The DmtA methyltransferase contributes to Aspergillus flavus conidiation, sclerotial production, aflatoxin biosynthesis and virulence

DNA methylation is essential for epigenetic regulation of gene transcription and development in many animals, plants and fungi. We investigated whether DNA methylation plays a role in the development and secondary metabolism of Aspergillus flavus, identified the DmtA methyltransferase from A. flavus, and produced a dmtA knock-out mutant by replacing the dmtA coding sequence with the pyrG selectable marker. The A. flavus dmtA null mutant lines produced white fluffy mycelium in liquid medium, and displayed a slightly flavescent conidial pigmentation compared with the normal yellow of the wild-type strain when grown on agar. The ΔdmtA lines exhibited decreased conidiation and aflatoxin (AF) biosynthesis, compared with the wild-type line, suggesting that the DmtA knock-out affected the transcriptional level of genes in the AF cluster. In particular, sclerotia development and host colonization were altered in the dmtA null mutants. Green fluorescent protein tagging at the C-terminus of DmtA showed that DmtA localized to the nucleus and cytoplasm. DNA methylation content measurements in the dmtA mutants revealed no widespread DNA methylation in the mutants or wild-type lines. Thus, our findings suggest that DmtA, apart from being a C-5 cytosine methyltransferase in A. flavus, contributes to asexual development, aflatoxin biosynthesis, sclerotial production and virulence.

1 replication foci, a Zinc finger (CXXC-type) domain, a Bromo adjacent homology domain and a C-5 cytosine methyltransferase domain (Fig. 1B). The other DNA methyltransferase, Dnmt3, usually has a PWWP domain, an ADD domain and a C-5 cytosine methyltransferase domain. The novel identified DNA methyltransferase Dnmt2, which only has a C-5 cytosine methyltransferase domain, functions mainly as a tRNA methyltransferase, but not as a DNA methyltransferase (Fig. 1A). The DNA methyltransferase DmtA identified in Aspergillus is highly divergent from the DNA methyltransferases that occur in species where DNA methylation has been validated. These proteins from Aspergillus also share a highly conserved structure (Fig. 1B, Figure S1). Of particular interest, we found that the DmtA proteins possessed by Aspergillus are closely related to two putative Masc1 DMTs in A. immerses and RID in Neurospora. Next, we investigated the transcriptional level of dmtA in the wild-type strain at different time points. Here, the gene AFL2G_08335 encoded histone H3-K79 methytransferease, which we are studying and showing important in A. flavus (data not shown), was used as a positive control for better understanding the expression pattern of DMTA. The results showed that DmtA transcripts were present at very low levels at the 24 h time point; however, during the time where AF production was high (after 48 h, Figure S2), the dmtA transcript level increased dramatically (Fig. 1C), indicating that DmtA might have some function(s) in A. flavus. Therefore, a gene knockout was carried out to study the function of DmtA in A. flavus.

Generation of A. flavus DmtA knockout mutants and phenotypic characterization. To gain insight
into the potential function of DmtA in A. flavus, we first made full-length deletion mutants of the DmtA gene using pyrG selection. The schematic diagram of the genomic region of the dmtA and pyrG genes is shown in Fig. 2A. Among the many transformants obtained, most displayed identical phenotypes. Two uracil/uridine autotrophy isolates (ΔdmtA1-6 and ΔdmtA1-7) were confirmed to be knockouts by Southern blot (Fig. 2B), and the failure of gene expression in ΔdmtA was also verified by RT-PCR (Fig. 2C). ΔdmtA1-6 and ΔdmtA1-7 were selected for further analysis. The colony phenotypes of the ΔdmtA mutants differed markedly from those of the control strain when grown on YES agar medium; they displayed a slightly flavescent conidial pigmentation compared with the yellow coloration of the wild-type line grown on the same medium, while the ΔdmtA mutants grown in YES liquid medium produced white fluffy mycelium (Fig. 2D) which is consistent with the phenotype produced by DNA methyltransferase inhibitor 5-azacytidine (5-AC) treatment 15 . However, the dmtA mutants started to produce some conidia after 7 days (data not shown). The difference between mutants and WT might be due to a delay of conidial formation.

