Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Candidate genes on murine chromosome 8 are associated with susceptibility to Staphylococcus aureus infection in mice and are involved with Staphylococcus aureus septicemia in humans

  • Qin Yan,

    Affiliation Division of Infectious Diseases & International Health, Department of Medicine, Duke University School of Medicine, Durham, North Carolina, United States of America

  • Sun Hee Ahn,

    Affiliation Department of Biochemistry School of Dentistry, Chonnam National University, Bukgu, Gwangju, Korea

  • Felix Mba Medie,

    Affiliation Division of Infectious Diseases & International Health, Department of Medicine, Duke University School of Medicine, Durham, North Carolina, United States of America

  • Batu K. Sharma-Kuinkel,

    Affiliation Division of Infectious Diseases & International Health, Department of Medicine, Duke University School of Medicine, Durham, North Carolina, United States of America

  • Lawrence P. Park,

    Affiliation Duke Global Health Institute, Duke University, Durham, North Carolina, United States of America

  • William K. Scott,

    Affiliation Dr. John T. Macdonald Foundation Department of Human Genetics and John P. Hussman Institute for Human Genomics, University of Miami, Miami, Florida, United States of America

  • Hitesh Deshmukh,

    Affiliation Department of Pediatrics, University of Cincinnati, Cincinnati, Ohio, United States of America

  • Ephraim L. Tsalik,

    Affiliations Division of Infectious Diseases & International Health, Department of Medicine, Duke University School of Medicine, Durham, North Carolina, United States of America, Emergency Medicine Service, Durham Veteran’s Affairs Medical Center, Durham, North Carolina, United States of America, Duke Clinical Research Institute, Durham, North Carolina, United States of America

  • Derek D. Cyr,

    Affiliation Duke Clinical Research Institute, Durham, North Carolina, United States of America

  • Christopher W. Woods,

    Affiliations Division of Infectious Diseases & International Health, Department of Medicine, Duke University School of Medicine, Durham, North Carolina, United States of America, Duke Clinical Research Institute, Durham, North Carolina, United States of America, Section on Infectious Diseases, Durham Veteran’s Affairs Medical Center, Durham, North Carolina, United States of America

  • Chen-Hsin Albert Yu,

    Affiliation Division of Infectious Diseases & International Health, Department of Medicine, Duke University School of Medicine, Durham, North Carolina, United States of America

  • Carlton Adams,

    Affiliation Division of Infectious Diseases & International Health, Department of Medicine, Duke University School of Medicine, Durham, North Carolina, United States of America

  • Robert Qi,

    Affiliation Division of Infectious Diseases & International Health, Department of Medicine, Duke University School of Medicine, Durham, North Carolina, United States of America

  • Brenda Hansen,

    Affiliation Division of Infectious Diseases & International Health, Department of Medicine, Duke University School of Medicine, Durham, North Carolina, United States of America

  • Vance G. Fowler Jr

    vance.fowler@duke.edu

    Affiliations Division of Infectious Diseases & International Health, Department of Medicine, Duke University School of Medicine, Durham, North Carolina, United States of America, Duke Clinical Research Institute, Durham, North Carolina, United States of America

Candidate genes on murine chromosome 8 are associated with susceptibility to Staphylococcus aureus infection in mice and are involved with Staphylococcus aureus septicemia in humans

  • Qin Yan, 
  • Sun Hee Ahn, 
  • Felix Mba Medie, 
  • Batu K. Sharma-Kuinkel, 
  • Lawrence P. Park, 
  • William K. Scott, 
  • Hitesh Deshmukh, 
  • Ephraim L. Tsalik, 
  • Derek D. Cyr, 
  • Christopher W. Woods
PLOS
x

Abstract

We previously showed that chromosome 8 of A/J mice was associated with susceptibility to S. aureus infection. However, the specific genes responsible for this susceptibility are unknown. Chromosome substitution strain 8 (CSS8) mice, which have chromosome 8 from A/J but an otherwise C57BL/6J genome, were used to identify the genetic determinants of susceptibility to S. aureus on chromosome 8. Quantitative trait loci (QTL) mapping of S. aureus-infected N2 backcross mice (F1 [C8A] × C57BL/6J) identified a locus 83180780–88103009 (GRCm38/mm10) on A/J chromosome 8 that was linked to S. aureus susceptibility. All genes on the QTL (n~ 102) were further analyzed by three different strategies: 1) different expression in susceptible (A/J) and resistant (C57BL/6J) mice only in response to S. aureus, 2) consistently different expression in both uninfected and infected states between the two strains, and 3) damaging non-synonymous SNPs in either strain. Eleven candidate genes from the QTL region were significantly differently expressed in patients with S. aureus infection vs healthy human subjects. Four of these 11 genes also exhibited significantly different expression in S. aureus-challenged human neutrophils: Ier2, Crif1, Cd97 and Lyl1. CD97 ligand binding was evaluated within peritoneal neutrophils from A/J and C57BL/6J. CD97 from A/J had stronger CD55 but weaker integrin α5β1 ligand binding as compared with C57BL/6J. Because CD55/CD97 binding regulates immune cell activation and cytokine production, and integrin α5β1 is a membrane receptor for fibronectin, which is also bound by S. aureus, strain-specific differences could contribute to susceptibility to S. aureus. Down-regulation of Crif1 with siRNA was associated with increased host cell apoptosis among both naïve and S. aureus-infected bone marrow-derived macrophages. Specific genes in A/J chromosome 8, including Cd97 and Crif1, may play important roles in host defense against S. aureus.

Introduction

An emerging body of evidence supports the concept that human genetic variation can influence host susceptibility to and outcome of infectious diseases. Examples of human genetic variation and susceptibility to specific infectious syndromes include susceptibility to severe sepsis in Chinese Han subjects with the rs1800629 variant of the TNF gene [1], genetic variants of TRAF6 and increased susceptibility to sepsis-induced acute lung injury [2], variants in β2-adrenocepter and an increased susceptibility to bacterial meningitis [3], Toll-like receptor variants associated with both infectious and autoimmune diseases [4], and IL17A variation in association with susceptibility to Gram-positive infection and severe sepsis [5].

However, the genetic basis for variation in host susceptibility to S. aureus remains largely unknown. We [68] and others [9, 10] have reported the different susceptibility to S. aureus in various inbred mouse strains. For example, A/J is highly susceptible to S. aureus infection while C57BL/6J is resistant [9]. These two strains thus provide a unique platform to investigate the host genetic determinants associated with susceptibility to S. aureus infection. Using these strains, we previously reported that the genetic determinants of susceptibility to S. aureus in A/J mice localized to chromosomes 8, 11, and 18 [6, 8] and identified candidate susceptibility genes on chromosomes 11 [8] and 18 [6].

In the present investigation, we used a multipronged strategy to identify genes associated with susceptibility to S. aureus infection on murine chromosome 8. We initially localized the region on chromosome 8 associated with S. aureus susceptibility by quantitative trait locus (QTL) mapping. Having narrowed down our investigation to this region of ~ 100 genes, we employed a comprehensive approach to address multiple potential mechanisms by which genetic variation could result in our phenotype of interest. First, we considered the possibility that genetic susceptibility to S. aureus was due to genes that were only differentially expressed between A/J and C57BL/6J in the setting of active S. aureus infection. Second, we considered the possibility that susceptibility to S. aureus infection was due to genes that were differentially expressed between susceptible A/J and resistant C57BL/6J mice in both uninfected and S. aureus infected states. In our third approach, we considered the possibility that damaging single nucleotide polymorphisms (SNPs) in key genes influenced susceptibility to S. aureus infection. We then used whole blood gene expression data from a cohort of patients with S. aureus blood stream infection (BSI) to identify relevance of the candidate genes in human infection [11]. Finally, we evaluated the biological plausibility of our top two priority genes, Cd97 and Crif1, as important determinants of susceptibility to S. aureus infection.

Results

QTL mapping identifies locus on chromosome 8 linked to susceptibility to S. aureus

Previously we demonstrated that C57BL/6J inbred mice were resistant to S. aureus sepsis (median survival >5 days) [6]. By contrast, A/J and Chromosomal Substitution Strain 8 (CSS8) mice, which contain A/J chromosome 8 but are otherwise genetically C57BL/6J, were susceptible to S. aureus challenge, with significantly lower median survival (<2 days) and significantly higher kidney bacterial load 24 hours following S. aureus challenge[6]. The A/J derived allele on chromosome 8 that is responsible for S. aureus susceptibility is dominant, and F1 mice of CSS8 × C57BL/6J were susceptible to S. aureus [6]. Mouse gender did not influence S. aureus susceptibility [6]. In the current manuscript, we employed QTL mapping to localize genetic regions on chromosome 8 that are associated with susceptibility to S. aureus. Using 337 S. aureus-infected N2 (F1 [C8A] × C57BL/6J) backcross mice, QTL mapping identified a region on chromosome 8 that was significantly linked to survival time after S. aureus infection. This region was located between 83180780–88103009 (GRCm38/mm10) and contained approximately 102 genes (Fig 1). A/J and CSS8 also demonstrate similar patterns of susceptibility to infection with Escherichia coli (S1 Fig).

thumbnail
Fig 1. Chromosome substitution strain 8 (CSS8) mice were susceptible to S. aureus infection and QTL mapping found a region with putative candidate genes on chromosome 8.

