The Lysine Deacetylase RpdA Is Essential for Virulence in Aspergillus fumigatus

Current suboptimal treatment options of invasive fungal infections and emerging resistance of the corresponding pathogens urge the need for alternative therapy strategies and require the identification of novel antifungal targets. Aspergillus fumigatus is the most common airborne opportunistic mold pathogen causing invasive and often fatal disease. Establishing a novel in vivo conditional gene expression system, we demonstrate that downregulation of the class 1 lysine deacetylase (KDAC) RpdA leads to avirulence of A. fumigatus in a murine model for pulmonary aspergillosis. The xylP promoter used has previously been shown to allow xylose-induced gene expression in different molds. Here, we demonstrate for the first time that this promoter also allows in vivo tuning of A. fumigatus gene activity by supplying xylose in the drinking water of mice. In the absence of xylose, an A. fumigatus strain expressing rpdA under control of the xylP promoter, rpdAxylP, was avirulent and lung histology showed significantly less fungal growth. With xylose, however, rpdAxylP displayed full virulence demonstrating that xylose was taken up by the mouse, transported to the site of fungal infection and caused rpdA induction in vivo. These results demonstrate that (i) RpdA is a promising target for novel antifungal therapies and (ii) the xylP expression system is a powerful new tool for in vivo gene silencing in A. fumigatus.


INTRODUCTION
Aspergillus fumigatus is the most common airborne mold pathogen, capable of causing systemic disease, termed invasive pulmonary aspergillosis, mostly in immunocompromised patients (van de Veerdonk et al., 2017). Difficulties in diagnosis and the emergence of azole-resistant clinical isolates result in high mortality rates associated with invasive pulmonary aspergillosis (Meis et al., 2016;Fisher et al., 2018). Additionally, antifungal drugs used in the clinic suffer from poor specificity and side effects in patients (Campoy and Adrio, 2017). Consequently, there is an urgent need for improvement of antifungal prophylaxis, diagnosis, and therapy (Denning and Bromley, 2015).
Chromatin modulators represent potential antifungal targets. Chromatin is composed of DNA, histones and other proteins ensuring compact organization of the genetic material. The highly conserved N-terminal tails of histones are subject to a variety of post-translational modifications, which significantly impact the expression of genes. One of these modifications is the reversible acetylation of distinct lysine residues, catalyzed by lysine acetyltransferases and their counterparts, lysine deacetylases (KDACs), originally termed histone deacetylases (HDACs). The term KDAC appears more appropriate as it became evident that non-histone proteins are also subject to acetylation by the very same enzymes (reviewed by Narita et al., 2019). One example for an A. fumigatus non-histone protein, whose acetylation status has been proposed to have significant implications on virulence, is the heat shock protein 90 (Lamoth et al., 2014).
In A. fumigatus, the KDAC RpdA was recently shown to be essential and was therefore proposed as a potential new target for antifungal therapies (Bauer et al., 2016). RpdA-type KDACs share a catalytic domain that is highly conserved throughout eukaryotes and that is druggable; several such inhibitors are FDA-approved for cancer treatment (West and Johnstone, 2014). Furthermore, RpdA proteins contain a conserved fungal-specific C-terminal motif required for nuclear localization, catalytic activity, and viability in axenic growth (Tribus et al., 2010;Bauer et al., 2016), which might allow development of fungal-specific KDAC inhibitors.
Here, we demonstrate for the first time, that (i) the xylP promoter allows the control of A. fumigatus gene expression during infection in a non-neutropenic murine pulmonary aspergillosis model and (ii) downregulation of rpdA renders A. fumigatus avirulent. These data confirm RpdA as promising target for KDAC inhibitors with potential as novel agents against invasive fungal infections.

