Filamentous fungal carbon catabolite repression supports metabolic plasticity and stress responses essential for disease progression

Aspergillus fumigatus is responsible for a disproportionate number of invasive mycosis cases relative to other common filamentous fungi. While many fungal factors critical for infection establishment are known, genes essential for disease persistence and progression are ill defined. We propose that fungal factors that promote navigation of the rapidly changing nutrient and structural landscape characteristic of disease progression represent untapped clinically relevant therapeutic targets. To this end, we find that A. fumigatus requires a carbon catabolite repression (CCR) mediated genetic network to support in vivo fungal fitness and disease progression. While CCR as mediated by the transcriptional repressor CreA is not required for pulmonary infection establishment, loss of CCR inhibits fungal metabolic plasticity and the ability to thrive in the dynamic infection microenvironment. Our results suggest a model whereby CCR in an environmental filamentous fungus is dispensable for initiation of pulmonary infection but essential for infection maintenance and disease progression. Conceptually, we argue these data provide a foundation for additional studies on fungal factors required to support fungal fitness and disease progression and term such genes and factors, DPFs (disease progression factors).


Introduction
Microbial pathogenesis is a complex, multifactorial process in which interactions between a microbe and host shape disease development and outcome [1,2]. While many studies have focused on defining the molecular determinants required for fungal pathogenesis, molecular pathogenesis studies often fail to address the spatial and temporal dynamics of an infection. These dynamics are the result of local changes in nutrients, stressors, and even substrate phase (solid tissue versus liquid environments) that occur over the course of a host-microbe interaction. How fungi navigate these rapidly changing infection dynamics to promote disease is largely unknown and difficult to study with traditional molecular pathogenesis approaches. However, identifying fungal and host factors essential for infection maintenance and disease progression may yield novel and potent therapeutic targets and approaches. With regard to invasive pulmonary aspergillosis (IPA), initiation of disease caused by A. fumigatus involves conidial deposition into the airways, conidia germination, and subsequent hyphal extension into the lung parenchyma that induces host damage as a result of both fungal growth and inflammation. Many fungal genes and attributes have been identified to be associated with virulence through robust animal model studies [3,4]. However, patients are typically diagnosed after this initiation phase of the infection when invasion of the lung parenchyma and even vasculature has already occurred. Consequently, it is unclear if fungal virulence factors and attributes identified through animal model virulence studies are relevant at these later stages of the host-pathogen interaction that must be targeted by clinically relevant therapeutics.
It has long been hypothesized that metabolic flexibility significantly contributes to A. fumigatus virulence; though specific tests of this hypothesis are difficult to achieve due to genetic redundancy in the fungus and the largely uncharacterized dynamic infection microenvironments [5][6][7][8]. While the role of several core metabolic pathways has been largely studied in the context of infection initiation (reviewed in [8,9]), the carbon metabolism regulatory mechanisms required for fungal proliferation within the host as the infection and disease progresses are ill defined. Previously, we observed that established infection microenvironments in murine models of invasive pulmonary aspergillosis (IPA) are depleted in oxygen and contain fungal metabolic by-products such as ethanol [10]. This result was surprising given that A. fumigatus and related filamentous fungi are obligate aerobes that use glucose as the preferred carbon source and raises important questions about fungal metabolism and bioenergetics throughout the course of an infection. These observations led us to question whether illdefined mechanisms of A. fumigatus infection metabolism dynamics are critical for in vivo fungal persistence, virulence, and ultimately disease progression.
One mechanism widely employed by microbes to regulate and optimize nutrient acquisition and metabolism is carbon catabolite repression (CCR) [11][12][13]. Transcriptional regulation of CCR controls central carbon metabolism in many microorganisms and allows prioritization of preferential carbon source usage to yield maximum energy and fitness. The role of CCR in microbial pathogenesis is established in bacteria, and in many species this system plays a role in virulence (reviewed in [14]). For example, in Streptococcus pyogenes, the transcriptional regulator CcpA is required for both colonization of the nasopharynx and virulence in invasive disease [15]. Intriguingly, transcriptional regulation of CCR seems dispensable for virulence in human pathogenic yeast based on host survival outcomes. Mig1, the CCR transcriptional regulator in Candida albicans, is dispensable for virulence in a systemic murine infection model of candidiasis [16]. Similarly, a mig1-null mutant in the basidiomycete yeast Cryptococcus neoformans causes wild type levels of disease in a murine inhalation model with no effect on growth within the host though an undefined role for Mig1 in macrophage-yeast interactions was suggested [17]. Yet, given our and other's previous in vivo observations in murine models of IPA, we questioned whether CCR would be dispensable for virulence in an environmental filamentous fungus capable of extreme metabolic flexibility.
To test this hypothesis, we generated a genetic null mutant of the predicted A. fumigatus CCR transcriptional repressor, CreA. Interrogation of this genetic null mutant in vivo and in vitro supports the hypothesis that CreA is a CCR transcriptional repressor in A. fumigatus and surprisingly revealed a crucial role for this system in established infection microenvironments.
Our data suggest a model whereby a clinically relevant steroid treatment promotes release of alternative de-repressing carbon sources utilized by A. fumigatus through CreA independent mechanisms during infection initiation. However, for the infection to proceed and cause lifethreatening host damage, dynamic spatial temporal changes in nutrient and oxygen availability at the site of infection drive a requirement for CreA activity in A. fumigatus. Taken together, our data emphasize the critical importance of spatial temporal dynamics in fungal-host interactions and reveal a novel role for fungal CCR in navigation of infection dynamics and disease progression. We propose that CreA and CCR represent a new class of fungal virulence factors with significant clinical relevance we term DPFs for disease progression factors.

