Refinement of a Methodology for Untargeted Detection of Serum Albumin Adducts in Human Populations

Covalently modified blood proteins (e.g., serum albumin adducts) are increasingly being viewed as potential biomarkers via which the environmental causes of human diseases may be understood. The notion that some (perhaps many) modifications have yet to be discovered has led to the development of untargeted adductomics methods, which attempt to capture entire populations of adducts. One such method is fixed-step selected reaction monitoring (FS-SRM), which analyses distributions of serum albumin adducts via shifts in the mass of a tryptic peptide [Li et al. (2011) Mol. Cell. Proteomics 10, M110.004606]. Working on the basis that FS-SRM might be able to detect biological variation due to environmental factors, we aimed to scale the methodology for use in an epidemiological setting. Development of sample preparation methods led to a batch workflow with increased throughput and provision for quality control. Challenges posed by technical and biological variation were addressed in the processing and interpretation of the data. A pilot study of 20 smokers and 20 never-smokers provided evidence of an effect of smoking on levels of putative serum albumin adducts. Differences between smokers and never-smokers were most apparent in putative adducts with net gains in mass between 105 and 114 Da (relative to unmodified albumin). The findings suggest that our implementation of FS-SRM could be useful for studying other environmental factors with relevance to human health.


Ion Trap Mass Spectrometry of Synthetic Peptides
The methods and instrumentation used for electrospray ionisation mass spectrometry (ESI-MS) and tandem mass spectrometry (ESI-MS/MS) were as described previously. 1 Briefly, diluted HPLC eluates containing peptides of interest were infused directly into a Thermo LTQ-XL ion trap mass spectrometer (Thermo Fisher Scientific, Hemel Hempstead, UK) using a syringe pump. The mass spectrometer was operated in the positive ion polarity mode. The scan types were "full" (wider/lower resolution) or "ultra-zoom" (narrower/higher resolution). Spectra were acquired using LTQ Tune Plus

FS-SRM
All instruments, parts and software used for FS-SRM were purchased from Thermo Fisher Scientific (Hemel Hempstead, UK). The Dionex UltiMate 3000 Series liquid chromatograph consisted of a binary pump and an autosampler. The system was controlled using Xcalibur (version 2.0.7, SP1) and DCMS Link (version 2.14). The injection valve of the autosampler was connected directly to the ion source of the mass spectrometer via a length of fused silica tubing (length = 1 m, i.d. = 20 µm), and sufficient system back-pressure to ensure correct pump function (~30 bar) was achieved by connecting a column (Acclaim PepMap 100, length = 2 cm, i.d. = 75 µm, particle size = 3 µm) to the flow meter outlet. The aqueous eluent (eluent A) was 0.1% (v/v) formic acid in water. The organic eluent (eluent B) was 0.1% (v/v) formic acid in acetonitrile. Both eluents were prepared using de- Transitions were organised into 30-second segments using the format described by Li et al. 4 Thirtytwo SPs were distributed randomly among 16 segments (two SPs per segment). The mass spectrometer cycled repeatedly through a list of seven transitions (Table S1) for the duration of a segment (20 or 21 cycles per segment). The first transition of every cycle was reserved for targeted SRM of the internal standard. The 16 segments were run consecutively as an 8-min block. The block was flanked by repeated targeted SRM of the internal standard, allowing the sample's arrival (between 0 min and 4 min) and disappearance (between 12 min and 15 min) to be monitored (see Fig.   S1 for an example "chromatogram"). Low dispersion mode was used to switch the injection valve of the autosampler back to the LOAD position after 11 min, thus stopping the delivery of further sample to the ion source. The sample loop and associated fluidics were washed after each injection with approximately 0.5 mL of eluent mixture (composition as above). The total duration of each analysis (injection and wash) was 26 min. performed over 16 × 30-second "segments" between 4 min and 12 min (2 SPs per segment; see Table   S1).

Preparation of Cam-iT3
Cam-iT3 was prepared as a 7.8 µM solution in HPLC effluent containing 34% (v/v) acetonitrile and 0.1% (v/v) formic acid. The methods used for synthesis, purification, structural characterisation and quantification were as described previously. 1 The analytical data were consistent with the previous report.

Preparation of Nns-T3
A 0.   Table S9). The concentration of Nns-T3 could not be estimated from its tyrosine fluorescence 1 because of interference from the naphthalenyl group. For practical purposes, the concentration was approximated relative to Cam-iT3 using peak areas from HPLC-UV (λ = 210 nm; one-point calibration).