Depletion of RpdA Leads to a Drastic but Reversible Growth Reduction Under Axenic Growth Conditions
To clarify the role of RpdA (Afu2g03390) during infection, we used for the first time the xylP promoter for conditional gene expression of rpdA in a murine aspergillosis model. Therefore, the endogenous rpdA promoter was replaced by the xylP promoter with concomitant integration of a pyrithiamine resistance cassette via homologous recombination in A. fumigatus AfS35, termed wild type here, resulting in rpdA xylP transformants, which were verified by Southern blot analysis (Supplementary Figure S1).
Although RpdA has been shown to be essential in A. fumigatus by heterokaryon rescue analysis (Bauer et al., 2016), rpdA xylP displayed weak growth without sporulation on solid minimal medium ( Figure 1B), solid complex medium ( Figure 1B) and in liquid medium without xylose-induction ( Figure 1C). Xylose rescued growth of rpdA xylP in a concentration-dependent manner confirming the functionality of this expression system. Interestingly, radial growth of rpdA xylP did not fully reach the wild-type degree (Figure 1B), which might reflect differences in the 5 -UTR of rpdA in the rpdA xylP strain, as the rpdA mRNA in rpdA xylP contains the 5 -UTR of the xylP mRNA. Alternatively, fine-tuned regulation of rpdA expression during growth via the endogenous promoter might be of importance. Northern analysis confirmed downregulation of rpdA in the absence of xylose and a xylose-concentration-dependent induction in strain rpdA xylP (Figure 1D), while xylose did not affect rpdA expression in the wild type.
The weak growth of rpdA xylP without xylose induction might be caused by slight leakiness of the xylP promoter and/or by  (Haas et al., 1993). Promoter motifs including putative binding sites for the xylose-sensing transcription factor XlnR have been predicted previously (Zadra et al., 2000). parental carry-over of xylose or RpdA via the conidia used for inoculation (see below). After 5 days of growth the colony area of rpdA xylP without xylose induction reached 75% of that of rpdA xylP with xylose induction ( Figure 1E). An impact of parental RpdA carry-over during germination cannot be completely excluded. Nevertheless, the presented data indicate that growth of rpdA xylP without xylose induction is caused by leakiness of the xylP promoter as dilution of parental RpdA via growth would be expected to decrease and finally block the growth rpdA xylP without xylose induction within this time frame.
Notably, radial growth of rpdA xylP without xylose displayed significant differences to rpdA xylP with xylose induction: The mycelial layer was significantly thinner (data not shown) and conidiation was largely blocked in rpdA xylP without induction ( Figure 1E). Similarly, RpdA was found to play a crucial role in conidiation in A. nidulans (Tribus et al., 2010). The KDAC inhibitor trichostatin A (TSA, 10 µM) decreased the growth of wild type and xylose-induced rpdA xylP but completely blocked growth of rpdA xylP without xylose induction ( Figure 1F). These data strongly indicate that RpdA is indeed the target for the growth-inhibitory activity of TSA as suggested previously in A. nidulans (Bauer et al., 2016).

Xylose Feeding of Mice Leads to Rising Xylose Concentrations in Plasma
To test if xylose is absorbed by mice to levels sufficient for in vivo induction of the xylP promoter, 0.2 ml of a 10% (w/v) xylose solution was administered to mice by oral gavage, followed by determination of the plasma xylose concentration 15-240 min after xylose ingestion (Figure 2). No xylose was detected in plasma samples taken directly after the gavage, indicating the absence of xylose under conventional nutritional conditions. Already after 30 min, a plasma xylose concentration of 0.055% was reached. The xylose concentration FIGURE 2 | Xylose plasma levels reach concentrations that are sufficient to promote growth of rpdA xylP in vitro. 0.2 ml of a 10% xylose solution were administered to a mouse by oral gavage. Blood samples were taken after the time points indicated and plasma xylose concentrations were measured in triplicates (mean ± standard deviation is indicated). A cubic spline interpolation line illustrating the course of xylose plasma levels is shown in gray.
decreased after 60 min and was hardly detectable after 120 min indicating efficient clearing. As reported previously, mice drink 4-8 ml water within 24 h (Bachmanov et al., 2002), resulting in approx. 0.25 ml/h calculated water uptake (Li et al., 2016). These data indicated that drinking of xylose results in sufficient concentrations for in vivo induction of the xylP promoter.

