Generating Phenotypic Diversity in a Fungal Biocatalyst to Investigate Alcohol Stress Tolerance Encountered during Microbial Cellulosic Biofuel Production

Consolidated bioprocessing (CBP) of lignocellulosic biomass offers an alternative route to renewable energy. The crop pathogen Fusarium oxysporum is a promising fungal biocatalyst because of its broad host range and innate ability to co-saccharify and ferment lignocellulose to bioethanol. A major challenge for cellulolytic CBP-enabling microbes is alcohol inhibition. This research tested the hypothesis that Agrobacterium tumefaciens - mediated transformation (ATMT) could be exploited as a tool to generate phenotypic diversity in F. oxysporum to investigate alcohol stress tolerance encountered during CBP. A random mutagenesis library of gene disruption transformants (n=1,563) was constructed and screened for alcohol tolerance in order to isolate alcohol sensitive or tolerant phenotypes. Following three rounds of screening, exposure of select transformants to 6% ethanol and 0.75% n-butanol resulted respectively in increased (≥11.74%) and decreased (≤43.01%) growth compared to the wild –type (WT). Principal component analysis (PCA) quantified the level of phenotypic diversity across the population of genetically transformed individuals and isolated candidate strains for analysis. Characterisation of one strain, Tr. 259, ascertained a reduced growth phenotype under alcohol stress relative to WT and indicated the disruption of a coding region homologous to a putative sugar transporter (FOXG_09625). Quantitative PCR (RT-PCR) showed FOXG_09625 was differentially expressed in Tr. 259 compared to WT during alcohol-induced stress (P<0.05). Phylogenetic analysis of putative sugar transporters suggests diverse functional roles in F. oxysporum and other filamentous fungi compared to yeast for which sugar transporters form part of a relatively conserved family. This study has confirmed the potential of ATMT coupled with a phenotypic screening program to select for genetic variation induced in response to alcohol stress. This research represents a first step in the investigation of alcohol tolerance in F. oxysporum and has resulted in the identification of several novel strains, which will be of benefit to future biofuel research.


Introduction
Lignocellulosic biomass is an abundant feedstock and attractive source of sugars for biofuel production. Large-scale utilisation is however challenged by the general lack of low-cost technologies that can overcome cellulose recalcitrance [1]. One potential route to eco-friendly sustainable energy production is the consolidated bioprocessing (CBP) of biomass into biofuels [2].
At present no organism can ferment cellulosic materials to ethanol at rates and titres necessary to achieve economic feasibility [5]. CBP requires robust microbes able to: (i) degrade lignocellulosic materials in the absence of supplementary exogenous enzymes, (ii) utilise at high efficiency both hexose and pentose sugars including their conversion into high-titer ethanol and (iii) tolerate toxic compounds notably the primary alcohol product formed during fermentation [16]. Whereas the endogenous ability of F. oxysporum to tolerate inhibitory compounds encountered during CBP including lignocellulosic hydrosylates (carboxylic acids, phenolic compounds, furan derivatives) [17] and the fermentation by-product acetic acid [18] has been reported, a knowledge deficit exists in regards to F. oxysporum's capability to tolerate ethanol.
Ethanol stress affects cell growth and viability in addition to productivity [19,20]. While other compounds compatible as biofuels such as n-butanol offer advantages over ethanol (e.g. high energy content and miscibility with gasoline), production is limited by the sensitivity of native producers such as Clostridium sp. to the alcohol [21]. This has led to the investigation of alternative butanol production hosts [22][23][24], prompting the investigation of n-butanol (hereafter referred to as butanol) tolerance in addition to ethanol in this study. For the cost-effective and commercially viable production of biofuels, product yields must exceed native microbial tolerance levels [25] therefore engineering of stress tolerant strains is critical to achieve high survival and production rates [25,26].
Several strategies have been investigated to improve alcohol tolerance in yeast and bacteria, including gene over-expression or random knockout libraries, genome shuffling and transcriptional or translational engineering [27]. For filamentous fungi ATMT has established itself as a valuable and powerful tool for genetic studies, greatly facilitating the identification and analysis of genes involved in complex biological processes such as host pathogenicity [28] and, as we hypothesise here, in regards to alcohol tolerance. As the gram-negative soil bacterium A. tumefaciens facilitates the stable transfer and integration of the tumour inducing (Ti) plasmid segment (known as T-DNA) into the targeted host cell [29], ATMT holds significant advantages over more traditional methods of gene disruption. For example, A. tumefaciens' broad host range, its potential to transform a range of starting materials and its propensity for low copy number insertion events [30]. ATMT also provides an ability to 'tag' the disrupted gene(s) thereby facilitating their identification via genome walking. In the case of F. oxysporum this is further assisted by the recent availability of the organism's genome sequence [31] (http:// www.broadinstitute.org).
In this study, we hypothesised that ATMT could be exploited as a tool to generate significant degrees of phenotypic diversity in F. oxysporum strain 11C in response to alcohol stress. To this end, a modified ATMT protocol was established to generate a library of random gene disruption transformants, which underwent a three tier physiological evaluation. As a result, the initial population of F. oxysporum transformants was reduced from 1,563 to 29 individuals, which were subjected to ethanol tolerance screening and cross-resistance to butanol. The level of recorded phenotypic variation confirmed the occurrence of ATMT-derived genotypes with several F. oxysporum transformants identified for future analysis. To the best of our knowledge this is the first report detailing the use of random insertional mutagenesis to investigate alcohol tolerance in a fungal CBP agent.

