Molecular Medicine Reports Molecular Medicine Reports 1791-2997 1791-3004 D.A. Spandidos 10.3892/mmr.2017.7013 mmr-16-03-3133 Articles Identification of key genes in Gram-positive and Gram-negative sepsis using stochastic perturbation LiZhenliang 1 ZhangYing 2 LiuYaling 1 LiuYanchun 1 LiYouyi 1 Intensive Care Unit, Pinggu Hospital Affiliated to Capital Medical University, Beijing 101200, P.R. China Department of Infection Diseases, Beijing Pinggu Hospital of Traditional Chinese Medicine, Beijing 101200, P.R. China Correspondence to: Dr Zhenliang Li, Intensive Care Unit, Pinggu Hospital Affiliated to Capital Medical University, 59 Xinping North Road, Pinggu, Beijing 101200, P.R. China, E-mail: lizhenliang1973@hotmail.com 032017 15072017 16 3 3133 3146 01092016 17052017 Copyright: © Li et al. 2017 This is an open access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made.

Sepsis is an inflammatory response to pathogens (such as Gram-positive and Gram-negative bacteria), which has high morbidity and mortality in critically ill patients. The present study aimed to identify the key genes in Gram-positive and Gram-negative sepsis. GSE6535 was downloaded from Gene Expression Omnibus, containing 17 control samples, 18 Gram-positive samples and 25 Gram-negative samples. Subsequently, the limma package in R was used to screen the differentially expressed genes (DEGs). Hierarchical clustering was conducted for the specific DEGs in Gram-negative and Gram-negative samples using cluster software and the TreeView software. To analyze the correlation of samples at the gene level, a similarity network was constructed using Cytoscape software. Functional and pathway enrichment analyses were conducted for the DEGs using DAVID. Finally, stochastic perturbation was used to determine the significantly differential functions between Gram-positive and Gram-negative samples. A total of 340 and 485 DEGs were obtained in Gram-positive and Gram-negative samples, respectively. Hierarchical clustering revealed that there were significant differences between control and sepsis samples. In Gram-positive and Gram-negative samples, myeloid cell leukemia sequence 1 was associated with apoptosis and programmed cell death. Additionally, NADH:ubiquinone oxidoreductase subunit S4 was associated with mitochondrial respiratory chain complex I assembly. Stochastic perturbation analysis revealed that NADH:ubiquinone oxidoreductase subunit B2 (NDUFB2), NDUFB8 and ubiquinol-cytochrome c reductase hinge protein (UQCRH) were associated with cellular respiration in Gram-negative samples, whereas large tumor suppressor kinase 2 (LATS2) was associated with G1/S transition of the mitotic cell cycle in Gram-positive samples. NDUFB2, NDUFB8 and UQCRH may be biomarkers for Gram-negative sepsis, whereas LATS2 may be a biomarker for Gram-positive sepsis. These findings may promote the therapies of sepsis caused by Gram-positive and Gram-negative bacteria.

 Gram-negative sepsis Gram-positive sepsis similarity network functional enrichment stochastic perturbation Introduction

Sepsis is a systemic and deleterious inflammatory response to noxious infection (1,2). Sepsis causes ~18 million new cases and millions of deaths worldwide annually; therefore, it is a major cause of morbidity and mortality globally in critically ill patients (3,4). The excessive activation of inflammation, complement and coagulation systems may damage the host's own tissues and organs, leading to multiple organ failure and death (5). In a group of patients diagnosed with sepsis, the most common causative agents are Gram-positive and Gram-negative bacteria (6,7).

Tang et al (8) used the microarray expression profile of GSE6535 to identify the differentially expressed genes (DEGs) between patients with Gram-positive and Gram-negative sepsis with univariate F test according to the cut-off criteria of false discovery rate (FDR) <0.05 and |log fold-change (FC)|>1.5 and determined that Gram-positive sepsis and Gram-negative sepsis had a common host response at the transcriptome level in critically ill patients (8). However, a previous study illustrated the different mechanisms of sepsis caused by Gram-positive bacteria and Gram-negative bacteria. Hypoxia-inducible factor 1α and Kruppel-like factor 2 have been identified to be involved in Gram-positive endotoxin-mediated sepsis by regulating cellular motility and proinflammatory gene expression in myeloid cells (9). In Gram-negative bacteria-induced sepsis, it has been determined that the inhibition of caspase-1 and defective interleukin 1β production are important immunological features (10). Additionally, α2-antiplasmin has been identified to be a protective mediator during Gram-negative sepsis by inhibiting bacterial growth, inflammation, tissue injury and coagulation (11). Furthermore, thrombomodulin-mediated protein C activation may contribute to protective immunity in severe Gram-negative sepsis by regulating inflammatory and procoagulant response (12). Despite the clinical importance of the disease and extensive research, no specific treatment is available for sepsis caused by Gram-positive and Gram-negative bacteria. Therefore, it is necessary to screen the biomarkers for sepsis.

The present study aimed to use the microarray data of Tang et al (8) to screen the DEGs in Gram-positive and Gram-negative samples compared with control samples using the limma package based on a wide range of thresholds (P<0.05 and |log2FC|>0.8). In addition, specific genes were collected as biomarkers for sepsis caused by Gram-positive and Gram-negative bacteria. A previous study has proposed that analyses based on differential statistical tests may lead to different outcomes (13). Therefore, the findings of the present study may differ to those of Tang et al (8).

 Materials and methods <sec> <title>Microarray data

The microarray dataset of GSE6535 (8) was downloaded from the database of gene expression omnibus (www.ncbi.nlm.nih.gov/geo), which was sequenced on the platform GPL4274 NHICU Human 19K version 1.0. Probe annotation information for mapping the probes into gene symbols was also downloaded. From GSE6535 dataset, 17 neutrophil samples from patients without sepsis, 18 neutrophil samples from patients with Gram-positive sepsis, and 25 neutrophil samples from patients with Gram-negative sepsis were selected. Tang et al (8) obtained whole blood samples from critically ill patients on admission to the intensive care unit of Nepean Hospital (Sydney, Australia). Using Ficoll-Paque density gradient separation, neutrophils were isolated from the whole blood. The patients with sepsis were diagnosed retrospectively according to their medical record. According to the criteria established by Calandra and Cohen (14), the patients with sepsis were divided into Gram-positive and Gram-negative sepsis groups through assessing various clinical features, including physical examination and history and microbiological cultures, such as bronchoalveolar washings, urine, blood and cerebrospinal fluid. GSE6535 was deposited by Tang et al (8). The study of Tang et al was approved by the Ethics Committee of Nepean Hospital and written informed consent was provided by the patients or their families (8).

