Abstract
Escherichia coli and Staphylococcus aureus are two of the most common bacterial species responsible for sepsis. While it is observed that they have disparate clinical phenotypes, the signaling differences elicited by each bacteria that drive this variance remain unclear. Therefore, we utilized human whole blood exposed to heat-killed E. coli or S. aureus and measured the transcriptomic signatures. Relative to unstimulated control blood, heat-killed bacteria exposure led to significant dysregulation (up and down-regulated) of >5000 genes for each experimental condition, with a slight increase in gene alterations by S. aureus. While there was significant overlap regarding pro-inflammatory pathways, Gene Ontology over-representation analysis of the most altered genes suggested biological processes like macrophage differentiation and ubiquinone biosynthesis were more unique to heat-killed S. aureus, compared to heat-killed E. coli exposure. Utilizing Ingenuity Pathway Analysis, it was demonstrated that nuclear factor erythroid 2-related factor 2 (Nrf2) signaling, a main transcription factor in anti-oxidant responses, was predominately up-regulated in S. aureus exposed blood relative to E. coli. Further, the use of pharmacologics that preferentially targeted the Nrf2 pathway led to differential cytokine profiles depending on the type of bacterial exposure. These findings reveal significant inflammatory dysregulation between E. coli and S. aureus and provide insight into the targeting of unique pathways to curb bacteria-specific responses.
Introduction:
Sepsis, an exaggerated host response to infection, is a considerable cause of morbidity and mortality worldwide 1. This response is initially propagated by exposures to pathogen-associated molecular patterns (PAMPs) and the release of cytokines that stimulate the innate immune system. This results in a pro-inflammatory state, clinically observed as a systemic inflammatory response syndrome (SIRS) as well as a nearly concurrent compensatory anti-inflammatory response syndrome (CARS) 2. These responses act in concert to recruit the appropriate immune cells to deal with the pathogen and, in turn, resolve the inflammatory process once the pathogen has been successfully contained 3. However, it has become increasingly recognized that the SIRS and CARS responses are not prototypical but instead a heterogenous constellation of immunological and biochemical reactions that are dependent on both the host and the pathogens encountered to create a unique immunological output 4,5.
Interest in these unique immunologic profiles and biologic signatures has grown as the ability to delineate and dissect the pathways involved in the inflammatory response has evolved. Indeed, one of the primary modalities to treat sepsis (i.e., antimicrobials) is tailored based on the type of pathogen encountered 6. Gram-positive and gram-negative bacteria make up an overwhelming majority of the pathogens encountered during sepsis. Staphylococcus aureus is the most common bloodstream pathogen encountered among gram-positive species, while Escherichia coli accounts for the most common gram-negative bloodstream infections 7. While the treatment for either pathogen requires timely antimicrobials, fluids, and possibly other supportive measures, each species provides unique PAMPs that differentially affect the inflammatory response 8. In the case of E. coli, PAMPs often come in the form of bacterial cell wall endotoxins, such as lipopolysaccharide (LPS), that have a well-established relationship with activation of a pathogen recognition receptor (PRR) known as toll-like receptor (TLR) 4. In contrast, S. aureus produces its own PAMPs, typically as enterotoxins or different cell membrane amphiphiles such as lipoteichoic acid (LTA), which stimulate their own PRRs.
Prior studies have attempted to discern unique transcriptomic profiles based on the pathogenic challenge. Stimulation of whole blood utilizing TLR2 and TLR4 specific PAMPs was demonstrated to show unique, temporally-associated transcripts, particularly concerning preferential interferon (IFN) signaling with a maximal transcript output of around 6 hours 9. Another study confirmed enhanced IFN, tumor necrosis factor (TNF), and interleukin (IL)-1 signaling in human whole blood exposed to heat-killed Pseudomonas aeruginosa as opposed to Streptococcus pneumonia 10. Additional studies performed in mouse macrophages showed enhanced, early TNF response in E. coli compared to S. aureus 11. Conversely, no unique expression patterns were demonstrated when transcriptional analysis was performed on patients with gram-negative sepsis compared to gram-positive sepsis 12. Instead, patients suffering from either pathogen type showed similar inflammatory and mitochondrial function pathway alterations. Thus, despite the well-known signaling pathways involved in PAMP and PRR activation, heterogeneity of the inflammatory response remains a challenge.
Given these prior studies and the possible unique host-pathogen interactions, we sought to perform comprehensive, non-targeted RNA sequencing in whole blood exposed to the two most common pathogens encountered in sepsis: E. coli and S. aureus. In addition, we hypothesized that deep sequencing would identify unique signaling pathways between the two bacterial types and offer a pharmacological means to shift the inflammatory response that would be pathogen-specific.