Downregulation of rpdA Causes Avirulence
Next, the rpdA xylP strain was tested for virulence in a nonneutropenic pulmonary infection model, in which mice are immunosuppressed with cortisone acetate. Three days after the first immunosuppression, cohorts of ten mice were infected intranasally with wild-type or rpdA xylP conidia and were supplied with drinking water with or without xylose (10% m/v). For mice infected with the wild-type strain, the xylose diet did not affect mortality (Supplementary Figure S2), weight loss, or drinking behavior (data not shown). The latter is interesting as xylose supplementation results in a sweet taste of the drinking water, which might affect drinking behavior.
The rpdA xylP strain with xylose supplementation showed mortality rates similar to that of the wild-type strain without xylose in the drinking water (Figure 3). In contrast, rpdA xylP without xylose was avirulent (P < 0.0001, Figure 3). Histological analysis 2 days post infection revealed a similar progress in germination of wild-type and rpdA xylP strains with and without xylose (Figure 4C), which is in agreement with the growth of rpdA xylP under both inducing and non-inducing axenic conditions (Figures 1B,C), although the growth was very limited under non-inducing conditions. The rpdA xylP mutant without xylose caused a lower lung fungal burden compared to rpdA xylP with xylose (P = 0.0292, Figure 4A) −xylose (xyl) compared to wild type (P = 0.0094) and rpdA xylP +xylose (P = 0.0292). (B) Hyphal lesions from rpdA xylP -xylose contain significantly less fungal hyphae (P < 0.0001) than those from rpdA xylP +xylose and wild type. The extend of fungal growth was measured by manually tracing the combined length of hyphae in each lesion (n = 10) using ImageJ software. (C) Representative lung sections of infected mice. Staining, performed with gomori methenamine silver (GMS), stains fungi black.
or wild type without xylose (P = 0.0094, Figure 4A). Moreover, examination of histological lung sections revealed less and smaller fungal lesions in mice infected with rpdA xylP maintained without xylose, as determined by measuring the total length of fungal hyphae within 10 infection foci each of rpdA xylP with and without xylose and wild-type (P < 0.0001, Figures 4B,C). These results match the partial growth of rpdA xylP in the absence of xylose during axenic growth conditions (Figures 1B,C).
Taken together, these data demonstrate that downregulation of rpdA (rpdA xylP under non-inducing conditions), which allows limited axenic growth (Figures 1B,C), permits germination during infection (Figures 4B,C) but is insufficient to support virulence (Figure 3). Moreover, the avirulence of rpdA xylP without xylose supplementation compared to the virulence of rpdA xylP with xylose supplementation (Figure 3) demonstrates that the xylP promoter allows control of expression during infection by xylose feeding.

