Development of a fumonisin-sensitive Saccharomyces cerevisiae indicator strain and utilization for activity testing of candidate detoxification genes

ABSTRACT Fumonisins, produced mainly by various Fusarium species, are sphinganine-analog mycotoxins and potent inhibitors of ceramide synthases in animals, plants, and other eukaryotes, leading to toxicity and programmed cell death. Wild-type baker’s yeast is naturally highly resistant to fumonisin B1 (FB1); on rich medium, it tolerates more than 1,000 µM. We report the construction of a fumonisin-sensitive Saccharomyces cerevisiae strain with six disrupted genes: SNQ2, PDR12, and YOR1, coding for ATP-binding cassette transporters involved in active efflux; CKA2, encoding casein kinase II, regulating ceramide synthase activity by phosphorylation; and LCB3 and VPS51, encoding proteins involved in the biosynthesis and recycling of sphingolipid precursors, respectively. The constructed FB1-sensitive strain YTKT33 was tested on complex yeast extract-peptone-dextrose (YPD) and synthetic complete medium alongside the wild-type strain and strains with only three (YRU74) or five (YRU94ML) disrupted genes. YTKT33, exhibiting full growth inhibition at a concentration below 5 µM FB1 on synthetic complete agar lacking uracil (SC-URA), was used as a host for heterologous expression of previously reported fumonisin detoxification genes. Transformants expressing either the Sphingopyxis fumD esterase, which removes removing tricarballylic acid side chains, or an Aspergillus amine oxidase, which converts the amino- to a keto-group, can grow at FB1 concentrations inhibitory for transformants containing the empty vector. YTKT33 was inhibited by other type B fumonisins (B2, B3, B4) at concentrations very similar to FB1, allowing rapid evaluation of total fumonisin toxicity in crude fungal extracts. IMPORTANCE Fumonisins can cause diseases in animals and humans consuming Fusarium-contaminated food or feed. The search for microbes capable of fumonisin degradation, or for enzymes that can detoxify fumonisins, currently relies primarily on chemical detection methods. Our constructed fumonisin B1-sensitive yeast strain can be used to phenotypically detect detoxification activity and should be useful in screening for novel fumonisin resistance genes and to elucidate fumonisin metabolism and resistance mechanisms in fungi and plants, and thereby, in the long term, help to mitigate the threat of fumonisins in feed and food.

barley (3)(4)(5).Fumonisin-producing Fusarium proliferatum can also infect rice (6), where fumonisin production is a virulence factor in rice spikelet rot disease (7).While fumonisin production contributes to virulence in maize seedlings (8), it is dispensable for the ability to cause maize ear rot (9).Climate change seems to increase the previously rather negligible threat of fumonisin contamination in central Europe (10).
To characterize the toxicity of fumonisins and to develop mitigation strategies, mostly chemical assay methods, requiring expensive equipment, are used since previously established bioassays for fumonisin toxicity have shortfalls.Highly purified toxin preparations are needed for mammalian cells cultured in vitro to avoid the unspecific toxic effects of other metabolites co-occurring in crude toxin extracts.The reported susceptibility of different human and animal cell lines is quite variable.The effective concentrations ranged mainly from 10 to 100 µM (17).The IC 50 value for primary rat hepatocytes was reported to be 2,000 µM (18), while human HepG2 cells are more sensitive, with an IC 50 value of 399 µM (19).Plants are typically very sensitive; e.g., Arabidopsis seed germination and root development are already strongly inhibited by 1 µM fumonisin B 1 (FB1) (20).However, such bioassays are rather time-consuming (e.g., requiring a 10-day observation period).Duckweed (Lemna minor) growth inhibition by 40 µM of fumonisin metabolites over a 5-day period was used to investigate the structure-activity relationship of different fumonisins and metabolites (21).Similarly, labor-intensive and low-throughput whole-animal test systems were described.For instance, a 4-day incubation period was needed to observe the death or disintegration of Hydra attenuata during exposure to 150 µg/mL (~208 µM) FB1 (22).Similarly, microscopic observation of non-motile or dead brine shrimp (Artemia salina) yielded IC 50 values for FB1 ranging from 1.7 µM (48 h) to about 10 µM after 24 h (23,24).
Adding detoxification enzymes to fumonisin-contaminated feed commodities (25) or expressing heterologous detoxification genes already in transgenic plants (26) are possible strategies to counteract and reduce mycotoxin contamination.However, testing enzyme activity in vitro is a major bottleneck in the development of such agents.As of now, in vitro efficacy is verified through the determination of the detoxified products [e.g., hydrolyzed FB1 (HFB1)], which requires costly and laborious liquid chromatographymass spectrometry (LC-MS).Since little is known about fumonisin metabolism in plants, useful endogenous detoxification genes may exist that could be selected by breeders or manipulated by gene editing-the search is ongoing.To the best of our knowledge, none of the previously mentioned test organisms can serve as convenient hosts for the expression and high-throughput screening of potential candidate detoxification genes.We therefore set out to engineer Saccharomyces cerevisiae as an indicator organism for monitoring the toxicity of fumonisins and to phenotypically detect and test expressed detoxification enzymes.

