Identification of haplotype tag single nucleotide polymorphisms within the nuclear factor-κB family genes and their clinical relevance in patients with major trauma

Introduction Nuclear factor-κB (NF-κB) family plays an important role in the development of sepsis in critically ill patients. Although several single nucleotide polymorphisms (SNPs) have been identified in the NF-κB family genes, only a few SNPs have been studied. Methods A total of 753 patients with major blunt trauma were included in this study. Tag SNPs (tSNPs) were selected from the NF-κB family genes (NFKB1, NFKB2, RELA, RELB and REL) through construction of haplotype blocks. The SNPs selected from genes within the canonical NF-κB pathway (including NFKB1, RELA and REL), which played a critical role in innate immune responses were genotyped using pyrosequencing method and analyzed in relation to the risk of development of sepsis and multiple organ dysfunction (MOD) syndrome. Moreover, the rs842647 polymorphism was analyzed in relation to tumor necrosis factor α (TNF-α) production by peripheral blood leukocytes in response to bacterial lipoprotein stimulation. Results Eight SNPs (rs28362491, rs3774932, rs4648068, rs7119750, rs4803789, rs12609547, rs1560725 and rs842647) were selected from the NF-κB family genes. All of them were shown to be high-frequency SNPs in this study cohort. Four SNPs (rs28362491, rs4648068, rs7119750 and rs842647) within the canonical NF-κB pathway were genotyped, and rs842647 was associated with sepsis morbidity rate and MOD scores. An association was also observed between the rs842647 A allele and lower TNF-α production. Conclusions rs842647 polymorphism might be used as relevant risk estimate for the development of sepsis and MOD syndrome in patients with major trauma. Electronic supplementary material The online version of this article (doi:10.1186/s13054-015-0836-6) contains supplementary material, which is available to authorized users.


Introduction
Trauma is a major public health problem worldwide, ranking as the third leading cause of death. With great improvements in the emergency care system, survival after major trauma is common, but it is often complicated with sepsis and multiple organ dysfunction syndrome (MODS) [1]. Therefore, preventing sepsis and MODS is crucial in the treatment of patients who survive major trauma. It has been demonstrated that inappropriate immune inflammatory response contributes to the development of sepsis and MODS in major trauma patients [2]. Increasing evidence suggests that genetic variants, particularly single nucleotide polymorphisms (SNPs), are critical determinants for interindividual differences in both inflammatory responses and clinical outcome in sepsis patients [3,4]. Delineating the variation in genes and associated differences in response to trauma might assist in the risk stratification of trauma patients at the early stage of trauma in relation to the development of posttraumatic sepsis and MODS. NF-κB family regulates genes required for both the innate and adaptive immune responses [5,6]. The human NF-κB family contains five members: p50, p52, p65 (RelA), RelB, and c-Rel, which are encoded by NFKB1 [GenBank: NC_000004], NFKB2 [GenBank: NC_000010], RELA [GenBank: NG_029971], RELB [GenBank: NC_000019] and REL [GenBank: NC_000002] genes, respectively. Recent studies have unraveled the complexity of NF-κB activation by identifying two parallel activation pathways: the canonical and non-canonical pathways [7]. The canonical pathway mainly activates the transcription of many proinflammatory cytokine and chemokine genes that initiate and propagate innate immune responses [8]. The noncanonical pathway mainly regulates adaptive immune responses and lymphoid development. Therefore, the pivotal role of NF-κB canonical pathway in the development of sepsis makes it an interesting candidate for genetic analysis.
Growing evidence indicates that allelic variation within the NF-κB family genes may influence the magnitude of proinflammatory response, thereby affecting susceptibility to acute and chronic inflammatory diseases [9]. Up to now, the majority of studies have focused on cancer susceptibility. To date, Adamzik [10] and Schafer [11] have reported that the deletion allele of the NFKB1 insertion/ deletion (−94 ins/delATTG) polymorphism is associated with increased 30-day mortality in patients with severe sepsis and septic shock. However, little is known about the association of NF-κB family gene polymorphisms with the risk of posttraumatic sepsis and MODS.
In this study, we surveyed common SNPs (minor allele frequency >0.05) located within and around the five NF-κB family genes. We then selected four SNPs (NFKB1 rs28362491 ins/delATTG, rs4648068 A/G, RELA rs7119750 C/T, and REL rs842647 G/A) from the canonical pathway of NF-κB and investigated their clinical relevance in relation to the development of sepsis and MODS.