Results
A. fumigatus employs a carbon catabolite repression system controlled by CreA nidulans (Anid_CreA), C. albicans (Calb_MIG1; 32% identity), Cryptococcus neoformans (Cneo_MIG1; 32% identity), Trichoderma reesei (Trees_Cre1; 56% identity) and Saccharomyces cerevisiae (Scere_MIG1; 29% identity), reveals little conservation outside of the zinc-finger domains of these 6 CCR transcriptional regulator homologs (S1 Fig). Outside of this region, there is large divergence in the amino acid sequences among these proteins, both in sequence and total length that may suggest the existence of functional differences across species. To study the role of this regulator in A. fumigatus, we generated genetic null mutant and reconstituted strains of creA (ΔcreA and creA R respectively; S2 Fig).
To determine whether CreA controls CCR in A. fumigatus, we used allyl alcohol as a measure of glucose repression of alcohol dehydrogenases that are well characterized targets of the CCR system in A. nidulans [18,19]. Allyl alcohol is metabolized by the alcA/aldA gene cluster [20,21] and broken down into the toxic byproduct, acrolein. When grown on 1% glucose media in the presence of 0.1% allyl alcohol the wild type strain shows a 27% inhibition in radial growth. However, under the same conditions ΔcreA fails to germinate resulting in 100% growth inhibition (Fig 1A). Re-introduction of creA (creA R ) restored growth on glucose in the presence of allyl alcohol to wild type levels. As expected, growth of all strains on 1% lactate with 0.1% allyl alcohol was inhibited 100% (S3A Fig), supporting the conclusion that derepression of glucose-repressed alcohol dehydrogenases is responsible for growth inhibition on allyl alcohol. We also tested the growth of wild type and ΔcreA on media containing the glucose analog 2-deoxyglucose (2-DG). While the wild type, ΔcreA, and creA R strains are not impaired on 1% glucose media with 0.1% 2-DG, the wild type and creA R strains are partially inhibited on 1% ethanol or 1% lactate with 0.1% 2-DG. However, ΔcreA growth is significantly less inhibited than the wild type strain, further supporting the conclusion that CreA controls CCR in A. fumigatus (S3B Fig). Collectively, these results support regulation of CCR by CreA in A. fumigatus.
Loss of creA reduces fitness in carbon repressing and depressing environments As expected for a transcriptional regulator of CCR, loss of creA results in de-repression of ethanol utilization genes as observed with increased allyl alcohol sensitivity ( Fig 1A). Given this role in CCR we tested growth of ΔcreA on several carbon and nitrogen sources. We observed that fitness of ΔcreA on solid media is reduced on both repressing (glucose, acetate) and derepressing carbon sources (ethanol, lactate), as well as on a variety of rich (glutamine, ammonium) and poor nitrogen sources (nitrate, urea) with glucose as a carbon source (Fig 1B; S4A  Fig). This result contrasts with CreA homologs in other environmental but non-human pathogenic fungi, such as CRE1 of T. reesei and Cre-1 of Neurospora crassa, where growth defects of the respective CCR mutant are only apparent on particular carbon sources [22,23]. We also observed comparable growth defects on ethanol, lactate, acetate, and glycerol when glutamine was supplied as the nitrogen source (S4A Fig) and on complete media (S4B Fig). Re-introduction of the creA coding sequence (creA R ) fully restored fitness of ΔcreA on all tested conditions. Taken together, these data suggest that CreA is critical for A. fumigatus fitness on solid substrates in both repressing and de-repressing conditions.

Regulation of CCR via CreA reveals spatial temporal requirements for in vivo fungal fitness and virulence
Given the role of CreA in CCR and growth in vitro, we hypothesized that this transcription factor would be critical for virulence. To test the contribution of CreA to A. fumigatus virulence, we used a triamcinolone-induced immune suppression murine model of IPA [24,25]. Surprisingly, given the in vitro phenotypes of CreA loss, after 2 days post-inoculation, mice inoculated with the wild type, ΔcreA, or creA R strains all exhibited symptoms of A. fumigatus infection. In fact, on day 3 post-inoculation the mortality between all 3 groups was essentially identical with 50% of mice having succumbed to the fungal challenge irrespective of the presence of CreA and a functioning CCR system (Fig 2A).
One potential hypothesis to explain this result is a difference in immunopathogenesis in the presence and absence of CreA. However, no obvious qualitative difference in the inflammatory response was observed between each strain with histopathology. Inflammation observed in all  animals by H&E staining of sectioned lung tissue collected 48 hours post inoculation (hpi) was mainly centered upon major airways with extension into the surrounding alveolar parenchyma ( Fig 2B). In support of the qualitative data, we observed no significant differences in expression of TNF-α, IL-10 and CXCL1 between mice inoculated with wild type or ΔcreA conidia, with the exception of increased CXCL1 in response to WT at 72 hpi, a result driven by Consistent with the survival curve and observed similar levels of inflammation, Gomori methenamine silver (GMS) staining indicated that the fungal burden of mice inoculated with wild type and ΔcreA was indeed similar. Moreover, the amount of tissue involvement in mice inoculated with ΔcreA both in terms of lesion size and frequency was similar to wild type ( Fig  2B). The histopathology data is supported by quantitative measurement of fungal burden using quantification of 18S rDNA 48 hpi that showed no significant difference between wild type and ΔcreA inoculated mice (Fig 2C). Similarly, we found that ΔcreA showed no defect in germination at 8 hpi in vivo, which further suggests that there is no defect in initiation of infection in the absence of CreA ( Fig 2D). These data suggest that the initial in vivo airway microenvironment complements the loss of CreA in A. fumigatus and therefore CCR as mediated by CreA is not necessary for establishment of infection and disease in the lung. To test whether the in vivo nutrient environment could complement the loss of CreA as suggested by the in vivo data, we tested the growth of ΔcreA in liquid lung homogenate and observed that growth was similar to the wild type strain ( Fig 2E). These data suggest the nutrients available within the steroid treated lung prior to initiation of infection and inflammation are adequate to support the germination and growth of ΔcreA.
At day 4 post inoculation a marked difference in the infection course emerged between animals inoculated with ΔcreA compared to the wild type and reconstituted strains. Ninety percent of mice inoculated with the wild-type and reconstituted strain succumbed to the infection at this time point while mice inoculated with ΔcreA had a significant increase in survival compared to wild type (p = 0.0217) and creA R (p = 0.0425) that extended out to 14 dpi (Fig 2A). Histopathological analysis of the lungs from the surviving animals inoculated with ΔcreA show persistent fungal abscesses that are walled off by a thick layer of immune cells that consisted predominantly of neutrophils (S5B Fig). Taken together, these data indicate that transcriptional regulation of filamentous fungal CCR is dispensable for early growth and initiation of disease but essential for virulence and disease progression after infection establishment.