Strain development with a crude extract containing fumonisins
Wild-type Saccharomyces cerevisiae is highly suited as a host for heterologous gene expression and for monitoring the toxicity of various substances on agar media or in liquid culture on microtiter plates.Yet, S. cerevisiae is highly resistant to many compounds, including FB1.To save costly pure FB1 during our initial characterization and strain improvement, we used a crude extract containing fumonisins, which was prepared from maize cultures inoculated with various Fusarium strains (see Materials and Methods), with an FB1 content of 3,180 mg/L as determined by LC-MS/MS.Based on this concentration value, the "wild-type" laboratory strain YPH500 (see Table 1 for relevant genotype and reference) tolerated more than 1,000 µM on rich medium (yeast extractpeptone-dextrose, YPD) (Fig. 1A and C) and more than 300 µM on synthetic complete (SC) medium (Fig. 1B and D).As it has been reported that the ATP-binding cassette (ABC) transporter YOR1 (Yeast Oligomycin Resistance) is involved in fumonisin resistance (27), we generated and tested strains that combined yor1 with mutations of other ABC transporters that were implicated in mediating resistance to charged compounds (28).We generated all double mutants and the triple mutant with the genes SNQ2 (Sensitivity to 4-NitroQuinoline-N-oxide) and PDR12 (Pleiotropic Drug Resistance), and continued strain improvement with the triple mutant YRU74 (see Table 1 for strain genotypes).This strain had reduced FB1 resistance both on YPD and SC medium compared to wild-type on solid agar plates (Fig. 1A and B).We also tested growth in liquid media in microtiter plates, and here (Fig. 1C and D) resistance was lower.
Since the resistance level of YRU74 was still unacceptably high, requiring concentra tions of FB1 exceeding naturally occurring levels, we next inactivated the CKA2 gene, encoding one of the catalytic subunits (alpha′) of casein kinase 2. It has been reported that such a mutant had 3-4× reduced ceramide synthase activity and reduced resistance to FB1 (29).We have furthermore knocked out LCB3, encoding one of the two long-chain base-1-phosphate phosphatases (paralog YSR3), which are needed for the utilization of exogenous long-chain base phosphates present in rich medium.The cka2 lcb3 double mutant YRU94 initially segregated small and large colonies upon revival from −80°C stocks, so a stable respiration-competent clone was selected on YPG, which also formed stable red colonies on YPD, indicating an unimpaired respiration needed for accumula tion of the red pigment in adenine-requiring ade2 mutants.Transformation markers (LEU2 and URA3) were removed from this strain, designated as YRU94*, using plasmid pOS4a (ADE2 GAL1-Cre), resulting in YRU94ML-A (ura − and leu − but still ADE + ).After the loss of the pOS4a plasmid, the strain YRU94ML was obtained, which clearly exhibited an over fivefold increase in sensitivity toward FB1 as compared to the yor1 snq2 pdr12 parental strain (Fig. 1C).
To further reduce FB1 resistance, we also disrupted VPS51 (Vacuolar Protein Sorting), a gene encoding a component of the Golgi-associated retrograde protein complex.Endosome-to-Golgi retrograde vesicular transport allows recycling of complex sphingoli pids (30), which would otherwise be degraded using the default pathway in the vacuole.The reduced sensitivity of the strain with the additional vps51 mutation to FB1 is evident from Fig. 1C (strain grown in liquid YPD) and in the spotting experiment with purified FB1 (Fig. 2), where strain YTKT33 (yor1 snq2 pdr12 cka2 lcb3 vps51) is more sensitive than its precursor YRU94ML with wild-type VPS51 both in YPD liquid culture (Fig. 2C) and on YPD plates (Fig. 2A).

