Skip to main content
Download PDF

Novel plasma protein biomarkers from critically ill sepsis patients

Clinical Proteomics volume 19, Article number: 50 (2022) Cite this article

Abstract

Background

Despite the high morbidity and mortality associated with sepsis, the relationship between the plasma proteome and clinical outcome is poorly understood. In this study, we used targeted plasma proteomics to identify novel biomarkers of sepsis in critically ill patients.

Methods

Blood was obtained from 15 critically ill patients with suspected/confirmed sepsis (Sepsis-3.0 criteria) on intensive care unit (ICU) Day-1 and Day-3, as well as age- and sex-matched 15 healthy control subjects. A total of 1161 plasma proteins were measured with proximal extension assays. Promising sepsis biomarkers were narrowed with machine learning and then correlated with relevant clinical and laboratory variables.

Results

The median age for critically ill sepsis patients was 56 (IQR 51–61) years. The median MODS and SOFA values were 7 (IQR 5.0–8.0) and 7 (IQR 5.0–9.0) on ICU Day-1, and 4 (IQR 3.5–7.0) and 6 (IQR 3.5–7.0) on ICU Day-3, respectively. Targeted proteomics, together with feature selection, identified the leading proteins that distinguished sepsis patients from healthy control subjects with ≥ 90% classification accuracy; 25 proteins on ICU Day-1 and 26 proteins on ICU Day-3 (6 proteins overlapped both ICU days; PRTN3, UPAR, GDF8, NTRK3, WFDC2 and CXCL13). Only 7 of the leading proteins changed significantly between ICU Day-1 and Day-3 (IL10, CCL23, TGFα1, ST2, VSIG4, CNTN5, and ITGAV; P < 0.01). Significant correlations were observed between a variety of patient clinical/laboratory variables and the expression of 15 proteins on ICU Day-1 and 14 proteins on ICU Day-3 (P < 0.05).

Conclusions

Targeted proteomics with feature selection identified proteins altered in critically ill sepsis patients relative to healthy control subjects. Correlations between protein expression and clinical/laboratory variables were identified, each providing pathophysiological insight. Our exploratory data provide a rationale for further hypothesis-driven sepsis research.

Background

Sepsis refers to life-threatening organ dysfunction caused by a dysregulated host response to infection [1, 2]. Septic shock is sepsis with compounded circulatory, cellular, and metabolic dysfunction associated with a higher risk of mortality. Sepsis is a predominant contributor to hospital admissions, as well as patient morbidity and mortality, representing as many as 50% of adult intensive care unit (ICU) admissions and 48% of ICU deaths [3].

Sepsis is manifested by complex immunologic, neuronal, autonomic, hormonal, metabolic and coagulation derangements. These are further influenced by host genetic predisposition and heterogeneity, as well as social determinants of health [4,5,6,7,8,9]. These dynamic interactions demonstrate the diversity in critical illness pathophysiology, exemplifying the need for investigations rooted in high-throughput technologies, to elucidate their complexity. To this end, we have recently used targeted proteomics to identify novel plasma biomarkers associated with disease outcome in critically ill COVID-19 patients; six novel proteins were identified that predicted ICU mortality with 100% accuracy [10]. Targeted proteomics is valuable for not only identifying proteomic associations with disease progression and outcome, but also for improved understanding of pathophysiology related to specific patient features.

Over the last decade, proteomic analysis of the blood from critically ill patients has provided novel insight into the molecular basis of sepsis [11]. One of the first studies identified inflammatory and acute phase response proteins upregulated in sepsis [11]. Others identified novel biomarkers of lipid transport and coagulation proteins [12], followed by a host of mass spectroscopy and immunologic assays that increased our knowledge of the sepsis proteomes [13,14,15,16,17]. Despite numerous sepsis investigations, biomarkers for disease surveillance and organ dysfunction remain elusive. Moreover, temporal investigations to better understand disease evolution, as well as the host response to therapies, are lacking.

In this study, we performed targeted proteomics with proximity extension assays (PEA) as previously described [18, 19]. The PEA approach relies on antibodies conjugated to complimentary oligonucleotides, which upon binding to a target protein are in close proximity for hybridization and deoxyribonucleic acid (DNA) polymerase-based extension. The complex can then be amplified using microfluidic quantitative polymerase chain reaction (qPCR). PEA offers unparalleled sensitivity and specificity, with measurement of 1161 distinct human plasma proteins in this study.

Our specific study objectives were to: [1] measure plasma protein expression between critically ill sepsis patients and age-/sex-matched healthy control subjects; [2] determine the leading proteins with greatest classification accuracies between cohorts; [3] investigate how these protein expression levels vary temporally in critically ill sepsis patients; and [4] determine correlations between these protein expression levels and relevant clinical/laboratory variables.

Methods

This study was reviewed and approved by the Western University, Human Research Ethics Board (HREB #6970, “Repository of biological specimens from patients with critical illness and/or traumatic conditions”, March 17, 2021; #17908E, “Analysis of inflammatory markers in blood samples from patients with serious illness”, March 19, 2021). A total 15 patients with sepsis based on Sepsis-3 criteria [2] were enrolled. Patient characteristics included age, sex, comorbidities, medications, hematologic labs, creatinine, PaO2/FiO2 (P/F) ratios (arterial to inspired oxygen), chest x-ray findings, Multiple Organ Dysfunction Scores (MODS), Sequential Organ Failure Assessment scores (SOFA), use of antibiotics, anti-virals, corticosteroids, vasopressors, thromboembolism prophylaxis, anti-coagulation status, renal replacement therapy, invasive and non-invasive ventilation, intubation duration and ICU duration. Survival until Day-3 was not an inclusion criterion. For comparison with sepsis patients, 15 healthy control subjects without disease or acute illness were identified from the Translational Research Centre, London, Ontario (HREB #6963, “Repository of control biological specimens from healthy volunteers”, March 22, 2021; Directed by Dr. D.D Fraser; https://translationalresearchcentre.com/) [20, 21]. This study was performed in accordance with the ethical standards of the responsible committee on human experimentation and with the Helsinki Declaration of 1975.

Blood draws

Standard phlebotomy procedures were used to collect blood. Samples were obtained from critically ill ICU patients via indwelling venous catheters and immediately placed on ice. Maximum volumes were not exceeded per standard phlebotomy protocols. Following collection, samples were transferred to a negative pressure hood; after centrifugation, isolated plasma was aliquoted in 250 µL increments and frozen at − 80 °C. All samples remained frozen until use to avoid repeated freeze–thaw cycles.

Proximity extension assay

Plasma was thawed for PEA testing (Olink Proteomics, Sweden) as previously described [18, 19]. Specifically, we measured a total of 1161 plasma proteins in the plasma of critically ill sepsis patients and age-/sex-matched healthy controls. Each protein was targeted with two antibodies, labelled with one oligonucleotide each, and having a region complementary to each other. The PEA was performed in three steps: [1] antibody pairs, labelled with unique DNA oligonucleotides, were attached to their target antigen in plasma; [2] oligonucleotides that were brought into proximity hybridized and were extended by a DNA polymerase; and [3] the newly formed DNA barcode was amplified using qPCR. Individual samples were screened based on quality controls for immunoassay and detection, as well as degree of hemolysis. Intra-assay variability was minimized via robotic pipetting for volume accuracy, and through normalization using three specifically engineered internal controls that were added to each sample, including one control for the incubation, one for the extension and one for the amplification. External negative control and plate control samples were included in each sample plate in triplicate to improve inter-assay precision. Following proteomic quality control, all 30 (15 healthy control subjects and 15 critically ill sepsis) participants were deemed suitable for analysis. The data generated were expressed as relative quantification on the log2 scale of normalized protein expression (NPX) values. NPX values were rank‐based normal transformed for further analyses.

