# MANAGEMENT OF *FUSARIUM* SPECIES AND THEIR MYCOTOXINS IN CEREAL FOOD AND FEED

EDITED BY: Thomas Miedaner, Daniela Gwiazdowska and Agnieszka Was´kiewicz PUBLISHED IN: Frontiers in Microbiology

#### *Frontiers Copyright Statement*

*© Copyright 2007-2017 Frontiers Media SA. All rights reserved. All content included on this site, such as text, graphics, logos, button icons, images, video/audio clips, downloads, data compilations and software, is the property of or is licensed to Frontiers Media SA ("Frontiers") or its licensees and/or subcontractors. The copyright in the text of individual articles is the property of their respective authors, subject to a license granted to Frontiers.*

*The compilation of articles constituting this e-book, wherever published, as well as the compilation of all other content on this site, is the exclusive property of Frontiers. For the conditions for downloading and copying of e-books from Frontiers' website, please see the Terms for Website Use. If purchasing Frontiers e-books from other websites or sources, the conditions of the website concerned apply.*

*Images and graphics not forming part of user-contributed materials may not be downloaded or copied without permission.*

*Individual articles may be downloaded and reproduced in accordance with the principles of the CC-BY licence subject to any copyright or other notices. They may not be re-sold as an e-book.*

*As author or other contributor you grant a CC-BY licence to others to reproduce your articles, including any graphics and third-party materials supplied by you, in accordance with the Conditions for Website Use and subject to any copyright notices which you include in connection with your articles and materials.*

> *All copyright, and all rights therein, are protected by national and international copyright laws.*

*The above represents a summary only. For the full conditions see the Conditions for Authors and the Conditions for Website Use.*

ISSN 1664-8714 ISBN 978-2-88945-294-1 DOI 10.3389/978-2-88945-294-1

#### About Frontiers

Frontiers is more than just an open-access publisher of scholarly articles: it is a pioneering approach to the world of academia, radically improving the way scholarly research is managed. The grand vision of Frontiers is a world where all people have an equal opportunity to seek, share and generate knowledge. Frontiers provides immediate and permanent online open access to all its publications, but this alone is not enough to realize our grand goals.

#### Frontiers Journal Series

The Frontiers Journal Series is a multi-tier and interdisciplinary set of open-access, online journals, promising a paradigm shift from the current review, selection and dissemination processes in academic publishing. All Frontiers journals are driven by researchers for researchers; therefore, they constitute a service to the scholarly community. At the same time, the Frontiers Journal Series operates on a revolutionary invention, the tiered publishing system, initially addressing specific communities of scholars, and gradually climbing up to broader public understanding, thus serving the interests of the lay society, too.

#### Dedication to Quality

Each Frontiers article is a landmark of the highest quality, thanks to genuinely collaborative interactions between authors and review editors, who include some of the world's best academicians. Research must be certified by peers before entering a stream of knowledge that may eventually reach the public - and shape society; therefore, Frontiers only applies the most rigorous and unbiased reviews.

Frontiers revolutionizes research publishing by freely delivering the most outstanding research, evaluated with no bias from both the academic and social point of view. By applying the most advanced information technologies, Frontiers is catapulting scholarly publishing into a new generation.

#### What are Frontiers Research Topics?

Frontiers Research Topics are very popular trademarks of the Frontiers Journals Series: they are collections of at least ten articles, all centered on a particular subject. With their unique mix of varied contributions from Original Research to Review Articles, Frontiers Research Topics unify the most influential researchers, the latest key findings and historical advances in a hot research area! Find out more on how to host your own Frontiers Research Topic or contribute to one as an author by contacting the Frontiers Editorial Office: researchtopics@frontiersin.org

# **MANAGEMENT OF** *FUSARIUM* **SPECIES AND THEIR MYCOTOXINS IN CEREAL FOOD AND FEED**

Topic Editors:

**Thomas Miedaner,** University of Hohenheim, Germany **Daniela Gwiazdowska,** Poznan´ University of Economics and Business, Poland **Agnieszka Was**'**kiewicz,** Poznan´ University of Life Sciences, Poznan´, Poland

Plot of bread wheat (*Triticum aestivum*) infected with *Fusarium culmorum* in the field.

Image: Thomas Miedaner.

Health and safety of food and feed are the most important criteria for their quality. The quality of feed is in turn important for animal health, the environment and for the safety of food from animal origin. Fungi belonging to the *Fusarium* genus are widespread in crops causing plant diseases and producing toxic metabolites. *Fusarium* species can colonize plants during their growth on the field and cause serious damage in terms of yield and quality of harvested grains. One of the most important fungal diseases of wheat and other cereals in the world is Fusarium head blight caused by the fungal pathogens *Fusarium graminearum* and *Fusarium culmorum* and others. In addition, these fungi produce mycotoxins, contaminating food and feed. The most important *Fusarium* mycotoxins include trichothecenes, zearalenone and fumonisins, primarily because of their prevalence, but also because of the toxic effect to humans and animals. However, these fungi produce also other mycotoxins such as moniliformin, beauvericin, enniantin or fusarins. Food and feed can be contaminated with mycotoxins at various stages in the production chain resulting in serious problems with health, safety and economic losses. It is estimated that 25% of the crop in the world each year are contaminated with these metabolites, the problem affects both industrialized countries and developing countries.

The aim of this Research Topic of Frontiers in Microbiology is to publish state of the art research about occurrence and genomics of *Fusarium* species and their mycotoxins in the whole food and feed chain starting from the crops as well as implications for health and economic aspects. This research topic highlights the current knowledge on the plant diseases caused by *Fusarium* fungi as well as all aspects of *Fusarium* mycotoxin contamination of crops, food and feed, taking into account decontamination methods.

**Citation:** Miedaner, T., Gwiazdowska, D., Was'kiewicz, A., eds. (2017). Management of *Fusarium* Species and Their Mycotoxins in Cereal Food and Feed. Lausanne: Frontiers Media. doi: 10.3389/978-2-88945-294-1

# Table of Contents

#### *06 Editorial: Management of* **Fusarium** *Species and Their Mycotoxins in Cereal Food and Feed*

Thomas Miedaner, Daniela Gwiazdowska and Agnieszka Was'kiewicz

#### **Chapter 1: Occurrence and genomics of Fusarium species and their mycotoxins**

## *09 A European Database of* **Fusarium graminearum** *and* **F. culmorum** *Trichothecene Genotypes*

Matias Pasquali, Marco Beyer, Antonio Logrieco, Kris Audenaert, Virgilio Balmas, Ryan Basler, Anne-Laure Boutigny, Jana Chrpová, Elz˙bieta Czembor, Tatiana Gagkaeva, María T. González-Jaén, Ingerd S. Hofgaard, Nagehan D. Köycü, Lucien Hoffmann, Jelena Levic', Patricia Marin, Thomas Miedaner, Quirico Migheli, Antonio Moretti, Marina E. H. Müller, Françoise Munaut, Päivi Parikka, Marine Pallez-Barthel, Jonathan Piec, Jonathan Scauflaire, Barbara Scherm, Slavica Stankovic', Ulf Thrane, Silvio Uhlig, Adriaan Vanheule, Tapani Yli-Mattila and Susanne Vogelgsang


## **Chapter 2: Preventive measures for controlling Fusarium in the field**


*119 Inoculum Potential of* **Fusarium** *spp. Relates to Tillage and Straw Management in Norwegian Fields of Spring Oats*

Ingerd S. Hofgaard, Till Seehusen, Heidi U. Aamot, Hugh Riley, Jafar Razzaghian, Vinh H. Le, Anne-Grete R. Hjelkrem, Ruth Dill-Macky and Guro Brodal

*134* **Fusarium** *Head Blight Resistance QTL in the Spring Wheat Cross Kenyon/86ISMN 2137*

Curt A. McCartney, Anita L. Brûlé-Babel, George Fedak, Richard A. Martin, Brent D. McCallum, Jeannie Gilbert, Colin W. Hiebert and Curtis J. Pozniak

*145 Identifying Rare FHB-Resistant Segregants in Intransigent Backcross and F2 Winter Wheat Populations*

Anthony J. Clark, Daniela Sarti-Dvorjak, Gina Brown-Guedira, Yanhong Dong, Byung-Kee Baik and David A. Van Sanford


Magdalena Fra˛c, Agata Gryta, Karolina Oszust and Natalia Kotowicz

#### **Chapter 3: Post-harvest biodegradation and detoxification of mycotoxins**


Naveen K. Kalagatur, Venkataramana Mudili, Chandranayaka Siddaiah, Vijai K. Gupta, Gopalan Natarajan, Murali H. Sreepathi, Batra H. Vardhan and Venkata L. R. Putcha

*245 Bacterial Epimerization as a Route for Deoxynivalenol Detoxification: the Influence of Growth and Environmental Conditions*

Jian Wei He, Yousef I. Hassan, Norma Perilla, Xiu-Zhen Li, Greg J. Boland and Ting Zhou

# Editorial: Management of Fusarium Species and their Mycotoxins in Cereal Food and Feed

Thomas Miedaner <sup>1</sup> \*, Daniela Gwiazdowska<sup>2</sup> and Agnieszka Waskiewicz ´ 3

*<sup>1</sup> State Plant Breeding Institute, University of Hohenheim (720), Stuttgart, Germany, <sup>2</sup> Department of Natural Science and Quality Assurance, Faculty of Commodity Science, Poznan University of Economics, Pozna ´ n, Poland, ´ <sup>3</sup> Department of Chemistry, Faculty of Wood Technology, Poznan University of Life Sciences, Pozna ´ n, Poland ´*

Keywords: Fusarium, mycotoxins, cereals, food contamination, feeding

**Editorial on the Research Topic**

#### **Management of Fusarium Species and their Mycotoxins in Cereal Food and Feed**

Fusarium species are pathogenic fungi that appear in all wheat and maize growing areas worldwide. They are only lowly specialized and one Fusarium species can infect several hosts and host organs. The worldwide most recognized species, Fusarium graminearum, for example, readily infects wheat, triticale, barley, oat, rye as well as maize and the infected organs are similarly diverse: seedling, roots, stems/stalks, ears (Becher et al., 2013). The most prevalent diseases are Fusarium head blight (FHB) in wheat and barley and Fusarium ear rot (FER) in maize caused by cereal infecting Fusarium pathogens.

#### Edited by:

*Alex Andrianopoulos, University of Melbourne, Australia*

#### Reviewed by:

*Kemal Kazan, Commonwealth Scientific and Industrial Research Organization (CSIRO), Australia*

> \*Correspondence: *Thomas Miedaner miedaner@uni-hohenheim.de*

#### Specialty section:

*This article was submitted to Fungi and Their Interactions, a section of the journal Frontiers in Microbiology*

Received: *30 November 2016* Accepted: *31 July 2017* Published: *17 August 2017*

#### Citation:

*Miedaner T, Gwiazdowska D and Waskiewicz A (2017) Editorial: ´ Management of Fusarium Species and their Mycotoxins in Cereal Food and Feed. Front. Microbiol. 8:1543. doi: 10.3389/fmicb.2017.01543*

Infection of cereals by Fusarium spp. reduces grain yield in the first line. A yield loss of 1 Mg ha−<sup>1</sup> was predicted to occur at 19% FHB incidence (Salgado et al., 2015). Additionally, grain size and baking quality are affected, and the harvested grain is contaminated with mycotoxins, especially A and B trichothecenes, fumonisins, and the estrogenic zearalenone that are harmful for humans and livestock (Gallo et al., 2015; Stoycho, 2015). Among these mycotoxins deoxynivalenol (DON) and its acetylated forms 3-ADON and 15-ADON, are considered to be the most important, but a whole spectrum of other mycotoxins differing in chemical structure and toxicity may appear (Stoycho, 2015), because several Fusarium species can naturally co-occur in the same host tissue. Due to their heat stability, Fusarium mycotoxins occur in the whole cereal supply chain from the farmer to the customer providing problems to all stakeholders. In the European Union and many other countries, maximum levels for DON, T-2 and HT-2 toxins, zearalenone, and fumonisins in human food and guidance levels in animal feeding are implemented (Ferrigo et al., 2016). Consequently, grain contaminated with DON above 1,250 µg kg−<sup>1</sup> , the threshold level allowed in food established by the EU, may be rejected or priced down by grain buyers. It is, therefore, of utmost importance to prevent the formation of mycotoxins or at least to reduce their concentration. Cumulative direct production and price impacts between 1998 and 2000 due to FHB on wheat and barley from the northern Great Plains and the Central States of the USA were estimated at \$871 million over the period, with additional secondary economic losses of \$1.8 billion (Nganje et al., 2004). The main drivers of the cereal supply chain concerning Fusarium diseases are the seed

and chemical industry, the farmers, and the storage and processing companies. In the field, weather, crop species, cultivars, Fusarium species/isolates and management practices are the main components affecting disease severity and mycotoxin contamination. To control Fusarium diseases and mycotoxins all these stakeholders should work together (Wegulo et al., 2015) as illustrated in this Research Topic.

At the beginning of the cereal supply chain stands the seed industry with the aim to improve host resistance. Growing resistant cultivars minimizes Fusarium incidence, severity and mycotoxin concentrations in an environmentally friendly and cost-effective way. This affords to detect resistance sources, to bring them into actual breeding material and to develop commercially successful cultivars. One strategy to achieve this aim is to search in native sources for FHB resistance and to identify the responsible genes (QTL) in a segregating population (McCartney et al.). This is a time-consuming procedure because FHB resistance is based on an array of genes with small effects that work together additively and are affected by environment. Consequently, the use of molecular markers should accelerate the introgression. An additional challenge in FHB resistance breeding is combining resistance with superior agronomic and quality characteristics (Clark et al.). Their results indicate that rare segregants within native wheat populations can be found that combine the most important traits when the populations are large enough. Although a lot of effort has been put into breeding resistant varieties in the last two decades, we still have no idea on the function of these resistance genes. Lahlali et al. showed that the lignification pathway and callose deposition can play a role in confining the fungus to the inoculation site. This might lead to candidates of biochemical markers for selecting FHB resistance in the lab. Another route for reaching the same goal is discussed by the review of Atanasova-Penichon et al. on the contribution of grain antioxidant secondary metabolites to the mechanisms of plant resistance to Fusarium and mycotoxin accumulation.

Another strategy for controlling FHB is the use of fungicides that have to be sprayed during flowering of wheat to be effective. Finding new fungicides by the chemical industry is a timeconsuming and extremely costly procedure. Frac et al. developed a microplate assay to efficiently test substances for their fungicidal activity toward several Fusarium isolates in a more efficient way. Also biological antagonists or molecules could play a role in this framework as reviewed by Alberts et al.

The main stage of Fusarium development and potential mycotoxin hazard are the commercial fields (Ferrigo et al., 2016). Here, we face the pathologist's square with climate and weather, Fusarium as pathogen, the crop as host and farmers' management practices as the major drivers. FHB infections need high humidity during flowering while the temperature is normally not a limiting factor. The results of Scala et al. clearly show that the environmental conditions at field level and soil management practices may drive FHB outbreak and mycotoxin contamination even in a growing area suitable for cropping durum wheat like in Southern Italy. Similarly, in Norwegian oats rainy periods are necessary for Fusarium dispersal. Additionally, soil and straw management by the farmer are important factors (Wegulo et al., 2015). Fusarium infections are largely reduced when soil tillage is practiced before sowing wheat or as less straw as possible from the previous crop remains on the soil surface (Hofgaard et al.).

Hofgaard et al. has emphasized the importance of Fusarium species given the fact that in Europe we have mostly several species infecting the same crop, like in Norwegian oats, where F. avenaceum, F. graminearum, F. culmorum, and F. langsethiae have been isolated. A similar topic was followed in a fundamental study of Pasquali et al. where he and his 31 coauthors collected data on the occurrence of trichothecene genotypes of F. graminearum and F. culmorum in Europe. In F. graminearum, the predominant genotype was 15-acetyldeoxynivalenol (15-ADON) (82.9%), and in F. culmorum 3-ADON (59.9%). In the latter species, however, the nivalenol (NIV) genotype accounted for the remaining 40.1%. These data are available from a freely accessible and updated database (http://www.catalogueeu.luxmcc.lu). A totally different occurrence of chemotypes was found in Brazilian wheat where isolates from the Fusarium graminearum species complex were classified as NIV (55%), 15-ADON (43%), and 3- ADON (2%) chemotypes by PCR (Tralamazza et al.) illustrating that these chemotype analyses have to be performed in each region. To understand the evolution and role of trichothecene chemotypes a comparative gene expression study of nine out of 16 genes was accomplished (Amarasinghe and Fernando). One important outcome was that relative expression of TRI genes was higher in 3-ADON producing strains compared to 15-ADON and NIV strains. Analyzing the transcriptomes of F. graminearum cells infecting living, actively defending wheat heads vs. dead wheat tissue showed that in the living plant much higher toxin production is promoted (Boedi et al.).

The interaction of the four factors in the field complex results in the actual DON level of the harvest. Rainy weather, aggressive Fusarium isolates, susceptible crops and suboptimal management by the farmer can lead to a DON contamination surpassing legislative limits. Foroud et al. provided a deeper insight into the role of DON in disrupting protein synthesis of the host plant. This might provide in future possibilities to develop trichothecene remediation strategies. Such strategies could be important steps for reducing mycotoxin concentrations in grain storage centers or food processing industries. They include the use of bacterial biodegradation pathways that are capable of transforming DON to a non-toxic stereoisomer (He et al.). Perczak et al. and Kalagatur et al. found significant effects of selected essential oils extracted from diverse plants on degradation of zearalenone in vitro and on downregulating the expression of genes involved in zearalenone production in maize, respectively. In an excellent review Vanhoutte et al. summarize several approaches to reduce mycotoxins by chemical removal, physical binding, or microbial degradation. At the end, they provide the features of an ideal biodegrading and detoxifying agent that has still to be detected.

An even broader focus on this topic is given by Alberts et al. who review pre-harvest and post-harvest strategies for controlling fumonisin-producing Fusaria and their toxins in maize, including essential oils and microbial biodegradation but also resistance breeding and genetic engineering. This review stretches over the whole cereal supply chain and brings us back to the main aim of Fusarium research: to enhance food and feed safety by avoiding Fusarium infection and/or decreasing their effects. We have today accumulated much knowledge and a large number of strategies that include breeding resistant varieties, developing fungicides or antagonists, best management practices of the farmers, and possibilities of mycotoxin binding or degradation in storage and processing facilities (McMullen et al., 2012). The main challenge that remains is to combine the most effective measures in an integrative approach for combating Fusarium species wherever they occur.

#### REFERENCES


## AUTHOR CONTRIBUTIONS

All authors listed have made a substantial, direct and intellectual contribution to the work, and approved it for publication.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2017 Miedaner, Gwiazdowska and Wa´skiewicz. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# A European Database of *Fusarium graminearum* and *F. culmorum* Trichothecene Genotypes

Matias Pasquali <sup>1</sup> \*, Marco Beyer <sup>1</sup> , Antonio Logrieco<sup>2</sup> , Kris Audenaert <sup>3</sup> , Virgilio Balmas <sup>4</sup> , Ryan Basler <sup>5</sup> , Anne-Laure Boutigny <sup>6</sup> , Jana Chrpová<sup>7</sup> , Elzbieta Czembor ˙ 8 , Tatiana Gagkaeva<sup>9</sup> , María T. González-Jaén<sup>10</sup>, Ingerd S. Hofgaard<sup>11</sup>, Nagehan D. Köycü<sup>12</sup> , Lucien Hoffmann<sup>1</sup> , Jelena Levic´ <sup>13</sup>, Patricia Marin<sup>10</sup>, Thomas Miedaner <sup>14</sup>, Quirico Migheli <sup>4</sup> , Antonio Moretti <sup>2</sup> , Marina E. H. Müller <sup>15</sup>, Françoise Munaut <sup>16</sup>, Päivi Parikka<sup>17</sup> , Marine Pallez-Barthel <sup>1</sup> , Jonathan Piec<sup>1</sup> , Jonathan Scauflaire<sup>16</sup>, Barbara Scherm<sup>4</sup> , Slavica Stankovic´ <sup>13</sup>, Ulf Thrane<sup>18</sup>, Silvio Uhlig<sup>19</sup>, Adriaan Vanheule<sup>3</sup> , Tapani Yli-Mattila<sup>20</sup> and Susanne Vogelgsang<sup>21</sup> \*

#### *Edited by:*

Alex Andrianopoulos, University of Melbourne, Australia

#### *Reviewed by:*

Vijai Kumar Gupta, NUI Galway, Ireland Stefan G. R. Wirsel, Martin-Luther-Universität Halle-Wittenberg, Germany

#### *\*Correspondence:*

Matias Pasquali matias.pasquali@list.lu; Susanne Vogelgsang susanne.vogelgsang@ agroscope.admin.ch

#### *Specialty section:*

This article was submitted to Fungi and Their Interactions, a section of the journal Frontiers in Microbiology

*Received:* 18 December 2015 *Accepted:* 14 March 2016 *Published:* 06 April 2016

#### *Citation:*

Pasquali M, Beyer M, Logrieco A, Audenaert K, Balmas V, Basler R, Boutigny A-L, Chrpová J, Czembor E, Gagkaeva T, González-Jaén MT, Hofgaard IS, Köycü ND, Hoffmann L, Levic J, Marin P, Miedaner T, ´ Migheli Q, Moretti A, Müller MEH, Munaut F, Parikka P, Pallez-Barthel M, Piec J, Scauflaire J, Scherm B, Stankovic S, Thrane U, Uhlig S, ´ Vanheule A, Yli-Mattila T and Vogelgsang S (2016) A European Database of Fusarium graminearum and F. culmorum Trichothecene Genotypes Front. Microbiol. 7:406. doi: 10.3389/fmicb.2016.00406 <sup>1</sup> Department of Environmental Research and Innovation, Luxembourg Institute of Science and Technology, Belvaux, Luxembourg, <sup>2</sup> Institute of Sciences of Food Production, National Research Council, Bari, Italy, <sup>3</sup> Department of Applied Biosciences, Faculty of Bioscience Engineering, Ghent University, Ghent, Belgium, <sup>4</sup> Department of Agriculture, University of Sassari, Sassari, Italy, <sup>5</sup> BIOGER UMR, INRA, Thiverval-Grignon, France, <sup>6</sup> ANSES, Plant Health Laboratory, Angers, France, <sup>7</sup> Division of Crop Genetics and Breeding, Crop Research Institute, Prague, Czech Republic, <sup>8</sup> Department of Grasses, Legumes and Energy Plants, Plant Breeding and Acclimatization Institute-National Research Institute, Radzikow, Poland, <sup>9</sup> Laboratory of Mycology and Phytopathology, All-Russian Institute of Plant Protection, St. Petersburg, Russia, <sup>10</sup> Department of Genetics, Faculty of Biology, Complutense University of Madrid, Madrid, Spain, <sup>11</sup> Norwegian Institute of Bioeconomy Research, Ås, Norway, <sup>12</sup> Department of Plant Protection, Agriculture Faculty, Namık Kemal University, Tekirdag, Turkey, <sup>13</sup> Laboratory of Phytopathology and Entomology, Maize Research Institute Zemun Polje, Belgrade, Serbia, <sup>14</sup> Plant Breeding Institute, University of Hohenheim, Stuttgart, Germany, <sup>15</sup> Leibniz Centre for Agricultural Landscape Research, Institute for Landscape Biogeochemistry, Müncheberg, Germany, <sup>16</sup> Applied Microbiology, Earth and Life Institute, Université Catholique de Louvain, Louvain-la-Neuve, Belgium, <sup>17</sup> Department Natural Resources and Bioproduction, Natural Resources Institute Finland (Luke), Jokioinen, Finland, <sup>18</sup> Section for Eukaryotic Biotechnology, DTU Systems Biology, Technical University of Denmark, Kongens Lyngby, Denmark, <sup>19</sup> Section for Chemistry and Toxicology, Norwegian Veterinary Institute, Oslo, Norway, <sup>20</sup> Molecular Plant Biology, Department of Biochemistry, University of Turku, Turku, Finland, <sup>21</sup> Research Division Grassland Sciences and Agro-Ecosystems, Institute for Sustainability Sciences, Agroscope, Zürich, Switzerland

Fusarium species, particularly Fusarium graminearum and F. culmorum, are the main cause of trichothecene type B contamination in cereals. Data on the distribution of Fusarium trichothecene genotypes in cereals in Europe are scattered in time and space. Furthermore, a common core set of related variables (sampling method, host cultivar, previous crop, etc.) that would allow more effective analysis of factors influencing the spatial and temporal population distribution, is lacking. Consequently, based on the available data, it is difficult to identify factors influencing chemotype distribution and spread at the European level. Here we describe the results of a collaborative integrated work which aims (1) to characterize the trichothecene genotypes of strains from three Fusarium species, collected over the period 2000–2013 and (2) to enhance the standardization of epidemiological data collection. Information on host plant, country of origin, sampling location, year of sampling and previous crop of 1147 F. graminearum, 479 F. culmorum, and 3 F. cortaderiae strains obtained from 17 European countries was compiled and a map of trichothecene type B genotype distribution was plotted for each species. All information on the strains was collected in a freely accessible and updatable database (www.catalogueeu.luxmcc.lu), which will serve as a starting point for epidemiological analysis of potential spatial and temporal trichothecene genotype shifts in Europe. The analysis of the currently available European dataset showed that in F. graminearum, the predominant genotype was 15-acetyldeoxynivalenol (15-ADON) (82.9%), followed by 3-acetyldeoxynivalenol (3-ADON) (13.6%), and nivalenol (NIV) (3.5%). In F. culmorum, the prevalent genotype was 3-ADON (59.9%), while the NIV genotype accounted for the remaining 40.1%. Both, geographical and temporal patterns of trichothecene genotypes distribution were identified.

Keywords: acetyldeoxynivalenol, chemotype, database, *Fusarium*, genotype, mycotoxin, nivalenol, trichothecene

#### INTRODUCTION

Fusarium head blight (FHB) is one of the most important cereal diseases worldwide. Severe outbreaks of FHB may result in significant yield losses of up to 50%, depending on the small grain cereal crop (Parry et al., 1995). McMullen et al. (2012) suggested that FHB in the United States might lead to economic losses in excess of one billion USD per year. More importantly is the production of secondary metabolites, specifically mycotoxins, contaminating the harvested products and thus jeopardizing food and feed safety (e.g., Snijders, 1990).

In cereals, FHB is usually caused by a set of different Fusarium species, with different life styles and different types of mycotoxins produced. Within the Fusarium graminearum species complex (FGSC; O'Donnell et al., 2000), which presently includes 16 species (Aoki et al., 2012), F. culmorum and F. cerealis are among the most dominant pathogens causing head blight on wheat and other cereals worldwide (Moss and Thrane, 2004; Osborne and Stein, 2007). Other frequently detected species are F. poae, F. avenaceum, F. langsethiae, F. tricinctum, F. sporotrichioides (Ioos et al., 2004; Xu et al., 2005; Xu and Nicholson, 2009; Somma et al., 2014), and the non-toxigenic species Microdochium nivale and M. majus (Glynn et al., 2005).

One of the main Fusarium mycotoxin classes are the trichothecenes, sesquiterpene epoxides that inhibit eukaryotic protein synthesis, which may cause severe toxicosis in humans and animals (Ueno, 1983; Maresca, 2013). Fusarium trichothecenes are grouped into two classes based on the presence (type B) vs. absence (type A) of a keto group at the C-8 position (Kristensen et al., 2005). Depending on differences in the core trichothecene cluster (TRI cluster), which includes two regulatory genes (TRI6 and TRI10) and most of the biosynthetic enzymes required for the production of trichothecenes (Kimura et al., 2007; Alexander et al., 2011), Fusarium species as well as individual strains may produce different types of trichothecenes.

Among the type B trichothecenes, the following are considered to have a significant impact on food and feed safety: deoxynivalenol (DON), nivalenol (NIV), and their acetylated derivatives 3-acetyldeoxynivalenol (3-ADON), 15-acetyldeoxynivalenol (15-ADON), and 4-acetylnivalenol (4-ANIV, syn. fusarenone-X; Eriksen et al., 2004; Desjardins, 2006).

Different Fusarium species chemotypes have been described: chemotype I, produces DON and/or its acetylated derivatives while chemotype II, produces NIV and/or 4-ANIV (Sydenham et al., 1991). The DON chemotype can be further split into chemotype IA (producing 3-ADON) and IB (producing 15-ADON; Miller et al., 1991). Structural differences among toxins from different chemotypes may have consequences on strain fitness, since the specific pattern of oxygenation and acetylation can modify the bioactivity and hence the (phyto) toxicity of these compounds (e.g. Ward et al., 2002; Alexander et al., 2011). As it has been shown in a large survey on Canadian grains, DON and NIV, being the two most abundant toxins detected, now represent the two major concerns for safety of wheat and barley products (Tittlemier et al., 2013).

Environmental factors may result in a geographical partitioning of subpopulations that may coincide with chemotypes. The success of a given chemotype, which is of importance with respect to FHB control, is often related to local factors (van der Lee et al., 2015). Based on chemotype characterization of Italian Fusarium species, Covarelli et al. (2015) suggested that climatic conditions have a strong impact on the occurrence of 3-ADON and 15-ADON whereas NIV contamination occurred irrespective of climatic conditions. Yli-Mattila et al. (2013) proposed that the prevalence of a specific chemotype may also be influenced by a certain host. For example, NIV-producing strains were found to be more aggressive towards maize compared with DON-producers (Carter et al., 2002) and were associated, in F. asiaticum, preferentially to maize in China (Ndoye et al., 2012). Maier et al. (2006) postulated NIV to be a virulence factor in maize, which is in line with findings that associate an increase in NIV populations in areas where the preceding crop was maize (Audenaert et al., 2009; Pasquali et al., 2010; Sampietro et al., 2011).

Two main reasons to determine the chemotype of a strain have been proposed (Pasquali and Migheli, 2014): (1) to obtain epidemiological information on the population colonizing a crop in a given area, using chemotype as a proxy in the field; (2) to inform on the toxigenic risk of contaminated food or feed determined by the presence of a certain chemotype, with the long term perspective of developing preventive models and strategies to decrease the risk.

At present, data on chemotype distribution of FGSC are available from all continents (reviewed in Pasquali and Migheli, 2014), being F. graminearum sensu stricto (s.s.) the most studied species. Less work has been devoted to chemotype determination in F. culmorum (Scherm et al., 2013). Shifts in species population have been reported in many surveys (Xu et al., 2005; Nielsen et al., 2011; Fredlund et al., 2013), but reports on chemotype shifts in certain areas are more recent (e.g. Nielsen et al., 2012; Beyer et al., 2014).

Despite the fact that information from all continents is now available, most reports do not include complete information on the strains analyzed, such as geographic origin, host from which it was isolated, method used for species identification, etc. In addition, precise characterization of the species is frequently lacking, being based only on morphological observations, hence, making it unfeasible to use the dataset for further studies.

The main goal of this joint study was to generate an accessible map of trichothecene genotypes from three FHB causing species with detailed information on how the data were obtained. This will allow, in the long term, the acquisition of consistent and homogenous datasets providing a valid comparison of the distribution of chemotypes during years and among countries. To reach this aim, research institutions from 17 European countries were inquired to provide data on how the sampling was performed as well as detailed information on cropping history and location. A more coordinated effort, leading to common protocols for sampling, chemotype determining and data reporting in a more accessible way would increase the standardization of epidemiological data. Furthermore, it could facilitate the effort of understanding which factors do favor establishment and persistence of a specific chemotype. This collective effort is now assembled in a fully accessible and upgradable dataset of chemotype diversity within FGSG and F. culmorum on cereals in Europe.

#### MATERIALS AND METHODS

#### Data Collection

An Excel template file was sent to research partners agreeing to participate in the initiative (**Supplementary File 1**). The information to be submitted (if applicable) were as follows: chemotype, year of isolation, whether the strain was obtained by a single spore or a single hyphae, the location including the geographic coordinates, the crop host and cultivar from which it was isolated, previous crop, method of isolation, method used for species attribution, primers and/or gene in case of PCR and sequencing, name of culture collection in case it was deposited, strain ID, and citation of the strain/s in a publication. Whenever genetic chemotype or species was unknown, strains were shipped to the Luxembourg Institute of Science and Technology (LIST) laboratory for genetic chemotyping (Pasquali et al., 2011) and species identification by sequencing EF1alpha (Geiser et al., 2004). The overall dataset (www.catalogueeu.luxmcc.lu; available as of mid-April 2016) was built through integrating data communicated by research partners and by laboratory results obtained with the procedures described below.

## DNA Extraction and Chemotype Determination

Fungal colonies were grown on PDA as described in Pallez et al. (2014) in order to directly extract DNA using a rapid procedure. Briefly, a 2–5 mm piece of miracloth tissue (Millipore, USA) covered by a 5 days old fungal culture, was collected and added to 100 µL TE (10 mM Tris-Cl, 0.05 mM pH 9 EDTA solution, Sigma-Aldrich, USA). After 5 min of microwave treatment at 900 W and a 30 s centrifugation at 13,000 g, 5 µL were then used for PCR reactions. When identification of the strain was reported to be putative by partners, EF1alpha amplification was carried out, followed by sequencing as described in Dubos et al. (2011). If the species was previously defined, tri12 multiplex PCR for genetic chemotyping (Ward et al., 2002) was carried out. All PCR reactions were performed in a 50 µL volume to avoid risk of PCR inhibition due to the quick extraction method using 2X Phusion master mix (Thermo, USA), 300 nM of each primer and water. Thermal cyclers used were Biometra T-professional and Veriti PCR Thermal cyclers (Life Technologies, USA) using the programme as described in Pasquali et al. (2011). All reactions included positive controls for the three chemotypes and a negative control for monitoring potential contaminations. Reactions were visualized on a Biorad agarose ready to use gel at 3%, using a UV spectra analyzer (Ingenius, Syngene, UK). When results were ambiguous the reaction was repeated at least once.

Data assembled from other laboratories were collected by the Excel template file and uploaded to the database page. When diverse methods for genetic chemotyping were used (Waalwijk et al., 2003; Jennings et al., 2004; Quarta et al., 2006; Starkey et al., 2007; Yli-Mattila and Gagkaeva, 2010) by the original isolating laboratory, this fact is specified directly in the database.

#### Statistical Analysis and Visualization Tools

Descriptive graphs on species and chemotype distribution were obtained using SigmaPlot version 12.5 **(**Systat Software Inc, USA) and SPSS version 19 (IBM, USA). The European maps generated for this study were prepared using the ArcMap platform (ESRI Inc., USA). A Multiple Correspondence Analysis tool (Broeksema et al., 2013) was used for studying the overall dataset and its homogeneity in relation to species and chemotype distribution. Logistic regressions were performed using SigmaPlot 12.5.

#### Database Construction

The European database was constructed by assembling the overall dataset on the database architecture developed by Piec et al. (2016). A filtering option for country and laboratory, the option to upload new datasets, with administrator validation, was added to the existing architecture. Functions of the database include the possibility to have a full or filtered download of the overall dataset.

## RESULTS

The current work represents the first collective attempt to compile information on chemotype diversity occurring in European countries. Moreover, the availability of a full open access database provides for the first time a centralized source of information for Fusarium disease records on cereals, which is of high value for researchers working in the mycotoxin/Fusarium biodiversity domain.

#### The Database

The overall dataset including all information collected for this work has been assembled in a database. Based on the previous architecture constructed for the LIST culture collection (Piec et al., 2016), a database with improved functionalities was built. The overall map with overlapping F. graminearum and F. culmorum species is shown on the first page of the database (www.catalogueeu.luxmcc.lu; **Figure 1**). Further uploading of data can be performed according to the instructions in **Supplementary File 2**. Researchers working on Fusarium toxigenic diversity are invited to contribute to the database or to download the dataset for further analysis.

## Data Description

Information of a total of 1147 F. graminearum and 479 F. culmorum strains was included in the dataset collected from the period 2000–2013 and plotted on the respective maps (**Figures 2A,B**). Years of isolation were close to homogeneity (**Figure 3A**). Luxembourg was the country where most strains were obtained, followed by Belgium and Russia (**Figure 3B**). At present, chemotype information from some countries is missing in the current dataset, therefore, further uploading of information will be important to obtain a more precise picture of chemotype distributions in Europe.

The major crop from where strains were isolated was wheat (66.7%) followed by maize (22.5%), barley (5.4%), and other crops (combined 5.3%; **Figure 3C**). As can be observed by the map of crop distribution, wheat was sampled in 16 out of the 17 countries, whereas other crops were sampled in a limited number of countries (maize n = 6; barley n = 7; oats = 3; **Figure 4**). Oats were sampled only in Northern Europe, including Norway, Finland and Russia, where oats are an important crop, while no

FIGURE 1 | Spatial distribution of chemotypes and *Fusarium* species in Europe. 3-ADON, 3-acetyldeoxynivalenol; 15-ADON, 15-acetyldeoxynivalenol; NIV, nivalenol. F. cortaderiae were isolated in Italy but cannot be visualized as they are overlapped by other strains.

FIGURE 2 | (A) Spatial distribution of *Fusarium culmorum* chemotypes in Europe. Red squares, genetic 3-ADON chemotype. Yellow squares, genetic NIV chemotype. (B) Spatial distribution of *Fusarium graminearum sensu stricto* chemotypes in Europe. Green circles, genetic 15-ADON chemotype. Red circles, genetic 3-ADON chemotype. Yellow circles, genetic NIV chemotype.

FIGURE 3 | Frequency distributions of (A) the years, when the fungal strains for the current database were isolated, (B) the countries of origin of the strains, (C) the host plants from which strains were isolated, (D) the previous crops ("Other" include mixtures of legumes and cereals, lucerne, lupines, perennial forages, spinach, and sugar beet), and (E) the species/chemotype combinations found. Only 3 Fusarium cortaderiae strains were included and are thus not visible in this figure. In the host plant figure, "other crops" include thistles, soya and potatoes. "Other poaceae" include forage grasses, einkorn wheat, triticale, wild type barley.

maize was sampled in Northern Europe, where the climate is not yet suitable for maize production.

Previous crop information was available only for a minority of samples (<50%). Therefore, the analysis of these data was postponed until enough data are uploaded to allow meaningful conclusions (**Figure 3D**).

With respect to the overall distribution of chemotypes per species for F. graminearum, the predominant genotype was 15-ADON (82.9%, 951 strains), followed by 3-ADON (13.6%, 156 strains), and NIV (3.5%, 40 strains). The 15-ADON genotype was most common in isolates from wheat and maize, while the 3-ADON genotype was most common in northern Europe and oats. For F. culmorum, the prevalent genotype was 3-ADON (59.9%, 287 strains), while the NIV genotype accounted for the remaining 40.1% (192 strains). Three F. cortaderiae with the NIV chemotype were also included (**Figure 3E**). The chemotype distribution within each country can be accessed through the filtering options available online (see **Supplementary File 3** for graphical representation). Interestingly, only 3-ADON isolates of F. graminearum and F. culmorum were found in the collected European isolates from oats.

## Data Analysis

Multiple correspondence analyses including year, country, and host plant showed no evidence for a preferential association of the species (F. graminearum or F. culmorum) with specific countries (both species were present in 16 out of 17 countries), sets of countries, years or crops (**Figure 5A**). On the contrary, when the corresponding analysis was performed on the chemotype dataset, it was evident that chemotypes were not randomly distributed over countries, years, and crops (**Figure 5B**).

For further analyses of the chemotype distribution, we focused on the most abundant population obtained from the same host. A total of 784 F. graminearum s.s. strains were isolated from wheat. The 15-ADON chemotype was rarely observed in Northern Europe (**Figure 2B**). The latitude that marked the Northern limit of the 15-ADON chemotype distribution in F. graminearum strains isolated from wheat in Europe was estimated by logistic regression: Hardly any 15-ADON chemotype strains of F. graminearum were found above 54.4 ± 10.8◦ Northern latitude while the probability for a 15-ADON chemotype strain in a F. graminearum population more southwards converged to 95.5 ± 0.85% (**Figure 6A**).

Similarly, in F. culmorum, the percentage of NIV chemotype strains drastically dropped off East of 7.5 ± 0.6◦ longitude **(Figures 2A**, **6C**). West of 7.5 ± 0.6◦ longitude, the probability of observing a NIV chemotype strain in a F. culmorum population isolated from wheat converged to 62.7 ± 12.6%. In F. graminearum, the probability of finding a NIV chemotype strain was 15.9 ± 2.5% West of 5.0 ± 0.8◦ longitude East, but dropped quickly further eastwards **Figures 2B**, **6B**.

The availability of temporal series of strains allowed also verifying the possible shift of species or chemotypes in regions with high data density. By selecting the densest sampled area from the available dataset (Luxembourg, year 2007– 2012), it was possible to observe how the NIV chemotype in

F. graminearum disappeared at the end of the sampling period (**Supplementary Video 1**).

## DISCUSSION

By assembling the dataset described here, we could establish a comprehensive collection of European data on Fusarium diversity on cereals. The analysis of the distribution of genetically determined chemotypes confirms the dominance of the 15- ADON chemotype in Western, Southern and Central Europe in F. graminearum (Pasquali and Migheli, 2014) but at the same time, we identified some current geographic limits to its distribution in wheat. Whether this depends on cropping and/or climatic factors merits further investigation given the fact that the 15-ADON chemotype is currently the major cause of DON accumulation in European wheat. Our data are generally in agreement with earlier reports on heterogeneous chemotype distribution in Europe (Yli-Mattila et al., 2009; Aamot et al., 2015; van der Lee et al., 2015). However, earlier reports did not estimate where the geographical limits of the chemotype spatial distribution are. Our study indicates a limited spread of 15-ADON trichothecene genotypes toward the Northern latitudes. This is confirmed by the results from a recent study in Norway where 3-ADON was the dominating trichothecene genotype (Aamot et al., 2015). However, the 15-ADON type was recently introduced into Norway, probably from other parts of Europe (Aamot et al., 2015), which exemplifies the need for a common database to monitor genotype shifts in Europe. A hypothesis worth testing would be to verify if parameters such as temperature, light irradiation, cropping practices, and/or host plants typical of certain latitudes may have an impact on the spread of the 15-ADON population by combining phenotypic tests with studies on the homogeneity of the fungal population.

The dominance of the 3-ADON chemotype in Northern Europe, which is in accordance with previous results of Yli-Mattila et al. (2009), Yli-Mattila and Gagkaeva (2010), and Yli-Mattila et al. (2013), was also confirmed. The role of oats as the potential preferential host for the 3-ADON population of F. graminearum seems to be confirmed in our dataset.

We also observed that the NIV chemotype was preferentially found in Western Europe. The fact that NIV chemotypes in F. culmorum were rarely found in Eastern locations, suggests that possibly different populations associated to chemotype diversity might be adapted to distinct cropping practices or to distinct climatic conditions, hence, further diversity studies on F. culmorum populations are needed.

Furthermore, we could identify temporal patterns of chemotype distribution that were partially associated to temporal changes in climatic conditions as observed in Luxembourg (Beyer et al., 2014). Enlarging the dataset would allow similar comparisons across European countries as has been done for example in North America and within China. In fact, analyses of the trichothecene chemotype distributions across Canada (Ward et al., 2008) revealed a dramatic longitudinal cline in which 3-ADON producers from wheat were significantly more common in Eastern Canada than in Western provinces (Ward et al., 2008), amounting to a 14-fold increase between 1998 and 2004. The authors suggested that the rapid increase in 3-ADON frequency in Western Canada might indicate that 3-ADON populations have a selective advantage over isolates from the resident 15-ADON population. The reason for the observed shift is unclear but they hypothesized that it could be due to changes in agricultural practices or environmental conditions. Guo et al. (2008) hypothesized that the shift occurring in Manitoba might have been produced by seed shipment and long distance transportation of spores. In the study by Zhang

et al. (2010), where more than 400 isolates of F. asiaticum (part of the F. graminearum species complex) from barley at 18 sites (10–2000 km apart) in three valleys of Southern China were analyzed, a significant cline of 3-ADON producers was observed in the middle valley, but no correlations with climate or agronomy factors were identified.

Certainly, the dataset presented here includes some gaps that should be filled in order to perform increasingly reliable analyses on the potential causes of change in species or chemotype distribution. It is expected that further updates which could include also sowing date, tillage regime, fungicide treatments, and other cultural practices will strengthen the dataset and allow a better understanding of the effects of cropping measures and environmental factors on strain distribution.

Nevertheless, it is evident that molecular genotyping is a powerful tool to support or refute epidemiologically generated hypotheses (Litvintseva et al., 2015). Numerous databases related to fungal diversity and bioproducts are available (Wackett, 2015), as well as fungal repositories for specimens (Abd-Elsalam et al., 2010), but they rarely integrate different sources of information on strain diversity. Our database includes molecular datasets associated to geographic and cropping practices and with the availability of GPS data, our database can be indirectly enriched with meteorological datasets related to the location where the strains were collected.

It has been suggested that "large datasets containing epidemiological data associated to genetic information can help understanding, recognizing and eventually, managing fungal

graminearum/F. culmorum population.

outbreaks" (Litvintseva et al., 2015). Hence, given the importance of shared data for fungal epidemiological studies and considering the interactions between environmental, cropping, and genetic factors, our up-scalable and fully open access database can possibly help addressing future risks of spread of toxigenic Fusarium populations on cereals.

This work also represents an example of a European participative and cooperative approach that can serve as an example for the establishment of other epidemiological studies profiting from the availability of large and well maintained datasets. Given the complexity of FGSC, future efforts will not only aim at inclusion of data from other countries and new fungal strains but will also have to integrate multi locus sequence genotyping information (MLST) to better characterize diversity and possibly even new species or the level of population diversity (Ward et al., 2002; Talas et al., 2011; Aamot et al., 2015; Talas and McDonald, 2015).

At the same time, this entirely accessible dataset is essential for allowing further targeted studies in order to fully differentiate all included strains at both species and subpopulation levels, assuming that F. graminearum as well as F. culmorum are constituted by different subpopulations (Liang et al., 2014; Balmas et al., 2015; van der Lee et al., 2015).

In the current version of the dataset, genetic determination of the chemotype was carried out to differentiate the three major known chemotypes in Europe. It is foreseeable to further analyze the strains by investigating the presence of the NX-2 chemotype that has been currently found only in the USA (Fruhmann et al., 2014; Liang et al., 2014; Varga et al., 2015).

Employing a flexible, up-scalable and upgradable database, our work represents the first attempt to build a global database in which strains, provided with information of GPS data and host, are also analyzed using multi-locus genotyping in combination with VNTR (Variable Number Tandem Repeat) screening and potentially whole genome sequencing (van der Lee et al., 2015). The success of such an effort will depend on the future contributions: the availability of a well-maintained and expanded database is a solid contribution to a shared approach philosophy of conducting research that will help to speed up scientific progress in fungal biology and agriculture (Abbà et al., 2015).

### AUTHOR CONTRIBUTIONS

MP conceived and performed the experiments, coordinated the assembly, analyzed the data, and wrote the manuscript. SV conceived the experiments, coordinated the assembly, assembled and analyzed the data, and wrote the manuscript. MB performed the experiments, analyzed the data and wrote the manuscript. JP assembled and managed the database. AL, KA, VB, RB, AB, JC, EC, TG, MG, IH, NK, LH, JL, PG, TM, QM, AM, MM, FM, PP, MP-B, JS, BS, SS, UT, SU, AV, and TY performed experiments and helped in writing the manuscript.

### FUNDING

The Luxembourg Institute of Science and Technology, LU, acknowledges the Ministère de l'Agriculture, de la Viticulture et de la Protection des Consommateurs-Administration des Services Techniques de l'Agriculture for financially supporting the Sentinelle project. The work on Italian strains has been financially supported through the M.I.U.R. Project AGROGEN (Laboratory of GENomics for traits of AGROnomic importance in durum wheat: Identification of useful genes, functional analysis and assisted selection by biological markers for the development of the national seed chain) (D. D. 14.03.2005 n. 602/Ric). Funding for the research of Ryan Basler was provided by Felix Thornley Cobbold Trust and the John Oldacre Foundation.

The work of JC was supported by the Ministry of Agriculture of the Czech Republic, Project No. RO0415. The research of MG and PG was supported by the Spanish Ministry MINECO (AGL2014-53928-C2-2-R). The Ministry of Agriculture and Food, Norway funded the work of IH. The research of TM was funded by the Federal Ministry of Education and Research (BMBF) (GABI-KANADA #FKZ 0313711A), Bonn and by the German Academic Exchange Service (DAAD), Bonn (code no.: A/06/92183). PP acknowledges the Finnish Ministry of Agriculture and Forestry for funding the project FinMyco on Fusarium and mycotoxins in Finland. The research of JS was funded by the Direction Générale de l'Agriculture, Direction de la Recherche (ref. D31-3159, D31-1162, D31-7055), in the framework of a project entitled "Caractérization et dynamique des fusarioses sur maïs en Région Wallonne." BS acknowledges support by P.O.R. SARDEGNA F.S.E. 2007–2013—Obiettivo competitività regionale e occupazione, Asse IV Capitale umano, Linea di Attività l.3.1 (research project "Identification of natural and natural-like molecules inhibiting mycotoxin biosynthesis by Fusaria pathogenic on cereals"). UT thanks the Danish Directorate for Food, Fisheries and Agri Business grant FFS05-3 for financial support. The work of TY was financially supported by the Academy of Finland (no. 126917, 131957, 250904, 252162, 267188, and 266984), Olvi Foundation, Turku University Foundation, a CIMO travel grant to Taha Hussien, and the Nordic network project New Emerging Mycotoxins and Secondary Metabolites in Toxigenic Fungi of Northern Europe (project 090014), which was funded by the Nordic Research Board.

### ACKNOWLEDGMENTS

We would like to acknowledge the MycoRed project for patronizing the activity in the framework of a scientific alliance. The MycoKey project is also acknowledged. Friederike Pogoda, Céline Brochot, Servane Contal, Boris Untereiner, Frédéric Giraud, Alyssa Cowles, and Tiphaine Dubos from the Luxembourg Institute of Science and Technology (LU) and Irene Bänziger and Eveline Jenny from Agroscope (CH) are acknowledged for their excellent technical support in fungal collection, isolation and management. Bertjan Broeksema (LU) is acknowledged for supporting the data analyses. Jana Palicová and Tat'ána Sumíková from the Crop Research Institute (CZ) are acknowledged for their excellent technical support in fungal collection, isolation and management. The Finnish Cereal committee is acknowledged for the research material. Veli Hietaniemi is acknowledged for project management and Marjaana Virtanen for the excellent technical support in fungal isolation and management [Natural Resources Institute Finland (Luke)], FI. The Kimen seed health laboratory is acknowledged for collecting Norwegian F. graminearum isolates for IH, NIBIO, NO.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fmicb. 2016.00406

#### Supplementary File 1 | Template for participants of the original initiative.

Supplementary File 2 | Instruction to upload new dataset on the database and an output example for a given strain.

### REFERENCES


Supplementary File 3 | (A) *Fusarium graminearum* and (B) *Fusarium culmorum* chemotype distribution within each participating country in % of all samples analyzed (n). 3-ADON, 3-acetyldeoxyniavlenol; 15-ADON, 15-acetyldeoxyniavlenol; NIV, nivalenol.

#### Supplementary Video 1 | *Fusarium graminearum* sampled during 6 years

in Luxembourg. Every year, at least 15 fields were sampled in different geographic locations in the country. Depending on the year, the species was more or less present. The NIV chemotype disappeared in the last 4 years of sampling in this species.


compared to those in Southern Europe. Microorganisms 1, 162–174. doi: 10.3390/microorganisms1010162

Zhang, H., Zhang, Z., van der Lee, T., Chen, W., Xu, J., Xu, J., et al. (2010). Population genetic analyses of Fusarium asiaticum populations from barley suggest a recent shift favoring 3ADON producers in Southern China. Phytopathology 100, 328–336. doi: 10.1094/PHYTO-100-4-0328

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2016 Pasquali, Beyer, Logrieco, Audenaert, Balmas, Basler, Boutigny, Chrpová, Czembor, Gagkaeva, González-Jaén, Hofgaard, Köycü, Hoffmann, Levi´c, Marin, Miedaner, Migheli, Moretti, Müller, Munaut, Parikka, Pallez-Barthel, Piec, Scauflaire, Scherm, Stankovi´c, Thrane, Uhlig, Vanheule, Yli-Mattila and Vogelgsang. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Trichothecene Genotypes of the Fusarium graminearum Species Complex Isolated from Brazilian Wheat Grains by Conventional and Quantitative PCR

#### Sabina M. Tralamazza\*, Raquel Braghini and Benedito Corrêa

Laboratory of Mycotoxins and Toxigenic Fungi, Institute of Biomedical Sciences, University of São Paulo, São Paulo, Brazil

We compared two well-established methods, fungal isolation followed by conventional PCR and DNA analysis by quantitative PCR (qPCR), to define trichothecene genotypes in Brazilian wheat grains from different locations. For this purpose, after fungal isolation from 75 wheat samples, 100 isolates of the Fusarium graminearum species complex (FGSC) were genotyped by PCR to establish their trichothecene profile. For profiling by qPCR, DNA was extracted from the wheat samples and analyzed. The methods provided similar and divergent results. The FGSC isolates were classified as NIV (55%), 15-ADON (43%), and 3-ADON (2%). Analysis by qPCR showed 100% contamination with 15-ADON strains in all wheat samples, 80% contamination with the NIV genotype, and only 33.3% contamination with 3-ADON strains. Further analysis revealed that 96% of all quantified DNA was attributed to the 15-ADON profile, while 3.4% was attributed to NIV and only 0.06% to 3-ADON. A positive correlation was observed between 15-ADON genotype DNA concentration and deoxynivalenol (DON) content in the wheat samples. The high frequency of fungi, DNA levels and positive correlation with DON strongly indicate that 15-ADON producers are the main trichothecene genotype in Brazilian wheat grains. Surprisingly, although many isolates (55%) carried the NIV genotype and this genotype was identified in 80% of the wheat samples, only 3.4% of fungal DNA was in fact from NIV producers. Although, our findings showed that each method provided a different perspective about the trichothecene profile, DNA analysis by qPCR gave us new insight about fungal contamination levels in Brazilian wheat grains. Nevertheless, both techniques should be used to obtain more robust results.

#### Keywords: genotype, trichothecenes, wheat, Fusarium, qPCR, deoxynivalenol

## INTRODUCTION

The continuous increase in wheat production in Brazil (USDA, 2014) and recent mycotoxin regulations (ANVISA, 2011) indicate the urgent need for more studies about fungal diversity and mycotoxin profiles found in these grains to ensure a good-quality and safe product for human and animal consumption. Members of the Fusarium graminearum species complex (FGSC) are the main fungal agents associated with Fusarium head blight (FHB) in wheat and other cereal

#### Edited by:

Thomas Miedaner, University of Hohenheim, Germany

#### Reviewed by:

Vijai Kumar Gupta, National University of Ireland, Galway, Ireland Lukasz Stepien, Polish Academy of Sciences, Poland

> \*Correspondence: Sabina M. Tralamazza sabina@usp.br

#### Specialty section:

This article was submitted to Fungi and Their Interactions, a section of the journal Frontiers in Microbiology

Received: 01 December 2015 Accepted: 15 February 2016 Published: 01 March 2016

#### Citation:

Tralamazza SM, Braghini R and Corrêa B (2016) Trichothecene Genotypes of the Fusarium graminearum Species Complex Isolated from Brazilian Wheat Grains by Conventional and Quantitative PCR. Front. Microbiol. 7:246. doi: 10.3389/fmicb.2016.00246

crops, a disease that causes severe grain losses for the industry every year (Goswami and Kistler, 2004; Osborne and Stein, 2007). In addition to causing FHB, FGSC produce type B trichothecenes and zearalenone, mycotoxins that pose hazardous health risks to humans and animals. The consumption of trichothecenes causes vomiting, feed refusal, anorexia and weight loss (Smith et al., 1997; Pestka and Smolinski, 2005), while zearalenone ingestion can induce significant changes in reproductive organs and fertility loss in animals and humans (Zinedine et al., 2007).

Type B trichothecenes synthesized by FGSC vary worldwide and their profile is normally divided into three categories: (1) nivalenol (NIV) and its acetylated form (4-ANIV); (2) deoxynivalenol (DON) and 15-acetyldeoxynivalenol (15- ADON); (3) DON and 3-acetyldeoxynivalenol (3-ADON) (Ward et al., 2002; Desjardins, 2006). Although these toxins are included in the same group, they may differ in terms of aggressiveness and toxicity (Lee et al., 2009; Puri and Zhong, 2010; Umpiérrez-Failache et al., 2013). In view of the possibility of different profiles, it is important to characterize potential mycotoxins in cereals and to analyze shifts in the toxigenic profile of fungal populations.

Trichothecene genotyping provides a rapid method to predict trichothecene production by Fusarium species. Current knowledge of TRI genes involved in the trichothecene biosynthetic pathway permitted to design specific primers for the identification of trichothecene genotypes. Primers based on the TRI3 and TRI12 genes were developed to differentiate 3-ADON, 15-ADON and NIV genotypes (Ward et al., 2002), and those based on the TRI13 and TRI7 genes to define DON and NIV genotypes (Chandler et al., 2003).

Most studies exclusively focus on genotyping strains isolated from wheat by PCR. Although less expensive, the technique is a qualitative analysis and requires a prior step, i.e., fungal isolation. On the other hand, qPCR permits qualitative and quantitative analysis and, more importantly, can be used to evaluate the genotype profile directly in the fungal substrate; however, the method is more expensive and requires a higher technical knowledge. Therefore, the objective of this study was to investigate the trichothecene profile of wheat grains using two methods, fungal isolation and conventional PCR and qPCR, in order to provide a more robust analysis of the trichothecene profile of the Fusarium population in Brazilian wheat.

#### MATERIALS AND METHODS

#### Wheat Samples

The wheat samples used in the study were previously analyzed for mycobiota and DON content (Tralamazza et al., 2016). Of the 150 samples collected in that study, 75 were randomly chosen for the present study. The freshly harvested wheat grains were collected in three different states (25 samples/region) in Brazil. The regions were chosen due to their importance for the wheat industry. At present, the states of Parana and Rio Grande do Sul are responsible for more than 90% of the total wheat production in Brazil (CONAB, 2015). Harvest occurred between September and November 2012 in Novo Itacolomi (Parana State, PR), Passo

#### Trichothecene Genotype Identification by Conventional PCR Fungal Isolation

As mentioned earlier, the wheat samples and subsequent fungal isolates were part of a previous work. For the present study, 100 FGSC strains were used. These strains were directly isolated from the 75 wheat samples used in the qPCR genotype investigation.

A small percentage (<1%) of Fusarium trichothecene producers from other Fusarium complex were found during the mycobiota study but none were isolated from the 75 samples used in the present study. Thus, for this study we only worked with fungal samples from the FGSC.

Briefly, subsamples (100 g) of the wheat grains were disinfected with commercial sodium hypochlorite solution (1%) for 1 min and washed two times with distilled water. Subsamples (100 grains) were transferred to PDA plates (10 grains/plate) and incubated for 5 days at 25◦C. For species identification, DNA was extracted as described in "Fungal DNA Extraction for Genotype Identification" and the elongation factor (EF-1α) gene was sequenced using the EF-1/EF-2 primers (O'Donnell et al., 1998). All amplification reactions were carried out in a volume of 25 µl containing 1x PCR buffer, 0.3 mM of each primer, 2.5 mM MgCl2, 0.2 mM dNTPs, 0.04 U/µl Taq DNA polymerase (Invitrogen, Carlsbad, CA, USA), and 100 ng template DNA. The PCR conditions were initial denaturation at 94◦C (5 min), followed by 35 cycles at 95◦C (30 s), 56◦C (30 s) and 72◦C (1 min), and a final extension step of 7 min at 72◦C. After DNA purification (Illustra ExoProStar, GE), sequencing was performed in an ABI 3730 DNA Analyzer (Applied Biosystems, Foster City, CA, USA) using the BigDye Terminator v3.1. kit (Applied Biosystems) according to manufacturer instructions.

#### Fungal DNA Extraction for Genotype Identification

Monosporic strains were cultured for 5 days at 25◦C on yeast extract sucrose (YES) agar. After growth, mycelia were scrapped off and DNA was extracted using the Easy-DNA kit (Invitrogen) according to manufacturer instructions.

#### PCR

Primers Tri13-F/Tri13DON-R were used for analysis of the DON profile and primers Tri13NIV-F/Tri13-R for NIV (Chandler et al., 2003). Multiplex PCR was performed to determine NIV, 3-ADON and 15-ADON genotypes using primers 12CON/12NF/12- 15F/13-3F (Ward et al., 2002). Amplification was carried out in a Veriti Thermal Cycler (Applied Biosystems) using the following cycle parameters: 95◦C (1 min), 25 cycles at 95◦C (30 s), 52◦C (30 s) and 72◦C (30 s), and a final extension at 72◦C (7 min). The amplification reactions was carried out in a volume of 25 µl containing 1x PCR buffer, 2.5 mM MgCl2, 0.3 dNTPs, 0.56 mM of each primer, 0.04 U/µl of Taq DNA polymerase (Invitrogen), and 100 ng template DNA.

## Trichothecene Genotype Identification by Quantitative PCR

#### DNA Extraction

A 50-mg aliquot of a 100-g ground wheat sample was transferred to a microtube with a 3-mm steel pearl and shaken for 3 min (50 rpm) on a TissueLyser LT (Qiagen, Venlo, The Netherland). Next, DNA was extracted using the Easy-DNA kit (Invitrogen) according to manufacturer instructions. Fungal isolates with defined trichothecene genotypes were cultured in YES agar medium for 5 days at 25◦C and used for the construction of efficiency and standard curves. Mycelia were scrapped from the medium and DNA was extracted using the Easy-DNA kit (Invitrogen) according to manufacturer instructions. Fungal and wheat DNA concentrations were determined in a Nanodrop 2000 UV-VIS spectrophotometer (Thermo Fisher, Waltham, MA, USA).

#### Validation of the Method

Quantitative PCR analysis was carried out according to Nielsen et al. (2012). Briefly, 5-point calibration curves were constructed for the fungal isolates of each genotype (3-ADON, 15-ADON, and NIV) using defined quantities of DNA. PCR efficiency was calculated from the slope of the linear relationship between the log10 values of DNA quantity and the cycle number (E = 10(−1/slope)−1). Target genes were amplified using the trichothecene genotype-specific primers (Nielsen et al., 2012), and primers for plant elongation factor (TEF1-α) (Nicolaisen et al., 2009) were used to determine DNA yield and possible nucleic acid degradation.

#### Quantification of Trichothecene Genotype DNA

The qPCR conditions and cycle protocol described by Nielsen et al. (2012) were used. The qPCR assays were carried out in a StepOnePlus Real-Time PCR System (Applied Biosystems). For the determination of genotype DNA concentration, the cycle threshold (Ct) of each sample was compared to the standard curve of the fungal isolate of each specific genotype. To obtain the final value, genotype DNA was normalized to the plant elongation factor, resulting in pg fungal DNA per µg plant DNA (Nicolaisen et al., 2009).

## Deoxynivalenol Analysis

#### Materials and Reagents

Mycotoxin standards (DON and deepoxydeoxynivalenol) were purchased from Sigma- Aldrich (São Paulo, Brazil). Acetonitrile, methanol and ammonium acetate were purchased from J. T. Baker (São Paulo, Brazil). Ultrapure water was obtained with a Milli-Q-System from Merck Millipore (Bedford, MA, USA).

#### Extraction of Deoxynivalenol

Three gram of wheat grains was ground and homogenized in 24 ml of a methanol/water solution (70: 30, v/v) and shaken for 20 min. The mixture was filtered through a Whatman No. 4 filter (18 cm). Prior to liquid chromatography-mass spectroscopy (LC-MS/MS) analysis, a 40-µl aliquot was transferred to a vial, mixed with 955 µl methanol/water (50: 50, v/v), and 5 µl of the internal standards previously diluted in methanol/water (50: 50, v/v) was added.

#### Chromatographic Conditions

The content of DON had been determined in a previous study (Tralamazza et al., 2016). DON was analyzed in an Agilent 1200 HPLC System (Agilent Technologies, Santa Clara, CA, USA) equipped with an API5000 triple quadruple mass spectrometer with an electrospray source (AB Sciex, Concord, ON, Canada). The LC column was a C8 Zorbax-XDB, 200 × 4.6 mm, 3 µm (Agilent Technologies) equipped with a C8 pre-column cartridge. For the mobile phase, methanol/water (60:40, v/v) containing 0.05 M ammonium acetate was used in an isocratic procedure at a flow rate of 1 ml/min. The column temperature was 35◦C and an injection volume of 5 µl was used. The MS sourcedependent parameters were: curtain gas 30 psi (240 kPa of maximum 99.5% nitrogen), dry gas (GS1) 50 psi (380 kPa of zero grade air), dry gas (GS2) 20 psi (105 kPa of zero grade air), collision-activated dissociation gas 12 (arbitrary unit), source temperature 360◦C, and ion spray voltage 5200 V. Detection was performed in the negative ion electrospray mode using multiple reaction monitoring. The retention time was 1.38 min. The declustering potential was set at −55 V, the collision energy at −30 eV, and the cell exit potential at −20V.

#### Statistical Analysis

Statistical analysis was performed using the GraphPad Prism software (GraphPad, 2014, v. 6.05). The Kruskal–Wallis test and Pearson's correlation test were used. A p-value < 0.05 was considered statistically significant.

## RESULTS

## Trichothecene Genotype Identification by Conventional PCR

All three genotypes were identified in the fungal isolates. However, wide variation in the frequency of the profiles was found. As can be seen in **Table 1**, 55% of the isolated fungi carried the NIV genotype, 43% the 15-ADON genotype, and only 2% the 3-ADON genotype. Mycobiota analysis of the wheat samples revealed that the 15-ADON genotype was attributed to Fusarium graminearum sensu stricto (s.s.), NIV to F. meridionale, and NIV or 3-ADON to F. cortaderiae and F. austroamericanum (Tralamazza et al., 2016), all belonging to the FGSC group. All F. graminearum s.s. isolates were characterized as 15-ADON, while all F. meridionale isolates were NIV. The F. cortaderiae and F. austroamericanum fungal isolates were found to be 3-ADON, but mainly of the NIV genotype (**Table 1**).

Regarding the origin of the fungal isolates, there were more NIV producers among strains from SP and RS, while the 15- ADON profile was more predominant among strains from PR. The 3-ADON genotype was not found in strains from RS (**Table 2**). Despite variations in frequency, the presence of the 15-ADON and NIV genotypes was constant in all three regions.



TABLE 2 | Frequency of trichothecene genotypes in members of the Fusarium graminearum species complex isolated in different wheat-producing regions of Brazil.


TABLE 3 | Frequency of trichothecene genotypes in wheat grains from different regions of Brazil.


### Trichothecene Genotype Identification by Quantitative PCR

The results showed that all wheat samples were contaminated with fungi carrying the 15-ADON genotype. The NIV and 3- ADON profiles varied according to region (**Table 3**). In SP, PR, and RS, the NIV genotype was detected in 68, 72, and 100% of the samples, respectively. The frequency of the 3-ADON genotype was lower and this genotype was found in 12, 36, and 52% of the samples from SP, PR, and RS, respectively.

In addition to frequency, qPCR permitted the quantification of fungal DNA. The results showed that most of the quantified DNA belonged to fungi carrying the 15-ADON genotype (96%), followed by a small portion of the NIV genotype (3.4%) and a non-significant quantity of the 3-ADON genotype (0.06%) (data not shown). Furthermore, significant DNA differences were observed between the regions studied (**Figure 1**). Samples from RS contained the highest concentrations of 15-ADON and NIV DNA, followed by samples from PR and SP. A different trend was observed for the 3-ADON genotype. Samples from PR were more contaminated than samples from RS. Nevertheless, the average amount of DNA was very low in all regions.

We also investigated the correlation between trichothecene genotype DNA and DON content in the wheat grains. The DON content in the wheat samples was determined in a previous study (Tralamazza et al., 2016). All 75 samples analyzed were contaminated. DON levels ranged from 183 to 1,903 µg/kg and varied across regions. The highest contamination was observed in RS (mean of 885 µg/kg), followed by PR (mean of 551 µg/kg) and SP (mean of 372 µg/kg) (data not shown). Our data showed a strong relationship between fungal DNA contamination and DON levels in the wheat samples. Grains from RS were the most contaminated by fungal DNA and showed the highest levels of DON, the same trend is seen with the samples from SP and PR (**Figure 2**). Pearson's correlation analysis with 15-ADON + 3- ADON genotype DNA and only with 15-ADON genotype DNA showed a positive and significant correlation (r = 0.68, p < 0.001, CI 0.53–0.79) between 15-ADON genotype DNA and DON content in wheat grains (**Figure 3**). Analysis also revealed that the small concentration of 3-ADON genotype DNA did not interfere with the correlation results.

#### DISCUSSION

Two methods were used to investigate the trichothecene genotype profile of Brazilian wheat grains. Both techniques demonstrated that 3-ADON is not a relevant genotype in Brazilian wheat grains. Although detected in some wheat samples, a very small portion of DNA (0.06%) was attributed to 3-ADON fungi and only two isolates were classified as 3-ADON producers.

Almost half the fungal isolates from the wheat samples were F. graminearum s.s. and carried the 15-ADON genotype. DNA analysis by qPCR revealed that the vast majority of the quantified DNA was from 15-ADON producers. Our results indicated that fungal DNA contamination had an impact in the DON levels content at each region studied. Also, analysis exhibited a significant correlation between 15-ADON DNA concentration and DON content in wheat grains. Other studies have reported a positive correlation between fungal biomass and DON using qPCR. Nielsen et al. (2012), working with wheat grains, found a strong correlation between DON and F. graminearum and between NIV and F. culmorum. Similar results have been reported in other studies correlating DON and F. graminearum biomass (Demeke et al., 2010; Horevaj et al., 2011).

Discrepant results were observed regarding the NIV genotype. More than half the fungal isolates carried the NIV genotype and were identified as F. meridionale, indicating a possible role of this species as a causative agent of FHB. However, different results were obtained when the wheat grains were analyzed by qPCR. The results showed the presence of NIV producers in most wheat samples analyzed, but the DNA concentration of fungi carrying the NIV genotype was proportionally smaller than that of the 15-ADON genotype.

Trichothecene genotype profiles have been reported for Brazilian wheat, but all studies have focused on fungal isolation analysis. Previous studies have reported F. graminearum s.s. (15-ADON) as the most frequent profile in wheat grains (93–83%), followed by small quantities of the NIV genotype (7– 13%), and no or <1% contamination with 3-ADON fungi (Scoz et al., 2009; Astolfi et al., 2012; Del Ponte et al., 2015). To our

knowledge, this is the first study to identify and quantify fungal genotypes directly in Brazilian wheat grains. The result of DNA quantification and the positive relationship with DON supports the role of F. graminearum s.s. (15-ADON) as the main FGSC species and strongly suggests that this species is responsible for the majority of DON production in Brazilian wheat grains in all three regions studied.

In the present study, strains carrying the NIV genotype were the most frequently isolated fungi in the wheat samples. QPCR analysis of the samples indicated that, although present in most samples, less than 4% of the target DNA belonged to NIV fungi. We could not explain why several F. meridionale strains were isolated, although 96% of the DNA detected in the samples was 15-ADON DNA. However, we hypothesize that the conditions of incubation and grain selection might have opened an opportunity for the emergence of other fungi.

Although the qPCR findings could be affected by DNA from other trichothecene-producing Fusarium species (e.g., F. culmorum, an NIV producer), the fungal isolation data together with existing studies on Brazilian Fusarium wheat mycobiota and genotype strongly indicate that most relevant species belong to the FGSC complex, especially those described in our study. Thus, even if present, it is unlikely that other trichothecene-producing species have significantly interfered with the results.

In the present study, a larger number of NIV genotype fungi were isolated than previously reported (Scoz et al., 2009; Astolfi et al., 2012; Del Ponte et al., 2015). Grain selection may have contributed to the differences found. The studies cited used FHB damaged kernels, while our wheat grains showed no signs or symptoms of FHB. In FHB damaged kernels, the tissue is heavily infected with the plant pathogen, a condition that decreases the chance of emergence of other fungi. Xu et al. (2007), studying the infection of wheat with different FHB pathogens (F. graminearum, F. poae, F. culmorum, and F. avenaceum), showed that the combination of pathogens led to competition between these species and to a substantial reduction in fungal biomass (>90% reduction) of the weaker pathogen.

Another possible explanation for the findings is that, under crop conditions, F. graminearum may be fitter than F. meridionale to infect and spread in plants, as suggested by some studies (Goswami and Kistler, 2005; Spolti et al., 2012). In vitro, fungal isolation at controlled temperature and on different substrates (PDA medium) may have interfered with factors such as fitness, aggression and growth rate and yielded the unexpected result. Nevertheless, the data should be viewed with caution. Although only 3.4% of the fungal biomass quantified was attributed to the NIV genotype, almost all wheat samples were to some extent positive for the NIV genotype. Del Ponte et al. (2012) detected NIV in 54/66 wheat grain samples from RS (mean level of 337 µg/kg), indicating a possible role of NIV fungi as a plant pathogen and mycotoxin producer in Brazilian wheat. Reports currently show F. graminearum (15-ADON) as the main causative agent of FHB and DON production in Brazilian wheat; however, it seems prudent to monitor the possible introduction of new genotypes or shifts.

In North America, the occurrence of 3-ADON strains has been increasing over the last decade, replacing the formerly predominant 15-ADON genotype as the main profile in wheat crop (Ward et al., 2008; Schmale et al., 2011). Some studies speculate that 3-ADON fungi are more aggressive and produce more DON than 15-ADON fungi (Puri and Zhong, 2010). Similar results were obtained when 3-ADON and NIV fungi were compared. In contrast, other authors did not find significant differences in aggressiveness or trichothecene production between fungi carrying the 3-ADON and 15-ADON genotypes (Spolti et al., 2012). The reason for this shift remains unclear. In Uruguay, F. graminearum s.s. (15-ADON genotype) is also the predominant FHB agent in wheat crop, but high levels of Fusarium asiaticum (NIV genotype) have been identified in new wheat crop areas near rice plantations (Umpiérrez-Failache et al., 2013). Rice crop has been reported to be highly infected with F. asiaticum (NIV genotype) in Brazil (Gomes et al., 2015).

#### REFERENCES


For a better understanding of disease symptoms and mycotoxin production, it is essential to determine the potential of the main fungal pathogens in cereal grains. Comparison of fungal isolation and conventional PCR with wheat DNA analysis by qPCR showed that both methods clearly made important contributions to the data obtained. Choosing two approaches permitted us to widen the perspective on the subject. The fungal isolation from the wheat samples provided a variety of isolated species and the possibility to work with the pathogen itself. In contrast, the use of qPCR gave us new information about trichothecene profiles in Brazilian wheat and showed for the first time a direct correlation between DON content and 15-ADON DNA levels, as well as high contamination with 15-ADON isolates in the grains, supporting the evidence that F. graminearum s.s. (15-ADON) is the main causative agent of FHB and producer of DON in Brazilian wheat. Although, the technique is more expensive and requires a higher technical level, the use of qPCR provided us with new information regarding the trichotheceneproducing Fusarium species impact, and in the future, we intend to study the correlation between genotype profiles and others types of trichothecenes to further understand the effect of different genotypes in Brazilian wheat crops.

### AUTHOR CONTRIBUTIONS

Conceived and designed the experiments: ST, RB, and BC. Performed the experiments: ST and RB. Analyzed the data: ST, RB, and BC. Wrote the paper: ST. Revised the paper: RB and BC.

#### FUNDING

This work was supported by the State funding agency Fundação de Amparo a Pesquisa do Estado de São Paulo (FAPESP).


on wheat and rice. Phytopathology 95, 1397–1404. doi: 10.1094/PHYTO-95- 1397


complex possessing either 15-ADON or NIV genotype. Eur. J. Plant Pathol. 133, 621–629. doi: 10.1007/s10658-012-9940-5


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2016 Tralamazza, Braghini and Corrêa. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Comparative Analysis of Deoxynivalenol Biosynthesis Related Gene Expression among Different Chemotypes of Fusarium graminearum in Spring Wheat

Chami C. Amarasinghe and W. G. Dilantha Fernando\*

Department of Plant Science, University of Manitoba, Winnipeg, MB, Canada

#### Edited by:

Daniela Gwiazdowska, Poznan University of Economics, ´ Poland

#### Reviewed by:

Sean Doyle, Maynooth University, Ireland Friday Obanor, Commonwealth Scientific and Industrial Research Organisation, Australia

#### \*Correspondence:

W. G. Dilantha Fernando dilantha.fernando@umanitoba.ca

#### Specialty section:

This article was submitted to Fungi and Their Interactions, a section of the journal Frontiers in Microbiology

Received: 14 April 2016 Accepted: 25 July 2016 Published: 08 August 2016

#### Citation:

Amarasinghe CC and Fernando WGD (2016) Comparative Analysis of Deoxynivalenol Biosynthesis Related Gene Expression among Different Chemotypes of Fusarium graminearum in Spring Wheat. Front. Microbiol. 7:1229. doi: 10.3389/fmicb.2016.01229 Fusarium mycotoxins, deoxynivalenol (DON) and nivalenol (NIV) act as virulence factors and are essential for symptom development after initial infection in wheat. To date, 16 genes have been identified in the DON biosynthesis pathway. However, a comparative gene expression analysis in different chemotypes of Fusarium graminearum in response to Fusarium head blight infection remains to be explored. Therefore, in this study, nine genes that involved in trichothecene biosynthesis were analyzed among 3-acetyldeoxynivalenol (3-ADON), 15-acetyldeoxynivalenol (15-ADON) and nivalenol producing F. graminearum strains in a time course study. Quantitative reverse transcription polymerase chain reaction revealed that the expression of all examined TRI gene transcripts initiated at 2 days post-inoculation (dpi), peaked at three to four dpi and gradually decreased at seven dpi. The early induction of TRI genes indicates that presence of high levels of TRI gene transcripts at early stages is important to initiate the biosynthetic pathway of DON and NIV. Comparison of gene expression among the three chemotypes showed that relative expression of TRI genes was higher in 3-ADON producing strains compared with 15-ADON and NIV strains. Comparatively higher levels of gene expression may contribute to the higher levels of DON produced by 3-ADON strains in infected grains.

Keywords: Fusarium graminearum, wheat, trichothecenes, chemotype, qRT-PCR, TRI genes

## INTRODUCTION

Fusarium head blight (FHB) is one of the major economically important fungal diseases in wheat, barley, corn, and other small grains worldwide. Wheat yield losses of up to 50% have been reported in North America due to FHB (McMullen et al., 1997; Goswami and Kistler, 2005). One of the major concerns of FHB is the contamination of infected grains with Fusarium mycotoxins. Fusarium mycotoxins represent the largest group of mycotoxins, which contains more than 140 known metabolites such as trichothecenes, zearalenone and fumonisins (Yazar and Omurtag, 2008; Sobrova et al., 2010). Among these mycotoxins, trichothecenes are one of the major Fusarium mycotoxins synthesized mainly by the members in the Fusarium graminearum species complex (FGSC), F. culmorum, F. sprotrichioides and F. poae (Desjardins et al., 1993; Foroud and Eudes, 2009; Wang et al., 2011). The fungi in the FGSC have the potential to devastate

a crop by reducing grain quality and quantity. After Fusarium infection, the grains become contaminated with trichothecene mycotoxins such as deoxynivalenol (DON), produced by the pathogen, making the crop unsuitable for food and feed. Trichothecenes produced by Fusarium spp. act as virulence factors in wheat plants. It has been reported that DON is important for the spread of F. graminearum beyond the point of infection within the host plant. Proctor et al. (1995) have shown that TRI5- mutants have reduced virulence compared to wild type strains in Wheat and Rye cultivars suggesting that trichothecene production contributes to the virulence of F. graminearum. Non-DON producing strains of F. graminearum can initiate the infection, but not spread within the host tissue (Proctor et al., 1995; Bai et al., 2002). A study done by Diamond et al. (2013) found that DON is capable of inhibiting the apoptosis– like programmed cell death in Arabidopsis cell cultures subjected to heat stress.

So far, 16 genes have been characterized in the DON biosynthesis pathway. These genes reside at four different loci on different chromosomes; the core TRI cluster consists of 12 genes located on chromosome 2, the TRI1-TRI16 loci on chromosome 1, TRI101 on chromosome 4, and TRI15 on chromosome 3, respectively (Gale et al., 2005; Alexander et al., 2009; Merhej et al., 2011). The first step in the DON biosynthesis pathway consists of the cyclization of the initial substrate, farnesyl pyrophosphate (FPP) to produce non-toxic trichodiene, by the trichodiene synthase enzyme encoded by TRI5 gene (Hohn and Beremand, 1989). The next nine reactions in the pathway are mediated by the enzymes encoded by TRI4, TRI101, TRI11 and TRI3, respectively. These reactions lead to the formation of calonectrin, which serves as a substrate for the production of 3-ADON, 15-ADON and 4-acetylnivalenol (4-ANIV) (Alexander et al., 2009; Foroud and Eudes, 2009; Merhej et al., 2011). The genes TRI7 and TRI13 are functional only in F. graminearum strains that are capable of producing NIV (Brown et al., 2001; Lee et al., 2002). The enzymes encoded by TRI7 and TRI13 genes mediate two common steps following calonectrin. In nivalenol producing F. graminearum strains, the pathway continues with the product of TRI1 to produce 4-ANIV and the final step mediated by TRI8 to give NIV. The TRI7 and TRI13 genes are not active in DON producers; therefore, DON biosynthesis proceeds directly from calonectrin with the enzymes encoded by TRI1 gene (McCormick and Alexander, 2002; Alexander et al., 2011; Merhej et al., 2011). The formation of 3-ADON or 15-ADON is strain specific and decided by the esterase coding sequence of TRI8 gene (Alexander et al., 2011). To date, limited research has been done on expression of TRI genes in different chemotypes of F. graminearum during wheat colonization.

Among the different TRI genes, TRI5 gene has received more attention and so far the majority of studies have focused on the expression of the TRI5 gene during Fusariumwheat colonization. A study done by Hallen-Adams et al. (2011) examined the expression of the TRI5 gene during wheat spike infection of susceptible and resistant cultivars and susceptible cultivars treated with strobilurin fungicides. The highest expression of the TRI5 gene was observed at the infection front. Gardiner et al. (2009) reported that TRI5 gene is strongly expressed in the rachis tissue of wheat. In this study they used a F. graminearum strain constructed by fusing a green fluorescent protein (GFP) marker to the promoter of TRI5 gene. Zhang et al. (2009) examined the expression of the TRI5 gene between carbendazim-resistant and sensitive F. graminearum in shake culture and reported a significant exponential relationship between trichothecene production and TRI5 gene expression. More recently Lee et al. (2014) compared the expression of TRI cluster genes in DON vs. NIV producing F. graminearum strains in liquid cultures. No study has been done to compare the level of expression of TRI genes in different chemotypes of F. graminearum during wheat colonization.

Therefore, in this study we have compared the level of expression of nine TRI genes in 3-ADON, 15-ADON and NIVproducing F. graminearum strains in a time course study both in resistant and susceptible wheat cultivars. The objective of this study was to evaluate the chemotype specific gene expression patterns in trichothecene biosynthesis related genes in different chemotypes of F. graminearum during wheat infection and colonization.

## MATERIALS AND METHODS

### Greenhouse Experiment and RNA Isolation

Two wheat cultivars with different levels of resistance to Fusarium head blight (FHB) were used in this study. A spring wheat cultivar, Roblin, which is highly susceptible (S) to FHB, and a FHB moderately resistant (MR) cultivar, Carberry, with resistance originating from the Chinese cultivar Sumai3 were used in the study. The Chinese cultivar Sumai3 have both Type I and II FHB resistance (Bai and Shaner, 1994). To prepare inoculum, two F. graminearum strains from each chemotype were cultured on Spezieller Nährstoffarmer agar (SNA) medium (0.2 g glucose, 0.2 g sucrose, 1 g KH2PO4, 1 g KNO3, 0.25 g MgSO4.7H2O, 0.5 g KCl, 14 g technical agar in 1 L of distilled water). F. graminearum strains used in this study were consisted of; Q-06-11 (designated as: 3-ADON1, isolated from wheat in Canada), A6-06-01 (3-ADON2, isolated from wheat in Canada), PH1 (15-ADON1, isolated from wheat in the USA), M2-06-02 (15-ADON2, isolated from wheat in Canada), W52516 (NIV1, isolated from maize in China) and W56604 (NIV2, isolated from maize in China). To produce liquid inoculum, 1.5 L of carboxymethyl cellulose (CMC) liquid media (15 g CMC, 1 g NH4NO3, 1 g KH2PO<sup>4</sup> monobasic, 5 g MgSO4.7H2O, 1 g yeast extract in 1 L of distilled water) was prepared and four SNA media (Leslie and Summerell, 2006) plates from each strain were divided into sections and added into each flask. Seven days after incubation at 25◦C under fluorescent light, the number of conidia per milliliter was determined by using a haemocytometer. The final conidial concentration was adjusted to 50,000 conidia/mL using distilled water. Seeds of spring wheat cultivars; Carberry and Roblin were planted in 15-cm plastic pots and maintained at 22–24◦C in the greenhouse at the Department of Plant Science, University of Manitoba, Winnipeg, MB, Canada. Inoculations were conducted at 30–50% anthesis. A 10 µL of F. graminearum

suspension (50,000 conidia/mL) was injected between the palea and lemma of spikelets per each spike according to the protocol described by Cuthbert et al. (2006). Five biological replicates for each strain and time point were conducted following a complete randomized design. Four to five spikes were inoculated per plant. FHB disease severity (DS) ratings were taken at 2, 3, 4, 7, 10, and 14 days post-inoculation (dpi) using the FHB disease scale by Stack and McMullen (1995). FHB DS readings were taken from five inoculated spikes for each replicate. The inoculated spikes were sampled at 2, 3, 4, 7, 10, and 14 dpi and stored at −80◦C freezer until RNA isolation. The mock inoculations were made using distilled water in both Roblin and Carberry for all time points. The inoculated spikes from five replicates were pooled and ground into fine powder in liquid nitrogen using a mortar and pestle. Total RNA was isolated using TRIzol <sup>R</sup> reagent (Invitrogen Life Technologies, Carlsbad, CA, USA) according to manufacturer's instructions. Extracted RNA was quantified using the NanoDrop 3300 (Thermo Scientific Inc., Wilmington, DE, USA). The integrity of RNA was analyzed using 1% agarose gel electrophoresis. To remove any DNA contaminations, RNA was treated with TURBOTM DNaseI (Invitrogen Life Technologies, Carlsbad, CA, USA) before cDNA synthesis. The first strand of cDNA was synthesized from 2 µg total RNA as the template using SuperScriptTM III First-Strand Synthesis System for reverse transcription-polymerase chain reaction (Invitrogen Life Technologies, Carlsbad, CA, USA).

#### FDK, DON and NIV Analysis

Kernels were harvested from inoculated spikes from both cultivars Carberry and Roblin, at 14 dpi. The percentage of Fusarium damaged kernels (FDK) was estimated by taking a pooled sample of 10 g from all replicates. The same kernels used for FDK analysis were used for DON or NIV analysis. Wheat kernels of each strain were pooled, ground and analyzed by Veratox <sup>R</sup> DON 5/5 kit (product no: 8331, Neogen Corp., Lansing, MI, USA) for DON analysis. NIV analysis was done using GC-MS according to the protocol described by Tittlemier et al. (2013).

#### Quantitative Reverse Transcription PCR

A total of nine genes (TRI4, TRI5, TRI6, TRI3, TRI8, TRI101, TRI9, TRI12 and FPP) in the DON biosynthetic pathway were examined using quantitative reverse transcription PCR (qRT-PCR). The level of expression of each gene was analyzed using a set of gene specific primers as described by Lee et al. (2014). As a house-keeping gene, translation elongation factor 1 alpha (EF-1α) from F. graminearum was selected (Kim and Yun, 2011). qRT-PCR reactions were performed in a CFX96 TouchTM Real Time PCR Detection System (Bio-Rad, Hercules, CA, USA) according to the protocol described by Lee et al. (2014). The qRT-PCR reaction cycles were consisted of initial denaturation at 95◦C for 3 min, followed by 45 cycles at 95◦C for 10 s, 60◦C for 20 s, 72◦C for 20 s, and finally 95◦C for 10 s and 65◦C for 5 s. The qRT-PCR reaction mixture contained 10 µL of 2× iQ SYBR <sup>R</sup> Green Supermix consisted of SYBR <sup>R</sup> Green I dye, 50 U/ml iTaqTM DNA polymerase, 0.4 mM each dNTPs, 6 mM MgCl2, 40 mM Tris-HCl (pH 8.4), 100 mM KCl, 20 nM fluorescein and stabilizers (Bio-Rad, Hercules, CA, USA), 0.5 µL of each primer (10 pM), 1 µL of template cDNA (10 ng), and RNase free water to a final volume of 20 µL. Quantification values were analyzed using the Bio-Rad CFX Manager v1.6, and the threshold cycle (Ct) values were determined. In all reactions, a non-template control (NTC) was set up to avoid any DNA contaminations in the reaction mixtures. Each reaction sample was amplified three times and final Ct values were calculated as an average of three replicates. The relative transcript abundance of the target genes was determined by the Pfaffl method (Pfaffl et al., 2002). qRT-PCR primer amplification efficiency was determined using the Ct slope method. In this method, serial dilutions of the template were prepared and Ct values were determined. Then a standard curve was generated by plotting the Ct values against the log cDNA concentrations. The amplification efficiency (E) of each primer was determined using the formula E = 10−1/slope. The percent amplification efficiency was determined using the formula %E = (E-1) <sup>∗</sup> 100%. The percent amplification efficiency of all genes were ranged between 95.6 and 101.3%.