Preparation of Nes-HSA
A 15-mL polypropylene centrifuge tube was charged with a solution of commercial HSA (8.0 mg, ~120 nmol) in PBS (2 mL). A solution of DTT (8.00 µmol) in PBS (2 mL) was added, and the tube was rotated end-over-end for one hour at ambient temperature. The crude mercapto-HSA was transferred to a Vivaspin 2 centrifugal concentrator (see below) and the low-molecular-weight solutes were exchanged with fresh PBS over three cycles of molecular weight cut-off (MWCO) filtration and re-dilution. After the third cycle, the tube was centrifuged until the volume of retentate was 1.5 mL.
The retentate was repeatedly aspirated in a micropipette tip until homogenous, and samples of retentate and filtrate were removed for quantification of thiol groups (see below). To the remaining retentate (6.7 mg of protein in 1.25 mL of PBS), was added a solution of NEM (2.70 µmol) in PBS (450 µL). The mixture was left to stand for one hour at ambient temperature. Low-molecular-weight solutes were exchanged with fresh PBS over three cycles of MWCO-filtration and re-dilution. After the third cycle, the retentate was concentrated to a volume of approximately 0.2 mL. The concentration of thiol groups in a diluted (4 g L −1 ) solution of this material was lower than the detection limit of the thiol quantification assay (3.4 µM; see below), indicating that at least 97% of thiol groups in the mercapto-HSA had been derivatised.
The number of NEM additions per molecule of HSA was inferred from the number of available thiol groups (assumption: quantitative conversion of thiols to succinimide thioethers). The concentration of thiol groups in MWCO-filtered mercapto-HSA solution (see above) was estimated relative to GSH using an assay similar to one described previously. 1 Briefly, Ellman's reagent was reacted with known concentrations of GSH and unknown concentrations of protein thiol groups. The amounts of coloured product were measured using a microplate spectrophotometer, and the concentrations of protein thiol groups were estimated via a calibration curve (absorbance at 405 nm as a function of GSH concentration). For the present study, the assay medium consisted of sodium phosphate buffer (± GSH), PBS (± mercapto-HSA) and a buffered solution of Ellman's reagent in aqueous acetonitrile.
The estimated concentration of protein thiol groups was calculated by subtracting the thiols detected in the MWCO filtrate (residual DTT) from the total thiols in the MWCO-retentate (protein + residual DTT). The concentration of HSA in the MWCO-retentate was estimated using a microplate BCA assay, and was converted from mass-per-unit-volume to molarity using M r = 66,500. The average number of thiol groups per molecule of HSA was estimated by dividing the protein thiol concentration by the HSA concentration. The first equivalent of thiol groups was attributed to fully-reduced Cys-34 and the excess (0.7 thiol groups per molecule) was attributed to the formation of some half-cystine.

Extraction of HSA from Serum or Plasma
Serum or plasma was centrifuged (10,000 × g, 2 min, ambient temperature) and 50-µL aliquots of clear supernatant were dispensed into 1.5-mL polypropylene microcentrifuge tubes (Eppendorf, Stevenage, UK). PBS was saturated with ammonium sulfate, diluted to 60% (v/v) with fresh PBS and added to the serum or plasma in portions (3 × 150 µL; vortex-mixing after each addition). The tubes were rotated end-over-end for 10 min at ambient temperature. Samples were freed of the resulting colourless precipitates by centrifugation (10,000 × g, 20 min, ambient temperature). Supernatant (180 µL) was transferred to a centrifugal concentrator (Vivaspin 2, capacity = 2 mL, regenerated cellulose membrane, molecular weight cut-off = 10,000 Da; Sartorius, Surrey, UK), which had been washed with Tris-HCl buffer (50 mM, pH 8.0) to remove preservatives. The crude HSA in the concentrator was diluted to 2 mL with the same Tris-HCl buffer, and then centrifuged (2,500 × g, ambient temperature) until the volume of retentate was approximately 0.2 mL. The filtrate was discarded and the retentate was subjected to three more cycles of dilution (same diluent) and filtration.
After the final cycle, the retentate was transferred to a clean 1.5-mL microcentrifuge tube (see above), and any residual protein was washed out of the concentrator with more of the Tris-HCl buffer. The total material recovered from the filter membrane was diluted to a fixed volume (550 µL for fresh plasma, max. = 554 µL, min. = 548 µL; 450 µL for control plasma, max. = 452 µL, min. = 448 µL).
These fixed-volume extracts were prepared gravimetrically (density of all solutions ~ 1 g L −1 ).

