'r'ba'fil’cfi—iJ--Ao.w" — - JT-WMWu- Wfl-wr—‘wv—w' In?“ '3: "' ' “xxx-gm" 2- ‘ ... —. - at: ‘ 3.4 \{1 I EAIE’i‘ifi’J 7:, V: . ' .. ‘51-‘1' . < ' ‘ ' 1 3 . ‘ .. .23- ‘ ‘ J r "I ”'4 v .I . - I I"; - w 17. . 4.... - I. I 1101‘; :‘k H’ V , . I I “r. I .1. I ,1, II. .Il‘lfii 9‘} I II' "I » . I 1,. , I. , " d .. ‘ I I: I 1‘: I I u .a w; .' , ' -: W In :1. III I' ." I" I 3]). “-l .1". . .I 4‘ .' I. v ' I, . “.14.” I ‘ . '- a :1. I:' "“4 M4: m: 32?} : “’3: a 5i“: L'I-I"“II"’I . I 4| :,I- H: " .I-uf:\’!it._‘, [I v t}; 1‘ ‘ . .‘ I ‘ A .IIIvI . III in. Lm I ”I, 'Is'3‘dsro‘m . 'Ij .{ -. ' “ .mflh“ ~. w! "a; . III'IIIII..~I:'N"I:"§II -: I" :I H2 .4: - . - I I I I. Ii 1" .zI-m, ~ .2 m 1 law -. leg: I ’ '1' v ." . , -..,....... . m ’ ~3’Jv‘2’m ~41. u-.. I -.. -I. %.5 |. fit my“ ‘vq --~4-_?-_.; 37-" .fig‘fi.’ 1:3,: This is to certify that the dissertation entitled Molecular analysis of HC-toxin biosynthesis in Cochliobolus carbonum presented by Yi-Qiang Cheng has been accepted towards fulfillment of the requirements for Ph.D. degree in Botany and Plant Pathology NW Major professor t/zI/I I MS U is an Affirmative Action/Equal Opportunity Institution 0- 12771 . LIBRARY Michigan State University PLACE IN RETURN BOX to remove this checkout from your record. TO AVOID FINES return on or before date due. MAY BE RECALLED with earlier due date if requested. DATE DUE DATE DUE DATE DUE 6/01 cJCIRC/Dateouepes-p. 1 5 MOLECULAR ANALYSIS OF HC-TOXIN BIOSYNTHESIS IN COCHLIOBOLUS CARBONUM By Yi-Qiang Cheng A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Botany and Plant Pathology 1999 ABSTRACT MOLECULAR ANALYSIS OF HC-TOXIN BIOSYNTHESIS IN COCHLIOBOLUS CARBONUM By Yi-Qiang Cheng HC-toxin is the molecular determinant in the interaction between the fungal pathogen Cochliobolus carbonum race 1 and its host plant, Zea mays L. (genotype hm1lhm1). HC-toxin exists as a small family of cyclic tetrapeptides. The major component of HC-toxin, HC-toxin I, has the structure cyclo(D-Pro-L- Ala-D-Ala-L-Aeo), where Aeo stands for 2—amino-9,10-epoxi.8-oxo-decanoic acid. The biosynthesis of HC-toxin is controlled by a genetic locus, TOX2, that is complex at the molecular level. Previous studies have identified four genes (HTS1, TOXA, TOXC, and TOXE; collectively called TOX2). These known TOX2 genes each exist as two or three functional copies per genome and are present only in HC-toxin-producing isolates (Tox2+) of C. carbonum. The TOX2 genes are distributed over ~540 kb on a special chromosome in a standard lab strain $8111 (one copy of TOXE is located on a different chromosome). When all copies of any individual gene are mutated, the fungus loses the capability to produce HC-toxin and thus pathogenicity. In this thesis research, two additional TOX2 genes, TOXF and TOXG, have been cloned by using bacterial artificial chromosomes (BACs). TOXF and TOXG are two tightly linked genes. Both have three copies in 88111, and two functional copies in isolate 164R10, both are exclusively present in Tox2+ isolates of C. carbonum, both map to the TOX2 locus, and both are regulated by TOXE. TOXF encodes a putative branched-chain amino acid aminotransferase that we hypothesize to aminate an a-keto acid in the pathway to make Aeo. A null mutant of TOXF failed to produce HC-toxin and lost the ability to causes severe leaf spot disease on maize. Therefore, TOXF can be regarded as a pathogenicity gene. TOXG encodes a novel alanine racemase that catalyzes the interconversion between L-alanine and D-alanine. D-alanine is a critical constituent in HC-toxin l, Ill and IV, but not HC-toxin II, which has glycine in place of D-alanine. A null mutant of TOXG failed to make HC-toxin I, III and IV, but still made HC-toxin ll. Feeding the TOXG null mutant with D-alanine restored the normal HC-toxin production profile. Compared to wild type, the TOXG mutant caused a delayed disease phenotype that eventually resulted in full symptoms due to the presence of HC-toxin ll. Therefore, TOXG can only be classified as a virulence factor. Genetic and biochemical experiments were successfully adopted to confirm that TOXG gene product functions as alanine racemase. Another TOX2 candidate gene, TOXD, was cloned based on its linkage to TOXC. TOXD shares similarities with other known TOX2 genes, including physical linkage, copy number, gene distribution, and regulation by TOXE. However, HC-toxin production and fungal pathogenicity were unchanged in a TOXD null mutant. It appears that TOXD has no apparent role in HC-toxin production or fungal pathogenicity. To my beloved parents. iv ACKNOWLEDGEMENTS I am grateful to Dr. Jonathan D. Walton for years of guidance, encouragement and sharing scientific insights. Many thanks to the rest of the Guidance Committee members, Drs. Hammerschmidt, Linz, Ohlrogge and Zeevaart, for their advice. Members of the Walton lab, especially Dr. Joong-Hoon Ahn, have been very helpful in technical assistance, collaboration and discussions. TABLE OF CONTENTS LIST OF TABLES.............. ....viii LIST OF FIGURES... .......ix LIST OF ABBREVIATIONS...... ......xl CHAPTER 1. SECONDARY METABOLISM AND FUNGAL TOXINS... .1 lntroduction...................... 