2.2.1 HRMS analysis

The untargeted analysis was performed on the Ultimate 3000 UHPLC system (Thermo Scientific) coupled with a linear ion-trap quadrupole-Orbitrap Velos Prohybrid mass spectrometry (Thermo Fisher Scientific, San Jose, CA, USA). The chromatography separation was carried out on a ACQUITY UPLC® BEH C18 column (2.1 mm × 100 mm, 1.7 μm), operated in the positive electrospray ionization (ESI + ) mode. The mobile phase was consisted of 0.1% (v/v) formic acid aqueous solution (A) and 0.1% formic acid (v/v) acetonitrile solution (B) with an optimized gradient program: 0–5 min, 5% (B); 5–55 min, 5%–25% (B); 55–75 min, 25%–35% (B); 75–83 min, 35%–95% (B); 83–87 min, 95% (B); 87–87.1 min, 95%–5% (B); 87.1–90 min, 5% (B).The column temperature was 30℃. The flow rate was 0.3 mL min⁻¹ and the injection volume was 4 μL.

The optimized source parameters were set as follows: spray voltage, 3 kV; capillary temperature, 300 °C; source heater temperature, 400 °C; sheath gas (N₂), 40 arbitrary units; and auxiliary gas (N₂), 10 arbitrary units. The Orbitrap analyzer scanned over a mass range of m/z 300–2000 at the resolution of 30,000 for MS¹ scan in profile format and the resolution of 7500 for MS² scan in centroid format. Multiple-charged ions dependent acquisition was applied for eliminating the interference of single-charged ions. The normalized collision energy (NCE) for MS/MS product ions was set at 10, 20, 30, 40 V, respectively. Dynamic exclusion was enabled by the following parameters: repeat count, 1; repeat duration, 20 s; exclusion list size, 50; and exclusion duration, 20 s. An isolation width of 2 Da was set. The minimum signal intensity that could trigger MS² fragmentation was set at 5000. Data processing were performed by Xcalibur 2.1 software (Thermo Fisher Scientific, San Jose, CA, USA).

2.2.2 Triple quadrupole mass spectrometer analysis

Targeted peptide analysis was performed on 1290 Infinity II UPLC system coupled with Agilent 6495 Triple Quad (Agilent, Santa Clara, USA) in positive ion mode. The chromatographic separation was performed on an ACQUITY UPLC® BEH C18 column (2.1 mm × 50 mm, 1.7 μm). The optimized gradient program was set as follows: 0–13 min, 5%–15% (B); 13–20 min, 15%–20% (B); 20–25 min, 20%–35% (B); 25–26 min, 35%–95% (B); 26–28 min, 95% (B); 28–28.1 min, 95%–5%; 28.1–30 min, 5% (B).

The ESI source parameters were set as follows: Gas temperature, 200 ℃; Gas flow (N₂), 14 L min⁻¹; Nebulizer, 35 psi; Sheath gas temperature, 250 ℃; Sheath gas flow, 11 L min⁻¹; Capillary voltage, 4000 V; Nozzle voltage, 500 V.

3.1.1 Construction of transition list

Peptide is polymers of amino acids linked by amide bonds. In ESI source mass spectrometry, peptides have unique ionization behaviors, which is different from metabolites and lipids. For MS¹ spectrum, the peptides are mainly characterized with multiple multi-charge ions simultaneously. For MS² spectrum, the fragmentation of peptide is dominated in bx-yz channels (Paizs et al., 2005) and the richness of its product ions information is greatly affected by the CE voltage.

The software Pep-MRMer was developed to extract and fuse the ion pairs. As shown in Fig. 2A, several crucial principles were set: for precursor ion, the ion fusion was performed to de-replicate the same ions with various charged forms, and the same ion in different acquisition parameters or different sample by the filter of tolerance of RT (tolRT) and tolerance of Mass (tolM), which were adjusted according to the experiment (e.g., 30 s, 0.02 Da, respectively). For product ion, the top two relative abundance ions were selected and the min MS² and max MS² were set to control the m/z within a range so as to obtain characteristic product ions. Notably, the minimum absolute m/z tolerance between precursor ion and product ion is 2 Da for eliminating isotope ion, which is different from that of 13.9 Da (—CH₂—) for the metabolites in MRM-Ion Pair Finder (Zheng et al., 2020). Another difference could be that the m/z of product ions might be higher than precursor ions’, which caused by the higher charge states of precursor ions. Eventually, the corresponding information of ion pairs with highest product ion abundance was remained and constructed a transition list.

Close

Fig. 2

The principles and performance of in-house software Pep-MRMer. A, The principles of ion fusion and transition list construction. B. The distribution of m/z of the highest product ion at CE voltage 10, 20, 30, 40, respectively. C, The statistic and identification of product ions with m/z less than 200. D, The 3D scatter diagram of ion pairs extracted from HRMS of three Pheretima. Different sizes represent different charges and different colors represent different abundance levels.

