Comparison of Internal Standard Approaches for ... - ACS Publications

Subscriber access provided by University of Florida | Smathers Libraries

Article

Comparison of internal standard approaches for SRM analysis of alpha-synuclein in cerebrospinal fluid Patrick Oeckl, Petra Steinacker, and Markus Otto J. Proteome Res., Just Accepted Manuscript • DOI: 10.1021/acs.jproteome.7b00660 • Publication Date (Web): 29 Nov 2017 Downloaded from http://pubs.acs.org on December 6, 2017

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Journal of Proteome Research is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Proteome Research

Comparison of internal standard approaches for SRM analysis of alphasynuclein in cerebrospinal fluid Patrick Oeckl1, Petra Steinacker1 and Markus Otto1* 1

Department of Neurology, Ulm University Hospital, Ulm, Germany

KEYWORDS: SRM, MRM, absolute quantification, alpha-synuclein, stable-isotope dilution, SIL, PSAQ, winged peptide, QPrEST, cerebrospinal fluid, biomarker

ACS Paragon Plus Environment

1


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 30

ABSTRACT

Absolute protein quantification by selected reaction monitoring (SRM, also MRM) is an alternative to immunoassays and the gold standard here is the addition of stable-isotope labeled (SIL) proteins (PSAQ). Cerebrospinal fluid (CSF) is the preferred source of biomarkers for neurological diseases and the recent improvements in mass spectrometry enable the quantification of disease-relevant proteins in CSF. We used an alpha-synuclein SRM to investigate alternatives to the PSAQ approach in human CSF regarding precision and accuracy including SIL peptides, winged SIL (WiSIL) peptides and quantitative protein epitope signature tags (QPrESTs).

All approaches yielded precise results in CSF with CV values 95%) was purchased from AJ Roboscreen GmbH (Leipzig, Germany) and used for external calibration. Exact protein concentration was determined by amino acid analysis (Alphalyse A/S, Odense, Denmark). 15N-labeled αSyn (purity >95%) was from rPeptide (Bogart, GA) and synthetic peptides (SIL and WiSIL) were purchased from Thermo Fisher Scientific. Protein and peptide sequences, position of labeled amino acids and isotope incooperation are given in Fig. 2A. The QPrEST for αSyn (#QPrEST24323) was kindly provided by Atlas Antibodies AB (Sweden). Trypsin/LysC mix was purchased from Promega GmbH, human serum albumin (HSA, #A3782) from Sigma and artificial CSF (aCSF) from EcoCyte Bioscience (Austin, TX).


7


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 30

Figure 2. Protein and peptide sequences and experimental workflow. (A) Amino acid sequence of full-length human αSyn (UniProt ID: P37840) which is identical in the recombinant αSyn and 15N-αSyn used for the PSAQ approach. 15N-αSyn was expressed in E. coli with 15N as the nitrogen source resulting in complete labeling of all nitrogens with 15N (rPeptide, Bogart, GA, #S-1004-1). Sequences of SIL and WiSIL peptides and the QPrEST are indicated and stable-isotope labeled amino acids are highlighted in bold and underlined. SIL peptides were labeled with 13C6,15N2-K or 13C3,15N1-A, WiSIL peptides and the QPrEST were labeled with


8

Page 9 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


13

C6,15N2-K. Level of isotope incooperation was >99% for all.. Colored peptides are proteotypic

peptides and are quantified by LC-SRM. Representative chromatograms of the CSF pool sample are shown. (B) Overview of the experimental design. Samples were prepared in 96-deep-well plates and one of the four internal standards (IS) was added depending on the run. Fractionation was also performed in the 96-well format allowing full automation of the sample preparation procedure. The samples in each run included two technical replicates of the calibration standards and five technical replicates of the CSF samples. CV values and accuracy were calculated for the CSF samples. All runs were performed on five different days as indicated. To ensure that interday variation does not influence the comparison of the IS approaches the PSAQ approach was included on all days why the number of PSAQ runs is higher.