2 Concepts of secondary metabolism and secondary metabolite... .3 Characteristics of fungal secondary metabolites and their biological activities... ...4 Biosynthesis offungal secondary metabolites... ........6 Functions offungal secondary metabolites...... ......8 Origin/evolution of secondary metabolism...... .....10 The existence of gene clusters for fungal secondary metabolite biosynthesis...... .......12 Selected fungal toxins: Molecular biosynthesis and roles in pathogenesis... ....13 1. Aflatoxins/sterigmatocystins... ...15 2. Fumonlsms16 3. Trichothecenes......... .....20 4. T-toxin......... ............23 5. Victorin... ......25 6. HC-toxm27 Objectives ofthis dissertation........... .....32 CHAPTER 2. CLONING AND ANALYSIS OF TOXD, A GENE THAT BELONGS TO TOX2 LOCUS BUT HAS NO APPARENT ROLE IN HC-TOXIN PRODUCTION .....34 Abstract .......35 lntroduction...... .........36 vi Materials and Methods... Results Discussion............... CHAPTER 3: A PUTATIVE BRANCHED-CHAIN AMINO ACID AMINOTRANSFERASE GENE REQUIRED FOR HC-TOXlN BIOSYNTHESIS AND PATHOGENICITY 39 ...41 .49 IN COCHLIOBOLUS CARBONUM... Abstract...... Introduction... Methods Results Discussion............. CHAPTER 4: A NOVEL EUKARYOTIC ALANINE RACEMASE GENE INVOLVED IN CYCLIC PEPTIDE BIOSYNTHESIS IN THE FUNGUS COCHLIOBOLUS CARBONUM... Abstract......... Introduction Materials and Methods... Results Discussmn CHAPTER 5: SUMMARY AND PERSPECTIVE... APPENDIX: ‘30 LABELING INDICATES THAT THE EPOXIDE- CONTAINING AMINO ACID OF HC-TOXIN IS BIO- SYNTHESIZED BY HEAD-TO-TAIL CONDENSATION OF ACETATE .............................................................. vii seas 63 .76 81 82 .83 91 106 .111 ..... 119 134 LIST OF TABLES Table 1. Biological activities of selected fungal secondary metabolites... Table 2. Biochemical activities required for HC-toxin production ............... Appendix Table 1. Proton NMR assignments for HC-toxin... Appendix Table 2. Carbon NMR assignments for HC-toxin... Appendix Table 3. Quantitative analysis of enrichment of relevant carbon atoms in HC-toxin after labelling with 1-[13C]acetate or 2-[13C]acetate........................ viii ..5 ...... 30 .124 .125 .127 LIST OF FIGURES Figure 1-1. The four major pathways for secondary metabolite biosynthesis ....... 7 Figure 1-2. Structures of selected fungal toxms17 Figure 2-1. Restriction fragment length polymorphism (RFLP) of TOXC... ..43 Figure 2-2. Presence of TOXD only In HC-toXIn-producmg (Tox2 .) isolates ofC. carbonum... 44 Figure 2-3. Chromosomal location of TOXD45 Figure 2-4. Sequence of TOXD47 Figure 2-5. Protein sequence alignment of TOXD (CcTOXD) and/ovC(AtLovC)... ..... 48 Figure 2-6. Disruption of TOXD in C. carbonum isolate 164R10... ...50 Figure 2-7. Phenotypes of TOXD null mutant T49252 Figure 3-1. Construction of pBACocta60 Figure 3-2. Genes identified on BAC clone A24C165 Figure 3-3. Presence of TOXF only In HC-toxm-producmg (Tox2 ) isolates ofC. carbonum... 67 Figure 3-4. DNA and deduced protein sequence of TOXF7O Figure 3-5. Sequence alignment of TOXFp” with BCATs from other organisms. . . 72 Figure 3-6. Targeted mutation of TOXF74 Figure 3-7. Analyses of TOXF mutants77 Figure 3-8. Pathogenicity assay of TOXF mutants.......................................78 Figure 4-1. Restriction map of the two copies of the TOXF/G region in 164R1092 Figure 4-2. Sequence of TOXG and its relationship to TOXF... .94 ix Figure 4-3. Multiple sequence alignment of TOXGp and homologs... ..96 Figure 4-4. TOXF and TOXG expression requires TOXE... ...98 Figure 4-5. Complementation of E. coli alanine racemase mutant by TOXG ...... 99 Figure 4-6. Southern blot analysis of TOXG mutants..................................1OO Figure 4-7. Northern blot verification of the independent mutation of TOXF and TOXG101 Figure 4-8. TLC Analysis of HC-toxin In WIId type (WT, 164R10m) and In TOXG null Mutant (T698) .. ...............104 Figure 4-9. Pathogenicity assay of 164R10(Wl'), single mutant ”(T697) and double mutant (T698). .. . ..105 Figure 5-1. Chromosomal map of TOX2 genes and the 3.5-Mb chromosome.......................................................115 Figure 5-2. Diagram of the HC-toxin (I) biosynthesis machinery... 1 16 Appendix Figure 1. Structure of HC-toxin..................... 124 Appendix Figure 2. 13C NMR spectra of (A) native HC-toxin and HC-toxin from culture filtrates of C. carbonum grown in the presence of either (B) 1-[13C]acetate or (o) 2-[13C]acetate.... ....125 LIST OF ABBREVIATIONS Aeo... .....2-amino-9,10-epoxi-8-oxo-decanoic acid BAC... ....bacterial artificial chromosome BCAT...... ......branched-chain amino acid aminotransferase bp...... base pair CHEF... ......contour—clamped homogenous electricfield FAS...... .....fatty acid synthase GSP... ...gene-specific primer HST...... ......host-specifictoxin HTS... HC-toxm synthetase kb......... .........kilobase kDa......... ...............ki|odalton Mb............ .......megabase NRPS... ...non-ribosomal peptide synthetase ORF...... open reading frame PCR....... .........polymorase chain reaction PLP... ......pyridoxal-5’-phosphate RFLP... ....restriction fragment length polymorphism TLCthIn layer chromatography xi CHAPTER 1 SECONDARY METABOLISM AND FUNGAL TOXINS Introduction It has been estimated that the Mycota, the Fifth Kingdom, has about one and a half million species existing in almost every ecological niche on earth; nearly 70,000 of them have been described (Hawksworth, 1991; Kendrick, 1992). This vast pool of biological and genetic resources represents a considerable portion of all living things on the planet and has a profound effect on human life in many different ways. Since ancient times, human beings have been eating mushrooms and consuming yeast-fermented food and drinks. Ergotism, resulted from the consumption of grain contaminated with ergot alkaloids of Claviceps purpurea, has been responsible for numerous epidemics throughout human history (Beardle and Miller, 1994). In 1970, a widespread plant disease called Southern corn leaf blight, caused by Cochliobo/us heterostrophus race T, which produces a highly specific phytotoxin known as T-toxin, resulted in substantial crop losses in the US. (Ullstrup, 1970). The most noteworthy event in contemporary fungal research has been the discovery of pharmaceutically active compounds. Nobel laureate Alexander Fleming observed that Penicillium notatum can inhibit the growth of the human pathogenic bacterium Staphylococcus aureus and subsequently discovered the first and still very important antibiotic, penicillin (Fleming, 1929; 1946). Later research showed that penicillins are also produced by P. chrysogenum and a variety of other organisms (Lechevalier, 1975). The immunosuppressant drug cyclosporin was first discovered as an antifungal agent produced by Tolypocladium inflatums and Cylindrocarpon Iucidum (Dreyfus et al., 1976). It was later found to have excellent immunosuppressive activity and has been widely used during organ transplantation to suppress immune rejection (Goodman Gilman et al., 1985). Since its initial discovery, cyclosporin has been reported from many strains of Tolypocladium inflatum, 7'. geodes, and T. niveum, as well as from species of Acremonium, Beauvan'a, Fusarium, Paecillomyces, and Verticillium (Sanglier et al., 1990). Among the most successful drugs derived from fungal sources are the cholesterol biosynthesis inhibitors related to lovastatin, which were initially reported by researchers at Merck & Co. from Aspergillus terreus (Vagelos, 1991 ). Related compounds were also reported from Penici/Iium brevicompactum and P. citrinum (Brown et al., 1976; Endo et al., 1976). Concepts of secondary metabolism and secondary metabolite Collectively, the above examples of fungal products with beneficial or hazardous biological activities are called secondary metabolites, which are generally unique to the producing strains but have no confirmed roles in growth and development. Secondary metabolite was first termed by plant natural product chemists in 1891 to describe the materials that were relatively unimportant in the overall physiology of plants (Czapek, 1921). It was not until 1961 that secondary metabolite was applied to microbial products (Bu’Lock, 1961). Many scientists had defined secondary metabolite and secondary metabolism in subtly different ways. Martin and Demain (1978) stated that “Secondary metabolites are those metabolites which are often produced in a phase subsequent to growth, have no function in growth (although they may have a survival function), are produced by certain restricted taxonomic groups of microorganisms, have unusual chemical structures, and are often formed as mixtures of closely related members of a chemical familY’. Campbell (1983) paraphrased that “materials that occur uniquely in a single strain or species, that are found in two or more closely related members of a single genus, or that are found sporadically in a limited number of evolutionarily unrelated species in different genera, families, orders, classes, phyla or kingdoms” should be called secondary metabolites (other terms are shunt metabolite, “special” metabolite, or idiolyte). Bennett and Bentley (1989) proposed a definition for secondary metabolite: “A metabolic intermediate or product, found as a differentiation product in restricted taxonomic groups, not essential to growth and life of the producing organism, and biosynthesized from one or more general metabolites by a wider variety of pathways than is available in general metabolism”. Lately, Vining (1992) summarized three criteria of secondary metabolites: “1. Secondary metabolites are not essential for growth and tend to be strain specific; 2. They have a wide range of chemical structures and biological activities; 3. They are derived by unique biosynthetic pathways from primary intermediates and metabolites". Characteristics of fungal secondary metabolites and their biological activities WIth few exceptions, secondary metabolites are small molecules with diverse structures. They can be as simple as aliphatic acids (e.g., itaconic acid, CsHaO4), or as complex as alkaloid toxins (e.g., palytoxin, C129H223N3054) (Bentley, 