Hypoxia inducible factor-1 alpha and prolinhydroxlase 2 polymorphisms in patients with severe sepsis: a prospective observational trial

Background Hypoxia-inducible-factor-1α (HIF-1α) and HIF-1 degrading prolyl-hydroxylases (PHD) are key regulators of the hypoxic-inflammatory response. Functionally active genetic variants in the HIF-1α (C/T; Single Nucleotide Polymorphism (SNP) rs11549465) and the PHD2 gene (EGLN1; C/T; SNP rs516651 and T/C; SNP rs480902) are associated with altered HIF-1α mRNA nuclear translocation and an altered adaptation to hypoxia. Furthermore, the HIF system is important in surviving inflammatory disorders and sepsis. Thus, we tested the hypotheses, that SNPs in the HIF-1α or PHD2 genes are (1) common in Caucasians, with 2) the HIF-1α genetic variant being associated with an altered HIF-1α mRNA expression; and 3) independent risk factors for 30-day mortality in severe sepsis. Methods After ethics approval, 128 septic patients (Caucasian descent) were included prospectively within 24 h after first diagnosing sepsis. Patients characteristics and severity of illness (simplified acute physiology score II), genotypes (Taqman assay), and their influence on leukocyte HIF-1α-mRNA-expression (Real-Time PCR) and 30-day mortality were determined. Results Frequencies were 0.8 % for homozygous HIF-1α TT-carriers (CT 17.6 %; CC 81.6 %), 2.5 % for homozygous PHD2 SNP rs516651 TT-allele carriers (CT 17.5 % and CC 80 %), and 9.4 % for homozygous PHD2 SNP rs480902 TT-allele carriers (CT 34.4 % and CC 56.3 %). While HIF-1α T-allele carriers had a borderline decrease in HIF-1α-mRNA-expression (p = 0.06) neither HIF-1α nor PHD2 SNPs were (independent) risk factors for 30-day mortality. Conclusions Genetic variants in HIF-1α and PHD2 genes exist in Caucasians but do not appear to alter 30-day mortality in sepsis.


Background
Hypoxia inducible factor-1α (HIF-1) and HIF-1 degrading prolyl-hydroxylases (PHD) are key regulators of the hypoxic response but are also involved in many cellular actions [1]. Furthermore, we could recently show that leukocyte HIF-1α-mRNA-expression is decreased in severe sepsis and inversely correlated with disease severity [2]. Thus, genetic variants in HIF-1 or HIF-degrading PHD2 could impact on HIF-1α mRNA expression and outcome in septic patients.
Recently, functionally active genetic variants in the HIF-1α (C/T; Single Nucleotide Polymorphism (SNP) rs11549465) and the PHD2 gene (EGLN1; C/T; SNP rs516651 and T/C; SNP rs480902) were identified [3][4][5] and were associated with increased HIF-1α nuclear translocation and a worse outcome compared to the TTgenotype in several tumors [6]. However, the HIF-1pathway plays an important role in adaptation to inflammation, too, as both HIF-1α mRNA and intracellular HIF-1α protein are highly increased following inflammatory stimuli [7,8]. With persistent inflammation, HIF-1α is suppressed in vitro and even in patients with sepsis, possibly to counter regulating an overwhelming inflammatory response [2,7,9]. Furthermore, the HIF-1α Tallele is associated with an adverse outcome in acute injury of the kidney, i.e. an organ with a HIF-mediated oxygen sensor, important in mediating erythropoiesis [1,10]. In high altitude populations, the PHD2 variant rs480902 CC-genotype is associated with an increased heart rate and higher oxygen saturation under hypoxia, and PHD2 variant rs516651 T-allele carriers are associated with a decreased incidence of acute mountain sickness [5]. Furthermore, PHD deficiency is associated with an altered inflammatory response, i.e., suppressed cytokine expression in mice [11].
Taken together, HIF-1α and PHD2 genetic variations are functionally active but their impact on critical illness has not yet been investigated. Accordingly, we examined potential effects of genetic variants of HIF-1 and HIF degrading PHD2, in severe sepsis. Specifically, we tested the hypotheses, that SNPs in the HIF-1α or PHD2 genes are 1) common in Caucasians with 2) the HIF-1α genetic variant being associated with an altered HIF-1α mRNA expression, and 3) independent risk factors for 30-day mortality in severe sepsis.