Chromosome 8 LOD score plot for susceptibility to S. aureus in N2 backcross mice (F1 [C8A] × C57BL/6J). A total of 337 intercross mice (both sexes; age 6 to 8 weeks) were injected via intraperitoneal route with 107 CFU/g S. aureus Sanger 476 and observed every 8 hours continuously for 5 days. Thresholds for significant (p = 0.05) and suggestive (p = 0.63) linkage are indicated by the horizontal dashed lines. LOD score was determined by the J/qtl permutation test using 1,000 permuted data sets. The microsatellite markers for determining genotypes of N2 backcross mice are marked along the X-axis.

https://doi.org/10.1371/journal.pone.0179033.g001

Next, we sought to localize the basis of susceptibility to S. aureus within the QTL region by testing three possible sources of genetic variation: 1) genes within the QTL that are differentially expressed between susceptible A/J and resistant C57BL/6J mice only in S. aureus infected state; 2) genes within the QTL that are differentially expressed between susceptible A/J and resistant C57BL/6J mice in both uninfected and S. aureus infected states; and 3) presence of damaging SNPs in genes within the QTL that result in a damaged or dysfunctional gene product (Fig 2).

thumbnail
Fig 2. Overall strategies for identifying genes associated with S. aureus susceptibility on chromosome 8 of A/J mice.

Flow chart of the strategies for identifying S. aureus susceptible genes on chromosome 8 of A/J mice. Three different strategies were applied. Strategy 1 identified genes within the QTL region that were differentially expressed between A/J and C57BL/6J mice only in the setting of S. aureus infection by microarray and qPCR. Five candidates were identified by Strategy 1: D8Ertd738e, Ier2, Junb, Tbc1d9 and Zfp423. Strategy 2 identified genes within the QTL region that were differentially expressed between A/J and C57BL/6J mice in both uninfected and S. aureus infected states by microarray and qPCR. Six candidates were identified by Strategy 2: Crif1, Farsa, Inpp4b, Klf1, Nfix and Tnpo2. Strategy 3 identified damaging non-synonymous SNPs in A/J vs. C57BL/6J. Ten candidates were identified by Strategy 3: Dcaf15, Cd97, Farsa, Hook2, Klf1, Lyl1, Mylk3, Olfr370, Syce2 and Inpp4b.

https://doi.org/10.1371/journal.pone.0179033.g002

Identification of candidate genes within the QTL region

Strategy 1: Five genes within the identified QTL are significantly differentially expressed between A/J and C57BL/6J only in S. aureus infection and are validated by qPCR.

First, we considered the possibililty that candidate genes within the QTL would exhibit similar expression levels between uninfected susceptible and resistant inbred strains but would be differentially expressed in the setting of S. aureus infection. After adjustment for multiple comparisons in microarray results, 8 genes, all in A/J, exhibited similar expression levels between uninfected susceptible and resistant inbred strains but were significantly differentially expressed between pre-infection state and at 2 hours following S. aureus infection (S1 Table). Of these genes, 5 were validated by qPCR: D8ertd738; Ier2; JunB; Tbc1d9; Zfp423. These 5 genes exhibited significantly different qPCR-measured expression in A/J at 2hr following S. aureus infection vs uninfected A/J: D8ertd738 (2.58fold, p<0.05); Ier2 (3.51 fold, p<0.05); JunB (6.42 fold, p<0.01); Tbc1d9 (7.05 fold, p<0.05); and Zfp423 (2.29 fold, p<0.05) (Fig 3A). These 5 genes comprised our initial candidate gene list for Strategy 1 (Fig 2).

thumbnail
Fig 3. qPCR validation of murine candidate genes.

(A) qPCR validation of genes identified by Strategy 1. Genes within the QTL region that were differentially expressed between A/J and C57BL/6J mice only in the setting of S. aureus infection by microarray (S1 Table) underwent qPCR validation. qPCR validated five genes (D8ertd738, Ier2, JunB, Tbc1d9, Zfp423) with expression patterns in A/J and C57BL/6J (n = 4 in each group) that were consistent with microarray results. At 2 hours post S. aureus infection, the fold difference of A/J-2hr vs A/J-0hr was D8ertd738 (2.58fold, p<0.05), Ier2 (3.51 fold, p<0.05), JunB (6.42 fold, p<0.01), Tbc1d9 (7.05 fold, p<0.05) Zfp423 (2.29 fold, p<0.05). “*” represents p<0.05, “**” represents p<0.01. (B) qPCR validation of genes identified by Strategy 2. Genes within the QTL region that were differentially expressed between A/J and C57BL/6J mice in both uninfected and S. aureus infected states by microarray (S2 Table) underwent qPCR validation. qPCR validated six genes (Crif1, Farsa, Inpp4b, Klf1, Nfix, and Tnpo2) with expression patterns in A/J and C57BL/6J (n = 5 in each group) that were consistent with microarray results. At 3 hours post S. aureus infection, the fold difference of A/J vs C57BL/6J was Crif1 (0.6fold, p<0.05), Farsa (2.4fold, p<0.05), Inpp4b (4.5fold,p<0.05), Klf1 (2.4fold, p<0.05) Nfix (2.8fold, p<0.05), and Tnpo2 (3.5fold, p<0.05). “*” represents p<0.05. All mice were 8-week old males.

https://doi.org/10.1371/journal.pone.0179033.g003

Strategy 2: Six genes within the QTL region are significantly differentially expressed between A/J and C57BL/6J at all pre-infection and post-S. aureus infection timepoints and are validated by qPCR.

A total of 12 genes within the identified QTL region were significantly differentially expressed between susceptible A/J and resistant C57BL/6J at all pre-infection and post-infection time points from microarray (S2 Table). Of these, 6 were validated by qPCR: Crif1, Farsa, Inpp4b, Klf1, Nfix, and Tnpo2 (Fig 3B). Using qPCR at 3hr post S. aureus infection, Crif1 expression was significantly lower in A/J as compared with C57BL/6J (0.6fold; p<0.05), while other 5 genes exhibited significantly higher expression in A/J (Farsa [2.4fold, p<0.05], Inpp4b [4.5fold, p<0.05], Klf1 [2.4fold, p<0.05], Nfix [2.8fold, p<0.05], and Tnpo2 [3.5fold, p<0.05]). These 5 genes comprised our candidate gene list for Strategy 2 (Fig 2).

Strategy 3: Identifying damaging SNPs within the QTL region of susceptible A/J or resistant C57BL/6J mice.

To consider the possibility that the genetic basis for susceptibility to S. aureus might be the presence of damaging SNPs, we performed S.I.F.T. analysis on all known non-synonymous mutations for genes within the QTL region on murine chromosome 8. A total of 10 genes within the QTL region were predicted to contain damaging non-synonymous sequence variants in the A/J or C57BL/6NJ mouse strains: Dcaf15, Cd97, Farsa, Hook2, Klf1, Lyl1, Mylk3, Olfr370, Syce2, Inpp4b (Fig 2). Damaging SNPs were identified in A/J in 7 of the 10 genes (Dcaf15, Farsa, Hook2, Lyl1, Mylk3, Olfr370, Syce2) and in C57BL/6J in 5 of the 10 genes (Cd97, Hook2, Klf1, Mylk3, Inpp4b). These 10 genes comprised our candidate gene list for Strategy 3 (Fig 2).

Evidence for involvement of candidate genes in human S. aureus infection.

To provide evidence for the relevance of our findings in human infections, we evaluated whether human orthologues of the murine candidate genes identified by the three approaches were differentially expressed in patients with S. aureus BSI as compared to healthy human controls using gene expression data. The demographic and clinical details of these patients have been previously published (https://doi.org/10.1371/journal.pone.0048979.t001) [11]. Eleven of the putative candidate genes had human orthologues that were significantly differentially expressed between patients with S. aureus BSI (n = 32) and healthy subjects (n = 44) (Ier2, Rnaseh2a, Tbc1d9, Crif1, Farsa, Inpp4b, Tnpo2, Cd97, Hook2, Lyl2, Mylk3) (Fig 4). Three genes had increased levels of expression in S. aureus BSI patients: Cd97 (1.16-fold, p<0.05), Crif1(1.87-fold; p<0.0001), and Hook2 (1.29-fold, p = 0.0001), while other candidate genes were significantly down-regulated (Ier2: 0.93-fold, p<0.05; Rnaseh2a: 0.80-fold, p = 0.001; Tbc1d9: 0.82-fold, p<0.001; Lyl2: 0.72-fold, p<0.001; Farsa: 0.66-fold; p<0.0001; Inpp4b: 0.54-fold; p<0.0001; and Tnpo2: 0.84-fold; p<0.0001). Eight genes also showed similar significant changes in patients with Escherichia coli BSI (n = 19) (Fig 4).

thumbnail
Fig 4. Human orthologues of 11 candidate genes were significantly differentially expressed between patients with blood stream infection (BSI) due to S. aureus (S. A.), E. coli (E.C.) and healthy subjects (Control).

Human orthologues of 11 candidate genes (Crif1, Farsa, Inpp4b, Tnpo2, Cd97, Hook2, Ier2, Lyl1, Mylk3, Rnaseh2a and Tbc1d9) were significantly differentially expressed between patients with S. aureus BSI and healthy subjects by microarray. Human blood RNA from patients with S. aureus BSI (n = 32) and healthy subjects with no infection (n = 44) were extracted and analyzed and applied to microarray. The expression of Cd97 (1.17 fold; p<0.05), Crif1 (1.87 fold; p<0.0001), Hook2 (1.30 fold; p = 0.0001) were significantly higher in S. aureus BSI patients as compared with healthy controls. By contrast, the expression of Farsa (0.66fold; p<0.0001), Ier2 (0.93 fold; p<0.05), Inpp4b (0.54fold; p<0.0001), Lyl1 (0.72 fold; p<0.001), Rnaseh2a (0.81 fold; p = 0.001), Tbc1d9 (0.51 fold; p<0.0001) and Tnpo2 (0.84fold; p<0.0001), were significantly lower in S. aureus BSI patients. All of the 11 genes except Cd97 showed similar expression changes in Escherichia coli BSI (n = 19) patients.

https://doi.org/10.1371/journal.pone.0179033.g004

Four genes are differentially expressed in S. aureus-challenged human neutrophils.

