Skip to main content
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

An integrative genomics approach identifies novel pathways that influence candidaemia susceptibility

  • Vasiliki Matzaraki,

    Roles Data curation, Formal analysis, Investigation, Methodology, Validation, Visualization, Writing – original draft

    Affiliation Department of Genetics, University Medical Center Groningen, Groningen, The Netherlands

  • Mark S. Gresnigt,

    Roles Investigation, Writing – review & editing

    Affiliation Department of Internal Medicine and Radboud Center for Infectious Diseases, Radboud University Medical Center, Nijmegen, The Netherlands

  • Martin Jaeger,

    Roles Investigation, Writing – review & editing

    Affiliation Department of Internal Medicine and Radboud Center for Infectious Diseases, Radboud University Medical Center, Nijmegen, The Netherlands

  • Isis Ricaño-Ponce,

    Roles Formal analysis

    Affiliation Department of Genetics, University Medical Center Groningen, Groningen, The Netherlands

  • Melissa D. Johnson,

    Roles Investigation, Resources, Validation, Writing – review & editing

    Affiliation Duke University Medical Center, Durham, North Carolina, United States of America

  • Marije Oosting,

    Roles Investigation

    Affiliation Department of Internal Medicine and Radboud Center for Infectious Diseases, Radboud University Medical Center, Nijmegen, The Netherlands

  • Lude Franke,

    Roles Formal analysis

    Affiliation Department of Genetics, University Medical Center Groningen, Groningen, The Netherlands

  • Sebo Withoff,

    Roles Formal analysis, Writing – review & editing

    Affiliation Department of Genetics, University Medical Center Groningen, Groningen, The Netherlands

  • John R. Perfect,

    Roles Investigation, Resources, Validation, Writing – review & editing

    Affiliation Duke University Medical Center, Durham, North Carolina, United States of America

  • Leo A. B. Joosten,

    Roles Investigation, Resources, Validation

    Affiliation Department of Internal Medicine and Radboud Center for Infectious Diseases, Radboud University Medical Center, Nijmegen, The Netherlands

  • Bart Jan Kullberg,

    Roles Investigation, Resources, Validation, Writing – review & editing

    Affiliation Department of Internal Medicine and Radboud Center for Infectious Diseases, Radboud University Medical Center, Nijmegen, The Netherlands

  • Frank L. van de Veerdonk,

    Roles Investigation, Resources, Validation

    Affiliation Department of Internal Medicine and Radboud Center for Infectious Diseases, Radboud University Medical Center, Nijmegen, The Netherlands

  • Iris Jonkers,

    Roles Writing – review & editing

    Affiliation Department of Genetics, University Medical Center Groningen, Groningen, The Netherlands

  • Yang Li,

    Roles Formal analysis, Methodology, Visualization

    Affiliation Department of Genetics, University Medical Center Groningen, Groningen, The Netherlands

  • Cisca Wijmenga,

    Roles Conceptualization, Funding acquisition, Project administration, Resources, Supervision, Writing – original draft

    Affiliations Department of Genetics, University Medical Center Groningen, Groningen, The Netherlands, K.G. Jebsen Coeliac Disease Research Centre, Department of Immunology, University of Oslo, Oslo, Norway

  • Mihai G. Netea ,

    Contributed equally to this work with: Mihai G. Netea, Vinod Kumar

    Roles Conceptualization, Funding acquisition, Investigation, Project administration, Resources, Supervision, Writing – original draft, Writing – review & editing

    Affiliations Department of Internal Medicine and Radboud Center for Infectious Diseases, Radboud University Medical Center, Nijmegen, The Netherlands, Human Genomics Laboratory, Craiova University of Medicine and Pharmacy, Craiova, Romania

  •  [ ... ],
  • Vinod Kumar

    Contributed equally to this work with: Mihai G. Netea, Vinod Kumar

    Roles Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Supervision, Visualization, Writing – original draft, Writing – review & editing

    v.kumar@umcg.nl

    Affiliation Department of Genetics, University Medical Center Groningen, Groningen, The Netherlands

  • [ view all ]
  • [ view less ]

Abstract

Candidaemia is a bloodstream infection caused by Candida species that primarily affects specific groups of at-risk patients. Because only small candidaemia patient cohorts are available, classical genome wide association cannot be used to identify Candida susceptibility genes. Therefore, we have applied an integrative genomics approach to identify novel susceptibility genes and pathways for candidaemia. Candida-induced transcriptome changes in human primary leukocytes were assessed by RNA sequencing. Genetic susceptibility to candidaemia was assessed using the Illumina immunochip platform for genotyping of a cohort of 217 patients. We then integrated genetics data with gene-expression profiles, Candida-induced cytokine production capacity, and circulating concentrations of cytokines. Based on the intersection of transcriptome pathways and genomic data, we prioritized 31 candidate genes for candidaemia susceptibility. This group of genes was enriched with genes involved in inflammation, innate immunity, complement, and hemostasis. We then validated the role of MAP3K8 in cytokine regulation in response to Candida stimulation. Here, we present a new framework for the identification of susceptibility genes for infectious diseases that uses an unbiased, hypothesis-free, systems genetics approach. By applying this approach to candidaemia, we identified novel susceptibility genes and pathways for candidaemia, and future studies should assess their potential as therapeutic targets.

Introduction

Genome-wide association studies (GWAS) have greatly contributed to the identification of susceptibility genes for human complex diseases. However, the need for large cohorts precludes the use of GWAS to identify susceptibility genes for infectious diseases for which only relatively small patient cohorts can be recruited. So far, relatively few GWAS studies have been performed in patients with infections, including studies of viral and bacterial infections [1]. Compared to the hundreds of genetic loci that have been associated to human complex diseases, the number of susceptibility genes identified as associated to infectious diseases remains low. Given that the risk of death due to infectious diseases has a large genetic heritability [2], methods other than GWAS that can be applied to smaller cohorts are crucial to make progress in understanding and treating these diseases.