Methods
The study was reviewed and approved by the Medical Faculty's ethic committee (no. 06-3078, University of Duisburg-Essen, Essen, Germany). Adult patients with sepsis admitted to an intensive care unit of the University Hospital Essen, Germany, between 2010 and 2014 were evaluated for inclusion. Patients were considered eligible when they fulfilled the sepsis criteria as defined by Bone et al. [12]. Patients who refused or withdrew study participation, underage persons, and all individuals with non-Caucasian ethnicity were excluded.
Arterial blood samples were taken for blood tests, microbiology cultures, genotyping, and mRNA extraction within the first 24 h after diagnosing sepsis. Furthermore, SAPS II (Simplified Acute Physiology Score II), length of hospitalization, and 30-day mortality were recorded [2,13].

RNA preparation, qPCR
Whole blood HIF-1α mRNA-expression was analyzed by qPCR following RNA-extraction from the last 62 consecutive patients with severe sepsis included into the study [2,6]. RNA-extraction was performed on patients' blood samples using the RNeasy Midi Kit (Qiagen, Hilden, Germany) following manufacturers protocol. One microgram of total RNA was reverse-transcribed to cDNA (complementary deoxyribonucleic acid). As described previously, primers for qPCR were designed using NCBI primer blast (Rockville Pike, MD) and obtained from Invitrogen (Invitrogen AG; Karlsruhe, Germany) [2].
First, qualitative PCR was performed for GAPDH (glyceraldehyde-3-phosphate-dehydrogenase) as a housekeeping gene to test the quality of cDNA synthesis [2,14]. Second, quantitative PCR was performed to estimate the amount of specific cDNA for HIF-1α [2]. Resulting cDNA bands were visualized using ethidium bromide-stained 1.5 % (w/v) agarose gels [7]. HIF-1α mRNA expression was quantified by qPCR using SYBR® green as fluorescent dye (Eurogentec, Verviers, Belgium) on a StepOnePlus PCR system (Applied Biosystems, Carlsbad, USA). The denaturation steps were performed at 95°C for 10 min, followed by 40 cycles at 95°C for 15 s, and at 60 or 61°C for 1 min, as described previously [2]. Amounts of specific cDNA were finally normalized to GAPDH using the ΔΔct method, and results are depicted as 2 ΔΔct values, as described [2].

