Identification and Deletion of Tft1, a Predicted Glycosyltransferase Necessary for Cell Wall β-1,3;1,4-Glucan Synthesis in Aspergillus fumigatus

Aspergillus fumigatus is an environmental mold that causes severe, often fatal invasive infections in immunocompromised patients. The search for new antifungal drug targets is critical, and the synthesis of the cell wall represents a potential area to find such a target. Embedded within the main β-1,3-glucan core of the A. fumigatus cell wall is a mixed linkage, β-D-(1,3;1,4)-glucan. The role of this molecule or how it is synthesized is unknown, though it comprises 10% of the glucans within the wall. While this is not a well-studied molecule in fungi, it has been studied in plants. Using the sequences of two plant mixed linkage glucan synthases, a single ortholog was identified in A. fumigatus (Tft1). A strain lacking this enzyme (tft1Δ) was generated along with revertant strains containing the native gene under the control of either the native or a strongly expressing promoter. Immunofluorescence staining with an antibody against β-(1,3;1,4)-glucan and biochemical quantification of this polysaccharide in the tft1Δ strain demonstrated complete loss of this molecule. Reintroduction of the gene into the knockout strain yielded reappearance in amounts that correlated with expected expression of the gene. The loss of Tft1 and mixed linkage glucan yielded no in vitro growth phenotype. However, there was a modest increase in virulence for the tft1Δ strain in a wax worm model. While the precise roles for β-(1,3;1,4)-glucan within A. fumigatus cell wall are still uncertain, it is clear that Tft1 plays a pivotal role in the biosynthesis of this cell wall polysaccharide.


Introduction
Aspergillus fumigatus is a ubiquitous, environmental fungus that is currently the most commonly encountered mold pathogen among severely immunocompromised patients [1]. With advancement in medical therapies that lead to immunosuppression, there has been an increase in the risk of serious, invasive Aspergillus infections [2]. This increased risk, coupled with Plasmid and strain generation.

PCR and Southern blot analysis of tranformants.
Genomic DNA (gDNA) was prepared from fresh mycelium as previously described [9], and gDNA concentration was determined using a NanoDrop (ND100) spectrometer. For PCR analysis of the transformed strains, gDNA was amplified with primers TFT1OL-9 and 10 within the TFT1 ORF to show presence or absence of the gene. In addition, PCR was performed with primers within the resistance cassettes along with primers outside of the construct area to show expected placement of the cassette within the genome. For Southern blot, 20 μg of gDNA from each strain was digested using ScaI and SspI (NEB). Digested gDNA was separated on 0.8% agarose gel and transferred onto nitrocellulose membrane. Hybridization was achieved using a 32 P labeled probes that spanned the entire length of the tft1Δ construct. Film was exposed overnight at -80°C.
Immunofluorescence staining of cell wall glucans.
Cell wall polysaccharide quantification.
Strains were grown for 24 hours at 37°C in Sabaroud dextrose broth with shaking at 220 RPM. Cultures were vacuum strained to collect hyphae and washed with sterile PBS. Hyphae were exhaustively disrupted using a Bead Beater (BioSpec Products) with sterile 0.5mm zirconium oxide beads. Disrupted hyphae were centrifuged at 10,000xg and the pellets were washed with NaCl and ddH 2 O 3 times each. Cell wall pellets were suspended in ddH 2 O, frozen, and lyophilized. Dry cell wall preparations were weighed. 150mg of cell walls from each strain were then incubated in 6ml 1N NaOH at 65°C for 1 hour then centrifuged as above. Soluble fractions were reserved for α-1,3-glucan quantification. Insoluble fractions were treated with 6ml 1N NaOH again. Remaining insoluble material was washed with 50mM sodium acetate, pH 5.6 to balance the pH. Samples were suspended in 6.0 ml of the sodium acetate buffer and split into 500μL aliquots. β-1,3-glucanase (Megazyme) and chitinase (Sigma) were diluted 1:10 prior to enzyme treatment. Samples were then treated exhaustively with β-1,3;1,4-glucanase (Megazyme), β-1,3-glucanase, or chitinase for 1 week at 37°C. For each digestion, three replicates were performed. Reducing sugar assay (4-hydroxybenzhydrazide method) was performed as previously described on each sample to evaluate the amount of reducing sugar released by the enzyme, thus estimating the β-1,3;1,4-glucan, β-1,3-glucan, and chitin content in each cell wall preparation [13]. For α-1,3 glucan quantification, 1 ml of soluble fractions reserved from above were analyzed. 200uL of 5N HCL was added to each tube to precipitate the soluble fraction. Pellets were washed with sodium acetate buffer. Samples were treated with α-1,3-glucanase (Megazyme) and incubated at 37°C overnight. Reducing sugar assay was performed as above on each sample to estimate the relative α-1,3-glucan content in each cell wall preparation [13]. Sigma-Plot using One Way ANOVA analysis was used to calculate significance between strains.