Candidaemia is the fourth most common systemic bloodstream infection in the United States (US) and is associated with mortality rates of up to 40% [3,4]. It is caused by opportunistic fungal pathogens belonging to the Candida species, particularly Candida albicans (C. albicans), and primarily affects patients with a compromised immune system [5]. However, not all at-risk patients develop candidaemia, indicating that individual differences—including genetic background—influence their susceptibility to the infection. Despite the availability of potent antifungal drugs, the mortality rate of systemic Candida infections remains unacceptably high [6]. In addition to current treatment strategies, only adjuvant immunotherapy such as the administration of recombinant cytokines is believed to improve outcome [7]. Therefore, an understanding of the molecular pathways involved in the human host defence and identification of susceptibility genes is crucial for the design of appropriate prophylactic and immunotherapeutic strategies.

To identify genes underlying susceptibility to Candida infection, we have applied a systems genomics approach that integrates genetic data from candidaemia patients genotyped with the Immunochip platform [8] with Candida-induced gene-expression profiles in human leukocytes, cytokine production from Candida-stimulated peripheral blood mononuclear cells (PBMCs), and circulating cytokine concentrations in candidaemia patients. Using this approach, we demonstrate that genetic susceptibility loci with suggestive associations (P < 9.99 x 10−5) could play a major role in candidaemia susceptibility and we identify genes involved in inflammation, innate immunity, complement and hemostasis as having an important role in determining susceptibility to candidaemia.

Materials and methods

Study populations

To identify genetic variants associated with candidaemia, we performed a two-stage Immunochip-wide analysis of a candidaemia cohort using two control groups in 2014: a population-based healthy cohort and disease-matched cohort (European ancestry), as previously described [9]. Briefly, for Immunochip-wide association analysis, we first used a cohort consisting of 217 candidaemia patients of European ancestry and 11,920 population-based healthy controls (Discovery stage). The demographic and clinical characteristics of the candidaemia cohort have been previously described [9]. Re-analysis of the data and prioritization of genes from susceptibility loci with suggestive associations (P < 9.99 x 10−5) was performed in 2017.

After the discovery of single nucleotide polymorphisms (SNPs) for susceptibility to candidaemia using population-based healthy controls, we then validated our findings using a validation control consisting of 146 disease-matched but candidiasis-free controls. These candidiasis-free control patients were recruited from the same hospital wards as the candidaemia patients so that co-morbidities and clinical risk factors for infection were as similar as possible between patients and controls.

Ethics statement

The study was approved by the institutional review boards at each study centre, and enrollment occurred between January 2003 and January 2009. The study centers are the Duke University Hospital (DUMC, Durham, North Carolina, USA) and Radboud University Nijmegen Medical Centre (Nijmegen, The Netherlands). All adult subjects provided informed written consent.

Genotyping and case-control analysis of candidaemia cohort

DNA was isolated using the Gentra Pure Gene Blood kit (Qiagen, Venlo, the Netherlands) according to the protocol of the manufacturer. Genotyping of candidaemia patients and both control groups was performed using Immunochip according to Illumina’s protocol [10]. Genotype data analysis and quality control of this cohort has been previously reported [9]. Briefly, in the discovery stage, the associations of Immunochip SNPs and susceptibility with candidaemia were tested by logistic regression using the first four components from the multidimensional scaling analysis as covariates. We considered a P value < 9.99 x 10−5 as the threshold for suggestive association to select 268 SNPs in 77 independent loci for validation. For validation analysis using candidaemia case-matched controls, the association between SNPs and candidaemia was tested by logistic regression using the first four multidimensional scaling analysis components as covariates. A validation P value < 0.05 was considered significant.

PBMC isolation and stimulation with Candida albicans

Isolation of PBMCs from eight healthy volunteers and stimulation of PBMCs was performed as described previously [11]. After cell counting with a hemocytometer, the cell number was adjusted to 5 x 106/mL. To identify the transcriptome upon Candida stimulation, 5 x 105 isolated PBMCs were incubated with 1 x 106/mL heat-killed C. albicans (UC 820) (C. albicans:PBMC ratio of 1:2.5) and RPMI culture medium as a control for 4 and 24 hours. C. albicans UC 820 is a well-described strain regarding its immune responses in PBMCs [12].

Analysis of RNA sequencing reads

The RNA sequencing analysis of this dataset was described previously [13]. Briefly, sequencing reads were mapped to the human genome using STAR (version 2.3.0). The aligner was provided with a file containing junctions from Ensembl GRCh37.71. The Htseq-count of the Python package HTSeq (version 0.5.4p3) was used (http://www-huber.embl.de/users/anders/HTSeq/doc/overview.html) to quantify the read counts per gene based on annotation version GRCh37.71, using the default union-counting mode. Differentially expressed genes were identified by statistical analysis using the DESeq2 package. The statistically significant threshold (False Discovery Rate P < 0.05 and Fold Change ≥ 2) was applied.

Pathway enrichment analysis

We performed gene enrichment analysis using ConsensusPathDB-human database (CPDB; http://cpdb.molgen.mpg.de/) [14]. The over-representation analysis is done using the default setting in which the database compares the predefined lists of functionally associated genes (pathways and Gene Ontology categories) to the list of differentially expressed genes and generates P values based on the hypergeometric test. The hypergeometric test P values are further corrected for multiple testing using the false discovery rate method.

MAP3K8 inhibition of cytokines in Candida-stimulated PBMCs

PBMCs (5 x 105) were placed in 96-well round-bottom plates in a final volume of 200μL. PBMCs were stimulated with (1 x 106/mL) heat-inactivated conidia of C. albicans in the presence or absence of variable concentrations (10μM, 50μM, and 200μM) of a MAP3K8 inhibitor (#5240, Tocris Cookson Ltd., Bristol, UK) or a vehicle (DMSO) control. PBMCs were stimulated for 24 or 48 hours at 37°C and 5% CO2. After stimulation, culture supernatants were collected and stored at -20°C until cytokine assays were performed. Cytokine levels from Candida-stimulated PBMCs were measured in cell culture supernatants using enzyme-linked immunosorbent assay (ELISA) according to the manufacturer’s instructions (R&D Systems, MN, USA). Differences in cytokine levels were compared for statistical significance using the Wilcoxon rank test.