DmtA is important for conidiation in A. flavus.
To examine the role play of DmtA in fungal development and conidiation, 10 3 conidia were pointed onto YES agar at 37 °C and grown for 3 days under both dark and light conditions. The results showed that when grown under such conditions, a significant decrease in conidiation production occurred in the ΔdmtA mutant colonies when compared with the wild-type line (Fig. 3A,B). For further analysis of the conidia defect, we examined the conidiophore formation, which showed that the ΔdmtA mutants produced fewer conidiophores than the wild-type line (Fig. 3C). The alterations observed in the conidiophore numbers for the ΔdmtA strain possibly resulted in decreased conidiation in the mutants. These results indicate that DmtA plays a role in conidiation in A. flavus.
DmtA contributes to aflatoxin biosynthesis. A. flavus is well-known as a saprophytic soil fungus because of its secondary AF metabolites, which are the most toxic and carcinogenic natural contaminants. To examine the effect of DmtA on AF biosynthesis, we tested AF production in the wild-type line and ΔdmtA mutants by thin layer chromatography (TLC) at 5 d, which showed a statistically significant decrease in the ΔdmtA mutants (Fig. 4A). For quantitative analysis of AF production, high-performance liquid chromatography (HPLC) was also used to confirm the presence of AF in the samples. The HPLC peak that eluted after 11.83 min corresponded to aflatoxin B1 (AFB1) and showed that AFB1 production in the wild-type strain was about 20-fold higher than that of ΔdmtA (Fig. 4B). These results indicate that inactivation of DmtA affected AF biosynthesis in A. flavus. To determine whether DmtA regulates AF biosynthesis at the transcript level, biosynthetic AF gene expression was analysed by qPCR at 36 and 72 h time intervals (Fig. 4C). We analysed three AF structure genes, aflC (pksA), aflM (ver-1) and aflO (omtB), and two globally regulated genes, aflR and aflS. Our current results showed that aflC, aflK, aflO, aflS and aflR were severely suppressed in the ΔdmtA mutants at the 36 h and 72 h time intervals (Fig. 4C). The transcript levels of these genes at the 72 h time interval increased concurrently, indicating that expression of these genes might be delayed and co-regulated in the ΔdmtA mutants. Interestingly, the transcript levels of aflO, which encodes O-methyltransferase B, were very low at the 36 h and 72 h time intervals. These results indicate that DmtA might regulate AF biosynthesis by repressing gene expression from the AF cluster.
DmtA is unlikely to be involved in the hyper osmotic stress response. DNA methylation has been shown to play a vital role in many biological processes. Here, we explored the function of DmtA in the morphological development and AF production of A. flavus. To examine the potential roles for DmtA in environmental stress responses, we tested fungal sensitivity to hyperosmotic stress factors and cell-wall damaging agents. Growth of the ΔdmtA mutants in the presence of the osmotic stress agents NaCl and KCl was analysed. Interestingly, the mutant and wild-type lines showed almost the same sensitivities to hyper osmotic stress (1 mol/L NaCl, 1 mol/L KCl and 1.2 mol/L sorbitol) (Fig. 5A). Furthermore, the growth of the ΔdmtA mutant has been analysed in the presence of the cell-wall perturbing agents, Congo red and Calcofluor white. However, the ΔdmtA mutant was not sensitive to these cell wall damaging agents (Fig. 5B). According to our current study DmtA is probably not involved in resistance to hyper osmotic stress in A. flavus. DmtA may play a negative role in sclerotial production in A. flavus. Studies have shown that sclerotial production is linked to AF synthesis 20,21 . To determine if loss of DmtA resulted in aberrations in sclerotial production, the strains were grown on sclerotial inducing Wickerham's (WKM) agar under conditions of dark or light at 37 °C for seven days. After spraying with 75% ethanol to wash away the conidia, the sclerotial phenotype was visualized. To our surprise, unlike the negative effects on AFs production, our results showed that the ΔdmtA had almost quadruple the sclertotia production of the wild type (Fig. 6A,B). Additionally, fewer sclerotia were produced by ΔdmtA under conditions of light than when grown in the dark. These results indicate that DmtA possibly plays a negative role in the formation of sclertotia in A. flavus.