Population statistics

Medians (interquartile ranges [IQRs]) and frequency (%) were used to report ICU patient baseline characteristics for continuous and categorical variables, respectively. Group differences in baseline characteristics between sepsis patients on Day-1 and Day-3 were examined with Wilcoxon signed-rank tests given the dependency of the data (GraphPad Prism Version 8.4.0; San Diego, California USA). Statistical differences in protein expression between healthy controls and sepsis patients on either Day-1 or Day-3 were examined with Mann–Whitney U tests. Temporal differences in protein expression amongst sepsis patients were analyzed using Wilcoxon signed-rank tests. Statistical significance for protein expression between critically ill sepsis patients and healthy control subjects was adjusted for multiple comparisons using the Bonferroni correction (P < 0.05 post correction). As fewer proteins were analyzed for temporal differences between ICU Day-1 and Day-3, a P < 0.01 was considered statistically significant. Heat maps depicting Pearson correlation values between proteins and clinical or biochemical parameters were created in R (http://www.r-project.org) using the ggplot2 version 3.3.3 package. Significant correlations had a Pearson R-value of either 0.5 to 1.0 or -0.5 to -1.0 and a P < 0.05. Pearson correlation values assessing dynamic change (ICU Day-3 values minus ICU Day-1 values) between proteins and clinical or biochemical parameters were examined using SPSS v.28 (IBM Corp., Armonk, NY, USA). There were no missing data for any clinical outcomes.

Machine learning

Machine learning was implemented to identify protein variation underlying differences in the proteomes of ICU Day-1 and Day-3 sepsis patients, versus healthy controls. Analyte data were visualized with a nonlinear dimensionality reduction on the full matrix using the t-distributed stochastic nearest neighbour embedding (t-SNE) algorithm we have previously characterized [22]. t-SNE assumes that the ‘optimal’ representation of the data lies on a manifold with complex geometry, but low dimension, embedded in the full dimensional space of the raw data [23]. For feature selection, the raw data for each subject were ingested across subjects. A random forest classifier was trained on the variables to predict sepsis status. A random forest is a set of decision trees and, consequently, we were able to interrogate this collection of trees to identify the features that had the highest predictive value, or those features that frequently appeared near the top of the decision tree, for determining sepsis patients versus healthy controls. The feature matrices for ICU Day-1 and Day-3 critically ill sepsis patients were classified for patient outcome using a three-fold cross validation with a random forest of one hundred trees and max depth of six trees to reduce overfitting [24]. A Boruta feature selection method, based on random forest classifiers, was used to develop a reduced model using a training set (70%) and verified with a testing set (30%). The hyperparameter for the Boruta algorithm “percentile” was set to 90. All machine learning analysis was done using Python 3.9, Scikit-Learn (v. 24.0) and Boruta Py (v. 0.3) [25, 26].

Results

Demographic, clinical and laboratory parameters for the 15 critically ill sepsis patients, as defined by Sepsis-3 criteria [2], are presented in Table 1. A pathogen was identified in one third of patients and consisted of Methicillin-Resistant Staphylococcus aureus (sputum), Methicillin-Susceptible Staphylococcus aureus (blood), Streptococcus pneumoniae (sputum), Metapneumovirus (sputum) or Micrococcus species (blood). All patients were intubated and mechanically ventilated at some point throughout their ICU admission (median 4 days; IQR 3–4), and three quarters required vasoactive medications. The median ICU stay was 5 days (IQR 4–6). Only 1 of the 15 sepsis patients died, occurring in the ICU on Day-3 after blood work was drawn and clinical variables collected.

Table 1 Demographic and clinical data
Full size table

When comparing ICU patient clinical data on Day-1 versus Day-3, statistically significant decreases in white blood cell count, neutrophils, lactate and the use of vasoactive medications were observed (P < 0.05), suggesting overall patient improvement. Sepsis patients were intubated for a median of 4 days (IQR 3–4), and the median length of hospital admission was 12 days (IQR 7–16).

Of the 1161 plasma proteins measured using PEA, feature selection identified those with a 90% or greater classification accuracy for identifying a critically ill sepsis patient when compared to healthy control subjects. Using the 90% classification model, 25 leading proteins were identified on ICU Day-1 (Table 2) and 26 leading proteins on ICU Day-3 (Table 3). Respective t-SNE plots of the leading classifying proteins illustrate that the proteome of critically ill sepsis patients was distinct, recognizable and easily separated from age- and sex-matched healthy control subjects (Fig. 1A, B). A total of 6 leading proteins overlapped on ICU Day-1 and Day-3 (PRTN3, UPAR, GDF8, CXCL13, NTRK3 and WFDC2; Fig. 1C).

Table 2 Median plasma expression of the top 25 classifying proteins, ICU Day-1
Full size table
Table 3 Median plasma concentration of the top 26 classifying proteins, ICU Day-3
Full size table
Fig. 1
figure 1

Targeted proteomics accurately differentiates critically ill sepsis patients from healthy controls. In the upper section, t-SNE plots depict the separation between sepsis patients and healthy controls. Age- and sex-matched subjects plotted in 2D following dimensionality reduction of their respective proteomes by t-distributed stochastic neighbour embedding. Axes are dimensionless. The dimensionality reduction shows that based on plasma proteome, the two cohorts are distinct and easily separable. A Black dots represent ICU Day-1 patients and green dots represent healthy controls. B Orange dots represent ICU Day-3 patients and green dots represent healthy controls. C Proteins that feature selection identified as classifying proteins on both ICU Day-1 and Day-3 are shown in the Venn-diagram with overlap. Proteins in Red text were upregulated when compared to healthy controls, whereas proteins in Blue text were downregulated when compared to healthy controls

Full size image

The median plasma protein expression of the classifying proteins in both healthy control subjects and critically ill sepsis patients are shown for ICU Day-1 (Table 2) and Day-3 (Table 3), with their respective Bonferroni corrections. An alphabetized list of these proteins with brief descriptions are provided in Additional file 1: Table S1. While the majority of normalized plasma protein expression were increased in critically ill sepsis patients compared to healthy control subjects, a small number were decreased. For the latter, CPA2, GDF8, NTRK3 and DSG4 all had lower median plasma expression on ICU Day-1 when compared to healthy control subjects (Table 2). Similarly, on ICU Day-3, CNTN5, GDF8, GCP5, NTRK3, ITGAV, GALNT7 and PLXDC1 had lower median plasma expression as compared to healthy controls (Table 3).

Only a small number of plasma proteins significantly changed in critically ill sepsis patients from ICU Day-1 to Day-3. Among the ICU Day-1 classifying proteins IL-10, CCL23, and TGFα1, significantly decreased in sepsis patient plasma by Day-3 (Fig. 2; Additional file 1: Table S2). Of the ICU Day-3 classifying proteins, ST2, CNTN5, and ITGAV decreased significantly in median level from ICU Day-1 to Day-3 and VSIG4 increased significantly in sepsis patient plasma between Day-1 and Day-3 (Fig. 2B).

Fig. 2
figure 2

Significant temporal differences in plasma protein expression from sepsis patients. Box plots illustrating statistically significant differences between sepsis patients on ICU Day-1 versus ICU Day-3. The green shading illustrates healthy control protein expression (5–95% percentiles). Three ICU Day-1 classifying proteins significantly decreased in concentration by Day-3 (IL-10, CCL23, and TGFα1), and three ICU Day-3 classifying proteins significantly decreased in concentration by Day-3 (ST2, CNTN5, and ITGAV). Only one ICU Day-3 classifying protein was upregulated by ICU Day-3 (VSIG4). Median differences in protein concentrations were analyzed using Wilcoxon signed-rank tests, with P < 0.01 deemed statistically significant

Full size image

We then investigated whether the normalized protein expression in critically ill sepsis patients were correlated with their clinical, hematologic and biochemical parameters. On ICU Day-1 (Fig. 3A), OPG positively correlated with increasing MODS. TPP1 and CCL20 positively correlated with the use of vasopressors. In terms of hematologic parameters, PRTN3, CCL23, S100A11, CXCL13, ELOA, and FGR all positively correlated with increased white blood cell and neutrophil counts. OPG and SRP14 also positively correlated with neutrophilia. CPA2 positively correlated with lymphocyte count, and OPG negatively correlated with lymphocyte count. In terms of biochemical parameters, OPG, UPAR, TGFα1 and WFDC2 positively correlated with increasing international normalized ratio (INR). UPAR positively correlated with creatinine. Correlations of dynamic change (Day 3 minus Day 1) between normalized plasma protein expression and sepsis patient parameters are provided in Additional file 1: Table S3. In this latter analyses, positive correlations reflect movement of both clinical parameter and protein in the same direction, either increased or decreased, whereas negative correlations indicate divergence of the clinical parameter and protein in opposite directions.