1999). Because secondary metabolites are deeply implicated in human life as antibacterial, antifungal, and antitumor compounds, antihelmintics, immunosuppressants, cholesterol-lowering agents, carcinogenic and tumorgenic chemicals, food contaminants, etc. (Table 1), they have a major impact on health, nutrition and economics. Table 1. Biological activities of selected fungal secondary metabolites. Data adapted from Monagnan and Tkacz (1990), VIning (1990), Walton (1996), and Pearce (1997). Name of metabolite Biological activity Producing genus Aflatoxin Carcinogen Aspergillus Apicidin Antiprotozoal Fusarium Auvericin Insecticidal Beauveria Calphostins Antineoplastic Cladosporium Cephalosporin Antibacterial Cephalosporium Cornexistin Herbicidal Paecilomyces Cyclosporin lmmunosuppressant Tolypocladium Ergotrate Vasosuppressant Claviceps Fusarielin Antimitotic Fusarium Gibberellic acid Plant growth stimulating Gibberella HC-toxin Phytotoxic Cochliobolus Ibotenic acid Narcotizing Amanita Lachnumon Antinematode Lachnum Lovastatin Antihypercholesterol Aspergillus Paraherquamide Antihelmintic Penicillium Penicillin Antibacterial Penicillium Viridiofungin Antitumor Tn'choderma Zaragozic acid Antifungal Leptodontidium Zearalenone Estrogenic Fusarium Biosynthesis of fungal secondary metabolites Fungi are prolific producers of secondary metabolites. Early compendia recorded a total of about 5000 fungal metabolites (Shibata et al., 1964; Turner 1971; Turner and Aldridge, 1983). The accelerating pace of new discoveries nowadays has simply made such encyclopedia works practically impossible. Secondary metabolism utilizes a limited number of metabolites from primary metabolism in novel ways. The structural diversity of fungal secondary metabolites reflects the complexity of their biosynthetic routes. In summary, there are four major pathways in secondary metabolite biosynthesis (Figure 1-1) (Adapted from Bentley, 1999). The “isoprene pathway" starts with mevalonic acid derived from acetyl- CoA. The subsequent steps occur by head-to-tail condensation of the “isoprene” unit -C-C(C)-C-C-, often with further biochemical modifications. Trichothecenes (discussed later in this Chapter), which share a trichodiene core made of three isoprene units (called sesquiterpene), are examples of fungal secondary metabolites derived from this pathway. The “polyketide pathway” typically uses acetyl-CoA as starter unit, and the Co-A forms of lower fatty acids and their carboxylated derivatives, such as malonyl-CoA, propionyI-CoA, methylmalonyI-CoA, butyryl-CoA, and ethylmalonyl-CoA, as extender units. Polyketides vary from each other in the length of backbone, degree of reduction, and numbers of modifications. The polyketide family is the largest known group of secondary metabolic compounds 63.65ng5 2:399: Emu—Boom .8 $6353 BEE 52 on... . Tr 9:9...— mo>=u>=ou use «20:958. _ $333.50 .28 32.9.2an o <0 .r I 9955”. AIQSEEoemo 922655. I 11 2685 « 9:8 9:86 6:292 wEom oEEm 25.5.2. I mEEEEm a «03623.. 22:35 _ _Imo>=u>tou can $23329. in fungi. Aflatoxins/sterigmatocystins, fumonisin and T-toxin (discussed in the later part of this Chapter) are polyketides produced by fungi. The “shikimate pathway" uses shikimic acid itself as the starter material, or an intermediate involved in shikimate formation, or a primary metabolite derived from shikimate (e.g., isochorismate). Many plant aromatic products originate in this way, as well as some nonaromatic, cyclohexane structures. Fungal toxins derived from this pathway, however, are not common. The fourth “polypeptide pathway” includes secondary metabolites synthesized from amino acids and amino acid derivatives in a non-ribosomal fashion. This class of compounds is synthesized by multifunctional (cyclic)- peptide synthetases. HC-toxin and victorin (discussed later in this Chapter) are examples of this pathway. While these four pathways are the major biosynthetic routes to secondary metabolites, there are still other possibilities, such as compounds derived directly from carbohydrates without cleavage of the carbon chain and those derived from intermediates of the tricarboxylic acid cycle. Moreover, many secondary metabolites are biosynthesized by a mixed process involving a combination of two or more of these major pathways. A classic example is ergot alkaloids, which consist of both amino acid units and isoprene units (Bentley, 1999). Functions of fungal secondary metabolites Why do fungi produce secondary metabolites? Alternatively, what are the biological functions of secondary metabolites to the producing fungi? There is yet no definitive answer to this question. On the one hand, as defined, metabolites produced through secondary metabolic pathways are not essential for the growth and development of the producing organisms. Early hypotheses favored the conclusion that the secondary metabolism was merely an overflow of energy and intermediates (Bu’Lock, 1961; Woodruff, 1966). On the other hand, from the Darwinian point of view, the presence in an organism of the genetic information and biosynthetic machinery to carry out a complex