supplementary appendix 1€¦ · srs groups was assessed using roast17, a rotation-based gene set...
TRANSCRIPT
Supplementary appendix 1This appendix formed part of the original submission and has been peer reviewed. We post it as supplied by the authors.
Supplement to: Davenport EE, Burnham KL, Radhakrishnan J, et al. Genomic landscape of the individual host response and outcomes in sepsis: a prospective cohort study. Lancet Respir Med 2016; published online Feb 22. http://dx.doi.org/10.1016/S2213-2600(16)00046-1.
Appendix Genomic landscape of the individual host response and outcomes in sepsis: a prospective cohort study
Davenport et al Overview of content Supplementary Methods Supplementary References Supplementary Figures
Figure S1. Principal components and eQTL mapping. Figure S2. MHC gene expression and T cell activation among differentially expressed genes between patient SRS groups. Figure S3. Hypoxia related network among differentially expressed genes between patient SRS groups. Figure S4. NF-B related network among differentially expressed genes between patient SRS groups. Figure S5. MYC and histone related networks identified among differentially expressed genes between patient SRS groups. Figure S6. Correlation of differential gene expression between SRS groups in derivation and validation cohorts. Figure S7. Correlation of differential gene expression between groups in the validation cohort determined using the SRS predictive gene set or an unsupervised clustering approach. Figure S8. PI3K signalling canonical pathway enrichment for sepsis cis-eQTL. Figure S9. Pathway enrichment for sepsis cis-eQTL (FDR <0.01). Figure S10. Trans-eQTL gene hub involving rs12148454. Figure S11. Sepsis-specific cis-eQTL. Figure S12. eQTL shared or specific to sepsis and distance from the transcription start site (TSS). Figure S13. Sepsis cis-eQTL involving genes differentially expressed between SRS groups and SRS-specific eQTL. Figure S14. MDS plots comparing the genetic ancestry of sepsis (GAinS) patients to 1000 Genomes Project populations.
Supplementary Tables
Table S1 Participating hospitals involved in patient recruitment and GAinS Investigators
Page 1 of 20
Supplementary Methods Case definitions and phenotyping The diagnosis of sepsis was based on the International Consensus Criteria (2003)1 with all patients reported here showing some degree of organ dysfunction during ICU admission.2 Community acquired pneumonia was defined as a febrile illness associated with cough, sputum production, breathlessness, leucocytosis and radiological features of pneumonia acquired in the community or within less than 2 days of hospital admission.3,4 Exclusion criteria were: patient or legal representative unwilling or unable to give consent; patient <18 years of age; pregnancy; an advanced directive to withhold or withdraw life sustaining treatment; admission for palliative care only; or immune-compromise. Ethics approval was granted nationally (REC Reference Number 05/MRE00/38 and 08/H0505/78) and for individual participating centres. Written, informed consent was obtained from all patients or a legal representative. An electronic case report form (eCRF) was used to record demographics and clinical covariates. Microbiological investigations were performed according to local policies and practices with organism(s) isolated, source and the use of serological methods recorded in the eCRF. Patients were followed up for 6 months following ICU discharge with date of death recorded. Sample collection Samples for RNA were obtained after ICU admission (at the time of study enrolment, a window up to day 5) by collecting 5 ml blood into Vacuette EDTA tubes (Becton Dickinson). The total blood leucocyte population was isolated on the ICU using the LeukoLOCK filter system (Ambion).5 Blood was passed across the leucocyte enrichment filter using a vacutainer system. The filtered leucocytes were stabilised with RNAlater. In addition, whole blood was collected for DNA extraction. RNA extraction Purified RNA, depleted of globin mRNA, was extracted from the LeukoLOCK filters using the Total RNA Isolation Protocol (Ambion). The contents of the filter were lysed and eluted with a guanidine thiocyanate-based solution. Cellular proteins and DNA were degraded using Proteinase K and DNase I respectively. The RNA was then purified using magnetic bead technology. Spectrophotometry (Nanodrop 2000; Thermo Scientific) was used to quantitate the RNA yield and the quality of a small subset was determined by on-chip electrophoresis (Biorad Bioanalyzer; Agilent). Genomic DNA extraction DNA was extracted from the buffy layer (or when not available, whole blood) using either the Qiagen DNA extraction protocol, the automated Maxwell 16 Blood purification kit (Promega) or the QIAamp Blood Midi kit protocol (Qiagen). The DNA yield was determined by fluorescence using the Quant-iT PicoGreen kit (Invitrogen). Microarray data processing, pathway analysis and hierarchical cluster analysis Genome-wide gene expression analysis was carried out on 1000ng RNA using the Illumina Human-HT-12 v4 Expression BeadChip gene expression platform comprising 47,231 probes. The report for analysis was generated by Illumina’s Genomestudio software. Data backgrounds were subtracted and probes that did not have a detection value greater than or equal to 0.95 in at least 5% of samples were removed. The raw data were transformed and normalised using the Variance Stabilisation and Normalisation method.6 Quality control (QC) checks including principal component analysis (PCA) to identify batch and array effects were carried out using R.7 Following QC, five samples were removed from the discovery cohort and eight samples from the validation cohort. In addition, in order to assign SRS using the gene expression model, the raw validation data were restricted to probes retained in the original cohort and normalised against the discovery data. Pre-processing, QC checks and statistical analyses were conducted separately for the discovery and validation datasets. We identified the optimal number of sepsis response signature (SRS) groups (2-4) using within-group sum of squares. Significant associations with early mortality (14-day survival) remained when the number of groups was increased to 3 (P value 0.036) or 4 (P value 0.0015). We therefore selected the minimum number of groups that explained the difference in survival in order to maximize power to detect phenotypic differences between groups. For a false discovery rate of 0.05 with 3000 genes differentially expressed between SRS groups, the power to detect a 1.5-fold difference in expression between SRS1 and SRS2 with a minimal sample size of 37 patients in one group will be 1. To select predictive gene sets, the data were first restricted to genes with moderate to high expression (log2(expression) >6.5 in a proportion of samples which equated to the smallest group of a comparison) and showed >2 fold change (FC) between SRS groups or >1.5FC between survivors and non-
Page 2 of 20
