Production of Destruxins from Metarhizium spp. Fungi in Artificial Medium and in Endophytically Colonized Cowpea Plants

Destruxins (DTXs) are cyclic depsipeptides produced by many Metarhizium isolates that have long been assumed to contribute to virulence of these entomopathogenic fungi. We evaluated the virulence of 20 Metarhizium isolates against insect larvae and measured the concentration of DTXs A, B, and E produced by these same isolates in submerged (shaken) cultures. Eight of the isolates (ARSEF 324, 724, 760, 1448, 1882, 1883, 3479, and 3918) did not produce DTXs A, B, or E during the five days of submerged culture. DTXs were first detected in culture medium at 2–3 days in submerged culture. Galleria mellonella and Tenebrio molitor showed considerable variation in their susceptibility to the Metarhizium isolates. The concentration of DTXs produced in vitro did not correlate with percent or speed of insect kill. We established endophytic associations of M. robertsii and M. acridum isolates in Vigna unguiculata (cowpeas) and Cucumis sativus (cucumber) plants. DTXs were detected in cowpeas colonized by M. robertsii ARSEF 2575 12 days after fungal inoculation, but DTXs were not detected in cucumber. This is the first instance of DTXs detected in plants endophytically colonized by M. robertsii. This finding has implications for new approaches to fungus-based biological control of pest arthropods.


Introduction
Despite concerns with negative impacts of chemical insecticides on human health, the use of these chemicals remains high. Consequently, the demand for alternatives is increased. Biological control of arthropod pests using entomopathogenic fungi is one promising alternative [1,2,3]. Entomopathogenic fungi from the genus Metarhizium are some of the most frequently studied biological control agents for use against insects and ticks [2,3,4].
Metarhizium spp. produce a wide array of small molecules including destruxins (DTXs), cyclic depsipeptides which are produced as well as by some other fungi, both insect (Aschersonia) and plant pathogens (Alternaria, Trichothecium) [5]. The effects of DTXs on insects include: tetanic paralysis [6,7], inhibition of DNA and RNA synthesis in insect cell lines [8], inhibition of Malpighian tubule fluid secretion [9], blocking H + ATPase activity [10], and suppression of insect defense responses [11,12,13,14,15]. DTXs also have antifeedant and repellent properties [16,17]. The insecticidal potential of these toxins has been confirmed in numerous reports of acute toxicity [5]. Despite demonstrated insecticidal activity of DTX, Donzelli et al. [18] showed that a Metarhizium robertsii mutant with disrupted DTX synthetases was as virulent as the wild type strain when fungus conidia were topically applied to insect larvae. This supports the conclusions of a previous report that Metarhizium spp. isolates could be pathogenic for insects whether they had the ability to produce in vitro DTXs or not [19]. Although these compounds have been detected in moribund, infected hosts [20,21], DTXs reportedly have little or no impact on virulence as measured in whole-insect bioassays [18,19].
DTXs also have negative effects on insect behavior, for example inducing phagodepression and repellence [16,17]. Metarhizium robertsii (ARSEF 2575) is plant-rhizosphere competent and has endophytic capability [22,23,24,25]; accordingly, if DTXs produced inside Metarhizium-colonized plants induced antifeedant effects on arthropod pests of those plants, then the presence of DTXs in planta may afford enhanced levels of Metarhiziumassociated biological control of these pest arthropods.
We report here a survey of virulence of 20 Metarhizium isolates against insect larvae, and the concentration of DTXs A, B, and E produced by these same isolates in vitro (submerged shake cultures). We then analyzed plants endophytically colonized by a high-DTX producing M. robertsii isolate and a low-or non-DTX producing M. acridum isolate [26,27] to search for DTXs in colonized plants.

Fungal isolates
Twenty Metarhizium spp. isolates were used in the present study: 18 isolates from different regions of Brazil, one from the USA and one from Australia (Table 1)