synthesis implies positive selection. Not only does the process require energy and therefore introduce a growth disadvantage, but also the occurrence of deleterious mutations in a multi- step pathway would be expected to eventually terminate the normal functions, unless the metabolic processes or its products were somehow beneficial to the organism (Vining, 1990). Many putative functions for secondary metabolites have been proposed. The simplest idea is that they dispose of metabolic waste to ease the toxic effects of shunt primary metabolites that are formed in abnormal amounts under nutritional stress (Dhar and Khan, 1971 ). Use of the biochemical machinery of secondary metabolism was postulated to keep the primary pathways, which supply energy and precursors, functional during times when they would otherwise be brought to a standstill and deteriorate. By maintaining its system in an idling mode, the organism would be in readiness to resume growth when the opportunity arose (Woodruff, 1980). Secondary metabolites functioning as chemical signals or sex hormones are generally thought to play roles in modulating cellular differentiation such as mating, sporulation, and dormant spore formation (Demain, 1984; Beppu, 1992). As to those that produce antibiotics, the secretion of such compounds may promote their own survival by inhibiting other antagonistic organisms in a competing environment (Vining, 1990). Secondary metabolites owe their antibiotic activities to their ability to inhibit essential primary metabolic processes. As a classical example, penicillin inhibits the synthesis of bacterial cell walls (Monaghan and Tkacz, 1990). Zaragozic acid is a strong inhibitor of squalene synthase of all yeasts and fungi, while the producer Leptodontidium elatius itself possesses a self-resistance mechanism (Pearce, 1997). Some symbiotic fungi associated with grasses produce neurotoxic compounds which serve to deter consumption by worms, birds, or animals (Clay and Cheplick, 1989; Rowan, 1993). Phytotoxins, especially host-selective toxins, produced by plant fungal pathogens, allow the colonization of healthy plant tissue and therefore confer an ecological advantage to the producers (Walton, 1996; Desjardins and Hohn, 1997). Origin/evolution of secondary metabolism It is argued that microorganisms have evolved the ability to produce secondary metabolites because of the selective advantages thereby acquired (Vining, 1992). In general, only those organisms that lack an immune system are prolific producers of secondary metabolites, which are proposed to act as an alternative defense mechanism (Stone and Williams, 1992). Genetically, it is a widely accepted hypothesis that genes involved in secondary metabolism evolved from those of primary metabolism though gene-duplication, random mutation, and horizontal-gene-transfer. Zahner et al. (1983) described the 10 biochemical arena in which secondary metabolic pathways arise as a “trial and error" mode in which a reasonable level of low-cost inventive evolution is tolerated. As useful products appear, the genes for their synthesis are adopted into the genotype and the process is re-examined by selection. Pathways that prove to have no value are eventually discarded. Genes for the pathways that prove advantageous may be further transferred horizontally and thereby placed in a different, less confining regulatory setting (Hutter, 1986). Molecular genetic data obtained though gene sequence analysis supports the concept that secondary metabolism has arisen by modification of existing primary metabolic reactions. Although amino acid sequence identity deduced from nucleotide sequences of genes from related primary and secondary metabolic pathways is often sufficient to indicate a common ancestry, the match is often better when genes in different rather than in the same species are compared. The information so far available suggests that gene transfer between organisms has been an important factor in the evolution of secondary metabolism. This notion has been exemplified by the discovery of a striking similarity in gene organization, protein structure, and enzymatic functions between polyketide syntheses of the secondary metabolic pathways and fatty acid syntheses of the primary metabolic pathways (Hopwood and Sherman, 1990; VIning, 1992). 11 The existence of gene clusters for fungal secondary metabolite biosynthesis Gene clustering can be broadly defined as the close linkage of two or more genes that participate in a common metabolic pathway (Keller and Hohn, 1997). Studies have shown that genes involved in fungal secondary metabolic pathways are, in most cases, clustered. The discovery of clustered genes for a secondary metabolic pathway in fungi was first demonstrated in the penicillin biosynthesis pathway (Diez et al., 1990; MacCabe et al., 1990). It was the clustering of pathway genes, coupled with the high level of sequence similarity between unrelated organisms, that facilitated the rapid isolation and cloning of genes in the cephalosporin pathway (Skatrud, 1991; Turner, 1992). The most striking examples of