Seed infection is altered in the DmtA mutants.
Although AF is not considered to be a virulence factor and changes in AF production have not been shown to be important in seed infections, we considered it possible that deletion of DmtA might affect seed infections. In this study, we examined the ability of the ∆ dmtA strains to colonize peanut seeds and maize kernel. The wild-type strain tended to produce fluffy mycelium when colonizing peanut seeds and maize kernel (Fig. 7A), but the ∆ dmtA strains appeared to grow more vigorously than the wild-type strain on crop seed (Fig. 7A). Moreover, we measured conidial production in the two strains on seed. Of interest, the ∆ dmtA mutants produced statistically more conidia than the wild-type strain (Fig. 7B).
DmtA is located in the nucleus and cytoplasm. So far, the subcellular location of DmtA has not been reported in Aspergillus. Therefore, to study the cellular location of DmtA, we generated a strain expressing a GFP tag at the C-terminus of DmtA (DmtA-GFP) under the control of its native promoter. During the time period of spore germination, DmtA::GFP accumulated in the tip of swellings (Fig. 8a,b). However, in the vegetative growth period, it showed a strong fluorescence signal in the nucleus and cytoplasm of the hyphae (Fig. 8d,f).

DmtA might not function as a DNA methyltransferase in A. flavus.
To determine whether changes in A. flavus development and AF biosynthesis induced by deletion of DmtA are caused by a reduction in DNA methylation, the 5-methyl-2-deoxycytidine (5-mdC) content of genomic DNA from A. flavus and the ΔdmtA mutants were both analysed by HPLC. The retention time for 2-deoxycytidine (dC) was 5.126 min and 7.771 min for 5-mdC (Fig. 9A). Here, we found that around the retention time of 7.97 min a peak was formed in both chromatograms from the wild-type and the ΔdmtA lines ( Fig. 9B-D), and there was a stronger signal than dC. To validate if it was 5mdC, we mixed the hydrolysed DNAs isolated from the wild-type strain or ΔdmtA mutants with the 5mdC standard. However, two independent peaks were seen around the 7.7 min retention time (Fig. 9E,F), thus indicating that the substance with a retention time of 7.97 min in the chromatograms of the wild-type strain and ΔdmtA mutants was not 5mdC. These results suggest that there was no detectable 5-mdC in the genomic DNA of the A. flavus wild-type strain and the ΔdmtA mutants, which is consistent with the results of a former study 11 . Because of the low level of DNA methylation in the genomic DNA of A. flavus, it seems that DmtA might not function as a DNA methyltransferase in A. flavus.