Study population
A total of 753 unrelated patients with major trauma were recruited in this study. All of the participants are ethnic Han Chinese and live in the Chongqing district, southwest China. The trauma patients were consecutively admitted to the Department of Trauma Surgery in the Daping Hospital and the Chongqing Emergency Medical Center between from January 1, 2005 to June 1, 2014. They were enrolled in the study if they met the following criteria: (1) age between 18 and 65 years, (2) expected injury severity score (ISS) greater than 16 and (3) probability of survival for longer than 48 hours. Patients were not eligible if they had penetrating injuries or preexisting cardiovascular, respiratory, renal, hepatic, hematologic, or immunologic diseases. ISS were calculated according to the 2005 abbreviated injury scale developed by independent evaluators [12]. All patients requiring operative intervention received standard surgical care and treatment postoperatively in the ICU. The protocol for this study was approved by the Ethical and Protocol Review Committee of the Third Military Medical University, and informed consent was obtained from the subjects or the patient's next of kin. Patient confidentiality was preserved according to the guidelines for studies of human subjects.

Clinical evaluation
The patients with major trauma were monitored prospectively after admission, by physicians who did not know the genotypes. A sepsis diagnosis was made if patients met all of the following criteria: clinical evidence of infection, body temperature >38.5°C or <36.5°C, and leukocyte count >10 × 10 9 /L or <4 × 10 9 /L. Infection was defined as a clinically obvious source or positive bacterial cultures. Pneumonia was diagnosed when a predominant organism was isolated from appropriately obtained sputum cultures in the setting of purulent sputum production and/or a new or changing pulmonary infiltrate on chest radiograph. Bloodstream infections were diagnosed based on isolation of a predominant organism from blood cultures obtained under sterile conditions. The criteria for urinary tract infections included the following: >10 white blood cells per highpower field on microscopic examination or isolation of >10 5 organisms/mL urine or >10 4 organisms and presence of symptoms. The criteria for catheter-related infections included isolation of >15 colony forming units from catheter tips cultured only in the setting of suspected infection. A wound infection was identified by drainage of purulent material from the wound. Daily physiologic and laboratory data were collected during the ICU stay and clinical events were recorded thereafter until death or hospital discharge. MOD scores were calculated as the sum of the simultaneously obtained individual organ scores on each hospital day [13]. Neurological scoring was not performed because every patient was sedated. The MOD scores and sepsis diagnoses were determined by individuals who did not know the patients' genotypes.

SNP selection
The full sequence of the human NFKB1, NFKB2, RELA, RELB and REL genes observed in the current study in-  [14] (www.ncbi.nlm.nih.gov/genbank/). Genetic variation data for the entire NFKB1, NFKB2, RELA, RELB and REL genes and their surrounding selected regions were obtained from the HapMap project [15] (www.hapmap.org) for 137 Han Chinese individuals from Beijing (CHB). From this database, a total of 325 SNPs (247 in NFKB1, 14 in NFKB2, 17 in RELA, 18 in RELB and 29 in REL) have been identified in the CHB population (Additional file 1), and 118 (91 in NFKB1, 3 in NFKB2, 7 in RELA, 13 in RELB and 4 in REL) were common SNPs with a minor allele frequency (MAF) more than or equal to 0.05 (Additional file 2). Haplotype blocks were constructed using Haploview, version 4.2 (Broad Institute of MIT and Harvard, Cambridge, MA, USA), a software package that provides computation of linkage disequilibrium (LD) statistics and population haplotype patterns from genotype data [16]. Haplotype blocks represent regions inherited without substantial recombination in the ancestors of the current population. The history of recombination between a pair of SNPs can be estimated with the use of the normalized measure of allelic association D' (value of D prime between the two loci) [17]. The criterion for the selected SNPs to construct a haplotype block is that all SNPs in one region must be in strong LD with D' of greater than 0.98 for the upper 95% confidence bound and greater than 0.7 for the lower bound. To determine the possible functionality of the haplotype tag SNPs (htSNPs) selected from the 5′-flanking region of the NF-κB family genes, online software was used to analyze the effect of these SNPs on potential transcription factor binding sites [18] (www.targetscan.org).

Genotyping of selected SNPs
Blood specimens were collected in tripotassium ethylenediaminetetraacetic acid sterile tubes from trauma patients immediately after admission, to avoid the effect of blood transfusion. The genomic DNA was isolated from whole blood using the Wizard genomic DNA purification kit (Promega, Madison, WI, USA) according to the manufacturer's protocol. DNA concentration in all samples was determined by ultraviolet spectrophotometry, adjusted with sterile distilled water to a 40 μg/mL concentration and stored at −80°C. Pyrosequencing was used for genotyping [19]. Genotyping was performed in a blinded fashion without knowledge of the patients' clinical data, and approximately 10% of the samples were genotyped in duplicate to monitor genotyping quality.