 Data preprocessing and differential expression analysis

Based on the probe annotation information, probe IDs were converted into their corresponding gene symbols. The average value of multiple probes (that were corresponding to the same gene) was used as the gene expression value. To eliminate inherent expression differences between genes, the gene expression values were performed with Z-score normalization as previously described (15). Subsequently, the limma package version 3.32.2 in R (16) was used to screen the DEGs in the Gram-positive and Gram-negative samples compared with the control samples. The P<0.05 and |log2FC| >0.8 were used as the cut-off criteria for screening DEGs. Using the VennDiagram in R (17), the common DEGs between Gram-positive and Gram-negative samples, as well as the specific DEGs in Gram-positive samples or Gram-negative samples were identified. Gene Ontology (GO; www.geneontology.org) is a bioinformatics resource that may be used to classify gene product function using controlled, structured vocabularies (18). Using the Database for Annotation, Visualization and Integrated Discovery (DAVID) (19), GO functional enrichment analysis was performed on the common DEGs. The hierarchical cluster analysis of the specific DEGs in Gram-positive or Gram-negative samples was conducted using cluster version 3.0 software (20) and then visualized using TreeView tool version 3 (21).

 Similarity network construction

Pearson's correlation coefficient (PCC) (22), which determines the correlation between two variables, was used to identify the positive or negative correlations among different samples, with the threshold of |PCC|>0.5. Using Cytoscape version 2.8 software (23), a similarity network was constructed for the Gram-positive, Gram-negative and control samples.

 Functional and pathway enrichment analyses

Kyoto Encyclopedia of Genes and Genomes (KEGG; www.genome.jp/kegg/), which integrates genomic, chemical and systemic functional information, is a useful resource for pathway mapping (24). Using the online tool DAVID (19), GO functional and KEGG pathway enrichment analyses were conducted for the DEGs. P<0.05 was used as the threshold.

 Identification of significantly differential functions using stochastic perturbations

The average expression value in Gram-positive or Gram-negative samples was calculated for each gene enriched in the same term (GO functions or KEGG pathways). Euclidean distance (25) was used to calculate the difference between the levels of all the terms between Gram-positive and Gram-negative samples, according to the following equation:

distance=∑i=1T(X¯Pi–X¯Ni)2

Where distance represents the Euclidean distance between Gram-positive samples and Gram-negative samples; X¯Pi stands for the average expression value of gene i in Gram-positive samples; X¯Pi represents the average expression value of gene i in Gram-negative samples; and T indicates the gene number in each term.

Subsequently, stochastic perturbations were used (26) to determine the significance findings. The 18 Gram-positive and 25 Gram-negative samples were randomly sorted. Subsequently, 18 samples were randomly selected and defined as Gram-positive samples and the remaining 25 samples were defined as Gram-negative samples. The Euclidean distance between the newly defined Gram-positive samples and Gram-negative samples was recalculated. This was repeated for 10,000 times and the Euclidean distance for 10,000 perturbations were sorted from small to large and used as the background distribution. The ranking order of the initial Euclidean distance in the background distribution was calculated and converted to a P-value. The terms with P<0.05 were considered significantly differential functions between Gram-positive and Gram-negative samples.

 Results <sec> <title>DEGs analysis

The gene distribution of Gram-negative (Fig. 1A) and Gram-positive samples (Fig. 1B) are presented using a volcano plot. Using the P<0.05 and |log2FC|>0.8 as thresholds, a total of 340 DEGs, including 181 upregulated genes, including large tumor suppressor kinase 2 (LATS2), NADH:ubiquinone oxidoreductase subunit S4 (NDUFS4) and 159 downregulated genes, including myeloid cell leukemia 1 (MCL1) and chitinase-like 1, were obtained in Gram-positive samples compared with control samples. A total of 485 DEGs were identified, 324 upregulated genes, including NDUFS4 and NADH:ubiquinone oxidoreductase subunit B2 (NDUFB2) and 161 downregulated genes, including MCL1 and ecotropic viral integration site 2B, were identified in Gram-negative samples compared with the control samples. The top 10 significantly upregulated genes and downregulated genes in the Gram-negative and Gram-positive samples are presented in Table I.

A total of 188 common DEGs, including 120 upregulated and 68 downregulated were identified between Gram-positive and Gram-negative samples. Additionally, 152 specific DEGs, including 61 upregulated and 91 downregulated genes in the Gram-positive samples and 297 specific DEGs, including 204 upregulated and 93 downregulated genes in Gram-negative samples were also screened (Fig. 2). GO functional enrichment analysis was performed on the common DEGs, consisted of 120 upregulated and 68 downregulated genes in the Gram-positive and Gram-negative samples and the top 5 terms for each sample type were presented in Fig. 2. The findings revealed that the common upregulated genes were primarily associated with the regulation of apoptosis and cell death, whereas the common downregulated genes were primarily associated with cellular respiration (Fig. 2). Hierarchical cluster analysis of the specific DEGs revealed that there were significant differences between control and sepsis samples. However, no significant difference was identified between the Gram-positive and Gram-negative samples (Fig. 3).

 Similarity network analysis

In the similarity network, positive associations were identified between the majority of the control and sepsis samples. However, negative associations were also identified between the control and sepsis samples (Fig. 4).

 Functional and pathway enrichment analyses

Functional enrichment analysis was performed on the upregulated and downregulated genes in the Gram-positive or Gram-negative samples separately. For the downregulated genes in the Gram-positive samples and Gram-negative samples, MCL1 was significantly associated with the functions of apoptosis and programmed cell death regulation. For the upregulated genes in the Gram-positive and Gram-negative samples, NDUFS4 was significantly associated with mitochondrial respiratory chain complex I assembly. Additionally, NDUFB2, NDUFB8 and ubiquinol-cytochrome c reductase hinge protein (UQCRH) were significantly enriched in the functions of cellular respiration, ATP synthesis coupled electron transport and respiratory electron transport chain in Gram-negative samples. LATS2 was significantly associated with the G1/S transition of the mitotic cell cycle in Gram-positive samples (Tables II and III). KEGG pathway enrichment analysis was also conducted for up and downregulated genes in Gram-positive and Gram-negative samples (Tables IV and V). NDUFS4 was significantly enriched in the pathway of oxidative phosphorylation.

