proteins (for select proteins multiple unique SIS peptides were available), were selected for MRM assay development.
Targeted MRM analysis of the final MRM assay panel was performed on an Agilent 6490 triple quadrupole mass spectrometer connected to a 1290 Infinity UHPLC system via a jet stream ESI source (Agilent Technologies). The column compartment was equipped with a Zorbax Eclipse Plus C18 rapid resolution HD 2.1x150mm 1.8-micron column (Agilent Technologies) and an upstream 0.3 µm in-line filter. UHPLC was performed by setting up a 60 min gradients at a flow rate of 0.4 mL/min. The mobile phases were 0.1% formic acid (FA) in water and 0.1% FA in ACN. The gradient was set as follows: 2% ACN, 0 min; 2% ACN, 4 min; 7% ACN, 6 min; 30% ACN, 54 min; 45% ACN, 57 min; 80% ACN, 57.5 min; 80% ACN, 59.5 min; 2% ACN, 60 min. Post-gradient equilibration time was set to 4 min and the stop-time was set to 60 min. The column compartment was kept at 50 ⁰C.
Targeted MS acquisitions were performed using 3-min detection windows, a 970-ms cycle time, and ≥10 ms and ≤ 400 ms dwell times. Three time segments were used for dynamic MRM as follows : at a start time of 0 min, the valve was diverted to waste, at 4 min the valve was diverted to the MS, and at 57.5 min valve was diverted to waste. A 15-min wash, where 8 µL of 3% aqueous ACN was injected, was included after each DBS sample run to minimize carryover. At the end of each worklist, ten 5-min wash steps were included, injecting 20 µL of 3% aqueous ACN to wash the LC capillary lines from the injection needle onwards, and for these injections, the MS valve was diverted to waste. The Dynamic MRM method for a single transition acquisition is included in Supplemental Table. The source settings were set as follows; gas temperature 150 ⁰C, gas flow 15 L/min, nebulizer 30 psi, sheath gas temperature 250 ⁰C, sheath gas flow 11 L/min, capillary 3500 V and 3000V (-ve) for positive and negative ion funneling, respectively, nozzle voltage was 300 V and 1500 V for positive and negative ionization, respectively, iFunnel parameters for high pressure RF was 200 V for positive mode and 150 V for Negative mode and low pressure RF was 100 V and 60 V for positive and negative, respectively.
Peptide Standard Synthesis
Peptide synthesis was performed as previously described . Briefly, SIS peptides were synthesized by incorporating 13C/15N heavy isotope L-arginine or L-lysine on the C-terminus. Peptides were purified using HPLC and the parent mass to charge ratio confirmed using an Ultraflex III-MALDI TOF/TOF Mass Spectrometer (Bruker LTD., Milton, ON). The purity of each peptide was determined using capillary zone electrophoresis on a 7100 Capillary Electrophoresis System (Agilent Technologies), as previously described [7, 8], and the absolute concentration of each peptide was measured in-house using LC/MS-based amino acid analysis . Briefly, peptide standards were acid hydrolyzed and the resulting amino acids were dansylated  and subjected to HPLC-MS/MS (1290 Infinity HPLC online with 6490 triple Quad MS) (Agilent-Technologies) operated in dynamic MRM mode. Quantification was performed using a 13C or 2H labeled internal standard for each amino acid (Cambridge isotope Laboratories, Inc., Tewksbury, MA). Synthesis and quantification of the corresponding NAT peptides were performed exactly as indicated above , but without incorporation of heavy-isotope labels. For some of the SIS or NAT peptides, which were not quantified via CZE and AAA, the corresponding NAT or SIS peptides which had been quantified via CZE and AAA were used to accurately measure the concentration of the stock peptide.