Fig. 3
figure 3

Correlations between normalized plasma protein expression and sepsis patient parameters on ICU Day-1 and Day-3. Heat maps of rank-based classifying proteins reported in Fig. 1 on ICU Day-1 (A) and ICU Day-3 (B) are illustrated (y -axis) along with patient parameters (x-axis). Only proteins that showed a significant correlation (P < 0.05) with at least one biochemical or clinical parameter are illustrated. Significant correlations had a Pearson R-value of ≥ 0.5 or ≤ -0.5 and P < 0.05, denoted by *. Positive correlations are depicted in red and negative correlations in blue. MODS Multiple Organ Dysfunction Score, SOFA Sequential Organ Failure Assessment, P/F arterial partial pressure of oxygen divided by the fraction of inspired oxygen concentration, MV mechanical ventilation, Vaso vasopressors, HgB hemoglobin concentration, WBC white blood cell count, PMN neutrophil count; Lymph lymphocyte count, PLT platelet count, PTT partial thromboplastin time, INR international normalized ratio, Creat creatinine concentration

Full size image

Significant correlations between classifying proteins and patient variables on ICU Day-3 are shown in Fig. 3B. NTproBNP positively correlated with MODS. CHI3L1 positively correlated, while ITGAV and PLXDC1 negatively correlated, with SOFA score. GCP5 negatively correlated with P/F ratio, and PLXDC1 negatively correlated with mechanical ventilation. ST2 positively correlated with stress-steroid administration. ANGPT2 negatively correlated, and ITGAV positively correlated, with hemoglobin. ANGPT2 also positively correlated with lymphocyte count. PLXDC1 positively correlated with platelets. OPN and ASGR1 positively correlated with INR. UPAR, TNFR1, TNFRSF10A and WFDC2 positively correlated, while KIM1 negatively correlated, with creatinine. ST2 negatively correlated with partial thromboplastin time, while positively correlated with lactate.

Discussion

In this study, we performed targeted, high-throughput plasma proteomic profiling of critically ill sepsis patients on ICU Day-1 and Day-3. We used PEA [18, 19] to measure 1161 plasma proteins, and confirmed distinct proteomes in sepsis patients. Feature selection identified proteins that differentiate between critically ill sepsis patients and healthy control subjects with greater than 90% accuracy, and provided a rank order of importance based on their ability to differentiate cohorts. Moreover, we identified correlations between normalized protein expression and clinical, hematological and biochemical parameters. These data will be valuable for future hypothesis-driven studies [22], particularly those investigating disease surveillance, organ dysfunction and outcome prediction. In addition, the proteins identified here can be further evaluated with bioinformatics to aid our understanding of signalling pathways and potential therapeutics targets.

The baseline characteristics and comorbidities recorded in our critically ill sepsis patients were similar to other studies that profiled plasma proteomes using a variety of immunoassay methods [10, 22]. Pathogens were identified in 33% of patients (4 bacteria and 1 viral). All sepsis patients required mechanical ventilation, with most patients showing acute lung injury. Vasoactive medications were required in 73% of sepsis patients. Despite relatively high MODS and SOFA scores, our sepsis cohort had a higher survival rate (14 out of 15 patients) than expected based on previous literature [27].

We identified novel sepsis protein biomarkers using PEA and feature selection. A threshold of 90% classification accuracy was used to narrow the number of clinically relevant proteins for correlation studies: 25 proteins on ICU Day-1 and 26 proteins on ICU Day-3. The following 6 proteins overlapped on ICU Day-1 and Day-3: PRTN3, UPAR, GDF8, NTRK3, WFDC2 and CXCL13.

Only a small number of plasma proteins changed significantly between ICU Day-1 and Day-3 in critically ill sepsis patients, with decreased expression in IL10, CCL23, TGFα1, ST2, CNTN5, and ITGAV, and increased expression of VSIG4. CCL23 is a chemokine that is highly chemotactic for T-cells and monocytes, and has reported age-related differences in sepsis [28]. IL10 is an anti-inflammatory cytokine previously reported to be upregulated in septic shock, whose impact on clinical outcome is still unclear [29, 30]. TGFα1 is a mitogenic peptide (signalling through epidermal growth factor receptors triggering a kinase cascade) that has been found in bronchoalveolar lavage samples in ARDS correlating with lung injury and mortality [31]. ST2 is a member of the IL1R family that has roles in cardiac disease and cardio protection, and serves prognostic value after myocardial infarction. CNTN5, a member of the immunoglobulin superfamily, is a glycosylphosphatidylinositol-anchored membrane protein known to mediate nervous system development, which is relatively unexplored in sepsis [32, 33]. ITGAV is an integrin mediating cell adhesion (discussed below). VSIG4 encodes immunoglobulin domains that are structurally similar to the B7 family proteins, acting as a negative regulator of T-cell responses [34]. Taken together, each of the above proteins help to coordinate inflammatory responses. Both IL10 and TGF family proteins have been previously investigated as therapeutic targets in interventional trials, largely due to their measured serum changes in sepsis patients [35]. Polymorphisms in IL10 genes have also been associated with altered sepsis mortality and risk of ICU admission, highlighting IL10 as a potential therapeutic target [36]. Given that little is known about temporal protein expression in sepsis patients receiving ICU therapies [37, 38], these identified proteins may represent a targeted pool that hold specific value to understanding the host responses. Temporal proteomic analysis may also identify specific proteins with unique roles in disease pathophysiology beyond the acute illness phase, and provide prognostic information for those patients who have longer ICU admissions [39].

Significant correlations occurred between several proteins and parameters of disease severity, ventilation, and need for vasoactive agents. On ICU Day-1, OPG uniquely correlated with MODS, consistent with previous studies showing OPG as a biomarker of disease severity and mortality in sepsis [40]. OPG was first implicated in osteoclast regulation, yet recent evidence suggests emerging roles in vascular and immune biology [41]. Recently, OPG levels in sepsis were shown to correlate with markers of endothelial dysfunction, raising the possibility that endothelial dysfunction is mediated by OPG and resulting in end-organ dysfunction [42]. By ICU Day-3, only NTproBNP positively correlated with MODS. NTproBNP is secreted by the cardiac ventricles in response to ventricular expansion and pressure overload to promote natriuresis [43, 44]. With sepsis, NTproBNP expression may be influenced by fluid resuscitation and volume overload, resulting in increased myocardial stretch. Indeed, NTproBNP levels are used clinically for cardiac and perioperative risk stratification [45], suggesting it may provide similar risk stratification in sepsis patients.

Only CHI3L1 positively correlated with SOFA score. CHI3L1 is known to regulate tissue injury, repair and inflammation with strong associations in asthma, cancer, diabetes, cirrhosis, and coronary artery disease [46]. Moreover, previous cohort studies have documented the ability of CHI3L1 to predict outcome in infectious and inflammatory diseases [47, 48], highlighting this protein as an inadequately studied modulator of sepsis [47]. In contrast, ITGAV and PLXDC1 negatively correlated with SOFA score on ICU Day-3. ITGAV is an integral membrane protein, which regulates signal transduction, gene expression, proliferation, apoptosis, angiogenesis, invasion and metastasis [49]. ITGAV is also thought to preserve hematopoietic stem cells in the bone marrow [50], and regulate adipocyte proliferation and differentiation [51]. PLXDC1 is expressed on vascular endothelium, serving as a receptor for PEDF, with involvement in capillary morphogenesis [52, 53]. Collectively, proteins that correlate with MODS and SOFA scores may complement these clinically validated prognostication scores to provide more accurate diagnostic certainty [54].

