A Heterogeneously Expressed Gene Family Modulates the Biofilm Architecture and Hypoxic Growth of Aspergillus fumigatus

Colony furrows increase oxygen diffusion within colony biofilms.A definitive feature of A. fumigatus H-MORPH and a common feature of hypoxia-grown colony morphotypes is the presence of furrows or invaginations within the colony biofilms (Fig. 1A; see also Fig. S1A). We hypothesize that these furrows increase colony surface area and oxygen diffusion into the colony. To test this hypothesis, a microelectrode oxygen sensor was utilized to quantify oxygen above, within, and below the colony biofilms of the reference N-MORPH strain AF293 grown in normoxia (normal oxygen, 21% O₂, nonfurrowing condition) or hypoxia (furrowing condition, 0.2% O₂) (Fig. 1B). The nonfurrowing normoxia-grown colony of AF293 shows a precipitous drop in oxygen within the 400 μm of the colony above the agar surface (0 μm) and 200 μm below the agar surface (embedded colony). In contrast, the furrowed, hypoxia-grown colonies of AF293, measured both within the furrowing (F) or nonfurrowing (NF) regions, show significantly increased oxygen levels within the colonies (Fig. 1A and B). To determine whether the furrows in the normoxia-grown H-MORPH colony of EVOL20 also impact oxygen diffusion, we utilized the same approach to quantify oxygen in AF293 normoxia-grown colonies and furrowing (F) and nonfurrowing (NF) regions of EVOL20 normoxia-grown colonies (Fig. 1A and C). Within the furrows of the EVOL20 colony oxygen is significantly increased above, at, and below the agar surface (0 μm). In addition, the nonfurrowing regions of the EVOL20 colony biofilm also have significantly increased oxygen compared to AF293 within the embedded colony (0 to 200 μm) (Fig. 1C). Together, these data suggest that colony furrowing of A. fumigatus occurs in hypoxia in part to increase oxygen diffusion into the colonies and that the furrows of the hypoxia-evolved H-MORPH strain EVOL20 develop even under normoxia to increase oxygen deep within the colonies. The increased oxygen diffusion within H-MORPH colonies coincides with altered hyphal architecture within biofilms, increased hypoxic growth, reduced adherence, and increased inflammation and virulence (see Fig. S1C) (26).

• Open in new tab
• Download powerpoint

FIG 1

A cryptic gene within the hrmA-associated gene cluster is necessary for the hypoxia-evolved phenotypes of EVOL20. (A) 72-h colony biofilms used for oxygen measurements, with furrows (F) and nonfurrowing (NF) regions labeled. Images are representative of three independent biological samples. (B) Oxygen quantification of AF293 colony biofilms grown in normoxia (21% O₂) or hypoxia (0.2% O₂). In hypoxia-grown colonies oxygen was measured both in furrows (F) and in nonfurrows (NF). n = 3 independent biological replicates. Error bars indicate standard errors around the mean. Multiple one-way analyses of variance (ANOVA) were performed with Dunnett’s posttest at each depth. (C) Oxygen quantification for normoxia-grown colonies of AF293 and EVOL20. EVOL20 colonies were measured in furrowing (F) and nonfurrowing (NF) regions. n = 3 independent biological replicates. Error bars indicate standard errors around the mean. Multiple one-way ANOVAs were performed with Dunnett’s posttest at each depth. (D) 96-h colony biofilms in normoxia (21% O₂) and hypoxia (0.2% O₂). Images are representative of three independent biological samples. (E) Quantification of colony biofilm morphological features from three independent biological samples. One-way ANOVA with Dunnett’s posttest for multiple comparisons was performed relative to EVOL20 within each oxygen environment. (F) The ratio of fungal biomass in hypoxia (0.2% O₂) relative to fungal biomass in normoxia (21% O₂) (H/N) in shaking-flask cultures. One-way ANOVA with Dunnett’s posttest for multiple comparisons was performed relative to EVOL20. n = 3 independent biological samples. (G) Adherence to plastic measured through a crystal violet assay. Dashed line marks the mean value for media alone. One-way ANOVA with Dunnett’s posttest for multiple comparisons was performed relative to ΔcgnA^EVOL. n = 6 independent biological replicates. (H) Gene expression measured by qRT-PCR for cgnA and the cryptic ORF. n = 3 independent biological replicates. One-way ANOVA with Tukey’s multiple-comparison test was performed.