Calibration standards for external calibration The proteins and peptides were dissolved in LC-MS grade water at concentrations of 5-350µM and stored at -80°C. Calibration standards were prepared freshly in a surrogate matrix (aCSF+200µg/mL HSA) using recombinant full-length αSyn at concentrations of 5, 10, 20, 50, 70, 100 and 200pM. Human CSF sample pool CSF samples were derived from patients with hydrocephalus by lumbar puncture, centrifuged and stored within 2h at -80°C. The samples were pooled and aliquoted so the same sample was analyzed for all comparisons. All patients or their relatives gave written informed consent. The study was approved by the Ethics Committee of Ulm University. Total protein concentration of


9


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 30

the CSF pool was determined using the Pierce BCA Protein Assay Kit (Thermo Fisher Scientific). Sample preparation Calibration standards, blanks (aCSF+200µg/mL HSA) and CSF samples (200µl each) were pipetted into a protein low-binding 96-deep-well plate (Eppendorf, Hamburg, Germany). Fourty microliters of an IS solution (containing 15N-αSyn, SIL peptides, WiSIL peptides or QPrEST in 0.5 M TEAB) and 12µL of Trypsin/LysC (0.1µg/µL in 100mM TEAB) were added (Fig. 2B). The samples were digested for 16h at 27°C and fractionated using SCX STAGE Tips as previously described14. After vacuum drying, fractions were resolubilized using 27µL of 0.5% TFA/6% acetonitrile (ACN) (fraction 1), 0.1% TFA/6% ACN (fraction 2) and 0.1% TFA/4% ACN (fraction 3) and stored in the autosampler at 4°C until LC-SRM analysis. Fig. 2C shows the set-up of the runs and each run included a single IS type only. Three independent runs were performed for each IS approach. The use of a 96-deep-well-plate allowed the simultaneous preparation of samples from three runs with different IS approaches. Because the PSAQ approach was always included as a reference, data from five runs were available for PSAQ (five plates necessary to analyze three runs per IS approach). LC-SRM analysis of αSyn LC-SRM analysis was performed as described previously14. In brief, 20µL of the fractions were loaded on a 0.3x5mm, C18 PepMap100 trap column (Thermo Fisher Scientific) with a mobile phase of 0.05% TFA/1% methanol and a flow rate of 200µL/min. Peptides were separated on a Eksigent HALO Fused-core C18, 2.7µm, 0.5x100 mm column (40°C, 15µL/min) with mobile


10

Page 11 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


phase A: 0.1% formic acid in 4% DMSO and mobile phase B: 0.1% formic acid, 4% DMSO and 96% ACN with a linear gradient over 9.75min from 5% (1% for fraction 1) to 30% B. Samples were analyzed with a QTRAP6500 mass spectrometer (AB Sciex, Framingham, MA) in positive ion mode with a source temperature of 175°C and curtain gas of 40 psi. MS settings and the measured transitions are listed in table 1.


11


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48

Page 12 of 30

Table 1. Transitions and MS settings of αSyn peptides Sequence (position)

Precursor ion mass

z

Product ion mass

Voltage

GS1

GS2

DP

CE

CXP

Fraction

(min)

Peptide

QGVAEAAGK (24-32)

2+

4900

20

30

61

Unlabeled

415.7

546.3 (y6), 346.2 (y4)

21, 21

29, 21

15

421.2

553.3 (y6), 352.2 (y4)

21, 21

29, 21

SIL, WiSIL, QPrEST

419.7

554.3 (y6), 354.2 (y4)

21, 21

29, 21

N-αSyn

2+

EGVVHGVATVAEK (46-58)

5100

20

30

100

Unlabeled

648.4

774.4 (y8), 911.5 (y9)

35, 34

41, 50

15

656.3

783.4 (y8), 923.5 (y9)

35, 34

41, 50

SIL, WiSIL, QPrEST

652.4

782.5 (y8), 919.5 (y9)

35, 34

41, 50

N-αSyn

EQVTNVGGAVVTGVTAVAQK (61-80)

2+

5500

35

30

120

Unlabeled

964.5

973.6 (y10), 874.5 (y9), 1072.6 (y11)

44, 44, 44

50, 50, 53

15

976.5

985.5 (y10), 885.5 (y9), 1085.6 (y11)

44, 44, 44

50, 50, 53

SIL, WiSIL, QPrEST

968.5

981.6 (y10), 882.5 (y9), 1080.7 (y11)

44, 44, 44

50, 50, 53

N-αSyn

2+

TVEGAGSIAAATGFVK (81-96)

5500

40

30

125

Unlabeled

739.9

764.4 (y8), 1021.6 (y11), 693.4 (y7)

34, 34, 34

40, 38, 55

15

748.4

773.4 (y8), 1033.5 (y11), 701.4 (y7)

34, 34, 34

40, 38, 55

SIL

745.9

776.5 (y8), 1033.6 (y11), 701.4 (y7)

34, 34, 34

40, 38, 55

WiSIL, QPrEST

743.9

772.4 (y8), 1029.6 (y11), 701.4 (y7)

34, 34, 34

40, 38, 55

N-αSyn

RT

1

3.7

3

5.4

1

8.9

2

8.1

CE, collision energy; CXP, cell exit potential; DP, declustering potential; GS1, gas 1 (sheath gas); GS2, gas 2 (heater gas); RT, retention time.


