An Alanine Aminotransferase is Required for Polysaccharide Regulation and Resistance of Aspergillus fumigatus Biofilms to Echinocandin Treatment

Alanine metabolism has been suggested as an adaptation strategy to oxygen limitation in organisms ranging from plants to mammals. Within the pulmonary infection microenvironment A. fumigatus forms biofilms with steep oxygen gradients defined by regions of oxygen limitation. A significant increase in alanine levels was observed in A. fumigatus cultured under oxygen limiting conditions. An alanine aminotransferase, AlaA, was observed to function in alanine catabolism and is required for several aspects of A. fumigatus biofilm physiology. Loss of alaA, or its catalytic activity, results in decreased adherence of biofilms through a defect in the maturation of the extracellular matrix polysaccharide galactosaminogalactan (GAG). Additionally, exposure of cell wall polysaccharides is also impacted by loss of alaA and loss of AlaA catalytic activity confers increased biofilm susceptibility to echinocandin treatment which is correlated with enhanced fungicidal activity. The increase in echinocandin susceptibility is specific to biofilms and chemical inhibition of alaA by the alanine aminotransferase inhibitor β-chloro-L-alanine is sufficient to sensitize A. fumigatus biofilms to echinocandin treatment. Finally, loss of alaA increases susceptibility of A. fumigatus to in vivo echinocandin treatment in a murine model of invasive pulmonary aspergillosis. Our results provide insight into the interplay of metabolism, biofilm formation, and antifungal drug resistance in A. fumigatus and describes a mechanism of increasing susceptibility of A. fumigatus biofilms to the echinocandin class of antifungal drugs. eLife Digest Aspergillus fumigatus is a ubiquitous filamentous fungus that causes an array of diseases depending on the immune status of an individual, collectively termed aspergillosis. Antifungal therapy for invasive pulmonary aspergillosis (IPA) or chronic pulmonary aspergillosis (CPA) is limited and too often ineffective. This is in part due to A. fumigatus biofilm formation within the infection environment and the resulting emergent properties, particularly increased antifungal resistance. Thus, insights into biofilm formation and mechanisms driving increased antifungal drug resistance are critical for improving existing therapeutic strategies and development of novel antifungals. In this work, we describe an unexpected observation where alanine metabolism, via the alanine aminotransferase AlaA, is required for several aspects of A. fumigatus biofilm physiology including resistance of A. fumigatus biofilms to the echinocandin class of antifungal drugs. Importantly, we observed that chemical inhibition of alanine aminotransferases is sufficient to increase echinocandin susceptibility and that loss of alaA increases susceptibility to echinocandin treatment in a murine model of IPA.

fumigatus forms biofilms with steep oxygen gradients defined by regions of oxygen limitation. A 23 significant increase in alanine levels was observed in A. fumigatus cultured under oxygen limiting 24 conditions. An alanine aminotransferase, AlaA, was observed to function in alanine catabolism 25 and is required for several aspects of A. fumigatus biofilm physiology. Loss of alaA, or its catalytic 26 activity, results in decreased adherence of biofilms through a defect in the maturation of the 27 extracellular matrix polysaccharide galactosaminogalactan (GAG). Additionally, exposure of cell 28 wall polysaccharides is also impacted by loss of alaA and loss of AlaA catalytic activity confers 29 increased biofilm susceptibility to echinocandin treatment which is correlated with enhanced 30 fungicidal activity. The increase in echinocandin susceptibility is specific to biofilms and chemical 31 inhibition of alaA by the alanine aminotransferase inhibitor β-chloro-L-alanine is sufficient to 32 sensitize A. fumigatus biofilms to echinocandin treatment. Finally, loss of alaA increases 33 susceptibility of A. fumigatus to in vivo echinocandin treatment in a murine model of invasive 34 pulmonary aspergillosis. Our results provide insight into the interplay of metabolism, biofilm 35 formation, and antifungal drug resistance in A. fumigatus and describes a mechanism of 36 increasing susceptibility of A. fumigatus biofilms to the echinocandin class of antifungal drugs. 37