The native 5′ sequence to cgnA is required to complement the loss of cgnA in EVOL20.We have previously characterized a role for the HAC gene cluster in the generation of H-MORPH in EVOL20, based on the observation that hrmA, the HAC regulator, and the annotated cgnA coding sequence are required for H-MORPH in EVOL20. However, while constitutive expression of hrmA in the reference strain AF293 is sufficient to elevate mRNA levels of the HAC genes and generate H-MORPH, the expression of cgnA alone is not sufficient to generate H-MORPH (26). Therefore, we hypothesized that elevated expression of multiple HAC genes may be required to generate the H-MORPH phenotype.

Since the majority of annotated HAC genes, with the exception of Afu5g14920, remain unaltered following the loss of cgnA in EVOL20 (ΔcgnA^EVOL) (26), we overexpressed cgnA in this background (ΔcgnA^EVOL; cgnA^OE) using an Aspergillus nidulans gpdA promoter to drive constitutive expression (Fig. 1D). We discovered that the overexpression of cgnA could not restore the H-MORPH phenotype in ΔcgnA^EVOL strains. Instead, the ΔcgnA^EVOL; cgnA^OE strain colony morphology is not significantly different compared to ΔcgnA^EVOL respective to furrowing and the percent vegetative mycelia (Fig. 1D and E). We next hypothesized that the native sequence 5′ of cgnA may be required to restore H-MORPH in ΔcgnA^EVOL strains. Ectopic integration of cgnA with its native promoter and 5′ sequence (cgnA^RECON) is able to reconstitute H-MORPH in ΔcgnA^EVOL strains with elevated colony furrows and an increased percentage of vegetative mycelia in normoxia (Fig. 1D and E). In addition to a transition from the H-MORPH phenotype of EVOL20 to N-MORPH, the ΔcgnA^EVOL strain has a significantly reduced ratio of hypoxic to normoxia growth (hypoxia fitness [H/N]) (Fig. 1F) and significantly increased hyphal adherence (Fig. 1G) compared to EVOL20 (26, 29). Where ΔcgnA^EVOL; cgnA^OE does not restore either of these phenotypes to the level of EVOL20, cgnA^RECON, where cgnA is reintroduced with its native 5′ sequence, restores both hypoxia fitness and adherence of ΔcgnA^EVOL similarly to that of EVOL20 (Fig. 1F and G). Although the integration loci of cgnA in the ΔcgnA^EVOL; cgnA^OE and cgnA^RECON strains may not be identical and contribute to some phenotypic variation, the absolute necessity of the native sequence 5′ of cgnA to complement the loss of cgnA in EVOL20 (ΔcgnA^EVOL) prompted us to investigate this genomic region more closely.

A cryptic gene is encoded 5′ of cgnA within HAC and is required for H-MORPH and HAC-related phenotypes.Utilizing published RNA-sequencing data, we identified a substantial region of mapped reads 5′ to cgnA in EVOL20 that were absent in AF293 (see Fig. S2A) (26). Neither the AF293 assembled reference genome nor the partially assembled genome of A1163 annotates a gene within this region (26, 29). It is unlikely these reads belong to the same transcript as cgnA since they map to the opposite strand. Therefore, we hypothesize that these reads map to an independent cryptic gene within HAC and that this gene may be important for H-MORPH and other EVOL20-related phenotypes (i.e., hypoxia fitness, adherence, and biofilm architecture) (26). To determine whether our strategies to delete cgnA interrupted the mRNA levels of this cryptic gene, we designed primers within the predicted open reading frame (ORF) to quantify relative expression in two isogenic strain sets: EVOL20/ΔcgnA^EVOL and hrmA^R-EV/hrmA^R-EV; ΔcgnA (see Table S1). In both cases, deletion of the cgnA coding sequence reduces cgnA mRNA levels and mRNA levels corresponding to the cryptic gene (see Fig. S2B and C).

With Integrative Genome Viewer and NCBI ORF Finder, we were able to predict a two-exon ORF of 579 bp from the region corresponding to the cryptic gene (see Fig. S2D). In the DNA construct used to generate ΔcgnA^EVOL; cgnA^OE, cgnA expression was driven by the constitutive A. nidulans gpdA promoter, and the native 5′ sequence containing the cryptic gene ORF was therefore not reintroduced (see Fig. S2E). In contrast, the DNA construct used to generate the cgnA^RECON strain utilized the native sequence 5′ to cgnA to drive expression. This region includes the entire predicted coding sequence of the cryptic gene (see Fig. S2E). Gene expression analysis confirmed that both ΔcgnA^EVOL; cgnA^OE and cgnA^RECON strains have cgnA mRNA levels equivalent to or greater than those of EVOL20, but only cgnA^RECON restores the mRNA levels of the cryptic gene similarly to EVOL20 (Fig. 1H). Only with the strain cgnA^RECON, where both cgnA and the cryptic gene are expressed, is H-MORPH restored (Fig. 1D and E), hypoxic fitness increased (Fig. 1F), and adherence reduced (Fig. 1G) in ΔcgnA^EVOL strains to resemble EVOL20. Thus, the cgnA sequence alone is not sufficient to generate the EVOL20 phenotypes but requires the 5′ cryptic gene. Based on the previously published phenotypes of EVOL20 and the data presented here, we propose the name biofilm architecture factor (bafA) for this cryptic gene.