Export to PPT

Herein, the software was used for the transition list construction for QC samples of three Pheretima acquired at CE voltage of 10 V, 20 V, 30 V, and 40 V, respectively. As shown in Fig. 3B, the distribution of m/z of the highest product ions varies with CE voltage when the range was not restricted. This can be interpreted as the lower CE is not enough to fragment the peptide ion and the MS² spectrum is mainly the quasi-molecular ion peak of the compound; as CE increases, a part of ions is fragmented and produce abundant product ions with a wide range of m/z; further, the richness and relative abundance of peptide fragments with small m/z (<2 0 0) increases gradually. In order to illustrate what was produced, we counted the frequency of 20,000 MS/MS spectra with highest product ion m/z less than 200 at CE 30 V (Fig. 2C), the main product ions were fragment ions of mono-amino acid (e.g., [P-COOH] − 70.0651; [L/I-COOH] − 86.0964; [K-H₂O] − 129.1022). Therefore, we recommend a MS² range 200–1200 to establish more specific ion pair, that is, a peptide-peptide ion pair system, while eliminating noise or impurities signal interference. Moreover, for experimental purposes, high-abundance features with the ion abundance exceeds pre-defined threshold were remained. 5884 non-redundant ion pairs were extracted with widely distributed over the whole chromatographic gradient time, with diverse m/z, charges, and intensity (Fig. 2D).

Close

Fig. 3

The establishment and performance of HAI-RT-cal method. A, The principle of HAI-RT-cal method. B, The ΔRT distribution histogram of the 3307 ions by HAI-RT-cal and IS-RT-cal.

Export to PPT

3.1.2 Retention time calibration

Internal standards-based retention time calibration algorithm (IS-RT-cal) has effective performance for transferring RT between instruments for metabolite analysis (Zheng et al., 2020; Wang et al., 2016). Adversely, the method is severely restricted due to over-reliance on internal standards and stringent conditions, such as equal distribution in chromatogram, no interference for constituents, diverse structures, and stability.

A novel method named HAI-RT-cal was proposed, which replaced IS with easily accessible high abundance ions in matrices. In detail, the transition ion pairs with high abundance and equally distributed in whole chromatographic time were firstly selected and detected in QQQ6495 in MRM mode. Subsequently, according on the rule that the elution order of compounds is theoretically invariant under similar chromatography elution system, the high content peptides were manually refined to establish a calibration curve, which led to that all peptides within a time range of 90 min were calibrated into a time range of 30 min and without change in distribution of peptide. (Fig. 3A).

To verify the feasibility of HAI-RT_cal, two methods are simultaneously used for ion pairs transfer in this paper. A total of 304 high abundance ions were obtained and provided to HAI-RT-cal, while 21 standard peptides (Purity > 95%) were synthesized for the IS-RT-cal. The 5884 ion pairs extracted from HRMS were calibrated and detected by dMRM methods, of which 4664 (79.2%) transitions were clearly detected in 1 min detection window and 3307 (56.5%) transitions exhibited canonical chromatographic peak shape (Figure S2). The experimental and theoretical RT deviations (ΔRT) of 3307 ions by both methods were compared. As shown in Fig. 3B, ΔRTs of most ions were in the range of ± 0.5 min, which is the maximum threshold that allows to be detected in 1 min MRM window. But a considerable fraction of the post-eluting ions produced a large deviation after calibration by IS-RT-cal, even up to ± 7 min. The inadequate RT distribution of the standards will inevitably impact the effect of calibration. Overall, this method intelligently resolves the reliance of ion pairs transference on internal standard and improves the efficiency of sample acquisition.

3.1.3 Construction of dMRM method

Several parameters (e.g., cycle time, dwell time) greatly influence the detecting number of MRM transitions and the accuracy of quantitation in a single run. Considering the upper limit of detection performance of QQQ6495, 960 high intensity transitions with species specificity were remained, and then used to construct the quantitative dMRM method. To evaluate the quantitative performance of the pseudotargeted peptidomics dMRM method, methodological investigation was conducted, including linearity, precision, and repeatability. The linearity of the 960 transitions was evaluated by calculating the R² value of the 2⁸-fold gradient diluted QC samples. As shown in Fig. 4A, >94% ions had an R² > 0.8, and > 76% ions had an R² > 0.95. Repeatability and instrument precision were evaluated by calculating relative standard deviation (RSD) of 6 repeated injections. 90% and 93% of the transitions had RSD less than 20%, respectively (Fig. 4B and C).

Close

Fig. 4

Quantitative performance of the pseudotargeted peptidomics method and untargeted omics. A, Linearity of 960 dMRM transitions in QC samples with 2⁸-fold series dilution. B, Repeatability in positive ion mode evaluated in 6 QC samples. C, Instrument precision in positive ion mode evaluated in QC samples with 6 replicates. D, Venn diagram of 960 transition screened by pseudotargeted and untargeted strategy. E, Linearity of 764 transitions screened by untargeted and pseudotargeted method.

Export to PPT

Compared with untargeted strategy, pseudotargeted strategy was demonstrated with better quantitative ability and simplified data post-processing. As shown in Fig. 4D, of the 960 components retained in pseudotargeted strategy, 764 components could be assigned by Progenesis QI in untargeted strategy. While among the remaining 196 ions, 178 ions were not detected, and 18 ions were extracted with wrong charge value. All these components can be manually checked in MS¹ scans of original *.raw data file, indicating that the data post-processing of untargeted method is more problematical. Furthermore, linearity comparison of the 764 transitions acquired by both untargeted and pseudotargeted methods showed that this strategy has a lower quantification threshold and higher accuracy (Fig. 4E). As reported, the minimum limit of detection (LoD) and limit of quantification (LoQ) of the targeted strategy are typically two orders of magnitude lower than that of untargeted strategy (Domon et al., 2010).

In brief, the pseudotargeted peptidomics successfully integrated advantages of high detection coverage of the untargeted strategy and quantitative accuracy of the targeted strategy.