Ex vivo TNF-α production
A human whole-blood assay was used as described by Majetschak et al. [20]. In brief, aliquots of whole blood collected from the trauma patients immediately after admission were mixed 1:1 with Roswell Park Memorial Institute (RPMI) 1640 culture medium (Thermo Scientific, Beijing, China), and incubated with 100 ng/mL lipopolysaccharide (LPS) (Escherichia coli O26:B6; Difco Laboratories, Detroit, MI, USA) in a sample mixer at 37°C for 4 hours. After centrifugation, the supernatants were aspirated and aliquoted for storage at −80°C. TNF-α in the supernatants was assayed with a sandwich ELISA, according to the manufacturer's instructions (Endogen, Woburn, MA, USA). The detection limits of the assay were 4 pg/mL.

Statistical analysis
Sample size was calculated using the online Power and Sample Size software program [21] (http://biostat.mc. vanderbilt.edu/wiki/Main/PowerSampleSize). The desired power of our study was set at 80% with a significance level of 0.05 calculated using a two-sided test. We chose the log-additive inheritance model, which is the most suitable one for polygenic diseases. On the basis of our calculations using the Power and Sample Size software program, our sample (n = 753) was considered adequate to study the selected SNPs of the NF-κB family genes.
Allele frequencies for each SNP were determined by gene counting. Genotype distribution of each SNP was tested for deviation from Hardy-Weinberg equilibrium (HWE) using chi-square (χ 2 ) analyses. The extent of pair-wise linkage disequilibrium (r 2 -value) between polymorphisms was determined by the Haploview version 4.2. The association between polymorphisms and MOD scores was performed using analysis of covariance testing with age, sex ratio, and injury severity to adjust for possible confounding effects. Three genetic models (allele dose, dominant, and recessive) were used. The association of genotypes with sepsis morbidity rate was determined by χ 2 analysis. Odds ratios (OR) with 95% CI were calculated by multivariate logistic regression models to estimate the relative risk of sepsis. All P-values were two-sided and were adjusted using the Bonferroni correction for multiple testing, with P <0.05 used to determine statistical significance. All statistical analyses were carried out using SPSS statistical software (version18.0; SPSS Inc, Chicago, IL, USA).

Construction of haplotype blocks and selection of SNPs
SNPs with a MAF ≥0.05 were shown within and around the NFKB1, NFKB2, RELA, RELB and REL genes from the HapMap database for the CHB population (Additional file 3 and Figure 1). rs28362491, a common insertion/ deletion polymorphism (−94 insertion/deletion ATTG), though it did not form any block with other SNPs, has been identified located between two putative key promoter regulatory elements in the NFKB1 gene. The presence of ATTG deletion (allele D) resulted in the loss of binding to nuclear protein, leading to reduced promoter activity [22], thus, rs28362491 was selected.
Taken together, eight SNPs (rs28362491, rs3774932, rs4648068, rs7119750, rs4803789, rs12609547, rs1560725 and rs842647) were selected from this study ( Table 1). The canonical NF-κB pathway, which mainly involves RelA/p50 and c-Rel, plays a critical role in innate immune responses [7]. Thus four SNPs (rs28362491, rs4648068, rs7119750 and rs842647) of NFKB1, RELA and REL genes were selected for genotyping. We did not select tSNP from NFKB1 block1, due to it including only two SNPS (rs3774932 and 3774933) in intron regions and little was known about the role of these two polymorphisms and immune response.

Allele frequencies and genotype distribution of the four selected SNPs from among trauma patients
The genotyping success rates of the four selected SNPs by pyrosequencing ranged from 98.3% to 100% in our study cohort. The MAF among the 753 trauma patients was 41.8% (rs28362491), 44.7% (rs4648067), 37.6% (rs7119750) and 13.4% (rs842647), respectively, which was quite similar to those observed in the 137 unrelated individuals in the CHB cohort in the HapMap database. The genotype distribution of all four selected SNPs was in agreement with Hardy-Weinberg equilibrium (P >0.05) ( Table 2), indicating that both allele and genotype frequencies of these selected SNPs in the population remain constant. That is to say, they are in equilibrium from generation to generation.