Introduction 52
Aspergillus fumigatus is a ubiquitous filamentous fungus with a prominent ecological role 53 in the decomposition of organic carbon, that is easily isolated from compost piles and similar 54 environments (Gugnani, 2003). Within compost piles a complex set of microenvironments can 55 emerge along temperature and nutrient gradients that naturally form as saprophytes become been associated with adaptation to oxygen limitation in numerous organisms ranging from plant 108 roots adapting to waterlogging (Lothier et al., 2020;Rocha et al., 2010) to exercise induced 109 oxygen deprivation in muscle cells (Felig, 1973). Importantly, alanine is also one of the handful of of more favorable nitrogen sources, as is seen with carbon metabolism described above, and/or 145 potentially indicative of nitrate fermentation as a strategy to recycle reducing potentials and allow 146 glycolysis to continue. The accumulation of alanine was of particular interest due to an association of 147 alanine and low oxygen adaptation in a wide-range of organisms including plants (Lothier et al., 2020; 8 Rocha et al., 2010), crustaceans (Harrison, 2015), flies (Feala et al., 2007), and mammals (Felig et al., 149 1970). Additionally, several transcriptomic studies involving A. fumigatus and oxygen limitation have 150 found that an alanine aminotransferase (Afu6g07770/AFUB_073730) is highly increased in mRNA 151 abundance upon exposure to low oxygen conditions ( Figure 1B)  The alanine aminotransferase, alaA, is required for efficient catabolism of L-alanine. 156 To assess the role of alaA in A. fumigatus metabolism and stress resistance, strains lacking the 157 alaA gene were generated in the Af293 (Af293ΔalaA) and CEA10 (CEA10ΔalaA) backgrounds along with 158 respective reconstituted strains in which alaA was ectopically re-introduced into the Af293ΔalaA and 159 CEA10ΔalaA genomes under control of its native promoter (Af293alaA rec and CEA10alaA rec ). Both 160 Af293ΔalaA and CEA10ΔalaA colony biofilms grew on glucose minimal media (GMM), where glucose is 161 the sole carbon source and nitrate is the sole nitrogen source, indicating that sufficient alanine was 162 generated for colony biofilm growth independent of alaA under these in vitro conditions ( Figure 1C, Figure  163 1-figure supplement 2A). However, radial growth of the alaA null strains on solid GMM was approximately 164 10% less than that of their respective WT and reconstituted strains at both ambient O2 and 0.2% O2 165 indicating a role for this protein in fungal metabolism in the presence of its preferred carbon source when 166 growing as a colony biofilm ( Figure 1C, Figure 1-figure supplement 2A). When the alaA null strains were 167 grown with L-alanine as the sole carbon or sole nitrogen source the strains displayed severe colony 168 biofilm growth defects ( Figure 1C, Figure 1-figure supplement 2A). Surprisingly, both the wildtype (WT) 169 and the alaA null strains grew more robustly at 0.2% O2 than ambient O2 when alanine was the sole 170 carbon or nitrogen source, despite alanine being a non-fermentable carbon source ( Figure 1C). Thus, 171 alaA plays an important role in A. fumigatus metabolism in multiple carbon, nitrogen, and oxygen 172 environments. 173 Figure 1: An alanine aminotransferase is required for alanine catabolism and normal biofilm physiology. A) Significantly changed amino acids upon acute exposure to a 0.2% oxygen environment. Ion counts were normalized to the mean ion count for each metabolite across all samples and log2 transformed. Each column is a replicate (n = 5 per condition). B) Reaction catalyzed by AlaA and its position in central carbon and nitrogen metabolism. Each metabolite and gene are color-coded according to their relative abundance upon exposure to a 0.2% oxygen environment. Metabolite data was obtained from the experiment in (A), and RNA-sequencing data was obtained from Chung, et al. 2014. C) Growth of Af293ΔalaA on minimal media containing the indicated sole carbon and nitrogen sources in ambient oxygen (normoxia) and 0.2% oxygen environments. Images are representative of four replicate cultures. D) Dry biomass of biofilms grown in minimal media containing the indicated sole carbon and nitrogen sources for 24 hours (n = 4). Each replicate is shown along with the mean +/-SD. E) Representative static growth assay of Af293ΔalaA over 24 hours of biofilm growth (n = 6 technical replicates). Experiment was repeated at least three times with similar results. F) Crystal violet adherence assay of biofilms grown for 12, 18, 24, and 30 hours (n = 3). G) Oxygen concentration as a function of distance from the air-liquid interface in 24-hour biofilms (n ≥ 7). Culture volumes are approximately 3000µm in depth. ** p < 0.01, *** p < 0.001, **** p < 0.0001, n.s. = not significant by either Two-Way ANOVA with a Tukey's multiple comparison test (D, F, and G) or One-Way ANOVA with a Tukey's multiple comparison test (E). All graphs show the mean +/-SD unless otherwise stated.

