Scholarly article on topic 'Introducing plasma/serum glycodepletion for the targeted proteomics analysis of cytolysis biomarkers'

Introducing plasma/serum glycodepletion for the targeted proteomics analysis of cytolysis biomarkers Academic research paper on "Chemical sciences"

Share paper
Academic journal
OECD Field of science
{Plasma / Protein / Biomarker / Cytolysis / Glycodepletion / Proteomics}

Abstract of research paper on Chemical sciences, author of scientific article — Pauline Maes, Sandrine Donadio-Andréi, Mathilde Louwagie, Yohann Couté, Guillaume Picard, et al.

Abstract A major class of clinical biomarkers is constituted of intracellular proteins which are leaking into the blood following ischemia, exposure to toxic xenobiotics or mechanical aggression. Their ectopic presence in plasma/serum is an indicator of tissue damage and raises a warning signal. These proteins, referred to as cytolysis biomarkers, are generally of cytoplasmic origin and as such, are devoid of glycosylation. In contrast, most plasma/serum proteins originate from the hepatic secretory pathway and are heavily glycosylated (at the exception of albumin). Recent advances in targeted proteomics have supported the parallelized evaluation of new blood biomarkers. However, these analytical methods must be combined with prefractionation strategies that reduce the complexity of plasma/serum matrix. In this article, we present the glycodepletion method, which reverses the hydrazide-based glycocapture concept to remove plasma/serum glycoproteins from plasma/serum matrix and facilitates the detection of cytolysis biomarkers. Glycodepletion was integrated to a targeted proteomics pipeline to evaluate 4 liver cytolysis biomarker candidates in the context of acetaminophen-induced acute hepatitis.

Academic research paper on topic "Introducing plasma/serum glycodepletion for the targeted proteomics analysis of cytolysis biomarkers"


Contents lists available at ScienceDirect


journal homepage:


Introducing plasma/serum glycodepletion for the targeted proteomics analysis of cytolysis biomarkers

Pauline Maesa,b,c, Sandrine Donadio-Andréia,b,c, Mathilde Louwagiea,b,c, Yohann Coutéa,b,c, Guillaume Picarda,b,c, Claire Lacosted,e, Christophe Bruleya,b,c, Jérôme Garina,b,c, Philippe Ichaid,f, Jamila Faivred,e,f, Michel Jaquinoda,b,c, Virginie Bruna,b,c,i

a Université Grenoble-Alpes, F-38000 Grenoble, France b CEA, BIG, Biologie à Grande Echelle, F-38054 Grenoble, France c INSERM, U1038, F-38054 Grenoble, France d INSERM, U1193, Centre Hépatobiliaire, Villejuif F-94800, France e Université Paris Saclay, Faculté de Médecine, Villejuif F-94800, France

f Assistance Publique-Hôpitaux de Paris (APHP), Hôpital Universitaire, Paul Brousse, Villejuif F-94800, France










A major class of clinical biomarkers is constituted of intracellular proteins which are leaking into the blood following ischemia, exposure to toxic xenobiotics or mechanical aggression. Their ectopic presence in plasma/ serum is an indicator of tissue damage and raises a warning signal. These proteins, referred to as cytolysis biomarkers, are generally of cytoplasmic origin and as such, are devoid of glycosylation. In contrast, most plasma/serum proteins originate from the hepatic secretory pathway and are heavily glycosylated (at the exception of albumin). Recent advances in targeted proteomics have supported the parallelized evaluation of new blood biomarkers. However, these analytical methods must be combined with prefractionation strategies that reduce the complexity of plasma/serum matrix. In this article, we present the glycodepletion method, which reverses the hydrazide-based glycocapture concept to remove plasma/serum glycoproteins from plasma/serum matrix and facilitates the detection of cytolysis biomarkers. Glycodepletion was integrated to a targeted proteomics pipeline to evaluate 4 liver cytolysis biomarker candidates in the context of acetaminophen-induced acute hepatitis.

1. Introduction

Biomarkers are biological parameters which are measured to

characterize a physiological or pathological state, the evolution of a

disease, or a response to treatment. Several types of biomarkers are

routinely exploited in clinical laboratories: small molecules (vitamins, hormones, etc.), nucleic acids and proteins. In the category of protein biomarkers, a study performed by Anderson in 2010 [1] listed over 200

protein analytes used in clinical biology, 109 of which have been

cleared or approved by the American Food and Drug Administration (FDA). These biomarkers can be divided between the following classes:

immunoglobulins, ligands, proteins with a function in plasma, abnor-

mal secretions, and cytolysis biomarkers. Cytolysis biomarkers are

defined as intracellular proteins which are released into the extracellular medium following cellular necrosis and an inflammation,

which increases vascular permeability. The presence of these proteins in the blood is indicative of tissue damage due to ischemia, exposure to xenobiotics or mechanical trauma [2].

The vast majority of protein biomarkers are assayed in plasma or serum samples. In biochemical terms, these two biological matrices are immensely complex and have a very wide dynamic range. Thus, plasma (and serum) contains more than 2000 proteins, but the 20 major proteins represent over 99% of the total protein mass [3]. Among these major proteins, albumin (50mg/mL) and immunoglobulins (17 mg/ mL) alone represent 90% of the total proteins. Because these proteins are so highly represented, the plasma concentrations of all the proteins in plasma/serum varies by over ten orders of magnitude (1010) [4]. This extremely broad dynamic range is a real challenge when analyzing biomarkers, as these proteins are generally present in medium- or low-abundance in the matrix.