Selection of transitions and initial assessment of specificity and interference
The selection of the MRM transitions was performed using Skyline-daily  by cross-referencing tandem mass spectral libraries available through the National Institute of Standards and Technology (NIST) . Using these resources, the top five most-intense transitions were selected for collision energy optimization  and retention time determination. If a tandem mass spectrum was not available for a target peptide, transitions were characterized empirically by monitoring all y and b-ion fragments using precursor charge states 2 through 5 and corresponding product ions with charge states 1 through 4. SIS peptides were then spiked into the DBS matrix (i.e., a tryptic-digested DBS sample) and the combination of retention time and transitions was used to confirm the detectability of each endogenous (END) peptide via MRM, and to eliminate transitions prone to interference. As an example, the top five transitions representing the proteotypic peptide (VGEFSGANK), for redox signaling protein thioredoxin (P10599), are shown in (Supplemental Figure 2). The NIST tandem mass spectrum for the same peptide is also included for reference. To maximize assay sensitivity, the collision energy (CE) used to fragment parent peptide ions was optimized as previously described . Briefly, the existing linear equation used for predicting the optimal CE for each peptide, in Skyline-daily , was first used to create a scheduled MRM method. The scheduled MRM method was then used to create a CE optimization method by setting the step count to 5 and the step size to 1 (5 V on each side of the predicted CE, incremental at 1 V). The CE which produced the highest peak area for each transition was used for downstream methods development. To increase downstream multiplexing, only three transitions were selected for subsequent assay development (Supplemental Table). The selection criteria used for selecting the three transitions included prioritizing y-ions over b-ions, since the former is more robustly fragmented in low-energy CID used in a triple quadrupole mass spectrometer , selecting the highest-intensity product ions to maximize sensitivity, and selecting transitions which displayed consistent peak-area ratios when comparing between the heavy and light versions of a peptide.
CPTAC experiment 1  was used as a guide to assess the lower limit of quantification (LLOQ) for each target peptide. Using SIS peptides to spike trypsin digested DBS (DBS matrix), the reverse curve [3, 4] approach was used to generate an eight point response curve (in addition to a blank), to determine the LLOQ. The 336 SIS peptides to be tested were used to make working solutions of SIS peptides containing 5000 fmol/µL of peptide in 3% ACN/H2O. The working solution of SIS peptide was used to make a dilution series for an eight-point response curve, covering the concentration range from 60 attomoles to 4000 femtomoles of SIS peptide injected on column. At the time the response curve was performed, counterpart NAT peptides were available for use as normalizers for 245 SIS peptides. For those SIS peptides where the corresponding NAT peptide was not available, the endogenous (END) analog was used as a normalizer. Additionally, the repeatability was assessed without normalization by measuring the coefficient of variation using the peak areas corresponding to the SIS peptides. Samples to be used for the response curve were prepared in triplicate and each was reconstituted by adding 2 µL of a SIS peptide solution (from the dilution series described above) and 2 µL of NAT from the working solution (containing 5000 fmol/µL in 3% aqueous ACN/H2O) to 46 µL of DBS matrix. Each replicate was injected once for MRM acquisition and three transitions were monitored per peptide. Only the highest intensity and interference-free transition was used for quantification of the LLOQ and for subsequent assay development.
Raw data files were analyzed using Skyline-daily , setting the light version of the peptide as the internal standard for normalization. The regression fit was set to linear and the regression weighting was set to 1/x^2. The MS level was set to 2 and the limit of detection was set to blank plus 3 standard deviations. The chromatographic peak-area ratio of the heavy to light peptides was plotted against the fmol of SIS peptide (heavy) injected on column. An R-squared value of ≥0.9 was used as the cut-off for an assay to be considered linear. Peak areas were then exported to excel and the coefficient of variation (CV) was calculated at each SIS dilution level. The LLOQ was determined using the peak areas of the SIS peptides with or without normalization, which was dependent on the approach which produced the lower CV. The lowest dilution point displaying a CVof <20% was defined as the LLOQ.