Correlations were not significant for acute lung injury, including both P/F ratio and rates of mechanical ventilation, on ICU Day-1; however, these correlations were significant by Day-3. GCP5 negatively correlated with P/F ratio. GCP5 is a component of the gamma tubulin complex facilitating microtubule nucleation and centrosome assembly [55], suggesting a role in cell division and migration. The exact clinical significance of GCP5 in lung injury remains unknown. PLXDC1 negatively correlated with mechanical ventilation, as well as SOFA (reported above). As a receptor for PEDF, PLXDC1 may play a protective role in the pulmonary system. For example, PEDF is associated with a variety of pulmonary diseases such as idiopathic pulmonary fibrosis, lung cancer and COPD [56,57,58,59]. PEDF also has been associated with capillary permeability and septic shock [60, 61]. Given the role of PLXDC1 endothelial regulation, further investigation in sepsis is warranted with particular focus on alveolar-capillary unit.

Vasoactive agents are often necessary to maintain adequate end-organ perfusion. TPP1 and CCL20 positively correlated with the need for vasopressors on ICU Day-1. The serine protease TPP1 regulates lysosomal proteins [62], as well as mitochondrial trafficking and cellular metabolism [63], suggesting a role in cellular metabolism [64].The chemokine CCL20 regulates innate immunity and inflammation in skin and mucosa [65], is upregulated by LDL in smooth muscle of vascular structures, and is a reported biomarker for aortic aneurysms [66, 67].

Multiple proteins expressed by immune cells positively correlated with either total white blood cell count or neutrophil count on ICU Day-1, including OPG, PRTN3, CCL23, SRP14, S100A11, CXCL13, ELOA, and FGR. Uniquely, OPG (described above), positively correlated with neutrophilia and negatively correlated with lymphocytosis. The serine protease PRTN3 is produced and secreted by granulocytes, and is implicated in endothelial dysfunction during sepsis [68], while also having roles in organizing immune responses and autoimmune disease [69,70,71]. The chemokines CCL23 and CXCL13 attract immune cells [72], with CXCL13 affecting lymph node organization, B-cell migration, and chronic inflammatory diseases [73, 74]. SRP14, which traffics secretory proteins through the endoplasmic reticulum, and ELOA, a subunit of the transcription factor B (SIII) complex, may aid protein synthesis by immune cells during sepsis [75, 76]. S100A11 and FGR both regulate cellular migration and motility [77,78,79]. Their positive correlations with leukocyte and neutrophil counts may reflect leukocytes extravasating into tissues during sepsis. CPA2, a carboxypeptidase expressed by pancreatic tissue [80], was positively correlated on ICU Day-1 with lymphocytosis. As there are few published studies on CPA2 and immune regulation, further investigations are warranted. Interestingly, OPG and CPA2, emerge as potential key early regulators of either a neutrophilic or lymphocytic immune response in sepsis. Collectively, the identified correlations highlight the complexity of host response in sepsis, and suggest that these proteins could play central roles in cellular activation and migration.

By ICU Day-3, there were far fewer significant hematologic correlations compared to ICU Day-1. ITGAV (described above) positively correlated with hemoglobin titre. ANGPT2 negatively correlated with hemoglobin titre, but positively correlated with lymphocytosis. ANGPT2 prognosticates shock and death in critically ill patients [81], and is expressed by the vascular endothelium where it interacts with Tie and VEGF proteins to regulate angiogenesis, while demonstrating both pro- and anti-inflammatory properties [82]. PLXDC1 (described above) correlated with platelets on ICU Day-3.

There were significant correlations with INR. On ICU Day-1, OPG, TGFα1, WFDC2 and UPAR positively correlated with INR. OPG (discussed above) has emerging roles in vascular and immunobiology [41]. TGFα1 regulates cell proliferation, differentiation and development through binding epidermal growth factor receptors, with a specific role in tissue regeneration [83, 84]. WFDC2, also known as HE4, is an anti-protease that promotes angiogenesis [85,86,87] and is associated with kidney [88, 89] and lung [90] fibrosis. Elevated UPAR has been associated with lupus, cardiovascular disease, pancreatitis, cirrhosis and COVID-19 pathogenesis [91]. On ICU Day-3, OPN and ASGR1 positively correlated with INR. OPN is a matricellular protein that interacts with integrins to mediate cell motility, and promotes cell-mediated immune responses; it is involved in atherosclerosis, glomerulonephritis, cancer, and chronic inflammation, while also serving in biomineralization and inhibiting vascular calcification [92]. OPN was reported to be elevated in sepsis patients previously [93]. ASGR1 is a transmembrane protein that mediates endocytosis and lysosomal degradation of glycoproteins [94]. INR correlations suggest that these proteins may contribute to the consumptive coagulopathies and hepatic injury that often occur in sepsis, which greatly impact mortality and morbidity [95, 96].

KIM1, TNFR1, TNFRSF10A, UPAR (described above) and WFDC2 (described above) were correlated with creatinine, suggesting an association of these proteins with renal injury. KIM1, a type-I cell surface glycoprotein, negatively correlated with creatinine. As an acute marker of early acute kidney injury, KIM1 is predictive of long-term renal outcome [97] and is reported to have protective functions in renal ischemia–reperfusion injury, promote renal graft recovery, have a role in renal cancer pathophysiology and serve as a biomarker for sepsis-induced acute kidney injury [98,99,100,101,102]. The tumor necrosis and transforming growth factor protein families have well-known roles in cellular migration, growth, differentiation, and tissue repair [103], as well as critical roles in mediating the cytokine storm in sepsis [104]. As renal failure is an independent risk factor for morality in sepsis, the identification of protein markers associated with renal dysfunction may provide important prognostic information [105, 106].

Lactate is a critical marker of end-organ perfusion in sepsis. ST2 positively correlated with lactate on ICU Day-3. ST2, has been best characterized in cardiac disease, whereby myocardial stretch or myocyte injury results in ST2 upregulation and secretion of soluble ST2 that binds with IL33 to regulate cardiac response to injury [107]. ST2 also has elucidated roles in T-cell regulation, through interplay with IL33 and downstream transcriptional effects [108]. Relevant to sepsis, ST2 correlates with mortality and facilitates endothelial permeability [109, 110]. Moreover, ST2 negatively correlated with PTT and positively correlated with steroid therapy in our study. The latter correlation between ST2 and steroid dosing may relate to hydrocortisone-induced shock reversal [111], and subsequent improvements in lactate. New biomarkers of end-organ perfusion could impact management decisions in sepsis, as lactate can be unreliable in certain circumstances such as administration of vasoactive agents or confounding causes of Type B and D lactic acidosis [112, 113].

Correlations of dynamic change from ICU Day-1 to Day-3 between normalized plasma protein expression and sepsis patient parameters revealed interesting temporal findings, particularly for plasma chemokines. For example, a temporal decrease in CCL23 correlated with increased P/F ratio and decreased lactate, suggesting improved lung function and oxygen delivery. A decrease in CCL20 expression with time also correlated with decreased lactate. In contrast, a temporal increase in CXCL13 expression correlated with increased MODS and both decreased P/F ratio and worsening coagulopathy (increased INR and PTT).