### Statistical Analysis

Analysis of variance (ANOVA) for FHB DS at seven and 14 dpi was performed using the PROC Mixed procedure of SAS software (SAS version 9.3, SAS Institute Inc., Cary, NC, USA). Cultivar, strain and cultivar\*strain were considered as fixed effects. The Bonferroni method was used to compare statistically significant differences in least squares (LS) means of all variables. The type 3 test of fixed effects was determined and those with p ≤ 0.05 were considered significant.

## RESULTS

Nine genes from the F. graminearum trichothecene biosynthesis pathway, TRI4, TRI5, TRI6, TRI3, TRI8, TRI101, TRI9, TRI12 and FPP along with the housekeeping gene EF-1α, were selected for gene expression analysis. Each selected gene was analyzed by quantitative reverse transcription PCR to examine the changes in transcript levels at different time intervals post-inoculation. Accumulation of TRI gene transcripts initiated as early as 2 dpi in most strains. Significant differences were observed for cultivar, strain, and dpi for all the analyzed genes. The qRT-PCR analysis showed that the FPP transcript accumulation initiated at 2 dpi, peaked at 3–4 dpi and rapidly decreased at 7 dpi in MR cultivar Carberry (**Figure 1A**). A similar transcript accumulation pattern was observed in the susceptible (S) cultivar Roblin, however, at 10 dpi there was a slight increase in transcript accumulation in 3-ADON1, 3-ADON2 and 15-ADON1 strains and then gradually decreased at 14 dpi (**Figure 1B**). In both MR cultivar and S cultivar, the abundance of FPP transcripts was higher in 3-ADON producing F. graminearum strains than 15-ADON and NIV producing strains at most time points.

The qRT-PCR analysis for TRI5 showed that, transcript accumulation initiated at 2 dpi and peaked at 4 dpi and rapidly decreased by day 7. In contrast to the expression pattern of

(F). All data were normalized to the EF-1α expression level. Values are means ± SE of three replicates.

FPP gene, TRI5 gene expression increased again after 7 dpi in both cultivars (**Figures 1C,D**). Similar to FPP and TRI5 genes, accumulation of TRI4 transcripts initiated at 2 dpi, peaked at 3–4 dpi and started decreasing after day 4 in both cultivars (**Figures 1E,F**). However, in cultivar Roblin (S), transcript accumulation again peaked at 10 dpi in 3-ADON producing strains. The accumulation of TRI6 transcripts also initiated at 2 dpi, peaked at 4 dpi and gradually decreased after day 4 in cultivar Carberry (MR) (**Figure 2A**). In cultivar Roblin (S), transcript accumulation initiated at 2 dpi, peaked at 3 dpi and start decreasing after day 3 (**Figure 2B**). In 3-ADON producing strains the level of gene expression again peaked at 10 dpi. For TRI8 gene, transcript accumulation was initiated at 2 dpi and peaked at 4 dpi in all strains in cultivar Carberry (MR**)** (**Figure 2C**). In cultivar Roblin (S), transcript accumulation peaked at 3 dpi in 3-ADON and 15-ADON strains whereas for NIV strains it was at 4 dpi (**Figure 2D**).

The expression of the TRI101 gene initiated at 2 dpi and remained relatively constant for 15-ADON1, 15-ADON2, NIV1 and NIV2 strains during the early time intervals at 2–7 dpi (no distinct peaks were observed) and started decreasing at 10 dpi in MR cultivar, Carberry (**Figure 2E**). In 3-ADON1, peaks were observed at 4 and 10 dpi. A similar pattern was observed in S cultivar Roblin, however, a distinct peak was observed for 3-ADON2 at 3 dpi (**Figure 2F**). Similar to other genes, the accumulation of TRI3 transcripts initiated at 2 dpi and peaked at 3–4 dpi in most strains in both cultivars; however, the transcript abundance started decreasing at 4 dpi (**Figures 3A,B**). Despite the earlier induction (2 dpi), the level of expression of TRI9 gene peaked at 7–10 dpi (in cultivar Carberry) and 10 dpi (in cultivar Roblin) and gradually decreased in both cultivars starting at 10 dpi (**Figures 3C,D**). Transcript accumulation of the TRI12 gene also initiated at 2 dpi, peaked at 4 dpi and gradually decreased at 7 dpi in most of the strains (**Figures 3E,F**). Based on the qRT-PCR data, F. graminearum 3-ADON strains showed a higher level of TRI gene expression compared to the other strains for genes FPP, TRI3, TRI4, TRI6, TRI8, TRI12, and TRI101, at most time points except for TRI5 gene in MR cultivar and TRI9 gene. In TRI5 gene, 3-ADON1 and 15-ADON2 strains showed higher levels of expression than other F. graminearum strains in MR cultivar Carberry. The level of gene expression in 15- ADON and NIV producing strains showed no specific pattern of higher or lower expression. In some genes and time points the level of transcript accumulation was higher in 15-ADON strains and lower in NIV strains and vice versa. Among the analyzed genes, the highest abundance of transcripts was observed for TRI4 and TRI12 genes for all the examined strains (**Figures 1E,F** and **3E,F**). Our data showed that relative expression of TRI genes was significantly higher in wheat cultivar Carberry (MR) compared with Roblin (S).

The FHB DS was analyzed at 7 and 14 dpi, terminal FDK and DON/NIV content were analyzed at 14 dpi. When FHB DS was considered, there were significant differences between the cultivars and among the strains. The two-way interaction cultivar<sup>∗</sup> strain was significantly different (**Table 1**). The highest FHB DS was shown by cultivar Roblin inoculated by 3-ADON strains followed by 15-ADON and NIV strains. FHB DS caused by 3-ADON strains was significantly different from the 15- ADON producing strains and NIV strains. A similar trend was observed in the MR cultivar Carberry, however, the FHB symptom development was slower than in cultivar Roblin which is highly susceptible to FHB (**Figures 4A,B**). The percentage of FDK was higher in cultivar Roblin than in cultivar Carberry (**Figures 5A,B**). Similarly, a higher total DON content was observed in cultivar Roblin inoculated with 3-ADON strains than the 15-ADON strains (**Figures 5A,B**). Cultivars inoculated with 3-ADON strains showed higher levels of FDK and DON content than 15-ADON strains. NIV producing strains showed the lowest FDK percentage and toxin contamination.

## DISCUSSION

The objective of this study was to identify the potential chemotype-specific gene expression patterns of the TRI genes during wheat- F. graminearum infection and colonization. The expression of most TRI genes required for trichothecene production in F. graminearum were strongly induced at early time points after infection (i.e., 2–4 dpi) and the expression levels gradually decreased at 7 dpi. Also 3-ADON producing strains showed a comparatively higher level of gene expression than 15- ADON and NIV producing strains, confirming their ability to produce higher amounts of toxin in infected wheat kernels.

Deoxynivalenol biosynthesis related gene expression profiling indicated that, the expression of most TRI genes were initiated at 2 dpi. This shows that a high level of TRI transcript accumulation is essential for initiating the biosynthetic pathway of DON or NIV during wheat infection and colonization. The early expression patterns of five TRI genes along with FPP gene (TRI4, TRI5, TRI6, TRI8 and TRI3) strongly suggested that TRI6 gene which encodes a transcriptional regulator, positively regulates the expression of other TRI genes in the DON biosynthesis pathway. Similar observations have reported by Lee et al. (2014) in liquid culture media. The level of TRI gene expression was significantly different among the three chemotypes analyzed. The level of expression of most of the examined genes was higher in 3-ADON producing strains in both cultivars compared to the 15-ADON producing strains and NIV producing strains. It has been reported that 3-ADON strains produce more trichothecenes than 15-ADON and NIV strains (Ward et al., 2008). Therefore, the higher levels of expression of trichothecene biosynthesis related genes in 3-ADON producing strains during colonization may mediate the production of high amounts of toxins. According to the total DON content at 14 dpi, kernels infected with 3-ADON strains showed higher total DON content than the 15-ADON strains in both cultivars. Also, in this study the level of transcript accumulation of TRI4 gene was comparatively higher than other genes (except for TRI12 gene). The TRI4 gene regulates multiple steps (four steps) in the trichothecene biosynthesis pathway (McCormick et al., 2006). The accumulation of TRI4 transcripts in higher amounts could be explained by the involvement of this gene in multiple steps during trichothecene production. The level of expression of the TRI12 gene initiated at 2 dpi, peaked at 4 dpi and gradually

decreased at 7 dpi, which was similar to the other analyzed TRI genes in the present study. It has been reported that TRI12 gene encodes for trichothecenes efflux pump, which gives selfprotection for the fungus from the produced trichothecenes (Alexander et al., 1999). Therefore, coherent gene expression patterns of TR112 genes with other analyzed TRI genes further supports the role of TRI12 gene as a self-protector against the produced trichothecenes.

TABLE 1 | Analysis of variance (ANOVA) table for cultivar, strain and their interaction for Fusarium head blight disease severity at 7 and 14 days post-inoculation.


The level of expression of TRI genes was significantly higher in the MR cultivar Carberry than in the S cultivar Roblin. Similar results have been reported in other studies (Boddu et al., 2007; Brown et al., 2011; Hallen-Adams et al., 2011). Still there is no clear reason to explain the higher levels of expression in trichothecene biosynthesis genes in MR cultivar compared to the S cultivar. However, when we analyzed the total DON content at 14 dpi it was higher in the susceptible cultivar Roblin than in the MR cultivar Carberry. DON is a virulence factor in wheat, necessary for the spread of the fungus beyond the point of infection (Proctor et al., 1995; Bai et al., 2002; Audenaert et al., 2013). The above contrary observations between the trichothecene biosynthesis gene expression and terminal DON content in MR and S cultivars can be explained as follows: F. graminearum enters the plant either through inoculation or natural infection; in the first stage, it grows biotrophically at the point of infection and start producing DON (Audenaert et al., 2013). Then the fungus becomes more aggressive and attempts to grow into adjacent spikelets. However, the resistance mechanisms in R or MR cultivars prevent the fungus invasion from the point of infection. In order to overcome the resistance and spread further from the point of infection, fungus increases DON production as DON acts as a virulence factor in wheat.

FIGURE 4 | Mean Fusarium head blight disease severity in (A) moderately resistant cultivar Carberry and (B) susceptible cultivar Roblin after inoculating with different chemotypes of Fusarium graminearum at 7 and 14 days post-inoculation. Means with the same letters for Fusarium head blight disease severity are not significantly different.

FIGURE 5 | Fusarium damaged kernel (FDK) percentage and total terminal deoxynivalenol (DON) or nivalenol (NIV) content in (A) moderately resistant cultivar Carberry and (B) susceptible cultivar Roblin after inoculating with different chemotypes of Fusarium graminearum at 14 days post-inoculation.

Finally, to increase the DON production, the fungus increases the level of expression of DON biosynthesis related genes in R or MR cultivars compared to S cultivars. Investigations are in progress to further understand the reasons for higher levels of TRI gene expression in MR cultivars than in S cultivars.

Although the level of expression of DON biosynthetic genes were higher in the MR cultivar than in the S cultivar the final DON content is higher in the S cultivar. It has been reported that during Fusarium infection there is a broad expression of genes related to the DON detoxification process (Muhovski et al., 2012). This may explain the low levels of DON contamination in MR cultivar compared to S cultivar. Gene expression studies have shown that the expression of DON detoxification transcripts such as UDP-glycosyltransferase family (UGTs), CYP450s, ABC transporters and multidrug resistance-associated protein (MRP) were more highly abundant in FHB resistant cultivars than in susceptible cultivars during Fusarium infection (Muhovski et al., 2012; Al-Taweel et al., 2014; Kosaka et al., 2015). Therefore, it can be hypothesized that, although the level of TRI gene expression is higher in the MR cultivar, the resistance mechanisms within the cultivar can more efficiently detoxify the produced DON than the susceptible cultivar.

This study provides evidence on the chemotype specific gene expression patterns in the DON biosynthesis pathway during wheat infection and colonization. The results from this study indicated that 3-ADON producing strains showed higher levels of gene expression compared to 15-ADON and NIV producing strains. However, use of only two strains

#### REFERENCES


representing a chemotype may not be sufficient to draw definitive conclusions. Therefore, this study suggests the use of more strains from each chemotype group to gain a more comprehensive understanding of chemotype specific gene expression patterns during F. graminearum infection and colonization.

#### AUTHOR CONTRIBUTIONS

CA performed the research, coordinated the experimental part of the project and wrote the manuscript. WF supervised the research project and critically reviewed the manuscript.

## FUNDING

This work was supported by Western Grains Research Foundation (WGRF), National Wheat Improvement Program and Agri-Food Research and Development Initiative Manitoba (ARDI).

### ACKNOWLEDGMENTS

We acknowledge Canadian Wheat Board and University of Manitoba Graduate Fellowship for the financial support provided to CA.



of relative expression results in real-time PCR. Nucleic Acids Res. 30:e36. doi: 10.1093/nar/30.9.e36


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2016 Amarasinghe and Fernando. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Comparison of *Fusarium graminearum* Transcriptomes on Living or Dead Wheat Differentiates Substrate-Responsive and Defense-Responsive Genes

Stefan Boedi <sup>1</sup> , Harald Berger 1, 2, Christian Sieber <sup>3</sup> , Martin Münsterkötter <sup>4</sup> , Imer Maloku<sup>5</sup> , Benedikt Warth5 †, Michael Sulyok <sup>5</sup> , Marc Lemmens <sup>5</sup> , Rainer Schuhmacher <sup>5</sup> , Ulrich Güldener <sup>6</sup> and Joseph Strauss 1, 2 \*

#### *Edited by:*

Daniela Gwiazdowska, Poznan University of Economics and ´ Business, Poland

#### *Reviewed by:*

Jose M. Diaz-Minguez, University of Salamanca, Spain Frank Ebel, Ludwig Maximilian University of Munich, Germany

> *\*Correspondence:* Joseph Strauss joseph.strauss@boku.ac.at

#### *†Present Address:*

Benedikt Warth, Institute of Food Chemistry and Toxicology, University of Vienna, Vienna, Austria

#### *Specialty section:*

This article was submitted to Fungi and Their Interactions, a section of the journal Frontiers in Microbiology

> *Received:* 08 April 2016 *Accepted:* 04 July 2016 *Published:* 26 July 2016

#### *Citation:*

Boedi S, Berger H, Sieber C, Münsterkötter M, Maloku I, Warth B, Sulyok M, Lemmens M, Schuhmacher R, Güldener U and Strauss J (2016) Comparison of Fusarium graminearum Transcriptomes on Living or Dead Wheat Differentiates Substrate-Responsive and Defense-Responsive Genes. Front. Microbiol. 7:1113. doi: 10.3389/fmicb.2016.01113 <sup>1</sup> Fungal Genetics and Genomics Unit, Division of Microbial Genetics and Pathogen Interactions, Department of Applied Genetics and Cell Biology, BOKU University, University and Research Centre Tulln, Tulln, Austria, <sup>2</sup> Bioresources, Austrian Institute of Technology GmbH, Tulln, Austria, <sup>3</sup> Department of Earth and Planetary Sciences, University of California, Berkeley, Berkeley, CA, USA, <sup>4</sup> Helmholtz Zentrum München, Deutsches Forschungszentrum für Gesundheit und Umwelt, Neuherberg, Germany, <sup>5</sup> Department for Agrobiotechnology (IFA-Tulln), BOKU University, Tulln, Austria, <sup>6</sup> Department of Genome-oriented Bioinformatics, Wissenschaftszentrum Weihenstephan, Technische Universität München, München, Germany

Fusarium graminearum is an opportunistic pathogen of cereals where it causes severe yield losses and concomitant mycotoxin contamination of the grains. The pathogen has mixed biotrophic and necrotrophic (saprophytic) growth phases during infection and the regulatory networks associated with these phases have so far always been analyzed together. In this study we compared the transcriptomes of fungal cells infecting a living, actively defending plant representing the mixed live style (pathogenic growth on living flowering wheat heads) to the response of the fungus infecting identical, but dead plant tissues (cold-killed flowering wheat heads) representing strictly saprophytic conditions. We found that the living plant actively suppressed fungal growth and promoted much higher toxin production in comparison to the identical plant tissue without metabolism suggesting that molecules signaling secondary metabolite induction are not pre-existing or not stable in the plant in sufficient amounts before infection. Differential gene expression analysis was used to define gene sets responding to the active or the passive plant as main impact factor and driver for gene expression. We correlated our results to the published F. graminearum transcriptomes, proteomes, and secretomes and found that only a limited number of in planta- expressed genes require the living plant for induction but the majority uses simply the plant tissue as signal. Many secondary metabolite (SM) gene clusters show a heterogeneous expression pattern within the cluster indicating that different genetic or epigenetic signals govern the expression of individual genes within a physically linked cluster. Our bioinformatic approach also identified fungal genes which were actively repressed by signals derived from the active plant and may thus represent direct targets of the plant defense against the invading pathogen.

Keywords: *Fusarium*, secondary metabolism, pathogenicity factors, defense genes, active plant, passive plant

## INTRODUCTION

Fusarium graminearum (telemorph: Giberella zeae) is a plant pathogenic ascomycete fungus causing various plant diseases on small-grain cereals such as Fusarium head blight (FHB or scrab) of wheat (Triticium aestivum) and barley (Hordeum vulgare) as well as ear and stalk rot of maize (Zea mays; McMullen et al., 1997; Bottalico and Perrone, 2002; Stack, 2003; Goswami and Kistler, 2004). Infection with F. graminearum leads to yield losses and reduction in grain quality due to shriveled, discolored kernels, or total failure of kernel development as well as to contamination with different mycotoxins, especially deoxynivalenol (DON) and zearalenone (ZEA; Trail, 2009; McMullen et al., 2012). The trichothecene mycotoxin DON is a potent inhibitor of protein biosynthesis in eukaryotes (McLaughlin et al., 1977) and causes intestinal irritation, poor weight gain and feed refusal in livestock (Eriksen and Pettersson, 2004) and may bear immunological and teratogenic effects toward human (Desjardins, 2006), whereas ZEA has estrogenic effects in human and animals (Kim et al., 2005; Gaffoor and Trail, 2006). FHB is a worldwide disease occurring among others in the United States, Canada, South America, Europe and China, whereby enormous economic losses and health threats are reported (Nganje et al., 2002).

Wheat is highly susceptible to FHB infection during the time period of anthesis (Pugh et al., 1933; Booth, 1971; Booth and Taylor, 1976; Sutton, 1982; Windels and Kommedahl, 1984; Khonga and Sutton, 1988; Trail, 2009), where temperatures ranging from 15 up to 29◦C and high humidity represent favorable environmental conditions (Tschanz et al., 1976; Dufault et al., 2006; McMullen et al., 2012). After spore germination postulated possible entering sites for the extending fungal hyphae include wounds or natural openings like stomata, inner surfaces of palea and lemma near the floret mouth as well as crevices between the palea and the lemma (Bushnell, 2001; Lewandowski et al., 2006). The infection process is accompanied by the formation of infection cushions (Boenisch and Schäfer, 2011), an agglomeration of fungal hyphae which secrete various hydrolyzing enzymes able to degrade components of the epidermal plant cuticle and the plant cell wall, such as e.g., cutinases, pectinases, hemicellulases, cellulases, and lipases (Kang and Buchenauer, 2000; Bushnell et al., 2003; Voigt et al., 2005; Cuomo et al., 2007; Walter et al., 2010). After this initial stage of surface colonization (between the time span of roughly 20 and 70 h after infection, abbreviated hai), asymptotic intercellular fungal growth occurs, which resembles the lifestyle of biotrophic fungi. Subsequently, and from around 70 hai onwards, plant tissue necrosis occurs which is triggered by mycotoxins and intracellular growth of the pathogen (Kang and Buchenauer, 1999; Bushnell et al., 2003; Jansen et al., 2005; Boddu et al., 2006; Boenisch and Schäfer, 2011). G protein- coupled receptors have been reported to be involved in host recognition followed by downstream signaling cascades involving the mitogen-activated protein kinases (MAPK) FgGPMK1 (Jenczmionka et al., 2003; Jenczmionka and Schafer, 2005) and FgMGV1 (Hou et al., 2002) as well as the heterotrimeric G protein subunits Gα (GzGPA1, 2, and 3), Gβ (GzGPB1), Gγ (GzGPG1), and the Ras-GTPase RAS2 (Bluhm et al., 2007; Yu et al., 2008; Walter et al., 2010). Finally, the classical FHB symptom of head bleaching occurs from chlorosis of the whole head. DON is necessary for this process because it allows the invading fungus to spread through the rachis from the infected to the adjacent spikelet. Through its function as translation inhibitor, DON suppresses the establishment of cell wall thickenings in the rachis node and thus inhibits this important defense response of the host (Cutler, 1988; Jansen et al., 2005; Boenisch and Schäfer, 2011). As DON induction plays such a critical role in pathogenesis its genetic regulation has been extensively studied (reviewed in Kimura et al., 2007; Walter et al., 2010). In planta DON production is believed to be triggered by a number of signals including specific metabolites which may pre-exist in the healthy plant or are induced in response to the pathogenic attack (such as H2O2). Some of these conditions can be mimicked in vitro in axenic liquid shake cultures and in fact induction of DON (and it's acetylated derivatives 15ADON and 3ADON) was found in the presence of compounds which are part of the polyamine and urea cycle pathways in plants (ornithine, arginine, agmatine, putrescine; Gardiner et al., 2009b) and/or at low pH (Gardiner et al., 2009c; Merhej et al., 2010). Polyamines and intermediates of amino acid metabolism (e.g., biogenic amines) are known to accumulate in plants attacked by pathogens (Walters, 2000, 2003) and thus it was hypothesized that F. graminearum might use this natural plant response to trigger DON induction (Gardiner et al., 2009b).

F. graminearum is among the most intensively studied fungal pathogens (Goswami and Kistler, 2004). Sequencing and annotation of its genome (Cuomo et al., 2007) as well as the development of an Affymetrix GeneChip (Güldener et al., 2006) laid the foundations of a variety of studies exploring the transcriptome of this facultative pathogen under a variety of in vitro growth conditions (different nutrient sources) and during different stages of infection on wheat and barley (Sieber et al., 2014).

Genome-wide expression profiles were investigated during the early developmental stages after spore germination in culture (Seong et al., 2008); under DON-inducing and non- inducing conditions in culture (Gardiner et al., 2009a); during in vitro growth on complete media, under nitrogen starvation and under carbon starvation. These conditions were compared with pathogenic growth during barley infection (Güldener et al., 2006) and, later on, to the transcriptome during wheat infection (Lysoe et al., 2011b). In additional surveys the transcriptome of F. graminearum was examined during early wheat infection (Erayman et al., 2015), during distinct infection phases of crown rot disease in wheat (Stephens et al., 2008), on dry wheat stems at different stages of colonization until perithecium formation (Guenther et al., 2009) as well as after laser capture microdissection of plant bulk material which allowed the transcriptomic examination of developmentally synchronized mycelia at distinct growth stages inside of wheat coleoptiles (Zhang et al., 2012). To test the contribution of DON production to pathogenicity-related gene expression the transcriptomes of TRI6 and TRI10 mutants unable to produce DON were analyzed in axenic cultures and during wheat infection (Seong et al., 2009). Developmental mutants carrying deletions in FgStuA were also studied in sporulation medium, on infected wheat heads as well as during secondary metabolites inducing culture conditions (Lysoe et al., 2011a). Finally, a deletion mutant in FGP1, a WOR-like protein was analyzed on wheat heads and under DON inducing culture conditions (Jonkers et al., 2012). All these transcriptome comparisons revealed that specific subsets of F. graminearum genes are exclusively expressed in planta and, based on these observations, were designated as "pathogenicityrelated" or "virulence" genes.

It has been recognized only recently from our work and others that virulence factors such as secondary metabolites or effector proteins are not only under genetic, but also under epigenetic control in various fungi, including F. graminearum (Gacek and Strauss, 2012; Reyes-Dominguez et al., 2012; Connolly et al., 2013; Wiemann et al., 2013; Chujo and Scott, 2014; Soyer et al., 2014) and thus this chromatin-based regulatory mechanisms might contribute to virulence. Because we want to further study the field of epigenetic regulation in the wheat-F. graminearum interactions by using chromatin-modification mutants, we first sought to better understand the transcriptional response of the fungal wild-type cells to the substrate it encounters during pathogenic attack. However, some of the fungal genes expressed only in planta may not be related to pathogenic processes but simply responding to the specific substrate (floral tissue) while others may be directly involved in the pathogenic process (overcoming resistance). In order to better define those fungal genes which are directly associated with pathogenicity by counteracting plant defense we compared in this study F. graminearum transcriptomes originating from Ph-1 wild type cultures growing on living flowering wheat heads (pathogenic conditions) or on identical, but cold-killed material (saprophytic conditions). For this comparison we inoculated ears of flowering wheat heads but one half of them had been inactivated prior to infection by cutting them off the plant and dipping them into liquid nitrogen. Both samples were further incubated under the same conditions thus presenting to the fungal cells basically identical substrates but in one case under pathogenic conditions on the living "active" plant and in the other case under saprophytic conditions on the cold-killed "passive" plant material. In this paper we show that only background levels of secondary metabolites are formed on the dead plant material and that this approach is able to differentiate pathogenicity related genes from plant matrix genes and also found genes repressed by the active plant potentially revealing novel fungal targets of active plant defense.

#### RESULTS AND DISCUSSION

#### Experimental Set Up

**Figure 1** provides a schematic overview of the experimental workflow. Three independent ears were inoculated on living plants of the susceptible wheat cultivar Remus representing the interaction of the fungus with the living host. This condition is subsequently referred to as "pathogenic growth." Another set of three ears, which were cut off the plant and shockfrozen in liquid nitrogen prior to spore application, was identically inoculated representing the same plant substrate but without active metabolism and defense responses. We called this condition subsequently "saprophytic growth." The inoculated living wheat heads were cultivated under standard conditions and the inactive heads were incubated at the same location under the same conditions (see Section Materials and Methods for details). Living and dead plant material was harvested 3 and 5 days after inoculation (dai) and further analyzed for mycotoxin levels and fungal transcriptomes. In addition to these plant-based experiments we performed a standard axenic culture experiment (subsequently termed "axenic growth") using the resting cell method (liquid minimal medium with L-ornithine as nitrogen source, growth for 3 days without shaking; see Section Materials and Methods for details).

### Active Plant Metabolism Restricts Fungal Growth

We first determined to which extent the fungal cells are able to proliferate on and in the wheat heads depending on whether there is active metabolism or not. **Figure 2** shows example pictures of F. graminearum Ph-1 inoculated Remus wheat heads and as seen in panel A of **Figure 2** a typical living wheat head 3 dai of the florets already presents clearly visible symptoms of infection (brownish lesions around the inoculation sites). Contrary to the living plant, the cold-killed wheat heads inoculated exactly the same way and incubated in the same location next to the living plant were totally overgrown by fungal mycelia. **Figure 2B** portrays some typical cold-killed wheat heads with extensively growing fungal mycelium around the plant material (right head). For further analysis of these saprophytically growing fungal cells we removed the portion of fluffy aerial mycelium (partial removal shown in the middle wheat head in **Figure 2B**) to ensure analysis of mainly plant tissue-associated fungal material (left wheat head of **Figure 2B**). We used qPCR to determine the relative proportion of fungal chromosomal DNA in total DNA extracted from the living or dead wheat heads using previously described methods (Brunner et al., 2009). According to this DNA-based analysis the average infection rates were 13.2% for the pathogenic and 99.6% for the saprophytic samples harvested 3 dai. After the longer incubation time (5 dai) 28.2% of the total DNA was of fungal origin in the pathogenic samples but in the saprophytic samples exclusively fungal and hardly any plant qPCR products were detected resulting in infection rates of around 100.0% (**Figure 2C**). This indicates that basically all plant DNA was degraded by nucleases or otherwise metabolized.

To assess, to which extent the ratio of fungal DNA to wheat DNA mirrors the activity of the fungal cells we also quantified the transcript levels of a constitutively transcribed fungal housekeeping gene relative to the transcript levels of an equivalent plant gene. Quantification of fungal and wheat GAPDH cDNA levels by qPCR (**Figure 2C**) gave similar results as the DNA quantifications 3 dai, i.e., 12.7% fungal GAPDH transcripts for the pathogenic and 100.0% for the saprophytic samples. Interestingly, after longer incubation (5 dai) of the pathogenic samples the relative fungal GAPDH mRNA was significantly higher (52.3%) than the equivalent DNA share

based on qPCR quantification of the proportion of fungal chromosomal DNA (chrDNA) within the fungus/wheat mixture (Brunner et al., 2009). Additionally to the plant experiment axenic media cultivation in presence of the DON inducing nitrogen source L-ornithine (Gardiner et al., 2009b) was carried out. All samples were subjected to chemical and molecular biological analysis. Quantitative secondary metabolite analysis was performed by LC/ESI-MS/MS against a set of analytical mycotoxin standards (Sulyok et al., 2006; Vishwanath et al., 2009). Sample processing was performed as detailed described in the text.

(28.2%) indicating that- despite of their growth restriction by the active plant tissue- the fungal cells were highly active.

## Low Mycotoxin Titers during Saprophytic Growth on Cold-Killed Wheat Heads (Passive Plant Tissue)

When the production of secondary metabolites was analyzed in the different samples, we found that DON (and its acetylated 15ADON and 3ADON as well as glycosylated DON-3-glucoside derivatives), butenolide and culmorin only accumulated in the living plant tissue but mycotoxin levels remained very low in the dead plant samples (**Figure 3A**). This is remarkable for two reasons. First, because the basic substrate for the fungal cells is identical in the living and the dead plant and second, there is much more fungal biomass accumulating during saprophytic compared to pathogenic growth. These results indicate that whatever the in vivo signals for mycotoxin induction are, they are not formed or unstable in the non-challenged plant cells. Obviously, for high induction of these SM biosynthetic genes the fungal invader needs host metabolites that are actively formed, most likely associated with the defense reaction, by the plant. Transcriptional analysis of the key biosynthetic genes and the transcription factors involved in the formation of the tested metabolites showed that the elevated mycotoxin levels detected during pathogenic growth on the living plant (**Figure 3A**) were consistent with enhanced transcription levels of these genes (**Figure 3B**). This demonstrates that the elevated mycotoxin levels are derived from a genetic induction event and can not only be based on biochemical changes, e.g., precursor availabilities. According to recent findings (Gardiner et al., 2010), possible candidates of such inducing plant metabolites are secondary amines (agmatine or putrescine) or intermediates of amino acid metabolism such as ornithine. It is remarkable that these metabolites are not present in the plant in sufficiently high amounts before the infection, but are produced or stabilized

arrows in (A) point toward brownish lesions visible on pathogenic samples. The red rectangle in (B) indicates the type of saprophytic material used for analysis. To avoid that the majority of the fungal material had no direct contact to the wheat tissue, extensive aerial hyphae were stripped off the wheat head and only the intimately connected fungal cells were used for further analysis. (C) Infection rates were analyzed according to the published method (Brunner et al., 2009) by DNA-based and cDNA (RNA)-based quantitative PCR. The proportion of fungal chromosomal DNA (chrDNA) was determined within the total fungal/plant DNA mixture and the proportion of fungal mRNA (GAPDH cDNA) was determined within the total fungal/plant cDNA mixture. In the saprophytic samples basically no plant-derived DNA or mRNA was detectable any more already 3 dai. Patho, pathogenic growth on living wheat heads; sapro, saprophytic growth on cold-killed wheat heads.

during the infection and are "converted" by the pathogen to a signal detrimental for the attacked host.

From our experimental results we can safely conclude that the concentrations of these amino acids in the dead plant material are not sufficiently high for induction and that the pathway generating ornithine or agmatine in wheat must be turned on by the fungal infection. Genes and metabolites of this pathway in fact have been identified in the combined transcriptomemetabolome studies by Nussbaumer et al. (2015). The authors found that 50 h after infection many genes involved in the polyamine/urea cycle are up-regulated in the infected wheat heads in comparison to the water-treated controls. The metabolic measurements revealed increased metabolite levels of ornithine and putrescine 96 h post infection. Our transcriptome data of the fungal cells suggest that the invading fungus reacts to these metabolic changes. We found many genes of the urea cycle and especially genes involved in the generation of ornithine and citrulline to be up-regulated (see **Figure 4** and **Table 1**). Interestingly, citrulline was not a strong DON inducer in the study by Gardiner et al. (2009b) suggesting that the upregulation of this branch may only be necessary to provide sufficient substrate for the catabolic pathways starting from ornithine and arginine and generating the inducers ornithine, putrescine, and agmatine. Which of these mentioned metabolites act directly as inducers for the diverse SM pathways is not precisely known but based on this pathway analysis it could be that the catabolic metabolites agmatine and putrescine are the true inducers. Gene inactivation studies with these candidates would clarify this point.

It is noteworthy that in the metabolomics analyses of the plant (Nussbaumer et al., 2015) ornithine, agamatine, and putrescine production was also induced by simply injecting DON into the flowering wheat heads. This demonstrates that the toxin itself triggers accumulation of these amino acids and catabolic products. For the Fusarium interaction this could mean that a positive feedback loop operates in which small amounts of DON produced by the initial pathogen infection induces agamatine/putrescine production which again accelerates the induction process for DON.

We also wanted to compare our in planta transcriptomes with axenic conditions and high DON production and consequently used 5 mM L-ornithine as sole nitrogen source. The axenic cultures were also harvested 3 dai from submerged batch cultivation. As expected, we found in these cultures high levels of DON (**Figure 3C**), its acetylated derivatives and culmorin (Gardiner et al., 2009b). In contrast, cultures grown on nitrate as sole nitrogen source (23.5 mM nitrogen) formed only very low levels of these metabolites (**Figure 3C**). L-ornithine, but not nitrate, was fully consumed during fungal biomass accumulation in these synthetic media (see Table S1) and additionally we observed a pH drop from the initial at pH 6.5 buffered media to pH 2.9 on L-ornithine whereas pH slightly rose to 6.8 on nitrate. Our data confirm published results that in axenic cultures the metabolism of L-ornithine and low pH are the responsible genetic triggers for DON induction (Gardiner et al., 2009c; Merhej et al., 2010, 2011).

#### Analysis of Specific Gene Sets Responding to the Different Growth Conditions

Total RNA was isolated from axenic as well as pathogenic and saprophytic samples harvested 3 dai and all samples were independently subjected to RNA-seq analysis. Transcript levels of genes are expressed as "FPKM" values (Fragments Per Kilobase of exon per Million reads mapped) which are normalized for both sequencing depth and gene length. **Figure 5A** shows the number of genes in each interval of

log2(FPKM) values for each condition. From this plot it becomes evident that under pathogenic and axenic conditions around 3500 genes of the F. graminearum genome are not or only rarely expressed (log<sup>2</sup> < 1). Interestingly, around 1000 additional genes are expressed under saprophytic conditions. Considering the distribution of expression values over the genome, we saw that the majority of genes are expressed between ≥1 log2(FPKM) ≤ 5. This range covers around 9000 genes or around 2/3 of the transcribed F. graminearum genome. Due to these findings we considered from all predicted 13.826 genes the 65% most strongly expressed ones in each condition for further analysis. Using this cut off, all considered genes showed an expression level of at least log2(FPKM) ≥ 1.92 under axenic, at least log2(FPKM) ≥ 1.99 under pathogenic and at least log2(FPKM) ≥ 2.46 under saprophytic growth conditions. Relative transcript abundances in these RNA-seq data very well

FIGURE 4 | Selection of arginine biosynthesis as well as arginine and proline metabolism reactions from KEGG database whose underlying genes showed at least a two-fold up-regulation during pathogenic in comparison to saprophytic growth. Genes which underlie corresponding reaction numbers are shown in Table 1.



E*-*values of appropriate BLAST hits are listed in the table and positions within the pathway environment of arginine metabolism are depicted schematically in *Figure 4* via assigned reaction numbers.

matched the RT-qPCR results obtained for the selected set of genes shown in **Figure 3B**, which were analyzed in the identical RNA extracts of all replicates and experimental conditions. The direct comparison of expression profiles of these selected five genes between RNA-seq and RT-qPCR using β-tubulin transcription for normalization of the PCR analysis is shown in Figure S1.

## Only 17% of the Transcribed Genome Flexibly Responds to Conditions

Using the settings described above we found that, from all 8987 significantly expressed genes, 7467 genes were transcribed in all three samples representing 83% of the transcribed genome (**Figure 5B**). This means that the vast majority of genes was always transcribed, irrespectively of the greatly

contrasting metabolic condition and their functions and, thus is likely associated with basic cellular processes. Overall, in our set-up, around 17% or 1528 genes were specific in their transcriptional profile for one or two of the three conditions.

The Venn diagram in **Figure 5B** shows that 452 genes are unique for the axenic culture conditions where growth occurs in liquid minimal medium containing sucrose as carbon and L-ornithine as nitrogen source. Consultation of the FunCat Database (https://www.helmholtz-muenchen.de/en/ ibis; Ruepp et al., 2004) showed significant enrichment of functional categories such as metabolism of amino acids, and more broadly, regulation of nitrogen assimilation (see register tab "Axenic\_only\_452" of Table S2). This exclusivity in gene expression is most likely due to the relatively high amount of L-ornithine providing the nitrogen source in the growth medium. The second highly significant category is cellular detoxification with additional functions predicted in glutathione, glutaredoxin, and thioredoxin metabolism. This is probably due to the high mycotoxin titers found in these cells (compare to **Figure 3C**), but still somewhat surprising as the genes would be expected to be shared with the pathogenic conditions rather than appearing in the axenic culture category. However, the relative enrichment of genes in this gene set compared to the presence of this category in the whole genome (see register tab "Axenic\_only\_452" of Table S2) indicates that different gene sets are active in detoxification whether the fungus grows pathogenically or in axenic cultures.

Only few expressed genes are functionally annotated in the gene set representing the overlap between the two "non-pathogenic" conditions, i.e., axenic and saprophytic (251 unclassified proteins within the 334 genes present in this set, see register tab "Sapro\_Axenic\_334" of Table S2). Nitrogen regulation and amino acid metabolism is not in this commonly expressed gene set indicating that the wide variety of different nitrogen sources expected to be available to the saprophytically growing fungus is stimulating a different gene expression network compared to the exclusive L-ornithine based diet in axenic conditions.

## Individual Genes within SM Clusters Are Induced by Different Signals

We used L-ornithine in the axenic cultures to exclude the known trichothecene biosynthesis genes from appearing in the gene set specific for the pathogenic growth condition. Consistent with this approach we find several genes belonging to the TRI cluster and coding for DON biosynthesis among the 734 gene overlap between axenic and pathogenic conditions. The same is true for genes involved in culmorin and, butenolide biosynthesis as well as for several other genes encoded within putative SM clusters involved in the biosynthesis of known as well as unknown products (**Figures 6A,B** and Figure S2). However, we find it intriguing that gene regulation within a given SM gene cluster does not appear to be homogenous. The TRI cluster expression profile under the three chosen conditions is given in detail in **Figure 6C**. As we found DON only under pathogenic and axenic conditions one could expect that all genes of the cluster are highly expressed in these conditions and not or weakly expressed in saprophytically grown cells. However, the thorough analysis of the cluster genes revealed a different picture. ORF-B (FGSG\_03530) coding for a putative acetylesterase of unknown specificity is basically not expressed under axenic [log2(FPKM) = 0.3] and weakly expressed under pathogenic conditions [log2(FPKM) = 1.3] but strong in saprophytic samples [log2(FPKM) = 2.7]. This is an interesting example of a gene residing within a physically-linked gene cluster but its co-regulation is condition-dependent. In the case of ORF-B the "synthetic" TRI cluster inducer L-ornithine is insufficient and the plant tissues—no matter if alive or dead must contain a compound to generate the induction signal for ORF-B.

This feature of non-homogenous gene expression is true also for many other SM gene clusters (see **Figure 6B**, Figure S2 and section below).

### The Saprophytic Gene Set Is Dominated by Plant Cell Wall-Degrading and Carbohydrate-Transporting Functions

Overall, there are 768 genes found to be exclusively expressed under saprophytic conditions and for 299 of these genes functions can be predicted. Not unexpectedly, around 100 of these genes exclusively expressed in the dead plant material seem to be involved in the degradation and uptake of the plant cell wall material (see register tab "Sapro\_only\_768" of Table S2). Functions such as enzymes for exogenous polysaccharide degradation, sugar metabolism and C-compound transport (sugar and amino acid transporters) are highly enriched functional categories in the saprophytic-specific gene set. These results suggest that a certain set of biomass degradation genes are not activated when the plant lives, i.e., in the pathogenic sample. The reason for this difference is not clear but either the specific inducer(s) for these genes are not present in the living plant or the plant defense actively prevents the transcription of these genes involved in cell wall degradation, probably by restricting the access to the substrate. However, it is surely relevant to consider for the interpretation of our saprophytic sample results that, inactivating the wheat head by dipping it into liquid nitrogen and subsequent thawing, the plant cuticles and cell walls might be partially damaged and thus easier to access by the invading fungus and its extracellular enzymes. This physical disruption might expose some substrates or liberate inducers which are not available to the fungus on the active plant.

There are also some virulence- and disease-related genes only appearing in the saprophytic sample and—similar to the cell walldegrading enzymes—it is surprising that some of these putative pathogenicity factors are not induced when the plant defense is on. The genes code mainly for predicted membrane proteins and secondary metabolite biosynthesis of so far unknown metabolites (register tab "Sapro\_only\_768" of Table S2).

## Genes Commonly Expressed in the Dead and the Living Plant Material Represent Typical "*In planta*" Genes

Consistent with our working hypothesis that some so-called "in planta" genes are not necessarily involved in overcoming the restriction posed by the defending host we found that a large portion of these genes were induced simply by the living and the dead plant tissue. The overlap between saprophytic and pathogenic conditions consists of 418 genes with many predicted functions associated with carbohydrate metabolism and transport (see register tab "Patho\_Sapro\_418" of Table S2).

The vast majority of genes assign to putative function in carbon compound metabolism and transport including sugar, glucoside, polyol, carboxylate, and polysaccharide metabolism. Genes coding for ion, sugar and amino acid transporters including di- and tripeptide transporters were also activated. Interestingly, the siderophore iron transport and biosynthesis system is active in dead and living plant samples although it could be hypothesized that in the saprophytic sample iron may be less limiting because the plant tissue is damaged and nutrients are exposed to the fungus. This finding indicates that the damage of the plant material by liquid nitrogen treatment was not too extensive and plant cells were not fully disrupted. On the other hand, also the pathogenically growing fungus destroys plant tissue and consistently, we found for example genes involved in vacuolar protein degradation such as alkaline proteases, aminopeptidases, and endopeptidases among the genes expressed in both pathogenic and saprophytic conditions.

Among the over-represented functional categories in this gene set were also several genes putatively involved in secondary metabolism. This is unexpected as **Figure 3A** shows that no

classical mycotoxins are produced under saprophytic conditions however, the SM genes identified in this gene set so far cannot be assigned to a specific metabolite produced by F. graminearum. More broadly, secondary metabolism does not seem to be exclusive to the pathogenic condition, but also occurs in the fungus growing saprophytically or in axenic cultures. **Figures 6A,B** and Figure S2 give an overview of the gene distribution above the 65% threshold of highest expression within all annotated experimentally verified and predicted SM biosynthesis gene clusters (Sieber et al., 2014). The transcriptional profile of these clusters shows that the genes involved in aurofusarin production are activated in all conditions as well as seven out of nine genes within the ferricrocin cluster (**Figure 6B**). Some of the predicted clusters producing putative unknown metabolites or intermediates show many of their genes expressed under all tested conditions whereas for some clusters this is not the case (Figure S2). Interestingly, all ten genes of cluster C16 and all eight genes of cluster C64 are expressed under pathogenic, none of them under saprophytic and only two genes in each cluster under axenic conditions. Both clusters may be responsible for biosynthesis of unknown SMs potentially involved in pathogenicity. Generally, we find it intriguing that not all genes within a given cluster respond to the same signal and axenic conditions mimic the signals necessary for the expression of a subset of cluster genes. The molecular basis of this difference might be associated with particular promoter sequences and be relevant to better understand the living plantspecific signals leading to the production of putative virulence factors.

## The Majority of Genes Expressed Exclusively under Pathogenic Conditions Are Not Associated with Substrate Degradation

Substrate degradation and transport activities are not enriched functional categories in pathogenic sample growing on the living plant (**Figure 5B**/see register tab "Patho\_only\_368" of Table S2). Instead we find strongly over-represented detoxification functions. In this category several multidrug resistance proteins, ABC transporters and other carriers appear in addition to enzymes putatively degrading plant defense molecules or fungal toxins produced during the infection process. This is a very clear indication that it is the active plant which triggers a fungal defense reaction and that there are basically no pre-formed or stable secondary plant or fungal metabolites which may lead to a permanent detoxification reaction in the unchallenged fungal cell.

Although we induced SM by L-ornithine in axenic cultures, there were still some specific SM genes exclusively expressed under pathogenic conditions. For example, six NRPSs with similarity to the AM-toxin forming enzymes are specific for the living plant and neither of them appear in any other sample. Thus, whereas PKS expression does not seem to be restricted to pathogenic growth, these indicated NRPS genes are. As they may code for the production of peptide toxins they might represent true virulence factors.

## 3-Way Comparison of Differential Gene Expression Defines the Impact of the Active-Plant for Gene Expression

To gain a clearer picture on the signals specifically originating from the living plant, we performed a 3-way differential gene expression analysis considering only genes which were at least four-fold differentially regulated between the conditions. This way we were able to define the main impact factors or "drivers" of gene expression in the active plant. According to these calculations which are graphically represented in the scheme in **Figure 7** we were able to define genes specifically induced or repressed by signals originating from the "active plant," the "passive plant" or the "DON-inducing" conditions of the axenic cultures. In contrast to the list of condition-specific genes discussed above which is based on simple expression values in the three conditions this 3-way comparison of differential gene expression identifies genes which are both up-regulated or downregulated in response to the most relevant impact a certain condition exerts on the analyzed gene. Based on this, "active plant" genes were defined as those at least four-fold higher (active plant induced) or lower (active plant repressed) expressed during pathogenic growth compared to the two other conditions. In case of these "active plant" (AP) regulated genes we assumed that they represent functions required for the pathogenic fungus to establish infections (up-regulated AP genes) or functions which may be targeted by the actively defending host and are thus down- regulated in the pathogenic fungal cells (down-regulated AP genes). In a similar reasoning we considered as "passive plant" (PP) genes those which are commonly up- or down regulated in both the pathogenic and the saprophytic samples—but differed at least four-fold from the axenic sample. Such genes are called "PP-genes" as they responded to the presence of plant tissues and metabolites, regardless if these are part of a living plant or not. This approach would also reveal only those SM genes in the AP gene sample whose induction truly requires a signal from the active plant and different to L-ornithine (present in axenic cultures). The genes differentially expressed in all three conditions fall into the overlap between AP, PP, or DI categories (see **Figure 7**) but as they do not respond to one specific signal, they will not be further discussed here.

Our main interest was to better define the gene set specific to pathogenic conditions. By taking out genes which respond to the plant tissue (PP genes) or to the SM-inducing conditions (DI genes) we were able to restrict the large number of genes expressed under pathogenic conditions to genes with "active-plant" profiles possibly revealing novel virulence factors. According to our calculations 306 F. graminearum genes were specifically responding during the infection process to the living, flowering wheat heads. From this at least four-fold regulated gene set 184 genes were induced, whereas 122 genes were repressed in the AP gene set (**Figure 8A**). As shown in **Figure 8B**, functional categories most significantly overrepresented in the up-regulated active-plant gene set are associated with secondary metabolism, lipid metabolism, cellular defense and detoxification.

## Only Certain Known SM Genes Are Exclusively Induced by Plant Signals during Pathogenesis

We summarized them in **Table 2** and found siderophore biosynthesis genes such as one gene from the ferricrocin cluster, the NRPS, and another gene from the malonichrome as well as the NRPS gene from the triacetylfusarinin cluster. Additionally one gene involved in carotenoid formation could be classified. Apart from the two NRPS-encoding genes involved

in malonichrome and triacetylfusarinin synthesis there are two other putative NRPS-encoding genes (belonging to cluster C37 and C66, respectively) which are specific for the active plant gene set but for both of them the corresponding products are not known. Furthermore, three genes of the 10-gene cluster C16 including a terpene cyclase (FGSG\_04591) and four genes of the eight-gene cluster C64 including a NRPS (FGSG\_10990) were classified as AP up-regulated—both clusters of which all genes were found to be expressed among the 65% of highest expressed genes during pathogenic growth (chapter "Genes commonly expressed in the dead and the living plant material represent typical 'in planta' genes").

### The AP Gene Set Defines Defense-Related Genes among the Previously Described "*In planta*" Genes

In order to categorize our genes in relation to published data, we compared our AP and PP gene set defined after 72 h after inoculation to genes identified previously to be expressed

inducing (DI) categories (*p* < 0.01) (A) and FunCat analysis of 184 exclusively AP up-regulated genes (B). In (A) the appendage "\_up" indicates up- and the appendage "\_down" stands for down-regulation by the respective impact factor.

specifically "in planta" at different time points in wheat and barley. The most comprehensive transcriptomic study taking also previous data sets into account is available from Lysoe et al. (2011b). They investigated the global gene expression of F. graminearum during wheat infection over time series ranging from 24, 48, 72, 96, 144, to 192 h after inoculation using microarray technology and compared their own results with published microarray data from other groups (Güldener et al., 2006). This global comparison resulted in a gene set of 591 genes exclusively expressed during infection (in wheat or both wheat and barley) whereas 9500 genes are expressed in barley, in complete or carbon or nitrogen starvation media or simultaneously during the other conditions. Importantly, fresh dead plant material like the one we used in our approach was not among the various tested substrates in the analyses carried out previously.

The Venn diagrams in **Figure 9** show the results of this overlay between our analysis (AP, PP) and the published datasets (in planta publ/others publ.). For the interpretation of this comparison it is important to know that the microarray-based analysis carried out by Lysoe and coworkers considered all genes which were expressed at a certain time point regardless to which level they were induced. Thus, their plant-specific gene set contained all genes which were below a certain threshold in any of the other conditions and expressed above this threshold in infected wheat, barley or in both of them. This analysis is similar with our approach shown in **Figures 5**, **6A,B** and Figure S2 where we considered condition-specific genes expressed above a certain threshold (contained within the 65% highest expressed gene set).

Here, we used our differential gene expression analysis which defined the AP and PP gene sets and looked for the overlap between the "plant-specific" gene set defined by Lysoe and colleagues. From the 591 "in planta expressed" fungal genes defined in Lysoe et al. (2011b) 31 genes could be classified as AP, 39 as PP up- regulated and 6 genes are induced regardless if the host is alive or dead whereas 515 genes do not match the upregulated AP or PP gene sets. One important functional category not overlapping between the gene sets are genes associated with SM functions (71 genes). This difference is consistent with the use of L-ornithine as SM inducer in our axenic cultures and this leads to a removal of almost all SM-associated functions from the plant-specific gene sets. The previously published transcriptome studies, in contrast, used complete media and nitrogen or carbon starvation conditions for non-plant controls which do not, or to a much lesser extent, induce SM-associated genes and, therefore, these functions appear in their analysis in the plantspecific gene set. It is noteworthy, that also genes of the DON biosynthetic pathway, such as TRI5, the trichodiene synthase, or TRI6 and TRI10, the two DON-cluster transcription factors, are not responding heavily to the starvation media used in these experiments and thus appear in the in planta-specific gene set in Lysoe et al. (2011b).

#### Mainly Genes Responding to Plant Tissues Are in the Overlap between Previously Defined Plant-Specific Genes and Our Up-Regulated PP Gene Set

From the 76 genes that do overlap between previously defined in planta-expressed genes and our at least four-fold induced condition-specific gene sets (**Figure 9**) we find genes belonging to functional categories such as regulation of directional cell growth, cell wall, degradation/modification of foreign (exogenous) compounds as well as sugar, glucoside, polyol,



\*Cluster numbers/ names are according to Sieber et al. (2014).

\*\*Signature enzymes (TPC, terpene cyclase; NRPS, nonribosomal peptide synthetase).

\*\*\*Tayloring enzyme (P450, cytochrome P450 monooxygenase).

FIGURE 9 | Venn diagrams showing overlaps between active plant (AP) and passive plant (PP) up- (A) and down-regulated (B) gene sets with published datasets comprising exclusively wheat or wheat and barley induced genes (*in planta* publ.) as well as genes expressed exclusively in barley, complete or carbon and nitrogen starvation media or combined wheat/barley/media genes (others publ.).

and carboxylate metabolism (register tab "PPup\_in\_planta\_39" of Table S3). The underlying genes are listed in **Table 3** and include genes such as ones coding for a probable pectate lyase (FGSG\_04430) or a cellulose binding protein (FGSG\_03968). Furthermore, there are several genes predicted to encode sugar degrading enzymes, a putative monosaccharide transporter (FGSG\_10921) or a probable CYB2—lactate dehydrogenase cytochrome b2 (FGSG\_01531; **Table 4**), predicted to contribute to C-2 compound and organic acid metabolism as well as lactate fermentation (register tab "APup\_PPup\_in\_planta\_6" of

TABLE 3 | Thirty nine passive plant (PP) up-regulated as well as *in planta* expressed genes.


Table S3). Taken together, using the described bioinformatic approach, we were here able to better describe 45 of the formerly "plant-specific" genes as activities which simply respond to the plant tissue and are thus very unlikely specific pathogenicity genes.

Additionally, we could identify within our experiment 30 new genes within the PP category which had not been found previously to be plant-specifically expressed (**Table 5**). The large proportion of genes with predicted functions among them act in cellular transport contributing to cellular import, carbohydrate transport, ion transport and homeostasis of cations and phosphate as well as peroxisomal transport (register tab "PPup\_30" of Table S3).

## Many Up-Regulated Genes in the PP Category Code for Secreted Proteins

Many of our PP-specific genes code for proteins which carry secretion signals and are thus likely to be externalized during growth. Consequently, we also compared this gene set to the proteins found in a proteomic study to occur in apoplastic fluids of infected wheat (Paper et al., 2007). From the 120 proteins previously identified in the proteomic study we found 118 genes in our experimental set up and by considering only genes that encode predicted secretion proteins we could identify 16 genes to be PP up-regulated and one gene to be positively regulated by AP and PP impact factors (see **Figure 10**). The PP up-regulated functional categories comprise genes coding for cytoplasmic, nuclear, lysosomal, and vacuolar protein degradation as well as for proteolytic protein processing and extracellular polysaccharide degradation. Examples are proteins similar to secreted alpha-Narabinofuranosidase/alpha-L-arabinofuranosidase (FGSG\_03003), to vacuolar aminopeptidase Y precursor (FGSG\_03027), to endo-1,4-beta-xylanase A precursor (FGSG\_03624) and a predicted cellulase (FGSG\_11184). The one gene that is positively affected by AP and PP impact factors, FGSG\_00028, is annotated as probable metalloprotease MEP1 likely to be involved in protein/peptide degradation.

### One Part of the AP Gene Set Overlaps with Known "*In planta*" Genes and Contains Many SM Precursor Functions

The 31 AP genes which overlap with the previously identified in planta genes show functional enrichment in categories of

TABLE 4 | Six active and passive plant (AP and PP) up-regulated as well as *in planta* expressed genes.



tyrosine, phenylpropanoid, and triterpene metabolism as well as metabolism of peptide antibiotics, defense-related proteins and secondary metabolism (register tab "APup\_in\_planta\_31" of Table S3 and **Table 6**). The genes have been previously identified as "in planta" expressed but it is interesting that the genes with SM-related functions appear in our analysis not together with the classical SM genes such as DON or culmorin cluster genes. This indicates that—among the known SM genes there are genes which cannot be induced by the pathway described above involving L-ornithine as inducer compound. This seems to be true for the farnesyltransferase involved in the provision of general precursor farnesylpyrophosphate (FPP) and the tyrosinase which is part of the shikimate pathway important for SM precursors. Among the "defense-related" functions there are the two AM toxin genes and the third one, FGSG\_11542 most likely encodes an aldehyde dehydrogenase involved in acetate formation. It thus may also be part of active-plant responsive SM biosynthesis as acetate can serve as a building

block for several SMs. Induction of all these SM-related functions is clearly dependent on the actively defending plant and it would be worthwhile to study gene regulation of this gene set to understand the difference between the "classical" SM genes induced by the ornithine pathway also in axenic cultures and these genes which are strictly dependent on the active plant.

## Expression Pattern of Known Pathogenicity Factors in Our Data Set

In their great review Walter et al. (2010) published a table that shows pathogenicity and virulence factors produced by F. graminearum and F. culmorum which are known in literature to reduce or even lead to loss of virulence when mutated. The trichodiene synthase-encoding gene TRI5 (Proctor et al., 1995a, 1997; Desjardins et al., 2000) and the transcription factor TRI6 (Proctor et al., 1995b; Hohn et al., 1999) are wellknown examples, which showed the highest transcription during axenic growth in our experiment and could be assigned to the DI up-regulated (TRI5) and PP down-regulated (TRI6) impact factor categories. The extracellular secreted lipase-encoding gene FgFGL1 (Voigt et al., 2005) was strongest transcribed in the saprophytic samples, whereas FgNPS6 (Oide et al., 2006) and FgSID1 (Greenshields et al., 2007) both involved in iron metabolism showed the highest transcription during infection and both could be classified as AP up-regulated genes. Furthermore, the F. graminearum homolog of FcABC1 (Skov et al., 2004), encoding a PDR ABC transporter possibly involved in transport of phytoanticipins, and GzARG2 (Kim et al., 2007), encoding an acetylglutamate synthase involved in arginine biosynthesis both showed highest transcription during pathogenic growth and Fg/FcABC1 could be assigned as AP upregulated, too. The other listed pathogenicity factors showed a

#### TABLE 6 | Thirty one active plant (AP) up-regulated as well as *in planta* expressed genes.


similar transcription in all three sample types, which does not mean that they are not important for the infection process (since their mutation causes reduced virulence) but might indicate, that their transcriptional regulation does not necessarily need signals from the active, living plant to be switched on. Additionally posttranscriptional regulation might take place modifying or fine-tuning activity of the gene product. Cuomo et al. (2007) published a list of possible virulence factors among predicted secreted proteins that were specifically expressed in planta and found in high SNP density regions within the fungal genome. The list contains putative homologs of known virulence factors, cell wall-degrading enzymes and cytochrome P450s. Most of the cell wall-degrading enzymes show similar transcription levels on the pathogenic and the saprophytic samples, some a higher level on the latter one. The pectate lyase FGSG\_03483, however, in addition to the secreted protease FGSG\_00028 and


the cytochrome P450 oxidase FGSG\_10991, show significant higher induction during infection of the living plant compared to saprophytic and axenic growth and, thus, may represent interesting candidate genes for further study (out of this three FGSG\_10991 could be classified as AP up-regulated). In our experimental setup we could dissect three factors that impact on the invading fungus and vary its transcriptional response accordingly. Targeting genes that are affected by the active plant response may be a promising strategy for countermeasures against this fungal disease.

## The AP Category Also Contains Putatively Novel Virulence Factors

The most interesting category in our analysis is surely the 21 genes in the AP category which have not yet been considered in any of the previous studies. As enriched functional categories we found here disease, virulence and defense as well as detoxification, drug/toxin transport, and detoxification by export (register tab "APup\_21" of Table S3) comprising genes such as FGSG\_02672 (probable cytochrome P450 monooxygenase (lovA)-encoding gene), FGSG\_03080 (related to ethionine resistance protein), FGSG\_00101 (related to integral membrane protein), FGSG\_16401 (probable ATP-binding cassette multidrug transport protein ATRC; **Table 7**). Isoprenoid and terpene metabolism was also enriched additionally including FGSG\_09863 (related to acyl-CoA cholesterol acyltransferase). Furthermore, the gene set includes FGSG\_02917 (related to cellobiose dehydrogenase) and FGSG\_04057 (probable potassium transporter TRK-1) as well as FGSG\_02263 (uncharacterized protein—related to ABC-type Fe3 transport system, periplasmic component) and FGSG\_03186 (related to oxidoreductase).

Novel SM-related genes in this category feature an acylcoA transferase with putative functions in sterol metabolism (FGSG\_09863). This type of acyltransferases are known to be membrane-bound proteins that utilize long-chain fatty acid acyl-CoA and sterols as substrates to form steryl esters as precursors of ergosterol biosynthesis (Madhosingh and Orr, 1981; Perkowski et al., 2008; Breakspear et al., 2011). The P450 monooxygenaseencoding gene (FGSG\_02672) is not part of any established SM cluster and may thus, have different non-secondary metabolic functions. Several transporters are part of this AP gene list including predicted iron and potassium transporters, multidrug resistance proteins as well as an efflux pump for the methionine analog ethionine (FGSG\_03080). The protein may be involved in methionine-related C1 metabolism, most likely via connection to SAM (S-adenosylmethionine) metabolism which was shown to be affected in ethionine-resistant mutants (Barra et al., 1996). An interesting gene is represented by FGSG\_02917, which might encode an extracellular protein of the haemoflavoenzyme family because proteins with significant similarity are able to convert cellobiose and other cellodextroses to their respective lactones (Henriksson et al., 2000). This function of the APinduced gene might hence be necessary to induce further plant tissue-degrading enzymes. Another yet unrecognized gene with a putative virulence function might be represented by FGSG\_00101 which encodes a protein similar to Pth11 from Magnaporthe oryzae, a GPCR localized to membranes and required for appressorium differentiation and, thus, pathogenesis (DeZwaan et al., 1999; Lu et al., 2005).

Until recently, not much was known about small secreted cysteine-rich proteins (SSCPs) or "effectors" in F. graminearum. This secreted protein family plays a prominent role in other pathosystems including Fusarium species (de Sain and Rep, 2015). A recent genomic analysis, however, predicted 190 candidate genes which conform to the characteristics of SSCPs with a possible effector function (Lu and Edwards, 2016). In this study, 25 out of the 190 predicted peptides were confirmed by proteomics in axenic media culture supernatants and 23 of these were found by semi-quantitative RT-PCR to be expressed to different degrees in planta during the infection process. We queried our dataset for the expression levels of these possible effectors and found 48 of the predicted genes in the AP, PP, and DI categories to be differentially expressed more than fourfold (**Table 8**). Only six of the confirmed 23 SSCPs appeared to be strongly regulated in our comparison. Three of them (FGSG\_04805, FGSG\_05714 and FGSG\_11205) were strongly down-regulated by the active plant response (category AP\_down) and, thus, may be targets of the plant defense system. Two were found to be up-regulated under DON-inducing conditions (DI-up, i.e., high in pathogenic and axenic conditions) indicating that they are co-regulated with DON and, thus, be part of the regulatory circuit for virulence factor expression. Finally, the last confirmed SSCP-encoding gene in our dataset is FGSG\_13952, which assigns to the category of down-regulated genes in the DON-inducing conditions. Thus, this secreted peptide seems to be inversely regulated compared to DON, very highly expressed under saprophytic conditions, most likely induced by a dead plant matrix component.

Interestingly, there are four genes of the 190 predicted SSCPs which assign to our AP\_up category because they are highly induced only in the actively defending plant (FGSG\_01239, FGSG\_07807, FGSG\_11047, and FGSG\_12554). Unfortunately, the corresponding peptides have not been found in the synthetic media secretome and thus, they are not confirmed SSCPencoding genes. However, the fact that they are highly induced only in the active plant indicates that they might have a signaling or regulatory function during the F. graminearumwheat interaction.

## Genes Repressed in the AP Gene Set Might Represent Important Fungal Targets of Plant Defense

So far we have mainly discussed genes up-regulated by the signals derived from the actively defending plant. But also genes specifically repressed by the active plant are interesting as they may represent possible targets of the plant defense system to keep the pathogen under control.

In this category (**Figure 9B**) we identified 46 genes repressed by signals originating from the active plant and which were not found by Lysoe et al. (2011b). Unfortunately, the vast majority in this gene set (42 genes) encodes putative proteins of unknown function whereby notable four genes in this set are predicted to encode short peptides with a predicted amino acid (AA) length between 100 and 210 AAs and a cysteine content above 3%: FGSG\_00860 (209 AA, 6.7% cys. content), FGSG\_12722 (143 AA, 5.6% cys. content), FGSG\_13097 (105 AA, 3.8% cys. content), and FGSG\_13598 (103 AA, 4.9% cys. content). Thus, they may represent small cysteine-rich effectors which may play a role in pathogen recognition or other processes and the fact that they are actively repressed by the living plant signals is certainly an interesting observation. If these peptides contribute to virulence, however, remains to be shown in future studies. The remaining four genes with predicted functions are listed in **Table 9**.

## CONCLUSIONS

The aim of our study was to better define the group of "in planta" expressed fungal genes because the pathogenic growth status represents a mixture of genes responding to the plant tissue as substrate already present (such as cell wall compounds and pre- formed metabolites) as well as genes responding to defense signals and metabolites built up in response to the pathogenic attack. This genetic differentiation is surely relevant because it has the potential to identify novel pathogenicity factors. In addition, potentially novel plant defense targets may be present in the set of down-regulated fungal genes identified uniquely in the living plant. Our approach has led to a differentiation of genes responding to these mixed life styles and we identified 184 genes upregulated and 122 genes repressed at least four-fold by the signals the active plant generates during the infection process.


TABLE 8 | Out of the 190 small secreted cysteine-rich proteins (SSCPs) identified in the genome of *F. graminearum* (isolate PH-1) by Lu and Edwards (2016), we could assign 48 exclusively to active plant (AP), passive plant (PP), or DON- inducing (DI) up- or down-regulated categories within our analysis (indicated by gray colored fields).

Of these, 12 genes are among the 34 SSCPs detected by Lu and Edwards (2016) during transcriptional analysis of infected wheat heads- FGSG numbers marked by one asterisk (\*). FGSG numbers marked by two asterisks (\*\*) are among the 15 in planta SSCPs with transcripts up-regulated at certain point. Underlined FGSG numbers are among the 25 SSCPs confirmed by nanoscale liquid chromatography-tandem mass spectrometry (nanoLC-MS/MS) analysis of the minimal medium-based in vitro secretome.

TABLE 9 | Only four out of 46 new AP down-regulated genes show a predicted gene function.


We found in this gene set a substantial overlap with previously characterized "in planta" expressed genes but we identified 21 so far unrecognized genes truly responsive only to the living plant. Among them we found several uncharacterized secondary metabolite gene clusters and putative membrane proteins with signaling functions. Interestingly, most of the downregulated transcripts potentially code for proteins with unknown functions making this category highly attractive for future research. It is not unreasonable to expect that some of these novel active-plant induced or repressed genes code for novel pathogenicity factors promoting the infection process or targeting of the plant defense system. Further studies by inactivation and overexpression of these genes will probably shed new light on the F. graminearumwheat interaction process during the pathogenic process and it will also be interesting to see which of those genes are altered in their expression profiles in the epigenetic mutants we are studying in parallel.

## MATERIALS AND METHODS

## *F. graminearum* Strain Maintenance, Axenic Culturing, and Spore Production

The F. graminearum Ph-1 wildtype (FGSC 9075, NRRL 31084) applied in this study was maintained on Fusarium minimal medium (FMM) agar plates according to standard methods (Reyes-Dominguez et al., 2012). Details on cultivation conditions and media are given in Datasheet 8 in Supplemental Materials and Methods.

## Wheat Infection Experiment

The highly susceptible cultivar Remus (pedigree: Sappo/Mex//Famos; Buerstmayr et al., 2002) was used in this study. On average 20 spikelets per wheat head were inoculated by pipetting 20,000 macroconidia (10 µL of a 2 <sup>∗</sup> 10<sup>6</sup> spores/ mL suspension) on the reproductive part between the lemma and palea of the two basal florets during anthesis (resulting in 8 <sup>∗</sup> 10<sup>5</sup> spores per wheat head). Mock ears were inoculated the same way, but using water instead of spore suspension. Three ears were inoculated on the living plant representing pathogenic growth of the fungus and three ears, which were cut off the plant and shock-frozen in liquid N prior to spore application, were inoculated as "non-response" control from the wheat side representing saprophytic growth of the fungus. After inoculation the wheat heads on the living plants were covered with moistened plastic bags for the first 24 h to provide high humidity. The inoculated dead wheat heads were placed in glass petri dishes (Ø 140 mm; h 20 mm). Incubation conditions were set at 20◦C, 50% relative humidity during daytime and 18◦C, 50% humidity during night with a 16 h photoperiod. Harvesting was performed 3 and 5 dai by freezing the plant material in liquid nitrogen. Only palea and lemma of the inoculated florets were sampled including the respective part of the rachis. Additionally completely untreated wheat heads were sampled. After freezing in liquid nitrogen samples were stored at −80◦C.

Cell disruption of the collected plant material was performed in the Retsch <sup>R</sup> Mixer Mill MM 301. The 50 mL steel grinding jars were used together with the 25 mm Ø steel balls (one ball per grinding jar). Before sample loading the grinding jars/ steel balls were baked at 270◦C for 1 h, left at room temperature for 1 h and then pre-cooled overnight at −80◦C. Milling was done at 30 Hz and room temperature for 20 s.

Independent RNA-extractions of three distinct pathogenic and three distinct saprophytic inoculated wheat heads harvested 3 dai were delivered for RNA-Seq analysis to the VetCORE—Facility for Research [University of Veterinary Medicine Vienna (VUW), Veterinärplatz 1, A-1210 Vienna, Austria].

## Chemical Analysis of Ornithine and Sugar Contents in Culture Supernatants

The levels of ornithine, fructose, glucose and sucrose in Fusarium culture supernatants were determined by applying a recently published GC-MS method (Warth et al., 2015). A concise description of the applied method and modifications is given in Datasheet 8 in Supplementary Material.

## Chemical Analysis of Nitrate Content in Culture Supernatants

In the assay 100 µL culture filtrate or standard solution were mixed with 100 µL of acidic VCl<sup>3</sup> (50.9 mM Vanadium(III) chloride [Sigma 2087272], 1 M HCl), 50 µL Griess Reagent I (N- (1-Naphthyl)ethylenediamine dihydrochloride [Sigma 33461]), and 50 µL Griess Reagent II (58 mM Sulfanilamide alias Amino-4 benzénesulfonamide [VWR 21156-237], 3 M HCl) incubated at 37◦C for 1 h followed by measurement of extinction at 540 nm. Culture filtrates were diluted 1: 1000 prior to the assay and as standard solutions liquid FMM media (Reyes-Dominguez et al., 2012) was prepared containing 0.2125% (w/v) NaNO<sup>3</sup> (25 mM) and diluted to final nitrate concentrations of 50, 40, 30, 20, and 10 µM.

## Chemical Analysis of Secondary Metabolites Culture Supernatants

In case of the axenic minimal media cultures 1 mL of each culture filtrate was directly analyzed by liquid chromatography/electrospray ionization-tandem mass spectrometry (HPLC/ESI-MS/MS) as previously described (Sulyok et al., 2006; Vishwanath et al., 2009). A more detailed description of the applied method and modifications is given in Datasheet 8 in Supplementary Material.

## Chromosomal DNA and RNA Extraction and RT-qPCR Quantification of Infection Rate

#### DNA and RNA Extraction and Measurement of Infection Rates

Details for the standard procedures of DNA and RNA extractions are given in Datasheet 8 in Supplemental Materials and Methods. The measurement of infection rates is based on a qPCR method that quantifies the proportion of fungal chromosomal DNA (chrDNA) within the fungus/wheat DNA mixture (Brunner et al., 2009). The qPCR protocol was set up as it was described by Brunner et al. (2009) with following modifications: The external standard curve for the TRI5 assay was generated by analyzing four-fold dilution series of F. graminearum DNA from pure fungal cultures starting with a concentration of 110 ng/µL (110, 27.5, 6.88, 1.72, 0.43 ng/µL), diluted in sterile water. The external standard curve for the EF-G assay was still done by measuring two-fold dilution series of wheat DNA starting from 100 ng/µL, but going down till 6.25 ng/µL (100, 50, 25, 12.5, 6.25 ng/µL), diluted in sterile water. For a single TaqMan reaction 2 µL template DNA was mixed with 7.5 µL Kapa Probe Fast qPCR Universal 2x Mastermix (Peqlab 07-KK4701), 4.78 µL sterile water, 0.24 µL dual-labeled probe, forward and reverse primer (each 100 pmol/ µL). Used primers and probes can be found in Brunner et al. (2009). The RT-qPCR analysis was performed on a BIORAD IQTM5 Multicolor Real-Time PCR Detection System. Cycling conditions included a single initial step at 95◦C for 1 min 50 s, followed by 45 cycles of 95◦C for 15 s, 52◦C for 20 s, and 60◦C for 15 s for the TRI5 assay. In case of the EF-G assay the initial denaturation was set at 95◦C for 1 min 50 s, followed by 45 cycles of 95◦C for 15 s, 57◦C for 20 s, and 62.5◦C for 20 s.

## RNA Extraction and RT-qPCR Analysis

RNA isolation was performed using RNeasy Plant Mini Kit (Qiagen, 74904) following the instructions of the manufacturer with one modification and one extension: Instead of eluting RNA once in 30 µL RNase-free water, elution was performed in 2 × 20 µL RNase-free water with a 3 min standing period at room temperature after each water application to the column. Additionally on-column DNase digestion was done during the extraction protocol by using RNase-Free DNase Set (Qiagen, 79254). RNA quantity was first checked on the NanoDrop 2000c Spectrophotometer (Thermo Scientific). Afterwards RNA quantity and integrity was validated by using the Agilent RNA 6000 Nano Kit on an Agilent 2100 Bioanalyzer machine following the instructions of the provider. The cDNA Synthesis was carried out with the RevertAid H Minus First Strand cDNA Synthesis Kit (Thermo Scientific, # K1632) using random hexamer primer and following the manufacturer's protocol. Details on procedure and conditions for cDNA synthesis and determination of infection rates are given in Datasheet 8 in Supplemental Materials and Methods.

## RNA-Seq Mapping and Quantification

The genome of F. graminearum and FGDB annotation version 3.2 was retrieved from http://www.helmholtz-muenchen.de/ en/ibis/institute/groups/fungal-microbial-genomics/resources/ index.html (Wong et al., 2011). RNA-seq reads were mapped on the reference genome using tophat2 (v2.0.8). The interval for allowed intron lengths were set to min 20 nt and max 1 kb (Trapnell et al., 2009; Kim et al., 2013). We used cufflinks to determine the abundance of transcripts in FPKM (Fragments Per Kilobase of exon per Million fragments mapped) and integrated the biological replicates using cuffdiff (Trapnell et al., 2010, 2012). The gene models were included as raw junctions. Genes with a minimum of two-fold increase or decrease in expression (|log2 of the FPKM values +1| ≥ 1) between the two experimental conditions were considered as regulated. Significant differential regulated genes of no functional annotation were manually re-visited. All gene models which were based on all current evidences including RNA-seq data (this study and unpublished) considered as spurious ORFs, were omitted from the downstream analysis. All data are available at NCBI GEO under the accession number GSE72124.

## Functional Classification

Genes with a two-fold increase or decrease in expression between the two experimental conditions were analyzed for overrepresented functions. We therefore used the FunCat catalog of protein function (Ruepp et al., 2004) in combination with Fisher's exact test (Fisher, 1922) and the MGSA-R package (Bauer et al., 2010). Resulting p-values were corrected for multiple testing using the Benjamini Hochberg procedure (Benjamini and Hochberg, 1995).

## Prediction of Putative Secreted Proteins

We computed secreted proteins in a pipeline approach. First we filtered on proteins targeted as secreted by TargetP (Emanuelsson et al., 2007) with an RC-score less than four. To add nonclassically secreted proteins we selected on predicted secreted proteins using SecretomeP (Bendtsen et al., 2004) with a cutoff score of 0.65. This set of proteins was further filtered for those which are predicted as extracellular by Wolfpsort (Tamura and Akutsu, 2007). To exclude extracellular, membrane bound proteins we utilized TMHMM (Krogh et al., 2001) for transmembrane domain prediction and excluded proteins with more than one predicted transmembrane domain.

## Extracting Factors from Combinations

Factor dependent transcription was calculated as follows: Factor AP was calculated according to formula AP = [patho] − [(sapro + axo)/2], where patho, sapro, and axo represent log2(FPKM) values of the experiments pathogenic- saprophyticand axenic- growth. Similarly, PP was calculated as PP = [(sapro + patho)/2] − [axo] and DI = [(axo + patho)/2] − [sapro]. Means and standard deviation were calculated by first order Taylor expansion in R. To determine if a gene is affected by a single factor t-tests between the groups of log2(FPKM) values between the specific contrasts, also indicated by square brackets, were calculated like for AP: patho vs. sapro & axo; PP: sapro & patho vs. axo; and DI: axo & patho vs. sapro.

The reasoning behind formula AP = [patho] − [(sapro + axo)/2]: patho is the only condition where a living plant is present so any difference between patho and the combination of sapro and axo is likely down to active plant action. PP = [(sapro + patho)/2] − [axo]: in both experiments sapro and patho is plant material but not in axo, so a deviation from zero may be cause by the plant. DI = [(axo + patho)/2] − [sapro]: both axo and patho experimental setups provide limited supply with nutrients vs. in sapro setup strong growth was observed what hints toward satisfactory supply. To determine if a factor is singly/mainly responsible we performed t-tests as described above.

#### AUTHOR CONTRIBUTIONS

SB, IM, ML, and JS planned and performed fungal and plant experiments. HB, SB, CS, MM, UG, and JS analyzed and

#### REFERENCES


interpreted transcriptome data. BW, MS, and RS analyzed and interpreted metabolomic data. SB, HB, and JS wrote the main parts of the manuscript with contribution from all other authors.

#### ACKNOWLEDGMENTS

We are grateful to Viktoria Preiser and Kurt Brunner for their help in quantitative infection assays by PCR. Work was supported by projects F3703, F3705, F3706, and LAP3714, which are projects within a special research area "SFB Fusarium" funded by the Austrian Science Fund.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fmicb. 2016.01113

Canty, J. Lewis, L. Siler, and R. W. Ward (Erlanger, KY: US Wheat & Barley Scab Initiative), 105.


bread wheat molecular response to Fusarium graminearum. G3 (Bethesda) 5, 2579–2592. doi: 10.1534/g3.115.021550


amino acid composition. BMC Bioinformatics 8:466. doi: 10.1186/1471-2105- 8-466


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2016 Boedi, Berger, Sieber, Münsterkötter, Maloku, Warth, Sulyok, Lemmens, Schuhmacher, Güldener and Strauss. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Solvent and Water Mediated Structural Variations in Deoxynivalenol and Their Potential Implications on the Disruption of Ribosomal Function

Nora A. Foroud<sup>1</sup> , Roxanne A. Shank <sup>2</sup> , Douglas Kiss <sup>2</sup> , François Eudes <sup>2</sup> \* and Paul Hazendonk <sup>1</sup> \*

<sup>1</sup> Lethbridge Research and Development Centre, Agriculture and Agri-Food Canada, Lethbridge, AB, Canada, <sup>2</sup> Department of Chemistry and Biochemistry, University of Lethbridge, Lethbridge, AB, Canada

#### Edited by:

Daniela Gwiazdowska, Poznan University of Economics, ´ Poland

#### Reviewed by:

Guilherme Lanzi Sassaki, Federal University of Paraná, Brazil Maciej Busko, ´ Poznan University of Life Sciences, ´ Poland

#### \*Correspondence:

François Eudes françois.eudes@agr.gc.ca Paul Hazendonk paul.hazendonk@uleth.ca

#### Specialty section:

This article was submitted to Fungi and Their Interactions, a section of the journal Frontiers in Microbiology

Received: 05 February 2016 Accepted: 25 July 2016 Published: 17 August 2016

#### Citation:

Foroud NA, Shank RA, Kiss D, Eudes F and Hazendonk P (2016) Solvent and Water Mediated Structural Variations in Deoxynivalenol and Their Potential Implications on the Disruption of Ribosomal Function. Front. Microbiol. 7:1239. doi: 10.3389/fmicb.2016.01239 Fusarium head blight (FHB) is a disease of cereal crops caused by trichothecene producing Fusarium species. Trichothecenes, macrocylicic fungal metabolites composed of three fused rings (A–C) with one epoxide functionality, are a class of mycotoxins known to inhibit protein synthesis in eukaryotic ribosomes. These toxins accumulate in the kernels of infected plants rendering them unsuitable for human and animal consumption. Among the four classes of trichothecenes (A–D) A and B are associated with FHB, where the type B trichothecene deoxynivalenol (DON) is most relevant. While it is known that these toxins inhibit protein synthesis by disrupting peptidyl transferase activity, the exact mechanism of this inhibition is poorly understood. The three-dimensional structures and H-bonding behavior of DON were evaluated using one- and two-dimensional nuclear magnetic resonance (NMR) spectroscopy techniques. Comparisons of the NMR structure presented here with the recently reported crystal structure of DON bound in the yeast ribosome reveal insights into the possible toxicity mechanism of this compound. The work described herein identifies a water binding pocket in the core structure of DON, where the 3OH plays an important role in this interaction. These results provide preliminary insights into how substitution at C<sup>3</sup> reduces trichothecene toxicity. Further investigations along these lines will provide opportunities to develop trichothecene remediation strategies based on the disruption of water binding interactions with 3OH.

Keywords: deoxynivalenol (DON), fusarium head blight (FHB), fusarium graminearum, NMR spectroscopy, mycotoxins, chemical structure

#### INTRODUCTION

Deoxynivalenol (DON) belongs to a class mycotoxins called trichothecenes and are produced by Fusarium species involved cereal crop diseases, such as Fusarium head blight (FHB; Foroud and Eudes, 2009). The responsible fungal species infect wheat and other small grains during flowering and kernel development stages and mycotoxins accumulate in the kernels of infected plants (Foroud et al., 2014). Ingestion of trichothecene-contaminated grain is harmful for human and animal consumers (Pestka, 2010). Trichothecenes are known to induce programmed cell death (apoptosis) by exerting ribotoxic effects on eukaryotic cells (Shifrin and Anderson, 1999; Rocha et al., 2005). Interestingly, like many antibiotics, trichothecenes interfere with ribosome function, and act as potent inhibitors of protein synthesis in eukaryotes (Ueno et al., 1968; McLaughlin et al., 1977). Earlier it was hypothesized that trichothecenes make direct contact with the ribosomal protein RPL3 (Gilly et al., 1985). Moreover, three domains of RPL3 function as a "rocker switch" that dynamically coordinates amino acyl-tRNA (aa-tRNA) and ribosome during translation elongation (Meskauskas and Dinman, 2008)—thus, an interaction of these toxins with RPL3 would result in an inhibition of protein synthesis. This hypothesis was validated in yeast where W225C or W225R mutations in the highly conserved W-finger of RPL3 conferred toxin resistance (Mitterbauer et al., 2004). The close proximity of RPL3 with the peptidyl transferase center (PTC) suggests that trichothecenes interfere directly with peptidyl transferase activity (Mitterbauer et al., 2004). Recent x-ray crystallography studies of toxin-bound yeast ribosomes, clearly shows trichothecene (DON, T-2 toxin and verrucarin A) binding to the A-site of the PTC (Garreau De Loubresse et al., 2014), which would impair peptide bond formation during translation elongation.

The most effective method to minimize trichothecene contamination of food/feed grain is to grow cultivars with FHB resistance and to employ strategic disease management practices, such as those previously described (Dill-Macky and Jones, 2000; Krupinsky et al., 2002; McMullen et al., 2008, 2012; Foroud et al., 2014). A major challenge is that "immunity" to FHB has not been identified in cultivated cereals, and the availability of highly resistant cultivars is limiting since resistance tends to be associated with poor agronomics (Foroud et al., 2014). Ongoing efforts have led to some improvements over the years (for an overview see McMullen et al., 2012, and other publications in the current issue of Frontiers in Microbiology), meanwhile FHB continues to have significant impact. Furthermore, no remediation strategies are available for detoxification or sequestration of trichothecenes. That being said, grain cleaning strategies can be employed to remove some of the contaminated roughage from the grain (Tittlemier et al., 2014) and biological mechanisms to detoxify trichothecenes have been identified (Fuchs et al., 2002; Poppenberger et al., 2003; Boutigny et al., 2008).

Trichothecenes are composed of three fused rings: the cyclohexene (A-ring) is fused to the tetrahydropyran (B-ring), which is bridged by a 2-carbon chain at C<sup>2</sup> and C<sup>5</sup> thereby forming a cyclopentyl moiety (C-ring). In addition, an epoxide functionality is attached at C<sup>12</sup> which is common to the Band C-rings (**Scheme 1**) (Cole and Cox, 1981). Side chains at C3, C4, C7, C8, and C<sup>15</sup> are variable, although primarily consist of H, OH, or OC(=O)CH3. Trichothecenes fall into four classes (types A–D) (McCormick et al., 2011), where either A and B are produced by Fusarium species. DON is a type B trichothecene, which is characterized by a ketone at C8, and has hydroxyl groups at C3, C7, and C15. The epoxide ring is essential for toxicity (Ehrlich and Daigle, 1987). which are unusually stable in the trichothecenes (Pronyk et al., 2006; Bullerman and Bianchini, 2007). Some bacterial species can open the epoxide ring, forming de-epoxynivalenol (DOM-1) (Fuchs et al., 2002; Schatzmayr et al., 2006). No de-epoxy trichothecenes have been reported in plants infected with trichothecene-producing Fusarium species. Other modifications, in planta, that show reduced toxicity include DON-thiol, and -glucoside conjugates (Gardiner et al., 2010; Kluger et al., 2013; Stanic et al., 2015). The former were observed at C9, disrupting the double bond in the A-ring (Kluger et al., 2013), and at C8, disrupting the R5 keto functional group at Gardiner et al. (2010). The latter are catalyzed by plant UDPglucostranferases that yield DON-3-glucosides (Poppenberger et al., 2003). Generally, modifications at C<sup>3</sup> reduces toxicity; for example, acetylation, produces 3-O-acetyl-DON which is significantly less toxic than DON (Desjardins et al., 2007; McCormick, 2009).

Structural knowledge of trichothecenes has driven detoxification/remediation strategies. Trichothecenes from all classes have been studied primarily in chloroform (Cole and Cox, 1981; Savard et al., 1987; Greenhalgh et al., 1989), which obviously does not emulate their biological environment. A single report on the effect of solvent DON and NIV structures provides compelling evidence for significant structural variation in different chemical environments. Jarvis et al. (1990) provided preliminary evidence for a second minor configuration in a single crystal X-ray diffraction (SCXRD) study of NIV recrystallized from a mixture of methanol-d<sup>4</sup> (CD3OD) and water. The authors compared changes in <sup>13</sup>C nuclear magnetic resonance (NMR) spectra of both DON and NIV in CDCl<sup>3</sup> with those in acetone-d6, CD3OD, and DMSO-d6, and proposed that the second configuration in both toxins is an isomer resulting from a rearrangement of the ketone at C<sup>8</sup> and hydroxyl at C<sup>15</sup> to a hemiketal linkage between C<sup>8</sup> to C15.

Historically, CDCl<sup>3</sup> was the solvent of choice as most trichothecenes are soluble in it, and it seems to minimize problems associated with aggregation which is at the heart of their poor solubility in water. Chloroform is a significant departure from the cellular environment. Although the cytosol is primarily aqueous, some hydrophobic environments with limited free water exist. As such, trichothecene structure and dynamics should be observed in both types of environments, where the presence of free water is controlled to gain a better understanding of how these toxins interact with and move through the cell.

