Background
The immune system plays a pivotal role in maintaining the balance between health and disease. Profiling immunological perturbation holds potential for elucidating the pathogenesis of a wide range of diseases. Currently available high-throughput profiling technologies and emerging systems immunology analysis approaches enable the study of clinical samples on a system-wide scale and provide unbiased tools for investigation of immune responses[
1,
2]. Whole blood transcriptome profiling has been employed to investigate a wide range of conditions[
3‐
5]. Plasma, which is a valuable source of potential biomarkers, is an attractive alternative for profiling molecular changes associated with disease pathogenesis and progression on a systems scale. However, robust, cost-effective and reproducible technologies needed for measuring plasma protein abundance on a systems scale are still lacking. Most prevalent is mass spectrometry, however this lacks well-established reference databases and is biased toward detecting high-concentration compounds, which are major limitations for assessment of the plasma proteome by this technology[
1,
6].
So-called “transcriptomic reporter assays” provide an alternate means to assess perturbations in plasma on a system-wide scale[
7]. This strategy consists of measuring whole genome transcriptional responses elicited in reporter cells exposed
in vitro to patient plasma. This type of approach has already proven useful in studies of several immunologically mediated diseases. It was employed to help unravel the pathogenesis of systemic onset juvenile idiopathic arthritis, eventually leading to the adoption of a novel therapeutic modality for treatment of this disease[
7,
8]. It has been used to identify candidate biomarker signatures in patients prior to the clinical onset of type 1 diabetes mellitus[
9]. It has contributed to identifying pathways of pancreatic islet cell destruction in islet cell transplantation[
10,
11]. Despite these and other successes, this approach has not yet been widely explored or adopted.
Sepsis is a clinical syndrome related to dysregulated systemic inflammation in response to an underlying infection. Uncontrolled production of cytokines and chemokines is believed to play a role in sepsis severity[
12]. Early recognition leading to targeted antimicrobial and supportive therapy is critical to survival and each hour that treatment is delayed can markedly increase mortality[
12‐
14]. However, due to an incomplete understanding of sepsis pathogenesis, criteria for rapid diagnosis and severity assessment are limited and based largely on non-specific clinical signs of systemic inflammation and organ dysfunction[
15,
16]. Several biomarkers have been studied in attempts to provide more simple, rapid, and accurate methods for diagnosis and prognosis of sepsis[
17]. These include C-reactive protein, procalcitonin, triggering receptor expressed on myeloid cells 1 (TREM-1) and others[
17‐
19]. While several small studies have shown correlation of such proteins to sepsis severity and outcomes, these proteins have not proven reliable on a larger scale and are not routinely used in clinical practice[
18,
20]. Recent studies suggest that combined use of multiple biomarkers may be more accurate, however at present these remain investigational[
19].
Our study evaluated responses of three different cell types to stimulation with septic plasma: polymorphonuclear cells (PMNs), peripheral blood mononuclear cells (PBMCs), and monocyte-derived dendritic cells (MoDCs). PMNs and PBMCs were selected as these constitute the primary types of leukocytes in peripheral blood and are key in control of infections. MoDCs were selected because they are known to play a central role in the immune system and are able to respond to diverse immune signals. Each of these cell populations functioned as a so-called “reporter cell system” to investigate the transcriptional response to septic plasma. We demonstrate the utility of a reporter cell system for elucidating immune pathogenesis of a complex disease such as sepsis and the potential relevance of this approach for predicting prognosis in sepsis.
Methods
Ethics statement
The study was approved by the ethical review committees of Khon Kaen University and Khon Kaen Regional Hospital (Khon Kaen, Thailand) and the Institutional Review Board of Benaroya Research Institute (Seattle, WA). Participants provided written informed consent to participate in this study. Written informed consent was obtained from parents or guardians on behalf of the minor/child participant in this study.