Multi-day validation of assay analytical repeatability
For peptides for which an LLOQ was determined, based on the criteria defined above, a multi-day assessment of repeatability was performed, using CPTAC experiment 2  as a guide. Repeatability was assessed using the reverse-curve approach [3, 4], which involves varying the SIS concentration and using NAT and/or END as normalizers, or by using the SIS peptide without normalization, for targets for which the NAT peptide was not available or where the endogenous peptide was contributing to the variability. DBS matrix spiked with NAT peptides served as the blank datapoints which would later be used to assess the limit of detection, defined as three standard deviations higher than the signal measured across the blank samples (three replicates). Repeatability was determined by analyzing 0.2 fmol to 750 fmol of each SIS peptide injected on column. Blank and dilution samples were prepared in three technical replicates -- a technical replicate being defined as DBS matrix spiked with NAT and SIS peptides -- and each technical replicate was injected once for multiplexed MRM, targeting a single transitions per peptide (the same transition was targeted as was used for determining the LLOQ). These experiments were repeated on five experiment-days with at least 16 h between experiment-days. Targeted MRM and raw data analyses were performed as described in the preceding methods section, entitled “response curve”. SIS:NAT and/or SIS:END peak-area ratios or SIS areas without normalization were used for the calculation of CVs related to the analytical variance. Intra-day CVs were calculated using the three technical replicates at each concentration point on each day, and subsequently averaging across all days. The inter-assay CVs were calculated using the first injection on each day and across all days. This calculation was performed for all three injections, and the resulting CVs were averaged and used as the inter-assay CV. The total variation was calculated as the root sum of squares (RSS) of the intra and inter day assay CVs. This CV was a measure of the analytical variance of the MRM assays.
Determination of assay selectivity
Assay selectivity was assessed using the CPTAC experiment 3 guidelines  with some modification. DBSs obtained from the capillary blood of 6 subjects were processed as described above. A calibration curve for each MRM assay was made by spiking all of the samples with a constant amount of SIS peptide, which served as the normalizer. The lowest point on each calibration curve was the peak-area ratio of END:SIS, and three additional higher datapoints were obtained by spiking DBS samples with successively increasing amounts of NAT. The amount of the NAT-peptide spike was a function of the corresponding endogenous peptide abundance. The amount of NAT spiked into the sample (equivalent to fmol injected on column) was 1.2 fmol to 3340 fmol for the lowest spikes, 13.3 fmol to 36740 fmol for the middle spikes and 147 fmol to 404808 fmol for the highest spikes. The calibration curve in Skyline-Daily was adjusted to assign the sample without a NAT spike as the blank, and all other samples were assigned as standards. The slope of the line in each matrix was obtained using only two of the three points in addition to the blank, either the low and medium or the medium and high datapoints, selected based which gave a better (higher) R-squared value. The peak areas selected for determining the slope of the line was kept consistent across the six matrices.
The CPTAC 3 guidelines for analysis of the peak-area ratios were modified to include a bias correction in peak areas across the matrices, before calculating the slope. This bias correction was included to compensate for differences in the amount of peptide standard spiked into the samples, which would cause an increase in variance between the calculated slope of calibration curves when comparing between matrices. To measure the bias between matrices, the NAT:SIS peak-area ratios as measured in each matrix were exported to GraphPad Prism. One matrix was arbitrarily defined as the reference and the other five matrices were compared to the “reference matrix”. The bias across all peptides was calculated using Bland-Altman plots by plotting the NAT:SIS ratio vs the average ratio of the two matrices being compared. Peak-area ratios obtained for each concentration point were independently bias-corrected, and the bias-corrected peak-area ratios were then used to calculate the slope in Microsoft Excel using the slope function. The CPTAC experiment 3 guidelines for data analysis require that the slope of the line in an individual matrix be <10% of the average of the slopes across all six matrices. The limitation to this approach is that MRM assays which have a repeatability of >10% could produce a slope of the line, in an individual matrix, that varies more than the cut-off (of 10 %) and these assays could falsely be deemed non-selective. To avoid this possibility, assays were deemed to be “selective” if the slopes of the line calculated in each of the six matrices was within two standard deviations of the full process or the analytical assay variance (whichever was higher) of the averaged slope across all six matrices. The difference between the slope in a matrix from the average slope was calculated as a Z-score 〖|(slope〗_matrix-〖slope〗_ave)÷σ|, where σ is the full process or the analytical variance. A Z-Score of <2 was indicative of a difference in slope within two standard deviations of the mean across all slopes, and was therefore considered to be within the acceptable variance of the assay.