A more comprehensive structural analyses and a dynamic understanding of the trichothecene-ribosome interaction and inhibition mechanisms will lead to more advancements in this area. With some exceptions, the majority of the structural analysis of trichothecenes reported to date was carried out 20– 30 years ago, and often with the purpose of basic structural determination and identification. Technological advances in structural determination have made it possible to perform more detailed structural analysis and dynamics. A small number of dynamic studies have been reported among the different classes of trichothecenes (Jarvis et al., 1990; Fragaki et al., 2008; Steinmetz et al., 2008; Chaudhary et al., 2011; Shank et al., 2011). Density functional theory (DFT) computations were carried out by Nagy et al. (2005), on the internal dynamics

of DON, which suggested that the lowest energy configuration contains an internal H-bonding network. The proposed Hbonding occurs between the C<sup>8</sup> carbonyl oxygen and the C<sup>7</sup> hydroxyl hydrogen, which in turn is linked to the hydroxyl hydrogen at C15. The energy of this interaction is significantly lower than all other locally optimal configurations considered; however, H-bonding and bridging with water was not taken into account in this study. Also recent evidence presented by our group on water binding in T-2 toxin (Chaudhary et al., 2011) demonstrates that the H-bonding networks, both intraand inter-molecular, of all the trichothecenes should be carefully considered.

The current study was designed to determine whether different solvent properties induce structural changes in DON, where the presence of free and bound water is controlled. High resolution <sup>1</sup>H and <sup>13</sup>C solution state spectra are presented which provides novel insights into the structure and dynamics of DON in different chemical environments. The structural implications on the toxicity mechanisms and biological interactions of DON are discussed.

TABLE 1 | Chemical shifts (ppm) for DON in CDCl3 compared with literature.


<sup>a</sup>Savard and Blackwell (1994).

<sup>b</sup>EXSY crosspeaks are in blue.

<sup>c</sup>Not observed.

<sup>d</sup>NOESY crosspeaks of note are in red.

## MATERIALS AND METHODS

#### Materials

4-Deoxynivalenol was purchased from Sigma, (DON; CAS 51481-10-8), with estimated purity >95%. The delivered sample, which had a flaky appearance, was dried under dynamic vacuum to remove residual water.

NMR samples of DON were prepared in CDCl<sup>3</sup> to a concentration of 1 mg mL−<sup>1</sup> with TMS as an internal reference for both <sup>13</sup>C and <sup>1</sup>H. CDCl<sup>3</sup> was dried over molecular sieves to prevent further introduction of water through the solvent. A sample was also prepared in dried DMSO-d6; however for the samples in acetone-d6, THF-d<sup>4</sup> and methanol-d<sup>4</sup> the solvents were not dried. All deuterated solvents were purchased from Sigma-Aldrich.

## <sup>1</sup>H Solution-State NMR Experiments

All NMR spectra were acquired using a Bruker Avance 2 300 MHz spectrometer, outfitted with a 5 mm HX PABBO BB probe. The magnetic field strength is 7.05 Tesla, with larmor frequencies of 300.13 MHz for <sup>1</sup>H. All experiments were performed at ambient temperature (25◦C). Typical 1D <sup>1</sup>H spectra were recorded as 128 transients, using a 90◦ pulse width of 12.4 µs, and a recycle delay of 1.5 s. Similarly typical 1D <sup>13</sup>C spectra were obtained in 4000 transients, using a 90◦ pulse width of 7.6 µs, and a recycle delay of 4.0 s, as 64 K points over a spectral window of 18 kHz.

The homonuclear magnitude gradient <sup>1</sup>H COSY spectrum was acquired in 256 increments over a spectral width of 1800 Hz (12.0 ppm) in both dimensions, using a recycle delay of 1.5 s. 64 transients were collected for each increment, having 4096 points. The direct and indirect dimensions have a digital resolution of 0.88 and 7.03 Hz, respectively, before zero filling.

The gradient <sup>1</sup>H NOESY spectrum was acquired in at least 256 increments covering a spectral width of 1802.45 Hz (12.0 ppm) in both dimensions, using a recycle delay of 20 s and an array of mixing times of 0.5, 1.0, 2, 5, 7, and 10 s. Eight transients were collected for each increment, having 4096 points. The direct and indirect dimensions have a digital resolution of 0.88 and 7.03 Hz, respectively, before zero filling. Inversion recovery experiments determined that the T1s of all DON and water signals were below 3 s.

The <sup>1</sup>H-13C HSQC spectrum was acquired in 128 increments, using a recycle delay of 2.0 s, and a spectral width of 4006.41 Hz (13.34 ppm) in the direct dimension and 12,500 Hz (165.62 ppm) in the indirect dimension. 152 transients were collected for each increment, having 1024 points, which was zero filled to 4096 points. The t<sup>1</sup> dimension was linear forward predicted to 256 points and further zero filled to 1024 points. The direct and indirect dimensions have a digital resolution of 3.9 and 98 Hz, respectively.

The <sup>1</sup>H-13C HMBC spectrum was acquired in 256 increments using a recycle delay of 2.0 s, and a spectral width covering 1951.60 Hz (6.50 ppm) in the direct dimension and 14,268 Hz (190.24 ppm) in the indirect dimension. One-hundred transients were collected for each increment, having 1024 points and the FID was zero-filled up to a value of 4096 points. The t<sup>1</sup> dimension was linear forward predicted to 512 points and further zero filled to 1024 points. The direct and indirect dimensions have a digital resolution of 0.52 and 56 Hz, respectively.

Drying experiments were performed at ambient temperature by sequential addition of individual molecular sieves to the NMR tube between consecutive NMR measurements, over the course of 3 h where the water to DON ratio was monitored over a period of 72 h.

#### Simulations

The SpinWorks processing and simulation software suite, developed by Marat (2009) at the University of Manitoba, was used to simulate the data obtained for the 300 MHz <sup>1</sup>H spectra. The FIDs were zero-filled to 256 K points and were subjected to Gauss-Lorentz apodization with a line broadening between –1.00 to –0.50 Hz, and a Gaussian broadening of 0.33, depending on the signal-to-noise ratio in the data. The spectra were simulated in two parts as the whole spin system could not be simulated at once. The sub-spectrum of the six-membered ring was simulated as a 10-spin ABC3DEFG system using 7OH, 7β, 10, 11, 15α, 15β, 15OH, 16. Similarly, the sub-spectrum associated with the remaining hydrogens: 2, 3, 3OH, 4α, 4β, 14, 13α, and 13<sup>β</sup> was simulated as a ten-spin ABCDEF3GH spin system. Long range couplings were considered up to five bonds, and an inherent line-width of 0.3 Hz along with Gaussian line shapes was used to fit the data. In the A-ring simulation typically 2000 transitions were assigned to within RMS deviations below 0.03 Hz, and a largest absolute differences of less than 0.08 hz Standard deviations (SD) in all the spectral parameters ranged from 0.1000 to 0.001 hz In the C ring simulation, typically 2200 transitions were assigned with an RMS deviations below 0.03 Hz, and largest absolute differences less

than 0.09 hz SD in all the spectral parameters range from 0.1000 to 0.001 Hz.

#### Structural Analysis

All proposed structures were calculated and optimized using ACD ChemSketch, from which internuclear distances were estimated for comparison to those obtained from the NOESY spectra in CDCl3, and DMSO-d6. All NOESY spectra were analyzed using Mestre NOVA software, using an LB of 3 Hz in each dimension. Integration of all the cross-peaks (CP) and autocorrelation peaks (ACP) was performed from which the CP to ACP ratio were computed and the cross-relaxation rate, s, was obtained using:

$$
\tan \mathbf{h}^{-1} \left( \frac{\mathbf{I}\_{\rm CP}}{\mathbf{I}\_{\rm CAP}} \right) = \sigma \,\mathbf{r}\_{\rm mix} \tag{1}
$$

The cross-relaxation rates were used to determine the internuclear distances, using 4<sup>α</sup> and 4<sup>β</sup> as well as 13<sup>α</sup> and 13<sup>β</sup> as references with internuclear distances of 181 and 186 pm respectively. The following relationship was used:

$$r\_{ij} = r\_{ref} \sqrt[6]{\frac{\sigma\_{ref}}{\sigma\_{ij}}} \tag{2}$$

The EXSY CP were used to compute exchange rates between water and OH signals using:

$$\tan \mathbf{h}^{-1} \left(\frac{\mathbf{I}\_{\rm CP}}{\mathbf{I}\_{\rm ACP}}\right) = k \mathbf{r}\_{\rm mix} \tag{3}$$

#### RESULTS

#### Full Spectral Assignment of DON in Chloroform

The assignment of the <sup>1</sup>H spectrum in CDCl<sup>3</sup> differed significantly from those in previous reports on the basis of spectral simulation and new evidence from the COSY and NOESY spectra. In addition to the revision of the methylene proton assignments, accurate chemical shifts of the hydroxyl protons are offered here.

The <sup>1</sup>H spectrum of DON in CDCl<sup>3</sup> is shown in **Figure 1** in which the labeling convention of Savard et al. (1987) was employed as shown in **Scheme 1**. Where significant π-electron delocalization is expected, coupling constants of up to five-bonds were considered. The assignment of the <sup>1</sup>H spectrum is given in **Table 1** where the chemical shifts and coupling constants given alongside the diagnostic homonuclear CP from the COSY and the NOESY spectra. **Figure 2** contains the NOESY spectrum of DON in CDCl<sup>3</sup> measured using a mixing time of 3 s. The most

TABLE 2 | Internuclear distances calculated from NOESY spectrum in CDCl3 , at a mixing time of 3 s, compared with computed boat and chair geometries.


<sup>a</sup>Hydrogen nucleus.

<sup>c</sup>Predicted estimates for internuclear distances using AM1 computations.

prominent CP are labeled, and used to compute exchange rates and internuclear distances, which are shown in **Table 2**. The heteronuclear 2D experiments are not shown. The assignment corresponds closely to previous reports (Savard et al., 1987; Nagy et al., 2005) with some notable exceptions described below.

The assignment of 4αand 4βis now reversed, where the chemical shift of the former is smaller than the latter (**Table 1**). This was confirmed by NOESY, where a strong cross-peak between 4<sup>β</sup> and 11 is observed and that between 4<sup>β</sup> and 14 is weak, which stands in contrast to the strong cross-peak 4<sup>α</sup> and 14 and that between 4<sup>α</sup> and 11 being weak. Furthermore, the crosspeak between 3 and 4<sup>β</sup> is much stronger than the 1 to 4α, and those from both 15<sup>α</sup> and 15<sup>β</sup> to 4<sup>β</sup> are much stronger than those to 4α. This reassignment clearly makes it easier to appreciate the relative order in the chemical shifts where 4<sup>β</sup> is larger than 4α, which is consistent with 4<sup>β</sup> being eclipsed by the hydroxyl oxygen on C<sup>3</sup> (see **Figure 3A**). Correspondingly 4<sup>α</sup> eclipses 3 as indicated by the vicinal coupling of 10.98 Hz, which could be easily confused with a trans vicinal coupling causing 4<sup>α</sup> and 4<sup>β</sup> to be interchanged.

The assignments of the protons on the epoxy ring were reversed with respect to previous reports (Savard and Blackwell, 1994), where 13α, at 3.195 ppm, has a significant cross-peak with 14, while 13β, at 3.121 ppm, has a prominent cross-peak with 2. Additionally, the larger cross-peak between 7 and 13<sup>α</sup> compared with 7 and 13<sup>β</sup> further supports this assignment. Hence, the larger chemical shift of 13<sup>α</sup> compared with 13βis due to its proximity to 15OH.

The methylene protons 15<sup>α</sup> and 15<sup>β</sup> are not easy to assign; however, NOESY CP provide the most convincing evidence that 15<sup>α</sup> is closer to 14, while 15<sup>β</sup> is closer to 11. Thus, 15<sup>α</sup> nearly eclipses C5, while similarly 15<sup>β</sup> is almost eclipses C<sup>11</sup> and thus is oriented underneath the B ring, placing 15OH in the correct orientation to undergo H-bonding to 7OH. The larger chemical shift of 15<sup>α</sup> can be ascribed to its closer proximity to 7OH. This indicates a preference for the gauche (with a dihedral angle C7- <sup>C</sup>6-C15-O<sup>15</sup> <sup>=</sup> <sup>330</sup>◦ ) rotamer, which is stabilized by transient H-bonding to 7OH.

The shifts and couplings involving of the hydroxyl protons went previously unreported presumably due to being obscured by chemical exchange broadening with excess free water in the solvent. To illustrate the loss of resolution in those signals, **Figure 4** compares the spectra for DON prepared with CDCl<sup>3</sup> that was either dried or not dried. The high resolution achieved in the OH signals of the dried sample indicates that very slow exchange takes place with water. Line broadening in the wet sample suggests faster exchange; however, 7OH remains relatively narrow. The EXSY cross-peak intensities give exchange rates in the dry sample for 7OH is 0.04 s−<sup>1</sup> while those for 15OH and 3OH are 0.12 s−<sup>1</sup> suggesting that strong H-bonding between 7OH and the carbonyl oxygen hinders water exchange.

The natural line width of the water and OH signal in the dry sample are below 0.2 Hz as their T2s occur near 1.5 s. Consequently, the effect of chemical exchange on the OH chemical shifts is negligible and thus they reflect the true shift of the OH environment. The OH shifts in the dry sample support preferential H-bonding to 7OH as its shift is much larger than those of 15OH and 3OH.

The water peak contains only one component in approximately equal stoichiometric proportions to DON.

b Internuclear distance in Å computed from NOESY crosspeaks using Equations (1) and (2) (green indicates distances to the bound water molecule).

<sup>d</sup>Mean deviation squared.

<sup>e</sup>Chi squared based on 33 distances in Å<sup>2</sup> , and s is the standard deviation in Å (red and blue highlights the largest contributor to the chi squared of boat and chair conformations, respectively).

The signal is extremely narrow lacking any significant fine structure, indicating that it has significant mobility. In an attempt to remove water bound to DON, a sample was prepared in dried solvent where molecular sieves were added over a period of 72 h. There was an initial slight decrease in the ratio of the water to DON, after which, the ratio remained constant with time, and the number of sieves added. The stoichiometric ratio of water to DON converged to 1:1, meaning that one water molecule is very strongly bound to each DON molecule.

The <sup>13</sup>C signals were assigned using HSQC and HMBC. **Table 3** shows the <sup>13</sup>C assignment for CDCl<sup>3</sup> along with all the diagnostic heteronuclear CP. All the <sup>13</sup>C chemical shifts are summarized in **Table 4** and are reported with respect to TMS in all 5 solvents. The <sup>13</sup>C chemical shift assignment in CDCl<sup>3</sup> is the same as previously reported. Additional structural observations can be made from when they are compared between the different solvents.

The internuclear distances from the NOESY spectra in CDCl<sup>3</sup> are compared with those from structures modeled with a boat and chair configuration of the B ring (**Figures 3A,B**). The results are shown in **Table 2**. The c<sup>2</sup> between the experimental values and the boat configuration is 17.6 Å<sup>2</sup> , while that with the chair configuration is 6.7 Å<sup>2</sup> , indicating a preference for the chair form. If one were to drop internuclear distances involving any mobile nuclei (i.e., OH, sidechain and CH<sup>3</sup> signals) from consideration, c <sup>2</sup> would dramatically decrease to 11.8 Å<sup>2</sup> for the boat and 2.1 Å 2 for the chair form. This significantly increases the confidence level in the match of the remaining 18 internuclear distance measurements to the chair form with a standard deviation of 0.35 Å as compared with the boat form at 0.83 Å.

## Full Spectral Assignment of DON in DMSO-d<sup>6</sup>

The spectrum in DMSO-d<sup>6</sup> is shown in **Figure 5** along with its simulation. **Figure 6** shows the corresponding NOESY spectrum. The spectral parameters obtained by simulation and fitting to the experimental spectrum are given in **Tables 5, 6** for all five solvents.

The assignment of the <sup>1</sup>H spectrum also departs from previous work concerning the methylene protons at C<sup>4</sup> and C15. The hydroxyl resonances have been previously reported and are consistent with our findings. As in CDCl3, 4α, and 4βare reversed with respect to previous reports (Savard and Blackwell, 1994), where the former has the lower shift, and the vicinal coupling between the 4<sup>α</sup> and 3 indicates they are eclipsed. In addition, the chemical shift of 4<sup>α</sup> decreases, while 4<sup>β</sup> increases significantly between CDCl<sup>3</sup> and DMSO-d6. The assignments of 13<sup>α</sup> and 13<sup>β</sup> is the same as in Savard et al. (1987), hence reversed with respect to CDCl3, where the shift of 13<sup>α</sup> decreases sufficiently to be lower than 13<sup>β</sup> which is less sensitive to the change in solvent environment. The assignment of 13<sup>α</sup> and 13<sup>β</sup> were verified by NOESY CP between 13<sup>α</sup> and 14 and between 13<sup>β</sup> and 2, respectively. Both 15<sup>α</sup> and 15<sup>β</sup> decrease by similar amounts, although their absolute assignment is difficult to verify by NOESY as they both have CP with 14, 11, and 3OH. However, 15<sup>α</sup> is the only one to have a cross-peak with 4β.

Other notable trends in <sup>1</sup>H shifts changes from CDCl<sup>3</sup> to DMSO-d<sup>6</sup> are that all the shifts corresponding to the position on the C-ring namely 2, 3, and 4<sup>α</sup> decrease, and similarly 7, 14, and 16 decrease while 11 increases. Most of the changes in the chemical shifts can be reconciled by the reorientation of 15OH to favor the transrotamer with respect to C6, as shown in **Figure 3C**. The 15OH now situated under the B-ring oriented toward the Cring side. This increases the distances from 15OH to 4α, 7, 14, and 16 hence significantly decreasing their shifts, while decreasing the distances to 4<sup>β</sup> and 11 hence increasing their shifts.

Again the slow water exchange allows OH shifts to be interpreted. In DMSO-d<sup>6</sup> the chemical shift of 7OH increases the least by 1 ppm, while 15OH and 3OH increase much more dramatically by approximately 3 ppm each. This can only occur if significant new H-bonding interactions take place involving 15OH and 3OH. In response to 15OH being sequestered 7OH is free to interact more strongly with the carbonyl oxygen as indicated by <sup>3</sup> J7,7OH changing from 2 to 4 Hz between CDCl<sup>3</sup> and DMSO-d6.

Resolution enhancement of the water signal indicates that there are two distinct water signals. One component has a splitting pattern that resembles a doublet of triplets with a linewidth of 0.3 Hz (**Figure 7**). This is actually an AMNX spin system, where A and X are 15OH and 3OH, and the water protons M and N are inequivalent by 0.013 ppm. The coupling constants are <sup>2</sup> JMN = −7.8 Hz (<sup>2</sup> JHH is estimated from HOD in deuterium exchange study on T2-toxin in CDCl<sup>3</sup> (Chaudhary et al., 2011), and the outer lines of the MN pattern have 1.3% intensity of inner lines and thus are lost in the noise), <sup>6</sup> JAX = 0 Hz, and 2 JAM = <sup>2</sup> JNX and <sup>4</sup> JAN = <sup>4</sup> JMX each ranging from −0.20 to −0.40 Hz. Such a coupling pattern is consistent with an immobile bridging water species, between 3OH and 15OH, spanning the mouth of a "binding pocket." When considering the B- and Crings together as a seven-membered ring that is bridged by the epoxide, the "binding pocket" is the bottom of the combined ring, opposite to the epoxide bridge.

The other component of the water signal exhibits no fine structure with linewidth of 0.86 Hz. With the water bridged (hence strained) the binding pocket retains essentially the same configuration as in CDCl<sup>3</sup> where 3 and 4<sup>α</sup> remain staggered. The exchange rate between water and all three hydroxyl groups is effectively the same near 0.07 to 0.08 s−<sup>1</sup> , which indicates



<sup>a</sup>Measured directly from the spectrum. No statistical error available from fitting. The experimental digital resolution is 0.55 Hz or 0.007.

TABLE 4 | Experimental and literature <sup>13</sup>C chemical shift data for DON in CDCl3 , acetone-d6 , and DMSO-d6 .


<sup>a</sup>All literature data obtained from Jarvis et al. (1990).

<sup>b</sup>Blue indicates increase and red decrease in chemical shift with respect to CDCl3. No error estimates available from fitting. The experimental digital resolution is 0.55 Hz or 0.007 ppm.

that binding to 15OH and 3OH is stronger in DMSO-d<sup>6</sup> than in CDCl3.

The <sup>13</sup>C shift changes occurring in DMSO-d<sup>6</sup> with respect to CDCl<sup>3</sup> are consistent with the conformation changes proposed, where δC5, δC7, and δC15 decrease when 15OH moves under the binding pocket which also causes δC4, δ12, and δ<sup>14</sup> to increase. The changes in δC3, δC4 and δC12 are also due in part to reorientation of 3OH.

Conformational changes were investigated further using NOESY CP in to compare cross relaxation rates s11,4β, s15α,4β, s13α,7, s13β,7, and s13β,<sup>2</sup> between the two solvents. The corresponding internuclear distances r11,4β, r13β,2, r13α,7, and r13β,<sup>7</sup> are consistently shorter in DMSO-d<sup>6</sup> by at least 0.10, 0.15, 0.10, and 0.25 Å. These changes indicate that water binding has caused slight changes in the binding pocket bringing C<sup>4</sup> and C<sup>3</sup> closer to C<sup>6</sup> and C11, effectively increasing the degree of folding between the B- and C-rings by decreasing the valence angles at C<sup>2</sup> and C5. As a result, the epoxide ring becomes further removed from the A- and C-rings. This is corroborated by the decrease in chemical shifts of 2, 3, 4α, and 7 which would experience deshielding effect of the epoxide ring less strongly. Also the reorientation of 15OH to underneath the binding pocket was confirmed by r15α,4<sup>β</sup> decreasing from 2.70 to 2.19 Å.

Additional evidence for changes in the geometry of the Band C-rings upon water binding is seen in the changes to the scalar couplings in DMSO-d6. The dihedral angle between 2 and 3 increases, and between 3 and 4<sup>α</sup> decrease as indicated by 3 J2, <sup>3</sup> decreasing and <sup>3</sup> J3,4<sup>α</sup> increasing. The valence angle between 4<sup>α</sup> and 4<sup>β</sup> increases and that between C<sup>3</sup> and C<sup>5</sup> decreases as indicated by <sup>2</sup> J4α,4<sup>β</sup> decreasing. Increased strain in the epoxide ring can be inferred by the increase in <sup>2</sup> J13α,13β.

## Analysis of the <sup>1</sup>H Spectra of DON in Acetone-d6, THF-d4, and Methanol-d<sup>4</sup>

The <sup>1</sup>H chemical shift trends in the remaining solvents are consistent with the findings in DMSO-d6. The shifts of 11 and 4<sup>β</sup> increase, whereas that of 4<sup>α</sup> decreases (**Table 5**). Even though the trends in 7, 2, 3, and 13<sup>α</sup> are not as extreme as in DMSO-d6, they do follow the same sense. Also, in acetone-d<sup>6</sup> and THF-d<sup>4</sup> the 3OH and 15OH shifts increase significantly by 2 ppm, which are not quite as large as in DMSO-d6. In methanol-d<sup>4</sup> the hydroxyl signals could not be observed as they were lost to deuterium exchange with the solvent.

Only the water signal in acetone-d<sup>6</sup> exhibited sufficient resolution such that two components could be found, corresponding to two bound water molecules; however, there were no fine structural features that indicate a bridging binding motif. Cross-relaxation measurements suggest that one of the water molecules was approximately 2.80 Å away from 11.

In all three solvents 15OH does favor the trans rotamer; however, not exclusively as in DMSO-d6. The remaining mobility in 15OH accounts for the smaller increases in the hydroxyl shifts as well as the weaker trends in 7, 2, 3, and 13α. The weaker bonding interaction in the binding pocket for methanol-d<sup>4</sup> and THF-d<sup>4</sup> is not just the result of differing solvent polarity as they also contain much more free water, where the water to DON ratio was 14 and 30 to one, respectively. Furthermore, methanol-d<sup>4</sup> is a hydrogen bonding solvent and therefore would compete with water thereby further weakening the water binding to DON. One would expect to see the same effect in bulk water.

#### TABLE 5 | The chemical shifts (ppm) and associated error (Hz) obtained by fitting the <sup>1</sup>H spectra of DON are in the various solvents.


RMS<sup>a</sup>

<sup>a</sup>Typically 4200 transition were assigned in total. RMS, Root mean square deviation. The largest absolute difference was typically below 0.1 Hz.

<sup>b</sup>The simulations give chemical shifts in Hz units with standard deviation of less than 0.002 to 0.1 Hz. This amounts 5–7 significant figures; therefore, when converting to ppm the shifts should have be recorded from 4 to 6 decimal places. In this case the errors are given in Hz units, as in ppm the error would be 0 up to the third decimal place.

<sup>c</sup>Red indicated increase and blue a decrease in chemical shift with respect to CDCl3. The more significant changes are underlined.


TABLE 6 | Scalar couplings (Hz) and associated error (Hz) obtained by fitting the <sup>1</sup>H spectra of DON in the various solvents.

#### DISCUSSION

## Water Binding in the Proposed Structure with NOESY in CDCl3and DMSO-d<sup>6</sup>

Previous work in our group showed that a water molecule is bound to T-2 toxin at a 1:1 ratio (Chaudhary et al., 2011). The same was observed here for DON in both chemical environments investigated. Water binding to DON was observed through crossrelaxation between water and protons of the binding pocket. Internuclear distances in CDCl<sup>3</sup> from water to 2, 3, 11, 15<sup>α</sup>

and 15<sup>β</sup> are 2.99, 2.84, 3.14, 2.93, and 3.22 Å, respectively. This places at least one water molecule beneath the binding pocket in the vicinity of 15OH and 3OH. The lack of fine structure in the high resolution water <sup>1</sup>H signal suggest significant mobility remains in the bound water and thus is unlikely to bridge 3OH to 15OH.

In DMSO-d<sup>6</sup> water binding involves one water molecule bridging the 3OH to 15OH. Additional water may be bound elsewhere; however, it would be more mobile in a similar manner to what is observed in CDCl3. Unfortunately, no CP between

water and 2, 11, or 15<sup>α</sup> and 15<sup>β</sup> could be resolved, or found to be adequately free from t<sup>1</sup> noise to obtain reliable distance estimates. It was possible to integrate the cross-peak with 3 giving a distance of rW,<sup>3</sup> = 2.25 Å which is significantly smaller than in CDCl<sup>3</sup> and supports the bridging binding mode.

Considerable conformational variation exists amongst the trichothecenes due to differences in the pucker of the A- and B-rings. X-ray studies indicate a preference for the B-ring to adopt a chair conformation (see **Scheme 2**) in the crystal phase (Greenhalgh et al., 1984), although boat configurations have been observed for the macrocyclic trichothecenes (Jarvis and Mazzola, 1982; Jarvis and Wang, 1999). Until now, direct evidence for the chair configuration in solution was scarce, apart from some long range scalar coupling interactions observed between 7<sup>β</sup> and 11 in some systems pointing toward the chair form for the B-ring (Greenhalgh et al., 1989). In the structure presented herein, the B-ring is shown to be in a chair configuration, and the water molecule is bound in the binding pocket in both CDCl<sup>3</sup> and DMSO-d<sup>6</sup> (**Figure 3**).

The major structural differences observed between the two environments occurred with respect to the OH groups, which undergo distinct internal H-bonding and water binding patterns. In CDCl<sup>3</sup> the 15OH has significant configurational flexibility and undergoes H-bonding with both 7OH and 3OH, although 7OH is primarily occupied with H-bonding with the carbonyl oxygen at C8. In DMSO-d6, the interaction between 15OH and 7OH is lost, and the 15OH interaction with 3OH involves a bridging through a water molecule. This bridged interaction leads to a much stronger binding of the water in DMSOd<sup>6</sup> as compared with CDCl3. Furthermore, water binding in DMSO-d<sup>6</sup> leads to slight conformational changes in the B- and C-rings, such that the opening of the combined rings is smaller, thus accommodating the water bridging. Therefore, in non-Hbonding and low polarity environments, in the absence of free water, the configuration of the binding pocket arranges the 3OH and 15OH optimally for donation, hence they will be very reactive to nearby acceptors, such as water, basic amino acid residues, and metal cations.

### Comparison of DON Structures in Solution with Reported Structures of Bound Trichothecenes

The crystal structure of DON and other trichothecenes bound to the yeast ribosome (Garreau De Loubresse et al., 2014) and to Fusarium trichothecene acetyltransferases, TRI101 (Garvey et al., 2008) and TRI3 (Garvey et al., 2009), were previously reported. In the bound crystal structures for DON (Garvey et al., 2008; Garreau De Loubresse et al., 2014) and in the solution structure in DMSO-d<sup>6</sup> described herein, the B-ring is in the chair configuration and the 15OH group is pointing underneath the ring. In DMSO-d6, the 15OH is poised underneath the C-ring, whereas when bound in TRI101 it is underneath the A-ring. In the ribosome-bound crystal structure it is unclear whether it is pointing inwards toward the A- or C-ring. By contrast, in CDCl3, the 15OH of DON is pointing outwards from the molecule. Interestingly, this is similar to the crystal structures of the ribosome-bound and the TRI101-bound T-2 toxin (Garvey et al., 2008; Garreau De Loubresse et al., 2014), where the 15OAc is unambiguously pointing outwards. Furthermore, a water molecule was observed in the vicinity of the binding pocket in the TRI101-DON and TRI3-15-decalonectrin interactions (Garvey et al., 2008, 2009), in approximately the same distance as observed for DON herein. Furthermore, in the TRI101-DON interact, a strong donor-acceptor interaction was observed between the 3OH and His-156 (Garvey et al., 2008).

Based on these structural insights, we propose the following mechanism of inhibition of protein synthesis. The epoxide ring, known to be essential for toxicity, was not reported to have direct interactions with ribosomal components in the crystal structure presented by Garreau De Loubresse et al. (2014). The specific nature of the epoxide ring making it essential for toxicity is unclear; however, it stands to reason that the epoxide holds the binding pocket in an ideal configuration to bring 3OH and 15OH in close proximity. The NMR structures presented here show

that 3OH and 15OH bind to a water molecule in various ways depending on the nature of the solvent. Thus, we hypothesize that in a cellular environment, where free water is limited, the 15OH and 3OH of DON interacts with a magnesium ion in the A-site in the PTC (which is aligned with the normal vector to the plane of the opening of binding pocket) thereby disrupting its activity by changing the local conformation of the PTC. It should be noted that the resolution of the toxin in the ribosome was not sufficient enough to determine whether or not any water was included in the binding. In the event that water is present, it is also possible that the water itself, held in place through the bridging interaction, interacts with the magnesium ion. If the magnesium ion were to be sequestered in either capacity, the local conformation in the PTC would be disrupted; hence, the peptidyl transferase activity would be inhibited.

## Implications of Water Binding on the Toxicity Of DON and Related Trichothecenes

The work presented herein provides new high resolution structural and dynamic information for the trichothecene toxin DON. Our results support previous structural works, but also give new insights with respect to the role of water. Of note, is the detailed hydrogen bonding interaction with water which revealed a bridging between 3OH and 15OH. This binding between water and DON is incredibly strong with exchange rates of 0.04– 0.12 s−<sup>1</sup> , which represents a Gibbs energy on the order of 77– 81 kJ mol−<sup>1</sup> . In a previous study (Chaudhary et al., 2011), we reported water binding with T-2 toxin where the 3OH was also shown to be important; however, in this case, the functional group at C<sup>15</sup> is substituted with an acetyl group in T-2 toxin, which prevents the formation of the water bridge observed here for DON.

The nature of the water binding is likely very important for toxicity, and the strength of this interaction with the molecule may explain differences in toxicity. For example, our work shows that 3OH plays an important role in water binding, meanwhile substitution at the 3OH in trichothecenes, such as acetylation or glycosylation, has been shown to reduce or eliminate toxicity (Alexander et al., 1999; Poppenberger et al., 2003; McCormick, 2009). It is possible that these modifications at C<sup>3</sup> increase the number of possible binding modes which are non-optimal for donor-acceptor interactions of the binding pocket with its immediate surroundings, thereby reducing its ability to effectively bind magnesium.

We hypothesize that the role of the epoxide ring in toxicity is in stabilizing the structure of the binding pocket for interactions with water and/or relevant molecules in its vicinity, and that differences in toxicity are related to (a) the reactivity of the binding pocket, which is based on substitution patterns on the trichothecene core, and (b) the orientation of the binding pocket with respect to the magnesium ion in the A-site of the PTC. This hypothesis outlines directions of future inquiries currently under investigation, such as effects of substitutions on water binding, the role of the epoxide ring on water binding, direct donor-acceptor interactions of these toxins with metal centers. The knowledge to be obtained from these studies can be exploited to initiate and/or support the development of detoxification strategies. Strategies to modify the C<sup>3</sup> have previously been proposed to reduce the impacts of trichothecene contamination of grain using transgenic approaches (Muhitch et al., 2000; Alexander, 2008; Shin et al., 2012). The work herein provides the first insights as to why substitutions at C<sup>3</sup> reduce toxicity, namely the role of 3OH in water binding. Further analysis of this interaction may lead alternative strategies to disrupt the interaction with water and could therefore be used in trichothecene remediation from contaminated grains.

### AUTHOR CONTRIBUTIONS

NF contributed to roughly half of the writing of this manuscript and brought her perspective as molecular biologist/biochemist to the study to put things into context with FHB research, bringing biological insights, and discussion from the data that was generated. RS was a M.Sc. student working under the cosupervision of PH and FE. She conducted the first round of NMR experiments and provided some ideas for background discussion. DK is a B.Sc. student who was working under the supervision of PH. He conducted the second round of NMR experiments and provided the majority of the data used for this manuscript. PH contributed to roughly half of the writing of this manuscript, provided detailed analysis of the data, prepared the figures and tables. He provided financial contributions to the project and expertise in chemistry and NMR. FE provided the main source of financial contributions to the project, and contributed expertise in FHB.

### FUNDING

This research was funded by Agriculture and Agri-Food Canada (AAFC) research grants awarded to FE and NSERC grants awarded to PH.

## REFERENCES


#### ACKNOWLEDGMENTS

This research was funded by NSERC and Agriculture and Agri-Food Canada (AAFC). Special thanks to Dr. Barbara Blackwell from AAFC-Ottawa for providing the DON sample and for her insights in NMR of trichothecenes, and the NMR facility manager Tony Montina at the University of Lethbridge for his assistance. From the Department of Chemistry and Biochemistry at University of Lethbridge we thank Dr. Peter Dibble for his valuable input and Dr. Nehal Thakor for his insights and critical reading of this manuscript.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2016 Foroud, Shank, Kiss, Eudes and Hazendonk. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Biologically Based Methods for Control of Fumonisin-Producing Fusarium Species and Reduction of the Fumonisins

#### Johanna F. Alberts <sup>1</sup> , Willem H. van Zyl <sup>2</sup> and Wentzel C. A. Gelderblom<sup>1</sup> \*

*<sup>1</sup> Mycotoxicology and Chemoprevention Research Group, Institute of Biomedical and Microbial Biotechnology, Cape Peninsula University of Technology, Bellville, South Africa, <sup>2</sup> Microbiology Department, Stellenbosch University, Stellenbosch, South Africa*

#### Edited by:

*Daniela Gwiazdowska, Poznan University of Economics, Poland*

#### Reviewed by:

*Jose M. Diaz-Minguez, CIALE - Universidad de Salamanca, Spain Jon Y. Takemoto, Utah State University, USA*

\*Correspondence:

*Wentzel C. A. Gelderblom gelderblomw@cput.ac.za*

#### Specialty section:

*This article was submitted to Fungi and Their Interactions, a section of the journal Frontiers in Microbiology*

Received: *28 January 2016* Accepted: *04 April 2016* Published: *26 April 2016*

#### Citation:

*Alberts JF, van Zyl WH and Gelderblom WCA (2016) Biologically Based Methods for Control of Fumonisin-Producing Fusarium Species and Reduction of the Fumonisins. Front. Microbiol. 7:548. doi: 10.3389/fmicb.2016.00548* Infection by the fumonisin-producing *Fusarium* spp. and subsequent fumonisin contamination of maize adversely affect international trade and economy with deleterious effects on human and animal health. In developed countries high standards of the major food suppliers and retailers are upheld and regulatory controls deter the importation and local marketing of fumonisin-contaminated food products. In developing countries regulatory measures are either lacking or poorly enforced, due to food insecurity, resulting in an increased mycotoxin exposure. The lack and poor accessibility of effective and environmentally safe control methods have led to an increased interest in practical and biological alternatives to reduce fumonisin intake. These include the application of natural resources, including plants, microbial cultures, genetic material thereof, or clay minerals pre- and post-harvest. Pre-harvest approaches include breeding for resistant maize cultivars, introduction of biocontrol microorganisms, application of phenolic plant extracts, and expression of antifungal proteins and fumonisin degrading enzymes in transgenic maize cultivars. Post-harvest approaches include the removal of fumonisins by natural clay adsorbents and enzymatic degradation of fumonisins through decarboxylation and deamination by recombinant carboxylesterase and aminotransferase enzymes. Although, the knowledge base on biological control methods has expanded, only a limited number of authorized decontamination products and methods are commercially available. As many studies detailed the use of natural compounds *in vitro*, concepts in reducing fumonisin contamination should be developed further for application *in planta* and in the field pre-harvest, post-harvest, and during storage and food-processing. In developed countries an integrated approach, involving good agricultural management practices, hazard analysis and critical control point (HACCP) production, and storage management, together with selected biologically based treatments, mild chemical and physical treatments could reduce fumonisin contamination effectively. In rural subsistence farming communities, simple, practical, and culturally acceptable hand-sorting, maize kernel washing, and dehulling intervention methods proved to be effective as a last line of defense for reducing fumonisin exposure. Biologically based methods for control of fumonisin-producing *Fusarium* spp. and decontamination of the fumonisins could have potential commercial application, while simple and practical intervention strategies could also impact positively on food safety and security, especially in rural populations reliant on maize as a dietary staple.

Keywords: Fusarium, fumonisins, prevention, biological control, reduction, sub-Saharan countries

## INTRODUCTION

**Fusarium** spp. are agriculturally important **plant pathogenic fungi** associated with disease and **mycotoxin** contamination of grain crops (Wild and Hall, 2000; Picot et al., 2011). Fusarium ear rot in maize is one of the major diseases affecting maize production worldwide and poses an enormous threat to the international trade of foods and feeds. Fungal species of Fusarium Section Liseola, including Fusarium verticillioides, Fusarium proliferatum, and Fusarium subglutinans are some of the most important causative fungal agents of Fusarium ear or kernel rot as well as symptomless infection of maize crops, leading to contamination with the **fumonisin** mycotoxins (Munkvold et al., 1997).

Fifteen Fusarium spp. have been reported to produce fumonisins. Eight species are from the Section Liseola, i.e., F. verticilloides, Fusarium sacchari, Fusarium fujikuroi, F. proliferatum, F. subglutinans, Fusarium thapsinum, Fusarium anthophilum, and Fusarium globosum (Rheeder et al., 2002). Another five species fall within Section Dlaminia, i.e., Fusarium nygamai, Fusarium dlamini, and Fusarium napiforme. Trace amounts of fumonisin were detected in culture material of two species, i.e., Fusarium andiyazi and Fusarium pseudonygamai. The remaining two fumonisin-producing Fusarium spp. are one species in Section Elegans, i.e., Fusarium oxysporum and one in Section Arthrosporiella, i.e., Fusarium polyphialidicum. The fumonisins are associated with several diseases in humans, animals, poultry, and fish (Marasas, 2001; Marasas et al., 2004; Kimanya et al., 2010) and are classified as Group 2B carcinogens (IARC, 2002). Home-grown maize is a major dietary staple in southern Africa and known to be frequently contaminated with unacceptable levels of fumonisins, with fumonisin B<sup>1</sup> (FB1) being the most prevalent natural occurring fumonisin (Marasas, 2001; Marasas et al., 2004; Shephard et al., 2007, 2013; Burger et al., 2010). The Eastern Cape Province of South Africa is one of the areas in the world where the highest levels of FB<sup>1</sup> were recorded in home-grown maize. As a result exposure to FB<sup>1</sup> in adults is more than four times above the provisional maximum tolerable daily intake (2µg FB1/kg body weight/day) set by the Joint Food and Agriculture Organization of the United Nations and the World Health Organization (FAO/WHO) Expert Committee on Food Additives (Bolger et al., 2001).

The fumonisins comprise a group of 28 characterized analogs, which can be separated into four main groups: fumonisin A, B, C, and P (Rheeder et al., 2002). The fumonisin B (FB) analogs, which includes FB1, FB2, and FB3, are the most abundant naturally occurring fumonisins, with FB<sup>1</sup> predominating and usually being found at the highest levels. Apart from FB, some of the other analogs may occur in naturally contaminated maize at relatively low levels. The complete fumonisin molecule plays an important role in toxic and cancer-initiating activities in vivo (Gelderblom et al., 1993). Studies evaluating the structureactivity relationship of fumonisin analogs, hydrolysis products and a monomethyl ester of FB<sup>1</sup> in short-term carcinogenesis in rats and cytotoxicity assays in primary rat hepatocytes, indicated that the free amino group plays a pivotal role in the toxicological effects of the fumonsins in vitro and in vivo. It was suggested that the tricarballylic acid moiety is required for effective absorption of the fumonisins from the gut. The fumonisins disrupt sphingolipid biosynthesis by inhibiting the enzyme ceramide synthase (Wang et al., 1991), and the tricarballylic acid moiety is required for maximal effect (Van der Westhuizen et al., 1998).

Fusarium infect maize in the field with the highest levels of fumonisins present at harvest, concentrated in the pericarp and embryo of the maize kernel (Fandohan et al., 2006; Kimanya et al., 2008; Burger et al., 2013). Kinetics of Fusarium growth and mycotoxin production are mainly affected by water activity, temperature, and atmospheric composition, while nutritional factors such as kernel endosperm composition and nitrogen sources also play an important role (Chulze, 2010; Picot et al., 2011). Fumonisin production strongly depends on the kernel stage, and may be regulated by physicochemical factors that vary during ear ripening. Insect damage of maize by the European corn borer (Ostrinia nubilalis Hübner) and the corn earworm (Helicoverpa zea Boddie) further favors Fusarium infection (Betz et al., 2000).

**Methods for reduction** of fumonisins in maize are applied pre-harvest or during harvesting and processing (Wild and Gong, 2010). These include several existing strategies to reduce Fusarium growth and production of fumonisins in food sources, i.e., controlled agricultural practices, ensiling strategies, breeding for insect and fungal resistance in maize cultivars, various physical-, chemical-, and biological treatment methods and genetic engineering approaches. Good agricultural management and hazard analysis and critical control point (HACCP) practices promote the general condition of crops, reducing but not eliminating fungal growth, and mycotoxin contamination, while resistance breeding strives to achieve a balance between developing resistant crops and maintaining high quality crop yield (Cleveland et al., 2003; Wild and Gong, 2010). However, optimization of agricultural management practices is not always possible due to high production costs, the geographical location or nature of the production systems, and challenging environmental conditions.

Several physical and chemical control methods for mycotoxins have been commercialized involving sorting and flotation, solvent extraction, chemical detoxification by alkalization (e.g., ammonia, sodium hydroxide, and sulfur dioxide treatments), oxidation (e.g., ozone), and irradiation and pyrolysis (He and Zhou, 2010). There are, however, several limitations, challenges, and concerns with regards to physical and chemical control methods (Schatzmayr et al., 2006). Physical methods generally have low efficacy and less specificity, while chemical methods are not always effective, are considered expensive and may decrease the nutritional value of foods, affect the sensory quality, and could produce toxic derivatives (Alabouvette et al., 2009; He and Zhou, 2010). Furthermore, methods involving fungicides pose a potential health, safety, and environmental risk as certain antifungal chemical compounds are not biodegradable or have a long degradation period, could contaminate soil and water and their effect on food quality and human health is a concern (Larkin and Fravel, 1998; da Cruz Cabral et al., 2013). Prolonged chemical treatment of grains can lead to the development of resistance in fungal strains, a demand for higher concentrations, and an increase in toxic residues in food crops. Increasingly more stringent regulation is enforced with regards to the use of chemical control methods together with a strong consumer demand to reduce the use of potentially harmful chemicals in the food supply (Liu et al., 2013). There is also an ecological and societal movement toward safe and natural food, without chemical treatments and/or preservatives (Edlayne et al., 2009).

Research over the past 25 years indicates support for agricultural management practices and a renewed interest in **practical and biological control methods** as possible alternatives. In this regard several methods for controlling fungal growth and mycotoxin production pre- and postharvest involving clay minerals, plant extracts and a variety of microbial taxa have been commercialized (He and Zhou, 2010). In rural **subsistence farming communities** a number of effective, practical, and culturally acceptable intervention methods have been developed (Kimanya et al., 2008; Van der Westhuizen et al., 2010). While the focus in the past was more on the most economically important mycotoxins, i.e., aflatoxin B<sup>1</sup> (AFB1), much less information is available on other important mycotoxins such as FB1, trichothecenes, zearalenone, citrinin, and patulin (Kabak et al., 2006). This paper presents a comprehensive overview of recent research on biological- and practical-based approaches for control of fumonisin-producing Fusarium spp. and **methods for reduction** thereof during pre- and post-harvest conditions. Current information on the application of natural clay adsorbents, biocontrol organisms, antioxidants, essential oils, plant extracts, and molecular approaches are reviewed; as well as practical and culturally acceptable methods for reduction of fumonisin exposure in rural subsistence farming communities.

## PRE-HARVEST BIOLOGICALLY BASED CONTROL METHODS FOR FUMONISIN-PRODUCING FUSARIUM Spp.

#### Biocontrol Microorganisms

This approach involves a three-way interaction between the host commodity, the pathogen and the antagonistic biocontrol microorganism together with dynamics such as competition for nutrients and space, parasitism of the pathogen, secretion of antifungal compounds, induction of systemic resistance (ISR), biofilm formation and involvement with reactive oxygen species in defense response (Larkin and Fravel, 1998; Alabouvette et al., 2009). Recent research also suggested that the aflatoxin biocontrol mechanism, employing atoxigenic strains of Aspergillus flavus, is triggered by physical contact or interaction between hyphae of the competing fungal strains (Damann, 2014). Essential criteria for effective biocontrol microorganisms include the ability to colonize the plant part infected by the pathogen organism, efficacy under the relevant environmental conditions and compatibility with other control methods that are applied (Bacon and Hinton, 2011; Liu et al., 2013). Niche overlap indices (NOIs) provide information on ecological similarity, coexistence, and competition between microorganisms in a specific niche and assists in identifying possible microbial antagonists against F. verticillioides colonization (Cavaglieri et al., 2004). Microorganisms naturally associated with and adapted to the vegetative parts of a specific plant, sharing the ecological niche with pathogen microorganisms, could hold advantages as biocontrol agents. One such a microorganism, Bacillus subtilis occupies the same ecological niche as F. verticillioides within the maize plant and effectively inhibits growth of the fungus, based on competitive exclusion (Bacon et al., 2001; **Table 1**). B. subtilis is considered generally regarded as safe (GRAS) by the United States Food and Drug Administration [US FDA, GRAS substances evaluated by the Select Committee on GRAS substances (SCOGS)], is easy to cultivate and manipulate genetically, and therefore suitable for industrial processes. A preharvest biological control system, involving B. subtilis RRC101, was developed on maize which reduces fumonisin accumulation during the endophytic growth phase of F. verticillioides (= F. moniliforme; Bacon et al., 2001). The endophytic phase of F. verticillioides is transferred vertically to the next generation through clonal infection of seeds. This phase is characterized by intercellular systemic infection of plants and seeds, which cannot be controlled with fungicides. Effective biocontrol has also been demonstrated with wild type and fusaric acid resistant mutant strains of the bacterial endophyte, Bacillus mojavensis, in vitro and in planta (Bacon and Hinton, 2011). Efficacy of these strains under field conditions could be influenced by fusaric acid produced by F. verticillioides. The mechanism of biocontrol by B. mojavensis is complex and still unclear, as indicated by broad differences in maize seedling protection by a range of strains evaluated.

Pediococcus pentosaceus, a lactic acid bacterial isolate from maize, inhibits F. verticillioides and F. proliferatum growth in vitro (Dalie et al., 2010; **Table 1**). Antifungal activity in P. pentosaceus culture supernatant was observed toward the end of the exponential phase of growth and was pH dependent. The antifungal metabolites produced proved to be heat stable and resistant to proteolytic enzymes. Culture fractions exhibiting antifungal activity contained compounds with molecular masses ranging from 500 to 1400 Da. P. pentosaceus has GRAS status, has been widely used in the fermentation of a variety of foods and could be suitable as biocontrol organism to improve the quality of ensilage. Clonostachys rosae, a fungal isolate from straw, stubble, seed surfaces, and the phylosphere or roots of cereal crops, effectively reduced sporulation of F. verticillioides and F. proliferatum on maize stalks in vitro and in field trials


 | Current information on reduction of fumonisin-producingFusariumspp. by biocontrol microorganismsin vitro, in planta, and

 in field trials.

TABLE

1

*(Continued)*


Frontiers in Microbiology | www.frontiersin.org April 2016 | Volume 7 | Article 548 |

TABLE

1


Continued

**80**

*(Continued)*



*FB*1*, Fumonisin B*1*; CFU, Colony forming units; PDA, Potato dextrose agar; MRS broth/agar, de Man, Rogosa and Sharpe broth/agar.*

TABLE

1


(Luongo et al., 2005). C. rosae exhibited potential to control Fusarium spp. in maize at the flowering ear stages and in crop residues post-harvest. Food-grade yeasts are also considered ideal biocontrol microorganisms, as they are generally genetically stable, effective at low concentrations, easy to cultivate, capable to survive under adverse environmental conditions, compatible with commercial processing, and resistant to pesticides.

#### Trichoderma spp.

Trichoderma spp. are considered effective biocontrol agents because of their repertoire of extracellular lytic enzymes that cause necrotrophic action through lysis of fungal cell walls as well as the role they play in ISR in plants (Bacon et al., 2001; Hermosa et al., 2012). Trichoderma mainly colonizes the rhizosphere and intercellular root areas of plants, and maintains interactions by promoting plant growth and providing protection against infections, while utilizing plant sucrose to facilitate root colonization (Hermosa et al., 2012). Plant disease severity is reduced in the presence of Trichoderma by inhibition of a wide range of plant pathogens through antagonistic and mycoparasitic action; ISR or induction of localized resistance. Trichoderma is also able to withstand toxic metabolites that are produced by the plant in response to invasion. Plants are able to detect pathogen- or microbe associated molecular patterns (MAMPs), which leads to activation of defense mechanisms and eventually synthesis of antimicrobial compounds. Certain Trichoderma strains produce a variety of MAMPs, contributing to activation of plant defense responses. Salicylic acid, jasponic acid and ethylene play a key role in plant immunity and hormone-signaling pathways as well as defense response pathways of the hormones abscisic acid, indole-3-acetic acid, and gibberellin (Pieterse et al., 2009). Indole-3-acetic acid produced by Trichoderma contributes to ethylene biosynthesis, which in turn stimulates abscisic acid biosynthesis. Depending on Trichoderma stimuli, phytohormone homeostasis will control plant development and immune responses. Trichoderma chitinases also release fungal chitin oligosaccharides, and elicit ISR by jasmonic acid/ethylene dependent pathways, thereby triggering defense responses in plants. A polyketide synthase/non-ribosomal peptide synthetase hybrid enzyme of Trichoderma virens is involved in plant interactions and was shown to induce plant defense responses (Mukherjee et al., 2012). Several Trichoderma spp. with GRAS status, including Trichoderma viride and Trichoderma harzianum, are capable of effectively reducing F. verticillioides (= F. moniliforme) growth and fumonisin production in vitro and in planta (Calistru et al., 1997; Larkin and Fravel, 1998; Yates et al., 1999; **Table 1**). The inhibitory effect on F. verticillioides growth when co-cultured with Trichoderma spp. can be attributed to antibiosis through production of volatile compounds, extracellular enzymes and antibiotics. The antagonistic fungal species T. viride is widely used in biofertilizers for biological control of soil borne plant-pathogenic fungi in crops.

### Non-Pathogenic Biocontrol Strains

Non-pathogenic strains of pathogenic species are often applied for biocontrol (Liu et al., 2013). In this regard, moderate

The development of Fusarium biocontrol strains with reduced mycotoxin production ability through RNA silencing technology may be a useful tool for reducing mycotoxin contamination in agricultural products (McDonald et al., 2005). Transformation of F. graminearum with inverted repeat transgenes (IRT) containing sequences of mycotoxin-specific regulatory genes results in suppression of mycotoxin production. Other gene silencing techniques involving deletion of ZFR1 of F. verticillioides, which regulates sugar transporter genes and in turn affect fumonisin biosynthesis during kernel colonization, resulted in significantly less growth on maize kernel endosperm tissue (Bluhm et al., 2008).

## Rhizobacteria

Fusarium verticillioides is the most prevalent Fusarium spp. present in the rhizoplane and endorhizosphere areas of maize, while Arthrobacteria and Azotobacter are the predominant bacterial genera (Cavaglieri et al., 2005a). Pathogens germinate and colonize roots within a few days of planting, while biocontrol rhizobacteria could be metabolically active during this period. A number of rhizobacterial isolates of maize plants sampled from a commercial maize field and exhibiting high NOIs with F. verticillioides, including Arthrobacter globiformis, Azotobacter armeniacus, Pseudomonas solanacearum, B. subtilis, Enterobacter cloacae, and Microbacterium eoleovorans exhibited antifungal activity in vitro by effectively reducing F. verticillioides growth and FB<sup>1</sup> production on maize meal extract agar (Cavaglieri et al., 2004, 2005a,b,c) (**Table 2**). Maize seeds pre-treated with A. armeniacus RC2, A. globiformis RC5, E. cloacae, M. eoleovorans, and Bacillus sp. CE1 and evaluated in planta, resulted in effective reduction of F. verticillioides growth in the rhizoplane and endorhizosphere areas. A good correlation was observed between results obtained from in vitro and in planta studies (Cavaglieri et al., 2005c). Enterobacter cloacae exhibited potential for biocontrol of root colonization by F. verticillioides. Inducible Type 1 fimbrae of E. cloacae may play a role in the colonization of roots (Hinton and Bacon, 1995). Rhizobacterial strains could have potential application as seed inoculants to reduce F. verticillioides colonization on root level, in the rhizoplane and endorhizosphere areas (Cavaglieri et al., 2005c). Effectiveness of a biocontrol organism to colonize the rhizosphere and its value as biocontrol agent could, however, be influenced by environmental conditions and the initial cell concentrations of the biocontrol organism and the pathogen.

## Antioxidants, Phenolic Compounds, and Essential Oils

Several natural phenolic compounds derived from plants are strong antioxidants and exhibit antimicrobial activity by inhibiting the activity of key fungal enzymes, and are applied as preservatives in the cosmetic, food and drug industries (**Table 3**). These compounds are also considered promising antifungal agents for controlling fungal growth and associated mycotoxin production in agricultural crops pre-harvest, post-harvest, and during storage.


TABLE 2 | Current information on reduction ofFusarium verticillioidesgrowth and fumonisin B1production by rhizobacteriain vitroandin

 planta.

*(Continued)*


TABLE

2


Continued

*(Continued)*


**86**



TABLE

3



TABLE

3


Continued

*(Continued)*


*(Continued)*

TABLE

3



*(Continued)*

TABLE

3



**92**

#### Frontiers in Microbiology | www.frontiersin.org April 2016 | Volume 7 | Article 548 |


 *US FDA, United States Food and Drug Administration; GRAS, Generally regarded as safe; MIC, Minimum inhibitory concentration; UV, Ultra violet; THCs, Tetrahydrocurcuminoid compounds; MMEA, Maize meal extract Czapek-Dox agar.*

TABLE

3


## Antioxidants

The food-grade antioxidants butylated hydroxyanisole (BHA) and propylparaben (PP) have shown potential for controlling F. verticillioides and F. proliferatum growth and fumonisin production at a variety of water activities and incubation temperatures in vitro (Etcheverry et al., 2002; **Table 3**). Both fungal species were more sensitive to BHA and PP than the other antioxidants evaluated, i.e., trihydroxybutyrophenone (THBP) and butylated hydroxytoluene (BHT). In another study, combination treatments of BHA and PP resulted in further reduction of fumonisin production (Reynoso et al., 2002). BHA, PP, and BHT alone or in combination also resulted in a significant (P < 0.001) reduction in hydrolytic enzyme activity, which is required for early fungal growth. Similar results were reported by Torres et al. (2003). BHA is produced naturally by Botryococcus braunii, Cylindrospermopsis raciborskii, Microcystis aeruginosa, and Oscillatoria sp., while PP is a natural compound extracted from plants. Both antioxidants are also produced synthetically, are considered GRAS by the US FDA and frequently employed as preservatives in the food and cosmetic industries (Reynoso et al., 2002; Rawal et al., 2010; US FDA, GRAS substances evaluated by SCOGS).

Tetrahydrocurcuminoids (THC), a class of phenolic antioxidants extracted from the roots of the non-toxic herbaceous plant Curcoma longa L. (Turmeric), inhibits F. proliferatum growth and FB<sup>1</sup> production in vitro (Coma et al., 2011; **Table 3**). THC1, a food-grade compound containing two guaiacyl phenolic subunits, exhibited high antifungal activity and inhibition of FB<sup>1</sup> production in liquid cultures at low inhibitory concentrations. FB<sup>1</sup> production was affected irrespective of the effect on fungal growth, indicating that fungal growth and FB<sup>1</sup> biosynthesis are independently modified by THC1. Comparative studies on THCs and related molecules n-propylguaiacol, eugenol, acetylacetone, and ferulic acid indicated that the presence of the benzene rings and guaiacyl groups play an important role in fungal inhibition (Beekrum et al., 2003; Samapundo et al., 2007). It was further noticed that the presence of hydroxyl and methoxy groups in the ortho position of the benzene ring of THC molecules affects the degree of antifungal activity, while the enolic part of the non-phenolic THC3 molecule could play a role in bioactivity. It was suggested that the biochemical mechanisms involved during antioxidant and antifungal activities differ between the respective THC compounds, as the presence of a phenol group in the meta- or para-position of the linking chain and a phenol or a methoxy group adjacent to it is required for antioxidant activity.

## Phenolic Compounds

Investigations into the effects of the natural phenolic compounds vanillic and caffeic acid on F. verticillioides and F. proliferatum growth and FB<sup>1</sup> production at different water activities in maize in vitro indicated that an increase in phenolic compound concentration results in an increase in the lag phase of growth, and a decrease in fungal growth rate and FB<sup>1</sup> production (Samapundo et al., 2007; **Table 3**). In general, complete inhibition of Fusarium growth was observed at relatively high phenolic concentrations and low water activities. F. proliferatum was more sensitive, exhibiting complete inhibition of growth in the presence of the compounds. Both compounds significantly reduced FB<sup>1</sup> production by F. verticillioides and F. proliferatum, with vanillic acid being more effective. No FB<sup>1</sup> was produced by F. verticillioides in the presence of vanillic acid at the lowest concentration tested.

F. verticillioides growth and FB<sup>1</sup> production are inhibited by several other plant phenolic compounds in vitro (**Table 3**). Chlorophorin, iroko, maakianin, vanillic acid, and caffeic acid inhibits F. verticillioides growth, while FB<sup>1</sup> production is inhibited by chlorophorin, iroko, vanillic acid, caffeic acid, and ferulic acid (Beekrum et al., 2003; **Table 3**). Flavonoids, phenolic acid, and terpine rich 70% ethanol extracts of the non-toxic food-grade plants Equisetum arvense (Horsetail) and Stevia rebaudiana (Candyleaf), effectively inhibited F. verticillioides growth, with S. rebaudiana being more effective (Garcia et al., 2012). However, fumonisin production was not affected. Extracts of the herbaceous climbing vine of the family Cucurbitaceae, Gynostemma pentaphyllum (Southern Ginseng), inhibited growth of F. verticillioides (Srichana et al., 2011). G. pentaphyllum is frequently applied as herbal medicine and exhibits high antioxidant activity. Fumigation by trans-2-hexanal (extracted from fruits and vegetables), carvacrol (extracted from oregano and thyme), and eugenol (extracted from cinnamon and clove) effectively inhibits F. verticillioides conidial germination and mycelial growth in maize kernels, with trans-2-hexanal the most effective (Menniti et al., 2010). Trans-2-hexanal fumigation was also effective in controlling the fungus in asymptomatic kernels. However, the treatment does not reduce fumonisin levels post-harvest, but reduces the germ-ability of maize kernels. The compound 6,7-dimethoxycoumarin, occurring in Penicillium digitatum infected Citrus sinensis cultivar Valencia fruit (Valencia orange), reduces F. verticillioides growth and FB<sup>1</sup> production (Mohanlall and Odhav, 2006). Possible mechanisms of inhibition by phenolic plant extracts include disruption of the fumonisin biosynthetic pathway; effects on colony morphology; granulation of the cytoplasm; and rupture of the cytoplasmic membrane (Garcia et al., 2012).

## Essential Oils

Essential oil and oleoresins extracted from Zingiber officinale (Ginger) rhizomes exhibit clear antimicrobial activity against F. verticillioides (= F. moniliforme) in vitro (Singh et al., 2008; **Table 3**). Ginger oil and carbon tetrachloride oleoresin extracts have shown highly effective inhibition of F. verticillioides growth. The antioxidative potential of the essential oil and oleoresins, in terms of peroxide content, anisidine and thiobarbituric acid values, 1,1-diphenyl-2-picrylhydrazyl free radical scavenging activity and total antioxidant activity was in general comparable to the antioxidants BHA and BHT, but not as effective as propyl gallate. The phenolic compound geranial is dominant in the essential oil component, while eugenol and singerone are dominant in the oleoresin extracts. The antioxidant activity could also be enhanced by a possible synergistic effect of the phenolic compounds.

Essential oils extracted from cinnamon, clove, oregano, palmarosa and lemongrass inhibit growth and FB<sup>1</sup> production by F. verticillioides and F. proliferatum in vitro (Velluti et al., 2003; **Table 3**). The inhibitory effect of the essential oils was overall more pronounced at higher water activities, probably due to more effective penetration of oils into kernels in the presence of water. The antimicrobial activity of these oils could be attributed to the presence of aliphatic alcohols and phenols in their chemical composition. Oils of cinnamon and oregano were most promising for control of fungal growth and FB<sup>1</sup> production by F. proliferatum, and cinnamon, oregano and lemongrass oils for F. verticillioides. These oils could be effective in controlling fungal growth and FB<sup>1</sup> production in maize under pre-harvest conditions.

#### Developing Resistant Crops through Breeding and Genetic Engineering

Studies in breeding and genetic engineering for resistance in crops are mainly aimed at preventing invasion by insects, contamination by mycotoxigenic fungi and detoxification of mycotoxins in planta through various molecular strategies (Duvick, 2001; Cleveland et al., 2003). Selection of resistant genotypes is complex, it requires sufficient genotypic variation within the breeding material; is affected by climatic conditions; and should be tested across several locations and years (Löffler et al., 2010). Lower mycotoxin levels measured in United States and Canadian maize, where no fungicide was introduced, was attributed to successes with breeding resistant maize varieties.

Extensive genomic resources are essential for investigations into the biochemical and regulatory pathways of mycotoxin biosynthesis, pathogenesis of fungal–plant interactions, and the development of targeted and innovative approaches for breeding and engineering crops for resistance (Cleveland et al., 2003; Brown et al., 2006; Desjardins and Proctor, 2007). Whole genome sequences and expression sequence tags (ESTs) are important tools for understanding disease caused by fungi, fungal lifecycles and secondary metabolism. Available genomic resources include genetic maps, genome sequences, an EST library, and an integrated gene index. Next-generation RNA sequencing was used to study transcriptional changes associated with F. verticillioides inoculation in resistant and susceptible maize genotypes by including an extensive range of maize inbred lines (Lanubile et al., 2014). The technique generated extremely useful data on genetic markers involved in recognition, signaling, and controlling host resistance mechanisms. It also provided quantification of expression, thus enabling interpretation of defense responses. The data provides an important genomic resource for the development of disease resistant maize genotypes. Genetic markers identified through this technique could be added to existing information on single nucleotide polymorphism markers.

#### Natural Resistance in Crops

Comprehensive knowledge on the biochemical and molecular mechanisms involved in natural resistance of crops is imperative for the further development of resistance to Fusarium infection and insect infestation in crops (Cleveland et al., 2003). The whole genome sequence of maize is available (Schnable et al., 2009), permitting genome-wide expression analysis of the maize– Fusarium interaction. Studying maize varieties with varying degrees of resistance enables researchers to associate resistant crops with specific genetic, biochemical and anatomical traits. Regions on chromosomes associated with natural resistance to insect invasion, fungal contamination, or mycotoxin production are identified, resistant traits mapped and resistant lines crossed with commercially acceptable lines. Chromosomal regions could be associated with resistance to fungal growth; with mycotoxin production; or with both traits, indicating the possibility of separate genetic control (Cleveland et al., 2003). Comparison of kernel protein profiles between susceptible and resistant genotypes through proteomic analyses contributes to identifying resistance associated proteins. Resistant inbred lines are distinguished from susceptible lines and serve as sources of resistant germplasm.

Expression profiles for maize genes during infection with F. verticillioides indicated up-regulation of genes encoding a range of proteins related to cell rescue, defense, and virulence in both resistant and susceptible maize lines, including pathogenesis related (PR) proteins [e.g., chitinase (reducing chitin in fungal membrane); permatin (fungal hyphae leak and rupture)]; proteins involved in detoxification response (e.g., cytochrome P450 monoxygenase, peroxidases, and glutathione-S-transferases); heat-shock proteins (regulating folding of resistance proteins); and proteinase inhibitors (Lanubile et al., 2010). Resistance in maize lines could be due to constitutive defense mechanisms that resist fungal infection (Lanubile et al., 2010; Campos-Bermudez et al., 2013). In resistant maize lines defense-related genes, encoding constitutively expressed PR, detoxification enzymes, and β-glucosidases, were transcribed at high levels before infection, and provided defense against the fungus. In susceptible maize lines, defense genes are induced as a response to pathogen infection, though not sufficiently enough to prevent progress of the disease.

Host–pathogen recognition and interaction processes underlie resistance and susceptibility (Campos-Bermudez et al., 2013). Sucrose is one of the compounds that play an important role in host-pathogen recognition and in the outcome of interactions. During fungal infection plant carbohydrate metabolism is manipulated by induced invertase and sucrose synthase enzymes and the formation of hexoses required for fungal growth. Maize lipoxygenase (ZmLOX) derived oxylipins (e.g., jasmonic acid) are known for regulating plant defense against pathogens, and also play an important role in recognition during host-pathogen interactions, as indicated by up-regulation of LOX genes ZmLOX5 and ZmLOX12 in a response to F. verticillioides infection (Maschietto et al., 2015).

Mapping of chromosomal regions encoding Fusarium ear mold resistance as quantitative trait loci (QTL) and the employment of marker-assisted QTL in selection for Fusarium ear mold resistance are valuable tools being developed for maize hybrid development (Duvick, 2001). Ear mold resistance can be mapped as QTL using large segregating plant populations. Molecular markers linked to these QTL could be valuable during inbred development. Other factors that enhance the susceptibility of maize genotypes include: late-maturing cultivars where grain moisture content decreases slowly; upright cobs and thin grain pericarp which increase susceptibility to fungal infection; tightness of husks; and the competitive advantage of F. verticillioides by having a broader optimum temperature range than F. graminearum (Butròn et al., 2006).

## Genetic Engineering for Resistance to Insect Infestation and Fusarium Infection in Crops

Natural fungal and insect resistance mechanisms could be further enhanced in commercially acceptable crops through genetic engineering (Cleveland et al., 2003). The role of hemicellulose, cysteine protease, peroxidase, α-amylase inhibitors, as well as maize ribosomal inactivating protein in insect resistance mechanisms are important focus areas. Genetically modified Bt maize expressing cry proteins from the bacterium Bacillus thuringiensis, has the potential to reduce insect damage and fumonisin levels compared to non-Bt hybrids. Furthermore, chitinase enzymes for digestion of chitin, an integral part of the exoskeleton of insects, have been applied for control of Sesamia cretica (corn borer; Osman et al., 2015). A chitinase gene from the cotton leaf worm, Spodoptera littoralis, was expressed in transgenic maize, and resulted in enhanced resistance against S. cretica. The development of transgene resistance to fungal disease appears to be more challenging than insect resistance (Duvick, 2001). Although, moderate resistance was demonstrated in model systems, no transgenic crops with effective resistance to fungal disease are commercially available. However, genetics of Fusarium infection of maize kernels, development of disease symptoms and biosynthesis of fumonisins is a rapid developing field and could provide more insights for developing transgenic resistance to Fusarium infection in the near future.

Genetic engineering approaches include the cloning and expression of genes encoding maize secondary metabolites with antifungal properties and the overexpression of pathway-limiting enzymes (Duvick, 2001). However, it should be kept in mind that diversion of metabolic pathways could compromise other vital biosynthetic routes. Expression of antifungal protein in tissue critical for fungal infection could be a strategy, while different types of resistance could be employed by pyramiding different types of resistance genes into commercial germplasm. Host plant–pathogen interactions are complex, involving multiple proteins and metabolites as well as competition for biomass and nutrients. Signaling pathway genes control a variety of cellular defense pathways involving protein-protein interactions. Engineering of the main signals controlling defense gene expression could result in more effective defense response including constitutive response or a chemically induced response and the development of enhanced disease resistance phenotypes.

Another approach involves the expression of catabolic enzymes to detoxify mycotoxins in situ before it accumulates in the plants (Duvick, 2001). Success depends on several factors: the extent to which the plant-produced enzyme reaches its target substrate and the stability of the detoxification step; enzyme localization in the seed in relation to mycotoxin accessibility; kinetic parameters of the enzyme in the context of its localization in the plant; stability and activity of the enzyme pre- and postharvest; and the identity and toxicity of breakdown products.

#### Bt Maize

Genetic modification of maize plants to express insecticidal Cry proteins of Bacillus thuringiensis (called Bt maize) provides a safe and highly effective method for insect control and accompanying Fusarium infection and fumonisin production (Betz et al., 2000). Corn borers cause considerable damage to maize stalk and ear tissue, which in turn stimulates germination of F. verticillioides spores, leading to progressive ear and kernel rot and eventually production of increased levels of fumonisins. A significant correlation was reported between the extent of insect damage and total fumonisin levels in maize (Dowd, 2001). Cry1Ab protein in Bt protected maize reduces corn-borer damage in maize dramatically, resulting in considerable less Fusarium infection and reduced fumonisin levels (Betz et al., 2000). Cry proteins are selectively active against a specific range of insects including lepidopteron and coleopteran insect pests. Extensive field trials across the USA and Europe confirmed frequently lower fumonisin concentrations detected in maize using Bt maize hybrids (Hammond et al., 2004), thereby increasing the percentage maize grain suitable for human consumption. In South Africa, there has been a decrease over the last 20 years in the amount of chemical insecticides used, due to the cultivation of Bt crops (Kunert, 2011). In the US States the annual benefits that Bt maize provides in terms of lower fumonisin and aflatoxin contamination are estimated at about \$23 million (Wu, 2006). Bt maize could especially be a useful tool in developing countries.

The insecticidal nature of the Cry proteins has led to the development of a variety of commercial Bt microbial pesticide products since 1961 (Betz et al., 2000). Extensive toxicological studies by the US Environmental Protection Agency (EPA) and the World Health Organisation (WHO) have proven the safety of Bt protected crops and products to humans, animals and the environment [US EPA, 1998a,b; International Programme on Chemical Safety (IPCS), 1999]. Food derived from Bt crops has also been fully approved by numerous regulatory agencies through-out the world. Safety considerations were further supported by the more than 50 years history of safe use of these products (McClintock et al., 1995). The potential for human and non-target exposure is extremely low, as Cry proteins exhibit a high degree of specificity toward the target insect species, should be ingested to activate in the target species and should have no contact activity (Betz et al., 2000). Bt products are considered to reduce the risks posed by insecticides, thereby impacting less on the environment. It also functions as a supplementary pest control by enhancing the presence of beneficial natural occurring non-target insects (Gianessi and Carpenter, 1999). The cultivation of Bt protected maize by growers increased rapidly throughout the world since its commercial introduction in 1996 (Betz et al., 2000). Grower approval could be ascribed to increased crop yields, reduced crop damage and input costs as a result of reduction in the use of chemical pesticides; and highly effective pest control. Cry proteins in the plant tissue are not affected by application timing, accuracy of application, concentration, rain or sunlight. Bt crops are entirely equivalent to non-recombinant

plants, except for the presence of cry genes and proteins. Bt protected crops and products meet important standards for biological control agents regarding technical viability, need, safety and efficacy.

Recently, increasing insect resistance and accompanied occurrence of resistance alleles in insects against first generation Bt crops have been reported (Kunert, 2011; Abbas et al., 2013). Efforts to reduce the development of target insect resistance to Bt crops include introduction of a refuge strategy, which involves the cultivation of non-Bt crops nearby Bt crops to prevent domination of resistant insect species. The effectiveness of Bt crops is also influenced by fluctuation of the Bt protein concentrations produced in plants, which in turn is determined by factors such as plant maturation and photosynthesis. Possible structural changes of Bt proteins, including changes in micro-RNA and protein profiles were also reported. Bt maize genotype plays a determining role in the efficacy of insect damage control (Clements et al., 2003). Bt (Cry1Ab protein) protected plants could reduce fumonisin concentration in maize during seasons when the European corn borer (O. nubilalis Hübner) dominates, but not in seasons when the corn earworm (H. zea Boddie) dominates. Tende et al. (2010) evaluated sensitivity of the stalk borer species Chile partellus (Lepidoptera, Crambidae) and Busseola fusca (Lepidoptera, Noctuidae) toward endotoxins constitutively produced by two Bt maize inbred lines frequently cultivated in Kenya. The Bt maize inbred lines (Event 223 cry1AB::Ubiquitin and Event 10 cry1Ba::Ubiquitin) reduced C. partellus survival significantly and sensitivity remained constant through eight generations. However, B. fusca invasion could not be sufficiently controlled by these inbred lines and remained unchanged through five generations. More efficient transgenic Bt crops could be produced through gene pyramiding (Kunert, 2011).

### POST-HARVEST BIOLOGICALLY BASED CONTROL METHODS FOR REDUCTION OF THE FUMONISINS IN FOOD AND FEED

#### Natural Clay Adsorbents

Introduction of natural clay adsorbents during food processing leads to detoxification of contaminated food through adsorption of mycotoxins (Aly et al., 2004; Robinson et al., 2012). The bioavailability of mycotoxins in animal feed is also reduced in this manner, thereby preventing toxic interactions and absorption across the gastrointestinal tract.

Montmorillonites are a group of phyllosilicate clay minerals that have the ability to adsorb organic compounds through cation-exchange (Aly et al., 2004). The adsorption abilities of montmorillonite clays are higher than other clay minerals due to their large molecular structure and surface area that increases considerably when wet. Their chemical structures are characterized by alternating layers of tetrahedral silicon and octahedral aluminum coordinated with oxygen atoms. Montmorillonite clay minerals effectively reduce FB<sup>1</sup> in aqueous solutions in vitro, and in human- and animal models in vivo through adsorption (**Table 4**). The adsorption is saturable and occurs largely within the interlaminar regions of the clay (Mitchell et al., 2013). Certain clay minerals, particularly naturally occurring aluminum oxides have structure-selective affinities for different mycotoxins and the degree of adsorption depends on the polarity of the molecules, while the particle size of clays could also influence binding affinity (He and Zhou, 2010). A correlation exists between the binding capacity of the clays and the ratio of their surface acidity to pore volume. In this regard, the slightly higher adsorption of AFB<sup>1</sup> than FB<sup>1</sup> to hydrated sodium calcium aluminum magnesium silicate hydroxide (Egyptian montmorillonite, EM) and hydrated sodium calcium aluminum silicate (HSCAS) in spiked malt extracts, could be ascribed to the difference in polarity between the molecules (Aly et al., 2004). The adsorption capacity of montmorillonite clays can be enhanced by addition of phosphate and polyphosphate salts, bentonite, or calcined attapulgite (He and Zhou, 2010). A combination of clay minerals (1–10%) and modified yeast cell wall extracts (90–99%) could be beneficial for adsorption of multiple mycotoxins, including the fumonisins (Howes and Newman, 2000).

Because natural clay mineral adsorbents are considered GRAS by the US FDA (2015), they could be applied effectively and economically in the food and feed industries and several clay minerals have been proven to be acceptable for commercial uses [US FDA, GRAS substances evaluated by the Select Committee on GRAS substances (SCOGS); He and Zhou, 2010]. However, application of clay minerals often requires high levels to be included into animal feed; interaction of natural clays with food- and gut-based nutrients remains unclear; and the possibility of accumulation of dioxin (a toxic trace component in montmorillonite) in animals remains a concern.

## Microbial Transformation of the Fumonisins

Development of control methods to detoxify the fumonisins through transformation should be directed toward deamination of the free amino group at C-2 and hydrolysis of the ester bonds at C-14 and C-15 (Gelderblom et al., 1993). Microorganisms capable of transforming FB<sup>1</sup> to less toxic end products include Exophiala spinifera ATCC 74269, Rhinocladiella atrovirens ATCC 74270, Bacterium ATCC 55552, and Sphingopyxis macrogoltabida MTA144 (Duvick et al., 1998a,b; Blackwell et al., 1999; Heinl et al., 2010). Transformation of FB<sup>1</sup> by the black-yeast E. spinifera was mainly achieved through decarboxylation by inducible extracellular esterase enzymes and amino oxidases converting hydrolysed fumonisin (HFB1) to unknown end products. Degradation by Bacterium ATCC 55552 and S. macrogoltabida MTA144 is achieved through de-esterification by carboxylesterases and subsequent deamination of HFB<sup>1</sup> by aminotransferases, with the formation of 2-keto HFB<sup>1</sup> (Heinl et al., 2010; Hartinger et al., 2011). The microbial gene sequences coding for these enzymes were determined by employing degenerate polymerase chain reaction (PCR) primers, inverse PCR and gene walking techniques. Carboxylesterase (FumD) and aminotransferase enzymes (FumI) of S. macrogoltabida MTA144 and Bacterium ATCC 55552 were expressed in Pichia pastoris



*HSCAS, Hydrated sodium calcium aluminum silicate; EM, Egyptian montmorillonite (Hydrated sodium calcium aluminum magnesium silicate hydroxide); NS, Calcium montmorillonite; UPSN, calcium montmorillonite Uniform particle size Novasil; FB*1*, Fumonisin B*1*; UFB*1, *Urinary FB*1;*AFB*1*, Aflatoxin B*1*; AFM*1*, Aflatoxin M*1*; Bw, Body weight.*



TABLE

5



TABLE

5


Continued

*(Continued)*


TABLE 5 | Continued

*FB*1*, Fumonisin B*1*; FB*2*, Fumonisin B*2*; FB*3*, Fumonisin B*3*; UFB*1*, Urinary FB*1*; UFB*1*C, urinary FB*1*creatinine; DON, deoxynivalenol; PDI, Probable daily intake; PMTDI, Provisional maximum tolerableFAO/WHO Expert Committee on Food Additives.* \**, Indicates combined treatments.*

and E. coli, respectively, by employing episomal pET-3a vectors. Production of the recombinant enzymes were induced in liquid cultures by isopropyl-beta-D-thiogalactopyranoside, where after degradation of FB<sup>1</sup> and HFB<sup>1</sup> was demonstrated with the recombinant culture supernatant as well as with purified enzyme preparations. HFB<sup>1</sup> prepared through enzymatic transformation by FumD carboxylesterases exhibited considerable less toxicity than FB<sup>1</sup> when evaluated in a pig intestine model as indicated by the modified sphinganine/sphingosine ratios in the liver and plasma, modified intestinal immune response, and absence of hepatotoxicity and impaired intestinal morphology (Oswald et al., 2012). Although, certain of these technologies are considered safe for humans, animals and the environment by the European Food Safety Authority (EFSA), applications of microbial enzymes are presently mainly directed toward the animal feed industry (Duvick et al., 1998b, 2003; Moll et al., 2011). Recombinant enzymes are mass produced in a bioreactor and are applied during storage and food-processing to incorporate into animal feed and act in the intestinal tract of animals, or for treatment of grains in the form of a wash, additive or spray. Other post-harvest methods involving microbial transformation include the engineering of ruminal organisms and supplementation to feed in the form of a probiotic inoculant.

## Commercialization of Biological Methods of Control

The lack of effective and environmentally safe chemical control methods against fungal growth and mycotoxin production in food crops has led to investigations into biologically safe alternatives to prevent these contaminants from entering the food chain (Beekrum et al., 2003). Biological pesticides and methods involving natural resources such as plants, microorganisms, genetic factors thereof, and clay minerals are popular alternatives being evaluated for control of mycotoxigenic fungi in grains (Alabouvette et al., 2009). Fusarium growth and fumonisin production pre-harvest and post-harvest are effectively reduced by several natural and biological methods involving plant material, microorganisms and minerals, as evident by the extensive research done on this subject in recent years.

Several commercial products for biological control of Fusarium diseases and the fumonsins have been developed for application alone, in combination or as part of an integrated control strategy. Products containing biocontrol microorganisms are mainly aimed at application as seed and soil treatments as outlined by Fravel et al. (1998) and Kahn (2013):


Although, there is an increased interest in biological control methods, much effort is put into details of natural compounds capable of controlling fungal growth and mycotoxins in vitro. However, the growing knowledge base on this subject should be further developed for application in planta and in the field preharvest, post-harvest, and during storage and food-processing. In order to develop the available information into appropriate methods for application in planta and in the field, there are many economic and technological hurdles to overcome. The effectiveness of antioxidants, essential oils, phenolic compounds and combinations for example, has been demonstrated at laboratory scale, and bioactivity in the vapor phase makes it promising as fumigant for protection of grains on the field immediately after harvest or during storage (Chulze, 2010). However, evaluation studies in grains are limited due to cost implications and the inhibitory effect in maize generally achieved with higher concentrations than in synthetic media, because of possible matrix interference and reduced bioavailability relating to distribution on kernel surfaces and penetration into the pericarp (Torres et al., 2003; Samapundo et al., 2007). In certain cases, high concentrations of phenolic compounds could also affect the sensory quality of the maize. Certain antioxidants such as BHA and PP, clay minerals, and plant extracts are considered GRAS, making it very promising for biocontrol purposes. Mixtures of antioxidants or combinations with other food preservatives (i.e., benzoic and sorbic acids) could further enhance the antifungal efficacy (Reynoso et al., 2002).

Even though biologically based treatments most likely will have a reduced effect than chemical methods on the desired nutritional value, quality, safety, or sensory attributes of foods and feed and impact on the environment, compliance to food safety assessment guidelines, such as those prescribed by the European Network on Safety Assessment of Genetically Modified Food Crops (ENTRANSFOOD) and the FAO/WHO, have to be met (He and Zhou, 2010). Assessments could include compositional analyses of key components of treated food including nutrients, micronutrients, and predictable secondary metabolites; assessment of possible toxicity, allergens; potential environmental impact; long-term nutritional impact; influence of food/feed processing; potential dietary intake and change in dietary pattern. While there are several opportunities for further exploring and developing biological control methods for Fusarium growth and fumonisins, each method has its own challenges. However, an integrated approach, involving good agricultural management practices, HACCP models and storage management, together with appropriately selected biologically based microbial treatments, mild chemical and physical treatments could reduce Fusarium diseases and fumonisins effectively pre- and post-harvest (da Cruz Cabral et al., 2013).

## Practical and Culturally Acceptable Methods for Mycotoxin Reduction—Approaches in Sub-Saharan Countries

Methods for prevention of chronic exposure to the fumonisins, particularly in low socio-economic rural subsistence farming communities, remain critically important. In developed countries high standards of the major food suppliers and retailers are upheld and the regulatory controls deter the importation and marketing of seriously contaminated products. In developing countries only a limited number of countries have legislative maximum levels for fumonisins, and implementation thereof is often poor. In rural subsistence farming communities, legislation is not applicable and with continued pressure on food security, an increased **mycotoxin exposure** on a daily basis is the norm. In addition, due to the stringent mycotoxin standards in developed countries, the best-quality food products are normally exported resulting in highly contaminated foods being utilized domestically which increases the risk of mycotoxin exposure and the associated adverse health effects (Pitt et al., 2012). High risk population groups include rural communities and/or subsistence farmers heavily reliant on maize as their staple diet. Although, commercial maize is contaminated with lower levels, daily exposure could be a risk factor for disease development in impoverished communities.

In developing countries, where resources are limited and sophisticated technologies are lacking, the importance of cost-effective and simple intervention methods, predominantly at population level, has been emphasized. In this regard, culturally acceptable simple, practical and biologically based methods of reduction are relevant, as a last line of defense

#### REFERENCES


in rural subsistence farming communities exposed to high levels of the fumonisins in their staple diet. Effective reduction has been demonstrated with hand sorting, flotation, washing, dehulling of maize kernels and combinations thereof in vitro and in field studies (**Table 5**). Dehulling and shelling of maize are common practices in West-Africa (Fandohan et al., 2006), with the removal of the pericarp an effective way to reduce mycotoxin contamination (Sydenham et al., 1994; Bullerman and Bianchini, 2007; Burger et al., 2013). The effectiveness of hand-sorting of maize by removing visibly infected and damaged kernels, resulting in a significant reduction of fumonisins has been demonstrated in several African countries, including Benin (Fandohan et al., 2005), Nigeria (Afolabi et al., 2006), Tanzania (Kimanya et al., 2008), South Africa (Van der Westhuizen et al., 2010), and Malawi (Matumba et al., 2015). In South Africa a simple, practical and culturally acceptable hand-sorting and washing intervention method was developed and implemented for reduction of fumonisin exposure in a subsistence maizefarming community (Van der Westhuizen et al., 2010, 2011b). The efficacy of the maize kernel wash method could possibly be further enhanced by incorporating clay minerals or fumonisin detoxifying enzymes. Advantages of interventions involving practical methods usually take the form of improved health outcomes rather than market outcomes (Wu and Khlangwiset, 2010a,b). Public health interventions should be culturally acceptable; be implemented through educational campaigns; and must have financial and infrastructural support to be feasible in remote rural areas where they are most needed. Sustainability of these reduction strategies is, however, dependent on the available maize supply (food security), as well as the socio-economic status and education of a community.

### AUTHOR CONTRIBUTIONS

Dr. JA, Wrote article; Prof. WG, Coordinated and assisted in writing article; Prof. WV, Assisted in writing article.

### ACKNOWLEDGMENTS

The authors thank the South African Maize Trust for their financial support of research on the use of biological methods for mycotoxin control.


European maize (Zea mays L.) inbred lines. Theor. Appl. Genet. 120, 1053–1062. doi: 10.1007/s00122-009-1233-9

Luongo, L., Galli, M., Corazza, L., Meekes, E., Haas, L., Plas, L. C., et al. (2005). Potential of fungal antagonists for bio-control of Fusarium spp. in wheat and maize through competition in crop debris. Biocontrol Sci. Technol. 15, 229–242. doi: 10.1080/09583150400016852

Marasas, W. F., Riley, R. T., Hendricks, K. A., Stevens, V. L., Sadler, T. W., Gelineau-van Waes, J., et al. (2004). Fumonisins disrupt sphingolipid metabolism, folate transport, and neural tube development in embryo culture and in vivo: a potential risk factor for human neural tube defects among populations consuming fumonisin-contaminated maize. J. Nutr. 134, 711–716.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2016 Alberts, van Zyl and Gelderblom. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Climate, Soil Management, and Cultivar Affect *Fusarium* Head Blight Incidence and Deoxynivalenol Accumulation in Durum Wheat of Southern Italy

Valeria Scala<sup>1</sup> , Gabriella Aureli <sup>2</sup> , Gaspare Cesarano<sup>3</sup> , Guido Incerti <sup>3</sup> , Corrado Fanelli <sup>4</sup> , Felice Scala<sup>3</sup> , Massimo Reverberi <sup>4</sup> \* and Giuliano Bonanomi <sup>3</sup>

#### *Edited by:*

*Thomas Miedaner, University of Hohenheim, Germany*

#### *Reviewed by:*

*Ruth Dill-Macky, University of Minnesota, USA Richard-Forget Florence, Institut National de la Recherche Agronomique, France*

*\*Correspondence: Massimo Reverberi massimo.reverberi@uniroma1.it*

#### *Specialty section:*

*This article was submitted to Fungi and Their Interactions, a section of the journal Frontiers in Microbiology*

*Received: 25 January 2016 Accepted: 14 June 2016 Published: 30 June 2016*

#### *Citation:*

*Scala V, Aureli G, Cesarano G, Incerti G, Fanelli C, Scala F, Reverberi M and Bonanomi G (2016) Climate, Soil Management, and Cultivar Affect Fusarium Head Blight Incidence and Deoxynivalenol Accumulation in Durum Wheat of Southern Italy. Front. Microbiol. 7:1014. doi: 10.3389/fmicb.2016.01014* *<sup>1</sup> Research Unit for Plant Pathology, Council for Agricultural Research and Economics, Rome, Italy, <sup>2</sup> Research Unit for Cereal Quality, Council for Agricultural Research and Economics, Rome, Italy, <sup>3</sup> Dipartimento di Agraria, University of Naples Federico II, Naples, Italy, <sup>4</sup> Plant Pathology, Dipartimento di Biologia Ambientale, Sapienza University of Rome, Rome, Italy*

*Fusarium* head blight (FHB) is a multifaceted disease caused by some species of *Fusarium* spp. A huge production of mycotoxins, mostly trichothecenes, often accompanied this disease. Amongst these toxic compounds, deoxynivalenol (DON) and its derivatives represent a major issue for human as well as for animal health and farming. Common and durum wheat are amongst the hosts of trichothecene-producing *Fusaria.* Differences in susceptibility to fungal infection and toxin accumulation occur in wheat cultivars. Recently, increasing incidence and severity of *Fusarium* infection and a higher DON accumulation in durum wheat were observed in Italy, especially in Northern regions. In this study, we analyzed wheat yield, technological parameters, the incidence of *Fusarium* infection and DON content in kernel samples of durum wheat coming from three locations of Southern Italy with different climatic conditions and grown during two seasons, with two methods of cultivation. Four different durum wheat cultivars prevalently cultivated in Southern Italian areas were chosen for this study. Our analysis showed the effects of environment and cultivar types on wheat productivity and key technological parameters for the quality level of the end-product, namely pasta. Notably, although a low rate of mycotoxin contamination in all study sites was assessed, an inverse relation emerged between fungal infection/DON production and durum wheat yield. Further, our study pinpoints the importance of environment conditions on several quality traits of durum wheat grown under Mediterranean climate. The environmental conditions at local level (microscale) and soil management practices may drive FHB outbreak and mycotoxin contamination even in growing area suitable for cropping this wheat species.