Because the neutrophil is the primary host defense cell for management of S. aureus infection, we next used publically available GEO data to evaluate which genes were differentially expressed in human neutrophils when challenged by S. aureus (http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE16837). Of our 11 candidate genes found to be differentially expressed in patients with S. aureus BSI, four were also significantly differentially expressed in S. aureus–challenged human neutrophils (Ier2, Crif1, Cd97, and Lyl1) (Fig 5). In S. aureus-challenged human neutrophils, Cd97 was down-regulated 0.4 fold at 6 hours (p = 0.0015); Lyl1 was down-regulated 0.5 fold at 2 hours (p = 0.0133), 3 hours (p = 0.0251) and 6 hours (p = 0.0339); Ier2 was upregulated 2.4 fold at 1 hour (p = 0.0030) and 1.8 fold at 2 hours (p = 0.0056); and Crif1 was upregulated 8.7 fold at 6 hours (p = 0.0251).

thumbnail
Fig 5. Expression of candidate genes in human neutrophils upon S. aureus infection.

Expression of Cd97, Crif1, Ier2 and Lyl1 in S. aureus-challenged human neutrophils were significantly different as compared as naïve status. Human neutrophil data from public data set GEO:GSE16837 (http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE16837) was analyzed. Cd97 (0.44 fold, p<0.0001) and Lyl1 (0.48 fold, p<0.05) were decreased at 6hr after S. aureus stimulation. By contrast, Crif1 (8.70 fold, p<0.05) and Ier2 (1.77 fold, p<0.05) were significantly increased at 6hr and 2h, respectively, after S. aureus stimulation.

https://doi.org/10.1371/journal.pone.0179033.g005

Allele specific expression analysis.

Four 8-week A/J, C57BL/6J and CSS8 male mice were infected with S. aureus for 3 hours by intraperitoneal route. White blood cell RNA was extracted and applied to RNA-seq analysis. RNA-seq data was processed using the TrimGalore toolkit[12]. Principal component analysis found the distance difference among the three strains (S2A Fig). For the 11 CSS8 candidate genes, there was fairly even parental origin from either A/J and C57BL/6J (S2B Fig), indicating that the S. aureus susceptible phenotype was due to multiple gene effect instead of a single candidate gene.

Cd97 has different ligand binding ability between A/J and C57BL/6J.

CD97 is a cell membrane G-protein coupled receptor and its ligand binding affects its biological function. In this study, we focused on CD97/CD55 and CD97/integrin α5β1 binding, as these two are closely associated with host immune function [13, 14]. CD97 from C57BL/6J peritoneal neutrophils had significantly stronger integrin α5β1 binding ability as compared with A/J (Fig 6A), while its binding ability to CD55 was weaker (Fig 6B) (p<0.05 for both). Integrin α5β1 serves as a membrane receptor for matrix fibronectin, which also interacts with fibronectin binding protein (FnBPs) from S. aureus [15, 16]. The stronger CD97/integrin α5β1 binding in C57BL/6J than in A/J, suggests a more robust host-pathogen interaction in the resistant strain (C57BL/6J) vs the susceptible strain (A/J). The weaker binding of CD97/integrin α5β1 in A/J identified a very important and interesting target for studying host susceptibility to S. aureus. Because the binding of CD97 and CD55 helps regulate immune cell activation and increase proliferation and cytokine production, the signaling induced by this binding may play a role in the exaggerated cytokine production upon S. aureus challenge in A/J strain.

thumbnail
Fig 6. CD97 binding assay. CD97/integrin α5 (ITGAV) and CD97/CD55 binding ability was compared in A/J and C57BL/6J at both naïve and S. aureus infection condition.

(A) CD97-Integrin α5 binding assay. At the naïve status, peritoneal neutrophils in both A/J and C57BL/6J (n = 3 in each group) had higher CD97-Integrin α5 binding ability than in the S. aureus infected condition (p<0.05 in A/J; p<0.05 in C57BL/6J). C57BL/6J peritoneal neutrophils had higher CD97-Integrin α5 binding ability as compared with A/J peritoneal neutrophils (p<0.05). (B) CD97-CD55 binding assay. S. aureus infection reduced CD97-CD55 binding ability in C57BL/6J peritoneal neutrophils (p<0.05) (n = 3 in each group). All mice were 8-week old males.

https://doi.org/10.1371/journal.pone.0179033.g006

Crif1 is consistently expressed in S. aureus infected mice and humans.

Crif1 was significantly upregulated in S. aureus-infected mice (Fig 7), S. aureus-infected humans (Fig 4), and S. aureus-challenged human neutrophils (Fig 5). In S. aureus-infected A/J mice, Crif1 was upregulated 6.7 fold at 3 hours (p<0.05) and 1.8 fold at 6 hours (p = 0.4) (Fig 7). In S. aureus-infected C57BL/6J mice, Crif1 was upregulated 4.4 fold (p<0.05) at 3 hours and 2.5 fold (p = 0.14) at 6 hours (Fig 7). Collectively, these data support Crif1’s role in influencing host susceptibility to S. aureus infection.

thumbnail
Fig 7. Crif1 expression is upregulated in S. aureus-infected A/J and C57BL/6J mice by qPCR.

Consistent expression pattern of Crif1in A/J and C57BL/6J mouse (n = 5 in each group). Crif1 showed consistent uninfected vs infected expression patterns between mouse and humans (Figs 4 and 5) At 3 hours and 6 hours post S. aureus challenge, Crif1 is upregulated 6.7 fold (p<0.05) and 1.8 fold (p = 0.4) respectively in A/J, and 4.4 fold (p<0.05) and 2.5 fold (p = 0.14) in C57BL/6J. The normalization was conducted within each strain for comparing different time points (e.g. C57BL/6J time points were normalized to C57BL/6J pre-infection time point and A/J time points were normalized to A/J pre-infection time point.) All mice were 8-week old males.

https://doi.org/10.1371/journal.pone.0179033.g007

Apoptosis is increased in A/J bone marrow derived macrophages (BMDM), CSS8 BMDM, and Crif1 siRNA transfected BMDMs.

Next, we considered the possible biological basis for the association of Crif1 expression and susceptibility to S. aureus. Given Crif1’s key role in regulating apoptosis [17], and the importance of apoptosis in host cellular immunity [1821], we hypothesized that Crif1 influenced susceptibility to S. aureus by increasing apoptosis of host immune cells. To test this hypothesis, we first compared apoptosis of Bone Marrow Derived Macrophages (BMDM) in our susceptible and resistant mice strains. Rates of apoptosis were significantly higher among BMDMs from susceptible mice (A/J and CSS8) as compared to resistant mice (C57BL/6J) in both uninfected (A/J: 25.8%; CSS8: 18.7%; C57BL/6J: 12.6%; p <0.05) and S. aureus-infected (A/J: 23.3%; CSS8: 16.7%; C57BL/6J: 10.2%; p <0.05) conditions (Fig 8A). Next, we disrupted Crif1 expression by siRNA transfection of BMDMs. Apoptosis was significantly higher in Crif1 siRNA knockdown BMDMs as compared to BMDMs transfected with scramble siRNA in both naïve (Crif1-knockdown 38.7% vs scramble siRNA 17.3%; p<0.05) and S. aureus-stimulated conditions (Crif1-knockdown 40.6% vs scramble siRNA 32.3%; p<0.05) (Fig 8B). S. aureus stimulation did not change the apoptosis rate in primary murine (Fig 8A) or siRNA transfected BMDMs (Fig 8B) in our experimental conditions. These findings suggest that reduced Crif1 expression in A/J and CSS8 mice may contribute to their susceptibility to S. aureus infection through enhanced cellular apoptosis.

thumbnail
Fig 8. Increased apoptosis is associated with lower Crif1 expression in A/J and CSS8 macrophages.

Apoptosis rates in Figs 8A and 8B cannot be compared because transfection experiments, which intrinsically elicit apoptosis, were performed in (B) but not (A). (A) Bone-marrow derived macrophages from A/J and CSS8 demonstrate higher apoptosis level as compared with C57BL/6J in both uninfected (25.8% and 18.7% vs 12.6%; p <0.05) and S. aureus-challenged conditions (23.3% for A/J, 16.7% for CSS8 and 10.2% for C57BL/6J; p<0.05) (n = 5 in each group). (B) Knockdown of Crif1 by siRNA enhances apoptosis in both uninfected and S. aureus-challenged conditions. Knockdown of Crif1 in BMDMs from C57BL/6J mice enhances apoptosis in uninfected status as compared with scramble siRNA (38.7% vs 17.3%; p<0.05). S. aureus stimulation exhibits similar patterns of apoptosis (40.6% vs 32.3%; p<0.05) (n = 5 in each group). Mice were 8-week old males.

https://doi.org/10.1371/journal.pone.0179033.g008

Discussion

The genetic factors associated with host susceptibility to S. aureus infection remain largely unknown. Our study overcame the well-known inconsistencies between murine and human sepsis [22] by using a stringent trans-species selection strategy to identify four candidates from murine chromosome 8, and to establish a biological basis for influencing susceptibility for our top two candidates. In this way, Crif1 and Cd97 were identified as promising candidate genes associated with S. aureus susceptibility from murine chromosome 8 (Fig 9).

thumbnail
Fig 9. Proposed mechanism of susceptibility to S. aureus in A/J mice. Summary of candidate gene function in A/J mice responsible for S. aureus susceptibility.