Testing the constructed strains with purified FB1
The constructed strains were subsequently exposed to media spiked with purified, commercially available FB1 (>98% purity, Fermentek).Compared to Fig. 1, in which only crude fumonisin extracts were used, it is evident in Fig. 2 that higher levels of the pure toxin are needed to achieve a similar degree of inhibition.One obvious explanation is that the crude extract contains other intermediates of FB1 synthesis, such as fumonisins B 2 (FB2), B 3 (FB3), and B 4 (FB4), and possibly additionally other forms, such as non-toxic forms (e.g., N-acetylated fumonisin A), that might be hydrolyzed and reactivated by the yeast.Yet, no fumonisin A metabolites were detected by LC-MS/MS in the crude extract, and later results (see below) indicate that the response to FB2, FB3, and FB4 was similar to FB1 (see Fig. 4).
The different behavior of the mutants in the two media is also of interest.While on SC medium, YRU94ML behaved similar to YTKT33 (Fig. 2B and D), on the rich medium, the latter strain was far more susceptible, indicating that components (sphingolipid metabolites not present in SC medium but presumably in the yeast extract in YPD) could reduce fumonisin toxicity in a VPS51-dependent manner.
After reaching (OD 600 >0.3), they were diluted again for spotting on (A) YPD and (B) SC plates containing crude FB1, as indicated.(C and D) The inoculum was pipetted into a 96-well microtiter plate with different concentrations of crude FB1 and put into an incubator, where the absorbance at OD 600 was measured after 24 h.At least three replicates were used for each strain.On the x-axis, the graph shows the final FB1 concentration that the strains were exposed to, while the y-axis shows the inhibition of growth in % compared to each strain growing without exposure to FB1.The standard deviation was calculated for the replicates and displayed in the error bars.

Dose-response of strain YTKT33 to B-type fumonisins in liquid synthetic medium
To determine the inhibition characteristics of different B-type fumonisins [which differ in the hydroxylation pattern; for review, see reference (31)], we obtained high-purity FB1, FB2, FB3, and FB4 and tested their potency to inhibit the growth of yeast strain YTKT33 in liquid SC medium in microtiter plates.The inhibition of strain YTKT33 upon exposure to fumonisins B 2 , B 3 , and B 4 was similar to the inhibition by FB1 (see Fig. 3).
As shown in Fig. 3, the IC 50 values for the four tested fumonisins, FB1, FB2, FB3, and FB4, were all observed to be between 24 and 26 µM.The relative composition of the different compounds in the extract can be variable depending on the Fusarium strain and harvesting time, but due to the similar dose-response, the toxicity reflects the sum of all B-type fumonisins.The inhibitory concentrations for YTKT33 obtained in liquid SC culture (in microtiter plates with limited aeration in still cultures) were clearly higher than the inhibitory concentrations observed for the empty vector transformant in spottings carried out on solid synthetic complete media lacking uracil (SC-URA, see Fig. 4).

Heterologous expression of fumonisin detoxification genes
Strain YTKT33 is growing slower than the YPH500 precursor strain.Alterations in sphingolipids have been reported to negatively affect heat shock resistance (32).However, YTKT33 is robust and easily transformable with the standard lithium transfor mation protocol, which includes a heat shock treatment.We utilized the strain as a host to express candidate fumonisin resistance genes.The empty vector pYes2-P TEF1 was used in a previous study to express candidate ceramide synthases from a fumoni sin-producing F. verticillioides strains behind the constitutive TEF1 promoter (33).In the aforementioned study, evidence was provided that the FvFUM18 gene, located in the fumonisin biosynthetic cluster, codes for an insensitive ceramide synthase providing fumonisin self-resistance (33).The empty vector and the FUM18 expression vector were used as negative and positive controls in our study.As evident from Fig. 4, the lowest tested fully inhibitory concentration is clearly below 5 µM FB1, at approx.2.5 µM FB1, which is about 10-fold lower than the IC 50 in liquid SC medium in microtiter plates.

