Metabolic competition between host and pathogen dictates inflammasome responses to fungal infection

Loading metrics

Open Access
Peer-reviewed

Research Article

Timothy M. Tucey, 

Jiyoti Verma, 

Françios A. B. Olivier, 

Tricia L. Lo, 

Avril A. B. Robertson, 

Thomas Naderer, 

Ana Traven

x

Published: August 4, 2020

https://doi.org/10.1371/journal.ppat.1008695

AbstractThe NLRP3 inflammasome has emerged as a central immune regulator that senses virulence factors expressed by microbial pathogens for triggering inflammation. Inflammation can be harmful and therefore this response must be tightly controlled. The mechanisms by which immune cells, such as macrophages, discriminate benign from pathogenic microbes to control the NLRP3 inflammasome remain poorly defined. Here we used live cell imaging coupled with a compendium of diverse clinical isolates to define how macrophages respond and activate NLRP3 when faced with the human yeast commensal and pathogen Candida albicans. We show that metabolic competition by C. albicans, rather than virulence traits such as hyphal formation, activates NLRP3 in macrophages. Inflammasome activation is triggered by glucose starvation in macrophages, which occurs when fungal load increases sufficiently to outcompete macrophages for glucose. Consistently, reducing Candida’s ability to compete for glucose and increasing glucose availability for macrophages tames inflammatory responses. We define the mechanistic requirements for glucose starvation-dependent inflammasome activation by Candida and show that it leads to inflammatory cytokine production, but it does not trigger pyroptotic macrophage death. Pyroptosis occurs only with some Candida isolates and only under specific experimental conditions, whereas inflammasome activation by glucose starvation is broadly relevant. In conclusion, macrophages use their metabolic status, specifically glucose metabolism, to sense fungal metabolic activity and activate NLRP3 when microbial load increases. Therefore, a major consequence of Candida-induced glucose starvation in macrophages is activation of inflammatory responses, with implications for understanding how metabolism modulates inflammation in fungal infections.
Author summary
Activation of the immune regulator NLRP3 inflammasome by microbial pathogens has been shown to play both protective and destructive roles in infection, underscoring the importance of tight control over NLRP3-driven inflammation to ensure host health. A key microbe recognised by NLRP3 is the human yeast commensal and pathogen Candida albicans, which is responsible for mucosal and invasive infections. We demonstrate that innate immune cells sense their metabolic status to trigger NLRP3 activation only when microbial numbers have reached dangerous levels. This regulation is a consequence of metabolic competition between C. albicans and macrophages for an essential nutrient–glucose. The NLRP3 inflammasome is activated when increased fungal load in the infection microenvironment drives down glucose levels, thereby causing glucose starvation in macrophages. Restoring glucose homeostasis in macrophages reduced NLRP3 activation and production of the proinflammatory cytokine IL-1β, suggesting that metabolism regulates NLRP3 inflammasome activity in fungal infections.

Citation: Tucey TM, Verma J, Olivier FAB, Lo TL, Robertson AAB, Naderer T, et al. (2020) Metabolic competition between host and pathogen dictates inflammasome responses to fungal infection. PLoS Pathog 16(8):
e1008695.

https://doi.org/10.1371/journal.ppat.1008695Editor: Tobias M. Hohl, Memorial Sloan-Kettering Cancer Center, UNITED STATESReceived: January 12, 2020; Accepted: June 7, 2020; Published: August 4, 2020Copyright: © 2020 Tucey et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.Data Availability: All relevant data are within the manuscript and its Supporting Information files.Funding: This work has been supported by Project Grants from the National Health and Medical Research Council of Australia, specifically APP1081072 (to AT and TN) and APP1158678 (to AT). TN and AT are Future Fellows of the Australian Research Council (FT170100313 to TN and FT190100733 to AT). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.Competing interests: The authors have declared that no competing interests exist. We note that MCC950 is a compound that is in the public domain. More broadly AABR is inventor on three patents on novel NLRP3 inhibitors:WO2016131098, WO2017140778, WO2018215818.