Materials and Methods:
Blood Collection, Bacterial Exposure, and Pharmacological Treatments
Human whole blood collection was approved by the Vanderbilt University Medical Center Institutional Review Board (IRB #202528, Supp. Table 1). Before blood collection, all participants in the study underwent signed informed consent. For RNA analysis, after obtaining consent, blood was drawn via venipuncture from each participant and immediately aliquoted into equal volume EDTA tubes containing 100 μl of either sterile water (control), 3x108 CFU/ml of heat-killed Escherichia coli (HKEC strain 0111:B4, InvivoGen, San Diego, CA, USA) or 3x108 CFU/ml heat-killed Staphylococcus aureus (HKSA, InvivoGen), each resuspended in equal volumes of endotoxin-free, sterile water. Blood was then incubated on an orbital shaker at 80 rpm at 37 °C for 6 hours. Afterward, blood was transferred to PAXgene Blood RNA Tubes (QIAGEN, Hilden, Germany) and stored in the freezer at −80 °C until RNA extraction. For cytokine analysis, blood was collected via venipuncture and then aliquoted into equal volumes in EDTA tubes. Blood was then exposed to a combination of HKEC or HKSA (3x108 CFU/ml) with or without vehicle (<1% DMSO), brusatol (60 ng/ml, Sigma-Aldrich, St. Louis, MO, USA), or dimethyl fumarate (6 μg/ml, Fisher Scientific, Hampton, NH, USA) for 16 hours on an orbital shaker at 80 rpm at 37 °C. In an additional set of experiments, blood was collected by venipuncture and treated with HKEC or HKSA as well as Nrf2 modulators (brusatol and dimethyl fumarate or <1% DMSO control) at the doses and temperatures listed above in the presence of either anti-human TLR4 antibody (5 μg/ml, clone W7C11, InvivoGen) to HKEC-treated blood, anti-human TLR2 antibody (5 μg/ml, clone B4H2, InvivoGen) to HKSA-treated blood or rat IgG2a control (5 μg/ml, clone R35-95, BD Biosciences, Franklin Lakes, NJ, USA) for 16 hours. Lastly, in some preliminary experiments, blood was collected and treated with the respective heat-killed bacteria in the presence or absence of auranofin (1.3 μg/ml, Cayman Chemical, Ann Arbor, MI, USA) for 6 hours. Afterward, plasma was isolated by centrifugation of whole blood specimens at 1500 xg for 10 minutes and stored at −80 °C until cytokine analysis.
RNA Isolation and Sequencing
Total RNA was extracted from the aforementioned whole blood samples using a PAXgene Blood RNA Kit (QIAGEN) as per the manufacturer’s instructions. RNA integrity was evaluated on an Agilent 2100 Bioanalyzer (Agilent, Santa Clara, CA, USA; RIN score ranging 7.7-9.3) and further treated for globin and rRNA depletion by NEB globin/ribo reduction kit performed at Vanderbilt Technologies for Advanced Genomics (VANTAGE). 150 basepair Paired-End (PE) sequencing was performed on the Illumina NovaSeq 6000 platform (Illumina, San Diego, CA, USA) to generate approximately 50 million reads per sample.
RNA-Sequencing Bioinformatic Analysis
RNA sequencing reads were adapter-trimmed and quality-filtered using Trimgalore v0.5 13. An alignment reference was generated from the GRCh38 human genome and GENCODE comprehensive gene annotations (Release 26), to which trimmed reads were aligned and counted using Spliced Transcripts Alignment to a Reference (STAR) v2.7.9a 14 with the –quantMode GeneCounts parameter. On average, 46 million uniquely mapped reads were acquired per sample. DESeq2 package v1.34.0 15 was used to perform sample-level quality control, low-count filtering, normalization, and downstream differential expression analysis. Genomic features counted fewer than five times across at least three samples were removed. False discovery rate adjusted for multiple hypothesis testing with Benjamini-Hochberg (BH) procedure p value < 0.05 and Log2 FoldChange >1 was used to define differentially expressed genes. Sex as a variable of patients was used as the batch factor in the DESeq2 design to increase the sensitivity of differential expression. Gene enrichment analysis was implemented by gene set over-representation analysis using Gene Ontology (GO) knowledge base 16 applied using the GeneTonic v1.6.4 package in R 17. Annotated gene sets GO were sourced from genome-wide annotations for “Human” 18. For GO, genes significantly up- or down-regulated in the tested conditions were used as input. Lastly, canonical pathway analysis of the differentially expressed genes was analyzed with Ingenuity Pathway Analysis (IPA, QIAGEN 19).
Cytokine Analysis
Cytokine analysis was performed on the extracted plasma previously mentioned. All cytokine examinations were completed using an enzyme-linked immunosorbent assay (ELISA). Each cytokine analyzed used a corresponding Human DuoSet ELISA Kits (R&D Systems, Minneapolis, MN, USA) following the manufacturer’s instructions. Plasma samples were subjected to a series of dilutions in 5% BSA to determine the optimal concentration for each analyte. ELISAs were analyzed using a FLUOstar Omega microplate reader via blank correction at an absorbance of 450nm (BMG Labtech, Cary, NC, USA).
Statistical Analysis
Global alterations in cytokine profiles were assessed using a Friedman ANOVA variance of rank and individual comparisons were performed using paired Wilcoxon Ranked Sum tests. Analysis was performed on GraphPad Prism v.9.0 (San Diego, CA, USA). Statistical significance was assigned based on a p value < 0.05.
Results:
Global Expression Differences in Human Blood Gene Transcripts Following E. coli or S. aureus Exposure
To understand how human whole blood exposure to heat-killed E. coli (HKEC) or heat-killed S. aureus (HKSA) differentially regulates gene expression, we performed a comparison between the transcriptomic profiles in whole blood exposed to the respective pathogens (3x108 CFU/ml) for 6 hours versus blood exposed to unstimulated, control conditions (sterile water) over the same time frame (Figures 1A, B). We performed differential gene expression between blood exposed to HKEC and HKSA versus the unstimulated controls. At a false discovery rate (FDR) <0.05, a significant number of differentially expressed genes (DEGs) was observed in blood treated with HKEC (5388 genes) and HKSA (6112 genes) (Figures 1C, SF1). Of these differentially expressed genes, 53% were common between the response to HKEC and HKSA DEGs suggesting nearly half of the blood transcriptomic alterations were shared by either pathogen exposure. Conversely, 28% of DEGs were unique to HKSA-treated blood and 19% of DEGs were unique to blood treated with HKEC, indicating unique responses elicited by each pathogen (Figure 1C). DEGs among up-regulated genes and down-regulated genes also showed 48% and 58% overlap, respectively, whereas the remaining DEGs were unique in the response to either pathogen.
