Abstract
Background
Candidemia, one of the most common causes of fungal bloodstream infection, leads to mortality rates up to 40% in affected patients. Understanding genetic mechanisms for differential susceptibility to candidemia may aid in designing host-directed therapies.
Methods
We performed the first genome-wide association study on candidemia, and we integrated these data with variants that affect cytokines in different cellular systems stimulated with Candida albicans.
Results
We observed strong association between candidemia and a variant, rs8028958, that significantly affects the expression levels of PLA2G4B in blood. We found that up to 35% of the susceptibility loci affect in vitro cytokine production in response to Candida. Furthermore, potential causal genes located within these loci are enriched for lipid and arachidonic acid metabolism. Using an independent cohort, we also showed that the numbers of risk alleles at these loci are negatively correlated with reactive oxygen species and interleukin-6 levels in response to Candida. Finally, there was a significant correlation between susceptibility and allelic scores based on 16 independent candidemia-associated single-nucleotide polymorphisms that affect monocyte-derived cytokines, but not with T cell-derived cytokines.
Conclusions
Our results prioritize the disturbed lipid homeostasis and oxidative stress as potential mechanisms that affect monocyte-derived cytokines to influence susceptibility to candidemia.
Keywords: arachidonic acid metabolism, C albicans, cytokine-QTLs, genetic variants, PLA2G4B
Understanding genetic mechanisms for candidemia susceptibility may help designing host-directed therapies. By performing the first candidemia GWAS and in vitro stimulation experiments using population-based cohorts, we prioritized lipid homeostasis and oxidative stress as potential susceptibility mechanisms that affect Candida-response cytokines.
Invasive fungal infections are an increasingly significant cause of sepsis in critically ill and immunocompromised patients, and they are associated with high morbidity and mortality. Candida albicans is the most commonly encountered Candida species associated with nosocomial invasive candidiasis globally and is mainly part of the normal gut flora [1, 2]. Candida albicans causes both invasive and mucosal infections. It is a dimorphic fungus because it grows both as yeast and filamentous cells, and it causes approximately 250 000 new systemic infections on a yearly basis, leading to more than 50 000 deaths [3, 4].
Both innate and adaptive immune mechanisms are crucial for an effective host defense against fungal pathogens. One of the defense mechanisms used by phagocytes is the use of reactive oxygen species (ROS) against various pathogens, including fungi [5]. Cytokines play a critical role during anti-Candida host immune defense, and recombinant cytokines have been proposed as adjunctive therapy in systemic Candida infection [6–9]. Therefore, identifying genetic loci that influence susceptibility to systemic Candida infections by modulating ROS production as well as cytokine levels will help us understand what drives susceptibility, a crucial step for the design of novel preventive and therapeutic strategies.
In the present study, we performed the first genome-wide association study (GWAS) of candidemia on the largest patient cohort published to date consisting of 178 candidemia patients and 175 case-matched controls of European ancestry. By making use of cytokine-quantitative trait loci (cQTL) datasets generated from our previous studies from the Human Functional Genomics Project (HFGP) consortium [10, 11], we assessed the impact of genetic polymorphisms on the cytokine production capacity of peripheral blood mononuclear cells (PBMCs), whole blood, and monocyte-derived macrophages in response to C albicans yeast or hyphae, in 500 healthy volunteers [10]. By integrating these cQTLs with the candidemia susceptibility data (Figure 1), we show that 35% of the independent susceptibility loci influence cytokine levels in response to C albicans, with several genes from these loci being known to be involved in lipid and arachidonic acid (AA) metabolic pathways. Furthermore, given that AA metabolism plays a role in the generation of ROS in various cell types [12], we show that the numbers of risk alleles at these loci are negatively correlated with ROS production in response to Candida, which is positively correlated with interleukin (IL)-6 levels, using an independent cohort of healthy volunteers. Taken together, our results suggest that increased risk to candidemia can be partly explained by a disturbed lipid homeostasis, which results in modulation of ROS and proinflammatory cytokines, which would ultimately lead to increased susceptibility to candidemia.
METHODS
Study Populations
Population-Based Cohorts
The study was performed using 2 independent population-based cohorts, 500FG and 200FG, composed of healthy individuals of Western European ancestry from the HFGP (www.humanfunctionalgenomics.org). For further description of these cohorts and ethics approval statements, see Supplementary Methods.
Candidemia Cohort
To identify single-nucleotide polymorphisms (SNPs) associated with candidemia susceptibility, we performed the first GWAS analysis of a well described candidemia cohort with corresponding case-matched controls [13]. In total, 178 candidemia cases and 175 case-matched controls of European ancestry were tested for disease association. The demographic and clinical characteristics of the candidemia cohort have been described previously [13].
Genotyping of the 500FG Cohort
Deoxyribonucleic acid (DNA) obtained from the 500FG cohort was genotyped using the commercially available Illumina HumanOmniExpressExome-8 v1.0 SNP chip. Genotype calling, quality control, and imputation were previously described elsewhere and in the Supplementary Methods [10].
Genotyping, Quality Control, and Imputation of the Candidemia Cohort
Isolated DNA obtained from case and control samples was genotyped using the commercially available SNP chips, HumanCoreExome-12 v1.0, and HumanCoreExome-24 v1.0 BeadChip from Illumina (https://www.illumina.com). Genotype calling, quality control, and imputation are described in Supplementary Methods.
Genome-Wide Case-Control Association Analysis
The associations between the genome-wide variants and candidemia susceptibility were tested by Fisher’s exact test using PLINK v1.9 (www.cog-genomics.org/plink/1.9/) [14]. A P value significance threshold of <5.0 × 10–8 was set to call genome-wide significant associations. We considered SNPs with a P value between 9.99 × 10–5 and 5 × 10–8 as variants showing a suggestive association with candidemia susceptibility.