In vitro production of DTXs and HPLC-UV analysis
For the analysis of in vitro DTXs production, fungal cultures were started with 1610 6 conidia/100 mL CZAPEK-DOX Broth (BD Difco) with bactopeptone (0.5%) and incubated in 250-mL flasks at room temperature (,22uC) on a rotary shaker at 150 rpm for 1, 2, 3, 4, or 5 days. Control isolates were M. robertsii ARSEF 2575 (a high DTX producer) and Metarhizium acridum ARSEF 324 (a low or non DTX producer) [26,27]. Production of DTX in the culture supernatants was determined by quantitative HPLC-UV analysis of the major components (DTXs A, B and E). All solvents used in the current study were HPLC grade. Cultures were separated into fungus mycelium and supernatant by centrifugation at 1000 6 g for 20 minutes. Mycelia were harvested, dried at 80uC for 48 hours, and weighed to obtain the amount of mycelial production for each isolate. Extraction of DTXs from culture supernatants was accomplished by loading 5 mL aliquots onto C18-SPE cartridges (100 mg; Agilent Bond Elut #12102001) that were previously conditioned with 10 column volumes of methanol followed by a similar volume of ultra-pure water. The loaded cartridges were rinsed with 10 mL ultra pure water and then eluted with 2 mL methanol [18].
Just prior to analysis the methanol extracts were diluted 1:1 with water and then 10-mL aliquots of extract were injected onto a reversed phase (RP) Betasil C18 column (100 mm62.1 mm, Thermo Fisher) with a guard column of the identical phase. Elution was with a gradient of acetonitrile and water using a modular HPLC system (Shimadzu Corp., Kyoto, Japan). The linear gradient conditions using the solvents A (acetonitrile) and B (water) were: 0-10 min (25% A increased to 60% A); 10-13 min (isocratic 60% A); 13-15 min (60% A decreased to 25% A) at a flow rate of 0.3 mL min 21 . Detection was by UV absorbance at 220 nm. After the run was complete, the column re-equilibration time was 5 min. DTXs A, B, and E were measured using standard curves for each compound. DTXs A, B, and E standards were purified using methods based on those of Krasnoff et al. [28] and standard solutions prepared at 1 mg/mL in methanol. Calibration standards were prepared by dilution of 20 mL of each standard stock solution into 0.940 mL of 50% methanol and then serial dilution to give standards at 20, 10, 5, 2.5, 1.25, 0.62 and 0.31 mg mL 21 . Limit of detection (LOD) was estimated to be 0.10 mg/mL based on a S/N ratio of 3 for UV detection at 220 nm.

Detection of DTXs in plants
(i) Fungal inoculation of plants. Seeds of cowpea (V. unguiculata) (organic seeds, Shangri-la Health Foods, Logan, UT, USA) and cucumber (C. sativus) (''Straight Eight'' untreated organic seeds, Snow Seed, Salinas, CA, USA) were weighed individually and only those weighing between 0.2500 g and 0.2599 g for cowpeas, and 0.0240 g and 0.0249 g for cucumber were used. Seeds were surface sterilized by immersion in 95% ethanol for 2 minutes, rinsed in sterile deionized water followed by immersion in 30% hydrogen peroxide for 1 minute. Disinfected seeds were then rinsed 3 times in sterile deionized water [29]. These axenic seeds were kept overnight at 4uC to synchronize growth. After synchronization, seeds were immersed for 1 h in conidial suspensions (1610 6 conidia mL 21 0.01% Tween 80) of ARSEF 2575 or ARSEF 324. Seeds were then individually set on sterile, moist filter paper in Petri plates and kept at 25uC for 12 days with a photoperiod of 16:8 (L:D) (white fluorescent tubes [30]). Sterile water was added as needed to keep the filter paper moist. Uninoculated seeds (no-fungus control) were immersed in sterile deionized water containing 0.01% Tween 80 [23]. After 12 days, presence or absence of M. robertsii or M. acridum in plants was confirmed by incubating surface sterilized leaves, stems and roots on artificial medium. Surface sterilization was by immersion for 2 minutes in 0.5% sodium hypochlorite, 2 minutes in 70% ethanol, rinsed in sterile deionized water 3 times and dried using sterile filter papers. The outer edges of the leaves were dissected and discarded [31]. The remaining parts were cut into pieces and cultured on PDAY medium supplemented with 0.05% chloramphenicol in a 60 mm Petri plate. Three plates from each treatment (ARSEF 2575 exposed, ARSEF 324 exposed, or not-infected plants) were incubated with 2 or 3 pieces of leaf, stem or root per plate. The plates were examined daily for 7 days. Fungi growing from plant tissues were isolated and characterized morphologically according to Tulloch [32].
(ii) Extraction and LC-MS/MS analysis. After 12 days of growth, 10 plants of each treatment (ARSEF 2575 exposed; ARSEF 324 exposed; and not exposed) were frozen in liquid nitrogen and ground with mortar and pestle to a powder. To verify the accuracy of the DTX detection method, pure DTX standards (A, B, and E, 16.5 mg each) were mixed (before liquid nitrogen freezing and homogenization) into ten additional not-fungusexposed 12-day-old plants. Methanol (5 mL) was added to each plant powder, followed by 15 mL ultra pure water. Plant suspensions were clarified by filtration (Whatman Nu 1). Extractions of filtrates (20 mL) were carried out with C18 SPE cartridges (as described before).
The concentrations of DTXs A, B, and E, were measured by liquid chromatography-mass spectrometry (LC-MS). The LC-MS system consisted of a Betasil C18 RP HPLC column (10062.1 mm, Thermo Fisher), coupled to a Surveyor MS Pump Plus, a Surveyor Auto Sampler Plus and a PDA UV-vis absorbance detector in-line with an LCQ Advantage Max mass spectrometer and electrospray (esi) ionization source (Thermo Electron Corp, San Jose, CA, USA). Sample injection size was 5-mL. The gradient-elution steps were the same as those used for LC-UV analysis (see section 2.2). Pseudomolecular ions [M+H] + of DTX A, B, and E were observed at m/z 578, 594 and 594 respectively with the following retention times: 7.05 min (DTX A), 9.12 min (DTX B) and 5.03 min (DTX E). DTXs A, B, and E concentrations were measured using a standard curve for each compound prepared by serial dilutions as previously described (section 2.2); but with the lowest standard at 0.15 mg mL 21 . The limit of detection (LOD) with the LC-MS system was estimated to be 0.010 mg mL 21 based on a S/N ratio of 3.