Figure 1:
(A) Overview of the experimental design for RNA-sequencing analysis in whole blood exposed to heat-killed E. coli (HKEC, 3x108 CFU/ml) or heat-killed S. aureus (HKSA, 3x108 CFU/ml) relative to control (untreated) blood among paired, healthy people (n=4 individuals). (B) Principal component analysis (PCA) showing clustering of samples by exposure to sterile water (Control), HKEC, and HKSA. (C) Venn diagrams showing overlap among differentially expressed genes (DEGs) between two pairwise conditions, HKEC vs Control and HKSA vs Control, among all DEGs as well as significantly upregulated genes and significantly downregulated genes. (D) Gene Ontology (Biological Processes) enrichment associated to DEGs is shown in Fig1C for both pairwise conditions.
We performed gene set over-representation analysis using Gene Ontology knowledge base 16 on the DEGs passing FDR <0.05 with a Log2 FoldChange (Log2FC) >2. For added “direction” of change for each biological process enriched, we computed z-scores utilized by GeneTonic R package 17. The z-score represents the standardized sum of the number of genes regulated in either direction indicating a positive z-score as an increased number of up-regulated genes and a negative z-score as an increased number of down-regulated genes. Using the top 20 enriched gene ontologies in HKEC and HKSA-exposed blood, we found that genes associated with neutrophil activation were mostly down-regulated and genes associated with leukocyte response, cytokine production, and lipopolysaccharide response were up-regulated in both HKEC and HKSA-treated samples compared to controls (Figures 1D, SF2).
Differential Regulation of Inflammatory Pathways by E. coli versus S. aureus
To evaluate the association in the whole blood transcriptomic response to HKEC and HKSA relative to controls, we performed a linear regression analysis to model the relationship between HKEC and HKSA-treated datasets by correlating expression changes (Log2FC values) of the DEGs from each differential gene expression analysis. A strong association, measured as an R-squared value of 0.911, confirms the similarities between HKSA and HKEC-treated analyses (Figure 2A). To investigate the dissimilarity between the DEGs of each dataset, we assessed for genes with positive residuals versus negative residuals. These were determined by evaluating genes that had a Log2FC in the HKSA vs Control (Ctrl) dataset greater than Log2FC in HKEC vs Control dataset, while negative residuals indicated genes that had a Log2FC in the HKSA vs Control data less than Log2FC in HKEC vs Control data. Gene Ontology over-representation analysis of the positive residual genes suggested biological processes like macrophage differentiation and ubiquinone biosynthesis were more unique in blood treated with HKSA compared to HKEC (Figure SF3). Gene Ontology over-representation analysis of the negative residual genes suggested TLR4 signaling pathways were more unique to HKEC-exposed blood compared to HKSA (Figure SF3). In particular, LINC00595, a long non-coding RNA with a previously undefined significance, was up-regulated in the HKSA-exposed blood (Figures SF1, SF3). In contrast, angiopoietin-like protein 4 (ANGPTL4), a serum hormone involved in glucose homeostasis and lipid metabolism, was up-regulated after HKEC exposure (Figures SF1, SF3).
Figure 2:
(A) Linear regression analysis showing the correlation between the two pairwise conditions, HKEC vs Control and HKSA vs Control, based on the fold changes measure of differential expression analysis within each pairwise condition. Color is by residual value range. For this plot, only Log2FC of genes that had an adjusted p value <0.05 in the differential expression test by DESeq2 are shown. (B) Ingenuity Pathway Analysis (IPA) showing the top 10 pathways enriched in HKEC compared to HKSA sorted by adjusted p value significance. Blue color represents more activated in HKSA-exposed blood, yellow indicates more activated by HKEC exposure. (C) Heatmap showing variable expression of Nrf2-mediated oxidative stress response pathway-associated genes. (D) Nrf2-mediated oxidative stress response pathway schematic generated by IPA.
Differential Impacts of Nrf2-Mediated Signaling During Exposure to S. aureus Compared to E. coli
Given the differences in multiple genes between the two pathogens, a direct differential expression analysis in the response to blood exposed to HKEC and HKSA was performed, followed by Ingenuity Pathway Analysis (IPA) to evaluate divergent canonical pathways stimulated. As a result, while a majority of pro-inflammatory pathways were more up-regulated during HKEC exposure, nuclear factor erythroid 2-related factor 2 (Nrf2) oxidative stress response, and to a lesser degree liver and retinoid X receptor (LXR/RXR) activation, were more strongly activated by HKSA (Figure 2B, SF4). Furthermore, alterations in HMOX1, SOD, TXN, and TXNRD1 as part of the anti-oxidant response element (ARE), and NQO, GCLM, and GCLC involved in NADPH and glutathione generation, had a predicted increase in expression during HKSA exposure compared to HKEC (Figures 2C, D).
To further assess what impact the modulation of Nrf2 would have on inflammatory outputs, whole blood was collected and subjected to the same CFUs of HKSA or HKEC in the presence or absence of brusatol (60 ng/ml), an Nrf2 inhibitor, or dimethyl fumarate (DMF, 6 μg/ml), a Keap1 inhibitor 20,21. DMF was selected as there are no known direct activators of Nrf2, but Keap1 is a negative regulator of Nrf2, and inhibition of Keap1 has been shown to promote Nrf2 activity 22. In addition, given the known differential cytokine kinetics between HKEC and HKSA (Figure SF5) 9,11, 16 hours was chosen to make the global cytokine production between the organisms more comparable. As shown in Figure 3, Nrf2 inhibition with brusatol significantly impaired IL-8 and IL-1β production in both HKEC and HKSA-exposed blood. In comparison, brusatol potentiated TNFα and IL-6 production in blood exposed to HKEC but did not affect those cytokines in blood exposed to HKSA. On the other hand, DMF globally had no impact on HKSA-mediated cytokine production compared to vehicle-control treated blood. However, DMF significantly enhanced IL-8 and IL-6 production in HKEC-exposed blood, suggesting that enhanced Nrf2 activation was possible after HKEC, but not after HKSA where activation was already higher. While the addition of HKEC or HKSA significantly reduced S100A8, consistent with the prior finding of negative regulation of neutrophils by both pathogens from the RNA sequencing data and likely a reflection of neutrophil cell death, the other cytokines and chemokines tested (IL-10, IFN-γ, TIMP-1, Syndecan-1, sVCAM) were not altered by the bacterial type nor by pharmacological manipulation of Nrf2 at the timepoint tested.