clustered genes for fungal secondary metabolism are those of aflatoxinslsterigmatocystins. Brown et al. (1996) identified 25 transcripts within a 60-kb region of the sterigmatocystin gene cluster in Aspergillus nidulans. The majority of genes could be assigned to the biochemically-defined metabolic pathway. Comparative analysis of the aflatoxin gene cluster in A. parasiticus and A. flavus has revealed many genes that are functionally and structurally similar to homologs in A. nidulans (Trail at al., 1995; Yu et al., 1995). Gene clusters for fungal secondary metabolism have also been identified in Alternan’a alternate for the biosynthesis of melanin (Kimura and Tsuge, 1993), in Fusarium sporotn'chioides and Myrothecium roridum for trichothecenes (Hohn et al., 1995; Trapp et al., 1998), in Gibberella fujikuroi for gibberellin (Tudzynski and Holter, 1998), and in Aspergillus terreus for lovastatin (Kennedy et al., 1999). Slightly 12 distant from this “doctrine” is that, in Cochliobulus carbonum, genes involved in HC-toxin biosynthesis, export, and regulation are not tightly clustered; instead, they are scattered over a region of at least 540 kb on a single chromosome (Ahn and Walton, 1996, 1998). Three functional categories of genes are commonly found in secondary metabolic pathway gene clusters. These are genes encoding enzymes and transporters, regulatory genes, and genes conferring self-resistance. The clustering of functionally related genes on a chromosome implies that at least part of their evolution has occurred as a unit. Stone and WIlliams (1992) rationalized that, “if the genes had occurred initially at distant positions on the chromosome, then the production of a secondary metabolite would be favored if the genes were moved closer together, since this would increase the probability for them being passed on as a unit to subsequent generations, or to other species via horizontal gene transfer, and would facilitate mechanisms for their simultaneous or perhaps coordinated expression. Such a clustering of genes would only be selected for if the product conferred a selective advantage to the organism”. Therefore the occurrence of gene clusters is evidence that the products have been advantageous to the producers. Selected fungal toxins: Molecular biosynthesis and roles in pathogenesis Among the secondary metabolites produced by fungi, there is a class of compounds that are notoriously toxic to animals (mycotoxins) orland to plants (phytotoxins). Since the discovery of aflatoxins and their causal role in Turkey X 13 disease in 1960 in England, over 100 toxigenic fungi and more than 300 mycotoxins have been identified from various sources (Wang and Groopman, 1999). These toxins are mainly produced by five genera of fungi: Aspergillus, Penicillium, Fusarium, Alternen'e, and Claviceps. The major mycotoxins are: aflatoxins, sterigmatocystins, ochratoxin, and cyclopiazonic acid made by Aspergillus; patulin, ochratoxin, citrinin, penitrem, and cyclopiazonic acid made by Penicillium; deoxynivalenol, nivalenol, zearalenone, T-2 toxin, diacetoxyscirpenol, fumonisins, moniliformin, and trichothecenes by Fusarium; tenuazonic acid, altemariol, and altemariol methyl ether by Altemaria; and ergot alkaloids by Clavieps (Steyn, 1995). Dozens of phytotoxins including nearly 20 host-specific toxins (HSTs) have also been purified from fungal cultures (reviewed by Scheffer and Livingston, 1984; Kohmoto and Otani, 1991; Walton, 1996). HSTs are produced mainly by two genera of fungi, Cochliobolus (Helminthospon’um) and Alternan'e. The best known HSTs are: HC-toxin from C. carbonum, T-toxin from C. heterostrophus, victorin from C. victon’ae, AM-toxin from A. alternate Apple pathotype, AAL-toxin from A. alternate Tomato pathotype, AF-toxin from A. alternate Strawberry pathotype, AK-toxin from A. alternate Japanese pear pathotype, ACR-toxin from Rough lemon pathotype, ACT-toxin from A. alternate Tangerine pathotype, and AT-toxin from A. altemeta Tobacco pathotype. The studies of fungal toxin toxicity, biogenesis, and mode of action not only have significance for crop protection, post-harvest storage, and food- process decontamination, but also have implications in carcinogenic and 14 medicinal biochemistry research. The remaining part of this Chapter will be devoted to several of the most important fungal toxins, with an emphasis on the molecular genetics of their biosynthesis, mechanisms of toxicity, and roles in phytopathogenesis. 