Origin and maintenance of fungi
Fusarium oxysporum strain 11C originated from Irish soils as previously described [13]. For ATMT, F. oxysporum was subcultured onto potato dextrose agar (PDA; Difco, UK) plates, incubated at 25°C for 5 days prior to producing fungal conidial inoculum in Mung bean broth [32].

Plasmid and ATMT transformation
The binary vector used for fungal gene disruption was pSK1019, which is equipped with the hph antibiotic resistance marker gene under the control of an Aspergillus nidulans TrpC promoter and was donated by Professor Seogchan Kang (The Pennsylvania State University, USA). Agrobacterium tumefaciens strain AGL-1 was transformed with the binary vector pSK1019 [33] and ATMT of F. oxysporum strain 11C was optimised based on a previously published protocol [34] (see Methods S1). Three independent ATMTs of F. oxysporum strain 11C were conducted.

Molecular analysis of transformants
For PCR analysis, fungal genomic DNA was extracted from mycelia using a modified method [35] (see Methods S1). DNA was quantified using a Qubit TM Quant-iT assay (Invitrogen, U.S) and stored at -80°C. PCR was used to detect the presence/ absence of the hph transgene in DNA extracts from putative transformants (see Methods S1). For Southern blot analysis, genomic DNA was extracted [36] and quantified using a NanoDrop® ND-1000 Spectrophotometer (NanoDrop Technologies, USA). Southern blot analysis was used to determine plasmid copy number in transformants (see Methods S1).

Primary alcohol tolerance screen
A total of 1,563 putatively transformed hygromycin-resistant colonies were isolated into microplates (96-well) containing 200 μl minimal medium (supplemented with hygryomcinB, 60 μg ml -1 , Sigma, UK) per well [37] which were then sealed and incubated at 25°C for 3 days and subsequently used as inoculum for primary screening of alcohol tolerance. For the primary screen (first of three), 96-well plates were prepared containing 140μl minimal medium [37] per well supplemented with six treatments either hygromycinB (60 μg ml -1 ), no ethanol or 0.5, 6.0 or 10% vv -1 ethyl alcohol (Sigma, UK) plus or minus hygromycinB (60 μg ml -1 ; Sigma, UK). To target different response levels, low (0.5% vv -1 ethanol), medium (6% vv -1 ethanol) and high (10% vv -1 ) ethanol concentrations were then selected. Wells were inoculated with 10 μl of fungal material (one well per treatment per transformant/wild type fungus; plate layout is shown in Figure S1A). Plates were sealed with a lid and parafilm and incubated at 25°C with fungal growth (OD 600nm ) measured after both 48 and 96 hours using a spectrophotometer (Safire 2 , Tecan, Austria). Putative transformants with a ≥ 2-fold higher OD 600nm in 10% (vv -1 ) ethanol relative to wild type fungus 11C were purified prior to secondary screening. This primary screening of the transformant collection was conducted once followed by the selection of putative transformants for purification and graduation to the secondary screening (see Methods S1).