Clinical association of the four selected SNPs with development of sepsis and MODS in trauma patients
The patient cohort comprised a total of 753 consecutive Han Chinese patients, 603 male and 150 female, with age 41.2 ± 13.3 years (mean ± SD) and ISS of 22.2 ± 9.5. The demographic and baseline characteristics and the clinical data of the patients are summarized in Table 3. All patients survived ≥48 hours after admission: 311 patients (41.3%) developed sepsis. Pathogens were identified as causative microorganisms in the blood cultures of 133 septic patients (42.8%). The common pathogens identified in this study cohort were Staphylococcus aureus, coagulase-negative staphylococci, Klebsiella pneumoniae, Acinetobacter baumanii, Pseudomonas aeruginosa, Escherichia coli, Enterococcussp., and Enterobacter cloacae.   Table 4, there were no significant differences in age, sex ratio or ISS among patients stratified according to the different genotypes of each selected SNP. Among the four genetic variants selected in this study, although it was found in only 15 patients with variant homozygotes in this study cohort, only the rs842647 was shown to be significantly associated with the risk for development of sepsis and MODS in patients with major trauma. The patients carrying the variant A allele revealed a significantly lower sepsis morbidity rate and MOD scores, when compared with those carrying the G allele (P = 0.024 for sepsis morbidity rate and P = 0.013 for MOD scores in the case of a recessive effect). Data from multiple logistic regression analyses further indicated that the patients with the rs842647 polymorphism had a lower risk for developing sepsis (OR = 0.673, 95% CI = 0.532 to 0.873; P = 0.012) after adjusting for possible confounders, including age, sex ratio and ISS. There were no significant associations with sepsis morbidity rate and MOD scores for the other three selected SNPs (rs28362491, rs4648068 and rs7119750).

Effect of rs842647 on LPS-induced TNF-α production
NF-κB is a key molecule for the maturation and secretion of TNF-α [23,24]. Therefore, we hypothesized that the rs842647 polymorphism might be associated with TNF-α production. As shown in Figure 2, the rs842647 polymorphism was well associated with the LPS responsiveness  of peripheral blood leukocytes. LPS-induced TNF-a production was significantly lower in patients with the variant A allele than that in those with the wild G allele (P = 0.027 for recessive effect) (Figure 2).