survivors. Models with a small number of predictors were then generated using a methodology developed for data sets with many more variables than observations, which fits response models with a sparsity prior and carries out simultaneous variable selection and parameter estimation.8 R packages7 used included limma9 (differential gene expression), FactoMineR10 (hierarchical clustering) and GeneRave CSIRO Bioinformatics version 3.0.88 (to identify clinical covariates or genes for use in prediction models). Genotyping Genotyping was performed for 730,525 SNPs using the Illumina HumanOmniExpress BeadChip. PLINK11 was used for the genotyping QC. Sample QC included discordant sex information, proportion of missing genotypes (>0.02), heterozygosity rate, identity by descent (pi hat >0.1875) and multi-dimensional scaling with 1000 Genomes populations (Figure S14). SNP QC included SNP missing data proportion (>0.02), minor allele frequency (MAF) (<0.01) and HWE (<1x10-10). For both the SRS cluster specific and trans-eQTL analysis, SNPs with MAF < 0.05 were excluded. eQTL analysis Following QC, 240 patients within the discovery cohort had good quality genome-wide gene expression and genotyping data for eQTL analysis. We conducted two further QC steps at the probe level to reduce potential confounding effects. Firstly, we excluded probes with sequences that mapped to more than one genomic location using BLAST. Secondly, we excluded probes corresponding to sequences with a SNP present at a MAF of at least 1% in European populations (1000 Genomes Project). eQTL analysis was performed using an additive linear model by the R package Matrix eQTL12. We included major principal components (PCs) of the gene expression data as covariates in the eQTL analysis to limit the effect of confounding factors. As noted by other investigators mapping eQTL in different contexts, this enhances detection of eQTLs13-15. The number of PCs to include was determined by mapping cis eQTL with increasing numbers of PCs to maximise the number of unique probes with cis eQTL associations. We included the first 30 PCs for the main eQTL (Figure S1) and the first 25 PCs for the individual SRS group eQTL. Comparison with whole blood eQTL meta-analysis Sepsis eQTL were compared to the Westra et al eQTL meta-analysis dataset.15 We restricted the analysis to the 13,590 genes shared across the Illumina Expression BeadChips used and compared the datasets for shared and specific eQTL at the gene level. eQTL were considered likely to be sepsis-specific if no significant association (FDR <0.05) was seen for a given gene in the Westra data. Annotation of eQTL with epigenomic features Following exclusion of Y chromosome SNPs, 3,643 unique peak eSNPs showing evidence of sepsis cis-eQTL (FDR<0.05) were annotated with chromatin features for lipopolysaccharide stimulated monocytes from the Blueprint Consortium16 downloaded from their ftp server. The data included histone marks (H3K27ac, H3K4me1 and H3K4me3) for two samples and DNase I hypersensitivity for four samples. Data were available for all autosomes and the X chromosome. Consensus peaks were defined by presence in at least two samples. The proportion of eSNPs that overlapped with each mark was compared to the proportion for SNPs within 1Mb of a probe (n=614,176) by Fisher’s Exact test, and distance to the nearest mark by Mann-Whitney test. Endotoxin tolerance Two human endotoxin tolerance datasets were accessed through NCBI GEO database (http://www.ncbi.nlm.nih.gov/geo/), accession GSE15219 and GSE22248. The transcriptomic response to a single lipopolysaccharide treatment was first established. The expression of these lipopolysaccharide response genes in cells stimulated once was then compared to their expression in cells from the same subject that had been treated twice to identify differentially responding genes. Combining these two studies defined an endotoxin tolerance gene signature comprising 398 genes (FDR <0.05, FC >1.5), of which 331 were measured in the sepsis patients. Enrichment of this tolerance signature within the genes differentially expressed between SRS groups was assessed using ROAST17, a rotation-based gene set test that takes account of directionality in addition to significance, enabling the likelihood of the SRS1 group expression signature being enriched for the endotoxin tolerance signature to be determined.
Page 3 of 20
References 1. Levy MM, Fink MP, Marshall JC, et al. 2001 SCCM/ESICM/ACCP/ATS/SIS International Sepsis Definitions Conference. Intensive Care Med 2003; 29(4): 530-8. 2. Vincent JL, Opal SM, Marshall JC, Tracey KJ. Sepsis definitions: time for change. Lancet 2013; 381(9868): 774-5. 3. Angus DC, Marrie TJ, Obrosky DS, et al. Severe community-acquired pneumonia: use of intensive care services and evaluation of American and British Thoracic Society Diagnostic criteria. Am J Respir Crit Care Med 2002; 166(5): 717-23. 4. Walden AP, Clarke GM, McKechnie S, et al. Patients with community acquired pneumonia admitted to European intensive care units: an epidemiological survey of the GenOSept cohort. Crit Care 2014; 18(2): R58. 5. Idaghdour Y, Storey JD, Jadallah SJ, Gibson G. A genome-wide gene expression signature of environmental geography in leukocytes of Moroccan Amazighs. PLoS Genet 2008; 4(4): e1000052. 6. Huber W, von Heydebreck A, Sultmann H, Poustka A, Vingron M. Variance stabilization applied to microarray data calibration and to the quantification of differential expression. Bioinformatics 2002; 18 Suppl 1: S96-104. 7. R Core Team. R: A language and environment for statistical computing (R Foundation for Statistical Computing, Vienna, Austria). url: http://www.Rproject.org/. 2013. 8. Kiiveri HT. A general approach to simultaneous model fitting and variable elimination in response models for biological data with many more variables than observations. BMC Bioinformatics 2008; 9: 195. 9. Smyth GK. Linear models and empirical bayes methods for assessing differential expression in microarray experiments. Stat Appl Genet Mol Biol 2004; 3: Article3. 10. Husson FJ, Josse J, Le S, Mazet J. FactoMineR: Multivariate Exploratory Data Analysis and Data Mining with R. 2013. 11. Purcell S, Neale B, Todd-Brown K, et al. PLINK: a tool set for whole-genome association and population-based linkage analyses. Am J Hum Genet 2007; 81(3): 559-75. 12. Shabalin AA. Matrix eQTL: ultra fast eQTL analysis via large matrix operations. Bioinformatics 2012; 28(10): 1353-8. 13. Biswas S, Storey JD, Akey JM. Mapping gene expression quantitative trait loci by singular value decomposition and independent component analysis. BMC Bioinformatics 2008; 9: 244. 14. Fehrmann RS, Jansen RC, Veldink JH, et al. Trans-eQTLs Reveal That Independent Genetic Variants Associated with a Complex Phenotype Converge on Intermediate Genes, with a Major Role for the HLA. PLoS Genet 2011; 7(8): e1002197. 15. Westra HJ, Peters MJ, Esko T, et al. Systematic identification of trans eQTLs as putative drivers of known disease associations. Nat Genet 2013; 45: 1238-43. 16. Saeed S, Quintin J, Kerstens HH, et al. Epigenetic programming of monocyte-to-macrophage differentiation and trained innate immunity. Science 2014; 345(6204): 1251086. 17. Wu D, Lim E, Vaillant F, Asselin-Labat ML, Visvader JE, Smyth GK. ROAST: rotation gene set tests for complex microarray experiments. Bioinformatics 2010; 26(17): 2176-82. 18. Hotchkiss RS, Swanson PE, Knudson CM, et al. Overexpression of Bcl-2 in transgenic mice decreases apoptosis and improves survival in sepsis. J Immunol 1999; 162(7): 4148-56. 19. Zhang DW, Shao J, Lin J, et al. RIP3, an energy metabolism regulator that switches TNF-induced cell death from apoptosis to necrosis. Science 2009; 325(5938): 332-6. 20. Rodrigue-Gervais IG, Labbe K, Dagenais M, et al. Cellular inhibitor of apoptosis protein cIAP2 protects against pulmonary tissue necrosis during influenza virus infection to promote host survival. Cell Host Microbe 2014; 15(1): 23-35. 21. Allam R, Kumar SV, Darisipudi MN, Anders HJ. Extracellular histones in tissue injury and inflammation. J Mol Med 2014; 92(5): 465-72. 22. Liberali P, Snijder B, Pelkmans L. A hierarchical map of regulatory genetic interactions in membrane trafficking. Cell 2014; 157(6): 1473-87.