The HAC cryptic gene shares significant similarity to genes encoded within putative clusters orthologous to HAC in an independent strain.Previous phylogenetic analysis of the HAC genes hrmA and cgnA for presence across A. fumigatus strains revealed heterogeneity across the species, where some strains did not encode hrmA or cgnA (26). However, some A. fumigatus strains, such as the well-studied CEA10, encoded putative orthologs to hrmA within putative orthologous HAC clusters where the neighboring genes encode proteins with domain and amino acid sequence similarities to the proteins encoded by the genes of HAC (26). This observation suggests that some strains, such as CEA10, may encode multiple HAC-like clusters in addition to HAC. Although these orthologous HAC-like clusters show no evidence of encoding a putative ortholog of cgnA, they encode a gene that is similar to the HAC cryptic gene bafA (see Fig. S3A). In the hrmB associated cluster (H_BAC) from CEA10, the predicted amino acid sequence of BafA in AF293 shares 78.35% identity with that of AFUB_044360 (see Fig. S3B). In addition, in the hrmC associated cluster (H_CAC) from CEA10, the predicted amino acid sequence of BafA in AF293 shares 45.41% identity with that of AFUB_096610 (see Fig. S3C). At the level of DNA sequence, the similarity is even greater, with 87% nucleotide identity between bafA and AFUB_044360 and 68% nucleotide identity between bafA and AFUB_096610. Based on these sequence similarities we propose the name bafB for AFUB_044360 and bafC for AFUB_096610. Notably, all three of these genes—bafA, bafB, and bafC—are located adjacent to a gene encoding a hypothetical protein with a conserved domain of unknown function DUF2841. The role of these DUF2841-containing proteins and their potential role in the development of H-MORPH remains the focus of future study.

AF293 and EVOL20 only encode HAC and there is no evidence for intact H_BAC or H_CAC in these genomes based on orthologs to hrmA. In contrast, CEA10 encodes all three putative clusters. Previously, we identified H-MORPH strains of A. fumigatus that do not encode hrmA/HAC and speculated that other genetic mechanisms, possibly these other orthologous clusters, may function to generate H-MORPH in these strains. Therefore, we sought to determine the abundance of HAC, H_BAC, and H_CAC throughout the A. fumigatus population. To do this, we looked for the presence of bafA (HAC), bafB (H_BAC), and bafC (H_CAC) across available sequenced A. fumigatus strains. We confirmed that similar to hrmA, the number of strains positive for the presence of bafA is low (n = 24) (Fig. 2). Similarly, bafB is present within ∼27% of the A. fumigatus genomes analyzed (n = 24) (Fig. 2). Strains positive for encoding bafC are more abundant (n = 35), but this is complicated by the presence of another bafC ortholog in AF293 (Afu1g00770) that is not encoded near a putative hrmC ortholog (Fig. 2). Afu1g00700 shares 91% nucleotide identity with AFUB_096610 in CEA10 and, while not syntenic, its high sequence similarity contributes to the positive identity of bafC in AF293 and potentially other strains as well (Fig. 2) (FungiDB) (29). Interestingly, there are strains similar to CEA10 that are positive for the presence of two or more of the baf genes (bafA, bafB, and bafC) (n = 21). However, this is likely an overestimate due to the presence of Afu1g00700 orthologs in some genomes.

• Open in new tab
• Download powerpoint

FIG 2

Phylogeny of 92 A. fumigatus strains with copy number of the cryptic gene and its putative orthologs. The A. fumigatus strain maximum-likelihood phylogeny was constructed from 71,513 parsimony informative SNPs identified across the strains. The heat map indicates the abundance of bafA, bafB, and bafC, as well as a baf pseudogene (Pseudobaf) across the phylogeny based on the genome sequences from CEA10. Strains which have been previously identified as encoding hrmA are indicated by a plus (+) sign.