Results

Candida albicans induces transcription of genes involved in inflammation and immune-hemostasis interaction

In our experiments, we have used heat-killed C. albicans instead of live C. albicans for two main reasons. The first one is that long incubation periods of more than 48 hours to obtain the T cell derived cytokines IL-17, IL-22 and IFNγ result in overgrowth of the fungal cells and, ultimately, in cell lysis. Therefore, to achieve cell viability, we opted for heat-killed C. albicans. Secondly, the immune response is enhanced when using heat-killed C. albicans instead of live C. albicans. Heat-killing enhances the accessibility of β-glucans, whereas in live C. albicans yeast cells, β-glucans are shielded from immune recognition by a mannan layer. The β-glucans, which can be found in the inner layer of the C. albicans cell wall represent the main conserved pathogen-associated molecular patterns (PAMPs) that are recognized by one of the most important innate receptors for the recognition of Candida species, known as C-type lectin receptors (CLRs) [15]. The recognitions of β-glucans by CLRs constitutes the first step in the development of an immune response to C. albicans. Therefore, enhanced accessibility of β-glucans results in a pronounced immune response when using heat-killed C. albicans instead of live cells.

The transcriptome of PBMCs upon Candida stimulation was profiled by next-generation RNA sequencing (Fig 1). A total of 312 (4 hours) and 1,476 (24 hours) protein-coding genes were identified that showed >1.5-fold higher expression in cells stimulated with C. albicans compared to unstimulated cells (adjusted P < 0.05) (S1 Table, S1A and S1B Fig). A total of 246 protein-coding genes were found to be more strongly induced both 4 h and 24 h after stimulation (S2 Table), while 66 genes showed increased expression after 4 h only and 1,230 genes were more strongly expressed after 24 h only.

thumbnail
Fig 1. Transcriptome of PBMCs upon Candida stimulation.

(A) Principal component analysis of RNA sequencing data obtained from eight healthy volunteers upon stimulation of their PBMCs with C. albicans for 4 or 24 hours or unstimulated. Four distinct groups were observed based on time-dependent exposure to C. albicans or culture medium alone using the first two principal components. Principal Component 1 (PC1, x-axis) represents 24% and PC2 (y-axis) represents 19% of total variation in the data. Green and red circles represent the stimulated samples with heat-killed C. albicans for 4 and 24 hours respectively and purple and blue circles represent the samples incubated with RPMI for 4 and 24 hours (used as controls) respectively.

https://doi.org/10.1371/journal.pone.0180824.g001

Among the significantly differentially expressed genes, pathway enrichment analysis showed that there is an overrepresentation of genes involved in cytokine signalling and of chemokine genes at both 4 h and 24 h after stimulation. Pathway enrichment analysis on differentially expressed protein-coding genes confirmed previously described Candida-response pathways including cytokine signalling (4 h: P = 1.36 x 10−12, 24 h: P = 3.19 x 10−9), Toll-like-receptor-mediated signalling [16] at 4 h (P = 0.00553), interferon signalling [17] (4 h: P = 4.90 x 10−10, 24 h: P = 1.41 x 10−5) and RIG-I/MDA5 mediated induction of IFN-alpha/beta pathways (4 h: P = 7.55 x 10−6, 24 h: P = 0.00753) (S3 and S4 Tables) [18]. The enrichment analysis also uncovered an overrepresentation of differentially expressed genes involved in dissolution of the fibrin clot at 4 h (P = 0.0018) and 24 h (P = 0.00422), alternative complement activation at 4h (P = 0.0022), platelet activation, signalling and aggregation at 24 h (P = 0.0016) and activation of C3 and C5 -a central step of complement activation- at 24h (P = 0.00479). An even stronger effect for hemostasis (P = 6.51 x 10−6) was observed at 24h (Figs 2 and 3). These results suggest that, along with genes involved in cytokine and interferon signalling, genes involved in the integration of immunity and hemostasis as biological processes may also be involved in the pathogenesis of systemic Candida infections.

thumbnail
Fig 2. Pathway enrichment analysis of the Candida-induced transcripts at 4 hours.

S3 Table displays all the signalling and biological processes as well as their enrichment p-values.

https://doi.org/10.1371/journal.pone.0180824.g002

thumbnail
Fig 3. Pathway enrichment analysis of the Candida-induced transcripts at 24 hours reveals a stronger effect for hemostasis at 24 hours than 4 hours.

S4 Table displays all the signalling and biological processes as well as their enrichment p-values.

https://doi.org/10.1371/journal.pone.0180824.g003

Immunochip-based genetic study identifies 18 candidaemia susceptibility loci

After identification of the transcriptome profile induced by Candida in human PBMCs, we explored whether these pathways could be validated through genetic association data from a clinical candidaemia cohort of 217 patients. We had previously reported three loci to be associated to Candida infection at genome-wide significance (P < 5 x 10−8) by performing a case-control study on the Immunochip platform [9]. In addition to these three loci, during the discovery stage, we identified 77 independent loci showing suggestive associations with P values lower than 9.99 x 10−5. It is these independent loci that we have followed up in this study by validating them against 146 disease-matched controls to exclude confounding effects of the patient’s clinical background. Following screening of the disease-matched controls, associations at 18 independent loci could be successfully validated (P < 0.05; S2 Fig and Table 1).

thumbnail
Table 1. Prioritization of putative causal genes of candidaemia using an Immunochip-based association study.

https://doi.org/10.1371/journal.pone.0180824.t001

eQTL and differential expression prioritize candidaemia causal genes enriched for inflammation and hemostasis

To identify causal genes from these 18 susceptibility loci, three different approaches were applied. In our first approach, we tested whether the 18 candidaemia SNPs were in linkage disequilibrium (LD) with variants that alter the protein-coding sequence of the gene (both synonymous and non-synonymous). Two candidaemia-associated SNPs were identified, rs1802141 and rs12491812, that were in strong LD with synonymous variants (S5 Table) and therefore may point to putative causal genes. SNP rs1802141 on chromosome 16 is in strong LD (R2 = 0.82, D’ = 1) with a synonymous variant, rs61747536, in the SPNS1 gene (Fig 4A). The SPNS1 gene, also known as HSpin1, has been described to induce a caspase-independent autophagic cell death in cultured human cells [19]. SNP rs12491812 on chromosome 3 is in strong LD (R2 = 0.89, D’ = 1) with two synonymous variants, rs2239753 and rs2239752, in the CISH gene (Fig 4B). The CISH locus has been associated with major infectious diseases such as bacteraemia, tuberculosis, malaria [20], and viral infections such as hepatitis B [21,22], suggesting shared susceptibility genes among different infections.

thumbnail
Fig 4. Prioritization of candidaemia genes enriched for blood coagulation based on eQTL and differential expression data.