Page 4 of 20
0 20 40 60 80 100
2000
2500
3000
3500
4000
Number of PCs
Num
ber o
f uni
que
prob
es w
ith c
is a
ssoc
iatio
ns
Figure S1. Principal components and eQTL mapping. Analysis of the effects of incorporation of major associated principal components (PC) of the expression data as covariates on observed cis-eQTL. PCs correlated with genotype were not used. The number of unique probes with a cis association is plotted.
Page 5 of 20
MHC class I MHC class IITAPBP NLRC5 CIITA CREB1 RFX5 NFYB
CD74
HLA-AHLA-B
HLA-CHLA-E
HLA-FB2M HLA-DRB5
HLA-DRB1HLA-DQA1
HLA-DQB1HLA-DOB
HLA-DMBHLA-DMA
HLA-DOAHLA-DPA1
HLA-DPB1CD40 CD80 CD86
OX40L
MHC/Antigen
Coinhibitor
PDCD1CTLA4 PAG1PTPRC
CD4 CD8A CD8BCD3E
CD3GTRA
TRBCD3D
CD3E LAT TRAT1CD28 ICOS CD40LGIL-2R OX40
P
SHP LCK
FYNPTPN22
PTPN6
CBL
Negativeregulators
CD247
ZAP70GRAP2 GRB2
Shc
PI3K(complex)
GRB2
GAB2
BCL-XL
BCL2
Cell survivalP38 MAPK SKAP1
Rap1 LCP2
NCK
Pak
PDPK1 Akt
GSK3B
Integrins, cell adhesionITK
VAV
CDC42
RHOA
Calcineurinprotein(s)
PLCG1
Sos MAP3K14MAP3K8
TRAF3 TRAF5 TRAF2
RASGRP1
Actin reorganisation
CARD11
Ras
RAF1
DAG
PRKCQ
BCL10MALT1
MAPK1
MAP2K1/2
MAPK3
MAP3K7
MAP2K7MAPK9
NFAT (complex)
JNK
IκB
Ap1 NF-κBc-JUN
Co-ordinated mobilisation of transcription factors needed for expression of genes
ProliferationDifferentiation
Immune response
MHC class I MHC class IITAPBP NLRC5 CIITA CREB1 RFX5 NFYB
CD74
HLA-AHLA-B
HLA-CHLA-E
HLA-FB2M HLA-DRB5
HLA-DRB1HLA-DQA1
HLA-DQB1HLA-DOB
HLA-DMBHLA-DMA
HLA-DOAHLA-DPA1
HLA-DPB1CD40 CD80 CD86
OX40L
MHC/Antigen
Coinhibitor
PDCD1CTLA4 PAG1PTPRCCD4 CD8A
CD8BCD3E
CD3GTRA
TRBCD3D
LAT TRAT1 CD28 ICOSCD40LG
IL-2R OX40
P
SHP LCK
FYNPTPN22
PTPN6
CBL
Negativeregulators
CD247
ZAP70GRAP2 GRB2
Shc
PI3K(complex)
GRB2
GAB2
BCL-XL
BCL2
Cell survivalP38 MAPK SKAP1
Rap1 LCP2
Pak
PDPK1 Akt
GSK3B
Integrins, cell adhesionITK
VAV
CDC42
RHOA
Calcineurinprotein(s)
PLCG1
Sos MAP3K14
MAP3K8 TRAF3 TRAF5 TRAF2
RASGRP1
Actin reorganisation
CARD11
Ras
RAF1
DAGBCL10MALT1
MAPK1
MAP2K1/2
MAPK3
MAP3K7
MAP2K7MAPK9
NFAT (complex)
JNK
IκB
Ap1 NF-κB
Co-ordinated mobilisation of transcription factors needed for expression of genes
ProliferationDifferentiation
Immune response
CD3E
c-JUN
Figure S2. MHC gene expression and T cell activation among differentially expressed genes between patient SRS groups. Differentially expressed genes (FC >1.5, FDR <0.05) between patients in groups SRS1 and 2 involving MHC and T cell activation are shown for (A) the discovery cohort (n=265) and (B) the validation cohort (n=106) with red shading indicating up-regulation, green shading down-regulation (relative to the SRS1 group). Pathways derived from manual annotation and Kegg pathway hsa04660.
B
A
Page 6 of 20
A
NOP56
NHP2L1
RCC2 DHX30
TSR1
RIOK1 FBL
-0.8042.2E-31
-0.8226.2E-18
-0.7838.7E-12
-1.0061.2E-35
-0.6373.8E-15
-1.2723.5E-39
-0.7991.6E-23
-0.6964.3E-26
-0.9741.1E-11
0.6522.7E-9
0.6013.4E-15
-1.2022.2E-38
-1.0297.0E-31
-1.0011.5E-18
-0.7473.2E-19
-0.7402.8E-17
0.6605.3E-14
-0.6639.1E-14
-0.9566.9E-29
0.7375.1E-8
-0.9241.8E-23
-0.9672.2E-22
-1.2111.6E-20
-0.6451.9E-5
0.8921.5E-6
-0.5883.5E-11
1.6193.6E-26
-0.8673.1E-22
0.9551.6E-29
1.5183.4E-35
-1.0621.5E-21
-1.2784.7E-16
-0.7472.1E-21
-0.9141.0E-40
-0.5971.5E-9
WBSCR22
UBE2O
NOP58
SF3A3 FHL1 HIST1H2BK GADD45B
HLA-DRB3
ACP5
SLC6A8
FAM13A SESN2RUVBL1
ABCF2
INPP5E
KIFC2
TMEM19HIF1A
FH LDHA PFKFB3
CTPS1
EPAS1
PLAC8
RPL3
SLC29A1
RPL10A
AQP3
NOP56
NHP2L1
RCC2 DHX30
TSR1
RIOK1 FBL
WBSCR22
UBE2O
NOP58
SF3A3 FHL1 HIST1H2BK GADD45B
HLA-DRB3
ACP5
SLC6A8
FAM13A SESN2RUVBL1
ABCF2
INPP5E
KIFC2
TMEM19HIF1A
FH LDHA PFKFB3
CTPS1
EPAS1
PLAC8
RPL3
SLC29A1
RPL10A
AQP3
-0.4429.4E-6
-0.4881.1E-5-0.654
5.3E-12
-0.7771.3E-11
-0.3381.9E-3
-0.8899.1E-14
-0.7301.2E-3
-1.0178.8E-13
-0.8116.3E-6
-0.6935.3E-8
-0.6804.6E-6
-0.3866.7E-4
0.7662.1E-6
-1.0291.8E-12
-0.4273.9E-41.221
7.5E-9
-0.3637.7E-3 2.054
7.8E-13
0.7977.5E-9
0.9678.6E-9
-0.7122.4E-3
-0.6092.5E-7
-0.7053.0E-7
-1.0214.8E-7
-0.7312.2E-3
-0.4264.9E-3
-0.4531.2E-4
0.5363.6E-2