Many of the analyzed genomes also encode a predicted pseudogene with high similarity to bafB (Pseudobaf) (Fig. 2). The pseudogenes are degenerate ORFs that have multiple stop codons throughout their sequence. Although AF293 does not encode bafB and bafC and their putative gene clusters, the presence of the pseudogene suggests that an ancestral strain of AF293 did encode bafB and H_BAC. A BLAST search with hrmB or bafB from CEA10 against the AF293 genome matches a region of ∼1,050 kb on chromosome 3, where no genes are annotated (between Afu3g03760 and Afu3g03770) (see Fig. S4A) (FungiDB) (29). The regions that map to hrmB and bafB are littered with stop codons truncating the ORFs and thus are likely pseudogenes (see Fig. S4B and C). If expressed, the chromosomal region that maps to hrmB in Af293 is predicted to generate a 123-amino-acid protein instead of 423 amino acids (see Fig. S4B), and the Pseudobaf chromosomal region that maps to bafB in AF293 is predicted to generate a 25-amino-acid protein instead of 193 amino acids (see Fig. S4C). Other degraded ORFs, or pseudogenes, similar to bafB are observed across the phylogeny in different copy numbers (Fig. 2), posing interesting questions about potential functions of these pseudogenes, how they arose in the population, and how they are maintained.

Introduction of the cryptic gene ortholog bafB is sufficient to complement the loss of cgnA and hrmA in EVOL20.To determine whether bafB from CEA10, whose protein sequence is 78.35% identical to bafA, could complement the loss of cgnA in EVOL20 (ΔcgnA^EVOL), we introduced bafB with the constitutively active gpdA promoter (ΔcgnA^EVOL; bafB^OE). The resulting strain reverted the N-MORPH phenotype of a ΔcgnA^EVOL strain to the H-MORPH phenotype of EVOL20 with significantly increased colony furrows and percent vegetative mycelia (Fig. 3A and B). As mentioned above, the majority of HAC genes are not altered in expression as a result of cgnA deletion (26), thus the expression of other HAC genes could still be required for bafB to generate H-MORPH. The loss of hrmA in EVOL20 (ΔhrmA^EVOL) reverts the colony to N-MORPH and mRNA levels of HAC genes are significantly reduced (26). To determine whether hrmA and subsequently the HAC cluster genes that rely on hrmA for expression are necessary to generate H-MORPH in the presence of bafB, we introduced bafB with the constitutive gpdA promoter into the ΔhrmA^EVOL strain (ΔhrmA^EVOL; bafB^OE). Even in the absence of hrmA, bafB is sufficient to generate H-MORPH and significantly increase colony furrows and the percent vegetative mycelia (Fig. 3A and B). In addition to H-MORPH, EVOL20 has elevated hypoxic fitness (H/N) and reduced surface adherence relative to AF293 that is dependent on both hrmA and cgnA/bafA (Fig. 1C and D) (26, 30). The overexpression of bafB significantly increases the hypoxic fitness of ΔhrmA^EVOL and ΔcgnA^EVOL strains (Fig. 3C) and significantly reduces the adherence of these strains to a plastic surface (Fig. 3D). Importantly, bafB is sufficient to complement these phenotypes in EVOL20 without increasing HAC gene mRNA levels (Fig. 3E). In fact, the mRNA levels of hrmA are slightly, but significantly, reduced as a result of constitutive bafB expression (Fig. 3E).

• Open in new tab
• Download powerpoint

FIG 3

The putative ortholog of the cryptic gene, bafB, is sufficient to complement the loss of the HAC genes cgnA and bafA. (A) 96-h colony biofilms from 21% O₂ where hypoxia-locked (H-MORPH) morphological features, furrows and vegetative mycelia, can be visualized. Images are representative of three independent biological samples. (B) Quantification of the H-MORPH features from colony biofilms of three independent biological samples. Student two-tailed nonparametric t tests were performed between each isogenic strain set. (C) Ratio of fungal biomass in hypoxia (0.2% O₂) relative to fungal biomass in normoxia (21% O₂) (H/N) in shaking-flask cultures. A Student two-tailed nonparametric t test was performed between isogenic strain sets. n = 3 independent biological samples. (D) Adherence to plastic measured through a crystal violet assay. Dashed line marks the mean value for media alone. A Student two-tailed nonparametric t test was performed between isogenic strain sets. n = 7 independent biological samples. (E) Gene expression measured by qRT-PCR for representative HAC genes as a result of bafB overexpression at 21% O₂. n = 3 independent biological samples. (F) Heat map displaying the architecture of the fungal biofilms measured as the deviation of the hyphae from a vertical axis. Each column is representative of a minimum of three independent biological samples. (G) Representative images (n = 3 biological samples) of submerged biofilms on the orthogonal plane (xz) that are quantified in the heat map in panel F. Scale bar, 200 μm. Error bars indicate standard errors around the mean.