Stability assessment of SIS, NAT, and END peptides
Stability testing of the SIS peptides and the corresponding NAT and END peptides was performed in accordance with the CPTAC experiment 4 guidelines . The DBS matrix (prepared as described above) was spiked with NAT and SIS peptides so that the equivalent of 0.73 fmol (NAT) to 100 fmol (NAT), and 2 fmol (SIS) to 197 fmol (SIS), were injected on column. Aliquots were prepared in 12 sample injection vials, and three of the vials were placed in the autosampler, which was maintained at 5 ⁰C, while the remaining aliquots were stored at -80 ⁰C. Each of the three aliquots on the autosampler was injected in duplicate, giving a total of 6 MRM acquisitions which would later be combined to serve as the 0-h time point. Eight hours after the first injection of the first replicate, the same injection sequence was repeated, and the 6 MRM acquisitions were grouped and used as time point 8-16 h. Thirty five hours after the first injection of time point zero, the injection sequence, MRM acquisitions, and grouping was repeated for time point 35- 43 h. After injection of the 8-16 hour time point, six aliquots from the aliquots stored at -80 ⁰C were thawed at room temperature. Three aliquots were placed back at -80 ⁰C while the remaining three aliquots were used for MRM acquisitions as above. These samples were grouped and designated as a single freeze-thaw cycle. The three aliquots which had been placed back in the -80 ⁰C were thawed a second time and analyzed as above (designated as two freeze-thaw cycles). The three remaining aliquots were removed from -80 ⁰C four weeks later, and analyzed as above.
The guidelines for data analysis in CPTAC experiment 4 suggest that variability in peak-area ratios should not exceed the variability in peak-area ratios measured at the 0-h time point nor the variability assessed in CPTAC experiment 2. The peak-area ratios were obtained setting NAT+END as the normalizer in Skyline-Daily and all peak-area ratios were compared to time-point zero using the Z-score which was calculated as follows: 〖|(Peak area ratio〗_(t=0)-〖Peak area ratio〗_(subsequent t))÷σ|, where σ is the full process or the analytical variance. A Z-score of <2 was indicative of a difference in peak area ratio within two standard deviations of the peak-area ratio at time point zero, and was therefore within the variance of the assay.
Assessment of repeatable quantification of END peptide
Venous blood draws from one male and one female donor were pooled and used to generate DBSs, spotting all five spots on five Whatman 903 protein saver cards. One card was processed per day on five separate days, and each spot was initially processed alone (no pooling) for MRM analysis. The full experiment, from obtaining the venous blood draws through to the MRM analysis of five Whatman 903 protein saver cards over five experiment-days, was performed in two separate experimental blocks (a total of ten experiment-days), in accordance with guidelines outlined in CPTAC experiment 5 . Due to repeatedly-observed high variance for a subset of proteins (which was attributed to migration in the filter paper), a partial validation of repeatability was performed (in accord with Bioanalytical method validation guidance for industry) by pooling punched DBS spots as follows: the first spot was punched West, the second Centre, the third South, the fourth North and the fifth East and all discs were pooled for processing. A partial validation was performed by processing pooled disc punches from a single Whatman 903 protein saver card, processed in duplicate technical replicates to perform parallel sample processing, over a total of four experiment-days.
Data analysis for calculation of intra and inter-assay CVs was as follows; peak-area ratios were measured using Skyline-Daily, setting the SIS as the normalizer and END as the numerator. Peak-area ratios were obtained from technical replicates which were each used for triplicate MRM analysis (six total injections per experimental day). Peak-area ratios from triplicate injections were averaged and the intra-day CV was calculated using the average peak area ratio obtained for technical replicates on each day. CVs across all 4 days were averaged to obtain the average intra-assay CV. The inter-day CV was calculated using the standard deviation and average using the average peak area ratio from each technical replicates. The total variation was calculated as the RSS of the average intra and inter-day CVs.