 Significantly differential functions screening

Based on the Euclidean distance of the biological functions, as well as the P-values of the 10,000 stochastic perturbations between Gram-positive samples and Gram-negative samples, a total of 10 significantly differential functions were obtained, including cellular respiration (P<1.00×10−8, Euclidean distance=1.156277), ATP synthesis coupled electron transport (P<1.00×10−8, Euclidean distance=1.156277) and G1/S transition of mitotic cell cycle (P=0.015, Euclidean distance = 0.554799; Table VI).

 Discussion

In line with the results of Tang et al (8), the present study determined that there was no significant difference in the expression profile between Gram-positive and gram-negative samples from hierarchical clustering analysis. In the Gram-positive and Gram-negative samples, the GO functional enrichment analysis revealed that MCL1 was significantly associated with the regulation of apoptosis and programmed cell death. A previous study has determined that the apoptosis of T-cells may induce the breakdown of defense mechanisms resulting in sepsis (27). Additionally, the inhibition of programmed cell death may reverse T-cell exhaustion and thus eradicate the invading pathogens which cause sepsis (28). Additionally, MCL1 may also be associated with the reduction of apoptosis of neutrophils in patients with sepsis (29). Therefore, it is possible for MCL1 to be involved in sepsis via the regulation of T-cell apoptosis and programmed T-cell death in both Gram-positive and Gram-negative sepsis.

Additionally, the present study also determined that NDUFS4 was significantly associated with mitochondrial respiratory chain complex I assembly. Mitochondrial dysfunction may lead to oxidative stress and failure of energy production, which may result in organ dysfunction in sepsis (30). The KEGG pathway enrichment analysis revealed that NDUFS4 was significantly enriched in oxidative phosphorylation. Lee and Hüttemann (31) have determined that the inhibition of oxidative phosphorylation may lead to a reduction of the mitochondrial membrane potential, resulting in a lack of energy, which may cause organ failure and death in septic patients (31). NDUFS4 has been previously reported to be an important subunit of complex I which has a key role in oxidative phosphorylation (32). Additionally, NDUFS4 may participate in the regulation of sepsis induced by Gram-negative and Gram-positive bacteria through regulation of oxidative phosphorylation.

However, the present study identified specific DEGs in Gram-positive and Gram-negative samples compared with normal samples. According to the Euclidean distance and the stochastic perturbations performed between Gram-positive and Gram-negative samples, NDUFB2, NDUFB8 and UQCRH were significantly upregulated in the Gram-negative samples, whereas they were not upregulated in the Gram-positive samples. In addition, functional annotation revealed that they were significantly associated with cellular respiration, ATP synthesis coupled electron transport and mitochondrial electron transport, ubiquinol to cytochrome c. NDUFB2 and NDUFB8 are parts of the multisubunit mitochondrial NADH ubiquinone oxidoreductase (complex I) which has an important role in mitochondrial functioning (33,34). A previous study determined that a dysfunction of respiratory chain complex I may be associated with reactive oxygen species (ROS) production (35). Additionally, previous studies reported that ROS are toxic oxygen-containing molecules that may damage the cells and the antioxidant defense system, which is the pathogenesis of sepsis (36,37). UQCRH, which encodes the cytochrome b-c1 complex subunit 6 of complexes III (cytochrome c-oxidoreductase), is involved in the mitochondrial oxidative phosphorylation and the dysfunction of UQCRH may lead to breast and ovarian cancer by altering the function of the mitochondria (38,39). To the best of our knowledge, this is the first study investigating the functions of NDUFB2, NDUFB8 and UQCRH in Gram-negative bacteria-induced sepsis. The present study concluded that NDUFB2, NDUFB8 and UQCRH may be involved in the Gram-negative bacteria-induced sepsis by altering mitochondrial oxidative phosphorylation and may also be potential targets for the treatment of Gram-negative bacterial sepsis.

In addition, the function of the G1/S transition of the mitotic cell cycle was also determined to be significantly different between the Gram-positive and Gram-negative samples. LATS2 was enriched in this function and was significantly upregulated in patients with Gram-positive sepsis, whereas it was not significantly expressed in Gram-negative patients. LATS2, encoding serine/threonine-protein kinase, has been identified to inhibit the G1/S transition in the cell cycle of tumor cells (40). Additionally, G1 cell cycle arrest may be important for the initiation of kidney injury in sepsis (41). Therefore, LATS2 may be associated with Gram-negative bacterial sepsis by the modulation of G1/S transition in cell cycle.

In conclusion, MCL1, NDUFS5 and NDUFS4 may be potential target genes for the treatment of Gram-positive and Gram-negative bacterial sepsis. Additionally, NDUFB2, NDUFB8 and UQCRH may also be associated with Gram-negative bacterial sepsis. LATS2 may contribute to the progression of Gram-negative bacterial sepsis. However, further studies are still required in order to elucidate their action mechanisms in sepsis.