Regional association plots are shown around candidaemia-associated SNPs identified in the Discovery cohort. (A) rs1802141 (P = 9.325 x 10−7) on chromosome 16 is in strong LD with the synonymous variant rs61747536 (R2 = 0.82, D’ = 1) in the SPNS1 gene and (B) rs12491812 (P = 1.67 x 10−5) on chromosome 3 is in strong LD (R2 = 0.89, D’ = 1) with two synonymous variants, rs2239753 and rs2239752, in the CISH gene. SNPs are plotted as the–log10 of the p-value. Local LD structure is reflected by the plotted estimated recombination rates (from HapMap) in the region around the associated SNP (purple diamond) and its correlated proxies. The correlation of the lead SNP to other SNPs at the locus is indicated by colour. (C) Regional association plot for rs7149309, which is located in a locus on chromosome 14, that harbors a cluster of serine protease inhibitor genes of the serpin family, with SERPINA1 being the most likely causal gene at this locus based on our expression data upon Candida stimulation. (D) Log2 fold change expression levels of all SERPIN genes upon Candida stimulation for 24 hours at a locus on chromosome 14 marked by the candidaemia-associated SNP rs7149309. SERPINA1 showed a log2 fold change of ~2.6.

https://doi.org/10.1371/journal.pone.0180824.g004

We should mention that the majority of candidaemia-associated SNPs and proxies fall within intergenic/intronic regions, suggesting that they could have a functional potential by affecting gene expression (S5 Table). Therefore, in our second approach, the 18 candidaemia-associated SNPs were mapped for cis-eQTLs using publicly available eQTL datasets from healthy blood donor samples [2326]. Cis-eQTL mapping pinpointed 8 potential causal genes at 6 loci (Table 1).

In our third approach, genes that were located within a 500 kb window around the 18 candidaemia-associated SNPs were extracted and tested as to whether they are differentially expressed after Candida stimulation for 4 and 24 hours (S6 and S7 Tables). Within these ‘susceptibility regions’, we found 18 and 28 genes that are significantly induced upon Candida-stimulation for 4 and 24 hours respectively in human PBMCs, raising the possibility that some of these genes could be regulated by the candidaemia-associated SNPs in the context of Candida infection. For example, one of the 18 candidaemia-associated SNPs, rs7149309, is located within a locus on chromosome 14 that harbours a cluster of ten serine protease inhibitor genes of the serpin family (Fig 4C). Serpin Family A Member 1 (SERPINA1) was significantly expressed in PBMCs in response to Candida stimulation for 24 hours compared to the other SERPIN genes in the same locus (Fig 4D). SERPINA1 encodes human alpha-1 antitrypsin (hAAT). A protective role of AAT has been demonstrated against different types of infectious pathogens, such as in bacterial peritonitis [27] and pulmonary Pseudomonas aeruginosa infection in cystic fibrosis patients [28].

Together, our three approaches prioritized 31 putative candidate genes for candidaemia susceptibility located in 18 loci (Table 1). Intriguingly, nine out of these 31 prioritized genes (LAT, CD19, F5, PROC, C5, SERPINA1, SELP, SELL and SELE) are enriched in the processes of complement and blood coagulation (Table 1, S3 Fig and S8 Table). As expected, pathway enrichment analysis for all 31 candidaemia genes showed enrichment for cytokine- and immune-related signalling pathways, which is in agreement with our pathway enrichment analysis of the Candida-induced transcripts at 4 and 24 hours (Figs 2 and 3, S3 and S4 Tables). In three out of 18 loci, genes that are located in close proximity to the top SNP were shown (Table 1), as we did not find any evidence from the above three approaches to prioritize causal genes.

MAP3K8 modulates cytokine production in patients and in experimental models

One of the candidaemia-associated SNPs, rs1360119, was mapped to the Mitogen-Activated Protein Kinase Kinase 8 (MAP3K8) locus (Fig 5A). In particular, rs1360119 is an intronic variant, suggesting that it may affect gene expression. However, rs1360119 does not show cis-eQTL effect using publicly available eQTL datasets from healthy blood donor samples (Table 1). That does not exclude the fact that this SNP may show an eQTL effect under context-specific conditions, for instance, upon Candida stimulation. In addition, an intronic variant can have functional effects on splicing and, therefore, we can speculate that this SNP may affect splicing.

thumbnail
Fig 5. Candidaemia-associated SNP rs1360119 affects the cytokine expression in serum from patients.

(A) Regional association plot for candidaemia-associated SNP rs1360119, which was mapped to MAP3K8 locus on chromosome 10. (B) Genotypes of candidaemia-associated SNP rs1360119 are consistently associated with expression levels of (B) IL6 (P = 0.018), (C) IL8 (P = 0.024) and (D) IFNɣ (P = 0.013) as measured in serum of candidaemia patients. The numbers of individuals per genotype are shown in parentheses. Cytokine levels were log transformed and P values were obtained using Kruskal Wallis test, and 0.05 was considered statistically significant (minor allele frequency of A = 2% in Europeans from 1000 Genomes Project Phase 3, http://www.internationalgenome.org).

https://doi.org/10.1371/journal.pone.0180824.g005

Of note, MAP kinases are notably attractive therapeutic targets [29]. Therefore, considering the regulatory role of MAP3K8 protein in the production of cytokines in response to lipopolysaccharide [30], bacterial [31,32], and viral pathogens [33], we tested whether the rs1360119 SNP correlates with levels of pro-inflammatory cytokines in serum of candidaemia patients. The T allele of the SNP rs1360119 is associated with decreased IL-6 (P = 0.018), IL-8 (P = 0.024), and interferon (IFNγ) (P = 0.013) cytokine levels (Figs 5B–5D). Five other candidaemia-associated SNPs showed a moderate association (P < 0.05) with circulating cytokine levels (S9 Table), suggesting that some of the candidaemia-associated variants could determine susceptibility to candidaemia by influencing cytokine production capacity.