Cell Collection and Candida Stimulation Experiments
For the cytokine profiling, PBMCs, monocyte-derived macrophages and whole blood were collected and stimulated with heat-killed C albicans yeast (strain ATCC MYA-3573, UC 820) or hyphae in a concentration of 106 colony-forming units (CFU)/mL as previously described [10]. To identify the transcriptome upon Candida stimulation, 5 × 105 PBMCs isolated from 8 healthy volunteers were incubated with 1 × 106/mL heat-killed C albicans (UC 820) and Roswell Park Memorial Institute (RPMI) culture medium as a control for 4 and 24 hours, as previously described [15]. For further details on the experiments, see Supplementary Methods.
Analysis of Ribonucleic Acid Sequencing Reads
Ribonucleic acid (RNA) sequencing analysis of the data generated from Candida-stimulated PBMCs for 4 and 24 hours was previously described (see Supplementary Methods) [15].
Cytokine-Quantitative Trait Loci Mapping
We made use of cQTL datasets generated from our previous studies from the HFGP consortium, which are publicly available at www.humanfunctionalgenomics.org [10]. To correct for multiple testing, we used Bonferroni and Benjamini-Hochberg approaches for the 26 susceptibility variants located in 24 independent loci. We considered P < 5 × 10–8 to be the threshold for significant cQTLs. Candida albicans response QTLs (P < .05) were intersected with SNPs associated with susceptibility to candidemia showing a P value between 9.99 × 10–5 and 5 × 10–8 to identify cQTLs associated with susceptibility.
Peripheral Blood Mononuclear Cell Fungal Killing Assays
Human PBMCs isolated from healthy volunteers were plated in 96 round-bottom plates (5 × 105 cells per well). After 24 hours of incubation, 25 nM of either nontargeting small interfering RNA (siRNA) control pool (D-001810-10-05) or PLA2G4B (L-187552-01-0005) targeting siRNA (SMARTpool, Thermo Fischer Scientific) were added for 24 hours. After the transfection period, 1 × 106/mL live Candida was added for 24 hours. After incubation, the well contents were serially diluted in sterile water and plated on Sabouraud agar for counting of CFUs after 24 hours of incubation. Experiments were performed 3 times with 2 donors (N = 6) in each individual experiment. The serial dilutions were plated in duplicate, and the mean value of both measurements was used for the analysis: P <0.05 = *, P < .01 = ** and P < .001 = ***.
Reactive Oxygen Species Production
Reactive oxygen species were measured in PBMCs of healthy volunteers from the 200FG cohort. Induction of ROS was measured by oxidation of luminol (5-amino-2,3-dihydro-1,4-phtalazinedione). The PBMCs (5 × 105) of healthy volunteers were resuspended in Hanks’Balanced Salt Solution and put in dark 96-well plates. Cells were exposed to 1 × 106/mL C albicans yeast (UC820, heat killed) together with 20 µL of 1 mM luminol (final concentration 50 μM). Chemiluminescence was measured in BioTek Synergy HTreader at 37°C for 1 hour with intervals of 2.23 minutes.
Quantitative Trait Loci Mapping of Reactive Oxygen Species Production
We selected 137 healthy individuals from 200FG for whom both genotype and measurements of ROS production were available. We applied a log2 transformation on raw measurements of ROS production. The correlation between the transformed phenotype with genotypes was tested by a linear regression model using age and gender as covariables, using the R-package Matrix-eQTL. We considered P < .05 to be the threshold for suggestive QTLs.
Constructing Allelic Scores
We constructed allelic risk scores for candidemia susceptibility using 122 individuals from the 200FG cohort for whom we have genotypes and ROS measurements available. We calculated an allelic score per individual as the sum of the number of minor alleles across all susceptibility variants multiplied by the log2 of the odds ratio (OR) of that particular SNP for candidemia susceptibility. Allelic scores were calculated using PLINK v1.9 (www.cog-genomics.org/plink/1.9/) [14].
Correlation Analyses
All correlations were tested by non-parametric Spearman’s rank correlation test using R version 3.3.2 (available at http://www.R-project.org). Both ROS and IL-6 measurements were log2 transformed. Measurements of IL-6 levels from Candida-stimulated PBMCs were available for 61 individuals from the 200FG cohort measured with enzyme-linked immunosorbent assay in 2017. P < .05 was considered as the threshold for significance.
Pathway Enrichment Analysis
We performed pathway enrichment analysis using a novel tool called Pascal (Pathway scoring algorithm). We applied Pascal to define gene-level P values for all genetic loci that showed association to candidemia susceptibility with a P value of <9.99 × 10–4 using default parameters. The Pascal tool is described in detail elsewhere [16]. Results reported in this paper are based on the empirical P values; similar results were obtained using the χ2 statistic.
RESULTS
Candidemia Genome-Wide Association Study Identifies a Susceptibility Locus on Chromosome 15
To identify genetic variants associated with candidemia susceptibility, we genotyped 178 well described candidemia patients and 175 case-matched controls of European ancestry [13] on a genome-wide SNP platform and imputed the genotype data using the Human Reference Consortium [17]. After quality control filtering per SNP and sample (Figure S1A and B), we obtained 161 patients and 152 case-matched controls for which we had genotype information on 5 326 313 SNPs, which were tested for disease association. This GWAS identified 235 SNPs in 69 independent loci that show suggestive disease associations (P < 9.99 × 10–5). Among these loci, an SNP on chromosome 15, rs8028958, showed the strongest association with susceptibility to candidemia (P = 7.97 × 10–8, OR = 0.4002, imputation quality score R2 = 0.97) (Figure 2A) and is located within the intronic region of the SPTBN5 gene (Figure 2B).