Abbreviations: ADH4, alcohol dehydrogenase 4; ALT1, alanine aminotransferase 1; BHMT, betaine-homocysteine S-methyltransferase; FABP1, fatty acid binding protein 1; PRM, Parallel Reaction Monitoring; SRM, Selected Reaction Monitoring; PSAQ, Protein Standard Absolute Quantification

* Correspondence to: Unité de Biologie à Grande Echelle, CEA/DRF/BIG/INSERM/UGA 1038, 17 Avenue des Martyrs, 38054 Grenoble cedex 9, France. E-mail address: (V. Brun).

Received 13 January 2017; Received in revised form 7 April 2017; Accepted 16 April 2017

Available online 19 April 2017

0039-9140/ © 2017 Published by Elsevier B.V.

In routine clinical applications, protein biomarkers are generally assayed by immunological methods based on antigen/antibody interactions or enzymatic tests. These assays have several advantages, including their sensitivity, simplicity of use, rapidity, low cost, and they are easily automated. However, during the process of developing new biomarkers, multiple candidates must be assessed, this can be done most effectively using liquid chromatography (LC) analysis and mass spectrometry in targeted mode, by Selected Reaction Monitoring (SRM) [5] or Parallel Reaction Monitoring (PRM) [6]. These analytical methods have an unparalleled multiplexing capacity, and thus allow several tens of target proteins to be simultaneously assayed in clinical samples without systematically requiring the development of antibodies. In addition, because peptides produced by digestion of target proteins can be selectively analyzed, these approaches are highly specific and can distinguish between protein isoforms [7,8]. However, the concentrations of most biomarker candidates in plasma/serum (from ^g/mL to pg/mL) are beyond the dynamic range of these analysis methods (105). Therefore, these approaches must be combined with upstream biochemical steps to prepare samples for analysis [9]. This pre-analytical step aims to accomplish the following two objectives: (i) to make it possible to detect candidate biomarkers by enriching them or by reducing the dynamic range and complexity of the matrix and (ii) to digest the proteins into peptides, some of which (signature peptides) will be analyzed as "surrogate proteins".

Various biochemical strategies are currently employed to reduce serum/plasma protein complexity and dynamic range [9-11]. Among these strategies, a very popular and straightforward approach consists in depleting highly abundant proteins using immunoaffinity devices (resin, spin or HPLC columns) [12]. This method significantly reduces the dynamic range of protein concentrations, thus improving detection of medium- and low-abundance proteins. However, this approach only eliminates a limited number of proteins from plasma or serum (in general between 1 and 20 proteins) and it is expensive when processing numerous sample or volumes exceeding 20 ^L [9]. Another limitation of this method is that proteins which have formed complexes with the captured abundant proteins will also be eliminated [13]. This scenario becomes problematic if a candidate biomarker interacts with the major proteins in serum or plasma, as its bound fraction will be retained on the depletion device resulting in reduced amounts available for MS analysis.

In this study, we have developed a novel method for the preparation of serum and plasma samples. This method is specifically adapted for proteomics analysis of cytolysis biomarkers (candidate or validated). The protocol is based on the depletion of plasma or serum glycopro-teins (glycodepletion) to facilitate access to non-glycosylated proteins such as cytolysis biomarkers. Following optimization of the plasma glycodepletion procedure, a proteomics analysis pipeline involving LC-SRM analysis was optimized and applied for the assessment of three novel candidate biomarkers of liver cytolysis.

2. Material and methods

2.1. Biological samples

Plasma and serum glycodepletion was optimized using human plasma and human serum purchased from Sigma Aldrich (Saint Quentin Fallavier, France). Serum samples from patients with acet-aminophen-induced acute hepatitis were collected at the Hepatology Department at Paul-Brousse Hospital (Villejuif, France) within the first 48 h after hospital admission. Samples were supplied by the Biological Resource Centre of the Paris-Sud Faculty of Medicine, France (approval number: 2011/39938). Research using blood samples was approved by the institutional review board of Paul-Brousse Hospital (Villejuif, France). All the patients provided written informed consent for participation. The French Blood Service (Etablissement Français du Sang, La Tronche, France) provided anonymous serum samples from

healthy donors. Serum samples were collected in non-treated tubes (BD Biosciences, Le Pont de Claix, France) and were centrifuged at 1000g for 15 min to obtain serum supernatant. Serum samples were aliquoted and frozen at -80 °C for proteomics analysis.

2.2. Recombinant proteins

Recombinant ADH4, ALT1, BHMT and FABP1 proteins were obtained from Abcam (references ab132569, ab113862, ab51280 and ab82994 respectively). Isotopically-labeled recombinant protein analogs (PSAQ standards) for these four proteins were synthesized using cell-free expression (RTS 500 Proteomaster E. coli HY kit, 5 Prime, Hamburg, Germany) in the presence of [13C6, 15N2] L-lysine and [13C6, 15N4] L-arginine (Eurisotop, Saint-Aubin, France) as previously described [14]. Production was scaled-up at Promise Advanced Proteomics (Grenoble, France). PSAQ standards were checked for isotope incorporation ( > 99%) and were quantified by amino acid analysis [15].

2.3. Glycodepletion

Plasma or serum samples (18 ^L) were prepared using the GlycoLink Immobilization Kit (Life Technologies, Villebon-sur-Yvette, France) according to the manufacturer's instructions. Plasma/serum glycoproteins were oxidized by 0.1 M meta-periodate. Then, samples were loaded on spin columns filled with hydrazide-activated resin. Unbound proteins were eluted by centrifugation using 400 ^L of urea 8 M. Eluates were concentrated to 50 ^L using 3000 Da cut-off ultrafiltration devices before digestion.