FIG 2
Sensitivity testing with pure FB1.For strain genotypes, refer to Fig. 1 or Table 1.Growth inhibition of yeast by a 98% pure FB1 stock on solid and liquid media.The wild-type yeast laboratory strain YPH500 and three derived multi-deletion strains YRU74 (triple ABC transporter mutant, yor1 pdr12 snq2), YRU94ML (yor1 pdr12 snq2 plus cka2 and lcb3), and YTKT33 (yor1 pdr12 snq2 cka2 lcb3 plus vps51) were grown in complex (YPD) and synthetic (SC) media overnight and rediluted to an OD 600 of 0.1 in the morning.After reaching OD 600 >0.3, they were diluted again for spotting on (A) YPD and (B) SC plates containing crude FB1, as indicated.(C and D) Inoculum was pipetted into a 96-well microtiter plate with different concentrations of pure FB1 and put into an incubator, where the absorbance at OD 600 was measured after 24 h.At least three replicates were used for each strain.
Plasmid-based expression of FUM18 in YTKT33 results in an approximately fourfold increase in resistance, up to 10 µM FB1.
In addition, we expressed two enzymes altering the chemical structures of fumonisin B 1 , leading to its detoxification.First, we expressed the Sphingopyxis macrogoltabida MTA144 fumD gene (34), which encodes a periplasmic secreted carboxylesterase that removes the tricarballylic acid side chains of fumonisins and leads to the less toxic metabolite HFB1 (34).We have constructed two different fumD expression vectors, with and without the N-terminal secretion signal.While complete fumD expression also leads to a low increase in resistance phenotype, the strain lacking the signal peptide (fumD_noL) exhibits high-level FB1 resistance (see Fig. 4).Previously, another detoxifi cation gene from Aspergillus, encoding a flavinedependent amine oxidase, had been described (35,36).This enzyme converts the amino group of fumonisins into a keto group.We have custom-synthesized the amine oxidase gene with an identical sequence as previously reported (35) and cloned it behind the TEF1 promoter.The expression of this gene (without the 6xHis epitope tag in Fig. 4) also led to increased fumonisin B 1 resistance, although at a lower level than YTKT33 expressing fumD_noL.In summary, these results show that the engineered yeast strain is a useful tool to validate candidate genes.
The strain YTKT33 was also used in other applications, for instance, in the classical agar diffusion assay.When seeded into the top agar, application of a crude fungal extract containing FB1 applied to the plate (on a paper disc) produced a clear inhibition zone.As shown in Fig. 5, application of a detoxification enzyme (the commercial product FUMzyme) (37) next to the toxin-containing disc resulted in a diminished and deformed halo as the enzyme diffused into the inhibition zone and inactivated the toxin.