Secondary alcohol tolerance screen
Conidial suspensions of the F. oxysporum 11C wild type strain and transformants of interest (20 μl from the from monoconidial 20% (vv-1 ) -80°C glycerol stock) were inoculated into 220 μl minimal medium in microtiter (96-well) plate wells and incubated at 25°C for 24 hours. This served as inoculum for the secondary screen. For secondary screening; 100 μl of fungal culture was transferred into microtiter plate wells containing 100 μl minimal medium [37] supplemented with either 0, 6, 7, 8, 9 or 10% vv -1 ethanol (single well per treatment per transformant/wild type; see Figure S1B for plate layout). Plates were sealed with a lid and parafilm and incubated at 25°C for 96 hours after which fungal growth (OD 600nm ) was recorded. Data was normalised relative to the control (wild-type strain 11C). All datasets (0-10 vv -1 ethanol) were used for selecting transformants of interest. From the secondary screen, F. oxysporum strain 11C putative transformants demonstrating either ethanol sensitive or ethanol tolerant phenotypes relative to the wild-type strain 11C control, at concentrations ranging between 6-10% (vv -1 ) were selected for tertiary screening. This experiment was conducted once.

Tertiary alcohol tolerance screen
In the tertiary screen, tolerance to both ethanol and butanol was assessed, with the goal of examining the potential for cross-correlation between ethanol and butanol tolerance. Conidial inoculum for the tertiary assessment was generated by individually inoculating each transformant into Mung bean broth [32] (100 ml conical flask) with three fungal plugs and incubating at 25°C and 150 rpm for 5 days. Inoculum was harvested by centrifugation and washed twice with sterile distilled water and resuspended in minimal medium [37] to a concentration of 10 6 ml -1 conidia. Microtiter (96-well) plates were prepared with wells containing 100 μl minimal medium supplemented with either no alcohol, ethanol [2,4,6,8 or 10% (vv -1 )] or butanol (0.25, 0.5, 0.75, 1 and 1.25% (vv -1 )] with a single treatment tested per 96-well plate (see Figure S1C for plate layout). Each plate was inoculated with 100 μl fungal conidia (10 6 ml -1 ) (eight wells per treatment per wild type/ transformant strain). Plates were sealed with a lid and parafilm and incubated at 25°C for 7 days. Fungal growth was measured using a spectrophotometer OD 600nm (Spectra Max 340 PC 96-well Plate Reader, Molecular Devices, USA), and percentage increase or decrease in growth relative to wild-type control was determined. For data analysis see Methods S1. This experimental screening was repeated three times.

Phenotypic analysis of F. oxysporum 11C and Tr. 259
A series of phenotypic assays of F. oxysporum 11C and Tr. 259 were conducted to investigate the effect of increasing alcohol concentration on growth, temporal analysis of alcohol tolerance, the effect of alcohol tolerance stress on spore germination and biomass production (see Methods S1). All experiments were conducted either twice or three times. For data analysis see Methods S1.

Genome walking and sequence analysis
Genome walking was conducted on Tr. 259 to characterise the degree of genetic disruption underpinning the transformant's observed phenotype (see Methods S1 & Table  S1).