To test whether MAP3K8 is involved in regulating Candida-induced cytokine production, we measured cytokine levels upon stimulation of PBMCs with C. albicans in the presence or absence of a MAP3K8 chemical inhibitor. Three different cytokines (TNFα, IL-6, and IFNγ) known to be involved in host defence against C. albicans were measured [15]. MAP3K8 activity was blocked using three different concentrations of the MAP3K8 inhibitor (10 μM, 50 μM, and 200 μM). Overall, we observed a dose-dependent reduction in all three cytokine levels upon increasing concentration of MAP3K8 inhibitor. A significant decrease in Candida-induced TNFα production was observed compared to the DMSO-control at 200 μM of MAP3K8 inhibitor (P = 0.03) upon 24-hour stimulation (Fig 6A). In addition, a significant decrease in IL-6 was observed upon stimulation with Candida conidia for 24 hours with 200 μM MAP3K8 inhibitor compared to DMSO (P = 0.04) (Fig 6B). We should note that IFNγ, one of the most crucial cytokines for efficient host defence for systemic candidiasis, decreased compared to DMSO after stimulation with Candida conidia (P = 0.057) for 48 hours at 200 μM of MAP3K8 inhibitor (Fig 6C). Overall, these experiments demonstrate that MAP3K8 regulates cytokine levels upon Candida infection.

thumbnail
Fig 6. MAP3K8 modulates cytokine production in PBMCs.

Median expression levels of (A) TNFα (B) IL6 and (C) IFNγ in Candida-stimulated PBMCs upon inhibition of MAP3K8 at three different concentrations (10 μM, 50 μM, and 200 μM). Candida conidia was added at a concentration of 1 x 106/ml. P values were obtained using Wilcoxon rank test (*P < 0.05). Data shown are from two independent experiments of PBMC stimulation for 24 and 48 hours with conidia of C. albicans.

https://doi.org/10.1371/journal.pone.0180824.g006

Discussion

Although Candida spp are the fourth most common cause of sepsis in the US, and the most common cause of fungal sepsis in Europe, the difficulty of recruiting sufficient numbers of patients has prevented the use of large GWAS studies to reveal the genetic factors involved in the pathophysiology of Candida infections. Our study demonstrates the potential of using a systems genomics approach to obtain novel insights into the genetic basis and host defence mechanisms involved in relatively rare infections in which classical GWAS studies are not applicable. By integrating transcriptomic, genetic and immunological studies, we have identified novel susceptibility genes for candidaemia and new potential therapeutic targets.

The first major group of pathways that are induced during stimulation of human PBMCs with C. albicans are genes important for inflammation and innate immunity, cytokine and chemokine synthesis, interferon and inflammasome signalling. This is not surprising as these pathways have been previously shown to have a crucial role for antifungal host defence [15]. The importance of inflammation and innate immunity was also validated by genetic studies that confirmed that the genes determining susceptibility to candidaemia were enriched for inflammatory genes. However, our analysis has also provided new insight in additional processes that apparently have an important impact on candidaemia. One of these processes is complement activation, which has been shown to modulate cytokine production induced by Candida stimulation [34].

Hemostasis is another important biological process suggested by our study to be involved in susceptibility to candidaemia. Hemostasis is known to be strongly activated during sepsis and to interact with the inflammatory responses [3538]. Strong interactions between immune defence and hemostasis are well documented for bacterial infections, and hemostasis has been previously associated with an increased susceptibility for bacterial sepsis [37,39,40]. Some of the hemostasis genes that we identified as candidaemia susceptibility genes (SERPINA1, LAT and F5) have also been shown to be important for bacterial sepsis (S4A Fig) [41]. Future studies should examine the involvement of these pathways and mechanisms in systemic Candida infection, especially as platelets contain several anti-fungal defensins [42], indicating their potential involvement in candidaemia. For example, an examination of protein-protein interaction data using the Plateletweb database found that 14 of our candidaemia susceptibility proteins were found in platelets (S4B Fig).

Another novel observation of the present study is the association of SERPINA1 gene with susceptibility to candidaemia. Mutations in this gene lead to AAT deficiency and predispose individuals to chronic obstructive pulmonary disease and liver diseases [43,44]. Considering the important role of AAT in modulation of inflammation, as well as the recent beneficial effect of human AAT against bacterial infections [45], hAAT could represent a potential novel adjuvant immunotherapy against systemic Candida infection.

MAP3K8 is known as a critical gene in innate immune responses linking pattern recognition receptors such as Toll-like receptors (TLRs) to TNF production through activation of extracellular signal-regulated kinase [30,46]. In this study we also demonstrate that MAP3K8 plays an important role in Candida infection by modulating cytokine production. The link of MAP3K8 with TLR-signalling suggests that the innate immune response is important in the context of Candida infection. In particular, non-synonymous SNPs in TLR1 have previously been associated with increased susceptibility to candidaemia, highlighting the role of TLRs in the recognition of Candida species [47]. Furthermore, the critical role of MAP3K8 in host immune defence has been demonstrated for other infectious pathogens including influenza virus [33] and Listeria monocytogenes [32]. Most importantly, MAP3K8 is an important and novel therapeutic target for inflammatory diseases [48]. A recent computational approach using publicly available transcriptome datasets for the discovery of common immunomodulators in fungal infections also pinpointed MAP3K8 and SERPINE1 in the top ten consistently perturbed gene sets [49]. However, further functional studies are needed to shed light on the role of these genes in host defence against Candida infection and to understand their potential as therapeutic targets.

Several limitations also apply to this study. First, the patient cohort study is relatively small, which limited our power to identify many of the genetic factors influencing susceptibility to candidaemia. Second, we lack genetic validation in an independent cohort of patients. This is because no such cohort is currently available: the cohort studied here is the largest candidaemia cohort currently available. In addition, a potential limitation of the present study is that the Immunochip platform covers only 5% of the human genome, i.e. we still lack genome-wide information. This means that further critical genetic variations for candidaemia remain to be discovered on a genome-wide scale. Furthermore, it should be mentioned that C.albicans is a polymorphic fungus and is encountered either as yeast or hyphal forms. The transition between conidia and hyphae is a virulence trait of C. albicans and mutants that are locked in the yeast form are less virulent in experimental models of disseminated candidiasis [50,51]. Differential cytokine expression and excretion by immune cells may explain the increased invasiveness of hyphae. For instance, hyphae form has been shown to be unable to induce IFNγ in either human PBMCs or murine splenic lymphocytes and conidia induced a much higher TNFα production than hyphae did [52]. This differential cytokine expression may be attributed to structural differences in the cell wall between yeast and hyphal forms and therefore, the human innate immune system can discriminate between yeast and hyphae [53]. In addition, the accessibility of different pathogen associated molecular patterns (PAMPs) between hyphae and conidia of C. albicans as well as the potential of only hyphae to activate the inflammasome can explain the induction of immune responses that discriminate between conidia and hyphae forms [53,54]. Thus, by using only heat-killed C. albicans conidia in the present study may not represent the full physiological conditions in humans, and further studies should address the differential effect of both forms in PBMCs. Last, C. albicans strains may vary in pathogenicity and, therefore, may elicit different host immune responses making it interesting to consider more clinical strains (in addition to UC 820) for validation in the future.