Figure 3:
Plasma cytokine profiles from individuals (n=6) whose whole blood was exposed to either heat-killed E. coli (HKEC) or heat-killed S. aureus (HKSA) in the presence of either brusatol (Nrf2 inhibitor) or dimethyl fumarate (Keap1 inhibitor) for 16 hours. Data is expressed as median with intraquartile range. Colored data points are for each respective individual across all treatments and measurements. * indicates a p value < 0.05 between designated groups and † indicates a p value < 0.05 between HKEC versus HKSA for the respective condition.
To further assess the impact of TLRs on Nrf2-mediated bacterial cytokine production, we applied Nrf2 modulators with the addition of anti-human TLR4 and TLR2 antibodies in the presence of HKEC or HKSA, respectively (Figure SF6). The application of anti-TLR4 antibodies to HKEC-exposed blood and anti-TLR2 antibodies to HKSA-exposed blood caused variable alterations in cytokine production compared to IgG control, though the application of anti-TLR4 antibodies caused a significant reduction related to IL-8. In the presence of anti-TLR antibodies to their respective pathogens, the addition of Nrf2 modulators (brusatol or dimethyl fumarate) added further variability across the bacteria-exposed blood samples. However, there were similarities to the prior findings, particularly related to reduced IL-8 production after brusatol exposure and an increase in IL-8 after dimethyl fumarate relative to Nrf2 inhibition. This was also noted with IL-10 production though not as robustly after HKEC exposure. On the other hand, IL-6 production was reduced by both brusatol and dimethyl fumarate in the setting of anti-TLR2 antibodies which was not previously seen. Despite the variability across the multiple treatments, the patterns of cytokine production were more similar than not to the prior finding in the absence of anti-TLR antibodies, suggesting that at least partially, the impact of Nrf2 modulation was TLR-independent.
Discussion:
The ability to examine the myriad of host responses to severe infections, known as sepsis, has provided researchers and clinicians with a deeper understanding of the unique inflammatory signatures generated when the immune system is confronted with a pathogenic challenge. These signatures, called endotypes, categorize patients into specific subclasses and provide insight into the body’s innate abilities to combat the pathogens but also allow for investigations into when the response may be aberrant or perturbed 23,24. The hope is that through the generation of endotypes, patients could be compared within a more homogenous group as opposed to the heterogeneous population in sepsis. Certainly, the genetics of the host play a role in sepsis outcomes, specifically with polymorphisms in immune cells and metabolism 25. However, sepsis is an acquired disease and other variables contribute to the presentation, clinical course, and eventual outcome. Included in those variables are the type of infection and its primary source 20,26,27. Thus, to assess the contribution of this variable to the genetic host response, we performed whole transcriptomic sequencing in paired human samples of blood exposed to either heat-killed Escherichia coli (HKEC) or Staphylococcus aureus (HKSA), two of the most common pathogens in sepsis and sepsis-related mortalities 20,28.
Global inflammatory changes in gene transcripts were observed with both pathogens, with HKSA inducing slightly more changes compared to HKEC at the time point of 6 hours of exposure. The genes affected were mainly associated with cytokine-mediated signaling pathways and cellular responses to bacterial components, including lipopolysaccharides. Exposure of whole blood to HKEC or HKSA induced many commonalities in gene expression changes but transcripts also exhibited unique activation patterns. Pathway analysis revealed that while pro-inflammatory pathways were generally up-regulated during HKEC exposure, HKSA induced the Nrf2 oxidative stress response pathway more readily. When modulating Nrf2 by dimethyl fumarate (activator) or brusatol (inhibitor), the cytokine expression profiles became more altered after HKEC exposure compared to HKSA. Particularly, impaired IL-8 and IL-1β production was observed in both pathogen exposures, but potentiated TNFα and IL-6 production was seen in only HKEC-exposed blood. These data highlight that different bacterial species potentiate unique transcriptomic profiles in humans. Further, differentially regulated pathways between the bacterial species may not be entirely amendable by pharmacological intervention despite a known pathway difference.
A consistent inflammatory response to pathogenic challenge is evolutionarily preserved across species in genetic pattern recognition receptors that trigger cytokine and interferon production 29. The multifaceted, complex, and often redundant signaling pathways, must work in concert to mitigate the impact of the pathogen on the host 30,31. Utilizing global human blood transcriptomic profiling, we found that the response to HKEC and HKSA demonstrated a similar influence on cytokine-mediated pro-inflammatory signaling pathways with a majority of signaling pathways and cascades altered in the same direction. Processes related to ribosomes and transcription were both up-regulated, indicating an overall increase in protein synthesis and gene expression as part of the immune response. In addition, there was a significant overlap in cytokine-mediated signaling pathways, including the pro-inflammatory IL-6 and CSF3, but also activation of pathways associated with anti-inflammatory type I interferon signaling. Similarities have been observed in other studies examining the host response to different pathogens or TLR ligands 9,10. These data demonstrating similarities also likely explain prior studies examining critically ill patients with sepsis where there was no discernable difference in RNA expression between those patients with a gram-positive organism versus those with a gram-negative 12,32. While the immune regulatory changes were globally similar between HKEC and HKSA-exposed blood, unique signatures did exist between the pathogen exposures.