Discussion
DNA methylation, which is an important epigenetic modification, plays vital roles in regulating gene transcription, transposable element silencing, genome imprinting, X-chromosome inactivation and development in higher eukaryotes, and in genome defence in fungi 1,4,8 . In our previous study, we found that 5-azacytidine, which is a DNA methyltransferase inhibitor, caused a drop in AF production and concurrent morphological changes 15 , thereby hinting that DNA methylation or DMTs might perform roles in development and secondary metabolite biosynthesis in A. flavus, a fungus for which negligible DNA methylation has been found. In this study, to explore the potential role that DNA methylation might have, a C-5 cytosine methyltransferase DmtA from A. flavus has been idetifed from A. flavus. In addition to this, the knock-out mutants of dmtA we generated showed a decrease in AF production and conidiation, and a change in seed infection.
In filamentous fungi, the two putative fungal DMTs, Masc1 and RID, which have low DMTase activities, were shown to function in a genome defence system RIP in Ascobolus immerses, and MIP in Neurospora 8,11,12 . The putative DMT-like protein DMTA from Aspergillus shows high similarity with Masc1 and RID in the phylogenetic tree we generated. In A. nidulans, the Masc1/RID DMT-like protein identified was found to be essential for sexual development 8 . In our current study, we reported that instead of sexual development, a transient cellular process in A. flavus, inactivation of dmtA in A. flavus led to a dramatic decrease in conidiation production. An alteration in conidiophore numbers was also found in the ΔdmtA strain, which might result in decreased conidiation in the mutants.
In addition to conidiation, we found that dmtA affected the expression of genes involved in AF biosynthesis. As one of the most toxic and carcinogenic natural contaminants, AF is biosynthesized by an extremely sophisticated mechanism 19 . To the best of our knowledge, this is the first report on the effect of dmtA on secondary metabolism in fungi. Our results show that AF production was markedly blocked in the dmtA mutants, and the transcript levels of the structure genes aflC, aflK and aflO, together with the regulator genes, aflR and aflS, were severely suppressed compared with those of the wild-type strain. These data suggest that DmtA may be required for activating the AF gene cluster.
To address the effect of DmtA on A. flavus physiology and pathogenicity, we also examined sclerotia formation and host colonization in the ΔdmtA mutants. Interestingly, our results showed that sclerotial production in the ΔdmtA mutants dramatically increased compared with the wild-type line, a result differing from a previous study where enhanced sclerotial production was accompanied with increased AF production 22 . However, there is no direct evidence of a close link between sclerotial formation and AF biosynthesis. In this study, we found that deletion of the dmtA gene resulted in a drop in conidiation but enhanced sclerotial production, findings consistent with previous studies 22,23 . We also noted that ΔdmtA appeared more aggressive in tissue maceration compared with the wild-type line, which produced much more conidia on peanut and maize crops. A previous study showed that degradative enzymes such as lipase and esterase played a significant role in fungal infections of plants by A. flavus 23 . A change in the activities of these degradative enzymes might have occurred in the DmtA deletion mutants, leading to increased pathogenesis in hosts.
Nevertheless, it is interesting to consider why a gene known to be related to RIP in Neurospora would be quite so conserved and functional in A. flavus, which apparently lacks active RIP and DNA methylation. Because a negligible level of DNA methylation in the genomic DNA of A. flavus was observed, it appears that DmtA might not function as a DNA methyltransferase in A. flavus. However, the inactivation of this DMTA did result in changes in A. flavus development and AF biosynthesis. Studies have shown that DMTs are associated with the initiation of chromatin remodelling and gene regulation 24,25 . The effect of Dnmt1 on gene expression may depend on histone deacetylase activity via its N-terminal non-catalytic domain binding to histone deacetylases 26 . Meanwhile, a positive correlation between transcriptional activation of AF cluster genes and histone H4 acetylation has been found in A. flavus 27 . It also has been shown that inactivation of the sterigmatocystin gene cluster in A. nidulans requires epigenetic control by H3K9 methylation and heterochromatin protein-binding to establish a repressive chromatin structure 28 . In Aspergillus, the known global regulator of many secondary metabolites, LaeA, which also functions as a methyltransferase, plays a role in chromatin remodelling at the site of secondary metabolite gene clusters 29,30 . Because DNA methylation in A. flavus is negligible, DmtA might function in chromatin remodelling in a way  In conclusion, our results provide evidence that DmtA, a putative C-5 cytosine methyltransferase, is essential for conidiation and AF metabolism in A. flavus. Moreover, inactivation of dmtA induced a change in seed infection, which produced statistically more conidia in crop seeds than the wild-type strain. Furthermore, no detectable DNA methylation in the genomic DNA of the A. flavus wild-type strain and ΔdmtA mutants was found. Thus, we propose that DmtA might not function as a DNA methyltransferase in A. flavus. The ΔdmtA mutant lines may be useful tools for exploring pre-or post-harvest control of AF in A. flauvs. Our results also provide valuable information that could advance our understanding of the epigenetic modification of AF biosynthesis in A. flavus and give a clue for furthermore study in the future.