Down-regulation of Dusp3 and Psme3 from chromosome 11 enhances inflammation upon S. aureus infection by activation of NF-κB signaling. Down-regulation of Crif1 from chromosome 8 compromises host immune defense against S. aureus by increasing apoptosis. Non-synonymous SNPs in Cd97 alters ligand binding ability. Down-regulation of Tnfaip8 and Seh1l from chromosome 18 increase the production of GM-CSF and lower expression of Tnfaip8 decreases the production of IL-1β.

https://doi.org/10.1371/journal.pone.0179033.g009

A robust body of evidence supports our conclusion that Crif1 contributes to host susceptibility to S. aureus infection. Crif1 was under-expressed in susceptible A/J mice as compared with resistant C57BL/6J mice. Crif1 expression was increased in response to S. aureus infection in both strains, in S. aureus-challenged human neutrophils, and in humans with S. aureus BSI. These findings suggest that Crif1 is important in host response to S. aureus and that its comparative under-production in A/J may contribute to that strain’s susceptibility to S. aureus. Substantial evidence also suggests that the biological basis for Crif1’s role in S. aureus susceptibility may involve apoptois [17]. BMDM from susceptible A/J and CSS8 mice were significantly more apoptotic than BMDM from resistant C57BL/6J strains. siRNA-mediated knockdown of Crif1 significantly increased apoptosis in both naïve and S. aureus challenged BMDMs. Because cell survival and death are fundamental parameters in the process of immune function [2325], Crif1 may be involved in host immune defense against S. aureus in both humans and mice.

The functional variety of Crif1 requires further detailed investigation on its biological relevance to S. aureus susceptibility. Crif1 plays a key role in mitochondrial homeostasis of host cell and oxidative phosphorylation. Crif1 is involved in regulation of oxidative phosphorylation and respiration by lymphocyte expansion molecule (LEM) to promote antigen-dependent CD8(+) T cell proliferation [26] and by lymphocyte-specific protein tyrosine kinase (Lck) to cause blood malignancies [27]. The fundamental effect of Crif1 on mitochondrial function is not limited to the immune system [28]. For example, reduced expression of Crif1 has been shown to play an important role in Alzheimer’s disease through regulation of Aβ-induced mitochondrial disruption [29], and Crif1 deficiency reduces adipose OXPHOS capacity and triggers inflammation and insulin resistance in mice [30]. Disruption of Crif1 in mouse islet beta cells leads to mitochondrial diabetes with progressive beta cell failure [31]. Finally, cardiomyocyte specific deletion of Crif1 causes mitochondrial cardiomyopathy in mice [32]. Collectively, these reports of Crif1’s effect on mitochondrial function suggest that its impact on susceptibility to S. aureus may occur in part by influencing central energy metabolism in the host.

Crif1 is also involved in major signaling pathways to modulate cell fate and functions. Crif1 affects PKA/CREB signaling pathway to promote adipogenic differentiation of bone marrow mesenchymal stem cells [33]. As PKA/CREB signaling is also fundamentally involved in different immune cells, Crif1 may indirectly affect host immune defense against S. aureus in both the innate [3436] and adaptive immune systems [37, 38]. For example, Crif1 is a novel transcriptional coactivator of STAT3 [17], a critical component of many cytokine receptor systems involved in pathogen resistance [3941]. Interestingly, mutations in STAT3 cause Job’s syndrome, which is characterized by recurrent S. aureus infections and hyper-IgE production [42, 43]. The current discovery suggests that Crif1 may indirectly affect S. aureus susceptibility through an autocrine or paracrine signaling pathway by affecting the JAK-STAT responsiveness to cytokines. Further, Crif1 is required in RNA interference and Dicer-2 stability [44] which are vital parts of host immune response to viruses and other foreign genetic material [45, 46].

Cd97 is a transmembrane G-Protein Coupled Receptor, characterized by an extended extracellular region to mediate cell-cell adhesion and interaction [47]. The function of Cd97 is critical for host immune defense, and upon activation lymphoid, myeloid cells and neutrophils increase Cd97 level to promote adhesion and migration [48]. The ligand binding of Cd97 initiates several important biological functions. For example, the binding between CD97 and CD55 has been shown to regulate granulocyte homeostasis [49], T-cell activation, proliferation and cytokine production [50, 51]. Interestingly, CD97 also binds to integrin α5β1, a cell surface fironection receptor, to regulate inflammatory cytokine production [14]. Because fibronection binding proteins are important mediators of S. aureus pathogenesis, the involvement of CD97 bridges the host and S. aureus interaction and serves as an important future target. Our flow cytometry study showed that peritoneal neutrophils from C57BL/6J have less CD55 binding than A/J, which indicates that the SNP of CD97 may predispose A/J to exaggerated cytokine production upon S. aureus challenge, rendering A/J mice susceptible to S. aureus infection. While on the other hand, the interaction between CD97 and integrin α5β1, the matrix fibronectin receptor, was enhanced in C57BL/6J, indicating that cells from C57BL/6J may have stronger capability to trap S. aureus through CD97-integrin α5β1-fibronetin-fibronection binding protein complex during S. aureus infection, providing a very promising candidate to study host- S. aureus interaction. Our future direction will mainly focus on the study of Cd97 function of initiating host immune defense against S. aureus.

Considerably less is known about the potential role of our remaining candidate genes, Ier2 and Lyl1, in determining host susceptibility to S. aureus. Ier2 is an immediate early response gene, affecting cell adhesion and motility [52, 53]. Although it has some DNA binding ability [54], nothing is known regarding its involvement in immunity or host defense. Lymphoblastomic leukemia 1 (Lyl1) is a transcription factor involved in hematopoietic stem cell function [55], and B cell differentiation [56]. The overexpression of Ly11 induces T- and B-cell lymphoma in mice [57]. We suspect that the function of Lyl1 in S. aureus susceptibility is mainly relevant to T- and B- cell. Although we were unable to evaluate biological plausibility for Lyl1 and Ier2 because of limited prior information about their role in immune function, future studies could expand our understanding of their role in this area. Thus, it is important that they not be ruled out as potential candidate genes.

The current study has limitations. First, it is possible that multiple genes within the QTL may contribute to the phenotype of interest. This possibility is supported by the robust number of differentially expressed genes in the QTL region, by the allele-specific expression analysis, and by our discovery of several potential candidate genes. Second, the essential function maintaining mitochondrial homeostasis of host cell and oxidative phosphorylation [26, 30, 32] provides the possibility that Crif1 is merely a general host responsive factor coping with stress during inflammation. Third, S. aureus colonization can increase a patient’s risk for subsequent S. aureus infection[58], and we did not limit control subjects to those who were colonized with S. aureus. However, given the fact that approximately one-third of all persons are colonized with S. aureus[58], it is likely that a many of our uninfected control subjects were in fact colonized with S. aureus. Fourth, the murine sepsis model can not fully address the full diversity of disease caused by S. aureus, such as skin and soft tissue infection, osteomyelitis and endocarditis [59, 60]. Thus, our consideration of survival as a dichotomous trait is likely to overly simplify the full complexity of susceptibility to bacterial infection. Alternately, our approach may disregard genes that contribute in combination to S. aureus susceptibility. Our current model of interaction for the identified S. aureus susceptibility genes on the three chromosomes is illustrated in Fig 9. To ultimately solve these limitations and mechanistically understand the biological relevance of candidate genes, additional experiments are underway in our lab, including defining the pathogenesis of Crif1, Dusp3, and Tnfaip8 using knockout mice. Fifth, our approach did not include the role of insertions/deletions in coding regions.

A growing number of studies have evaluated genetic susceptibility to S. aureus infections in humans. We reported a case-control Genome-wide Association study (GWAS) of 361 patients with SAB who were matched to 699 controls[61]. Ye et al reported a similar design with an outcome of any S. aureus infection (309 cases, 2925 controls)[62]. Neither identified SNPs reaching genome-wide significance, probably due to relatively small sample size. More recently, however, we have identified genetic variants within the HLA class II region in two distinct study populations that were associated with increased susceptibility to S. aureus infection at a level of genome-wide significance. Using a population of over 50,000 White subjects (4701 cases with S. aureus infection and 45,344 matched controls), we identified two imputed SNPs near HLA-DRA and HLA-DRB1 genes that were genome wide significant (rs115231074: p = 1,3 x 10−10 and rs35079132: p = 3.8 x 10−8) and one genotyped SNP that almost achieved genome-wide significance (rs4321864: p = 8.8 x 10−8).[63] Finally, we used admixture mapping to evaluate the impact of genetic variation on susceptibility to S. aureus infection in a cohort of African-Americans with SAB.[64] After empirical multiplicity adjustment, a single region in HLA class II was found to exhibit a genome-wide statistically significant increase in European ancestry, providing additional evidence for genetic variation influencing HLA-mediated immunity. Taken together with the findings of the current manuscript, it is likely that genetic susceptibility to S. aureus infection is complex and syndrome-specific.[65] Thus, the genetic variation found to be important in S. aureus sepsis may differ from that influencing pneumonia, soft tissue infection, or endocarditis.