To test whether bafB expression alters biofilm architecture, a HAC-dependent phenotype of EVOL20, we cultured submerged biofilms for 24 h and imaged the bottom ∼300 μm of the biofilm. As a metric for biofilm architecture, we measured the angle of hyphal deviation from the vertical axis. As has been described for the N-MORPH AF293, ΔcgnA^EVOL, and ΔhrmA^EVOL strains, at 24 h the bottom ∼50 μm of the biofilm features filaments that grow along the surface and have a high deviation from the vertical (26). At depths above 50 μm for these N-MORPH strain, the hyphae orient vertically and grow polarized toward the air-liquid interface with little deviation from the vertical axis. In contrast, the H-MORPH strain EVOL20 features hyphae throughout all 300 μm that are oriented with a high deviation from the vertical (26). When bafB is overexpressed in the N-MORPH strains cgnA^EVOL and ΔhrmA^EVOL, the resulting H-MORPH strains (Fig. 3A) develop biofilms that also resemble the architecture of EVOL20 (Fig. 3F and G). There is greater hyphal deviation from the vertical axis above 50 μm in the biofilms of ΔcgnA^EVOL; bafB^OE and ΔhrmA^EVOL; bafB^OE strains (Fig. 3F and G). Thus, introduction of a constitutively expressed bafB is sufficient to complement the HAC-dependent phenotypes of EVOL20.

The mechanisms underlying hyphal arrangement and ultimately the shift from vertically oriented hyphal growth to a more horizontal hyphal growth in the biofilm remains undefined. Previously, we had hypothesized that this was a consequence of the altered hyphal surface of H-MORPH strains (26). The BafB protein is predicted to have a signal sequence at its N terminus (SignalP, FungiDB) (see Fig. S5A) (29, 31). To gain insight into how bafB could directly impact the biofilm architecture of the ΔcgnA^EVOL strain, we generated a C-terminal green fluorescent protein (GFP)-tagged allele of bafB in the ΔcgnA^EVOL strain. Introduction of the GFP-tagged allele, like the native bafB allele, is able to revert the N-MORPH colony morphotype of the ΔcgnA^EVOL strain to H-MORPH (see Fig. S5B). In mature hyphae, the localization of the GFP signal is present both in the cytosol within circular structures that resemble trafficking endosomes or vacuoles previously described in A. nidulans (32) (see Fig. S5C) and concentrated toward the distal hyphal region (see Fig. S5D). At the distal region, the GFP signal is present within circular structures, or puncta, as well as localized along the sides of the hyphae (see Fig. S5D). Time-lapse imaging reveals that these BafB puncta are dynamic and move rapidly within the hyphae (see Video S1 and Fig. S5E). Costaining with the membrane dye FM4-64 indicate overlap in the patterns of BafB localization and endosome localization (see Fig. S5F). This subcellular pattern and the presence of the N-terminal secretion signal peptide (see Fig. S5A) support the hypothesis that BafB localizes extracellularly at the hyphal tips or is secreted (33). Although the GFP signal corresponding to BafB is largely absent from the hyphal edges where the cell wall is more stable, the signal is abundant at the tip where cell wall modeling is in progress. This is evidenced by the absence of Dectin-1 binding, which specifically interacts with β-1,3-glucan, at the hyphal tip where BafB is abundant (see Fig. S5G). Cell wall irregularities are a feature of H-MORPH, and the actively growing hyphal tip directs cell polarity (26, 34). Since colony morphology is a consequence of polarized growth and structure of the cell wall, this localization pattern indicates that BafB could be acting as the H-MORPH effector (26, 35). The high amino acid identity shared between bafB and the HAC-resident gene bafA raise the question of whether bafA is the HAC effector and is sufficient to generate H-MORPH in the parental strain AF293.

Overexpression of bafA generates H-MORPH and elevated hypoxic growth in the absence of HAC induction in two independent strain backgrounds.In the parental strain AF293, the basal expression of HAC is low, and previous RNA-sequencing data reveal no mapped reads to the predicted bafA ORF in AF293 (see Fig. S2A) (26). In addition, quantitative reverse transcription-PCR (qRT-PCR) for bafA mRNA revealed no detection above background in AF293, but overexpression of an additional bafA allele results in detectable bafA mRNA (see Fig. S6A). The synthetic, elevated expression of bafA in AF293 results in H-MORPH colony morphology with significantly increased colony furrows and the percent vegetative mycelia relative to AF293 (Fig. 4A and C). Interestingly, the colony morphology in hypoxia (0.2% O₂) is also distinctly different as a result of bafA overexpression. Unlike AF293, the colony in hypoxia is small, dense and lacks furrows and conidiation (Fig. 4A), resembling the previously published colony morphology resulting from constitutive hrmA expression (26).