12

Page 13 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


The correct transition pattern of each peptide was checked using the Skyline 3.7 software21. Analyst software 1.6.2 (AB Sciex) was used for peptide quantification. The peak area of all transitions per peptide were summed up and αSyn peptides were quantified using the peak area ratio and an external calibration curve, calculated using the calibration standards and a weighting of 1/x2. Statistics Statistical analysis was performed using Analyst 1.6.2 software and GraphPad Prism 5.0. CVs and accuracy were calculated within each run. A CV and deviation ≤15% was regarded acceptable based on recommendation from EMA and FDA.

RESULTS AND DISCUSSION We used the protein αSyn to compare different IS approaches for quantitative SRM analysis in CSF in terms of their accuracy and precision, i.e. PSAQ, SIL, WiSIL and QPrESTs. A pooled CSF sample was analyzed in several independent runs at different days (five replicates in each run) using the four IS approaches (Fig. 2C) and endogenous concentrations of four proteotypic αSyn peptides (two peptides for QPrEST) were quantified (Fig. 2A) using a previously established and validated SRM method14. Precision (%CV) and mean accuracy (% deviation) of the five replicates in each run is shown in Fig. 3. The mean CV of all runs for the PSAQ approach was in the range of 3.1-6.8% for the different αSyn peptides whereas the peptide with the lowest signal intensity (αSyn61-80) showed the highest CV (Fig. 3A).


13


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 30

Figure 3. Precision and accuracy of internal standard approaches for αSyn in CSF. (A) CV and (B) accuracy of endogenous αSyn peptide concentrations in CSF using the different internal standard approaches. Three independent runs (on different days) were performed for each approach (five for PSAQ). Each run consisted of five CSF samples and the dots are the CV (A) and mean accuracy (B) of these five samples. Bars are mean values of all runs. The grey area indicates a range of ±15%.


14

Page 15 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


The other three IS approaches showed higher CV values compared with PSAQ (SIL 4.9-8.6%, WiSIL 3.5-8.7%, QPrEST 9.5-10.9%) depending on the αSyn peptide measured but all approaches are within the acceptable range of ≤15%. Notably, there was no difference between SIL and WiSIL peptides. Our observation in CSF is in agreement with studies in other biological matrices such as plasma or tissue samples which also observed good precision of SIL and WiSIL peptides9,22,23. This is unexpected for the SIL approach because it cannot reflect the digestion process but our and previous data9,24 indicate that when performed under controlled conditions, the variability of digestion and pre-digestion steps is low and SIL peptides yield acceptable precision. This is the first study using QPrESTs for protein quantification in CSF and we observed good precision for the QPrEST approach as well (Fig. 3A). Thus, QPrESTs are a reliable alternative to PSAQ, SIL and WiSIL peptides in terms of precision. For accuracy determination the mean concentration of the first four PSAQ runs was used as the reference value (“true value”) for all αSyn peptides. Based on these values, the total αSyn concentration in the CSF pool calculated as the mean concentration of the four peptides was 18.6pM (equal to 269pg/mL) which is in the normal range of patients without neurodegenerative diseases14. The values of the PSAQ runs varied within the acceptable ±15% (Fig. 3B) but both the SIL and WiSIL approach showed higher deviation from the PSAQ reference value of up to +54% although it greatly depended on the αSyn peptide measured (mean accuracy SIL 103137%, WiSIL 111-130%). A higher deviation of values using SIL peptides has been shown previously in blood9,25 and other matrices5,22 but greatly depends on the peptide. The digestion efficiency for different peptides within a protein can vary considerably26,27 resulting in concentration differences of up to 100x between peptides when using the SIL approach28. In addition, peptide decay during and after digestion can significantly influence accuracy of the SIL