 Abbreviations DEGs

differentially expressed genes

 FDR

false discovery rate

 FC

fold-change

 ICU

intensive care unit

 PCC

Pearson's correlation coefficient

 KEGG

Kyoto Encyclopedia of Genes and Genomes

 References TaylorFBJrKinasewitzGTLupuFPathophysiology, staging and therapy of severe sepsis in baboon modelsJ Cell Mol Med16672682201210.1111/j.1582-4934.2011.01454.x219729703263329 BrodskáHMalíčkováKAdámkováVBenákováHŠťastnáMMZimaTSignificantly higher procalcitonin levels could differentiate Gram-negative sepsis from Gram-positive and fungal sepsisClin Exp Med13165170201310.1007/s10238-012-0191-822644264 DeutschmanCTraceyKSepsis: Current dogma and new perspectivesImmunity40463475201410.1016/j.immuni.2014.04.00124745331 LyleNHPenaOMBoydJHHancockREWBarriers to the effective treatment of sepsis: Antimicrobial agents, sepsis definitions, and host-directed therapiesAnn N Y Acad Sci1323101114201410.1111/nyas.1244424797961 HotchkissRSKarlIEThe pathophysiology and treatment of sepsisN Engl J Med348138150200310.1056/NEJMra02133312519925 KohEMLeeSGKimCKKimMYongDLeeKKimJMKimDSChongYMicroorganisms isolated from blood cultures and their antimicrobial susceptibility patterns at a university hospital during 1994–2003Korean J Lab Med272652752007(In Korean)10.3343/kjlm.2007.27.4.26518094587 GlikJKaweckiMGaździkTNowakMThe impact of the types of microorganisms isolated from blood and wounds on the results of treatment in burn patients with sepsisPol Przegl Chir84616201222472489 TangBMMcLeanASDawesIWHuangSJCowleyMJLinRCGene-expression profiling of gram-positive and gram-negative sepsis in critically ill patientsCrit Care Med3611251128200810.1097/CCM.0b013e3181692c0b18379237 MahabeleshwarGHQureshiMATakamiYSharmaNLingrelJBJainMKA myeloid hypoxia-inducible factor 1α-Krüppel-like factor 2 pathway regulates gram-positive endotoxin-mediated sepsisJ Biol Chem28714481457201210.1074/jbc.M111.31270222110137 Giamarellos-BourboulisEJvan de VeerdonkFLMouktaroudiMRaftogiannisMAntonopoulouAJoostenLAPickkersPSavvaAGeorgitsiMvan der MeerJWNeteaMGInhibition of caspase-1 activation in Gram-negative sepsis and experimental endotoxemiaCrit Care15R27201110.1186/cc9974212446703222063 KagerLMWeehuizenTAWiersingaWJRoelofsJJMeijersJCDondorpAMvan't VeerCvan der PollTEndogenous α2-antiplasmin is protective during severe gram-negative sepsis (melioidosis)Am J Respir Crit Care Med188967975201310.1164/rccm.201307-1344OC23992406 KagerLMWiersingaWJRoelofsJJde BoerOJWeilerHvan't VeerCvan der PollTA thrombomodulin mutation that impairs active protein C generation is detrimental in severe pneumonia-derived gram-negative sepsis (melioidosis)PLoS Negl Trop Dis8e2819201410.1371/journal.pntd.0002819247627403998929 AfsariBGemanDFertigEJLearning dysregulated pathways in cancers from differential variability analysisCancer Inform13(Suppl 5)S61S672014 CalandraTCohenJInternational Sepsis Forum Definition of Infection in the ICU Consensus Conference: The international sepsis forum consensus conference on definitions of infection in the intensive care unitCrit Care Med3315381548200510.1097/01.CCM.0000168253.91200.8316003060 MurieCBaretteCLafanechèreLNadonRControl-plate regression (CPR) normalization for high-throughput screens with many active featuresJ Biomol Screen19661671201410.1177/108705711351600324352083 RitchieMEPhipsonBWuDHuYLawCWShiWSmythGKlimma powers differential expression analyses for RNA-sequencing and microarray studiesNucleic Acids Res43e47201510.1093/nar/gkv007256057924402510 ChenHBoutrosPCVennDiagram: A package for the generation of highly-customizable Venn and Euler diagrams in RBMC Bioinformatics1235201110.1186/1471-2105-12-35212695023041657 Gene Ontology ConsortiumBlakeJADolanMDrabkinHHillDPLiNSitnikovDBridgesSBurgessSBuzaTGene ontology annotations and resourcesNucleic Acids Res41(Database issue)D530D535201310.1093/nar/gks105023161678 da HuangWShermanBTLempickiRASystematic and integrative analysis of large gene lists using DAVID bioinformatics resourcesNat Protoc44457200810.1038/nprot.2008.211 ZhouJDongXZhouQWangHQianYTianWMaDLiXmicroRNA expression profiling of heart tissue during fetal developmentInt J Mol Med3312501260201424604530 ZhaiYTchieuJSaierMHJrA web-based tree view (TV) program for the visualization of phylogenetic treesJ Mol Microbiol Biotechnol46970200211763971 WeaverBWuenschKLSPSS and SAS programs for comparing Pearson correlations and OLS regression coefficientsBehav Res Methods45880895201310.3758/s13428-012-0289-723344734 DemchakBHullTReichMLiefeldTSmootMIdekerTMesirovJPCytoscape: The network visualization tool for GenomeSpace workflowsF1000Res31512014251655374133763 KoteraMMoriyaYTokimatsuTKanehisaMGotoSKEGG and GenomeNet, New DevelopmentsMetagenomic Analysis3293392015 LibertiLLavorCMaculanNMucherinoAEuclidean distance geometry and applicationsQuantitative Biol563692012 BódaiTAltmannEGEndlerAStochastic perturbations in open chaotic systems: Random versus noisy mapsPhys Rev E Stat Nonlin Soft Matter Phys87042902201310.1103/PhysRevE.87.04290223679484 SchmidtMVPaulusPKuhnAMWeigertAMorbitzerVZacharowskiKKempfVABrüneBvon KnethenAPeroxisome proliferator-activated receptor γ-induced T cell apoptosis reduces survival during polymicrobial sepsisAm J Respir Crit Care Med1846474201110.1164/rccm.201010-1585OC21471100 ChangKSvabekCVazquez-GuillametCSatoBRascheDWilsonSRobbinsPUlbrandtNSuzichJGreenJTargeting the programmed cell death 1: Programmed cell death ligand 1 pathway reverses T cell exhaustion in patients with sepsisCrit Care18R3201410.1186/cc13176243876804056005 HärterLMicaLStockerRTrentzOKeelMMcl-1 correlates with reduced apoptosis in neutrophils from patients with sepsisJ Am Coll Surg197964973200310.1016/j.jamcollsurg.2003.07.00814644285 GalleyHOxidative stress and mitochondrial dysfunction in sepsisBr J Anaesth1075764201110.1093/bja/aer09321596843 LeeIHüttemannMEnergy crisis: The role of oxidative phosphorylation in acute inflammation and sepsisBiochim Biophys Acta184215791586201410.1016/j.bbadis.2014.05.031249057344147665 PetruzzellaVVergariRPuzziferriIBoffoliDLamanteaEZevianiMPapaSA nonsense mutation in the NDUFS4 gene encoding the 18 kDa (AQDQ) subunit of complex I abolishes assembly and activity of the complex in a patient with Leigh-like syndromeHum Mol Genet10529535200110.1093/hmg/10.5.52911181577 UgaldeCHinttalaRTimalSSmeetsRRodenburgRJUusimaaJvan HeuvelLPNijtmansLGMajamaaKSmeitinkJAMutated ND2 impairs mitochondrial complex I assembly and leads to Leigh syndromeMol Genet Metab901014200710.1016/j.ymgme.2006.08.00316996290 BottleyAPhillipsNMWebbTEWillisAESpriggsKAeIF4A inhibition allows translational regulation of mRNAs encoding proteins involved in Alzheimer's diseasePLoS One5pii: e13030201010.1371/journal.pone.0013030209273852946912 DröseSBrandtUMolecular mechanisms of superoxide production by the mitochondrial respiratory chainMitochondrial Oxidative PhosphorylationSpringer145169201210.1007/978-1-4614-3573-0_6 Van RaamsdonkJMHekimiSSuperoxide dismutase is dispensable for normal animal lifespanProc Natl Acad Sci USA10957855790201210.1073/pnas.1116158109224519393326508 SchulteJStruckJKöhrleJMüllerBCirculating levels of peroxiredoxin 4 as a novel biomarker of oxidative stress in patients with sepsisShock35460465201110.1097/SHK.0b013e3182115f4021283059 OwensKMKulawiecMDesoukiMMVanniarajanASinghKKImpaired OXPHOS complex III in breast cancerPLoS One6e23846201110.1371/journal.pone.0023846219011413162009 ModenaPTestiMAFacchinettiFMezzanzanicaDRadiceMTPilottiSSozziGUQCRH gene encoding mitochondrial Hinge protein is interrupted by a translocation in a soft-tissue sarcoma and epigenetically inactivated in some cancer cell linesOncogene2245864593200310.1038/sj.onc.120647212881716 LiYPeiJXiaHKeHWangHTaoWLats2, a putative tumor suppressor, inhibits G1/S transitionOncogene2243984405200310.1038/sj.onc.120660312853976 YangQHLiuDWLongYLiuHZChaiWZWangXTAcute renal failure during sepsis: Potential role of cell cycle regulationJ Infect58459464200910.1016/j.jinf.2009.04.00319428114