DISCUSSION
It has been shown that the class 1 KDAC RpdA is essential for growth of both A. nidulans and A. fumigatus under axenic conditions (Tribus et al., 2010;Bauer et al., 2016). Concordantly, pharmacological KDAC inhibition resulted in strong growth inhibition (Bauer et al., 2016). Here, we employed for the first time the xylP promoter, which has previously been shown to allow xylose-mediated induction of gene expression in several mold species during axenic conditions, for modulation of A. fumigatus gene expression during murine infection.
The observed wild-type-like virulence of rpdA xylP under xylose supplementation demonstrates that the xylP inducer xylose indeed reaches the infected tissue in amounts sufficient to induce rpdA to a level that can rescue virulence. The employed xylP system therefore appears to be a suitable new tool to conduct virulence studies as an interesting alternative to gene deletion mutants or expression systems that are induced or repressed by tetracycline/doxycycline (Tet-On/Tet-Off promoter systems). The latter systems have been introduced for A. fumigatus in several axenic approaches (Vogt et al., 2005;Helmschrott et al., 2013;Kalb et al., 2015;Macheleidt et al., 2015;Wanka et al., 2016) and there are four studies employing these promoters for control of A. fumigatus gene expression during infection: three for Tet-On (Lin et al., 2015;Sasse et al., 2016;Souza et al., 2019) and one for the Tet-Off system (Peng et al., 2018). A disadvantage of administration of tetracycline-type antibiotics might be their potential side effects in host mitochondria and on the animal microbiome (Ahler et al., 2013;Chatzispyrou et al., 2015;Moullan et al., 2015). Moreover, the Tet-On/Tet-Off systems require not only exchange of the promoter, as in the xylP system, but additionally incorporation of the Tet transcription factor (Lin et al., 2015;Sasse et al., 2016). The half-life of doxycycline in blood was determined to be approximately 170 min in healthy mice (Böcker et al., 1981(Böcker et al., , 1984. Due to the shorter in vivo half-life of xylose in blood of approximately 35 min, the xylP promoter system might be an attractive alternative, providing different kinetics. As RpdA has been shown previously to be essential for growth (Bauer et al., 2016), the limited growth of the rpdA xylP strain in the absence of xylose indicates slight leakiness of the xylP promoter. Expression of various genes under control of the xylP promoter in several mold fungi caused a loss-of-function phenotype under non-inducing conditions (Pongsunk et al., 2005;Hartmann et al., 2010;Sigl et al., 2010;Fazius et al., 2012;Gsaller et al., 2012;Bugeja et al., 2013;Baldin et al., 2015;Vaknin et al., 2016), e.g., growth of mutant strains expressing essential genes of the iron sulfur cluster biosynthetic machinery under control of the xylP promoter was completely blocked in non-inducing conditions in A. fumigatus (Misslinger et al., 2018). In contrast, use of the xylP promoter for expression of the essential monothiol glutaredoxin encoding grxD gene in A. fumigatus indicated leakiness of the xylP promoter dependent on the grxD allele (Misslinger et al., 2019). Most likely, the consequence of promoter leakiness, as monitored by growth pattern, depends on the gene expressed; i.e., it is affected by the required amount of the gene product, whereby the efficiency of expression is influenced by mRNA and protein stability, which are gene-specific features. Moreover, the genomic integration site might affect promoter leakiness. Similar to A. fumigatus, xylP promoter-driven rpdA without induction allowed minimal growth also in A. nidulans (Tribus et al., 2010).
At first glance, promoter leakiness might appear unfavorable. Under certain circumstances, however, it actually might be advantageous. The avirulence of rpdA xylP without xylose induction underlines that even a decrease and not only absence of rpdA expression is sufficient to impair pathogenicity. This result actually emphasizes the importance of RpdA as a virulence factor. Moreover, residual RpdA activity in rpdA xylP strains under repression mimics a realistic drug treatment scenario much more than an entire blocking of gene expression, as drug-mediated therapies hardly ever cause total inhibition of the respective targets. Consequently, only few therapy regimens achieve the effect of a total inactivation of the corresponding targets or physiological pathways. Furthermore, in studies evaluating potential drug targets with molds, generally conidia are used for pathogen inoculation. Very often it is observed that the avirulence of mutants is based on the lack of germination of the conidia in the lung. However, this does not reflect the real scenario for therapeutic intervention, as treatment of patients has to combat hyphae but not conidia. In this respect, xylP promoter-mediated downregulation of gene expression represents an interesting alternative to gene deletion, not only for the evaluation of the role in virulence of essential, but also for non-essential genes. On the other hand, regarding a potential broader application, promoter leakiness could indeed impede significant knock-down, particularly of weakly expressed genes.
Growth of rpdA xylP without xylose induction was completely blocked by the KDAC inhibitor TSA in a concentration that caused only partial growth inhibition of wild type or xyloseinduced rpdA xylP . These data indicate that the antifungal effect of TSA is mainly due to inhibition of catalytic RpdA activity.
The avirulence of the rpdA xylP strain under non-inducing conditions might be attributed to its reduced growth. Remarkably, deletion of the mitogen-activated protein kinase MpkA-encoding gene caused a dramatic growth defect without diminishing virulence (Valiante et al., 2008). The latter indicates that growth rate and virulence are not always strictly linked. If avirulence of rpdA xylP under non-inducing conditions is caused by the growth defect or a deregulation of virulence-relevant pathways remains to be elucidated.
Irrespective of the mode of action of RpdA, several KDAC inhibitors have already been clinically approved by the FDA for therapeutical use against certain types of cancer: e.g., vorinostat, panobinostat, belinostat, and romidepsin (West and Johnstone, 2014). It is, however, not clear yet, to which extent these compounds inhibit RpdA activity or affect growth of A. fumigatus in vitro or in vivo. Importantly, RpdA itself seems to be a promising candidate for the development of fungal-specific KDAC inhibitors due to specific structure variations within its catalytic domain and the presence of a conserved fungal-specific C-terminal motif (Bauer et al., 2016).
Taken together, in this study we demonstrate the crucial role of RpdA in virulence of A. fumigatus employing a novel expression system, which enables tuning of expression during the infection process.

Strains and Media
For this study, the rpdA promoter was exchanged by the xylP promoter in the strain AfS35, which lacks non-homologous end joining due to deletion of the akuA gene (Krappmann et al., 2006). Information on these strains is given in Table 1. Fungal strains were propagated in Aspergillus minimal medium (AMM: 1% (m/v) glucose, 20 mM glutamine, salt solution and trace elements according to Pontecorvo et al., 1953) and in Aspergillus complete medium (CM: 2% (m/v) glucose, 0.2% (m/v) peptone, 0.1% (m/v) yeast extract, 0.1% (m/v) casamino acids, salt solution and trace elements). Addition of xylose is indicated separately.