Enrichment of HSA Adducts Using Covalent Chromatography (Test-Scale Only)
Protein was extracted from Nes + serum essentially as described above, except that the usual buffer (50 mM Tris-HCl, pH 8.0) was replaced with "binding buffer" (100 mM Tris-HCl, 500 mM NaCl, pH 7.4). A microplate BCA assay indicated that the concentration of total protein in the extract was 3.04 ± 0.06 g L −1 (mean ± s.d., three measurements). HPLC-UV analysis (gradient elution, λ = 280 nm) indicated that 80% (w/w) of the total protein was HSA [assumptions: (1)  Buckinghamshire, UK; 0.30 g, washed with water, slurried in de-gassed "binding buffer"). The headspace of the tube was purged with nitrogen and the tube was sealed. The contents of the tube were mixed by end-over-end rotation for 19 h, followed by filtration to remove the resin. The drained resin was washed with buffer (50 mM Tris-HCl, pH 8.0; 3 × 2 mL). The filtrates were combined, buffer-exchanged fully (50 mM Tris-HCl, pH 8.0) and concentrated to a known volume (~0.75 mL).
The total protein recovered in each experiment was quantified using a microplate BCA assay and analysed using HPLC-UV (see above). Amounts of HSA (pre-and post-enrichment) were estimated by multiplying the amount of total protein by the purity of the HSA (purity of adduct-enriched HSA: mean ± s.d. = 74.9 ± 0.1%, four experiments). In a control experiment where no resin was added, >95% of the total protein was recovered (one experiment).   (Table S2) and eluates were collected manually into glass vials (Sigma-Aldrich, Dorset, UK). The times at which to start and stop collecting eluate were defined using reference chromatograms for mixtures of Cam-iT3 and Nns-T3 (Fig. S2). New reference chromatograms were acquired on each day that eluates were collected.

Evaluation of Step Size
Nes + control serum was prepared for FS-SRM as described above (clean-up method: SPE). The format of the FS-SRM method was as described above, but the SPs were fewer in number, closer together and centred on d = 125. Throughput Analysis Table S3. Effect of method development on the time required for sample preparation and TQ-MS. A distinction is made between "manual" operations (e.g., liquid handling, centrifugation, short incubations) and "unsupervised" operations (long incubations and automated analyses). The time required for protein quantification is not included, but would be the same in either case (ca. 1 h per 20 samples). Step

SDS-PAGE of Plasma Extracts
A randomly-selected pair of plasma extracts (one from a smoker and one from a never-smoker) were analysed alongside an extract of Nes + plasma and a reference sample (commercial HSA). Plasma extracts and commercial HSA were diluted with Tris-HCl buffer (50 mM, pH 8.0) to a concentration of 0.2 g L −1 (total protein). Ten microlitres of each diluted sample were combined with 4 µL of a 10% (

Processing of FS-SRM Data
The MATLAB code used to process the raw data was as essentially as described by Li et al. 4 A minor modification was made to account for a difference in the method of data acquisition (targeted SRM of Cam-iT3 occurring at the beginning of every cycle rather than sometimes in the middle). Wilcoxon rank sum tests (significance level: p = 0.10) were used to ascertain the presence/absence of a sequence tag. Each set of raw data was compared to the equivalent data for a simple matrix of 0.1% (v/v) formic acid in 1:3 (v:v) water:acetonitrile. Partially validated responses were normalised using eq S1, in which i is a technical replicate (A or B); j is a subject; k is the SP; ̅ is a constant (protein concentration, mean, all extracts); and c ij is the concentration of protein extracted from subject j's plasma in replicate i. Random imputation of null responses was done using values sampled from a distribution representing "pure background". This was a uniform distribution of responses ranging from zero magnitude to that of a typical false positive. The typical false positive was defined as the mean of any partially-validated responses observed in an analysis of bovine serum albumin.

Preparation of Nns-T3
Analytical HPLC resolved three putative isomers of Nns-T3, each of which exhibited a UV spectrum resembling that of N-(naphthalen-1-yl)succinimide (i.e., the unfurnished chromophore 6 ; Fig. S4). The mixture of putative isomers eluted as a single peak under ASP-HPLC conditions, and the mass spectral data for the mixture were generally consistent with a single substance. When the mixture was analysed by ESI-MS, most of the prominent peaks could be accounted for by multiply-charged Nns-     Table   S9).    The variation intrinsic to the assay (see above) was not corrected for, nor was the variation among the volumes of the extracts (coefficient of variation = 0.3%). The colour-coding is as per the key in part (C), but the terms "among", "within", "both" and "either" are used with respect to batches rather than plates. Square markers are used as in part (D).
Table S10. Sources of technical variation in the measured amounts of protein extracted from human plasma.