Assessment of in vitro growth.
Serial dilutions of strains were created at 10 6 , 10 5 , and 10 4 spores per ml. 5 μl of each dilution (i.e. 5000, 500, 50 spores) from each strain was plated onto various media, including Aspergillus minimal media (AMM), and AMM containing Farnesol (25μg/mL), Calcoflour White (100μg/mL) or Congo Red (55μg/mL) [14]. Plates were grown for 2 days at 37°C and then evaluated for any growth phenotypes. In addition, spores and hyphae from each strain were subjected to cold conditions (4°C) prior to either inoculation onto AMM and incubation at 37°C (for spores) or moving from the cold conditions to 37°C (for hyphae) to assess the effects of cold incubation on subsequent growth.

Antifungal susceptibility.
Antifungal susceptibility was tested for each strain using standard methods [15]. RPMI 1640 with HEPES (Invitrogen) was used as the medium for the assay. Caspofungin and Nikkomycin (Sigma) were used in concentrations from 64μg/mL to 0.0625μg/mL via serial dilution. 100μL of 1.0x10 4 spores/ml were used to inoculate each well. Growth was assessed visually. Concurrently, MM agar plates with either 1.0 μg/mL of Caspofungin or Nikkomycin were inoculated for observation of radial growth at 37°C for 3-5 days.

Galleria Mellonella challenge.
The Galleria worm model of A. fumigatus virulence was performed as previously described [16]. Briefly, spores suspensions of each strain were made at 1.0x10 8 and 2.0x10 8 spores/mL in 0.01M PBS. G. mellonella (Vanderhorst Wholesale) worms were chosen by weight at 0.2-0.4g and placed in a petri dish with a small amount of wood chips at 15 worms per dish. Hamilton 10μL syringes were used to inject 5.0μL of desired spore suspension into the left proleg of each worm [16]. Worms were incubated at 37°C and observed daily for mortality over a 5 day period. Dead worms were determined by the lack of response to tactile stimulation and removed from the plate. The acumΔ control strain was a kind gift from Scott Filler [17]. SigmaPlot was utilized to analyze data via Kaplan-Meier survival-log rank analysis.

Results
Homology searches and topology.
CSLF and CSLH are plant enzymes that have been shown to have β-1,3;1,4 glycosyltransferase activity in rice and barley, respectively. As such, A. fumigatus orthologs, if present, would potentially have similar activity and likely involved in cell wall synthesis. The two plant protein sequences were blasted against the A. fumigatus database in NCBI. Only one protein in the A. fumigatus database appeared to have any homology to these two enzymes. This protein (XP_748682/Afu3g03620), which was renamed Tft1, was annotated in NCBI as a putative glycosyltransferase. There was only about 30% homology between either plant sequence and Tft1, but the homology present was confined entirely within the "CesL" (Cellulose Synthase-like) domains of both plant sequences. CesL is the putative active site of the CSL glycosyltransferase enzymes [7,8]. In addition, to determine if there were orthologs to Tft1 in other organisms, the Tft1 protein sequence was blasted back into the general NCBI database. Interestingly, there were close orthologs (70-95% identity) in other fungal organisms. All the organisms with orthologs were not only fungi, but all were ascomycetes. These included Neosartorya (the sexual anamorph of A. fumigatus), other species of Aspergillus including A. flavus and A. clavatus, Fusarium oxysporum, and numerous species of Trichoderma and Metarhizium. Finally, the protein sequence of Tft1 was submitted to transmembrane prediction software (TMPRED: http://www.ch.embnet.org/software/TMPRED_form.html). It is predicted to be a 7 pass transmembrane protein, a common feature of many membrane bound glycosyltransferases in fungi [18].