Protein stability testing in DBS
Protein stability in DBSs was tested at three temperatures and up to 57 days of storage prior to processing. A pooled venous draw from 5 donors was used to spot all five spots of Whatman 903 protein saver cards using 50 µL of blood per spot and spotting a total of 46 cards (230 spots). After drying overnight, each card was individually placed in a heat sealable bag with desiccant and sealed. Bags were placed in cardboard freezer boxes and stored at either -20 ⁰C, ambient laboratory temperature (~25 ⁰C), or 40 ⁰C. After 24 h of storage, one card from each temperature was processed and analyzed using MRM. These samples served as time point zero and all subsequent peak-area ratios measured at future time points would be compared to this time point to assess whether a protein concentration had a measurable change after storage. The subsequent time points measured were Days 3, 4, 7, 13, 21, 28, 35, 42, and 57. Peak-area ratios (END:SIS), measured at each time point, were compared to time point zero using the Z-score which was calculated as follows: 〖|(Peak area ratio〗_(Day 1)-〖Peak area ratio〗_(Day n))÷σ|, where σ is the full process or the analytical variance. A Z-score < 2 was indicative of a difference in peak area ratio within two standard deviations of the peak area ratio at Day 1 and therefore within the variance of the assay.
Determination of correction factors
A correction factor for each DBS MRM assay was determined to convert the fmol of peptide measured on column from a DBS sample, to a concentration in whole blood. To determine these correction factors, MRM assays were performed to measure END:SIS peak-area ratios in whole-spot DBS, 6 mm disc punched DBS, whole blood, and lyophilized blood (in parallel). Pooled blood was used to generate DBSs by spotting either 50 µL or 13 µL of blood on Whatman 903 protein saver cards, each in triplicate. Whole blood in the liquid phase was processed, adding 13 µL directly to 237 µL of extraction buffer (2% DOC in 50 mM ammonium bicarbonate), in triplicate. In parallel, 13 µL of blood was lyophilized in triplicate. The 50-µL DBSs were used to punch out 6-mm discs using a pneumatic punch, while the 13-µL DBS spots were cut out whole. Each spot was processed without pooling by performing solid to liquid phase extraction in 250 µL of extraction buffer. Similarly the lyophilized blood samples were each processed using 250 µL of extraction buffer. Samples were then processed for MRM analysis as described in the previous sections. Data were analyzed in Skyline-Daily as described in the previous sections, and the END:SIS peak-area ratios were used to calculate the correction factors. The measured peak-area ratios in each liquid blood technical replicate was divided by the corresponding peak-area ratio in each 13 µL whole spot DBS technical replicate, and averaged across all three ratios to give a unique correction factor for each MRM assay. Using the identical approach MRM assay specific correction factors were calculated for the whole spot and 6 mm disc punched DBS samples. Lastly, to make the 6 mm disc punch a volumetric sample processing strategy, the whole spot to disc punch correction factor was divided by the whole spot to blood correction factor. Using the resulting correction factors offered the option of reporting protein concentration per volume of blood even though the concentration measurement was performed using a 6 mm disc punch from a DBS sample. The advantage of this is that volumetric spotting on Whatman 903 filter papers would then not be needed -- instead, blood could be sampled directly from a capillary puncture. The identical data analysis was performed using lyophilized blood to determine whether the drying process was the main contributing factor to the higher concentration measurements that was observed in DBS relative to whole liquid blood samples.