DISCUSSION
Wild-type yeast is highly resistant to fumonisins, but we were successfully able to engineer a susceptible strain, YTKT33, by deleting several genes.Fumonisins are highly charged and water-soluble molecules (two tricarballylic acid side groups and a basic amine residue).An interesting, still unsolved question is how fumonisin enters the cell, as it is too polar to pass directly through lipid membranes.Obviously, yeast utilizes ABC transporters to reduce the concentration at the target through active afflux.We combined the described yor1 mutation with a disruption of SNQ2 (implicated in resistance to a broad range of mostly cationic compounds).In addition, we inactivated the ABC transporter encoded by PDR12, which is required for resistance to various weak acids.We settled for the triple mutant, yet these alterations may also sensitize the strain for other toxins or potentially so-far undescribed fumonisin synergists that might be present in crude extracts.
Fumonisin FB1 accounted for about 75% of the total B-type fumonisins in the crude extract.The additionally present FB2, FB3, and FB4 cannot fully explain the observed higher toxicity of the crude extract.In principle, other toxins or yet unknown low-toxicity substances acting as synergists of fumonisin might explain the higher toxicity of the crude extract.This effect limits the utility of semi-quantitative determination using the indicator strain in a bioassay.Since the inhibitory activity of the crude extract could be abolished by preincubation with FUMzyme (data not shown), the increased IC 50 in the crude extract compared to pure fumonisin is likely due to substances that do not confer toxicity on their own but rather act as synergists of fumonisin.
Based on previous studies using mammalian cells, there does not appear to be any evidence for a synergism between fumonisin B 1 and frequently co-produced Fusarium toxins, e.g., between the cyclic depsipeptides enniatins or beauvericin (38,39).Poten tially, moniliformin (40) or fusaproliferin might also be involved, which should be investigated in future work.Medium composition and growth conditions have a strong influence on fumoni sin toxicity.The binding of fumonisins, acting as competitive inhibitors of ceramide synthase, could be outcompeted by the concentration of the natural substrates of the enzyme that may be abundant, especially in rich media containing yeast extract.Particularly important in rich mediums was the targeting of the recycling pathway.As of yet, little is known about the degradation of complex sphingolipids and their reintegra tion into the biosynthesis pathway.Yeast produces inositol phosphosphingolipids, which can be hydrolyzed by the ISC1 gene product (inositol phosphosphingolipid phospholi pase C) back to phytoceramide and dihydroceramide, the products of ceramide synthase.This gene might be a candidate for an additional knockout, aiming to further decrease fumonisin resistance on YPD medium.
Yeast can be screened at high density on microtiter plates, where oxygen supply is low compared to growth on an agar surface.However, the slower (fermentative) growth under oxygen limitation seems to increase FB1 resistance.The concentration to inhibit growth was about one order of magnitude higher in SC liquid medium than in SC-URA agar plates.
The previously described fumonisin detoxification genes and a gene encoding an insensitive ceramide synthase all conferred FB1 resistance when expressed in YTKT33, validating the utility of the developed strain.The observed resistance level cannot be taken as direct evidence that one enzyme detoxifies better than the other.In a study where Lemna minor (duckweed) was exposed to HFB1 and deaminated FB1 (FPy1), the latter exhibited less toxicity than HFB1 (21).However, the final resistance level is not only caused by the reduced toxicity of the product but is also dependent on the catalytic properties and the expression level of the enzyme, which we have not determined in this study.Nevertheless, it is clear from our results that cytosolic Sphingopyxis fumD lacking the bacterial secretion signal can counteract a high level of FB1 entering the cell.It has been reported that hydrolyzed FB1 can be acylated and used as a substrate by ceramide synthases present in rat liver microsomes, forming toxic compounds (41).Under our experimental settings, even if N-acylation of HFB1 took place, it was insufficient to show toxicity.Hydrolysis of FB1 was also clearly shown to be a useful detoxification strategy in piglets through reduction of intestinal inflammation (37,42).
Strain YTKT33 can be employed in the classical agar diffusion assay.Compounds that inhibit fumonisin biosynthesis at concentrations that are not inhibitory for growth have been described, e.g., pyrrocidine (43).The developed strain could be useful in screens for such substances or to study the mode of action of natural products or plant extracts that seemingly counteract fumonisin toxicity in ill-understood ways, e.g., reference (44).We have shown that by placing the commercially available detoxifica tion enzyme FUMzyme next to the toxin, the deformed halo reveals the detoxification ability of the enzyme.In a more simple setup, preincubating amounts of FB1 with a detoxication agent and observing a diminished inhibition zone would be a possible embodiment of the utilization of the developed strain.Overall, the generated strain is clearly an improvement over previously reported indicator organisms (brine shrimp, hydra, duckweed), with the advantage that it can be used for heterologous expression of foreign genes and for high-throughput screening.It should be useful to identify currently unknown detoxification genes.For instance, it is unclear which gene encodes the fungal acetyltransferase converting FB1 to non-toxic acetylated fumonisin A 1 .Furthermore, there is a lack of knowledge about whether plant enzymes are involved in the produc tion of hidden and bound fumonisins (45).The yeast indicator strain could be useful to rapidly test mutants of detoxification genes generated by site-directed mutagene sis or to select variants with superior kinetic properties from randomly mutagenized plasmids based on their ability to confer higher levels of resistance than the starting material.This can be achieved through high-throughput screening of the candidate genes using YTKT33.Potentially, novel genes could also be identified by expressing cDNA libraries and selecting resistance-conferring clones, similar to the approach leading to the identification of a deoxynivalenol-inactivating gene (46).
To inactivate VPS51 in YRU94ML, primers VPS51_KO_pUG_767_fw and VPS51_KO_pUG_767_rv, providing the sequence overhangs for homologous recombi nation, were used to amplify a loxP− Kluyveromyces lactis LEU2-loxP construct from pUG73.The correct insertion of the marker (vps51Δ::loxPKlLEU2-loxP) in the transformants was confirmed via colony PCR using oligonucleotides flanking the disruption cas sette (VPS51_KO_upstream_outside and VPS51_KO_downstream_outside) and primers contained within the KlLEU2 gene (KlLEU2fwd and KlLEU2 rev), leading to strain YTKT1.Plasmid pBS49 (GAL1-Cre URA3) (54) was used for marker removal.The resulting strain after plasmid loss was designated YTKT33.