Volcano plots indicate the gene distribution of (A) Gram-negative samples and (B) Gram-positive samples. DEGs, differentially expressed genes; FC, fold-change.

Summary of the common differentially expressed genes between Gram-negative and Gram-positive samples, and their respective enriched functions.

Hierarchical clustering of specific DEGs in Gram-P, Gram-N and control samples. Gram N, Gram-negative samples; Gram P, Gram-positive samples.

Correlation network of the Gram-positive, Gram-negative and control samples. Green circles indicate control samples, brown circles indicate Gram-negative samples and purple circles indicate Gram-positive samples. The red lines indicate a positive correlation between the two samples and blue lines indicate a negative correlation between the two samples.

Top 10 upregulated and downregulated genes in patients with Gram-negative and Gram-positive sepsis.

A, Gram-negative
DEGs -log2 (P-value) logFC
Upregulated genes
  RPL27 3.735442   1.654937
  TM4SF1 2.691156   1.541493
  SEC11A 2.258718   1.518767
  PLOD2 2.255042   1.494549
  UQCRH 2.154078   1.446315
  AFP 1.889918   1.429409
  CDK5RAP2 3.965291   1.413672
  EPB41L4A-AS1 4.122053   1.411759
  SOD1 2.91784   1.40033
  CANX 4.013587   1.392226
Downregulated genes
  EVI2B 4.312471 −2.08489
  MME 5.521434 −1.73605
  ZBP1 5.974694 −1.56361
  LITAF 3.113137 −1.54349
  CYTH4 2.874971 −1.53818
  FBXL5 2.808339 −1.53
  CHI3L1 4.498941 −1.45407
  QPCT 4.411504 −1.45154
  TREM1 4.12983 −1.43918
  MXD1 3.392031 −1.41323
B, Gram-positive
DEGs −log2 (P-value) logFC
Upregulated genes
  SSBP1 1.896196279   1.5341469
  LAIR1 3.8569852   1.4491332
  MRPS18A 2.377785977   1.4338711
  NDUFC2 3.935542011   1.4093212
  CTSC 3.982966661   1.3851966
  MT1L 2.991399828   1.3843636
  TM4SF1 2.249491605   1.3772427
  FCHSD2 1.694648631   1.301164
  CYP1B1 2.460923901   1.297874
  NDUFA4 1.876148359   1.267824
Downregulated genes
  CHI3L1 6.359519 −1.92246
  EVI2B 3.271646 −1.71993
  MME 4.36251 −1.68036
  KCNB1 2.300162 −1.56111
  LITAF 2.415669 −1.44533
  FUS 2.767004 −1.42285
  QPCT 3.090444 −1.38819
  CKAP4 4.251812 −1.35146
  MCL1 4.221126 −1.34034
  EFHC2 3.458421 −1.32892

DEGS, differentially expressed genes; FC, fold-change.

Top 10 enriched GO terms for the upregulated and the downregulated genes in Gram-positive samples.