Gene Cassette and Strain Generation
For generation of the rpdA xylP allele, protoplast fusion (Tilburn et al., 1983) combined with the split-marker technique (Nielsen et al., 2006) was used. A. fumigatus AfS35 was co-transformed with 2 DNA fragments, each containing 0.5-kb overlapping but incomplete fragments of the pyrithiamine resistance-conferring ptrA allele from Aspergillus oryzae (Kubodera et al., 2000) ligated to 1.2-and 1-kb of A. fumigatus rpdA 5 -UTR and rpdA coding regions, respectively. These fragments were amplified from plasmid pIB24, which was constructed as follows: First, the rpdA 5 -UTR was amplified using primers AfrpdA5UTRfwdKpnI (AAATGGTACCATCCAACGACTGTTTACCATCCAACGACT GTT) and AfrpdA5UTRrevNdeI (TTTATAATACATATGTGAG AAAGGTCGTGTATGTGAGAAAGGTCGTG). The resulting product was digested with KpnI and NdeI and cloned into pSK275 (Szewczyk and Krappmann, 2010) harboring the pyrithiamine resistance cassette digested with the same enzymes,

AfS35
This study yielding pSK275-rpdA5 UTR. Next, a fragment containing the P. chrysogenum xylP promoter (Zadra et al., 2000) fused to the rpdA coding sequence was inserted at the SmaI site of pSK275-rpdA5 UTR and screened for proper orientation by restriction analysis. To this end, the xylP promoter fragment was amplified using primers xylPf2 (CCCGGGCACTGATGCGAGC AACAG) and AfrpdAxylPr (CGGAGGAGCAGCCATGTT GGTTCTTCGAGTCG) with pxylPrpdATRU0 (Tribus et al., 2010) as template and then spliced by overlap extension (Shevchuk et al., 2004) to a part of the coding sequence of rpdA that was amplified from A. fumigatus genomic DNA with primers xylPAfrpdAf (CGACTCGAAGAACCAACATGGCTGCTCCTC CGATTG) and AfrpdA7r (ACTGTCACCACCACACTG). All PCR reactions were performed using the Pfu-X7 polymerase (Nørholm, 2010).

RNA Isolation and Northern Blot Analysis
RNA was isolated with TRI Reagent (Sigma) according to the manufacturer's manual. 10 µg of RNA was used for electrophoresis on a 0.6 M formaldehyde agarose gel and blotted on Hybond-N-membranes (Amersham). A digoxigenin (DIG)-labeled rpdA probe was amplified by PCR using primers AfrpdA5f (CAAGTACGGAGAATACTTCC) and AfrpdA6r (CTTCGTCAAGGTCATCCAG) and detected with an anti-DIG-AP conjugate and the chemiluminescent substrate CSPD (Sigma).

Southern Blot Analysis
Genomic DNA was extracted by PCI and precipitated with isopropanol. For verification of correct xylP promoter integration (xylP(p):rpdA), DNA was digested with PstI and signals were detected with DIG-labeled probes against the 5 part of the rpdA coding sequence. Correct integration at the rpdA locus yielded a signal at 3422 bp. The DIGlabeled rpdA probe was amplified from pIB24 by PCR using oligonucleotides xylPf3 (TTCATCGACTCGAAGAAC) and AfrpdAr2 (GTCGCATTTGTTGCGGTTAAG).

Plasma Xylose Measurements
Mouse plasma xylose concentrations were determined spectrophotometrically using the commercially available D-Xylose-Assay-Kit (Megazyme, Ireland) according to the manufacturer's instructions.

Virulence Studies
Six-week-old female ICR mice were immunocompromised by subcutaneous injection with cortisone acetate (300 mg/kg) 3 days prior to infection, on the day of infection, and 3 and 7 days after infection. Conidia were collected in PBS with 0.2% Tween 20. Mice were infected intranasally with 5 × 10 5 dormant conidia, suspended in 20 µl of PBS with 0.2% Tween 20 (10 µl in each nostril). Mortality was monitored for 21 days. The statistical differences for mouse survival were calculated using the log-rank (Mantel-Cox) test. Differences were considered significant at P-values ≤0.05. For histology and lung fungal burden estimation, 5 mice/group were treated and infected as described above and sacrificed after 48 h for lung extraction. For histology, lungs were paraffin-embedded, sectioned (3-4 µm) and stained with Gomori methenamine silver. Histology section preparations were observed using a BX51 microscope (Olympus, Milan, Italy), and images were captured using a high-resolution DP71 camera (Olympus). Lung fungal burden was assessed by homogenizing the lungs and plating dilutions on Sabouraud dextrose agar (Difco) +0.5% xylose plates for CFU (colony forming unit) enumeration. This study was carried out in accordance with the recommendations of the Ministry of Health (MOH) Animal Welfare Committee, Israel. The protocol was approved by the MOH Animal Welfare Committee, Israel.

DATA AVAILABILITY STATEMENT
All datasets generated for this study are included in the article/Supplementary Material.

ETHICS STATEMENT
The animal study was reviewed and approved by Ministry of Health (MOH) Animal Welfare Committee, Israel.