Treatment with crude or purified fumonisins
The crude fumonisin extract used during the construction of strains was produced as follows: autoclaved maize media (10 g cracked maize, 1.9 g saccharose, 0.125 g casein hydrolysate, and 5 mL of water) in a 220 mL CultureJar G9 Glass Plant Tissue Culture Vessel (PhytoTech Labs) with Magenta B-Cap vented lids (Bioworld 309300051) were inoculated with conidia of various Fusarium verticillioides and F. proliferatum strains from our collections and incubated without shaking in the dark at 25°C for 29 days.After the addition of 25 mL of water, the cultures were homogenized, 25 mL of methanol was added, and extraction was performed by shaking for 30 minutes at 140 rpm at room temperature.Then, 1 mL aliquots were cleared by centrifugation (14,680 × g for 10 minutes at room temperature).Samples were diluted 1:400 with methanol/water (1:1) for analysis using high pressure liquid chromatography (HPLC).The remaining homogenates were frozen at −80°C until workup.
All homogenates containing ≥250 mg/L of FB1 were thawed, pooled, and centrifuged.Methanol (and partly water) was evaporated using a rotavapor at 35°C until precipitation occurred in the flask.The contents of the flask were centrifuged, and the precipitate was discarded.The fumonisin concentration of the remaining 160 mL of the red-brown, slightly viscous aqueous raw extract was determined, as described below, to be 5.3 g FB1/L.The extract was then slightly diluted and filter sterilized, and the working stock used in experiments was determined to contain 3.18 g/L FB1.
For the determination of fumonisins, an Agilent 1100 series HPLC was coupled to a Sciex 3500 triple quadrupole mass spectrometer.Separation was achieved on a Phenomenex Gemini C18 column (150× 4.6 mm, 5 µm particle size) at 30°C.Mobile phase A consisted of methanol/water/acetic acid (40/59.9/0.1, vol/vol/vol), while mobile phase B consisted of 0.1% acetic acid in methanol.The initial gradient (0% B) was maintained for 0.5 min before 73% B was reached at 5.9 min.The column was washed with 100% B from 6.0 to 9.9 min before re-equilibration with 0% B from 10.0 to 13.0 min.Then, 1 µL of samples (diluted with 50% aqueous methanol if needed) was injected into a flow of 800 µL/min.Ionization was achieved using a TurboV electrospray source in positive ion mode, applying 4,000 V of ionspray voltage at 550°C, curtain gas of 35 psi, and gas 1 and gas 2 settings of 50 psi each.Selected ion monitoring was employed with 20 ms dwell times for each transition, followed by 5 ms pause times.The following transitions were used [analyte: Q1 mass (Da) >Q3 mass (Da), declustering potential (V), collision energy (eV)]: FB1: 722.Solid FB1 was either provided by Dr. G. Jaunecker (Romer Labs, Tulln, Austria) or obtained from Fermentek (Jerusalem, Israel).For the comparison of the responses to different B-type fumonisins (Fig. 3), 1 mg of each of FB1, FB2, FB3 (>99%), and FB4 (>98%) was purchased from Fumizol Ltd. (Szeged, Hungary).