A, Upregulated genes
Term Function Count P-value Gene symbol
GO:0006412 Translation 22 3.52×10−11 AIMP1, EEF1A2, GARS, MRPS10, RPS27L, MRPS18C, MRPL15, MRPS18A, RPL22, MRPL27, EEF1E1, NARS2, EIF2S2, HARS, MRPL19, MRPL36, RPL10, RPL11, RSL24D1, MRPL32, EEFSEC, MRPL33
GO:0006120 Mitochondrial electron transport, NADH to ubiquinone   7 4.78×10−6 NDUFA4, NDUFS6, NDUFS5, NDUFS4, NDUFA8, NDUFA9, NDUFC2
GO:0042773 ATP synthesis coupled electron transport   7 2.62×10−5 NDUFA4, NDUFS6, NDUFS5, NDUFS4, NDUFA8, NDUFA9, NDUFC2
GO:0022904 Respiratory electron transport chain   7 5.65×10−5 NDUFA4, NDUFS6, NDUFS5, NDUFS4, NDUFA8, NDUFA9, NDUFC2
GO:0045333 Cellular respiration   8 7.34×10−5 NDUFA4, NDUFS6, NDUFS5, NDUFS4, NDUFA8, NDUFA9, NDUFC2, MDH1
GO:0044267 Cellular protein metabolic process 44 1.09×10−4 FASTK, FKBP3, PRDX4, MRPS10, RPS27L, CANX, LATS2, VRK1, PSMB7, PSMB6, MRPL15, PLOD2, NARS2, MRPL36, B3GNT1, MRPL19, RPL10, RPL11, RSL24D1, MRPL32, LOXL1, FGF2, MRPL33, HSP90AA1, AIMP1, EEF1A2, GARS, SOD1, LRPAP1, IKBKE, MAST4, HSP90B1, PPIH, MRPS18C, PSMA5, MRPL27, RPL22, MRPS18A, EEF1E1, EIF2S2, HARS, DSP, EEFSEC, FKBP2
GO:0016310 Phosphorylation 19 1.87×10−3 NDUFA4, NDUFA8, MVD, NDUFA9, FASTK, HK2, NDUFC2, PRDX4, SOD1, LATS2, NDUFS6, IKBKE, MAST4, VRK1, NDUFS5, NDUFS4, ATP5C1, ATP5A1, FGF2
GO:0000079 Regulation of cyclin-dependent protein kinase activity   5 2.51×10−3 GTPBP4, CDKN2C, CKS2, CDKN3, LATS2
GO:0033365 Protein localization in organelle   7 4.53×10−3 PPIH, MTX2, NDUFA13, FGF2, TIMM44, SEC61G, NR5A1
GO:0010257 NADH dehydrogenase complex assembly   3 4.73×10−3 NDUFAF4, NDUFS5, NDUFS4
B, Downregulated genes
Term Function Count P-value Gene symbol
GO:0010942 Positive regulation of cell death 17 1.51×10−6 PTGS2, PREX1, STK17B, RRAGA, PRKDC, NLRP3, NLRP1, SERINC3, NOTCH1, EI24, SSTR3, CASP4, DUSP1, CASP8, BNIP3L, FGD3, KALRN
GO:0043065 Positive regulation of apoptosis 16 6.31×10−6 PTGS2, PREX1, STK17B, PRKDC, NLRP3, NLRP1, SERINC3, NOTCH1, EI24, SSTR3, CASP4, DUSP1, CASP8, BNIP3L, FGD3, KALRN
GO:0043068 Positive regulation of programmed cell death 16 6.86×10−6 PTGS2, PREX1, STK17B, PRKDC, NLRP3, NLRP1, SERINC3, NOTCH1, EI24, SSTR3, CASP4, DUSP1, CASP8, BNIP3L, FGD3, KALRN
GO:0042981 Regulation of apoptosis 20 8.53×10−5 PTGS2, MCL1, PREX1, STK17B, PRKDC, PIM2, NLRP3, NLRP1, SERINC3, NOTCH1, EI24, SSTR3, CASP4, DUSP1, IGF2R, BNIP3L, CASP8, DLG5, FGD3, KALRN
GO:0043067 Regulation of programmed cell death 20 9.73×10−5 PTGS2, MCL1, PREX1, STK17B, PRKDC, PIM2, NLRP3, NLRP1, SERINC3, NOTCH1, EI24, SSTR3, CASP4, DUSP1, IGF2R, BNIP3L, CASP8, DLG5, FGD3, KALRN
GO:0012502 Induction of programmed cell death 12 1.38×10−4 SERINC3, EI24, CASP4, SSTR3, PREX1, BNIP3L, CASP8, STK17B, NLRP3, NLRP1, FGD3, KALRN
GO:0009966 Regulation of signal transduction 20 2.68×10−4 LITAF, PREX1, KLK5, CYTH4, RABGAP1L, PIM2, TBC1D22A, OSM, ECE1, CXCR4, SOSTDC1, CASP8, GPSM1, RAMP1, RAPGEF1, RUNX2, ARAP1, FGD3, GNG7, KALRN
GO:0008277 Regulation of G-protein coupled receptor protein signaling pathway   5 1.10×10−3 ECE1, KLK5, GPSM1, RAMP1, GNG7
GO:0051056 Regulation of small GTPase mediated signal transduction   8 7.45×10−3 PREX1, CYTH4, RABGAP1L, RAPGEF1, FGD3, ARAP1, TBC1D22A, KALRN
GO:0010647 Positive regulation of cell communication   8 2.82×10−2 OSM, LAMA2, ECE1, PTGS2, LITAF, KLK5, CASP8, PIM2

GO, gene ontology.

Top 10 enriched GO terms for the upregulated and the downregulated genes in Gram-negative samples.