Prioritization of PLA2G4B as Plausible Causal Gene
To identify potential causal genes affected by rs8028958, we made use of expression-QTL (eQTL) datasets derived from whole blood of healthy donors. In addition, we tested the expression levels of all genes located within a 500-kilobase cis-window of all 69 candidemia variants with suggestive association (P < 9.99 × 10–5) in PBMCs stimulated with C albicans yeast for 4 and 24 hours. For eQTL analysis, we made use of the largest cis- and trans-expression QTL study in blood from 31 684 individuals through the eQTLGen Consortium [18]. The results from this eQTL mapping at rs8028958 identified a significant effect on expression levels of PLA2G4B gene (P = 7.9 × 10–158). In addition, we observed significant downregulation of PLA2G4B expression (P = 1.02 × 10–13) upon C albicans stimulation for 24 hours in PBMCs, which helped to prioritize PLA2G4B as a plausible candidemia susceptibility gene (Figure S2A). Furthermore, publicly available transcriptomic data from PBMCs isolated from patients with chronic mucocutaneous candidiasis (CMC), which were stimulated with RPMI and heat-killed C albicans, showed a trend of decreased expression of PLA2G4B in response to C albicans compared with RPMI control medium (Figure S2B) [19]. The CMC patients have increased susceptibility to mucosal fungal infections and often carry gain-of-function mutations in STAT1 [20, 21].
PLA2G4B (cytosolic phospholipase A2 group IVB; also termed cPLA2-beta) encodes a member of the cytosolic phospholipase A2 protein family [22]. It is interesting to note that the other member of this family, cPLA2-alpha, has also been shown to modulate inflammation and immune responses to Candida, which suggests a critical role of phospholipid metabolism pathway in candidemia [23]. Therefore, we conducted specific PLA2G4B siRNA or control siRNA experiments in human PBMCs to silence the gene before stimulation with C albicans. Although it was not statistically significant (P = .08), we observed that targeted silencing of PLA2G4B led to an increased trend of PBMCs to kill C albicans yeast (Figure S2C), suggesting a functional role for PLA2G4B in host defense against C albicans.
Candidemia Genetic Loci Are Enriched for Phospholipid Metabolism
Given the function of PLA2G4B in phospholipid metabolism, we hypothesized that genetic loci that show suggestive association with candidemia (P < 9.99 × 10–4) are also involved in phospholipid metabolism. To test this, we ran a pathway enrichment analysis using the Pascal tool. Pascal first allows mapping individual SNPs to genes, and their association P values are combined into gene scores (Supplementary Table S1); second, genes are grouped into pathways, and their gene scores are combined into pathway scores (Supplementary Table S2) [16]. We found a significant enrichment of candidemia genes in alpha-Linolenic acid (P = .006), linoleic acid (P = .007), phospholipid (P = .007), and AA metabolism (P = .03) (Figure 3, Supplementary Table S2). Arachidonic acid is a polyunsaturated fatty acid present in the phospholipids of cell membranes, conferring them with fluidity and flexibility, and alpha-Linolenic and linoleic acids are major dietary polyunsaturated fatty acids in the Western diet and are metabolic precursors to AA [24, 25].
Next, we prioritized potential causal genes from these candidemia loci (P < 9.99 × 10–5) by using 3 different approaches (Supplementary Methods). Our 3 approaches prioritized 129 candidate genes (Supplementary Table 3) located in 69 susceptibility loci (P < 9.99 × 10–5). Of note, 19 of these 131 prioritized genes are known to be involved in lipid metabolic processes and AA metabolism. In addition to the PLA2G4B gene, these loci encompassed genes such as SLC16A11, GGT5, GSTT1, AGPAT5, and AGPAT1, which are known to be involved in lipid metabolism. Genes, other than the PLA2 family, such as ALOX12B, ALOXE3, and ALOX15B encode members of the family of enzymes called arachidonate lipoxygenases, which are known to metabolize AA to different metabolites. Other interesting genes relevant to phospholipid synthesis and metabolism are TTC7B and HSD17B4 genes [26, 27]. Finally, genes in the candidemia-associated locus rs4369966 on chromosome 22 encoding Apolipoprotein L (APOL) proteins, which belong to the high-density lipoprotein family, were found to be significantly differentially expressed after 4-hour Candida stimulation (APOL1, APOL6) and 24-hour Candida stimulation (APOL1, APOL2, APOL3, APOL4, APOL6). The APOL proteins are known to play a central role in cholesterol transport. Taken together, our gene prioritization pinpointed several genes that are known to be involved in lipid and AA metabolic processes, suggesting a crucial role for these metabolic processes in regulating inflammation and susceptibility to candidemia.