Materials and Methods
Strain and culture conditions. Aspergillus flavus PTS Δku70ΔpyrG, a uracil auxotrophic, purchased from the Fungal Genetics Stock Center, School of Biological Sciences, University of Missouri, Kansas City, USA, was used for gene disruption. A. flavus was cultured on YES (2% yeast extract, 150 g/L sucrose, 1 g/L MgSO 4· 7H 2 O and 20 g/L agar) or YGTUU agar (5 g/L % yeast extract, 20 g/L glucose, 1 mL of trace element solution per litre of medium, 1 mg/mL uracil, 1 mg/mL uridine and 15 g/L agar) at 37 °C. DNA methyltransferase domain architecture and phylogenetic tree generation. DmtA, a C-5 cytosine methyltransferase in A. flavus was identified in the NCBI using the Basic Local Alignment Search Tool (http://blast.ncbi.nlm.nih.gov/Blast.cgi). Its coding sequence was cloned into the pET-28a prokaryotic expression vector (+ ) and sequenced. The sequence obtained was compared with the coding sequence reported in NCBI. Then, the conservative domain of the C-5 cytosine methyltransferase was identified using InterProScan 31  Construction of ΔdmtA mutants. All the primers used in this study are listed in Table 1. To generate ΔdmtA mutants, a 1226 bp fragment upstream from DMTA was amplified with dmtA/P1 and dmtA/P3 primers. Next, a 1049 bp fragment downstream from DMTA was amplified with dmtA/P4 and dmtA/P6 primers. To generate the fragment containing the upstream fragment, the pyrG selectable marker and downstream fragment were added sequentially, using the PCR fusion approach and the PCR program described by Szewczyk 33 . To generate the dmtA mutants, the fusion PCR product was transformed into protoplasts of the wild-type CA14 (Δku70ΔpyrG).

Preparation of protoplasts and fungal transformation.
In the initial transformation, the culture of spores harvested from YGTUU plates was inoculated into 150 ml of YGTUU liquid medium [34][35][36] , and the resultant culture was shaken at 180 revolutions/min for about 12 h at 37 °C. The cultures were collected on sterile filter paper, transferred to a 250 mL flask, and resuspended in 20 mL of filter-sterilized enzyme solution (10 mM NaH 2 PO 4 , pH5.8, 20 mM CaCl 2 , with 0.2 ml β -glucuronidase at 8500 U/mL, 200 mg lysing enzyme from Trichoderma harzianum (Sigma, MO, USA), 50 mg Driselase (Sigma) and 1.2 M NaCl). Digestion was performed at 70 revolutions/min at 30 °C for 3.5-4 h. Protoplasts, harvested by filtration through miracloth, were collected using a micro centrifuge. The protoplasts obtained were washed twice with STC solution (1.2 M sorbitol, 50 mM CaCl 2 , 50 mM Tris-HCl, pH7.5, autoclaved and stored at 4 °C).
Fungal transformation was performed as previously described with minor modifications 33,34 . Polyethylene glycol (PEG) buffer (50% PEG4000, 0.6 mM KCl, 10 mM NaH 2 PO 4 , pH5.8, 50 mM CaCl 2 and 10 mM Tris-HCl, pH7.5) was substituted. The transformation mixture was mixed well with 20 mL of regeneration medium (Czapek's Agar: 50.0 g/L of Czapek Solution Agar (BD Bioscience, NJ, USA), 1.0 M sucrose and 0.5% agar), plated onto selective plates, and 10 ml of regeneration medium (1.5% agar) was poured in the first layer. The plates were incubated at 37 °C for 2-3 d. Uracil/uridine autotrophy transformants were screened by PCR with primers dmtA/ UA and P801/R and primers dmtA/OF and dmtA/OR (Table 1.  Physiology experiments. Conidial production, sclerotial formation and relative colony diameter measurements were recorded for the wild-type and ΔdmtA strains. In order to eliminate the effect of illumination on A. flavus development, the physiology experiments were performed under dark and light condition. The diameter was measured from point inoculation of a 1 μL of a 106 spores/mL suspension of A. flavus conidia onto YES media. To analyse conidia production, YES media was overlaid with 5 mL of a 106 spore/mL suspension of A. flavus conidia in molten agar according to the former description by Shubha 23 . Cultures were grown for 5 d at 37 °C, and three 1.5 cm diameter cores were harvested from the centre of each plate and homogenized in 3 mL of distilled water. The spore number was counted haemocytometrically. For sclerotial production analysis, we used sclerotial inducing WKM medium 37 . Cultures were grown at 37 °C for 5 d under dark and light conditions. The plates were then sprayed with 75% ethanol to kill and wash away conidia to aid in enumeration of sclerotial.