1. Aflatoxinslsterigmatocystins Aflatoxins and sterigmatocystins are a group of polyketide mycotoxins produced by several genera of fungi, mainly the genus Aspergillus. They are derived from highly homologous metabolic pathways (Keller and l-lohn, 1997). Aflatoxins are the end products in A. flavus and A. parasiticus; whereas sterigmatocystins are the end products in A. nidulans. The four naturally occurring aflatoxins are BI, 82, G1 and G2, with B1 (Figure 1-2) usually being found at the highest concentration in contaminated food and feed (Sweeney and Dobson, 1998). Aflatoxins are potent teratogenic, mutagenic, and carcinogenic agents. Afiatoxin B1 is by far the most toxic form of the family and the most studied in mammalian toxicity. Upon entering the mammalian liver, the unsaturated terminal furan ring of afletoxins is converted into an epoxide group by a cytochrome P450. Covalent binding of this epoxide to DNA bases to form a drug-DNA adduct is the nature of aflatoxin tocixity (Wang and Groopman, 1999). Aflatoxins/sterigmatocystins are synthesized by condensation of acetate units, a typical polyketide-forming process. It is estimated that at least 19 gene products are required for aflatoxin biosynthesis, and at least 17 for sterigmatocystins (Brown et al., 1999). The sterigmatocystin biosynthetic IS pathway in A. nidulans is one of the best-studied pathways of fungal secondary metabolism. A sequencing and transcript mapping study of the 60-kb region of the gene cluster revealed 25 coordinately regulated genes (Brown et al., 1996). Most of those genes are predicted to encode enzymes with roles in sterigmatocystin biosynthesis. Among them, one gene encodes a polyketide synthase, two encode a pair of fatty acid synthase a and [3 subunits, five encode P450 monooxygenases, several encode oxido-reductases and dehydrogenases, one encodes a methyltransferase and another an esterase. Specially, one gene (aflR) encodes a transcription factor that regulates the expression of the most, if not all, of the genes in the cluster. There are still a number of genes having no significant homology to sequences in the database. Comprehensive reviews of the biochemistry, genetics, molecular biology and physiology of aflatoxin/sterigmatocystin biosynthesis are available (Minto and Townsend, 1997; Payne and Brown, 1998; Woloshuk and Prieto, 1998; Brown et al., 1999). Aflatoxin production occurs on maize kernels, nuts, and cotton bolls when infected with Aspergillus, mainly A. flevus. However, afiatoxins do not increase fungal virulence to crops because both toxigenic and atoxigenic strains of the fungus colonize plant tissues equally well (Cotty et al., 1994). Therefore, aflatoxins seem to have no role in pathogenesis. 2. Fumonisins Fumonisins are amino polyalcohol mycotoxins produced by several species of the genus Fusarium. Among the most common and highest producing 16 .958. 333 360.3 to 622035 d; 952". 0 c2205 2: _ 5x961 £0 12 1000 if 9% o I .56. _ _ _ _ _ _ _ _ 3.0V . C O O I. O 0 I. O O I. O O I. I. 000: _ 59:95 6:89:05: m . . u_ o Fm c_x2m=< IO IO 0 m < £2 :0 o _ /_\ a n O . _ / O 0001 / I o o 17 species are F. moniliforme (teleomorph Gibberella fujikuroi mating population A). and F. proliferetum, both of which are frequently found on maize (Ross et al., 1990). Fumonisin 81 (Figure 1-2) was first isolated from cultures of F. moniliforme by Gelderblom et al. (1988) and structurally characterized by Benzuidenhout et al. (1988). Fumonisin B1 is the most prevalent component of fumonisins in naturally contaminated maize. Fumonisins 82, Ba, and 8.; were later identified as minor forms (Plattner et al., 1992; Marasas, 1996). Fumonisins induce a wide range of adverse effects in experimental animals. Among the observed toxicological symptoms are equine leucoencephalomalacia (Kellerman et al., 1990), porcine pulmonary edema (Calvin and Harrison, 1992), and hepatic cancer in rats (Gelderblom et al., 1993). Epidemiological study has suggested a linkage between high rate of esophageal cancer in humans and high levels of consumption of fumonisins in maize products (Marasas, 1995). The mechanism of action of fumonisins appears to involve binding to proteins, probably thereby causing the conversion of an amino acid side chain to the anhydride followed by covalent coupling to other proteins (Shier et al., 1997). Fumonisins are structurally similar to the long-chain backbones of sphingolipids. Studies have found that fumonisins inhibit the activity of sphingosine N-acetyltransferase, which results in the accumulation of toxic sphingoid bases. Sphingolipids are essential components of biological membranes. Sphingosine, the backbone of Sphingolipids, acts as an intracellular regulator and second messenger mediating normal cellular development (Shiel et al., 1997). Therefore, the correlation between fumonisin toxicity and 18 tumorogenesis is rationalized, but more study is needed to verify the current working model. The biosynthesis of fumonisins is poorly understood. The structural similarity between fumonisins and sphingosine prompted the idea that they may be biosynthetically related (Plattner and Shackelford, 1992). Sphingosine biosynthesis begins with the condensation of palmitoyI-CoA with serine. Fumonisin B1 has an alanine instead of serine, and an 18—carbon fatty acid instead of 16-cerbon fatty acid. Isotope-feeding studies demonstrated that alanine is indeed a biosynthetic precursor of fumonisin B1, and that the polyalcohol moiety is derived from acetate (Blackwell et al., 1996). Genetic studies indicated that at least some of the genes involved in fumonisin production are closely linked to form a putative gene cluster on chromosome 1 of G. fujikuroi (Desjardins et al., 1996; Xu and Leslie, 