Despite our study’s limitations, the present investigation makes several key observations. First, we have identified one QTL on chromosome 8 that is significantly linked to survival after infection with S. aureus. Among the 102 genes in the QTL that was associated with susceptibility to S. aureus, four show evidence of association in both S. aureus-infected mice and humans. Of these 4 genes, Crif1 and Cd97 also exhibit biological evidence for their relevance in S. aureus infection. Crif1 exhibited differential expression between naïve and S. aureus-infected mice; differential expression between susceptible and resistant mice; and had human orthologues that exhibited a consistent pattern of expression in patients with S. aureus BSI and in human neutrophils challenged with S. aureus. Biologically, several lines of evidence suggest that Crif1 influences susceptibility to S. aureus by apoptosis of host defense cells. Cd97 has damaging SNPs in C57BL/6J and had significant human orthologue expression in patients with S. aureus BSI. Ligand binding assay also shows the stronger CD97/integrin α5 binding ability in resistant strain but not in susceptible strain. Collectively, our results support a potential role of Crif1 and Cd97 in host response to S. aureus by affecting host cell fate during inflammation caused by S. aureus.

Materials and methods

Ethics statement

All animal experiments were approved by the Institutional Animal Care and Use Committee (IACUC Protocols A191-12-07/143-15-05) of Duke University. The most current edition of the Guide For The Care And Use of Laboratory Animals was followed when developing SOPs and policies. The the human studies referenced in this work were approved by Duke University Medical Center Institutional Review Board (Durham, NC), Durham VA Medical Center Institutional Review Board (Durham NC), and Henry Ford Hospital Institutional Review Board (Detroit MI). Written informed consent was obtained for all human subjects.

Human subjects

Subjects were enrolled at Duke University Medical Center (DUMC; Durham, NC), Durham VAMC (Durham, NC), and Henry Ford Hospital (Detroit, Michigan) as part of a prospective, NIH-sponsored study to develop novel diagnostic tests for severe sepsis and community-acquired pneumonia as mentioned before [6668]. All participants were adults. Detailed clinical information about these patients, including age and gender, has been previously published [11]. RNA was obtained from blood drawn at the time patients initially presented to the Emergency Department with sepsis. RNA expression data from patients who were ultimately found to have BSI with either S. aureus (n = 32) or E. coli (n = 19) were used in this study. Healthy controls were defined as uninfected human (n = 44), enrolled as part of a study on the effect of aspirin on platelet function among healthy volunteers [69]. Subjects were recruited through advertisements posted on the Duke campus. Blood used to derive gene expression data in these healthy controls was drawn prior to aspirin challenge. Human orthologs of murine genes were identified by Chip comparer (http://chipcomparer.genome.duke.edu/) as reported before [11]. When there were multiple orthologs, we preferentially used the anti-sense target probes that shared the fewest probes with other genes as identified by the probe label.

Mouse strains

C57BL/6J, A/J, and CSS8 mice were purchased from the Jackson Laboratory (Bar Harbor, ME). All the mice were allowed to acclimate for more than 7 days before experiments. For generation of F1 progeny, CSS8 mice were mated with C57BL/6J in reciprocal crosses [C57BL/6J male × CSS8 female and C57BL/6J female × CSS8 male] to generate an F1 population with heterozygous chromosome 8. To generate N2 backcross mice for QTL linkage analysis, F1 (C8A) mice were backcrossed with C57BL/6J to produce a total of 337 progeny that were used for phenotyping. Specific numbers of mice employed in experiments are provided in the tables and figures presenting the data.

Preparation of bacteria

S. aureus clinical strain, Sanger 476 was used in the mortality and infection studies. For preparation of S. aureus for injection, overnight culture of S. aureus was diluted 100 folds with fresh tryptic soy broth (TSB) and shake at 37°C with aeration to log-phase (OD600 ≈ 0.8). S. aureus was harvested by centrifugation at 3000rpm for 10 minutes at 4°C, washed once in DPBS and re-suspended in DPBS.

Murine sepsis experiment and bacterial load quantification

For murine peritonitis-sepsis experiments, 8-week-old male mice (n = 8) in each strain of C57BL/6J, A/J, and CSS8 were injected via intraperitoneal route with 107CFU/g S. aureus (Sanger 476) or 2×105 CFU/g E. coli (K1H7) and observed every 6 hours for morbidity continuously for 5 days.

QTL linkage analysis

Polymorphic microsatellite markers on chromosome 8 between C57BL/6J and A/J were chosen from a database maintained by Mouse Genomic Informatics (http://www.informatics.jax.org/). Seventeen microsatellite markers were selected with an average inter-marker distance of 0.65 cM covering chromosome 8. A total of 337 N2 backcross mice were generated, all of which were genotyped for each microsatellite marker by PCR amplification and gel electrophoresis. J/qtl software was used to analyze phenotype and genotype data for linkage of survival time after infection with S. aureus Sanger 476 and marker location. Phenotypes were defined as either sensitive or resistant based on the dichotomization of survival data (survival of less than 2 day is “0” and survival of longer than 2 days is “1”, respectively). All linkage analysis results were expressed as LOD scores. LOD score was considered “suggestive” if > = 0.43 (p = 0.63) and “significant” if > = 1.51 (p = 0.05). Threshold values for linkage were determined by a 1,000 permutation test by using J/qtl.

Microarray

Accession numbers for murine genes and their human orthologs were identified in NCBI and are provided in Table 1. The microarray data have been deposited in the NCBI GEO and are accessible through GEO series accession no. GSE19668 [6]. RNA integrity numbers (RIN) were calculated for all samples and found to be within tolerance limits (RIN > 7) (S3 Table). Post-processing of microarray data included several steps. The microarray gene expression data was analyzed using Partek Genomic Suite 6.4 software (Partek Inc., Louis, MO). All Affymetrix CEL files were imported and normalized using robust multiarray averaging (RMA). Analysis was performed using Analysis of Variance (ANOVA) and multi-test correction for p-values in Partek Genomic Suite. Differentially expressed genes between susceptible A/J and resistant C57BL/6J were identified a) at all pre-infection and post-infection time points (for Strategy 2) and b) only in S. aureus infection state (for Strategy 1). Student’s t-test was used to test for differential expression between 2 groups (eg, AJ mice at time 0 vs. time 2 hours, etc.). The lists of significant differentially expressed genes were generated based on a criterion of ≥2 relative fold change at a false discovery rate (FDR) of ≤ 5% as previously described[70].

Quantitative PCR

Blood was obtained by cardiac puncture from all mice strains. Total RNA was isolated using RNeasy kits (Qiagen) primed with random hexamer oligonucleotides and reversely transcribed using Invitrogen SuperScript II. Real-time quantitative PCR was performed using SYBR Green Mastermix (ABI). The qPCR primers for candidate genes are provided in S4 Table. All data were normalized to 18s rRNA (S5 and S6 Tables).

Flow cytometry analysis of apoptosis

Bone marrow-derived macrophages (BMDMs) were differentiated from male A/J, C57BL/6J and CSS8 as before [8]. 2 × 106 BMDMs were seeded to 6-well plate and cultured overnight. On the next day, PBS with S. aureus or same volume of PBS was added to BMDMs at MOI 10 and cultured for 1 hour. After washing twice with PBS, BMDMs were detached from plate by EDTA [71] and stained with FITC-Annexin-V and 7-AAD for 20 minutes before analysis through BD FACSCanto II. Double positive of Annexin-V and 7-AAD were determined as late apoptotic cells and further analyzed for their apoptosis rate.

Small interfering RNA (siRNA) experiments

siRNAs were purchased from Invitrogen and transfected into BMDMs of C57BL/6J by Lipofectamine RNAiMAX (Invitrogen) according to the manufacturer’s instructions as before [8]. Twenty-four hours post-transfection, cells were infected with S. aureus by MOI 10 for 1 hour and further analyzed of apoptosis. In parallel experiments, cells were harvested for RNA and qPCR analysis for their level of candidate genes. A full list of gene names and siRNA ID numbers are listed in S7 Table. Knockdown efficiency of SiRNA is shown in S3 Fig.

QTL region SNP functional analysis

The 1505 version of the Sanger Mouse Genomes Project sequence variation tool was used to identify all known non-synonymous sequence variants for the genes in the QTL region for the A/J and C57BL/6NJ mouse strains. This analysis utilized the C57BL/6J mouse strain as a reference to identify variants and was carried out at1 kbp resolution. These SNPs were then processed through the S.I.F.T. (Sorting Intolerant From Tolerant) program which is a platform that can be used to predict whether a specific amino acid substitution is a functionally damaging alteration [72]. Analyses for these non-synonymous SNPs were done using the program’s default threshold settings (cutoff = 0.05).

CD97 ligand binding assay

BD FACSCanto was used to evaluate ligand binding. Mouse peritoneal cells with or without S. aureus challenge were harvested and incubated with a mixture of recombinant mouse CD55 protein (Thermo Fisher) plus phycoerythrin-conjugated anti-mouse CD55 (BioLegend); or recombinant mouse integrin α5β1 (R&D Systems) plus Alexa Fluo 488-conjugated anti-mouse CD49e (α5) (BioLegend) for 30 minutes at room temperature following manufacturer’s instruction. Cells were assayed using the respective fluorescence channel.