IntroductionThe innate immune system plays important roles in recognising and responding to pathogenic microbes, so that infections are prevented, controlled and resolved. A key immune pathway that responds to infection is orchestrated by the NLRP3 inflammasome [1, 2]. Activation of the cytoplasmic NOD-like receptor NLRP3 by microbial invasion of macrophages results in the assembly of a higher-order inflammasome complex that contains oligomers of NLRP3, the adaptor protein ASC and the protease caspase 1, reviewed in [3]. Assembled inflammasome mediates auto-proteolytic activation of caspase 1, which in turn leads to two functional consequence: (i) cleavage of the inflammatory cytokines IL-1β and IL-18 to their active versions, and (ii) programmed cytolysis of macrophages called pyroptosis, which is mediated by cleavage of gasdermin D and formation of gasdermin D pores in the plasma membrane [4–9]. These caspase 1-driven events are highly pro-inflammatory and orchestrate the immune response by producing and releasing bioactive cytokines, expelling intracellular pathogens and recruiting and activating neutrophils to promote microbial clearance [1, 10].
Potent inflammation can lead to tissue destruction and pathology. Therefore, macrophages have to carefully regulate the activity of NLRP3. In the context of infection, this means that the NLRP3 pathway needs to respond to dangerous microbial colonisation (i.e. infection), but there should be limited activation in response to benign, harmless colonisation. Indeed, while NLRP3 inflammasome contributes to host defences against microbial pathogens [11–15], if not tamed, hyper-inflammatory pathology can actually drive disease [15–21]. Our understanding of how the NLRP3 infammasome is controlled during infection remains incomplete, particularly since there is little evidence that NLRP3 binds to microbial ligands directly. Instead, NLRP3 is a sensor of physiological disturbances, such as potassium efflux, increased levels of reactive oxygen species (ROS) and breaches of plasma membrane integrity that result from microbial invasion or other dangers [3, 22]. Although the precise mechanisms remain to be understood, these “signals” are thought to lead to a conformational change in NLRP3 that initiates assembly of the inflammasome complex [23].
An important human pathogen recognised by the NLRP3 inflammasome is the yeast Candida albicans. Normally a benign coloniser of the human gut and mucosal surfaces, C. albicans is also an opportunistic pathogen causing oral, genital and invasive infections [24]. Precise regulation of NLRP3 activity is thought to be critical in fungal infections and it needs to be carefully balanced. In fact, both protective and pathological roles for NLRP3 inflammasome-driven inflammation have been found in models of C. albicans infection [11–13],[17–20]. Yet, precisely how macrophages regulate NLRP3 when responding to C. albicans is unclear.
It has been suggested that the NLRP3 inflammasome responds to virulence traits of C. albicans, and is activated when Candida transitions from the yeast morphotype (which is less invasive and more aligned with the benign state), to the invasive hyphal morphotype (which causes tissue damage and promotes pathogenesis) [25, 26]. Proposed mechanisms by which fungal hyphae could be sensed by the inflammasome are breaching of the phagosome membrane resulting from hyphal growth (but this has recently been disputed [27]), expression of a hyphae-specific membrane-lysing toxin or via recognition of cell surface properties that differ between yeast and hyphal cells [2, 11, 25, 26, 28–32]. The problem with this model is that recent work has uncoupled the strict requirement for hyphae in inflammasome activation. First, hyphae are not necessary to trigger caspase 1-dependent pyroptotic cell death of macrophages (C. albicans mutants locked in yeast form can do it) [31]. Second, hyphae are also not sufficient to trigger inflammasome activation, as there are mutants that can form hyphae but are compromised in causing pyroptosis and IL-1β production [32–34].
Recent work has put macrophage immunometabolism centre stage in infectious diseases [35]. Upon sensing of microbes, major reprogramming of macrophage metabolism leads to upregulation of glycolysis, increased glucose consumption and accumulation of TCA cycle intermediates, which together drive pro-inflammatory antimicrobial responses [36–38]. While these metabolic mechanisms are thought to promote clearance of infection, we have recently identified that microbial pathogens can fight back, taking advantage of the metabolic reprogramming of macrophages and turning it into a weakness [39]. In C. albicans-infected macrophages, the pathogen competes with immune cells for glucose and triggers glucose starvation and massive macrophage cell death [39]. The exact consequence of glucose starvation and macrophage killing by C. albicans remains to be understood. In other words, is this an immune evasion mechanism by C. albicans? An alternative possibility is that this constitutes a host response, by which macrophage glucose starvation and cell death regulate immune responses are needed to control C. albicans. These two possibilities are not mutually exclusive–Candida might try to evade immunity via glucose competition, which is then sensed by macrophages to trigger protective responses.
Here we demonstrate that the major consequence of Candida-induced glucose starvation in macrophages is activation of the NLRP3 inflammasome. We determine the mechanistic requirements for metabolic activation of the inflammasome by C. albicans and show that the metabolic mechanism of inflammasome activation is broadly conserved in diverse C. albicans clinical isolates. Our study demonstrates that increased pathogen load leading to glucose starvation is the central signal for immune cells to trigger inflammatory responses to C. albicans, while recognition of “virulence factors” in the form of hyphae plays a more restricted role. Taken together, our results show that metabolic competition between macrophages and C. albicans results in inflammasome responses, indicating that macrophage killing by glucose starvation triggers host protective processes. Sensing metabolic homeostasis as a mechanism for controlling immune activation is likely to be broadly relevant to microbial pathogenesis.