RT-PCR of putative hexose transporter (FOXG_09625) under alcohol stress
An experiment was designed to investigate the temporal accumulation of a putative hexose transporter under alcohol stress ( Figure S2) (see Methods S1& Table S1). Each experiment was conducted either three times (no alcohol, ethanol) or twice (butanol) with three replicates per strain per treatment per time point. Real-time quantification of the target and housekeeping gene respectively were performed in separate reactions. The threshold cycle (C T ) values obtained from RT-PCR were used to calculate the accumulation of the target gene (relative mRNA accumulation) relative to β-tubulin transcript by the 2 -ΔΔC T method, where ΔΔC T = (C T , target gene -C T , β-tubulin) [38]. Results were based on the average obtained for two replica RT-PCR reactions per sample. For data analysis see Methods S1.

Phylogenetic analysis of putative sugar transporters in F. oxysporum
A phylogram of putative F. oxysporum strain 4287 sugar transporters was constructed using amino acid sequences from transcripts showing ≥30% homology to FOXG_09625 (See Methods S1).

Data analysis
The primary/secondary screen dataset was analysed using the Boxplot function in R (R v2.15.2 R Development Core Team, 2012) and a Box and Whisker plot was used to depict the distribution of transformant response across various treatments. The primary/secondary screen dataset was nonnormally distributed as determined within Minitab (Minitab release © 16, 2011 Minitab Inc.). The significance between treatments at each time point (48 and 96 hours respectively) or ethanol treatments was analysed within the Statistical Package for the Social Sciences (SPSS 18, SPSS Inc.) using Kruskal-Wallis H-test for non-parametrical data. The significance between treatments and transformants amongst the 29 individuals selected from secondary screening was determined using one-way ANOVA as the data could be transformed to fit a normal distribution using Johnson transformation within Minitab. For tertiary screen analysis see Methods S1.

ATMT library construction for primary alcohol screening
A transformation platform was established for F. oxysporum strain 11C to generate a library of gene disruption transformants (n=1,563) ( Figure S3). The library was screened using a high throughput alcohol screen based on microtiter (96well) plates to discriminate against low to high ethanol (0.5-10% (vv -1 )) response levels measured across 48 and 96 hours ( Figure 1A & 1B & Figure S3). Transformant selection for second -tier screening, as previously described [Methods], was based on recordings at 96 hours for which greater phenotypic variation was observed compared to 48 hours. An established threshold [Methods] led to the isolation of 182 putative transformants, which when purified (4 monoconidial cultures produced per putative transformant) yielded 443 viable monoconidial-derived cultures for PCR analysis and secondary alcohol screening.

PCR analysis and secondary alcohol screening
Over 90% (402/443) of the monoconidial lines generated from primary transformants were PCR-positive for the hph transgene ( Figure S4A). On re-screening the ethanol tolerance of this population, a significant difference between treatments was observed between the derived monoconidial lines (P<0.001) ( Figure 1C). A total of 29 PCR-verified transformants indicating hypo-or hyper-ethanol growth phenotypes were selected for tertiary screening.

Genomic analysis of Tr. 259
Genome walking was conducted to determine the potential genomic region associated with Tr. 259's hypo-tolerant phenotype ( Figure S7). This resulted in the cloning of a 1031nt sequence external to the T-DNA right border (RB) (sequence analysis confirmed the amplicon was flanked by the universal FOXG_09625 showed homology (percent identity ≥ 30%) with 30 different proteins all containing MFS and sugar transporter domains respectively, encoded within the genome of F. oxysporum f.sp.lycospersici (strain 4287) (FGCD). Five of these proteins were annotated as hexose transporters (FOXG_02491, FOXG_09722, FOXG_05876, FOXG_13263, FOXG_014666) including FOXG_09625, and the remaining as hypothetical proteins. Comparison of these proteins showed considerable sequence diversity indicating potentially diverse roles of these proteins in F. oxysporum ( Figure S9A). In total, the genome of F. oxysporum f.sp. lycospersici (strain 4287) encodes 13 proteins annotated as hexose transporters ( Figure  S9B).
RT-PCR analysis was used to analyse FOXG_09625 transcript levels in strain 11C compared to Tr. 259, following short-term (0.5-12 hours) cultivation of germinated spores (precultured for 12 hours) in the absence or presence of alcohol ( Figure 5). In the absence of alcohol, basal expression was low in both 11C and Tr. 259 (P>0.05). ( Figure 5A). Reduced transcript levels were recorded for both strains under ethanol stress compared to normal conditions from 0.5 to 12 hours with a significant difference between strains at 2 hours (P<0.01) ( Figure 5B). In the presence of butanol, a significant difference between Tr. 259 and 11C was recorded for all time points tested (P<0.001) with transcript levels highest at 0.5 hours ( Figure 5C).