2.4. Depletion of abundant proteins

Plasma or serum samples (14 ^L) were depleted of the six most abundant proteins using the Affinity Removal System (MARS) Hu-6 spin cartridge (Agilent Technologies, Les Ulis, France) according to the manufacturer's instructions. Depleted samples were concentrated to 50 ^L and buffer was exchanged using a 3000 Da cut-off ultrafiltration device (Merck Millipore, Molsheim, France) in 4 M urea and 50 mM NH4HCO3. The resulting concentrates were then submitted to digestion.

2.5. Protein digestion

In-solution digestion was performed using an endoLysC/Trypsin mix (Promega, Charbonnières les Bains, France) at a protein/enzyme ratio of 1:30 (w/w) in 4 M urea and 50 mM NH4HCO3 at 37 °C. After 3 h of incubation, samples were diluted (4X) and incubated overnight at 37 °C. Digestion was stopped by adding formic acid (0.1% final concentration). Samples were purified on C18 Macro SpinColumns (Harvard Apparatus, Les Ulis, France) and dried by vacuum centrifu-gation. Digests were resolubilized in 15 ^L of 2% acetonitrile, 0.1% formic acid.

2.6. LC-MS/MS analysis

Peptides were analyzed by nanoliquid chromatography coupled to tandem mass spectrometry (Ultimate 3000 coupled to LTQ-Orbitrap Velos Pro, Thermo Scientific). Peptides were sampled on a 300 ^mx5 mm PepMap C18 precolumn and separated on a 75 ^mx250 mm C18 column (3 ^m beads, PepMap, Thermo Scientific) using a 120 min gradient ranging from 5% to 37% acetonitrile in 0.1% formic acid during 114 min before reaching 72% acetonitrile in 0.1% formic acid for the last 6 min. Spray voltage and heated capillary were set at 1.4 kV and 200 °C, respectively. MS and MS/MS data were acquired using Xcalibur (Thermo Scientific). Survey full-scan MS spectra (m/z=400-1600) were acquired in the Orbitrap

with a resolution of 60,000 after accumulation of 106 ions (maximum filling time, 500 ms). The 20 most intense ions from the preview survey scan delivered by the Orbitrap were fragmented by collision-induced dissociation (collision energy, 35%) in the LTQ after accumulation of 104 ions (maximum filling time, 100 ms). Peptides and proteins were identified through concomitant searches against Uniprot (Homo sapiens taxonomy, November 2016 version) and classical contaminants database (126 sequences, homemade) and the corresponding reversed databases using Mascot (version 2.5.1). The Proline software (http:// was used to filter the results (conservation of rank 1 peptides, peptide identification FDR< 1% as calculated on peptide scores by employing the reverse database strategy, minimum peptide score of 25, and minimum of 1 specific peptide per identified protein group) before performing a compilation, grouping and comparison of the proteins from the 2 fractions. Proteins from the contaminants database and additional keratins were discarded from the final list of identified proteins. Only proteins identified with a minimum of 2 specific spectral counts in 1 fraction were further considered.

2.7. LC-SRM analysis

LC-SRM analyses were performed on a 6500 QTrap mass spectrometer (AB Sciex, Les Ulis, France) 400-1000 m/z range) equipped with a TurboV source and controlled by Analyst software (version 1.6.1, AB Sciex). The instrument was coupled to an Ultimate 3000 LC-chromatography system (Thermo Scientific, Courtaboeuf, France). Chromatography was performed using a two-solvent system combining solvent A (2% acetonitrile, 0.1% formic acid) and solvent B (80% acetonitrile, 0.1% formic acid). Peptide digests were concentrated on a 1x15 mm C18 PepMap precolumn (Thermo Scientific) before separation on a Kinetex XB-C18 column (2.1x100 mm, 1.7 pm, 100 Ä; Phenomenex, Le Pecq, France). A 40 min gradient from 3% B to 35% in 30 min and from 35% to 90% in 10 min was applied at a flow rate of 50 pL/min. MS data were acquired in positive mode with an ion spray voltage of 4300 V and the interface heater temperature was set to 320 °C. Collision exit, declustering and entrance potentials were set to 18, 14 and 55 V, respectively. For each signature peptide, collision energy was calculated based on the following equations: CE (Volts) =0.44*m/z+4 for doubly-charged precursors and CE (Volts)=0.5*m/z +5 for triply-charged precursors. Scheduled SRM acquisitions were performed with Q1 and Q3 quadrupoles operating at unit resolution, the acquisition time windows and target scan time were set to 120 s and 1 s, respectively. Parameters for LC-SRM acquisitions can be found in Supplementary Table 1.

2.8. LC-SRM data analysis

LC-SRM data analysis was performed using Skyline software [16]. Peptide peak picking was performed using the mProphet algorithm and the second best peak model. A Q-value of 0.01 (1% FDR) was set as the cut-off for peptide signal analysis. In addition to peptide signal scoring (composite signal), all transitions were individually inspected and excluded if they were found to be unsuitable for quantification (low signal to noise ratio, obvious interferences). Unlabeled/labeled peak area ratios were calculated for each SRM transition. These ratios were then used to determine the corresponding average peptide ratio, and finally the average protein ratio was calculated from the ratios for the different signature peptides. Biomarker concentration was calculated from the average protein ratio. The lower limit of detection (LLOD) was determined according to the calibration curve method [17] and the lower limit of quantification (LLOQ) was defined as 3 X LLOD.