Effect of DTXs on plant dry weight
After DTXs extraction, plant powders were held for 48 hours at 80uC and their dry weights' determined. Dry weights of fungustreated plant groups and the non-treated groups were analyzed by analysis of variance (ANOVA) followed by the Tukey test with a significance level of 5% (P#0.05) [33].
Commercially produced G. mellonella larvae (last instar; 239 mg average weight) and T. molitor larvae (at least ninth instar; 95 mg average weight) (Fluker Farms, Port Allen, LA, USA) were treated either with 1610 7 conidia mL 21 or 1610 5 conidia mL 21 . Two groups of 8 last instar G. mellonella larvae and 2 groups of 10 T. molitor larvae were placed in 60615 mm polystyrene Petri dishes lined with a 5.5 cm P4 filter paper (Fisherbrand, Porosity: Medium -Fine, Flow rate: Slow) moistened with 0.5 mL sterile distilled water. Each plate containing larvae was sprayed with 0.5 mL fungal suspension. Control plates were sprayed with 0.01% Tween 80 solution. Plates were incubated at 28uC and $80 RH. Insect mortality was assessed daily for 10 days. The bioassays were repeated 3 times.
Mean larval mortalities at 3 days with G. mellonella and 5 days with T. molitor were compared using the non-parametric Kruskal-Wallis test for statistical differences. Comparison between the mean mortalities was performed using Student-Newman-Keuls (SNK) test. Data analyses were conducted using BioEstat software, version 4.0. P-values less than 0.05 were considered to be significant [34].

Results
Conidial viability (percent germination) of all suspensions used in in vitro DTX production, plant-seed inoculations, and insect bioassays was at least 98%.
Generally, the earliest detection of DTXs in in vitro cultures was at day 3; the exception being ARSEF 759, which produced DTX E at day 2 (0.55 mg DTX E/g d.w. mycelium) (Figure 2

Detection of Metarhizium-produced DTXs in plants
Endophytic growth in 12-day-old cowpea (V. unguiculata) and cucumber (C. sativus) by M. robertsii and M. acridum was confirmed ( Figure 4). In each case, the isolated fungus colonies presented the key morphological features consistent with Metarhizium isolates.
Detectable levels of DTXs A, B, and E were identified in combined roots, stems and leaves of cowpea plants cultured for 12 days after exposure of their seeds to M. robertsii conidia ( Figure 5B). The concentrations of each compound followed by its respective standard error were: 5.7360.29 mg DTX E/g d.w. cowpeas; 1.5660.29 mg DTX A/g d.w. cowpeas; and 0.8260.11 mg DTX B/g d.w. cowpeas. With cucumber, however, despite confirmation of M. robertsii endophytic colonization, no DTXs were detected in extracts of these plants. Also, no DTXs were detected in plants (cowpeas or cucumber) colonized by M. acridum, nor in control plants (not-infected plants). DTXs were detected in all positive controls (not-infected cowpea and cucumber plant tissues spiked with DTX standards) ( Figure 5C).