1.3701.2E-10
1.1325.6E-11-0.460
3.9E-5
-0.4102.5E-2
-0.6206.9E-10
-0.0796.4E-1
-0.4533.5E-6
1.0E-5
1.8E-14
2.3E-5
5.2E-7
1.4E-4
4.2E-13 2.3E-4
2.3E-4
1.9E-28
2.5E-61.2E-6
1.5E-61.1E-4
RCC2 DHX30
TSR1
NHP2L1
NOP56
FBL
WBSCR22
UBE2O
NOP58
SF3A3 FHL1 HIST1H2BK GADD45B
HLA-DRB3
ACP5
SLC6A8
FAM13A SESN2
ABCF2
INPP5E
KIFC2
TMEM19
HIF1A
FHLDHA
PFKFB3
CTPS1
EPAS1
PLAC8
RPL3
SLC29A1
RPL10A
AQP3
RUVBL1
RIOK1
CDC14A
NME3
IREB2VBP1
HIF1A
-0.6933.9E-5
-0.7911.3E-11
1.1325.6E-110.788
1.2E-7
0.6138.2E-6
0.6611.1E-7
-0.7252.2E-2
0.6766.4E-6
-0.7194.8E-7
-1.0291.8E-12
-0.7053.0E-7
2.0547.8E-13
0.6231.4E-3
0.7662.1E-6
1.2217.5E-9
-0.7202.1E-12
0.7123.8E-7
-0.6361.5E-6
0.6921.6E-6
-0.7122.4E-3
-0.7678.7E-5
0.7977.5E-9
-0.9491.4E-6
0.9678.6E-9
0.7882.3E-7
1.3701.2E-10
-0.7598.3E-4
0.8075.0E-6
-0.8274.0E-4
0.7491.6E-3
-1.1347.7E-10
BA
C D
LDHACollagen type IV
ALDOA
PROK1
TCEB1CHKAPLAC8 VHL
KIFC2PIM2 MYL9
HLA-DRB3
SESN2
TLR6
DNAJA3RUNX1
ETS1FHL1CDK11A
GADD45BCABIN1
HIST1H2BK
POR
PFKFB3
AMPK
CYP4F3
ERGIC1
MCL1
IGF2BP3
BCL11A
0.5922.5E-5
0.6984.0E-8
Figure S3. Hypoxia related network among differentially expressed genes between patient SRS groups. (A) The most significant network (P 1x10-35, indicating likelihood of the genes in a network being found together due to random chance) identified on network analysis of differentially expressed genes between SRS groups 1 and 2 in the discovery cohort included 35 genes of which HIF1A (HIF1α) and EPAS1 (HIF2α) are nodal genes. Log fold change is shown below molecular symbols together with FDR. (B) Gene expres-sion data for differentially expressed genes in the validation cohort overlaid on the same gene network. Log fold change is shown below molecular symbols together with FDR. (C) Sepsis cis-eQTL involving genes in network (where cis-eQTL are found, molecules are shaded and P values shown). (D) In the validation cohort, the second most significant network identified (P 1x10-28) was also related to hypoxia with HIF1A as a nodal gene and included 9 other genes shared with (a). Log fold change is shown below molecular symbols together with FDR.
Page 7 of 20
ERAP2USPL1 LSP1
WWP2
NQO2
CARD91.0E-47
1.1E-4
2.2E-41.3E-24
2.1E-4
4.7E-8
5.8E-6
7.5E-10 4.1E-8
2.5E-12
1.9E-4
3.0E-9 6.5E-5 3.6E-9
5.9E-6
A
C
B
D
-0.6327.8E-5
ERAP2USPL1 LSP1
WWP2
ZDHHC3OPTN
CARD6CARD9
SLC2A5
NQO2PASK
TBC1D4CAMK2D
PIM3FASTKD1
UBE2E2
AMPH
NOP14 CCL3L1
NFκB (complex)CIRH1A
GTPBP4MLKL
PPP1R16B
G0S2
IL1R1IL-1R
Fc gamma receptor
IL1
INF2
RNMTL1
FUCA1
EMR1
ZNF385A
VOPP1
REL RELA RELB NFκB1 NFκB2 NFκBIA NFκBIB NFκBIE
ERAP2USPL1 LSP1
WWP2
ZDHHC3OPTN
CARD6CARD9
SLC2A5
NQO2PASK
TBC1D4CAMK2D
PIM3FASTKD1
UBE2E2
AMPH
NOP14 CCL3L1
NFκB (complex)CIRH1A
GTPBP4MLKL
PPP1R16B
G0S2
IL1R1IL-1R
Fc gamma receptor
IL1
INF2
RNMTL1
FUCA1
EMR1
ZNF385A
VOPP1
REL RELA RELB NFκB1 NFκB2 NFκBIA NFκBIB NFκBIE
-0.7604.4E-22
0.6257.9E-10
1.2855.0E-25
-0.9195.7E-6
0.7391.0E-11
-0.9593.3E-22
0.7589.4E-4
-0.8096.0E-20
0.7661.4E-7
-0.9942.1E-13
-1.1452.8E-38
-0.5892.7E-8
0.7734.3E-7
-0.6462.1E-13
1.4104.8E-18
-0.5862.7E-24
-1.1842.6E-35
0.9264.5E-12
0.9102.8E-16
-1.0041.8E-25
0.9031.8E-9
-0.9122.6E-20
-0.7912.4E-15
-1.1181.6E-34
0.9125.4E-27
-0.8034.9E-5
-0.6692.1E-23
0.8181.1E-28
-0.8931.3E-24
-0.9402.1E-27
0.4301.0E-5
0.0691.9E-1
-0.0834.5E-1
0.2792.3E-12
-0.3612.2E-4
0.5001.3E-17
0.4176.3E-5
-0.6686.1E-12
-0.5246.8E-2
-0.2892.3E-2
0.6823.7E-6
0.8102.5E-7
-0.4163.9E-1
0.6395.6E-4
-0.4943.4E-6
0.4502.6E-1
-0.5563.3E-4
0.2353.4E-1
-1.0126.8E-7
-0.6991.4E-10
-0.2887.2E-2
0.5551.2E-1
-0.8217.9E-6
1.4567.9E-9
-0.4055.8E-5
-1.0516.1E-8
1.2582.4E-13
0.9501.2E-9
-0.7311.0E-4
1.0108.9E-9
-0.6037.9E-5
-0.6651.2E-4
-0.8123.2E-12
0.8604.1E-10
-0.2854.7E-1
-0.8024.4E-11
0.7261.1E-10
-0.3975.4E-4
-0.5026.3E-4
0.5802.0E-5
0.1994.5E-2
0.1245.5E-1
0.2501.8E-4
-0.3276.6E-2
0.2721.7E-2
0.3133.9E-3
-0.4275.8E-3
ZDHHC3OPTN
CARD6SLC2A5
PASKTBC1D4
CAMK2D
PIM3FASTKD1
UBE2E2
AMPH
NOP14 CCL3L1
NFκB (complex)CIRH1A
GTPBP4MLKL
PPP1R16B
G0S2
IL1R1IL-1R
Fc gamma receptor