CreA is critical for fitness in liquid environments containing repressing carbon sources
Despite the ability of ΔcreA to establish infection, the observed temporal virulence defect prompted us to examine the in vitro growth phenotypes of this strain more closely to understand underlying mechanisms. In contrast to growth on solid media, ΔcreA shows a partial growth defect in repressing carbon source liquid media, but not under liquid de-repressing conditions. To test growth initiation in liquid conditions that may better reflect the early infection air-surface liquid interface in the lung, we measured absorbance of cultures grown in 1% glucose, 1% lactate and 1% Tween-80 ( Fig 3A, 3B and 3C) and calculated the change in absorbance over time (ΔAbs 405 /min). In 1% glucose, the ΔAbs 405 /min of ΔcreA is significantly lower than the wild type and creA R . However, under carbon source de-repressing conditions (Tween-80, lactate), the ΔAbs 405 /min of ΔcreA is strikingly similar to the wild type and creA R strains ( Fig 3D). These results suggest that the initial host environment encountered by A. fumigatus is enriched for de-repressing carbon sources that promote fungal growth in the absence of transcriptional CCR regulation.
In support of this hypothesis, a UPLC-MS/MS based steady-state metabolite profile of whole murine lungs of steroid-treated mice versus healthy controls revealed a significant increase in alternative carbon and nitrogen sources upon steroid treatment (Fig 4). Of 630 detected metabolites, 242 metabolites were significantly (p 0.05) altered in the triamcinolone-treated lungs compared to healthy, untreated controls (Fig 4A, S1 File). Some key findings from this comparison include a significant decrease in glucose (0.71-fold change; p = 0.0042; Fig 4B) and glutamate (0.77-fold change; p = 0.0133; Fig 4C), preferred carbon and nitrogen sources, respectively, for A. fumigatus, and a concomitant increase in amino acids (leucine, asparagine, valine, methionine, alanine, glutamine, threonine, proline and 5-oxoproline), long chain fatty acids (myristate, arachidate, maragarate, oelate; Fig 4D and 4E), and the alternative nitrogen source, urea. While the steroid treated lungs represent the lung environment first encountered by conidia upon inhalation, we also hypothesized that the environment would change rapidly upon fungal inoculation due to the host immune response. Thus, we also measured the steady-state metabolite profile of steroid-treated animals inoculated with wild type fungal conidia. At 8 hours post inoculation, 168 of 630 detected metabolites significantly (p<0.05) changed between steroid-treated un-inoculated lungs and steroid-treated fungalinoculated lungs (Fig 4A). While many changes induced by steroid treatment were reversed by inoculation with fungus, glucose remained significantly lower than naïve lungs, and alternative carbon sources such as fatty acids, remained significantly increased (Fig 4B-4D, S1 File). Overall, the nutrient landscape of the lungs at the time of inoculation and during initiation of fungal growth is enriched with alternative carbon sources. These data support the hypothesis that the host environment is complex and dynamic and support the observation that filamentous fungal CCR is dispensable for fungal fitness early in lung infection. CreA plays a major role in transcriptional regulation of fungal central carbon metabolism To gain a deeper understanding of the function of CreA under repressing and de-repressing conditions we used an RNA-Sequencing based approach. Wild type and ΔcreA were cultured in 1% liquid glucose minimal media (GMM) overnight, then shifted to fresh 1% glucose (repressing) or 1% ethanol (de-repressing) minimal media (EMM) for 2 hours before sample preparation for RNA-sequencing. Importantly, this time point was chosen because growth rate is similar between strains (S4C Fig). Moreover, our previous in vivo metabolomics of an IPA murine model revealed significant ethanol levels in the lungs suggesting this alcohol is a potential carbon source seen by A. fumigatus [10].
In the wild-type strain in de-repressing conditions (EMM), the abundance of 860 transcripts significantly (p<0.05) increased and the abundance of 741 transcripts significantly decreased by at least two-fold (S1 File). As expected, we observed an increase in transcripts of genes involved in the utilization of alternative carbon sources, including the glyoxylate cycle, gluconeogenesis, succinate-fumarate antiporter, aldehyde dehydrogenase and acetyl-coA synthase encoding genes. The induction of these gene indicates generation of anabolic intermediates through gluconeogenesis and subsequent energy production through respiration. As expected, transcript levels of the ethanol utilization alcohol dehydrogenase, alcA, increased significantly in EMM, while transcript levels of the fermentation associated alcohol dehydrogenase, alcC, decreased [10].
In glucose, loss of CreA resulted in a significant reduction (! 2 fold) in transcript levels of 450 genes (p<0.05) and significant increase in transcript levels of 402 genes (S1 File). We verified transcript levels of a subset of genes with qRT-PCR and found that patterns of transcript levels between wild type and ΔcreA in both glucose and ethanol conditions were consistent between RNA-sequencing and qRT-PCR results (S6 Fig). We applied FunCat analysis to the genes that significantly increase and decrease in ΔcreA compared to wild type genes using Fun-giFun2 (S7 Fig) [26]. The categories of increased genes include a large number with predicted functions in transport of metabolites, carbohydrate metabolism, and secondary metabolism. Included in these genes are the alcohol utilization genes alcA and alcS, the glyoxylate cycle enzymes, isocitrate lyase, acuD, and malate synthase, acuE, as well as acuF, which encodes PEP carboxykinase, and fbp1, encoding fructose bisphosphatase aldolase, the rate-limiting steps of gluconeogenesis. Consistent with a role of CreA in regulation of CCR, this transcriptome profile of ΔcreA suggests that the glyoxylate cycle and gluconeogenesis are active in this strain despite abundant glucose in the environment. The representative categories of decreased genes also include secondary metabolism and c-compound and carbohydrate metabolism in addition to degradation of amino acids. Genes that changed significantly in 1% ethanol between ΔcreA and wild type were assigned to similar FunCat categories (S1 File; S7 Fig). These data demonstrating significant reductions in mRNA levels in many genes in the absence of CreA may also suggest that CreA has additional roles as a transcriptional activator, but further analyses are required to test this hypothesis. Consequently, these data suggest that loss of creA in an environment rich in de-repressing carbon sources such as the steroid treated airway would likely be dispensable for fungal fitness. To further explore the mechanism underlying the observed in vitro growth defects of CreA loss we turned to a metabolomics approach.