Robust adherence and growth of A. fumigatus biofilms is dependent on alaA. 175
A. fumigatus submerged and colony biofilms naturally become increasingly oxygen deprived as 176 they mature, albeit to different degrees where submerged biofilms form steeper oxygen gradients than 177 colony biofilms (Kowalski et al., 2020(Kowalski et al., , 2021. We next investigated the role of alaA in A. fumigatus 178 submerged biofilms. To assess biofilm formation on GMM, and further quantify the role of AlaA in alanine 179 metabolism, we quantified the dry biomass of submerged biofilms grown for 24 hours in GMM and with 180 alanine as a sole carbon or nitrogen source. Loss of alaA resulted in a 40-50% decrease in submerged 181 biofilm biomass in GMM. This growth defect was exacerbated when alanine was the sole carbon or 182 nitrogen source, with no biomass recovered when alanine was the sole nitrogen source ( Figure 1D). 183 Additionally, we utilized a static growth assay to assess biofilm growth kinetics, which revealed that alaA 184 null strains had a longer lag phase than their respective WT or reconstituted strains, indicative of a delay 185 in conidial germination ( Figure 1E, Figure 1-figure supplement 2B). To further determine if alaA had broad 186 physiological impacts on A. fumigatus submerged biofilm formation, a crystal violet adherence assay was 187 utilized to quantify adherence of the alaA null biofilms to abiotic surfaces. To account for any impacts of 188 the germination delay on biofilm formation, the adherence of Af293ΔalaA was measured over a time 189 course from an immature biofilm at 12 hours to a highly mature biofilm at 30 hours. At all timepoints after 190 12 hours Af293ΔalaA had a severe defect in adherence compared to the WT and reconstituted strains 191 ( Figure 1F). CEA10ΔalaA was also tested for adherence and showed a similar inability to strongly adhere 192 to surfaces (Figure 1-figure supplement 2C). Finally, we quantified oxygen levels within 24-hour biofilm 193 cultures of Af293ΔalaA. The alaA null strain cultures were significantly more oxygenated than the WT 194 and reconstituted strains' biofilms ( Figure 1G). However, the portion of the culture containing the bulk of 195 the biofilm's biomass, depth ~2000μm -3000μm based on previous microscopy studies ( Kowalski et al., 196 2020), was still below 5% O 2 and thereby experiencing hypoxia. Therefore, while the loss of alaA has an 197 impact on colony biofilm growth, alaA appears to play a greater role in A. fumigatus submerged biofilm 198 physiology where steep oxygen gradients naturally occur (Kowalski et al., 2020). 199 Catalytic activity of AlaA is required for adherence and alanine growth, but not mitochondrial  to be the case in tumor cells (Beuster et al., 2011). Therefore, AlaA catalytic activity is required for 218 adherence and alanine catabolism, but not mitochondrial localization. 219 The primary adherence factor for A. fumigatus submerged biofilms studied to date is the 223 extracellular matrix polysaccharide galactosaminogalactan (GAG) (Gravelat et al., 2013;Lee et al., 2015Lee et al., , 224 2016, and the lack of adherence observed in the alaA null strain suggests a role in GAG production or 225 maturation. To examine GAG production, we first utilized a fluorescently labeled lectin specific to N-226 acetyl-D-galactosamine (GalNAc) residues found in the GAG polysaccharide, FITC-Soybean Agglutinin 227 (SBA). Biofilms of Af293, Af293ΔalaA, and Af293alaA rec were stained with SBA to visualize GAG and 228 calcofluor white, which binds chitin, to visualize biomass at 12, 18, 24, and 30 hours of growth ( Figure  229 3A, D, G). Spinning-disk confocal microscopy was utilized to image the first 300μm of the biofilms, 230 followed by quantification using BiofilmQ (Hartmann et al., 2021). As seen in growth curve experiments 231 ( Figure 1F), Af293ΔalaA had a lower biovolume at 12 and 18 hours ( Figure 3A-C). Total SBA staining of 232 the GAG polysaccharide was quantified as the sum intensity of the SBA stain in each image, revealing 233 that Af293ΔalaA biofilms had less total SBA staining than the WT and reconstituted strains starting at 18 234 hours of growth ( Figure 3D-F). While the SBA staining tightly associated with the cell wall at all timepoints 235 in Af293ΔalaA, at 30 hours in the WT and reconstituted strains the SBA staining pattern shifted from 236 hyphal associated to primarily staining the extracellular milieu ( Figure 3D, G). We quantified the hyphal 237 associated SBA staining as the sum intensity of SBA stain that overlapped with the segmented calcofluor 238 white stain, therefore showing GAG in relation only to hyphal biovolume. In the WT and reconstituted 239 strain biofilms hyphal associated SBA peaked at 18 hours and decreased at 24 and 30 hours as matrix 240 was shed from the hyphae into the extracellular milieu ( Figure 3G-I). This was in contrast to total SBA 241 staining, which remained relatively consistent from 18-30 hours of growth in the WT and reconstituted 242 strains ( Figure 3E). 243 To chemically define how GAG was being altered in the alaA null strain, monosaccharide analysis 244 of ECM polysaccharides and an enzyme-linked lectin assay (ELLA) were conducted. Monosaccharide 245 WT strain ( Figure 4A). This finding suggests that the altered ECM is primarily due to a change in the 247 maturation of the ECM polysaccharides rather than a difference in the base polysaccharides produced. 248 After GAG has been synthesized, partial deacetylation by the Agd3 deacetylase is necessary for 249 functional adherence (Bamford et al., 2020;Lee et al., 2016). To test if GAG maturation was altered, we 250 utilized an ELLA in combination with treatment of the ECM by recombinant Agd3. In principle, 251 deacetylated GAG in supernatants allows for adherence to the walls of a polystyrene plate, whereas fully 252 acetylated GAG is unable to adhere and is easily removed by washing. Adherent, and therefore 253 deacetylated, GAG can then be quantified by binding of a biotinylated SBA lectin coupled to a 254 streptavidin-conjugated horseradish peroxidase. Additionally, the presence of fully acetylated GAG can 255 be detected by pre-treating samples with recombinant Agd3, producing de-acetylated, adherent, GAG 256 that can then be detected by SBA. A strain lacking the agd3 gene, which only produces fully acetylated 257 GAG (Lee et al., 2016), was utilized as a control. The alaA and agd3 null strains both yielded low levels 258 of adherent (deacetylated) GAG compared to the WT, and this was rescued by treatment of ECM with 259 recombinant Agd3 protein ( Figure 4B). Therefore, alaA is not required for GAG production, but rather is 260 required for deacetylation and maturation of the GAG polysaccharide into its functional form. Finally, 261 mRNA abundance of uge3 and agd3 was measured from RNA isolated from 24-hour biofilms of Af293, 262 Af293ΔalaA, and Af293alaA rec to begin to distinguish if the observed differences in GAG are through a 263 transcriptional or post-transcriptional mechanism of regulation. No differences in expression of uge3 were 264 observed ( Figure 4C). While a statistically significant decrease of ~20% in agd3 mRNA levels was 265 observed ( Figure 4D), it is unclear if that level of mRNA difference could cause the degree of altered 266 GAG deacetylation observed. Thus, while loss of alaA has a modest impact on agd3 at the transcriptional 267 level, the impact of alaA on GAG maturation is likely to be primarily post-transcriptional. 268 Representative image renderings of biovolume in the first 300µm of biofilms grown for 12, 18, 24, and 30 hours. Biofilms were stained with calcofluor white and FITC-SBA followed by fixing with paraformaldehyde. Biovolume was determined by segmentation of the calcofluor white stain of each image. B) Heatmap of biovolume as a function of height from the base of the biofilm. C) Global segmented biovolume quantifications of each biofilm. D) Representative image renderings of FITC-SBA staining intensity corresponding to biomass images in (A). Renderings show FITC-SBA matrix intensity mapped onto the segmented FITC-SBA stain. E) Heatmap of FITC-SBA intensity as a function of height from the base of the biofilm. F) Sum intensity quantification of FITC-SBA staining for each biofilm. G) Representative merged image renderings of the segmented biovolume (calcofluor white), shown in blue, and segmented FITC-SBA stain shown in orange. Hyphal associated SBA staining will appear green as a result of the overlap between the two channels. SBA was considered hyphal-associated or non-hyphal associated based on overlap with the segmented biomass. H) Heatmap of hyphal associated FITC-SBA intensity as a function of height from the base of the biofilm. I) Sum intensity quantification of hyphal associated FITC-SBA staining for each biofilm. Each graph and heatmap shows the individual replicates for each timepoint (n = 6). For (C, F, and I), the line goes through the mean of each timepoint. * p < 0.05, *** p < 0.001, **** p < 0.0001, n.s. = not significant as determined by Two-Way ANOVA with a Tukey's multiple comparison's test for (C, F, and I). Deletion of alaA leads to cell wall changes and increased susceptibility of biofilms to 272