Despite similarities in the initiation of inflammation, it is clear that not all pathogens, nor the host response to them, are created equally. In particular, animal studies have demonstrated that mice challenged with S. aureus versus E. coli demonstrate subtle differences in cytokine and gene expression profiles 11,33,34. Likewise, several studies utilizing either TLR ligands or heat-killed pathogens, have shown differences in transcriptomic expression profiles 9,10. Our data would undoubtedly agree with those prior studies. When comparing the two pathogens against each other with at least a Log2 FoldChange in gene expression, HKEC exposure differentially up-regulated 386 unique genes and down-regulated 396 genes compared to HKSA. Utilizing Ingenuity Pathway Analysis, we found in particular that HKSA exposure more strongly activated HMOX1, SOD, TXN, TXNRD1, NQO, GCLM, and GCLC. These genes comprise part of the anti-oxidant response element (ARE) that are transcribed by the nuclear factor erythroid 2-related factor 2 (Nrf2), an essential promotor of anti-oxidant defense. This suggested that Nrf2 may be an important discriminator between the host response to HKSA versus HKEC.
Nuclear factor erythroid 2-related factor 2 is a transcription factor that plays a crucial role in cellular defense against oxidative stress 35,36. In an uninflamed state, Nrf2 is kept contained by the oxidant-sensor Kelch-like ECH-Associated protein 1 (Keap1) 22. Upon its release from Keap1, it moves to the nucleus, leading to the transcription of ARE-related genes for redox defense. In this capacity, modulating Nrf2 has been shown to improve outcomes in animal models of acute lung injury and burns, for which infections are a significant contributing factor to those pathologies 37,38. Given the differential impact of HKEC and HKSA on whole blood Nrf2-ARE transcripts via IPA analysis, we sought to assess if modulation of this pathway with known inhibitors would make HKSA and HKEC immune-mediated cytokine output more similar to one another. For this purpose, whole blood was exposed to HKSA or HKEC in the presence or absence of two well-studied compounds: brusatol, an Nrf2 inhibitor, and dimethyl fumarate (DMF), a Keap1 inhibitor (Keap1 being a negative regulator of Nrf2)39. Initially, we postulated that Nrf2 enhancement via DMF would make the inflammatory cytokine output to HKEC more like HKSA, while inhibition via brusatol would have the opposite effect. However, these global trends were not observed. Instead, the results demonstrated that the inhibition of Nrf2 with brusatol significantly reduced IL-8 and IL-1β production in blood exposed to both HKEC and HKSA, suggesting that the regulation of these cytokines are at least partially under the control of Nrf2 activation independent of the pathogen species. IL-8 and IL-1β are essential cytokines in the immune response, particularly in inflammation. Interestingly, the impact of Nrf2 modulation on these cytokines has been confirmed in other studies suggesting that in some instances, activation of Nrf2 can have a pro-inflammatory effect through inflammasome activation 40-42. On the other hand, Nrf2 activation via DMF did not alter the cytokine profiles in HKSA-exposed blood. Instead, it was observed that in HKEC-exposed blood, brusatol potentiated TNFα and IL-6 production. Additionally, DMF significantly enhanced the production of IL-6 and IL-8, the latter of which is the opposite finding compared to the HKSA-exposed blood. Thus, the question remains as to why Nrf2 would be more upregulated in HKSA-exposed blood compared to that exposed to HKEC. One postulated reason could be that S. aureus is a catalase-positive bacteria. This enzyme is an anti-oxidant enzyme that converts hydrogen peroxide to water and oxygen and is directly related to Staphylococcal species virulence43. In addition, deficiency in NADPH oxidase 2 (Nox2) is associated with chronic granulomatous disease. In this setting, the absence of Nox2 prevents the production of the free radical superoxide, leading to an increased risk of S. aureus infections44. Thus, the increased Nrf2 pathway activation seen in the blood exposed to HKSA may be a mechanism through which S. aureus can manipulate the oxidant defense systems of the immune system to promote pathogen survival. While we did not test this theory directly, it could be postulated that the addition of a pharmacological agent to block Nrf2 activation during S. aureus bacteremia to create a pro-oxidant environment may allow for better bacterial killing and reduced severity.
Beyond cytokine profiles, the modulation of Nrf2 to inflammatory patterns had an additional critical finding. Despite controlled blood exposure to bacteria, pharmacologic agents, and TLR-directed antibodies, the data suggest that the application of targeted therapy can not necessarily overcome inherent, baseline differences within the hosts. Part of this can be recognized in the inability of anti-TLR antibodies to produce consistent effects in modulating cytokine outputs among a diverse group of individuals, particularly for antibodies directed at TLR2, where deficiency does not prevent Staphylococcus aureus-mediated TNFα production or pathogen phagocytosis 45,46. Further difficulties were demonstrated in the matched pairs of individuals exposed to Nrf2 modulators, wherein four of the individuals, Nrf2 inhibition potentiated IL-6 production to HKSA, similar to HKEC, but in the other two individuals, inhibition reduced IL-6. One potential explanation for this variability could be the host colonization of S. aureus. While we did not assess for colonization, S. aureus colonizes approximately 1 in every 3 people, providing unique host-pathogen interactions in those individuals 47. This variability in humans, either inherent or acquired, makes it challenging to assess what unforeseen factors can influence the immune system. For a majority of patients who survive sepsis through proactive interventions and timely antibiotics, this variability may not be of clinical significance. Alternatively, for those patients for whom timely interventions prove to be insufficient or for whom the course becomes long and complex eventually leading to a patient’s demise, understanding this heterogeneity from a host-pathogen standpoint may prove crucial. It is this heterogeneity, however, that will likely continue to provide hurdles in applying precision therapeutics to alleviate pathogen-related disease burden, but it remains a pursuit of significant value if we can impact the individual lives of the sickest of patients 48.