CreA regulates fungal bioenergetics and cell wall homeostasis
Global steady-state metabolomics analysis of ΔcreA mycelia compared to wild-type revealed striking alterations in glucose metabolism in ΔcreA (Fig 5A; S1 File). We observed a significant increase (5.31-fold) in the amount of intracellular glucose in ΔcreA, however no significant changes in metabolites of the early steps of glycolysis (glucose-6-phoshate and fructose-6-phosphate) were observed. In contrast, metabolites of the late steps in glycolysis, including the sugar-phosphates 3-phosphoglycerate (3-PG), 2-phosphoglycerate (2-PG), and phosphenolpyruvate (PEP), were significantly increased in ΔcreA. In addition, we observed significant changes in tricarboxylic acid (TCA) cycle intermediates. The oxidative portion of the cycle shows significantly altered metabolite levels with citrate, isocitrate and aconitate decreased in the absence of CreA. However, fumarate and malate significantly increase in ΔcreA compared to wild type ( Fig 5A). This pattern of changes in the TCA cycle is consistent with over-expression of isocitrate lyase in A. niger [27], suggesting that the glyoxylate cycle is active in ΔcreA despite the presence of glucose. Consistent with this observation, transcript levels of both the predicted isocitrate lyase (acuD; 4.7-fold) and malate synthase (acuE; 2.39-fold) in the glucosegrown ΔcreA are significantly increased compared to wild type. Furthermore, transcripts of key gluconeogenic enzymes, PEP carboxykinase (acuF; 4.53-fold) and fructose bisphosphatase (fbp1; 3.2-fold) are increased in ΔcreA grown in glucose, suggesting that carbon is being fluxed from the TCA cycle, through the glyoxylate shunt and fed into gluconeogenesis. The over-lap of transcriptomics and metabolomics data sets support the conclusion that CreA is a central regulator of fungal bioenergetics. Given the significant impact of CreA loss on carbon metabolism, we hypothesized that the polysaccharide rich cell wall, closely associated with virulence, Metabolites that are boxed in red and green are significantly increased and decreased, respectively, in ΔcreA, and black are not significantly changed (p 0.05 by Welch's two-sample t-test). mRNA levels of key metabolic enzymes are given as fold-change in ΔcreA compared to wild type. B) Expression of acuF as measured by qRT-PCR from cultures grown in GMM overnight, then shifted to fresh glucose media with samples taken at indicated time-points post shift. acuF expression is normalized to actA and tub2. a p = 0.0046, b p = 0.0006, c p = 0.0120, d p = 0.001, e p = 0.0054, n.s. = not significant by unpaired, two-would be significantly altered in ΔcreA. In support of this hypothesis, we observed that ΔcreA is more sensitive to the cell wall perturbing agents calcoflour white (CFW), congo red (CR) and to a lesser extent caspofungin (CF) when on a glucose containing medium, indicating that cell wall integrity of ΔcreA is perturbed in these conditions (Fig 6).
To test whether changes in transcript levels of these metabolic enzymes at the time of harvest for RNA-sequencing are a result of true dysregulation or a product of a shift in kinetics of expression following the shift to fresh glucose media, we examined transcript levels of acuF over two hours following the shift to fresh glucose. Consistent with perturbation in regulation of these enzymes, we observed transcripts of the gene encoding this enzyme are significantly higher than wild type at nearly all time-points ( Fig 5B). Thus, we conclude that loss of CreA results in dysregulation of genes encoding glyoxylate shunt and gluconeogenic enzymes. Moreover, transcript levels of a putative fructose bisphosphate aldolase are 12.72-fold reduced in ΔcreA, rendering transcript levels of this enzyme nearly absent. We conclude from these data that carbon is being moved through the glyoxylate shunt back into gluconeogenesis despite the already high levels of glucose intracellularly. Consequently, the lack of an FBPaldolase generates a block in glycolysis/gluconeogenesis, which is responsible for the accumulation of the sugar-phosphate intermediates 3-PG, 2-PG and PEP. The accumulation of these intermediates not only traps cellular carbon, but also lowers available phosphate for cellular bioenergetics and fitness.
In support of this hypothesis, we observed a 5.79-fold increase in the ADP/ATP ratio in ΔcreA as compared to wild type ( Fig 5C). From our metabolomics data, we noted a 5.67-fold increase in the amount of AMP present in ΔcreA compared to wild type (S1 File). Together, both the increase in AMP and ADP/ATP ratio suggest a defect in mitochondrial output resulting in insufficient cellular bioenergetics. In further support of these data, ΔcreA is significantly (p<0.0001) more sensitive to the respiratory inhibitors targeting complex II (thenoyltrifluoroacetone; TTFA), complex III (Antimycin A), complex V (Oligomycin A) and the alternative oxidase (Salicylhydroxamic acid; SHAM) when grown on both 1% glucose and 1% ethanol minimal media (S8 Fig). Overall, transcriptomics and metabolomics data suggest that CreA is critical for fungal fitness in environments with repressing carbon sources such as glucose.

CreA activity supports disease progression in part through promoting fungal fitness in low oxygen infection microenvironments
With the observed increase in alternative carbon sources upon steroid treatment, we sought to test the hypothesis that reduced repressing carbon sources early during A. fumigatus host interaction results in de-repression of CreA-regulated genes. We extracted RNA from triamcinolone treated mice inoculated with fungal conidia 24 and 72 hpi. In support of the model, the CreA-regulated genes, acuD and acuF, do not increase over time from 24 to 72 hours and the trend was for higher levels early in the infection consistent with increased levels of alternative carbon sources (Fig 7A and 7B). In contrast, the transcript levels of acuF and acuD, are significantly increased in ΔcreA compared to wild type at 24 and 72 hpi, which strongly supports the conclusion that these CreA-target genes identified in our RNA-sequencing data are regulated in part by CreA in vivo (Fig 7A and 7D).
To further test our model, we measured the transcript levels of genes expressed in response to severe oxygen depletion, a known characteristic of fungal lesions in the triamcinolone tailed t-test as compared to WT of respective time-point. Data represents the mean of three biological replicates ± SEM. C) ADP/ATP ratio of cultures grown in GMM overnight then shifted to fresh GMM for 2 hours. Data represents mean of biological triplicates ± SEM; **p<0.0001, n.s. = not significant by unpaired, two-tailed t-test, as compared to WT. https://doi.org/10.1371/journal.ppat.1006340.g005 Carbon catabolite repression is required for fungal virulence model of IPA [10]. These genes, including the pyruvate decarboxylase, pdcA, and alcohol dehydrogenase alcC, involved in ethanol production, are significantly increased from 24 hpi to 72 hpi in both wild type and ΔcreA (Fig 7C and 7D). This increase in hypoxia responsive genes led us to hypothesize that SrbA and SrbB activity, known regulators of the hypoxia response and carbohydrate metabolism in A. fumigatus [28,29], increases over the course of infection. However, we did not observe any change in srbA mRNA levels from 24 to 72 hpi in the wild Carbon catabolite repression is required for fungal virulence type strain perhaps consistent with the absolute requirement of SrbA for A. fumigatus virulence ( Fig 7E). Strikingly, we observed a significant increase in the mRNA levels of srbA and srbB over the course of the ΔcreA infection (Fig 7E and 7F). We interpret these data to suggest that ΔcreA experiences increased reductive stress due to an inability to engage the fungal hypoxia response. Taken together, these data suggest that CreA is critical for maintaining fitness in the low-oxygen environment of established fungal lesions in the lung.
To test this hypothesis, we measured the growth rate of point inoculated colonies that were germinated in normoxia for 24 hours and then shifted to low oxygen (hypoxia inducing) environment for 96 hours. The ratio of colony growth in hypoxia to normoxia was significantly higher for ΔcreA compared to wild type and creA R at the first day post-shift to hypoxia, however, after 48 hours, hypoxia/normoxia growth ratio of ΔcreA was dramatically lower than wild type, and continued to decrease over time (Fig 8A), which indicates that increased exposure to low oxygen conditions in the presence of a repressing carbon source results in a strong reduction in ΔcreA growth. Based on our model of CreA interactions with the host and the inability to adapt to the severe low oxygen conditions of the host environment, we tested the susceptibility of ΔcreA to killing by neutrophils. We grew wild type, ΔcreA and creA R germlings in both repressing (glucose) and de-repressing (Tween-80) conditions, then shifted cultures to a low oxygen environment and added bone marrow derived neutrophils. After 2 hours of incubation with neutrophils, fungal damage was measured by XTT. While we observed no significant difference between damage of wild type and ΔcreA germlings, or germlings grown in repressing versus de-repressing conditions, all strains, including ΔcreA, are very sensitive to neutrophil damage under these conditions (Fig 8C). CreA is required for adaptation to changing oxygen environments. A) Ratio of hypoxia to normoxia growth of fungal colonies grown in normoxia for 24 hours, then shifted to hypoxia (0.2% O 2 , 5% CO 2 ) for 96 hours. *p<0.05; **p<0.01 by unpaired, two-tailed t-test compared to WT of respective time points. Data represents mean of biological triplicates ± SEM. B) Host survival curve of wild type, ΔcreA and creA R in a chemotherapy model of IPA. Mice were treated with 175mg/kg Cyclophosphamide on day -2 and 40mg/kg Kenalog on day -1 then inoculated with 1x10 6 conidia in 40uL PBS intranasally. n = 12/group, n = 4 for mock. C) Percent fungal damage by bone marrow derived neutrophils of germlings grown in 1% glucose (GMM) or 1% Tween-80 minimal media. Germlings were grown in normoxia conditions, then shifted to hypoxia (0.2% O 2 , 5% CO 2 ) upon addition of neutrophils for 2 hours. No significant differences were observed between any group by Wilcoxon rank-sum test. https://doi.org/10.1371/journal.ppat.1006340.g008 Carbon catabolite repression is required for fungal virulence As our data suggest infection site hypoxia plays an important role in disease progression and a requirement for CreA activity, we utilized the observation that loss of leukocytes reduces the severity of oxygen depletion at the site of fungal infection [10]. Consistent with our data, in this chemotherapy model of IPA, where mice are largely leukopenic, inoculation with ΔcreA results in 100% mortality by 4 dpi, while wild type results in 100% mortality by 3 dpi (Fig 8B). Therefore, in a host environment with reduced hypoxic and immune cell mediated stress, ΔcreA is more fit, and produces mortality with near wild type kinetics.
Together, these data suggest that fungal cells are responding to the infection site microenvironment in part through CreA activation which is essential for supporting fungal bioenergetics in the face of oxygen depletion, the dynamic nutrient environment and host immune cell interactions. Therefore, we conclude that CreA represents a novel class of fungal virulence attributes that are dispensable for the initiation of infection but required in established infection microenvironments, which we term disease progression factors (DPFs). Further interrogation of CreA direct genes is thus likely yield additional DPFs in future studies.