Our study has several limitations. First, due to the cost of proteomics, we studied a relatively small sample size; however, our sample size is on par with a subset of other published exploratory studies investigating proteomic variation in sepsis [114]. Second, the limited dataset could also result in the random forest model overfitting; however, our use of conservative parameters would help mitigate this potential risk (three-fold cross-validation, number of trees was limited to 100 and the maximum depth of the trees was limited to 6). Third, given our incomplete knowledge of sepsis, and the proteins identified here within, we cannot propose direct links or mechanisms. Nonetheless, our exploratory study provides excellent background for hypothesis-generation. Fourth, given the improved SOFA scores, lactates and vasopressor requirements, the majority of our sepsis patients clinically improved from ICU Day-1 to Day-3, raising the possibility that our data might not have adequately captured proteome changes in sepsis patients who clinically deteriorate, and highlights the need for future studies with subgroup analysis. Finally, we identified pathogens (culture positive sepsis) in only 33% of all sepsis patients; however, identification of pathogens by culture ranges from 28 to 89% of patients [115, 116].

Conclusions

In conclusion, we present an exploratory study of high-throughput, targeted plasma proteomics from critically ill sepsis patients. Utilizing both machine learning and conventional statistics, we narrowed protein number for correlative investigations with clinical and laboratory parameters. Our exploratory data identified proteins that may serve as the targets for future hypothesis generating studies, including those investigating disease surveillance, organ dysfunction and outcome prediction. Furthermore, these data may be used to better understand the host response to both infection and interventions, including future studies investigating cellular and organ deconvolution, signalling pathways and drug repurposing.

Availability of data and materials

The datasets generated and/or analysed during the current study are available from the corresponding author on reasonable request.

Abbreviations

ICU:

Intensive care unit

PEA:

Proximity extension assay

P/F:

PaO2/FiO2

MODS:

Multiple Organ Dysfunction Scores

SOFA:

Sequential Organ Failure Assessment Scores

IQR:

Interquartile ranges

t-SNE:

T-distributed stochastic nearest neighbour embedding

PRTN3:

Proteinase 3

UPAR:

Urokinase plasminogen activator surface receptor

GDF8:

Growth differentiation factor 8

NTRK3:

Neurotrophic receptor tyrosine kinase 3

WFDC2:

WAP 4-disulphide core domain 2

CXCL13:

Chemokine (C-X-C motif) ligand 13

IL6:

Interleukin 6

IL10:

Interleukin 10

CCL23:

Chemokine (C–C motif) ligand 23

CCL20:

Chemokine (C–C motif) ligand 20

CHI3L1:

Chitinase 3 like 1

CPA2:

Carboxypeptidase A2

DSG4:

Desmoglein 4

CNTN5:

Contactin 5

GCP5:

Gamma-tubulin complex component 5

ITGAV:

Integrin subunit alpha V

GALNT7:

Polypeptide N-Acetylgalactosaminyltransferase 7

PLXDC1:

Plexin domain containing 1

OPG:

Osteoprotegerin

TPP1:

Tripeptidyl peptidase 1

S100A11:

S100 calcium binding protein A11

ELOA:

Elongin A

FGR:

Gardner-Rasheed feline sarcoma viral (v-fgr) oncogene homolog

SRP14:

Signal recognition particle 14

TGFα1:

Transforming growth factor α1

WFDC2:

WAP four-disulfide core domain 2

INR:

International normalized ratio

NTproBNP:

NT-proB-type natriuretic peptide

ST2:

Suppression of tumorigenicity 2 protein

ANGPT2:

Angiopoietin-2

OPN:

Osteopontin

ASGR1:

Asialoglycoprotein receptor 1

TNFR1:

Tumor necrosis factor receptor 1

TNFRSF10A:

Tumor necrosis factor receptor superfamily member 10A

KIM1:

Kidney injury molecule-1

References

  1. Rhodes A, et al. Surviving sepsis campaign: international guidelines for management of sepsis and septic shock: 2016. Crit Care Med. 2017;45:486–552.

    Article  Google Scholar 

  2. Singer M, et al. The third international consensus definitions for sepsis and septic shock (Sepsis-3). JAMA. 2016;315:801–10.

    Article  CAS  Google Scholar 

  3. Blanco J, et al. Incidence, organ dysfunction and mortality in severe sepsis: a Spanish multicentre study. Crit Care. 2008;12:R158.

    Article  Google Scholar 

  4. Hotchkiss RS, et al. Apoptotic cell death in patients with sepsis, shock, and multiple organ dysfunction. Crit Care Med. 1999;27:1230–51.

    Article  CAS  Google Scholar 

  5. Wiersinga WJ, Leopold SJ, Cranendonk DR, van der Poll T. Host innate immune responses to sepsis. Virulence. 2014;5:36–44.

    Article  Google Scholar 

  6. Singer M, De Santis V, Vitale D, Jeffcoate W. Multiorgan failure is an adaptive, endocrine-mediated, metabolic response to overwhelming systemic inflammation. Lancet. 2004;364:545–8.

    Article  Google Scholar 

  7. Hotchkiss RS, Monneret G, Payen D. Sepsis-induced immunosuppression: from cellular dysfunctions to immunotherapy. Nat Rev Immunol. 2013;13:862–74.

    Article  CAS  Google Scholar 

  8. Deutschman CS, Tracey KJ. Sepsis: current dogma and new perspectives. Immunity. 2014;40:463–75.

    Article  CAS  Google Scholar 

  9. Angus DC, van der Poll T. Severe sepsis and septic shock. N Engl J Med. 2013;369:840–51.

    Article  CAS  Google Scholar 

  10. Fraser DD, et al. Novel Outcome Biomarkers Identified With Targeted Proteomic Analyses of Plasma From Critically Ill Coronavirus Disease 2019 Patients. Crit Care Explor. 2020;2: e0189.

    Article  Google Scholar 

  11. Cao Z, Robinson RA. (2014) The role of proteomics in understanding biological mechanisms of sepsis. PROTEOMICS–Clinical Applications 8: 35–52.

  12. Shen Z, et al. Sepsis plasma protein profiling with immunodepletion, three-dimensional liquid chromatography tandem mass spectrometry, and spectrum counting. J Proteome Res. 2006;5:3154–60.

    Article  CAS  Google Scholar 

  13. Kalenka A, et al. Changes in the serum proteome of patients with sepsis and septic shock. Anesth Analg. 2006;103:1522–6.

    Article  CAS  Google Scholar 

  14. Paiva RAd, David CM, Domont GB. Proteomics in sepsis: a pilot study. Revista Brasileira de terapia intensiva. 2010;22:403–12.

    Article  Google Scholar 

  15. Triantafilou M, et al. Serum proteins modulate lipopolysaccharide and lipoteichoic acid-induced activation and contribute to the clinical outcome of sepsis. Virulence. 2012;3:136–45.

    Article  Google Scholar 

  16. Soares AJ, et al. Differential proteomics of the plasma of individuals with sepsis caused by Acinetobacter baumannii. J Proteomics. 2009;73:267–78.

    Article  CAS  Google Scholar 

  17. Cao Z, Yende S, Kellum JA, Angus DC, Robinson RA. Proteomics reveals age-related differences in the host immune response to sepsis. J Proteome Res. 2014;13:422–32.

    Article  CAS  Google Scholar 

  18. Lundberg M, Eriksson A, Tran B, Assarsson E, Fredriksson S. Homogeneous antibody-based proximity extension assays provide sensitive and specific detection of low-abundant proteins in human blood. Nucleic Acids Res. 2011;39: e102.

    Article  CAS  Google Scholar 

  19. Assarsson E, et al. Homogenous 96-plex PEA immunoassay exhibiting high sensitivity, specificity, and excellent scalability. PLoS ONE. 2014;9: e95192.

    Article  Google Scholar 

  20. Gillio-Meina C, Cepinskas G, Cecchini EL, Fraser DD. Translational research in pediatrics II: blood collection, processing, shipping, and storage. Pediatrics. 2013;131:754–66.

    Article  Google Scholar 

  21. Brisson AR, Matsui D, Rieder MJ, Fraser DD. Translational research in pediatrics: tissue sampling and biobanking. Pediatrics. 2012;129:153–62.