Candidemia Single-Nucleotide Polymorphisms Affecting Phospholipid Metabolism Regulate Cytokine Levels in Response to Candida albicans
It is known that the phospholipid metabolism influences inflammation in response to infections [28–30]. Therefore, we hypothesized that candidemia-associated SNPs, together with phospholipid metabolism, may also affect cytokine production in response to Candida infection. We therefore tested whether SNPs associated with susceptibility to candidemia (P < 9.99 × 10–5) influence cytokine levels in response to C albicans yeast or hyphae using the results of the previous cQTL mapping [10]. We found 26 SNPs from 24 independent candidemia-susceptibility loci that modulate cytokine levels upon C albicans yeast or hyphae stimulation in any of the cell systems (PBMCs, whole blood, and macrophages) (Table 1, Supplementary Table 4). Of note, the candidemia-associated SNP on chromosome 15, rs12593397 (missense variant), is correlated (D’ = 0.94, r2 = 0.49) with rs8028958 in a European population, which showed the strongest association to candidemia susceptibility (Supplementary Table 5). Genes within 7 independent candidemia-associated loci are known to be involved in lipid and AA metabolism pathways (Table 1). In addition, this analysis also identified important innate immune genes such as NOD2 and IL21R for candidemia susceptibility. These observations led us to the hypothesis that susceptibility to candidemia can be partly explained by defects in phospholipid metabolism and, subsequently, to AA metabolic processes, which could ultimately lead to dysregulated cytokine responses.
Table 1.
P value | ||||||||||
---|---|---|---|---|---|---|---|---|---|---|
Chr | SNP | Susceptibility | cQTL | OR | Cytokine | Stimulant | Cell system | Time | Gene(s) | |
1 | rs11588087 | 4.34 × 10–6 | 0.014 | 3.07 | IL-1β | Yeast | PBMCs | 24h | GLUL(a), RGS16(a) | |
2 | rs34089869 | 4.26 × 10–5 | 0.019 |
0.35 IFNγ |
Yeast | PBMCs | 7days | |||
2 | rs62159917 | 3.74 × 10–5 | 0.049 | 0.36 | TNF-α | Yeast | Macrophages | 24h | ||
3 | rs12493413 | 3.54 × 10–5 | 0.021 | 0.49 | IL-22 | Hyphae | PBMCs | 7days | ARL8B(a), BHLHE40 (a), SUMF1(g) | |
3 | rs9879846 | 4.79 × 10–5 | 0.036 | 0.44 | IFNγ | Yeast | Whole blood | 48h | C3orf14(a,c) | |
5 | rs78969277 | 4.69 × 10–5 | 0.026 | 3.29 | TNF-α | Yeast | Macrophages | 24h | ADAM19(a,b), HAVCR2(a), NIPAL4(a), ITK(c) | |
5 | rs4895365 | 6.94 × 10–5 | 0.019 | 0.46 | IL-1β | Hyphae | PBMCs | 24h | HSD17B4(c,g), TNFAIP8(b,g), DMXL1(c) | |
5 | rs6872016 | 4.76 × 10–5 | 0.015 | 0.39 | IL-1β | Yeast | Whole blood | 48h | ||
6 | rs11155859 | 1.06 × 10–5 | 0.044 | 0.48 | IFNγ | Yeast | PBMCs | 7days | SYNE1(c) | |
6 | rs11155859 | 1.06 × 10-5 | 0.031 | 0.48 | IL-6 | Yeast | Macrophages | 24h | SYNE1(c) | |
6 | rs11155859 | 1.06 × 10-5 | 0.048 | 0.48 | IL-6 | Yeast | Whole blood | 48h | SYNE1(c) | |
7 | rs3807184 | 7.63 × 10–5 | 0.013 | 0.45 | IL-17 | Hyphae | PBMCs | 7days | NME8(a,c,e), EPDR1(b,c,d),GPR141(c),TXNDC3(c), STARD3NL(g) | |
7 | rs3807184 | 7.63 × 10–5 | 0.036 | 0.45 | IL-22 | Hyphae | PBMCs | 7days | NME8(a,c,e), EPDR1(b,c,d),GPR141(c),TXNDC3(c), STARD3NL(g) | |
8 | rs2725008 | 6.17 × 10–5 | 0.015 | 0.43 | IL-6 | Yeast | Macrophages | 24h | ||
8 | rs2725008 | 6.17 × 10–5 | 0.015 | 0.43 | TNF-α | Yeast | Macrophages | 24h | ||
10 | rs10884489 | 1.02 × 10–5 | 0.032 | 0.48 | TNF-α | Yeast | Macrophages | 24h | ||
11 | rs1945666 | 7.37 × 10–5 | 0.038 | 1.99 | IL-17 | Yeast | PBMCs | 7days | ||
11 | rs1945666 | 7.37 × 10–5 | 0.038 | 1.99 | TNF-α | Yeast | Macrophages | 24h | ||
12 | rs7306373 | 1.33 × 10–5 | 0.034 | 0.47 | IFNγ | Hyphae | PBMCs | 7days | CPM(a,b), LYZ(a,b) | |
13 | rs9530375 | 5.45 × 10–5 | 0.049 | 2.329 | IFNγ | Yeas | PBMCs | 7days | PLEKHA1(d) | |
13 | rs7322708 | 2.08 × 10–5 | 0.026 | 2.34 | IL-22 | Hyphae | PBMCs | 7days | CMIP(d),HSPA9(d), PLEKHA1(d), PPP1R8(d) | |
13 | rs6563046 | 4.35 × 10–6 | 0.001 | 2.12 | IL-6 | Yeast | Macrophages | 24h | RNF219(c) | |
13 | rs6563046 | 4.35 × 10–6 | 0.002 | 2.12 | TNF-α | Yeast | Macrophages | 24h | RNF219(c) | |
13 | rs7337751 | 7.61 × 10–5 | 0.017 | 2.05 | TNF-α | Yeast | Macrophages | 24h | ||
14 | rs71423384 | 1.29 × 10–5 | 0.041 | 0.39 | IL-1β | Hyphae | PBMCs | 24h | SLC8A3(a), KIAA0247, SLC10A1(g) | |
15 | rs12593397 | 3.02 × 10–5 | 0.037 | 0.51 | IFNy | Yeast | Whole blood | 48h | EHD4(b,c,d), JMJD7-PLA2G4B(c,g),RPAP1(c), PLA2G4B(c,g), SPTBN5(c), RP11-23P13.6(c), RP11-107F6.3(c), JMJD7(c), MAPKBP1(c), MGA(c), CTD-2382E5.3(c), PLA2G4D(g), PLA2G4E(g), PLA2G4F(g) | |
16 | rs2189521 | 9.14 × 10–5 | 0.048 | 0.52 | IL-1β | Yeast | PBMCs | 24h | GTF3C1(c), IL21R(c) | |