A, Upregulated genes
Term Function Count P-value Gene symbol
GO:0045333 Cellular respiration 15 8.97×10−9 UQCRC2, NDUFA8, NDUFB8, NDUFA6, NDUFA7, CYCS, NDUFC2, NDUFB2, NDUFS6, UQCR10, NDUFS5, NDUFS4, UQCRH, UQCRB, MDH1
GO:0042773 ATP synthesis coupled electron transport 12 1.31×10−8 NDUFS6, NDUFS5, UQCR10, NDUFS4, NDUFA8, UQCRH, NDUFB8, NDUFA6, NDUFA7, NDUFC2, UQCRB, NDUFB2
GO:0022904 Respiratory electron transport chain 12 5.68×10−8 NDUFS6, NDUFS5, UQCR10, NDUFS4, NDUFA8, UQCRH, NDUFB8, NDUFA6, NDUFA7, NDUFC2, UQCRB, NDUFB2
GO:0006412 Translation 24 2.78×10−7 MRPL3, RPL19, EEF1B2, EEF1A2, HBS1L, RPL15, MRPS10, RPL27, RPS27L, RPL22L1, IARS2, RPS7, MRPS18C, MRPL15, MRPS18A, RPL22, MRPL27, EEF1E1, MRPL19, RPL11, RSL24D1, MRPL32, MRPL33, RPS23
GO:0006457 Protein folding 17 6.33×10−7 HSP90AA1, FKBP5, FKBP4, FKBP3, TTC9, PDIA5, CCT3, LMAN1, CANX, LRPAP1, CCT7, PPIH, HSP90B1, SIL1, RUVBL2, HSPD1, FKBP2
GO:0006120 Mitochondrial electron transport, NADH to ubiquinone   9 1.70×10−6 NDUFS6, NDUFS5, NDUFS4, NDUFA8, NDUFB8, NDUFA6, NDUFA7, NDUFC2, NDUFB2
GO:0051437 Positive regulation of ubiquitin-protein ligase activity during mitotic cell cycle   9 6.75×10−5 CDK1, PSMB7, PSMD14, PSMB6, PSMA6, PSMA5, PSMC2, PSMA4, PSMD1
GO:0051351 Positive regulation of ligase activity   9 1.12×10−4 CDK1, PSMB7, PSMD14, PSMB6, PSMA6, PSMA5, PSMC2, PSMA4, PSMD1
GO:0051438 Regulation of ubiquitin-protein ligase activity   9 1.80×10−4 CDK1, PSMB7, PSMD14, PSMB6, PSMA6, PSMA5, PSMC2, PSMA4, PSMD1
GO:0051436 Negative regulation of ubiquitin-protein ligase activity during mitotic cell cycle   8 3.37×10−4 PSMB7, PSMD14, PSMB6, PSMA6, PSMA5, PSMC2, PSMA4, PSMD1
B, Downregulated genes
Term Function Count P-value Gene symbol
GO:0043067 Regulation of programmed cell death 22 3.47×10−6 IFIH1, PTGS2, MCL1, PREX1, TGFBR1, BCL2A1, STK17B, IFI16, NLRP3, NLRP1, TNFRSF9, CASP4, DUSP1, BTG2, IGF2R, BNIP3L, CHST11, CASP8, LRRK2, MX1, IFI6, KALRN
GO:0010942 Positive regulation of cell death 16 3.57×10−6 PTGS2, PREX1, TGFBR1, STK17B, RRAGA, IFI16, NLRP3, NLRP1, TNFRSF9, CASP4, DUSP1, CASP8, BNIP3L, MX1, LRRK2, KALRN
GO:0042981 Regulation of apoptosis 21 1.10×10−5 IFIH1, PTGS2, MCL1, PREX1, TGFBR1, BCL2A1, STK17B, IFI16, NLRP3, NLRP1, TNFRSF9, CASP4, DUSP1, BTG2, IGF2R, BNIP3L, CHST11, CASP8, MX1, IFI6, KALRN
GO:0043068 Positive regulation of programmed cell death 15 1.61×10−5 PTGS2, PREX1, TGFBR1, STK17B, IFI16, NLRP3, NLRP1, TNFRSF9, CASP4, DUSP1, CASP8, BNIP3L, MX1, LRRK2, KALRN
GO:0043065 Positive regulation of apoptosis 14 6.62×10−5 PTGS2, TGFBR1, PREX1, STK17B, IFI16, NLRP3, NLRP1, TNFRSF9, CASP4, DUSP1, CASP8, BNIP3L, MX1, KALRN
GO:0012502 Induction of programmed cell death 12 8.32×10−5 TNFRSF9, CASP4, PREX1, TGFBR1, BNIP3L, CASP8, STK17B, IFI16, MX1, NLRP3, NLRP1, KALRN
GO:0031401 Positive regulation of protein modification process   7 5.07×10−3 OSM, CCND3, TGFBR1, CD4, RICTOR, UBE2D1, LRRK2
GO:0010562 Positive regulation of phosphorus metabolic process   5 1.02×10−2 OSM, CCND3, TGFBR1, CD4, RICTOR
GO:0045937 Positive regulation of phosphate metabolic process   5 1.02×10−2 OSM, CCND3, TGFBR1, CD4, RICTOR
GO:0019048 Virus-host interaction   3 1.22×10−2 IRF7, RRAGA, CD4

Enriched pathways for the upregulated and the downregulated genes in Gram-negative samples.

A, Upregulated genes
Term Function Count P-value Gene symbol
hsa05012 Parkinson's disease 19 9.00×10−9 UQCRC2, NDUFA8, SLC25A5, NDUFA4L2, NDUFB8, SLC25A6, NDUFA6, COX7B, NDUFA7, CYCS, NDUFC2, NDUFB2, NDUFS6, UQCR10, NDUFS5, NDUFS4, UQCRH, ATP5A1, UQCRB
hsa05016 Huntington's disease 21 7.73×10−8 UQCRC2, NDUFA8, SLC25A5, NDUFA4L2, NDUFB8, POLR2K, SLC25A6, NDUFA6, CYCS, COX7B, NDUFA7, NDUFC2, SOD1, NDUFB2, NDUFS6, UQCR10, NDUFS5, NDUFS4, UQCRH, ATP5A1, UQCRB
hsa00190 Oxidative phosphorylation 17 4.25×10−7 UQCRC2, NDUFA8, NDUFA4L2, NDUFB8, NDUFA6, COX7B, NDUFA7, NDUFC2, NDUFB2, NDUFS6, UQCR10, NDUFS5, NDUFS4, UQCRH, ATP5A1, ATP5I, UQCRB
hsa05010 Alzheimer's disease 18 1.96×10−6 UQCRC2, NDUFA8, NDUFA4L2, NDUFB8, NDUFA6, COX7B, NDUFA7, CYCS, NDUFC2, NAE1, NDUFB2, NDUFS6, UQCR10, NDUFS5, NDUFS4, UQCRH, ATP5A1, UQCRB
hsa03050 Proteasome 9 3.61×10−5 PSMB7, PSMD14, PSMB6, PSMA6, PSMA5, PSMC2, PSMA4, SHFM1, PSMD1
hsa03010 Ribosome 10 6.12×10−4 RPL19, RPL22, RPL15, RPL27, RPS27L, RPL11, RSL24D1, RPL22L1, RPS23, RPS7
hsa00620 Pyruvate metabolism 6 4.60×10−3 LDHA, ACYP1, GLO1, ACAT2, PCK1, MDH1
hsa04260 Cardiac muscle contraction 7 2.05×10−2 UQCRC2, UQCR10, UQCRH, COX7B, ATP1A2, TNNI3, UQCRB
hsa04110 Cell cycle 9 2.21×10−2 CDK1, YWHAG, CDKN2C, YWHAQ, TFDP2, PCNA, CDK6, GADD45A, SMC3
hsa04115 p53 signaling pathway 6 3.94×10−2 CDK1, CYCS, CDK6, PERP, IGFBP3, GADD45A
B, Downregulated genes
Term Function Count P-value Gene symbol
hsa04622 RIG-I-like receptor signaling pathway 5 5.32×10−3 IFIH1, ISG15, IRF7, CASP8, IFNA8
hsa04612 Antigen processing and presentation 5 9.21×10–3 HSPA6, CD4, IFNA8, CTSS, HLA-F
hsa04620 Toll-like receptor signaling pathway 4 1.99×10−2 IRF7, CASP8, IFNA8, CD14
hsa04660 T cell receptor signaling pathway 4 2.33×10−2 PTPN6, RAF1, CD4, MAP3K14

Enriched pathways for the upregulated and the downregulated genes in Gram-positive samples.