Calibration curves for measuring peptide concentration
The concentration of endogenous peptide was measured using the linear regression equation for the line obtained by generating calibration curves in a surrogate matrix background. The surrogate matrix has been described in the methods section entitled pre-analytical sample preparation. The regression equation was generated by plotting NAT:SIS ratio on the y-axis and the NAT concentration on the x-axis. Wherever possible, the SIS peptide concentration was adjusted to fall within a factor of 10 of the concentration of its endogenous counterpart. The NAT peptide was used to make a dilution series to give a seven point calibration curve with a three-orders-of-magnitude dynamic range, covering the endogenous peptide concentration. The calibration curves were generated using Skyline-Daily by setting the regression fit and weighting to linear and 1/x*x, respectively. Using the equation of the line and the END:SIS peak-area ratio, the fmol of END injected on column can be calculated for an unknown sample. Finally, the concentration of the peptide in the original blood sample can be calculated using the correction factor for each analyte, as described in the method section entitled determination of correction factors.
Details of the donor cohort and collection of capillary blood from finger sticks are described in the methods sections on collection of human whole blood and pre-analytical sample preparation. DBS samples were processed within 13 days after collection (of capillary blood as DBS) as described in the methods section on pre-analytical sample preparation. MRM analysis was performed on each sample and the measured protein concentration was the average of three MRM analyses.
1. Ostler, M.W., J.H. Porter, and O.M. Buxton, Dried Blood Spot Collection of Health Biomarkers to Maximize Participation in Population Studies. Journal of Visualized Experiments, 2014(83).
2. Eshghi, A. and C.H. Borchers, Multiple Reaction Monitoring Using Double Isotopologue Peptide Standards for Protein Quantification. Methods Mol Biol, 2018. 1788: p. 193-214.
3. Michaud, S.A., et al., Molecular phenotyping of laboratory mouse strains using 500 multiple reaction monitoring mass spectrometry plasma assays. Communications Biology, 2018. 1: p. 78.
4. Percy, A.J., et al., Multiplexed MRM-based assays for the quantitation of proteins in mouse plasma and heart tissue. Proteomics, 2017. 17(7).
5. UniProt: the universal protein knowledgebase. Nucleic Acids Research, 2017. 45(D1): p. D158-D169.
6. Mohammed, Y., et al., PeptideTracker: A knowledge base for collecting and storing information on protein concentrations in biological tissues. Proteomics, 2017. 17(7).
7. Kuzyk, M.A., et al., Multiple reaction monitoring-based, multiplexed, absolute quantitation of 45 proteins in human plasma. Molecular and Cellular Proteomics, 2009. 8(8): p. 1860-77.
8. Kuzyk, M.A., et al., Development of MRM-based assays for the absolute quantitation of plasma proteins. Methods in Molecular Biology, 2013. 1023: p. 53-82.
9. Han, J., et al., Short-Term Stabilities of 21 Amino Acids in Dried Blood Spots. Clinical Chemistry, 2018. 64(2): p. 400-402.
10. MacLean, B., et al., Skyline: an open source document editor for creating and analyzing targeted proteomics experiments. Bioinformatics, 2010. 26(7): p. 966-8.
11. Yang, X., P. Neta, and S.E. Stein, Extending a Tandem Mass Spectral Library to Include MS2 Spectra of Fragment Ions Produced In-Source and MSn Spectra. Journal of The American Society for Mass Spectrometry, 2017. 28(11): p. 2280-2287.
12. MacLean, B., et al., Effect of Collision Energy Optimization on the Measurement of Peptides by Selected Reaction Monitoring (SRM) Mass Spectrometry. Anal. Chem., 2010. 82(24): p. 10116–10124.
13. Sherwood, C.A., et al., Correlation between y-type ions observed in ion trap and triple quadrupole mass spectrometers. J Proteome Res, 2009. 8(9): p. 4243-51.
14. Whiteaker, J.R., et al., CPTAC Assay Portal: a repository of targeted proteomic assays. Nature Methods, 2014. 11(7): p. 703-4.
15. U.S. Department of Health and Human Services Food and Drug Administration Center for Drug Evaluation and Research (CDER) Center for Veterinary Medicine (CVM). Bioanalytical Method Validation Guidance for Industry. 2018, Office of Communications, Division of Drug InformationCenter for Drug Evaluation and ResearchFood and Drug Administration: Silver Spring, MD.