Plasma collection
Septic patients were enrolled from Khon Kaen Regional Hospital, Khon Kaen, Thailand. Patients who met at least two of the criteria for severe inflammatory response syndrome (SIRS) were enrolled in the study[
3,
15]. As part of the routine investigations, clinical specimens were collected for bacterial culture within 24 h following SIRS diagnosis. Only blood samples obtained from patients who were retrospectively diagnosed with culture-proven sepsis were retained for further analyses. Patients with negative blood cultures were excluded. Severe sepsis was defined according to the current guidelines from the Surviving Sepsis Campaign[
15]. These criteria include several clinical and laboratory findings of sepsis-induced tissue hypoperfusion or organ dysfunction. We used the subset of these criteria for which the necessary data had been collected for our patient cohort: elevation in creatinine to >2.0 mg/dl, elevation in bilirubin to >2.0 mg/dl, platelet count <100,000/μL, and sepsis induced hypotension (septic shock). In hospital death was also counted as severe sepsis. Demographic and clinical data were recorded for all subjects (Additional files
1,
2, and
3). Uninfected healthy controls were selected as individuals who had no signs of acute infectious diseases during the previous 3 months or at the time of the study. Uninfected controls also had to have normal blood counts, normal fasting blood glucose, and normal glycosylated hemoglobin. Three milliliters of whole blood was collected from each patient and healthy control into heparinized tubes (BD Biosciences). For sepsis patients, samples were grouped as drawn either in the first 48 h of admission, or at >48 h after admission. To separate plasma, blood samples were centrifuged at 2,000 rpm for 10 minutes and the plasma component was transferred into a cryogenic vial and stored at -80°C until used.
PMNs and PBMCs isolations from healthy volunteers
Blood samples from three additional healthy volunteers were used in subsequent cell isolation procedures. PMNs were isolated from heparinized venous blood by 3.0% dextran T-500 sedimentation and Ficoll-Paque PLUS centrifugation (Amersham Biosciences) as previously described[
21]. The purity of isolated cells was generally more than 95% as determined by flow cytometry (FACSCalibur, Becton Dickinson)[
22]. PBMCs were isolated from whole blood samples by centrifugation through a Ficoll-Paque Plus (Sigma Aldrich) density gradient.
Generation of MoDCs
A portion of the isolated PBMCs was subsequently used for MoDCs generation as previously described[
22,
23]. MoDCs were harvested and resuspended in serum-free RPMI-1640 medium (Gibco), 5 × 10
5 cells/well were plated into a 24-well tissue culture plate (Corning) for 24 h. The resulting cells were determined to be >95% CD11c
+ by flow cytometry.
Cell culture
Cell cultures were performed as described by Pascual
et al.[
7]. Two million PMNs or one million PBMCs were resuspended in serum-free RPMI-1640 medium (Gibco) and added to either 5 ml or 2 ml culture tubes (Becton Dickinson), respectively. Five hundred thousand MoDCs were seeded into 24 well tissue culture plates (Corning) at 1 × 10
6 cells/ml and rested for 24 h before the experiments. Cells were cultured with medium alone or a plasma sample in a final concentration of 20%. After 6 h incubation at 37°C in 5% CO
2, cells were harvested, washed twice with phosphate buffered saline, homogenized in RLT buffer (RNeasy mini kit; QIAGEN), and stored at -80°C until use.
RNA preparation and microarray
Total RNA was isolated using the RNeasy Mini kit (QIAGEN) according to the manufacturer’s instructions. RNA integrity number (RIN) was determined by using an Agilent 2100 Bioanalyzer (Agilent). Qualified samples (RIN >6 or presence of 28s and 18s rRNA) were retained for further processing. Total RNA was amplified and labeled using the Illumina TotalPrep RNA Amplification Kit (Ambion). Labeled cRNA was hybridized overnight to Human HT-12 V4 BeadChip array (IIlumina), washed, blocked, stained and scanned on an Illumina HiScan instrument following the manufacturer’s protocols.
Data acquisition and background subtraction
GenomeStudio was used to generate signal intensity values from the scans and perform background subtraction. Post-hybridization quality controls were done by the standard metrics provided by the manufacturer. Data from each cell type and each culture experiment were processed independently. All possible outliers were excluded from the expression data set by metrics for post-hybridization quality controls.