Growth assays on solid and liquid media
Both complex (YPD) and SC (synthetic complete) media (47) supplemented with FB1 were used to conduct growth inhibition tests.For plate assays, liquid overnight cultures were diluted back to an OD 600 nm of 0.1.After reaching again about 0.3, they were diluted to an OD 600 nm of 0.1, 0.01, and 0.001, and 3 µL of these suspensions was spotted on the agar plates.Photographs were taken after incubation at 30°C for 5 days.For comparing the sensitivity to FB1 of different strains in liquid media, CELLSTAROneWell flat bottom plates (Greiner Bio-One GmbH, Kremsmünster, Austria) were placed in a Biotek Synergy H1 Hybrid microplate reader.Plates were incubated at 30°C for 24 h, and OD 600 nm was measured every hour using a fully automated set-up.A 0.6 N dilution series of the toxin was prepared manually by adding 120 µL of the previous toxin solution (200 µL in well 1) to 80 µL of prepared toxin-free media (in wells 2) and repeating this step in the following wells (wells 3-11) step by step.Then, 80 µL of inoculum was added to each well (1:1 dilution), leading to 11 different concentrations of FB1 and a no-toxin control (well 12).The initial OD 600 of the strains was ~0.05 when diluted 1:1 with the FB1-YPD/SC mixture inside the well plate.Three replicates were used for each strain, and the growth after 24 h was analyzed by calculating the average of the OD 600 of these replicates along with the standard deviation.The value of a blank (medium without yeast) was subtracted.In order to determine the response of the YTKT33 strain to different B-type fumonisins again, a dilution series of 0.6 N starting at a concentration of 500 µM was prepared in a microtiter plate and SC medium with 250 µM set as the highest one and wells containing only the yeast strain serving as a control.Four replicates were used, and the mean was calculated based on the OD values measured after 24 h, along with the standard deviation.

Expression of candidate genes
Plasmid pYes2-P TEF1 was used to express different candidate genes expected to confer resistance against FB1.It contains URA3 as a selection marker, the TEF1 promoter, and a N-terminal 6xHis-tag (33).The empty vector, as well as one containing the Fusarium verticillioides self-resistance gene (33), FUM18, as a positive control, were kindly provided by Dr. Vito Valiante (Leibniz Institute for Natural Product Research and Infection Biology, Hans Knöll Institute, Jena, Germany).The reading frames of other genes of interest were PCRamplified using a polymerase with proof-reading activity (Phusion High-Fidel ity DNA Polymerase, Thermo Fischer Scientific, Vienna, Austria) and about 50-bp-long primers with 35-bp homology to the plasmid and the gene at both the 5′ and 3′ ends.The candidate genes were recombined behind the TEF1 promoter, and the resulting plasmids were recovered from yeast and transformed into Escherichia coli by electropora tion.For sequence confirmation, primers pVT-U_R4 (see Table 2) and CYC1-RV were used.
The original Sphingopyxis sp.MTA144 fumD esterase gene recoded for Pichia expression (34) was used as a template for PCR amplification.For generating the construct with the natural bacterial secretion signal, primers SmFumD_rv and SmFumD_fw_Leader were combined.In order to generate the fumD construct without a leader, primers SmFumD_rv and SmFumD_fw_noLeader_MAQ were used.The respective PCR products were co-transformed with the BamHI + Eco721 cut vectors for in vivo recombination.The Aspergillus amine oxidase gene as described (35) was ordered from BioCat (Heidelberg, Germany), flanked by a sequence containing BamHI and Eco721 for cloning into vector pYes2-P TEF1 , yielding pTAK37.Modifications were added to this plasmid later on to test for potential negative effects of the 6xHis-tag.Therefore, pTAK37 was cut with BamHI and PvuII (removing the 6xHis-tag) and blunt ends religated after Klenowfill in.This vector, pTAK38, was used to clone other detoxification genes and transformed into YTKT33 for spotting.

FIG 3 FIG 4
FIG 3 Growth inhibition of YTKT33 in microtiter plates in SC medium by B-type fumonisins FB1-FB4.The sensitive yeast strain YTKT33 was grown in SC medium and exposed to different concentrations of FB1, FB2, FB3, and FB4.Strain inoculum was pipetted into a microtiter well plate with 0.6 N dilutions of the respective fumonisins.After 24 h at 30°C, the optical density at 600 nm (OD 600 ) was measured to monitor growth.The blank (medium without yeast) was subtracted from the measured OD 600 values.Means and standard deviations were calculated for four replicates each, as depicted in the error bars of the graphs.

FIG 5
FIG5 One hundred microliters of YTKT33 (OD 600 = 0.3) was embedded in SC top agar after equilibration at 48°C and left at room temperature to solidify on SC media (bottom agar).Five microliters of crude FB1 (3,180 mg/L) was added to paper discs (left), while water or FUMzyme was added on the right.

TABLE 1
Yeast strains used or generated in this

TABLE 2
Sequences of primers used in this study a The underlined part of the sequences is complementary to the beginning or end of the resistance cassette amplified from different plasmids, while the non-underlined part is complementary to the 5′ or 3′ end of the gene to be knocked out. a