Discussion
In the present study, to the best of our knowledge, we are the first investigators to identify the potential clinical relevance of the rs28362491, rs4648068, rs7119750 and rs842647 polymorphisms, which are identified within the NFKB1, RELA, and REL genes in the Han Chinese population. Tag SNPs are a subset of all variants within a chromosome region within a disease study. Use of tag SNPs is a powerful approach to investigate genetic association studies, for the genetic effect of SNPs that are not genotyped in the study can be detected through LD with a tag SNP [16]. In this case, the four SNPs selected in the current study theoretically reflect the biological significance of most genetic variations across the NFKB1, RELA, and REL genes because of their strong LD with other non-assayed variants. The NF-κB, a collection of dimeric transcription factors, was first identified as a B cell factor that binds to a site in the enhancer region on the gene encoding the immunoglobulin κ light chain in 1986 [25]. The NF-κB subunits form homo-and heterodimers that control a broad spectrum of biological processes, including development, apoptosis, and the immune response [8,26]. Mice deficient in NF-κB family members exhibit defects in macrophage activation and a compromised immune response to pathogens [27]. NF-κB1 signaling is known to amplify and perpetuate inflammatory and coagulatory mechanisms prevailing in sepsis [28]. Also, NF-κB1 has also been shown to influence tissue factor in the bloodstream and, thus, activates coagulation [28][29][30]. In addition, the major importance of the p50 and p65 subunits in the pathophysiology of sepsis has been highlighted in both animal models and human pathology [31]. NF-κB2 is a critical mediator in the development and function of a variety of organs and cell lineages, as shown by studies of genetically modified mice in which NF-κB2 function was ablated or modified [32]. RelB is responsible for LPSmediated attachment [33]. Moreover, c-Rel is reported to be a key NF-κB member required for host antimicrobial defenses and a regulatory transcription subunit that ISS, injury severity score; MOD, multiple organ dysfunction; a, recessive effect (variant homozygotes versus heterozygotes + wild-type homozygotes) as analyzed by one-way analysis of variance. a1, P = 0.024; a2, P = 0.013. Figure 2 Effect of the rs842647 polymorphism on lipolysaccharide (LPS)-induced TNF-α production. The whole-blood samples collected from trauma patients (n = 30 for the GG and GA genotype, n = 15 for the AA genotype) immediately after admission were mixed at a 1:1 ratio (vol/vol) with RPMI 1640 culture medium and incubated with 100 ng/mL bacterial LPS at 37°C for 4 hours. TNF-α production was determined by a sandwich ELISA. The data are presented as mean ± SD. One-way analysis of variance was used to assess statistical significance. *P = 0.027 for recessive association (AA versus GG + GA).
controls the inflammatory and immune responses in severe infection [34]. In this study, we hypothesized that common genetic variants within the canonical pathway could contribute to the development of sepsis and MODS in patients with major trauma. Four SNPs (rs28362491, rs4648068, rs7119750 and rs842647) within the canonical pathway were selected for genotyping, and they capture most of the genetic variation of NFKB1, RELA and REL genes and might represent the possible biological significance of their genetic variation.
Our results indicate that among the four SNPs selected in our study cohort, only the rs842647 polymorphism reveals a strong clinical relevance, with a lower rate of sepsis morbidity and MOD scores in the patients with the variant A allele. Of the other three SNPs (rs28362491, rs4648068 and rs7119750) evaluated, we did not observe a significant association. Although it has been reported to have a regulatory influence on NFKB1 gene expression and is known to be associated with cancer [35][36][37][38][39], acute respiratory distress syndrome [40], periodontitis [41], autoimmune and inflammatory diseases [42], and inflammatory bowel diseases [43], the functional polymorphism rs28362491 (also known as the −94 insertion/deletion ATTG polymorphism in the promoter of NFKB1) was not shown to be associated with risk of sepsis and MODS in our study. Different ethnic populations and diseases, interactions or the effects of combination with non-genetic risk factors might account for these contrasting results.
Given the clinical relevance of the rs842647 polymorphism, we further hypothesized that this SNP might be associated with TNF-α production in patients with major trauma. The peripheral blood was taken immediately after admission in an attempt to avoid the potential effects of infection and fluid resuscitation. Our results reveal that the rs842647 polymorphism could affect the capacity of peripheral leukocytes to produce TNF-α, showing much lower levels of TNF-α among patients who carry the variant A allele. This is in accordance with its clinical relevance. Recently, the same variant was found to be associated to other immune-related diseases. Trynka et al. found that rs842647 was associated with risk of coeliac disease [44]; moreover, Amundsen et al. observed that coeliac disease risk polymorphisms, containing rs842647, could affect gene regulation in the thymus [45].
How might the rs842647 variant affect susceptibility to sepsis and MODS? This variant is located in the intron of the REL gene, which is implicated in T cell differentiation [46] and it appears to play a critical role in promoting immune and inflammatory responses [34]. A potential mechanism of the rs842647 effect is alteration of gene regulation. Online splicing prediction by SplicePort [47,48] (http://spliceport.cbcb.umd.edu/) showed that this mutation might not change gene splicing, thus, might not lead to the loss of splicing of the intron. It is possible that this variant simply amplifies the initial innate immune inflammatory response to bacterial products beyond a threshold that is tolerated by the host organ systems. Alternatively, other SNPs in high LD with this SNP may have a biological function. In the Han Chinese population the at-risk allele G is the major allele of rs842647 G/A, potentially marking a haplotype that includes many rare functional SNPs that increase risk, or the rarer A allele may actually be protective. Another, more nuanced possibility involves the modulation of regulatory T (Treg) cell function. Levels and function of Treg cells have been associated with mortality in human sepsis and have been shown to play an important role in regulating the inflammatory and antimicrobial response in animal models of sepsis [49]. Research has shown that at the Foxp3 locus encode information defining the size, the pioneer element conserved non-coding DNA sequence 3, which acts to potently increase the frequency of Treg cells generated in the thymus and the periphery, binds c-Rel in in vitro assays [50]. We therefore speculate that excessive REL-mediated stimulation of Treg cells conferred by this mutation could lead to dysregulation of Treg-cell function in patients with sepsis and lead to organ dysfunction.
Limitations of this investigation should be mentioned. First, although the sample comprised 753 trauma patients in total, the group of patients homozygous with the variant allele of the rs842647 polymorphism was relatively small (15 patients). Second, difficulties in obtaining enough blood samples from patients did not allow us to investigate the kinetics of NF-κB mRNAs levels in patients with major trauma. Therefore, the association of the four selected SNPs with changes of NF-κB mRNA expression could not be confirmed in this study. Furthermore, the patients recruited into this study cohort were limited to the Chongqing district. The generalizability of the association to populations in other regions in China is therefore not ensured. Future studies in a larger patient population might be needed to further validate their clinical relevance.

Conclusion
In the current study, we investigated the clinical relevance of the genetic variants within the entire NFKB1, RELA, and REL genes by means of tag SNPs. We have demonstrated that the rs842647 polymorphism affects TNF-α production and might be used to estimate risk for sepsis and MODS in trauma patients. Further studies, both clinical and experimental, are therefore needed to confirm the significance of these findings and to investigate their synergistic effect with other genetic polymorphisms in relation to the development of sepsis in and the outcomes of trauma patients.