16 | rs4785438 | 8.50 × 10–5 | 0.039 | 0.53 | IL-1β | Yeast | Whole blood | 48h | CARD15(c), NOD2(a,c,d), CYLD(c), NKD1(c), RP11-401P9.4(c), RP11-327F22.6(c) | |
17 | rs3027232 | 9.27 × 10–5 | 0.019 | 0.47 | IL-17 | Yeast | PBMC | 7days |
C17orf44(c), TMEM107(c), TMEM88(b), ALOX15B(a,g), ALOXE3(a,f,g), LINC00324(c), ALOX12B(g) |
|
17 | rs3027232 | 9.27 × 10–5 | 0.009 | 0.47 | IL-22 | Yeast | PBMCs | 7days |
C17orf44(c), TMEM107(c), TMEM88(b), ALOX15B(a,g), ALOXE3(a,f,g), LINC00324(c), ALOX12B(g) |
|
17 | rs3027232 | 9.27 × 10–5 | 0.005 | 0.47 | IFNy | Yeast | Whole blood | 48h | C17orf44(c), TMEM107(c), TMEM88(b), ALOX15B(a,g), ALOXE3(a,f,g), LINC00324(c), ALOX12B(g) | |
17 | rs3027232 | 9.27 × 10–5 | 0.046 | 0.47 | IL-1β | Yeast | Whole blood | 48h | C17orf44(c), TMEM107(c), TMEM88(b), ALOX15B(ag), ALOXE3(a,f,g), LINC00324(c), ALOX12B(g) | |
17 | rs3027232 | 9.27 × 10–5 | 0.036 | 0.47 | IL-6 | Yeast | Whole blood | 48h | C17orf44(c), TMEM107(c), TMEM88(b), ALOX15B(a,g), ALOXE3(a,f,g), LINC00324(c), ALOX12B(g) | |
18 | rs72987764 | 3.01 × 10–6 | 0.042 | 0.38 | IL-17 | Yeast | PBMCs | 7days | ||
18 | rs72987756 | 2.65 × 10–5 | 0.04 | 0.449 | IL-6 | Yeast | Macrophages | 24h | ||
18 | rs72987756 | 2.65 × 10–5 | 0.03 | 0.449 | TNF-α | Yeast | Macrophages | 24h | ||
22 | rs113413 | 8.04 × 10–5 | 0.029 | 2.03 | IL-17 | Hyphae | PBMC | 7days | AP000350.4(c), AP000350.5(c), AP000350.6(c),AP000351.10 (c), DDT(c,d), DDTL(c),GSTT1(c,d),KB-226F1.1(c),KB-226F1.2(c), MIF(d), GGT5(a,g), DDT22(c), UPB1(c), |
Abbreviations: Chr, chromosome; cQTL, cytokine quantitative trait loci; IFN, interferon; IL, interleukin; OR, odds ratio; PBMCs, peripheral blood mononuclear cells; SNP, single-nucleotide polymorphisms; TNF, tumor necrosis factor.
a Differentially-expressed genes after 24-hour Candida stimulation.
b Differentially-expressed genes after 4-hour Candida stimulation.
c Candidemia-associated SNPs or proxies (r2 > 0.8) showed an eQTL effect in whole blood based on publicly available eQTL datasets [18,47,48].
d Candidemia-associated SNPs or proxies (r2 > 0.8) showed an eQTL effect in monocytes [46,49].
e Candidemia-associated SNP in strong LD (r2 > 0.8) with a synonymous variant.
f Candidemia-associated SNP is a missense variant.
g Gene is involved in lipid metabolism (Reactome).
Genetic Basis for Interplay Among Phospholipid Metabolism, Cytokine Responses, and Reactive Oxygen Species Production in Candidemia
Given that ROS production is one of the downstream products of AA metabolism, we tested whether the candidemia SNPs (P = 9.99 × 10–5) have an impact on ROS production in response to C albicans [31, 32]. For this, we generated genotypes and measurements of ROS production in PBMCs of 122 healthy volunteers from a 200FG cohort. Due to the small cohort size, only the candidemia-associated SNP, rs2725008, on chromosome 8 was found to significantly influence ROS production (P = .007) (Supplementary Table S6). Five additional candidemia-associated SNPs influenced ROS production with a P value of nominal significance (P < .05) (Supplementary Table S6). Next, to examine the overall impact of candidemia susceptibility SNPs on ROS production, we constructed an allelic risk score for candidemia susceptibility based on the 24 independent susceptibility variants that affect cytokine levels and correlated with ROS levels (Figure 4A). We observed a strong negative correlation between the abundance of ROS and allelic scores (P = .0065, r = −0.24). This correlation remained significant only when calculating allelic scores based on 16 independent candidemia-associated SNPs that affect only monocyte-derived cytokine levels (IL-6, IL-1β, and tumor necrosis factor [TNF]α) (P = .00063, r = −0.31) (Figure 4B). The correlation did not remain significant when using 12 independent candidemia-associated variant SNPs that affect T cell-derived cytokines (IL-17, IL-22, and interferon [IFN]γ) (Figure 4C). No significant correlations observed when allelic scores constructed as the total number of risk alleles either from 49 susceptibility loci (P = 0.57, r = −0.052) (excluding those that influence Candida-induced cytokines) (Figure S3A) or 69 susceptibility SNPs (P < 9.99 × 10–5), including those that influence Candida-induced cytokines (P = 0.15, r = −0.13) (Figure S3B). Furthermore, we correlated the ROS measurements with the levels of the IL-6 produced in Candida-stimulated PBMCs. We observed a strong positive correlation between IL-6 and ROS levels (r = 0.27, P = .036) (Figure S4). Overall, these findings indicate that candidemia-associated loci have a functional role in the anti-Candida defense mechanism by influencing cytokine and ROS production upon Candida infection.