• Open in new tab
• Download powerpoint

FIG 4

Introduction of the HAC cryptic gene bafA is sufficient to generate H-MORPH in AF293 and impacts biofilm architecture in the baf⁺ strain CEA10. (A) 96-h colony biofilms in normoxia (21% O₂) and hypoxia (0.2% O₂) of AF293 and AF293 with the overexpression of bafA. Images are representative of three independent biological samples. (B) 96-h colony biofilms in normoxia (21% O₂) and hypoxia (0.2% O₂) of CEA10 and CEA10 with the overexpression of bafA. Images are representative of three independent biological samples. (C) Quantification of the H-MORPH features from colony biofilms of three independent biological samples. Student two-tailed nonparametric t tests were performed between each isogenic strain set. (D) Representative images of submerged biofilms (n = 3 biological samples) on the orthogonal plane (xz). Scale bar, 200 μm. (E) Heat map displaying the architecture of the fungal biofilms measured as the deviation of the hyphae from a vertical axis. Each column is representative of a minimum of three independent biological samples. (F) Adherence to plastic measured through a crystal violet assay. Dashed line marks the mean value for media alone. A Student two-tailed nonparametric t test was performed between isogenic strain sets. n = 6 independent biological samples. (G) Ratio of fungal biomass in hypoxia (0.2% O₂) relative to fungal biomass in normoxia (21% O₂) (H/N) in shaking-flask cultures. Student two-tailed nonparametric t test performed between isogenic strain sets. n = 4 independent biological samples. (H) Gene expression measured by qRT-PCR for bafA, bafB, and bafC in AF293 and CEA10. n = 4 independent biological samples. Error bars indicate standard errors around the mean.

The strain CEA10 contains HAC, H_BAC, and H_CAC but, like AF293, bafA expression is below the level of detection by qRT-PCR in biofilm cultures but can be detected after introduction of a second overexpressed bafA allele (see Fig. S6B). Elevated expression of bafA in CEA10 qualitatively alters the colony morphology in normal (21% O₂) and low (0.2% O₂) oxygen and significantly increases the percent vegetative mycelia (Fig. 4B and C). However, no colony furrows are present as a result of bafA constitutive expression in CEA10 (Fig. 4C). Despite the absence of this macroscopic H-MORPH feature, overexpression of bafA in CEA10 and in AF293 impacts biofilm architecture by increasing the deviation of hyphae from the vertical axis above the bottom 50 μm of the biofilm (Fig. 4D and E). Unlike AF293, even during hypoxic growth CEA10 colonies do not feature furrows, and instead abundant aerial hyphae develop, generating a “fluffy” colony morphotype. We speculate that perhaps there is a dichotomy among strains of A. fumigatus where some respond to low oxygen by forming aerial hyphae (i.e., CEA10) and others develop furrows (i.e., AF293).

H-MORPH in EVOL20, as well as other clinical isolates, coincides with reduced adherence and increased hypoxic fitness (hypoxic growth relative to normoxia growth, H/N) (26). In both CEA10 and AF293, overexpression of bafA significantly reduces hyphal adherence to plastic (Fig. 4F). Despite documented differences in hypoxic growth between AF293 and CEA10, bafA overexpression also significantly increases the hypoxic fitness of both strains, though to a lesser extent in CEA10 (Fig. 4G) (30). The inability for bafA expression to impact CEA10 colony morphology and its apparent reduced impact on adherence and hypoxic growth relative to AF293 may be explained by the presence of the other baf genes encoded in the CEA10 genome. Although bafA mRNA levels are undetectable in CEA10 during normal oxygen growth, mRNA for both bafB and bafC is detected (Fig. 4H). Since the amino acid identity between these three proteins ranges from 45 to 78%, we hypothesize that bafB and bafC are also sufficient to impact colony and biofilm morphology.

Overexpression of the bafA orthologs bafB and bafC generates H-MORPH-like phenotypes and impacts hypoxic growth.To determine whether bafB and bafC are sufficient to generate H-MORPH phenotypes in the independent reference strains AF293 and CEA10, we used a constitutive promoter to drive expression of these genes and assessed colony morphology, adherence, and biofilm architecture. Introduction of either bafB or bafC in AF293 generates features of H-MORPH in normoxia with significantly increased furrows and the percent vegetative mycelia (Fig. 5A and C). Similar to bafA overexpression in CEA10, bafB overexpression did not induce H-MORPH features of colony furrows and increased the percent vegetative mycelia in CEA10 (Fig. 5B and D). However, bafB expression significantly reduced overall conidiation in normoxia (21% O₂) and hypoxia (0.2% O₂), a complementary metric to the percent vegetative mycelia (Fig. 5E). Overexpression of bafC in CEA10 is unique in that it does significantly increase colony furrows in normoxia relative to CEA10 (Fig. 5B and D). However, the percent vegetative mycelia is not significantly increased (Fig. 5D).