3. Results

3.1. Plasma glycodepletion procedure

Our study started with an analysis of the biochemical properties of cytolysis biomarkers. This category of biomarkers is almost exclusively composed of non-glycosylated proteins which are usually present in the cytoplasm. When these proteins are released into the bloodstream following cellular necrosis, they contribute to the non-glycosylated protein fraction of the plasma or serum. This fraction represents around 43% of the total proteins in these two biological matrices [18]. We hypothesized that analysis of non-glycosylated proteins such as cytolysis biomarkers could be improved by depletion of glycosylated plasma/serum proteins, as this step reduces the complexity of the matrix.

To achieve depletion of glycosylated proteins from the plasma/ serum, we adapted the concept of glycocapture which was initially developed to allow the analysis of glycoproteins [19]. In our procedure, plasma/serum glycoproteins were first oxidized with sodium periodate to convert saccharide ring cis-diol groups into aldehyde groups. Then, glycoproteins carrying periodate-created aldehydes were mixed with hydrazide-activated resin (in a spin column) in the presence of aniline to form hydrazone bonds. This aniline-catalyzed reaction results in covalent linkage between glycoproteins and the resin. Non-glycosylated proteins could then be eluted with 8 M urea. This denaturing elution buffer disrupts any protein/protein interactions that may have formed and could otherwise result in retention of some non-glycosylated proteins on the resin. This process resulted in efficient elution of cytolysis biomarkers (Fig. 1A). It should be noted that glycodepletion has no effect on albumin concentrations in samples as albumin is a non-glycosylated protein.

Glycodepletion can be applied to both plasma and serum samples. The depletion yield of this procedure was evaluated in these two

Fig. 1. Principle and characterization of plasma protein glycodepletion. (A)

Principle of plasma glycoprotein depletion using hydrazide-activated beads. Cytolysis biomarkers (non-glycosylated) are present in the unbound fraction. (B) Venn diagram showing the distribution of plasma proteins identified in the resin-bound and unbound fractions, as determined by nanoLC-MS/MS analysis (For more details, see Supplementary Table 2).

Fig. 2. Detection and quantification of liver cytolysis biomarker candidates in plasma using PSAQ standards, plasma glycodepletion and LC-SRM analysis. (A) Extracted ion chromatogram obtained after plasma glycodepletion, digestion and analysis using scheduled LC-SRM. For better clarity, only 8 out the 12 monitored peptides have been assigned. Calibration curves obtained for ADH4 (B), BHMT1 (C), FABP1 (D) and ALT1 (E) in plasma after spiking with PSAQ standards, glycodepletion, proteolysis and scheduled LC-SRM analysis.

matrices by quantifying proteins before and after glycodepletion using a micro-bicinchoninic protein assay. In plasma and serum samples (18 ^L initial volume), 43% of the protein content was retained on the hydrazide-activated resin. Interestingly, we also assessed glycodepletion efficiency to fractionate plasma from a patient with a strong inflammatory state (elevated acute phase glycosylated proteins). In this sample, 40% of the protein content was trapped onto the resin.

To further characterize glycodepletion, the resin-bound and the unbound protein fractions obtained after plasma processing were analyzed using nanoLC-MS/MS. These analyses allowed us to identify 124 plasma proteins exhibiting at least 2 specific spectral counts in one fraction (Fig. 1B, Supplementary Table 2). A comparison of the proteins identified in the two fractions confirmed that glycocapture was very effective to decomplexify plasma: numerous validated glyco-proteins were identified only in the resin-bound fraction or strongly enriched in this fraction. In particular, we identified a number of well-known high-abundance plasma glycoproteins (1-200 ^mol/L) including alpha-2-macroglobulin, alpha-1-antitrypsin, serotransferrin and haptoglobin. Proteins present at lower concentrations (0.1-10 ^mol/ L), such as coagulation factors and kallistatin, were also depleted by the

procedure (Supplementary Table 2) [20]. In the unbound fraction, few non-glycosylated plasma proteins were identified including albumin, vitamin D-binding protein, retinol-binding protein 4 and 3-hydroxya-cyl-CoA dehydrogenase type-2 (Supplementary Table 2) but the depth of analysis - as classically performed for a discovery experiment - was hampered by the high amount of albumin present in the fraction. Interestingly, some glycoproteins were also identified in the unbound fraction including immunoglobulins and apolipoproteins. The presence of these proteins may be explained by the low level and/or low accessibility of saccharide rings on these proteins.

3.2. Application of the assay combining glycodepletion and LC-SRM to quantify biomarkers of liver cytolysis in plasma

To examine whether glycodepletion can contribute to targeted proteomics analysis of cytolysis biomarkers, we selected and studied four proteins with the biochemical characteristics of liver cytolysis biomarkers. The four proteins were: alcohol dehydrogenase 4 (ADH4), betaine homocysteine methyltransferase (BHMT), fatty acid binding protein 1 (FABP1) and alanine aminotransferase 1 (ALT1). The first

three of these proteins are candidate cytolytic biomarkers which are linked to three major hepatic functions: ADH4 is involved in detoxification of xenobiotics, BHMT participates in the metabolism of sulfated amino acids, and FABP1 contributes to lipid metabolism [21]. The fourth protein, alanine aminotransferase, is already used in clinical biology; its enzymatic activity is measured in serum or plasma as an indicator of hepatic cytolysis. In these assays, the specific activities of the ALT1 and ALT2 isoforms cannot be distinguished, but ALT1 is the isoform specific to the liver. In the study described here, thanks to the careful selection of proteolytic peptides, we were able to specifically quantify the ALT1 isoform and to compare its levels to those of the pool of the two isoforms (ALT1 and ALT2).