Discussion
This study has shown that ATMT is an effective tool to generate phenotypic diversity in response to alcohol stress within F. oxysporum strain 11C. Although the generation of fungal insertion libraries is well documented, the integration of a three-tier alcohol screen into an ATMT platform is novel, and led to the generation and isolation of phenotypically diverse transformants from the library, indicating genotype-dependent variation and a functional ATMT system.
Large-scale screening of filamentous fungi can be challenging as manipulating fungal material (e.g. mycelium) can interfere with pipette-aided transfer between plates [43]. Fungal growth rates or hyphal growth patterns can vary greatly among isolates with some strains growing poorly in low volume space-limited microplates. Regardless, microplate-based assays facilitate the screening of large populations: a significant advantage over the labour-intensive regime of individual shake flasks or tubes [43]. Screening of F. oxysporum ATMT-derived transformants for alcohol tolerance has not yet been described. In other ascomycetous fungi notably the rice blast fungus Magnaporthe grisea, large-scale ATMT mutagenesis coupled with phenotypic plate-based screens has been used in pathogenicity related studies [44,45].
The three-tier screening system was designed to specifically discriminate either alcohol sensitive or tolerant phenotypes.
Integration of the primary screen into the ATMT protocol provided a course but effective method to instigate a logistically feasible sorting of the library, thereby avoiding the laborious and time-consuming task of purifying thousands of putative transformants prior to screening. Non-purified, putative transformants underwent primary screening in contrast to the secondary screen, which used single-spored hph-PCR positive transformants. It cannot be ruled out that false negatives were lost to the analysis during these two stages. These primary rounds of screening were based on single reads to accelerate the phenotyping process and isolate transformants of interest for further analysis using an intensive confirmatory tertiary screen. This latter screen highlighted transformant Tr. 259 which underwent phenotypic and genomic analysis to confirm its hypo-alcohol tolerant phenotype and the ATMT-based disruption of a coding region showing homology to a putative sugar transporter (FOXG_09625) in F. oxysporum f.sp lycospersici (strain 4287). Southern hybridisation suggested a single insertion site within Tr. 259 with genome walking indicating partial T-DNA integration but it cannot be ruled out Eukaryotic transporters are best characterised in the yeast 20-member family of major hexose transporter genes (HXT1-4, HXT6-7, HXT8-17) which includes two sensor-coding genes (SNF3, RGT2) and the GAL2 gene [46]. In contrast to yeasts (S. cerevisiae, Pichia stipitis, Schizosaccharomyces pombe, Kluveromyces lactis), few sugar transporters have been identified and characterised in filamentous fungi [47,48]. The genome of F. oxysporum f.sp lycospersici (strain 4287) has at least 13 putative hexose transporters including FOXG_09625, similar to Aspergillus nidulans reported to have 17 [47]. RT-PCR analysis indicated FOXG_09625 was differentially expressed in Tr. 259 compared to WT during alcohol-induced stress and it would be interesting to investigate by Western blotting whether FOXG_09625 is also differentially expressed at the protein level. In yeast, the closest protein homologue of FOXG_09625 is HXT13, which is induced in the presence of non-fermentable carbon sources (e.g. alcohols), low sugar levels and repressed under high sugar concentrations [49]. It was hypothesised that FOXG_09625 would be poorly expressed under normal conditions and highly expressed in the presence of ethanol or butanol. Whilst this hypothesis was not observed under ethanol stress, results recorded for butanol were as expected. For