• Open in new tab
• Download powerpoint

FIG 5

Introduction of bafB and bafC impact colony and submerged biofilm morphology in independent strain backgrounds. (A) 96-h colony biofilms in normoxia (21% O₂) and hypoxia (0.2% O₂) of AF293 and AF293 with the overexpression of bafB or bafC. Images are representative of three independent biological samples. (B) 96-h colony biofilms in normoxia (21% O₂) and hypoxia (0.2% O₂) of CEA10 and CEA10 with the overexpression of bafB or bafC. Images are representative of three independent biological samples. (C) Quantification of the H-MORPH features from colony biofilms of AF293, AF293 bafB^OE, and AF293 bafC^OE with three independent biological samples. One-way ANOVAs with Dunnett’s posttest for multiple comparisons relative to AF293 were performed. (D) Quantification of the H-MORPH features from colony biofilms of CEA10, CEA10 bafB^OE, and CEA10 bafC^OE with three independent biological samples. One-way ANOVAs with Dunnett’s posttest for multiple comparisons relative to CEA10 were performed. (E) Quantification of conidiation from three independent biological samples of CEA10 and CEA10 bafB^OE in normoxia (21% O₂) or hypoxia (0.2% O₂). Student two-tailed nonparametric t tests were performed between CEA10 and CEA10 bafB^OE for each time point. (a, P = 0.0004; b, P = 0.0006; c, P < 0.0001). (F) Adherence to plastic measured through a crystal violet assay. Dashed line marks the mean value for media alone. A one-way ANOVA with Dunnett’s posttest for multiple comparisons was performed between isogenic strain sets relative to AF293 or CEA10. n = 6 independent biological samples for AF293 strains and n = 8 independent biological samples for CEA10 strains. (G) Representative images of submerged biofilms (n = 3 biological samples) on the orthogonal plane (xz) of AF293, AF293 bafB^OE, and AF293 bafC^OE. Scale bar, 200 μm. (G) Representative images of submerged biofilms (n = 3 biological samples) on the orthogonal plane (xz) of CEA10, CEA10 bafB^OE, and CEA10 bafC^OE. Scale bar, 200 μm. (I) Heat map displaying the architecture of the fungal biofilms measured as the deviation of the hyphae from a vertical axis. Each column is representative of a minimum of three independent biological samples.

Despite variation in how the baf genes impact colony morphology in the two strain backgrounds, in both AF293 and CEA10 overexpression of bafB or bafC results in significantly reduced adherence to plastic (Fig. 5F). CEA10 adheres less well to plastic compared to AF293, and the difference in adherence is smaller as a result of bafB or bafC overexpression. Since these two genes are already present and expressed in CEA10 (Fig. 4H), it is possible that this native baf expression contributes to this difference between CEA10 and AF293.

As putative biofilm architecture factors, we sought to confirm an impact of bafB and bafC on biofilm architecture, similar to that observed with elevated expression of bafA (Fig. 3D and E). In AF293, overexpression of bafB visibly impacts biofilm architecture and formation in the xz (Fig. 5G) and xy (see Fig. S6C) dimensions. The xy dimension reveals dense hyphal growth and abundant hyphal branching (see Fig. S6C). The xz dimension shows a stunted 24-h biofilm that reaches heights of only 200 to 250 μm (Fig. 5G). Similarly, regions of the 24-h biofilms generated by the overexpression of bafC in AF293 (AF293 bafC^OE) are also stunted with evidence of hyphae that are hyperbranching (Fig. 5G; see also Fig. S6C). In regards to biofilm architecture as defined by hyphal orientation to the vertical axis, overexpression of bafC but not bafB in AF293 results in increased deviation from the vertical axis above 50 μm (Fig. 5G and H). Notably, constitutive expression of bafA, bafB, or bafC in AF293 also impacts morphology during liquid growth similar to that of EVOL20 (see Fig. S6D).

In CEA10 biofilms, overexpression of bafB and bafC results in increased deviation from the vertical axis above 50 μm in 24 h biofilms (Fig. 5I and J). There is also qualitative evidence for hyper branching as a result of elevated bafB or bafC expression in CEA10 (see Fig. S6E). These data support a role for all three proposed baf genes in biofilm architecture, through multiple metrics, in two independent strain backgrounds of A. fumigatus.