Discussion
Spatial temporal mechanisms of filamentous fungal pathogenesis are under studied. These mechanisms have underappreciated clinical relevance as current antimicrobial therapies largely must thwart pathogens in complex established infection microenvironments to ameliorate disease progression. Yet, in animal models of invasive fungal infections, an emphasis is placed on survival curves with fungal mutants that exhibit marked virulence attenuation from the initiation of infection. Consequently, it is unclear to what extent fungal virulence factors identified from animal model studies mediate infection maintenance and disease progression. In this regard, discovery and critical analysis of regulatory mechanisms essential for fungal fitness and persistence after the initiation of infection and host damage is a promising approach to identify new virulence factors and consequently new and perhaps more clinically relevant drug targets.
Here, we propose that metabolic flexibility allows an environmental microbe, A. fumigatus, to navigate infection site dynamics and allow disease progression. Our model proposes that during an invasive fungal infection there are at least two distinct fungal metabolic phases driven by changes in nutrient and oxygen availability. For full pathogenic potential and disease progression, cells must undergo metabolic reprogramming to adapt to the changing microenvironments to support proliferation and continued host damage. Our model is not the first to consider this type of adaptation to changing infection dynamics. Saville et al. (2003, [30]) used an engineered C. albicans strain to show that yeast-locked cells could initiate infection, as evidenced by the ability to disseminate to target organs, however, the transition to hyphal morphology was required to cause disease, as measured by murine survival. This study elegantly demonstrates that distinct fungal morphologies are required during different stages of infection and that a morphological transition is essential for C. albicans virulence.
Our data suggest that early in infection, the steroid treated airway microenvironment contains sufficient gluconeogenic carbon sources and oxygen for fungal conidia germination and growth (Fig 9). In this environment, CCR as mediated by CreA is dispensable. As the infection and disease progresses, carbon source and oxygen availability changes drive the fungal response toward a more glycolytic and hypoxia based metabolism. In this established infection microenvironment, CCR as mediated by CreA significantly contributes to virulence and disease progression. CreA contributes to disease by modulating carbon metabolism and loss of this regulator results in persistent activity of the glyoxylate shunt, which causes metabolic dysregulation in the presences of repressing carbon sources [31]. Importantly, the creA-mutant is unable to thrive in the low oxygen environment within the host, as demonstrated through the increase in mRNA levels of the hypoxia response transcription factors, SrbA and SrbB, and the decrease in growth rate upon shift to hypoxia.
A critical consequence of hypoxia in eukaryotic cells is the increase in reducing equivalents. Consequently, the increase in these hypoxia associated gene transcript levels in vivo in the absence of CreA strongly suggests a role for CreA in mitigating reductive stress that occurs in where upon infection initiation, the presence of oxygen and alternative carbon sources allows for gluconeogenesis and oxidative phosphorylation. However increased fungal growth and influx of host immune cells results in depletion of local oxygen concentration, shifting metabolism towards glycolytic fermentative metabolism. Fungal metabolic adaptation is required for progression to invasive disease, and this requires CreA to mediate disease progression. We propose the concept of disease progression factors (DPFs) as factors required to navigate the dynamic microenvironments that occur during infection and disease progression. these conditions. Although the signals for SREBP activation in mammalian systems are well characterized and depend on cellular sterol levels sensed by the sterol sensor, SCAP [32], SREBP activation mechanism(s) in A. fumigatus remain unknown [29,33]. The lack of a SCAP homolog in A. fumigatus, suggests a distinct mechanism of activation from mammals. An intriguing possibility is regulation of this transcriptional network in response to carbohydrate metabolism and cellular redox status. Support for this hypothesis comes from the observed role of SrbA and SrbB in carbohydrate metabolism and the tight connections between hypoxia and central carbon metabolism in many organisms [34,35].
Notably, CreA homologs are not required for virulence in the pathogenic yeasts C. albicans and C. neoformans as measured by murine mortality [16,17]. There are several potential reasons for the contrasting results between the yeast systems and our results which highlight differences between the biology and pathogenesis of yeasts and filamentous fungi. First, the models used to test the virulence of the C. albicans and C. neoformans mig1-null mutants are significantly different than IPA models. Neither yeast study uses a model that involves the use of corticosteroid immune suppression, which we observed to dramatically alter the metabolite and oxygen landscape of the lungs. Furthermore, C. albicans murine models often use tail vein delivery of inocula, where yeasts enter the blood, an environment that is very different from the airways in terms of available nutrients and structural landscapes. In particular, the airways contain far less glucose compared to the bloodstream [36][37][38]. Beyond the models utilized in each study, some observed phenotypes between yeast Mig1 mutants and our CreA mutant are also different, underscoring differences in basal metabolism between these fungi. In C. albicans, the loss of Mig1 does not affect growth rate in glucose, galactose or glycerol containing media [16] and similarly, growth of C. neoformans Δmig1 shows wild type growth on solid YPD and YES media [17]. Thus, although some observed stress phenotypes between A. fumigatus and C. neoformans are similar, the loss of Mig1/CreA has different effects on fungal metabolism across species. For example, in some filamentous fungi, loss of CCR leads to lethality. For example, clean full gene replacements of the CCR transcriptional regulator CreA has not been possible in Fusarium oxysporum [39], Penicillium chrysogenum [40] and Colletotrichum gloeosporoides [41] and results in extremely severe growth defects in Aspergillus nidulans [42,43]. Consequently, the ability to generate a creA null mutant in A. fumigatus and its subsequent phenotypes on de-repressing carbon sources further highlights the diversity of metabolism and bioenergetics regulation in fungi. Whether yeast and other filamentous fungal CCR homologs are DPFs would require additional experiments including assessment of disease in relevant immune compromised animal models.
Although CreA homologs have been shown to be dispensable for virulence in pathogenic yeasts [16,17], alternative regulation of metabolism through catabolite inactivation, the proteolytic degradation of metabolic enzymes in response to glucose, is important for growth of C. albicans in vivo. C. albicans isocitrate lyase (ICL) is subject to transcriptional regulation in response to glucose [44], however, contrary to S. cerevisiae, ICL is not subject to catabolite inactivation, allowing simultaneous assimilation of glucose and lactate [45]. Consequently, the ability to simultaneously assimilate multiple carbon sources results in higher fungal burden in kidneys during disseminated infection and in feces and kidneys in a gastrointestinal colonization model [46]. Intriguingly, our results suggest that like C. albicans, A. fumigatus can integrate both lactate and glucose simultaneously, as inferred from the ability of A. fumigatus to grow on lactate or ethanol in the presence of 2-DG, suggesting that this fungus is metabolically plastic, an attribute which likely contributes to spatial temporal fitness within the host and ultimately disease maintenance and progression. Therefore, a further understanding of the divergence of the CCR pathway across fungi can yield insights into the mechanistic differences of pathogenicity and virulence across pathogenic fungi for a better understanding of how to treat these infections.
Lastly, we argue that further studies to identify DPFs in pathogenic microbes will identify novel therapeutic targets distinct from existing virulence factors. Though existing virulence factors may certainly be critical for disease progression throughout the course of infection, our data strongly suggest additional factors that may not always be evident from traditional animal model survival curve studies exist and remain to be identified. Moreover, if the metabolic state of the fungus or microbe can be precisely identified during disease progression, in vitro systems that induce the same metabolic state are expected to enhance antifungal drug discovery. Moving forward pathogenesis studies should consider the spatial temporal dynamics of infections and leverage developing tools to probe fungal and microbial gene function in established infection microenvironments in vivo for high impact therapeutic target identification [47,48].