Keywords: FHB, deoxynivalenol, durum wheat, yield, environmental parameters, minimum tillage

## INTRODUCTION

Durum wheat (Triticum durum Desf.) is the most widespread crop in the Mediterranean area. Sixty-seven percent of the Italian production of durum wheat comes from the Southern regions and it is used mainly for producing pasta (Fagnano et al., 2012). Quaranta et al. (2010) confirmed the importance of environmental local conditions in driving mycotoxin contamination in durum wheat. They reported that Southern Italy is an area particularly suitable for producing high quality durum wheat with a low content of Fusarium-toxins. Some species of Fusarium are the causal agents of the Fusarium Head Blight (FHB), a disease of great concern for wheat and for other cereal crops (Kelly et al., 2015). FHB disease causes direct economic losses including reduced yield and quality of grains and indirect loss due to mycotoxin contamination (Shephard, 2008; Berthiller et al., 2009; Zain, 2011). Climatic conditions, especially during wheat anthesis, consistently affect composition of Fusarium species causing FHB (Bernhoft et al., 2012; Covarelli et al., 2015; Kelly et al., 2015). Fusarium spp. associated to FHB disease (Xu et al., 2008) may change throughout the years (Covarelli et al., 2015), but Fusarium graminearum and Fusarium culmorum are the most commonly found species (Häller et al., 2008; Edwards, 2009). In Norway, Bernhoft et al. (2012) reported, "Agronomic and climatic factors explained 10– 30% of the variation in Fusarium species and mycotoxins." Since cereals from organic farming resulted less infected by Fusarium species than cereals from conventional farming systems, the authors conclude that this difference is mainly due to lack of crop rotation and use of mineral fertilizers and pesticides in conventional systems. In general, precipitation during anthesis is particularly conductive for cereal contaminations by Fusarium spp. (De Wolf et al., 2003; Fedak et al., 2007; Visconti and Pascale, 2010). The impact of FHB can be limited by adopting measures for reducing the inoculum and preventing its dispersal such as the cultural, biological, and chemical control and use of resistant varieties (Sutton, 1982; Magan et al., 2002). Czaban et al. (2015) suggested that winter wheat kernel infection by Fusarium spp. depends primarily by weather conditions and then by the wheat genotype.

Previous crop residue and tillage practices differentially affected the incidence and severity of FHB disease (Dill-Macky and Jones, 2000). Schaafsma et al. (2005) confirmed that planting wheat after corn or wheat, together with minimal or no-tillage practices increased the potential for FHB epidemics across South-Western Ontario. Miller et al. (1998) also reported that zero tillage resulted in increased seed infection compared to conventional tillage in Canada. Further, data concerning the influence of environment on several technological parameters of durum wheat kernels are of great interest (Flagella et al., 2010; Pinheiro et al., 2013). In Italy, FHB has been reported mostly in the Northern–Central regions. Disease incidence and Fusarium species involved, varied depending on the year, cultivation area, and wheat cultivar. Infections increase gradually from the South to the North and are closely related to the amount of precipitations during wheat anthesis (Pancaldi et al., 2010; Covarelli et al., 2015).

Several aspects concerning the effect on quality traits of durum wheat by species of fungi involved in FHB disease are already assayed, as for damage on protein fractions of kernels (Dexter et al., 1997; Nightingale et al., 1999; Brzozowski et al., 2008). Other than direct damaging effects on kernels, FHB may produce different types of trichothecenes: Fusarium spp. producing DON and/or its acetylated derivatives are described as chemotype I, whereas those produce nivalenol (NIV) and/or 4-acetyl-NIV are included in chemotype II (Pasquali and Migheli, 2014). DON is the most frequent Fusarium-toxin in Italy, as well as in other European countries. Its occurrence in durum wheat increases from Southern to Northern areas in Italy, with a heavy influence of some factors such as year and area of cultivation (Aureli et al., 2015). DON is toxic for humans, animals, and contributes to the aggressiveness of F. graminearum during wheat infection. Resistance to DON is an important aspect of wheat resistance to FHB (Rocha et al., 2005; Gauthier et al., 2013; Pasquali and Migheli, 2014). Since durum wheat is more susceptible to FHB than common wheat (Covarelli et al., 2015) mycotoxin accumulation in kernels is of particular concern in Italy as food safety issue (Boutigny et al., 2008; Covarelli et al., 2015). The European Commission established maximum thresholds (EU Commission Regulation No. 1881/2006 and 1126/2007) and "indicative levels" (Recommendation 2013/165/EU) for the T-2 and HT-2 Fusarium-toxins content in cereals and cereal based products.

In this paper, we study the influence of climatic conditions on durum wheat grown in cropping trials performed in Southern Italy that is known as a suitable growing area for growing this species of wheat. Several quality traits, infection by trichothecene-producing Fusarium spp. and accumulation of deoxynivalenol (DON) in grains were assessed in four cultivars of durum wheat grown with two techniques in three locations during two growing seasons.

### MATERIALS AND METHODS

## Study Sites Description

Experiments were conducted at three study sites in Southern Italy in the area of Lacedonia (Avellino) included in the Campania region (Figure S1) during two growing seasons (2011–2012 and 2012–2013). The three sites are characterized by different microclimatic conditions, related to altitude and aspects, but similar soil types (Calcixerert Vertisols; Soil Survey Staff, 1998). Soil characteristics including pH, organic C, total N, available <sup>P</sup>2O5, K+, and electrical conductivity were rather similar at the three sites (Table S1). This selection of study sites allows an evaluation of the microclimate impact on wheat production and disease incidence by keeping constant soil type as ecological factor. All sites share a Mediterranean climate, with differences related to altitude and aspect. Site A is located at 827 m a.s.l. (41◦ 02′ 00.99′′N, 15◦ 27′ 10.82′′E) on an almost flat hilltop. Site B is at 520 m a.s.l. (41◦ 01′ 20.65′′N, 15◦ 30′ 05.55′′E) with a western face aspect, while site C is at 513 m a.s.l. (41◦ 03′ 16.98′′N, 15◦ 30′ 32.45′′E) with East exposure. Mean annual air temperature is 13.1, 13.6, and 14.5◦C and the mean annual rainfall is 512, 510, and 523 mm at study sites A, B, and C, respectively (WorldClim 1.4.; Hijmans et al., 2005). To assess microclimate conditions at the three sites we installed monitoring meteorological stations (Vantage Pro2 Plus, Davis, USA), each of which was equipped with a data logger and integrated sensors to collect the following hourly weather data: air temperature, air relative humidity and rainfall (Figure S2).

## Experimental Design, Crop Management, and Sampling

Each experimental trial was carried out according to a randomized complete block experimental design with three replications totalling 24 permanent plots (50 × 25 m) were established at each site; this experimental protocol was the same both for year and within each site (A, B, and C) but not overlapping the plots areas sowed in the previous year. Two tillage treatments: (i) conventional tillage (CT), consisting of mouldboard plowing to 40 cm depth followed by a soil grubber and a disk harrow passage for seedbed preparation, and (ii) minimum-tillage (MT) consisting of a single passage by disk harrow to a depth of 8–10 cm for seedbed preparation. Mouldboard plowing was used because is still the most common tillage in the study area. On both CT and MT soils the following durum wheat cultivars with different growth cycle length, from early (E) to medium (M) were sown: Svevo (E), Simeto (ME), Claudio (M), and Normanno (M). The four cultivars employed in this assay were chosen based on their large diffusion in the area considered and the different length of the growth cycle. The sowing period ranged from 15th to 17th Nov. 2011 and from 11th to 27th Nov. 2012; the harvesting period ranged from 23th Jun. to 11th Jul. 2012 and from 6th to 25th Jul. 2013. Crop management was carried out according to local agronomic professional practice. The seeding rate used was of 400 seeds/mq/plot. Previous crop rotation at all sites was wheatfaba bean (Vicia faba minor). Based on soil analyses (Table S1), mineral nitrogen (120 kg N ha−<sup>1</sup> ) was split applied, at the rate of 1/3 before sowing and 2/3 N top-dressed applied during wheat tillering as ammonium nitrate. Weeds were periodically controlled during the growing season by means of specific and selective herbicides. At harvest time, approximately 30 kg grain samples grains were random collected from each plot directly from the threshing machine. After homogenization, subsample of 5 kg was taken from which subsequent subsample of 1 kg was taken again. A final grain sample of 100 g was milled for the analyses.

All sampling operations were based substantially on the criteria of representativeness reported by European legislation (EU Commission Regulation No. 401/2006).

#### Analysis of Quality Parameters

The semolina samples obtained by a pilot milling plant (Buhler MLU 202) were employed for the following quality analyses: protein content carried out by Dumas combustion method (ICC method n. 167) with automatic instrument Leco FP 428 (USA), gluten content (EN ISO 21415), gluten index (ICC 158), Glutomatic System (Perten, Sweden), rheological parameters (alveographic P/L and alveographic W; alveograph Chopin—UNI 10453 method), yellow and brown indices by reflection colorimeter (Minolta Chromameter CR-400). Pasta samples (spaghetti shape, 1.65 mm diameter) were produced by an experimental press (Namad, Italy) and by an experimental drying system (AFREM-France) at low-temperature (50◦C) drying diagram. The overall judgment was carried out evaluated by a score ranging from 10 to 100 (D'Egidio et al., 1993). The results showed are the average values of replicate analyses as specified for each method employed.

## Fungal Identification and Infection Incidence

Fifty kernels were randomly selected from each sample and surface disinfected accordingly to Giorni et al. (2015). Grain kernels were plated on Petri dishes (Ø 9 cm) containing Potato Dextrose Agar (PDA, Oxoid Ltd., Basingstoke, Hampshire, UK) added with 0.1% streptomycin (Sigma-Aldrich, St. Louis, MO, USA) and incubated at 25◦C for 7 days with a 12 h light photoperiod. After incubation, kernels showing mold development were counted and incidence percentage calculated as in Giorni et al. (2015). Fusarium, Aspergillus, and Penicillium growing colonies were identified at genus level (Raper and Fennell, 1965; Pitt, 1979; Summerell et al., 2003). All Fusarium isolates were sub-cultured on water agar (2% of Bacto agar, Difco) using the single spore technique. The Fusarium spp. mycelia used for the DNA extraction were grown on PDA. Glumes and spikelets were evaluated for FHB severity at harvest. FHB severity was estimated by counting infected spikelets in a head and expressed as percentages (Burlakoti et al., 2007). FHB severity values were calculated from 90 wheat heads (30 heads per replicate and a total of 90 heads) per treatment derived from the experimental design previously described.

## Fungal Growth by qPCR

Total DNA was extracted from wheat kernels as described in Reverberi et al. (2013) and its concentration and quality was determined using by spectrophotometer (DU-530, Beckman Instrument Inc.). The total DNA extract from kernels was used as template to monitor Fusarium trichothecene-producer (TR-producing Fusarium) growth in durum wheat kernels. At this aim, a specific SYBR green qPCR method was set by using Tri5 primers (for\_CAGATGGAGAACTGATGGT; rev\_GCACAAGTGCCACGTGAC) as described by Edwards et al. (2001). Tri5 primers yielded a 260-bp fragment. Standard calibration was performed plotting the Real-time PCR signals obtained for genomic DNA of Fusarium spp. mycelia, harvested from 7-day-old single-spore cultures (see description in chapter 2.4), in the concentration range 0.01 pg–100 ng. The equation, describing the increase of Fusarium trichotheceneproducer (TR-producing Fusarium) DNA concentration, was calculated (y = −0, 9829x+27,921, R <sup>2</sup> = 0.994) and used for extrapolating quantitative information of DNA targets of unknown concentrations in wheat kernels. The efficiency of the PCR reaction (98.8%) was obtained from the calibration curve slope (E = 10−<sup>1</sup> /slope−<sup>1</sup> ). In all samples, DNA was extracted in triplicate and each biological replicate was technical repeated three times.

#### Deoxynivalenol Analysis

From each grain sample a representative subsample was milled (particle size ≤0.5 mm) by the use of Cyclotec 1093 (FOSS, Sweden) and submitted to the DON analyses by Enzyme-Linked Immuno-Sorbent Assay (ELISA). DON determination was made according to the Ridascreen <sup>R</sup> DON method (R-Biopharm AG, Germany). The limit of detection (LOD) was 18.5 µg/kg. The range of recovery declared in the method was between 85 and 110%. Data were acquired on such as samples as mean of double analysis (CV≤ 10%). The Basic Robotic Immunoassay Operator (BRIO, SEAC, Radim Group, Italy) was used and the absorbance values were read using Sirio-S Microplate Reader (SEAC, Radim Group, Italy). The RIDA <sup>R</sup> Soft Win software (R-Biopharm AG, Germany) was employed for quantitation of DON in samples. Distilled water was obtained from Water Purification System Zeener Power I (Human Corporation, Korea).

#### Statistics

We considered the following 11 dependent variables: content of DNA from TR-producing Fusarium and of deoxynivalenol toxin (DON) in wheat kernels, yield (total wheat production), percent content of proteins and gluten, gluten index, alveographic W and P/L ratio, yellow and brown indices, and sensorial assessment score, assessed in wheat samples undergoing 48 different combinations of cultivation treatments. First, a cross-correlation matrix was calculated among the dependent variables. Then, for each dependent variable, a combined analysis by Generalized Linear Model (GLM) was fitted, including main and second order interactive effects of treatments: wheat cultivar (four levels: Claudio, Normanno, Simeto, Svevo), soil management practice (either conventional tillage, CT or minimum tillage, MT). Experimental replication factors including harvesting year (either 2012 or 2013) and experimental field (three sites: A, B, and C) were considered as additional treatments in the GLMs after preliminary verification, for each dependent variable, of the homogeneity of variances for different years and for different sites. In the case of harvesting years (N = 2) the homogeneity of variances were tested by the F-test, while in the case of sites (N = 3) the Bartlett's chi-square test was used. In the GLMs, pairwise significant differences between treatment combinations were assessed by Duncan's post-hoc test. Statistical significance was tested at the conventional threshold of α = 0.05. The data matrix of dependent variables and treatments was submitted to Principal Component Analysis. Loading vectors of variables and factorial scores of treatment combinations were plotted in 3D biplots representing the first three principal components (Jolliffe, 2002). All statistical analyses were carried out using the software package STATISTICA 7 (StatSoft Inc., Tulsa, Oklahoma).

### RESULTS

#### Quality Parameters Assessment

Results of GLM and post-hoc Duncan tests showed that crop yields were significantly affected by all experimental factors, with also significant interactions of wheat cultivar with cultivation year and study site (**Table 1**; Table S2; Supplementary Datasheet 1). In detail, yield was significantly higher in 2012 compared to 2013, in sites B and C compared to site A and lower for the cultivar Simeto (**Figure 1**); concerning soil tillage regime, wheat yield was significantly higher in CT compared to MT (**Figure 1**).

The protein and gluten content were significantly affected by all experimental factors with the exception of soil management whereas the cultivar factor had the major influence on gluten index parameter (**Table 1**). Protein content was higher in 2013, at site A, and was significantly lower for cultivar Claudio compared to Simeto and Svevo (**Figure 1**). The year and the study site factors had a clear and significant influence on the alveographic parameters W and P/L ratio, while their interaction was significant only for P/L parameter.

The environmental factors influenced the brown index, a negative characteristic of semolina that is also influenced by the mineral content of grain, whereas only cultivar and year factors affected the yellow index color. With regard to this last character, the results obtained indicate the significant influence of genotype and environmental factors on grain quality parameters. Finally, the overall judgment score was not affected by any of the experimental factors (**Table 1**; Table S2).

#### *Fusarium* spp. Infection and DON Accumulation

The kernels sampled in 2012 and 2013 showed a different incidence of total fungal infection, with Fusarium spp. as the main fungal contaminants in 2013 (Table S3). GLM analysis showed that the content of TR-producing Fusarium DNA in kernels was significantly affected by all experimental factors, with the exception of soil tillage regime; the year and its interaction with the study site and cultivar type had the major significant influence (**Table 1**). In particular, fungal DNA content as well as FHB severity was higher in 2013 compared to 2012 with a major presence of fungal DNA content in grain samples of cultivar Normanno, especially at the coldest and wettest study site (A; **Figure 1**). Deoxynivalenol accumulation in wheat kernels was significantly affected by year, study site and by their interactions but not by the cultivar type (**Table 1**). DON reached higher concentrations in 2013 compared to 2012 at the site (A; **Figure 1**). The sole soil tillage regime partly (p = 0.056) affected DON content in kernel, whereas its interaction with the study site significantly influences the toxin content, with higher values of concentration in MT compared to CT (**Figure 1**).

#### Relationship among Fungal Infection, DON Contamination and Grain Quality Parameters

The multivariate PCA ordination of dependent variables and treatments showed a positive association of the first principal component with most of the grain quality parameters (**Figure 2**). The second principal component was positively associated with alveographic P/L ratio and negatively with yellow index and gluten index (**Figure 2**), while the third ordination axis was positively related with P/L and negatively with overall sensorial assessment score.

Yield was the only variable negatively associated with the first principal component, indicating that this parameter was

TABLE 1 | Results of Generalized Linear Models (GLMs) for 11 dependent variables assessed in 48 samples of durum wheat undergoing different cultivation treatments, including harvesting year (Y, either 2012 or 2013), wheat cultivar (Cv, four levels), experimental field (S, three sites) and soil management practice (M, either conventional tillage or minimum tillage).


*(Continued)*

#### TABLE 1 | Continued


*df, degrees of freedom; SS, sum of squares; MS, mean square; F and p, statistics for main and second order interactive effects of treatments.*

\**For best visualization, SS and MS of TR-producing Fusarium DNA data have been multiplied by 10<sup>10</sup> for content in kernels.*

inversely proportional to some indicators of high-quality (e.g., protein and gluten content and W), as well as to other ones related to low-quality (i.e., brown index, Table S4). The negative correlation between protein content and yield parameters was in agreement with previous data (Mangini, 2006; Blanco et al., 2011). Yield, being unrelated to the second and third components, was not associated with other quality indices (i.e., gluten index, P/L, yellow index, sensorial assessment score, Table S4). DON content and fungal infection parameters (content of TR-producing Fusarium DNA in kernels) were both positively correlated with the first principal component, indicating a negative effect of fungal infection on yield even with low levels of contamination.

#### Weather Conditions

Weather conditions during wheat flowering, a process occurring in May at our study sites, are the key factor for Fusarium infection and disease spread (De Wolf et al., 2003). Air temperature and RH as well as daily rainfall in May 2012 and 2013 are summarized in Figures S2–S4. In 2012, the anthesis stage occurred between 20th and 25th (site A), 11th and 15th (site B), and 10th and 14th day (site C) of May whereas in 2013 between the 28th of May and 2th of June (site A), 10th and 16th of May (site B), 14th and 19th of May (site C). Temperature and rainfall were consistently different between the two years and among study sites. Site A, located at the highest elevation, was the coldest with relatively high humidity levels in both years (Figure S2). Mean monthly temperature in 2012 was consistently lower than in 2013 at all study sites (Figure S2). In addition, lower cumulated rainfall, fewer rain events along the flowering periods, and higher air relative humidity were recorded in 2012 compared to 2013 (Figures S2, S3). Some of the rainy events recorded in 2012 and 2013 were followed by several days of dry weather, with low air relative humidity (Figures S2, S3). However, the fewer rain events recorded at site A in 2013 from the 25th day to the end of the month were followed by higher air relative humidity as well as lower temperature trend compared to 2012: these are to the most suitable conditions for FHB spread during the anthesis phase.

## DISCUSSION

Durum wheat is the most widespread crop in the Mediterranean area. Our study suggests that environmental conditions at local level (microscale) and soil management practices are determinant factors in controlling potential FHB outbreak and mycotoxin contamination. This study, based on a two-year field experiment at three study sites of Southern Italy, confirms the critical role of weather conditions in promoting the development of Fusarium species, producers of trichothecenes, even in areas

deviation. Different letters indicate statistically significant differences within each plot (Duncan test, *P* < 0.05, for statistical detail see Table 1).

suitable for cropping durum wheat. Intriguingly, it appears that data related to different climates such as those in USA (De Wolf et al., 2003), Canada (Hooker et al., 2002), and Northern Europe (Chandelier et al., 2011; West et al., 2012) can be applied to Mediterranean conditions of Southern Italy. Previous crop residues and tillage practices can also affect incidence and severity of FHB (Dill-Macky and Jones, 2000). In Italy, FHB as well as DON contamination have been reported in several regions with different intensity depending on the year, cultivation area and durum wheat variety (Pancaldi et al., 2010; Aureli et al., 2015). Infection rates increase with a South-North gradient and closely relate to precipitations during wheat anthesis (Pancaldi et al., 2010; Covarelli et al., 2015).

assessed in 48 samples differing by wheat cultivar and cultivation year, site and type of soil management. Data refer to loading vectors of variables (blue vectors) and factorial scores of samples (orange dots).

Our study, reporting the highest presence of TR-producing Fusaria and DON contamination in the wettest study site (site A) and year (2013). These results de facto extend to Southern Italy area, previous data obtained in other studies. In fact, FHB incidence and the amount of DNA of trichothecene-producing Fusarium into durum wheat kernels significantly varied between the two observation periods, being higher in 2013 than in 2012 as well as the DON content in kernels. Apparently, the two different soil management regimes had no influence on the fungal DNA amount in the wheat kernel, whereas the climatic features of the cultivation areas consistently affected it. In the site A, the highest values of cumulated rainfall and mean air humidity were recorded and associated with the highest values of fungal DNA content and DON contamination in wheat kernels. Besides environmental conditions, the wheat cultivar influenced fungal infection and the cultivar Normanno showed the higher content of TR-producing Fusarium DNA (**Figure 1**).

DON is the most frequent Fusarium-mycotoxin detected in Italy, as well as in other European countries; nevertheless, FHBassociated species of Fusarium can produce different types of trichothecenes (Pasquali and Migheli, 2014). DON occurrence in our grain samples, showed very low levels of contamination (maximum: 734 µg/kg), even if a slight difference between 2012 and 2013 harvest was measured (**Figure 1**). Incidence of positive samples (DON concentration ≥18.5 µg/kg) on the total samples analyzed was 21% in 2012 and 71% in 2013 whereas the average of contamination values of the positive samples was 24 µg/kg (2012) and 276 µg/kg (2013). Negligible amount, or absence, of DON was detected all over the trials during the two cropping years with the exception of the wetter and colder field site A, especially in 2013. In this site, the highest average values of DON were achieved both in the tilled (414 µg/kg) and in minimum-tillage management (92 µg/kg). However, data collected about the levels of DON contamination were all very far from the maximum limit (1750 µg/kg) set for unprocessed durum wheat (EC Regulation 1881/2006 and 1126/2007). These results were in agreement with the meteorological data collected all over the 2-year period that showed a rainy condition in the year 2013 generally higher than in 2012. The suitable conditions for an outbreak of FHB in field and DON production, such as rainfall occurrence and high percentage of relative humidity, raised from meteorological mean data collected. By all data collected, a slight positive—but highly significant—correlation (p < 0.001) was found between the parameters DON and fungal DNA detected in kernels. Several studies reported positive correlations between disease incidence and mycotoxin accumulation (Burlakoti et al., 2007); however this question might be more complex due to the not always positive and significant relationship between wheat varieties with high FHB resistance and low levels of DON detected (Boutigny et al., 2008).

Regarding the quality aspects assessed with protein and gluten content, alveographic W, and alveographic (P/L) ratio, our analysis confirmed the effects of the environment and cultivar type on key technological parameters for the quality level of the end-product (pasta; Mariani et al., 1995; Raciti et al., 2005). The protein content in durum wheat is influenced by environment parameters, with high temperature and water shortage that can significantly affect both the content and protein composition in Mediterranean climate (Flagella, 2006). The negative correlation between protein content and yield parameters was in agreement with previous data (Mangini, 2006; Blanco et al., 2011). These results confirm the importance of the genotype-environment interaction in determining the protein content particularly influenced by additive effects of environment (Mariani et al., 1995). Apparently, this interaction has scarce effect on gluten index parameter (Ames et al., 1999) which however is significantly affected by the temperature trend during grain filling (Flagella et al., 2010; Fois et al., 2011).

In the same way the year and site of cultivation had a significant "weight" (p < 0.001) on the brown index, a negative quality trait subjected also to the environmental effect. As expected the yellow index was influenced (p < 0.001) by environmental factor (year) but also the role of genotype influenced was confirmed by the significant effect (p < 0.001) of cultivar parameter. However, no significant differences was found by the overall judgment of the end-product (pasta) probably due to the minimal differences of mean values within the study sites.

#### CONCLUSIONS

This study suggested that also in Southern Italy, a growing area under Mediterranean climate suitable for durum wheat cropping, weather conditions, and soil management may affect not only several quality traits of durum wheat, but also infection by trichothecene-producing Fusarium and accumulation of DON in kernels. More in general, differences emerging among the sites and the cultivars in relation to fungal growth and DON content (even if both kept at low level) apparently suggest how these parameters can be finely regulated by environmental parameters at micro-scale.

Mediterranean climatic conditions are mostly unsuitable for outbreak of FHB and field contamination by DON. This likely occurs because of the scarcity of rainfall events and amount during the anthesis. However, FHB outbreak can still occur particularly in moist years in production areas located at "high" elevation. Under such circumstances, the wetter and colder conditions promote infection and disease spreading also in suitable growing areas.

In conclusion, weather parameters, especially rainfall and air humidity with low values of temperature appear the main drivers to settle conductive or non-conductive conditions for TR-producing Fusarium spp. development, although the choice of type of tillage and the selection of suitable cultivars can significantly affect the level of disease spreading.

#### AUTHOR CONTRIBUTIONS

VS: conceived of the study, participated in its design and coordination, drafts, and revises critically the manuscript. Conception and design of the nucleic acid extraction and Real Time PCR methods, Fusarium spp. infection and FHB index; acquisition, analysis, interpretation of data for the work; GA: contributes to the conception or design of the work. Acquisition, analysis, or interpretation of data for the mycotoxins analysis and yield and grain quality. Drafting the work and revising it critically for important intellectual content; GC: contributions to the acquisition, analysis, or interpretation of data for the work. Drafting and the work or revising it critically; GI: contributions

REFERENCES


to data analysis, interpretation of data for the work, Drafting the work; CF: Drafting the work, revising it critically for important intellectual content, conceived of the study, participated in its design and coordination; FS: Drafting the work, revising it critically for important intellectual content, conceived of the study, participated in its design and coordination; MR: conceived of the study, participated in its design and coordination, drafts and revises critically the manuscript. Conception and design of the nucleic acid analysis, Fusarium spp. infection and FHB index; acquisition, analysis, interpretation of data for the work; GA: contributes.

#### FUNDING

The present study was funded by the MiPAF through the project "VAFRUMIC" within the OIGA (DM 18829/2009)— "Progetti Ricerca e Sviluppo" framework and by Progetti Ateneo Sapienza University of Rome (Year 2014—prot. C26A 14AFZ7).

#### ACKNOWLEDGMENTS

The authors acknowledge Cristina Cecchini, Ester Gosparini, Alessandra Arcangeli, and Roberto Mortaro (CREA-QCE) for technical support.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fmicb. 2016.01014

Fusarium graminearum in wheat inoculated with isolates collected from potato, sugar beet, and wheat. Phytopathology 97, 835–841. doi: 10.1094/PHYTO-97-7- 0835


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

The handling editor declared an intended collaboration with the author MR, but states that the process nevertheless met the standards of a fair and objective review.

Copyright © 2016 Scala, Aureli, Cesarano, Incerti, Fanelli, Scala, Reverberi and Bonanomi. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Inoculum Potential of Fusarium spp. Relates to Tillage and Straw Management in Norwegian Fields of Spring Oats

Ingerd S. Hofgaard<sup>1</sup> \*, Till Seehusen<sup>1</sup> , Heidi U. Aamot<sup>1</sup> , Hugh Riley<sup>1</sup> , Jafar Razzaghian<sup>1</sup> , Vinh H. Le<sup>1</sup> , Anne-Grete R. Hjelkrem<sup>1</sup> , Ruth Dill-Macky1,2 and Guro Brodal<sup>1</sup>

<sup>1</sup> Division of Biotechnology and Plant Health, Norwegian Institute of Bioeconomy Research, Ås, Norway, <sup>2</sup> Department of Plant Pathology, University of Minnesota, St. Paul, MN, USA

#### Edited by:

Thomas Miedaner, University of Hohenheim, Germany

#### Reviewed by:

Päivi Parikka, Natural Resources Institute Finland, Finland Matthias Heinrich Herrmann, Julius Kühn-Institut, Germany

> \*Correspondence: Ingerd S. Hofgaard ingerd.hofgaard@nibio.no

#### Specialty section:

This article was submitted to Fungi and Their Interactions, a section of the journal Frontiers in Microbiology

Received: 29 January 2016 Accepted: 04 April 2016 Published: 22 April 2016

#### Citation:

Hofgaard IS, Seehusen T, Aamot HU, Riley H, Razzaghian J, Le VH, Hjelkrem A-GR, Dill-Macky R and Brodal G (2016) Inoculum Potential of Fusarium spp. Relates to Tillage and Straw Management in Norwegian Fields of Spring Oats. Front. Microbiol. 7:556. doi: 10.3389/fmicb.2016.00556 The increased occurrence of Fusarium-mycotoxins in Norwegian cereals over the last decade, is thought to be caused by increased inoculum resulting from more cereal residues at the soil surface as a result of reduced tillage practices. In addition, weather conditions have increasingly promoted inoculum development and infection by Fusarium species. The objective of this work was to elucidate the influence of different tillage regimes (autumn plowing; autumn harrowing; spring plowing; spring harrowing) on the inoculum potential (IP) and dispersal of Fusarium spp. in spring oats. Tillage trials were conducted at two different locations in southeast Norway from 2010 to 2012. Oat residues from the previous year's crop were collected within a week after sowing for evaluation. IP was calculated as the percentage of residues infested with Fusarium spp. multiplied by the proportion of the soil surface covered with residues. Fusarium avenaceum and F. graminearum were the most common Fusarium species recovered from oat residues. The IP of Fusarium spp. was significantly lower in plowed plots compared to those that were harrowed. Plowing in either the autumn or spring resulted in a low IP. Harrowing in autumn was more effective in reducing IP than the spring harrowing, and IP levels for the spring harrowed treatments were generally higher than all other tillage treatments examined. Surprisingly low levels of F. langsethiae were detected in the residues, although this species is a common pathogen of oat in Norway. The percentage of the residues infested with F. avenaceum, F. graminearum, F. culmorum, and F. langsethiae generally related to the quantity of DNA of the respective Fusarium species determined using quantitative PCR (qPCR). Fusarium dispersal, quantified by qPCR analysis of spore trap samples collected at and after heading, generally corresponded to the IP. Fusarium dispersal was also observed to increase after rainy periods. Our findings are in line with the general understanding that plowing is a means to reduce the IP of Fusarium spp. in cereal fields. The main inoculum source for F. langsethiae remains unclear. Our results will be useful in the development of forecasting tools to calculate the risk of Fusarium in cereals.

Keywords: Fusarium langsethiae, spore traps, qPCR, straw residues, Fusarium graminearum, Fusarium avenaceum

## INTRODUCTION

fmicb-07-00556 April 22, 2016 Time: 15:11 # 2

Fusarium head blight (FHB) is an important fungal disease of cereals (Parry et al., 1995). It can cause significant yield losses and reduced grain quality. The disease is caused by several Fusarium species, which survive largely in soil and on crop residues. The species abundance in the field is influenced by environmental conditions (Xu et al., 2008). Fusarium spp. produce a range of different mycotoxins, and if consumed, contaminated grain can be harmful for animals and humans (Desjardins, 2006).

Soil tillage is important to loosen the soil, to prepare a good seedbed, for the incorporation of plant residues, and to control weeds (Håkansson et al., 1998) and plant diseases (Bockus and Shroyer, 1998). However, due to increased risks from erosion and nutrient runoff from tilled fields, the Norwegian authorities encourage farmers to reduce soil tillage operations. Therefore, primary tillage operations in Norwegian cereal fields are more commonly performed in spring, and reduced tillage operations have become more prevalent (Tørresen et al., 2012). In Scandinavia, no-till systems can be used successfully under a wide range of soil types and weather conditions (Rasmussen, 1999). Reports from long-term trials in Norway, indicate that there is little difference in grain yields between reduced tillage and plowed treatments on loamy soil although average yields may be significantly lower on silt, sandy loam, and some clay soils where reduced tillage practices are implemented (Riley et al., 1994, 2005, 2009, 2014). The use of reduced tillage practices does, however, lower the labor requirement and machinery costs (Riley et al., 1994), and the practice has been demonstrated to be a profitable practice in a German study of a crop rotation systems in wheat (Verch et al., 2009). Thus, under certain conditions, reduced tillage practices may be a good way to ensure sustainable crop production with little negative influence on grain yield.

Over the last 15 years, the average Fusarium infection levels of spring cereal seeds in Norway have been more than doubled compared to the previous 30 years, and the infection levels have been positively correlated with July rainfall, which is the flowering month of Norwegian spring cereals (Norwegian Scientific Committee for Food Safety, 2013). The FHB/mycotoxin situation has become a serious challenge for the Norwegian grain industry, especially in oats, which is an important crop in Norway covering approximately 25% of the cereal cultivation acreage.

In a Norwegian survey of oats sampled from 2004 to 2009, the following ranking of Fusarium species was made based on the DNA concentrations of the Fusarium spp. analyzed (from high to low): Fusarium graminearum = F. langsethiae = F. avenaceum > F. poae > F. culmorum (Hofgaard et al., 2016). F. graminearum is more prevalent than it has been at any time (Bernhoft et al., 2010; Aamot et al., 2015; Hofgaard et al., 2016). The increase in Fusarium spp., particularly F. graminearum, is thought to have resulted directly from the increase of cereal crop residues remaining on the soil surface, combined with weather conditions that promote Fusarium growth and infection of these cereals that has promoted inoculum survival and production (Norwegian Scientific Committee for Food Safety, 2013). Diseases caused by residue-borne pathogens, including FHB, are reported to increase with increasing amounts of crop residues (Bockus and Shroyer, 1998; Dill-Macky and Jones, 2000). In order to minimize the risk of erosion and nutrient runoff and at the same time ensure suitable grain quality, it is important to identify tillage practices that are suitable for Norwegian conditions but which do not promote the development of residue-borne diseases.

In areas with low Fusarium inoculum pressure from surrounding fields, residues from the previous crop are considered an important source of inoculum within a field (McMullen et al., 2012). FHB severity and the mycotoxin contamination of cereals has been reported to be influenced by the type and quantity of previous crop residues (Dill-Macky and Jones, 2000). Similarly, the presence of Fusarium spp. in association with plant residues has been demonstrated to vary with the crop species, plant tissue, decomposition stage, soil biota, and microclimate (Champeil et al., 2004a; Pereyra and Dill-Macky, 2008). The pathogenic Fusarium species have a temporary advantage as they can colonize host plant tissues ahead of the saprophytic fungi that only colonize plant residues after they are incorporated into the soil (Bruehl and Lai, 1966). F. graminearum has been reported to survive for years in crop residues and for longer periods on residues left on the soil surface than on buried residues (Pereyra et al., 2004). Similarly the formation of perithecia and macroconidia has been observed to be reduced if the residues have been buried for some time (Khonga and Sutton, 1988). Residues that are buried decompose more quickly when in contact with the soil with residue decomposition being influenced by temperature and moisture as well as the activity of antagonistic microorganisms (Leplat et al., 2013).

Fusarium avenaceum, F. culmorum, F. graminearum. F. poae, and F. sporotrichioides are Fusarium species often identified on cereal crop residues (Dill-Macky and Jones, 2000; Köhl et al., 2007; Fernandez et al., 2008; Golkari et al., 2008; Pereyra and Dill-Macky, 2008; Postic et al., 2012). In a Canadian study, Fusarium spp. were isolated from more than 50% of the cereal residues collected from producers' fields (Fernandez et al., 2008). The various Fusarium species are differentially influenced by environmental conditions. F. graminearum is most prevalent where humid and relatively warm conditions prevail whereas F. avenaceum and F. culmorum is more prevalent under cool and wet or humid conditions (Xu et al., 2008). The infestation of residues by F. avenaceum is reported to be more stable over time compared to that of F. graminearum (Hogg et al., 2010; Palazzini et al., 2013).

Many studies have focused on the effect of tillage regimes on the subsequent development of Fusarium and mycotoxins in cereal grains, mainly wheat (Henriksen, 1999; Dill-Macky and Jones, 2000; Guo et al., 2010; Munger et al., 2014), and plowing is often considered as the best tillage practice to reduce the risk of Fusarium disease development in cereals. However, it is sometimes difficult to find a direct link between tillage practices and the occurrence of mycotoxins, as Fusarium inoculum may be dispersed aerially over large distances (Lori et al., 2009; Prussin et al., 2014a). Several studies have shown that the development of Fusarium and mycotoxins in cereals is related more to the amount of residues than to the tillage regime (Maiorano et al.,

2008). Only a few studies have focused on the effect of tillage regimes on the presence of Fusarium spp. in crop residues within a specific field (Dill-Macky and Jones, 2000; Munger et al., 2014). In Norway, no studies have been published on the presence of Fusarium in cereal residues. Little information is published on the influence of weather conditions on the development and spread of Fusarium from cereal crop residues within a field and none of these studies have been conducted in regions where F. avenaceum and F. langsethiae are among the prevalent pathogens associated with FHB of oat. Such information would be valuable for developing best practices for reducing FHB including models to predict the effect of weather in combination with tillage regime and previous cropping on the development and spread of Fusarium species.

The objective of this study was to elucidate the influence of various tillage and straw coverage regimes on the IP of Fusarium spp. in spring oats in Norway.

#### MATERIALS AND METHODS

#### Field Trials

Two tillage trials with continuously grown oats were conducted at two locations in southeast Norway (Solør and Østfold) over 3 years 2010–2012. The trial at Solør was established on silty soil following a precrop of oat, and the trial at Østfold was established on clay soil following a precrop of winter wheat. The dates of seeding, tillage operations and harvesting are presented in Seehusen (2014).

Each oat trial had a randomized split-plot design with two replicate blocks. The two main residue treatments (plot size 42 m × 15 m) comprised I: most of the residues removed and II: all residues chopped and retained on the field. The main plots were separated by a minimum border of 6 m to allow for the operation of tillage implements. Within each main treatment plot, split-plots (6 m × 15 m) with different tillage regimes were established. Plant material from four of these regimes were used in this study: shallow harrowing (5–6 cm) conducted in spring (SSH), shallow plowing (12–15 cm, furrow plow) conducted in spring (SSP), shallow harrowing (5–6 cm) conducted in autumn (SAH), and deep plowing (25 cm, furrow plow) conducted in autumn (DAP). The type of machinery used varied between locations, but at each location the same implements were used in most years. All trials were harvested with a stubble height of 10–15 cm. The straw was baled and removed from those plots where the treatment called for residues to be removed. In the treatments where the residues were retained, the straw was cut to an average length of 6–7 cm, using a straw cutter mounted on the combine harvester or a stubble chopper, and spread evenly over the whole plot surface. Final seedbed preparation was done by harrowing to <5 cm, before sowing with a combined fertilizer and seed drill and rolling with a Cambridge roller. The location of the plots was fixed throughout the experimental period (2010– 2012). The proportion of the soil surface area covered with straw residue was recorded within a week after sowing each year at all locations using the line-transect method which involves the use of a cord with 100 equally spaced knots (Morrison et al., 1993). No fungicide, insecticide or plant growth regulators were used in these trials. Additional details of the various tillage treatments and yield parameters examined are presented in Seehusen (2014).

## Assessment of Fusarium on Straw Residues

For assessment of Fusarium spp. on residues, straw of oats was collected each year at all field locations within a week of sowing. In 2010, the 1st year of the experiment, residues were collected across the whole field area, in order to calculate the background level of Fusarium spp.. In 2011 and 2012, oat straw residues were collected from each treatment. Within each experimental plot, residues were collected from four 1 m × 1 m quadrats outside the area designated to be harvested. The residues were dried at 25◦C for 24 h and stored at room temperature until used for the recovery of Fusarium spp.

For the recovery of Fusarium species, 50 pieces of straw from each plot were analyzed, except from Solør in 2011 where 100 pieces were used. The straw pieces, 1.5–2 cm long and mostly including a node, were surface disinfected in 0.5% NaOCl for 30 s, transferred to 70% alcohol for 15 s, then rinsed three times with sterile distilled H2O and then finally transferred to sterile filter paper to remove surface water. The straw pieces were then plated onto Petri dishes containing a modified CZID (Abildgren et al., 1987) in which iprodione was replaced with propiconazole (0.75 mg/l). The plated residues were then incubated for 7– 10 days, under alternating 12 h darkness and 12 h near ultra violet light ('black light') and white light at 20◦C. Fusarium mycelium, observed following the incubation period, was transferred to SNA (Nirenberg, 1976) containing chlortetracycline to reduce bacterial growth. A small piece of sterile filter paper was placed on the agar surface to promote sporulation. The SNA cultures were incubated for 10–14 days (incubation conditions as above) and used for the morphological identification of Fusarium species (Leslie and Summerell, 2006). The percentage of Fusariuminfested straw residues was calculated as the number of residue pieces infested with Fusarium as a proportion of the total number of residue pieces analyzed.

Inoculum potential (IP) was calculated for each plot as the percentage of the residues infested with Fusarium spp. multiplied with the proportion (0–1) of the plot surface covered by residues after sowing. The percentage of the plot surface for the four treatments covered with residue after sowing are presented in **Table 1**, and also in Seehusen (2014).

## Assessment of Fusarium DNA in Straw Residues

Surprisingly low percentages of F. langsethiae-infested residues (average field levels 0–1%) were recorded in our morphological analyzes of the Fusarium spp. associated with the oat residues examined, despite this species being commonly detected in Norwegian oat grains (Hofgaard et al., 2016). Therefore, the content of Fusarium DNA was quantified in the samples remaining after much of each sample was utilized for the morphological analysis.


TABLE 1 | The average percentage of soil area covered with residues of the previous years' oat crop, measured after sowing in spring and following the implementation of combinations of two residue treatments and four tillage treatments (Seehusen, 2014).

<sup>∗</sup>Tillage regimes: DAP, deep autumn plowing; SSP, shallow spring plowing; SAH, shallow autumn harrowing; SSH, shallow spring harrowing.

The plant residues were milled using a ZM 200 Mill with a 0.2 mm sieve (Retsch, Haan, Germany). Total genomic DNA from one gram of milled residues was extracted using PowerMax <sup>R</sup> Soil (MO Bio Laboratories, Inc., Carlsbad CA, USA) according to the manufacturers' description. DNA was eluted in a volume of 5 ml and up-concentrated as follows: NaCl was added to a final concentration of 0.2 M, and the tube inverted 3–5 times to mix. Twenty microliters of linear acrylamide (5 mg/ml) was added, followed by the addition of 2.5 × the mix volume of cold 100% ethanol. The mix was inverted by 3–5 times and stored on ice overnight. The following day the mix was spun at 4500 × g for 35 min to pellet the DNA. The pellet was washed with 5 ml 70% ethanol and centrifuged 4500 × g for 10 min to re-pellet. The ethanol was decanted and the tube was inverted to drain, then turned right side up and air dried. The pellet was resuspended in 200 µl of Solution C6 (MO Bio Laboratories), and cleaned using NucleoSpin <sup>R</sup> Gel and PCR Clean-up (Macherey-Nagel, Dueren, Germany). DNA was eluted twice, in 30 µl each time, combined (total volume of 60 µl) and stored at −20◦C prior to analysis by quantitative PCR (qPCR). Fungal standard DNA was extracted from pure cultures according to Divon et al. (2012) from the following isolates from the NIBIO collection: F. avenaceum (isolate ID 201081), F. culmorum (ID 201064), F. graminearum (ID 200630), and F. langsethiae (ID 201087).

The genomic DNA from plant residues were analyzed by TaqMan qPCR to determine the content of F. avenaceum, F. culmorum, F. graminearum, and F. langsethiae DNA, using the primers and probes described in Halstensen et al. (2006; F. avenaceum), Waalwijk et al. (2004; F. culmorum and F. graminearum), and Hofgaard et al. (2016; F. langsethiae). qPCR was performed in a total volume of 25 µl that consisted of 4 µl 10-fold diluted genomic DNA, 300 nM of each primer (Invitrogen by Thermo Fisher Scientific, Waltham, MA, USA), 100 nM of each probe, and 1 × SsoAdvancedTM Universal Probes Supermix (Bio-Rad, Hercules, CA, USA), in a C1000 Touch Term Cycler combined with a CFX96TM Real-Time System (Bio-Rad). The probes for detection were labeled with 6-FAM (Applied Biosystems, by Thermo Fisher Scientific, Waltham, MA, USA) in case of F. graminearum, F. langsethiae, and F. culmorum, or Cy5 (Sigma–Aldrich, St. Louis, MO, USA) in case of F. avenaceum. All reactions were performed with the following parameters: 95◦C for 3 min followed by 45 cycles of 95◦C for 10 s and 60◦C for 30 s. The data were analyzed using the Bio-Rad CFX manager software version 3.1 (Bio-Rad). In case of F. langsethiae, qPCR was performed on undiluted genomic DNA in addition to the 10 fold diluted DNA. The amount of fungal DNA in the samples was quantified as a mean of two technical qPCR replicates differing by a Cq ≤ 1, using a standard curve algorithm with five dilutions of known amounts of DNA of the respective Fusarium species in the range of 0.001–4 ng. The fungal content is presented as pg fungal DNA per g plant residue (pg/g).

## Assessment of Fusarium DNA in Air Samples

Two Automatic Multi-Vial Cyclone Samplers (Burkard Manufacturing Co. Ltd., Rickmansworth, UK) were placed in each oat field in 2011 and 2012. One sampler was placed in plots with shallow harrowing in spring (SSH) where the straw was chopped, the other in deep autumn plowing (DAP) plots where the straw had been removed. The air intake of the spore traps was at about 1 m height above ground. Bio-aerosols were collected in 1.5 ml tubes at an air movement rate of 16.5 l/min, and the tubes were automatically replaced every 24 h (at 00.00 h). At Solør, air samples were collected from week 25 to 33 (the oat crop reached heading in week 28) in 2011, and from week 23 to 37 (heading: week 31) in 2012. At Østfold, samples were collected from week 19 to 33 (heading: week 26) in 2011, and from week 21 to 35 (heading: week 27) in 2012. The tubes were collected from the samplers once a week and stored at −20◦C prior to DNA extraction.

Total genomic DNA was extracted from air samples collected over 1 week periods (equivalent to eight 1.5 ml tubes as the tubes were sampling only <sup>1</sup>/<sup>2</sup> day in the beginning and end of the period) using the FastDNA <sup>R</sup> SPIN Kit for Soil (MP Biomedicals, Santa Ana, CA, USA). A volume of 489 µl of sodium phosphate buffer was successively transferred between all tubes included in one extraction, combined with vigorous vortexing of the buffer in the individual tubes, before transferring it to a Lysing Matrix E tube. The procedure was repeated once using the same amount of buffer, followed by a quick spin down of the tubes so that any remaining amounts of sodium phosphate buffer was also collected. Both buffer washtroughs were combined into one extraction. DNA was extracted according to the manufacturer's protocol, and eluted in 100 µl DNase/pyrogen-free water. Samples were stored at −20◦C prior to analysis by qPCR.

Conidia from the standard isolates of F. avenaceum, F. graminearum, and F. langsethiae (see isolate ID's above) were

produced on mung bean agar medium (Dill-Macky, 2003) at 22–23◦C under a combination of white and black light with 12 h photoperiod, and harvested after 10–15 days growth. The conidia were suspended in sterile water, and the spore concentrations were measured using KOVA <sup>R</sup> Glasstic <sup>R</sup> Slide 10 with Grids (Hycor Biomedical Inc.). Spore suspensions were stored at −20◦C. For DNA extraction, spores suspensions were thawed on ice and 3–5 ml of spore suspension was centrifuged for 15 min at 3220 × g. Most of the water was removed without disturbing the spore pellet, and DNA was extracted using the FastDNA <sup>R</sup> SPIN Kit for Soil (MP Biomedicals) as described above. The final concentration of these DNA standards was measured using a Qubit <sup>R</sup> dsDNA HS Assay (Invitrogen) and a Qubit <sup>R</sup> 2.0 Fluorometer (Invitrogen). The DNA standards were stored at −20◦C prior to use in qPCR.

Fusarium avenaceum, F. graminearum, and F. langsethiae DNA was quantified by qPCR as described above, except that in this case the analysis was performed on undiluted DNA extracts from air samples. The amount of fungal DNA in the air samples was quantified using a standard curve algorithm of five dilutions of conidial DNA from the respective species' standard. The dilutions were in the range of 0.1–400 pg F. graminearum DNA (equivalent to 0.6–2,565 conidia), and 0.2–800 pg of both F. avenaceum and F. langsethiae DNA (equivalent to 10–40,000 conidia for each species). The amount of fungal DNA was expressed as pg Fusarium DNA per week.

Little Fusarium DNA was detected in the spore traps prior to heading. In case this may have been a result of the spore-traps' inability to catch spores released from plant residues near the soil surface, only data from Fusarium DNA recorded from the week of heading onward were used in the subsequent analyses. The sampling in each field was terminated from 1 to 6 weeks prior to harvest.

#### Weather Data

For both field locations, weather data were collected from the nearest Agrometeorology Norway<sup>1</sup> weather station. Historical values, covering 1961 to 1990, for the weeks sampled, were provided by the Norwegian Meteorological Institute (Supplementary Table S1).

In order to identify possible associations between weather conditions and the weekly amounts of Fusarium DNA collected in spore traps, weather data were collected at each location from 2 weeks prior to cereal heading till the end of the spore trapping period. Hourly recorded data on temperature ( ◦C), precipitation (mm), relative humidity (%) and wind speed (m/s), were summarized into different weather variables using MATLAB (2013b). Daily precipitation and temperatures in the period from 2 weeks prior to cereal heading until end of spore trapping are presented in **Table 2**.

#### Statistical Analysis

The percentage infestation of F. avenaceum and F. graminearum on residue and the calculated IP per plot within each field experiment were subjected to statistical analysis. No analyses

<sup>1</sup>http://lmt.nibio.no

were performed for the IP of F. culmorum or F. langsethiae due to the low levels recorded in straw residues. Data from 2011 and 2012 were analyzed separately. Tillage treatment (nested within whole plots) and straw removal (whole plots, nested within blocks) were used as factors in the statistical model in which block was used as a random factor. Significant treatment effects were separated by applying proc mixed in SAS program for Windows (version 9.3, SAS institute inc.) creating pairwise comparisons and 95% confidence intervals according to Tukey– Kramer's method.

Regression analysis (R 3.2.0) were performed in order to identify possible associations between weather conditions and the weekly amount of F. avenaceum and F. graminearum DNA collected in spore traps. No analysis was performed for F. langsethiae due to the low DNA levels recorded. Weather data for air temperature (◦C), precipitation (mm) and relative humidity (%) was collected on an hourly basis but the regression conducted on blocks of data combined in three formats: from the week of spore collection, from the week prior to the start of the spore collection period, and from the 2 weeks prior to the start of the spore collection period. The average amount of Fusarium DNA detected in the two spore traps sampling within a single field during a 1 week period was used in the analysis.

Regression analysis (Minitab 16) were performed in order to identify possible associations between the percentage of oat straw residues within a field plot infested with F. avenaceum versus F. graminearum, and the amount of F. avenaceum DNA vs. F. graminearum DNA per gram straw residues within a field plot. As well to identify possible associations between the DNA content of F. avenaceum, F. graminearum, F. culmorum, or F. langsethiae per gram straw (determined by qPCR), and the percentage of the straw infested by these fungal species, respectively, within the same field plot. In the regression analysis, the natural logarithm (ln) of Fusarium DNA (pg DNA per gram plant residue) was used. qPCR values of F. langsethiae and F. culmorum were added a value of 1 prior to ln transformation due to low DNA levels.

## RESULTS

## Assessment of Straw Residues

The percentage of the soil area covered by oat straw residues differed between residue treatment and tillage regimes (**Table 1**). At both locations, a higher amount of straw was recorded in 2011 compared to 2012. In 2011, the average percentage of residue cover at Solør ranged from 2 to 49% and at Østfold from 1 to 45%, depending on the residue treatment and tillage regimes. In 2012, the average percentage of residue cover ranged from 0 to 23% at Solør, and from 0 to 20% at Østfold.

#### Fusarium Isolated from Residues

Fusarium avenaceum was the most prevalent Fusarium species isolated from straw residues at both field locations over the 3-year period (**Figures 1** and **2**). For both fields, the yearly average F. avenaceum infestation of residues increased during the experimental period (**Figure 2**). At Solør the percentage of straw residues infested with F. avenaceum increased from 42 to 94%


TABLE 2 | Minimum, maximum and mean daily precipitation (mm) and temperature (◦C) in the period from 2 weeks prior to cereal heading until end of spore trapping at two locations (Solør and Østfold) for 2 years (2011 and 2012).

FIGURE 1 | Percentage of oat straw residues from the previous years' crop from which Fusarium was isolated (A). The relative amount of DNA of different Fusarium species given in percentage of total Fusarium DNA quantified in these residues (B). The straw residues were collected from the soil surface at sowing at Solør and Østfold in 2011 and 2012.

during the experiment, and at Østfold, an increase from 58 to 75% was observed (**Figure 2**). The proportion of F. avenaceum infested straw residues ranged from 40 to 100% between the different treatments (median 76%) in 2011 and 2012 (**Figure 3A**).

Fusarium graminearum was the second most common Fusarium species isolated in this study (**Figure 1A**). The percentage of F. graminearum infestation on straw residues in spring declined from 2011 to 2012 at each location (**Figure 2**). The proportion of oat straw infested with F. graminearum ranged from 4 to 80% between the different treatments (median of 31%) across 2011 and 2012 (**Figure 3A**). In some location-years, a slightly lower Fusarium infestation of straw was observed on average for treatments where the straw was removed compared to those where straw was retained (**Figure 2**). This effect was most evident in the case of F. graminearum. However, the percentage of straw within a field infested by all Fusarium spp. was not significantly influenced by tillage or residue management treatments. The percentage of F. avenaceum infested straw was negatively correlated with the percentage of F. graminearum infested straw (R 2 adj = 27%, p < 0.001, **Figure 3A**).

Only low levels (less than 5%) of F. culmorum were detected on the straw residues at both locations (**Figure 2**). F. langsethiae was only detected in Østfold (2011 and 2012) and then only at average levels below 1% (**Figure 2**). F. tricinctum, F. cerealis, F. sporotrichioides, F. poae, and F. equiseti were also detected, but their average levels generally fell below 2% (data not shown). One exception was Østfold in 2012, where F. tricinctum was isolated from 5% of the oat straw residues sampled.

#### Fusarium DNA Quantified in Residues

The DNA of F. avenaceum, F. graminearum, F. culmorum, and F. langsethiae was quantified in straw residues by using qPCR (**Figure 1B**). In general, the highest levels of DNA were detected for F. avenaceum followed by F. graminearum. DNA of F. culmorum and F. langsethiae was detected at low levels, and accounted on average for less than 1% of the total DNA quantified for these four Fusarium species. DNA of F. langsethiae was sporadically detected in all fields, but always at low levels (0– 2.2 pg per gram straw residue within plot). DNA of F. culmorum was detected in all fields, and at variable levels (0–144 pg per gram straw residue within plot). For most samples, the DNA content of a particular Fusarium species was below 10,000 pg per gram straw residue (**Figures 4A,B**). One exception was the field at Solør in 2012, where the DNA content of F. avenaceum ranged from 12,620 to 51,695 pg per gram straw residue (**Figure 4A**). By comparison, DNA levels of F. langsethiae never exceeded 2.2 pg per gram straw residue (data not shown). A positive relationship was calculated between the quantities of DNA detected for F. avenaceum and F. graminearum calculated as ln pg Fusarium DNA per g plant residue (R 2 adj = 16%, p = 0.001, **Figure 3B**). For samples in which less than 90% of the straw residues were infested with a specific Fusarium species, the DNA content per gram straw ranged from 0 to 10,000 pg per gram straw (**Figures 4A,B**). For samples where more than 90% of the residues were infested with F. avenaceum, concentrations of F. avenaceum DNA were consistently above 10,000 pg/g (**Figure 4A**). A significant, positive association (R 2 adj = 54%, p < 0.001) was found between the DNA content for F. avenaceum

(ln pg DNA per g plant residue) and the percentage of residues infested by F. avenaceum within the same field plot. A low but significant positive association (R 2 adj = 24%, p < 0.001) was also found between the DNA content for F. culmorum ln (pg DNA per g plant residue + 1) and the percentage of residues infested by F. culmorum within the same field plot. No significant associations were found between the DNA content for F. graminearum and F. langsethiae and the percentage of residues infested by these Fusarium species within the same field plot.

### Inoculum Potential of Fusarium in Relation to Tillage and Residue Management

Tillage significantly influenced the IP of both F. avenaceum and F. graminearum at most locations, whereas no significant effects of straw removal were evident. However, significant interactions between straw removal and tillage regime were found within many fields. Therefore, a Tukey analysis was performed to identify significant differences in IP of F. avenaceum and F. graminearum between combined treatments (straw removal combined with tillage treatment).

Generally, higher IPs were estimated for harrowed compared to plowed plots (**Figure 5**). Within individual locations, the highest IPs were found on spring-harrowed plots where the straw was retained after harvest. In fields where the IPs exceeded 10% in spring-harrowed plots (mostly recorded in 2011, **Figure 5**), significant differences in IP were often recorded between harrowed and plowed plots. Moreover, in some of these spring-harrowed plots, straw removal significantly reduced the IP of F. avenaceum (Solør, 2011) and F. graminearum (Østfold, 2011; **Figure 5**). Hardly any significant effect of tillage on the IP was detected in fields where the Fusarium spp. were isolated from less than 25% of the residues sampled, as was the case for F. graminearum in Østfold and Solør in 2012 (**Figures 2** and **5**). Similarly, hardly any significant effect of soil cultivation was detected when the different tillage treatments resulted in a maximum plant residue cover below 20% as Østfold 2012 (**Table 1** and **Figure 5**). The lowest IPs (<5%) were estimated for plowed plots where the removal of straw residue or timing of tillage treatment had no apparent impact upon the Fusarium IP. No significant differences in IPs were found between spring and autumn plowed plots.

#### Fusarium in Air Samples

As no consistent differences were detected in the amount of Fusarium DNA collected in the two spore traps situated in plots with different tillage treatments within a field, average values of fungal DNA collected per week are presented for each field. F. langsethiae DNA was only detected at low levels, with an average maximum level of 12 pg DNA in week number 36 at Solør 2012 (**Figure 6**). F. langsethiae DNA was only recorded in the spore traps at late time points, from 4 weeks after heading. DNA of F. avenaceum and F. graminearum was detected at higher levels than that of F. langsethiae. At both locations, the amount of F. graminearum DNA was, however, less in 2012 than in 2011 and only low levels of F. graminearum DNA were detected at Østfold in 2012. For F. avenaceum, the highest DNA levels were detected at Solør, with an average maximum of 51 pg in week

FIGURE 5 | Effects of tillage treatment and residue regimes on the inoculum potential (IP) of F. graminearum and F. avenaceum in two experimental fields of oats in southeast Norway (Solør and Østfold) in 2011 and 2012. Tillage treatments include DAP, deep autumn plowing; SSP, shallow spring plowing; SAH, shallow autumn harrowing; SSH, shallow spring harrowing. Data from the two residue regimes examined are indicated by either hatched bars, where the straw was removed, or filled bars, where the straw was chopped and retained in the field. IP was calculated as the relative soil area covered with oat residues after sowing in spring (0–1) multiplied by the percentage of the residues infested with the respective Fusarium species. Different letters indicate significant treatments effects at p = 0.05.

36, 5 weeks after heading, in 2012. Only low levels were detected at Østfold in both years. The average amount of F. avenaceum and F. graminearum DNA recorded increased significantly in the sampling periods after heading in spore-traps at both locations in 2011 and 2012 (**Figure 6**). The highest F. graminearum DNA amounts were recorded in 2011, with an average maximum of 100 pg DNA in week number 31 in the two spore traps located at Østfold (**Figure 6**).

## Fusarium Dispersal in Relation to Weather Conditions

Wetness and moisture in the period prior to spore sampling influenced the amount of F. avenaceum DNA in spore traps sampling from heading onward. The best regression model to explain the variations in F. avenaceum DNA comprised the number of days with precipitation in the two consecutive weeks prior to spore sampling, the mean daily hours with relative humidity exceeding 70% during the week of spore sampling and their interaction. This model accounted for 53% of the variation in the data, with an R 2 adj = 47%. Both main terms in the model were positive, giving a positive relationship for both factors separately. However, only the precipitation term was significant (p < 0.05). The interaction term was negative (p < 0.05), indicating that the effects of these factors were not additive.

Lower associations were generally found between weather conditions and the DNA of F. graminearum collected. The best model to explain the variation in F. graminearum DNA included

data on the total precipitation during the two consecutive weeks prior to spore sampling, the mean air temperature during the week of spore sampling and the mean air temperature 1 week before spore sampling, in addition to the four interaction terms. The model described 59% of the variation in the data, with an R 2 adj = 46%. All the main factors in the model were positive, but only the precipitation factor was significant (p < 0.05). The three factor interaction was positive and significant (p < 0.05), while all the two factor interactions were negative and only significant when precipitation was included in the interaction. The relative amount of airborne Fusarium inoculum increased after rainy periods and corresponded, in most cases, with the IP calculated within a field in spring.

#### DISCUSSION

The objective of this work was to elucidate the influence of different tillage and straw coverage regimes on the IP and dispersal of Fusarium spp. in Norwegian spring oats. In general, the amount of crop residues left on the soil surface and thus, the IP, was significantly lower on plowed compared to harrowed plots. On harrowed plots, the removal of straw prior to harrowing reduced the IP. Differences in the relative Fusarium spp. infestation of oat straw were detected between locations and years. The relative amount of airborne Fusarium inoculum increased after rainy periods and corresponded in most cases with the IP calculated within a field in spring. F. langsethiae was only detected sporadically and at low levels in residues (DNA and fungal growth) and spore traps (DNA).

Soil cultivation clearly influenced the Fusarium IP. For both F. avenaceum and F. graminearum, the lowest IPs were found on the plowed plots and the highest IPs on unplowed, spring harrowed plots. This is in agreement with other studies that, based on analysis of crop stubble, reported reduced inoculum levels of F. graminearum in plowed fields /experimental plots compared to those established with reduced or minimum tillage (Dill-Macky and Jones, 2000; Guo et al., 2010). In our study, the overall Fusarium IP was most closely related to the level of Fusarium infested straw within a field, with no large differences in the percentage of Fusarium infested straw recorded between plots subjected to different tillage treatments within a field. The observed differences in Fusarium IP between treatments within a field were more closely related to the amount of straw residues on the soil surface. This finding is supported by Maiorano et al. (2008), who observed that the amount of Fusarium spp. in harvested wheat grain mainly corresponded to the amount of residues lying on the soil surface of the production field, rather than the tillage regime. In our study, significant effects of tillage treatments on IPs were largely recorded in 2011. The amount of straw residues left on the soil surface after tillage operations in the spring of 2011 was greater than in spring 2012, which likely explains the greater difference observed in 2011. Similarly Koch et al. (2006) reported only slight differences in Fusarium disease development between tillage treatments when residue cover was below 30%. In our study, significantly lower IPs were often recorded for autumn harrowed plots compared to spring harrowed plots within the same field, although the time of plowing (autumn vs. spring) did not significantly influence the IP of the plowed treatments. The observed differences thus largely reflect the variation in straw cover resulting from these various tillage methods (Seehusen, 2014).

Removal of straw residues generally reduced the IP in harrowed treatments, though the effect was not always significant. By contrast, the removal of straw residues did not significantly influence the IP on the plowed treatments. The reason for the different effect of straw removal in harrowed vs. plowed treatments is most probably the generally low amount of residues, and the correspondingly low IPs, in the plowed treatments. Our results are in line with the general understanding of plowing as the best means of reducing the occurrence, and thereby the IP, of Fusarium spp. in cereal fields (Dill-Macky and Jones, 2000; Champeil et al., 2004b; Guo et al., 2010). The effect of tillage on FHB development may not always be evident, as weather factors play an important role in the dispersal and infection of Fusarium spp. (Lori et al., 2009; Prussin et al., 2014a). According to our data, spring plowing appears to be the best option to both reduce the risk of soil erosion and at the same time minimize Fusarium diseases. Removal of cereal straw in autumn would be the best practice for reducing the IP of Fusarium spp. where harrowing is the preferred tillage system.

In our study, F. avenaceum was the dominant Fusarium species recovered from oat residues, followed by F. graminearum. In contrast, F. culmorum was detected in only a few of the residues sampled. These Fusarium species have been recorded on wheat residues elsewhere (Köhl et al., 2007; Golkari et al., 2008; Landschoot et al., 2011; Postic et al., 2012), as well as on oat residues in Canada (Golkari et al., 2008). High levels of F. sporotrichioides were also detected on oat residues in the Canadian study. The discrepancy between the Norwegian and Canadian studies may be explained by the generally low levels of F. sporotrichioides normally detected in Norwegian cereals (Bernhoft et al., 2010). F. langsethiae was detected in our field trials, but at exceedingly low levels despite the fact that this fungal species and the mycotoxins it produces are commonly detected in oats grown in Norway (Aamot et al., 2013; Hofgaard et al., 2016). The Fusarium species detected on the oat residues in this study reflected those most commonly detected in Norwegian grain, other than F. langsethiae (Kosiak et al., 2003; Henriksen and Elen, 2005; Bernhoft et al., 2010; Norwegian Scientific Committee for Food Safety, 2013; Hofgaard et al., 2016). This supports previous findings that have identified crop residues as an important source of inoculum for FHB (Dill-Macky and Jones, 2000).

The general increase in F. avenaceum infestation of crop residues in our fields during the project period is in agreement with another report of increased F. avenaceum infestation of crop residues in monoculture cereals (Fernandez et al., 2008). F. avenaceum infestation of wheat residues is reported to be more stable over time than that of F. graminearum (Köhl et al., 2007; Hogg et al., 2010; Palazzini et al., 2013). Similarly while we observed a yearly increase in the relative prevalence of F. avenaceum, we did not see F. graminearum increase on straw residues over the period of this study. Infestation of crop residues

by plant pathogens may be related to the establishment of fungi in host tissues prior to crop senescence (Bruehl and Lai, 1966; Köhl et al., 2007; Hogg et al., 2010). The low F. graminearum infestation of residues in 2012 does not appear to be explained by a lack of inoculum the previous year, as the DNA levels in spore traps were high, but the lower infestation of residues by F. graminearum may be explained by the relatively late peak spore dispersal, compared to the cereal flowering period, that would likely have resulted in relatively few FHB infections.

Fusarium avenaceum, had less marked peaks of spore dispersal than F. graminearum in 2011, but there was a relative stable dispersal of spores from shortly before heading till the end of the spore sampling period. Since we were not able to differentiate between spore types (ascospores vs. conidia), we elected to use the amount of DNA (in pg), as an indication of the relative abundance of the different species. The use of DNA as a measure makes it difficult to extrapolate the relative number of spores of the two species as the DNA level in spores may vary with species and spore type. Our result do, however, suggest that F. avenaceum produced inoculum more consistently and over a longer period, than F. graminearum, in 2011.

The differences we detected in the relative prevalence of Fusarium species on crop residues in spring, may be related to the establishment of the fungi as pathogens in the living plant. The infection of above ground plant tissues by Fusarium, and thus the infestation of crop residues detected later, is likely dependent upon the prevailing weather conditions from flowering until harvest (McMullen et al., 2012). The monthly average temperature during cereal flowering (July) in our field trials in 2010–2012, ranged from 14.8 to 17.4◦C. This is closer to the optimal growth temperature for F. avenaceum (20◦C) compared to F. graminearum (25◦C; Brennan et al., 2003). Although F. graminearum has a higher in vitro growth rate compared to F. avenaceum in the range 10–30◦C, the growth rate of F. avenaceum is reported to be less affected by temperature than that of F. graminearum (Brennan et al., 2003). Consequently, in addition having temperatures closer to the optimum for growth of F. avenaceum, small fluctuations in the temperature likely influenced the growth and establishment F. avenaceum less than F. graminearum in our trials.

The lower proportion of F. graminearum infested straw at all locations in 2012 compared to 2011 may also be explained by the influence of weather conditions between harvest and the sowing of the subsequent crop. The competition between Fusarium and other microorganisms in crop residues is influenced by environmental conditions, especially temperature and moisture (El-Naggar et al., 2003; Lakhesar et al., 2010; Leplat et al., 2013). A sharp reduction in Fusarium infestation on cereal residues has been observed in spring, and survival of F. graminearum seems inversely related to the decomposition of residues (Pereyra et al., 2004; Köhl et al., 2007). The recovery and sporulation of plant pathogens from residues have been reported to decrease in warm and wet conditions, probably due to an increased decomposition rate combined with increased antagonistic activity (Zhang and Pfender, 1992; Lakhesar et al., 2010). In our study, the weather at Solør and Østfold in September 2011 was warmer (11.6 and 12.7◦C, respectively) and wetter (119 and 236 mm, respectively) compared with the previous year with average temperatures of 9.6C and 10.9◦C and rainfall of 44 and 99 mm, respectively. In addition, the precipitation before sowing in the spring (April) at Solør and Østfold was higher in 2012 (28 and 87 mm, respectively) than in 2011 (15 and 45 mm, respectively) and this may also have increased the activity of antagonistic microorganisms. We surmise that the relatively warm and moist conditions during autumn 2011 along with the moist conditions in spring 2012 may have facilitated both the decomposition of plant residues and competition by other saprophytes that reduced the survival of F. graminearum in those residues.

Although not significant, we observed a consistent trend of a higher percentage of F. graminearum infested oat straw residues on plots where the straw was retained after harvest than on plots where most of the straw was removed. This may be explained by a relatively lower proportion of the residues in direct contact with the soil when the straw is retained after harvest as F. graminearum survives and reproduces better on surface residue than on buried residue (Khonga and Sutton, 1988; Pereyra et al., 2004). A comparison of F. graminearum infested residues in autumn vs. spring would have been of interest.

We found a negative association between the infestation of residues by F. avenaceum and F. graminearum. This finding is similar with the relative abundance reported for these two species in Norwegian grain determined using morphological methods (Kosiak et al., 2003; Bernhoft et al., 2010). The negative association may be indicative of competition between these two fungal species. F. graminearum has a faster growth rate than F. avenaceum at the temperatures used during the incubation in our study (Brennan et al., 2003), which may explain why we detected a reduced F. avenaceum infestation in residues that were highly infested with F. graminearum. This explanation is supported by the qPCR analysis, where a low but significant positive association was detected between the DNA content of F. avenaceum vs. F. graminearum in the residues examined. This suggests that the observed data on infestation of cereals or plant residues by different Fusarium species may be influenced by growth rates of the individual Fusarium species in the sample.

Despite the use of both morphological and molecular methods, little F. langsethiae was detected in the residues collected in our study. Furthermore, we could not find any association between the IP of F. langsethiae in crop residues and the amount of fungal DNA detected in the spore traps. The epidemiology of F. langsethiae is unclear (Imathiu et al., 2013a), and our study adds little to our understanding of the sources of inoculum of F. langsethiae.

The relative amounts of total Fusarium spp. isolated from the residues in this study corresponded with the total DNA quantified of the Fusarium species in residues. A significant association was found between the DNA content of F. avenaceum, and the percentage of straw infested by F. avenaceum. DNA content of F. culmorum was slightly associated with the percentage of straw infested by F. culmorum. However, no associations were found between the DNA content of F. graminearum or F. langsethiae and the percentage of straw infested by these fungal species. This poor correlation may be partly due to variation within the sample as the Fusarium DNA was quantified from the residues

remaining after straw was selected from the sampled straw for isolation and morphological analysis. The qPCR was initiated to investigate if F. langsethiae, detected at extremely low levels in our isolations was being out-competed by the faster growing Fusarium species. Additionally the qPCR enabled finer scale quantification of Fusarium species, whereas in the isolations the percentage of straw infested with a specific Fusarium was limited by the number of residue pieces examined. On the contrary, qPCR quantified the amount of fungal DNA irrespective of how this DNA was distributed in a sample or whether the fungus was dead or alive. qPCR has been used by others to record the change in the relative abundance of Fusarium species in plant residues over time (Köhl et al., 2007; Hogg et al., 2010). However, it is recognized that obtaining amplifiable DNA from crop residues may be challenging (Imathiu et al., 2013b). We suggest that additional studies need to be undertaken to investigate whether molecular methods such as qPCR are appropriate to quantify the IP of Fusarium in cereal residues.

According to our regression models, the DNA and thus the airborne spores of F. avenaceum and F. graminearum increased after rain periods. The amount of F. avenaceum DNA collected in the spore traps increased with the number of rain days in the previous 2 weeks, and with higher mean daily hours with RH > 70% during the week of spore sampling. Increases in the amount of airborne F. graminearum spores was, by contrast, related more closely to the total precipitation in the 2 weeks prior to spore sampling. Although the variables that influenced spore dispersal varied, an increase in the DNA sampled of both these fungal species was related to an increase of humid/wet conditions, which corresponds well with other studies (de Luna et al., 2002; Xu et al., 2008; Manstretta and Rossi, 2015). For F. graminearum, model performance also increased when data for mean air temperature the week before and during spore sampling were included, whereas the strength of the F. avenaceum model did not increase when temperature data was included. This may indicate that growth and sporulation of F. avenaceum was less affected by temperature within the temperature range (9– 21◦C) of this study than was F. graminearum. F. avenaceum is known to have a lower optimum temperature for conidia production than F. graminearum (Rossi et al., 2002). However, since we did not examine the samples morphologically, we cannot know if conidia or ascospores were being collected in our study. Differences in temperature response was also reflected in reports of in vitro growth of these two fungal species (Brennan et al., 2003). Our findings are supported by studies demonstrating that F. graminearum perithecia number and development was enhanced by increasing temperatures between 12 and 20◦C (Dufault et al., 2006). Our study indicate that the release of spores of F. graminearum increased with increasing rain combined with increasing temperatures up to 21◦C. The higher levels of F. graminearum detected in 2011, may be explained by higher temperatures during the period of spore trapping in this year of the study.

The relative amount of airborne Fusarium inoculum recorded in a field in our study from before heading till the end of the spore trapping period, generally corresponded to the IP in the treatments that were harrowed in the spring. The Fusarium DNA was most abundant in the air samples collected after heading. As the air intake of the spore traps was at 1 m height above ground level, our spore traps may be unsuitable for capturing spores released from Fusarium-infested residues in the immediate vicinity. Inoculum generated from residues may be better captured by using sampling nearer to the residue (Munger et al., 2014; Prussin et al., 2014b). We detected no consistent differences in the amount of Fusarium DNA collected in the spore traps situated in plots with different tillage treatments, which may indicate that the data are more representative of the background inoculum in the field rather than the inoculum of individual treatments. Small and inconsistent differences in F. graminearum colony-forming units were also detected in Petri dish spore traps placed in plots with different tillage and cropping systems in a Canadian study (Munger et al., 2014). A comparison of F. graminearum and F. avenaceum DNA amounts from different types of spore-traps co-located would be of interest, as well as a more thorough study of the kind of spores (conidia or ascospores) collected by the different spore traps.

Disease development and the contamination of grain by Fusarium spp. and mycotoxins were not examined in our study, and therefore, we can only speculate on whether the differences in Fusarium IPs influenced these variables. However, other studies have described relationships between cropping practices, inoculum levels and development of FHB in monoculture cereals (Guo et al., 2010). An effect of tillage regimes was also reported to explain differences in Fusarium spp. isolated from cereal grains of barley and oats in a Norwegian study (Henriksen, 1999), which indicated that the IP within a field impacts the level of Fusarium spp. detected in grain harvested from that field. The effect of plowing on reducing the Fusarium inoculum within a field is frequently unclear (Miller et al., 1998; Munger et al., 2014). That Fusarium inoculum may be transported aerially over large distances is well recognized (Lori et al., 2009; Prussin et al., 2014a) suggesting that inoculum delivered by long-distance spore dispersal may overwhelm local sources of inoculum. Presumably, inoculum pressure over a large area can be reduced if sources of inoculum within a region are diminished through the cumulative effects of reducing inoculum at the individual field level.

## CONCLUSION

The Fusarium species most prevalent on oat residues in this study reflected those species most commonly detected in Norwegian oat grains although F. langsethiae, which is often isolated from Norwegian oats, was rarely detected. Our findings support previously published work that have identified crop residues, principally straw, as an important source of Fusarium inoculum. Our results are also in line with the general understanding of plowing as a means of reducing the IP of Fusarium spp. In areas where spring plowing is feasible, this appears to be the best option for Norwegian growers to reduce the risk of soil erosion while minimizing the risk of Fusarium infection. If harrowing is preferred, removal of the straw of a cereal crop prior to harrowing should aid in reducing the IP of Fusarium spp. This is the first report of F. graminearum, F. avenaceum, and F. langsethiae

recorded in air sampled in cereal fields in Norway. The amount of airborne Fusarium increased after rainy periods, a finding which corresponds well with other studies. As airborne Fusarium detected at flowering generally corresponded with the IP of straw residues measured in the spring of our study, it appears that tillage practices aimed at reducing straw residues could be recommended to reduce the subsequent risk of Fusarium infection of cereal hosts and the subsequent contamination of grain by mycotoxins. Our results will be valuable in the development of agricultural decision support systems aimed at reducing the risk of Fusarium infection in cereals in Norway.

#### AUTHOR CONTRIBUTIONS

IH: Responsible for coordinating the work (planning, implementation, and interpretation). Responsible for writing this manuscript. TS: Responsible for coordinating the work connected the field trials. Responsible for recording the proportion of the soil surface area covered with straw residue. Involved in discussions and the writing of this manuscript. HA: Responsible for all DNA analysis. Involved in planning, discussions and in the writing of this manuscript. HR: Involved in the work connected to planning and implementation of the field trials, discussions and in the writing of this manuscript. JR: Responsible for all morphological analysis of Fusarium. Involved in planning, discussions and in the writing of this manuscript. VL: Responsible for the technical equipment and for coordinating the work connected to the collection of debris for Fusarium analysis and of bio-aerosols from spore traps. Involved in planning and discussions. A-GH: Responsible for analyzing the association between weather conditions and Fusarium inoculum

#### REFERENCES


collected in spore traps. Involved in discussions and in the writing of this manuscript. RD-M: Involved in the planning of this work, discussions, interpretation of the results, and in the writing of this manuscript. GB: Project leader. Involved in the planning of this work, in discussions regarding methods and implementation of the results, and in the writing of this manuscript.

### FUNDING

This work was financed by The Foundation for Research Levy on Agricultural Products/Agricultural Agreement Research Fund/Research Council of Norway (research grant 199412/E50) with support from the industry partners Animalia, Bayer Crop Science, Braskereidfoss kornsilo, Felleskjøpet Agri, Felleskjøpet Rogaland og Agder, Fiskå Mølle, Flisa Mølle og Kornsilo, Graminor, Lantmännen Cerealia, Norgesfôr, Norgesmøllene, Norkorn.

#### ACKNOWLEDGMENT

We thank the agricultural extension service groups at Østfold and Solør and the technical staff at NIBIO for the implementation of the field experiments.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fmicb. 2016.00556

and the production of mycotoxins by Fusarium in wheat grains. Plant Sci. 166, 1389–1415. doi: 10.1016/j.plantsci.2004.02.004


different temperatures. Acta Phytopathol. Entomol. Hungarica 38, 275–280. doi: 10.1556/APhyt.38.2003.3-4.8


cereals on silty clay loam and sandy loam soils in the cool, wet climate of central Norway. Soil Tillage Res. 80, 79–93. doi: 10.1016/j.still.2004.03.005


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2016 Hofgaard, Seehusen, Aamot, Riley, Razzaghian, Le, Hjelkrem, Dill-Macky and Brodal. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Fusarium Head Blight Resistance QTL in the Spring Wheat Cross Kenyon/86ISMN 2137

Curt A. McCartney <sup>1</sup> \*, Anita L. Brûlé-Babel <sup>2</sup> , George Fedak <sup>3</sup> , Richard A. Martin<sup>4</sup> , Brent D. McCallum<sup>1</sup> , Jeannie Gilbert <sup>1</sup> , Colin W. Hiebert <sup>1</sup> and Curtis J. Pozniak <sup>5</sup>

<sup>1</sup> Agriculture and Agri-Food Canada, Morden Research and Development Centre, Morden, MB, Canada, <sup>2</sup> Department of Plant Science, University of Manitoba, Winnipeg, MB, Canada, <sup>3</sup> Agriculture and Agri-Food Canada, Ottawa Research and Development Centre, Ottawa, ON, Canada, <sup>4</sup> Agriculture and Agri-Food Canada, Charlottetown Research and Development Centre, Charlottetown, PEI, Canada, <sup>5</sup> Crop Development Centre, University of Saskatchewan, Saskatoon, SK, Canada

Fusarium head blight (FHB), caused by Fusarium graminearum, is a very important disease of wheat globally. Damage caused by F. graminearum includes reduced grain yield, reduced grain functional quality, and results in the presence of the trichothecene mycotoxin deoxynivalenol in Fusarium-damaged kernels. The development of FHB resistant wheat cultivars is an important component of integrated management. The objective of this study was to identify QTL for FHB resistance in a recombinant inbred line (RIL) population of the spring wheat cross Kenyon/86ISMN 2137. Kenyon is a Canadian spring wheat, while 86ISMN 2137 is an unrelated spring wheat. The RIL population was evaluated for FHB resistance in six FHB nurseries. Nine additive effect QTL for FHB resistance were identified, six from Kenyon and three from 86ISMN 2137. Rht8 and Ppd-D1a co-located with two FHB resistance QTL on chromosome arm 2DS. A major QTL for FHB resistance from Kenyon (QFhb.crc-7D) was identified on chromosome 7D. The QTL QFhb.crc-2D.4 from Kenyon mapped to the same region as a FHB resistance QTL from Wuhan-1 on chromosome arm 2DL. This result was unexpected since Kenyon does not share common ancestry with Wuhan-1. Other FHB resistance QTL on chromosomes 4A, 4D, and 5B also mapped to known locations of FHB resistance. Four digenic epistatic interactions were detected for FHB resistance, which involved eight QTL. None of these QTL were significant based upon additive effect QTL analysis. This study provides insight into the genetic basis of native FHB resistance in Canadian spring wheat.

Keywords: Fusarium head blight, Fusarium graminearum, wheat, Triticum aestivum L., SNP, QTL, linkage

## INTRODUCTION

Fusarium head blight (FHB), primarily caused by Fusarium graminearum Schwabe (teleomorph: Gibberella zeae (Schwein.) Petch), is one of the most serious diseases of wheat. FHB lowers grain yield, grain quality, and contaminates grain with the trichothecene mycotoxin deoxynivalenol, and its acetylated derivatives 3-ADON and 15-ADON (Ward et al., 2008). FHB damage reduces functional performance of wheat for bread and noodle production (Dexter et al., 1996; Hatcher et al., 2003) and durum wheat (Dexter et al., 1997). Trichothecenes are a virulence factor for the pathogen and have multiple inhibitory effects on eukaryote cells, which are harmful to the plant

#### Edited by:

Daniela Gwiazdowska, Poznan University of Economics and ´ Business, Poland

#### Reviewed by:

Chandra Nayak, University of Mysore, India Venkataramana M, DRDO-Bharathiar University Center for Life Sciences, India

> \*Correspondence: Curt A. McCartney curt.mccartney@agr.gc.ca

#### Specialty section:

This article was submitted to Fungi and Their Interactions, a section of the journal Frontiers in Microbiology

Received: 29 April 2016 Accepted: 15 September 2016 Published: 13 October 2016

#### Citation:

McCartney CA, Brûlé-Babel AL, Fedak G, Martin RA, McCallum BD, Gilbert J, Hiebert CW and Pozniak CJ (2016) Fusarium Head Blight Resistance QTL in the Spring Wheat Cross Kenyon/86ISMN 2137. Front. Microbiol. 7:1542. doi: 10.3389/fmicb.2016.01542 host and any humans and animals consuming contaminated grain (Proctor et al., 1995; Rocha et al., 2005). FHB has been a problem in eastern Canadian wheat since the 1980s, and only became a significant problem in western Canada in 1993, particularly in the province of Manitoba (Gilbert and Tekauz, 2000). In 2014, FHB caused substantial damage in the province of Saskatchewan (https://www.grainscanada.gc.ca/str-rst/fusarium/ data/frequency-en.htm, accessed 25 April 2016), which had been largely unaffected by FHB until 2014.

Host FHB resistance is an important control measure and wheat breeding objective. The genetic basis of FHB resistance in Asian spring wheats has been the focus of many genetic studies. Buerstmayr et al. (2009) reported 52 QTL for FHB resistance in a comprehensive review of published research. Similarly, (Liu et al., 2009) and (Löffler et al., 2009) performed meta-QTL analyses and identified 43 QTL clusters and 19 important QTL, respectively. A few FHB resistance QTL have been studied in isolation from other FHB resistance QTL, which enabled the resistance controlled by these QTL to be treated as a qualitative trait and mapped as discrete loci, Fhb1 (Cuthbert et al., 2006; Liu et al., 2008) and Fhb2 (Cuthbert et al., 2007).

The genetic basis of FHB resistance in Canadian spring wheat germplasm is not well understood. Native FHB resistance is a topic that has gained interest as wheat breeders have struggled to make progress in improving FHB resistance using exotic resistance sources. There is a strong need for wheat breeders to understand the basis of FHB resistance already present in their programs so that introgression of FHB resistance QTL from exotic germplasm is targeted. In western Canada, a number of hard red spring wheat varieties have been identified that possess an intermediate level of FHB resistance relative to more highly susceptible wheat varieties, but do not have Asian sources of FHB resistance in their pedigrees. Such Canadian varieties include AC Barrie, CDC Bounty, AC Cadillac, AC Cora, Journey, Kane, Katepwa, McKenzie, Neepawa, 5500HR, 5601HR, and 5602HR (Gilbert, unpublished data). This FHB resistance may come from Frontana, which is present in the pedigree of many of these wheats (Gilbert and Tekauz, 2000). This study examines the genetic basis of FHB resistance in the Canadian spring wheat variety Kenyon.

### MATERIALS AND METHODS

#### Population

A F9-derived recombinant inbred line (RIL) population consisting of 125 lines from the cross Kenyon x 86ISMN 2137 was tested in this study. Each RIL was generated by single seed descent from a unique F2 individual. Kenyon is a Canada Western Red Spring (CWRS) variety with the pedigree Neepawa<sup>∗</sup> 5/Buck Manantial (Hughes and Hucl, 1991). Neepawa was the most widely grown CWRS variety in the 1970s to mid-1980s in western Canada (McCallum and DePauw, 2008), and a prominent in the pedigree of many current CWRS wheat varieties. Buck Manantial was the source of the leaf rust resistance gene Lr16 in Kenyon. 86ISMN 2137 is a spring wheat line with resistance to tan spot (Singh and Hughes, 2005), Septoria nodorum blotch (Feng et al., 2004), and Septoria tritici blotch (McCartney, unpublished data), however, its pedigree is unknown.