echinocandins. 273
Given that GAG maturation was substantially impacted by loss of alaA and that alaA plays a role 274 in metabolism, we asked if loss of alaA impacts additional cell wall polysaccharides. To test this 275 hypothesis, germlings of Af293, Af293ΔalaA, and Af293alaA rec were stained with calcofluor white (to 276 measure total chitin), wheat-germ agglutinin (WGA) (to measure surface exposed chitin), and soluble 277 Dectin-1 Fc (to measure surface exposed β-glucans). Curiously, Af293ΔalaA germlings had lower WGA 278 staining (exposed chitin), despite no difference in calcofluor white staining (total chitin) ( Figure 5A lower levels of total cell wall β-glucans and that alaA is required for WT cell wall organization. 284 To determine if these cell wall changes translated to functional phenotypes, biofilms of Af293, 285 Af293ΔalaA, and Af293alaA rec were tested for sensitivity to the cell wall perturbing agents calcofluor white 286 and the echinocandin class of antifungal drugs. Biofilms were grown to maturity (24 hours) prior to 287 application of cell wall stress for 3 hours. Viability was then compared to untreated biofilms via reduction 288 of the metabolic dye XTT. Af293ΔalaA biofilms were significantly more susceptible to damage by 289 calcofluor white at all concentrations tested, with greater than 90% inhibition beginning at 12.5 μg/mL 290 ( Figure 5D). In contrast, the WT and reconstituted strains maintained at least 30% viability at even the 291 highest concentration tested (100 μg/mL) ( Figure 5D). We next tested the strains for susceptibility to the 292 echinocandin caspofungin. The WT and reconstituted biofilms displayed minimal damage regardless of 293 concentration. Quantification also reveals signs of the paradoxical effect in this system, where the fungus 294 will recover growth as the concentration of drug increases beyond a MEC ( Figure 5E). Thus, similar to 295 MEC and agar colony biofilm plate assays that begin with conidia, mature WT A. fumigatus biofilms are 296 caspofungin tolerant. However, alaA null biofilms were highly susceptible to caspofungin and reached >90% inhibition at a concentration of 0.0625 μg/mL. Unlike WT biofilms, the alaA null biofilms did not 298 display evidence of a paradoxical effect, with increasing concentrations of caspofungin yielding 299 equivalent or greater damage ( Figure 5E). 300 To test if this increased susceptibility to caspofungin extended to other echinocandins, these 301 experiments were validated with another echinocandin, micafungin. Similar to caspofungin treatment, the 302 biofilms of WT and reconstituted strains were highly resistant to treatment with micafungin, whereas 303 Af293ΔalaA was inhibited >90% at even the lowest concentration of drug tested, 0.015625 μg/mL ( Figure  304 5F). The catalytically inactive strain (Af293alaA K322A ) was also tested for caspofungin sensitivity and 305 displayed the same phenotype as Af293ΔalaA ( Figure 5G). Finally, to ensure this phenotype was not 306 specific to the Af293 reference strain, these phenotypes were validated in another reference background, 307 CEA10. CEA10, CEA10ΔalaA, and CEA10alaA rec biofilms were tested for susceptibility to caspofungin 308 and again the loss of alaA resulted in increased echinocandin susceptibility ( Figure 5H). Moreover, the 309 increased susceptibility of the alaA null mutant was confirmed by measuring adenylate kinase release 310 We next tested if the increased susceptibility of Af293ΔalaA to echinocandins was biofilm specific 317 or if it extended to more traditional measures of drug resistance and tolerance that begin with exposing 318 dormant conidia to the drug. Tolerance, or the ability to grow in the presence of a fixed concentration of 319 drug, to caspofungin was measured using radial growth of conidia on agar plates at three concentrations 320 of caspofungin (0.25 μg/mL, 1 μg/mL, and 4 μg/mL). No significant differences in colony biofilm growth 321 were observed between the strains at any concentration tested on agar surfaces ( Figure 5-figure  322 supplement 2A-C). Additionally, the paradoxical effect was observed in all three strains tested, with increased growth as the concentration of caspofungin increased. Intriguingly, no difference in resistance 324 to caspofungin was observed when a minimal effective concentration (MEC) assay was utilized with 325 conidia of Af293, Af293ΔalaA, and Af293alaA rec ( Figure 5-figure supplement 2D). Therefore, the 326 increased susceptibility of Af293ΔalaA to echinocandins is a biofilm specific phenomenon, as no 327 difference in resistance or tolerance to caspofungin is observed when the drug is applied to dormant 328 conidia. 329 Figure 5: Loss of alaA leads to cell wall changes and increased susceptibility of biofilms to echinocandins. Germlings were stained with calcofluor white to quantify total chitin content (A), FITC-wheat germ agglutinin (WGA) to quantify surface exposed chitin (B), and Dectin-1 Fc to quantify surface exposed β-glucans (C). Each data-point represents an individual germling across three independent cultures per strain for each cell wall stain and the lines correspond to the mean. * p < 0.05, **** p < 0.0001, n.s. = not significant as determined by One-Way ANOVA with a Tukey's multiple comparisons test. D-F) 24-hour biofilms were established in the absence of drug and treated with calcofluor white (D), caspofungin (E), or micafungin (F) at the indicated concentrations for 3 hours and viability was determined by XTT assay. Mean +/-SD are shown for n ≥ 3 for each experiment. **** p < 0.0001 as determined by Two-Way ANOVA with a Tukey's multiple comparison test. The highest p-value for Af293ΔalaA compared to both Af293 and Af293alaA rec is shown. G) Af293alaA K322A -GFP biofilms were grown for 24-hours, treated with caspofungin at the indicated concentrations for 3 hours, and viability was determined by XTT assay. Mean +/-SD are shown for n = 3 replicates. **** p < 0.0001 as determined by Two-Way ANOVA with a Tukey's multiple comparison test. The highest p-values for Af293alaA K322A -GFP and Af293ΔalaA in comparison to Af293alaA-GFP are shown. No significant difference was observed between Af293alaA K322A -GFP and Af293ΔalaA. H) CEA10ΔalaA biofilms were grown for 24-hours, treated with caspofungin at the indicated concentrations for 3 hours, and viability was determined by XTT assay. Mean +/-SD are shown for n = 3 replicates. *** p < 0.001, **** p < 0.0001 as determined by Two-Way ANOVA with a Tukey's multiple comparison test. The highest p-values for CEA10ΔalaA compared to both CEA10 and CEA10alaA rec are shown. I) Adenylate kinase release assay as a quantification of cell lysis. 24-hour biofilms were treated with 1µg/mL of caspofungin (left) or micafungin (right) for 3-hours and supernatant adenylate kinase activity was quantified. Each replicate and the mean are shown (n = 4). ** p < 0.01, *** p < 0.001 as determined by Two-Way ANOVA with a Tukey's post-test. adherence of the WT and reconstituted strains was inhibited to slightly above that of the alaA null strain. 343 Importantly, adherence of the Af293ΔalaA was unaltered by any concentration of β-chloro-L-alanine 344 tested indicating some level of chemical specificity for AlaA ( Figure 6A). Additionally, treatment of the 345 GFP-tagged AlaA and catalytically inactive strains with β-chloro-L-alanine yielded similar results to the 346 WT and alaA null strains, respectively, indicating that the compound is acting through the catalytic activity 347 of the AlaA enzyme ( Figure 6B). 348 To test if β-chloro-L-alanine treatment increases susceptibility of biofilms to echinocandins, 349 biofilms were grown in the presence of 10μM and 100μM β-chloro-L-alanine to represent a range of 350 values that encompass both the EC50 value determined by the adherence assay results (10μM) and the 351 concentration that yielded an alaA deletion-like phenotype (100μM) ( Figure 6A). The β-chloro-L-alanine 352 treated biofilms of Af293, Af293ΔalaA, and Af293alaA rec were tested for sensitivity to caspofungin 353 treatment. In the WT and reconstituted strains efficacy of caspofungin increased as the concentration of 354 β-chloro-L-alanine increased ( Figure 6C). Curiously, even in the alaA null strain basal XTT reduction 355 decreased as the concentration of β-chloro-L-alanine increased. However, the alaA null strain remained highly susceptible to caspofungin regardless of the concentration of β-chloro-L-alanine ( Figure 6C). 357 Additionally, treatment of CEA10 with β-chloro-L-alanine increased caspofungin susceptibility of biofilms 358 validating that this is not specific to the Af293 strain background ( Figure 6D). Together these data 359 establish the proof of concept that chemical inhibition of AlaA is a possible strategy for increasing 360 susceptibility of A. fumigatus biofilms to echinocandins. 361 362 Altered extracellular matrix is not the primary factor impacting caspofungin susceptibility. 