    Article  Google Scholar 

  22. Fraser DD, et al. Inflammation profiling of critically ill coronavirus disease 2019 patients. Crit Care Explor. 2020. https://doi.org/10.1097/CCE.0000000000000144.

    Article  Google Scholar 

  23. Van der Maaten L, Hinton G: Visualizing data using t-SNE. J Mach Learn Res. 2008; 9:2579–605.

  24. Tang C, Garreau D, von Luxburg U: When do random forests fail? Proceedings of the 32nd International Conference on Neural Information Processing Systems. December 2018: 2987–997.

  25. Pedregosa F, et al. Scikit-learn: machine learning in python. J Mach Learn Res. 2011;12:2825–30.

    Google Scholar 

  26. Kursa MB, Rudnicki WR. Feature selection with the Boruta package. J Stat Softw. 2010;36:1–13.

    Article  Google Scholar 

  27. Goulden R, et al. qSOFA, SIRS and NEWS for predicting inhospital mortality and ICU admission in emergency admissions treated as sepsis. Emerg Med J. 2018;35:345–9.

    Article  Google Scholar 

  28. Ginde AA, et al. Age-related differences in biomarkers of acute inflammation during hospitalization for sepsis. Shock (Augusta, Ga). 2014;42:99.

    Article  CAS  Google Scholar 

  29. Scumpia PO, Moldawer LL. Biology of interleukin-10 and its regulatory roles in sepsis syndromes. Crit Care Med. 2005;33:S468–71.

    Article  Google Scholar 

  30. Oberholzer A, Oberholzer C, Moldawer LL. Interleukin-10: a complex role in the pathogenesis of sepsis syndromes and its potential as an anti-inflammatory drug. Crit Care Med. 2002;30:S58–63.

    Article  CAS  Google Scholar 

  31. Madtes DK, et al. Elevated transforming growth factor-α levels in bronchoalveolar lavage fluid of patients with acute respiratory distress syndrome. Am J Respir Crit Care Med. 1998;158:424–30.

    Article  CAS  Google Scholar 

  32. Deneault E, et al. CNTN5-/+ or EHMT2-/+ human iPSC-derived neurons from individuals with autism develop hyperactive neuronal networks. Elife. 2019. https://doi.org/10.7554/eLife.40092.

    Article  Google Scholar 

  33. Dauar MT, et al. CNTN5 is associated with disease risk and pathology throughout the Alzheimer’s disease continuum. Alzheimers Dement. 2021;17: e052359.

    Google Scholar 

  34. Vogt L, et al. VSIG4, a B7 family–related protein, is a negative regulator of T cell activation. J Clin Investig. 2006;116:2817–26.

    Article  CAS  Google Scholar 

  35. Zanotti S, Kumar A, Kumar A. Cytokine modulation in sepsis and septic shock. Expert Opin Investig Drugs. 2002;11:1061–75.

    Article  CAS  Google Scholar 

  36. Montoya-Ruiz C, et al. Variants in LTA, TNF, IL1B and IL10 genes associated with the clinical course of sepsis. Immunol Res. 2016;64:1168–78.

    Article  CAS  Google Scholar 

  37. Gyawali B, Ramakrishna K, Dhamoon AS. Sepsis: The evolution in definition, pathophysiology, and management. SAGE Open Med. 2019;7:2050312119835043.

    Article  Google Scholar 

  38. Abraham E, Singer M. Mechanisms of sepsis-induced organ dysfunction. Crit Care Med. 2007;35:2408–16.

    Article  Google Scholar 

  39. Wong DT, Gomez M, McGuire GP, Kavanagh B. Utilization of intensive care unit days in a Canadian medical-surgical intensive care unit. Crit Care Med. 1999;27:1319–24.

    Article  CAS  Google Scholar 

  40. Kemperman H, et al. Osteoprotegerin is higher in sepsis than in noninfectious SIRS and predicts 30-day mortality of SIRS patients in the intensive care. J Appl Lab Med. 2019;3:559–68.

    Article  CAS  Google Scholar 

  41. Baud’huin M, et al. Osteoprotegerin: multiple partners for multiple functions. Cytokine Growth Factor Rev. 2013;24:401–9.

    Article  CAS  Google Scholar 

  42. Schaalan M, Mohamed W. Predictive ability of circulating osteoprotegerin as a novel biomarker for early detection of acute kidney injury induced by sepsis. Eur Cytokine Netw. 2017;28:52–62.

    Article  CAS  Google Scholar 

  43. Steiner J, Guglin M. BNP or NTproBNP? A clinician’s perspective. Int J Cardiol. 2008;129:5–14.

    Article  Google Scholar 

  44. Hall C. NT-ProBNP: the mechanism behind the marker. J Card Fail. 2005;11:S81-83.

    Article  CAS  Google Scholar 

  45. Mueller C, Breidthardt T, Laule-Kilian K, Christ M, Perruchoud AP. The integration of BNP and NT-proBNP into clinical medicine. Swiss Med Wkly. 2007;137:4–12.

    CAS  Google Scholar 

  46. Zhao T, Su Z, Li Y, Zhang X, You Q. Chitinase-3 like-protein-1 function and its role in diseases. Signal Transduct Target Ther. 2020;5:1–20.

    Google Scholar 

  47. Kornblit B, et al. Plasma YKL-40 and CHI3L1 in systemic inflammation and sepsis—Experience from two prospective cohorts. Immunobiology. 2013;218:1227–34.

    Article  CAS  Google Scholar 

  48. Kronborg G, et al. Serum level of YKL-40 is elevated in patients with Streptococcus pneumoniae bacteremia and is associated with the outcome of the disease. Scand J Infect Dis. 2002;34:323–6.

    Article  Google Scholar 

  49. Varner JA, Cheresh DA. Integrins and cancer. Curr Opin Cell Biol. 1996;8:724–30.

    Article  CAS  Google Scholar 

  50. Khurana S, et al. Outside-in integrin signalling regulates haematopoietic stem cell function via Periostin-Itgav axis. Nat Commun. 2016;7:1–14.

    Article  Google Scholar 

  51. Morandi EM, et al. ITGAV and ITGA5 diversely regulate proliferation and adipogenic differentiation of human adipose derived stem cells. Sci Rep. 2016;6:28889.

    Article  CAS  Google Scholar 

  52. Yamaji Y, et al. TEM7 (PLXDC1) in neovascular endothelial cells of fibrovascular membranes from patients with proliferative diabetic retinopathy. Invest Ophthalmol Vis Sci. 2008;49:3151–7.

    Article  Google Scholar 

  53. Cheng G, et al. Identification of PLXDC1 and PLXDC2 as the transmembrane receptors for the multifunctional factor PEDF. Elife. 2014;3: e05401.

    Article  Google Scholar 

  54. Herridge MS. Prognostication and intensive care unit outcome: the evolving role of scoring systems. Clin Chest Med. 2003;24:751–62.

    Article  Google Scholar 

  55. Murphy SM, et al. GCP5 and GCP6: two new members of the human gamma-tubulin complex. Mol Biol Cell. 2001;12:3340–52.

    Article  CAS  Google Scholar 

  56. Cosgrove GP, et al. Pigment epithelium–derived factor in idiopathic pulmonary fibrosis: a role in aberrant angiogenesis. Am J Respir Crit Care Med. 2004;170:242–51.

    Article  Google Scholar 

  57. Shin ES, Sorenson CM, Sheibani N. PEDF expression regulates the proangiogenic and proinflammatory phenotype of the lung endothelium. Am J Physiol Lung Cell Mol Physiol. 2014;306:L620–34.

    Article  CAS  Google Scholar 

  58. Li L, et al. Pigment epithelial-derived factor (PEDF)-triggered lung cancer cell apoptosis relies on p53 protein-driven Fas ligand (Fas-L) up-regulation and Fas protein cell surface translocation. J Biol Chem. 2014;289:30785–99.

    Article  CAS  Google Scholar 

  59. Li X, et al. Elevated plasma levels of pigment epithelium-derived factor correlated with inflammation and lung function in COPD patients. Int J Chron Obstruct Pulmon Dis. 2015;10:587.