Strain generation and confirmation.
Three strains were developed using Agrobacterium tumefaciens Mediated Transformation (ATMT). One strain with the Tft1 gene deleted (tft1Δ) and two additional revertant strains where the native Tft1 gene was added back to the tft1Δ tranformant under control of either the native promoter (Revtft1) or the strongly constitutive SSA1 promoter (SSA1-Revtft1) [19]. Strain confirmations were achieved by PCR and Southern Blot. PCR was used to amplify the Tft1 gene in each strain. Af293, Revtft1 and SSA1-Revtft1 were all positive for the presence of the Tft1 gene, while the tft1Δ strain was negative for the Tft1 gene (S1 Fig.). To confirm proper insertion of each construct, PCR was performed with a primer outside the construct integration site, coupled with a primer within the construct. All strains had the correct amplification patterns and were determined to have the proper orientation at the site of integration (S1 Fig.). Southern blot was used to determine that only a single, correct integration event was present throughout the genome. Genomic preparations from each strain were digested with restriction enzymes SspI and ScaI, which allowed for the differentiation between the strains with probes aimed at the tft1Δ construct. As shown in S2 Fig., all strains presented the correct banding patterns without any additional, unexpected bands.
Immunofluorescence staining of β-1,3;1,4-glucan on the surface of each strain was performed using an antibody. This antibody has been shown in numerous studies in plants to be specific for this β-1,3;1,4 glucan, with no binding to β-1,3 glucan or other plant cell wall polysaccharides [6,20]. As shown in Fig. 1, in WT (Af293), β-1,3;1,4-glucan staining was present throughout the hyphae in a speckled pattern. On rare occasion, fruiting structures were observed (expected to be rare in the growth conditions used). The fruiting structures stained more brightly than hyphae with this antibody. Pre-treatment of WT hyphae with β-1,3;1,4-glucanase prior to immunostaining abrogated the WT staining. Pretreatment of the other strains yielded the same negative result. Consistent with the proposed role of Tft1, the hyphae from the tft1Δ strain had no staining, appearing identical to results from the no primary antibody control and the β-1,3;1,4-glucanase digestion samples. Reintroduction of the WT gene under the control of the native promoter (Revtft1) yielded staining very similar to the WT strain. When TFT1 was replaced under the control of the strong, constitutive SSA1 promoter, the immunostaining was considerably increased across the entire surface of the hyphae (Fig. 1). Staining was also performed on the WT and tft1Δ strains with an antibody that specifically binds β-1,3 glucan. Very strong staining was observed for both strains, preventing observation of subtle differences between the strains for this polysaccharide (S3 Fig.).
Cell wall polysaccharide quantification.