IL1
INF2
RNMTL1
FUCA1
EMR1
ZNF385A
VOPP1
REL RELA RELB NFκB1 NFκB2 NFκBIA NFκBIB NFκBIE
REL RELA RELB NFκB1 NFκB2 NFκBIA NFκBIB NFκBIE
HLA-DRB
CD82
HLA-DMB
PIM3
ZFAND6TRAF5 TNFAIP8 IL36G
LSP1NAIP
EMR1
ZDHHC3 NQO2 NLRC4CIRH1A PPP1R16B
CARD9
CARD6
S100A12OPTN NOP14
MLKLZNF385A
RFTN1RNMTL1
TNFSF15 TNFRSF25PASK WWP2
BAG4
BNIP2
INF2
VOPP1TNFRSF4
0.9504.1E-10
-1.5965.5E-11 1.010
8.9E-9
0.8796.2E-8 0.682
3.7E-6 1.2432.6E-7
1.1955.5E-14
0.8105.0E-6
0.7141.6E-4
-1.1102.8E-7
-0.8123.2E-12
0.7261.1E-10
1.4567.9E-9
0.6395.6E-4
-1.0516.1E-8
-0.6651.2E-4
0.7292.2E-9
0.8604.1E-10
-0.8024.4E-11
-0.7311.0E-4
-0.8217.9E-6
-1.0455.4E-11
0.9501.2E-9
-0.7611.9E-4 -1.293
2.4E-7
-0.6991.4E-10
-0.6037.9E-5
0.6307.9E-11
0.8102.5E-7
0.8208.8E-13
1.2582.4E-13
-1.0126.8E-7
-0.6338.2E-4
0.5802.0E-5
0.1994.5E-2
0.1245.5E-1
0.2501.8E-4
-0.3276.6E-2
0.2721.7E-2
0.3133.9E-3
-0.4275.8E-3
Figure S4. NF-κB related network among differentially expressed genes between patient SRS groups. (A) Gene network identi-fied on network analysis of differentially expressed genes between SRS groups 1 and 2 in the discovery cohort included 31 genes with NF-κB complex as the node (individual NF-κB genes shown below nodal network) (P 1x10-25). Log fold change is shown below molec-ular symbols together with FDR. (B) Gene expression data for differentially expressed genes in the validation cohort overlaid on the same gene network. Log fold change is shown below molecular symbols together with FDR. (C) Sepsis cis-eQTL involving genes in discovery data network (where cis-eQTL are found, molecules are shaded and P values shown). (D) In the validation cohort, the most significant network identified was also related to NF-κB (P 1x10-28) involving 33 genes. Log fold change is shown below molecular symbols together with FDR.
Page 8 of 20
CCR3
-1.8663.4E-15
0.6792.0E-10
LDLRAP1-1.4602.1E-29
EEF2-0.6614.4E-33
CRIP2-1.1981.9E-18 STOM
1.2771.1E-22
LDLR0.6501.2E-7
LRPAP1
1.0031.1E-22
AHSP-1.1577.9E-9
PHF1-0.6262.7E-16
LARP1B-1.1804.6E-28
RPS6KA30.9243.5E-12
NFATC2IP-0.7184.0E-32
HNRNPDL-1.0287.9E-28
SF3B3-0.6757.2E-25
PNO1-0.8121.2E-21
-0.9625.1E-24 -0.852
2.5E-27
PTCD2-0.6292.1E-18
ADD30.7169.7E-17
DDX47
-0.9657.4E-41
SUN1-1.4088.7E-27
SYNE2-0.8662.3E-12
Histone H3
ACKR3
-0.8813.6E-13
ANK1-0.6041.3E-5
MRPS9-1.2936.8E-40
PPP2R2B-2.0616.1E-34
STXBP21.1886.8E-25
NCAPD3-0.6912.9E-12
SMC2-0.9153.5E-13
SUV39H1
-0.6241.4E-11
VAMP8-0.6014.4E-22
NCAPH2-0.5871.9E-19
STX110.6041.8E-13
CCR3
-0.7675.9E-2
PCSK9
0.9205.6E-6
LDLRAP1-1.0451.9E-9
EEF2-0.5411.0E-8
CRIP2-0.4606.3E-3 STOM
1.3873.6E-9
LDLR1.1246.6E-8
LRPAP1
0.9522.6E-8
AHSP-1.0343.6E-3
PHF1-0.5087.0E-5
LARP1B-0.5303.7E-5
RPS6KA31.1465.2E-9
NFATC2IP-0.6084.1E-10
HNRNPDL-0.7249.9E-10
SF3B3-0.5104.2E-6
PNO1-0.3931.2E-3
MGMT
-0.6994.9E-8
PRMT6
-0.4531.7E-4
PTCD2-0.2442.1E-2
ADD30.6601.3E-7
DDX47
-0.6011.4E-8
SUN1-0.6927.9E-5
SYNE2-0.5626.5E-3
Histone H3
ACKR3
-0.2382.3E-1
ANK1-0.4208.6E-2
MRPS9-0.7553.3E-9
PPP2R2B-0.9331.2E-4
STXBP20.7115.7E-6
NCAPD30.0109.6E-1
SMC20.1862.6E-1
SUV39H1
-0.1074.6E-1
VAMP8-0.4726.7E-6
NCAPH2-0.3731.7E-4
STX110.7628.8E-9
IGF2BP2-0.6561.6E-8
PITX2
SSBP2
CCND2
RRN3
TAF1B
SH3KBP1
SHMT1FBXO32 SLC25A19
BCAT1
EXOSC8
ACACA ALDH18A1
MYC
HSPH1 ST3GAL4
SCPEP1 PREPDKC1
BCORODC1
GCLMDNA-methyltransferase
GEMIN2
GAR1
NHP2
SLC22A4
BRF1TIAM1SHMT2
RGS10PTMA
RPS14RPL15
IGF2BP2
PITX2
SSBP2
CCND2
RRN3
TAF1B
SH3KBP1
SHMT1FBXO32 SLC25A19
BCAT1
EXOSC8
ACACA ALDH18A1
MYC
HSPH1 ST3GAL4
SCPEP1 PREPDKC1
BCORODC1
GCLMDNA-methyltransferase
GEMIN2
GAR1
NHP2
SLC22A4
BRF1TIAM1SHMT2
RGS10PTMA
RPS14RPL15
0.6854.7E-21
-1.1309.7E-27
A B
C DCCR3
-0.7675.9E-2
PCSK9
0.9205.6E-6
LDLRAP1-1.0451.9E-9
EEF2-0.5411.0E-8
CRIP2-0.4606.3E-3 STOM
1.3873.6E-9
LDLR1.1246.6E-8
LRPAP1
0.9522.6E-8
AHSP-1.0343.6E-3
PHF1-0.5087.0E-5
LARP1B-0.5303.7E-5
RPS6KA31.1465.2E-9
NFATC2IP-0.6084.1E-10
HNRNPDL-0.7249.9E-10
SF3B3-0.5104.2E-6
PNO1-0.3931.2E-3
MGMT
-0.6994.9E-8
PRMT6
-0.4531.7E-4
PTCD2-0.2442.1E-2
ADD30.6601.3E-7
DDX47
-0.6011.4E-8