Introduction of A. fumigatus bafA into Aspergillus niger generates H-MORPH and simultaneously increases hypoxic growth.We have previously reported that among aspergilli, hrmA is absent from the notable species of A. nidulans, A. oryzae, A. flavus, and A. niger based on available genome sequences (26). However, Aspergillus niger strain CBS 513.88 encodes a gene, An08g12010, with 69% nucleotide identity to A. fumigatus bafA and 41.03% amino acid identity to the predicted protein sequence of BafA (see Fig. S7A). This suggests that the role of baf or baf-like genes may be conserved in other Aspergillus species. We sought to determine whether A. fumigatus bafA (AfbafA) could influence colony morphology, biofilm architecture, hypoxic growth, and adherence in the A. niger reference strain A1144. This strain was selected for its robust growth at 37°C and the ease at which it is genetically manipulated.

We overexpressed A. fumigatus bafA in A. niger with the constitutive gpdA promoter to generate An AfbafA^OE (see Fig. S7B). Overexpression of bafA in A. niger generated H-MORPH colonies with significantly increased colony furrows and percent vegetative mycelia compared to the control A1144 (Fig. 6A and B). Intriguingly, the overexpression of AfbafA in A. niger resulted in the production of a bright yellow pigment, shown here in two independent transformants (Fig. 6A). The production of yellow pigments by A. niger has been noted in the literature for decades as a result of various growth conditions and genetic manipulations (36).

• Open in new tab
• Download powerpoint

FIG 6

Introduction of A. fumigatus bafA is sufficient to generate H-MORPH in A. niger. (A) 96-h colony biofilms in normoxia (21% O₂) and hypoxia (0.2% O₂) of A. niger reference strain A1144 and two independent strains of A1144 with the overexpression of A. fumigatus bafA (AfbafA^OE). Images are representative of three independent biological samples. (B) Quantification of the H-MORPH features from colony biofilms of three independent biological samples in normoxia (21% O₂). One-way ANOVA with Dunnett’s posttest was performed for multiple comparisons relative to A1144. (C) Representative images of submerged biofilms (n = 3 biological samples) on the orthogonal plane (xz). Scale bar, 200 μm. (D) Heat map displaying the architecture of the fungal biofilms measured as the deviation of the hyphae from a vertical axis. Each column is representative of a minimum of three independent biological samples. (E) The ratio of fungal biomass in hypoxia (0.2% O₂) relative to fungal biomass in normoxia (21% O₂) (H/N) in shaking-flask cultures. A Student two-tailed nonparametric t test was performed between isogenic strain sets. n = 3 independent biological samples. (F) Adherence to plastic measured through a crystal violet assay. A Student two-tailed nonparametric t test was performed between samples within each media type. n = 6 independent biological samples for minimal media and n = 7 independent biological samples for complex media (minimal media with yeast extract).

Similar to A. fumigatus, the A. niger reference strain A1144 forms a submerged biofilm with dense filaments within the first 50 μm that are oriented perpendicular to the vertical axis (Fig. 6C and D). Above the ∼50 μm at the base of the biofilm, filaments become oriented more closely along the vertical axis, similar to what has been observed with N-MORPH strains of A. fumigatus (i.e., AF293) (Fig. 6C and D). Introduction of the constitutively expressed AfbafA alters the biofilm of A1144. At 24 h, the hyphae are stunted reaching heights of only 200 to 250 μm in height (Fig. 6C). These stunted filaments highly deviate from the vertical axis throughout the height of the biofilm indicating that AfbafA is capable of impacting biofilm architecture across fungal species (Fig. 6D).

Not only does AfbafA impact the colony morphology to generate H-MORPH and modulate the biofilm architecture, but it also generates other H-MORPH and EVOL20 associated phenotypes, including increased hypoxia fitness and reduced adherence. In AF293 and CEA10 expression of bafA results in increased hypoxia fitness (hypoxic growth normalized to normoxic growth); similarly, the hypoxia fitness of A1144 significantly increases with constitutive expression of AfbafA (Fig. 6E). Adherence of A. fumigatus is quantified in minimal media; however, the adherence of the reference A. niger strain A1144 is low in minimal media. Thus, we opted to quantify the impact of AfbafA on A. niger adherence in both minimal and complex media, where A1144 adherence is more robust. Under both conditions, adherence was significantly reduced with expression of AfbafA compared to A1144 (Fig. 6F). Not only do reduced adherence and increased hypoxia fitness track with H-MORPH on the macroscale and microscale, as has been observed previously, but they do so as a result of bafA expression across different Aspergillus species. The ability of bafA alone to generate these phenotypes in the two independent species of Aspergillus supports its role as the effector protein of HAC and supports its application to modify biofilm architecture and function in Aspergillus species.