363 In other fungi, the biofilm extracellular matrix has been shown to be a major factor in reducing 364 antibiotic efficacy against biofilms (Taff et al., 2013). Therefore, we wanted to test if the increased 365 susceptibility to caspofungin could be attributed to the altered GAG composition of the alaA null strain. 366 To do this we utilized a strain lacking the UDP-glucose-4-epimerase required to produce GAG 367 All three strains were highly susceptible to caspofungin when AlaA was inhibited by β-chloro-L-alanine 377 ( Figure 6E, 6F). Untreated Af293Δuge3 was inhibited by caspofungin treatment to a greater extent than 378 the WT strain. However, this increased susceptibility was far less severe than observed in β-chloro-L-379 alanine treated biofilms and was not observed in the deacetylase deficient strain (Af293Δagd3) (Figure  380 6E). Therefore, GAG contributes to caspofungin resistance to some degree, but it is not the primary factor 381 responsible for the increased susceptibility when AlaA is chemically inhibited or genetically altered. Biofilms were treated with the indicated concentrations of caspofungin for 3-hours and viability was assessed by XTT assay. Mean +/-SD are shown (n = 3). D) Susceptibility of 24-hour CEA10 biofilms established in the presence or absence of 100µM β-chloro-L-alanine to caspofungin treatment. Biofilms were treated with the indicated concentrations of caspofungin for 3-hours and viability was assessed by XTT assay. Mean +/-SD are shown (n = 3). ** p < 0.01, **** p < 0.0001 as determined by Two-Way ANOVA with Tukey's multiple comparisons test. E-F) Susceptibility of 24-hour Af293Δuge3 (E) and Af293Δagd3 (F) biofilms established in the presence or absence of 100µM β-chloro-L-alanine to caspofungin treatment. Biofilms were treated with the indicated concentrations of caspofungin for 3-hours and viability was assessed by XTT assay. Mean +/-SD are shown (n = 3). ** p < 0.01, *** p < 0.001, **** p < 0.0001 as determined by Two-Way ANOVA with a Tukey's multiple comparisons test. The highest p-values for β-chloro-L-alanine treated groups vs their respective untreated groups are shown. alaA is required for echinocandin resistance in vivo. 384 Finally, we sought to determine if alaA plays a role in echinocandin resistance in vivo within lung 385 infection microenvironments. To address this question, we utilized a chemotherapy murine model of 386 invasive pulmonary aspergillosis (IPA). Outbred CD1 mice were immunosuppressed with 387 cyclophosphamide and triamcinolone then challenged with conidia of Af293, Af293ΔalaA, or 388 Af293alaA rec . The infection was allowed to establish for 24 hours followed by three treatments with either 389 0.9% NaCl or 1mg/kg micafungin every 24 hours ( Figure 7A). 12 hours after the final micafungin treatment 390 relative fungal burden was determined by qPCR quantification of A. fumigatus 18S rDNA. The dosage of 391 micafungin treatment used had no significant impact on fungal burden in mice inoculated with the WT or 392 reconstituted strains. Moreover, loss of alaA at the time point examined did not significantly impact fungal 393 burden levels in the untreated groups ( Figure 7B). In contrast, there was a 4-fold reduction in fungal 394 burden in mice inoculated with Af293ΔalaA and treated with micafungin compared to untreated mice 395 ( Figure 7B). Thus, loss of alaA in vivo significantly increases the susceptibility of A. fumigatus to sub-396 effective concentrations of the echinocandin micafungin. 397 Figure 7: alaA is required for echinocandin resistance in vivo. A) Experimental outline for determining in vivo echinocandin resistance using a chemotherapy model of invasive aspergillosis. Outbred CD1 mice were immunosuppressed with 150mg/kg cyclophosphamide 48 prior to fungal challenge and 40mg/kg triamcinolone 24 hours prior to fungal challenge. Mice were challenged with A. fumigatus or PBS (mock) at D0, and infection was allowed to establish for 24 hours. Mice were treated with either 1mg/kg micafungin or 0.9% NaCl every 24 hours from D1 to D3. 12 hours after the final micafungin treatment mice were sacrificed for fungal burden determination by qPCR. B) qPCR quantification of total ng fungal DNA in lungs of mice challenged with the indicated A. fumigatus strains and treated or untreated with 1mg/kg micafungin according to design in (A). Each datapoint and the mean are shown (n ≥ 12 for each experimental group and n = 5 for mock infected mice across two independent experiments). * p < 0.05, n.s. = not significant as determined by Kruskal-Wallis with a Dunn's multiple comparisons test. We originally found alaA through investigation of datasets associated with low oxygen adaptation. 422 AlaA catalyzes interconversion of pyruvate and alanine without direct involvement of reducing potentials 423 or any high energy molecules, such as ATP. However, the reduction of nitrate to alanine would consume five reducing potentials and this pathway is suggested to be important for a variety of systems in the 425 adaptation to low oxygen (Feala et al., 2007;Felig et al., 1970;Harrison, 2015;Lothier et al., 2020;Rocha 426 et al., 2010). It is possible that in these systems alanine serves as a nitrogen sink to prevent toxic 427 ammonium accumulation during the conversion of nitrate to ammonium. Therefore, we had originally 428 hypothesized that AlaA function was a critical means of recycling reducing potentials during low oxygen 429 growth and were surprised to find that AlaA plays a significant role in polysaccharide regulation and 430 biofilm formation despite the minimal impact on growth in a low oxygen environment. This result could be 431 due to a high redundancy in the number of mechanisms encoded by the fungus to balance reducing 432 potentials, or it could suggest that alanine metabolism has a more specific role in adaptation to natural 433 oxygen gradients formed by respiration and/or adaptation to stochastic fluctuations in environmental 434 oxygen that naturally occur during filamentous fungal biofilm growth. 435 The involvement of an alanine aminotransferase in polysaccharide regulation is even more 436 perplexing when one considers that none of the components of the reaction (alanine, pyruvate, 437 glutamate, and α-ketoglutarate) have any obvious or direct role in any known biochemical pathways to 438 generate A. fumigatus cell wall or extracellular matrix components. AlaA also does not appear to be 439 essential for maintaining basal levels of alanine for protein production, as genetic disruption of alaA did 440 not yield alanine auxotrophy. Here we describe that the catalytic activity of the AlaA protein is essential 441 for polysaccharide regulation. Additionally, transcriptional data corresponding to the uge3 and agd3 442 genes essential for synthesis and maturation of GAG, respectively, suggests that the mechanism of 443 regulation is predominantly post-transcriptional and potentially indirect from alterations in fungal 444 metabolism ( Figure 4C-D). However, the exact mechanism through which this reaction regulates 445 polysaccharide biosynthesis and maturation remains to be further studied. Further investigation into this 446 mechanism could yield significant insight into the interplay between metabolism, biofilm formation, and 447 antifungal drug resistance to help inform development of novel biofilm targeted antifungal drugs. construct. The catalytically inactive mutation was generated using nested PCR from the mutation site to 483 immediately before the stop codon in order to modify the AAG lysine codon to a GCC alanine codon. 484 This fragment was then fused with 500bp upstream of the point mutation, along with the in frame gfp-485 trpCterminator ptrA fragment, and ~1kb downstream of the alaA stop codon. The two alleles were 486 transformed into Af293 protoplasts and mutants were selected using pyrithiamine. Sanger sequencing 487 was used to confirm each allele. 488 Protoplasts were generated using lysing enzyme from Trichoderma harzianum (Sigma) and 489 transformed as previously described (Willger et al., 2008). Protoplasts were plated on sorbitol stabilized 490 minimal media (GMM + 1.2M sorbitol) containing pyrithiamine. For hygromycin selection protoplasts were 491 allowed to recover without hygromycin selection until germtubes were visible by inverted microscope 492 (overnight at 37°C). At which point 0.6% agar media containing hygromycin was added to a final 493 concentration of 175μg/mL. All strains were single spored and checked for correct integration, or 494 presence of construct in the case of the ectopic reconstituted strains, via PCR and southern blot. 495 Additionally, the basal expression of alaA was checked by RT-qPCR on RNA extracted from 24-hour 496 biofilms for the reconstituted strains using the alaA null mutants as negative controls (Figure 1-figure  497 supplement 3). 498