15


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 16 of 30

approach. This is because conditions for SIL peptides and peptides generated from the endogenous protein are different depending on the time point when SIL peptides are added and on peptide-release rate during digestion29. The peptide αSyn61-80 showed the strongest deviation using SIL (Fig. 3B). The peptide contains an N-terminal Glu residue which is prone to pyroGlu formation30. In the intact αSyn protein, this residue is protected from pyroGlu formation until digestion but not in the SIL peptide which is added to the sample at the beginning of sample preparation. This could explain the higher calculated concentrations based on the increased peak area ratios which are calculated with the unmodified SIL peptide. The higher accuracy of WiSIL for this αSyn peptide (Fig. 3B) where the Glu-residue is also protected until digestion is in support of this hypothesis. Interestingly, the peptide αSyn46-58 also contains an N-terminal Glu residue but is less affected which may results from the different peptide sequence or fast release during digestion. On the other hand, our data in CSF indicate that for some peptides the SIL approach can yield accurate results which is expected for peptides with 100% digestion efficiency and has been shown before24,31. This highlights the importance of a conscious peptide selection and thorough method validation when using the SIL approach. Higher deviations with the SIL approach where expected but one of the main aims of our study was to investigate whether WiSIL peptides can improve accuracy in CSF SRM analyses due to the addition of the natural tryptic cleavage sites. As indicated in Fig. 3B, the WiSIL approach did not perform better than the SIL approach in our study with CSF samples although other peptides were characterized by a high deviation. Several reasons are possible: Although WiSIL peptides contain a tryptic cleavage site the conformational effects of a protein on digestion are not


16

Page 17 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


mimicked. Barnidge and colleagues32 showed diverging digestion kinetics of WiSIL peptides and the intact protein with a very fast release of the signature peptides from the WiSIL precursor compared with the intact protein. Variations in digestion conditions might therefore differently affect the cleavage of WiSIL peptides and the intact protein. Scott et al. investigated how the number of amino acids used for the extension in WiSIL peptides influences the digestion27. The number of flanking amino acids in WiSIL peptides varies between studies22,23,32–34 and an improved performance of peptides with longer extensions (6 amino acids) has been observed for WiSIL peptides27 and QConCATs35,36 especially if the sequence contains amino acids which are known to affect trypsin cleavage such as Asp and Glu C- and N-terminal to Lys/Arg36. The WiSIL peptides in our study were extended by two amino acids at the N- and C-terminus (Fig. 2A) which could be a reason for the lack of improvement compared with the SIL approach. It is unlikely that differences in sample preparation are responsible for the deviation of the SIL and WiSIL approaches from the reference value since it would affect all peptides similarly but the deviation is different between peptides with some showing good accuracy. Since the IS peptides/proteins were added at the first step of sample preparation, differences during sample fractionation can also be ruled out. In addition to PSAQ, SIL and WiSIL we tested a new approach for absolute quantification of proteins in CSF. QPrESTs were originally designed for the generation of an antibody library in the Human Protein Atlas project and are protein fragments of 50-150 amino acids length12. Thus, they do not only contain the original cleavage sites but due to their size they might also reflect some secondary and tertiary structural features of the intact protein. The available QPrEST for αSyn (QPrEST24323, Atlas Antibodies AB) contained three of the αSyn peptides measure here (αSyn46-58, αSyn61-80 and αSyn81-96, Fig. 2) but αSyn46-58 was not detectable. Since this


17


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 18 of 30

peptide represents the N-terminus of the QPrEST (Fig. 2A) and contains an N-terminal Glu residue we assume a high degree of N-terminal pyroGlu formation as the reason for the missing (unmodified) peptide although we could not prove this experimentally by LC-MS/MS analysis (data not shown). Indeed, we observed good accuracy for the QPrEST peptides within the acceptable range of ±15% deviation (mean accuracy 98-115%) and accuracy was higher compared with the SIL and WiSIL approach. This indicates that the QPrEST at least in part reflects structural feature of the protein better compared with SIL or WiSIL peptides and shows that QPrESTs can be accurate alternatives for PSAQ. The inter-individual variability of CSF composition can considerably affect sample preparation in SRM analyses. The total protein concentration which mainly depends on the blood-CSFbarrier function is one important factor which could significantly influence the digestion process especially in our method where we use a fixed amount of trypsin/LysC for all samples. Our CSF sample pool used for the comparative study had a total protein concentration of 340µg/mL which is in the normal range of

Comparison of Internal Standard Approaches for ... - ACS Publications

Recommend Documents