SUN1-0.6927.9E-5
SYNE2-0.5626.5E-3
Histone H3
ACKR3
-0.2382.3E-1
ANK1-0.4208.6E-2
MRPS9-0.7553.3E-9
PPP2R2B-0.9331.2E-4
STXBP20.7115.7E-6
NCAPD30.0109.6E-1
SMC20.1862.6E-1
SUV39H1
-0.1074.6E-1
VAMP8-0.4726.7E-6
NCAPH2-0.3731.7E-4
STX110.7628.8E-9
-0.8037.0E-23
-0.8716.2E-27
-0.8491.2E-14
0.6067.8E-4
-0.6008.0E-20-0.984
2.9E-30-0.6621.7E-23
-1.7483.5E-36
-0.8662.7E-10
-1.3382.3E-25-0.746
9.2E-14
-0.8411.7E-14
0.6631.9E-15
-1.0916.7E-23
-1.4891.6E-27
0.8541.3E-19
-0.8642.6E-18
-0.6153.7E-15
0.9047.3E-23
-0.8922.1E-18
-0.6722.1E-15
-1.0787.3E-14
0.6201.8E-13
-0.6331.2E-15
-0.5975.1E-33
-0.9802.0E-35
-0.9853.6E-26
-0.6622.3E-18
-0.8204.4E-25
-0.5897.8E-34
-0.7827.7E-33
0.5159.6E-6
-0.6031.5E-8
-0.6441.4E-3
-0.4891.1E-3 -0.282
1.4E-2-0.1811.9E-1
-0.9623.4E-6 0.744
9.3E-3-0.2552.9E-1
-0.4372.8E-4
-0.6953.3E-6
-0.9853.6E-26
-0.3631.9E-4
-0.6122.9E-5
-0.5062.4E-6
-0.4881.1E-2
0.4555.6E-3
-0.3724.8E-3
-0.5457.3E-4
-0.6663.1E-5
0.9212.4E-9
1.0878.4E-8
-0.3272.0E-2
0.5202.22E-3
-0.1264.7E-1
-0.3255.7E-6
-0.6331.2E-8
-0.4652.2E-3
-0.3877.4E-4
-0.4373.9E-5
-0.3194.2E-5
-0.6543.9E-7 -0.685
2.3E-8
-0.4371.9E-3
-0.5083.2E-5
PCSK9
MGMTPRMT6
Figure S5. MYC and histone related networks identified among differentially expressed genes between patient SRS groups. In each network log fold change and FDR are shown below molecular symbols. (A) Gene network identified on network analysis of genes differentially expressed between SRS groups in the discovery cohort included 34 genes with MYC as the node (second most significant network identified, P 1x10-27). MYC, a transcription factor involved in cell proliferation, apoptosis and survival, was significantly down-regulated in SRS1 patients, as was the anti-apoptotic gene BCL2, overexpression of which improves survival in sepsis18. Expres-sion of the serine/threonine kinase RIP3 (RIPK3) which plays an essential role in necroptosis19 and its substrate MLKL were increased in SRS1 patients, while BIRC3 encoding cIAP2, a critical inhibitor of necroptosis conferring protection during influenza virus infec-tions20, was down-regulated. (B) Gene expression data for differentially expressed genes in the validation cohort overlaid on the same MYC gene network. (C) Gene network identified on network analysis of differentially expressed genes between SRS groups in the discovery cohort included 34 genes with histone H3 gene complex as the node (P 1x10-27). The majority of assayed histone genes were up-regulated in SRS1 patients (all 7 genes with >1.5 FC up-regulated). Extracellular histones have potent toxic effects, notably in the context of neutrophil extracellular traps (NETs) resulting in NETosis, which in the context of sepsis can promote endothelial dysfunc-tion, hypoxia and death21. (D) Gene expression data for differentially expressed genes in the validation cohort overlaid on the same histone gene network.
Page 9 of 20
Figure S6. Correlation of differential gene expression between SRS groups in derivation and validation cohorts. The log fold change for differential gene expression of all measured genes between SRS groups 1 and 2 in the validation cohort is plotted against log fold change for the same genes between SRS groups 1 and 2 in the discovery cohort. The log fold changes are highly correlated (r2 0.82 P <2.2x10-16).
−2 −1 0 1 2 3
−2−1
01
23
Discovery cohort log2(fold change)
Valid
atio
n co
hort
log 2
(fold
cha
nge)
r2=0.82
Page 10 of 20
Figure S7. Correlation of differential gene expression between groups in the validation cohort determined using the SRS predictive gene set or an unsupervised clustering approach. The log fold change for differential gene expression of all measured genes between groups determined by unsupervised clustering in the validation cohort is plotted against log fold change for the same genes between SRS groups 1 and 2 defined by our predictive gene set in the same cohort. The log fold changes are highly correlated (r2 0.84 P <2.2x10-16).