yeast, the association of hexose transport and glycolysis associated genes with higher expression levels under 4% (vv -1 ) ethanol stress led to the proposal that the cell enters a pseudo-starvation state during ethanol stress, whereby sugar-based nutrients are no longer accessible to the cell despite being present in the culture medium [50,51]. In this study, the highest FOXG_09625 transcript levels were recorded under butanol stress (0.5 hours) suggesting that cells may have immediately entered a pseudostarvation state in response to the increased toxicity of butanol, possibly inhibiting sugar up-take. In yeast, increased expression of hexose transporters and glycolysis genes improve both rapid production and consumption of ethanol [52].
A previous study [53] identified two sugar transporters overexpressed in strain 11C during CBP. In the analysis of additional transformants we would expect to identify an overlap with the SSH library for genes holding dual function i.e. relating to both production and tolerance. However, ethanol stress studies to date have identified genes belonging to diverse functional categories hence the identification of novel genes could also be anticipated. Interestingly, FOXG_09625 showed 29% protein homology to the SSH -identified hexose transporter and no homology to the high affinity glucose transporter indicating potentially different roles for all three genes in either tolerance or production. Yet, a recent study [54] has indicated FOXG_09625 is associated (directly/indirectly) with ethanol production since mutants overexpressing a hexose transporter resulted in increased ethanol yield coupled with compensatory changes in the expression of other transporters notably the up-regulation of FOXG_09625. Investigations, including overexpression/knockout studies of FOXG_09625, while not possible due to a limitation in  [32] and resuspended in minimal medium [37] at a concentration of 10 6 ml -1 . A volume 100 µl conidia (10 6 ml -1 ) was added to microtiter (96-well; Sarstedt, Germany) plates with either no alcohol (A), ethanol (Ethyl absolute, Sigma, UK) at a concentration of 4% (vv -1 ) (B) or butanol (n-Butanol Butyl alcohol, Sigma, UK) at a concentration of 0.75% (vv -1 ) (C). Fungi were maintained at 25°C for 168 hours and growth (OD 600nm ) was measured every 24 hours. Fungal growth (OD 600nm ) was measured using a spectrophotometer (Spectra Max 340 PC 96-well plate reader, Molecular Devices, USA). Bars indicate SEM. Growth of Tr. 259 significantly different from 11C is highlighted with an asterisk (level of significance: *<0.05, ** < 0.01). doi: 10.1371/journal.pone.0077501.g004 resources for this study will be required to fully elucidate the complexity of this gene interaction with alcohol and determine the degree to which over-expression/silencing of FOXG_09625 could improve consolidated bioprocessing yield.
Filamentous fungal transporters appear to hold diverse functional roles notably in plant-fungi symbiosis [55][56][57] pathogenicity [58,59], and ethanol production and tolerance [48,50,[60][61][62]. In regards to tolerance, transcriptional studies have identified alcohol responsive genes associated with sugar transport [50,63,64]. This is not surprising since the rate of sugar transport is limited by accumulation of alcohol in the fermentation broth, which can become a significant stress during fermentation resulting in a 'survival versus production' conundrum for the producing microbe [65]. While major advances have been made investigating alcohol tolerance in model systems (yeast, E. coli), a knowledge deficit exists for filamentous fungi [66][67][68]. A better understanding of the molecular basis of alcohol stress and tolerance in such fungi is important if enhanced tolerance and production is to be achieved. Previous work [54] coupled with this study point towards sugar transporters as key targets for improving fungalenabled CBP.