#### Phenotyping

The RIL population, parents, and checks were tested in six environments over 2012 and 2013 (Carman MB 2012, Winnipeg MB 2012, Carman MB 2013, Charlottetown PEI 2013, Ottawa ON 2013, and Winnipeg MB 2013). The check lines were 5602 HR, 93 FHB 37, AC Barrie, AC Cora, AC Morse, AC Vista, CDC Teal, ND 2710, Neepawa, Roblin, Shaw, and Snowbird. Entries were randomized in an alpha lattice with 12 incomplete blocks of 12 plots each, with 3 replicates per environment. The experimental unit was a 1 m row.

The date of 50% anthesis was recorded for each plot in all locations, except Charlottetown, PEI. On this date, anthesis had begun for 50% of the main tillers in that plot. Carman and Winnipeg tests were spray inoculated twice [on the recorded anthesis date and 2–3 days later (confirmed for Carman and Winnipeg)]. The inoculum consisted of the isolates M7-07-1 (3- ADON), M9-07-1 (3-ADON), M1-07-2 (15-ADON) and M3- 07-2 (15-ADON) for the Carman nurseries and the Winnipeg 2012 nursery. In Winnipeg 2013, the isolates used were M9- 07-01 (3-ADON) and M1-07-02 (15-ADON). The inoculum concentration was adjusted to 50,000 macroconidia per L and applied at a rate of 50 ml per row with a backpack sprayer. During the period the population was being inoculated, a misting system irrigated each day for 10 min every hour over a 12 h period (6:00 pm–6:00 am) in Carman. This misting system kept the plots wet overnight to promote F. graminearum infection. In Winnipeg, plots were irrigated with a sprinkler head irrigation system for 1 h following inoculation. The goal of this irrigation was to maintain wet soil to result in a humid micro-environment in the nursery, rather than to directly apply water to the canopy over the entire night.

In PEI, five F. graminearum isolates were used to produce macroconidia for inoculation. Inoculum was prepared to 50,000 macroconidia per ml and inoculated on plots three times per week over the course of flowering in the trial. Inoculum was applied with a standard pesticide sprayer delivering 200 L/ha water. The field was misted for 2 min bursts at a rate of 650–700 L/ha; misting was done every half hour from hour 7:00–10:30, at 15 min intervals thru to 19:00, half hour intervals to 21:00 and then on an hourly basis until 7:00. Misting nozzles were Naan Dan 501/2 (yellow) (Southern Drip Irrigation Ltd, Lethbridge, Alberta) spaced 10 feet apart in rows which ran down the center of the plots to ensure complete and overlapping coverage.

The Ottawa FHB nursery was inoculated by spreading infected corn and barley kernels on the soil surface as described by Xue et al. (2006). The corn and barley was inoculated with a mixture of three F. graminearum isolates (DAOM178148, DAOM232369, and DAOM212678; Canadian Collection of Fungal Cultures, AAFC, Ottawa, Canada), colonized, and then dried down. Inoculated corn is spread between the rows at a rate of 80 grams/meter<sup>2</sup> at about 6 weeks after planting. Plots were irrigated twice daily for 30 min with irrigation sprinklers to promote conditions favorable for infection by F. graminearum.

Plant height, anthesis date, FHB incidence, and FHB severity data were collected. Plant height was measured from the soil surface to the top of main tiller spikes (excluding awns if present). FHB incidence was the percentage of spikes with visual FHB symptoms. FHB severity was the percentage of the spike with visual FHB symptoms, when only considering diseased spikes. FHB incidence and severity data were converted into FHB visual rating index (VRI). VRI = (FHB incidence <sup>∗</sup> FHB severity)/100.

### Statistics

Least squares means were calculated for anthesis date, plant height, and VRI with JMP Genomics 6.0 (SAS Institute Inc.) using a mixed model. Wheat lines were considered fixed effects, and environment, rep, and incomplete block were considered random effects.

#### Genotyping

Ninety-seven RILs were genotyped with a combination of microsatellite, diversity array technology (DArT) (Akbari et al., 2006), and single nucleotide polymorphism (SNP) markers. Genomic DNA was extracted from freeze-dried leaf tissue with the DNeasy 96 Plant Kit (Qiagen, Toronto, Canada). DNA was quantified with PicoGreen stain (Molecular Probes, Inc., Eugene, Oregon, USA). SNP markers were genotyped on the RIL population and parents using the Illumina Infinium 9K wheat SNP beadchip (Illumina, San Diego, CA) (Cavanagh et al., 2013). The raw data were analyzed with GenomeStudio V2011.1 software (Illumina, San Diego, CA). The genotype calls from GenomeStudio were converted into allele scores for linkage mapping in Excel. Markers with greater than 10% missing data or strong segregation distortion were excluded from mapping. Microsatellite markers were tested as previously described (McCartney et al., 2004).

#### Linkage and QTL Analysis

The linkage map was developed with MapDisto version 1.7.7 (Lorieux, 2012). Linkage groups were initially formed with stringent LOD and recombination fraction thresholds and gradually relaxed to a minimum LOD score of 4 and a maximum recombination fraction of 0.20 cM. Loci were ordered using a combination of the "AutoMap," "Order sequence," and "Compare all order" functions. The "Branch and Bound II" and "Seriation II" ordering methods were used in combination with the sum of adjacent recombination fractions (SARF) and count of crossover events (COUNT) as fitting criteria. For each linkage group, the shortest linkage map was selected from the linkage map solutions generated using these different mapping algorithms and criteria. The Kosambi mapping function was used to calculate map distances (cM) from recombination fractions.

QTL analysis was conducted with QTL IciMapping version 4.0.6.0 (Li et al., 2008) using interval mapping (IM) and inclusive composite interval mapping (ICIM). For linkage bins with more than one marker, a single marker was selected with the least missing data to represent the linkage bin for QTL analysis. Analysis for additive effect QTL was conducted with 0.1 cM steps and the 5% LOD significance threshold was calculated with 10,000 permutations. The LOD threshold for declaring QTL was 3.22 for additive effect QTL analysis for both IM and ICIM based upon this permutation analysis. Additive effect QTL were reported when the LOD score exceeded 3 in two or more environments, or one or more environments plus the pooled dataset, based upon IM or ICIM. For the reported QTL, QTL statistics were reported for environments in which the LOD score exceeded 2.

Analysis for epistatic QTL was conducted with 2.0 cM steps and a default LOD significance threshold of 5.0. Determining a 5% LOD significance threshold by permutation analysis was not possible for epistatic QTL analysis because of the computational power required. Linkage maps and QTL scans were illustrated with MapChart v. 2.2 (Voorrips, 2002). For anthesis date and plant height, epistatic QTL were reported when the LOD exceeded 3.5 in four or more environments, or three environments plus the pooled dataset, based upon IM or ICIM. For FHB resistance, epistatic QTL were reported when the LOD exceeded 3.5 in seven or more combinations of environment (individual environments or pooled dataset) by traits (VRI, FHB incidence, FHB severity). The more stringent criteria was applied to FHB resistance because the three FHB resistance traits were pooled for considering QTL. For the reported QTL, QTL statistics were reported for environments in which the LOD score exceeded 3.5.

## Cytology

To confirm the presence of a reciprocal translocation, immature spikes were harvested from F<sup>1</sup> plants. Spikes were fixed (9 95% ethanol: 6 chloroform: 1 glacial acetic acid) at −20◦C for 24 h and stored in 70% ethanol at −20◦C, changing the ethanol once a day for 3–4 days. Anthers were macerated in acetocarmine to liberate pollen mother cells (PMCs) and stain the chromatin. PMC preparations were warmed on a hot plate and gently squashed to spread the chromosomes. Cells in metaphase I of meiosis were analyzed under a compound microscope to confirm the presence of quadrivalents.

## RESULTS

### Phenotypic Analysis

Trait data histograms for pooled datasets of the Kenyon/86ISMN 2137 RIL population are reported in **Figure 1**. Histograms for anthesis date, plant height, VRI, FHB incidence, and FHB severity from each FHB nursery are presented in Supplementary Figures S1–S5, respectively. All traits were approximately normally distributed. The earliest and latest RILs flowered within 8.3 days of each other (**Figure 1**, Supplementary Table S1). Kenyon and 86ISMN 2137 had very similar flowering dates, but some transgressive segregation was present in the population. Plant height varied widely with the shortest and tallest RILs differing by 36.5 cm (**Figure 1**, Supplementary Table S1). There was some transgressive segregation for plant height, but most RILs had means within the means of the parents.

The resistant checks 93FHB37 and ND 2710 had the lowest FHB VRI, incidence, and severity in the field FHB nurseries (**Table 1**, Supplementary Table S1). AC Vista, Roblin, and CDC Teal were the most susceptible to FHB based upon the VRI,

incidence, and severity data, which was consistent with previous experience with these lines. The varieties 5602HR, AC Barrie, Neepawa, Snowbird, and AC Cora had intermediate FHB data. AC Morse, the sole durum wheat check, had intermediate to moderate susceptibility based upon VRI data, which is also typical. However, it should be noted that AC Morse would typically have very high Fusarium-damaged kernels (FDK) and deoxynivalenol (DON) accumulation relative to bread wheats with similar VRI scores. Cumulatively, these findings were all consistent with past experience with these wheat lines in FHB nurseries.

The wheat line 86ISMN 2137 was amongst the most susceptible to FHB based on VRI in these field nurseries (**Table 1**). Kenyon was more resistant to FHB than 86ISMN 2137 on average, and had a VRI score less than 86ISMN 2137 in four of the six field tests. Kenyon also had a higher VRI score than Neepawa in every field test, which was unexpected given its pedigree (Neepawa<sup>∗</sup> 5/Buck Manantial). The mean of Kenyon/86ISMN 2137 RIL population was similar to its parents, but there were individual RILs which were more resistant and susceptible to FHB. This result suggested transgressive segregation for FHB resistance in this population.

Correlation analysis revealed a consistent negative correlation between FHB VRI and plant height (r = −0.55) (Supplementary Table S2). Anthesis date and VRI were not highly correlated, and the correlation was not consistent (Supplementary Table S2). For instance, FHB VRI and anthesis date were negatively correlated in Winnipeg 2012, Ottawa 2013, and Winnipeg 2013, but positively correlated in Carman 2013 and completely uncorrelated in Carman 2012.

#### Linkage Map

The Kenyon/86ISMN 2137 linkage map was 2647 cM in length and consisted of 25 linkage groups and 3081 loci. Most of the wheat genome was covered by 22 of the linkage groups, with chromosome 1A consisting of two relatively large linkage groups of 82 and 29 cM (Supplementary Table S3). The three remaining linkage groups were small (7, 15, 13 cM) and were assigned to 1D (as linkage group 1D.1), 3D (as linkage group 3D.2), and 5D (as linkage group 5D.1) chromosomes based upon comparison to published maps (Cavanagh et al., 2013).

A problem was identified in the 5B and 7B linkage maps. Under the linkage mapping conditions described above in the Materials and Methods, markers on chromosomes 5B and 7B formed a single linkage group. A reciprocal translocation was suspected in one of the parents. This was investigated by examining chromosome pairing during metaphase I in pollen mother cells. Cytology revealed the presence of quadrivalents during this phase of meiosis (**Figure 2**), which was also consistent with a translocation in one of parents. Markers on chromosomes 5B and 7B were subsequently separated with the maximum recombination fraction of 0.02 and a minimum LOD score of 5. This strategy successfully separated the markers for the two chromosomes, except for a single linkage group that represented TABLE 1 | Least-squares means of checks and descriptive statistics of the Kenyon/86ISMN 2137 RIL population for FHB Visual Rating Index in field nurseries and pooled over environments.


<sup>a</sup>Car, Carman, MB; Ot, Ottawa, ON; PEI, Charlottetown, PEI; Wpg, Winnipeg, MB; 12, 2012; 13, 2013.

the translocation breakpoint. The 5B and 7B markers, that were successfully separated based upon the stringent linkage group criteria above, were then mapped as outlined in the Materials and Methods. This resulted in four linkage groups that represented chromosomes 5B and 7B (i.e., a linkage group on either side of the translocation breakpoint for each chromosome). The markers in the translocation breakpoint linkage group were sorted into their appropriate chromosomes based upon the chromosome that these markers were previously mapped in other populations (Somers et al., 2004; Cavanagh et al., 2013). The linkage maps for chromosomes 5B and 7B were then re-calculated and a

single linkage group was developed for each chromosome. The markers involved in the translocation breakpoint are indicated in Supplementary Table S3. These results were consistent with a reciprocal translocation in one of the parents of the mapping population, such that the 5B and 7B markers were genetically linked near the translocation breakpoint.

#### Anthesis Date

Three additive effect QTL for anthesis date were identified (**Table 2**). The QTL were located on chromosomes 2D, 4A, and 5B, and were named QAnth.crc-2D, QAnth.crc-4A, and QAnth.crc-5B. Kenyon alleles increased days to anthesis for QAnth.crc-2D and QAnth.crc-5B, and decreased days to anthesis for QAnth.crc-4A. QAnth.crc-2D and QAnth.crc-5B exceeded the LOD threshold more frequently than QAnth.crc-4A.

Analysis for epistatic QTL identified additional QTL for anthesis date. Three digenic epistatic interactions were identified (Supplementary Table S4). The most consistent identified epistatic interaction was between loci on chromosomes 5B and 5D. The r 2 value and estimated additive<sup>∗</sup> additive effect of the interaction was generally higher for IM than ICIM. Other epistatic interactions were identified between loci on chromosomes 3B and 3D, and 3D and 7D. None of the additive effect QTL for anthesis date were detected as an epistatic QTL, and vice-versa.

#### Plant Height

Four additive effect QTL for plant height were identified (**Table 2**). These QTL were located on chromosome 1B, 2D, 3D (linkage group 3D.1), and 7B, and were named QHt.crc-1B, QHt.crc-2D, QHt.crc-3D, and QHt.crc-7B. The Kenyon allele increased plant height for each of these QTL, which is consistent with Kenyon being 15 cm taller than 86ISMN 2137 in field tests. The QHt.crc-1B, QHt.crc-2D, and QHt.crc-3D were detected by IM (exceed LOD 3.22 in at least one environment), while all four height QTL were detected by ICIM. One epistatic QTL interaction was detected between loci on chromosomes 2B and 6B. This interaction was consistently identified by IM, but half of the time by ICIM.

The minor 7B height QTL was located within the translocation breakpoint. The peak of this QTL (38.0 cM) was located 1.9 cM from the most likely position of the translocation breakpoint (36.1 cM). Given this, it is strange that a height QTL was not detected on chromosome 5B at the translocation breakpoint (33.2 cM). Examination of the allele scores between markers at position 33.2 cM on chromosome 5B and position 38.0 cM on chromosome 7B revealed five allele score differences for these locations amongst the RILs. The relatively weak effect of this QTL and the number of allele score differences between these locations was apparently sufficient to prevent detection of the height QTL at the translocation breakpoint on chromosome 5B. Given these results, we recommend caution regarding the accuracy of the location of this QTL (i.e., the QTL could be located at the translocation breakpoint on chromosome 7B, or possibly 5B).

#### FHB Resistance

Nine additive effect QTL for FHB resistance were identified by QTL analysis using IM and ICIM with the additive effect module of QTL IciMapping (**Table 2**). These FHB resistance QTL were located on chromosomes 2D, 4A, 4D, 5B, and 7D, and were named QFhb.crc-2D.1 (chromosome 2D at 8.5 cM), QFhb.crc-2D.2 (chromosome 2D at 20.7 cM), QFhb.crc-2D.3 (chromosome 2D at 37.5 cM), QFhb.crc-2D.4 (chromosome 2D at 98.2 cM), QFhb.crc-4A.1 (chromosome 4A at 51.1 cM), QFhb.crc-4A.2 (chromosome 4A at 124.4 cM), QFhb.crc-4D, QFhb.crc-5B, and QFhb.crc-7D. Kenyon alleles decreased FHB symptoms for QFhb.crc-2D.1, QFhb.crc-2D.2, QFhb.crc-2D.3, QFhb.crc-2D.4, QFhb.crc-5B, and QFhb.crc-7D, and increased VRI for QFhb.crc-4A.1, QFhb.crc-4A.2, and QFhb.crc-4D. QFhb.crc-2D.3 was the most consistently detected. QFhb.crc-4D and QFhb.crc-5B were the least significant FHB resistance QTL. The other FHB resistance QTL were reasonably consistent and detected by both IM and ICIM.

Additional QTL for FHB resistance were identified by digenic epistasis QTL analysis. Four digenic epistatic interactions were identified between loci on chromosomes 1A and 4B, 1B (near QHt.crc-1B) and 7B (QHt.crc-7B), 2B and 6B, and 2B and 6D (Supplementary Table S4). All four interactions were detected in approximately the same number of environments. None of the FHB resistance QTL detected by digenic epistasis loci were identified by additive effect QTL analysis. Similar to anthesis date, the r 2 value and estimated additive<sup>∗</sup> additive effect of the interactions was generally higher for IM than ICIM.

## DISCUSSION

This study is the first study of native FHB resistance in western Canadian spring wheat. Nine FHB resistance QTL were detected in total. Kenyon contributed resistance at six of these QTL, which is consistent with Kenyon being more resistant than 86ISMN 2137 in these field FHB nurseries. QTL for FHB resistance were generally independent of QTL for anthesis date or plant height, except chromosome 2D. This is discussed in greater detail below. Kenyon contributed the resistance allele at QFhb.crc-7D, one of the most consistently detected QTL in this study. QFhb.crc-7D mapped to the same location as Qfhb.sdsu-7D (Eckard et al., 2015), where the resistant allele came from the wheat lines Wesley-Fhb1-BC56 and AL-107-6106. QFhb.crc-7D mapped to the same location as a minor FHB resistance QTL in the Wangshuibai/Alondra's population (Jia et al., 2005).

The anthesis date QTL QAnth.crc-2D is likely caused by photoperiod sensitivity gene Ppd-D1. QAnth.crc-2D mapped to ∼37.1 cM on the 2D linkage group in the Kenyon/86ISMN 2137 RIL population, which is between Xgwm261 and Xgwm484 (Xgwm261−19.7 cM–QAnth.crc-2D−10.3 cM–Xgwm484).

This is the expected location of Ppd-D1 based upon the Cappelle-Desprez (Mara 2D) RIL population (Xgwm261−22.3 cM–Ppd-D1−12.4 cM–Xgwm484) (Gasperini et al., 2012). Likewise, the plant height QTL QHt.crc-2D was detected in all environments in which plant height was measured, and in the pooled dataset. Rht8 was suspected to be responsible for QHt.crc-2D, which is known to map near Xgwm261 (Korzun et al., 1998). 86SIMN 2137 carries the 192 bp allele of Xgwm261, which is associated with the Rht8 reduced height allele (Korzun et al., 1998; Worland et al., 1998). Xgwm261 mapped to position 17.4 cM on chromosome 2D in the Kenyon/86ISMN 2137 population, which places QHt.crc-2D approximately 9.5 cM proximal of distal of Xgwm261. Rht8 maps 1.95 cM proximal of Xgwm261 based upon fine mapping (Gasperini et al., 2012). The location of QHt.crc-2D in this population is most likely due to the combined effect of Rht8 and Ppd-D1, since Ppd-D1 is known to have a pleiotropic effect on plant height by shortening the life cycle (Worland and Law, 1986).


TABLE 2 | Additive effect QTL detected for plant height, anthesis date, FHB visual rating index, incidence, and severity in the Kenyon/86ISMN 2137 RIL population.

(Continued)

#### TABLE 2 | Continued


(Continued)

#### TABLE 2 | Continued


a IM, interval mapping; ICIM, inclusive composite interval mapping.

<sup>b</sup>Chrom, chromosome.

<sup>c</sup>Pos, position on linkage group (cM).

<sup>d</sup>LOD, peak LOD score; LOD threshold (IM), 3.22, LOD threshold (ICIM), 3.22.

<sup>e</sup>PVE, phenotypic variation explained (r<sup>2</sup> ; %).

<sup>f</sup>Add, additive effect of allele substitution. The units are those of the respective trait. A positive sign indicated that the 'Kenyon' allele increased the respective quantitative trait, and vice-versa.

Three FHB resistance QTL (QFhb.crc-2D.1, QFhb.crc-2D.2, and QFhb.crc-2D.3) mapped to a relatively small region of chromosome arm 2DS. Kenyon contributed FHB resistance at all three loci, and carries a tall allele at Rht8 and Ppd-D1b (daylength sensitive allele). Given the close proximity of these FHB resistance QTL, it is difficult to conclusively determine whether they are truly distinct based on the present data. QFhb.crc-2D.2 mapped 1.3 cM distal of the expected location of Rht8 (position 19.4 cM) based on upon the position of Xgwm261. QFhb.crc-2D.3 mapped approximately 2.1 cM distal of QAnth.crc-2D (i.e., the location of Ppd-D1). Given these results, QFhb.crc-2D.2 and QFhb.crc-2D.3 are likely due to the pleiotropic effects of Rht8 and Ppd-D1. If true, QFhb.crc-2D.2 and QFhb.crc-2D.3 would be distinct from each other. QFhb.crc-2D.1 mapped about 8.9 cM distal of Xgwm261, or about 11 cM distal of Rht8. This suggests that QFhb.crc-2D.1 may be a distinct FHB resistance QTL from QFhb.crc-2D.2. FHB resistance QTL has been previously detected near Rht8 (Somers et al., 2003; Handa et al., 2008; Löffler et al., 2009). Further, genetic study is needed to clarify the number of QTL affecting FHB resistance on chromosome arm 2DS and to differentiate genetic linkage vs. pleiotropy between FHB resistance, plant height, and photoperiod response.

QFhb.crc-2D.4 mapped to the same region of chromosome 2D as the FHB resistance QTL present in Wuhan-1 (Somers et al., 2003). Kenyon carries the FHB resistance allele for this QTL. This result was unexpected since Kenyon has no common ancestry with Wuhan-1. It should be noted that QFhb.crc-2D.4 was not consistently identified in all field tests in this study. This contrasts with the Wuhan-1 2DL FHB resistance QTL, which was a strong QTL in past research (Somers et al., 2003; McCartney et al., 2007). Additional research is underway to study native FHB resistance in western Canadian germplasm, which will hopefully confirm the presence of QFhb.crc-2D.4 in other Canadian germplasm. It would also be valuable to know whether the Wuhan-1 2DL QTL has a stronger effect that the Kenyon allele at QFhb.crc-2D.4.

The relationship between the QTL identified in this study and previously published QTL was explored through comparative mapping. This relied upon simple sequence repeat (SSR) loci common between the maps and the SSR consensus map by Somers et al. (2004). 86ISMN 2137 contributed three FHB resistance QTL (QFhb.crc-4A.1, QFhb.crc-4A.2, and QFhb.crc-4D). QFhb.crc-4A.1 mapped to a similar location as an FHB resistance QTL derived from T. macha in a "Hobbit Sib" (T. macha 4A) single recombinant chromosome doubled haploid (DH) population (Steed et al., 2005). QFhb.crc-4A.2 mapped to a similar location as an FHB resistance QTL from Arina (Paillard et al., 2004). QFhb.crc-4D mapped near a FHB resistance QTL from DH181 (pedigree: Sumai 3/HY368) and CS-SM3-7ADS (Chinese Spring Sumai 3 chromosome 7A disomic substitution line) (Ma et al., 2006). Unfortunately the origin of 86ISMN 2137 is not known, but DNA marker data suggests that this line is not closely related to Canadian spring wheats. It is unlikely that these three QTL from 86ISMN 2137 are present in Canadian spring wheats. Kenyon contributed the FHB resistance QTL QFhb.crc-5B, which is approximately the same map location as a FHB resistance QTL detected in Patterson (Bourdoncle and Ohm, 2003). Patterson is a soft red winter wheat from Purdue University, USA.

The variety Kenyon (pedigree: Neepawa<sup>∗</sup> 5/Buck Manantial) is a backcross derived line of the variety Neepawa, in which the leaf rust resistance gene Lr16 was introgressed from Buck Manantial. Lr16 was previously mapped to the short arm of chromosome 2B (McCartney et al., 2005). Interestingly, Neepawa was more resistant to FHB than Kenyon in all six environments tested. This suggested that the higher VRI score of Kenyon relative to Neepawa could be due to the Lr16 introgression. However, no FHB resistance QTL was detected on chromosome 2B in this study, which indicates that the introgression carrying Lr16 does not have a major effect on FHB resistance. This is fortunate since Lr16 is a useful leaf rust resistance when pyramided with Lr34 (Hiebert et al., 2010). Presumably, other portions of the genome must differ between Neepawa and Kenyon that are responsible for the difference in FHB resistance. It should be noted that Lr16 has been confirmed to segregate in the Kenyon/86ISMN 2137 RIL population and was mapped to its expected location on chromosome arm 2BS (McCartney unpublished data).

Interestingly, the digenic epistatic interaction between loci on 1B and 7B for FHB resistance (Supplementary Table S4) corresponded to two additive effect QTL for plant height QHt.crc-1B and QHt.crc-7B. This result supports that this digenic epistasis interaction is valid and is not a statistical artifact. None of the other epistatic QTL were detected as additive effect QTL. This raises the question whether these epistatic QTL are real or statistical artifacts. Given the relatively small RIL population used in this study, it is quite possible that some of the detected epistatic interactions could be false. Additional research is needed to resolve this issue. None of the additive effect FHB resistance QTL were involved in epistatic interactions. This is likely good news for the deployment of these QTL in wheat breeding because this would suggest that each of these QTL appear to function independently of each other.

This study provides insight into the genetic basis of native FHB resistance in western Canada's CWRS marketing class, which is Canada's largest marketing class of wheat. A thorough knowledge of FHB resistance is needed to retain native FHB resistance from Canadian wheats and pyramid this with FHB resistance from Asian spring wheats, such as Sumai 3, Wangshuibai, and others. The FHB resistance QTL from Kenyon are likely to be valuable to wheat breeders in other growing regions, who would like to utilize FHB resistance QTL from wheats with excellent bread making properties.

## REFERENCES


#### AUTHOR CONTRIBUTIONS

CM planned and organized the study. AB, GF, RM, BM, and JG collected Fusarium head blight data. CH conducted the cytology experiments examining chromosome pairing during meiosis. CM and CP conducted DNA marker analyses. CM developed the linkage map and conducted QTL analysis. All authors contributed to and approved the final manuscript.

#### ACKNOWLEDGMENTS

The authors thank technical staff from the participating labs for their contributions to this research. This research was funded under the AgriInnovation Program Agri-Science Cluster entitled "National Wheat Improvement Program" funded by Western Grains Research Foundation and Agriculture and Agri-Food Canada, and as part of CTAG and CTAG2, Genome Prairie projects funded by Genome Canada, Saskatchewan Ministry of Agriculture, and Western Grains Research Foundation.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fmicb. 2016.01542


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2016 McCartney, Brûlé-Babel, Fedak, Martin, McCallum, Gilbert, Hiebert and Pozniak. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Identifying Rare FHB-Resistant Segregants in Intransigent Backcross and F<sup>2</sup> Winter Wheat Populations

Anthony J. Clark <sup>1</sup> , Daniela Sarti-Dvorjak <sup>2</sup> , Gina Brown-Guedira<sup>3</sup> , Yanhong Dong<sup>4</sup> , Byung-Kee Baik <sup>5</sup> and David A. Van Sanford<sup>1</sup> \*

*<sup>1</sup> Department of Plant and Soil Sciences, University of Kentucky, Lexington, KY, USA, <sup>2</sup> Syngenta Seeds, Des Moines, IA, USA, <sup>3</sup> Plant Science Research Unit, United States Department of Agriculture - Agricultural Research Service, Raleigh, NC, USA, <sup>4</sup> Department of Plant Pathology, University of Minnesota, Saint Paul, MN, USA, <sup>5</sup> Soft Wheat Quality Laboratory, United States Department of Agriculture - Agricultural Research Service, Wooster, OH, USA*

#### *Edited by:*

*Thomas Miedaner, University of Hohenheim, Germany*

#### *Reviewed by:*

*Hermann Buerstmayr, University of Natural Resources and Life Sciences, Vienna Mariana Ittu, National Agricultural Research Development Institute Fundulea, Romania*

> *\*Correspondence: David A. Van Sanford dvs@uky.edu*

#### *Specialty section:*

*This article was submitted to Fungi and Their Interactions, a section of the journal Frontiers in Microbiology*

*Received: 16 December 2015 Accepted: 22 February 2016 Published: 07 March 2016*

#### *Citation:*

*Clark AJ, Sarti-Dvorjak D, Brown-Guedira G, Dong Y, Baik B-K and Van Sanford DA (2016) Identifying Rare FHB-Resistant Segregants in Intransigent Backcross and F2 Winter Wheat Populations Front. Microbiol. 7:277. doi: 10.3389/fmicb.2016.00277* Fusarium head blight (FHB), caused mainly by *Fusarium graminearum* Schwabe [telomorph: *Gibberella zeae* Schwein.(Petch)] in the US, is one of the most destructive diseases of wheat (*Triticum aestivum* L. and *T. durum* L.). Infected grain is usually contaminated with deoxynivalenol (DON), a serious mycotoxin. The challenge in FHB resistance breeding is combining resistance with superior agronomic and quality characteristics. Exotic QTL are widely used to improve FHB resistance. Success depends on the genetic background into which the QTL are introgressed, whether through backcrossing or forward crossing; QTL expression is impossible to predict. In this study four high-yielding soft red winter wheat breeding lines with little or no scab resistance were each crossed to a donor parent (VA01W-476) with resistance alleles at two QTL: *Fhb1* (chromosome 3BS) and *QFhs.nau-2DL* (chromosome 2DL) to generate backcross and F<sup>2</sup> progeny. F<sup>2</sup> individuals were genotyped and assigned to 4 groups according to presence/ absence of resistance alleles at one or both QTL. The effectiveness of these QTL in reducing FHB rating, incidence, index, severity, *Fusarium*-damaged kernels (FDK) and DON, in F2-derived lines was assessed over 2 years. *Fhb1* showed an average reduction in DON of 17.5%, and conferred significant resistance in 3 of 4 populations. *QFhs.nau-2DL* reduced DON 6.7% on average and conferred significant resistance in 2 of 4 populations. The combination of *Fhb1* and *QFhs.nau-2DL* resistance reduced DON 25.5% across all populations. Double resistant lines had significantly reduced DON compared to double susceptible lines in 3 populations. Backcross derived progeny were planted in replicated yield trials (2011 and 2012) and in a scab nursery in 2012. Several top yielding lines performed well in the scab nursery, with acceptable DON concentrations, even though the average effect of either QTL in this population was not significant. Population selection is often viewed as an "all or nothing" process: if the average resistance level is insufficient, the population is discarded. These results indicate that it may be possible to find rare segregants which combine scab resistance, superior agronomic performance and acceptable quality even in populations in which the average effect of the QTL is muted or negligible.

Keywords: resistance breeding, Fusarium head blight, deoxynivalenol, soft red winter wheat, marker-assisted selection

## INTRODUCTION

Fusarium head blight (FHB), caused by several Fusarium species, is a destructive disease of wheat (Triticum aestivum L. and T. durum L.) and barley (Hordeum vulgare L.) worldwide (Bai and Shaner, 1994; Mesterhazy, 1995). In North America, Fusarium graminearum was primarily responsible for scab epidemics since 1993 in the spring, soft red winter and hard red winter wheat regions of the US. It is estimated, based on published literature (Windels, 2000; Johnson et al., 2003; Nganje et al., 2004) and anecdotal reports from millers, pathologists and breeders, that direct losses to scab in the US from 1993 to 2014 total \$ 4.8 B. As Nganje et al. (2004) note, the number would be increased by the inclusion of secondary (indirect) losses. Many wheat breeding programs focus on, along with high yield, the development of FHB resistance in commercial cultivars (McMullen et al., 2012). The incorporation of genetic resistance reduces the need for fungicide applications and, consequently, reduces production costs and environmental pollution while increasing food safety.

A major concern associated with FHB in wheat and barley is the production of mycotoxins, especially deoxynivalenol (DON) and its derivatives. High levels of DON in grains have negative effects on animal production, causing vomiting in non-ruminant animals leading to serious feeding problems and economic losses (McMullen et al., 1997; Pestka, 2007). There is support for the premise of a close linear relationship between FHB resistance and DON concentration in the infected grain (Mesterházy et al., 1999). Regulation of DON accumulation is challenging and depends on the host and fungal genotypes as well as environmental conditions (Mesterházy et al., 1999). Deoxynivalenol concentration is also the most important FHB trait because of the discount imposed on contaminated grain at the elevator. FDK is another indicator of kernel damage and relates to test weight reduction and partly accounts for yield reduction. FHB index is the best overall indicator of field symptoms and is a product of severity and incidence. Ratings theoretically should be very similar to index when given by very experienced workers and are attractive because they can be recorded quickly. Near infrared reflectance (NIR) measurements can be an efficient way to measure grain symptoms.

One of the major discussions within the US soft red winter (SRW) wheat community concerns the use of exotic (i.e., from outside the pool of elite US lines) resistance QTL (e.g., Fhb1) vs. so-called "native" resistance found in adapted SRW wheat cultivars and breeding lines. The existence or extent of linkage drag associated with exotic FHB resistance QTL on agronomic traits has not been widely documented but each new variety, with improved yield over its predecessors, represents the pinnacle of development and testing of thousands of new segregants. If linkage drag were present, exotic QTL-containing lines would likely be out yielded. The genetics of native resistance are largely unknown, and likely involve numerous genes of small effect. While the exotic QTL are easy to track with DNA markers, the level of FHB resistance conferred by a single QTL is often insufficient to reduce losses in grain yield and quality. Further, QTL by environment interactions and the effects of different genetic backgrounds on gene expression complicate the story (Van Sanford et al., 2001; Balut et al., 2013).

Previous research in our lab has shown that two FHB resistance QTL, Fhb1 and QFhs.nau-2DL vary in their expression levels according to genetic background (Verges et al., 2006; Agostinelli et al., 2012; Balut et al., 2013). This finding is in accord with other research on Fhb1 (Pumphrey et al., 2007; Buerstmayr et al., 2009) and QFhs.nau-2DL (Jiang et al., 2007a,b) and the 2DL QTL studied by Kang et al. (2011). In a single population study, Agostinelli et al. (2012) reported that QFhs.nau-2DL was more effective than Fhb1, reducing FDK and DON by 40 and 55%, respectively, in comparison to Fhb1-associated reductions of 32% for FDK and 25% for DON. Balut et al. (2013) observed that Fhb1 reduced FDK by 32%, and DON concentration by 20% across five populations and QFhs.nau-2DL reduced FDK and DON by as much as 29 and 24%, respectively in some backgrounds. The lines chosen for the latter study were all homozygous for Fhb1 but while the populations segregated for QFhs.nau-2DL many of the lines tested derived from heterozygotes. The current study is the first to look at the effect of QFhs.nau-2DL in several populations using lines homozygous at that locus.

The constraints associated with the exotic QTL, e.g., dependence on genetic background, plus the need to combine exotic QTL-based resistance with native resistance, are daunting. These issues do not encourage a resource-intensive backcrossing effort, lest the recurrent parent be in the wrong background or in a background devoid of native resistance. This study was undertaken to evaluate an approach to population development that might yield useful segregants independent of genetic background.

The objectives of the study were to: (1) determine whether useful segregants for FHB resistance and agronomic fitness could be recovered from populations in which overall resistance was not high enough to recommend backcrossing, and (2) whether F<sup>2</sup> or BC1F<sup>1</sup> populations were superior sources of such segregants.

### MATERIALS AND METHODS

#### Initial Population Development

The study began as a marker-assisted backcrossing project in which FHB resistance alleles from two QTL, Fhb1 and QFhs.nau-2DL, from VA01W-476 were introgressed into four elite SRW breeding lines, KY97C-0321-05-2, KY97C-0519-04-05, KY97C-0540-01-03, and KY97C-0508-01-01A. These recurrent parents were high yielding, FHB-susceptible wheat lines. The FHBresistance donor, VA01W-476, was a double haploid line derived from "Roane" and "W14" (Perugini, 2007; Agostinelli et al., 2012; Balut et al., 2013). VA01W-476 in addition to providing resistant Fhb1 and QFhs.nau-2DL alleles, also likely has additional native resistance genes from its parents (Jiang et al., 2007a; Agostinelli et al., 2012). All crosses used in this study were made by A.J. Clark.

The intent was to use genotyped F2—derived lines to validate the resistance QTL while at the same time examine and compare the agronomic and quality characteristics of a separate set of BC1F<sup>1</sup> and F<sup>2</sup> derived lines from the same four

crosses that had undergone agronomic selection during their development.

## F2-derived Populations used in QTL Validation Study in Scab Nursery

Single crosses were made in the greenhouse in spring 2008. KY97C-0321-05-2/VA01W-476 was developed as population 2, KY97C-0519-04-05/VA01W-476 as population 3, KY97C-0540-01-03/VA01W-476 as population 4 and KY97C-0508-01- 01A/VA01W-476 as population 6. Three additional backcrossed populations, 1, 5, and 7 were made but F<sup>2</sup> lines were not derived from them. Populations 2, 3, 4, and 6 were advanced to the F<sup>2</sup> generation in a greenhouse at Spindletop Research Farm, near Lexington, KY, in the fall of 2008 and F2:<sup>3</sup> and F2:<sup>4</sup> lines tested in the scab nursery.

## Development of Populations Used in Yield Trials and for Quality Testing

Agronomic BC1F1<sup>−</sup> Derived Population Development After crosses between recurrent and donor parents were made, plants positive for the presence of resistance alleles at both loci were selected and crossed back to the recurrent parent. BC1F<sup>1</sup> seedlings were grown in the greenhouse and BC1F<sup>1</sup> heads harvested and planted in BC1F1:<sup>2</sup> head-rows in 2009 at Spindletop Research Farm near Lexington, KY. For populations 2, 3, 4 and 6, 65, 23, 42 and 32 selections were made and thoserows harvested and threshed and BC1F1:<sup>3</sup> and BC1F1:<sup>4</sup> tested in 2011 and 2012 respectively. These BC1F1:<sup>4</sup> populations were also characterized in the scab nursery in 2012.

#### Agronomic F2-derived Population Development

Tillers of the F<sup>1</sup> plants produced in 2008 were allowed to selffertilize, to produce F<sup>2</sup> seed. Heads were threshed in bulk and planted in F<sup>2</sup> plots in Lexington, KY, 2009. From each population approximately 60–100 heads were selected and planted the following year in 1.2 m long F2:3headrows, spaced 30 cm apart, in Princeton, KY. Headrow selections were made for agronomic potential and threshed independently. For populations 2, 3, 4, and 6, 21, 19, 44 and 36 F2:<sup>4</sup> lines were selected respectively.

## Genotyping

DNA was isolated according to Pallota et al. (2003). Markers used were UMN10 (Liu et al., 2008) and gwm533 (Röder et al., 1998) for Fhb1; and cfd233 (Grain genes 2.0 at http://wheat.pw.usda.gov/GG3/ verified Nov 19th 2015) and gwm539 (Röder et al., 1998) for QFhs.nau-2DL. The genotyping process was divided between two laboratories: the University of Kentucky Wheat Breeding Laboratory and the USDA/ARS Regional Small Grains Genotyping Lab (RSGGL) (http://www. ars.usda.gov/Main/docs.htm?docid=19522) at Raleigh, NC. At the University of Kentucky, PCR products were separated using an ABI 3730 DNA Analyzer (Applied Biosystems) and sized using GeneMapper v4.0. Following backcrossing seedlings heterozygous for both Fhb1 and QFhs.nau-2DL resistant alleles were selected for growth in the greenhouse. For QTL validation, F<sup>2</sup> seedlings homozygous for markers at both QTL were selected in all combinations, RR, RS, SR, SS, at Fhb1 and QFhs.nau-2DL respectively.

#### Scab Nursery

For the 2011 season, 78, 79, 131, and 91 lines from Populations 2, 3, 4, and 6, respectively, along with parents were planted in replicated F2:<sup>3</sup> headrows in a misted, inoculated scab nursery at the UK Spindletop Research Farm (38◦ 7 ′ 37.81′′N, 84◦ 29′ 44.85′′W; Maury silt loam [fine, mixed, semiactive, mesic Typic Paleudalfs]) near Lexington, KY on 11 October, 2010. Each row was evaluated for FHB traits, harvested by row, and screened for grain disease levels. The following year, F2:<sup>4</sup> headrows were planted in the scab nursery on 17 October, 2011. The experiment planted each year was a randomized complete block design (RCBD) with 2 replications. In addition in 2012 the BC1F1:<sup>4</sup> lines used for agronomic testing were also seeded in 2 rep RCBD experiments in the Lexington scab nursery. Unfortunately space constraints prevented inclusion of the F2-derived lines used for agronomic testing. Rows were 1 m long, spaced 30 cm apart. Rows were misted with an overhead mist irrigation system on an automatic timer, from May to June, for periods of 5 min, every quarter hour from 8:00 pm to 8:45 pm, 11:00 pm to 11:45 pm, 2:00 am to 2:45 am, 5:00 am to 5:30 am, and for one time at 8:30 am so the integrity of the system could be monitored.

The scab nursery was inoculated with Fusarium graminearum—infected corn (Zea mays L.) (Verges et al., 2006). Inoculum source and preparation were exactly as described in Balut et al. (2013). The scabby corn was distributed between rows at a rate of 11.86 g/m−<sup>2</sup> , approximately 3 week prior to heading, on 14 April and 31 March of 2011 and 2012, respectively. Liquid nitrogen fertilizer (28% UAN) was applied in the spring at a rate of 105 kg N/ha in split applications. Harmony Extra herbicide was applied on 20 April 2011 and 20 March 2012.

### Phenotyping

Heading dates were recorded for each headrow in the scab nursery, when 50% of the spikes in the row had emerged from the flag leaf sheath. Plant height was measured at the soft dough stage. Effectiveness of QTL in reducing FHB was assessed through several resistance traits. These traits were measured approximately 21 days after anthesis and consisted of: rating, severity, incidence, and FHB index. Ratings were visual estimates on a 0–9 scale, where 0 = 0–10% and 9 = 91–100% of diseased spikelets within a row. Incidence was the count of spikes showing any disease among 20 randomly selected spikes in a headrow, expressed as a percentage. Severity, the number of visually infected spikelets divided by the total number, expressed as a percentage was also counted, in 10 randomly selected blighted heads per row. FHB index is the product of severity and incidence divided by 100.

Each headrow was hand harvested with a sickle and threshed in a small thresher with low air flow to avoid loss of tombstones (infected kernels, blighted, and lighter than healthy grains). Fusarium damaged kernels (FDK) percentages were estimated from carefully cleaned samples run through an air separation machine (Agostinelli, 2009; Agostinelli et al., 2012). FDK was expressed as the weight of scabby kernels divided by total weight. DON concentrations were estimated in these same samples by the University of Minnesota DON Testing Laboratory using gas chromatography with mass spectrometry (GC-MS) (Mirocha et al., 1998; Fuentes et al., 2005).

#### Near-Infrared Reflectance Spectroscopy

Near-Infrared Reflectance Spectroscopy (NIR) predictions of FDK and DON were also generated (Delwiche and Hareland, 2004) using the Perten Instruments DA7200. We ran samples of cleaned grain (15–20 g) through the instrument to predict FDK and DON (FDKNIR, DONNIR, respectively) and compared predictions with actual values.

#### Data Analysis of QTL Validation Study

Analysis of variance (ANOVA) was performed using the General Linear Model procedure (PROC GLM; SAS, 2011). The model used was:

$$\begin{array}{c} \text{Y}\_{\text{ij}} = \mu + \text{ENV}\_{\text{i}} + \text{R(ENV)}\_{\text{ij}} + \text{QTL} + \text{G}\_{\text{k}} \text{(QTL)}\\ + \text{ENV}\_{\text{i}} \text{"QTL} + \text{E}\_{\text{ij}} \end{array}$$

Where:

Yij = observation in the kth genotype in the jth rep in the ith environment,

µ = overall mean,

Gk(QTL) = effect of the kth genotype within QTL,

QTL = effect of the QTL,

R(ENV)ij = effect of jth rep within ith environment,

ENVi <sup>∗</sup> QTL = effect of the interaction of the ith environment with the QTL,

Eij = residual error.

Fisher's Least Significant Difference (LSD) was used to corroborate significant differences among QTL combination classes.

Broad sense heritability of FHB and agronomic traits estimates were based on entry means using the following model:

$$\text{Yij} = \mu + \text{ENVi} + \text{R(ENV)ij} + \text{Gk} + \text{Gk}^\* \text{ENVi} + \text{Eij}$$

Where:

Yij = the observation in the kth genotype in the jth rep in the ith environment,

µ = the overall mean,

G<sup>j</sup> = the effect of the kth genotype,

R(ENV)ij = the effect of jth rep within ith environment,

Gk <sup>∗</sup> ENV<sup>i</sup> <sup>=</sup> the effect of the interaction of the kth genotype with the ith environment,

Eij = the residual error.

Data were analyzed using PROC GLM (SAS, 2011). Genotypic and phenotypic variances were estimated from the expected mean squares (EMS) and heritability estimates were computed as:

$$\mathbf{h}^2 = \mathbf{v}\_{\mathbb{g}} / \mathbf{V}\_{\mathbb{P}}$$

Where:

h <sup>2</sup> = broad sense heritability,

V<sup>g</sup> = genotypic variance,

V<sup>p</sup> = phenotypic variance.

Confidence intervals (90%) for the heritability estimates were calculated after Knapp et al. (1985).

PROC CORR (SAS, 2011) was used to analyze the relationship among traits on an entry mean basis. Multiple comparisons of least squares means presented in **Tables 3**–**5** were handled by performing a t-test on every pair of means.

#### Yield Measurement

BC1F1:<sup>3</sup> plots were planted on 13th October 2010 in a randomized complete block design, with two replications, at Spindletop Farm, near Lexington, KY. In 2011, BC1F1:<sup>4</sup> plots along with F2:<sup>4</sup> plots were planted in a randomized complete block design, with two replications at two locations, Lexington, KY (1 Nov. 2011) and Princeton, KY (10 Oct. 2011). Each population's parents were planted along with four commercial varieties as yield checks. The experimental material was grown in conventional yield plots 6 rows wide and 3 m long, with a row spacing of 17.8 cm. All plots received 105 kg ha−<sup>1</sup> of actual N applied in the spring; recommended wheat production practices for Kentucky were followed (Lee et al., 2009). Plant height, yield and test weight were measured in each plot.

#### Milling and Baking Quality

In 2012, a 100-g sample from each replication was analyzed for milling and baking quality at the USDA-ARS Soft Wheat Quality Laboratory, Wooster, OH. Grain was tempered to 15% moisture before milling and milled using a Brabender Quadrumat Senior laboratory mill (South Hackensack, NJ). Flour yield was determined as the percent total flour weight (break flour + middlings) over tempered grain weight. Softness equivalent was the percentage weight break flour over the total flour (break flour + middlings). Water SRC, sucrose SRC, sodium carbonate SRC, and lactic acid SRC were determined using approved AACC International method 56-11.02 (American Association of Cereal Chemists, 2010) and were used to calculate the gluten performance index (GPI), defined as GPI = lactic acid SRC/(sodium carbonate SRC + sucrose SRC), as described by Kweon et al. (2011).

#### RESULTS

#### Weather Conditions and disease Levels

Weather conditions during 2011 were favorable for scab development, while unusually warm temperatures in 2012 from March through May accelerated wheat growth and reproductive development and resulted in heading dates 3–4 weeks earlier than normal. A severe April 2012 freeze followed by below normal rainfall led to drought-like conditions which minimized disease pressure; natural scab levels were much lower in Kentucky than in 2011 (Bruening et al., 2012). A uniform disease epidemic was achieved in the irrigated nursery but the level was also lower in 2012 (**Table 1**). Weather and scab intensity differences in any two growing seasons are not uncommon in Kentucky; 2011 and 2012 allowed us to observe QTL effects in very different environments.

In both years of the study, susceptible parents showed higher disease levels than the resistant parent in all populations. F<sup>2</sup>


#### TABLE 1 | Range of FHB traits in four F<sup>2</sup> derived wheat populations and their parents, Lexington, KY, 2011 and 2012.

*Recurrent parent in population 4, KY97C-054, not planted 2011.*

*DON, dexynivalenol; FDK, Fusarium-damaged kernels; Sev, severity; Inc, incidence; Ind, index; Rating, Fusarium head blight rating.*

derived progeny with DON levels lower than the resistant parent were observed in three populations in 2011 and four populations in 2012 (**Table 1**).

#### Heritability Estimates

Broad sense heritabilities and their corresponding 90% confidence intervals were estimated on an entry mean basis for each population separately and also for all populations combined to provide an overall heritability for each disease trait (**Table 2**). DON h<sup>2</sup> estimates were relatively high and consistent and ranged from 0.54 to 0.75. FDK h<sup>2</sup> ranged from 0.16 tp 0.48. Incidence h<sup>2</sup> was moderate ranging from 0.34 to 0.57 while heritability of severity was ≥0.30. FHB index h<sup>2</sup> estimates were moderate, with an average of 0.40. Heritabilities of FHB ratings varied widely across populations from 0.19 to 0.65 (**Table 2**).

## QTL effects on DON, FDK, and FHB Disease Traits

#### Fhb1

Under heavy scab pressure in 2011 and reduced pressure in 2012, the effectiveness of Fhb1 in reducing DON varied among populations. In populations 3, 4, and 6 significant (P ≤ 0.05) DON reductions were observed in both 2011 and 2012 ranging from 11 to 32% (**Table 3**). Significant (P ≤ 0.05) ENV∗QTL interactions were seen for DON accumulation for populations 4 and 6. In both populations the percentage reduction shown by Fhb1 resistant lines was higher in 2012


TABLE 2 | Heritabilities and their 90% confidence interval (in parentheses) of four F<sup>2</sup> derived wheat populations based on 2 year entry means, Lexington, KY 2011-2012.



\**FDK, Fusarium damaged kernels; Sev, severity; Inc, incidence; Ind, FHB index.*

when overall DON levels were lower. In population 3, resistant lines were lower in 2012 than in 2011 but there was no significant ENV∗QTL effect. In population 2, resistance alleles at this QTL had no effect on DON level (**Table 3**). Natural field infections typically result in disease pressure similar to that in the inoculated scab nursery in 2012; therefore one can expect Fhb1 to be effective in reducing typical DON levels in farmers' fields. FDK was significantly reduced in both years in populations 4 and 6 (**Table 3**). Compared to the widespread and consistent-between-years differences between resistant and susceptible Fhb1 lines in grain symptoms, differences in detailed head symptoms were less frequent or consistent. For example, a significant (P < 0.05) reduction in incidence and index was seen in population 4 in 2012 but not 2011 (**Table 3**). Population 2 rating, based on the overall row saw a significant increase in Fhb1 resistant lines in 2011 only. The lack of consistency between the patterns of incidence/severity/index and FDK/DON show that the former should not substitute for grain measurements. In 2012 Fhb1 resistant lines were significantly (P < 0.05) earlier however this was not seen in 2011.

#### QFhs.nau-2DL

QFhs.nau-2DL was similar to Fhb1 in that DON levels were significantly reduced in populations 3 (21%), 4 (19%), and 6 (18%) in 2012; though we only observed a significant reduction in population 6 in 2011 (**Table 4**). Unlike Fhb1 resistance, the largest DON reduction was observed in Population 3 in 2012, where DON levels were 21% lower in QFhs.nau-2DL resistant lines (**Table 4**). As we observed with Fhb1, the R allele at


TABLE 4 | Means for FHB traits evaluated according to the presence of resistance (R) or susceptible (S) alleles at *QFhs.nau-2DL* for four F<sup>2</sup> derived wheat populations, Lexington, KY, 2011 and 2012.

QFhs.nau-2DL did not reduce DON in population 2 (**Table 4**). In the other populations, QFhs.nau-2DL effects ranged from 14 to 17% varying across years (**Table 4**). As we noted with Fhb1, effects of QFhs.nau-2DL on incidence, severity and index were scarce and inconsistent between years. In contrast to Fhb1 there were no consistent effects on rating with only population 2 showing a significant (P < 0.05) benefit in 2011 and the opposite in 2012 (**Table 4**). Average heading dates were significantly (P < 0.05) increased in resistant lines of populations 3 (0.9 days), 4 (0.4 days), and 6 (0.6 days) in 2012 when heading was unusually early (**Table 4**). None of these differences was seen in the more typical 2011.

#### Fhb1 plus QFhs.nau-2DL

For populations 3, 4, and 6, the double QTL combination RR (Fhb1, QFhs.nau-2DL respectively) was always lowest or in the lowest significant (P ≤ 0.05) grouping in both years for DON (**Table 5**). Conversely the double susceptible (SS) lines always had the highest mean DON, or were in the statistical grouping with the highest DON for both years (**Table 5**). In 2012 in all populations, RR and SS were significantly (P ≤ 0.05) different (**Table 5**). In 2011 populations 4 and 6 RR and SS lines mean DON were also significantly (P ≤ 0.05) different. For population 3, lines resistant at Fhb1 only (RS lines) were significantly (P ≤ 0.05) different from SS lines (**Table 5**). No allele combination had any significant effect on DON in population 2 in either year (**Table 5**). For populations 3, 4, and 6 the reduction in DON in RR vs. SS lines was 10.2, 26.8, and 33.9% respectively in 2011 (**Table 5**). The same trend was seen in 2012 with 29.3, 43.9, and 46.7% reductions respectively (**Table 5**). It seemed that the resistances were also additive in 2012, and the greater individual effects of Fhb1 and especially QFhs.nau-2DL in the milder epidemic in that year combined to produce a greater effect in RR lines compared to 2011. The remarkable consistency in effect on DON was also seen for FDK. In populations 3, 4, and 6 in 2011 the reductions were 19.3, 30.5, and 33.9% respectively (**Table 5**). In 2012 they were 10.7, 29.6, and 30.4% respectively (**Table 5**). In neither year was a significant (P < 0.05) difference in FDK seen between any of the allele combinations for population 2 (**Table 5**). For incidence, severity and index, more significant differences were seen among the four allele combinations than were seen when looking at the QTL individually (**Tables 3**–**5**). Significant (P < 0.05) differences in incidence were seen in both 2011 and 2012 in populations 3 and 4. In 2011 significant (P < 0.05) differences in severity were seen among populations 2, 3, and 6 (**Table 5**), while in 2012 significant (P < 0.05) differences were again seen in populations 2 and 3. In population 3 in 2012 the lowest severity seen in SR lines, significantly (P < 0.05) different only from SS lines. The detailed pattern for index was similarly complex, varying by population and year (**Table 5**). Ratings were more consistent between the years in populations 4 and 6. For both populations, RS was usually next


TABLE 5 | Means for FHB traits evaluated according to the presence of resistance (R) or susceptible (S) alleles at two QTL (*Fhb1* and *QFhs.nau-2DL* respectively), for four F<sup>2</sup> derived wheat populations, Lexington, KY, 2011 and 2012.

lowest to RR and often not significantly (P < 0.05) different. Ratings over years were less consistent in Populations 2 and 6 (**Table 5**).

#### Resistance Alleles in Highly Resistant Lines

We focused on the 20 most resistant lines in each population in each year with rankings based on DON concentration in 2011 and 2012. We especially looked at those lines that were the lowest 20 in both years

Population 2, in which the average impact of either QTL was not significant (P < 0.05), presented a neutral picture consistent with the lack of impact of either QTL. Of the 20 lowest DON lines in each year, 12 were common to both years; 3 of these had RR alleles, 3 had RS, 3 had SR and 3 had the SS genotype (**Table 6**). Line 274508 from this population, with RS alleles, had the lowest mean DON for both years, 4.5 ppm. Populations 3 and 4 had 5 and 7 RR lines common to both years respectively, the highest numbers of RR lines common to both years (**Table 6**). TABLE 6 | Number of the 20 lowest DON F<sup>2</sup> - derived lines from four wheat populations with resistance alleles at either, both, or neither of two QTL: *Fhb1* and *2DL.*


*Lines were evaluated in inoculated, irrigated scab nurseries 2011 and 2012, Lexington, KY.*

The high number of RR lines was interesting given the lack of significant (P < 0.05) differences between RR and RS lines in both populations in both years (**Table 5**). Population 6 had just 2 RR lines consistently in the lowest 20 for DON but 6 RS lines, no SR lines and 1 SS line (**Table 6**). The predominance of RS genotypes among consistently highly ranked lines was unexpected. Population 6 was the only population that in 2012 had showed a significant (P < 0.05) improvement of RR over RS lines (**Table 5**). Surprisingly for a population that showed a significant (P < 0.05) average benefit from both QTL, but not for a population with so mostly RS genotypes in the lowest 20, the lowest line for mean DON both years was an RS line.

#### QTL x Environment Interaction in Highly Resistant Lines

In populations 2, 3, and 4, where no QTL x environment interactions were seen, the majority of the two-resistance-allele lines that were among the lowest 20 DON lines in each year were common to both years (**Table 6**). In population 6, where QTL x environment interaction was seen, 5 and 6 RR lines were among the lowest 20 in 2011 and 2012 respectively, and of these, 2 were common to both years (**Table 6**). The consistency of low DON SS lines also varied by population. For example, in 2011 and 2012, 7 and 5 of the lowest DON lines in population 2 were SS and of these 3 were common to both years. On the other hand, in population 3, 5, and 6 of the top 20 lines in 2011 and 2012 respectively had neither resistance allele and 2 of these were common to both years.

#### Low DON Segregants

Segregants with low DON were observed in all populations, but were much more frequent in Population 3 where 19 lines were found to have DON levels lower than the resistant parent VA01W-476, in the 2012 scab nursery. Across all populations there were 31 lines with numerically lower DON than VA01W-476: 4, 4, 19, and 4 in populations 2, 3, 4, and 6 respectively (data not shown). Seventeen lines had DON levels that were more than two standard errors lower than the 1.6 ppm measured in VA01W-476.

#### NIR Predictions

Correlations between FDKNIR and actual FDK in this study were 0.49 for all populations combined (**Table 7**) with highest value in Population 6 (r = 0.73; data not shown). Correlations between DONNIR and FDK were 0.53 (**Table 7**). Correlations between DONNIR and DON ranged from 0.55 for population 2 to 0.82 for population 6 (data not shown), with a value of 0.63 among all populations combined (**Table 7**). Interestingly, FDKNIR was better correlated to DON (0.64) than FDK itself (0.47; **Table 7**). Visual ratings also proved to be predictive of DON overall, with a correlation of 0.67 (**Table 7**). The correlations indicate NIR of grain and even visual ratings would have been good predictors of FDK and DON values of the soft red winter wheat populations in this study. We wanted to see how many of the lowest DON lines identified by GC-MS would also have been selected using NIR or ratings. For this comparison of 2 year means, in every population DONNIR identified the highest, or joint highest, number of the 20 lowest DON lines compared to FDK, FDKNIR, rating, and incidence, severity or index (**Table 8**). The usefulness of the DON-substitute measurements varied between populations (**Table 8**).

#### Agronomic Performance of F<sup>2</sup> and BC1F1-derived Lines

Yield trial data from BC1F<sup>1</sup> derived line tests at Lexington and Princeton in 2011 and 2012 is presented in **Table 9**. Yield trial data for BC1F<sup>1</sup> and F<sup>2</sup> derived lines in 2012 is shown in **Tables 10**, **11**. In population 2 in 2012, 23% of the 86 lines tested were not significantly different from the high yielding commercial cultivars that were used as checks (data not shown) a number similar to that seen for BC1F<sup>1</sup> derived lines in both years (**Table 9**). The highest yielding lines were all BC1F<sup>1</sup> derived (**Table 10**), while F<sup>2</sup> derived lines did not fare as well. The backcross lines were also screened in the scab nursery in 2012. Several of the lines at the very top of the yield trial also performed well in the scab nursery, with DON concentrations of 8 ppm for example, in comparison to the susceptible recurrent parent in which the DON level was 15.9 ppm (**Table 11**). These results indicate that it may be possible to combine scab resistance and superior agronomic performance even in populations in which the apparent effect of the QTL is muted or negligible.

Two year yield averages for Populations 2, 3, 4, and 6 were close to 4237 kg ha−<sup>1</sup> (Sarti-Dvorjak, 2014). This value is numerically but not significantly lower than the yield average of the check cultivars (4371 kg ha−<sup>1</sup> ). When measured in yield plots, F<sup>2</sup> derived lines in Populations 2, 4, and 6 were 10, 9, and 13% taller than their respective susceptible parent's average height (Sarti-Dvorjak, 2014). Both F<sup>2</sup> and BC1F<sup>1</sup> populations were effective sources of useful segregates, though it varied



*Data comprised 2 year entry means in four F*2*-derived wheat populations, Lexington, KY 2011–2012.*

TABLE 8 | Number of the 20 lowest DON F2—derived lines from four wheat populations identified using FDK, DONNIR, FDKNIR, scab rating, incidence, severity, and index\*.


\**FDK, Fusarium damaged kernels; DONNIR, DON predicted by NIR; FDKNIR, FDK predicted by NIR.*

by population. In population 2, only one F2-derived line was competitive with the yields of commercial check cultivars while maintaining an acceptably low DON concentration, in contrast to 11 BC1F1-derived lines that met these criteria (**Table 11**). However, in population 6 the top yielding lines in the same LSD group as the checks were BC1F<sup>1</sup> and F<sup>2</sup> derived in roughly equal numbers (9 and 11 respectively, data not shown). In population 3 all of the top yielding experimental lines were BC—derived, while in population 4, there were 7 F<sup>2</sup> derived and 4 BC1F1—derived lines that out yielded the commercial checks.

#### Milling and Baking Quality

In a previous study of five SRW populations, Balut et al. (2013) found little impact of resistance alleles at either Fhb1 or QFhs.nau-2DL on milling and baking quality traits. In this study, in all populations, the BC1F<sup>1</sup> derived lines have better quality scored than the F<sup>2</sup> derived lines as one would expect with 75 vs. 50% of recurrent parent in the pedigree (**Table 12**). The challenge in this instance is finding lines that combine superior agronomic performance with acceptable FHB resistance and acceptable milling and baking quality. In population 2, there were eight such lines selected for comparison because they had an average FDK of 10.0 vs. 9.6% for the highly resistant donor parent. The yield and test weight averages of these lines were equivalent to those of the checks (4375 vs. 4419 kg ha−<sup>1</sup> , and 73.5 vs. 72.8 kg hl−<sup>1</sup> ) and the quality, based on the traits listed in **Table 11**, was actually slightly higher than that of the check cultivars (data not shown).

## DISCUSSION

The high DON h<sup>2</sup> estimates seen in these populations were similar to results reported by Balut et al. (2013) but lower than Agostinelli et al. (2012). FDK h<sup>2</sup> were lower than estimates reported by Agostinelli et al. (2012) and Balut et al. (2013), which exceeded 0.60.

Resistance alleles and the interaction among FHB resistance QTL have distinct behavior in different genetic backgrounds in wheat. The best-validated gene for FHB resistance, Fhb1 on chromosome 3BS, showed an average reduction of 17.5% in DON, effecting significant improvement of FHB resistance in 3 of 4 populations. This is the first study to validate QFhs.nau-2DL resistance using homozygous lines. In this study the effect was modest, significant in 2 of 4 populations and reducing DON 6.7% overall. QFhs.nau-2DL seemed to combine well with Fhb1 however. The combined resistances reduced DON 25.5% across all populations.

Previous investigators have questioned whether exotic QTL will provide sufficient resistance to progeny in the absence of native resistance (Balut et al., 2013). The recurrent parent in Population 2, KY97C-0321-05-2, had been increased for possible release for its yield potential, but FHB susceptibility derailed this effort. Thus, it seemed the perfect candidate to ascertain whether backcrossed resistance QTL would lead to adequate levels of FHB resistance. While the overall impact of either QTL was negligible in this genetic background (**Tables 3**–**5**) breeders are always looking for rare segregants that perform the best in each population. In assessing each of the four populations it made sense to look beyond the average effect, at individual F<sup>2</sup> lines. At the outset of this study, it was our assumption that it would be difficult, if not impossible, to identify FHBresistant segregants from populations in which the expression of the introgressed QTL was muted. In contrast to our expectations, resistant segregants were seen for DON accumulation, even in population 2 as well as the other populations. This underscores our assertion that usable segregants can be recovered in many


TABLE 9 | Two year mean yield and test weight of BC1F<sup>1</sup> derived wheat lines, total number of lines and percentage of lines with yields not significantly different from checks, Lexington and Princeton, KY, 2011–2012.

TABLE 10 | Mean yield and test weight of BC1F<sup>1</sup> and F<sup>2</sup> derived wheat lines, total number of lines and percentage of lines with yields not significantly different from checks, Lexington and Princeton, KY, 2012.


recalcitrant populations. Furthermore, we found that there were individual lines in all populations in which agronomic fitness was combined with acceptable levels of FHB resistance. In population 2, for example, such segregants were identified with and without the R alleles at one or both QTL, even though the only apparent native resistance came from the donor parent. In sum, our results give cause for optimism concerning the utility of even the most intractable populations in FHB resistance breeding.

The key message that emerges from this complex picture is that in every population there were lines with low DON levels both years, under very heavy (2011) and rather light (2012) scab pressure. These results indicate that if a population with a scab resistant and scab resistance QTL-containing parent has the potential to deliver agronomically promising breeding lines it is critical to evaluate all of the progeny from crosses that may appear to be generally intransigent for scab resistance.

Previous studies of Fhb1 and QFhs.nau-2DL in our lab have revealed significant QTL by year interaction in several populations (Agostinelli et al., 2012; Balut et al., 2013). The nature of such interaction is important: significant changes in rank (crossovers) across environments would have a major impact on marker-assisted selection and reduce the utility of the QTL derived resistance. However, significant QTL x environment interaction does not preclude the existence of lines with consistent QTL effects in very different environments. As noted earlier, the donor parent, VA01W-476, probably contributed native resistance as well as Fhb1 and QFhs.nau-2DL and it is possible this was the source of resistance that consistently lowered DON in low DON SS lines.

TABLE 11 | Yield, test weight and DON concentrations of top yielding backcross-derived and F2 derived lines in population 2 along with several commercial checks and donor and recurrent parents.


*n.d., Not determined.*

Scab resistant progeny were found only through arduous phenotypic screening and further investigation and optimization of methods of efficiently collecting phenotypic scab data are needed. Our results also assert the importance of grain-based measures of FHB resistance. It has long been our assumption that the time and labor-consuming activities of assessing incidence and severity are of limited value if they are not effective in predicting FDK and DON, because FDK and DON are the best indicators of loss to growers, processors and end users. In this study, overall the best predictor of DON was, surprisingly, FHB rating (0.67; **Table 7**). This 0–9 "eyeball" measure is taken quickly and integrates incidence and severity, though it is essential that the ratings are completed by only one individual for a given test. A close second was FDKNIR (0.64; **Table 7**), which is a compelling reason to continue to look at NIR as a way of estimating scab damage. Initially, the purpose of this comparison was to estimate the correlation between scab damage and DON levels with NIR predictions, to determine whether NIR might eventually replace expensive, time consuming and destructive techniques like DON gas chromatography mass spectrometry (GC-MS). Alternatively, NIR might be useful to prescreen samples before using GC-MS. Previous studies in our lab have shown strong positive correlations between FDK and DON values measured with traditional methods and NIR estimations. Balut et al. (2013) reported FDK—FDKNIR correlations of 0.70 and 0.73 and DON—DONNIR of 0.56 and 0.63 in 2010 and 2011, respectively. Tibola et al. (2010) tested the ability of NIR to predict DON levels in both 125 grams whole grain and milled samples in the Southern Brazil and reported 0.89 and 0.91 coefficient of determination (R 2 ). While DONNIR was not entirely successful as a replacement for GC-MS DON measurement in the study described here, it did perform better than FDK in every population. In breeding populations that would be screened by FDK but not GC-MS, substitution of DONNIR should continue to be investigated.

Finally, we were interested in the value of BC1F<sup>1</sup> vs. F2 derived populations in generating usable progeny. In this study both kinds of populations were effective in this regard, though it was clear when we looked at milling and baking quality traits


TABLE 12 | Mean values of milling and baking quality traits from wet lab analyses and predicted by NIR from BC1F<sup>1</sup> and F2—derived wheat populations, parents, and commercial check cultivars grown at Lexington, KY 2012.

that BC1F1—derived populations had the edge. This fact is a good reason for breeders to consider creating BC1F<sup>1</sup> populations for routine breeding and extraction of inbred lines in addition to pursuing the more customary route through selection in F<sup>2</sup> populations.

## AUTHOR CONTRIBUTIONS

DV overall responsibility for design of the study and resultant manuscript, contributed text, data analysis, table preparation, editing, and revising. AC primary responsibility for writing text and preparing tables, data analysis, revising, editing. DS carried out the research for her PhD; contributed data and text, revised as needed. GB: carried out genotyping; contributed text describing same, revising as needed. BB: conducted milling baking analyses; contributed text describing same, revising as needed. YD conducted mycotoxin analyses;

### REFERENCES


contributed text describing same, revising as needed. All authors participated in discussions that related to the design of the study and all contributed significant intellectual content to the manuscript.

#### ACKNOWLEDGMENTS

DS conducted this research in partial fulfillment of the requirements to obtain a PhD degree at the University of Kentucky. This material is based on work supported by the U.S. Department of Agriculture, under Agreement No. 59-0206- 9-054. This is a cooperative project with the U.S. Wheat and Barley Scab Initiative. Any opinions, findings, conclusions, or recommendations expressed in this publication are those of the author(s) and do not necessarily reflect the view of the U.S. Department of Agriculture. We thank John Connelley and Sandy Swanson for their technical assistance.


lines developed from breeding populations. Crop Sci. 47, 200–206. doi: 10.2135/cropsci2006.03.0206


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2016 Clark, Sarti-Dvorjak, Brown-Guedira, Dong, Baik and Van Sanford. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Cell Wall Biomolecular Composition Plays a Potential Role in the Host Type II Resistance to Fusarium Head Blight in Wheat

Rachid Lahlali <sup>1</sup> \*, Saroj Kumar <sup>1</sup> , Lipu Wang<sup>2</sup> , Li Forseille<sup>2</sup> , Nicole Sylvain<sup>3</sup> , Malgorzata Korbas <sup>1</sup> , David Muir <sup>1</sup> , George Swerhone<sup>4</sup> , John R. Lawrence<sup>4</sup> , Pierre R. Fobert 2, 5, Gary Peng<sup>6</sup> and Chithra Karunakaran<sup>1</sup> \*

*<sup>1</sup> Canadian Light Source, Saskatoon, SK, Canada, <sup>2</sup> National Research Council Canada, Saskatoon, SK, Canada, <sup>3</sup> Department of Surgery, College of Medicine, University of Saskatchewan, Saskatoon, SK, Canada, <sup>4</sup> Environment Canada, Saskatoon, SK, Canada, <sup>5</sup> National Research Council Canada, Ottawa, ON, Canada, <sup>6</sup> Agriculture and Agri-Food Canada, Saskatoon Research Centre, Saskatoon, SK, Canada*

#### *Edited by:*

*Agnieszka Waskiewicz, ´ Poznan University of Life Sciences, ´ Poland*

#### *Reviewed by:*

*Radames J. B. Cordero, Johns Hopkins Bloomberg School of Public Health, USA Daniela Bellincampi, Sapienza University of Rome, Italy*

#### *\*Correspondence:*

*Rachid Lahlali rachid.lahlali@lightsource.ca; Chithra Karunakaran chithra.karunakaran@lightsource.ca*

#### *Specialty section:*

*This article was submitted to Fungi and Their Interactions, a section of the journal Frontiers in Microbiology*

*Received: 05 January 2016 Accepted: 27 May 2016 Published: 27 June 2016*

#### *Citation:*

*Lahlali R, Kumar S, Wang L, Forseille L, Sylvain N, Korbas M, Muir D, Swerhone G, Lawrence JR, Fobert PR, Peng G and Karunakaran C (2016) Cell Wall Biomolecular Composition Plays a Potential Role in the Host Type II Resistance to Fusarium Head Blight in Wheat. Front. Microbiol. 7:910. doi: 10.3389/fmicb.2016.00910* Fusarium head blight (FHB) is a serious disease of wheat worldwide. Cultivar resistance to FHB depends on biochemical factors that confine the pathogen spread in spikes. Breeding for cultivar resistance is considered the most practical way to manage this disease. In this study, different spectroscopy and microscopy techniques were applied to discriminate resistance in wheat genotypes against FHB. Synchrotron-based spectroscopy and imaging techniques, including focal plane array infrared and X-ray fluorescence (XRF) spectroscopy were used to understand changes in biochemical and nutrients in rachis following FHB infection. Sumai3 and Muchmore were used to represent resistant and susceptible cultivars to FHB, respectively, in this study. The histological comparison of rachis showed substantial differences in the cell wall thickness between the cultivars after infection. Synchrotron-based infrared imaging emphasized substantial difference in biochemical composition of rachis samples between the two cultivars prior to visible symptoms; in the resistant Sumai3, infrared bands representing lignin and hemicellulose were stronger and more persistent compared to the susceptible cultivar. These bands may be the candidates of biochemical markers for FHB resistance. Focal plane array infrared imaging (FPA) spectra from the rachis epidermis and vascular bundles revealed a new band (1710 cm−<sup>1</sup> ) related to the oxidative stress on the susceptible cultivar only. XRF spectroscopy data revealed differences in nutrients composition between cultivars, and between controls and inoculated samples, with substantial increases observed for Ca, K, Mn, Fe, Zn, and Si in the resistant cultivar. These nutrients are related to cell wall stability, metabolic process, and plant defense mechanisms such as lignification pathway and callose deposition. The combination of cell wall composition and lignification plays a role in the mechanism of type II host resistance to FHB. Biochemical profiling using the synchrotron-based spectroscopy holds potential for screening wheat genotypes for FHB resistance.

Keywords: Fusarium head blight, wheat, type II resistance, Fourier transform infrared spectroscopy, X-ray fluorescence spectroscopy, cell wall, synchrotron

## INTRODUCTION

Wheat (Triticum aestivum L.) is the most extensively grown food crop worldwide (Curtis et al., 2002; Mcmillan et al., 2014) and one of the most important crops in western Canada (Curtis et al., 2002). Fusarium head blight (FHB), caused by the fungus Fusarium graminearum [teleomorph: Gibberella Zeae (Schwein) Petch] is a devastating disease of wheat due to its negative impact on yield (Walter et al., 2010) and grain quality. The mycotoxin accumulation, such as deoxynivalenol (DON) in cereal grains, results in grain quality issues in food and feed, consequently exacerbating economic losses (Goswami and Kistler, 2004; Osborne and Stein, 2007).

It is generally difficult to control FHB in wheat with any single management tool and an integrated approach using multiple control options is recommended (Krupinsky et al., 2002; Osborne and Stein, 2007). The options used most frequently include genetic resistance and fungicide application. Tillage, crop rotation, and staggered planting of small grain crops may be used to reduce fungal survival on residues (Stack, 2003). Using genetic resistance is the most desirable option; it is practical and can also reduce the need for fungicide application, thus reducing input costs and environmental impact (Von Der Ohe et al., 2010). Current breeding strategies against FHB focus on the combination of desirable agronomic traits and type I and/or type II resistance mechanisms (Bai and Shaner, 2004), which refer to responses against initial infection and spread of the pathogen within the host, respectively (Schroeder and Christensen, 1963). Almost all reports on FHB resistance have been type II, and complete resistance cultivars have not been developed in wheat. The majority of studies have concluded that FHB resistance is quantitative and its inheritance involves several loci on different chromosomes (Buerstmayr et al., 2008; Steiner et al., 2009). Some of these quantitative trait loci (QTL) have been associated with certain transcriptomes and proteomes which may be used as gene markers for plant defense responses against FHB (Bai and Shaner, 2004). However, using these QTL markers for FHB-resistance breeding in wheat can still be difficult due to potentially low yield and quality drag linked to these QTLs (Buerstmayr et al., 2009). The interaction between genotype and environment can further complicate the phenotypic selection, making identification of FHB resistance time consuming and unreliable. Additional selection criteria, complementing genetic markers and phenotypic selection, may improve the efficiency and reliability of FHB resistance screening.

In wheat, symptoms of FHB begin with small water soaked brown spots at either the middle or base of the glume (Goswami and Kistler, 2004; Osborne and Stein, 2007). The discoloration or pre-mature bleaching spreads outwards from the point of infection, and white or magenta mycelium may appear around the edges of infected glumes. Eventually, the majority of inflorescence can become blighted and awns, if present, may become deformed, twisted, and curved downward. Infected kernels or Fusarium damaged kernels (FDKs) are gray and white in color, often with a magenta hue, appear shrunken, and have a floury interior. Many refer these FDKs as "tombstone kernels." Often the physiological conditions of host plant influenced by nutrition, hydration and plant age can play an important role in FHB development (Osborne and Stein, 2007); wheat crop is highly susceptible between the antithesis and soft dough stage of seed development (Mcmullen et al., 1997; Shaner, 2003). Few studies have characterized biochemical changes in wheat heads, especially in relation to different levels of resistance, during this most susceptible growth stage. Fourier transform mid infrared (FTIR) spectroscopy is a powerful tool for examining biochemical changes during FHB development (Lahlali et al., 2015), which may provide additional criteria for resistance identification. FTIR spectroscopy is a label-free and non-invasive technique that exerts an enormous attraction in biology and medicine, since it offers a rapid way to sample biomolecular contents (Kacurakova et al., 2000; Santos et al., 2010; Largo-Gosens et al., 2014). Its potential for detection and identification of fungal pathogens in plants promises to be of a great value because of the sensitivity, rapidity, low cost, and simplicity (Kummerle et al., 1998; Martin et al., 2005; Erukhimovitch et al., 2010; Peiris et al., 2012; Lahlali et al., 2015). FTIR spectral properties of infected wheat rachis may help reveal compositional differences due to infection between resistant and susceptible wheat cultivars before visible symptoms. The biochemical information may also be related to the type I and II resistance, making the data more versatile. We hypothesize that FHB resistance mechanisms involve type II resistance, which are related to biochemical composition in the cell wall including lignin, pectin, hemicellulose, and nutrients such as calcium, potassium, iron, and zinc in the internodes of wheat rachis.

The objectives of this study were to: (i) identify any structural and anatomical differences in the cell wall of rachis during infection based on microscopy, (ii) assess FTIR spectral absorption between control and inoculated rachis samples of both resistant and susceptible wheat cultivars, (iii) localize the cell compositions on the cross section of rachis of both cultivars with and without fungal infection, and (iv) characterize the nutrients composition in rachis of two contrasting cultivars without and with fungal infection using X-ray fluorescence (XRF) spectroscopy. The goal was to achieve a better understanding of biochemical and nutritional mechanisms related to FHB resistance. Some of the unique spectroscopic bands can be used as biomolecular markers by breeders to identify FHB resistance during routine screening of wheat genotypes.

#### MATERIALS AND METHODS

#### Fungal Culture and Inoculum Preparation

The wild-type F. graminearum (Fg) isolate DAOM 180379 (Canadian collection of fungal cultures, Ottawa, ON) was used in this study. For the production of macroconidia, a plug of actively growing Fg was placed in the center of a petri dish containing Soft Nutrient Agar (SNA). For spore production, cultures were grown in CMC (Carboxymethyl cellulose) medium and incubated at 28◦C for 2 days. Conidial suspension was harvested in sterile water, filtered through cheesecloth. A working concentration of 5 × 10<sup>4</sup> macroconidia/mL was used for inoculation.

#### Plant Material and Inoculation Procedures

All experiments were conducted in the environment-controlled growth chamber due to restrictions on using a transformed fungal pathogen in the field. Seeds of resistant "Sumai3" (SU3) and susceptible "Muchmore" (MM) wheat cultivars were sown in peat pots (12.7 cm, diameter) and maintained in a growth chamber at a 20◦C: 16◦C cycle (day: night), with a 16-h photoperiod until flowering. Pots were watered by hand at the base of the plants. At mid-anthesis, single floret inoculation with the Fg strain was carried out by pipetting 10 µl of the macroconidia suspension (5 × 10<sup>4</sup> /mL) between palea and lemma. Inoculated plants were placed in a dew chamber for 2 days and then moved back to the growth chamber for the rest of the experiment (Lahlali et al., 2015). Plants used as controls were inoculated with a drop of sterile distilled water.

## Visual and Microscopic Observations of Infection

This experiment was designed to compare the anatomy of the rachis nodes, cell arrangement, and cell wall thickness of both resistant and susceptible cultivars as well as the colonization pattern of F. graminearum. Cross sections of fresh and frozen rachis nodes were cut using a microtome and mounted on microscope slides for light microscopy. Intact cross sections were chosen to visualize structural and anatomical aspects of the control and inoculated rachis samples using a fluorescence microscope system (Leica Microsystems AF 6000, Leica Canada Ltd. Scarborough, ON). A Zeiss LSM710 confocal microscope (Carl Zeiss Canada Ltd. North York, ON) was used to assess changes in cell wall thickness and epidermis cells at different depths. The tissue samples were optically sectioned at different depth intervals and a Z-stack of images were generated for each fluorescence emission wavelength range. Confocal images were acquired at excitation wavelengths 488, 544, and 633 nm and emissions at 500–550, 573–613, and 650–1000 nm, respectively, for green, red, and blue colors. Images were processed using the Imaris software (Bitplane USA, South Windsor, CT), and histogram stretching and gamma adjustments were used to optimize the visual quality of images.

### Bulk FTIR Spectroscopy to Characterize the Infection and Resistance

In order to determine the differences and changes in the biochemical composition of control and infected rachis of both resistant and susceptible cultivars, FTIR spectra of bulk samples were collected at the mid infrared beamline (Canadian Light Source Inc.), using a globar (silicon carbide) as the infrared source. An IFS 66V/S spectrophotometer (Bruker Optics, Ettlingen, Germany) was used with a deuterated triglycine sulfate (DTGS) detector.

Rachis samples from control and inoculated spikes were prepared by the method described before (Naumann et al., 1991). Samples were freeze dried and then ground to fine powder. About 1–2 mg of the powder were homogenized with about 92 mg of dry potassium bromide (KBr) using pestle and mortar, and the mixture was compressed into a pellet. Transmission infrared spectrum was obtained from replicated pellet samples. Each IR spectrum was recorded in the mid infrared range of 4000–800 cm−<sup>1</sup> wavenumbers at a spectral resolution of 2 cm−<sup>1</sup> . The spectrum of each sample was an average of 64 scans and pure KBr spectra (average of 128 scans) was recorded to normalize all sample spectra. The normalized spectra were then baseline corrected using the rubber band correction (64 points) and vector normalized using the OPUS software (version 7.0, Bruker Optics Inc., Billerica, MA). The FTIR peaks cited in **Table 1** were determined using the Quick Peaks routine in OriginPro (version 9.1, OriginLab Corporation, MA) with the settings of local maximum at 0% threshold height, no baseline, and area at Y = 0 (Lahlali et al., 2015). The determination of components such as proteins, lignin, cellulose, hemicellulose, and pectin were made by integrating the area under specific bands. The area was determined using the OPUS integration method C, in which the area of interest was determined after considering two baseline points on the left and the other two on the right side of the peak/band. Statistical analysis was performed on integrated areas and ratios of lignin to other bands using ANOVA procedure (SAS Institute, Cary, NC). When the infection effect was revealed to be significant, the LSD test was employed for mean separation at P ≤ 0.05.

#### Synchrotron-Based FTIR Spectroscopy and Focal Plane Array Imaging

Previous results based on bulk FTIR spectra indicated substantial biochemical differences in the control rachis of resistant and susceptible cultivars (Lahlali et al., 2015). This experiment was designed to better characterize the infection and resistance based on assessing biochemical differences at the cellular level of epidermis using the cross sections of rachis. A synchrotron based FTIR spectroscopy (sFTIR) with a Hyperion 3000 microscope (Bruker Optics Inc., Milton, ON, Canada), equipped with a single element (Mercury Cadmium Telluride, MCT) detector and a 15X objective was used. The spatial resolution of the images was 15µm × 15 µm. The spectroscopic imaging data

TABLE 1 | Assignment of bands in the bulk FTIR spectra of rachis of the resistant (Sumai3) and susceptible (Muchmore) wheat cultivars inoculated with *Fusarium graminearum*a.


*<sup>a</sup>Dokken et al., 2005; Martin et al., 2005; Mann et al., 2009; Largo-Gosens et al., 2014; Lahlali et al., 2015.*

were acquired in transmission mode from samples (10 µm thick) mounted on 25 mm diameter and 1 mm thick CaF<sup>2</sup> slides. Frozen rachis sections were cut with a microtome-cryostat using a diamond knife. The sections deposited on CaF<sup>2</sup> slides were air dried at room temperature before sFTIR examination. Each sample spectrum was an average of 128 scans and a background of 128 scans was recorded for normalizing all sample spectra. Rachis samples used in this experiment were from 4 days postinoculation with 10 spectra collected for each sample. Data were analyzed using the principal component analysis as described by Jiang et al. (2015).

In order to assess changes in molecular composition in the large area of rachis cross sections including epidermis, xylem and phloem, the Hyperion imaging system was used (Heraud et al., 2007) with the globar source. The microscope was equipped with a 64 × 64 pixels Focal Plane Array (FPA) detector and a 15X objective. The spatial resolution was 2.7µm × 2.7 µm per pixel. One FPA grid measurement would result in 4096 spectra (scan area of 184µm × 184 µm), with each spectrum being an average of 256 scans. Principal component analysis (PCA) was carried out using the Matlab-based program Kinetics (version R2008, Mathworks Inc.), with respective loading graphs generated for epidermis and vascular bundles (Lahlali et al., 2014). Rachis samples were evaluated at 10 days post-inoculation and two replicates were used for each measurement.

## Bulk XRF Spectroscopy for Nutrients Composition in Rachis

Until now, there is no report on the involvement of nutrients in the type II resistance to FHB in wheat. The hard X-ray fluorescence data were collected at the Industry Development Education Applications Students (IDEAS) beamline of the Canadian Light Source. Dried rachis samples were mounted inside a vacuum sample chamber equipped with a linear paddle drive. The in-vacuum setup (at about 6 × 10−<sup>1</sup> Torr pressure) was used to detect low energy fluorescence photons such as phosphorus and sulfur, and to suppress the argon signal from the air that overlaps with potassium peak. Both sides of a rachis segment with 2 nodes long (∼10 mm) were exposed to 13 Kev X-rays, and the Acquaman software was used to collect data, with a dwell time of 180 s. The XRF data was normalized initially to the standard ring current of 250 mA, and the XRF spectrum was plotted using the OriginPro software (OriginLab Corporation Inc., USA). Each side of rachis sample was scanned three times at three different regions (six spectra/sample). Background scan of the vacuum chamber with no sample was also recorded. The data were collected from samples collected at 4 and 10 days postinoculation with two replicates per treatment and a silicon drift detector (Ketek AXAS-M M5T1T0-H0-ML5BEV) was used.

#### RESULTS

#### Visual and Microscopic Comparison of Infected Wheat Rachis

Considerable visual differences were observed between infected rachis of resistant and susceptible cultivars at 10 (**Figure S1**) and 15 days (**Figure 1**) post-inoculation; those of MM were completely infected whereas those of SU3 were only slightly discolored with browning on the edge (**Figure 1, Figure S1**). The cross sections of infected nodes under confocal microscopy (**Figure 2**) showed that the cell wall surrounding meta xylem (mx), protoxylem (px), and phloem (ph) degraded more prominently in MM than in SU3. The cell integrity of SU3 also remained almost intact (**Figure 2D**). The cell wall was thicker in MM compared to SU3 (**Figures 2B,D** and **Figure S2**). No fungal penetration into phloem, xylem, pith or any other cells was observed in infected wheat cultivars, SU3, and MM (**Figure 2C** and **Figure S3**).

### Bulk FTIR Spectroscopy for Potential Resistance Related Biochemical Marker Identification in Rachis

Bands associated with different chemical groups in the fingerprint region (1800–800 cm−<sup>1</sup> ), including pectin (C=O at 1740 cm−<sup>1</sup> ) and hemicellulose (1248 cm−<sup>1</sup> ) were identified and assigned in **Table 1**. The band at 1655 (Amide I) correspond to C=O and N-H vibrations and at 1546 cm−<sup>1</sup> (Amide II) to N-H and C-N vibration. The C=C stretching of the aromatic ring of lignin was at 1610-1590 cm−<sup>1</sup> and 1515-1505 cm−<sup>1</sup> . The cellulose bands were at 1372, 1161, and 1060 cm−<sup>1</sup> , while the band at 995 cm−<sup>1</sup> was associated with C-C ring vibration.