Effect of DTXs on dry weights of plants
No differences in total dry weights were noted between endophytic Metarhizium-colonized and not-colonized plants (P$ 0.05). Similarly, for both plant species (V. unguiculata and C.

Insect virulence assays
The virulence of 20 Metarhizium spp. isolates (Table 1) was surveyed using two different insect hosts: G. mellonella (waxworm) and T. molitor (mealworm). Natural mortality of untreated (control) G. mellonella larvae was always higher than with T. molitor larvae; e.g., waxworm control mortality reached 16.67% at day 5 while mealworm mortality was 1.6% at the same time. There were variations in the virulence of the isolates, and in the susceptibility of the different host species (Table 2). T. molitor larvae were less susceptible than G. mellonella larvae. For this reason, T. molitor bioassay data at day 5 after treatment were used for comparisons, and with G. mellonella day 3 data were used ( Table 2). According to this evaluation system, several isolates (e.g., ARSEFs 724, 1448, 1885, 2575, 3641, and 3643) had similar levels of virulence for both species of insect.

Discussion
The present study investigated 20 Metarhizium spp. isolates as to their virulence against two insect species and their levels of DTXs production in artificial liquid medium. A wide variation in DTXs production in vitro was observed among the 13 isolates in the present study ( Figure 1). Two Metarhizium isolates were used to analyze fungus-colonized cowpea and cucumber plants for DTXs production. This is the first report of the presence of DTXs in cowpea plants colonized by the entomopathogenic fungus M. robertsii.
Not all fungal isolates tested here have been classified according to the Bischoff et al. [35] protocol, but based on the species names attributed to these isolates in the ARSEF catalog (Table 1), we note that the production of DTXs is not strictly correlated with Metarhizium species. For example: ARSEF 552 (M. pingshaense) and ARSEF 2575 (M. robertsii) are producers of DTXs in vitro, while isolates ARSEF 1448 (M. pingshaense) and ARSEF 724 (M. robertsii) are not. In the present study, M. acridum isolate ARSEF 324 did not produce detectable levels of DTXs after culture in vitro for 5 days; which is similar to the findings of Wang et al. [27] with this isolate in vitro. In contrast, Kershaw et al. [19] and Moon et al. [26] reported low levels of DTXs A and E production in vitro by isolate ARSEF 324 with longer incubation periods and higher temperatures.
Comparisons of DTXs production in vitro with virulence to insects of the 20 Metarhizium spp. isolates did not indicate a close association of the two traits. The most virulent isolates for T. molitor were ARSEF 3643, ARSEF 1448 (both isolates caused 100% mortality 5 days after treatment with 1610 7 conidia mL 21 ). Interestingly, ARSEF 3643 was the best DTX producer in vitro, while there were no detectable levels of DTXs produced by ARSEF 1448 in liquid culture. The same occurred with G. mellonella, i.e., of the 5 most virulent isolates (e.g., ARSEFs 724, 760, 1885, 2575, 3641, and 3643) two did not produce DTXs in vitro (ARSEF 724 and ARSEF 760), one was a very weak producer in vitro (ARSEF 1885), and three were good DTXs producers in vitro (ARSEF 2575, ARSEF 3641, and especially ARSEF 3643). Another trait that might relate to virulence is the first time (date) that DTXs were detectable in culture supernatants.
For most DTXs producing isolates, these compounds were detected in liquid culture on day 3 of fungal growth; however, ARSEF 759 had detectable levels of DTX E at day 2 in culture ( Figure 2). ARSEF 759 did not demonstrate higher potency or a shorter lethal time in comparison to fungal isolates that only showed detectable levels at days 3 or 4 in culture (e.g., ARSEF 2575 and ARSEF 1885). These observations suggest that the presence or the absence of DTXs A, B, and E in in vitro culture supernatants had little or no correlation with percent mortality or speed of insect kill.
Arthropod pathogens such as B. bassiana [31,36,37]; Lecanicillium lecanii ( = Verticillium lecanii) [37,38]; Isaria farinosa ( = Paecilomyces farinosus) [39]; and M. robertsii [23] have been reported as endophytes. According to O'Brien [40], M. acridum, an acridid specialist, is not rhizosphere-competent; and Pava-Ripoll et al. [29] reported that germination of this fungal species in plant root exudates was significantly lower than with M. robertsii ( = M. anisopliae). On the other hand, as reported in the current study, M. acridum colonized endophytically either cowpea or cucumber when surface sterilized seeds were inoculated with conidia in the laboratory. It currently is not known if spraying leaves of plants with this fungus will permit endophytic establishment in leaves, stems and roots.