Seed infections.
The ability of the ΔdmtA mutants to infect crop seeds was assayed as described previously 23,38 . Then, peanut cotyledons were inoculated with a 105 spores/mL of A. flavus conidia for 30 min, with continuous shaking (50 revolutions/min). A blank control was performed by immersing the cotyledons in sterile water. Twenty treatments were placed in petri dishes lined with two pieces of moist filter paper to maintain high AF analysis. To analyse AF production, a 5 mL aliquot of a 10 6 spore/mL suspension of A. flavus conidia was incubated in YES medium in the dark at 28 °C for 5 d. AF was extracted from the 500 μL filtrate with an equal volume of chloroform. The chloroform layer transferred to a new 1.5 mL tube was evaporated to dryness at 70 °C. Next, TLC was used to analyse AF biosynthesis. A solvent system consisting of acetone and chloroform (1: 9, v/v) was used, and the TLC result was observed under ultra violet (UV) light at 365 nm. For quantitative analysis of AF production, HPLC (High Performance Liquid Chromatography) was also used to confirm the presence of AF in the samples. The AF extract was filtered (0.22 mm) and analysed by HPLC Construction of the DmtA-GFP vector. To localize DmtA, the 2644-bp fragment including the 1204-bp predicted promoter region and 1440-bp dmtA coding sequence was amplified using dmtA-GFP/F and dmtA-GFP/R primers ( Table 1). The resultant 2644-bp PCR product was cloned into the pKNTG vector, which has a GFP tag and pyrG selectable marker, using a pEASY-Uni Seamless Cloning and Assembly Kit (TransGen Biotech, Beijing, China) to generate the pKN-DmtA-GFP vector. The pKN-dmtA-GFP vector was transformed into the CA14 Δku70ΔpyrG strain, and the transformants were PCR-verified.

Reverse transcriptase (RT)-PCR and quantitative real-time PCR.
Both wild-type and ΔdmtA mutant mycelia were harvested at growth stages (36 h and 72 h incubated on YES Agar). RNA molecules were isolated with TRIzol reagent (Biomarker Technologies, Beijing, China) and purified with the DNA-free kit (TransGen Biotech, Beijing, China). TransScript ® All-in-One First-Strand cDNA Synthesis SuperMix was used to synthesize the first strand cDNA, and qRT-PCR was performed with the Thermo Fisher Scientific Real-time PCR System (Finland) using TransStart Top Green qPCR SuperMix (TransGen Biotech, Beijing, China). In the quantitative real-time PCR, AF aflC, aflK and aflO structural genes and aflR and aflS regulator genes were amplified by the primer pairs shown in Table 2. As an endogenous control, the β -tubulin gene was amplified with 9F/9R primers. The relative quantification of the transcripts was calculated by the 2 −ΔΔCt method 39 . All qRT-PCR assays were conducted with technical triplicates for each sample, and the experiment was repeated twice.