Statistical analyses
Data are presented as median with quartiles unless indicated otherwise, and in case of categorical variables as numbers and percentages. Data are presented as means ± standard deviation (SD) unless indicated otherwise. The two tailed t-test for independent samples or, in case of violations of the normality assumption (as tested by Kolmogorov-Smirnov and Shapiro-Wilk tests), the Wilcoxon signed rank test was used. Potential deviations from the Hardy-Weinberg equilibrium were ruled out using Court Lab calculator (Court MH; Court-laboratory Hardy-Weinberg calculator; Tufts University; www.tufts.com) for MS Excel 2011 software (Microsoft Deutschland, Unterschleißheim, Germany).
Furthermore, we investigated associations between the patients' clinicopathologic characteristics and of the HIF-1α and PHD2 genotypes with 30-day survival, defined as the interval from time of diagnosis of severe sepsis until day 30 or death, whichever occurred first. Patients alive after the 30-day follow-up were regarded as censored. Kaplan-Meier plots were used to display the overall 30-day survival data in the respective subcohorts followed by log-rank tests for comparison of the different genotypes. Thereafter, we computed hazard ratios and 95 %-CI using multivariate Cox regression models to assess the impact of the respective genotypes (HIF-1α and PHD2 rs516651/rs480902), sex, age, Simplified Acute Physiology Score II, and requirement for dialysis as predictors for the clinical outcome (30-day survival). The proportional hazards assumption was checked for all variables included in the models and no deviation was found (data not shown). An a priori alpha error p of less than 0.05 was considered statistically significant. In case of more than two groups one-way ANOVA with post hoc Student's t-test and correction for multiple testing (p = 0.05/n) was used. Interdependence between HIF-1α-mRNA-expression and age was analyzed using Pearson correlation and linear regression analysis.
Statistical analyses were performed using SPSS 21.0 (SPSS Inc., Chicago, IL) and Graph Pad Prism 5 (Graph Pad Software, La Jolla, CA) software. Cox regression models were analysed with SAS Version 9.4 PHREG procedure (SAS Institute, Cary, NC, USA).  . 1) and multivariate Cox-regression-analyses (Table 4) showed no influence of this genetic variation on 30-day mortality from septic shock. Of note, only a single patient (survivor) carried the homozygous TT-genotype, as shown in Table 1 and Fig. 1. However, this single individual was not included into the statistical analyses. We then analyzed in 62 samples from consecutive septic patients whether this genetic variant influences HIF-1α-mRNA-expression. Of these patients 17.7 % (n = 11) were T-allele carriers of the HIF-1α SNP rs11549465, and 82.3 % were homozygous CC-genotypes (51 patients). HIF-1α-mRNA-expression was decreased by 62 % in Tallele carriers compared to CC-genotypes (Δct-values: T-  Fig. 2) but this was of borderline statistical significance p = 0.06), with CT-allele carriers being older (68 years; Q3; 53-72) than individuals with a CC (54 years; Q1-Q3; 48-66) genotype (p = 0.04). Of note, HIF-1α-mRNAexpression did not correlate with age (Pearson correlation: r = −0.1, p = 0.42), and HIF1-α-mRNA expression did not differ between genotypes even after adjusting for age (data not shown). Further clinicopathogenic characteristics are displayed in Table 1 and did not differ between groups.