The entomopathogenic fungus B. bassiana is an endophyte in naturally colonized plants [36], and also has been isolated after artificial inoculation in many important agricultural crops such as bananas, bean, coffee, corn, cotton, tomato and wheat [37]. Bing and Lewis [41] reported that tunneling in corn plants by Ostrina nubilalis larvae, the European corn borer, was reduced when plants were endophytically colonized by B. bassiana. Although the overwhelming majority of publications on the use of arthropodpathogenic fungi against insects discuss the reduction of insect damage through insect death due to direct fungal infection by conidia, Vega et al. [36] suggested that this suppression of insect damage in response to B. bassiana plant colonization [41] may be the result of feeding deterrence or antibiosis. Such deterrence by some fungi is related to their production of metabolites. More recently, Gurulingappa et al. [37] studied the effect of endophytes (B. bassiana, L. lecanii and Aspergillus parasitucus) on the reproduction and growth of Aphis gossypii and Chortoicetes terminifera. They reported that endophytes significantly reduced aphid reproduction and locust growth rate, but no direct mortality was observed. Amiri et al. [16] reported residual and antifeedant activities of DTXs A, B, and E when leaf discs of Chinese cabbage were immersed in these toxins and submitted to larvae of crucifer pests P. xylostella and P. cochleariae; as a result, leaf area ingested by these larvae was greater for untreated leaves than DTXstreated leaves in doses higher than 3 mg/g [16]. According to the study [16] with crucifer pest larvae, the amount of DTXs detected in cowpeas in the current study should be slightly toxic. However, DTXs amounts in plants older than those that we studied probably would vary, and DTXs susceptibility of other insect species also are likely to vary. The mechanisms involved in feeding suppression of insects by contact and/or ingestion of DTXs remains unclear.
Metarhizium spp. have been indicated as mediators of interactions among plants, insects and soil: e.g., Behie et al. [22] showed that plants can receive significant amounts of nitrogen from Metarhizium-infected soil insects. Sasan and Bidochka [23] reported that endophytic establishment of M. robertsii in roots induced growth of plant roots and root hairs. The ability of some Metarhizium isolates to produce DTXs within plants, as reported here, suggests another potentially important benefit to plants from endophytic association with these fungi.
DTXs production by AP fungi in plants depends not only on the fungal isolate but also on the plant species. Our results showed that even when colonized with M. robertsii ARSEF 2575 (an isolate . The cowpea seeds, both fungus-inoculated and control (not colonized) were incubated on moist filter paper under optimal light (16L:8D) and temperature (25uC) conditions for 12 days at which time the germlings had developed roots, stems, cotyledons and two true leaves. DTXs were extracted from entire plants using methanol 100% and SPE-C18 cartridges. doi:10.1371/journal.pone.0104946.g005 that produces DTXs in vitro and also in cowpeas), cucumber extracts did not have detectable levels of DTXs. A plant pathogen Alternaria brassicae, the causative agent of Alternaria blackspot, is known to produce DTX B that is used to facilitate plant colonization. DTX B is a selective toxin, in that only plant cultivars susceptible to the toxin are damaged by the fungus [5,42]. Resistant plants have enzymes that detoxify DTX B [42]. The current study did not investigate whether cucumber plants hydrolyzed DTX or if this host plant did not support DTX production.
Further studies on the effects of per os DTXs exposure in vertebrate organisms are needed to support the use of entomopathogenic fungi inoculated in crop seeds to control insect pests. In an instance where there is some hesitancy by regulating agencies about allowing DTXs in a food product, a non-DTXs producing isolate of Metarhizium could be selected for use in biological control on that crop to avoid such DTXs production, or plant cultivars that detoxify DTXs could be selected.
In planta production of secondary metabolites by endophytic Metarhizium may be an exploitable feature of this fungus in its use against agricultural arthropod pests. The production of DTXs in M. robertsii-colonized plants reported here clearly indicates that further investigation is warranted on the antifeedant or repellent properties of fungal metabolites expressed in planta.

Supporting Information
Table S1 Destruxin production (mg/L) by 12 Metarhizium spp. isolates in vitro. Destruxin production is represented by mean values 6 standard error after 5 days in submerged shaken cultures. (DOC)