In order to biochemically quantify the relative amount of a cell wall polysaccharide, a common strategy is to specifically enzymatically digest the polysaccharide in cell wall preparations and measure the amount of new reducing ends formed by the specific enzymatic digestion [13]. This was done for all four strains on the alkali insoluble fraction of the cell wall using a specific endo β-1,3;1,4 glucanase [21], β-1,3 glucanase, and chitinase. Equal masses of the cell wall preparations from all four strains were extracted exhaustively with sodium hydroxide. The alkali soluble portion was reserved and later used subsequently to quantify α-1,3-glucan as a loading control. The alkali insoluble portion derived from each strain was then split and either left untreated or exhaustively digested with the β-1,3;1,4-glucanase, β-1,3 glucanase, or chitinase. As shown in Fig. 2A, the pattern of relative β-1,3;1,4 glucan observed in the cell wall of each strain correlates well with the immunostaining results. Cell walls from tft1Δ had no observable new reducing ends formed after digestion. The revertant Revtft1 strain with replacement of TFT1 with the native promoter had a relative glucose release of 115% that of the parent Af293 strain; and SSA1-revtft1 with expression driven by the strong SSA1 promoter had a relative glucose release of 215% that of Af293. As shown in Fig. 2B, β-1,3 glucanase digestion demonstrated an increase in glucose release in revtft1 and SSA1 -revtft1, while tft1Δ demonstrated no change when compared to Af293. No significant difference in glucose release was observed between any of the four strains after treatment with chitinase (S3 Fig.). To ensure equal amounts of cell wall preparations were used for each strain in the β-1,3;1,4-glucan quantification, the alkali soluble fraction from above was treated with α-1,3-glucanase and evaluated in the same manner. After overnight incubation, all strains were shown to be roughly equal in reducing ends formed by this enzyme (S4 Fig.). Considering that this cell wall component is in a different cell wall compartment and is most likely quantitatively unrelated to the insoluble fraction, this suggests that the cell wall input was similar for all four strains.
Assessment of in vitro growth.
As shown in Fig. 3, both sporulation and radial growth were identical for all four strains on Aspergillus Minimal Media (AMM). Addition of farnesol, an inhibitor of the cell wall integrity pathway [22], into the AMM plates resulted in a slight decrease in sporulation and radial growth, but this effect was observed equally across all strains. Addition of the cell wall stress agents Congo red or calcofluor white into the AMM plates resulted in significant stress on all colonies, but no phenotypes were observed for any of the tft1 strains. In addition, both spores and hyphae shown to release zero glucose when exposed to endo β-1,3;1,4-glucan hydrolase. from the strains were subjected to cold conditions prior to incubation at 37°C. Regardless of the cold incubation time or if hyphae or spores were subjected to cold, there were no differences between the strains in tolerating cold incubation (not shown).
In order to determine the effect of cell wall changes on antifungal susceptibility, minimum effective concentration (MEC) was determined for two cell wall active antifungal agents. Nikkomycin, an inhibitor of chitin synthesis, was shown to have no effect on any strain at any concentration. Caspofungin was shown to have a minimum effective concentration of .0625μg/mL for all strains tested. Wells containing concentrations above .0625μg/mL of caspofungin demonstrated increased growth inhibition. However, no visible differences were observed between strains in susceptibility to this β-1,3 glucan synthase inhibitor.