To assess our procedure, we first optimized the LC-SRM method for detection of the target proteins in plasma. The peptides monitored for the four target proteins were selected based on sequence specificity (BLAST search against Uniprot database) and analytical detectability. The "best-flying" signature peptides and the best SRM transitions were selected experimentally based on LC-SRM analysis of recombinant proteins spiked into glycodepleted plasma and digested with an endoLysC/trypsin mixture. Three specific peptides were retained for ADH4 and FABP1. For BHMT, two peptides shared by isoforms 1 and 2 were selected along with a peptide specific for isoform 1. The two isoforms of BHMT are expressed in the liver and they can therefore both be considered candidate liver cytolysis biomarkers. Two specific peptides were selected for ALT1, and a peptide shared by isoforms ALT1 and ALT2 was also tracked (Supplementary Table 1).

We next developed a quantitative assay combining glycodepletion and LC-SRM to quantify the candidate cytolysis biomarkers in plasma (Fig. 2A). We used PSAQ standards (Protein Standard for Absolute Quantification) to perform absolute quantification [22]. These standards were isotopically-labeled analogs of ADH4, FABP1, BHMT1 and ALT1. Analytical performance of the assay was determined based on calibration curves, as recommended by the health authorities and the proteomics community (Fig. 2B to E). Five calibration points were created by adding increasing amounts of unlabeled proteins (FABP1, ADH4, BHMT1 and ALT1) and constant amounts of PSAQ standards to plasma samples. The upper limit of the calibration curve (concentration of unlabeled proteins added) was adjusted to correspond to expected pathological levels. These levels were estimated from preliminary experiments with two serum samples from patients with acetaminophen-induced acute hepatitis. Zero samples, containing only PSAQ standards, were also included in the curves. Technical replicates of all calibration points were produced (n=3; n=4 for the two lowest calibration points). After spiking, plasma samples were treated by the workflow combining glycodepletion, digestion with endoLysC/trypsin, and LC-SRM analysis. Linear calibration curves over the concentration ranges tested were obtained for all 12 peptides monitored. Comparison of the quantification results obtained for different peptides from each target protein gave highly correlated results. The accuracy (trueness) of the calibration curves was found to be excellent for FABP1 and ALT1, at between 92% and 117%. For ADH4 and BHMT1, the signature peptides monitored slightly overestimated levels in plasma, with accuracies between 136% and 151%. Of the 12 peptides monitored, 11 provided excellent analytical precision (< 20%), a level which conforms to the most stringent recommendations [23]. The details of analytical performance of the assay including LLOD and LLOQ for each signature peptide can be found in Table 1.

3.3. Comparison of glycodepletion and abundant protein immunodepletion for the quantification of liver cytolysis biomarker candidate in plasma

Depleting abundant proteins using immunoaffinity devices is a very popular method to fractionate plasma samples before LC-SRM analysis. We compared the analytical performances obtained with glycode-pletion and abundant protein depletion (using the MARS Hu-6 spin

cartridge) for the analysis of ADH4, FABP1, BHMT1 and ALT1 in plasma. For this, defined quantities of ADH4, FABP1, BHMT1 and ALT1 were mixed with their corresponding PSAQ standards and were added to plasma samples. Plasma samples were prepared using either glycodepletion or immunodepletion before digestion and LC-SRM analysis. All signature peptides (labeled and unlabeled versions) could be detected with both depletion methods. However, for ADH4 and FABP1, quantification results were more accurate when glycodepletion was used upstream LC-SRM analysis (Fig. 3). Likely, the use of urea to elute the non-glycosylated protein fraction from the resin equalized the biochemical behavior of the unlabeled proteins and their quantification standards. In the case of abundant protein immunodepletion, slight differences in sequence or structure between the unlabeled proteins and their labeled versions might have influenced protein complex formation and retention on the cartridge. Importantly, besides analytical performance assessment, glycodepletion of plasma samples was easy to parallelize compared to the use of immunodepletion cartridge that required individual processing of each sample.

3.4. Glycodepletion for the analysis of clinical samples

To validate the use of glycodepletion combined with LC-SRM in studies assessing biomarkers we applied our workflow to clinical samples. Thus, the four liver cytolysis biomarkers were assayed in serum samples from patients with acetaminophen-induced acute hepatitis (n=7) and levels were compared to those detected in serum from healthy donors (n=7). Table 2 shows the serum concentrations measured for the proteins of interest in serum samples from patients with acute hepatitis after addition of PSAQ standards, glycodepletion, digestion and LC-SRM analysis. In these serum samples, ADH4, BHMT (which corresponds to the sum of the two isoforms BHMT1 and BHMT2) and FABP1 were detected at extremely high concentration levels, sometimes exceeding the highest points on the pre-established calibration curves. In contrast, none of the proteins could be quantified in serum from healthy donors, either because of an absence of endogenous signal or because the signal fell below the LLOD, as determined based on the calibration curves. These results allow us to conclude that ADH4, BHMT (isoforms 1 and 2) and FABP1 are effectively released in large amounts into the circulation following acetaminophen-induced liver damage. An increase in serum levels of ALT1 was detected in six of seven patients, but, once again, it could not be quantified in healthy donors. Finally, for the five of seven patients for whom ALT (the sum of isoforms ALT1 and ALT2) could be measured by LC-SRM, concentrations correlated well with ALT enzymatic activity measured in serum (R2 =0.91).