To conclude, the application of an unbiased, hypothesis-free, systems integrative genomics approach has the power to identify novel susceptibility genes for infectious diseases such as systemic candidiasis. In this study, this approach highlighted genes in inflammation, innate immunity and hemostasis pathways that contribute to the genetic susceptibility against candidaemia. We therefore believe that such integrative approaches are an important tool for future identification of genetic susceptibility to rare infectious diseases.

Supporting information

S1 Fig.

Heatmaps showing the expression of protein-coding genes, which showed >1.5-fold higher expression, upon (A) 4 and (B) 24-hour stimulation with C. albicans in PBMCs from healthy volunteers. RPMI medium was used as control. (adjusted P < 0.05).

https://doi.org/10.1371/journal.pone.0180824.s001

(TIF)

S2 Fig. Immunochip-wide association analysis with candidaemia.

Manhattan plot highlighting the 18 independent loci showing suggestive association with candidaemia (P < 9.99 x 10−5) using a second set of case-matched controls. The y-axis represents the–log10P values of 122,779 SNPs. Their chromosomal positions are shown on the x axis. The dotted line represents the suggestive threshold for association (P < 9.99 X 10−5). P values were not corrected for multiple testing when testing for association with candidaemia susceptibility at 18 independent loci identified in the discovery stage.

https://doi.org/10.1371/journal.pone.0180824.s002

(TIF)

S3 Fig. Pathway enrichment analysis based on KEGG and Reactome sources of all 31 candidaemia genes prioritized based on eQTL, differential expression upon Candida stimulation and proximity to top SNP.

Candidaemia genes showed an expected enrichment for cytokine signalling pathways and showed a strong enrichment for complement and coagulation pathways. Each node represents a separate pathway whose number of genes and P-value are encoded as node size and node colour, respectively. Two nodes are connected by an edge if they share members. The edge width reflects the relative overlap (corresponding to the Fowlkes-Mallows index) between the nodes, while the edge colour encodes the number of shared gene members. (see.tif image)

https://doi.org/10.1371/journal.pone.0180824.s003

(TIF)

S4 Fig.

(A) Heatmap depicts the false discovery rate (FDR) of differentially expressed genes in sepsis as identified by Davenport E et al in their discovery and validation cohort. These genes were differentially expressed in response to Candida stimulation as well. (B) Proteins encoded by 14 candidaemia susceptibility genes detected in platelets using plateletWeb (http://plateletweb.bioapps.biozentrum.uni-wuerzburg.de/plateletweb.php). (see.tiff image)

https://doi.org/10.1371/journal.pone.0180824.s004

(TIF)

S1 Table. Differentially expressed protein-coding genes in response to 4 and 24 hour-Candida stimulation that showed >1.5-fold higher expression compared to RPMI medium used as control.

https://doi.org/10.1371/journal.pone.0180824.s005

(DOCX)

S2 Table. A total of 246 protein-coding genes were differentially expressed at both 4 and 24 hour-Candida stimulation, showing >1.5-fold higher expression compared to RPMI medium used as control.

https://doi.org/10.1371/journal.pone.0180824.s006

(DOCX)

S3 Table. Pathway enrichment analysis on differentially expressed protein-coding genes that showed >1.5 higher expression compared to RPMI medium in response to 4 hour-Candida stimulation.

https://doi.org/10.1371/journal.pone.0180824.s007

(DOCX)

S4 Table. Pathway enrichment analysis on differentially expressed protein-coding genes that showed >1.5 higher expression compared to RPMI medium in response to 24 hour-Candida stimulation.

https://doi.org/10.1371/journal.pone.0180824.s008

(DOCX)

S5 Table. Candidaemia-associated SNPs and variants with r2> = 0.8.

SNPs rs12491812 and rs1802141 were in strong linkage disequilibrium (LD) with synonymous variants located in CISH and SNPS1 genes respectively. (source: Haploreg http://archive.broadinstitute.org/mammals/haploreg/haploreg.php).

https://doi.org/10.1371/journal.pone.0180824.s009

(DOCX)

S6 Table. Differential expression of genes that are located within a 500 kilobase (kb) window around the candidaemia-associated SNPs upon Candida stimulation at 4 hours.

Bolded genes show a log2 fold change > 1.5.

https://doi.org/10.1371/journal.pone.0180824.s010

(DOCX)

S7 Table. Differential expression of genes that are located within a 500 kilobase (kb) window around the candidaemia-associated SNPs upon Candida stimulation at 24 hours.

Bolded genes show a log2 fold change > 1.5.

https://doi.org/10.1371/journal.pone.0180824.s011

(DOCX)

S8 Table. All thirty-one prioritized susceptibility genes for candidaemia showed a strong enrichment for complement and coagulation pathways along with cytokine- and immune- related pathways based on KEGG and Reactome sources.

https://doi.org/10.1371/journal.pone.0180824.s012

(DOCX)

S9 Table. Five additional candidaemia SNPs showed a moderate association with circulating cytokine levels as measured in serum from candidaemia patients.

Cytokine levels were log transformed and statistical significance was tested with Kruskal Wallis test.

https://doi.org/10.1371/journal.pone.0180824.s013

(DOCX)

Acknowledgments

The authors would like to thank Kate Mc Intyre for editing the manuscript.