In summary, utilizing whole blood transcriptomic responses to pathogenic challenge by two of the most common bacteria encountered in sepsis, E. coli and S. aureus, we found commonalities in their stimulation of pro- and anti-inflammatory pathways that are temporally dependent. We also found they induced unique transcriptomic profiles in whole blood, particularly around their modulation of Nrf2, an important anti-oxidant pathway. When applying targeted therapeutics to Nrf2 in the setting of pathogenic challenge, inhibition or activation of Nrf2 altered HKEC-treated blood more than HKSA-treated blood, but inhibition of Nrf2 significantly reduced the production of the inflammatory cytokines irrespective of the pathogen type or the presence of TLR-directed neutralizing antibodies. These data suggest that blockade or activation of Nrf2 may differentially impact the inflammatory profiles depending on the type of bacteria and further, that the effects of the modulation are dependent not only on the pathogen but the host as well. Thus, when considering precision medicine in severe infections such as sepsis, a deep understanding of the heterogeneity of host-pathogen interactions will be crucial if there is ever to be successful, personalized implementation of therapeutics leading to a desired outcome.
Supplementary Material
Supp. Table 1: Demographic data (age, gender, and race/ethnicity) of all participants at any time point.
Supp. Figure 1: Heatmaps of top differentially expressed genes sorted by adjusted p value significance computed from the DESeq2 differential expression test. Volcano plots displaying genes differentially expressed between control and HKEC/HKSA samples. Differential expression is defined by a Log2 FoldChange >1 and an adjusted p value <0.05.
Supp. Figure 2: Dot plots of the top 25 positive z-score and negative z-score gene ontology (biological processes) sorted by adjusted p value. Gene Ontology analysis performed by GeneTonic R package 17. Z-score represents the standardized sum of the number of genes regulated in either direction indicating a positive z-score as an increased number of upregulated genes and a negative z-score as an increased number of downregulated genes.
Supp. Figure 3: List of genes with positive and negative residuals from linear regression analysis (Figure 2A). Gene Ontology overrepresentation analysis of positive and negative residual genes are as analyzed by EnrichR 49.
Supp. Figure 4: Overview of the major biological themes performed by Ingenuity Pathway Analysis (IPA) Core Analysis that includes canonical pathways, upstream regulators, diseases, and biological functions predicted based on findings in the QIAGEN’s IPA knowledge base. Input DEGs for IPA Core Analysis were DEGs obtained from a direct differential expression analysis between HKEC and HKSA-treated samples.
Supp. Figure 5: Plasma cytokine profiles from individuals (n=4) whose whole blood was exposed to either heat-killed E. coli (HKEC) or heat-killed S. aureus (HKSA) in the presence or absence auranofin (thioredoxin reductase 1 inhibitor (TXRN1), 1.3 μg/ml) for 6 hours. Data is expressed as median with intraquartile range. Colored data points are for each respective individual across all treatments and measurements.
Supp. Figure 6: Plasma cytokine profiles from individuals (n=6) whose whole blood was exposed to either heat-killed E. coli (HKEC) with anti-human TLR4 antibodies (5 μg/ml) or heat-killed S. aureus (HKSA) with anti-human TLR2 antibodies (5 μg/ml) in the presence of either brusatol (Nrf2 inhibitor, 60 ng/ml) or dimethyl fumarate (Keap1 inhibitor, 6 μg/ml) for 16 hours. Data is expressed as median with intraquartile range. Colored data points are for each respective individual across all treatments and measurements. * indicates a p value < 0.05 between designated groups.
Acknowledgements:
We thank all the volunteers who consented to blood draws for the completion of this study.
Funding:
RJS was supported by a grant from the National Institutes of Health (R35GM138191).
Data Availability
RNA sequencing data is publicly available in the Gene Expression Omnibus (GEO) as dataset GSE237960 as required by the National Institutes of Health Data Sharing Policy.
References:
- 1.Rudd KE, Johnson SC, Agesa KM, et al. Global, regional, and national sepsis incidence and mortality, 1990-2017: analysis for the Global Burden of Disease Study. Lancet. 2020;395(10219):200–211. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Thompson KB, Krispinsky LT, Stark RJ. Late immune consequences of combat trauma: a review of trauma-related immune dysfunction and potential therapies. Mil Med Res. 2019;6(1):11. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Efron PA, Mohr AM, Bihorac A, et al. Persistent inflammation, immunosuppression, and catabolism and the development of chronic critical illness after surgery. Surgery. 2018;164(2):178–184. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Vincent JL, van der Poll T, Marshall JC. The End of "One Size Fits All" Sepsis Therapies: Toward an Individualized Approach. Biomedicines. 2022;10(9). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Stark R. Protein-mediated interactions in the dynamic regulation of acute inflammation. Biocell. 2023;47(6):1191–1198. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Niederman MS, Baron RM, Bouadma L, et al. Initial antimicrobial management of sepsis. Crit Care. 2021;25(1):307. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Kern WV, Rieg S. Burden of bacterial bloodstream infection-a brief update on epidemiology and significance of multidrug-resistant pathogens. Clin Microbiol Infect. 2020;26(2):151–157. [DOI] [PubMed] [Google Scholar]