Results
Glucose starvation is the dominant mechanism of macrophage killing by diverse C. albicans clinical isolates, while only some isolates trigger robust pyroptosis
Following infection in vitro, C. albicans kills the entire macrophage population even at a low multiplicity of infection. This process encapsulates two key facets of C. albicans-macrophage interactions: NLRP3 inflammasome activation and competition for glucose between host and pathogen [32, 33, 39]. After infection with C. albicans, the first wave of macrophage death is driven by activation of NLRP3 inflammasome-dependent pyroptosis and has been termed “Phase I death” [32, 33]. The second wave is due to C. albicans starving macrophages of essential glucose [39], and has been termed “Phase II death” [32]. The two phases of macrophage death are depicted in Fig 1C, top left panel.
Fig 1. Clinical C. albicans isolates show different inflammasome-dependent killing of macrophages, but broadly conserved killing by glucose starvation.(A) Illustration of hyphal index calculation. For a given cell, the length from end-to-end (including all connecting septal junctions) was measured and divided by the width–defined as the widest measurement at the distal tip. True hyphae have a large hyphal index (great in length, narrow in width); pseudohyphae are reduced in length and greater in width; and yeast cells have a low hyphal index (similar proportions for both length and width). (B) Microscopy images of C. albicans clinical isolates 3 h post-phagocytosis in murine bone marrow-derived macrophages (BMDMs). The MOI was 6 Candida to 1 macrophage. Values in top-left corner of each image indicate the hyphal index, with the benchmark strain SC5314 normalized to 5. For each isolate, 200 cells were measured. Scale bar is 20 μm. These same images are also displayed in S1A Fig for direct comparison of hyphal index in the same tissue culture media but without macrophages. (C) BMDMs were infected with C. albicans strains at 6:1 MOI and assessed for cell death by live cell imaging over 24 h. Each graph compares a clinical isolate with the benchmark strain SC5314 and uninfected macrophages control, all assayed in the same live cell imaging experiment (note that P57055 and P78042 were assayed in the same experiment and therefore are being compared to the same control data (i.e. the data for strain SC5314 and the uninfected control is the same in these two graphs; the isolates are displayed in separate graphs for clarity in presentation). Data are the mean values and SEM from 2 independent experiments involving 4 different colonies (each colony was treated as a biological replicate) of the indicated Candida strain and at least 2000 macrophages surveyed for each strain, per experiment. The distinction between Phase I (inflammasome dependent killing) and Phase II (glucose starvation-dependent killing) is indicated in the first graph (labeled “I” and II”). The hyphal index (h.i.) is indicated for the clinical isolate being compared to SC5314 (SC5314 has an index of 5). A selection of clinical strains is shown here, and the entire set is displayed in S3 Fig. (D) Correlation plot of the time (in hours) until 50% macrophage death was reached versus hyphal index in macrophages, with the Pearson’s correlation coefficient (r) and p values.