DISCUSSION
In a given at-risk population, not all patients develop infections, which indicates that there is a strong genetic influence on individual susceptibility to different infections [33]. Genetic association studies that identify risk genes for systemic fungal infections are extremely challenging because of the difficulty of enrolling large numbers of patients and of using appropriate controls, such as case-matched controls. To circumvent these challenges, we combined traditional GWAS strategy with systems biology approaches. By testing whether SNPs associated with susceptibility to candidemia influences cytokine levels and ROS production in response to C albicans in an unbiased manner, we prioritize critical genes and pathways for candidemia susceptibility.
We identified a novel strong genetic association of candidemia with polymorphisms in PLA2G4B. It has been suggested that PLA2G4B-F enzymes may control phospholipid and/or arachidonate metabolism in a tissue-specific manner [34]. However, the role of PLA2G4B has never been studied in the context of infectious diseases. We observed that specific downregulation of PLA2G4B using siRNA resulted in a tendency to improve the candidacidal properties of PBMCs, suggesting a role of this gene in the host response against C albicans. Of note, genes from other susceptibility loci showed a significant enrichment in alpha-Linolenic acid and linoleic acid (metabolic precursors of AA), phospholipid, as well as AA metabolism.
Furthermore, we demonstrated the moderate impact of Candida-induced cQTLs on susceptibility to candidemia. As shown in Table 1, there are 24 independent cQTLs that modulate candidemia susceptibility. Genes well known to be involved in lipid and AA metabolic processes were found in these loci. For instance, genes encoding for members of the lipoxygenase family, such as ALOX15B, ALOXE3 and ALOX12B, in candidemia-associated locus rs3027232, are known to catalyze the attachment of molecular oxygen O2 to polyunsaturated fatty acids, such as AA. It has been previously reported that a mutation in ALOXE3 gene among nonbullous congenital ichthyosiform erythroderma patients predispose the patients to high risk of severe cutaneous fungal infections [35]. Many metabolic changes, including lipid synthesis, are shown to be important in regulating cytokine secretion to coordinate the immune response [36–38]. In mouse studies, hyperlipoproteinemia increased the susceptibility to systemic candidiasis due to an increased fungal outgrowth in their organs [39]. In human studies, infusion of lipoproteins enhanced the growth of C albicans in the plasma of volunteers [40], suggesting that hyperlipidemia have deleterious effects by enhancing the growth of C albicans in both species
One of the well known mechanisms of action of AA is triggering the generation of superoxides (O2−) in leukocytes, which act as microbicidal agents in phagocytes and a second messenger in many cell types [41, 42]. Thus, one would expect that ROS decrease or depletion would be associated with increased susceptibility to infection. Indeed, mice deficient in ROS production have been described to be more susceptible to different infections [5]. In addition, chronic granulomatous disease patients, who lack a functional NOX2 protein, show an increased susceptibility to various pathogens, including bacteria and fungi such as Candida and Aspergillus [43]. In this study, we observed that the numbers of risk alleles at candidemia susceptibility loci are negatively correlated with ROS production in response to C albicans. Of note, the correlation remained significant when using susceptibility variants that only influence monocyte-derived cytokines (IL-6, IL-1β, and TNFα), but not T cell-derived cytokines (IL-17, IL-22, and IFNγ). These data show that individuals with a higher genetic risk to candidemia susceptibility show a decreased ROS as well as cytokine production in response to C albicans infection. In the future, stratifying patients based on their genetic profiling would allow us to identify those at high-risk, who will benefit most from the treatment.
CONCLUSIONS
There are some limitations in this study. First, although it is difficult to gather large cohorts of patients, we need independent candidemia cohorts to replicate the genetic associations. Second, the publicly available eQTL datasets from whole blood used miss cell-type specific [44–46] or context-specific eQTLs upon Candida infection. Nevertheless, our approach to correlate the polygenic risk score for candidemia with ROS and cytokine levels in an independent cohort prioritized the disturbed lipid homeostasis and oxidative stress as potential genetic mechanisms that affect cytokine production to influence susceptibility to candidemia.
SUPPLEMENTARY DATA
Supplementary materials are available at The Journal of Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.
Notes
Acknowledgments. We thank all volunteers from the 500 and 200 Functional Genomics cohort of the Human Functional Genomics Project for participation in the study. We also thank K. McIntyre for editing the final text. Finally, we thank the UMCG Genomics Coordination center, the UG Center for Information Technology, and their sponsors BBMR-NL and Target for storage and compute infrastructure.