4. Discussion

Candidate biomarkers remain difficult to assess by LC-SRM or LC-PRM in plasma/serum due to the pre-analytical steps which must be applied to gain access to the target proteins [9-11]. Immunodepletion of abundant proteins very effectively reduces the dynamic range of serum and plasma, but it has a certain number of disadvantages (cost, difficulty with automation when using spin column formats, etc.) [24,25]. In addition, as this approach is based on affinity interactions between antibodies and proteins, minority proteins cannot be eluted using denaturing buffers. In this article, we present a biochemical method for the preparation of plasma or serum samples which drastically reduces the protein complexity of these two matrices by separating the glycosylated protein fraction from the non-glycosylated fraction. Several previous studies have described the use of a mixture of lectins to enrich and analyze serum and plasma glycoproteins [26]. However, like with immunodepletion, these multi-lectin glycocapture approaches are based on affinity interactions, and are therefore not compatible with the use of denaturing buffers to elute the non-glycosylated fraction (destabilization of lectin/sugar interactions). To

Table 1

Analytical performance of the assay combining glycodepletion and LC-SRM for the quantification of liver cytolysis biomarker candidates in plasma.

Targeted protein

Signature peptide

Specificity Range of tested

concentrations tested (^g/mL)


Accuracy Truenessa (%)

LLODb LLOQc (^g/ Precision at (^g/mL) mL) LLOQ

(CV in %)




0.25-24.8 0.25-24.8 0.25-24.8

1.00 1.00 1.00

139 148 136

1.59 1.75

4.77 5.25 4.81

9-16 8-10 0-8





0.60-79.5 0.60-79.5 0.60-79.5

1.00 1.00 1.00

143 151 136

6.53 4.15 6.17

19.58 12.45 18.52

ND 2-6





0.18-3.69 0.18-3.69 0.18-3.69

0.97 0.98 0.98

110 114 108

0.71 0.64 0.66

2.13 1.91 1.99

2-13 5-14 1-8





0.40-8.10 0.40-8.10 0.40-8.10

0.96 0.99 0.99

96 92 117

1.21 1.10 2.17

3.62 3.29 6.52

6-14 15

a Trueness corresponds to the slope value (%) of the calibration curve for the peptide considered. b LLOD was defined according to the "calibration curve" method [17]. c LLOQ was defined as 3 X LLOD.

Trypsin digestion and LC-SRM analysis

Glycodepletion (18 ц1)


Unlabeled PSAQ

Mixed in defined amounts

Spiking in plasma samples

6 most-abundant

protein depletion: MARS (14 pi)

CHECK RATIOS (unlabelled/PSAQ)


Unlabelled/labelled signal ratio (mean of signature peptides) in buffer 0,19 0,22 0,68 0,29

Unlabelled/labelled signal ratio (mean of signature peptides) in plasma after glycodepletion 0,24 0,30 0,77 0,22

Biais 26% 36% 13% -24%

Unlabelled/labelled signal ratio (mean of signature peptides) in plasma after MARS depletion 0,07 0,18 0,33 0,44

Biais -63% -18% -51% 52%

Fig. 3. Comparison of glycodepletion and abundant protein depletion for the quantification of liver cytolysis biomarker candidates in plasma. (A) Principle of parallelism tests. Defined quantities of the 4 targeted proteins (ADH4, BHMT1, FABP1 and ALT1) were mixed with their corresponding PSAQ standards and were added to a plasma sample. This sample was prepared using either glycodepletion or immunodepletion (MARS Hu-6 cartridge) before digestion and LC-SRM analysis. Labeled to unlabeled SRM signal ratio (obtained from the different signature peptides) was compared to that obtained from a protein/PSAQ mix digested in 50 mM NH4HCO3 buffer. If the biochemical behavior of the target protein and its PSAQ standard are similar during plasma prefractionation, these ratios should be equivalent. (B) Results of the parallelism tests. For each protein, the difference between the unlabeled to labeled SRM signal ratios obtained in buffer and plasma samples was evaluated (bias in %).

overcome this drawback we developed a protocol to create covalent links between glycosylated proteins and a solid support. Non-glycosy-lated proteins could then be eluted for analysis using a denaturing

buffer. The method was inspired by the glycocapture method using hydrazide-activated resin, which was initially described by Zhang [19] for analysis of glycosylated proteins. This approach, based on chemical

Table 2

Quantification of liver cytolysis biomarker candidates in serum samples from patients with acute hepatitis induced by acetaminophen overdose.

Patient ADH4 Og/mL)

(BHMT1 and BHMT2) (Mg/mL)

BHMT1 (^g/mL) FABP1 (^g/mL) ALT1 (^g/mL)

ALT1 and ALT2 (Mg/mL)

ALT activity (U/L)

Patient 1 Patient 2 Patient 3 Patient 4 Patient 5 Patient 6 Patient 7

43.1 100.6

95.0 62.5

79.0 85.7

51.1 23.0

93.2 158.1



1489 4565 6410 2653 2571 7165 15480

reagents, is cheaper than immunodepletion of major proteins, can easily be parallelized, and is compatible with large volumes of plasma or serum. However, as serum albumin is not glycosylated it will still be present in the eluted fraction; it is therefore necessary to optimize the LC step (column capacity, gradient) and the SRM acquisition parameters to ensure that target signature peptides are not affected by the presence of peptides derived from albumin (competition at ionization, SRM signal interference).