    Google Scholar 

  60. Nakamura T, Yamagishi S-I. PEDF and septic shock. Curr Mol Med. 2010;10:312–6.

    Article  CAS  Google Scholar 

  61. He T, et al. Pigment epithelium-derived factor regulates microvascular permeability through adipose triglyceride lipase in sepsis. Clin Sci. 2015;129:49–61.

    Article  CAS  Google Scholar 

  62. Haney MJ, et al. TPP1 delivery to lysosomes with extracellular vesicles and their enhanced brain distribution in the animal model of batten disease. Adv Healthc Mater. 2019;8: e1801271.

    Article  Google Scholar 

  63. Shlevkov E, et al. A high-content screen identifies TPP1 and aurora B as regulators of axonal mitochondrial transport. Cell Rep. 2019;28:3224-3237.e3225.

    Article  CAS  Google Scholar 

  64. Schulz A, et al. Study of intraventricular cerliponase alfa for CLN2 disease. N Engl J Med. 2018;378:1898–907.

    Article  CAS  Google Scholar 

  65. Schutyser E, Struyf S, Van Damme J. The CC chemokine CCL20 and its receptor CCR6. Cytokine Growth Factor Rev. 2003;14:409–26.

    Article  CAS  Google Scholar 

  66. Calvayrac O, et al. CCL20 is increased in hypercholesterolemic subjects and is upregulated by LDL in vascular smooth muscle cells: role of NF-κB. Arterioscler Thromb Vasc Biol. 2011;31:2733–41.

    Article  CAS  Google Scholar 

  67. Soto B, et al. Circulating CCL20 as a new biomarker of abdominal aortic aneurysm. Sci Rep. 2017;7:17331.

    Article  CAS  Google Scholar 

  68. Patterson EK, et al. Proteinase 3 contributes to endothelial dysfunction in an experimental model of sepsis. Exp Biol Med (Maywood). 2021. https://doi.org/10.1177/15353702211029284.

    Article  Google Scholar 

  69. Almansa R, et al. Quantification of immune dysregulation by next-generation polymerase chain reaction to improve sepsis diagnosis in surgical patients. Ann Surg. 2019;269:545–53.

    Article  Google Scholar 

  70. Zhang S, Li N, Chen W, Fu Q, Liu Y. Time series gene expression profiles analysis identified several potential biomarkers for sepsis. DNA Cell Biol. 2020;39:1862–71.

    Article  CAS  Google Scholar 

  71. Martin KR, Witko-Sarsat V. Proteinase 3: the odd one out that became an autoantigen. J Leukoc Biol. 2017;102:689–98.

    Article  CAS  Google Scholar 

  72. Tamassia N, et al. Cytokine production by human neutrophils: Revisiting the “dark side of the moon.” Eur J Clin Invest. 2018;48(Suppl 2): e12952.

    Article  Google Scholar 

  73. Schiffer L, et al. Elevation of serum CXCL13 in SLE as well as in sepsis. Lupus. 2011;20:507–11.

    Article  CAS  Google Scholar 

  74. Kazanietz MG, Durando M, Cooke M. CXCL13 and its receptor CXCR5 in cancer: inflammation, immune response, and beyond. Front Endocrinol (Lausanne). 2019;10:471.

    Article  Google Scholar 

  75. Ardehali MB, Damle M, Perea-Resa C, Blower MD, Kingston RE. Elongin A associates with actively transcribed genes and modulates enhancer RNA levels with limited impact on transcription elongation rate in vivo. J Biol Chem. 2020. https://doi.org/10.1074/jbc.RA120.015877.

    Article  Google Scholar 

  76. Lakkaraju AK, Mary C, Scherrer A, Johnson AE, Strub K. SRP keeps polypeptides translocation-competent by slowing translation to match limiting ER-targeting sites. Cell. 2008;133:440–51.

    Article  CAS  Google Scholar 

  77. Continolo S, et al. The proto-oncogene Fgr regulates cell migration and this requires its plasma membrane localization. Exp Cell Res. 2005;302:253–69.

    Article  CAS  Google Scholar 

  78. Meng M, Sang L, Wang X. S100 calcium binding protein A11 (S100A11) promotes the proliferation, migration and invasion of cervical cancer cells, and activates wnt/β-catenin signaling. Onco Targets Ther. 2019;12:8675–85.

    Article  CAS  Google Scholar 

  79. Liu L, et al. S100A11 regulates renal carcinoma cell proliferation, invasion, and migration via the EGFR/Akt signaling pathway and E-cadherin. Tumour Biol. 2017;39:1010428317705337.

    Article  Google Scholar 

  80. Nakano E, et al. Variants in pancreatic carboxypeptidase genes CPA2 and CPB1 are not associated with chronic pancreatitis. Am J Physiol Gastrointest Liver Physiol. 2015;309:G688–94.

    Article  CAS  Google Scholar 

  81. David S, et al. Angiopoietin-2 may contribute to multi-organ dysfunction and death in sepsis. Crit Care Med. 2012;40:3034.

    Article  CAS  Google Scholar 

  82. Scholz A, Plate KH, Reiss Y. Angiopoietin-2: a multifaceted cytokine that functions in both angiogenesis and inflammation. Ann NY Acad Sci. 2015;1347:45–51.

    Article  CAS  Google Scholar 

  83. Schultz G, Clark W, Rotatori DS. EGF and TGF-α in wound healing and repair. J Cell Biochem. 1991;45:346–52.

    Article  CAS  Google Scholar 

  84. Singh B, Coffey RJ. From wavy hair to naked proteins: the role of transforming growth factor alpha in health and disease. Semin Cell Dev Biol. 2014. https://doi.org/10.1016/j.semcdb.2014.03.003.

    Article  Google Scholar 

  85. James NE, et al. The biomarker HE4 (WFDC2) promotes a pro-angiogenic and immunosuppressive tumor microenvironment via regulation of STAT3 target genes. Sci Rep. 2020;10:8558.

    Article  CAS  Google Scholar 

  86. Hellström I, et al. The HE4 (WFDC2) protein is a biomarker for ovarian carcinoma. Can Res. 2003;63:3695–700.

    Google Scholar 

  87. Chen Y, et al. WFDC2 contributes to epithelial-mesenchymal transition (EMT) by activating AKT signaling pathway and regulating MMP-2 expression. Cancer Manag Res. 2019;11:2415–24.

    Article  CAS  Google Scholar 

  88. Nakagawa S, et al. Molecular markers of tubulointerstitial fibrosis and tubular cell damage in patients with chronic kidney disease. PLoS ONE. 2015;10: e0136994.

    Article  Google Scholar 

  89. Korbut AI, Klimontov VV, Vinogradov IV, Romanov VV. Risk factors and urinary biomarkers of non-albuminuric and albuminuric chronic kidney disease in patients with type 2 diabetes. World J Diabetes. 2019;10:517.

    Article  Google Scholar 

  90. Bingle L, et al. WFDC2 (HE4): a potential role in the innate immunity of the oral cavity and respiratory tract and the development of adenocarcinomas of the lung. Respir Res. 2006;7:1–10.

    Article  Google Scholar 

  91. D’Alonzo D, De Fenza M, Pavone V. COVID-19 and pneumonia: a role for the uPA/uPAR system. Drug Discov Today. 2020;25:1528–34.

    Article  CAS  Google Scholar 

  92. Lund SA, Giachelli CM, Scatena M. The role of osteopontin in inflammatory processes. J Cell Commun Signal. 2009;3:311–22.

    Article  Google Scholar 

  93. Castello LM, et al. The role of osteopontin as a diagnostic and prognostic biomarker in sepsis and septic shock. Cells. 2019;8:174.