Financial support. This work was funded by the following: a research grant (2017) from the European Society of Clinical Microbiology and Infectious Diseases and a Radboudumc Hypatia Grant (2018) (to V. K.); a European Research Council (ERC) Consolidator Grant (FP/2007–2013/ERC Grant 2012-310372) and a Netherlands Organization for Scientific Research (NWO) Spinoza Prize Grant (NWO SPI 94-212; to M. G. N.); an ERC Advanced Grant (FP/2007–2013/ERC Grant 2012-322698) and an NWO Spinoza Prize Grant (NWO SPI 92-266; to C. W.); and a European Union Seventh Framework Program Grant (EU FP7) TANDEM Project (HEALTH-F3-2012-305279; to C. W. and V. K.). Y. L. and M. O. were supported by VENI Grants (863.13.011 and 016.176.006) from the NWO. M. D. J. was supported by a grant from the National Institutes of Health (AI-51537).
Potential conflicts of interest. All authors: No reported conflicts of interest. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest.
References
- 1. Guinea J. Global trends in the distribution of Candida species causing candidemia. Clin Microbiol Infect 2014; 20 (Suppl 6):5–10. [DOI] [PubMed] [Google Scholar]
- 2. Brown GD, Denning DW, Gow NA, Levitz SM, Netea MG, White TC. Hidden killers: human fungal infections. Sci Transl Med 2012; 4:1–9. [DOI] [PubMed] [Google Scholar]
- 3. Boyce KJ, Andrianopoulos A. Fungal dimorphism: the switch from hyphae to yeast is a specialized morphogenetic adaptation allowing colonization of a host. FEMS Microbiol Rev 2015; 39:797–811. [DOI] [PubMed] [Google Scholar]
- 4. Campion EW, Kullberg BJ, Arendrup MC. Invasive candidiasis. N Engl J Med 2015; 373:1445–1456. [DOI] [PubMed] [Google Scholar]
- 5. Paiva CN, Bozza MT. Are reactive oxygen species always detrimental to pathogens? Antioxid Redox Signal 2014; 20:1000–37. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. van de Veerdonk FL, Kullberg BJ, Netea MG. Adjunctive immunotherapy with recombinant cytokines for the treatment of disseminated candidiasis. Clin Microbiol Infect 2012; 18:112–9. [DOI] [PubMed] [Google Scholar]
- 7. Armstrong-James D, Brown GD, Netea MG, et al. . Immunotherapeutic approaches to treatment of fungal diseases. Lancet Infect Dis 2017; 17:e393–402. [DOI] [PubMed] [Google Scholar]
- 8. Delsing CE, Gresnigt MS, Leentjens J, et al. . Interferon-gamma as adjunctive immunotherapy for invasive fungal infections: a case series. BMC Infect Dis 2014; 14:166. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Buddingh EP, Leentjens J, van der Lugt J, et al. . Interferon-gamma immunotherapy in a patient with refractory disseminated candidiasis. Pediatr Infect Dis J 2015; 34:1391–4. [DOI] [PubMed] [Google Scholar]
- 10. Li Y, Oosting M, Smeekens SP, et al. . A functional genomics approach to understand variation in cytokine production in humans. Cell 2016; 167:1099–110.e14. [DOI] [PubMed] [Google Scholar]
- 11. Li Y, Oosting M, Deelen P, et al. . Inter-individual variability and genetic influences on cytokine responses to bacteria and fungi. Nat Med 2016; 22:952–60. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Cho KJ, Seo JM, Kim JH. Bioactive lipoxygenase metabolites stimulation of NADPH oxidases and reactive oxygen species. Mol Cells 2011; 32:1–5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Kumar V, Cheng SC, Johnson MD, et al. . Immunochip SNP array identifies novel genetic variants conferring susceptibility to candidaemia. Nat Commun 2014; 5:4675. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Chang CC, Chow CC, Tellier LC, Vattikuti S, Purcell SM, Lee JJ. Second-generation PLINK: rising to the challenge of larger and richer datasets. Gigascience 2015; 4:7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Matzaraki V, Gresnigt MS, Jaeger M, et al. . An integrative genomics approach identifies novel pathways that influence candidaemia susceptibility. PLoS One 2017; 12:e0180824. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Lamparter D, Marbach D, Rueedi R, Kutalik Z, Bergmann S. Fast and rigorous computation of gene and pathway scores from SNP-based summary statistics. PLoS Comput Biol 2016; 12:e1004714. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. McCarthy S, Das S, Kretzschmar W, et al. . A reference panel of 64,976 haplotypes for genotype imputation. Nat Genet 2016; 48:1279–83. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Võsa U, Claringbould A, Westra HJ, et al. . Unraveling the polygenic architecture of complex traits using blood eQTL meta-analysis. bioRxiv 2018; 447367. [Google Scholar]
- 19. Smeekens SP, Ng A, Kumar V, et al. . Functional genomics identifies type I interferon pathway as central for host defense against Candida albicans Sanne. Nat Commun 2013; 4:1–18. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Liu L, Okada S, Kong XF, et al. . Gain-of-function human STAT1 mutations impair IL-17 immunity and underlie chronic mucocutaneous candidiasis. J Exp Med 2014; 8:1635–48. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. van de Veerdonk FL, Plantinga TS, Hoischen A, et al. . STAT1 mutations in autosomal dominant chronic mucocutaneous candidiasis. N Engl J Med 2011; 365:54–61. [DOI] [PubMed] [Google Scholar]
- 22. Song C, Chang XJ, Bean KM, Proia MS, Knopf JL, Kriz RW. Molecular characterization of cytosolic phospholipase A2-beta. J Biol Chem 1999; 274:17063–7. [DOI] [PubMed] [Google Scholar]