https://doi.org/10.1371/journal.ppat.1008695.g001Our current understanding of C. albicans-induced macrophage death comes predominantly from studying a prototype clinical isolate called SC5314 and laboratory strains derived from it. However, patients are infected with their own commensal strains of C. albicans, which display great phenotypic diversity [40]. To overcome this limitation, we used live cell imaging to systematically analyse macrophage death with SC5314 and 20 other phylogenetically diverse strains, which originate from a number of body sites in patients and healthy individuals (oral, bloodstream and vaginitis) [40] (S1 Table).
First, we performed a set of control experiments to compare key phenotypes that could affect the ability of C. albicans isolates to kill macrophages: hyphal formation, growth rates and phagocytosis rates. Hyphal formation was assessed in tissue culture medium and during infection of bone marrow-derived macrophages (BMDMs). This set of strains has been profiled for hyphal formation in vitro [40], but they have not been assessed during in macrophages. A hyphal index was calculated after 3 h of infection by dividing the total length of hyphae with the width at the broadest point along the hyphal tip, and then normalizing to give a ratio. A hyphal index of 5 corresponds to true hyphae, while a hyphal index approaching 0 represents primarily yeast cells (Fig 1A). The isolates exhibited a range of filamentation phenotypes in macrophage co-incubations, with only some able to form robust hyphae and many showing more modest to low hyphal indexes (Fig 1B). Of the 21 strains, only SC5314 and P87 showed rapid, near-uniform true hyphae formation across the entire population of macrophages. Overall, there was a high level of correlation of the hyphal index in macrophages with the hyphal index during growth in vitro in liquid tissue culture media (see S1A Fig for images and S1B Fig for the correlation plots). Exceptions were strains P60002, P76067, P37037, and P76055, which were capable of forming substantial hyphae in liquid media, but were poor in macrophages, with most of the cells remaining trapped in macrophages after 3 h (Fig 1B and S1A Fig). These outliers tend to have slow doubling time in vitro, as measured by growth at 37°C in minimal media using glucose or the non-fermentable carbon source glycerol (note that the phagosome is likely glucose poor) (S1C Fig and S1D Fig). Using a selection of strains, growth rates determined by measurement of optical density mainly correlated with metabolic activity measured by the XTT assay (S1E Fig). There was no correlation of doubling time with hyphal formation in vitro (S1C Fig and S1D Fig). However, there was a significant correlation of doubling time in glucose with hyphal formation during macrophage infections (S1F Fig), although no significant correlation was found with doubling time in glycerol which was much slower for all strains compared to glucose (S1G Fig). While it is not known what the growth of these strains is in the phagosome, our data suggests that macrophages are capable of restraining hyphal formation of diverse C. albicans strains, particularly those which are slower growing. Phagocytosis of the strains was robust, and there were no major differences in macrophage infection rates (S2A Fig and S2B Fig, 10 strains with a range of hyphal indexes were analysed). Collectively, our data show that hyphal formation in macrophages, the key invasive and pathogenic phenotype thought to drive NLRP3 activation and macrophage killing, is highly variable between C. albicans clinical strains.
