Development and Validation of a Capillary Electrophoresis Coupled to Mass Spectrometry Pipeline for Comparable Assessment of the Plasma Peptidome

Abstract

Although capillary electrophoresis coupled to mass spectrometry (CE-MS) holds promise for urinary peptide profiling, only a limited number of studies have used CE-MS to study plasma peptides. Here we describe the establishment of a workflow, including sample preparation, CE-MS analysis, data processing and normalization optimized for the analysis of plasma peptides. Using 291 plasma samples from 136 patients with end stage kidney failure (including pre- and post-dialysis samples) and 20 patients with chronic kidney disease, we identified and quantified the abundance of 3920 unique plasma peptides. The repeatability and intermediate precision of the analysis were high (with a coefficient of variation of 5% on average for all peptides). Six hundred sixty-one out of 3920 peptides were sequenced by CE-MS/MS. These peptide fragments belonged to 135 parent proteins. Using the pipeline, we identified 169 sequenced plasma peptides with different plasma abundance pre- and post-dialysis. These peptides combined in a support vector machine (SVM) classifier successfully discriminated between pre- and post-dialysis samples in a blinded validation cohort of 45 dialysis patients. Enriched peptides post-dialyses were for the major part associated to inflammation and the coagulation contact systems that may serve as signatures for optimizing dialysis materials. In conclusion, this high-throughput strategy focuses on the plasma peptidome, an understudied component of the plasma, as a promising area for further exploration. Due to their close proximity to the vascular bed, plasma peptides hold significant potential to serve as reliable biomarkers for systemic complications associated with kidney disease.

Keywords: biomarkers; cardiovascular disease; chronic kidney disease; peptidome; plasma.

FIGURE 1

Study design and participants. For this study, we included patients with ESKF on dialysis (n = 181) and non‐dialysis patients with CKD (n = 20). Patients on dialysis were randomly divided into a discovery and a validation cohort and plasma samples were collected before (Pre‐dial.) and immediately after (Post‐dial.) dialysis session. (A) Study flow chart. (B) Patient characteristics. ACEi/ARB, angiotensin converting enzyme inhibitors/angiotensin receptor blockers; BMI, body mass index; eGFR, estimated glomerular filtration rate; HBP, high blood pressure; hsCRP, high sensitivity C reactive protein; Q1, lower quartile; Q3, upper quartile.

FIGURE 2

Processing and normalization of plasma samples using new analysis pipeline. Plasma samples were analyzed with CE‐MS, processed, and then normalized according to the described pipeline. (A) Box‐whisker plots for the deviation in migration time of the analysis of 30 representative samples before and after migration time alignment against the reference dataspace. The purple line corresponds to the median migration time in the reference sample. (B) Representative distribution profile of plasma peptides before and after migration time alignment against the reference dataset. (C) The box‐whisker plot of peptide abundance for 30 representative samples before and after intensity normalization. The purple line corresponds to the median intensity across samples. The dotted lines correspond to the median intensity value of Min and Max across samples. (D and E) Short‐term and long‐term intra‐assay precision: the same sample was analyzed in nine consecutive runs to assess short‐term variability and then analyzed at three different points in time (in 2021, 2023, and 2024) to assess long‐term variability. The coefficient of variation in the nine consecutive runs was calculated for each peptide. The mean frequency of all peptides was plotted against the coefficient of variation (D). The intensity of 10 randomly selected peptides was plotted across the runs and across time (E). Non dial, non dialysis; pre‐dial., before dialysis session; post‐dial., after dialysis session.

FIGURE 3

Representation of the plasma peptidome content. (A) Number of peptides identified by CE‐MS/MS and database search for collagenic or noncollagenic proteins among all identified plasma peptides (n = 661). (B) Number of peptides identified by CE‐MS/MS and database search for noncollagenic proteins. (C) Top 20 Reactome pathways based on noncollagenic proteins. SERPINA1, alpha‐1‐antitrypsin; SERPINA3, alpha‐1‐antichymotrypsin; SERPIND1, heparin cofactor 2; SERPINF2, alpha‐2‐antiplasmin.

FIGURE 4

Peptidome analysis in plasma samples from dialysis patients before and immediately after the dialysis session. The plasma peptidome from 122 patients was analyzed pre‐ and post‐dialysis. (A) The frequency of the peptides detected in at least 50% of the samples in each group was plotted against the intensity values set to log(2). Histograms of the distribution of pre‐dialysis samples (red) and post‐dialysis samples (blue) both display a bimodal distribution with a point‐mass at zero and a continuous component. (B) Bar plot of average peptide intensity in paired plasma samples (pre‐dialysis, red and post‐dialysis, blue) of 122 patients. Inset, zoom on six patients. (C) Principal component analysis of peptide intensity in plasma samples taken pre‐ (red) or post‐dialysis session (blue) or in samples from non‐dialysis CKD patients (green). (D) Volcano plot showing fold‐changes (Log2) between pre‐ and post‐dialysis groups as well as statistical significance (‐Log 10 of adjusted p value) for 700 peptides. Points above the dashed line have a q‐value < 0.05 and corresponding peptides are considered significantly differentially abundant after dialysis session. (E) Peptides with significantly different abundances derived from collagenic and non‐collagenic proteins (y‐axis) and corresponding Log2 fold changes (x‐axis).

FIGURE 5

Development and validation of a plasma peptide classifier to stratify samples pre‐ and post‐dialysis. A SVM peptide classifier consisting of 169 sequenced plasma peptides that were significantly different between samples pre‐ and post‐dialysis was developed in the discovery cohort and tested in the independent validation cohort. (A) ROC curve in the discovery cohort. (B) Scores of the SVM classifier and ROC curve in the validation cohort. ****p < 0.0001 using One‐way Anova test.