We verified the performance and applicability of this workflow including glycodepletion to assess (non-glycosylated) candidate liver cytolysis biomarkers in serum samples. In hepatology, the commonly used cytolysis biomarkers are blood transaminases: aspartate aminotransferase (AST or SGOT) and alanine aminotransferase (ALT or SGPT). However, these biomarkers sometimes lack specificity [27], and they are therefore assayed in combination with other biological indicators, for example coagulation parameters (functional biomarkers). Using serum glycodepletion and LC-SRM, we investigated the potential value of three new biomarker candidates, namely ADH4, BHMT and FABP1 to reflect liver cytolysis. ALT1 and the pool of ALT1 and ALT2 were also studied. These protein targets were investigated in the context of acute hepatitis due to acetaminophen poisoning. This condition is characterized by the release of cytolysis biomarkers. Indeed, upon overdose, a toxic metabolite of acetaminophen (N-acetyl-p-benzoquinone imine) is produced in the liver parenchyma [28] which triggers hepatocyte necrosis and release of their contents into the bloodstream. Using our assay, which combines PSAQ standards, glycodepletion and LC-SRM, we confirmed that these proteins are effectively found in the blood following acetaminophen-induced liver toxicity. These candidate biomarkers appear to be highly sensitive as their serum levels increased dramatically following liver damage; the increase was much more marked for these proteins than for ALT. These initial results indicate that very extensive variations can be found between patient samples and control samples. In line with the guidelines formulated by the health authorities, we are working on extending the dynamic range of our test, both for higher concentrations (> 100 ^g/mL) to allow analysis of samples from patients with extensive hepatic cytolysis, and for lower concentrations (< 50 ng/mL) to attempt to cover physiological levels. However, as ADH4, BHMT and FABP1 are specifically expressed in the cytoplasm of hepatocytes, their circulating levels in samples from healthy donors are expected to be extremely low.

5. Conclusion

In conclusion, we have adapted the hydrazine resin glycocapture method to decomplexify plasma and serum matrices and to analyze the non-glycosylated fraction of these two sample types. This fraction is of particular interest as it contains a major class of biomarkers: cytolysis proteins. These proteins can be useful indicators of ischemic, toxic or mechanical damage to a number of tissues. The glycodepletion method developed, in combination with PSAQ standards and LC-SRM, was used to investigate the clinical relevance of three new candidate liver

cytolysis biomarkers for use in screening patients with acetaminophen-induced acute hepatitis.


We thank the team at EDyP for scientific discussions and technical support. We thank Maighread Gallagher-Gambarelli for editing services. This study was supported by grants from the 7th Framework Programme of the European Union (Contract no. 262067- PRIME-XS), by the COST Action CliniMark (CA16113) supported by COST (European Cooperation in Science and Technology), by the French National Research Agency in the framework of the "Investissements d'avenir" program (ANR-15-IDEX-02, LIFE project) and by the "Investissement d'Avenir Infrastructures Nationales en Biologie et Santé" program (ProFI project, ANR-10-INBS-08).

Appendix A. Supporting information

Supplementary data associated with this article can be found in the online version at doi:10.1016/j.talanta.2017.04.042.


[1] N.L. Anderson, The clinical plasma proteome: a survey of clinical assays for proteins in plasma and serum, Clin. Chem. 56 (2) (2010) 177-185.

[2] O.F. Laterza, V.R. Modur, J.H. Ladenson, Biomarkers of tissue injury, Biomark. Med. 2 (1) (2008) 81-92.

[3] T. Farrah, E.W. Deutsch, G.S. Omenn, D.S. Campbell, Z. Sun, J.A. Bletz, P. Mallick, J.E. Katz, J. Malmstrom, R. Ossola, J.D. Watts, B. Lin, H. Zhang, R.L. Moritz,

R. Aebersold, A high-confidence human plasma proteome reference set with estimated concentrations in PeptideAtlas, Mol. Cell. Proteom. 10 (9) (2011) (M110 006353).

[4] P. Mitchell, Proteomics retrenches, Nat. Biotechnol. 28 (7) (2010) 665-670.

[5] V. Lange, P. Picotti, B. Domon, R. Aebersold, Selected reaction monitoring for quantitative proteomics: a tutorial, Mol. Syst. Biol. 4 (2008) 222.

[6] A. Bourmaud, S. Gallien, B. Domon, Parallel reaction monitoring using quadrupole-Orbitrap mass spectrometer: principle and applications, Proteomics 16 (15-16) (2016) 2146-2159.

[7] N.R. Barthelemy, A. Gabelle, C. Hirtz, F. Fenaille, N. Sergeant, S. Schraen-Maschke, J. Vialaret, L. Buee, C. Junot, F. Becher, S. Lehmann, Differential mass spectrometry profiles of tau protein in the cerebrospinal fluid of patients with Alzheimer's disease, progressive supranuclear palsy, and dementia with Lewy bodies, J. Alzheimer's. Dis.: JAD 51 (4) (2016) 1033-1043.

[8] G. Picard, D. Lebert, M. Louwagie, A. Adrait, C. Huillet, F. Vandenesch, C. Bruley, J. Garin, M. Jaquinod, V. Brun, PSAQ standards for accurate MS-based quantification of proteins: from the concept to biomedical applications, J. Mass Spectrom.: JMS 47 (10) (2012) 1353-1363.

[9] T. Shi, D. Su, T. Liu, K. Tang, D.G. Camp 2nd, W.J. Qian, R.D. Smith, Advancing the sensitivity of selected reaction monitoring-based targeted quantitative proteo-mics, Proteomics 12 (8) (2012) 1074-1092.

[10] H.A. Ebhardt, A. Root, C. Sander, R. Aebersold, Applications of targeted proteomics in systems biology and translational medicine, Proteomics 15 (18) (2015) 3193-3208.

[11] M.A. Gillette, S.A. Carr, Quantitative analysis of peptides and proteins in biomedicine by targeted mass spectrometry, Nat. Methods 10 (1) (2013) 28-34.