RNA-seq

RNA quality and concentration were assessed with a Fragment Analyzer (Advanced Analytical) and Qubit 2.0 (ThermoFisher Scientific). For each sample, two hundred nanograms of total RNA was used for library construction. poly(A) mRNA capture and construction of stranded mRNA-seq libraries from intact total RNA (RIN numbers >7) was achieved using the commercially available KAPA Stranded mRNA-Seq library preparation Kit. In brief, mRNA transcripts were first captured using magnetic oligo-dT beads, fragmented using heat and magnesium, and reverse transcribed to produce dscDNA. Illumina standard sequencing adapters were then ligated to the dscDNA fragments and amplified to produce the final RNA-seq library. Libraries were indexed using a molecular indexing approach allowing for multiple libraries to be pooled and sequenced on the same sequencing lane on a HiSeq 4000 Illumina sequencing platform. After quality check of each individual library, the indexed libraries were diluted to 10nM, pooled at equimolar ratios and sequenced on 2 lanes of HiSeq 4000 with 50bp Single Read protocol. Data was demultiplexed and Fastq files were generated using BcltoFastq 2.19 script provided by Illumina.

RNA-seq data analysis

RNA-seq data was processed using the TrimGalore toolkit[12], which employs Cutadapt[73] to trim low quality bases and Illumina sequencing adapters from the 3’ end of the reads. Only reads that were 20nt or longer after trimming were kept for further analysis. Reads were mapped to the GRCm38v73 version of the mouse genome and transcriptome[74] using the STAR RNA-seq alignment tool[75]. Reads were kept for subsequent analysis if they mapped to a single genomic location. Gene counts were compiled using the HTSeq tool [76]. Only genes that had at least 10 reads in any given library were used in subsequent analysis. Normalization and differential expression was carried out using the DESeq2[77] Bioconductor[78] package with the R statistical programming environment[79]. We included batch and sex as cofactors in the differential expression model. The false discovery rate was calculated to control for multiple hypothesis testing. Gene set enrichment analysis[80] was performed to identify differentially regulated pathways and gene ontology terms for each of the comparisons performed.

Statistical analyses

Group differences in the distributions of continuous measures for candidate gene expression at times 0 and 2 hours were tested with Student’s T-test and the distributions for apoptosis were evaluated with the Mann Whitney-U test. The primary question of interest for the septic human cohort gene expression data was the presence of differences in the S. aureus-infected patients. For this reason, pairwise comparisons were made between healthy control subjects and either patients with S. aureus BSI or E. coli BSI. Differences in survival times of mice were examined with Kaplan-Meier plots and statistical differences in survival across different mouse strains were tested with the log-rank test. P-values smaller than 0.05 were considered statistically significant.

Supporting information

S1 Fig. Kaplan-Meier curve of E. coli intraperitoneal sepsis (n = 8 in each group).

Mice were 8-week old males.

https://doi.org/10.1371/journal.pone.0179033.s001

(TIF)

S2 Fig. Allele-specific expression analysis by RNA-seq of A/J, C57BL/6J, and F1 (CSS8 x C57BL/6J).

(A) PCA-plot of A/J, C57BL/6J and F1 (CSS8 x C57BL/6J). Principal component analysis for RNA-seq data. (B) Allele specific expression of the 11 candidate genes. For the 11 candidate genes in A/J chromosome 8 QTL, an even distribution of parental origins was observed in the F1 (CSS8 x C57BL/6J). (N = 4 male mice [8 week age] for each group.)

https://doi.org/10.1371/journal.pone.0179033.s002

(TIF)

S3 Fig. Knockdown efficiency of candidate genes.

https://doi.org/10.1371/journal.pone.0179033.s003

(TIF)

S1 Table. Genes exhibiting significantly different expression in response to S. aureus infection in A/J vs. C57BL/6J by microarray.

Expression data are provided for all 8 genes that were significantly differentially expressed in A/J but not C57BL/6J. No genes were significantly differentially expressed at 0 vs. 2 hours in only C57BL/6J. Multiple comparisons adjustments were applied using False Discovery Rates of ≤ 5%. 8-week old male A/J mice (n = 5 in each group) were used for experiments.

https://doi.org/10.1371/journal.pone.0179033.s004

(TIF)

S2 Table. Genes significantly differentially expressed between A/J and C57BL/6J mice at 0, 2, 4, 6, and 12 hours after infection with S. aureus.

Data previously published in S Table 1 of Ahn et al (https://doi.org/10.1371/journal.ppat.1001088.s007) [6].

https://doi.org/10.1371/journal.pone.0179033.s005

(TIF)

S4 Table. List of qPCR primers for candidate genes.

https://doi.org/10.1371/journal.pone.0179033.s007

(TIF)

S5 Table.

A. Ct values for qPCR results of 5 candidate genes identified by Strategy 1. Male 8 week-old mice were used (n = 4 in each group). B. 18s rRNA normalized Ct values for qPCR results of 5 candidate genes identified by Strategy 1. Male 8 week-old mice were used (n = 4 in each group).

https://doi.org/10.1371/journal.pone.0179033.s008

(TIFF)

S6 Table.

A. Ct values for qPCR results of 6 candidate genes identified by Strategy 2. Male 8 week-old mice were used (n = 5 in each group). B. 18s rRNA normalized Ct values for qPCR results of 6 candidate genes identified by Strategy 2. Male 8 week-old mice were used (n = 5 in each group).

https://doi.org/10.1371/journal.pone.0179033.s009

(TIFF)

S7 Table. List of siRNAs used in this study.

https://doi.org/10.1371/journal.pone.0179033.s010

(TIF)

Acknowledgments

We thank Dr. Deepak Voora for providing access to the gene expression data from the healthy human subjects used as controls in this study. We thank Dr. Sunil Suchindran and Dr. Thomas Burke for generous help with human orthologue studies.

Author Contributions

  1. Conceptualization: QY VGF SHA WKS HD.
  2. Data curation: FMM BKS ELT CWW C-HAY CA RQ BH.
  3. Formal analysis: DDC LPP QY CA SHA ELT.
  4. Funding acquisition: VGF.
  5. Investigation: QY SHA FMM BKS LPP ELT HD DDC CWW CA.
  6. Methodology: VGF QY BH SHA HD DDC WKS LPP.
  7. Project administration: VGF QY.
  8. Resources: VGF CWW ELT CA SHA HD.
  9. Software: QY SHA WKS DDC C-HAY CA.
  10. Supervision: VGF.
  11. Validation: QY VGF DDC LPP BKS.
  12. Visualization: VGF QY FMM BKS LPP WKS DDC RQ.
  13. Writing – original draft: VGF QY SHA WKS ELT DDC LPP CA.
  14. Writing – review & editing: FMM BKS HD ELT DDC CWW C-HAY RQ.