Metabolomics 499
Cultures for metabolomics were performed with 100mL of 10^6 conidia/mL of CEA10. Shaking 500 liquid cultures were performed in baffled flasks with a foam stopper to allow rapid environmental 501 acclimation to changes in oxygen tension. Cultures were grown for 24 hours at 37°C 200RPM in ambient 502 oxygen followed by either continued incubation at ambient oxygen or a shift to 0.2% O2 for two hours. 503 Biomass was harvested by filtering through Miracloth, washed thoroughly with water, and flash frozen in 504 liquid nitrogen. The biomass was then lyophilized and 100mg dry weight was submitted to Metabolon for 505 LC-MS/MS analysis, metabolite identification, and relative quantification. Data processing and figure  506 generation was performed in R using the ComplexHeatmap (Gu et al., 2016) and Pathview (Luo & 507 Brouwer, 2013) packages based on the relative ion counts and statistical measures given by Metabolon. 508

Growth Assays 509
For assays of growth on alanine as a carbon or nitrogen source, an equal molarity of carbon or 510 nitrogen atoms were added for each indicated molecule to our base minimal media lacking NaNO3 and 511 glucose. Agar plates were inoculated with 10^3 conidia and incubated for 72 hours at 37°C 5% CO2 in 512 either ambient oxygen or in a chamber that maintained oxygen at a concentration of 0.2% (InvivO2 400  513 Workstation, Ruskinn Baker). Biofilm biomass cultures were inoculated with 20mL of 10^5 conidia/mL in 514 GMM and grown in petri plates for 24 hours at 37°C 5% CO2. Supernatants and air-liquid interface growth 515 were removed, and biofilms were harvested using a cell scraper. Biomass was washed 2X with ddH2O 516 with centrifuging at 5000RPM for 10 minutes to spin down biomass, frozen at -80°C, lyophilized, and dry 517 weight was measured. For liquid kinetic growth assays of biofilms, 200μL of 10^5 conidia/mL in GMM 518 was inoculated in six technical replicates per strain in a 96 well plate. Plates were incubated statically in 519 a plate reader at 37°C with Abs405 readings every 15 minutes over the first 24 hours of growth. 520 Crystal Violet Adherence Assay 521 U-bottomed 96-well plates were inoculated with 10^5 conidia/mL in GMM and incubated statically 522 for the indicated time at 37°C 5% CO2 to allow biofilms to form. To remove non-adherent cells, media 523 was removed, and biofilms were washed twice with water via immersion followed by banging plate onto 524 a stack of paper towels. Adherent biomass was stained with 0.1% (w/v) crystal violet for 10 minutes and 525 biofilms were washed twice with water to remove excess crystal violet. Crystal violet was then dissolved 526 in 100% ethanol, supernatants were transferred to a flat-bottomed plate, and absorbance at 600nm was 527 quantified. Dose-response assays testing the impact of β-chloro-L-alanine on adherence were fit with a 528 non-linear regression based on a dose-response model (GraphPad Prism 9) to calculate EC50 values. 529 For assays testing the impact of collagen coating, 50μL of Collagen Coating Solution (Sigma) was applied 530 to half the wells of a 96-well U-bottom plate overnight at room temperature. The solution was removed, 531 and wells were washed one time with PBS prior to inoculation. All crystal violet adherence assays were 532 performed with 3-6 technical replicates and data presented represent at least three biological replicates. 533