−2 −1 0 1 2 3 4 5
−2−1
01
23
45
SRS log2(fold change)
Uns
uper
vise
d cl
uste
ring
log 2
(fold
cha
nge)
r2= 0.84
Page 11 of 20
PIP3
Extracellular space
Cytoplasm
Apoptosis
FcγRIIB
ABL CD79A
CD79B
C3
CD81
6.9E-9
CR2 CD19 CD45
LPS IL-4
TLR4 IL4R CD40CD180
RAS RAC12.4E-41.8E-16
3.6E-82.0E-6FYN
BLKSYK
VAVVAV2 VASP
LYN
1.3E-4
2.4E-5IRS CBL
p85PI3K
PIK3CDPIK3C2A
PIK3C2BATM
4.7E-8 2.2E-29 3.2E-5 1.2E-14
2.7E-5 7.2E-13
CDC42 PRKCZaPKC
SHIP BCAP
PLEKHA1
PIP2 PIP3BLNK BTK PIP3
PIP2
DAPP1TAPP
PLCL1 PLCH2 PLCD3 PLCL2
Cytoskeletal reorganisation
p110PI3KAKT1
c-RAF
MEK1/2
NF-κB
PKCβ
FOXO3
PPP3R1Cell survival IκB
NF-κB
NF-κBIE NF-κB1
ER
CREB1ATF
ATF6 ATF6B
Cellgrowth
Celldi fferentiation
Homeostasis Immuneresponse
8.8E-5
2.7E-8
1.7E-13 5.0E-9 1.6E-6 1.1E-5
3.1E-62.7E-7 2.5E-6
1.2E- 11
3.0E-62.1E-7
1.7E-6
3.2E-5
3.2E-5
2.2E-6
2.0E- 113.0E-93.6E-9
2.0E-6
2.3E-5 2.5E-7
PTEN
MAPK3
ERK1/2
AKTPDK1
IκB
MALT1BCL10 BIMP1 IKK
DAG
IP3
CAMK2GAKT Signalling
Ca2+ CALM
IκB
PPP3CACALCINEURIN
NFAT
IP3R
Ca2+
ITPR1ITPR2
P
ATF2Nucleus
NFAT NF-κB ELK1 c-JUN c-FOS3.8E-7
PLC
PIP3
Extracellular space
Cytoplasm
Apoptosis
FcγRIIB
ABL CD79A
CD79B
C3
CD81
6.9E-9
CR2 CD19 CD45
LPS IL-4
TLR4 IL4R CD40CD180
RAS RAC12.4E-41.8E-16
3.6E-82.0E-6FYN
BLKSYK
VAVVAV2 VASP
LYN
1.3E-4
2.4E-5IRS CBL
p85PI3K
PIK3CDPIK3C2A
PIK3C2BATM
4.7E-8 2.2E-29 3.2E-5 1.2E-14
2.7E-5 7.2E-13
CDC42 PRKCZaPKC
SHIP BCAP
PLEKHA1
PIP2 PIP3BLNK BTK PIP3
PIP2
DAPP1TAPP
PLCL1 PLCH2 PLCD3 PLCL2
Cytoskeletal reorganisation
p110PI3KAKT1
c-RAF
MEK1/2
NF-κB
PKCβ
FOXO3
PPP3R1Cell survival IκB
NF-κB
NF-κBIE NF-κB1
ER
CREB1ATF
ATF6 ATF6B
Cellgrowth
Celldi fferentiation
Homeostasis Immuneresponse
8.8E-5
2.7E-8
1.7E-13 5.0E-9 1.6E-6 1.1E-5
3.1E-62.7E-7 2.5E-6
1.2E- 11
3.0E-62.1E-7
1.7E-6
3.2E-5
3.2E-5
2.2E-6
2.0E- 113.0E-93.6E-9
2.0E-6
2.3E-5 2.5E-7
PTEN
MAPK3
ERK1/2
AKTPDK1
IκB
MALT1BCL10 BIMP1 IKK
DAG
IP3
CAMK2GAKT Signalling
Ca2+ CALM
IκB
PPP3CACALCINEURIN
NFAT
IP3R
Ca2+
ITPR1ITPR2
P
ATF2Nucleus
NFAT NF-κB ELK1 c-JUN c-FOS3.8E-7
PLC
Figure S8. PI3K signalling canonical pathway enrichment for sepsis cis-eQTL. Genes showing eQTL shaded red with P values for cis-eQTL shown below. IPA canonical pathway analysis showed significant enrichment (P 3.77x10-4) for PI3K signalling involving 28 molecules.
Page 12 of 20
Figure S9. Pathway enrichment for sepsis cis-eQTL (FDR <0.01). Genes showing eQTL shaded red with P values for cis-eQTL shown below. (A) MHC genes and antigen presentation. IPA canonical pathway analysis showed significant enrichment (P 5.3x10-4) for antigen presentation involving 12 molecules. Figure shows classical HLA molecules and associated proteins. There is evidence of eQTL involving antigen loading/processing genes including TAP, invariant chain (CD74) as well as eQTL involving classical class I and II molecules, and regulators such as NLRC5. (B) Viral respiratory infection. Analysis of IPA disease pathways showed most signifi-cant enrichment for viral respiratory infection (P 3.4x10-8) involving 37 molecules.
MHC class IIA
B
MHC class I
PSMB9
1.9E-27
PSMB5 TAP1 TAP2 NLRC5 CIITA CREB1 RFX5 NFYB
CD74
HLA-A HLA-B HLA-C HLA-E HLA-F B2M HLA-DRB5 HLA-DRB1 HLA-DQA1 HLA-DQB1 HLA-DOB HLA-DMB HLA-DMA HLA-DOA HLA-DPA1HLA-DPB1
1.4E-5 1.2E-5 1.1E-7 1.7E-10 3.8E-7
5.5E-28 1.7E-11
8.1E-6
1.5E-10 1.1E-5 1.2E-17
HP
SPARC
TIMP1
TNF
TIMP2
LILRA3
SERPINA1
SLCO3A1CD63
CLEC4EZYX BPI
ITGA4 FPR1 ITGAM C3AR1
STXBP2
CD9
IMPA2
LTA4H
PADI4
RRAGD
CTSZ
IER3
ST6GAL1ANXA1
RAB31
S100A12
HBE1
RNASE2 G3BP2
S100P
HIST1H2BD EIF1AY NFE2 MED16 KDM5D
1.6E-7
1.3E-7
9.8E-9
6.5E-81.8E-34
1.1E-5
2.4E-10
2.2E-64.2E-16
2.5E-55.6E-22 3.0E-5
6.1E-9 9.5E-6 1.6E-61.7E-33 1.3E-7
2.3E-8
6.1E-72.9E-8
6.9E-11
1.0E-116.9E-7
2.0E-6
4.1E-7 2.2E-5
4.8E-11
7.3E-10
4.3E-10
1.6E-5
3.0E-5
1.1E-33
2.1E-20 8.6E-19 5.2E-20 6.2E-13 3.5E-17
Page 13 of 20
WA
SF2
PP
CS
SSR
2GRK6
MYO
1G
TBC1D2B
CASC3
CNN2HDG
FRP2
MK
L1S
DF4
rs12148454
1
2
3
4
5
6
7
89
10
11
12
13
14
15
16
17
18
19
20
21
22
Figure S10. Trans-eQTL gene hub involving rs12148454. Circos plot showing trans-eQTL for rs12148454. Blue lines show trans associations (FDR < 0.05) with gene names indicated. This SNP shows a cis-eQTL (P 2.9x10-7) for VPS18 encoding vacuole protein sorting 18, a subunit of the homotypic fusion and vacuole protein sorting complex (HOPS) involved in endocytic and autophagocyt-ic pathways. VPS18 has been shown to be critical to interactions between signalling and membrane trafficking22. Related trans associated genes include WASF2 encoding a WAS protein family member involved in transduction of signals relating to cell shape, motility and function; SSR2, encoding signal sequence receptor 2; and MYO1G, encoding plasma membrane myosin IG which is abundant in lymphocytes and involved in Fc-gamma receptor mediated phagocytosis.
Page 14 of 20
0
2.5
5.0
7.5
10.0
Den
sity
(x10
-6)
SharedSepsis specific
−1Mb −500kb 0 500kb 1MbDistance from TSS
Figure S12. eQTL shared or specific to sepsis and distance from the transcription start site (TSS).