References

  1. 1. Newport MJ, Finan C. Genome-wide association studies and susceptibility to infectious diseases. Brief. Funct. Genomics. 2011;10:98–107. pmid:21436306
  2. 2. Obel N, Christensen K, Petersen I, Sørensen TIA, Skytthe A. Genetic and environmental influences on risk of death due to infections assessed in Danish twins, 1943–2001. Am. J. Epidemiol. 2010;171:1007–13. pmid:20375195
  3. 3. Wisplinghoff H, Bischoff T, Tallent SM, Seifert H, Wenzel RP, Edmond MB. Nosocomial bloodstream infections in US hospitals: analysis of 24,179 cases from a prospective nationwide surveillance study. Clin. Infect. Dis. 2004;39:309–17. pmid:15306996
  4. 4. Campion EW, Kullberg BJ, Arendrup MC. Invasive Candidiasis. N. Engl. J. Med. 2015;373:1445–56. pmid:26444731
  5. 5. Wisplinghoff H, Seifert H, Wenzel RP, Edmond MB. Inflammatory response and clinical course of adult patients with nosocomial bloodstream infections caused by Candida spp. Clin. Microbiol. Infect. 2006;12:170–7. pmid:16441456
  6. 6. Brown GD, Denning DW, Gow NAR, Levitz SM, Netea MG, White TC. Hidden Killers: Human Fungal Infections. Sci. Transl. Med. 2012;4:1–9.
  7. 7. 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. pmid:22032929
  8. 8. Cortes A, Brown M a. Promise and pitfalls of the Immunochip. Arthritis Res. Ther. 2011;13:101. pmid:21345260
  9. 9. Kumar V, Cheng SC, Johnson MD, Smeekens SP, Wojtowicz A, Giamarellos-Bourboulis E, et al. Immunochip SNP array identifies novel genetic variants conferring susceptibility to candidaemia. Nat. Commun. 2014;5:4675. pmid:25197941
  10. 10. Trynka G, Hunt K a, Bockett N a, Romanos J, Castillejo G, Concha EG De, et al. Dense genotyping identifies and localizes multiple common and rare variant association signals in celiac disease. Nat. Genet. 2012;43:1193–201.
  11. 11. Netea MG, Gow NAR, Munro CA, Bates S, Collins C, Ferwerda G, et al. Immune sensing of Candida albicans requires cooperative recognition of mannans and glucans by lectin and Toll-like receptors. J. Clin. Invest. 2006;116:1642–50. pmid:16710478
  12. 12. Lehrer RI, Cline MJ. Interaction of Candida albicans with human leukocytes and serum. J. Bacteriol. 1969;98:996–1004. pmid:4182532
  13. 13. Li Y, Oosting M, Deepen P, Ricaño-Ponce I, Smeekens S, Jaeger M, et al. Inter-individual variability and genetic influences on cytokine responses to bacteria and fungi. Nat. Med. 2016;22:952–60. pmid:27376574
  14. 14. Kamburov A, Stelzl U, Lehrach H, Herwig R. The ConsensusPathDB interaction database: 2013 Update. Nucleic Acids Res. 2013;41 (Database issue):D793–D800. pmid:23143270
  15. 15. Netea MG, Joosten LA, van der Meer JW, Kullberg BJ, van de Veerdonk FL. Immune defence against Candida fungal infections. Nat Rev Immunol. 2015;15:630–42. pmid:26388329
  16. 16. Gow N a. R, van de Veerdonk FL, Brown AJP, Netea MG. Candida albicans morphogenesis and host defence: discriminating invasion from colonization. Nat. Rev. Microbiol. 2011;10:112–22. pmid:22158429
  17. 17. Smeekens SP, Ng A, Kumar V, Johnson MD, Plantinga TS, van Diemen C, et al. Functional genomics identifies type I interferon pathway as central for host defense against Candida albicans. Nat Commun. 2013;4:1342. pmid:23299892
  18. 18. Jaeger M, van der Lee R, Cheng SC, Johnson MD, Kumar V, Ng A, et al. The RIG-I-like helicase receptor MDA5 (IFIH1) is involved in the host defense against Candida infections. Eur. J. Clin. Microbiol. Infect. Dis. 2015;34:963–74. pmid:25579795
  19. 19. Yanagisawa H, Miyashita T, Nakano Y, Yamamoto D. HSpin1, a transmembrane protein interacting with Bcl-2/Bcl-xL, induces a caspase-independent autophagic cell death. Cell Death Differ. 2003;10:798–807. pmid:12815463
  20. 20. Wang Y, Wang W. CISH and susceptibility to infectious diseases. N. Engl. J. Med. 2010;363:1676; author reply 1676–7.
  21. 21. Hu Z, Yang J, Wu Y, Xiong G, Wang Y, Yang J, et al. Polymorphisms in CISH gene are associated with persistent hepatitis B virus infection in Han Chinese population. PLoS One. 2014;9(6): e100826. pmid:24964072
  22. 22. Tong H V., Toan NL, Song LH, Kremsner PG, Kun JFJ, Velavan TP. Association of CISH-292A/T genetic variant with hepatitis B virus infection. Immunogenetics. 2012;64:261–5. pmid:22033525
  23. 23. Ward LD, Kellis M. HaploReg: A resource for exploring chromatin states, conservation, and regulatory motif alterations within sets of genetically linked variants. Nucleic Acids Res. 2012;40 (Database issue):D930–4. pmid:22064851
  24. 24. Bonder MJ, Luijk R, Zhernakova D V, Moed M, Deelen P, Vermaat M, et al. Disease variants alter transcription factor levels and methylation of their binding sites. Nat Genet. 2017;49:131–138. pmid:27918535
  25. 25. Westra HJ, Peters MJ, Esko T, Yaghootkar H, Schurmann C, Kettunen J, et al. Systematic identification of trans eQTLs as putative drivers of known disease associations. Nat. Genet. 2013;45:1238–43. pmid:24013639
  26. 26. GTEx Consortium. Human genomics. The Genotype-Tissue Expression (GTEx) pilot analysis: multitissue gene regulation in humans. Science. 2015;348:648–60. pmid:25954001
  27. 27. Kaner Z, Ochayon DE, Shahaf G, Baranovski BM, Bahar N, Mizrahi M, et al. Acute phase protein α1-antitrypsin reduces the bacterial burden in mice by selective modulation of innate cell responses. J. Infect. Dis. 2015;211:1489–98. pmid:25389308