References

  1. 1. Song Z, Song Y, Yin J, Shen Y, Yao C, Sun Z, et al. Genetic variation in the TNF gene is associated with susceptibility to severe sepsis, but not with mortality. PLoS ONE [Electronic Resource]. 2012;7(9):e46113. pmid:23029405.
  2. 2. Song Z, Yao C, Yin J, Tong C, Zhu D, Sun Z, et al. Genetic variation in the TNF receptor-associated factor 6 gene is associated with susceptibility to sepsis-induced acute lung injury. Journal of Translational Medicine. 2012;10:166. pmid:22901274.
  3. 3. Adriani KS, Brouwer MC, Baas F, Zwinderman AH, van der Ende A, van de Beek D. Genetic variation in the beta2-adrenocepter gene is associated with susceptibility to bacterial meningitis in adults. PLoS ONE [Electronic Resource]. 2012;7(5):e37618. pmid:22624056.
  4. 4. Netea MG, Wijmenga C, O'Neill LA. Genetic variation in Toll-like receptors and disease susceptibility. Nature Immunology. 2012;13(6):535–42. pmid:22610250.
  5. 5. Nakada TA, Russell JA, Boyd JH, Walley KR. IL17A genetic variation is associated with altered susceptibility to Gram-positive infection and mortality of severe sepsis. Critical Care (London, England). 2011;15(5):R254. pmid:22026963.
  6. 6. Ahn SH, Deshmukh H, Johnson N, Cowell LG, Rude TH, Scott WK, et al. Two genes on A/J chromosome 18 are associated with susceptibility to Staphylococcus aureus infection by combined microarray and QTL analyses. PLoS Pathogens. 2010;6(9):e1001088. pmid:20824097
  7. 7. Johnson NV, Ahn SH, Deshmukh H, Levin MK, Nelson CL, Scott WK, et al. Haplotype Association Mapping Identifies a Candidate Gene Region in Mice Infected With Staphylococcus aureus. G3-Genes Genomes Genet. 2012;2(6):693–700.
  8. 8. Yan Q, Sharma-Kuinkel BK, Deshmukh H, Tsalik EL, Cyr DD, Lucas J, et al. Dusp3 and Psme3 Are Associated with Murine Susceptibility to Staphylococcus aureus Infection and Human Sepsis. PLoS Pathogens. 2014;10(6):e1004149. Epub 2014/06/06. pmid:24901344
  9. 9. von Kockritz-Blickwede M, Rohde M, Oehmcke S, Miller LS, Cheung AL, Herwald H, et al. Immunological mechanisms underlying the genetic predisposition to severe Staphylococcus aureus infection in the mouse model. American Journal of Pathology. 2008;173(6):1657–68. pmid:18974303
  10. 10. Sugi N, Whiston EA, Ksander BR, Gregory MS. Increased resistance to Staphylococcus aureus endophthalmitis in BALB/c mice: Fas ligand is required for resolution of inflammation but not for bacterial clearance. Infection & Immunity. 2013;81(6):2217–25. pmid:23569113.
  11. 11. Ahn SH, Tsalik EL, Cyr DD, Zhang Y, van Velkinburgh JC, Langley RJ, et al. Gene expression-based classifiers identify Staphylococcus aureus infection in mice and humans. PLoS ONE. 2013;8(1):e48979. pmid:23326304.
  12. 12. http://www.bioinformatics.babraham.ac.uk/projects/trim_galore.
  13. 13. Hamann J, Stortelers C, Kiss-Toth E, Vogel B, Eichler W, van Lier RA. Characterization of the CD55 (DAF)-binding site on the seven-span transmembrane receptor CD97. European Journal of Immunology. 1998;28(5):1701–7. pmid:9603477.
  14. 14. Wang T, Ward Y, Tian L, Lake R, Guedez L, Stetler-Stevenson WG, et al. CD97, an adhesion receptor on inflammatory cells, stimulates angiogenesis through binding integrin counterreceptors on endothelial cells. Blood. 2005;105(7):2836–44. pmid:15576472.
  15. 15. Chen D, Wang X, Liang D, Gordon J, Mittal A, Manley N, et al. Fibronectin signals through integrin alpha5beta1 to regulate cardiovascular development in a cell type-specific manner. Developmental Biology. 2015;407(2):195–210. pmid:26434918.
  16. 16. Tian H, Mythreye K, Golzio C, Katsanis N, Blobe GC. Endoglin mediates fibronectin/alpha5beta1 integrin and TGF-beta pathway crosstalk in endothelial cells. EMBO Journal. 2012;31(19):3885–900. pmid:22940691.
  17. 17. Kwon MC, Koo BK, Moon JS, Kim YY, Park KC, Kim NS, et al. Crif1 is a novel transcriptional coactivator of STAT3. EMBO Journal. 2008;27(4):642–53. pmid:18200042.
  18. 18. Taylor JJ, Pape KA, Steach HR, Jenkins MK. Humoral immunity. Apoptosis and antigen affinity limit effector cell differentiation of a single naive B cell. Science. 2015;347(6223):784–7. pmid:25636798.
  19. 19. Pedersen J, LaCasse EC, Seidelin JB, Coskun M, Nielsen OH. Inhibitors of apoptosis (IAPs) regulate intestinal immunity and inflammatory bowel disease (IBD) inflammation. Trends in Molecular Medicine. 2014;20(11):652–65. pmid:25282548.
  20. 20. Jaworska J, Coulombe F, Downey J, Tzelepis F, Shalaby K, Tattoli I, et al. NLRX1 prevents mitochondrial induced apoptosis and enhances macrophage antiviral immunity by interacting with influenza virus PB1-F2 protein. Proceedings of the National Academy of Sciences of the United States of America. 2014;111(20):E2110–9. pmid:24799673.
  21. 21. Coulombe F, Jaworska J, Verway M, Tzelepis F, Massoud A, Gillard J, et al. Targeted prostaglandin E2 inhibition enhances antiviral immunity through induction of type I interferon and apoptosis in macrophages. Immunity. 2014;40(4):554–68. pmid:24726877.
  22. 22. Seok J, Warren HS, Cuenca AG, Mindrinos MN, Baker HV, Xu W, et al. Genomic responses in mouse models poorly mimic human inflammatory diseases. Proceedings of the National Academy of Sciences of the United States of America. 2013;110(9):3507–12. pmid:23401516.
  23. 23. Stenqvist AC, Nagaeva O, Baranov V, Mincheva-Nilsson L. Exosomes secreted by human placenta carry functional Fas ligand and TRAIL molecules and convey apoptosis in activated immune cells, suggesting exosome-mediated immune privilege of the fetus. Journal of Immunology. 2013;191(11):5515–23. pmid:24184557.
  24. 24. Beug ST, Cheung HH, LaCasse EC, Korneluk RG. Modulation of immune signalling by inhibitors of apoptosis. Trends in Immunology. 2012;33(11):535–45. pmid:22836014.
  25. 25. Rolig AS, Carter JE, Ottemann KM. Bacterial chemotaxis modulates host cell apoptosis to establish a T-helper cell, type 17 (Th17)-dominant immune response in Helicobacter pylori infection. Proceedings of the National Academy of Sciences of the United States of America. 2011;108(49):19749–54. pmid:22106256.
  26. 26. Okoye I, Wang L, Pallmer K, Richter K, Ichimura T, Haas R, et al. T cell metabolism. The protein LEM promotes CD8+ T cell immunity through effects on mitochondrial respiration.[Erratum appears in Science. 2015 Oct 23;350(6259):aad6462; 26494762]. Science. 2015;348(6238):995–1001. pmid:25883318.
  27. 27. Vahedi S, Chueh FY, Chandran B, Yu CL. Lymphocyte-specific protein tyrosine kinase (Lck) interacts with CR6-interacting factor 1 (CRIF1) in mitochondria to repress oxidative phosphorylation. BMC Cancer. 2015;15:551. pmid:26210498.
  28. 28. Kim SJ, Kwon MC, Ryu MJ, Chung HK, Tadi S, Kim YK, et al. CRIF1 is essential for the synthesis and insertion of oxidative phosphorylation polypeptides in the mammalian mitochondrial membrane. Cell Metabolism. 2012;16(2):274–83. pmid:22819524.
  29. 29. Byun J, Son SM, Cha MY, Shong M, Hwang YJ, Kim Y, et al. CR6-interacting factor 1 is a key regulator in Abeta-induced mitochondrial disruption and pathogenesis of Alzheimer's disease. Cell Death & Differentiation. 2015;22(6):959–73. pmid:25361083.
  30. 30. Ryu MJ, Kim SJ, Kim YK, Choi MJ, Tadi S, Lee MH, et al. Crif1 deficiency reduces adipose OXPHOS capacity and triggers inflammation and insulin resistance in mice. PLoS Genetics. 2013;9(3):e1003356. pmid:23516375.
  31. 31. Kim YK, Joung KH, Ryu MJ, Kim SJ, Kim H, Chung HK, et al. Disruption of CR6-interacting factor-1 (CRIF1) in mouse islet beta cells leads to mitochondrial diabetes with progressive beta cell failure. Diabetologia. 2015;58(4):771–80. pmid:25660120.
  32. 32. Shin J, Lee SH, Kwon MC, Yang DK, Seo HR, Kim J, et al. Cardiomyocyte specific deletion of Crif1 causes mitochondrial cardiomyopathy in mice. PLoS ONE [Electronic Resource]. 2013;8(1):e53577. pmid:23308255.
  33. 33. Zhang X, Xiang L, Ran Q, Liu Y, Xiang Y, Xiao Y, et al. Crif1 Promotes Adipogenic Differentiation of Bone Marrow Mesenchymal Stem Cells After Irradiation by Modulating the PKA/CREB Signaling Pathway. Stem Cells. 2015;33(6):1915–26. pmid:25847389.
  34. 34. Li G, Harton JA, Zhu X, Ting JP. Downregulation of CIITA function by protein kinase a (PKA)-mediated phosphorylation: mechanism of prostaglandin E, cyclic AMP, and PKA inhibition of class II major histocompatibility complex expression in monocytic lines. Molecular & Cellular Biology. 2001;21(14):4626–35. pmid:11416140.
  35. 35. Cao A, Ramos Martinez JI, Barcia R. Implication of PKA and PKC in the activation of the haemocytes of Mytilus galloprovincialis Lmk by LPS and IL-2. Molecular Immunology. 2004;41(1):45–52. pmid:15140574.
  36. 36. Wall EA, Zavzavadjian JR, Chang MS, Randhawa B, Zhu X, Hsueh RC, et al. Suppression of LPS-induced TNF-alpha production in macrophages by cAMP is mediated by PKA-AKAP95-p105. Science Signaling [Electronic Resource]. 2009;2(75):ra28. pmid:19531803.
  37. 37. Pasqualucci L, Kitaura Y, Gu H, Dalla-Favera R. PKA-mediated phosphorylation regulates the function of activation-induced deaminase (AID) in B cells. Proceedings of the National Academy of Sciences of the United States of America. 2006;103(2):395–400. pmid:16387847.