Data normalization
All data analyses were performed using R (version 2.14.0;
http://cran.r-project.org/bin/windows/base/old/2.14.0/). Data pre-processing of background subtracted data was performed by using the preprocessCore package from Bioconductor. Pre-processing included rescaling intensity by quantile normalization. After normalization, expressions were floored with intensities <10 set to 10. Transcripts with detection p-value of less than or equal to 0.01 in at least one sample (PALO) were selected for further analysis. Samples from the same cell type and batch were normalized to the average intensity of samples cultured in medium alone. A filter was set to include only transcripts that had at least two-fold changes and 100 intensity differences compared to medium control. Background subtracted and processed data from these experiments have been deposited at NCBI’s Gene Expression Omnibus database (
http://www.ncbi.nlm.nih.gov/geo/), with accession numbers GSE49758. To facilitate data sharing and interactive data analysis, we created a data portal (
https://gxb.benaroyaresearch.org/tra/tra-paper/tra-landing.gsp) to store and analyze background subtracted data from all three experiments (see[
24] for tutorial).
Unsupervised analysis
Principle component analysis (PCA) was performed using the R function “prcomp”. The first two principal components, PC1 and PC2, were plotted against each other. Each colored dot represents an individual sample. Euclidean distances were calculated by measuring the distance from each sample to the average of samples stimulated with uninfected plasma. The comparison between severe and non-severe sepsis was performed using the Mann-Whitney U-test. Hierarchical clustering analysis was performed using the function “heatmap.2” from the R package “gplots”. Euclidean distance and complete linkage methods were used by default.
Feature selection
Transcripts that were differentially expressed between study groups were selected by Random k-Nearest Neighbor – Feature Selection (RKNN-FS) using the R package “rknn”[
25]. A RKNN classifier consists of an ensemble of base k-nearest neighbor models (number of neighbors = 5), each constructed from a random subset of the input variables. Features were selected by ranking the importance of the PALO transcripts.
Pathway analysis
Gene ontology analyses were performed using GeneGo MetaCore pathway analysis tool (Thomson Reuters, NY). The default background gene list was used for all enrichment analyses including process networks, pathway map folders, and pathway maps. Pathway maps are the collection of pathways grouped into folders according to main cell processes, protein functions, and diseases. Map Folders are the collection of pathways grouped into folders according to main biological processes. Statistical significance was ascertained by using a threshold of false discover rate (FDR) <0.05. The network builder tool using the shortest possible path with no more than 2 steps was used to represent functional interactions. Upstream transcription factors were identified for lists of over-expressed genes using the Transcription Factors tool.
Class prediction
To determine the performance of our predictive signature, class prediction of binary variables was carried out by support vector machine (SVM; package “e1071”) and random forest (RF; package “randomForest”) algorithms. These machine learning methods are robust, well-accepted and commonly used methods for class prediction. Receiver operating characteristic (ROC) curves were constructed using the R package “ROCR”. Area under the curve (AUC) and confidence intervals were calculated using the R package “Hmisc”.
Discussion
This study reports the development and implementation of a “transcriptomic reporter assay” designed to investigate the immunogenicity of septic patient plasma and to our knowledge is the first published study employing immune cells in a whole genome transcriptomic reporter assay for an infectious process. One previous study investigated a focused transcriptional response of 1700 transcripts in cardiac myocytes cultured with septic serum[
46]. Whole transcriptomic reporter assays have been employed successfully in previous studies of autoimmune diseases and have contributed to the understanding of disease pathogenesis in systemic onset juvenile idiopathic arthritis and type 1 diabetes mellitus[
7‐
11]. Our work helps further extend this technique into the infectious disease field and also confirms the utility of comparing performance of multiple leukocyte subsets.
Transcriptomic reporter assays such as this one are based on the fact that plasma carries diverse circulating immune mediators. Stimulation of immune cells by plasma can demonstrate the biological processes triggered by the immune responses of the host[
7,
9]. In the context of infection, a transcriptomic reporter assay may be detecting in part the responses triggered by exogenous molecules including pathogen associated molecular patterns (PAMPs) such as LPS, lipoproteins, and peptidoglycans[
47]. However such an assay also reflects the responses triggered by diverse endogenous signaling molecules including cytokines and damage associated molecular patterns (DAMPs) among others[
48].