    Article  CAS  Google Scholar 

  94. Hoober JK. ASGR1 and its enigmatic relative, CLEC10A. Int J Mol Sci. 2020. https://doi.org/10.3390/ijms21144818.

    Article  Google Scholar 

  95. Woźnica EA, Inglot M, Woźnica RK, Łysenko L. Liver dysfunction in sepsis. Adv Clin Exp Med. 2018;27:547–51.

    Article  Google Scholar 

  96. Levi M, van der Poll T. Coagulation and sepsis. Thromb Res. 2017;149:38–44.

    Article  CAS  Google Scholar 

  97. Song J, et al. Understanding kidney injury molecule 1: a novel immune factor in kidney pathophysiology. Am J Transl Res. 2019;11:1219.

    CAS  Google Scholar 

  98. Ismail OZ, et al. Kidney injury molecule-1 protects against Gα12 activation and tissue damage in renal ischemia-reperfusion injury. Am J Pathol. 2015;185:1207–15.

    Article  CAS  Google Scholar 

  99. Lee JY, et al. Donor kidney injury molecule-1 promotes graft recovery by regulating systemic necroinflammation. Am J Transplant. 2018;18:2021–8.

    Article  CAS  Google Scholar 

  100. Lee JC, et al. Kidney injury molecule-1 inhibits metastasis of renal cell carcinoma. Sci Rep. 2021;11:1–12.

    Google Scholar 

  101. Han WK, Bailly V, Abichandani R, Thadhani R, Bonventre JV. Kidney Injury Molecule-1 (KIM-1): a novel biomarker for human renal proximal tubule injury. Kidney Int. 2002;62:237–44.

    Article  CAS  Google Scholar 

  102. Wang K, et al. Biomarkers of sepsis-induced acute kidney injury. BioMed Res Int. 2018. https://doi.org/10.1155/2018/6937947.

    Article  Google Scholar 

  103. Wajant H, Pfizenmaier K, Scheurich P. Tumor necrosis factor signaling. Cell Death Differ. 2003;10:45–65.

    Article  CAS  Google Scholar 

  104. Spooner CE, Markowitz NP, Saravolatz LD. The role of tumor necrosis factor in sepsis. Clin Immunol Immunopathol. 1992;62:S11–7.

    Article  CAS  Google Scholar 

  105. Oppert M, et al. Acute renal failure in patients with severe sepsis and septic shock—a significant independent risk factor for mortality: results from the German Prevalence Study. Nephrol Dial Transplant. 2008;23:904–9.

    Article  Google Scholar 

  106. Neveu H, et al. Prognostic factors in acute renal failure due to sepsis. Results of a prospective multicentre study. Nephrol Dial Transplant. 1996;11:293–9.

    Article  CAS  Google Scholar 

  107. Ciccone MM, et al. A novel cardiac bio-marker: ST2: a review. Molecules. 2013;18:15314–28.

    Article  CAS  Google Scholar 

  108. Griesenauer B, Paczesny S. The ST2/IL-33 axis in immune cells during inflammatory diseases. Front Immunol. 2017;8:475.

    Article  Google Scholar 

  109. Hoogerwerf JJ, et al. Soluble ST2 plasma concentrations predict mortality in severe sepsis. Intensive Care Med. 2010;36:630–7.

    Article  CAS  Google Scholar 

  110. Choi Y-S, et al. Interleukin-33 induces angiogenesis and vascular permeability through ST2/TRAF6-mediated endothelial nitric oxide production. Blood. 2009;114:3117–26.

    Article  CAS  Google Scholar 

  111. Gibbison B, et al. Corticosteroids in septic shock: a systematic review and network meta-analysis. Crit Care. 2017;21:78.

    Article  Google Scholar 

  112. Gomez H, Kellum JA. Lactate in sepsis. JAMA. 2015;313:194–5.

    Article  CAS  Google Scholar 

  113. Andersen LW, et al. Etiology and therapeutic approach to elevated lactate levels. Mayo Clin Proc. 2013. https://doi.org/10.1016/j.mayocp.2013.06.012.

    Article  Google Scholar 

  114. Pierrakos C, Vincent JL. Sepsis biomarkers: a review. Crit Care. 2010;14:R15.

    Article  Google Scholar 

  115. Phua J, et al. Characteristics and outcomes of culture-negative versus culture-positive severe sepsis. Crit Care. 2013;17:1–12.

    Article  Google Scholar 

  116. Gupta S, et al. Culture-negative severe sepsis: nationwide trends and outcomes. Chest. 2016;150:1251–9.

    Article  Google Scholar 

Download references

Acknowledgements

We are grateful to the front-line health care providers who assisted with collection of blood samples from acutely ill sepsis patients.

Funding

DDF received study funding from the London Health Sciences Foundation (https://lhsf.ca/), the London Community Foundation and the AMOSO Innovation Fund.

Author information

Authors and Affiliations

  1. Medicine, Western University, London, ON, Canada

    Logan R. Van Nynatten, Marat Slessarev, Claudio M. Martin & Aleks Leligdowicz

  2. Lawson Health Research Institute, London, ON, Canada

    Marat Slessarev, Claudio M. Martin, Aleks Leligdowicz, Michael R. Miller, Mark Daley, Eric K. Patterson, Gediminas Cepinskas & Douglas D. Fraser

  3. Pediatrics, Western University, London, ON, Canada

    Michael R. Miller & Douglas D. Fraser

  4. Computer Science, Western University, London, ON, N6A 3K7, Canada

    Maitray A. Patel & Mark Daley

  5. The Vector Institute for Artificial Intelligence, Toronto, ON, M5G 1M1, Canada

    Mark Daley

  6. Medical Biophysics, Western University, London, ON, N6A 3K7, Canada

    Gediminas Cepinskas

  7. Clinical Neurological Sciences, Western University, London, ON, Canada

    Douglas D. Fraser

  8. Physiology and Pharmacology, Western University, London, ON, Canada

    Douglas D. Fraser

  9. London Health Sciences Centre, Room C2-C82, 800 Commissioners Road East, London, ON, N6A 5W9, Canada

    Douglas D. Fraser

Authors
  1. Logan R. Van Nynatten
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Marat Slessarev
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Claudio M. Martin
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Aleks Leligdowicz
    View author publications

    You can also search for this author in PubMed Google Scholar

  5. Michael R. Miller
    View author publications

    You can also search for this author in PubMed Google Scholar

  6. Maitray A. Patel
    View author publications

    You can also search for this author in PubMed Google Scholar

  7. Mark Daley
    View author publications

    You can also search for this author in PubMed Google Scholar

  8. Eric K. Patterson
    View author publications

    You can also search for this author in PubMed Google Scholar

  9. Gediminas Cepinskas
    View author publications

    You can also search for this author in PubMed Google Scholar

  10. Douglas D. Fraser
    View author publications

    You can also search for this author in PubMed Google Scholar

Contributions

DDF conceived and designed the study. DDF, LRVN, MS and CM collected human samples and clinical data. DDF, AL, MRM, MAP, MD, EKP and GC analyzed all data. LRVN and DDF wrote the manuscript with input from all other authors. All authors read and approved the final manuscript.

Corresponding author

Correspondence to Douglas D. Fraser.

Ethics declarations

Ethics approval and consent to participate

This study was approved by the Western University, Human Research Ethics Board (HREB): acutely ill sepsis patients (HREB #6970, renewed March 17, 2021) and volunteer healthy control subjects (HREB #16986E, renewed March 9, 2021).

Consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Additional file 1: Table S1.

Protein function. Table S2. Median plasma expression of the top classifying proteins from sepsis patients for ICU Day-1 and ICU Day-3. Table S3. Correlations of dynamic change (Day 3 minus Day 1) between normalized plasma protein expression and sepsis patient parameters.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Van Nynatten, L.R., Slessarev, M., Martin, C.M. et al. Novel plasma protein biomarkers from critically ill sepsis patients. Clin Proteom 19, 50 (2022). https://doi.org/10.1186/s12014-022-09389-3

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/s12014-022-09389-3

Keywords

Clinical Proteomics

ISSN: 1559-0275