- 8.Ramachandran G. Gram-positive and gram-negative bacterial toxins in sepsis: a brief review. Virulence. 2014;5(1):213–218. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Blankley S, Graham CM, Howes A, et al. Identification of the key differential transcriptional responses of human whole blood following TLR2 or TLR4 ligation in-vitro. PLoS One. 2014;9(5):e97702. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Le KTT, Chu X, Jaeger M, et al. Leukocyte-Released Mediators in Response to Both Bacterial and Fungal Infections Trigger IFN Pathways, Independent of IL-1 and TNF-alpha, in Endothelial Cells. Front Immunol. 2019;10:2508. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Cui W, Morrison DC, Silverstein R. Differential tumor necrosis factor alpha expression and release from peritoneal mouse macrophages in vitro in response to proliferating gram-positive versus gram-negative bacteria. Infect Immun. 2000;68(8):4422–4429. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Tang BM, McLean AS, Dawes IW, Huang SJ, Cowley MJ, Lin RC. Gene-expression profiling of gram-positive and gram-negative sepsis in critically ill patients. Crit Care Med. 2008;36(4):1125–1128. [DOI] [PubMed] [Google Scholar]
- 13.Krueger FJ F; Ewels P; Afyounian E; Weinstein M; Schuster-Boeckler B; Hulselmans G; Sclamons. FelixKrueger/TrimGalore: v0.6.10 - add default decompression path (0.6.10). https://zenodo.org/record/7598955. Published 2023. Updated Feburary 2023. AccessedJuly 11 2023. [Google Scholar]
- 14.Dobin A, Davis CA, Schlesinger F, et al. STAR: ultrafast universal RNA-seq aligner. Bioinformatics. 2013;29(1):15–21. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Love MI, Huber W, Anders S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 2014;15(12):550. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Ashburner M, Ball CA, Blake JA, et al. Gene ontology: tool for the unification of biology. The Gene Ontology Consortium. Nat Genet. 2000;25(1):25–29. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Marini F, Ludt A, Linke J, Strauch K. GeneTonic: an R/Bioconductor package for streamlining the interpretation of RNA-seq data. BMC Bioinformatics. 2021;22(1):610. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Carlson M. org.Hs.eg.db: Genome wide annotation for Human. R package version 3.10.0. https://bioconductor.org/packages/release/data/annotation/html/org.Hs.eg.db.html. Published 2019. Accessed July 11 2023. [Google Scholar]
- 19.Kramer A, Green J, Pollard J Jr., Tugendreich S. Causal analysis approaches in Ingenuity Pathway Analysis. Bioinformatics. 2014;30(4):523–530. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Vincent JL, Sakr Y, Sprung CL, et al. Sepsis in European intensive care units: results of the SOAP study. Crit Care Med. 2006;34(2):344–353. [DOI] [PubMed] [Google Scholar]
- 21.Olayanju A, Copple IM, Bryan HK, et al. Brusatol provokes a rapid and transient inhibition of Nrf2 signaling and sensitizes mammalian cells to chemical toxicity-implications for therapeutic targeting of Nrf2. Free Radic Biol Med. 2015;78:202–212. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Lee S, Hu L. Nrf2 activation through the inhibition of Keap1-Nrf2 protein-protein interaction. Med Chem Res. 2020;29(5):846–867. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Leligdowicz A, Matthay MA. Heterogeneity in sepsis: new biological evidence with clinical applications. Crit Care. 2019;23(1):80. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Tian Y, Wang L, Chen W, Zhong W, Hu Y. Screening of Potential Core Genes in the Peripheral Blood of Adult Patients with Sepsis Based on Immunoregulation and Signal Transduction Functions. Shock. 2023;59(5):708–715. [DOI] [PubMed] [Google Scholar]
- 25.Engoren M, Jewell ES, Douville N, Moser S, Maile MD, Bauer ME. Genetic variants associated with sepsis. PLoS One. 2022;17(3):e0265052. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Leligdowicz A, Dodek PM, Norena M, et al. Association between source of infection and hospital mortality in patients who have septic shock. Am J Respir Crit Care Med. 2014;189(10):1204–1213. [DOI] [PubMed] [Google Scholar]
- 27.Shald EA, Erdman MJ, Ferreira JA. Impact of Clinical Sepsis Phenotypes on Mortality and Fluid Status in Critically Ill Patients. Shock. 2022;57(1):57–62. [DOI] [PubMed] [Google Scholar]
- 28.Patterson SG, Lamb CK, Gong W, et al. Pediatric Persistent Inflammation, Immunosuppression, and Catabolism Syndrome Prevalence in Sepsis-Related Mortalities: A 23-Year Institutional History. Chest. 2023. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Jenner RG, Young RA. Insights into host responses against pathogens from transcriptional profiling. Nat Rev Microbiol. 2005;3(4):281–294. [DOI] [PubMed] [Google Scholar]
- 30.Hotamisligil GS. Foundations of Immunometabolism and Implications for Metabolic Health and Disease. Immunity. 2017;47(3):406–420. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Taguchi T, Mukai K. Innate immunity signalling and membrane trafficking. Curr Opin Cell Biol. 2019;59:1–7. [DOI] [PubMed] [Google Scholar]
- 32.Tang BM, McLean AS, Dawes IW, Huang SJ, Lin RC. Gene-expression profiling of peripheral blood mononuclear cells in sepsis. Crit Care Med. 2009;37(3):882–888. [DOI] [PubMed] [Google Scholar]
- 33.Duan J, Xie Y, Yang J, Luo Y, Guo Y, Wang C. Variation of Circulating Inflammatory Mediators in Staphylococcus aureus and Escherichia coli Bloodstream Infection. Med Sci Monit. 2016;22:161–171. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Yu SL, Chen HW, Yang PC, et al. Differential gene expression in gram-negative and gram-positive sepsis. Am J Respir Crit Care Med. 2004;169(10):1135–1143. [DOI] [PubMed] [Google Scholar]
- 35.Li Y, Yang M, Xie L, Zhang G, Xu J, Xu S. Sulforaphane Alleviates Postresuscitation Lung Pyroptosis Possibly Via Activating the Nrf2/Ho-1 Pathway. Shock. 2023;60(3):427–433. [DOI] [PubMed] [Google Scholar]