Oxygen Quantification 534
Oxygen was quantified as previously described (Kowalski et  were coated with 2mL of 0.6% agar GMM to protect the microelectrode from breaking when performing 540 deep profiling into the biofilms. 3mL of 10^5 conidia/mL in GMM was inoculated into the plates and 541 incubated for 24 hours at 37°C 5% CO2. The meniscus of the culture was ~3mm above the surface of 542 the agar pad, and thus oxygen was measured at the center of each culture in 200μm steps, with 543 technical duplicates at each step, from the air-liquid interface to 2800μm into the culture. Oxygen 544 quantification was performed immediately upon removal of the culture from the incubator. At least 545 seven independent biofilms were measured for each strain across two experiments along with three 546 media only cultures that lacked fungus. 547

Fluorescent Microscopy 548
Fluorescent confocal microscopy was performed on an Andor W1 Spinning Disk Confocal with a Nikon 549 Eclipse Ti inverted microscope stand. 550

AlaA Localization Studies 551
Af293alaA-GFP, Af293alaA K322A -GFP, and Af293 were cultured in GMM on MatTek ® dishes at 552 37°C in GMM until germlings were visible on an inverted light microscope, ~9 hours for Af293alaA-GFP 553 and Af293, and ~10 hours for Af293alaA K322A -GFP. Media was removed and replaced with fresh GMM 554 containing 100nm MitoTracker™ Deep Red FM (ThermoFisher). Cultures were incubated for 30 minutes 555 at 37°C to allow mitochondrial staining. Images were acquired with a 60X oil-immersion objective at 556 488nm (GFP) and 637nm (MitoTracker) on the Andor W1 Spinning Disk Confocal. Images were 557 deconvolved and max intensity z-projections were generated using the Nikon NIS-Elements AR software. 558 Experiment was also performed with WT Af293 as a negative control for autofluorescence ( All germlings were imaged on the Andor W1 Spinning Disk Confocal with a 60X oil-immersion 599 objective using 405nm for calcofluor white and 488nm for FITC-WGA and Dectin-1-Fc. Cell wall staining 600 was quantified using Fiji (ImageJ). Z-stacks were assembled using a sum-intensity Z-projection. Regions 601 of interest (ROI's) were drawn around each individual germling within a given image, along with a region 602 lacking any germlings to account for background fluorescence. Within each ROI the area, sum intensity, 603 and mean intensity were quantified. To obtain corrected mean intensity measurements, the mean 604 background intensity was multiplied by the area of the ROI to calculate total background contribution. 605 The total background contribution was subtracted from the ROI's sum intensity and this value was divided 606 by the area of the ROI yielding the final corrected mean fluorescence intensity. Each cell wall stain was 607 performed in triplicate cultures and at least three fields of view were obtained for each culture. For FITC-608 WGA the staining pattern was almost entirely absent from the germ-tube and enough natural size 609 heterogeneity was found both within and between cultures to act as a confounding variable. Thus, for 610 FITC-WGA ROIs were drawn around each conidial body, where the staining was present, rather than the 611 entire germling. 612

Extracellular matrix monosaccharide analysis and ELLA 613
Enzyme Linked Lectin Assay (ELLA) 614 100µL of 10^5 conidia per mL in GMM were inoculated into wells of 96 well plate and incubated 615 for 24h. Culture supernatants were then transferred to a 384 well plate Immulon 4HBX with or without 616 500pM of recombinant Agd3. After a 1-hour incubation period, wells were washed three times with 1X 617 TBS -0.05% Tween20. A preincubated solution of 30nM soybean agglutinin lectin coupled to biotin and 618 1/700 avidin-HRP in TBS-T was added to the wells and incubated for 1 hour. After 3 TBS-T washes, 619 detection was performed using Ultrasensitive TMB read at 450nm. Normalization of the values were 620 performed reporting the absorbance reads to the absorbance of Af293. 621

Extracellular Matrix Monosaccharide Composition by Gas Chromatography Coupled to Mass 622
Spectrometry 623 100ml of GMM was inoculated with 10^4 conidia per mL and incubated for 3 days at 37°C at 624 200rpm. Culture supernatants were filtered by Miracloth prior to being dialyzed for 3 days against Milli-625 Q ® water and lyophilized. About 0.5mg of dried material was then derivatized into TriMethylSilyl 626 derivatives. Samples were hydrolyzed with either 2 M trifluoroacetic acid for 2 hours at 110°C or 6 M 627 hydrochloric acid (HCl) for 4 hours at 100°C. Monosaccharides were then converted in methyl glycosides 628 by heating in 1 M methanol-HCl (Sigma-Aldrich) for 16 hours at 80°C. Samples were dried and washed 629 twice with methanol prior to re-N-acetylating hexosamine residues. Re-N-acetylation was performed by 630 incubation with a mix of methanol, pyridine, acetic anhydride (10:2:3) for 1 hour at room temperature. 631 Samples were then treated with hexamethyldisilazane-trimethylchlorosilane-pyridine solution (3:1:9; 632 ThermoFisher) for 20 min at 110°C. The resulting TMS methyl glycosides were dried, resuspended in 1 633 ml of cyclohexane, and injected in the Agilent 7890B GC -5977A MSD. Identification and quantification 634 of the monosaccharides was performed using a mix of monosaccharide calibrants injected at different 635 concentrations as a reference. Quantification was finally normalized to an equivalent of 1mg of material 636 before comparison between groups. 637

RNA Extraction and RTqPCR 638
RNA was extracted from 24-hour biofilm cultures in a 6-well plate. Supernatant was removed 639 and 500μL of TRIsure TM (Bioline Reagents) was immediately applied to the biofilms. Biofilm 640 suspensions were centrifuged, and supernatant was removed. Biomass was resuspended in 200μL 641 TRIsure TM flash frozen in liquid nitrogen, and bead beat with 2.3mm beads. Homogenate was brought 642 to a final volume of 1mL with TRIsure TM , bead beaten a second time, and RNA was extracted following 643 the manufacturer's protocol. 5μg of RNA was DNase treated with TURBO DNA-free TM kit (Invitrogen) 644 according to manufacturer's protocol. 500ng of DNase-treated RNA was run on an agarose gel to 645 ensure RNA integrity. 500ng of DNase-treated RNA was used for cDNA synthesis as previously 646 described (Beattie et al., 2017). The RTqPCR data were collected on a CFX Connect Real-Time PCR 647 Detection System (Bio-Rad) with CFX Maestro Software (Bio-Rad). Gene expression was normalized 648 to tefA expression for all experiments. Primers used for RTqPCR are listed in Table S2.