Next, we asked how these strains kill macrophages. There was a strong inverse correlation between the hyphal index in macrophages and the time taken for 50% of macrophage killing to occur (Fig 1D). In other words, strains with a high hyphal index tend to kill macrophages faster due to their ability to trigger a steeper, more rapid Phase I death (see Fig 1C for a selection of strains and S3 Fig for all strains). We have previously shown that around 50% of Phase I death is attributable to inflammasome-dependent pyroptosis [32]. There was a correlation between growth rates of the strains in vitro with the speed of macrophage killing, but this was not statistically significant (S2C Fig). Some outliers were also evident. For example, P78048 had a relatively low hyphal index (Fig 1B and S1A Fig), yet it was capable of killing macrophages as fast as the robustly hyphal strain SC5314 during Phase I (Fig 1C). Our analysis further revealed that several C. albicans strains were very poor or incapable of triggering pyroptotic Phase I death altogether (e.g. P37037, P75010, P78042) (Fig 1C and S3 Fig). Therefore, the ability to trigger inflammasome-dependent pyroptosis is highly variable between C. albicans clinical strains, questioning the broad relevance of this process in C. albicans-macrophage interactions.
In contrast, all clinical strains, including those that form poor hyphae, were able to trigger robust glucose starvation-dependent Phase II death (Fig 1C and S3 Fig), showing that glucose starvation is a broadly conserved mechanism of death for C. albicans-infected macrophages. The timing of Phase II death differed between the strains, but once it started, it proceeded with fast kinetics (Fig 1C and S3 Fig). This means that all strains show some level of escape from macrophages, which likely enables glucose depletion during replication of C. albicans in the extracellular infection environment. Even strain P94015, which cannot form hyphae, could occasionally escape from macrophages sufficiently to kill the entire population, although this took well over 24 h to occur (S1 Movie).
Using strain SC5314, we show that the two mechanisms of macrophage cell death, namely NLRP3-dependent pyroptosis and glucose starvation-dependent death, are recapitulated in primary human monocyte-derived macrophages (hMDMs) and follow similar kinetics (Fig 2A and 2B). The NLRP3 inhibitor MCC950 rescued hMDMs from C. albicans-induced Phase I death, showing NLRP3-dependent pyroptosis (Fig 2A). Glucose supplementation delayed Phase II death of hMDMs, showing glucose starvation-dependent death (Fig 2B). As in mouse BMDMs, only some C. albicans strains triggered inflammasome-dependent pyroptosis of hMDMs depending on their hyphal index: SC5314 (hyphal index 5) triggered robust inflammasome-dependent pyroptosis, P57077 (hyphal index 1.4) triggered more modest pyroptosis and P78042 (hyphal index 0.3) did not trigger pyroptosis (Fig 2C). In contrast, all three strains triggered glucose starvation-dependent Phase II death of hMDMs (Fig 2C).
Fig 2. Distinct inflammasome and glucose-dependent killing by C. albicans isolates is recapitulated in human primary macrophages.(A) Primary human monocyte-derived macrophages (hMDMs) infected with C. albicans strain SC5314 at 6:1 Candida:macrophage MOI. hMDMs were collected from two donors and were assayed on the same day in the same experiment. The two donors are plotted separately. Each graph shows the average and SEM of two technical repeats. At least 1500 macrophages were counted per condition, per donor. The NLRP3 inhibitor MCC950 was added at 10 μM to assess the involvement of NLRP3 inflammasome-dependent pyroptosis during Phase I death of Candida-infected hMDMs. (B) Same experiment as in A, this time comparing media with distinct glucose concentrations to assess glucose starvation during Phase II death of hMDMs. All of the conditions (+/-MCC950 shown in A and medium with increasing glucose concentrations shown in B) were assayed together in the same experiment but are plotted separately for clarity. Therefore, data for uninfected controls and infections in medium with 10 mM glucose are the same in panel A and panel B for each donor. (C) hMDMs infected with C. albicans strains SC5314, P57055 and P78042 at 6:1 Candida:macrophage MOI. hMDMs were collected from two different donors and are plotted separately. Each graph shows the average and SEM of two technical repeats. At least 1500 macrophages were counted per condition, per donor.