Based on the absorption in the bulk FTIR spectra, the infection by F. graminearum caused biochemical changes in the rachis of MM than in those of SU3 (**Table 2**) relative to the respective

FIGURE 1 | Symptoms of Fusarium head blight on the rachis of the susceptible and resistant wheat cultivars, Muchmore (A) and Sumai3 (B) at 15 days post-inoculation.


TABLE 2 | Integrated absorption bands in the bulk FTIR spectra of rachis of the susceptible (MM) and resistant (SU) wheat cultivars with and without *Fusarium gramineraum* (FHB) inoculation at 4 and 10 days post-inoculation (*n* = 3).

*C, control not inoculated with FHB; F, Inoculated with FHB; 1740 (1760–1720 cm*−*<sup>1</sup> ); 1654 (1710–1620 cm*−*<sup>1</sup> ); aromatic ring of lignin [1605 (1615–1590 cm*−*<sup>1</sup> ), 1460 (1480–1455 cm*−*<sup>1</sup> ), and 1425 (1445–1410 cm*−*<sup>1</sup> )]; 1248 (1261–1200 cm*−*<sup>1</sup> ); and 1056 (1090-1022 cm*−*<sup>1</sup> ). Bold indicates most persistent bands in the resistant cultivar after inoculation with FHB. Means in the same column followed by the same letter are not significantly different according to the LSD test P* ≤ *0.05.*

*Fusarium graminearum*. The images were created using the emission wavelength of 650–1000 nm (far red range) and the excitation at 633 nm. mx: metaxylem, px: protoxylem, and ph: phloem. Red arrows show changes in cell wall thickness following infection. Scale bar = 50 µm.

controls. In the resistant SU3, the content of the integrated area of the absorption bands arising from C=C stretching of the aromatic ring vibration of lignin (1615–1590, 1460, and 1425 cm−<sup>1</sup> ), and hemicellulose (1248 cm−<sup>1</sup> ), were persistent after infection. The changes resulted from infection were observed at wavenumbers 1740, 1654, and 1054 cm−<sup>1</sup> , which may be from pectin, proteins, and cellulose, respectively. In the susceptible cultivar MM, however, a decrease in the integrated area of lignin bands, hemicellulose, and cellulose were observed after infection while no change was seen for pectin at 4 days post-inoculation. Meanwhile, an increase was detected for proteins at both 4 and 10 post-inoculations and this may arise from the fungus. The ratio of associated integrated area of lignin (1515 cm−<sup>1</sup> ) to pectin (1740 cm−<sup>1</sup> ), proteins (amide I), lignin, hemicellulose, and cellulose between inoculated and control rachis were relatively consistent for SU3, but not so much with MM (**Table 3**). Between inoculated SU3 and MM rachis, the differences in these ratios were more pronounced at 4 than at 10 DPI. The use of the absorption ratios help minimize the variability of data caused by changes in sample thickness in the pellets.

## Synchrotron-Based FTIR Spectroscopy and Focal Plane Array Imaging for Resistance Related Biochemical Marker Identification

The synchrotron-based infrared data from the epidermis of control rachis showed that most of the variations between the cultivars was explained by PC1 and PC2 (86%, **Figure 3A**). The prominent peaks that contributed to the variations were those encoded for lignin (1606 and 1510 cm−<sup>1</sup> ), pectin (1752 cm−<sup>1</sup> ), hemicellulose (1250, and 1238 cm−<sup>1</sup> ), and cellulose (1394 and 1035 cm−<sup>1</sup> ; **Figure 3B**).

The focal plane array imaging was used to compare epidermis between SU3 and MM (**Figure 4**), and the infrared images of cross sections from control and inoculated samples confirmed the results of bulk FTIR and sFTIR described above. An increase in lignin and proteins content was observed in epidermal and vascular bundle cells following infection with FHB, while pectin and cellulose decreased. The second derivative spectra in the fingerprint region (1800–800 cm−<sup>1</sup> ) showed differences in the epidermal cell wall composition (**Figure 5A**) and vascular bundles (**Figure 5B**) between inoculated cultivars. The impact TABLE 3 | Ratios between the integrated band of lignin (at 1515 cm−1) to pectin (1760–1720 cm−1), proteins (1710–1620 cm−1), lignin (1615–1590 cm−1, 1480–1455 cm−1, and 1445–1410 cm−1), hemicellulose (1261–1200 cm−1), and cellulose (1090–1022 cm−1) in bulk FTIR spectra of the rachis of Muchmore and Sumai3 with and without *Fusarium graminearum* infection.


*Data were collected at 4 and 10 days post-inoculation (n* = *3). I1* = *1515/(1760–1720); I2* = *1515/(1710–1620); I3* = *1515/(1615–1590); I4* = *1515/(1480–1455); I5* = *1515/(1445–1410); I6* = *1515/(1261–1200); and I7* = *1515/(1090–1022). Means in the same row followed by the same letter are not significantly different according to the LSD test P* ≤ *0.05.*

of infection was more significant on the epidermis of MM than that in SU3, where most of the changes were observed at 1735 cm−<sup>1</sup> (pectin), 1657 cm−<sup>1</sup> (proteins), 1461 cm−<sup>1</sup> (lignin), and 1370 and 1160 cm−<sup>1</sup> (cellulose). However, little change was recorded in vascular bundles following infection with FHB. In the epidermis of MM the band of carbonyl ester (1735 cm−<sup>1</sup> ), which is mainly associated to pectin, was converted to the lipid aldehyde (1709 cm−<sup>1</sup> ) due to the oxidative stress referred as lipid peroxidation induced potentially by plant defense mechanisms in response to FHB (Shoaib et al., 2013). The score plot for this dataset as a function of two principal components, PC1 (accounting for 78% of variation) vs. PC2 (13%) is shown in **Figure 6A**. Two distinct clusters are observed; on the positive side of PC1 is susceptible cultivar MM, while on the negative side of PC1 and positive side along PC2 is the resistant cultivar SU3. The loadings plot for PC1 indicates the key differences in wavenumbers that are responsible for grouping the samples along PC1, more specifically the chemical feature that distinguishes the epidermal layer of SU3 rachis from that of MM (**Figure 6A**). The variables (wavenumbers) with high peak intensities (positive or negative) contributed the most to the separation between the cultivars are associated with pectin (1735 cm−<sup>1</sup> ), lignin (1515, and 1478 cm−<sup>1</sup> ), cellulose (1498 and 1168 cm−<sup>1</sup> ), and hemicellulose (1230 cm−<sup>1</sup> ; **Figure 7A**). In contrast, datasets from vascular bundles did not show substantial differences between the cultivars (**Figure 6B**); PC1and PC2 together explained 80% of observed variance between infected and non-infected vascular bundles and most of the differences were shown at the wavenumbers associated with pectin, lignin, cellulose and hemicellulose (**Figure 7B**). The predominance of the bands in aromatic ring of lignin implies cell wall lignification relating to the type II resistance to FHB.

#### Bulk XRF Spectroscopy for Nutrients Composition in Rachis

The nutrients composition in rachis of resistant and susceptible wheat cultivars with and without F. graminearum was compared

at 4 and 10 DPI (**Figure 8**) and the average X-ray spectra showed an entire elemental map, with the content of calcium (Ca), iron (Fe), potassium (K), manganese (Mn), and zinc (Zn)

being detected readily. In both cultivars, K was present in the highest amount in rachis of both control and inoculated samples, followed by Fe, Zn, Ca, and Mn. At 4 DPI, the rachis of the susceptible cultivar, MM showed lower amounts of Ca, Fe, K, Mn, and Zn, relative to those of SU3 (**Figure 8A**). Fe concentration decreased while Zn concentration increased in SU3 due to the fungus infection. Both Fe and Zn increased in MM due to the fungus infection. Fe, Zn, and Mn however were present in substantially higher amounts in the resistant cultivar SU3 at 10 DPI than in MM (**Figure 8B**). Ca was higher in the control SU3 but its concentration changed little with inoculation, while the amount increased substantially in MM at 4 DPI. Small differences were observed for other elements, including phosphorous (P), sulfur (S), chlorine (Cl), and copper (Cu) between the cultivars regardless of inoculation. Interestingly, the silicon (Si) was present only in the resistant cultivar SU3 at 10 DPI and increased after infection with FHB.

## DISCUSSION

Of the five types of resistance to FHB reported in wheat, only three have been extensively studied (Schroeder and Christensen, 1963; Miller et al., 1985), including the type I (to initial infection of spikelets), type II (to spread of infection in spike), and type III (to DON). The modes of action, however, are not wellunderstood for any of the resistance mechanisms. Because cereal breeders are most interested in the type II resistance against FHB, our study focused on this particular aspect of mechanism using a range of synchrotron-based technologies, including bulk FTIR spectroscopy, sFTIR imaging, focal plane array FTIR imaging, and X-ray spectroscopy to compare the structural, biochemical, and nutritional changes in the rachis internodes. The information will help to identify key factors associated with the resistance and establish additional criteria for efficient screening of wheat germplasms against FHB.

## Visual and Microscopic Assessment of Rachis Infected by *F. graminearum*

The infection of rachis leads to destruction of cell wall and starch granules, affecting endosperm storage proteins and consequently resulting in a poor grain quality (Snijders, 2004). Microscopic observations in the current study showed cell wall thickening and cell deterioration surrounding vascular bundles in the inoculated rachis of the susceptible cultivar MM, whereas these changes were not observed for the resistance cultivar SU3. This incompatible interaction in SU3 could be due to the recognition of fungal infection and the defense mechanism can provide a barrier to the pathogen invasion. Several modes of action are possible for the incompatibility. The reinforcement of cell wall can be activated rapidly in response to pathogen penetration and may involve lignituber, callose, silicon, and lignin deposition between the cell wall and membrane directly below the point of penetration (Jacobs et al., 2003; Luna et al., 2011; Bellincampi et al., 2014). In a recent study, a particular composition of lignin-structural component was suggested to play a role in the cell wall reinforcement for SU3 (Lionetti et al., 2015), and the microscopic observations in the current study provide cell and cell wall structure evidence supporting

the reinforcement theory. Lignification is a common mechanism for disease resistance in plants (Vance et al., 1980; Bhuiyan et al., 2009). During defense responses, lignin or lignin-like phenolic compound accumulation has been shown in a variety of resistant plant-microbe interactions. Lignification also enables the plant cell wall more resistant to mechanical pressure exerted by penetrating fungal pathogens. Additionally, it can increase the resistance to water, thus lessening the effect of cell wall degrading enzymes from the pathogen (Vance et al., 1980; Nicholson and Hammerschmidt, 1992; Bhuiyan et al., 2009).

## Bulk/Synchrotron-Based FTIR Spectroscopy and Focal Plane Array Imaging

The type II resistance to FHB, which is responsible for halting the infection within the spike, is favored by breeders because it can be readily identified under greenhouse conditions (Buerstmayr et al., 2002, 2003). However, it is not well-understood what comprises the type II resistance against FHB. Phenols and triticenes from wheat can be toxic to F. graminearum (Spendley and Ride, 1984), although phytoalexines often are discounted for FHB resistance in wheat but lignification after infection is considered important (Ride and Barber, 1987). In this study, FTIR on bulk samples and cross sections of rachis effectively differentiated the cell wall composition between SU3 and MM;

the most indicative peaks between cultivars are related to lignin, cellulose, and hemicellulose. These results are consistent with our previous observations (Lahlali et al., 2015), and together showed that FTIR spectroscopy may be used to help identify FHB resistant germplasms based on a range of biochemical differences in fingerprint regions (Alonso-Simon et al., 2004; Lahlali et al., 2015). FTIR has been used to identify biochemical modifications in plants under environmental and biotic stresses (Alonso-Simon et al., 2004; Erukhimovitch et al., 2010; Lahlali et al., 2015). In this study, synchrotron-based FTIR analysis underlines increased amount of lignin, pectin, and hemicellulose in the cell wall of epidermis and highlights these structural components may play a role in the cell wall reinforcement. Interestingly, on the rachis cross sections, the focal plane array imaging also identified a new band around 1710 cm−<sup>1</sup> related to oxidative stress as a response following the invasion of epidermis and vascular bundles. Because this band was much more intense in MM than in SU3, it may serve as another biomarker for FHB resistance.

## Bulk XRF Spectroscopy for Nutrients Composition in Rachis

Although the disease resistance in plants is primarily a function of genetics, the ability of a plant to express its genetic potential can be affected by nutrition and nutrients can be important factors in disease resistance (Agrios, 2005). Interactions between

plants, nutrients, and pathogens can be complex and are not well-understood. Essential nutrients have a major role for plants to develop strong cell walls which can affect disease severity (Huber and Graham, 1999; Steinkellner et al., 2005), and the differences in nutrients composition can possibly influence the type II resistance of rachis to FHB. The X-ray analysis has many potential applications in studying plant-pathogen interaction, but the use has rarely been reported due likely to limited access to the equipment. The current data showed that K, Zn, and Fe are present in greater quantities in the rachis of both wheat cultivars. The function of metal ions in plant disease resistance varies; it is known that the fungal-spore germination and plant infection is stimulated by compounds exuded from the plant (Brenner and Romeo, 1986; Steinkellner et al., 2005). In this study, the Ca content was substantially higher in SU3 than in MM after infection, which may indicate the relevance of this element to FHB resistance. As known, Ca play an essential role in formation of healthy and stable cell walls (Dordas, 2008; War et al., 2012). The divalent Ca also plays critical role in signaling transduction upon infection through the calciumdependent protein kinases pathway (Romeis et al., 2001) and inhibits the formation of cell wall degrading enzymes. In a recent study, an increase in cellular Ca concentration was found to be one of the earliest events of induced plant defense response against many pathogens (Thuleau et al., 2013). The authors highlighted that sphingoid long-chain bases and Ca ions may be inter connected to regulate cellular processes leading to plant susceptibility or resistance mechanisms. The current results also suggest that K, Fe, Zn, Si, and Mn possibly play a role in the resistance to FHB. The critical role of K in plant stress

response has been recognized (Wang et al., 2013). Potassium is an essential element affecting most of the biochemical and physiological processes relating to plant growth and metabolism (Dordas, 2008; Wang et al., 2013), altering the host-parasite compatibility by changing the interactive environment. As a facultative parasite, the FHB pathogen invades senescing tissues more rapidly and may create the condition by releasing toxins. Nutrients which support the metabolic activities of host cell and delay the senescence of tissue would potentially increase the resistance or tolerance of plant to facultative parasites by promoting the development of thicker outer walls in epidermal cells (Agrios, 2005). The balance between K and other elements is also important for resistance. For example, the ratios of N/K and K/Ca may affect the susceptibility of plant to diseases (Marschner, 1995). A substantial increase of other elements, including Fe, Zn and Mn in inoculated relative to control MM, may also suggest their involvement in the susceptibility to FHB. Fe is essential for the growth of almost all living organisms; it acts as a catalyst in many metabolic processes such as respiration and photosynthesis (Kieu et al., 2012). Fe can also activate the enzymes involved in plant infection by pathogen or host defense response by promoting antimycosis. Fe does not seem to affect the lignin synthesis, however Fe is a component in peroxidase which stimulates enzymes involved in lignin biosynthesis pathways (Graham and Webb, 1991). Fe appears to be required for three critical defense responses: the formation of cell wall appositions, oxidative burst, and production of pathogenesis-related proteins (Huber and Jones, 2013). In this study, the amount of Fe is very high in SU3 but increased also in infected MM rachises, which may indicate its involvement in infection but its role in FHB resistance is uncertain based on these results. The micronutrient Mn has been linked to the resistance to foliar diseases (Dordas, 2008) while the contribution of Zn to disease resistance is uncertain, but positive and negative effects suggested, depending on the disease (Graham and Webb, 1991; Grewal et al., 1996). Therefore, the higher Mn content in MM relative to that in SU3 and its further increase in MM after inoculation may point to its relatedness to susceptibility of FHB. This observation is in contrast to some of the reports which suggested the importance of Mn in resistance to other diseases (Graham and Webb, 1991; Huber, 1996; Hammerschmidt and Nicholson, 2000; Vidhyasekaran, 2004). Despite its uncertain role in plant disease resistance, Zn plays an important role in proteins and starch biosynthesis. At low concentrations, Zn induces amino acids but reduces sugars in plant tissues (Dordas, 2008). It is also involved in membrane protection against oxidative damage through the detoxification of superoxide radicals (Cakmak, 2000). However, the increase in Zn content during infection of MM links this element more to susceptibility than resistance to FHB. Surprisingly, silicon presence in infected SU3 at 10 DPI, may indicate its crucial role in the resistance to FHB. Although, different reports highlight the importance of Si in plant disease resistance against pathogenic fungi, but the mechanism of Si in plant resistance to diseases is still unclear (Fauteux et al., 2005, 2006).

Comparative FTIR and XRF spectroscopy analyses indicate that lignin, cell wall polymers such hemicellulose as well as nutrients composition in rachis like K, Ca, and Fe are possibly important factors contributing to the strength of cell wall and consequently the type II resistance to FHB. Further investigations on the localization of these nutrients on phloem, xylem, and companion cell walls may better identify their roles in FHB resistance. The results from this work demonstrate the capability of synchrotron-based technique to assess plant histological changes relating to disease resistance mechanisms. Some of the recent studies used XRF to investigate foliar nutrient changes during infection (Pereira and Milori, 2010; Tian et al., 2014) or caused by fungal endophytes (Gonzalez-Chavez et al., 2014; Nayuki et al., 2014). Considering the importance of nutrient composition in several plant biochemical and physiological processes (Dordas, 2008) and based on our findings on wheat rachis, we believe that these synchrotronbased techniques can revolutionize the field of plant-microbe interactions by identifying nutrients involved in plant response mechanisms to abiotic and biotic stresses based on their localization, distribution, and motility at cellular spatial scale.

## CONCLUSION

This work is the first report using chemometric tools to determine the differences in biochemical and nutrients composition in the tissue of resistant and susceptible hosts. The changes identified in chemical composition following the infection with FHB during this study may be used as biomarkers for screening FHB resistance. The FTIR spectroscopy and imaging indicated the role of cell wall lignification in FHB resistance associated with SU3, and changes in cell wall composition and cell integrity were substantially more pronounced in epidermis and vascular bundles of the susceptible MM after infection. The oxidation stress band (∼1710 cm−<sup>1</sup> ) was more intense in MM than in SU3 after inoculation. XRF identified higher increase in the elements Ca, K, Si, and Fe in SU3 than in MM after infection, indicating possible involvement of these nutrients in FHB resistance. Further investigation is required to better characterize the roles of these nutrients in FHB resistance based on their localization, distribution, and motility at cellular levels, and use them as biochemical markers for screening FHB resistance.

## AUTHOR CONTRIBUTIONS

RL, LW, PF, and CK conceived this research. RL, CK, and SK collected and analyzed the data. NS, MK participated in sample preparation (microtome sectioning) and in fluorescence microscopy. DM helped during XRF data collection. LF helped in plant growth and inoculation. GS and JL helped in collecting confocal fluorescence data. CK, GP, and PF supervised the work. RL wrote the manuscript and all authors contributed to the manuscript revision.

## ACKNOWLEDGMENTS

We acknowledge Agriculture Development Fund of Saskatchewan for funding this research work. The research described in this paper was performed at the Canadian Light Source which is funded by the Canada Foundation for Innovation, the Natural Sciences and Engineering Research Council of Canada, the National Research Council Canada, the Canadian Institutes of Health Research, the Government of Saskatchewan, Western Economic Diversification Canada, and the University of Saskatchewan. Research by PF and LW. was funded by the Wheat Improvement Flagship Program, which is the National Research Council of Canada's contribution to the Canadian Wheat Alliance. The NRCC No.: 56191.

## SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fmicb. 2016.00910

Figure S1 | Asymptomatic and symptomatic infected rachis of the resistant and susceptible cultivars Sumai3 and Muchmore in comparison with control rachis at 10 days post-inoculation with FHB. Scale bar 4 mm. (A): control Sumai3, (B): inoculated Sumai3 with FHB, (C) control Muchmore, and (D) inoculated Muchmore with FHB.

Figure S2 | Fluorescent (A–D) microscope images from cross section (10 µm) of control and inoculated wheat cultivars Sumai3 (SU3, A,B) and Muchmore (MM, C,D) after 4 days of infection with FHB. Red arrows show

### REFERENCES


changes in cell wall thickness following the pathogenic infection with FHB. Scale bar = 75 µm.

Figure S3 | Confocal microscope movie showing cell wall structures of the infected wheat rachis of the resistant cultivar Sumai3 (A) and the susceptible cultivar Muchmore (B) with Fusarium head blight at 4 days post-inoculation.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2016 Lahlali, Kumar, Wang, Forseille, Sylvain, Korbas, Muir, Swerhone, Lawrence, Fobert, Peng and Karunakaran. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Antioxidant Secondary Metabolites in Cereals: Potential Involvement in Resistance to Fusarium and Mycotoxin Accumulation

Vessela Atanasova-Penichon\*, Christian Barreau and Florence Richard-Forget

MycSA, Institut National de la Recherche Agronomique, Villenave d'Ornon, France

#### Edited by:

Daniela Gwiazdowska, Poznan University of Economics, Poland

#### Reviewed by:

Falk Hillmann, Leibniz Institute for Natural Product Research and Infection Biology e.V. - Hans-Knöll-Institute, Germany Ana Butron, Consejo Superior de Investigaciones Científicas, Spain

#### \*Correspondence:

Vessela Atanasova-Penichon vessela.atanasova-penichon@ bordeaux.inra.fr

#### Specialty section:

This article was submitted to Fungi and Their Interactions, a section of the journal Frontiers in Microbiology

Received: 29 January 2016 Accepted: 04 April 2016 Published: 22 April 2016

#### Citation:

Atanasova-Penichon V, Barreau C and Richard-Forget F (2016) Antioxidant Secondary Metabolites in Cereals: Potential Involvement in Resistance to Fusarium and Mycotoxin Accumulation. Front. Microbiol. 7:566. doi: 10.3389/fmicb.2016.00566 Gibberella and Fusarium Ear Rot and Fusarium Head Blight are major diseases affecting European cereals. These diseases are mainly caused by fungi of the Fusarium genus, primarily Fusarium graminearum and Fusarium verticillioides. These Fusarium species pose a serious threat to food safety because of their ability to produce a wide range of mycotoxins, including type B trichothecenes and fumonisins. Many factors such as environmental, agronomic or genetic ones may contribute to high levels of accumulation of mycotoxins in the grain and there is an urgent need to implement efficient and sustainable management strategies to reduce mycotoxin contamination. Actually, fungicides are not fully efficient to control the mycotoxin risk. In addition, because of harmful effects on human health and environment, their use should be seriously restricted in the near future. To durably solve the problem of mycotoxin accumulation, the breeding of tolerant genotypes is one of the most promising strategies for cereals. A deeper understanding of the molecular mechanisms of plant resistance to both Fusarium and mycotoxin contamination will shed light on plant-pathogen interactions and provide relevant information for improving breeding programs. Resistance to Fusarium depends on the plant ability in preventing initial infection and containing the development of the toxigenic fungi while resistance to mycotoxin contamination is also related to the capacity of plant tissues in reducing mycotoxin accumulation. This capacity can result from two mechanisms: metabolic transformation of the toxin into less toxic compounds and inhibition of toxin biosynthesis. This last mechanism involves host metabolites able to interfere with mycotoxin biosynthesis. This review aims at gathering the latest scientific advances that support the contribution of grain antioxidant secondary metabolites to the mechanisms of plant resistance to Fusarium and mycotoxin accumulation.

Keywords: Fusarium, mycotoxins, cereals, antioxidants, resistance

## INTRODUCTION

Fusarium Head Blight (FHB) of small-grain cereals such as wheat and barley and Gibberella Ear Rot (GER) and Fusarium Ear Rot (FER) of maize are three devastating fungal diseases affecting crops worldwide. Both FHB and GER are caused by the same Fusarium species on wheat and maize respectively, Fusarium graminearum and Fusarium culmorum being the most predominant in Europe (Bottalico and Perrone, 2002). FER is caused by Fusarium species belonging to the Gibberella fujikuroi complex, including Fusarium proliferatum and Fusarium verticillioides. These three fungal diseases lead to huge economic losses, resulting from reduced yields, deteriorated grain quality and contamination of grains with mycotoxins.

F. graminearum and F. culmorum can produce zearalenone and type B trichothecenes (TCTB). TCTB include deoxynivalenol (DON) and its two acetylated forms, 3-acetyl-deoxynivalenol (3-ADON) and 15-acetyl-deoxynivalenol (15-ADON), as well as nivalenol (NIV) and its acetylated form 4-acetylnivalenol or fusarenon X (FX). F. proliferatum and F. verticillioides are major sources of maize contamination with fumonisins (FB) among which fumonisin B1 (FB1), FB2 and FB3 are predominant. All these Fusarium toxins exhibit various acute and chronic effects on humans and animals (Bennett and Klich, 2003). Consequently, thresholds for maximal DON, FB1+FB2 and zearalenone content in foodstuffs have been set up in Europe: Commission regulation published in 2005 (EC number 856/2005) and amended in July 2007 (EC number 1126/2007) for DON and zearalenone and Commission regulation number 1126/2007 for FB1+FB2. Published surveys on the mycotoxin status of European cereals and derived products clearly show that mycotoxins produced by Fusarium species are ubiquitously present and that contamination levels exceeding the EU maximum levels or guidance values are likely to occur, leading to significant economic losses (Streit et al., 2012; Schatzmayr and Streit, 2013; Nordkvist and Haggblom, 2014). There are no data on the economic costs of mycotoxins in Europe with the exception of one study in Hungary where these costs were estimated to be 100 million euros in 1998, consecutively to a severe FHB outbreak (Milicevic et al., 2010). Economic costs directly result from: (1) yield loss due to fungal diseases, (2) reduced crop value resulting from mycotoxin contamination, (3) losses in animal productivity and (4) trade impacts. Additional costs include the cost of management linked to prevention, sampling, analysis, mitigation, litigation, and research. For instance, the annual cost for monitoring aflatoxin alone in the US is estimated to be 30–50 million dollars.

Production of DON, FB and zearalenone by Fusarium spp. occurs during infection of crops. Structurally, DON is defined as an epoxide containing sesquiterpenoid skeleton. The epoxide group at position 12–13 allows DON to bind to ribosomes leading to the activation of various protein kinases, the modulation of gene expression, the inhibition of protein synthesis and cell toxicity (Maresca, 2013). The chemical structure of FB consists of an aminopentol backbone with one tricarballylic acid on each side chain and one or more hydroxyl groups (Bezuidenhout et al., 1988). Due to their structural similarity with sphinganine, FB may act as specific inhibitors of sphingolipid biosynthesis, which are major constituents of cell membranes and important components of many signaling pathways (Merrill et al., 2001). Zearalenone is a phenolic resorcyclic acid lactone and its toxicity is mainly related to its ability to competitively bind to estrogen receptors. Excellent reviews describing the detailed biosynthesis pathway for DON and FB have been recently published (Brown et al., 2007; Alexander et al., 2009) whilst there are still significant knowledge gaps in the understanding of zearalenone biosynthesis. During the last decades, tremendous progress has also been made in identifying the environmental factors that significantly impact the regulation of DON and FB biosynthesis during the colonization of plant tissues (Picot et al., 2010; Merhej et al., 2011; Montibus et al., 2015). Temperature, water availability, pH variations, nutrient sources and plant defense metabolites were pointed out as key factors regulating DON and FB production.

TCTB and FB toxins are heat-stable molecules that are not fully eliminated during food and feed processing (Hazel and Patel, 2004; Humpf and Voss, 2004). As a result, the best way to reduce or avoid contamination of food and feed is to control the biosynthesis of these mycotoxins at the field level during plant cultivation. Three major factors have been reported to significantly influence fungal development and mycotoxin accumulation in grains: (i) environmental conditions, (ii) agricultural practices, and (iii) susceptibility of cereal genotypes (Edwards, 2004). Several cultural practices such as crop rotation, tillage, use of chemicals as well as breaking the fungal disease cycle by adapting the sowing period or using resistant hosts have been shown to reduce efficiently the level of primary pathogen inoculum (Pirgozliev et al., 2003). More recently, integrated management studies have demonstrated the improvements that can be gained by combining multiple control strategies (Blandino et al., 2012). Plant breeding strategies are among the most promising and performing approaches to durably fight against Fusarium diseases and the contamination of cereals with mycotoxins. Undoubtedly, such strategies will be among the most important pillars of any integrated disease management programs (Terzi et al., 2014).

Plant resistance to Fusarium and mycotoxin accumulation is a highly complex mechanism. Five major types of resistance have been classified for wheat, and are transferable for barley and maize. However, mechanisms associated with one of these five types can be host specific. In wheat and barley, type I resistance operates against initial infection of the floret (Schroeder and Christensen, 1963), and in maize, it may be associated with silk resistance. Type II resistance limits spreading of the infection within the host. Unlike in wheat, fungal infection in barley usually does not spread from initially infected spikelets to adjacent spikelets. Type II resistance has therefore little meaning for barley. Type III concerns resistance to grain infection; type IV, tolerance and ability to maintain yields and finally type V resistance gathers all mechanisms of resistance to mycotoxin accumulation (Miller et al., 1985; Mesterhazy, 1995, 2002). Boutigny et al. (2008) proposed to divide the type V resistance into two components. The first one, called type V-1, represents resistance to toxin accumulation operated by metabolic transformation involving biochemical modification catalyzed by enzymes such as UDPglycosyltransferases, gluthatione-S-transferases or cytochrome P450 mono-oxygenases (Karlovsky, 2011; De Boevre et al., 2014). The second one (type V-2) corresponds to resistance via inhibition of mycotoxin biosynthesis through the action of plant endogenous compounds. These compounds include both constitutively synthetized compounds and those induced in response to pathogen infection.

In addition to genetic approaches aiming at identifying and characterizing Quantitative Trait Loci (QTL) for FHB, FER and GER resistance, recent biochemical studies have been attempted to decipher the biochemical defenses that contribute to FHB, FER, and GER resistance and low mycotoxin accumulation. Mainly based on comparative approaches of metabolite composition of resistant and susceptible varieties, challenged or not with Fusarium, these attempts have implemented targeted analytical approaches and non-targeted global metabolomic developments (Siranidou et al., 2002; Bollina et al., 2011; Atanasova-Penichon et al., 2012; Picot et al., 2013; Sampietro et al., 2013; Gunnaiah and Kushalappa, 2014). A large set of metabolites potentially acting in cereals to counteract toxigenic Fusaria and reduce mycotoxin accumulation has been highlighted by these studies. These metabolites derive from primary and secondary plant metabolism and can be roughly classified in six major groups: fatty acids, amino acids and derivatives, carbohydrates, amines and polyamines, terpenoids and phenylpropanoids (Gauthier et al., 2015). Plant secondary metabolites with antioxidant properties, mainly terpenoids and phenylpropanoids are among the most frequently reported for their potential involvement in plant defense against fungal pathogens (Balmer et al., 2013). In addition to their key role as plant defense mediators and their participation to cell wall reinforcement, these compounds display antifungal properties and some of them can interfere with mycotoxin biosynthesis (Gauthier et al., 2015).

Here, we review the latest scientific advances that support the potential contribution of grain antioxidant secondary metabolites to cereal resistance to Fusarium and mycotoxin accumulation focusing on (i) in vitro studies on the effect of antioxidants on fungal development and mycotoxin production by Fusarium, (ii) identification of the major antioxidant metabolites that Fusarium can encounter during ear infection process, from anthesis to grain maturity, and (iii) relation between resistance to Fusarium and antioxidant content in cereals.

## PRINCIPAL ANTIOXIDANT SECONDARY METABOLITES IN CEREALS

In cereals, the main secondary metabolites with antioxidant activity belong to three groups including phenolic compounds, carotenoids and tocopherols (Boutigny et al., 2008). An additional group, consisting of benzoxazinoid derivatives, less abundant in grains but with multiple recognized biological activities, needs also to be addressed.

#### Phenylpropanoids

Phenolics are considered the major contributors to total antioxidant capacity of cereal grains (Awika et al., 2003; Gorinstein et al., 2008). Phenolic compounds derive from the phenylpropanoid pathway and are divided into two groups: flavonoid phenylpropanoids including flavones, flavonols, flavanones, flavanols, anthocyanins and chalcones, and non-flavonoid phenylpropanoids such as stilbenes, lignans, and phenolic acids.

#### Non-Flavonoid Phenylpropanoids

Among non-flavonoid phenylpropanoids, phenolic acids are predominant in cereals (Dykes and Rooney, 2007; Gauthier et al., 2015). Phenolic acids are derivatives of either benzoic or cinnamic acids. In cereals, benzoic acid derivatives include gallic, p-hydroxybenzoic, vanillic, syringic, and protocatechuic acids while cinnamic acid derivatives include caffeic, chlorogenic, p-coumaric, sinapic, and ferulic acids. Phenolic acids found in cereals exist in both soluble (free) and insoluble (cell-wallbound) forms. Soluble phenolic acids are either free acids or esterified to sugar conjugates. Insoluble phenolic acids are linked to various polysaccharides and to lignin through ester and ether bonds. Soluble forms are compartmentalized within the plant cell vacuoles and insoluble forms are distributed in cell walls. Phenolic compounds are concentrated in the outer layers of the grain, the pericarp and the aleurone, and in the germ, and are less abundant in the endosperm (Bily et al., 2003; Das and Singh, 2015).

Studies comparing composition of phenolic compounds in cereals reveal significant differences between cereal types, within varieties as well as within grain fractions (Adom and Liu, 2002; Ndolo and Beta, 2014; Pihlava et al., 2015). This variability associated to the large set of extraction protocols and analytical procedures that can be used when addressing the phenolic composition of grains explains the frequent discrepancies observed in published data. For instance, in the study of Adom and Liu (2002), maize grains were reported as the richest in total phenolic acids, followed by wheat, oat and rice while, in the report of Irakli et al. (2012), the highest levels in both free and bound phenolic acids were found in oat, followed by maize, wheat and rice. Nevertheless, in all published studies, the major portion of phenolics in grains exists as bound forms: 85% in maize, 75% in oat and wheat and 62% in rice (Adom and Liu, 2002; Boz, 2015; Das and Singh, 2015).

Among free phenolic acids, ferulic acid is by far predominant, followed by p-coumaric and vanillic acids (Adom and Liu, 2002; Bakan et al., 2003; Santiago et al., 2007). Caffeic, phydroxybenzoic and sinapic acids are also present but at very low concentrations (0.5–1.5µg/g) (Irakli et al., 2012). In addition to phenolic acid monomers, hydroxycinnamic polyamines such as p-coumaroyl-feruloylputrescine (CFP) and diferuoylputrescine (DFP) have been quantified in significant amounts in free phenolic maize extracts. Their concentrations can reach 330 µg equiv. 8-5′ -benzofuran-diferulic acid/g (Moreau et al., 2001; Atanasova-Penichon et al., 2012; **Figure 1A**).

Among cell-wall-bound phenolic acids, ferulic acid is the most abundant one in common cereals and represents up to 90% of the total phenolic compounds (Adom and Liu, 2002; Boz, 2015). Concentrations of this compound can reach 3000µg/g for some maize varieties (Li et al., 2007). Ferulic acid and its oxidatively coupled products named ferulic acid dehydrodimers or diferulic acids (DiFA) are found in greater concentrations in cereal brans. Ferulic acid dehydrodimers are potent antioxidants and are

ester-linked to the cell wall polymers. The most important ferulic acid dehydrodimers in common cereals are: 8–5 DiFA (open form), 5-5′DiFA, 8-O-4′DiFA and 8-5′benDiFA (benzofuran form) and their sum represents 20–30% of the total phenolic acids (Atanasova-Penichon et al., 2012; Boz, 2015) (**Figure 1A**). The highest levels of ferulic acid dehydrodimers in cereals ranged between 250 and 475µg/g (Jilek and Bunzel, 2013).

#### Flavonoid Phenylpropanoids

The second group of phenolic compounds with significant concentrations in cereal grains is the class of flavonoids, located in the pericarp and the germ (Dykes and Rooney, 2007; Das and Singh, 2015). Flavonoids are also major active ingredient in corn silks (Hu et al., 2010). As phenolic acids, most grain flavonoids are found in the cell-wall-bound fraction: 93% of the total flavonoids in wheat, 91% in maize, 65% in rice and 61% in oat (Adom and Liu, 2002). According to Adom and Liu (2002), maize grains contain the highest level in total flavonoids followed by wheat, oat and rice. The most frequently cited flavonoids in cereal grains are the flavonols kaempferol and quercetin for maize (Das and Singh, 2015), the flavanone naringenin and its glycosylated forms and the flavanols catechin and epicatechin for barley (Bollina et al., 2010, 2011; Zilic et al., 2011), the flavones vitexin and luteolin for rye (Pihlava et al., 2015) and the anthocyanins in colored grains (Dykes and Rooney, 2007). According to the report of Reid et al. (1992), corn silks are characterized by high concentrations of flavones including luteolin and apigenin and flavone glycosides such as maysin, iso-orientin, and iso-vitexin.

## Lipophilic Compounds

In cereals, the major lipophilic secondary metabolites with antioxidant properties include tocols (or commonly referred to as tocopherols) and carotenoids. The latest group consists of carotenes, of which α-carotene and β-carotene are the major representatives, and xanthophylls, mostly lutein and zeaxanthin. According to the results of two-year field studies, lipophilic antioxidant secondary metabolites represent less than 0.25% of total antioxidant secondary metabolites in mature maize grains (Atanasova-Penichon et al., 2012; Picot et al., 2013; **Figure 1A**). Concentrations in carotenoids in grains significantly vary according to the cereal type, from 1.8µg/g in oat to 18.2µg/g in maize (Ndolo and Beta, 2013). The major carotenoids in maize are concentrated in the endosperm fraction, ranging from 14.2 to 31.2µg/g of endosperm, while the major carotenoids in smallgrain cereals are found in germ and range from 3.2 to 14.8µg/g of germ (Ndolo and Beta, 2013). In wheat, barley and oat grains, lutein was reported as the major xanthophyll and zeaxanthin as the minor one (Ndolo and Beta, 2013).

Tocol composition of cereals includes tocopherols (α-, β-, δ- and γ-tocopherol) and tocotrienols (α-, β-, δ- and γtocotrienols). The α-forms are predominant (Gutierrez-Gonzalez et al., 2013). Tocopherols are mainly present in the germ fraction while tocotrienols are present in the pericarp and endosperm fractions (Falk et al., 2004). In small-grain cereals such as oat, barley and wheat, tocotrienols are the main tocols and their concentrations range between 40 and 60µg/g depending on the cereal type and the variety (Falk et al., 2004). Conversely, maize grains contain more tocopherols than tocotrienols, with concentrations ranging between 34–70µg/g and 20–25µg/g respectively (Das and Singh, 2015).

## Benzoxazinoid Derivatives

Benzoxazinoids are a group of secondary metabolites found in maize, rye, wheat and triticale, but not in sorghum and rice (Niemeyer, 2009; Andersson et al., 2014). The mono and dihexose conjugates of cyclic hydroxamic acid DIBOA (2,4-dihydroxy-1,4-benzoxazin-3-one) and the corresponding lactam HBOA (2-hydroxy-1,4-benzoxazin-3-one) are the major benzoxazinoids in rye, while in wheat and maize, DIMBOA (2,4-dihydroxy-7-methoxy-1,4-benzoxazin-3-one) and its glycosylated derivative dominate (Etzerodt et al., 2015; Pihlava et al., 2015). Rye is by far the crop characterized by the highest content in benzoxazinoids, with concentration values more than 20-fold higher to that found in whole grain wheat (Andersson et al., 2014; Pihlava et al., 2015). Benzoxazinoids are located in all fractions of seeds, with greater concentrations in bran and germ (Pihlava et al., 2015).

## ANTIOXIDANTS MODULATE FUNGAL DEVELOPMENT AND MYCOTOXIN PRODUCTION BY FUSARIUM SPP.

The biosynthetic pathways that lead to the production of TCTB and FB by Fusarium species have been well established and characterized by several oxygenation steps (Proctor et al., 2003; Desjardins, 2007). Therefore, changes in the oxidative parameters of the medium are likely to interfere with the fungal secondary metabolism and to modulate the level of mycotoxin production (Ponts, 2005; Montibus et al., 2015). Due to the antioxidant properties of cereal secondary metabolites, several studies have been devoted to their anti-fungal and anti-mycotoxin effect.

## Antifungal Properties of Cereal Antioxidants

Phenolic acids are toxic toward many fungi including Fusarium species (Guiraud et al., 1995; Ponts et al., 2011; Gauthier et al., 2016). Their fungicidal efficiency has been characterized against different Fusarium species and IC<sup>50</sup> values (concentration that inhibits 50% of fungal growth) ranging between 0.7 to >10 mM have been reported (**Table 1**). The comparison of IC<sup>50</sup> values has to be considered with caution as these values are method and condition dependent. IC<sup>50</sup> values gathered in **Table 1** also illustrate the great variability in phenolic acid sensitivity between F. graminearum strains. Similar variations seem to occur among F. culmorum strains; however there is insufficient data to assert this hypothesis. When comparing results obtained in the same conditions by Gauthier et al. (2016) and Ponts et al. (2011) for F. graminearum and F. culmorum, it also appears that F. culmorum strains (IC<sup>50</sup> between 8.8 to >10.0 mM) could be less susceptible to caffeic acid than F. graminearum strains (IC<sup>50</sup> between 4 to 10.1 mM). Phenolic acids could be ranked in ascending order of toxicity toward F. graminearum as follows: chlorogenic acid < p-hydroxybenzoic acid < caffeic acid < syringic acid < p-coumaric acid < ferulic acid (**Table 1**). Chlorogenic acid, a cinnamic-derived phenolic acid, displays a lower fungicidal activity than p-hydroxybenzoic and syringic acid, which is contradictory with the assumption that cinnamicderived phenolic acids are roughly more toxic than benzoic acid-derived ones (Beekrum et al., 2003; Ponts et al., 2011). Chlorogenic acid displays however weak lipophilic properties, which, according to Guiraud et al. (1995) and Ponts et al. (2011) are primary factors in the antifungal efficiency of phenolic acids.

Compared to IC<sup>50</sup> values ascribed to phenolic acids, those determined for flavones and flavanones against different Fusarium species including F. culmorum and F. graminearum are substantially weaker (**Table 1**). According to Treutter (2005), the efficiency to inhibit fungal growth directly results from the ability of flavonoids to irreversibly combine with nucleophilic amino acids in fungal proteins. Among the different groups of the flavonoid subclass, i.e., flavanones, flavones and flavanols, data reported in **Table 1** suggest that unsubstituted flavones and flavanones (IC<sup>50</sup> between <0.05 to 1.6 mM) display a more efficient antifungal activity than hydroxylated flavones, i.e., flavonol (IC<sup>50</sup> between 2.9 to 4.8 mM). The promising ability of flavonoids to inhibit spore development and restrain mycelium hyphae elongation of plant pathogens have been the subject of numerous investigations (Treutter, 2006; Mierziak et al., 2014). Flavonoids are also the subject of intensive medical research, with the aim of identifying alternatives to synthetic drugs for counteracting human fungal pathogens that increasingly display resistance to commonly used antifungal agents such as triazole ones (Cushnie and Lamb, 2005).

As regards to benzoxazinoids, their antifungal activities have been the subject of numerous publications (Glenn et al., 2001; Martyniuk et al., 2006). According to the results of Glenn et al. (2001), Fusarium species responsible for GER and FER in maize show a wide range of sensitivity to 6 methoxybenzoxazolin-2(3H)-one (MBOA) and benzoxazolin-2(3H)-one (BOA), with the most tolerant being F. verticillioides, Fusarium subglutinans and F. graminearum. As demonstrated by Glenn et al. (2001), differences in tolerance can be ascribed do different abilities to metabolize and therefore detoxify these antimicrobial compounds.

## Cereal Antioxidants Inhibit Mycotoxin Biosynthesis by Fusarium

In addition to displaying antifungal properties, several antioxidant secondary metabolites of cereals can modulate the production of mycotoxins by various fungal pathogens. According to the report of Boutigny (2007), cinnamic acid derivatives such as sinapic, caffeic, p-coumaric, chlorogenic, and ferulic acids are efficient inhibitors of TCTB production by F. graminearum and F. culmorum while benzoic acid derivatives, with the exception of syringic acid, have an activating effect. It is noteworthy that the effect of phenolic compounds is strain and molecule dependent (Boutigny et al., 2009; Gauthier et al., 2016). Increasingly, phenolic acids are becoming the subject of anti-mycotoxin research and many groups have demonstrated their efficiency to modulate in vitro the biosynthesis of various mycotoxins, including type A trichothecenes (Ferruz et al., 2016), fumonisins (Beekrum et al., 2003; Samapundo et al., 2007; Atanasova-Penichon et al., 2014), ochratoxin (Palumbo et al., 2007), and aflatoxins (Norton, 1999).

Similarly, several studies illustrated the potential impact flavonoids could exert on mycotoxin production. Recently, rutin was demonstrated as a potent inhibitor of aflatoxin B1 production by Aspergillus flavus (Norton, 1999; Chitarrini et al., 2014) and naringin, hesperidin and some glucosides were characterized for their capacity to restrain the production of patulin by Penicillium expansum, Aspergillus terreus, and Byssochlamys fulva (Salas et al., 2012). As regards to TCTB, effects of flavonoids on their biosynthesis have been poorly documented with exception of the publication of Desjardins et al. (1988) describing an inhibitory effect of flavones on the biosynthetic step that catalyzes the conversion of trichodiene (the first chemical intermediate in trichothecene biosynthesis) to oxygenated trichothecenes. In addition to phenolic compounds, carotenoids and tocopherols are potent cereal antioxidant compounds, but their antifungal and antimycotoxin activities against Fusarium are poorly documented. Recent works have shown that sub-lethal doses of α-tocopherol significantly affected fumonisin production (Picot et al., 2013) and that 50µg/ml of β-carotene added to the culture medium led to a significant decrease (close to 50%) in TCTB accumulation (Boutigny, 2007). A few additional studies investigated the impact of carotenoids on other mycotoxin production but they led to opposite results,

#### TABLE 1 | IC<sup>a</sup> <sup>50</sup> values of non-flavonoids and flavonoids against different Fusarium species.


<sup>a</sup>Concentration that inhibits 50% of growth.

depending on the mycotoxin targeted. While capsanthin (a major carotenoid in paprika) has been shown to inhibit aflatoxin yield (Masood et al., 1994), more recent results demonstrated its lack of inhibitory effect on ochratoxin production (Santos et al., 2010).

The toxin suppressive effects of benzoxazinoids have also been addressed in several publications. This effect was first suggested by Miller et al. (1996) who reported that 4-acetyl-benzoxazolin-2-one (4-ABOA) and related compounds present in an active maize fraction were able to reduce trichothecene and aflatoxin productions by F. culmorum and A. flavus, respectively. Antimycotoxin activities of benzoxazinoids were recently confirmed by Etzerodt et al. (2015) who demonstrated that a 250 µM concentration of DIMBOA caused 50% inhibition of 15-ADON production by F. graminearum.

## Mechanisms of Fungal Toxicity and Inhibition of Mycotoxin Production

Regardless the phytochemical considered (phenolic compound, carotenoid, tocopherol, and benzoxazinoid), the exact mechanisms by which fungal growth and mycotoxin production are inhibited remain unclear. While few published studies have focused on plant fungal pathogens, the mechanism of action of phenolic acids on human pathogens and mainly on candida species have been the subject of extensive research. Last and most significant insights on the anti-adhesion, anti-biofilm effects of phenolic acids together with their inhibitory activity on morphogenesis and fungal exoenzymes production have been recently gathered in the review of Teodoro et al. (2015). Interestingly, phenolic acids have been evidenced for their ability to breakdown the fungal membrane permeability barrier, probably through a perturbation of the lipid bilayers causing the leakage of ions and other chemicals as well as the formation of pores and modification of the electric potential of membranes (Sung and Lee, 2010).

As regards to the effects of phenolic acids against plant fungal pathogens, Guiraud et al. (1995) and more recently, Boutigny (2007) and Ponts et al. (2011) indicated that the toxicity of phenolic compounds is related to their lipophilicity as well as their strong antioxidant properties. Accordingly, Pani et al. (2014) and Roleira et al. (2010) suggested that the balance among lipophilicity and antioxidant activity can be a key factor to predict the capacity of a phenolic to inhibit mycotoxin production. However, it is essential to keep in mind that fungal cultures are multi-component systems, where the media can be considered as both lipidic and emulsion systems and that in such biological media, several physicochemical parameters including pH, light or temperature can affect the lipophilicity and antioxidant capacity of phytochemicals. Thus, correlating theoretical antioxidant potential and lipophilicity values with experimental data is not a straightforward approach. Nonetheless, the hypothesis that antioxidant properties of cereal metabolites can be primary factors for their antimycotoxin activity is highly consistent with the assumed activating effect of oxidative stress on the biosynthesis of mycotoxins. Indeed, an increasing body of work, recently gathered in the review of Montibus et al. (2015), emphasizes the modulation of fungal secondary metabolism by oxidative stress and the enhancement of mycotoxin production, including DON and FB, after exposure to reactive oxygen species. Thus, due to their capacity to quench oxygen free radicals, antioxidant metabolites may reduce or suppress upstream signals, such as oxidative stress, that modulate toxin biosynthesis. According to Guiraud et al. (1995), toxicity of phenolic acids can also be linked to their interaction with various intra or extracellular fungal enzymes, including phenol oxidases and several hydrolytic activities (El Modafar et al., 2000; Paul et al., 2003). Moreover, Passone et al. (2009) mentioned that antioxidant compounds interfere with mycotoxin production probably indirectly via their capacity to perturb the membrane function and modify its permeability. Lastly, the results of Boutigny et al. (2009) and Etzerodt et al. (2015) that indicate a downregulation of the expression of the genes involved in DON biosynthesis by F. graminearum when ferulic acid and DIMBOA is added to in vitro culture media are in accordance with a transcriptional control exerted by phenolic acids and benzoxazinoids. A similar conclusion was evidenced by the study of Sanzani et al. (2009) that proved that quercetin and umbelliferon reduced patulin accumulation by acting on the transcription level of biosynthetic genes.

## ANTIOXIDANT SECONDARY METABOLITES ENCOUNTERED BY FUSARIUM SPP. DURING THE EAR INFECTION PROCESS

To date, most of the attempts aiming at clarifying the contribution of cereal secondary metabolites to the in planta control of Fusarium mycotoxin accumulation have targeted mature grains. However, during plant development, grain antioxidant composition is likely to be dramatically modified. Fusarium commonly infects cereal ears shortly after anthesis, and the compounds the fungus has to face at the onset of infection are certainly extremely different from those found in the mature grain. There are very few dynamical studies that have addressed the composition of the grain in the early stages of grain development, when the biosynthesis of mycotoxin is initiated. In recent field experiments on maize inoculated with F. graminearum or F. verticillioides, the kinetics of fungal development and the accurate stage at which mycotoxin production is initiated were established (Picot et al., 2011, 2013; Atanasova-Penichon et al., 2012). In planta TCTB and FB accumulation were found to start between 10 and 20 days after flowering, i.e., at the milk-dough stage (**Figure 2**). Major free and bound antioxidant secondary metabolites present at the milkdough stage were quantified and are detailed in **Figure 1B**. Cellwall-bound ferulic acid, which represents 80–85% of the analyzed antioxidants, is the predominant compound, followed by ferulic acid dehydrodimers (10–15%) and free chlorogenic acid (2–7%). A particular attention was paid to free antioxidants, particularly to chlorogenic acid that represents almost 80% of the total free phenolic acids in the early stages of maize grain development. Indeed, free antioxidant compounds are more likely to interfere first with Fusarium. Phenolic compounds present in kernels at early stages are likely to alleviate fungal infection in a manner similar to that observed in the in vitro inhibition studies. In maize grains at the milk-dough stage, lipophilic antioxidants such as carotenoids and tocopherols are present at much lower levels than phenolic acids and represent only 0.02–0.03% of the total antioxidant content (**Figure 1B**). However, because their antioxidant properties are much higher than that of phenolic compounds, it cannot be excluded that, despite their low concentrations, they also significantly contribute in planta

FIGURE 2 | (A) Relative quantification of fungal DNA in maize kernels, expressed as a log10 (F. graminearum DNA/maize DNA) ratio of a susceptible variety (blue triangles, right Y-axis), and level of trichothecenes accumulated in maize kernels (red squares, left Y-axis) after silk inoculation with F. graminearum. Vertical bars show standard error of the mean. Top X-axis: thermal time from inoculation (mean value of 2 years), bottom X-axis: days after inoculation for each sampling (mean value of 2 years). Data from the 2 years and two repetitions were pooled (mean values ± SEM, n = 4). (B) Relative quantification of fungal DNA in maize kernels, expressed as a log10 F. verticillioides DNA/log10 maize DNA ratio of a susceptible variety (blue triangles, right Y-axis), and level of fumonisins accumulated in maize kernels (red squares, left Y-axis) after silk inoculation with F. verticillioides. Vertical bars show standard deviations. Top X-axis: thermal time from inoculation, bottom X-axis: days after inoculation for each sampling. Data from the two sowing date treatments were pooled. Kinetics are established with data published by Atanasova-Penichon et al. (2012) and Picot et al. (2011).

to the inhibition of Fusarium toxin biosynthesis. In addition to highlight the milk-dough stage as a critical step, the studies of Atanasova-Penichon et al. (2012) and Picot et al. (2013) provided information on the evolution of antioxidant secondary metabolites during maize ear ripening (**Figure 3**). Similar evolution patterns of phytochemicals were reported by the previous authors for the two years of experimentation, suggesting that they may correspond to an intrinsic characteristic of maize genotypes not dependent on environmental factors. Free and cellwall-bound phenolic acid as well as carotenoid and tocopherol contents show large fluctuations during the ripening of maize grains. Composition in free phenolic acid evolves qualitatively over time whereas the composition in cell-wall-bound phenolic acids, carotenoids and tocopherols remains unchanged and only shows quantitative variation at the different grain stages. Kinetic of free chlorogenic acid, cell-wall-bound ferulic acid and ferulic acid dehydrodimers as well as xanthophylls, carotenes and tocopherols during maize ear ripening is presented in **Figure 3**. Except for tocopherols, all antioxidant secondary metabolites are found at higher concentrations in the grain at early stages, suggesting that these compounds are the main antioxidants that F. graminearum and F. verticillioides potentially encounter when their mycotoxin production is initiated. **Figure 3** indicates that, after a rapid increase from anthesis to the silking-blister stage (with exception of chlorogenic acid), levels of cell-wallbound monomers represented by ferulic acid, of free phenolic acids represented by chlorogenic acid, of xanthophylls and of carotenes decrease to reach traces at maturity. As regards the ferulic acid dehydrodimers, their concentration exhibits a pattern similar to monomeric phenolic acids in the first stages of grain development and then increases until the mature stage. This increase reflects the contribution of ferulic acid dehydrodimers to cell wall structure through their role in forming bridges between hemicellulose chains. A similar pattern for evolution of bound ferulic acid has also been reported for wheat (Shewry et al., 2012) and rice (Lin and Lai, 2011). Similarly, a decrease in free ferulic acid in rice (Lin and Lai, 2011) and total free phenols in oat (Alfieri and Redaelli, 2015) has been described, supporting old data on soft and durum wheat grain (McCallum and Walker, 1990; Régnier and Macheix, 1996; McKeehen et al., 1999). However, considering that most of the studies mentioned above were conducted with few genotypes, caution should be taken in generalizing the results.

The declining concentrations of phenolic acids during grain ripening can be ascribed to several rationales. First, the activity of phenyl-alanine ammonia-lyase and L-tyrosine ammonialyase, two crucial enzymes for the initial committed step in the biosynthesis of phenylpropanoids, have been shown to be maximal only during the early stages of grain development (McCallum and Walker, 1990; Régnier and Macheix, 1996). Second, the rate of endosperm development surpasses the rate of synthesis of the outer coverings during grain ripening which leads to a dilution of the overall phenolic constituents within the grain. Third, the decrease in phenolic acids can also result from their oxidative degradation involving phenoloxidases and peroxidases, induced by the breakdown of cellular structure in the pericarp at the end of the milk stage and during further maturation (Régnier and Macheix, 1996). Finally, the decrease in cell wall-bound phenolic acid contents can be correlated with the formation of alkali-resistant bounds occurring in cross-linked polymers in cell walls not extractable with the method commonly used to analyze phenolic acids (Iiyama et al., 1994).

**Figure 3** also indicates a decrease in xantophylls and carotenes from the dough stage until maturity, in accordance with the pattern reported by Rodríguez-Suárez et al. (2014) and Sreenivasulu et al. (2010) for carotenoids in durum wheat and barley. Reduced levels in carotenes and xanthophylls during ripening may be due to their oxidation sensitivity, as a result of the high degree of unsaturation present in their structure. According to Mellado-Ortega and Hornero-Mendez (2016) and Sandmann et al. (2006), carotenes are likely to be more prone to oxidation than xanthophylls. This oxidation is caused by reactive oxygen species and especially singlet oxygen or free radicals generated by enzymatic systems such as lipoxygenases. Lipoxygenases catalyze the hydroperoxidation of polyunsaturated fatty acids, preferentially non-esterified polyunsaturated fatty acids, to form conjugated diene hydroperoxides (Loiseau et al., 2001). These hydroperoxides react with carotenoids, breaking down the carbon backbone into smaller compounds, including volatile molecules and apocarotenoids (e.g., epoxyaldehydes, ketones; Mellado-Ortega and Hornero-Mendez, 2016). Lipoxygenase is widely distributed in cereals and located in the germ and bran of the grain (Loiseau et al., 2001). A second explanation to the declining levels of carotenoids during grain ripening, could be linked to their esterification with fatty acids that produce mono and/or diesters and is catalyzed by xanthophyll acyltransferase enzymes. However, this second hypothesis is unlikely to occur based on the fact that xanthophyll esters seem to be absent or at very low levels in cereals and particularly in durum wheat (Mellado-Ortega and Hornero-Méndez, 2015).

Unlike kinetics of free and bound phenolic acids and carotenoids, the kinetic of tocopherols reported on **Figure 3F** indicates a gradual accumulation during the course of maize grain development. A similar pattern of total tocopherols was established by Gutierrez-Gonzalez et al. (2013) in oat seeds. According to the results of Falk et al. (2004) in developing barley kernels, tocopherols reach a maximum level at milk stage and remain stable until final harvest time. In rice, the total tocopherols in immature grains is about 2-fold higher than in mature ones (Lin and Lai, 2011).

Altogether, data describing the time course of F. graminearum infection and reporting the evolution of phytochemical levels in grains, provide evidence that the main antioxidant metabolites F. graminearum is likely to encounter when the production of mycotoxin starts in planta are free phenolic acids such as chlorogenic acid and bound ferulic and diferulic phenolic acids. Although present in lower concentrations, xantophylls, carotenes and benzoxazinoids, which show a high fungal toxicity, could also interfere with the fungus. Additional information on the impact antioxidant phytochemicals could exert on accumulation of mycotoxins in grains is provided by the results of recent studies that attempted to link plant resistance to Fusarium and antioxidant content of cereal grains (Siranidou et al., 2002; Bollina et al., 2011; Picot et al., 2013; Atanasova-Penichon et al., 2014).

## RELATION BETWEEN RESISTANCE TO FUSARIUM SPP. AND ANTIOXIDANT CONTENT IN CEREALS

Host resistance is one of the primary traits that can be used as a control measure, and its manipulation is recognized as one of the best economic and ecological strategies to reduce damage caused by Fusarium (Bai and Shaner, 2004). Several authors argued that the use of cereal genotypes resistant to Fusarium infection (Champeil et al., 2004) and mycotoxin accumulation (Boutigny et al., 2008) is one of the most promising ways to reduce or prevent contamination. Combined with genetic approaches and the detection of QTL linked with FHB, GER, or FER resistance, biochemical ones aiming at deciphering the chemical mechanisms plants use to fight against F. graminearum and reduce toxin production hold great potential for assisting breeding programs (Gauthier et al., 2015). Most of the biochemical approaches that have addressed cereal resistance to Fusarium spp. and have been published to date, are based on a comparative analysis of the metabolite composition of resistant and susceptible cultivars, challenged or not with Fusarium. Targeted analytical tools and, more recently, metabolomics strategies were implemented. However, while these approaches can provide interesting insights that need to be further validated through genetic studies, they cannot allow conclusive evidence on the involvement of metabolite(s) or group(s) of metabolites in resistance. Indeed, the experimental designs frequently considered a set of limited and genetically unrelated genotypes and very rarely near isogenic lines, as done in the study of Gunnaiah et al. (2012). Moreover, data delivered through metabolomics approaches require to be considered with caution since differences in metabolic profiles of the studied genotypes may actually be confounding with the effects resulting of environment, cultivation practices and developmental stage. Lastly, it should be borne in mind that chemical identification remains a significant bottleneck in plant metabolomics studies and that most of the proposed identification are putative ones.

A number of studies focusing on phenolic acids supports the assumption that, in cereals, cell-wall-bound ferulic acid together with its dehydrodimers and free chlorogenic acid could be key components of the chemical defense against toxigenic Fusarium species (Siranidou et al., 2002; Atanasova-Penichon et al., 2012; Sampietro et al., 2013). Bily et al. (2003) highlighted that ferulic acid and its dehydrodimers in maize act as resistance factors to F. graminearum through type I resistance (resistance to initial penetration) and type II resistance (resistance to propagation due to a lower degradability of the cell wall). This hypothesis was further corroborated by Picot et al. (2013), Atanasova-Penichon and Richard-Forget (2014) and Atanasova-Penichon et al. (2014) who revealed the highest concentrations of ferulic acid, ferulic acid dehydrodimers and chlorogenic acid in immature grains of the more resistant varieties of a panel of maize genotypes with different susceptibility to FER or GER. Correlations between levels of GER resistance and phenolic acid contents in maize grains are reported on **Figures 4A–C**. When addressing maize resistance to GER, mechanisms resistance of silk that can slow down the process of infection need also to be addressed. Cao et al. (2011) have investigated the role of hydroxycinnamic acids in silk resistance and observed that, unlike data gathered in **Figure 4**, high concentrations in hydroxycinnamic acids were not related with a delayed progression of F. graminearum through silks. In wheat, positive relations between both free and cell wall bound phenolic acid levels and wheat resistance to FHB were reported by Siranidou et al. (2002). By the same reasoning, Choo et al. (2015) hypothesized that the high level of black barley resistance to FHB is linked to its richness in phenolic compounds. In addition to phenolic acids but with less conclusive evidence, many other phenylpropanoid compounds have been suggested to contribute to the chemical defense to FHB, GER, or FER. This potential contribution was mainly highlighted through comparative metabolomic profiling of grains issued from resistant and susceptible genotypes, challenged or not with toxigenic Fusarium strains (Hamzehzarghani et al., 2005, 2008; Browne and Brindle, 2007; Paranidharan et al., 2008; Bollina et al., 2010, 2011; Kumaraswamy et al., 2011a,b; Gunnaiah et al., 2012; Cajka et al., 2014; Chamarthi et al., 2014; Gunnaiah and Kushalappa, 2014). A large set of constitutive as well as inducible defense metabolites potentially related to Fusarium resistance was highlighted in the afore-mentioned studies. Among these metabolites, phenylpropanoids (approximately 180 compounds), including flavonoids and non-flavonoids, represent more than 50% of the total reported metabolites (Gauthier et al., 2015). Among these 180 phenylpropanoid candidates, more than 56% are putatively assigned as flavonoids that encompasse anthocyanins, flavones, flavonols, flavanones, flavanols, isoflavones, isoflavanones, isoflavonols, and chalcones. The remaining 44% is mainly composed by phenolic acids and derivatives, including benzoic and cinnamic ones.

Indeed, the role of phenylpropanoids in disease resistance has been the subject of intensive research (Treutter, 2006). In response to pathogen infection, they are released from the cell wall or massively synthesized by the plant accumulating rapidly at the site of infection (Nicholson and Hammerschidt, 1992). The main role ascribed to these compounds in plant defense mechanisms results from their antioxidant properties (Dykes and Rooney, 2007; Agati et al., 2012), that allow them to quench reactive oxygen species (ROS), generated by both the pathogen and the plant during infection. In addition, phenolic compounds operate in defense response through direct interference with the fungus, or through the reinforcement of plant structural

components to act as a mechanical barrier against the pathogen (Siranidou et al., 2002; Treutter, 2006). Phenolic compounds such as flavonoids can also protect plant cell wall integrity upon fungal infection by inhibiting the activity of several plant cell wall degrading enzymes secreted by fungal pathogens to penetrate plant tissues (Treutter, 2005).

Concerning carotenoids and tocopherols, they are rarely regarded in comparative metabolomic studies due to their lipophilicity and the requirement of a specific extraction protocol and analytical equipment. While compilation of the metabolomic studies reported above results in a list of about 30 terpenoids (Gauthier et al., 2015), no carotenoid or tocopherol were among these terpenoid candidates. Targeted approaches aiming at relating lipophilic antioxidant composition of grains and resistance to Fusarium have also been implemented and showed positive or negative correlations, depending on the addressed group of compounds, carotenoids or tocopherols. Thus, based on the use of a set of maize genotypes with moderate to high susceptibility to GER, the experimentations we performed in our laboratory indicated higher levels of carotenoids (lutein + zeaxanthin + β carotene) in immature grains of the more susceptible genotypes (**Figure 4E**). A similar trend but not statistically significant was also observed between the level of FER resistance and carotenoid contents in maize grains (Picot et al., 2013). Accordingly, positive correlations between the levels of lutein and DON accumulation in durum wheat cultivars were reported by Delgado et al. (2014). As regard to tocopherols, while an absence of correlation between resistance levels of maize and their tocopherol contents in immature grain was observed by Picot et al. (2013), the data of Iqbal et al. (2014) showed the existence of a negative correlation between the concentrations of tocopherols and aflatoxins in rice cultivars. Interestingly, the report of Boba et al. (2011), based on the use of transgenic flax overproducing carotenoids, indicates that a general level of lipophilic antioxidants rather than the content of any particular compound is the most important factor in resistance to F. culmorum infection. The main role ascribed to carotenoids and tocols in plant/Fusarium interactions directly results from their ability to quench the free radicals produced by plant cells (the so-called "oxidative-burst") as a first response to the fungal pathogen attack (Boba et al., 2011; Gutierrez-Gonzalez et al., 2013). Moreover, carotenoids are directly linked to abscisic acid, which level in wheat and barley was shown to increase after F. graminearum or F. culmorum inoculation (Gunnaiah et al., 2012; Kumaraswamy et al., 2012; Petti et al., 2012). In fact, abscisic acid is an apocarotenoid synthetized from the cleavage of carotenoids (Tan et al., 1997). 9-cis-epoxycarotenoid dioxygenases cleave 11,12 double bonds of the cis isomers of violaxanthin and neoxanthin to form the C15 product xanthin, the first committed and key regulatory step in the abscisic acid biosynthesis pathway (Sreenivasulu et al., 2010). Abscisic acid is well known for its roles in orchestrating stress response as well

as grain maturation in plants. In addition, its role in resistance of wheat to FHB has been linked to a regulatory effect on callose deposition in the transition zone between the spikelet's rachilla and rachis. This was shown to contribute to the type II resistance (Kang and Buchenauer, 2000; Flors et al., 2005). Besides, the involvement of abscisic acid in FHB resistance has also been ascribed to its negative interaction with the signaling ethylene pathway (Flors et al., 2005) since, according to Chen et al. (2009), F. graminearum can exploit ethylene signaling to enhance colonization in wheat tissues. Lastly, the possibility that abscisic acid could limit F. graminearum penetration through its control of stomatal aperture cannot be omitted (Mauch-Mani and Mauch, 2005).

As regards to benzoxazinoids, while several reports have indicated their ability to inhibit fungal activities linked with FER and GER (Miller et al., 1996; Glenn et al., 2001; Etzerodt et al., 2015), very few studies have investigated their concentration in cereals in relation to Fusarium sensitivity. One study (Søltoft et al., 2008) has revealed positive correlations between the susceptibility of wheat to FHB and the concentrations of some benzoxazinoid derivatives, suggesting that the capacity of wheat to produce these secondary metabolites could contribute to resistance mechanisms. However, as emphasized above, the results of Søltoft et al. (2008) based on the use of a set of unrelated germplasms are not sufficient to draw conclusive evidences on the involvement of benzoxazinoids in FHB resistance.

## CONCLUSION

Cereal diseases caused by pathogenic and toxigenic Fusarium species are responsible for major economic damage worldwide. Hence, the developments of sustainable strategies to avoid Fusarium and mycotoxin contamination have been the issue of intense research over past years and decades and a broad consensus has emerged to acknowledge that the use of FHB, GER or FER resistant genotypes is one of the primary pillars of any disease management programs. However, to date, knowledge of the complex mechanisms governing cereal resistance remains insufficient, and selection for resistant genotypes is still challenging.

#### REFERENCES


Considering the available data on the interactions between antioxidant metabolites in grains and Fusarium species, we can assume that some of these compounds could significantly contribute to the protection of grains against toxigenic Fusaria and mycotoxin accumulation. Five main classes of antioxidant metabolites, phenolic acids, flavonoids, carotenoids, tocopherols and benzoxazinoids, have been evidenced for the pivotal role they could play in FHB, GER, or FER resistance. A first shared argument in favor of the involvement of phenolics, carotenoids and tocopherols is linked to their ability to quench reactive oxygen species, thus protecting biological cells. In addition, tocopherols and carotenoids have the capacity to scavenge lipid peroxyl free radical and therefore to stop the chain propagation of the lipid peroxidation cycle (Das and Roychoudhury, 2014). A second shared argument rests on the fungal toxicity exhibited by cereal antioxidant metabolites. Indeed, as demonstrated by the present review, there are numerous studies illustrating the efficiency of phenolic compounds, carotenoids, tocopherols and even benzoxazinoids to restrain the growth of toxigenic Fusaria and their production of toxins. Lastly, phenolic compounds are known to participate to the reinforcement of plant structures and contribute therefore to the establishment of a physical barrier against fungal infection.

However, while involvement of antioxidant metabolites in resistance mechanism to Fusarium spp. has been highly suggested, this involvement is far from being elucidated. One major challenge for the coming years will be to obtain conclusive proofs. Even though the genetic architecture underlying the synthesis and regulation of secondary metabolites in cereals is extremely complex, pieces of evidence can certainly come from extensive genetic and functional genomic studies.

### AUTHOR CONTRIBUTIONS

VA and FR have made substantial, direct and intellectual contribution to the work. CB has made intellectual contribution. All authors approved this work for publication.

#### ACKNOWLEDGMENTS

We greatly acknowledge Dr N. Ponts for her valuable comments.


affects FUM gene expression and fumonisin production. Eukaryot. Cell. 6, 1210–1218. doi: 10.1128/EC.00400-06


through suppression of Tri6 expression. Int. J. Food Microbiol. 214, 123–128. doi: 10.1016/j.ijfoodmicro.2015.07.014


an anaerobic fungus isolated from wild nil gai (Baselophus tragocamelus). Lett. Appl. Microbiol. 36, 377–381. doi: 10.1046/j.1472-765X.2003.01331.x


Gibberella stalk rot (Fusarium graminearum Schwabe). J. Agric. Food Chem. 55, 5186–5193. doi: 10.1021/jf070641e


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2016 Atanasova-Penichon, Barreau and Richard-Forget. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Fast and Accurate Microplate Method (Biolog MT2) for Detection of Fusarium Fungicides Resistance/Sensitivity

#### Magdalena Fr ˛ac\*, Agata Gryta, Karolina Oszust and Natalia Kotowicz

Laboratory of Molecular and Environmental Microbiology, Department of Soil and Plant System, Institute of Agrophysics, Polish Academy of Sciences, Lublin, Poland

The need for finding fungicides against Fusarium is a key step in the chemical plant protection and using appropriate chemical agents. Existing, conventional methods of evaluation of Fusarium isolates resistance to fungicides are costly, time-consuming and potentially environmentally harmful due to usage of high amounts of potentially toxic chemicals. Therefore, the development of fast, accurate and effective detection methods for Fusarium resistance to fungicides is urgently required. MT2 microplates (BiologTM) method is traditionally used for bacteria identification and the evaluation of their ability to utilize different carbon substrates. However, to the best of our knowledge, there is no reports concerning the use of this technical tool to determine fungicides resistance of the Fusarium isolates. For this reason, the objectives of this study are to develop a fast method for Fusarium resistance to fungicides detection and to validate the effectiveness approach between both traditional hole-plate and MT2 microplates assays. In presented study MT2 microplate-based assay was evaluated for potential use as an alternative resistance detection method. This was carried out using three commercially available fungicides, containing following active substances: triazoles (tebuconazole), benzimidazoles (carbendazim) and strobilurins (azoxystrobin), in six concentrations (0, 0.0005, 0.005, 0.05, 0.1, 0.2%), for nine selected Fusarium isolates. In this study, the particular concentrations of each fungicides was loaded into MT2 microplate wells. The wells were inoculated with the Fusarium mycelium suspended in PM4-IF inoculating fluid. Before inoculation the suspension was standardized for each isolates into 75% of transmittance. Traditional hole-plate method was used as a control assay. The fungicides concentrations in control method were the following: 0, 0.0005, 0.005, 0.05, 0.5, 1, 2, 5, 10, 25, and 50%. Strong relationships between MT2 microplate and traditional hole-plate methods were observed regarding to the detection of Fusarium resistance to various fungicides and their concentrations. The tebuconazole was most potent, providing increased efficiency in the growth inhibition of all tested isolates. Almost all among tested isolates were resistant to azoxystrobin-based fungicide. Overall, the MT2 microplates method was effective and timesaving, alternative method for determining Fusarium resistance/sensitivity to fungicides, compering to traditional hole-plate approach.

Keywords: Fusarium fungicides resistance, MT2 microplates, hole-plate method, chemical sensitivity, fungicides, phenomic profiles

#### Edited by:

Daniela Gwiazdowska, Poznan University of Economics, Poland

#### Reviewed by:

Emil Rekanovic, Institute of Pesticides and Environmental Protection, Serbia Jadwiga Wyszkowska, University of Warmia and Mazury in Olsztyn, Poland

> \*Correspondence: Magdalena Fr ˛ac m.frac@ipan.lublin.pl

#### Specialty section:

This article was submitted to Fungi and Their Interactions, a section of the journal Frontiers in Microbiology

Received: 15 February 2016 Accepted: 24 March 2016 Published: 06 April 2016

#### Citation:

Fra˛c M, Gryta A, Oszust K and Kotowicz N (2016) Fast and Accurate Microplate Method (Biolog MT2) for Detection of Fusarium Fungicides Resistance/Sensitivity. Front. Microbiol. 7:489. doi: 10.3389/fmicb.2016.00489

## INTRODUCTION

fmicb-07-00489 April 6, 2016 Time: 15:20 # 2

The genus Fusarium contains over 70 cosmopolitan species, occurring in natural conditions in different regions of the world. They are common in soil, as saprophytes, but they can also grow on plant residues and other organic substances (Kotowicz et al., 2014). Fusarium species may cause plant diseases of both parts underground and aboveground (Leslie and Summerell, 2006). The application of fungicides plays an important role in Fusarium diseases management (Amarasinghe et al., 2013). Fungicides application is one of the available measurement that may indirectly contribute reduce the risk of pathogen infection, however, it does not guarantee getting rid of the pathogen (Baturo-Cie´sniewska et al., 2011).

In general, sterole biosynthesis inhibitors, which includes triazoles fungicides are reported to be the most effective chemicals against Fusarium ssp. (Pirgozliev et al., 2002; Amarasinghe et al., 2013). The increasing usage of triazoles in Europe and Asia leads to selection of less sensitive Fusarium isolates (Klix et al., 2007; Yin et al., 2009; Becher et al., 2010), which is considered as a key determinant of the relatively low fungicide efficacy in the field conditions (Spolti et al., 2012). The strobilurins can cause increase of mycotoxin deoxynivalenol accumulation (Haidukowski et al., 2005). However, the carbendazim resistance of Fusarium can cause diseases even after 2–3 years from first fungicide application (Zhou and Wang, 2001). Therefore, the efficacy of fungicide use for the control of Fusarium diseases and mycotoxins production varies from being highly effective to even increasing mycotoxins levels (Mullenborn et al., 2008). Strobilurin fungicides block electron flow through one of the protein complexes, and disrupt energy supply. Several pathogens have quickly developed resistance to strobilurins. The reason of this resistance can be an alternative oxidase (AOX) which is a strobilurin-insensitive terminal oxidase that allows electrons from ubiquinol to bypass appropriate protein complex. Its synthesis is constitutive in some fungi but in many others is induced by inhibition of the main pathway. Salicylhydroxamic acid (SHAM) is a characteristic inhibitor of AOX, and several studies have explored the potentiation of strobilurin activity by SHAM (Wood and Hollomon, 2003).

The need for finding fungicides against Fusarium and monitor of their sensitivity is a key step in the chemical plant protection and using appropriate chemical agents by farmers. For these reasons the rapid, accurate method for evaluation of Fusarium fungicides resistance in monitoring study is critical. Existing conventional methods of qualitative and quantitative evaluation of fungal isolates resistance to fungicides are costly, timeconsuming and even more they could be environmentally harmful due to usage of high amounts of potentially toxic chemicals (Arikan, 2007; Rekanovic et al., 2010 ´ ; Panek et al., 2016). Therefore, the development of fast, accurate and effective Fusarium fungicides resistance detection method is urgently required. MT2 microplates method is traditionally used for bacteria identification and the evaluation of their ability to utilize different carbon substrates (Kadali et al., 2012). This system was also successfully used as a method in fast screening of native lignocellulosic-straw-degrading bacteria (Taha et al., 2015), identifying of bacteria, which are able to decompose hydrocarbon fractions (Kadali et al., 2012) and metabolize microcystin-LR (Manage et al., 2009). However, to the best of our knowledge, there are no reports concerning the use of this technique to determine fungicides resistance of the Fusarium isolates. Because the wells of the MT2 microplate do not contain any carbon sources, they can be loaded by any set of carbon sources or their mixtures into the wells. Therefore the MT2 microplates can be designed and optimized as approach for different applications, including study of fungi. Additionally, each well already contains the buffered nutrient medium and the tetrazolium chemistry suitable for growth of wide range of microorganisms (Bochner, 1989). In recent years, there has been exploration of several alternative nutrient sources and nutrient media for fungal culture (Basu et al., 2015). Therefore, the nutrient medium in MT2 plates can be also used for fungal growth after implementation with other components and inoculating fluids. Tetrazolium violet is used as a redox dye to colorimetrically indicate utilization of the carbon sources. However, the reading can also include optical density measurement without color development analysis, which can be used for fungal growth intensity evaluation instead of colorimetric measurement, allowing optimizing MT2 microplates for these organisms.

Hence, the objectives of this study were to: (1) develop a fast Fusarium fungicides resistance detection method; (2) compare the approach between both traditional hole-plate method and MT2 microplate-based assay; (3) evaluate the efficacy of three different fungicides against nine Fusarium isolates; (4) determine relationship between Fusarium fungicide sensitivity and metabolic profile of the tested isolates.

#### MATERIALS AND METHODS

#### Fungal Strains and Growth Conditions

Nine Fusarium strains (G15/14, G17/14, G18/14, G21/14, G22/14, G24/14, G25/14, G27/14, and G29/14) were selected among fungal collection of Laboratory of Molecular and Environmental Microbiology, Institute of Agrophysics Polish Academy of Sciences (Lublin, Poland). Fungal strains were isolated from soil (G15/14, G21/14, G22/14, G25/14, G27/14) and wheat (G17/14, G18/14, G24/14, G29/14) in eastern Poland (N 50◦ 590 , E 23◦ 080 ). Fungal strains were isolated from soil by selective Bengal Rose medium (Biocorp, Poland) and antibiotics (chlortetracycline – 2 mg dm−<sup>3</sup> and streptomycin – 30 mg dm−<sup>3</sup> ) using dilution plate method. To isolate the fungi from wheat (Muszelka var.) the small pieces of plants were placed on a potato dextrose agar (PDA, Biocorp, Poland) in Petri dishes. The plates were then incubated for 10 days at 27◦C. Fungal colonies were purified by repeated sub-culturing on PDA medium. The fungal isolates were identified as Fusarium sp. based on macroscopic and microscopic observations on PDA and Spezieller Nährstoffarmer Agar (SNA) media (Leslie and Summerell, 2006). The fungi were further cultured on PDA.

FIGURE 1 | The optimization results of inoculating fluids (IF) composition, fungicides concentrations and incubation time on the susceptibility of Fusarium using MT2 microplates method.

### Fungicides

The three different commercially available fungicides, qualified for three different groups of fungicides: triazoles, benzimidazoles and strobilurins were tested in the study. Each group of tested fungicides contained different substances as active compounds: tebuconazole, carbendazim, and azoxystrobin, respectively. The fungicides were based on three different mode of action affecting the fungi by: inhibiting specific enzymes in fungi that play a role in production of ergosterol necessary for the development of fungal cell walls, inhibiting mitosis and cell division and inhibiting respiratory processes of fungi, respectively.

## The Optimization of MT2 Microplates Approach

Traditionally MT2 microplates (BiologTM) were used for the evaluation and identification of bacterial species based on their ability to utilize a range of different carbon sources (Kadali et al., 2012). In this study, the MT2 microplates were

used as an alternative method for rapid assessment of the sensitivity of fungi to different fungicides. The appropriate composition of inoculating fluid for testing the sensitivity was selected. Three different types of inoculating fluids were tested: PM4-IF, PM9-IF and FF-IF. The inoculating fluids had the following composition: PM4-IF – Tween 40, Phytagel and D-glucose, PM9-IF – Tween 40, Phytagel, D-glucose and additive solution containing yeast nitrogen base and FF-IF – Tween 40 and Phytagel. The study included the optimization of the amount of mycelium and each fungicide added into the wells. The optimization step included the selection of appropriate concentrations of fungicides. The following four concentrations of each fungicides: 0.0005, 0.005, 0.05, 0.5% were tested. Aliquots of the individual fungicide suspension were loaded into the wells of the MT2 microplates. Two doses of fungicide 50 and 100 µl were tested while optimization. The wells were then inoculated with the resuspended fungal mycelium using 100 and 50 µl, respectively. To optimize method the Fusarium isolate G18/14 was selected. The optimization was prepared in triplicates. Appropriate positive and negative controls were provided for each isolate and each fungicide. The microplates were incubated at 27◦C for 12 days and the optical density was monitored at 750 nm by every day readings using a microstation reader (BiologTM).

## Fusarium Fungicides Resistance/Sensitivity Evaluation by MT2 Microplates Method

The assessment of Fusarium fungi resistance/sensitivity on selected fungicides was carried out using an alternative optimized method of MT2 microplates. The fungicides were tested at six concentrations: 0, 0.0005, 0.005, 0.05, 0.1, and 0.2%. The study included nine isolates of Fusarium described at Fungal strains and growth conditions section. The tested concentrations of each fungicide were applied in triplicate to the wells of MT2 microplates. Subsequently, the wells were inoculated with the mycelium of Fusarium suspended in PM4-IF inoculating fluid with glucose addition. To prepare the inoculum for MT2 microplates, the mycelial cells were harvested by loops from the surface of agar, homogenized and suspended in sterile inoculating fluid. Then, the suspension of each isolate was standardized into 75% of transmittance using turbidimeter (BiologTM). Appropriate controls were set up for each isolate by loading the mycelium suspension into the wells without any fungicide and loading sterilized water instead. MT2 microplates were incubated at 27◦C for 11 days. The optical density was monitored at 750 nm every day using a microstation reader (BiologTM).

## Fusarium Fungicides Resistance/Sensitivity Evaluation by Hole-Plate Method

To verify the results of Fusarium fungicide resistance/sensitivity performed by MT2 microplates a hole-plate method based on zone of growth inhibition analysis was conducted. The fungicides were suspended in water as the following concentrations: 0, 0.0005, 0.005, 0.05, 0.5, 1, 2, 5, 10, 25, and 50% and afterward the fungicide sensitivity of the isolates were determined by assessing diameter of growth inhibition zones. Fusarium isolates (10-dayold cultures) were inoculated on 90 mm Petri dishes with PDA. Then, 8 mm diameter holes were cut in triplicates in PDA medium of each plate. After that, 100 µl of tested fungicides in each concentrations were added to individual holes. The study for each combination isolate-dose-fungicide were conducted in three replications. The results, as a diameter of growth inhibition zone, were collected after 1, 2, 3, and 4 days of incubation at 27◦C.

## The Catabolic Profile of Fusarium Fungi Using FF Microplates

The utilization of particular substrates by each of the Fusarium isolates based on 95 low molecular weight carbon sources were assessed using the FF microplates (BiologTM). The inoculation procedure was based on the original FF microplate (BiologTM) method according to manufacturer's protocol modified by Fr ˛ac (2012) and Janusz et al. (2015). To prepare inoculum, mycelia of each isolate were obtained by cultivation on PDA in the dark at 27◦C through 10 days. After the suspension of the mycelium in inoculating fluid (FF-IF, BiologTM) homogenization the transmittance was adjusted to 75% using a turbidimeter (BiologTM). Hundred microliter of the above-mentioned mycelium suspension was added to each well and the inoculated microplates were incubated at 27◦C through 10 days. The optical density at 750 nm was determined using a microplate reader (BiologTM) every day, in triplicates. Functional diversity was determined by the number of different substrates utilized by the individual isolates and expressed as substrate richness (R). Phenotype profiles of Fusarium isolates were generated from FF microplates based on the growth intensity of the organism on particular substrates. Dendrogram was performed to show the correlation between the Fusarium isolates in relation to utilization of C-sources from the FF microplates.

#### Statistical Analyses

Analysis of variance (ANOVA) was used to determine the differences in inhibiting effect of individual fungicides and their concentrations on the fungal isolates and in substrate richness of particular isolates. Post hoc analyses were performed using a Tukey test (HSD). The data were presented as 95% confidence intervals. Statistical significance was established at p < 0.05. Cluster analysis for substrate utilization, metabolic profiles and tested Fusarium isolates was used to detect groups in the data set and was made with Euclidian distance using Ward method

FIGURE 5 | The effect of three fungicides on the selected Fusarium isolates growth during eleven incubation days by MT2 microplates method.

approach. All statistical analysis was performed using Statistica software (version 10.0).

## RESULTS

## The Optimization of MT2 Microplates Approach

To optimize MT2 method for evaluation Fusarium susceptibility to fungicides the various composition of inoculating fluids, different concentrations of fungicides and the ratio of mycelium to fungicide were tested. The assay was performed by addition of different concentrations of fungicides suspended in particular inoculating fluids into the wells of MT2 microplates. Further, the inoculum of Fusarium mycelia was added into each well and mixed by pipetting. Then, microplates were incubated for 11 days in the dark, at 27◦C and optical density was measured every day. The trend in Fusarium growth during incubation time is presented at **Figure 1** and was mostly depended on used inoculating fluid.

To select a suitable inoculating fluid to evaluate fungal fungicides sensitivity/resistance, including the PM4-IF with glucose, PM9-IF with glucose and supplements and FF-IF known as standard inoculating fluid for filamentous fungi analysis were tested. Fungal growth was observed in all tested inoculating fluids, but in PM4-IF and PM9-IF was more intensive and significantly improved compared to FF-IF fluid (**Figure 2**). Moreover, in both inoculating fluids (PM4 and PM9) the fungal growth intensity in control wells (without fungicide addition) was significantly higher than in the wells with fungicides. Considering the fungal growth in FF-IF inoculating fluid, the growth intensity was very weak and almost at the same level for control wells with no fungicide addition and all tested fungicides concentrations (**Figure 1**). The highest differentiation between applied fungicides concentrations was found for the PM4-IF inoculating fluid, followed by the PM9-IF, whereas the lowest differentiation was obtained for the FF-inoculating fluid, suggesting significant influence of the inoculating fluid type on fungal growth, especially in stress conditions caused by fungicides presence. For this reason, the PM4-IF was selected after optimization for further study. The observed differences in fungal growth can be due to the inoculating fluid composition, but the transparency of fungicides (related to their color and consistency) might also have been responsible for observed artifacts, as the highest fungicides concentration (0.5%) was tested (**Figures 1** and **3**). If the dissolutions of fungicides causes optical density artifacts and to avoid such artifacts, those concentrations were omitted. Detailed the following lower

fmicb-07-00489 April 6, 2016 Time: 15:20 # 6

FIGURE 6 | The relationship between type of fungicide, doses of fungicide, action time of fungicide and Fusarium susceptibility expressed as growth intensity in MT2 microplates method.

concentrations were tested 0.1 and 0.2% (results described below).