A, Upregulated genes
Term Function Count P-value Gene symbol
hsa05012 Parkinson's disease 10 3.37×10−5 NDUFA4, NDUFS6, NDUFS5, NDUFS4, NDUFA8, SLC25A5, NDUFA9, NDUFC2, ATP5C1, ATP5A1
hsa05016 Huntington's disease 11 9.07×10−5 NDUFA4, NDUFS6, NDUFS5, NDUFS4, NDUFA8, SLC25A5, NDUFA9, NDUFC2, ATP5C1, ATP5A1, SOD1
hsa00190 Oxidative phosphorylation   9 2.43×10−4 NDUFA4, NDUFS6, NDUFS5, NDUFS4, NDUFA8, NDUFA9, NDUFC2, ATP5C1, ATP5A1
hsa05010 Alzheimer's disease   9 1.11×10−3 NDUFA4, NDUFS6, NDUFS5, NDUFS4, NDUFA8, NDUFA9, NDUFC2, ATP5C1, ATP5A1
hsa03010 Ribosome   5 2.58×10−2 RPL22, RPL10, RPS27L, RPL11, RSL24D1
hsa04612 Antigen processing and presentation   4 9.23×10−2 HSP90AA1, IFI30, HSPA4, CANX
hsa00970: Aminoacyl-tRNA biosynthesis   3 9.85×10–2 NARS2, HARS, GARS
B, Downregulated genes
Term Function Count P-value Gene symbol
hsa04650 Natural killer cell mediated cytotoxicity   6 9.37×10−3 PTPN6, ICAM2, RAF1, IFNA8, NFATC2, KLRD1
hsa04621 NOD-like receptor signaling pathway   4 2.25×10−2 IL8, CASP8, NLRP3, NLRP1
hsa04660 T cell receptor signaling pathway   4 2.91×10−2 PTPN6, RAF1, NFATC2, MAP3K14

Top 10 significant differential functions between Gram-negative samples and Gram-positive samples.

GO ID Term Euclidean distance P-value Gene symbols
GO:0006120 Mitochondrial electron transport, NADH to ubiquinone 1.156277 <1.00×10−8 NDUFB8, NDUFB2, NDUFA6, NDUFA7, NDUFA9, NDUFA4
GO:0042773 ATP synthesis coupled electron transport 1.156277 <1.00×10−8 UQCRH, NDUFB8, NDUFB2, UQCRB, NDUFA6, NDUFA7, UQCR10, NDUFA9, NDUFA4
GO:0022904 Respiratory electron transport chain 1.156277 <1.00×10−8 UQCRH, NDUFB8, NDUFB2, UQCRB, NDUFA6, NDUFA7, UQCR10, NDUFA9, NDUFA4
GO:0045333 Cellular respiration 1.156277 <1.00×10−8 UQCRH, NDUFB8, NDUFB2, UQCRC2, UQCRB, NDUFA6, NDUFA7, UQCR10, CYCS, NDUFA9, NDUFA4
GO:0016310 Phosphorylation 1.413364 1.00×10−3 UQCRH, PAK4, FGFR1, NDUFB8, NDUFB2, CDK1, CDK6, UQCRC2, GHR, IGFBP3, UQCRB, NDUFA6, NDUFA7, UQCR10, MET, ATP5I, MVD, NDUFA9, LATS2, MAST4, NDUFA4, IKBKE, ATP5C1
GO:0042981 Regulation of apoptosis 1.508092 2.70×10−3 IFI16, TNFRSF9, IFIH1, MX1, BTG2, CHST11, TGFBR1, BCL2A1, IFI6, LGALS1, PRDX1, PHLDA1, HSPD1, SORT1, MAL, DHCR24, GLO1, ITGB3BP, CDK1, GHR, SERPINB2, NQO1, ANXA1, IGFBP3, SMO, CADM1, CD44, KRT18, CYCS, NAE1, PERP, DLG5, EI24, PRKDC, NOTCH1, SERINC3, FGD3, PIM2, SSTR3
GO:0043067 Regulation of programmed cell death 1.517301 3.60×10−3 IFI16, TNFRSF9, IFIH1, MX1, BTG2, CHST11, LRRK2, TGFBR1, BCL2A1, IFI6, LGALS1, PRDX1, PHLDA1, HSPD1, SORT1, MAL, DHCR24, GLO1, ITGB3BP, CDK1, GHR, SERPINB2, NQO1, ANXA1, IGFBP3, SMO, CADM1, CD44, KRT18, CYCS, NAE1, PERP, DLG5, EI24, PRKDC, NOTCH1, SERINC3, FGD3, PIM2, SSTR3
GO:0000082 G1/S transition of mitotic cell cycle 0.554799 1.50×10−2 LATS2
GO:0044267 Cellular protein metabolic process 2.267426 2.37×10−2 RPS23, IARS2, HSPD1, PAK4, FGFR1, SEC11A, HEXB, FKBP5, PSMA4, MRPL3, RPL27, PSMC2, SIL1, SUPT3H, RPL27, RPS7, RPL19, FKBP4, LMAN1, PTPN1, CDK1, CDK6, GHR, RPL22L1, PDIA5, CCT7, FBXO7, ANXA1, IGFBP3, PSMA6, PSMD1, CCT3, HERC3, TTC9, MET, NAE1, PSMD14, HBS1L, RABGGTB, EEF1B2, RPL15, RUVBL2, AIMP1, LATS2, MAST4, EEFSEC, LOXL1, NARS2, HARS, B3GNT1, MRPL36, IKBKE, EIF2S2, DSP, RPL10, GARS
GO:0007517 Muscle organ development 1.050407 4.15×10−2 FAM65B, LAMA2, ANKRD2, FOXP1, CACNB4

GO, gene ontology.