Strain construction
CreA-null mutants were generated by replacing~1 kb of the creA coding sequence (~1.2 kb) with A. parasiticus pyrG in the CEA17 background. The replacement construct was generated using overlap extension PCR [49] to join~1 kb of creA 5' and 3' UTR with pyrG. The resulting construct was used for transformation of protoplasts for selection on media without uracil and uridine supplements. Reconstitution of creA in the null mutant was achieved by amplification of the entire creA coding sequence including~1 kb of the flanking 5' and 3' regions and the dominant selectable marker ptrA, which confers resistance to pyrithiamine. The creA coding sequence and ptrA were joined with overlap extension PCR, and the linear construct was used to transform ΔcreA for ectopic expression. The reconstituted strain is denoted as creA R .
Protoplast transformations were carried out as previously described [10,29]. Briefly, 5-10ug of construct was incubated on ice with protoplasts generated with Lysing Enzyme from Trichoderma harzianum (Sigma). PEG/CaCl 2 solution was added to protoplasts, then incubated at room temperature (RT). Protoplasts were plated on sorbitol stabilized media plates embedded in sorbitol stabilized media agar. Transformants were selected on appropriate media, and screened using PCR to determine if correct integration of the creA replacement construct with primers designed to bind within the pyrG sequence and outside the integrated construct. Single spores were isolated from positive colonies, and correct integration was confirmed using Southern blotting with the digoxigenin-anti-digoxigenin system (Roche Diagnostics), as previously described [10,29].

Growth assays
For alternative carbon and nitrogen plates, the indicated nutrients were added to our minimal media base (nitrogen-free trace elements and salt solution) then plates were inoculated with 1x10 3 conidia and incubated at 37˚C for 72 hours. Allyl alcohol (Sigma) experiments were completed using 1% agar, which was allowed to cool substantially before addition of allyl alcohol to prevent excess evaporation. 1x10 3 conidia were spotted on plates and incubated at 37˚C for 48 hours. Control plates were incubated in a separate incubator, under the same conditions, to prevent inhibition of growth by volatile allyl alcohol. Mitochondrial inhibitors were used in the following concentrations: 0.1mM thenoyltrifluoroacetone (TTFA; Sigma), 15ug/ mL Antimycin A (Sigma), 10uM Oligomycin A (Sigma), or 5mM Salicylhydroxamic acid (SHAM; Sigma) in GMM or 1% ethanol minimal media (EMM) agar. Plates were inoculated with 1x10 3 conidia and incubated at 37˚C for 72 hours. Percent inhibition was calculated by measuring radial growth of colonies in the presence of drug, compared to GMM/EMM plates without drug in biological triplicate. All growth assays were completed in biological triplicates.
Liquid growth assays were performed with conidia adjusted to 2x10 4 conidia in 20uL 0.01% Tween-80 in 96-well dishes, then 180uL of minimal media with indicated carbon source was added to each well. Plates were incubated at 37˚C for 7 hours, then Abs 405 measurements were taken every 15 minutes for the first 24-36 hours of growth with continued incubation at 37˚C. Lung homogenate media was prepared as follows: lungs were harvested from healthy CD-1 female mice (20-24g) and homogenized through a 100uM cell strainer in 2mL PBS/lung. Homogenate was diluted 1:4 in sterile PBS, then filter sterilized through 22uM PVDF filters. Each curve represents six technical replicates. ΔAbs405/min was calculated using the linear regression of each individual technical replicate.
Hypoxia shift experiments were performed by inoculation of 1x10 3 conidia on GMM plates, which were incubated at 37˚C for 24 hours in normoxia (21% O 2 , 5%CO 2 ) for 24 hours then moved to hypoxia (0.2% O2, 5% CO2; INVOVO2 400 hypoxia chamber [The Baker Company, Sanford, ME]) for 4 days. Radial growth was measured every 24 hours and compared to normoxia controls. Growth is presented as a ratio of hypoxia colony diameter to normoxia colony diameter for three biological replicates.