  28. 28. Griese M, Latzin P, Kappler M, Weckerle K, Heinzlmaier T, Bernhardt T, et al. alpha1-Antitrypsin inhalation reduces airway inflammation in cystic fibrosis patients. Eur. Respir. J. 2007;29:240–50. pmid:17050563
  29. 29. Patterson H, Nibbs R, Mcinnes I, Siebert S. Protein kinase inhibitors in the treatment of inflammatory and autoimmune diseases. Clin. Exp. Immunol. 2014; 176:1–10. pmid:24313320
  30. 30. Dumitru CD, Ceci JD, Tsatsanis C, Kontoyiannis D, Stamatakis K, Lin JH, et al. TNF-alpha induction by LPS is regulated posttranscriptionally via a Tpl2/ERK-dependent pathway. Cell. 2000;103:1071–83. pmid:11163183
  31. 31. Xiao N, Eidenschenk C, Krebs P, Brandl K, Blasius AL, Xia Y, et al. The Tpl2 mutation Sluggish impairs type I IFN production and increases susceptibility to group B streptococcal disease. J. Immunol. 2009;183:7975–83. pmid:19923465
  32. 32. Mielke L a Elkins KL, Wei L, Starr R, Tsichlis PN, O’Shea JJ, et al. Tumor progression locus 2 (Map3k8) is critical for host defense against Listeria monocytogenes and IL-1 beta production. J. Immunol. 2009;183:7984–93. pmid:19933865
  33. 33. Kuriakose T, Tripp RA, Watford WT. Tumor progression locus 2 promotes induction of IFNλ, interferon stimulated genes and antigen-specific CD8+ T cell responses and protects against influenza virus. PLoS Pathog. 2015;11:1–22.
  34. 34. Zipfel PF, Skerka C. Complement, Candida, and cytokines: The role of C5a in host response to fungi. Eur. J. Immunol. 2012; 42(4):822–5. pmid:22531909
  35. 35. Delvaeye M, Conway EM. Coagulation and innate immune responses: Can we view them separately? Blood. 2009; 114(12):2367–74. pmid:19584396
  36. 36. Esmon CT, Xu J, Lupu F. Innate immunity and coagulation. J. Thromb. Haemost. 2011;9:182–8. pmid:21781254
  37. 37. Esmon CT. The interactions between inflammation and coagulation. Br. J. Haematol. 2005;131(4):417–30. pmid:16281932
  38. 38. Aird WC. Sepsis and coagulation. Crit. Care Clin. 2005;21(3):417–31. pmid:15992665
  39. 39. Khakpour S, Wilhelmsen K, Hellman J. Vascular endothelial cell Toll-like receptor pathways in sepsis. Innate Immun. 2015;21:827–46. pmid:26403174
  40. 40. Opal SM EC. Bench-to-bedside review: functional relationships between coagulation and the innate immune response and their respective roles in the pathogenesis of sepsis. Crit Care. 2003;7(1):23–38. pmid:12617738
  41. 41. Davenport EE, Burnham KL, Radhakrishnan J, Humburg P, Hutton P, Mills TC, et al. Genomic landscape of the individual host response and outcomes in sepsis: A prospective cohort study. Lancet Respir. Med. 2016;4:259–71. pmid:26917434
  42. 42. Speth C, Rambach G, Lass-Flörl C. Platelet immunology in fungal infections. Thromb. Haemost. 2014;112:632–9. pmid:24990293
  43. 43. Stoller JK, Aboussouan LS. A review of a1-antitrypsin deficiency. Am. J. Respir. Crit. Care Med. 2012;185:246–59. pmid:21960536
  44. 44. Lomas DA, Mahadeva R. α1-antitrypsin polymerization and the serpinopathies: Pathobiology and prospects for therapy. J. Clin. Invest. 2002;110:1585–90. pmid:12464660
  45. 45. Kanera Ziv, Ochayona David E., Shahaf Galit, Baranovski Boris M., Bahar Nofar, Mark Mizrahi ECL. Acute Phase Protein α1-Antitrypsin Reduces the Bacterial Burden in Mice by Selective Modulation of Innate Cell Responses. J Infect Dis. 2015;211:1489–98. pmid:25389308
  46. 46. Waterfield MR, Zhang M, Norman LP, Sun SC. NF-kB1/p105 regulates lipopolysaccharide-stimulated MAP kinase signaling by governing the stability and function of the Tpl2 kinase. Mol. Cell. 2003;11:685–94. pmid:12667451
  47. 47. Plantinga TS, Johnson MD, Scott WK, Van De Vosse E, Velez Edwards DR, Smith PB, et al. Toll-like receptor 1 polymorphisms increase susceptibility to candidemia. J. Infect. Dis. 2012;205:934–43. pmid:22301633
  48. 48. George D, Salmeron A. Cot/Tpl-2 protein kinase as a target for the treatment of inflammatory disease. Curr Top Med Chem. 2009;9(7):611–622. pmid:19689369
  49. 49. Kidane YH, Lawrence C, Murali TM. Computational approaches for discovery of common immunomodulators in fungal infections: towards broad-spectrum immunotherapeutic interventions. BMC Microbiol. 2013;13:224. pmid:24099000
  50. 50. Krueger KE, Ghosh AK, Krom BP, Cihlar RL. Deletion of the NOT4 gene impairs hyphal development and pathogenicity in Candida albicans. Microbiology. 2004;150:229–40. pmid:14702416
  51. 51. Warenda AJ, Kauffman S, Sherrill TP, Becker JM, Konopka JB. Candida albicans septin mutants are defective for invasive growth and virulence. Infect. Immun. 2003;71:4045–51. pmid:12819094
  52. 52. Graaf C a a Van Der, Netea MG, Meer , Kullberg BJ, Verschueren I. Differential cytokine production and Toll-like receptor signaling pathways by Candida albicans blastoconidia and hyphae. Infection and immunity. 2005;73:7458–64. pmid:16239547
  53. 53. Lowman DW, Greene RR, Bearden DW, Kruppa MD, Pottier M, Monteiro MA, et al. Novel structural features in Candida albicans hyphal glucan provide a basis for differential innate immune recognition of hyphae versus yeast. J. Biol. Chem. 2014;289(6):3432–43. pmid:24344127
  54. 54. Cheng S-C, van de Veerdonk FL, Lenardon M, et al. The dectin-1/inflammasome pathway is responsible for the induction of protective T-helper 17 responses that discriminate between yeasts and hyphae of Candida albicans. Journal of Leukocyte Biology. 2011;90(2):357–366. pmid:21531876