https://doi.org/10.1371/journal.ppat.1008695.g002

C. albicans triggers inflammasome activation in two temporally distinct events
Given that several C. albicans strains failed to induce robust inflammasome-dependent Phase I killing of macrophages (Fig 1), the next question was whether all C. albicans isolates are recognised by macrophages to activate the inflammasome. To answer this, we imaged inflammasome activation over time using immortalized BMDMs expressing fluorescently labelled ASC, the adaptor subunit of the NLRP3 inflammasome (ASC-mCerulean) [41]. Activation of the inflammasome complex is associated with oligomerization of ASC and the formation of a visible large “speck” (~1 μm), which allows for quantification of inflammasome activation events in macrophage populations over time using live cell imaging. In parallel, we examined host mitochondrial health by staining mitochondria with TMRM (signal intensity depends on mitochondrial membrane potential), and we also quantified macrophage cell death. The live cell imaging results for all 21 C. albicans isolates, along with the positive control nigericin (a potent inducer of the NLRP3 inflammasome) and the uninfected negative control are shown in S4 Fig.
Unexpectedly, all C. albicans strains triggered inflammasome activation (S4 Fig), even those that formed poor or no hyphae and could not trigger “Phase I” inflammasome-dependent pyroptotic death in macrophages (Fig 1). Interestingly, depending on morphological phenotype and infection loads, the kinetics and mechanisms of inflammasome activation were distinct. Below we describe our conclusions by comparing the prototype highly hyphal strain SC5314 and a selected bloodstream infection isolate with poor hyphae, P78042.
In the presence of robust hyphae (strain SC5314), the bulk of the inflammasome activation events (i.e. ASC speck formation) occurred early, in the first 6 h following the start of infection (Fig 3A). The peak of ASC speck-positive cells in the population was ~15% by 6 h, after which it plateaued. ASC speck formation was preceding cell death (typically ~15–30 minutes before), consistent with activation of inflammasome-dependent pyroptosis during Phase I death of macrophages. This data is consistent with our previous study using a laboratory strain derived from SC5314 [42]. Similar to our previous finding with primary macrophages (BMDMs) [39], C. albicans infection of the ASC-mCerulean immortalised BMDMs triggered hyperpolarisation of macrophage mitochondria after several hours of infection preceding the rapid Phase II death (Fig 3B). This indicates that macrophages are experiencing mitochondrial dysfunction. During the early, Phase I activation of the inflammasome some ASC specks were occurring well in advance of mitochondrial hyperpolarization, but additional ASC specks formed coinciding with mitochondrial hyperpolarisation (Fig 3B).
Fig 3. Inflammasome activation in response to C. albicans occurs in two temporally distinct events.(A) ASC-mCerulean-expressing immortalized macrophages following challenge with Candida strains SC5314 or P78042 at MOI 6:1. Left panel: representative images from live cell microscopy at 6 and 16 h post-phagocytosis, with ASC-mCerulean detection displayed in cyan. Scale bar is 50 μm. Right panel: graph showing quantification of ASC speck formation over time. Data are the mean values and SEM from 4 independent experiments. (B) Experimental conditions as in A. Shown are % macrophages containing an ASC speck, % macrophages with hyperpolarized mitochondria and % dead macrophages quantified in live cell imaging experiments. ASC speck data is the same as in panel A. Data are the mean values and SEM from 4 independent experiments (see S4 Fig for complete set of data involving all 21 clinical isolates). (C) Correlation plot of the time post-phagocytosis when peak ASC speck formation occurs versus hyphal index of the Candida isolates in macrophages, with the Pearson’s correlation coefficient (r) and p values as indicated. (D) Correlation plot of peak percentage of macrophages in the population that contain an ASC speck versus hyphal index in macrophages, with the Pearson’s correlation coefficient (r) and p values as indicated. (E) IL-1β detection following challenge with Candida strains SC5314 or P78042 at MOI 6:1. Primary murine BMDMs were primed with LPS (50 ng/ml) for 3 h, followed by challenge with Candida strains SC5314 or P78042 (MOI 6:1), nigericin (10 μM), or left uninfected. At 12 h post-challenge, supernatants were analysed by immunoblot for IL-1β. Shown is one representative immunoblot from two independent experiments involving different mice. Note that this immunoblot contains additional lanes that were spliced out (indicated by the black lines); see S5 Fig for the entire, uncropped immunoblot from this experiment. (F) BALBc mice were injected with 1×106 CFU of the indicated C. albicans strains: SC5314 (highly hyphal, triggers the early wave of inflammasome activation) or P78042 (poorly hyphal, triggers only the second wave of inflammasome activation). Mock injection was with PBS. Levels of IL-1β in kidney were measured by ELISA (n=10 animals/group for C. albicans infection, and 5 for PBS injection). *** p Read More