ADP/ATP ratio measurement
ADP/ATP ratios were measured from cultures (1x10 6 conidia/mL) grown in liquid GMM at 37˚C for 16 hours with shaking at 250rpm. Fungal tissue was washed with sterile water, and transferred to fresh GMM for 2 additional hours at 37˚C with shaking at 250rpm. Tissue was lyophilized, ground with glass beads with a Mini-BeadBeater (BioSpec Products, Inc.). Metabolites were extracted using buffer from ApoSENSOR ADP/ATP Ratio Bioluminescent Assay Kit (K255-200, BioVision) and quantified following manufacturers protocol. Tissue lysates were filtered through Ultracel 10K centrifugal filters (Millipore) prior to processing to remove proteins.

Isolation of total RNA
Cultures (1x10 6 conidia/mL) were grown in GMM at 37˚C for 16 hours with shaking at 250rpm. Fungal tissue was washed with sterile water, and transferred to fresh media (GMM or EMM) for 2 additional hours at 37˚C with shaking at 200rpm. Mycelia were collected with vacuum filtration and immediately frozen with liquid nitrogen. Tissue was lyophilized, then ground with glass beads with a Mini-BeadBeater (BioSpec Products, Inc). RNA was extracted in 1mL of TriSure (Bioline Reagents), then 200uL of chloroform was added, and the aqueous phase was collected after centrifugation. The aqueous phase was in an equal volume of 80% ethanol, then eluted using RNeasy columns (Qiagen) following manufacturer's instructions.
Quantitative real-time PCR 5ug of RNA was treated with Ambion Turbo DNAse (Life Technologies) according to manufacturer's protocol. 500ng of DNAse treated RNA was used for cDNA synthesis with Quanti-Tech Reverse Transcription kit (Qiagen), then cDNA was diluted 1:5 with ddH 2 O. qRT analysis was performed in 20uL reactions using 2uL of dilute cDNA per reaction with iQ SYBR Green Supermix (BioRad). Gene expression was measured with a BioRad MyiQ Real Time PCR Detection System. For all expression studies, gene expression was normalized to actA and tub2 expression.
RNA-sequencing RNA, collected as described above, was analyzed by Qubit and a Fragment Analyzer (Advanced Analytical) for quality control. cDNA libraries were prepared with Illumina's Tru-Seq Directional polyA+ library prep kit following the manufacturer's protocol, with TruSeqLT sequencing adapters/indices. cDNA libraries were multiplexed and loaded at 1.6pM on a Next-Seq500 for sequencing. The Illumina raw RNA-seq reads for each replicate were downloaded and then aligned to the A. fumigatus A1163 genome (CADRE 30) using Tophat v2.1.0 [50] under default parameter settings. Transcripts were then assembled using Cufflinks v2.2.1 [51] and annotated using the A1163 (CADRE 30) GTF file from ENSEMBLFungi. Differential expression analysis was performed for each experimental group separately using DESeq2 [52] and p-values were corrected for multiple testing using Benjamini-Hochberg. NCBI GEO accession number pending.

Metabolomics sample preparation
Spores were inoculated in GMM at a concentration of 10 6 conidia/mL and incubated for 16 hours at 37˚C with shaking at 250 rpm, then collected with vacuum filtration and washed with sterile water. Fungal tissue was transferred to fresh GMM and further incubated at 37˚C, with shaking at 200rpm, for 2 hours. Mycelia were collected with vacuum filtration, washed with sterile water and frozen in liquid nitrogen and stored at -80˚C until samples were sent for processing and analysis. Murine metabolomics samples were prepared from whole lungs. CD-1 female mice, 20-24 grams, were administered 40mg/kg Kenalog-10 (Bristol-Myers Squibb) subcutaneously. Lungs of steroid-treated mice were collected 24 hours post injection. For fungal inoculation, immune suppressed mice were inoculated with 2x10 6 CEA10 conidia in 40uL PBS and lungs were collected 8 hours post inoculation. Healthy controls were given no drug. All lung samples were immediately frozen in liquid nitrogen and stored at -80˚C until processing.

Global metabolomic profiling
Samples were prepared using the automated MicroLab STAR system from Hamilton Company. Several recovery standards were added prior to the first step in the extraction process for QC purposes. To remove protein, dissociate small molecules bound to protein or trapped in the precipitated protein matrix, and to recover chemically diverse metabolites, proteins were precipitated with methanol under vigorous shaking for 2 min (Glen Mills GenoGrinder 2000) followed by centrifugation. The resulting extract was divided into five fractions: two for analysis by two separate reverse phase (RP)/UPLC-MS/MS methods with positive ion mode electrospray ionization (ESI), one for analysis by RP/UPLC-MS/MS with negative ion mode ESI, one for analysis by HILIC/UPLC-MS/MS with negative ion mode ESI, and one sample was reserved for backup. Samples were placed briefly on a TurboVap (Zymark) to remove the organic solvent. The sample extracts were stored overnight under nitrogen before preparation for analysis.
All methods utilized a Waters ACQUITY ultra-performance liquid chromatography (UPLC) and a Thermo Scientific Q-Exactive high resolution/accurate mass spectrometer interfaced with a heated electrospray ionization (HESI-II) source and Orbitrap mass analyzer operated at 35,000 mass resolution. The sample extract was dried then reconstituted in solvents compatible to each of the four methods. Each reconstitution solvent contained a series of standards at fixed concentrations to ensure injection and chromatographic consistency. One aliquot was analyzed using acidic positive ion conditions, chromatographically optimized for more hydrophilic compounds. In this method, the extract was gradient eluted from a C18 column (Waters UPLC BEH C18-2.1x100 mm, 1.7 μm) using water and methanol, containing 0.05% perfluoropentanoic acid (PFPA) and 0.1% formic acid (FA). Another aliquot was also analyzed using acidic positive ion conditions, however it was chromatographically optimized for more hydrophobic compounds. In this method, the extract was gradient eluted from the same afore mentioned C18 column using methanol, acetonitrile, water, 0.05% PFPA and 0.01% FA and was operated at an overall higher organic content. Another aliquot was analyzed using basic negative ion optimized conditions using a separate dedicated C18 column. The basic extracts were gradient eluted from the column using methanol and water, however with 6.5mM Ammonium Bicarbonate at pH 8. The fourth aliquot was analyzed via negative ionization following elution from a HILIC column (Waters UPLC BEH Amide 2.1x150 mm, 1.7 μm) using a gradient consisting of water and acetonitrile with 10mM Ammonium Formate, pH 10.8. The MS analysis alternated between MS and datadependent MS n scans using dynamic exclusion. The scan range varied slighted between methods but covered 70-1000 m/z.
Peaks were quantified using area-under-the-curve. For studies spanning multiple days, a data normalization step was performed to correct variation resulting from instrument interday tuning differences. Essentially, each compound was corrected in run-day blocks by registering the medians to equal one (1.00) and normalizing each data point proportionately. For studies that did not require more than one day of analysis, no normalization is necessary, other than for purposes of data visualization. 'Scaled Input' values represent the data rescaled with wild type values set to have a median of one, for ease of visualization. For in vitro fungal samples, metabolites were normalized to protein content as determined by a Bradford assay. In vivo lung samples represent metabolite content of whole lung without normalization.