Discussion
In this prospective study we show, that allele frequencies in Caucasian for the functionally active SNPs of the hypoxic-inflammatory response, namely HIF-1α SNP rs11549465, PHD2 rs516651 and rs480902 highly differ from Asian ethnicities. However these SNPs did not   have a major impact on 30-day mortality in patients with severe sepsis in our sample. Of note, HIF-1α-mRNA-expression showed a trend towards lower expression in HIF-1α T-allele carriers. HIF-1α SNP frequencies in our Caucasian cohort with sepsis revealed only a single homozygous TT-genotype carrier (0.8 %), whereas 17.6 and 81.6 % carried the heterozygous CT-and homozygous CC-genotypes, respectively. Notably, in people from India the TT-genotype is common with a frequency of 25 %, and 50 % CTgenotypes. T-allele carriers show an association with worse outcomes in metastatic prostate cancer or acute kidney injury [3,15]. In this prospective study we could not confirm an influence of this SNP on 30-day mortality from severe sepsis. While this might relate to the very low frequency of homozygous TT-genotype carrier in our cohort, 30-day survival also did not differ between HIF-1α CT-and CC-genotypes.
However, HIF-1α mRNA expression showed a trend towards lower expression in T-allele carriers and on average was decreased by 62 %. As we reported previously, decreased HIF-1α mRNA expression inversely correlates with sepsis severity, i.e., with greater HIF-1α mRNA expression and intracellular HIF-1α protein suppression in patients with higher Simplified Acute Physiology Score II (SAPS II) [2]. As a high SAPS II is associated with increased mortality, one might expect increased mortality in HIF-1α T-allele carriers as well, however one might emphasize that sepsis itself contributes stronger to alterations in HIF-1α mRNA expression    than this genetic variant does [13]. However, sepsis is a heterogeneous disease with multiple factors contributing to the patients' mortality. Even, in our previous study HIF-1α mRNA-expression correlated with disease severity but not with outcome [2]. In a next step we analyzed genetic variants in the HIF-1 degrading PHD2 gene, as these SNPs are associated with an increased oxygen saturation and heart rate in response to hypoxia and thus improved adaptation [5]. Furthermore, a PHD2 knockout in mice inhibits TNFα and ICAM-1 expression with reduced cell apoptosis and macrophage infiltration [11]. Furthermore, inhibition of PHD2 attenuates cardiac dysfunction in mice undergoing a high fat diet by a mechanism involving suppression of the MYD88/nuclear factor Kappa B inflammatory pathway [11]. Thus, functionally active genetic variants in the PHD2 gene could also impact on patients outcome in severe sepsis, e.g., by altering the inflammatory response.
Therefore, we determined the genotype frequencies of PHD2 SNP rs516651 carriers in our Caucasian cohort, with homozygous TT-genotypes found in 2.5 % of septic patients and heterozygous CT-genotypes in 17.5 %. Thus, we show that this SNP is present in Caucasians whereas in Tibetan high altitude residents, T-allele carriers do not exist [5]. Given this distribution in Caucasians, the PHD2 SNP rs516651 turned out not to be an independent predictor for 30-day mortality in a multivariate Cox-regression analysis including common known risk factors for sepsis mortality in our rather small patients cohort.
The PHD2 SNP rs480902, was common in Caucasians as well, however frequencies for homozygous TTgenotypes (9.4 %) and CT-genotypes (34.4 %), were lower when compared to Tibetan high altitude residents with frequencies of T-allele carriers of up to 75 % [5].
However, the PHD2 SNP rs480902 genotype did not impact on 30-day mortality. PHD2 is a key-regulator of HIF-1α in erythropoietin producing cells and, therefore, crucial for regulation of erythropoietin secretion [16]. In mice, PHD2 deficiency leads to pronounced erythrocytosis due to a suppressed HIF-1α degradation and hence activation of the erythropoietin pathway, [16] as well as to increased angiogenesis [17]. Whereas regulation of erythrocytosis usually represents monocausal mid to long-term changes, severe sepsis is an acute lifethreatening disorder depending on many issues. This might explain why PHD2 genetic variants did not impact on survival in our cohort.
Our study has limitations: First, given the low genotype frequencies for the homozygous HIF-1α and PHD2 rs516651 TT-genotype, 128 septic patients still need to be considered a small sample, reducing the explanatory power of our analysis. Thus, reanalyzing the influence of HIF-1α and PHD2 SNPs in a larger cohort of septic patients is warranted. Furthermore, in this study we did not analyze the impact of either genetic variant on the patients phenotype i.e. regarding immune cell function or blood coagulation. This is of particular interest as other genetic variants, i.e. the NFKB1 promoter polymorphism -94ins/delATTG, turned out to be an independent predictor for 30-day mortality from septic shock, in a similar patients cohort of 143 patients [18]. In detail, the Deletion-allele was associated with altered blood coagulation and hyperinflammation [18]. Second, due to ethics issues and restrictions we only could analyze HIF-1α mRNA expression from the last 62 consecutive patients only at one time point. Analyzing HIF-1α mRNA expression in the course of sepsis depending on genetic variants might help to understand the regulation of the hypoxic-inflammatory pathway in the critically ill. More important, leukocyte HIF-1α mRNA expression was suppressed in T-allele carriers when compared to the CC-genotype, but this failed to reach statistical significance (p = 0.06). However, such variations in HIF-1α mRNA expression in T-allele carriers, are in accordance with the large variation of leukocyte HIF-1α expression in septic patients, [2] as well as a worse outcome of T-allele carriers with metastatic prostate cancer [6]. Third, septic patients are a very heterogeneous cohort due to various entities of sepsis or comorbidities. Thus, it might be favorable to analyze these polymorphisms in a cohort with a more homogenous sepsis entity, e.g., in patients with acute respiratory distress syndrome.

Conclusions
In conclusion, genetic variants in HIF-1α and PHD2 genes exist in Caucasians. However, homozygous TTgenotypes are infrequent and range from 0.8 % for HIF-1α SNP rs516651 to 9.4 % for PHD2 SNP rs480902. Furthermore, even in our rather small sample, T-allele carriers of the HIF-1α SNP show a trend towards further suppressed HIF-1α mRNA expression compared to CC genotypes. However, neither of the analyzed genetic variants of HIF-1α and PHD2 genes turned out to be independent risk factors for 30-day mortality of severe sepsis. Reanalyzing the influence of these genetic variants in a larger and more homogenous patients cohort i.e. patients with Acute respiratory distress syndrome (ARDS) is warranted to confirm the findings of our study.