- 36.Teng Y, Li N, Wang Y, et al. NRF2 Inhibits Cardiomyocyte Pyroptosis Via Regulating CTRP1 in Sepsis-Induced Myocardial Injury. Shock. 2022;57(4):590–599. [DOI] [PubMed] [Google Scholar]
- 37.Seim RF, Mac M, Sjeklocha LM, et al. Nuclear Factor-Erythroid-2-Related Factor Regulates Systemic and Pulmonary Barrier Function and Immune Programming after Burn and Inhalation Injury. Shock. 2023;59(2):300–310. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Audi SH, Jacobs ER, Taheri P, Ganesh S, Clough AV. Assessment of Protection Offered By the NRF2 Pathway Against Hyperoxia-Induced Acute Lung Injury in NRF2 Knockout Rats. Shock. 2022;57(2):274–280. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Robledinos-Anton N, Fernandez-Gines R, Manda G, Cuadrado A. Activators and Inhibitors of NRF2: A Review of Their Potential for Clinical Development. Oxid Med Cell Longev. 2019;2019:9372182. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Zhang X, Chen X, Song H, Chen HZ, Rovin BH. Activation of the Nrf2/antioxidant response pathway increases IL-8 expression. Eur J Immunol. 2005;35(11):3258–3267. [DOI] [PubMed] [Google Scholar]
- 41.Zhou J, Wang T, Dou Y, et al. Brusatol ameliorates 2, 4, 6-trinitrobenzenesulfonic acid-induced experimental colitis in rats: Involvement of NF-kappaB pathway and NLRP3 inflammasome. Int Immunopharmacol. 2018;64:264–274. [DOI] [PubMed] [Google Scholar]
- 42.Kuang F, Wang B, You T, et al. Circ_0001818 Targets Mir-136-5p to Increase Lipopolysaccharide-Induced Hk2 Cell Injuries by Activating Txnip/Nlrp3 Inflammasome Pathway. Shock. 2023;60(1):110–120. [DOI] [PubMed] [Google Scholar]
- 43.Kanafani H, Martin SE. Catalase and superoxide dismutase activities in virulent and nonvirulent Staphylococcus aureus isolates. J Clin Microbiol. 1985;21(4):607–610. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Buvelot H, Posfay-Barbe KM, Linder P, Schrenzel J, Krause KH. Staphylococcus aureus, phagocyte NADPH oxidase and chronic granulomatous disease. FEMS Microbiol Rev. 2017;41(2):139–157. [DOI] [PubMed] [Google Scholar]
- 45.Watanabe I, Ichiki M, Shiratsuchi A, Nakanishi Y. TLR2-mediated survival of Staphylococcus aureus in macrophages: a novel bacterial strategy against host innate immunity. J Immunol. 2007;178(8):4917–4925. [DOI] [PubMed] [Google Scholar]
- 46.Yimin, Kohanawa M, Zhao S, et al. Contribution of toll-like receptor 2 to the innate response against Staphylococcus aureus infection in mice. PLoS One. 2013;8(9):e74287. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47.Byrd AL, Belkaid Y, Segre JA. The human skin microbiome. Nat Rev Microbiol. 2018;16(3):143–155. [DOI] [PubMed] [Google Scholar]
- 48.Shah FA, Meyer NJ, Angus DC, et al. A Research Agenda for Precision Medicine in Sepsis and Acute Respiratory Distress Syndrome: An Official American Thoracic Society Research Statement. Am J Respir Crit Care Med. 2021;204(8):891–901. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 49.Chen EY, Tan CM, Kou Y, et al. Enrichr: interactive and collaborative HTML5 gene list enrichment analysis tool. BMC Bioinformatics. 2013;14:128. [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.
Supplementary Materials
Supp. Table 1: Demographic data (age, gender, and race/ethnicity) of all participants at any time point.
Supp. Figure 1: Heatmaps of top differentially expressed genes sorted by adjusted p value significance computed from the DESeq2 differential expression test. Volcano plots displaying genes differentially expressed between control and HKEC/HKSA samples. Differential expression is defined by a Log2 FoldChange >1 and an adjusted p value <0.05.
Supp. Figure 2: Dot plots of the top 25 positive z-score and negative z-score gene ontology (biological processes) sorted by adjusted p value. Gene Ontology analysis performed by GeneTonic R package 17. Z-score represents the standardized sum of the number of genes regulated in either direction indicating a positive z-score as an increased number of upregulated genes and a negative z-score as an increased number of downregulated genes.
Supp. Figure 3: List of genes with positive and negative residuals from linear regression analysis (Figure 2A). Gene Ontology overrepresentation analysis of positive and negative residual genes are as analyzed by EnrichR 49.
Supp. Figure 4: Overview of the major biological themes performed by Ingenuity Pathway Analysis (IPA) Core Analysis that includes canonical pathways, upstream regulators, diseases, and biological functions predicted based on findings in the QIAGEN’s IPA knowledge base. Input DEGs for IPA Core Analysis were DEGs obtained from a direct differential expression analysis between HKEC and HKSA-treated samples.
Supp. Figure 5: Plasma cytokine profiles from individuals (n=4) whose whole blood was exposed to either heat-killed E. coli (HKEC) or heat-killed S. aureus (HKSA) in the presence or absence auranofin (thioredoxin reductase 1 inhibitor (TXRN1), 1.3 μg/ml) for 6 hours. Data is expressed as median with intraquartile range. Colored data points are for each respective individual across all treatments and measurements.
Supp. Figure 6: Plasma cytokine profiles from individuals (n=6) whose whole blood was exposed to either heat-killed E. coli (HKEC) with anti-human TLR4 antibodies (5 μg/ml) or heat-killed S. aureus (HKSA) with anti-human TLR2 antibodies (5 μg/ml) in the presence of either brusatol (Nrf2 inhibitor, 60 ng/ml) or dimethyl fumarate (Keap1 inhibitor, 6 μg/ml) for 16 hours. Data is expressed as median with intraquartile range. Colored data points are for each respective individual across all treatments and measurements. * indicates a p value < 0.05 between designated groups.
Data Availability Statement
RNA sequencing data is publicly available in the Gene Expression Omnibus (GEO) as dataset GSE237960 as required by the National Institutes of Health Data Sharing Policy.