  38. 38. Yu SH, Chiang WC, Shih HM, Wu KJ. Stimulation of c-Rel transcriptional activity by PKA catalytic subunit beta. Journal of Molecular Medicine. 2004;82(9):621–8. pmid:15197457.
  39. 39. Yu H, Pardoll D, Jove R. STATs in cancer inflammation and immunity: a leading role for STAT3. Nature Reviews Cancer. 2009;9(11):798–809. pmid:19851315.
  40. 40. Sansone P, Bromberg J. Targeting the interleukin-6/Jak/stat pathway in human malignancies. Journal of Clinical Oncology. 2012;30(9):1005–14. pmid:22355058.
  41. 41. Neurath MF, Finotto S. IL-6 signaling in autoimmunity, chronic inflammation and inflammation-associated cancer. Cytokine & Growth Factor Reviews. 2011;22(2):83–9. pmid:21377916.
  42. 42. Holland SM, DeLeo FR, Elloumi HZ, Hsu AP, Uzel G, Brodsky N, et al. STAT3 mutations in the hyper-IgE syndrome. New England Journal of Medicine. 2007;357(16):1608–19. pmid:17881745
  43. 43. Sowerwine KJ, Holland SM, Freeman AF. Hyper-IgE syndrome update. Annals of the New York Academy of Sciences. 2012;1250:25–32. pmid:22268731.
  44. 44. Lim SJ, Scott A, Xiong XP, Vahidpour S, Karijolich J, Guo D, et al. Requirement for CRIF1 in RNA interference and Dicer-2 stability. Rna Biology. 2014;11(9):1171–9. pmid:25483042.
  45. 45. Cullen BR, Cherry S, tenOever BR. Is RNA interference a physiologically relevant innate antiviral immune response in mammals? Cell Host & Microbe. 2013;14(4):374–8. pmid:24139396.
  46. 46. Fritz JH, Girardin SE, Philpott DJ. Innate immune defense through RNA interference. Science's Stke [Electronic Resource]: Signal Transduction Knowledge Environment. 2006;2006(339):pe27. pmid:16772641.
  47. 47. Langenhan T, Aust G, Hamann J. Sticky signaling—adhesion class G protein-coupled receptors take the stage. Science Signaling [Electronic Resource]. 2013;6(276):re3. pmid:23695165.
  48. 48. Leemans JC, te Velde AA, Florquin S, Bennink RJ, de Bruin K, van Lier RA, et al. The epidermal growth factor-seven transmembrane (EGF-TM7) receptor CD97 is required for neutrophil migration and host defense. Journal of Immunology. 2004;172(2):1125–31. pmid:14707087.
  49. 49. Veninga H, Hoek RM, de Vos AF, de Bruin AM, An FQ, van der Poll T, et al. A novel role for CD55 in granulocyte homeostasis and anti-bacterial host defense. PLoS ONE [Electronic Resource]. 2011;6(10):e24431. pmid:21984892.
  50. 50. Capasso M, Durrant LG, Stacey M, Gordon S, Ramage J, Spendlove I. Costimulation via CD55 on human CD4+ T cells mediated by CD97. Journal of Immunology. 2006;177(2):1070–7. pmid:16818763.
  51. 51. Abbott RJ, Spendlove I, Roversi P, Fitzgibbon H, Knott V, Teriete P, et al. Structural and functional characterization of a novel T cell receptor co-regulatory protein complex, CD97-CD55. Journal of Biological Chemistry. 2007;282(30):22023–32. pmid:17449467.
  52. 52. Xu Z, Zhu L, Wu W, Liao Y, Zhang W, Deng Z, et al. Immediate early response protein 2 regulates hepatocellular carcinoma cell adhesion and motility via integrin beta1-mediated signaling pathway. Oncology Reports. 2016;37(1):259–72. pmid:27840969.
  53. 53. Wu W, Zhang X, Lv H, Liao Y, Zhang W, Cheng H, et al. Identification of immediate early response protein 2 as a regulator of angiogenesis through the modulation of endothelial cell motility and adhesion. International Journal of Molecular Medicine. 2015;36(4):1104–10. pmid:26260137.
  54. 54. Neeb A, Wallbaum S, Novac N, Dukovic-Schulze S, Scholl I, Schreiber C, et al. The immediate early gene Ier2 promotes tumor cell motility and metastasis, and predicts poor survival of colorectal cancer patients. Oncogene. 2012;31(33):3796–806. pmid:22120713.
  55. 55. Souroullas GP, Goodell MA. A new allele of Lyl1 confirms its important role in hematopoietic stem cell function. Genesis: the Journal of Genetics & Development. 2011;49(6):441–8. pmid:21387538.
  56. 56. Capron C, Lecluse Y, Kaushik AL, Foudi A, Lacout C, Sekkai D, et al. The SCL relative LYL-1 is required for fetal and adult hematopoietic stem cell function and B-cell differentiation. Blood. 2006;107(12):4678–86. pmid:16514064.
  57. 57. Zhong Y, Jiang L, Hiai H, Toyokuni S, Yamada Y. Overexpression of a transcription factor LYL1 induces T- and B-cell lymphoma in mice. Oncogene. 2007;26(48):6937–47. pmid:17486074.
  58. 58. Peacock SJ, de Silva I, Lowy FD. What determines nasal carriage of Staphylococcus aureus? Trends in Microbiology. 2001;9(12):605–10. pmid:11728874
  59. 59. Jin W, Ibeagha-Awemu EM, Liang G, Beaudoin F, Zhao X, Guan le L. Transcriptome microRNA profiling of bovine mammary epithelial cells challenged with Escherichia coli or Staphylococcus aureus bacteria reveals pathogen directed microRNA expression profiles. BMC Genomics. 2014;15:181. pmid:24606609.
  60. 60. Jin T, Lu Y, He QX, Wang H, Li BF, Zhu LY, et al. The Role of MicroRNA, miR-24, and Its Target CHI3L1 in Osteomyelitis Caused by Staphylococcus aureus. Journal of Cellular Biochemistry. 2015;116(12):2804–13. pmid:25976273.
  61. 61. Nelson CL, Pelak K, Podgoreanu MV, Ahn SH, Scott WK, Allen AS, et al. A genome-wide association study of variants associated with acquisition of Staphylococcus aureus bacteremia in a healthcare setting. BMC Infectious Diseases. 2014;14:83. Epub 2014/02/15. pmid:24524581
  62. 62. Ye Z, Vasco DA, Carter TC, Brilliant MH, Schrodi SJ, Shukla SK. Genome wide association study of SNP-, gene-, and pathway-based approaches to identify genes influencing susceptibility to Staphylococcus aureus infections. Front. 2014;5:125. https://dx.doi.org/10.3389/fgene.2014.00125. pmid:24847357.
  63. 63. DeLorenze GN, Nelson CL, Scott WK, Allen AS, Ray GT, Tsai AL, et al. Polymorphisms in HLA Class II Genes Are Associated With Susceptibility to Staphylococcus aureus Infection in a White Population. Journal of Infectious Diseases. 2016;213(5):816–23. https://dx.doi.org/10.1093/infdis/jiv483. pmid:26450422.
  64. 64. Cyr DD, Allen AS, Du GJ, Ruffin F, Adams C, Thaden JT, et al. Evaluating genetic susceptibility to Staphylococcus aureus bacteremia in African Americans using admixture mapping. Genes & Immunity. 2017;18(2):95–9. https://dx.doi.org/10.1038/gene.2017.6. pmid:28332560.
  65. 65. Messina JA, Thaden JT, Sharma-Kuinkel BK, Fowler VG Jr. Impact of Bacterial and Human Genetic Variation on Staphylococcus aureus Infections. PLoS Pathogens. 2016;12(1):e1005330. https://dx.doi.org/10.1371/journal.ppat.1005330. pmid:26766507.
  66. 66. Zaas AK, Burke T, Chen M, McClain M, Nicholson B, Veldman T, et al. A host-based RT-PCR gene expression signature to identify acute respiratory viral infection. Science Translational Medicine. 2013;5(203):203ra126. Epub 2013/09/21. pmid:24048524.
  67. 67. Glickman SW, Cairns CB, Otero RM, Woods CW, Tsalik EL, Langley RJ, et al. Disease progression in hemodynamically stable patients presenting to the emergency department with sepsis. Academic Emergency Medicine. 2010;17(4):383–90. pmid:20370777
  68. 68. Langley RJ, Tsalik EL, van Velkinburgh JC, Glickman SW, Rice BJ, Wang C, et al. An integrated clinico-metabolomic model improves prediction of death in sepsis. Science Translational Medicine. 2013;5(195):195ra95. Epub 2013/07/26. pmid:23884467;
  69. 69. Voora D, Cyr D, Lucas J, Chi JT, Dungan J, McCaffrey TA, et al. Aspirin exposure reveals novel genes associated with platelet function and cardiovascular events. Journal of the American College of Cardiology. 2013;62(14):1267–76. pmid:23831034.
  70. 70. Benjamini Y, Drai D, Elmer G, Kafkafi N, Golani I. Controlling the false discovery rate in behavior genetics research. Behavioural Brain Research. 125(1–2):279–84. pmid:11682119.
  71. 71. Yan Q, Ahn SH, Fowler VG Jr. Macrophage Phagocytosis Assay of Staphylococcus aureus by Flow Cytometry. Bio-protocol. 2015;5(4):pii: e1406 pmid:25879046
  72. 72. Kumar P, Henikoff S, Ng PC. Predicting the effects of coding non-synonymous variants on protein function using the SIFT algorithm. Nature Protocols. 2009;4(7):1073–81. pmid:19561590.
  73. 73. Martin M. Cutadapt removes adapter sequences from high-throughput sequencing reads. 2011. 2011;17(1). pp. 10–12.
  74. 74. Kersey PJ, Staines DM, Lawson D, Kulesha E, Derwent P, Humphrey JC, et al. Ensembl Genomes: an integrative resource for genome-scale data from non-vertebrate species. Nucleic Acids Research. 2012;40(Database issue):D91–7. pmid:22067447.
  75. 75. Dobin A, Davis CA, Schlesinger F, Drenkow J, Zaleski C, Jha S, et al. STAR: ultrafast universal RNA-seq aligner. Bioinformatics. 2013;29(1):15–21. pmid:23104886.
  76. 76. http://www-huber.embl.de/users/anders/HTSeq/.
  77. 77. Love MI, Huber W, Anders S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biology. 2014;15(12):550. pmid:25516281.
  78. 78. Huber W, Carey VJ, Gentleman R, Anders S, Carlson M, Carvalho BS, et al. Orchestrating high-throughput genomic analysis with Bioconductor. Nature Methods. 2015;12(2):115–21. pmid:25633503.
  79. 79. www.r-project.org.
  80. 80. Mootha VK, Lindgren CM, Eriksson KF, Subramanian A, Sihag S, Lehar J, et al. PGC-1alpha-responsive genes involved in oxidative phosphorylation are coordinately downregulated in human diabetes. Nature Genetics. 34(3):267–73. pmid:12808457.