- 23. Jayaraja S, Dakhama A, Yun B, et al. . Cytosolic phospholipase A2 contributes to innate immune defense against Candida albicans lung infection. BMC Immunol 2016; 17:27. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Tallima H, El Ridi R. Arachidonic acid: physiological roles and potential health benefits - a review. J Adv Res 2018; 11:33–41. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Rett BS, Whelan J. Increasing dietary linoleic acid does not increase tissue arachidonic acid content in adults consuming Western-type diets: a systematic review. Nutr Metab (Lond) 2011; 8:36. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Lemmon MA. Membrane recognition by phospholipid-binding domains. Nat Rev Mol Cell Biol 2008; 9:99–111. [DOI] [PubMed] [Google Scholar]
- 27. Martin TF. Phosphoinositide lipids as signaling molecules: common themes for signal transduction, cytoskeletal regulation, and membrane trafficking. Annu Rev Cell Dev Biol 1998; 14:231–64. [DOI] [PubMed] [Google Scholar]
- 28. Hubler MJ, Kennedy AJ. Role of lipids in the metabolism and activation of immune cells. J Nutr Biochem 2016; 34:1–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Kominsky DJ, Campbell EL, Colgan SP. Metabolic shifts in immunity and inflammation. J Immunol 2010; 184:4062–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Khovidhunkit W, Kim MS, Memon RA, et al. . Effects of infection and inflammation on lipid and lipoprotein metabolism: mechanisms and consequences to the host. J Lipid Res 2004; 45:1169–96. [DOI] [PubMed] [Google Scholar]
- 31. Aratani Y, Kura F, Watanabe H, et al. . Critical role of myeloperoxidase and nicotinamide adenine dinucleotide phosphate-oxidase in high-burden systemic infection of mice with Candida albicans. J Infect Dis 2002; 185:1833–7. [DOI] [PubMed] [Google Scholar]
- 32. Frohner IE, Bourgeois C, Yatsyk K, Majer O, Kuchler K. Candida albicans cell surface superoxide dismutases degrade host-derived reactive oxygen species to escape innate immune surveillance. Mol Microbiol 2009; 71:240–52. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Chapman SJ, Hill AV. Human genetic susceptibility to infectious disease. Nat Rev Genet 2012; 13:175–88. [DOI] [PubMed] [Google Scholar]
- 34. Long JZ, Cravatt BF. The metabolic serine hydrolases and their functions in mammalian physiology and disease. Chem Rev 2011; 111:6022–63. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. Wang T, Xu C, Zhou X, et al. . Homozygous ALOXE3 nonsense variant identified in a patient with non-bullous congenital ichthyosiform erythroderma complicated by superimposed bullous Majocchi’s granuloma: the consequences of skin barrier dysfunction. Int J Mol Sci 2015; 16:21791–801. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Dimeloe S, Burgener AV, Grählert J, Hess C. T-cell metabolism governing activation, proliferation and differentiation; a modular view. Immunology 2017; 150:35–44. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37. Lachmandas E, Boutens L, Ratter JM, et al. . Microbial stimulation of different Toll-like receptor signalling pathways induces diverse metabolic programmes in human monocytes. Nat Microbiol 2016; 2:16246. [DOI] [PubMed] [Google Scholar]
- 38. Netea MG, Dinarello CA. More than inflammation: interleukin-1β polymorphisms and the lipid metabolism. J Clin Endocrinol Metab 2011; 96:1279–81. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39. Netea MG, Demacker PN, de Bont N, et al. . Hyperlipoproteinemia enhances susceptibility to acute disseminated Candida albicans infection in low-density-lipoprotein-receptor-deficient mice. Infect Immun 1997; 65:2663–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40. Netea MG, Curfs JH, Demacker PN, Meis JF, Van der Meer JW, Kullberg BJ. Infusion of lipoproteins into volunteers enhances the growth of Candida albicans. Clin Infect Dis 1999; 28:1148–51. [DOI] [PubMed] [Google Scholar]
- 41. Suzuki YJ, Forman HJ, Sevanian A. Oxidants as stimulators of signal transduction. Free Radic Biol Med 1996; 22:269–85. [DOI] [PubMed] [Google Scholar]
- 42. Pompéia C, Cury-Boaventura MF, Curi R. Arachidonic acid triggers an oxidative burst in leukocytes. Braz J Med Biol Res 2003; 36:1549–60. [DOI] [PubMed] [Google Scholar]
- 43. Ben-Ari J, Wolach O, Gavrieli R, Wolach B. Infections associated with chronic granulomatous disease: linking genetics to phenotypic expression. Expert Rev Anti Infect Ther 2012; 10:881–94. [DOI] [PubMed] [Google Scholar]
- 44. Fairfax BP, Makino S, Radhakrishnan J, et al. . Genetics of gene expression in primary immune cells identifies cell type-specific master regulators and roles of HLA alleles. Nat Genet 2012; 44:502–10. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45. Brown CD, Mangravite LM, Engelhardt BE. Integrative modeling of eQTLs and cis-regulatory elements suggests mechanisms underlying cell type specificity of eQTLs. PLoS Genet 2013; 9:e1003649. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46. Zeller T, Wild P, Szymczak S, et al. . Genetics and beyond–the transcriptome of human monocytes and disease susceptibility. PLoS One 2010; 5:e10693. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47. GTEx Consortium. The Genotype-Tissue Expression (GTEx) pilot analysis: multitissue gene regulation in humans. Science 2015; 348:648–60. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 48. Westra HJ, Peters MJ, Esko T, et al. . Systematic identification of trans eQTLs as putative drivers of known disease associations. Nat Genet 2013; 45:1238–43. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 49. Fairfax BP, Humburg P, Makino S, et al. . Innate immune activity conditions the effect of regulatory variants upon monocyte gene expression. Science 2014; 343:1246949. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.