Galleria Mellonella challenge.
To gain a general understanding of the effect of loss of mixed linkage glucan on virulence, the increasingly popular Galleria model was utilized [17,23]. For each strain tested, 60 worms were injected with 1 x 10 6 spores, and the worms were observed over 5 days for mortality. As shown in Fig. 4, the placebo group that was injected with diluent survived all 5 days of the experiment with no signs of morbidity. The acumΔ strain is a known hypovirulent strain in this model [17] and was used as a hypovirulent control. Morbidity for this control strain was observed on days 3-5, with some worms surviving, as compared to WT (Af293) for which most worms were dead by day 4 and nearly all dead by day 5. Worms injected with tft1Δ were visibly sick by day 1, and the group was completely dead by day 2/3. This hypervirulent phenotype was significant compared to WT (p-value: .00019). Revtft1 and SSA1-revtft1 were shown to have no significant virulence difference as compared to WT.

Discussion
Aspergillus fumigatus is an important filamentous fungal pathogen that can cause life threatening infections in immunocompromised patients. New treatment options for invasive fungal infections, including aspergillosis, are a key need. The cell wall is a structure that is essential for the survival of fungi, and humans do not make an analogous structure. However, while cell wall synthesis represents a promising area to find new anti-fungal drug targets, our understanding of how this structure is built, especially in molds, is crude. The cell wall of A. fumigatus is primarily composed of various cross-linked polysaccharides, including a novel mixed linkage glucan [4]. This β-1,3;1,4 glucan represents approximately 10% of the A. fumigatus cell wall glucan content, but the roles that this molecule plays for this mold are unknown [5]. Further, the enzyme(s) that synthesize this polysaccharide in A. fumigatus are also not known. Mixed linkage glucan is also found in various plants, and it is an important molecule in the agricultural industry, as it is undesirable for various downstream food processing applications [6]. As such, there has been great interest in finding the enzyme that makes this molecule in plants, and there are specific commercially available reagents to detect the presence of mixed linkage glucan. The identity of two such mixed linkage glucan synthases in rice and barley were discovered in 2006 and 2009, respectively [7,8]. Using the sequence of these two enzymes, we found one ortholog in the A. fumigatus genomic database. While this ortholog had only minimal homology with the two plant sequences, the relatively short stretch of homology was completely within the putative active site of the plant enzymes [7,8]. We named this putative synthase Tft1 for Three-four transferase 1. This was annotated as a predicted glycosyltransferase and the TMPRED transmembrane prediction software predicted the sequence to have seven transmembrane passes. The predicted transmembrane localization is consistent with a potential role in cell wall synthesis, as enzymes shown to synthesize cell wall polysaccharides are commonly localized to membranes [18]. Using our recently published gene deletion system for A. fumigatus [9], we deleted the Tft1 gene. With the same system, we also replaced the WT Tft1 gene in the deletion strain under the control of either the WT or a strongly expressing promoter. Loss of the gene led to complete loss of β-1,3;1,4 glucan in the cell wall as detected by both immunofluorescence and biochemical quantitation. Staining of WT hyphae with an antibody specific to β-1,3 glucan yields smooth, linear staining of the entire hyphal surface. In contrast, the pattern of immunofluorescence staining in the WT strain using the antibody specific to β-1,3;1,4 glucan was consistently punctate and relatively weak. This could be due to the comparatively lower prevalence of the molecule in the cell wall and/or the length of β-1,3;1,4 glucan needed for antibody binding. In addition, we also observed much brighter staining on the rarely encountered conidiophores, suggesting that this molecule might play a role in spore formation. While staining of spore preparations was negative, this is not unexpected due to the outer rodlet protein layer on spores that prevents antibody binding to the polysaccharides in the wall beneath [24]. Finally, significantly increased staining of hyphae with the β-1,3;1,4 glucan antibody was observed in the strain overexpressing TftT1. This was reinforced by a 2-fold increase of this molecule when biochemically quantified within the cell wall. As above, this further supports the direct role of this enzyme in the synthesis of the cell wall mixed linkage glucan.
Another interesting observation was that the β-1,3 glucan content in the cell wall appeared to increase in the strain overexpressing β-1,3;1,4 glucan. This could merely represent some digesting of the increased mixed linkage glucan with the β-1,3 glucanase. However, if there is a role for mixed linkage glucan in triggering or anchoring for β-1,3 glucan synthesis, this would also potentially be observed. Unfortunately, there are not good experimental tools to currently test this possibility.
In vitro growth of the tft1Δ strain was essentially the same as both WT and the strain overexpressing the gene on multiple media types and under a number of various conditions, including in the presence of cell wall stress agents or cell wall active antifungals. This would strongly suggest that cell wall β-1,3;1,4 glucan is dispensable for the growth and fitness of A. fumigatus and that a 2-fold excess of the molecule is not deleterious, either. Even though this enzyme does not appear to be a suitable antifungal drug target, we did assess virulence in the strains. Virulence was assessed with the Galleria mellonella wax worm model [16]. Interestingly, the loss of Tft1, and thus mixed linkage glucan, resulted in a slight, albeit significant, hyper-virulent phenotype. Increase of the molecule in the overexpressing strain had the same virulence pattern as WT, suggesting the specific glucanases that degrade this molecule would also not likely be suitable therapeutic targets. While the synthesis or degradation of mixed linkage glucan does not appear to harbor a promising drug target, this does represent the first report of a β-1,3;1,4 glucan synthase in fungi.