The ratio between mycelium and fungicide volume in the microplates wells should also be defined in order to establish standard and controlled experimental conditions for assay. To select the best ratio of mycelial inoculum and fungicide the two schemes were used as follows: 50 µl of mycelium with 100 µl of fungicide and 100 µl of mycelium with 50 µl of fungicide. Growth inhibition became smaller with the increase of inoculum size, however, this effect was observed only for the lowest concentration of fungicide (0.0005%). Furthermore, the addition of 100 µl of mycelium suspension caused linear growth inhibition with increasing fungicides concentration (**Figure 3**). This effect was not observed for the ratio of 50 and 100 µl of mycelium and fungicide, respectively. Although the addition of 100 µl fungicides was twofold higher compared with the amount of 50 µl, the inhibition of fungal growth with increasing doses of fungicides was not clearly defined in this conditions, making difficult its precise measurement due to insufficient transparency of the mixture (**Figure 3**). For this reason, to properly determine the Fusarium growth, further approaches were decided to be performed with suspension containing a higher volume of mycelium (100 µl) and lower volume of fungicide (50 µl).

According to the results obtained, the Fusarium fungicide sensitivity/resistance assay was defined in PM4-IF inoculating fluid, using lower concentrations of fungicides (0.0005, 0.005, 0.05, 0.1, 0.2%) and the following 100:50 volume ratio of mycelium and fungicide. In order to assess and validate the usefulness and adequacy of the bioassay, the three fungicides (triazole, benzimidazole, and strobilurins) were tested against nine Fusarium isolates. The method was also validated in comparison to the traditional hole-plate zone of growth inhibition agar diffusion method.

## The Effect of Fungicides on Growth of Fusarium Strains Using MT2 Microplates Method

Among nine Fusarium isolates tested for sensitivity to three fungicides groups seven isolates showed high, one medium and one law resistance to strobilurins (azoxystrobin), five medium resistance to benzimidazole (carbendazim) and most of isolates were sensitive to triazole (tebuconazole) (**Figure 4**). The Fusarium susceptibility was dependent not only of isolates and used fungicide but also of the exposure time to toxic compound which is presented at **Figure 5**. The resistance to azoxystrobins could be connected with phenomena of AOX

existing which was thoroughly discussed by Wood and Hollomon (2003). The most sensitive isolateto triazole- and benzimidazolebased fungicides was G21/14, whereas to strobilurins G27/14 (both soil isolates). The fungicide concentration, which inhibits mycelial growth by 50% relative to growth in unamended medium (EC50) was determined based on all readings of tested Fusarium isolates in particular days of analyses and presented as fungal growth intensity at **Figure 6**. The Fusarium growth inhibiting concentration of fungicides evaluated by MT2 microplates method was dependent on tested fungicide and their transparency, which might have been responsible for artifacts in measurements as false positive fungal growth. This effect was observed for benzimidazole-based fungicide applied at 0.1 and 0.2%. Although, all tested concentrations of tebuconazole caused fungal growth inhibition (determined as EC50), the Fusarium isolates were capable of good growth at 0.0005% tebuconazole concentration, and were severely inhibited by higher fungicide content (from 0.005 to 0.2%). The lowest tebuconazole concentration (0.0005%) inhibited fungal growth comparing to the control without fungicide, but with the increasing action time of tebuconazole, the fungi became resistant, which was expressed as increase of their growth. In the case of carbendazim four out of tested fungicide concentrations (0.0005, 0.005, 0.05, 0.1%) showed fungal growth inhibition. However, among above mentioned fungicide doses, the 0.1%

concentration had the lowest value of EC50 which was connected with artifacts caused by low transparency of benzimidazole-based fungicide. The investigated Fusarium isolates demonstrated ability to tolerance azoxystrobin at all tested concentrations compare to the control without fungicide addition during 11 days of incubation (**Figure 6**).

Basing on the MT2 microplates method results, the highest fungal resistance was found for the azoxystrobin-based fungicide. The Fusarium isolates showed moderate resistance to carbendazim- and were sensitive to tebuconazole-based fungicides.

### The Effect of Fungicides on Growth of Fusarium Strains Using Hole-Plate Method

Results from the conventional assay using hole-plate method showed that fungal growth of tested Fusarium isolates was strongly inhibited by azoxystrobin-based fungicide. However, the fungitoxicity of carbendazim and tebuconazole-based fungicides, expressed as diameter of fungal growth inhibition zone, was moderate and low, respectively (**Figure 7**). The inhibition zone diameter decreased during incubation time, indicating that all tested isolates were more or less able to grow in the presence of fungicide (**Figure 8**). The Fusarium resistance to

azoxystrobin was observed at the same level for all tested isolates. The results indicated the differentiation of Fusarium sensitivity to tebuconazole and carbendazim depended on fungal isolate and time exposure to fungicide. The inhibition zone varied considerably among the isolates under the influence of triazole and benzimidazole-based fungicides during incubation time and no significant differences for growth inhibition was detected between isolates in the strobilurins presence. The Fusarium G17/14 isolated from wheat, showing the largest growth inhibition zones for both tebuconazole and carbendazim, differed significantly from all other isolates during all days of incubation. Both tebuconazole and carbendazim reduced fungal growth expressed as higher diameter of inhibition zone but tebuconazole provided the best fungal control (**Figures 9** and **10**). When tebuconazole was used the lowest (0.0005%) or above fungicide concentrations were defined as EC50 causing 50% increase of growth inhibition zone relative to control with water instead of fungicide addition. EC50 concentrations were 0.005% or above for carbendazim. Nevertheless, any of tested fungicides concentrations, even the highest (50%), cannot be defined as EC50 for azoxystrobin. With respect to sensitivity of Fusarium to fungicides tested in this work, we observed decrease of growth inhibition zone with increasing action time of fungicide for all concentrations, indicating that isolates adapted to fungicide exposition which can lead to development of fungal resistance to fungicides. The traditional hole-plate method can be used for testing both low and high fungicides concentrations.

The results indicated that similarly as in the MT2 microplates method the highest fungal resistance by hole-plate method was found for the azoxystrobin-based fungicide and moderate resistance for carbendazim. The Fusarium isolates were sensitive to tebuconazole-based fungicides.

### The Catabolic Profile of Fusarium Fungi Using FF Microplates

The application of the FF microplates allowed comparing the functional diversity of the nine Fusarium isolates. The substrate utilization abilities for the isolates tested revealed variability, indicating significant differences (up to 16 carbon sources) in the substrate richness values (**Figure 11**). However, all Fusarium isolates showed high catabolic activity, utilizing more than 50% of tested substrates. The highest capabilities

to decompose of carbon sources were found for the following isolates G25/14, G24/14, G18/14, G21/14, and G29/14, which utilized 71, 70, 68, 69 and 69 out of 95 tested substrates, respectively. G17/14, G15/14, G22/14, G27/14 isolates were able to assimilate 61, 60, 59 and 55 out of 95 tested carbon sources, respectively (**Figure 11**). To determine the type of substrates that Fusarium isolates most utilized the optical density values indicating mycelial growth on the different carbon sources were subjected to cluster analysis and presented as heat map, according to the resulting growth (**Figure 12**). A detailed comparison of the carbon sources utilized by the Fusarium isolates revealed the most intensive differences in utilization of amino acids, carboxylic acids and carbohydrates. Among tested isolates three (G17/14, G22/14, G27/14) were able to utilized almost all among 95 tested carbon sources (**Figure 12**). Remaining isolates (G15/14, G18/14, G21/14, G24/14, G25/14, G29/14) in comparison to above-mentioned isolates were not capable of good growth at N-acetyl-D-galactosamine, N-acetyl-Dmannosamine, sedoheptulosan, N-acetyl-L-glutamic acid, betacyclodextrin, L-fucose, maltose and maltotriose. However, all isolates grew much better on amino acids and carbohydrates (L-alanine, L-ornityne, L-proline, L-phenylalanine, D-trehalose, turanose, D-sorbitol) than amides/amines and carboxylic acids (alaninamide, L-lactic acid, succinic acid). All isolates were unable to use glucuronamide and sebacic acid, although they utilized other substrates included into the groups of amines/amides and carboxylic acids (**Figure 12**). In cluster analysis two main distinct cluster groups were revealed. They included isolates which were capable of very intensive growth on most tested carbon sources (Cluster I) and isolates which utilized smaller number of tested substrates with less efficiency (Cluster II) (**Figure 13**). Cluster I contained two isolates from soil (G27/14, G22/14) grouped together and one isolate from plant (G17/14) as a separate branch. The clustered isolates were more sensitive to all tested fungicides than the rest of Fusarium strains. Cluster II assembled isolates, which also utilized large number of tested substrates but the intensity of their assimilation was significantly lower. This cluster was subdivided into two groups (IIA, IIB), of which subcluster IIA contained two the most sensitive isolates to tested fungicides (G18/14, G21/14) and one rather fungicides resistant isolate, as a separate branch. Subcluster IIB grouped two isolates which were able to cause the decomposition of

FIGURE 10 | Growth of selected Fusarium isolate on PDA with addition of various fungicides at different doses by hole plate method. Explanations: in columns – fungicide type: (1) strobilurins – azoxystrobin; (2) benzimidazole – carbendazim; (3) triazole – tebuconazole; in rows fungicide concentrations: (A) 0%; (B) 0.0005%; (C) 0.005%; (D) 0.05%; (E) 0.5%; (F) 1%; (G) 2%; (H) 5%; (I) 10%; (J) 25%; (K) 50%.

dextrin and were not capable of good growth on D-xylose and one isolate decomposing D-xylose and not growing on dextrin, as a separate branch. Biolog FF analysis of the nine Fusarium isolates identified possible phenetic differences between the strains isolated from soils and plants with different fungicides sensitivity.

### DISCUSSION

The monitoring of Fusarium with respect to fungicides resistance and sensitivity is important due to high quality food production. Traditional methods of qualitative and quantitative evaluation of Fusarium isolates resistance to fungicides are costly, time-consuming and could be harmful for environment due to usage of high amounts of potentially toxic chemicals (Pereira et al., 2013; Fr ˛ac et al., 2015). Therefore in this study approach to accurate and effective Fusarium fungicides resistance detection was designed and optimized using MT2 microplates (BiologTM). MT2 microplates method has been described as screening and evaluation tool for identification and metabolic characterization of bacteria strains (Kadali et al., 2012; Taha et al., 2015). However, MT2 microplates method for fungal sensitivity have never been used before. This is the first study done using MT2 microplates (BiologTM) to compare the efficacy of fungicides in controlling Fusarium isolates growth. The results of this study suggest that new MT2 microplates method can be used to determine Fusarium fungicides sensitivity/resistant efficiently, especially when low fungicides concentrations are tested. Using this method, Fusarium sensitivity to fungicides was best detected in PM4-IF inoculating fluid, for lower

fungicides concentration and in the mixture 100 and 50 µl of inoculum and fungicide, respectively. The main features of optimized method are simplicity, cost efficiency and data reproducibility.

The efficacy of fungicides may change based on the level of resistance connected with the interactions between fungicides, cultivars and isolates (Amarasinghe et al., 2013). In our study, significant differences were found between the tested fungicides containing tebuconazole, carbendazim, or azoxystrobin, their concentrations and selected Fusarium isolates. According to our study the Fusarium isolates were highly sensitive to tebuconazolebased fungicide, which was indicated by both used methods (MT2 microplates, hole-plate agar). On the one hand, the obtained results are in agreement with other reports (Homdork et al., 2000; Matthies and Buchenauer, 2000; Amarasinghe et al., 2013), confirming the effectiveness of triazole fungicides in controlling Fusarium caused diseases. The results of Avozani et al. (2014) also indicated that F. graminearum isolates were susceptible to tebuconazole. On the other hand, triazole fungicides have been used as long as 30 years for controling fungal pathogens, therefore some of active substances have not been effective and emerging due to the lack of anti-resistance strategy (Spolti et al., 2014). A gradual reduction in Fusarium sensitivity to tebuconazole was observed in Germany. This effect concerned the isolates exposed to increasing fungicide doses (Klix et al., 2007). Tebuconazole-resistant F. graminearum isolates were also detected in China (Yin et al., 2009) and in United States (Spolti et al., 2014). Biological and agronomic factors such as the high metabolic and genetic diversity of the Fusarium pathogens and increasing use of triazole fungicides may pose a risk of selection of isolates that are both resistant to triazole fungicides and produce large amount of dangerous mycotoxins (Spolti et al., 2014). Contrary, in our study tebuconazole was the most effective among all tested fungicides. The disagreement between particular studies with our results may be due to differences in time exposure to fungicides and their concentrations. Moreover, when sublethal fungicides concentrations are used for long time the fungi can adapt to stress conditions and they become resistant to fungicides (Becher et al., 2010). Thus, the selection of fungicides concentrations plays a key role in the fungicides resistance development and mycotoxins production. In our study, a strong relationship between fungicides concentration and effectiveness of fungicides was observed. Further study on monitoring of Fusarium isolates resistant are crucial in order to better understand the risk for plant diseases and food security connected with fungi, which are less sensitive to triazole-based fungicides.

Although according to Avozani et al. (2014) azoxystrobin was the most potent fungicide to inhibit spore germination of five F. graminearum isolates, the results of our study showed high resistance to the fungicide containing this substance. This effect was confirmed by both new MT2 microplates

and traditional hole-plate methods. Limitation of strobilurin effectiveness can be caused by AOX once an infection is established, but AOX is unable to interfere significantly with strobilurin action during germination (Wood and Hollomon, 2003). Audenaert et al. (2010) reported that sublethal doses of fungicides can induce hydrogen peroxide as catalyser of toxin deoxynivalenol biosynthesis and can cause the development of fungal resistance and adaptation to the fungicide influence (Becher et al., 2010). For these reasons some fungicides treatment can cause increase of mycotoxins, which was found after strobilurin-based fungicides application (Pirgozliev et al., 2002). However, to date there has not been sufficient evidence to explain the role of fungicides in increasing mycotoxins in infected plants (Amarasinghe et al., 2013). Our study indicated that most of Fusarium isolates were resistant to azoxystrobin, which represent the group of strobilurin-based fungicides.

In this work we addressed the potential of new MT2 microplates method in Fusarium fungicides sensitivity assessment and studied the catabolic profiles of tested isolates. Further studies are needed to evaluate an adaptation of Fusarium isolates to fungicides exposure by changes of carbon sources utilization and phenomic profiles. Becher et al. (2010) indicated that exposure to tebuconazole resulted in the emergence of two morphologically separate phenotypes differing in the degree of tebuconazole resistance, fitness, virulence and mycotoxins production. Due to the prelevance of fungicides in the control of Fusarium pathogens it is important to assess and monitor the isolates fungicides sensitivity and investigate the mechanisms leading to fungicides resistance.

#### CONCLUSION

This study demonstrated that the MT2 microplates (BiologTM) method developed here represents a cheap (less materials are used), environmentally friendly (significant reduction of amount of used fungicides, compering to hole-plate method), not as much time-consuming as for hole-plate approach) and effective tool for the rapid fungicides sensitivity/resistance assessment of Fusarium isolates. Results from traditional hole-plate method based on zone of growth inhibition and MT2 microplates method were summarized. The substantial relationship between results obtained using both methods was found, what clearly indicate that MT2 microplates method might be used successfully vice traditional hole-plate technique. Comparing the results of the traditional hole-plate and MT2 microplates methods we found strong relationship in whole Fusarium fungicide resistance, although there were some differences with respect to individual fungal isolates.

All tested fungicides showed toxicity for mycelium growth, but this effect was dependent on type, concentration and action time of fungicide and Fusarium isolate. The tebuconazole was most potent, providing increased efficiency in the growth inhibition of all tested isolates. Nevertheless, the isolates showed different sensitivity to the fungicide concentration. The carbendazim caused Fusarium growth inhibition, but some of isolates were resistant to this fungicide. Almost all among tested isolates were resistant to azoxystrobin-based fungicide.

The sensitivity determined in this study constitutes a critical step for monitoring changes in metabolic profile and identification of risk factors related to increase or decrease in particular carbon sources utilization intensity leading to decline in sensitivity.

#### REFERENCES


#### AUTHOR CONTRIBUTIONS

Conceived and designed the experiments: MF, AG, KO, NK. Performed the experiments: MF, AG, KO, NK. Analyzed the data: MF, AG, KO, NK. Contributed reagents/materials/analysis tools: MF, AG, KO, NK. Wrote the paper: MF, AG, KO, NK.

#### ACKNOWLEDGMENTS

The studies were performed using equipment bought with European Union funds – The Eastern Poland Development Programme 2007–2013, (Regional Laboratory of Renewable Energy), IA PAS.

strains. World J. Microbiol. Biotechnol. 31, 121–133. doi: 10.1007/s11274-014- 1769-y


from New York wheat and competitiveness of a tebuconazole-resistant isolate. Plant Dis. 98, 607–613. doi: 10.1094/PDIS-10-13-1051-RE


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2016 Fr ˛ac, Gryta, Oszust and Kotowicz. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Biodegradation of Mycotoxins: Tales from Known and Unexplored Worlds

#### Ilse Vanhoutte, Kris Audenaert and Leen De Gelder\*

*Department of Applied BioSciences, Faculty Bioscience Engineering, Ghent University, Ghent, Belgium*

Keywords: detoxification, microorganisms, mycotoxins, biodegradation, metabolite

Exposure to mycotoxins, secondary metabolites produced by fungi, may infer serious risks for animal and human health and lead to economic losses. Several approaches to reduce these mycotoxins have been investigated such as chemical removal, physical binding, or microbial degradation. This review focuses on the microbial degradation or transformation of mycotoxins, with specific attention to the actual detoxification mechanisms of the mother compound. Furthermore, based on the similarities in chemical structure between groups of mycotoxins and environmentally recalcitrant compounds, known biodegradation pathways and degrading organisms which hold promise for the degradation of mycotoxins are presented.

Edited by:

*Daniela Gwiazdowska, Poznan University of Economics, Poland*

#### Reviewed by:

*Augusto Schrank, Federal University of Rio Grande do Sul, Brazil Ozgur Bayram, Maynooth University, National University of Ireland Maynooth, Ireland*

> \*Correspondence: *Leen De Gelder leen.degelder@ugent.be*

#### Specialty section:

*This article was submitted to Fungi and Their Interactions, a section of the journal Frontiers in Microbiology*

Received: *29 January 2016* Accepted: *04 April 2016* Published: *25 April 2016*

#### Citation:

*Vanhoutte I, Audenaert K and De Gelder L (2016) Biodegradation of Mycotoxins: Tales from Known and Unexplored Worlds. Front. Microbiol. 7:561. doi: 10.3389/fmicb.2016.00561*

## INTRODUCTION

The presence of mycotoxins is inherent to many food and feed products worldwide (Bhat et al., 2010; Marroquín-Cardona et al., 2014). Hallmarks of their presence and their impact on animal and human health are encountered throughout history. Ergotism, also known as "St. Anthony's fire" occurred in several areas in Europe during the tenth century (Schiff, 2006) and was caused by the consumption of rye containing ergot alkaloids, produced by the fungus Claviceps purpurea (Bové, 1970; Beardall and Miller, 1994). In Siberia, a delayed harvest due to the second world war resulted in grains heavily contaminated with trichothecenes produced by Fusarium spp. People later consuming the grain were afflicted with number of nonspecific disorders and mortality mounted up to 10% (Manahan, 2002). In 1962, 100,000 turkeys died in London of Turkey X disease, linked to aflatoxins from Aspergillus flavus (Binder, 2007). These examples mentioned above illustrate the acute impact of high loads of singular mycotoxins on human and animal health. However, longtime exposure to low concentrations of mycotoxins also entail chronic toxicities which often result in non-specific symptoms, difficult to track-and-trace down to mycotoxins. These toxicities include estrogenic gastrointestinal, urogenital, vascular, kidney, and nervous disorders. Some mycotoxins are carcinogenic or immuno-compromising, and as such also promote the development of infectious diseases (Peraica et al., 1999; Hussein and Brasel, 2001; Creppy, 2002; Richard, 2007; Da Rocha et al., 2014).

For many years the research community focused on the occurrence of singular mycotoxins but nowadays scientific interest shifts to studies involving multiple mycotoxins. This new approach is highly relevant as large scale multi-toxin surveys show that a number of mycotoxins tend to cooccur with other sometimes structurally not-related mycotoxins (Gerding et al., 2014; Storm et al., 2014; Vanheule et al., 2014; and many more). In addition, mycotoxins are known to have additive and synergistic effects on human- and animal health (Alassane-Kpembi et al., 2013; Klaric et al., 2013; Clarke et al., 2014).

Research efforts progressively increase to develop mitigation strategies based on risk monitoring, risk characterization, prevention, intervention, and remediation strategies for multiple mycotoxins, which start from critical points along the production chain comprising field, storage, processing, and transportation. However, monitoring and good agricultural, storage, and transportation practices along with an effective Hazard Analysis and Critical Control Point approach do not completely prevent mycotoxin presence in the food or feed chain (Bhat et al., 2010). Decontamination technologies then offer a last resort to salvage contaminated batches along the production chain.

Decontamination strategies to reduce mycotoxins in foodand feed commodities are technologically diverse and based on physical, chemical, or biochemical principles. Some physical processes aim to remove highly contaminated fractions from bulk material (Bullerman and Bianchini, 2007; Cheli et al., 2013; Kaushik, 2015) through sorting (Scudamore et al., 2007), milling (Castells et al., 2007; Khatibi et al., 2014), dehulling (Fandohan et al., 2006; Rios et al., 2009; Matumba et al., 2015), cleaning (van der Westhuizen et al., 2011), heating, irradiation, or combinational approaches (Fandohan et al., 2005; Matumba et al., 2015). Another physical removal strategy is the use of inorganic or organic mycotoxin binders (Ramos et al., 1996; Kolosova and Stroka, 2011). Although these adsorbing binders have some promising features, some may have adverse nutritional effects due to binding of vitamins and minerals (Huwig et al., 2001; Yiannikouris et al., 2006) or reducing the efficacy pharmacokinetics of antibiotics (De Mil et al., 2015).

Chemical remediation strategies involve the conversion of mycotoxins via chemical reactions. Ammoniation (Norred et al., 1991), alkaline hydrolysis, peroxidation, ozonation, and the use of bisulphites are reported to be effective on one or more mycotoxins but a detailed insight into the toxicity of eventual end products or the impact on palatability and nutritive quality is questionable.

Microbial based methods comprise mycotoxin decomposition, transformation, or adsorption. The latter strategy has already been mentioned under physical measures and will not be considered in detail in this review. Focus in this review will be on transformation and biodegradation of the main mycotoxins by microorganisms. Although there are some excellent reviews on biodegradation (Zinedine et al., 2007; Wu et al., 2009; Awad et al., 2010; Jard et al., 2011; Devreese et al., 2013; McCormick, 2013; Hathout and Aly, 2014; Adebo et al., 2015), this review is timely because of two reasons:

Firstly, studies often wrongly identify biodegradation with detoxification, or do not test for toxicity of potential metabolites. Indeed, not all transformation or degradation products are detoxification products. This is nicely illustrated for aflatoxins and zearalenone (ZEN). Aflatoxin M1 (AFM1) is the hydroxylated metabolite of AFB1 and is categorized as possible carcinogenic to humans (Group 2B) by the International Agency for Research on Cancer (IARC; IARC, 2002). Aflatoxicol (or aflatoxin R0), a reduction product of AFB1, has been detected as degradation product by Corynebacterium rubrum, Aspergillus niger, Trichoderma viride, Mucor ambiguous, and Dactylium dendroides (Mann and Rehm, 1976; Wong and Hsieh, 1976). However, Karabulut et al. (2014) concluded that AFB1 and aflatoxicol have similar potency to form an exo-epoxide analog which can bind to DNA. Assessing the ZEN biodegradation capacity of several microorganisms, Hahn et al. (2015) found that many strains were able to convert ZEN to α- and/or β-ZEL, showing similar estrogenic activity compared to ZEN. Aerobic and anaerobic degradation to other uncharacterized metabolites with unidentified toxicity was obtained as well. These results demonstrate the importance of in vitro experiments to critically screen agents claiming mycotoxin detoxification.

Secondly, the available set of mycotoxin degrading microorganisms is limited and their performance is often doubtful when considering multiple mycotoxin degradation. This issue was also nicely illustrated by Hahn et al. (2015). Using an in vitro screening approach, 20 commercially available agents claiming mycotoxin detoxification were tested for their efficacy to inactivate and/or degrade the two structurally not related mycotoxins DON or ZEN. The majority of the agents were not effective or converted the toxins to equally toxic metabolites. Only one of the products efficiently inactivated or degraded the two considered mycotoxins under the tested conditions.

New insights on actual microbial detoxification routes are needed and can be based on known biodegradation metabolisms of non-mycotoxins found in diverse microbial communities, which we chose to identify as "unexplored worlds" to be discovered for the mycotoxin research field. Indeed, many hazardous, undesirable, deleterious, or recalcitrant molecules in other research fields share structural analogies with diverse mycotoxins and are reported to be successfully degraded by microorganisms. These unexplored worlds may serve as resource for cutting edge research in the field of mycotoxin remediation or in the field of metagenomics screening surveys in search for new microbial degraders of mycotoxins.

In this review, We are not only focusing on Fusarium mycotoxins, but also on Aspergillus, Penicillium, and other mycotoxins. This is relevant as independently of the producing genus, mycotoxins often share key-chemical groups responsible for their toxicity and thus biodegrading organisms for one mycotoxin can have their relevance for other mycotoxins produced by distinct fungal genera.

## TOXICITY AND DEGRADATION OF MYCOTOXINS

In order to assess detoxification by microorganisms, it is important to pinpoint the actual groups within the chemical structure of each mycotoxin which infer the toxic effects (**Table 1**). Next to the main toxic structural groups occurring in mycotoxins, structural similarities between mycotoxins are also highlighted; aflatoxins and ochratoxins are both composed of a coumarin moiety, whereas the main structure of aflatoxins, ZEN and ochratoxins is based on a lactone ring (**Table 1**— Red). Carboxyl derivatives (ester bonds), often playing a role in toxicity, are frequently present, as well in the lactone, as in side groups (**Table 1**—Red) (observed in fumonisins, ZEN, ochratoxins, and acylated trichothecenes). Each mycotoxin is

#### TABLE 1 | Chemical structural groups inferring toxicity in mycotoxins.


*Red, carboxylic derivatives (lactone rings and ester bonds) in red; Blue, specific groups responsible for toxicity.*

further characterized concerning specific groups responsible for its toxicity (**Table 1**—Blue).

From our perspective, there are two ways in which detoxification of the mother compound in a degradation study can be confirmed: (i) the confirmation of reduced toxicity after degradation through one or more actual toxicity assays on particular organisms or cell lines, this is the most convincing proof; (ii) the detection and identification of detoxification products, for which in independent literature has been shown that they confer an lower toxicity to the mother compound. Of course, a combination of both ways provides the most holistic approach. The decreased toxicity of the degradation metabolites listed in **Tables 2**–**7** can therefore be found in **Table 1** in Supplementary Material.

## Fumonisins

#### Toxicity

Fumonisins (most importantly FB1, FB2), first described by Gelderblom et al. (1988), are mainly produced by F. verticillioides and F. proliferatum and are structurally similar to sphingolipid long-chain bases such as sphinganine and sphingosine. This feature is tightly related to their toxicity mechanism through the inhibition of the sphingolipid biosynthesis (Merrill et al., 1993a; Soriano et al., 2005) and exposure to fumonisins

#### TABLE 2 | Degradation and/or detoxification products of fumonisins.


based on increased pH *Bacillus* sp. and yeast strain Camilo et al., 2000

\**, growth on fumonisin as sole carbon source.*

# *, growth on fumonisin as sole carbon and nitrogen source.*

has been associated a wide variety of diseases in animals as reviewed by Voss et al. (2007), such as liver cancer in rats, equine leukoencephalomalacia, and porcine pulmonary edema.

Specifically, fumonisins are comprised of a 22 carbon aminopentol with two tricarballylate (TCA) side groups, where two structural groups are important in their toxicity mechanism (**Table 1**). Firstly, the unsubstituted primary amino group at C2 competitively inhibits ceramide synthase, thereby disrupting the de novo biosynthesis of ceramide and sphingolipid metabolism (Voss et al., 2007). This free primary amino group of fumonisin-like compounds is a prerequisite for ceramide synthase inhibition, since N-acetylation of FB1 diminished or removed the toxicity effects in rat liver slices (Norred et al., 1997) and in jimsonweed and several mammalian cell lines (Abbas et al., 1993b). Secondly, the TCA side groups seem to have varying effects on the toxicity. On the one hand, absence of these side groups has been found to reduce both phytotoxicity and mammalian cytotoxicity (Abbas et al., 1995), and the resulting corresponding aminopentol (AP1, AP2) backbones were only 30–40% (Norred et al., 1997) or 10% as potent as the parent toxins (Merrill et al., 1993b). In contrast, removal of the TCA side groups has also been shown to enhance cytotoxicity in certain mammalian cell lines (Abbas et al., 1993b) and AP1 displays renal toxicity comparable to that of FB1 (Voss et al., 1996a).

#### Degradation: Organisms and Pathways

Only a few microorganisms are known to degrade and thereby detoxify fumonisins (**Table 2**), mostly by removal of the TCA groups as well as the free amino group. Although none of these studies actually determined detoxification of fumonisin B1 by these microorganisms through in vitro assays, based on what is known regarding the role of the TCA groups and the free amino group in inferring the toxicity of FB1 (**Table 1**, **Table 1** in Supplementary Material), we can safely assume detoxification was indeed achieved.

Sphingomonas sp. ATCC 55552 was isolated from fieldgrown, moldy maize kernels, and stalk tissue (Duvick et al., 1998a) and been shown to degrade fumonisin B1 through the consecutive action of a carboxylesterase (Duvick et al., 2003) and an aminotransferase (Heinl et al., 2011). The same pathway was found in Sphingopyxis sp. MTA 144 isolated from composted earth (Täubel, 2005), in which the gene cluster responsible for fumonisin degradation was identified with fumD, encoding the carboxylesterase and fumI encoding the aminotransferase (Hartinger et al., 2009; Heinl et al., 2009, 2010).

Degradation by Exophiala sp., also isolated from field-grown, moldy maize kernels, and stalk tissue (Duvick et al., 1998a), was shown to be conferred by a carboxylesterase and, in contrast to ATCC 55552 and MTA 144, by an oxidative deaminase. Two degradation products were identified: a new compound, 2-oxo-12,16-dimethyl-3,5,10,14,15-icosanepentol hemiketal, and in smaller amounts the N-acetylated aminopentol backbone (NacetylAP1).

Strain NCB 1492, isolated from maize field soil and related to the Delftia/Comamonas group, gave rise to four tentative degradation products of fumonisin B1 (C34H59NO15): heptadecanone (C17H34O), isononadecene (C19H38), octadecenal (C18H34O), and eicosane (C20H42) (Benedetti et al., 2006). The first degradative steps are thought to occur extracellularly, with deamination (and possibly esterase) activities followed by a slower degradation of the aliphatic chain.

Insights into the detoxification of fumonisins can also be useful for mycotoxins produced by other fungal genera. In

#### TABLE 3 | Degradation and/or detoxification products of ZEN.


\**uses ZEN as sole carbon source.*

this light, we would like to draw the focus on the Alternaria toxins AAL-T<sup>A</sup> en -TB, which share with fumonisins a distinct structural similarity and toxicity mechanism (Abbas et al., 1993a; Tsuge et al., 2013). Fumonisins have two TCA side chains esterified to the aminopentol backbone, whereas AALtoxins have only one TCA side group, and are therefore collectively referred to as sphinganine-analog mycotoxins. To the best of our knowledge, there have not been any reports of microbial strains capable of degrading AAL-toxins, but based on their structural similarity it is likely that fumonisin degrading organisms as described above might also be capable of degrading AAL-toxin.

## Zearalenone

#### Toxicity

ZEN is mainly produced by fungi belonging to the genus Fusarium such as F. graminearum and F. culmorum and possesses estrogenic activity in pigs, cattle and sheep (Zinedine et al., 2007). The toxicity of ZEN is mainly conferred by its lactone group and the free C-4 hydroxyl group (**Table 1**) which is necessary for binding the estrogen receptor (El-sharkawy S. H. and Abul-hajj Y. J., 1988). Many derivatives of ZEN are known and some exhibit a higher estrogenicity than the mother compound (Shier et al., 2001), such as α-zearalenol, α- and β-zearalanol, and zearalanone. Several studies described the

#### TABLE 4 | Degradation and/or detoxification products of acylated trichothecenes.

#### TABLE 4 | Continued


microbial transformation of ZEN to such derivatives, but as they do not represent a true detoxification of the compound they are not discussed in this review. Also, cases in which no clear evidence is presented (yet) for true detoxification (e.g., Pseudomonas strains in Tan et al., 2014, 2015) are not discussed in detail.

#### Degradation: Organisms and Pathways

To date, two main detoxification mechanisms are known for ZEN, both cleaving a ring structure (**Table 3**). The lactone ring can be cleaved by several fungal species through two mechanisms. Degradation by Gliocladium roseum NRRL1859 (El-sharkawy S. and Abul-hajj Y. J., 1988) resulted in a 1:1 mixture of 1-(3,5-dihydroxyphenyl)-10′ -hydroxy-1-undecen-6′ -one, and 1- (3,5-dihydroxyphenyl)-6′ -hydroxy-1-undecen-10′ -one. Matthies et al. (2001) showed that production of the ZEN-degrading enzyme in G. roseum DSM 62726 was induced the highest by the derivatives zearalanol and α-zearalanol. Almost similarly, only the first metabolite was observed after degradation by a near isogenic strain of NRRL1859, Clonostachys rosea (synonym: G. roseum, teleomorph: Bionectria ochroleuca) IFO 7063 (Kakeya et al., 2002), resulting in the loss of estrogenic activity in MCF-7 cancer cells (**Table 1** in Supplementary Materials), through the activity of a ZEN lactonohydrolase enzyme (zhd101) which catalyzes the hydrolysis of ZEN at the ester bond in the lactone ring, followed by spontaneous decarboxylation (Takahashi-Ando et al., 2004). Based on this knowledge, Popiel et al. (2014) searched a collection of Trichoderma and Clonostachys isolates for functional lactonohydrolase homologs, to find a functional ZEN lactonohydrolase in mycoparasitic Trichoderma aggressivum. A similar pathway might also exist in Bacillus sp., as cell culture extracts of B. natto CICC 24640 and B. subtilis 168 showed complete degradation of ZEN in conjunction with CO2-emmission, indicative of decarboxylation (Tinyiro et al., 2011).

A second cleavage pathway is exhibited by the yeast Trichosporum mycotoxinivorans (Molnar et al., 2004) to ZOM-1 intermediate (cleavage at the C6-ketone group), suggested to take place through a lactone intermediate and subsequent activity by unspecified a/b-hydrolase, but without the decarboxylation as seen in C. rosea. ZOM-1 did not show any estrogenic activity in a yeast bioassay, nor interaction with the human estrogen receptor (Vekiru et al., 2010), nor estrogenic activity with MCF-7 cells (Liu et al., 2001). It is important to notice that T. mycotoxinivorans is well-known in medicine, since it can cause opportunistic infections or induce summer-type

#### TABLE 5 | Degradation and/or detoxification products of non-acylated tricothecenes.


\**growth as sole carbon source.*

hypersensitivity pneumonitis in immune-deficient cystic fibrosis patients (Tintelnot et al., 2011) which can be an impediment for applications.

Detoxification of ZEN contaminated corn steep liquor by A. niger strain FS10 and its culture filtrate, exemplified by less severe liver and kidney damage in rats, was recently reported (Sun et al., 2014). Two intermediate products, ZEN-A and ZEN-B, which inferred reduced liver and kidney damage in rats compared to ZEN, were detected, of which the latter the authors suggested the benzene ring might be cleaved because the UV absorption of ZEN was lost in ZEN-B. Somewhat similarly, two degradation products (ZEN-1 and ZEN-2) were detected after degradation by Acinetobacter sp. SM04 isolated from agricultural soil, for which no equally estrogenic activity could be detected on the basis of the MTT (tetrazolium salt) cell proliferation assay in MCF-7 cell line. Also, UV-Vis spectroscopy indicated cleavage of the benzene ring in these products (Yu et al., 2011a). Interestingly, ZEN and its estrogenic properties were only reduced when degradation tests were performed with extracellular extracts from M1 medium cultures, where sodium acetate is the only extra carbon source, and not from Nutrient Broth cultures, where many different extra carbon sources are present (Yu et al., 2011b). This indicates that the transcription of genes responsible for ZEN degradation may be regulated by catabolite repression.

Pseudomonas sp. ZEA-1, isolated from the rhizosphere of a corn plant, was shown to harbor the responsible degradation genes on a 120 kb plasmid mediating the transformation of ZEN and its derivatives α- and β- ZEN into less toxic products to Artemia salina. The transformation product was not elucidated,

#### TABLE 6 | Degradation and/or detoxification products of aflatoxins.


other than the specific absorption maximum at 400 nm (Altalhi, 2007). A 5.5 kb fragment containing the gene(s) encoding for ZEN degradation was cloned and actively expressed in Escherichia coli (Altalhi and El-Deeb, 2009).

The complete loss of ZEN estrogenic activity was obtained by several degrading Rhodococcus strains (Kriszt et al., 2012; Cserháti et al., 2013), without the identification of possible metabolites. R. pyridinovorans K408 showed a biodegradation potential of up to 85% and decreased the estrogenicity with 76%. Several strains also simultaneously degraded AFB1, ZEN, and T2-toxin (Cserháti et al., 2013), confirming the status Rhodococcus as a metabolically highly versatile genus with a large potential for degradation of aromatic and other pollutants (Larkin et al., 2005).

#### Trichothecenes Toxicity

Trichothecenes are sesquiterpenoids produced by mainly the genera Fusarium, Trichothecium, Myrothecium, Trichoderma, and Stachybotrys fungi (Sudakin, 2003; Kimura et al., 2007; Li

#### TABLE 7 | Degradation and/or detoxification products of ochratoxins.


et al., 2011). High doses lead to emesis, whereas low doses induce decreased feed consumption and weight gain (Eriksen and Pettersson, 2004). Trichothecenes are characterized by a 12,13-epoxy-trichothec-9-ene nucleus (Hussein and Brasel, 2001). Type A trichothecenes do not contain carbonyl function at C8 (T-2 toxin, HT-2 toxin, T-2 tetraol, T-2 triol, 15 monoacetoxyscirpenol, DAS, neosolaniol, and scirpentriol). Type B trichothecenes have a carbonyl group at C8 [deoxynivalenol (DON), 15-acetyl DON, 3-acetyl DON, nivalenol (NIV), 4-acetyl NIV]. Type C trichothecenes include another epoxide group and type D trichothecenes contains an additional ring system between C4 and C15 position (Zhou et al., 2008; McCormick et al., 2011).

The 12,13-epoxide ring in trichothecenes is essential for their toxicity (Zhou et al., 2008) and has been linked to the cytotoxicity of trichothecenes, namely inhibition of protein, RNA and DNA synthesis (Hussein and Brasel, 2001; Rocha et al., 2005). Trichothecenes bind with the 60S subunit of the ribosome and interfere with the action of peptidyltransferase (Ehrlich and Daigle, 1987). However, the degree of toxicity is dependent on the presence of substituents on C15 and C4 (Cundliffe et al., 1974; Cundliffe and Davies, 1977). The most potent mycotoxin T-2 toxin has acetyl or acyl side groups on C4, C8, and C15 of the basic structure. Loss of a side group from either of these positions resulted in reduced protein synthesis inhibition (T-2 toxin to HT-2 toxin, neosolaniol, or DAS). Further removal of side groups weakens their effect (T-2 triol, T-2 tetraol, 15 monoacetyl DAS, scirpentriol, fusarenon X, and DON) and reduction of hydroxyl groups, forming verrucarol, reduced their effectiveness greatly (Thompson and Wannemacher, 1986; **Table 2** in Supplementary Material). De-acylation is clearly a first step toward detoxification, illustrated in **Figure 1** in Supplementary Material. This reduced effect of de-acylation of T-2 toxin is also confirmed with human melanoma SK-Mel/27 cell lines (Babich and Borenfreund, 1991) and β-galactosidase activity of Kluyveromyces marxianus (Engler et al., 1999) showing the same tendency (**Table 1** in Supplementary Material).

#### Degradation: Organisms and Pathways

The toxicity of trichothecenes is, next to their epoxide-group, also dependent on their acylated side chains. Therefore, two main groups are distinguished; acylated (e.g., T-2 toxin) and non-acylated trichothecenes (e.g., DON).

As previously described, de-acylation is the first step in detoxification of acylated trichothecenes. Degradation of T-2 toxin to HT-2 toxin and subsequently to T-2 triol was performed by Curtobacterium sp. strain 114-2 of which the reduced toxicity of T-2 triol was once more confirmed resulting in 23 and 13 times less toxic than, respectively T-2 toxin and HT-2 toxin (Ueno et al., 1983; **Table 4**). Still, the epoxide group in trichothecenes remains responsible for their toxicity. De-epoxidation is the next step of detoxification trichothecenes. Several studies focuses on the degradation of multiple trichothecenes and the differences between their metabolism by the same organism(s). Young et al. (2007) studied the metabolism of diverse trichothecenes by chicken intestinal microbes. For the nonacylated trichothecenes (4-DON, NIV, and verrucarol) their deepoxidized metabolites were observed, for DAS, neosolaniol and T-2 toxin only de-acylation was exhibited and for the monoacetyl trichothecenes (3-acetyl DON, 15-acetyl DON, and fusarenon X), de-acylation was the predominant pathway. In another study, pig gastrointestinal microflora transformed 3 acetyl DON into DON and which was further de-epoxidized (Eriksen et al., 2002). Rat intestinal microflora was also able to deepoxidize T-2 tetraol and scirpentriol, transform T-2 toxin into de-epoxy HT-2 toxin and de-epoxy T-2 triol and DAS into deepoxymonoacetoxyscirpenol and de-epoxyscirpentriol (Swanson et al., 1987). All above mentioned cases concern degradation by mixed cultures. In contrast, Eubacterium BBSH 797 has the ability to degrade several trichothecenes as pure culture isolated from bovine rumen fluid (Fuchs et al., 2000, 2002; Binder and Binder, 2004) and has been developed into a commercial product (Biomin <sup>R</sup> BBSH 797) for detoxifying trichothecenes in animal feed (He et al., 2010). It is known for its detoxification capacities of DON into DOM-1 and de-epoxidization of NIV, T-2 tetraol, scirpentriol, and HT-2 toxin. T-2 toxin was de-acylated into HT-2 toxin, whereas degradation of T-2 triol involved the competition of two reactions; (1) de-epoxidation or (2) deacylation into T-2 tetraol and subsequently de-epoxidation into de-epoxy T-2 tetraol (Fuchs et al., 2002). Further, 4-acetyl NIV and 3 acetyl NIV was de-acetylated and/or de-epoxidized (Fuchs et al., 2000).

Degradation of DON occurs through de-epoxidation, oxidation, or isomerization (**Table 5**). Microbial culture C133 of fish guts transformed DON to DOM-1 (Guan et al., 2009). Eubacterium BBSH 797 is known to degrade DON into DOM-1 anaerobically (Binder and Binder, 2004). Citrobacter freundii could transform DON into DOM-1 aerobically (Rafiqul, 2012). DON can also be oxidized to 3-keto DON which is 10 times less toxic than DON evaluated with a bioassay based on mitogen-induced and mitogen-free proliferations of mouse spleen lymphocytes (Shima et al., 1997). The bacterium strain E3-39 which degraded DON to 3-keto-DON, is belonging to the Agrobacterium-Rhizobium group. A mixed culture from environmental sources could degrade DON into 3-keto-DON, whereas 15-acetyl DON, 3-acetyl DON and fusarenon-X were also transformed (Volkl et al., 2004). Subsequently, He (2015) found the soil bacterium Devosia mutans 17-2-E-8 which transformed DON into 3-epi-DON (major product) and 3-keto-DON (minor product). These metabolites have also been tested on their toxicity with two assays. The IC<sup>50</sup> values of 3-epi-DON and 3-keto-DON were 357 and 3 times higher, respectively, than that of DON on the basis of a MTT bioassay using Caco-2 cell line to asses cell viability, and were 1181 and 5 times higher, respectively, than that of DON on the basis of a cell proliferation BrdU bioassay using 3T3 fibroblast cell line to asses DNA synthesis (**Table 1** in Supplementary Material). Toxicological effects of 14-day oral exposure of B6C3F<sup>1</sup> mouse to DON and 3-epi-DON were also investigated concluding that 3-epi-DON was at least 50 times less toxic than DON (He, 2015). The metabolite 3-epi-DON was also formed by degradation of DON through Nocardioides sp. strain WSN05-2 isolated from a wheat field (Ikunaga et al., 2011). And lastly, nine Nocardioides strains (Gram-positive) and four Devosia strains (Gram-negative) produced 3-epi-DON aerobically. The Gram-positive strains showed DON assimilation, whereas the Gram-negatives did not (Sato et al., 2012).

Further, hydroxylation and glycosylation of trichothecenes are also known for their detoxification capability (He et al., 2010), however these derivatives can be rehydrolyzed or regenerated in the digestive tract of animals and humans losing their detoxification capacity.

## Aflatoxins

#### Toxicity

Aflatoxins are furanocoumarins produced by mainly Aspergillus species (Wu et al., 2009; Samuel et al., 2013). Naturally occurring aflatoxins are categorized by IARC as carcinogenic to humans (Group 1; IARC, 2002). Aflatoxin B1 (AFB1) is activated by cytochrome P450 system to a highly reactive AFB1-8,9-epoxide which can react with DNA (Eaton and Groopman, 1994; Guengerich et al., 1996).

The toxicity of AFB1 is mainly caused by the lactone ring. Cleavage of the lactone ring leads to a non-fluorescent compound with reduced biological activity (Lee et al., 1981). The residual component has a 450 times reduced mutagenicity (measured with the Ames test) and a 18 times reduced toxicity (measured with chicken embryo test; Lee et al., 1981). Also the difuran ring moiety, especially the presence of the double bond in the terminal furan ring, contributes to the toxicity (Wogan et al., 1971) as evidenced by comparing the toxicity of aflatoxins with similar coumarin molecules. Wong and Hsieh (1976) concluded by comparing several aflatoxins and metabolites with the Ames test that the double bond was also involved in both mutagenic and carcinogenic activity of aflatoxins leading that the aflatoxins AFB2 and AFG2 (without a double bond) are much less toxic than AFB1 and AFG1 (with a double bond).

#### Degradation: Organisms and Pathways

To our knowledge, a first report on the microbial detoxification of AFB1 has been published in 1966, mediated by Flavobacterium aurantiacum (now called Nocardia corynebacterioides; Ciegler et al., 1966; Teniola et al., 2005). Although no detoxification products were analyzed, residual toxicity to ducklings was found to be absent indicating true detoxification (Ciegler et al., 1966). The biosafety of the microorganism was confirmed using an in vivo trial with chickens (Tejada-Castañeda et al., 2008). Since this first report, many studies have focused on the detoxification of AFB1. However, only a few studies detected the degradation products and analyzed their toxicity. Generally, two main detoxification pathways are observed: modification of the difuran ring or modification of the coumarin structure.

Firstly, modification of the difuran ring moiety was reported in several studies. Degradation of AFB1 into AFB1-8,9 dihydrodiol was performed by manganese peroxidase from the white rot fungi Phanerochaete sordida (Wang et al., 2011) and the "aflatoxin-detoxifizyme (ADTZ)" of fungus Armillariella tabescens (Liu et al., 1998b; **Table 6**). The authors suggested that AFB1 degradation initially involves formation of AFB1-8,9-epoxide, after which a hydrolysis resulted in a dihydrodiol-derivate. Detoxification was confirmed with a reduced mutagenicity measured by the Ames Salmonella-based test (Liu et al., 1998b, 2001; Wang et al., 2011; **Table 1** in Supplementary Material) and reduced toxicity measured with rat liver (Liu et al., 1998b) and chicken embryos (Liu et al., 1998a). Another metabolite was detected with the white rot fungus Pleurotus ostreatus GHBBF10 which degraded 91.76% of AFB1 into a component which could be a hydrolyte of AFB1, namely dihydrohydroxyaflatoxin B1 (AFB2a) (Das et al., 2014; **Table 2**). AFB2a has also a reduced mutagenicity (Wong and Hsieh, 1976; **Table 1** in Supplementary Material).

Secondly, the lactone ring in the coumarin moiety of AFB1 can be changed. A Pseudomonas putida strain has been discovered degrading AFB1 into AFD1 and subsequently into AFD2 (**Table 6**). The metabolite AFD1 had been previously discovered through ammonization and acidifying AFB1, whereas the difuran ring stays unchanged and the lactone ring is cleaved. AFD1 has a lower mutagenicity and toxicity measured by respectively the Ames Salmonella-based test (Méndez-Albores et al., 2005) and HeLa cells with the MTT [3-(4,5-dimethylthiazole-2-yl)-2,5 diphenyltetrazolium bromide] method (Samuel et al., 2014). The metabolites AFD2 (an aflatoxin metabolite lacking the lactone and cyclopentenone ring) and AFD3 also showed a lower toxicity toward HeLa cells (**Table 1** in Supplementary Material; Samuel et al., 2014).

In certain studies, no degradation product was identified, but toxicity tests were performed on the treated AFB1. Similarly to F. aurantiacum as mentioned before, a pure laccase enzyme from Trametes versicolor and a recombinant laccase enzyme produced by A. niger degraded, respectively, 87.34 and 55% of AFB1 with a significant loss of mutagenicity evaluated in the Ames Salmonella-based assay (Alberts et al., 2009). Extracellular enzymes of Rhodococcus eryhtropolis were also able to detoxify AFB1 with a loss of mutagenicity (Alberts et al., 2006).

## Ochratoxins

#### Toxicity

Ochratoxins (OT) are a group of mycotoxins sharing an isocoumarin moiety substituted with a phenylalanine group (OTA, OTB, hydroxyl-OTA), a phenylalanine ester group (OTC, OTA methylester, OTB methyl ester, OTB ethyl ester), or a hydroxyl group (OTα and OTβ). OTA is the most important OT because of its incidence in food- and feed commodities. It is composed of a 7-carboxy-5-chloro-8-hydroxy-3,4-dihydro-3-R-methylisocoumarin (OTα) moiety and the amino acid L- phenylalanine group. Both structures are linked through a carboxy group via an amide bond. OTA is produced by Aspergillus and Penicillium species (Richard, 2007; McCormick, 2013). The mode of action of OTA is broad and therefore the molecule has nephrotoxic, mutagenic, teratogenic, neurotoxic, hepatotoxic, and immunotoxic properties (Pfohl-Leszkowicz and Manderville, 2007). The toxicity of OTA is mainly attributed to its isocoumarin moiety and probably not to the phenylalanine moiety (**Table 1**; Xiao et al., 1996). The carboxyl group of the phenylalanine moiety and also the Cl group of the other moiety seem to be conducive for the toxicity of OTA.

#### Degradation: Organisms and Pathways

The main detoxification pathway of OTA is the hydrolyzation of the amide bond between the isocoumarin residue and phenylalanine by a carboxypeptidase. Two classes of carboxypeptidases have been associated with degradation of OTA namely Carboxypeptidase A (CPA) (Stander et al., 2001; Chang et al., 2015) and Y (CPY) (Dridi et al., 2015). The main difference between both is the use of a zinc ion within the protein for hydrolysis of the peptide at the C-terminal of the amino acid. Almost all strains that are reported to degrade OTA use this pathway resulting in the formation of L-βphenylalanine and OTα the former being less toxic than OTA (**Table 7**; Bruinink and Sidler, 1997). Although this is a very straightforward way of reducing the amount of OTA in food and feed samples, it is important to highlight that the efficient degradation of OTA is depending on the activity of the peptidase enzyme. With this respect, several research groups showed that these carboxypeptidase enzymes tend to have high optimal temperatures (30◦C or higher) which might hamper practical applications, observed with Pediococcus parvulus and several yeasts such as Pfaffia rhodozyma (Péteri et al., 2007; Patharajan et al., 2011; Abrunhosa et al., 2014). Other enzymes are also able to carry out this reaction: Deoxygenases, lipases, amidases, and several commercial proteases (Abrunhosa et al., 2006), have also been identified as carrying out this reaction. Although depending on the enzyme, intermediates can be different, the end product is always OTα.

Some interesting strains are highlighted here. Trichosporon mycotoxinivorans was demonstrated to deactivate OTA by conversion into the nontoxic OTα. Even more intriguingly, T. mycotoxinivorans was also able to decarboxylate ZEN (Molnar et al., 2004; Vekiru et al., 2010). After 24 h, ZEN was degraded to carbon dioxide or into metabolites that neither showed fluorescence nor did absorb UV-light. Neither α- nor β-ZEL, other equally estrogenic metabolites of ZEN, could be detected. It is commercially applied as feed additive under the commercial name Biomin <sup>R</sup> MTV.

Phenylobacterium immobile (Wegst and Lingens, 1983) was also found to convert OTA to OTα through a dioxygenase step on the phenylalanine moiety, a dehydrogenation to catechol, a ring cleavage, and the final formation of OTα via a hydrolase.

## UNEXPLORED WORLDS THAT MIGHT HARBOR VALUABLE MYCOTOXIN DEGRADING MICROORGANISMS

#### Targeting Carboxyl Esters

As stated above, fumonisins and acylated trichothecenes share carboxyl-esters which are involved in their toxicity. Detoxification of fumonisins is realized by removal of the tricarballylate side groups via carboxylesterases (EC 3.1.1.1). Similarly, acylated trichothecenes have several side groups where carboxylesterases could attack on the carboxyl group, as observed with carboxylesterases from rat liver microsomes (categorized as EC 3.1.1.1) degrading T-2 toxin into HT-2 toxin (Ohta et al., 1977; Johnsen et al., 1986). Carboxylesterases are multifunctional enzymes that catalyze the hydrolysis of substrates containing ester, amide, and thioester bonds with relatively broad substrate specificity (Bornscheuer, 2002) which is attributed to a large conformable active site that permits entry of numerous structurally diverse substrates. Microbial carboxylesterases have been reported in the degradation of pesticides; some hydrolyze pyrethroids and bind stoichiometrically to carbamates and organophosphates reviewed by Singh (2014). Several organisms have been isolated which degradative capacities of these compounds inferred by the expression of carboxylesterases, for which, bearing in mind their general broad substrate specificity, it might be worthwhile to screen for degradation of fumonisins and acylated trichothecenes. For example, broad-spectrum pyrethroid-hydrolyzing carboxylesterases were identified in the lambda-cyhalothrin degrading Ochrobactrum anthropic YZ-1 strain (Zhai et al., 2012) and Bacillus sp. DG-02, isolated from a pyrethroid-manufacturing wastewater treatment system (Chen et al., 2014). Similarly, an Acinetobacter baumannii strain was shown to degrade a wide range of organophosphorus compounds and evidence for a novel carboxylesterase in this strain was presented. Taking an environmental DNA (eDNA) isolation approach, Rashamuse et al. (2009) screened a microbial community to access novel carboxylesterases from environmental genomes: a carboxylesterase gene with 60% sequence identity to the gene from Ralstonia eutropha was identified, along with subsequent heterologous expression in Escherichia coli in a biologically active form. A similar approach might be taken to discover more mycotoxin-active carboxylases based on sequences of carboxylases present in Sphingomonas sp. ATCC 55552, Exophiala sp., or Sphingopyxis sp. MTA 144.

## Targeting a Lactone Ring

The presence of a lactone moiety is shared by OTA, aflatoxins, and ZEN. Lactone chemicals are well-known as auto-regulators in both eukaryotic and prokaryotic cells. A well-known example is acyl homoserine lacton which is a quorum sensing molecule associated with biofilm formation. Because of the detrimental effects of biofilms in many industrial applications, high throughput research initiatives have been undertaken in the past and present in search for enzymes able to degrade these lactone molecules. These, often metagenomics, approaches result in the characterization of new and more efficient lactonase enzymes (Shimizu et al., 2001; Riaz et al., 2008; Schipper et al., 2009). The potential activities of these lactonases with respect to mycotoxins remains elusive but scientific fields studying biofilm issues might offer new microbial consortia ready to be explored for their mycotoxin degrading capacities.

Also targeted analyses can result in the characterization of new and efficient lactonase enzymes. In a screening assay for enzymes able to degrade bio-active lactones, a novel lactonohydrolase, an enzyme that catalyzes the hydrolysis of aldonate lactones to the corresponding aldonic acids, was purified from Fusarium oxysporum AKU 3702. The enzyme irreversibly hydrolyzes a broad spectrum of aromatic lactones, such as dihydrocoumarin and homogentisic-acid lactone (Shimizu et al., 1992; Kobayashi et al., 1998).

New insights for biodegradation of mycotoxins with estrogenic effects such as ZEN might come from studies on the microbial degradation of steroidal estrogens. Several strains have been isolated which are able to degrade the steroidal estrogen estrone (E1), also harboring a lactone ring (Yu et al., 2013), among which Sphingomonas sp. KC8 (Yu et al., 2007), Bacillus subtilis E2Y4 (Jiang et al., 2010), and several Rhodococcus sp. (Yoshimoto et al., 2004), remarkably all isolated from activated sludge. Also, cometabolic degradation of ethinyl estradiol (EE2) was obtained with nitrifying activated sludge (Vader et al., 2000). Therefore, activated sludge might prove to be a rich source of degradation potential for lactone-harboring mycotoxins.

## Targeting an Epoxide Moiety

For trichothecenes, the epoxide moiety is an important chemical group associated with toxicity. Microbial transformation of epoxides was studied by Swaving and de Bont (1998) who demonstrated that two types of enzymes were responsible for detoxification of epoxides: glutathione transferases as a class of general detoxifying enzymes and epoxide hydrolases which are specific for detoxification of epoxides. Glutathione transferases (dependent on glutathione as cofactor) are mostly found in aerobic eukaryotes and prokaryotes, such as E. coli and Rhodococcus sp. which degrades a range of epoxides. Epoxide hydrolases are found in many microorganisms, like Flavobacterium, Pseudomonas, Corynebacterium, and Stigmatella species. Other enzymes can also convert an epoxide intermediate via a certain pathway (e.g., alpha-pinene oxide lyase from Nocardia sp. strain P18.3 and Pseudomonas fluorescens NCIMB 11671, styrene oxide isomerase of Pseudomonas species,

Xanthobacter 124X or Exophilia jeanselmei, or epoxyalkanedegrading enzyme in Xanthobacter Py2 (Swaving and de Bont, 1998). Broudiscou et al. (2007) proved that mono-and sesquiterpenes were degraded in the presence of mixed rumen microorganisms, corresponding with the isolation origin mostly found for microorganisms degrading trichothecenes. Ptaquiloside, also a sesquiterpene toxin, could be degraded by soil microorganisms (Engel et al., 2007) which can be a new source for biodegradation of trichothecenes.

#### Targeting Poly-Aromatic Ring Structures

White rot fungi are frequently found for degrading aflatoxins, such as A. tabescens, P. sordida, P. ostreatus, T. versicolor, and Peniophora sp. (Liu et al., 1998b; Motomura et al., 2003; Alberts et al., 2009; Wang et al., 2011; Das et al., 2014; Yehia, 2014). White rot fungi are well-known for their degrading capabilities of their natural substrate lignin and a broad spectrum of structurally diverse toxic environmental pollutants (e.g., munitions waste, pesticides, polychlorinated biphenyls, polycyclic aromatic hydrocarbons, bleach plant effluent, synthetic dyes, synthetic polymers, and wood preservatives; Reddy, 1995; Pointing, 2001). Lignin peroxidases, manganese peroxidases and laccases are the major enzymes involved in lignin degradation based on oxidative mechanisms (Tuor et al., 1995). Laccases and manganese peroxidases of white rot fungi have been reported for degrading aflatoxins which possible can lead to different metabolites (Motomura et al., 2003; Wang et al., 2011). Peniophora sp. SCC0152, P. ostreatus St2-3, and several Trametes sp. strains demonstrated the degradation of Poly R-478 dye and AFB1 (Alberts et al., 2009). Next to white rot fungi, the genus Rhodococcus is also known to have promising degradation capability for xenobiotics (Martínková et al., 2009). Alberts et al. (2006) and Eshelli et al. (2015) suggested that degradation of AFB1 (polyaromatic compound) by a Rhodococcus erythropolis strain could be degraded in a similar way of degrading polyaromatic compounds of which their degradation occurs through a cascade of enzyme reactions (e.g., ring cleavage biphenyl dioxygenases, dihydrodiol dehydrogenases, and hydrolases). Degradation of a wide range of aromatic compounds results in a limited number of central intermediates (catechol, protocatechuate, gentisate) which are further degraded through central pathways for finally entering the citrate cycle (Martínková et al., 2009). In addition, R. erythropolis NI1 strain was found which was capable of degrading AFB1, ZEN, and T-2 toxin at the same time (Cserháti et al., 2013). Hence, various organisms have the potential for degrading multiple mycotoxins or other components, exemplified by. Stenotrophomonas maltophilia, Stenotrophomonas sp. NMO-3, and Pseudomonas aeruginosa which can degrade AFB1 and coumarin (Guan et al., 2008; Liang et al., 2008; Sangare et al., 2015), and Mycobacterium fluoranthenivorans FA4T which can degrade AFB1 and also grow on the polycyclic aromatic hydrocarbon fluoranthene (Hormisch et al., 2004).

Supporting the notion that microorganisms are able to metabolize structurally comparable chemicals from vastly different origins, a mixed enrichment culture capable of removing ZEN as sole carbon source, without the presence of derivatives, was obtained from soil collected at a coal gasification site, which are generally known to be associated with polycyclic aromatic hydrocarbon contamination. Removal of ZEN was enhanced in the presence of phenanthrene through enhanced microbial growth, indicating that organisms capable of using ZEN were also able to metabolize phenanthrene (Megharaj et al., 1997). Building further on this notion, cleaving the aromatic ring of ZEN by A. niger FS10 (Sun et al., 2014) and Acinetobacter sp. SM04 (Yu et al., 2011a), for which no enzymes have been identified, might bear resemblance to the degradation of resorcinol (1,3-dihydroxybenzene) for which degradation is known by P. putida (Chapman Ribbons and Ribbons, 1976) and Azotobacter vinelandii (Groseclose and Ribbons, 1981).

## Targeting a Carboxyl/Amide Moiety

Carboxypeptidase A and Y belong to the group of protease enzymes. Great interest in these enzymes comes from the field of wastewater treatments as these enzymes play a vital role in the extracellular catabolism of organic matter in activated sludge. In search of these enzymes, progressively more culture independent screening approaches are being employed as up to 90% of bacteria present in wastewater cannot be cultured and in this way a large reservoir of enzymes is overlooked. In matrices harboring a vast set of microorganisms that cannot be cultured, metagenomics analyses are often the solution to get an in depth insight into the complexity of these enzymes in a certain matrix. Pursuing this approach, a metagenomics analysis of waste water revealed a highly diverse phylogenetic diversity of carboxypeptidase gene sequences including previously undescribed types of carboxypeptidases which might be interesting to be applied for diverse biotechnological applications such as the remediation of OTA contaminated batches (Jin et al., 2014).

## PERSPECTIVES

Although fumonisins, trichothecenes, ZEN, OTA, and aflatoxins comprise the major mycotoxin groups in food- and feed commodities, there are several other mycotoxins that were not addressed in present review because knowledge on biodegradation and detoxification is scarce. Cyclodepsipeptides such as beauvericin and enniatins are increasingly reported in many countries in several commodities. However, to our knowledge, no reports are available on their biodegradation and detoxification by microorganisms.The same accounts for ergot alkaloids such as lysergic acid, ergine, and ergopeptines. They occur widely but to date, only one paper has recently reported on a R. erythropolis isolate able to degrade these compounds (Thamhesl et al., 2015). Finally, for the Penicllium expansum mycotoxin patulin, recent papers report on biodegradation of this mycotoxin by Pichia caribbica (Cao et al., 2013), Metschnikowia pulcherrima (Reddy et al., 2011), Kodameae ohmeri (Dong et al., 2015), Rhodosporidium spp. (Castoria et al., 2011; Zhu et al., 2015), and Saccharomyces cerevisiae (Moss and Long, 2002). Nevertheless, for these emerging and also for the other mycotoxins, there is still a considerable need for concerted research initiatives to identify new high-performance strains which can be implemented in practice.

Many surveys around the globe illustrate that mycotoxin contaminated batches of food and feed products often contain multiple both structurally related and non-related mycotoxins. An emerging approach to tackle this issue is biodegradation of mycotoxins by microorganisms. In our opinion an ideal biodegrading and detoxification agent should meet following features: (i) a fast and efficient degradation, (ii) of a broad spectrum of toxins, (iii) into non-toxic end products, (iv) by a non-pathogenic strain or consortium (v) under conditions that are relevant for the matrix in which the mycotoxin problem occurs. In order to do so, we urge researchers to look beyond the disappearance of the mother compound to rule out the creation of any lesser evils, and to explore strange new worlds, seek out new organisms and new metabolic pathways.

#### REFERENCES


#### AUTHOR CONTRIBUTIONS

KA and LD conceived the idea and scope for the review. IV, KA, and LD all equally contributed to gathering and summarizing the literature, designing the tables and figures, and writing and editing of the paper.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fmicb. 2016.00561

Figure 1 | Reduced toxicity of T-2 toxin by subsequent de-acylation.

Table 1 | Toxicity data on mycotoxins and their metabolites.

Table 2 | Toxicity of trichothecenes (from most potent to almost no effectiveness observed) (Thompson and Wannemacher, 1986).


equi isolates from activated sludge in wastewater treatment plants. Appl. Environ. Microbiol. 70, 5283–5289. doi: 10.1128/AEM.70.9.5283-5289.2004


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2016 Vanhoutte, Audenaert and De Gelder. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Degradation of Zearalenone by Essential Oils under In vitro Conditions

Adam Perczak<sup>1</sup> , Krzysztof Jus´ 2 , Katarzyna Marchwinska ´ 2 \*, Daniela Gwiazdowska<sup>2</sup> , Agnieszka Waskiewicz ´ <sup>1</sup> and Piotr Golinski ´ 1

<sup>1</sup> Department of Chemistry, Poznan University of Life Sciences, Pozna ´ n, Poland, ´ <sup>2</sup> Department of Natural Science and Quality Assurance, Faculty of Commodity Science, Poznan University of Economics and Business, Pozna ´ n, Poland ´

Essential oils are volatile compounds, extracted from plants, which have a strong odor. These compounds are known for their antibacterial and antifungal properties. However, data concerning degradation of mycotoxins by these metabolites are very limited. The aim of the present study was to investigate the effect of essential oils (cedarwood, cinnamon leaf, cinnamon bark, white grapefruit, pink grapefruit, lemon, eucalyptus, palmarosa, mint, thymic, and rosemary) on zearalenone (ZEA) reduction under various in vitro conditions, including the influence of temperature, pH, incubation time and mycotoxin and essential oil concentrations. The degree of ZEA reduction was determined by HPLC method. It was found that the kind of essential oil influences the effectiveness of toxin level reduction, the highest being observed for lemon, grapefruit, eucalyptus and palmarosa oils, while lavender, thymic and rosemary oils did not degrade the toxin. In addition, the decrease in ZEA content was temperature, pH as well as toxin and essential oil concentration dependent. Generally, higher reduction was observed at higher temperature in a wide range of pH, with clear evidence that the degradation rate increased gradually with time. In some combinations (e.g., palmarosa oil at pH 6 and 4 or 20◦C) a toxin degradation rate higher than 99% was observed. It was concluded that some of the tested essential oils may be effective in detoxification of ZEA. We suggested that essential oils should be recognized as an interesting and effective means of ZEA decontamination and/or detoxification.

#### Keywords: essential oils, zearalenone, degradation, mycotoxin, HPLC analysis

## INTRODUCTION

Fusarium species and their toxic secondary metabolites – mycotoxins – are responsible for plant diseases and colonize cereal grains (including wheat, barley, rice, and maize) before harvest, causing significant yield losses and reducing their quality (Fletcher et al., 2006; Kalagatur et al., 2015). Fusarium mycotoxins currently considered of importance from the toxicological point of view include zearalenone (ZEA), trichothecenes (mainly deoxynivalenol) and fumonisins, and their occurrence is now regulated by EU legislations. Mycotoxins have a negative impact on the health of humans and animals such as hepatotoxic, haemotoxic, nephrotoxic, estrogenic, and genotoxic effects (Ghedira-Chekir et al., 1999; Riley et al., 2001; Gelderblom et al., 2002; Abid-Essefi et al., 2004). The toxicity of mycotoxins differs depending on the kind of toxin, dose ingested, exposure, gender and age of animals or humans (Przybylska-Gornowicz et al., 2015; Szabó-Fodor et al., 2015).

#### Edited by:

Vijai Kumar Gupta, National University of Ireland, Galway, Ireland

#### Reviewed by:

Magdalena Frac, Institute of Agrophysics Polish Academy of Sciences, Poland Vishal Prasad, Banaras Hindu University, India

\*Correspondence:

Katarzyna Marchwinska ´ katarzyna.kluczynska@ue.poznan.pl

#### Specialty section:

This article was submitted to Fungi and Their Interactions, a section of the journal Frontiers in Microbiology

Received: 28 April 2016 Accepted: 22 July 2016 Published: 11 August 2016

#### Citation:

Perczak A, Jus K, Marchwi ´ nska K, ´ Gwiazdowska D, Waskiewicz A and ´ Golinski P (2016) Degradation ´ of Zearalenone by Essential Oils under In vitro Conditions. Front. Microbiol. 7:1224. doi: 10.3389/fmicb.2016.01224

Zearalenone – a macrocyclic β-resorcyclic acid lactone – is known as a non-steroidal estrogen biosynthesized through a polyketide pathway by Fusarium spp., including F. graminearum, F. culmorum, F. cerealis, F. equiseti, and F. semitectum (Golinski ´ et al., 2009). ZEA is detected in different cereals and – as a thermostable compound – is also present in some final grain products, such as breakfast cereals, bread, pasta, beer, or processed feeds (Bullerman and Bianchini, 2007). ZEA may cause physiological alterations of the reproductive tract of domestic animals, especially pigs (Zinedine et al., 2007; Hueza et al., 2014) and it is also connected with precocious puberty of prepubertal girls (Jakimiuk et al., 2009; Massart and Saggese, 2011). Apart from its estrogenic properties, haematotoxic and genotoxic effects are also reported (Abid-Essefi et al., 2004; Zinedine et al., 2007; Hueza et al., 2014).

The most effective method to control growth of fungi as well as mycotoxin biosynthesis is prevention including preand postharvest strategies. The main solution for controlling fungal diseases in crops is chemical fungicides, with several negative implications after their application, e.g., observed and induced resistance to introduced substances reducing pathogen growth (Steffens et al., 1996; Aguin et al., 2006; Ishii, 2006). Therefore there is a need to find alternative ways for the control of fungi on crops such as biological agents or natural substances including essential oils. There is also a tendency to use essential oils as preservatives in the food industry to protect food from pathogenic or spoilage microorganisms both alone and in combination with other preservative methods (Tiwari et al., 2009; Somolinos et al., 2010). This is due to their strong antimicrobial properties.

Essential oils – as natural products – are complex, volatile, fragrant substances commonly occurring in secretory tissue cells. These odorous oily liquids are plants' aromatic secondary metabolites, mostly obtained by extraction or steam distillation of leaves, fruits, bark, flowers, buds, twigs, seeds, roots and other various plant organs (Bakkali et al., 2008). Essential oil in terms of composition is a mixture of 20–60 different chemical compounds such as ketones, aldehydes, esters, alcohols, terpenes, terpenoids, lactones, and other organic substances. The amount as well as the presence of various ingredients in the essential oil depends mainly on the type of its source material, variety and quality. Major components determine properties of each essential oil, as these natural substances show a broad spectrum of antagonistic activity (Bakkali et al., 2008; Nerio et al., 2010).

In vitro and in vivo assays have proved that essential oils exhibit broad spectrum of application including cosmetic (Arung et al., 2011; Bertuzzi et al., 2013; Binic et al., 2013), antibacterial (Prabuseenivasan et al., 2006; Sokovic et al., 2010 ´ ; Elaissi et al., 2012), fungicidal (Velluti et al., 2004), antiparasitical (Moon et al., 2006), antiprotozoal (Escobar et al., 2010), medical (Warnke et al., 2009; Buckle, 2014), insecticidal, sanitary (Ayvaz et al., 2010), repellent (Nerio et al., 2010) and agricultural (Lo Cantore et al., 2009) as well as in the food industry (Burt, 2004; Bakkali et al., 2008; Prakash et al., 2015). Several studies have reported the significant antifungal as well as antimycotoxigenic effect of essential oils (Tyagi and Malik, 2010; Xing et al., 2014; Rao et al., 2015). The above compounds are also considered as natural alternative to preservatives (Burt, 2004; Speranza and Corbo, 2010; Prakash et al., 2015).

Recently application of essential oils in prevention against fungi and mycotoxin production was reported (Aldred et al., 2008). Volatile essential oil of Salvia fruticosa demonstrated an antifungal effect against Rhizoctonia solani, Sclerotinia sclerotiorum and Fusarium solani (Pitarokili et al., 2003). On the other hand essential oils of Zataria multiflora, Cuminum cyminum, Foeniculum vulgare, Pinaceae and Heracleum persicum had an inhibitory effect on 11 non-toxigenic (F. solani and F. oxysporum) and 10 toxigenic (F. verticillioides, F. poae and F. equiseti) isolates (Naeini et al., 2010). In contrast, Velluti et al. (2003) found that cinnamon, clove, oregano, lemongrass and palmarosa essential oils inhibited growth of F. proliferatum and had an inhibitory effect on fumonisin B<sup>1</sup> biosynthesis. The above were also recognized as inhibiting the growth rate of F. graminearum, ZEA and DON synthesis, although their inhibitory effect on the toxin production depended on environmental conditions (Velluti et al., 2004). It is also worth stressing that inhibition of fungal growth and mycotoxin biosynthesis does not always take place at the same time. Hope et al. (2005) found that cinnamon oil was effective in controlling growth of F. culmorum and F. graminearum, but enhanced toxin formation, which confirms that mycotoxin biosynthesis is associated with the primary metabolism pathway (energy).

Recently, some reports have indicated the possibility of mycotoxin degradation by essential oils, e.g., reduction of fumonisin B<sup>1</sup> by cinnamon oil (Xing et al., 2014) or ochratoxin A by (among others) eucalyptus or neem oils (Rao et al., 2015). However, the data concerning detoxification of mycotoxins with these odorous natural complex compounds are limited, insufficient and need to be elucidated. Therefore the aim of the study was to evaluate the influence of different essential oils on degradation of ZEA under in vitro conditions.

### MATERIALS AND METHODS

## Chemicals

Zearalenone standard was purchased with a standard grade certificate from Sigma–Aldrich (Steinheim, Germany). Organic solvents (HPLC grade) and all the other chemicals were also purchased from Sigma–Aldrich (Steinheim, Germany). Water for the HPLC mobile phase preparation was purified using a Milli-Q system (Millipore, Bedford, MA, USA). The stock solution of ZEA was prepared in methanol (1 mg/mL) and stored at −20◦C. In experiment, three buffer solutions were used and prepared on the basis of citrate (pH 3 and 6) and ammonia (pH 9) buffer solutions.

#### Essential Oils

Eleven essential oils were used in the study: cedarwood (Juniperus virginiana, USA), cinnamon leaf oil (Cinnamomum zeylanicum, Sri Lanka), cinnamon bark (Cinnamomum zeylanicum, Indonesia), lemon peel (Citrus limonum, Italy), pink grapefruit peel (Citrus paradisi, Argentina), white grapefruit peel (Citrus grandis, Argentina), lavender flower (Lavandula angustifolia,

France), eucalyptus leaf oil (Eucalyptus radiata, China), thyme flower and leaf (Thymus vulgaris, Spain), rosemary flower and leaf (Rosmarinus officinalis, Spain) and palmarosa leaf oil (Cymbopogon martinii) (Ecospa Rita Kozak-Chaber Artur Chaber s.c., Poland). Solutions of essential oils were prepared by mixing with water and Tween 80 (10%) as an emulsifying agent. Depending on the experiment, a concentration of 100 and 200 µl/mL of essential oils was used.

## Effect of Essential Oils on Zearalenone Degradation

The effect of eleven different essential oils was examined by mixing ZEA with solutions of oils (100 µl/mL) in a concentration of 5 µg/mL. The mixture contained 100 µl of essential oil, 100 µl of Tween 80, 5 µl of ZEA stock solution and buffer solution at pH = 6 to a final volume of 1 mL. Samples were shaken and incubated for 72 h at 20◦C and the concentration of toxin was assayed at 0, 24, and 72 h by HPLC analysis.

## Effect of Temperature, pH, Concentration of Essential Oils and/or Zearalenone on the Toxin Degradation

For the evaluation of the effect of different factors on ZEA degradation, seven essential oils were chosen: cinnamon leaf, cinnamon bark, lemon peel, pink grapefruit peel, white grapefruit peel, eucalyptus leaf and palmarosa leaf. Different process conditions temperature of samples' incubation (4 and 20◦C), pH (3, 6, and 9), concentration of essential oils (100 and 200 µl/mL) and concentration of the toxin (0.5 and 5 µl/mL of ZEA) were investigated while the time of incubation was 72 h. The mixture contained 100 µl (or 200 µl) of essential oil, 100 µl of Tween 80, 0.5, or 5 µl of ZEA stock solution and appropriate buffer solution to final volume 1 mL.

## HPLC Analysis

After incubation time, 1 mL of each reaction mixture was homogenized for 3 min with 5 mL of acetonitrile:water (90:10, v/v). ZEA was extracted and purified on a ZearalaTest column (Vicam, Milford, CT, USA) according to a procedure described in detail previously (Golinski et al., 2010 ´ ). The elute was evaporated to dryness at 40◦C under a stream of nitrogen. Dry residue was stored at −20◦C until HPLC analyses. Evaporated extracts were dissolved in a 200 µL mixture of acetonitrile:methanol:water (70:20:10, v/v/v), homogenized in an ultrasonic bath (Ultron, type U-505, Dywity, Poland), filtered through a syringe filter of 0.2 µm mesh and applied onto the chromatographic column.

The chromatographic system used in the study consisted of a Waters 2695 high-performance liquid chromatograph (Waters, Milford, CT, USA) with detectors – Waters 2475 Multi λ Fluorescence Detector (λex = 274 nm, λem = 440 nm) and Waters 2996 Photodiode Array Detector – and a Nova Pak C-18 column (150 × 3.9 mm). Data were processed using the Empower software (Waters, Milford, CT, USA). Quantification of ZEA was performed by measuring the peak areas at the retention time according to the relevant calibration curve. A Photodiode Array Detector (PDA) was used to confirm the presence of ZEA on the basis characteristic spectra of this compound. The limit of detection was 0.01 µg m L−<sup>1</sup> .

## Statistical Analysis

The presented data are the mean (±standard deviation) of three replicate trials and obtained results of the mycotoxin degradation were subjected to Student's t-test at p < 0.05 to test for significant differences between different tested samples. The influence of above-mentioned process conditions (pH, temperature, essential oil concentrations) on ZEA degradation was examined by multivariate analysis of variance (ANOVA). Analyses were carried out using STATISTICA for Windows version 10.

## RESULTS

## Effect of Essential Oils (EO) on Zearalenone Degradation Depending on Incubation Time

In the first part of our work ten essential oils were tested for their ability to reduce ZEA content (**Figure 1**). All oils except lavender, thymic and rosemary oils reduced the concentration of toxin. Lemon, grapefruit, eucalyptus, and palmarosa oils were the most effective and the most interesting results were observed for lemon oil, as the amount of the toxin was reduced by up to 31.93% and 46.46% after 24 and 72 h, respectively. Cinnamon leaf essential oil only slightly, but statistically significantly, reduced ZEA by 20.02% and 24.70% after 24 and 72 h, respectively. Incubation time had a significant positive effect on ZEA degradation by tested essential oils. The greatest reduction of ZEA content was observed after 24 h of incubation. During the next 48 h toxin degradation was still observed, but the degree of reduction was lower. For example, cedarwood, grapefruit and eucalyptus essential oils reduced the amount of toxin by 25.36, 30.30, and 32.74% after 24 h of incubation, respectively, while after the next 48 h reduction by several percent was observed, reaching values of 28.73, 33.59, and 37.41%.

## Effect of EO Concentration on ZEA Degradation

On the basis of first stage results, five essential oils were chosen for the next step: cinnamon, grapefruit, lemon, eucalyptus and palmarosa. Preliminary results did not indicate high effectiveness in reduction of ZEA content by cinnamon oil, but it was chosen mainly due to current literature data. Taking into account the differences between essential oils according to their origin, the study was expanded to include additional variants in the case of grapefruit and cinnamon oil. In the experiment the ZEA concentration used was 0.5 µg/mL assuming that a lower amount of toxin would reveal greater differences concerning the influence of different concentrations of essential oils. The results showed that the concentration of essential oil significantly influenced degradation of ZEA by cinnamon bark, white grapefruit, lemon and palmarosa oil (**Figure 2**). The greater the amount of oil, the higher was the rate of reduction of toxin concentration. However

FIGURE 1 | Reduction of zearalenone content by essential oils at different incubation time. The experiment was conducted up to 72 h at 20◦C. Amount of zearalenone was determined by HPLC method. The concentration of essential oils was 100 µl/mL. Data were analyzed by Student's t-test at p < 0.05 (a, b, c – significantly different in column).

it is worth underlining that lemon oil was more efficient at its 10% addition to the samples. Nevertheless, the exact mechanism of such behavior is difficult to explain. Three of the tested essential oils, cinnamon leaf, pink grapefruit and eucalyptus, reduced ZEA with a similar rate, independently of the content of oil in the sample.

## Effect of pH, Temperature and Zearalenone Concentration on Toxin Degradation

Zearalenone degradation by selected essential oils under different conditions including pH (3, 6 and 9), temperature (4 and 20◦C) and ZEA concentration (0.5 and 5 µg/mL) was investigated (**Figures 3** and **4**). All examined parameters influenced the effectiveness of ZEA degradation, but it depended on the kind of EO. The results presented in **Figure 3** show that the degree of ZEA reduction at 20◦C was generally higher in samples containing 5 µg/mL (**Figure 3B**) and ranged from 59.56 to 99.29%, depending on the pH value as well as the EO used. At a concentration of 0.5 µg/mL (**Figure 3A**) the percentage of ZEA reduction showed greater variation. The weakest effect was observed for the pink grapefruit EO, where the highest percentage reduction was 22.68% at pH 9. The remaining essential oils reduced the amount of toxin by 52.8–93.81%. The value of pH at 20◦C significantly influenced the ZEA degradation by all essential oils at a concentration of 5 µg/mL, but at a concentration of 0.5 µg/mL the effect of pH was not observed for palmarosa and eucalyptus oils. Considering the degradation degree of ZEA depending on the pH value it was noted that the highest reduction of toxin amount at the initial concentration of 0.5 µg/mL (**Figure 3A**) was observed at pH 9, excluding eucalyptus and palmarosa oils, which were more effective at pH 3. The lowest degradation rate was observed at pH 6 for all essential oils except for white grapefruit, which was the least effective at pH 3. At a concentration of 5 µg/mL (**Figure 4B**) such a dependency was not observed. Cinnamon bark, pink grapefruit, lemon and palmarosa oils demonstrated the highest reduction of ZEA amount at pH 6, white grapefruit and eucalyptus oils at pH 9 and cinnamon leaf oil at pH 3.

**Figure 4** presents results concerning the influence of the essential oils on the degradation rate of ZEA at 4◦C depending on the pH value and toxin concentration. At a concentration

of 0.5 µg/mL (**Figure 4A**) the percentage of toxin reduction ranged from 6.06 to 89.41%, while at 5 µg/mL (**Figure 4B**) the degradation rate ranged from 49.39 to 99.11%, depending on the kind of EO and pH value. Similarly to the effect observed at 20◦C, it was found that the pH value at 4◦C significantly influenced the ZEA degradation by all essential oils at both concentrations, excluding palmarosa and eucalyptus oils at a concentration of 0.5 µg/mL. Taking into account the degree of ZEA reduction depending on the pH value, it was noted that the highest degradation rate at the initial concentration of 0.5 µg/mL (**Figure 4A**) was observed at pH 9 for all essential oils. The smallest decrease of toxin amount showed greater variation. Four essential oils – white grapefruit, pink grapefruit, eucalyptus and palmarosa were least effective at pH 3, while three of them – cinnamon leaf, cinnamon bark and lemon oils – were least effective at pH 6. At the initial concentration of 5 µg/mL (**Figure 4B**) the influence of essential oils on the degradation rate of ZEA was more variable than that observed at 20◦C. The highest effectiveness in reducing the toxin amount was observed at pH 6 for all essential oils except for white grapefruit, which was most effective at pH 9. The weakest effect of essential oils on ZEA degradation was observed at pH 9 in the case of cinnamon oils (leaf and bark) as well as lemon and palmarosa oils, while grapefruit oils (pink and white) and eucalyptus oil were least effective at pH 3.

Considering the effect of temperature on the level of ZEA degradation, it is worth underlining that the highest percentage reduction was observed at 20◦C. In the majority of samples the toxin degradation effectiveness was higher by a few percent at 20◦C, when compared to 4◦C.

Considering the mycotoxin concentration and pH value, ZEA degradation abilities of essential oils may be ranked in decreasing order as shown in **Table 1**. The highest degradation rate in samples at ZEA concentration of 0.5 µg/mL was achieved for cinnamon bark oil (82.45–90.93%), while in samples with 5 µg/mL of ZEA palmarosa oil caused the highest metabolite reduction (96.75–99.29%).

## DISCUSSION

Mycotoxins generate several problems because due to their physico-chemical stability the metabolites are transported and introduced into successive steps of the food chain. Therefore elimination of mycotoxins is of prime concern in research and food safety programs. Methods of detoxification of the above toxic metabolites include physical, chemical and biological treatment. One effective methods for ZEA detoxification is microbial degradation by species of Rhodococcus, Clonostachys, Trichosporon or Nocardia (El-Sharkawy and Abul-Hajj, 1988a; Duvick and Rood, 1998; Molnar et al., 2004). Microorganisms

FIGURE 3 | The effect of the pH and zearalenone concentration on the toxin degradation at temperature 20◦C. The experiment was conducted 72 h at the toxin concentration level 0.5 µg/mL (A) and 5 µg/mL (B) and 100 µl/mL of essential oil. Amount of zearalenone was determined by HPLC method. Data were analyzed by Student's t-test at p < 0.05 (a, b, c – significantly different in column).

FIGURE 4 | The effect of the pH and zearalenone concentration on the toxin degradation at temperature 4◦C. The experiment was conducted 72 h at the toxin concentration level 0.5 µg/mL (A) and 5 µg/mL (B) and 100 µl/mL of essential oil. Amount of zearalenone was determined by HPLC method. Data were

analyzed by Student's t-test at p < 0.05 (a, b, c – significantly different in column).



may transform toxins to metabolites of lower estrogenic properties, e.g., Cunninghamella bainieri converts ZEA to 2,4 dimethoxyl ZEA (El-Sharkawy and Abul-Hajj, 1988b) and Rhizopus arrhizus transforms ZEA to ZEA 4-sulfate (El-Sharkawy et al., 1991). However, it is worth underlining that biotransformation of ZEA may metabolize the molecule to compound(s) of comparable or even higher estrogenic effects. For example, Rhizopus sp. and Alternaria alternata, chemically transform ZEA to α- and β-zearalenol, without a decrease in oestrogenicity (Kamimura, 1986; El-Sharkawy et al., 1991), while El-Sharkawy and Abul-Hajj (1988b) reported conversion of ZEA by Aspergillus ochraceous to α- and β-zearalanols with estrogenic properties at a higher level when compared to ZEA. There are limited data concerning the possibilities of mycotoxin degradation by essential oils. In our studies 8 of 11 tested essential oils reduced the content of ZEA in the model experiment, with the highest effectiveness demonstrated by lemon, grapefruit, eucalyptus and palmarosa oils. Lavender, thymic and rosemary oils did not degrade the examined toxin. Similarly, Xing et al. (2014) reported that six of seven oils degraded fumonisin B<sup>1</sup> and the most effective was cinnamon oil, followed by citral, eugenol, eucalyptus and camphor oil, while peppermint oil was ineffective in their studies.