Fungal burden
Outbred female CD-1 mice (Charles River Laboratory, Raleigh, NC), 20-24 grams, were given subcutaneous injections of Kenalog-10 (triamcinolone acetonide, Bristol-Myer Squibb) at 40mg/kg to induce immune-suppression. 24 hours post injection; mice were inoculated with 2x10 6 fungal spores in 40uL PBS via intranasal inoculation. Mock mice were given 40uL sterile PBS. Mice were sacrificed 48 hours post inoculation and lungs were collected for DNA extraction. Lung tissue was freeze-dried and homogenized with 2.3mm zirconia/silica beads on a Mini-BeadBeater (BioSpec Products, Inc.). Total DNA was extracted using a phenol-chloroform extraction and RNAse treated DNA was used for quantitative PCR, as previously described [24], with primers to amplify 18S region.

Murine virulence assays
Outbred female mice were immune-suppressed as described above. Inocula of 2x10 6 CEA10, ΔcreA, creA R conidia were prepared in 40uL PBS and delivered via intranasal inoculation. Mock mice were given 40uL PBS in the absence of fungal spores. For the chemotherapeutic murine model outbred CD-1 female mice, 6 weeks old, were immunosuppressed with intraperitoneal (i.p.) injections of 175mg/kg cyclophosphamide (Baxter Healthcare Corporation) 48 hours before fungal inoculation and subcutaneous (s.c.) injections of 40 mg/kg Kenalog-10 (triamcinolone acetonide, Bristol-Myer Squibb24 hours before fungal inoculation. Mice were inoculated with 1x10 6 conidia in 40uL PBS intranasally or PBS alone for mock. n = 12 mice/ group; n = 4 mice/mock. Mice were housed 4 per cage and had access to food and water ad libitum. Mice were monitored for 14 days following challenge with A. fumigatus. Percent survival was plotted on a Kaplan-Meier curve and a Log-rank test was used to assess statistical significance of the curves. Histopathology CD-1 female mice, 20-24 grams, were immune suppressed as described above. Mice were inoculated with 2x10 6 conidia of CEA10, ΔcreA, or creA R and lungs were harvested 48 hours post inoculation for histopatholoical sectioning and staining. For early germination studies, immune suppressed mice were inoculated with 1x10 7 conidia and lungs were harvested 8 hpi for sectioning and staining. Briefly, lungs were perfused with 10% buffered formalin solution upon collection, then fixed in 10% buffered formalin overnight. Lungs were blocked in paraffin, sectioned, and stained with Gömöri methenamine silver (GMS) and hematoxylin and eosin (H&E) stains. Images were obtained with a Zeiss Axioplan 2 imaging microscope (Carl Zeiss Microimaging, Inc.) fitted with Qimiging RETIGA-SRV Fast 1394 RGB camera using Phylum Live 4 imaging software.

In vivo gene expression analysis
Outbred female mice were immune-suppressed as described above and given 5x10 6 CEA10 or ΔcreA conidia intranasally in 40uL PBS, 24 hours post injection with steroids. Lungs were harvested 24 and 72 hours post inoculation, and immediately frozen in liquid nitrogen. Lungs were freeze dried and homogenized with 2.3mm zirconia/silica beads on a Mini-BeadBeater (BioSpec Products, Inc.). RNA was extracted using TriSure (Bioline Reagents) then eluted using RNeasy columns (Qiagen) following manufacturer's instructions. cDNA was synthesized with Qiagen Quantitect Reverse Transcription kit (Qiagen) using 500ng of DNAse treated RNA, and 25ng of resulting cDNA was used in each qRT-PCR reaction. Gene expression was measured with qRT-PCR using BioRad iQ SYBR Green Supermix with a BioRad MyiQ Real Time PCR Detection System (BioRad). To ensure our primers amplified A. fumigatus genes specifically, we used cDNA from mock mice that had been treated with steroids but given PBS in lieu of fungal spores. All primer pairs tested failed to amplify from mock cDNA.
Neutrophil killing assay BMDN isolation. Bone-marrow derived neutrophils (BMDNs) were isolated from tibias and femurs of CD-1 females, 6 weeks of age, and cultured for neutrophils in murine neutrophil buffer (HBSS containing 0.1% FBS and 1% glucose). BMDNs were suspended in 3 ml 45% Percoll (GE Healthcare) and isolated from a 30 min 1600x g Percoll gradient in a Sorvall Legend Mach 1.6R benchtop centrifuge, with a BIOshield 600 rotor-75002005 (Thermo Scientific). BMDNs were collected and washed with HBSS before counting as previously described [53].
Fungal cultures of 1x10 5 conidia were grown in 1mL 1% glucose or 1% Tween-80 minimal media until all strains were germinated to the same extent (time adjusted for slow growth of ΔcreA in glucose media) in 24-well plates. Media was removed and 1x10 6 bone marrow derived neutrophils were added to each well in 1mL RPMI, then cultures were shifted to hypoxia (0.2% O 2 , 5% CO 2 ) at 37˚C for 2 hours. Fungus only (no neutrophils) samples were grown simultaneously for normalization of fungal damage. Following incubation, cells were cold water lysed and the remaining live germlings were quantitated by measuring the OD at 450 nm following a 1 h incubation with 300 μL/well of 0.4 mg/ml XTT (2,3-bis(2-methoxy-4-nitro-5[(sulphenylamino)carbonyl)]-2H-tetrazolium-hydroxide) solution with 50 μg/ml of the electron-coupling agent coenzyme Q. The percent fungal damage was defined by the equation: (1-[A450 of fungi with BMDNs-A450 of BMDNs alone]/[A450 of fungi alone])/100 [54].

Ethics statement
The Guide for the Care and Use of Laboratory Animals of the National Research Council was strictly followed for all animal experiments. The animal experiment protocols were approved by Institutional Animal Care and Use Committee at Dartmouth College (protocol: cram.ra.1).  (ScMig1p). Grey color indicates percent identity across all species, with darker color indicative of higher identity. Alignment was created using Clustal Omega [EMBL-EBI, [55]]. Alignment image was generated using Jalview [56]. creA R B shows two insertion sites, one which inserted back at the creA locus (recombination at 5' UTR), resulting in loss of the 6174bp band. This strain is used for all subsequent experiments unless otherwise noted, however, both creA R strains reconstitute all growth phenotypes tested. (TIF)

S3 Fig. Loss of CreA results in increased resistance to 2-DG on alternative carbon sources.
A) Growth of strains on 1% lactate minimal media with or without 0.1% allyl alcohol (AA) and incubated for 48 hours. b) Growth inhibition of CEA10, ΔcreA and creA R on indicated carbon source with 0.1% 2-deoxyglucose (2-DG) for 72 hours. Data represents mean of biological triplicates ± SEM; ÃÃÃ p<0.0001 by unpaired, two-tailed t-test as compared to WT of respective condition. creA R strains used for this experiment is creA R A. All assays use 1x10 3 spore dilutions incubated at 37˚C.