Our work identified that for investigation of sepsis, PMNs served as the best reporter cells in a side-by-side comparison with PBMCs and MoDCs. PMNs were better sensors of immunostimulatory factors present in plasma, displaying improved ability to discriminate septic from uninfected subjects, and PMNs mobilized the most robust immune transcriptional program. This finding was not initially expected given that PBMCs have been to date the preferred “serum sensing” cell reporter system[
9], and given that dendritic cells are well known for their sentinel role in the immune system and have ability to respond to a wide range of immune triggers. However our findings are consistent with the role of PMNs, which serve as the first line of the cellular innate immune response and are a major source of acute phase immune mediators[
49,
50]. Further our results support previous work investigating the robust transcriptional response of PMN in sepsis. For example, Wong
et al. demonstrated from leukocyte transcriptional profiling in pediatric septic shock that the number of differentially expressed genes in PMNs was greater than in monocytes or lymphocytes suggesting repression of adaptive immunity gene programs in early sepsis[
51]. De Kleijn
et al. demonstrated a robust set of functional gene networksdifferentially expressed by PMNs after both
in vivo and
ex vivo exposure to LPS that related to extended survival and the regulation of inflammatory responses[
52]. It should be noted that the arrays of receptors expressed by different immune cell types vary widely, and that while our report indicate that neutrophils are especially well equipped to respond to plasma from septic patients it may not be the most appropriate cell type in other settings.
The PMN cell reporter system coupled with whole transcriptome readout allowed identification of a severity signature for sepsis that was highly accurate in two independent datasets. Current guidelines for diagnosing sepsis and grading severity are based on multiple clinical parameters, which can be inaccurate and do not predict prognosis well[
15]. Although many potential biomarkers including cytokines, coagulation factors, and several others have been investigated for sepsis diagnosis and prognosis, none have proved reliable enough to enter routine clinical practice and so there is a need for better markers for risk stratification of septic patients to guide treatment and prognosis[
17‐
19]. Elucidating the biologic mechanisms that differ among sepsis patients will help advance this field[
12,
20]. The promising performance of the PMN transcriptomic reporter assay presented here to stratify patients with sepsis by severity offers a novel and attractive platform for the development of biomarker signatures in sepsis. Similarly, the ability to select a gene panel that was specific to stimulation with plasma from severe sepsis patients shows the utility of this method to better understand the immunopathogenesis of sepsis.
Additional investigation is warranted by these results. The use of three independent cohorts of patients demonstrates reproducibility, but investigation with profiling of longitudinal samples will be necessary to further validate our severity assessment and to determine its potential value in monitoring disease progression and the response to treatment. Seeing consistent results using three separate PMN donors suggests our results can be reproducible independent of donor source, however further studies are necessary to determine how much variability donor source could introduce. Similarly extending our study to stable cell lines such as the neutrophil-like HL-60 line could be useful to develop a standardized assay, an approach that has been previously investigated in the context of type 1 diabetes research[
53]. Perhaps most important, comparison of transcriptional profiles elicited from similar clinical conditions due to sterile inflammation (e.g. non-infectious SIRS or major surgery) is essential to determine the specificity of the severity signature described in this study. Also further work towards using this approach to identify a causative pathogen could hold potential – in our case the relatively small sample size and diversity of pathogens made it difficult to address this issue.
While at present a transcriptomic PMN reporter assay is not ideal for applications at the bedside given challenges for standardization of cell lines and data processing, technological advances in automation of sample processing and availability of polymerase chain reaction (PCR)-based amplification would make the implementation of a similar but more targeted assay feasible[
54,
55]. This approach could also serve as a novel platform for biomarker discovery and the development of novel clinical tests that could improve diagnosis and prognosis in sepsis. Moreover, this type of neutrophil transcriptomic reporter assay is likely to prove valuable for the investigation of other immunologically mediated diseases.
Acknowledgements
We thank all patients and volunteers who participated in the study, the staff of Khon Kaen Regional Hospital for patient recruitment, Brenda Norris for reviewing and submitting the manuscript, and all members of the Systems Immunology Division, BRI for helpful discussions. This work was supported by the National Institutes of Health (U01 Grant no. U01AI082110 to DC).
Competing interests
The authors declare that they have no competing interest.
Authors’ contributions
DC and GL designed and conceived research; DR, DS, SB, CK, KO, and QAN performed experiments. WS provided specimens and clinical data. PK, EW, MCA, SP, and MM analyzed data. VG contributed new reagents/analytic tools. PK, DR, LC, MCA, PL, GL, and DC wrote the paper. All authors read and approved the final manuscript.