We have observed that effectiveness of ZEA degradation depends on several factors including incubation time, temperature, pH and both essential oil and toxin concentrations. Temperature and pH had a significant effect on ZEA (at ZEA concentration of 5 µg/mL) degradation by all tested essential oils except palmarosa. In the case of lower ZEA content the influence of incubations conditions was more complex. Xing et al. (2014) reported that incubation time as well as temperature and concentration of cinnamon oil influenced the degradation of fumonisin B1. The authors stated that the degradation rate of toxin increased gradually with the increasing temperature in the range of 20–35◦C. The degradation rate of fumonisin B<sup>1</sup> was incubation time dependent and increased gradually with incubation time (up to 120 h). In our studies, a higher degradation rate was also observed at a higher temperature. Moreover, we also found that degradation effectiveness increased with incubation time (up to 72 h). The essential oil concentration effect, in contrast was dependent on the kind of the oil. This parameter was significant for cinnamon bark, white grapefruit, lemon and palmarosa oil. Similar observations were reported for the cinnamon oil concentration effect in the research of Xing et al. (2014). Their results showed that the content of the essential oil was higher the greater was the toxin degradation rate. Finally, the authors found that the optimal conditions for reduction of fumonisin B<sup>1</sup> were temperature 30◦C, incubation time 120 h and concentration of essential oil 280 µg/mL. The final toxin content was decreased by 94.06% (from 15.03 to 0.89 µg/mL).

In our research there were also observed significant differences in the degradation rate between tested essential oils. This could be due to different composition of essential oils and possible relationships (e.g., synergistic or antagonistic effects) between the components. Essential oils are very complex mixtures of different constituents and may contain 20 – 60 compounds. Their antimicrobial and antioxidant properties are usually associated with substances such as α-pinene, β-pinene, γ-terpinene and p-cymene, carvacrol, eugenol, geraniol, thymol, vanillic acid, camphor, linalool trans-cinnamaldehyde and trans-cinnamic acid (Marino et al., 2001; Delaquis et al., 2002; Si et al., 2006; Sonboli et al., 2006; Goze et al., 2009). Generally, components are included in two main groups with one formed on the basis of terpene and terpenoid chemical structure and the second formed on the basis of aromatic and aliphatic compounds (Bakkali et al., 2008). Authors often describe the relationship between structure of compounds dominating in the essential oils and their antifungal activity. For example, Bluma et al. (2008) reported that antifungal activity of tested oils was associated with monoterpenic phenols, mainly thymol eugenol and carvacrol. Moreover, it has been reported that antifungal properties are not related to one specific component but are associated rather with a

mixture of constituents included in the essential oil (Ranasinghe et al., 2002; Hashem et al., 2010). Considering that there is a lack of data concerning the mode of action of essential oils as degrading factors and participation of individual components, the investigation presented here is planned to be continued.

#### CONCLUSION

To our best knowledge this is the first report on degradation of ZEA by essential oils. Our results reveal that essential oils may reduce the content of toxin under a wide spectrum of process conditions. The results presented here as well as the literature data indicate that antifungal as well as antimycotoxigenic activity of essential oils should not be generalized based on these trials. It is difficult to indicate which essential oil has the highest degradation properties due to the variety of factors influencing the effectiveness of the process. However, based on the obtained results it can be stated that palmarosa EO showed good properties in reducing ZEA content. In contrast, in most cases the lowest ZEA degradation properties were demonstrated by pink grapefruit oil.

To obtain a general picture of essential oils' mycotoxin degradation potential and abilities, larger scale laboratory studies

#### REFERENCES


must be conducted involving a large range of mycotoxins as well as detailed information on quantitative and qualitative composition and activity of individual components of oils. It would be useful to expand the range of essential oils including their various composition, varying their number and ratio, to draw more precise conclusions more useful in practice. Complex studies are needed to fully understand the mechanism of action of essential oils as well as the mode of their application considering protection of foods and feeds.

Essential oils seems to be a good alternative to other detoxification method since they are recommended as safe substances by the Food and Drug Administration (FDA, 2015). Moreover, these volatile compounds are recognized as part of the concept of "green pesticides," which favors the use of environmentally friendly, natural substances rather than synthetic chemical compounds.

#### AUTHOR CONTRIBUTIONS

Design of the work: DG, AW, KJ, AP, KM, and PG. Conducting experiments: AP, KJ, and KM. Interpretation of data: AP, KJ, KM, DG, AW, and PG. Drafting the work: AP, KJ, KM, DG, AW, and PG. Final approval: DG, AW, and PG.


a short-term carcinogenesis model in rat liver. Toxicology 171, 161–173. doi: 10.1016/S0300-483X(01)00573-X


oxidative deterioration of agri-food commodities – Potentials and challenges. Food Control 47, 381–391. doi: 10.1016/j.foodcont.2014.07.023


Zinedine, A., Soriano, J. M., Molto, J. C., and Manes, J. (2007). Review on the toxicity, occurrence, metabolism, detoxification, regulations and intake of zearalenone: an oestrogenic mycotoxin. Food Chem. Toxicol. 45, 1–18. doi: 10.1016/j.fct.2006.07.030

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2016 Perczak, Ju´s, Marchwinska, Gwiazdowska, Wa ´ ´skiewicz and Golinski. This is an open-access article distributed under the terms of ´ the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Antagonistic activity of *Ocimum sanctum* L. essential oil on growth and zearalenone production by *Fusarium graminearum* in maize grains

*Naveen K. Kalagatur1, Venkataramana Mudili2\*, Chandranayaka Siddaiah3, Vijai K. Gupta4, Gopalan Natarajan5, Murali H. Sreepathi1\*, Batra H. Vardhan1 and Venkata L. R. Putcha2*

*<sup>1</sup> Food Microbiology Division, Defence Food Research Laboratory, Siddarthanagar, India, <sup>2</sup> Toxicology and Immunology Division, DRDO-Bharathiar University Centre for Life Sciences, Coimbatore, India, <sup>3</sup> Department of Biotechnology, University of Mysore, Mysore, India, <sup>4</sup> Discipline of Biochemistry, National University of Ireland Galway, Galway, Ireland, <sup>5</sup> Food Biotechnology Division, Defence Food Research Laboratory, Siddarthanagar, India*

The present study was aimed to establish the antagonistic effects of *Ocimum sanctum* L. essential oil (OSEO) on growth and zearalenone (ZEA) production of *Fusarium graminearum*. GC–MS chemical profiling of OSEO revealed the existence of 43 compounds and the major compound was found to be eugenol (34.7%). DPPH free radical scavenging activity (IC50) of OSEO was determined to be 8.5 μg/mL. Minimum inhibitory concentration and minimum fungicidal concentration of OSEO on *F. graminearum* were recorded as 1250 and 1800 μg/mL, respectively. Scanning electron microscope observations showed significant micro morphological damage in OSEO exposed mycelia and spores compared to untreated control culture. Quantitative UHPLC studies revealed that OSEO negatively effected the production of ZEA; the concentration of toxin production was observed to be insignificant at 1500 μg/mL concentration of OSEO. On other hand ZEA concentration was quantified as 3.23 μg/mL in OSEO untreated control culture. Reverse transcriptase qPCR analysis of ZEA metabolic pathway genes (*PKS4* and *PKS13*) revealed that increase in OSEO concentration (250–1500 μg/mL) significantly downregulated the expression of *PKS4* and *PKS13*. These results were in agreement with the artificially contaminated maize grains as well. In conlusion, the antifungal and antimycotoxic effects of OSEO on *F. graminearum* in the present study reiterated that, the essential oil of *O. sanctum* could be a promising herbal fungicide in food processing industries as well as grain storage centers.

Keywords: *O. sanctum* essential oil, GC–MS, *F. graminearum*, micro-well dilution method, scanning electron microscope, zearalenone, reverse transcriptase qPCR, UHPLC

#### *Edited by:*

*Agnieszka Waskiewicz, Poznan University of Life Sciences, ´ Poland*

#### *Reviewed by:*

*Ian A. Cleary, University of Tennessee at Martin, USA Zonghua Wang, Fujian Agriculture and Forestry University, China*

#### *\*Correspondence:*

*Venkataramana Mudili, Toxicology and Immunology Division, DRDO-Bharathiar University Centre for Life sciences, Bharathiar University Campus, Coimbatore 641046, Tamil Nadu, India ramana.micro@gmail.com; Murali H. Sreepathi, Food Microbiology Division, Defence Food Research Laboratory, Siddarthanagar, Mysore 570011, Karnataka, India drmuralihs@gmail.com*

#### *Specialty section:*

*This article was submitted to Fungi and Their Interactions, a section of the journal Frontiers in Microbiology*

*Received: 11 May 2015 Accepted: 17 August 2015 Published: 03 September 2015*

#### *Citation:*

*Kalagatur NK, Mudili V, Siddaiah C, Gupta VK, Natarajan G, Sreepathi MH, Vardhan BH and Putcha VLR (2015) Antagonistic activity of Ocimum sanctum L. essential oil on growth and zearalenone production by Fusarium graminearum in maize grains. Front. Microbiol. 6:892. doi: 10.3389/fmicb.2015.00892*

**Abbreviations:** OSEO, *Ocimum sanctum* L. essential oil; SDA, sabouraud dextrose agar; SDB, sabouraud dextrose broth; ZEA, zearalenone.

#### Introduction

The existence of fungal species and their toxic secondary metabolites viz. mycotoxins in food and feed commodities is a major concern for microbiological safety and food security (Pitt and Hocking, 2009). Fungi and their mycotoxins could cause massive financial loss to global economy, because of their deleterious effects in cereal crops besides humans and farm animals (Rocha et al., 2014). Among the toxigenic and pathogenic fungal species, *Fusarium graminearum* has been given special attention in agriculture commodities due to its ability to grow in diverse climatic conditions, and to produce different mycotoxins including ZEA and type-B trichothecenes (Morgavi et al., 2007; Bernhoft et al., 2012). ZEA also known as F-2 or RAL is a temperature stable, persistent, moderately water soluble, nonsteroidal estrogenic mycotoxin (Zinedine et al., 2007). As one of the most prevalent mycotoxins, its presence has frequently been analyzed in agricultural as well as environmental products (Zinedine et al., 2007). In a recent study, Cano-Sancho et al. (2012) and Tomoya et al. (2014) reported that, cereals and food grain samples have been found to be heavily contaminated with ZEA (concentration ranging from 3.1 to 5.9 μg/kg) and at the maximum concentration of 153 μg/kg ZEA was reported in Job's tears product. In another study, Pleadin et al. (2012) reported a maximum concentration of 5.11 mg/kg ZEA in maize.

Zearalenone binds to estrogen receptors (ERs) ending up in estrogenicity, which occasionally brings hyperestrogenism in livestock and humans, particularly in females (Zinedine et al., 2007). The toxic effects of ZEA as noticed both in the laboratory and household animals include endocrine interruption leading to induction of the expansion of estrogen-sensitive cells and tissues, abnormal feminization of male gonads or reproductive system disorders, skeletal distortion, cancer, weakening of bones, and myelofibrosis (Zinedine et al., 2007). In our recent study Venkataramana et al. (2014) reported the neurotoxic and genotoxic effects of ZEA in human neuronal (SH-SY5Y) cells. Also in another study Richard (2007) reported genotoxic role of ZEA by *in vitro* methods through SOS repair, chromosomal aberration and sister chromatid substitution. Owing the to its estrogenic and carcinogenic effects of ZEA to humans and other farm animals, International Agency for Research on Cancer (IARC) has defined ZEA as a Group 3 carcinogen (IARC, 1999). Due to the above said implications, many regulatory agencies proposed the maximum permissible limits for ZEA in several food matrices. The European Union (EU) has established allowed limits for ZEA in unprocessed cereals as 100 μg/kg excluding maize and the permitted level in unprocessed maize was 350 μg/kg (European Commission, 2007). The Joint FAO/WHO Expert Committee on Food Additives (JECFA) has set up a Provisional Maximum Tolerable Daily Intake (PMTDI) of 0.5 μg/kg body weight (JECFA, 2000).

To date, many synthetic antifungal agents have been licensed for successive control of *F. graminearum* as well as other fungi. Unfortunately, the application of synthetic antifungal agents in agricultural commodities was responsible for a multitude of negative health impacts in livestock and humans and also resulted in upsurge of resistant organisms as well (da Cruz Cabral et al., 2013). Therefore, there is a need to propose proper food grain management practices, including the application of herbal antifungal and mycotoxin controlling agents, thus to reduce the growth of toxigenic *F. graminearum* as well as the production of ZEA in agricultural commodities. In this instance, wide varieties of secondary metabolites produced by plants are offering options for synthetic antifungal agents because of their bio-degrading nature as well as non-toxic to the environment. Essential oils obtained from the plants have always been an excellent source of antioxidant compounds such as polyphenols, flavonoids, etc., which have been believed to be the basis of their antifungal capabilities (da Cruz Cabral et al., 2013). Essential oils are aromatic liquids attained through hydrodistillation from the plant material and they are constituted by a great diversity of compounds confers many advantages, such as possessing unique modes of action like antioxidant, antimicrobial, and anticancerous properties against many infectious pathogens and life-threatening diseases (Raut and Karuppayil, 2014).

*Ocimum sanctum*, also known as the Holy Basil or Tulsi is widely used as a medicinal plant in Ayurveda. The plant has been traditionally used in management of common cold, asthmatic illnesses, skin problems, urogenital infections, digestive, neurological and cardiovascular disorders (Gupta et al., 2002). The chemical constitution of OSEO includes a variety of active compounds, including eugenol, methyl eugenol, methyl chavicol, α-terpineol, germacrene D, β-caryophyllene, camphor, camphene, β-ocimene, β-elemene, linalool, 1,8-cineole etc. (Kothari et al., 2005; Khan et al., 2010; Joshi, 2013). Kumar et al. (2010, 2013) reported the inhibitory activity of OSEO on growth of *Aspergillus flavus* NKDHV8 and *A. flavus* LHPRS7 isolated from raw materials of *Rauvolfia serpentina*, respectively. Sethi et al. (2013) reported strong inhibitory effect of OSEO (MIC of 62.5 μg/mL) against *Rhizoctonia solani* and Khan et al. (2010) reported the antifungal activity of OSEO on *Candida sp.* by the mechanism of disrupting ergosterol biosynthesis and membrane integrity. In contrast, the application of OSEO as the antifungal agent in management of toxigenic *F. graminearum* and ZEA production are still unexplored. Also, no previous study investigated the applicability of OSEO on grain samples.

The objective of the present study is to establish the antagonistic effects of OSEO on growth and ZEA production of *F. graminearum*. OSEO was extracted from aerial parts of *O. sanctum* L. by hydrodistillation method and chemical profile was carried out by GC–MS method. Antioxidant and antifungal activities of OSEO were carried out by DPPH free radical scavenging activity, and micro-well dilution and scanning electron microscope methods, respectively. Effect of OSEO on ZEA production was analyzed by Rt-qPCR and UHPLC methods from broth culture of *F. graminearum*.

#### Materials and Methods

#### Chemicals and Reagents

All the chemicals (analytical grade) and solvents (HPLC grade) were procured from Merck (Bangalore, India). Plasticware used in the study were obtained from Eppendorf (Hamburg, Germany). ZEA standard was obtained from Sigma-Aldrich (Bangalore, India) and stock solution of ZEA was prepared in acetonitrile and stored at −20◦C until use. Sabouraud Dextrose media, peptone and nystatin were purchased from HiMedia (Mumbai, India).

#### Plant Material Collection and Essential Oil Extraction

The aerial parts of *O. sanctum* were collected from Mysore, Karnataka state, India. The identification and verification of plant was carried at the Botanical Survey of India (Southern Regional Centre, Coimbatore, India) and the voucher was safeguarded at Food Microbiology Division, Defence Food Research Laboratory (Mysore, India). The collected material was air-dried in the shade at 37◦C for 4 weeks and used for analysis. The essential oil was extracted from 250 g of the dried plant materials by hydrodistillation using a Clevenger-type device in accordance with the technique approved by the European Pharmacopoeia (Council of Europe, 1997). The oil gathered was separated and dried over anhydrous sodium sulfate to remove water and further stored at 4◦C in the dark until use.

#### GC–MS Analysis

The GC–MS analysis of OSEO was carried out on PerkinElmer Clarus 600 C (Waltham, MA, USA) analytical system equipped with DB-5MS (30 m × 0.25 mm; 0.25 μm film thickness) merged silica capillary column and attached to flame-ionization detector (FID). OSEO was diluted in acetone (10 μL/mL) and 1 μL solution was injected in a split-mode (1:30). Working conditions were as follows: carrier gas was He (1 mL/min); temperatures were set as follows: injector at 250◦C and sensor at 280◦C, whilst the column temperature was linearly scheduled 40–280◦C at 4◦C/min. Mass spectra were documented in EI mode (70 eV) with a range of m/z 40–450. Turbo Mass software application was adapted to operate and acquire data from the GC–MS. The detection of the individual components was achieved by comparison of their mass spectra (MS) with those from accessible libraries (NIST/Wiley) and experimentally determined retention indices (RI) with data from the literature (Adams, 2007). The percentage of the constitution of the individual components was derived out from the peak areas, devoid of correction factors.

#### Determination of Antioxidant Activity by DPPH**∗** Assay

2,2-diphenyl-1-picrylhydrazyl (DPPH) is a constant free radical which responds with compounds, which tend to be able to donate a hydrogen atom (HAT). Therefore, the hydrogen donating capability of OSEO to DPPH free radical (DPPH∗) was determined from the change in absorbance at 517 nm in accordance with the method of Ojeda-Sana et al. (2013) with slight modifications. For the radical scavenging measurements, 1 mL of methanol, 1 mL of 0.1 M acetate buffer (pH 5.5), 0.5 mL of a 250 μM methanolic solution of DPPH∗ were blended with various concentrations of OSEO (2– 18.5 μg/mL). The control was made from the reaction blend

without test sample and quercetin was considered as the reference sample. Subsequently the absorbance was determined using BioSpectrometer (Eppendorf, Germany) after 30 min of incubation at 37◦C in the dark. The result was stated as IC50 (μg/mL), which means that the quantity of sample required to reduce the absorbance of DPPH by 50%. DPPH∗ scavenging activity was calculated by the formula,

$$\% \text{Scavenging activity} = \frac{\text{A(control)} - \text{A(test)}}{\text{A (control)}} \times 100$$

Where, A (control) was the absorbance of the control (without test sample) and A (test) was the absorbance of the test sample.

#### Antifungal Activity

Zearalenone producing *F. graminearum* (MTCC, 1893) was obtained from the Microbial Type Culture Collection and Gene Bank, Chandigarh, India (MTCC). *F. graminearum* was grown on SDA for 7 days at 28◦C and spores were recovered using peptone water containing 0.001% Tween 80 with soft scrape. The spore number was determined using haemocytometer and spore suspension was adjusted to 1 <sup>×</sup> 106 per mL. Antifungal activity of OSEO was determined by micro-well dilution method and further it was validated quantitatively by scanning electron microscope observation.

#### Micro-Well Dilution Method

Minimum inhibitory (MIC) and minimum fungicidal (MFC) concentrations were determined by micro-well dilution technique in 96 well microtitre plates with minor modifications (Clinical and Laboratory Standards Institute [CLSI], 2008; Vieira et al., 2014). SDB was used as the media in the well to which 0.001% Tween 80, different concentrations of the OSEO and a volume of 10 μL spore suspension (1 × 106 spores/mL) were added, and the final volume was 100 μL per well. The wells without OSEO were referred as control and microplates were incubated for 3 days at 28◦C. The minimum concentration without detectable growth was determined as the minimal concentration which absolutely inhibited fungal growth (MIC). A volume of 10 μL from each well was inoculated into the SDA plates and incubated at 28◦C for 3 days, and the minimum concentration with no detectable growth was determined as the MFC, specifying 99.5% killing of the original inoculum in comparison to nystatin, used as a positive control.

#### Effect of OSEO on Spore Germination

Effect of OSEO on spore germination of *F. graminearum* was analyzed by the method of Rana et al. (1997) with minor modifications. A volume of 10 μL fungal spore suspension (1 <sup>×</sup> 106 spores per mL) was inoculated on SDA slides containing different concentrations of OSEO (100–1800 μg/mL) and incubated at 28◦C for 24 h. SDA slide alone with fungal spores and without OESO were considered as control. Following incubation period, each slide was fixed with lactophenol-cotton blue and observed under microscope (Leica DM1000 LED, Leica Microsystems, Wetzlar, Germany) for the spore germination. About 200 spores were examined from each slide and percentage of spore germination was calculated by the formula,

% Spore germination = ST*/*SC × 100

Where, SC was number of spores germinated in control and ST was number of spores germinated in test.

#### Scanning Electron Microscopic Observation

To know the effects of OSEO on mycelial and spore structure of *F. graminearum*, scanning electron microscope (SEM) analysis was performed according to the method of Yamamoto-Ribeiro et al. (2013) with minor modifications. A seven day old mycelia was inoculated aseptically into SDA dishes that contained 1250 and 1800 μg/mL concentration of OSEO and the dishes were incubated at 28◦C for 7 days. The control was performed in SDA medium without OSEO. After the incubation period, mycelia disk of 1 cm<sup>2</sup> was collected and rinsed in phosphatebuffer saline (0.01 M) and fixed with 2.5% glutaraldehyde in 0.1 M sodium cacodylate buffer, pH 6.5 and dehydrated with gradient ethanol (20, 40, 70, 90, and 100%, keeping the mycelia for a longer duration in 100%). The sample was pasted on dual side glue carbon tape and it was fixed to the surface of aluminum stubs. Further, the stubs were exposed toward criticalpoint dry out in CO2 and sputter-coated with gold to increase its conductivity. The morphological quality of the mycelia was observed under scanning electron microscope (FEI, USA) at 20.0 KV in environmental mode.

#### Determination of Antimycotoxic Activity of OSEO in Liquid Cultures

Different concentrations of OSEO including, 250, 500, 1000, 1500, and 2000 μg/mL were added to each 250 mL Erlenmeyer flask individually that contained 100 mL of SDB. A volume of 10 <sup>μ</sup>L fungal suspension (1 <sup>×</sup> <sup>10</sup><sup>6</sup> spores/mL) of 7-dayold culture was inoculated in these flasks and the flask without OSEO was considered as untreated control. Inoculated flasks were incubated under shaking condition (140–160 rpm) at 28◦C for 14 days.

#### Determination of Mycelial Biomass

Following the incubation period the culture media was separated from fungal biomass by filtering through Whatman no. 1 paper and the broth was used for ZEA determination. The fungal mycelia was washed twice with sterile distilled water and 10 mg of mycelia was employed for RNA extraction, and leftover mycelia was dried out on pre-weighed Whatman no. 1 filter paper at 60◦C for 24 h and weighed (Denver instruments, India).

#### UHPLC Determination and Quantification of Zearalenone

Detection and quantification of ZEA were carried out as reported by the method of Ibáñez-Vea et al. (2011) with slight modifications. An equivalent quantity of acetonitrile was added to each culture broth and blended thoroughly for 30 min. Subsequently, the sample was centrifuged for 12 min at 6000 rpm and 15 mL of the supernatant was transferred through an immunoaffinity column of ZEA (Vicam, USA), pre-conditioned under 10 mL of phosphate buffer saline (PBS). After the sample had passed away, the column was washed with 5 mL of PBS and 10 mL of distilled water. Finally, the column was dried out with air and ZEA was eluted with 5 mL of acetonitrile, after retaining in contact between acetonitrile and column antibodies for 5 min. The extract was dried out completely over water bath at 60◦C and final residue was redissolved in 1 mL of acetonitrile and filtered through 0.22 μm syringe filter. The filtrate was used for the UHPLC determination and quantification of ZEA.

The Nexera UHPLC system (Shimadzu, Kyoto, Japan) attached with the column C18, 5 μm, 250 × 4.6 mm (Phenomenex, USA) was employed for detection and quantification of ZEA carried out in reverse-phase with a fluorescence detector that was adjusted at 334 nm excitation and 450 nm emission. The mobile phase was acetonitrile-water (50:50 v/v) with a flow rate of 1 mL/min. A standard ZEA (100 ng– 500 μg/mL) was used to construct a five-point calibration curve of peak areas versus concentration. The injection volume was 25 μL for both the standard solution and sample extracts. The sensing limitation of the technique was 100 ng/mL.

#### Reverse Transcriptase qPCR (Rt-qPCR) Analysis of Zearalenone Metabolic Synthesis Genes

Reverse transcriptase qPCR (Rt-qPCR) evaluation was done to analyze the impact of OSEO on gene expression of *PKS4* and *PKS13*, which are involved in ZEA biosynthesis in *F. graminearum* (Kim et al., 2005; Gaffoor and Trail, 2006) and *GAPDH* was used as endogenous reference gene. Primers were designed against target genes using the GeneRunner software version 5.0.47 Beta (**Table 1**) and synthesized primer sequences were obtained from Sigma-Aldrich (Bangalore, India). Total RNA was extracted using RNA easy plant Mini kit following manufacturer's guidelines (Qiagen, USA). Briefly, mycelia were flash-frozen in liquid nitrogen and grounded into a fine powder with a porcelain mortar. The total RNA was extracted and quantified by NanoDrop 8000 Spectrophotometer (Thermo Scientific, USA). The analysis of Rt-qPCR was carried out in the Light cycler 480 (Roche, USA) using iScript One-Step RT-PCR Kit with SYBR Green (BIO-RAD). In concise, 50 μL volume of reaction mixture consists of 25 μL of 2X SYBR Green RT-PCR reaction mix, 1 μL of iScript reverse transcriptase for one-step RT-PCR, 1 μL of primer (450 nM), 1 μL of template RNA (100 ng) and 22 μL of nuclease-free water (PCR grade). The thermal conditions for reaction include 10 min of cDNA synthesis at 50◦C for 1 cycle, 5 min of polymerase activation



at 95◦C and followed by 35 cycles of PCR at 95◦C for 10 s, 60◦C for 30 s for combined annealing and extension. For each and every PCR product, an individual narrow peak was attained through melting curve analysis of the distinct temperatures. The relative quantification levels of expression had been quantified making use of second derivative maximum analysis with the determination of the crossing points for every single transcript. Crossing point values for each gene were normalized to the particular crossing point values for the reference gene *GAPDH*. Data are shown as normalized ratios of genes together with standard error by means of Roche Applied Science E-Method (Tellman and Olivier, 2006).

#### Antimycotoxic Property Evaluation of OSEO Onto Maize

Antimycotoxic efficacy of OSEO was assessed directly onto *F. graminearum* inoculated maize grains. The seeds were sterilized by autoclave and dried in hot-air oven at 60◦C for 2 h. One 100 g of sterilized maize grains was treated with various concentrations (250, 500, 1000, 1500, and 2000 μg/g) of OSEO in 500 mL conical flask and a volume of 10 μL fungal spore suspension (1 <sup>×</sup> 106 spores/mL) of 7-day-old culture was inoculated in each conical flask. The grains not treated with OSEO were referred as control and incubated for a period of 14 days at 28◦C in a dark condition. Following the incubation period, total RNA was extracted from fungal mycelia and Rt-qPCR evaluation for *PKS4* and *PKS13* genes were carried out as mentioned in "Reverse Transcriptase qPCR (Rt-qPCR) Analysis of Zearalenone Metabolic Synthesis Genes." Further, maize grains were ground into fine powder and dissolved in 500 mL of acetonitrile and centrifuged at 6000 rpm for 30 min. A volume of 15 mL supernatant was transferred through ZEA specific immunoaffinity column and ZEA was quantitatively determined by UHPLC as mentioned in "UHPLC Determination and Quantification of Zearalenone."

#### Statistical Analysis

All experiments were carried out in six independent replicates and the results were statistically evaluated applying one-way ANOVA for multiple comparisons adapted by the Tukey's test. Differences were considered statistically significant at a value of *p <* 0.05. Statistical program GraphPad Prism 5.0 (GraphPad Software, Inc., USA) was used to draw graphs.

#### Results and Discussion

#### Chemical Composition

Based on the dry weight calculation, the yield of OSEO was determined as 1.79% (w/w). Chemical profile of OSEO was revealed by GC-MS analysis and a total of 43 compounds were identified accounting to 98.03% of the total weight (**Table 2**). Among the identified compounds, eugenol (34.7%) was the major compound together with other active compounds with varied concentrations such as thymol (2.98%), linalool (4.94%), β-phellandrene (4.71%), α-phellandrene (3.79%), limonene


<sup>a</sup>*Compounds are indexed in order of their elution.*

<sup>b</sup>*Retention indices of compounds determined based on n-alkanes (C-9–C-24) on DB-5MS column.*

<sup>c</sup>*Retention indices of compounds on DB-5 column in accordance with Adams (2007).*

(3.73%), germacrene D (2.89%), β-pinene (2.59%) and *trans*pinocarveol (2.37%). Kothari et al. (2005) that methyl eugenol was the major compound (72.5, 75.3, 83.7, and 65.2%) and β-caryophyllene was the second most dominant compound (5.5, 6.4, 2.7, and 12.0%) in the essential oils extracted from whole herb, leaf, stem, and inflorescence of the *O. sanctum* L. from southern India. On the other hand, Kumar et al. (2010) with their study on chemical composition of OSEO from North Indian plants reported that eugenol (61.30%) was a major compound. Recently Verma et al. (2013) explored the chemical profile diversity of OSEO from Indian flora and distinguished the profile into three chemotypes i.e., eugenol, methyl eugenol, and caryophyllene. From the present study, it can be deduced that the chemotype of the *O. sanctum* L. plant in the present study was eugenol chemotype. Our results were also supported by the available literature with respect to chemical composition as there was no new compound observed. The significant differences in the concentration of the determined compounds in comparison with existing reports could be explained by the variation in climatic conditions and harvesting period, luminosity as well as oil extraction method (Kothari et al., 2004; Gobbo-Neto and Lopes, 2007; Verma et al., 2013).

#### Antioxidant Activity

*In vitro* antioxidant potential of OSEO was carried out by DPPH free radical scavenging assay based on single electron (SET) and a HAT transfer reactions (Huang et al., 2005). In the present study, free radical scavenging activity of the OSEO was directly proportional to the OSEO concentration (**Figure 1**) and antioxidant potential of the OSEO showed a higher free radical scavenging activity compared with that of reference antioxidant quercetin. The value for 50% scavenging activity (IC50) of OSEO was 8.5 μg/mL whereas quercetin was 12 μg/mL. Joshi (2013) reported antioxidant activity (IC50) of OSEO by DPPH<sup>∗</sup> assay as 219.16 ± 1.01 μg/mL. Interestingly, in the present study reported antioxidant potential of OSEO was quite high compared to earlier report of Joshi (2013). The chemical profile of OSEO in the present study revealed that, in addition to eugenol many other active compounds such as α-thujene, α-pinene, camphene, camphor, limonene, α-phellandrene, β-phellandrene, linalool, linalyl acetate, *trans*-pinocarveol, pinocarvone, germacrene D, β-caryophyllene and thymol were reported, which were absent

in the report of Joshi (2013). This might be the reason for enhanced antioxidant activity of OSEO in the present study. Trevisan et al. (2006) also reported DPPH∗ scavenging activity (IC50) of OSEO as 0.26 μL/mL. The antioxidant potential of essential oil is mainly associated with the presence of phenolic compounds participating in SET and/or HAT reactions. This may be due to the extent of structural conjugation and the presence of electron-donating and electron-accepting substituents on the ring structure of phenolic compounds (Ložiene et al., 2007 ˙ ). Earlier *in vitro* studies proved that oxidant stressors enhanced the mycotoxin biosynthesis and the use of plant derived antioxidant supplements are effective in down regulating the production of mycotoxin (da Cruz Cabral et al., 2013). Similar results were obtained by Kumar et al. (2010) in the case of effect of OSEO in controlling growth and aflatoxin B1 in *Aspergillus flavus*. The observed high antioxidant activity of OSEO indicated its applicability in affecting the mycotoxin production by fungi.

#### Antifungal Activity

The antifungal property of OSEO although was highly exploited in previous studies, no reports ever existed on antifungal activity on *F. graminearum* and ZEA production. In the present study OSEO showed MIC and MFC at a range of 1250 and 1800 μg/mL, respectively against *F. graminearum*. This was significantly higher (*p <* 0.05) when compared to reference drug nystatin that showed MIC and MFC activity at 2200 and 3000 μg/mL, respectively (**Figure 2A**). Alternatively, we also carried out fungal spore germination susceptibility assay to support the antifungal activity of OSEO. The results of the spore germination studies reveled that studied concentrations of OSEO showed significant control in *F. graminearum* spore germination on SDA compared to the untreated cultures (**Figure 2B**). A decrease in spore germination was observed with increasing the concentration of OSEO and 100% inhibition of spore germination was observed at 1800 μg of OSEO.

The variability and diversity of the chemical composition and the wide spectrum antimicrobial activities of OSEO, as supported by the present study and previous literature, could be indicative for further studies to be undertaken to delineate the specific compound in OSEO, which might be a major responsible factor for the aforementioned activities. In Campaniello et al. (2010) reported the role of eugenol in antifungal activity on *Penicillium*, *Aspergillus*, and *Fusarium* species. These findings further supported by Dalleau et al. (2008) and da Cruz Cabral et al. (2013), who reported that, phenolic compounds have an ability to disrupt the lipid bilayer of cell membrane and mitochondria, thus to cross the cell membrane and interacting with the enzymes and proteins of the membrane leads to functional alterations in the cell as well as mitochondrial dysfunction this in turn lead to apoptotic cell death.

In the present study, effect of OSEO on micro morphology of *F. graminearum* was confirmed and validated by SEM observations at MIC (1250 μg/mL) and MFC (1800 μg/mL). Significant effects of OSEO on micro morphology of *F. graminearum* mycelial structure as well as spore structure were observed. The control hyphae showed healthy morphology with

smooth, turgid, homogenous surface without any discernible change (**Figure 3A**). On the other hand, morphology of OSEO treated hyphae underwent noteworthy alterations and showed evident modifications in both apical regions and throughout the length of hyphae. The cell wall displayed an irregular surface appearance with craters and protuberances. In addition, hyphae were severely collapsed and squashed due to lack of cytoplasm and a few small vesicles were observed on the apical surface of the mycelia (**Figures 3B,C**). Spores treated with OSEO at concentrations of MIC and MFC values exhibited wrinkled, disrupted, and dispersed appearance compared to the untreated sample (**Figures 3D–F**). Earlier study by Wang et al. (2010) on effect of eugenol on *Botrytis cinerea* morphology revealed similar damages to the structure of hyphae. Also, Yamamoto-Ribeiro et al. (2013) observed similar morphological aberrations in *Zingiber officinale* essential oil (ZOEO) treated *F. verticillioides* hyphae. When the composition of ZOEO was analyzed, 7.73% of the oil was composed of β-phellandrene, a significant compound (4.71%) observed in OSEO also. These observations suggest that in the present study, the damage may be mediated via the most abundant component of the OSEO viz., eugenol and another significant compound, β-phellandrene.

#### Antimycotoxic Activity of OSEO on Liquid Cultures

The ultimate aim of the present study was to control the ZEA production by *F. graminearum* from contaminated cereal grains and other food samples intended for consumption. Hence, as an objective of the present study, effects of OSEO on mycotoxin (ZEA) production in liquid cultures was determined. Results of the present study, showed that OSEO has multifaceted efficacy in inhibiting ZEA production in *F. graminearum*. The mycelial biomass, gene expression (*PKS4* and *PKS13*) and ZEA production exhibited a significant declining trend with increasing concentration of OSEO, i.e., reduction of mycelial biomass and metabolic pathway gene expression causing significant reduction in ZEA production (**Figure 4**). Mycelial biomass was reduced in a dose-dependent manner upon the treatment of OSEO compared to untreated control. The dry weight of OSEO untreated control fungal biomass was estimated as 65.0 mg, while OSEO treated samples showed significant decrease in mycelia dry weight as 53.5, 37.0, 25.1, and 6.6 mg in OSEO concentrations of 250, 500, 1000, and 1500 μg/mL, respectively. However, there was no significant growth observed at a concentration of 2000 μg/mL or more of OSEO (**Figure 4A**).

The analysis of the culture broth for ZEA production by quantitative UHPLC analysis showed that, there was a significant decline in the ZEA production as followed by the OSEO concentration. The concentration of ZEA varied with varying in OSEO concentration viz, 2.62 μg/mL in 250 μg/mL, 1.66 μg/mL in 500 μg/mL and 1.05 μg/mL in 1000 μg/mL in comparison with OSEO untreated control (3.23 μg/mL of ZEA). However, at OSEO concentration of 1500 μg/mL or more, no ZEA production was observed. Results of the present study clearly indicated that, fungal growth was reduced significantly upon increasing OSEO concentration and at higher concentrations (2000 μg/mL) no observable fungal growth was recorded. This in turn is directly proportional to the ZEA concentration as analyzed by UHPLC. To know the relation between mycelial biomass and ZEA concertation upon OSEO treatment, correlation analysis was undertaken and these results suggested that, there is significant linearity between the reductions in fungal biomass versus reduction in ZEA concentration (**Figure 4B**) with that of OSEO concentrations used. The amount of toxin produced was normalized by the biomass of mycelium collected and the percentage of ZEA content in OSEO untreated control was considered as 100%. The reduction in ZEA content of OSEO treated samples were expressed as a percentage of ZEA content in untreated control and the percentage of reduction in ZEA content observed as 98.55, 90.29, and 84.18% at OSEO concentrations of 250, 500, and 1000 μg/mL, respectively (**Figure 4B**).

To further assess the mechanism behind this inhibition, we studied the effects of OSEO on regulation of target metabolic pathway genes (*PKS4* and *PKS13*) recorded in terms of mRNA expression and results were expressed as a fold change in normalization with reference control gene *GAPDH*. Results of the Rt-qPCR revealed that, upon treatment of *F. graminearum* with OSEO, *PKS4* and *PKS13* expression levels were downregulated with increase in OSEO concentration compared to the OSEO untreated *F. graminearum* culture. The downregulated gene

expression levels of *PKS4* and *PKS13* were observed as 0.9 and 1.2 fold in 250 μg/mL, 1.6 and 2.0 fold in 500 μg/mL, 2.8 and 4.4 fold in 1000 μg/mL and 3.4 and 5.6 in 1500 μg/mL concentrations of OSEO, respectively compared with OSEO untreated *F. graminearum* control (as expressed as 0 μg/mL in graphical representation **Figure 4C**).

Targeted genes encoding proteins were very significant in ZEA metabolism in synthesis of ZEA as well as release of ZEA to the media. This concluded that OSEO inhibited the production of ZEA by decreasing the mycelial biomass and also by downregulating the ZEA metabolic pathway genes (*PKS4* and *PKS13*). This could be explained by the effects of phenolic compounds on the structure and function of chromosomes. Significant reduction in the ZEA concentration at lower than MIC and MFC values of OSEO observed in the present study were clearly evidenced that, OSEO is more effective against ZEA production when compared with its antifungal activity. The results of the present study showed that, there was a clear lineage among mycelia biomass, target gene expression as well as ZEA production in liquid cultures.

Kumar et al. (2010) reported that OSEO was less toxic to mice (*Mus musculus* L.) compared to (LD50 value of OSEO as 4571.43 μg/kg) bio-preservatives like pyrethrum (350– 500 mg/kg) and carvone (1640 mg/kg) (Isman, 2006). Extracts of *O. sanctum* L. possess immunomodulatory effects against *Salmonella typhimurium* infection in rat model by increase in TNF-α, IFN-γ and IL-2 cytokines generation (Goel et al., 2010) and also showed ameliorative effects on in sciatic nerve transection-induced neuropathy in rats (Muthuraman et al., 2008). In a review, Gupta et al. (2002) summarized various beneficial properties of *O. sanctum* L. and its use in ancient medicine including Ayurveda, Greek, Roman, Siddha, and Unani to treat several diseases. Keeping in view the importance of the *O. sanctum* in medicine and non-mammalian toxicity, its essential oil has the potential for use as a safe bio-fungicide of the agricultural commodities. Moreover, at 1500 μg/mL of OSEO concentration, the sudden fourfold decrease in mycelial biomass as well as insignificant ZEA production as recorded in present study is still uncertain. To understand the exact mechanism of OSEO on ZEA production at this concentrations, further studies are required to know the influence of OSEO on ZEA metabolic pathway regulation at genetic level.

#### Application of OSEO onto Artificially Inoculated Maize Grains

To know the reliability and real time application of OSEO as an antagonist on growth and ZEA production of *F. graminearum*, studies on artificially contaminated maize grains were undertaken. Subsequently, UPHLC determination of ZEA revealed that, the decreased levels of ZEA in dose-dependent exposure of OSEO (**Figure 5A**). The concentration of ZEA production was significantly decreased with increasing in OSEO concentration viz, 5.12 μg/g in 250 μg/g, 2.83 μg/g in 500 μg/g and 1.55 μg/g in 1000 μg/g in comparison with untreated control (6.07 μg/g of ZEA). However, ZEA was not observed at a concentration of 1500 and 2000 μg/g of OSEO and the study clearly showed the inhibition of ZEA production significantly below MIC and MFC value of OSEO. The gene expression

levels were downregulated upon treatment with OSEO to the samples, these results were in agreement with the liquid culture studies. The inhibitory effect of OSEO on target ZEA metabolic pathway genes *PKS4* and *PKS13* were depicted in (**Figure 5B**). A gradual decrease in the relative fold change in target gene (*PKS4* and *PKS13*) expression were observed upon increasing the concentration of OSEO. The observed fold change in target genes *PKS4* and *PKS13* were 1.0 and 1.5 fold in 250 μg/g, 2.5 and 3.0 fold in 500 μg/g, 3.5 and 5.6 fold in 1000 μg/g and 5.0 and 7.5 in 1500 μg/g concentrations of OSEO, respectively compared to OSEO untreated control. These results suggested that, upon increasing the concentration of OSEO, the expression levels of *PKS4* and *PKS13* genes were downregulated. The level of gene expression was low at the concentration of 1500 μg/g and no fungal growth was observed at the concentration of 2000 μg/g. The grain culture studies were well supported with the results of liquid culture analysis for ZEA production as well as target gene expression.

#### Conclusion

In conclusion, by assessing the obtained results in the present study OSEO can be used as an herbal antagonistic agent against fungal infestation and ZEA production by *F. graminearum*. Owing the potential ill health effects of ZEA and *F. graminearum* on humans, animals and plant systems, the revelations of the present study indicates OSEO as an important intervention in food safety and processing industries where the fungal infestation is a main concern.

### References


## Author Contributions

Design of the work; NK, VM, MS, BV, and VP. Interpretation of data for the work; NK, VM, GN, VG, and VP. Drafting the work; NK, VM, GN, VG, and VP. Final approval version to be published; NK, VM, and VP.

#### Acknowledgments

The first author was thankful to the University Grants Commission [Grant number: F. 2-14/2012(SA-I)], Ministry of Human Resource Development, Government of India, for providing the junior research fellowship for pursuing Ph.D. The authors were also thankful to DRDO-BU-Centre for Life sciences, Bharathiar university campus, Coimbatore-641046, Tamil Nadu, India, for providing the instrumentation facility to carry out the present study.


different *Thymus pulegioides* L. chemotypes. *Food Chem.* 103, 546–559. doi: 10.1016/j.foodchem.2006.08.027


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Copyright © 2015 Kalagatur, Mudili, Siddaiah, Gupta, Natarajan, Sreepathi, Vardhan and Putcha. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

# Bacterial Epimerization as a Route for Deoxynivalenol Detoxification: the Influence of Growth and Environmental Conditions

Jian Wei He1,2† , Yousef I. Hassan<sup>1</sup> \* † , Norma Perilla1,3, Xiu-Zhen Li<sup>1</sup> , Greg J. Boland<sup>2</sup> and Ting Zhou<sup>1</sup> \*

<sup>1</sup> Guelph Research and Development Centre, Agriculture and Agri-Food Canada, Guelph, ON, Canada, <sup>2</sup> School of Environmental Sciences, University of Guelph, Guelph, ON, Canada, <sup>3</sup> Micotox Ltd., Bogota, Colombia

#### Edited by:

Daniela Gwiazdowska, Poznan University of Economics, Poland

#### Reviewed by:

Ian A. Cleary, University of Tennessee at Martin, USA Sean Doyle, National University of Ireland Maynooth, Ireland

#### \*Correspondence:

Ting Zhou ting.zhou@agr.gc.ca; Yousef I. Hassan yousef.hassan@agr.gc.ca †These authors have contributed equally to this work.

#### Specialty section:

This article was submitted to Fungi and Their Interactions, a section of the journal Frontiers in Microbiology

Received: 29 January 2016 Accepted: 06 April 2016 Published: 21 April 2016

#### Citation:

He JW, Hassan YI, Perilla N, Li X-Z, Boland GJ and Zhou T (2016) Bacterial Epimerization as a Route for Deoxynivalenol Detoxification: the Influence of Growth and Environmental Conditions. Front. Microbiol. 7:572. doi: 10.3389/fmicb.2016.00572 Deoxynivalenol (DON) is a toxic secondary metabolite produced by several Fusarium species that infest wheat and corn. Food and feed contaminated with DON pose a health risk to both humans and livestock and form a major barrier for international trade. Microbial detoxification represents an alternative approach to the physical and chemical detoxification methods of DON-contaminated grains. The present study details the characterization of a novel bacterium, Devosia mutans 17-2-E-8, that is capable of transforming DON to a non-toxic stereoisomer, 3-epi-deoxynivalenol under aerobic conditions, mild temperature (25–30◦C), and neutral pH. The biotransformation takes place in the presence of rich sources of organic nitrogen and carbon without the need of DON to be the sole carbon source. The process is enzymatic in nature and endures a high detoxification capacity (3 µg DON/h/10<sup>8</sup> cells). The above conditions collectively suggest the possibility of utilizing the isolated bacterium as a feed treatment to address DON contamination under empirical field conditions.

Keywords: deoxynivalenol, 3-epi-deoxynivalenol, transformation, epimerization, Devosia, growth conditions

## INTRODUCTION

Deoxynivalenol (DON, also known as vomitoxin) is a toxic secondary metabolite produced by several Fusarium species including Fusarium graminearum and F. culmorum (Audenaert et al., 2014). It is one of the most frequently detected mycotoxins in human foods and animal feeds worldwide (Lee and Ryu, 2015). DON presence within animal feed is connected with a myriad of immunological, reproductive, and developmental effects (Hassan et al., 2015d). The most characteristic toxicological symptoms of DON exposure in animals are feed refusal, body-weight loss and emesis (Pestka, 2010). DON is also a human health hazard that causes both acute and chronic effects associated with changes at the molecular and phosphoproteome levels (Wang et al., 2014). In plants, DON is believed to act as a virulence factor and was found essential for symptom development (Moretti et al., 2014). The toxicity of DON is conventionally attributed to its ability to inhibit ribosomal protein biosynthesis but recent studies have reported other novel mechanisms that further explain DON's toxicological profile such as the ability to induce an oxidative-stress response and the involvement in intestinal barrier dysfunction (Hassan et al., 2015d).

The chemical nature of DON and its relative heat stability pose technical challenges to the management of DON-contaminated grains. Physical and chemical detoxification methods have

been explored in the past and microbial detoxification represents an alternative approach that may provide a practical and effective solution for addressing DON-contaminated products (He and Zhou, 2010; McCormick, 2013). Several aerobic and anaerobic microorganisms selected from ruminants, swine, poultry, fish, and other agricultural commodities showed various DON transformation capabilities (Shima et al., 1997; Fuchs et al., 2000, 2002; Volkl et al., 2004; He and Zhou, 2010; Ikunaga et al., 2011; Ito et al., 2013). Despite the promising capabilities of these isolates, most of the reported bacteria require restrictive conditions for growth and DON bio-transformation, such as an anaerobic atmosphere (Fuchs et al., 2000, 2002) and/or the presence of DON as a sole carbon source (Ikunaga et al., 2011; Ito et al., 2013), which pose challenges for their empirical utilization.

The present study reports on the isolation and characterization of a unique bacterial strain capable of bio-transforming DON under aerobic conditions at mild temperatures. The bacterium was initially isolated from an alfalfa soil sample enriched with F. graminearum and moldy corn for several weeks. Microbiological and molecular characterization confirmed the affiliation of this bacterium with the Devosia genus.

The abrogation of toxicity of the biotransformation products was confirmed earlier using different human cell lines and mouse models (He et al., 2015a). The bacterium acts on the C-3 carbon in DON to epimerize the -OH group and produce 3-epi-DON (He et al., 2015b), eliminating the adverse toxicological effects associated with the consumption of DON in food/feed samples. It is hypothesized that the noticed isomeric changes influence 3-epi-DON ability to form hydrogen bonds within the peptidyl transferase center of ribosomes (A-site).

In addition, the role of different key growth and environmental factors such as pH, media formulations, nitrogen and carbon sources, and trace metals on the detoxification performance was elucidated confirming that DON epimerization takes place under a wide range of conditions and without the need for DON as the sole source of carbon in growth media.

## MATERIALS AND METHODS

## Chemicals and Growth Media

Deoxynivalenol, glucose, sucrose, (NH4)2SO4, (NH4)2HPO4, K2HPO4, KH2PO4, MgSO4, K2SO4, FeSO4, MnSO4, NH4NO37H2O were obtained from Sigma–Aldrich (Oakville, ON, Canada) or TripleBond (Guelph, ON, Canada). Methanol was obtained from Caledon Labs (Georgetown, ON, Canada). DIFCO Luria Bertani broth (LB), DIFCO Nutrient broth (NB), DIFCO Tryptic Soy Broth (TSB), DIFCO peptone, DIFCO tryptone, and DIFCO yeast extract were all purchased preparations.

A total of 11 different media formulations were evaluated and Corn meal broth (CMB) was used as the reference. To prepare corn meal broth without salts (CMB-WO-S): 40 g corn meal was soaked in 1 L water at 58◦C for 4 h and allowed to stand for 2 h, followed by a filtration (Whatman No. 1, Whatman; Maidstone, Kent, UK). CMB refers to 1 L of CMB-WO-S supplemented with 3 g (NH4)2SO4, 1 g K2HPO4, 0.5 g MgSO4, 0.5 g K2SO4, 0.01 g FeSO4, 0.007 g MnSO4, and 5 g yeast extract. When 1.5% agar was added to CMB broth, it was referred to as Corn meal agar (CMA).

The following carbon sources were tested: glucose, sucrose, and corn starch. In addition, two categories of nitrogen sources were used: organic such as corn steep liquor, peptones, yeast extract, and urea; inorganic sources such as ammonium sulfate and ammonium nitrate. The concentrations of carbon and nitrogen sources were 10 g/L.

When minerals were added, the following formulation was used: 3 g (NH4)2SO4, 1 g K2HPO4, 0.5 g MgSO4, 0.5 g K2SO4, 0.01 g FeSO4, 0.007 g MnSO4, and 5 g yeast extract was incorporated per 1 L of final broth.

Other media compositions were: Yeast + glucose (YG): 1 L water with 5 g yeast and 10 g glucose. BYE: 1 L water with 0.5 g of NH4NO3, 0.2 g of yeast extract, 50 mg of H3BO4, 40 mg of MnSO4·4H2O, 20 mg of (NH4)6Mo7O24, 4 mg of CuSO4·5H2O, 4 mg of CoCl<sup>6</sup> ·6H2O and 5 mM potassium phosphate buffer (adjusted to pH 7.0 with NaOH) (Shima et al., 1997). Minimal medium (MM): 1 L medium contained 10 g sucrose, 2.5 g K2HPO4, 2.5 g KH2PO4, 1 g (NH4)2HPO4, 0.2 g MgSO4·7H2O, 0.01 g FeSO4, and 0.007 g MnSO4. MM + yeast medium (MMY): MM medium with 0.5% yeast extract. Rice medium (RM) was prepared in a similar fashion to CMB. Corn meal broth + peptone + dextrose medium (CMBPD): CMB with 2% peptones and 2% dextrose. MM + peptones + tryptone medium (MMPT): MM medium with 1% peptones and 1% tryptone.

## Isolation of the Bacterium 17-2-E-8

In 2007 and during the characterization of microorganisms isolated from an alfalfa soil sample that was enriched with F. graminearum-infested moldy corn for 6 weeks, a bacterial isolate 17-2-E-8 was selected on CMA incubated at 28◦C for 72 h. In essence, soil suspensions were serially diluted using CMB medium and both DON reduction and bacterial growth were monitored. For DON reduction, 100 µL of each serial dilution was sub-cultured with 900 µL CMB containing DON (100 µL of 1000 µg/mL DON) at 28◦C on a rotary shaker at 200 rpm for 72 h. Cultures were analyzed later as described below.

For bacterial growth, 100 µL of each serial dilution were cultivated on a CMA plate and incubated at 28◦C. After 48–72 h incubation, the total number of CFU was calculated. Serial dilutions showing the lowest number of microorganisms yet exhibiting a reduction in DON concentrations were selected to be further pursued. The above procedure was repeated eight times and single colonies were picked from plates corresponding to the highest dilution which still showed a reduction in DON concentrations. These colonies were sub-cultured in CMB and their activities were evaluated individually using the methods described earlier. Ultimately, a bacterium showing transparent to white-colored colonies with strong DON-bio-transformation capabilities was isolated.

To follow how the cell numbers of the purified bacterium correlated with its ability to reduce DON concentrations, a tube containing 10 mL of CMB medium was inoculated with a loop of bacterial cell suspension (1 µL). The culture was incubated aerobically at 28◦C with shaking at 200 rpm. This culture was monitored for bacterial growth, decrease in DON concentrations, and formation of the major compound resulting from DON biotransformation, 3-epi-DON (He et al., 2015b). The microbial growth was determined using serial dilution plating methods while DON and 3-epi-DON concentrations were tracked using high performance liquid chromatography (HPLC) (as described later).

### The Influence of Environmental Factors on the Bio-transformation of DON by the Isolate 17-2-E-8

To follow how different environmental factors affect the growth of isolate 17-2-E-8 and its ability to reduce DON concentrations, 10 mL CMB medium were inoculated with a loop of 17-2-E-8 bacterial cells (1 µL). The culture was incubated at 28◦C for 72 h with continuous shaking at 200 rpm. This culture was then adjusted to a cell concentration of 1 × 10<sup>6</sup> CFU/mL using CMB based on optical density (OD600) and served as a seed culture.

To test the effect of aerobic/anaerobic conditions, shaking, and culture media ingredients on the growth and DON reduction by isolate 17-2-E-8, testing mixtures were prepared by adding 100 µL bacterial culture from the above seed culture at a cell concentration of 1 × 10<sup>6</sup> CFU/mL, and 100 µL of DON solution containing 1000 µg/mL, into 800 µL of the respective media, resulting in the final bacterial concentration of 1 × 10<sup>5</sup> CFU/mL and final DON concentration at 100 µg/mL. These mixtures were incubated at 28◦C for 72 h under aerobic conditions on a rotary shaker at 200 rpm, and also under anaerobic conditions (5% H<sup>2</sup> and 10% CO<sup>2</sup> balance N2) at 28◦C with hand-mixing approximately every 6 h.

The temperature effect on the bacterial growth and DON epimerization was tested in 1.5 mL tubes with 1 mL CMB medium containing isolate 17-2-E-8 (∼1 × 10<sup>5</sup> CFU/mL) and 100 µg/mL DON. The tubes were incubated at 5, 10, 15, 20, 25, 30, 35, and 40◦C, respectively, on a rotary shaker at 200 rpm. In a similar fashion, pH effect on the bacterial growth and reduction of DON was elucidated by preparing aliquots of CMB with adjusted pH values of 3, 4, 5, 6, 7, 8, 9, and 10, respectively, using NaOH or HCl 1 mol/L. After inoculation using the above seed culture, test tubes were incubated at 28◦C and 200 rpm shaking under aerobic conditions for 48 h. The samples were extracted and analyzed as described later.

## Mechanism of DON Bio-transformation by Isolate 17-2-E-8

The interactions of DON with the inactivating microorganisms have been reported to take different forms spanning the binding of bacterial cells to enzymatic bio-transformations (Volkl et al., 2004; Awad et al., 2010; Islam et al., 2012; Hassan and Bullerman, 2013). In order to narrow down the nature of the observed interactions between DON and isolate 17-2-E-8, we tested the ability of heat and acid-inactivated cells to reduce DON levels in growth medium and tracked the accumulation of 3-epi-DON in broth at the same time.

A 30 mL dense culture (OD<sup>600</sup> > 2) of isolate 17-2-E-8 was split to three equal parts. The first part (10 mL) represented the living cells while the two other parts (10 mL each) represented the heat and acid-inactivated cells. The heat-inactivated cells were subjected to autoclaving (15 min. at 121oC) while the acidinactivated cells were pelleted by centrifugation and treated with 1N HCl for 2 h. After a washing step with sterile phosphate buffer saline (PBS), acid-inactivated cells were re-suspended to the original volume (10 mL). 0.5 mL of each of the above treatments was mixed with 0.5 mL of fresh LB broth containing 50 µg/mL DON (to yield a final DON concentration of 25 µg/mL). Final tubes (in triplicates) were incubated at 28◦ overnight at 120 rpm. Actively growing cells of Devosia riboflavina Strain IFO13584 were included as a negative control. All the samples were analyzed for DON reduction/epimerization as described below.

#### Analysis of DON and 3-epi-DON

To each 500 µL bacterial culture, 500 µL methanol (analytical grade) was added. The mixture was allowed to shake for 2 h and filtered through a 0.45 µm polyvinylidine fluoride (PVDF) syringe filter (Whatman; Maidstone, Kent, UK) before analyzing by HPLC. DON and 3-epi-DON separation was achieved according to He et al. (2015b) using an Agilent Technologies 1100 Series HPLC system equipped with a Luna C18 column (150 × 4.6 mm, 5 µm) (Phenomenex, Torrance, CA,USA). The binary mobile phase consisted of solvent A (methanol) and solvent B (water) and the gradient program started at 22% A, increased linearly to 41% A at 5 min, 100% A at 7 min, held at 100% A from 7 to 9 min, and returned to 22% A at 11 min. There was a 2 min post-run column reconditioning phase under the starting conditions. The flow rate was 1.0 mL/min and the detector was set at 218 nm. Identification of DON and 3-epi-DON (He et al., 2015b) was achieved by comparing the retention times and UV-Vis spectra. DON/3 epi-DON concentrations were quantified based on reference to a calibration curve of DON/3-epi-DON standards (He et al., 2015a,b).

### Microbiological Characterization

The cell morphology of the isolate was observed using scanning electron microscopy (SEM) and transmission electron microscopy (TEM) for cells grown at 28◦C in TSB. The SEM analysis was conducted in the Electron Microscope Lab, Department of Food Science, University of Guelph; while image acquisition with TEM was conducted at the Guelph Regional Integrated Imaging Facility, Department of Molecular, and Cell Biology, University of Guelph.

The motility of isolate 17-2-E-8 was tested by stabbing 15 mL slants containing either motility test medium (beef extract 3 g, peptone 10 g, sodium chloride 5 g, and agar 4 g/L) or soft CMA (CMB prepared as described above and supplemented with 4 g/L agar). Slants were kept at 28–30◦C and evaluated after 36–48 h of

fmicb-07-00572 April 19, 2016 Time: 16:40 # 3

incubation. Disk diffusion antimicrobial susceptibility tests were performed at 28–30◦C for 24 h either on Mueller Hinton or LB agar plates using the following antibiotic disks: streptomycin (10 µg), chloramphenicol (30 µg), tetracycline (30 µg), penicillin G (10 IU), and kanamycin (30 µg).

The gas chromatographic analysis of fatty acid methyl esters (GC-FAME) was conducted twice independently. Isolate 17-2- E-8 cells were either grown in LB broth for 7 days before inoculating 500 mL (LB broth) with the actively growing cells and incubating the flask on an orbital shaker (120 rpm) at room temperature (23–25◦C) for 48 h. The lyophilized cells were sent later to the Identification Services at Deutsche Sammlung von Mikroorganismen und Zellkulturen (DSMZ) (Braunschweig, Germany) for analysis. Alternatively, the cells were streaked on Tryptic Soy Agar (TSA) and grown for 5 days at 28◦C and the cells were then analyzed by the University of Guelph, Laboratory Services (Guelph, ON, Canada). The obtained chromatographic results (patterns and recognition) were compared with the Sherlock library (Version 4.5; MIDI Inc.; Nework, DE, USA; 2002).

Respiratory quinones were analyzed at DSMZ on freeze-dried cells using the two stage method described by Tindall (1990a,b).

The ability of isolate 17-2-E-8 to oxidize different carbon sources was investigated using BIOLOG bacterial identification system (BIOLOG, Hayward, CA, USA). Briefly, the bacterium was grown on Biolog Universal Growth agar plates supplemented with 5% sheep blood, harvested, and re-suspended in Gram Negative/Gram Positive (GN/GP) inoculating fluid. Suspensions (150 µL) were pipetted into each well of the GN2 MicroPlate and incubated at 30◦C overnight. Carbon and amino acids utilization patterns were assessed and tolerance toward lactic acid, the reducing power, and sensitivity against a large array of chemical compounds were determined. Isolate 17-2-E-8 metabolic fingerprint was compared with the MicroLog database (BIOLOG).

A Matrix Assisted Laser Desorption/Ionization-Time of Flight (MALDI-TOF) bacterial identification based on the mass spectrometry signature of ribosomal proteins (Chalupova et al., 2014) was completed at MIDI Labs, Inc. (Newark, DE, USA) to match 17-2-E-8 to the closest bacterial species.

## 16S rRNA Gene Sequencing and Phylogenetic Analysis

The 16S rRNA gene sequence of 17-2-E-8 was obtained by preparing genomic DNA using Puregene Yeast/Bacteria Kit B (Qiagen, Toronto, ON, Canada) from a dense overnight culture. Primers fD1 and rD1 (Bresler et al., 2000) were used to amplify the 1.5 kb gene with 45◦ annealing temperature. Gel-purified polymerase chain reaction (PCR) products were then used in sequencing reactions using the same amplification primers. The almost complete (1421 bp) retrieved-sequence was searched against the entire collection of 16S ribosomal RNA sequences (bacteria and archaea) within NCBI GenBank to determine the most closely related genus/species (Zhang et al., 2000). For the reconstruction of phylogenetic trees; MEGA version 6.0 software package was used after multiple sequence alignments/comparisons using CLUSTAL\_W (Gap opening penalty = 15, Gap extension penalty = 6.66). Trees were reconstructed using neighbor-joining (NJ) and maximumparsimony (MP) methods with 1000 bootstrapping, Kimura2 parameter (for NJ), and subtree pruning/re-grafting (for MP) models. DNASTAR lasergene 8 software package (DNASTAR, Madison, WI, USA) was also used to confirm tree outcomes. The default settings of MegAlign were: gap penalty = 15.00, gap length penalty = 6.66, delay divergent seqs (%) = 30 and DNA transition weight 0.50. Bootstrap trials were 1000 with seed = 111 (Burland, 2000). Sequence similarity was calculated using the SIAS server<sup>1</sup> . The DNA G + C content of isolate 17-2-E-8 was calculated from a de novo genome assembly conducted and deposited recently (Hassan et al., 2014).

#### Next-Generation Whole-Genome Sequencing and Species Comparisons

Recent advancements in next-generation sequencing platforms have added a new dimension for bacterial isolates comparisons (Hassan et al., 2015c). Using the advantages of such a technique we compared the genome sequence of isolate 17-2-E-8 with other available type strains. The de novo sequencing of the entire genome of isolate 17-2-E-8 was accomplished as reported earlier (Hassan et al., 2014). Other type strains representing different Devosia species were obtained from DMSZ culture collection (Braunschweig, Germany) and the whole-genome sequencing of these strains was conducted as reported (Hassan et al., 2014, 2015a,b).

Pair-wise comparisons of multiple Devosia type-strains genomes were conducted using BRIG (Alikhan et al., 2011) with the default parameters. The entire genome of Devosia 17-2-E-8 was aligned with D. geojensis (DSM19414), D. psychrophila (DSM22950), D. chinhatensis (DSM24953), D. soli (DSM17780), D. limi (DSM17137), D. epidermidihirudinis (DSM25750), and D. riboflavina (IFO13584) genomes.

### Statistical Analysis

For DON concentrations and bacterial cell numbers, samples were analyzed in triplicate and the means were determined. The relevant reduction of DON was calculated as the following: Reduction in DON concentration (%) = (CDON added–CDON residual)/CDON added × 100. Data were analyzed using SAS (SAS for Windows, Version 9.1, SAS institute, Cary, NC, USA), SigmaStat Version 3.11 (Systat Software, Point Richmond, CA, USA), or Sigmaplot 12.5 (Systat Software Inc). Data were tested for normality using the Kolmogorov–Smirnov method and equal variance (P value to reject was set for 0.05). Multiple group comparisons of normally distributed data were conducted by One Way Analysis of Variance (One Way ANOVA), followed by post hoc pairwise comparisons using the Holm–Sidak test or Fisher's protected least significant difference (PLSD). Multiple group comparisons of non-parametric data were conducted using the Kruskal–Wallis ANOVA on Ranks, followed by post hoc pairwise comparisons using the Dunn's method.

<sup>1</sup>http://imed.med.ucm.es/Tools/sias.html

#### RESULTS

fmicb-07-00572 April 19, 2016 Time: 16:40 # 5

## Reduction of DON Concentrations Correlated with the Growth of Bacterial Isolate 17-2-E-8

The effect of incubation times on the bacterial growth and DON concentrations in CMB cultures (containing 100 µg/mL DON) was tested. A coincidence of increasing cell numbers of isolate 17-2-E-8 and decreasing DON concentrations was observed (**Figure 1A**). The maximum increase in bacterial cells and decrease in DON concentrations was observed within 72 h of inoculating CMB medium. During the bacterium exponential growth (6–36 h), a substantial decrease in DON concentrations within growth medium was observed. By the time the bacterium reached the stationary growth phase (almost 48 h

FIGURE 1 | Isolate 17-2-E-8 is capable of transforming DON into 3-epi-DON as a major product under aerobic conditions. (A) The effect of incubation times on the bacterial growth and DON concentrations in corn meal broth (CMB) cultures containing 100 µg/mL DON and inoculated with bacterium 17-2-E-8. The cultures were kept shaking (200 rpm) at 28◦C. The PLSD(0.05) of the tests for concentration of DON and 3-epi-DON, and cell concentration were 7.6 and 0.38, respectively. (B) The effects of shaking on bacterial cell numbers of isolate 17-2-E-8 and reduction of DON levels in CMB cultures containing 100 µg/mL DON. The PLSD(0.05) of the tests for DON concentration and viable cell numbers were 8.1 and 0.38, respectively.

after inoculation), DON concentrations fell drastically to low levels (5 µg/mL). As reported earlier, the bacterium 17-2-E-8 transforms DON to 3-epi-DON as the major metabolite under aerobic conditions (He et al., 2015b). A matched accumulation in 3-epi-DON within growth media developed in parallel with DON disappearance from growth medium. As shown by (**Figure 1A**), 3-epi-DON levels kept accumulating within test tubes in parallel to DON reduction until reaching the maximum levels noted around 60–72 h of incubation. Thereafter, 3-epi-DON levels did not change even with extended incubation times (up to 132 h). The initial trials indicated that isolate 17-2-E-8 had the capacity to detoxify DON at rates close to 3 µg/h/10<sup>8</sup> cells (**Figure 1A**) within the exponential growth phase (24– 36 h).

#### Bio-transformation of DON Proceeded under Aerobic Conditions

The effects of shaking/aerobic/non-aerobic conditions on bacterial cell numbers of isolate 17-2-E-8 and reduction of DON levels in CMB cultures were checked. No increase in bacterial cell numbers or any associated reduction in DON concentrations was observed when isolate 17-2-E-8 was incubated for 72 h in CMB medium under anaerobic conditions (data not shown). However, increases in bacterial cell numbers and reduction in DON concentrations were observed when isolate 17-2- E-8 was incubated for 72 h in CMB medium under aerobic conditions (**Figure 1B**). The rotary action (200 rpm) of the orbital shaker impacted the cell's growth, assumingly by altering the oxygen supply, and at the same time affected the DONtransformation activities of isolate 17-2-E-8. With continuous shaking, maximum cell numbers and maximum DON reduction levels were observed around 48 h of the incubation period (**Figure 1B**).

#### Growth of Isolate 17-2-E-8 and DON Bio-transformation Proceeded Favorably at Mild Temperature

A wide range of temperatures spanning 5–40◦ were selected to check their effect on supporting isolate 17-2-E-8 growth and DON epimerization. The effect of growth temperatures on DON levels in CMB cultures spiked with 100 µg/mL DON and inoculated with isolate 17-2-E-8 are shown in **Table 1**.

As shown, the highest growth was observed within the 25– 35◦ range after 48 h incubation. Isolate 17-2-E-8 cell numbers were significantly higher in this range compared to lower or higher temperatures. On the other side, the bio-transformation of DON continued at maximum velocity within the 20–35◦ range (**Table 1**). Unfavorable temperatures (as low as 15◦ or as high as 40◦ ) limited DON bio-transformation capabilities of isolate 17- 2-E-8 cells. The influence of low/high temperatures was more drastic on the epimerization function in comparison to cell growth. For example, the cells reached 1.6 × 10<sup>9</sup> CFU/mL at 40◦ (**Table 1**), but DON transformation merely exceeded a level of 1.3%.

TABLE 1 | Growth and DON biotransformation activities of isolate 17-2-E-8 at selected cultivation temperatures.


<sup>1</sup>Determined after 48 h in CMB (with minerals) at pH 6.9.

<sup>2</sup>Values in the same column with different superscripts differ significantly according to Tukey's multiple range test (p = 0.05).

#### TABLE 2 | The growth and biotransformation of DON<sup>1</sup> by isolate 17-2-E-8 at selected initial media pH values.


<sup>1</sup>Determined after 48 h bacterial cultivation in corn meal broth (CMB) at 28◦C and 200 rpm.

<sup>2</sup>Values in the same column with different superscripts differ significantly according to Tukey's multiple range test (p = 0.05).

TABLE 3 | Residual DON and its bio-transformation products after culturing isolate 17-2-E-8 in different media preparations supplemented with 100 µg/mL DON, at 28◦C with continuous shaking at 200 rpm for 72 h.


<sup>1</sup>Values in the same column with different superscripts differ significantly according to Tukey's multiple range test (p = 0.05).

<sup>∗</sup> BDL, below the detection level.

A follow-up experiment (data not shown) indicated that incubation at 28◦C demonstrated the same bio-transformation efficiency as 25◦C, whereas the required time was shorter. Therefore, a temperature of 28◦C was selected as the optimum temperature for the following experiments.

## Neutral pH Values Maximized Isolate 17-2-E-8 Bio-transformation Capacity

pH is considered as an influential factor that affects bacterial growth/function. The effect of this factor was assessed by growing isolate 17-2-E-8 in CMB with adjusted pH values and determining growth and DON biotransformation patterns. **Table 2** shows how different pH values affected both isolate 17- 2-E-8 growth and DON bio-transformation capacity. The total number of viable cells indicated that isolate 17-2-E-8 increased in numbers as the acidity decreased within the media and pH values came closer toward neutrality (pH = 7). Shifting toward alkalinity (pH = 9–10) reversed the observed increase in cell numbers causing a significant inhibitory effect but to a milder degree compared to the acidic pH range. In contrast to the gradual decline of the total number of viable cells observed due to the deviation from the neutral pH range, DON epimerization function was affected substantially in comparison. The most noticeable reductions in DON concentrations were only observed at pH values ranging from 6 to 8 in general and around pH = 7 more specifically (**Table 2**).

## Media Affected the Growth of Isolate 17-2-E-8 and DON Bio-transformation

Media composition had a strong influence on DONtransformation capacity of isolate 17-2-E-8. For example, both CMB and YG media supported the highest accumulation of 3-epi-DON with 49.2 and 89.3 µg/mL, respectively, and the lowest DON residual with 4.6 µg/mL and below the detection limit (BDL) concentrations, respectively (**Table 3**). In contrast, MMPT medium did not show any support for the epimerization function of DON (**Table 3**).

Our observation related to the influence of different media compositions on DON biotransformation rates of isolate 17-2-E-8 (**Table 3**) triggered a more systematic approach to understand the influence of different carbon and nitrogen sources in addition to the role of minerals. As indicated in **Table 4**, sources rich in complex organic nitrogen (particularly peptones and yeast extract) supported isolate 17-2-E-8 growth and maximum DON transformation with levels close to 99% of total DON concentration (**Table 4**) while inorganic nitrogen sources (such as ammonium sulfate, and ammonium nitrate) were limiting in supporting the biotransformation capacity (1–22%). This might be attributed to the role of organic nitrogen in supporting protein biosynthesis (essential and non-essential amino acids) needed for enzyme functionality/activity.

The role of minerals, mainly as co-factors in the enzymatic reactions responsible for epimerization of DON to 3-epi-DON is illustrated by the effect of incorporating mineral mixtures on growth and DON conversion rates (**Table 4**). While minerals did not influence the growth rate within corn steep liquor or peptone broth (with similar bacterial counts within the 1.5–3.5 × 10<sup>9</sup> range), it did significantly influence the bio-transformation rates enhancing them from 0.7 to 48% in corn steep liquor and from 17 to 99% in peptones broth, respectively (**Table 4**). This effect was also observed in the media supplemented with inorganic nitrogen where DON biotransformation was efficiently enhanced (independent from bacterial growth) from 1.0–1.4 to 18–22% range in both ammonium sulfate and ammonium nitrate broths.

## Bio-transformation of DON by Isolate 17-2-E-8 Was Enzymatic in Nature

In order to further track the mechanistic nature of DON reduction by isolate 17-2-E-8; both viable and inactivated cells (heat/acid) were tested separately. Cells of D. riboflavina Strain IFO13584 were included as a negative control. **Figure 2** shows that DON concentrations were reduced to an undetectable levels only when incubated overnight with viable cells (**Figure 2A**). This

TABLE 4 | The effect of carbon, nitrogen, and minerals sources on growth and DON biotransformation activity of Devosia mutans Strain 17-2-E-8<sup>1</sup> .


<sup>1</sup>Determined after 72 h in shaken culture (200 rpm) at 28◦C.

<sup>2</sup>Values of media pH are shown.

<sup>3</sup>Minerals (per liter): K2HPO4, 1.0 g; MgSO4, 0.5 g; K2SO4, 0.5g; FeSO4, 0.1 g, and 0.07 MnSO4, 0.07 g.

<sup>4</sup>Values in the same column with different superscripts differ significantly according to Tukey's multiple range test (p < 0.05).

<sup>5</sup>The percentage of DON bio-transformation was estimated by subtracting the remaining DON after incubation from the initial concentration, multiplied × 100.

reduction in DON was accompanied by the accumulation of 3 epi-DON in the culture media (**Figure 2B**). At the same time, none of the heat or acid-inactivated cells led to any detectable decrease in DON levels or the appearance of 3-epi-DON within growth media. As expected, the viable cells of D. riboflavina Strain IFO13584 were not able to bio-transform DON to 3-epi-DON nor influence DON concentrations within the test tubes (**Figure 2**). Collectively, these results support the notion of an enzyme-carried function.

## Microbiological Characterization Confirmed the Affiliation with the Devosia Genus

Our earlier observations indicated that the isolate 17-2-E-8 belongs to the genus Devosia possibly representing a new species (He et al., 2015a). A further phenotypical characterization coupled with more genomic data was needed to confirm these observations.


The acquired SEM and TEM images, in addition to Gramstaining patterns, clearly indicated that cells of isolate 17-2- E-8 had an oval to rod-shaped morphology (**Figure 3A**) that stained negative with Gram-staining. Cell dimensions were 1– 1.8 µm in length and 0.4–0.8 µm in width (**Figure 3B**). The bacterial cells were able to form 1–4 polar flagella (**Figure 3C**). These cells formed circular, raised, and transparent to whitecolored colonies with smooth edges when grown on CMA, TSA, and nutrient agar. The colonies seemed to lack the shiny reflection noticed with D. riboflavina (IFO 13584, J.W. Foster 4R3337) colonies growing on the same media (**Table 5**). Using D. riboflavina (Foster) Nakagawa et al. (1996) (ATCC 9526) as a reference/control, isolate 17-2-E-8 was resistant to kanamycin and chloramphenicol but susceptible to penicillin G and tetracycline. It showed intermediate sensitivity toward streptomycin. Isolate 17-2-E-8 was motile in both motility/soft CM slants.

The obtained fatty acid profile of 17-2-E-8 is shown in **Figure 4**, and was mainly composed of C16:<sup>0</sup> (11.64%), C18:<sup>0</sup> (8.68%), C18:1ω7c (31.78%), C10:<sup>0</sup> 3-OH (1.0%), C18:<sup>0</sup> 3- OH (3.04%), 11-methyl C18:1ω7c (26.26%), C19:0cycloω8c (13.19%), and C17:<sup>0</sup> (1.5%). This profile was close in nature to several members of the Devosia genus (**Table 6**). While we had some concerns about the elevated levels of 11-methyl C18:1ω7c and C19:0cycloω8c in the final profile, the identical values present in two independent analyses (25.1%, 13.3% in the first analysis and 26.26%, 13.19% in the second analysis for 11-methyl C18:1ω7c and C19:0cycloω8c, respectively) confirmed the validity of the obtained profile in this study, despite the slightly differing growth conditions/times. The predominant isoprenoid quinone was identified as Q10 (100%) while Q11 was not detectable in our sample. This is in accordance with most validly reported Devosia species where Q10 is the principle detected ubiquinone except for D. insulae (Yoon et al., 2007). The analysis of polar lipids reflected the presence of phosphatidylglycerol and di-phosphatidylglycerol in addition to two unidentified glycolipid bands labeled as GL1 and GL2 (**Figure 5**).

Metabolic profiling indicated that isolate 17-2-E-8 utilized α-D-glucose, α-D-lactose, maltose, D-mannose, Larabinose, D-cellobiose, L-fucose, D-galactose, gentiobiose, and melibiose, but not D-fructose, lactulose, or D-sorbitol (**Table 5**). The bacterium thrived in the presence of D, L-lactic, D-gluconic, hydroxybutyricacid, and D-saccharic acids while growth was inhibited in the presence of acetic, citric, formic, malonic, propionic, and succinic acids, respectively.

The MALDI-TOF ribosomal signature failed to reliably match 17-2-E-8 to Devosia at the genus level (matching score was lower than the acceptable 1.700 threshold) and the closest hit correlated to Lactobacillus plantarum ssp. plantarum (with a matching score of 1.399). This finding was not surprising as the representation of the Devosia genus within the searched MALDI Library (v.081213) is small with only D. riboflavina representing the entire Devosia genus.

.

 −,

fmicb-07-00572 April 19, 2016 Time: 16:40 # 9

(bacterial methyl ester) standards.

## Phylogenetic Analysis, Whole-Genome Sequencing, and Strain Comparisons

The 16S rRNA gene sequence alignment mapped isolate 17- 2-E-8 to other species within the Devosia genus with the closest species match to D. insulae DS-56<sup>T</sup> at 95% sequence similarity. Multiple sequence alignments (Clustal\_W) and the construction of phylogenetic trees with 1000 bootstraps showed how this isolate related to other species within the Devosia genus (**Figure 6A**). The results supported clustering 17-2-E-8 with other members of the Devosia genus yet forming an independent lineage adjacent to D. insulae DS-56T (EF012357). The 16S DNA gene sequence (1421 bp) was deposited within the NCBI nucleotides collection under accession number (KJ572863). The G + C content of isolate 17-2-E-8 was calculated at 63.95%, falling within the range that has been previously reported (59.5 to 66.2%) for the genus Devosia. The pair-wise comparisons of multiple genomes of Devosia type-strains (**Figure 6B**) assembled using the de novo approach clearly confirmed the uniqueness of Devosia 17-2-E-8 at the genome level in relation to the closest validated Devosia species (**Figure 6B**).

## DISCUSSION AND CONCLUSION

Considering the dynamics of ecological communities, strategies that adopt the enrichment of growth media with natural sources of detoxifying bacteria are more likely to succeed in isolating microorganisms capable of bio-transforming the target toxins (Volkl et al., 2004; Ito et al., 2013). In such cases, it is hypothesized that DON-transforming microbes will be able to grow in environments that contain a high population of the plant pathogen Fusarium spp. with high rates of host plants infection. In the present study, a bacterium 17-2-E-8 that can detoxify DON was isolated from soil samples enriched with F. graminearuminfested corn for 6 weeks. This approach may also be applied to select microorganisms that transform other natural toxins or environmental contaminants.

The isolated bacterium was confirmed to belong to the Devosia genus (Nakagawa et al., 1996; Rivas et al., 2003; Vanparys et al., 2005; Yoo et al., 2006; Lee, 2007; Yoon et al., 2007; Kumar et al., 2008; Ryu et al., 2008; Verma et al., 2009; Bautista et al., 2010; Zhang et al., 2012; Dua et al., 2013; Galatis et al., 2013; Romanenko et al., 2013). The results from 16S rRNA gene sequence similarity and phenotypic characterization support that isolate 17-2-E-8 represents a new species, for which the name Devosia mutans (mu'tans, L. part. adj; mutans, pertains to the ability of this species to transform or convert deoxynivalenol) is established, with the type strain 17-2-E-8 (IDAC 040408-1 = ATCC PTA-121309).

The ability of D. mutans 17-2-E-8 to grow and transform DON in various media (such as CMB and LB, etc) confirmed that this bacterium does not require DON to be the sole source of carbon to commit DON to the bio-transformation pathways, which highlights the possibility for an empirical use of such a strain within the feed industry. Strain 17-2-E-8 plasticity of transforming DON was maintained even with broths rich in other organic source of carbon (such as LB, CMB, peptones and corn steep liquor). This is different from most previously reported isolates such as Agrobacterium-Rhizobium sp. strain E3-39 (Shima et al., 1997), Sphingomonas sp. strain KSM1(Ito et al., 2013), and Nocardioides sp. strain WSN05- 2 (Ikunaga et al., 2011) that demand the presence of DON as the sole source of carbon for detoxification. Based on the results presented above, further investigations about the nature of DON detoxification capabilities of D. mutans 17-2-E-8 are

fmicb-07-00572 April 19, 2016 Time: 16:40 # 10


FIGURE 5 | Polar lipids of isolate 17-2-E-8 as separated by two dimensional silica gel thin layer chromatography. The first direction is developed in chloroform:methanol:water (65:25:4, v/v/v), and the second in chloroform:methanol:aceticacid:water (80:12:15:4, v/v/v/v). Four spots were detected reflecting glycolipids (GL1 and GL2), diphosphatidylglycerol (DPG),

proposed to explore its full-potential especially in the light of the recently confirmed abrogation of toxicity of 3-epi-DON (He et al., 2015a).

and phosphatidylglycerol (PG) fractions.

Growth and environmental factors can have a strong influence on bacterial cultures and their related functions/phenotypes (von Stetten et al., 1999; Coggan and Wolfgang, 2012). In the current study, we examined the influence of several such factors on the growth of, and DON epimerization by, isolate 17-2-E-8. The presented data showed that the isolated bacterium was involved in transforming DON to 3-epi-DON under aerobic conditions and in the presence of oxygen where it maintains optimal growth and functionality. Previously reported isolates originating from animal digestive-systems showed the need of strict anaerobic conditions to survive and transform DON (He et al., 1992; Fuchs et al., 2000, 2002) and, hence, limiting their practical use.

Equally important is that D. mutans 17-2-E-8 grows at mild temperature (25–28◦C) whereas other bacterial strains isolated from ruminants and poultry guts grow at higher temperature (i.e., 37◦C) (He et al., 1992). The results obtained in the present study showed that D. mutans 17-2-E-8 is not well adapted to temperature below 20◦C, but it grew fairly well in a broad range of temperature spanning 20–40◦C. As soil microorganisms are classified into three groups according to temperature tolerances, D. mutans 17-2-E-8 should be classified as a mesophilic bacterium. Similarly, DON bio-transformation of D. mutans 17-2-E-8 appeared to be very efficient within the range of 20–35◦C where the bacterium exhibited the highest DON detoxification activity at 28–30◦C. As the temperature increased and while the bacterium retained its growth capacity at 40◦C,

fmicb-07-00572 April 19, 2016 Time: 16:40 # 11

\*Summed features represent groups of two or three fatty acids that could not be separated by GLC with the MIDI system. Summed feature 3 contained C16:1ω7c and/or iso-C15:0 2-OH.

(DSM24953), D. soli (DSM17780), D. limi (DSM17137), D. epidermidihirudinis (DSM25750), and D. riboflavina (IFO13584) genomes (inner to outer circles, respectively). Nucleotide sequence comparisons were conducted using BRIG. The comparison clearly shows the uniqueness of Devosia 17-2-E-8 at the genome level in relation to the closest validated Devosia species.

no DON-bio-transformation was detected at this temperature suggesting a low correlation between the biomass and DONbio-transformation at such elevated temperatures. These findings are in agreement with the study by Volkl et al. (2004) in which a mixed bacterial culture transformed DON at temperature ranging from 20 to 30◦C but the functionality was lost when temperatures were above 37◦C.

In a similar fashion, the initial pH of the media had a significant effect on the growth of D. mutans 17-2-E-8 and DON bio-transformation. In general, soil bacteria grow optimally at a pH near neutrality (Feng et al., 2008). D. mutans17-2- E-8 reached its maximal DON biotransformation activity at pH = 7–8; however, it could not tolerate extreme alkaline or acidic conditions which resulted in minimal/no DON biotransformation. The effects of pH levels may be attributed to the interference of hydrogen atoms with enzyme-assisted epimerization (Goese et al., 2000).

When individual nitrogen and carbon sources were used in the media, changes in the isolate's growth and DON biotransformation were observed regardless of the presence of minerals. In the presence of organic nitrogen sources (yeast extract and peptones), the bacterium grew well while inorganic nitrogen sources (ammonium sulfate and ammonium nitrate) resulted in poor growth. Yeast extract and peptones are well known for supporting growth of microorganisms (Zhang et al., 2011). These substrates not only contain balanced levels of amino acids and peptides but also water-soluble vitamins, minerals, and carbohydrates. Yeast extract and peptones supported the rapid growth of D. mutans 17-2-E-8 independent of the supplementation with minerals, however, differences were found in DON bio-transformation capacity when minerals were added to corn steep liquor and peptones media. Peptone-containing media exhibited a small increase while corn steep liquor broth showed a significant improvement in DON biotransformation. This increase in the bio-transformation could be related to particular elements such as Mg2<sup>+</sup> and Fe2<sup>+</sup> considered as enzyme cofactors (Boll and Fuchs, 1995). Alternatively, the addition of minerals may also have shifted the pH of the medium drastically increasing DON bio-transformation by D. mutans 17-2-E-8 as it was the case in corn starch medium and ammonium sulfate broth (**Table 4**). The inability of urea broth to support DON biotransformation can be logically explained by the unsupportive pH

#### REFERENCES


value (alkaline conditions) observed in this media even in the presence of minerals.

The epimerization function in the above studies correlated positively with the numbers of viable bacterial cells suggesting an enzymatic pathway responsible for the noted activity. It should be noted that most of the enzymatic processes are dependent on the availability of specific co-factors (Lyagin et al., 2011) including metals, the current study clearly demonstrates that mineral mixtures addition substantially enhanced DON biotransformation capabilities of D. mutans 17-2-E-8.

In summary, the optimal growth and DON biotransformation conditions of D. mutans 17-2-E-8 include: temperatures close to 28◦C, close to neutral pH of 7, and an organic source of nitrogen and carbon in the presence of aerobic atmosphere. Under the aforementioned conditions, 3-epi-DON is the primary product of the conversion in a process that is fundamentally enzymatic in nature. The observed accumulation of 3-epi-DON within the growth medium keeps the question about the fate of 3-epi-DON open for future investigation. Collectively, these conditions and the efficient capacity to detoxify DON (3 µg/h/10<sup>8</sup> cells) distinguish D. mutans 17-2-E-8 from previously reported bacterial isolates. The present research serves as a foundation for development of a feed treatment to detoxify DON in contaminated grains for industrial application, under mild and empirical field conditions particularly in liquid-feeding systems or prior to fermentations with lacto-bacteria.

#### AUTHOR CONTRIBUTIONS

Design of the work: JWH, YH, NP, GB, and TZ. Conducting experiments: JWH, YH, NP, and X-ZL. Interpretation of data: JH, YH, NP, GB, and TZ. Drafting the work: JWH, YH, NP, GB, and TZ. Final approval: JWH, YH, GB, and TZ.

### ACKNOWLEDGMENTS

Authors would like to thank Drs., Xianhua Yin, Qing Yu, and Dion Lepp for their valuable technical advice and Agriculture and Agri-Food Canada (AAFC) for the financial support of this research work.


toxin deoxynivalenol. Appl. Environ. Microbiol. 79, 1619–1628. doi: 10.1128/AEM.03227-12


fmicb-07-00572 April 19, 2016 Time: 16:40 # 14

glacier cryoconite, and an emended description of the genus Devosia. Int. J. Syst. Evol. Microbiol. 62, 710–715. doi: 10.1099/ijs.0.023937-0


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2016 He, Hassan, Perilla, Li, Boland and Zhou. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.