Biofilm Assays 651
To test susceptibility of biofilms to inhibition by calcofluor white and caspofungin, 500μL of 10^5 652 conidia per mL in GMM were inoculated into wells of 24 well plates and biofilms were grown statically for 653 24 hours at 37°C 5% CO2. Any air-liquid interface growth was removed using a sterile pipette tip, the 654 supernatant was removed from each well, and fresh media containing the indicated concentration of 655 calcofluor white or caspofungin was added to the biofilm. Biofilms were incubated for a further 3 hours at 656 37°C 5% CO2, washed with PBS, and 300μL of XTT solution was added to each well (0.5mg/mL XTT 657 [2,3-bis-(2-methoxy-4-nitro-5-sulfophenyl)-2H-tetrazolium-5-carboxanilide] (VWR) with 25μM menadione 658 in PBS). XTT solution was incubated for 1 hour at 37°C to allow reduction of the dye. 150μL of 659 supernatants were transferred to a flat-bottomed 96 well plate and the absorbance at 450nm was read 660 on a plate reader. Abs450nm values of the treated samples were compared to untreated biofilms to calculate 661 the relative metabolic activity as the percent of untreated control for each strain. All XTT assays were 662 performed on at least three biological replicates. 663 For the adenylate kinase release assay biofilms were grown and treated in the same manner as 664 above. After three hours of echinocandin treatment supernatants were collected and allowed to cool to 665 room temperature. XTT assay was performed on the biofilms according to the protocol above for matched 666 XTT data. Relative adenylate kinase levels were measured on 40μL of supernatants via the ToxiLight TM 667 Non-Destructive Cytotoxicity BioAssay Kit (Lonza) according to the manufacturer's instructions. 668 Chemiluminescence was measured on a Synergy Neo2 multi-mode plate reader (BioTek). Experiment 669 was performed on four independent biological replicates. 670

Conidia Assays 671
Minimal effective concentration (MEC) assays were performed by inoculating 100μL of 2*10^5 672 conidia/mL in GMM into 96 well flat-bottomed plates containing 100μL serial 2-fold dilutions of 673 caspofungin in GMM from 8μg/mL to 0.015625μg/mL along with a no drug control. Cultures were grown statically for 24 hours and viewed under an inverted light microscope for the concentration at which gross 675 morphological changes characteristic of caspofungin treatment became visible. This concentration was 676 deemed the MEC. Radial growth assays were performed by inoculating GMM agar plates containing the 677 indicated quantity of caspofungin with 10^3 conidia in 2μL 0.01% Tween-80 and incubating at 37°C 5% 678 CO2 for 72 hours. Images are representative of four biological replicates. We thank A. Lavanway (Dartmouth) for their microscopy expertise. We also thank the Imaging 710 Facility at Dartmouth and the Biomolecular Targeting Core (P20-GM113132) for use of equipment. This 711 work was supported by funding from the NIH National Institute of Allergy and Infectious Diseases (NIAID) 712 (grant no. R01AI130128 and R01AI146121), a pilot award from the Cystic Fibrosis Foundation (CFF) 713 Research Development Award (STANTO15RO), and a CFF research award (CRAMERGO19). J.K. was 714 supported by the Molecular Pathogenesis Training Grant (T32AI007519). 715 Figure 1-figure supplement 1: Acute exposure to a low oxygen environment significantly changes A. fumigatus metabolism. A) Outline of the metabolomics experiment. Shaking culture were grown for 24 hours in liquid GMM followed by either a shift to a 0.2% oxygen environment or continued incubation in ambient oxygen for two hours. Biomass was then harvested, flash frozen to quench metabolic reactions, lyophilized, and submitted for metabolomics. B) Overview of all significantly altered metabolites detected in experiment. C) Significantly changed carbohydrate related metabolites in the two conditions. For C-D ion counts were normalized to the average ion count for the respective metabolite across all samples and log2 transformed. Each column corresponds to an individual sample (n = 5 per condition). D) Average relative abundance of metabolites in 0.2% vs ambient oxygen (log2 transformed) mapped onto the Arginine Biosynthesis KEGG map using the Pathview R package. Red indicates greater abundance in 0.2% oxygen, blue indicates greater abundance in ambient oxygen, white indicates no difference between conditions, and metabolites not detected are in grey. Figure 1-figure supplement 2: alaA is required for alanine catabolism and biofilm physiology in the CEA10 strain background. A) Growth of CEA10ΔalaA on minimal media containing the indicated sole carbon and nitrogen sources in ambient oxygen and 0.2% oxygen environments. B) Static growth assay of CEA10ΔalaA over the first 24 hours of biofilm growth. Mean +/-SD of 6 technical replicates is shown. Experiment was repeated a minimum of 3 times with similar results. * p < 0.05, *** p < 0.001 by One-Way ANOVA with a Tukey's multiple comparisons test C) Crystal violet adherence assay of 24-hour biofilms (n = 6). **** p < 0.0001, n.s. = not significant as determined by One-Way ANOVA with a Tukey's multiple comparisons test.   (Af293alaA K322A -GFP). B) Phylogeny of AlaA relative to human, murine, and Saccharomyces cerevisiae alanine aminotransferases. All species except A. fumigatus encode two alanine aminotransferases in their genomes. Proteins were aligned in MEGA X using MUSCLE and a maximum-likelihood tree was generated. Scale bar and branch lengths refer to substitutions per site. C) Percent identity matrix of the alanine aminotransferases shown in (B).  Supernatants were used to quantify adenylate kinase activity ( Figure 5I) and an XTT assay was performed to measure viability of biofilm biomass. Each replicate and mean are shown (n = 4). ** p < 0.01, *** p < 0.001 as determined by Two-Way ANOVA with a Tukey's multiple comparisons test.  . Each replicate along with the mean +/-SD are shown (n ≥ 3). * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001 as determined by Two-Way ANOVA with a Tukey's multiple comparisons test.