[12] L.A. Echan, H.Y. Tang, N. Ali-Khan, K. Lee, D.W. Speicher, Depletion of multiple high-abundance proteins improves protein profiling capacities of human serum and plasma, Proteomics 5 (13) (2005) 3292-3303.

[13] Y. Shen, J. Kim, E.F. Strittmatter, J.M. Jacobs, D.G. Camp 2nd, R. Fang, N. Tolie, R.J. Moore, R.D. Smith, Characterization of the human blood plasma proteome, Proteomics 5 (15) (2005) 4034-4045.

[14] D. Lebert, A. Dupuis, J. Garin, C. Bruley, V. Brun, Production and use of stable isotope-labeled proteins for absolute quantitative proteomics, Methods Mol. Biol. 753 (2011) 93-115.

[15] M. Louwagie, S. Kieffer-Jaquinod, V. Dupierris, Y. Coute, C. Bruley, J. Garin, A. Dupuis, M. Jaquinod, V. Brun, Introducing AAA-MS, a rapid and sensitive method for amino acid analysis using isotope dilution and high-resolution mass spectrometry, J. Proteome Res. 11 (7) (2012) 3929-3936.

[16] B. MacLean, D.M. Tomazela, N. Shulman, M. Chambers, G.L. Finney, B. Frewen, R. Kern, D.L. Tabb, D.C. Liebler, M.J. MacCoss, Skyline: an open source document editor for creating and analyzing targeted proteomics experiments, Bioinformatics 26 (7) (2010) 966-968.

[17] D.R. Mani, S.E. Abbatiello, S.A. Carr, Statistical characterization of multiple-reaction monitoring mass spectrometry (MRM-MS) assays for quantitative pro-teomics, BMC Bioinforma. 13 (Suppl. 16) (2012) S9.

[18] F. Clerc, K.R. Reiding, B.C. Jansen, G.S. Kammeijer, A. Bondt, M. Wuhrer, Human plasma protein N-glycosylation, Glycoconj. J. 33 (3) (2016) 309-343.

[19] H. Zhang, X.J. Li, D.B. Martin, R. Aebersold, Identification and quantification of N-linked glycoproteins using hydrazide chemistry, stable isotope labeling and mass spectrometry, Nat. Biotechnol. 21 (6) (2003) 660-666.

[20] G.L. Hortin, D. Sviridov, N.L. Anderson, High-abundance polypeptides of the human plasma proteome comprising the top 4 logs of polypeptide abundance, Clin. Chem. 54 (10) (2008) 1608-1616.

[21] S. Qin, Y. Zhou, L. Gray, U. Kusebauch, L. McEvoy, D.J. Antoine, L. Hampson, K.B. Park, D. Campbell, J. Caballero, G. Glusman, X. Yan, T.K. Kim, Y. Yuan, K. Wang, L. Rowen, R.L. Moritz, G.S. Omenn, M. Pirmohamed, L. Hood, Identification of organ-enriched protein biomarkers of acute liver injury by targeted quantitative proteomics of blood in acetaminophen- and carbon-tetrachloride-treated mouse models and acetaminophen overdose patients, J. Proteome Res. 15

(10) (2016) 3724-3740.

[22] V. Brun, A. Dupuis, A. Adrait, M. Marcellin, D. Thomas, M. Court, F. Vandenesch, J. Garin, Isotope-labeled protein standards: toward absolute quantitative proteomics, Mol. Cell. Proteom. 6 (12) (2007) 2139-2149.

[23] S.A. Carr, S.E. Abbatiello, B.L. Ackermann, C. Borchers, B. Domon, E.W. Deutsch, R.P. Grant, A.N. Hoofnagle, R. Huttenhain, J.M. Koomen, D.C. Liebler, T. Liu,

B. MacLean, D.R. Mani, E. Mansfield, H. Neubert, A.G. Paulovich, L. Reiter,

O. Vitek, R. Aebersold, L. Anderson, R. Bethem, J. Blonder, E. Boja, J. Botelho, M. Boyne, R.A. Bradshaw, A.L. Burlingame, D. Chan, H. Keshishian, E. Kuhn,

C. Kinsinger, J.S. Lee, S.W. Lee, R. Moritz, J. Oses-Prieto, N. Rifai, J. Ritchie, H. Rodriguez, P.R. Srinivas, R.R. Townsend, J. Van Eyk, G. Whiteley, A. Wiita, S. Weintraub, Targeted peptide measurements in biology and medicine: best practices for mass spectrometry-based assay development using a fit-for-purpose approach, Mol. Cell. Proteom. 13 (3) (2014) 907-917.

[24] R.L. Gundry, M.Y. White, J. Nogee, I. Tchernyshyov, J.E. Van Eyk, Assessment of albumin removal from an immunoaffinity spin column: critical implications for proteomic examination of the albuminome and albumin-depleted samples, Proteomics 9 (7) (2009) 2021-2028.

[25] T. Ichibangase, K. Moriya, K. Koike, K. Imai, Limitation of immunoaffinity column for the removal of abundant proteins from plasma in quantitative plasma proteomics, Biomed. Chromatogr.: BMC 23 (5) (2009) 480-487.

[26] S. Fanayan, M. Hincapie, W.S. Hancock, Using lectins to harvest the plasma/serum glycoproteome, Electrophoresis 33 (12) (2012) 1746-1754.

[27] D.N. Bateman, Changing the management of paracetamol poisoning, Clin. Ther. (2015).

[28] S. Tujios, R.J. Fontana, Mechanisms of drug-induced liver injury: from bedside to bench, Nat. Rev. Gastroenterol. Hepatol. 8 (4) (2011) 202-211.