Page 16 of 20
IL18
RA
P e
xpre
ssio
n
11.0
11.5
12.0
12.5
13.0
13.5
CC CT TT
P 2.5x10-12
P 5.9x10−9
rs7616215
CC
R1
expr
essi
on
8.5
9.0
9.5
10.0
10.5
TT TC CC
rs7616215
CC
R3
expr
essi
on
4.0
5.0
6.0
7.0
8.0
9.0
TT TC CC
rs2110735
A B
C D
E F
Obs
erve
d (−
log 10
P)
Obs
erve
d (−
log
10 P
)
Figure S13
IL18RAP region
CCR1 region
Chromosome 3 position (kb)46000 46300 46600
024
6
8
10
12
0
20
40
60
Rec
ombi
natio
n ra
te (c
M/M
b)
CCR9 CXCR6 CCR3CCR1 CCR2 CCR5 LTF
CCR3 region
Chromosome 3 position (kb)45600 46000 46400
0
2
4
6
8
10
0
20
40
60
Rec
ombi
natio
n ra
te (c
M/M
b)
LARS2 LIMD1 CCR9CXCR6 CCR3CCR1 CCR5LTF
rs7616215
rs7616215
0.8
0.5
0.2
1.0r2
Peak eSNP
G H
4.5
5.0
5.5
6.0
4.5
5.0
5.5
6.0
14.0
3.0
rs1741820
HS
F2 e
xpre
ssio
n S
RS
2
GG GA AA
HS
F2 e
xpre
ssio
n S
RS
1
0
2
4
6
Obs
erve
d (−
log 10
P) rs1741820
P 2.47x10−06
Chromosome 6 position (kb)122300 122700 123100
0
2
4
6
Obs
erve
d (−
log 10
P)
rs1741820P 0.039
PKIBHSF2 SERINC1 FABP7
SRS1 group
SRS2 group
0
20
40
Rec
ombi
natio
n ra
te (c
M/M
b)
0
20
40
Rec
ombi
natio
n ra
te (c
M/M
b)
Chromosome 2 position (kb)102900 103000 103100
02468
101214161820
0
20
40
60
Rec
ombi
natio
n ra
te (c
M/M
b)
IL1RL2 IL1RL1 IL18R1 IL18RAP
P 3.6x10-20rs2110735
Obs
erve
d (−
log 10
P)
HSF2 region
Page 17 of 20
Figure S13. Sepsis cis-eQTL involving genes differentially expressed between SRS groups and SRS-specific eQTL. Cis-eQTL illustrated by boxplots showing expression by genotype (A,C,E,G) and local regional association plots (B,D,F,H). (A-B) IL18RAP shows evidence of a cis-eQTL involving rs2110735 (P 3.6x10-20 FDR 1.7x10-16) and is up-regulated in SRS1 patients (FC 1.9, FDR 3.2x10-22). (C-F) rs7616215 is the lead eSNP for cis-eQTL involving chemokine receptor genes CCR1 (P 2.5x10-12 FDR 3.7x10-9) and CCR3 (P 5.9x10-9 FDR 4.8x10-6). CCR1 is significantly up-regulated in SRS1 patients (FC 1.8 FDR 3.1x10-14) while CCR3 is significantly down-regulated (FC 0.3 FDR 3.4x10-15). (G-H) HSF2 shows evidence of an eQTL specific to SRS1 involving rs1741820 (SRS1 P 2.47x10-6 FDR 0.0040 SRS2 P 0.039 FDR 0.74)
Page 18 of 20
A B
−0.05 0.00 0.05 0.10 0.15
−0.10
−0.05
0.00
All populations
MDS1
MD
S2 ASW
CEUCHBCHSCLMFINGAinSGBRIBSJPTLWKMXLPURTSIYRI
−0.050 −0.048 −0.046 −0.044 −0.042 −0.040
0.025
0.030
0.035
European populations
MDS1
MD
S2
GBRFINIBSCEUTSIGAinS
Figure S14. MDS plots comparing the genetic ancestry of sepsis (GAinS) patients to 1000 Genomes Project populations. The sepsis (GAinS) cohort is restricted to the 240 patients used for the eQTL analysis. (A) All populations. (B) European populations.
Page 19 of 20
Table S1. Participating hospitals involved in patient recruitment and GAinS Investigators
Site Principal Investigator Research Nurses/Fellows Other investigatorsSt Bartholomew's/Royal London Hospitals Professor Charles Hinds Eleanor McLees Dr Chris Garrard, Dr D. WatsonJohn Radcliffe Hospital, Oxford Dr Stuart McKechnie Paula Hutton, Penny Parsons, Alex Smith Dr Julian Millo, Dr Duncan YoungRoyal Victoria Infirmary, Newcastle Dr Simon Baudouin Charley Higham, Helen WalshQueen Elizabeth Hospital, Birmingham Professor Julian Bion Joanne Millar, Elsa Jane PerrySouthend Hospital Dr David Higgins Sarah AndrewsLeicester Royal Infirmary Dr Jonathan Thompson Sarah Bowrey, Sandra KazembeBroomfield Hospital, Chelmsford Dr D Arawwawala Karen Swan, Sarah Williams, Susan Smolen Dr Alasdair ShortRoyal Berkshire Hospital, Reading Dr Atul Kapila Nicola Jacques, Jane Atkinson, Abby BrownUniversity College London Hospital (UCLH), London Dr Geoffrey Bellingan Jung Hyun Ryu, Georgia Bercades Dr Richard Marshall, Dr Hugh MontgomeryNorfolk & Norwich University Hospital, Norwich Dr Simon Fletcher Melissa Rosbergen, Georgina GlisterWythenshawe Hospital, Manchester Dr Andrew Bentley Katie MccalmanThe James Cook University Hospital, Middlesbrough Dr Stephen Bonner Keith Hugill, Victoria GoodridgeAberdeen Royal Infirmary Professor Nigel Webster Jane Taylor, Sally Hall, Jenni Addison Dr Helen GalleyRoyal Hallamshire and Northern General Hospitals, Sheffield Dr Gary Mills John Humphreys, Kelsey ArmitageSt Mary's Hospital, London Dr Martin Stotz Adaeze Ochelli-OkpueRoyal Sussex County Hospital, Brighton Dr Stephen Drage Laura Ortiz-Ruiz De GordoaCharing Cross Hospital, London Dr Tony Gordon Emily Errington, Maie TempletonThe Whittington Hospital, London Dr Martin Kuper Sheik PaharyHuddersfield Royal Infirmary Dr Peter Hall Jackie HewlettCoventry and Warwickshire University Hospital, Coventry Dr Pyda Venkatesh Geraldine Ward, Marie McCauleyNorth Middlesex Hospital, London Dr Jeronimo Moreno CuestaHammersmith Hospital, London Dr Stephen Brett David KitsonFreeman Hospital, Newcastle Dr Simon Baudouin Charley HighamRoyal Preston hospital, Preston Dr Shond Laha Jacqueline Baldwin, Angela WalshBlackpool Victoria Hospital Dr Achyut Guleri Natalia WaddingtonRoyal Blackburn Hospital Dr Anton Krige Martin Bland, Lynne Bullock, Donna HarrisonKettering General Hospital Dr Jasmeet Soar Parizade RaymodeSouthmead Hospital, Bristol Dr Jasmeet Soar Sally GrierFrenchay Hospital, Bristol Dr Jasmeet Soar Sally Grier
Page 20 of 20