1996). Very recently, Proctor et al. (1999) cloned a polyketide synthase gene (FUM5) by degenerate PCR approach, and showed that the new gene is involved in fumonisin biosynthesis in Gibberella fujikuroi mating population A. FUM5 has an ORF of 7.8 kb. The predicted gene product is highly similar to bacterial and fungal Type I polyketide syntheses. Transformation of a cosmid clone carrying FUM5 into G. fujikuroi enhanced production in three strains and restored wild-type production in a fumonisin non- producing mutant. Disruption of FUM5 reduced fumonisin production by over 99%. FUM5 is the first fumonisin biosynthetic gene obtained. Fumonisin-producing Fusarium species are routinely found in maize tissues, such as root, stalks, kernels, and seeds. Species of Fusarium have also 19 been found in sorghum, millet, rice, wheat, cotton, etc. The high frequency of fumonisin contamination in maize suggests the possibility that fumonisins play a role in pathogenesis (Desjardins and Hohn, 1997). Low concentrations (1-100 uM) of pure fumonisins have been shown to cause necrosis and other disease symptoms in maize and other plant seedlings (Lamprecht et al., 1994). Genetic analysis has also shown an association between production of fumonisin and high levels of virulence on maize seedlings (Desjardins et al., 1995). Additionally, the structural similarity of fumonisins to AAL-toxin, the host-specific toxin of Alternan’a alternate Tomato pathotype, suggests a role for fumonisins in pathogenesis (\NInter et al., 1994). The future cloning of the fumonisin biosynthesis genes and creation of null mutants will unambiguously define the role of fumonisins in this regard. 3. Trichothecenes Trichothecenes are a large family of sesquiterpenoids produced by several genera of fungi, mainly Fusarium and Myrothecium, and by at least two species of the plant genus Bacchen’s (Jarvis, 1991; Desjardins et al., 1993). All trichothecenes contain a tricyclic nucleus named trichothecene (Figure 1-2) and usually have an epoxide at 0-12 and C-13 that is essential for toxicity. The total number of naturally occurring trichothecenes known today exceeds 60 (Desjardins et al., 1993). Structurally, trichothecenes can be classified into two distinct categories. Linear trichothecenes, such as T-2 toxins, diacetoxyscirpenol and deoxynivalenol, are produced by Fusarium species and are the major source 20 of trichothecene contamination in agricultural products. Macrocyclic trichothecenes, such as verrucarin A and baccharinoid B7, are produced by Myrothecium species and the plant species Baccheris (Trapp et al., 1998). The discovery that some higher plant species make highly similar trichothecenes prompted the hypothesis that plants originally obtained the genes for trichothecene biosynthesis though horizontal-gene-transfer from a trichothecene- producing fungus (Jarvis, 1991). Trichothecenes are potent inhibitors of protein synthesis in eukaryotic cells (McLaughlin et al., 1977). They bind to the peptidyl transferase active site within the 608 subunit of the ribosome, thereby inhibiting the initiation, elongation and termination reactions (Feinberg and McLaughlin, 1989). Macrocyclic trichothecenes are about 10-fold more toxic than the Fusarium trichothecenes. In general, trichothecenes induce toxic effects in animals, such as vomiting, oral lesions, dermatitis, and hemorrhaging, and they have been implicated in mycotoxicosis in both humans and animals (Sharma and Kim, 1991). Trichothecene biosynthesis starts with the cyclization of the isoprenoid biosynthetic intermediate farnesyl pyrophosphate to trichodiene, catalyzed by the central enzyme trichodiene synthase. Subsequent steps involve a variety of oxygenations, isomerizations, cyclizations, and esterifications (Desjardins et al., 1993). Genes for trichothecene biosynthesis in F. sporotn'chioides and M. ron’dum are clustered. A total of 10 clustered trichothecene biosynthesis genes have been identified within a 25-kb region in F. sporotn‘chioides. Among them, eight are predicted to encode biosynthetic enzymes (but only three have been 21 biochemically studied), one gene encodes a pathway transcription factor, and another encodes a transport protein. Apparent homologs of several Fusarium pathway genes have been found within a 40-kb region in M. roridum. However, the distances between genes in M. ron'dum are greater, compared to F. sporotrichioides. The gene order in M. roridum is also different from that in F. sporotrichioides (Hohn et al., 1993; Keller and Hohn, 1997; Trapp et al., 1998). Trichothecenes appear to contribute to the virulence of certain phytopathogenic Fusarium species on some host plants (Desjardins et al., 1989, 1992). At very low concentrations (1 pM to 1 mM), trichothecenes cause wilting, chlorosis, necrosis, and other disease symptoms in a wide variety of plants (Cutler, 1988). For example, simple trichothecenes such as T-2 toxin and deoxynivalenol inhibit protein synthesis in maize leaf disks and kernel sections and the growth of wheat coleoptiles (Casale and Hart, 1988; Wang and Miller, 1988). Complex macrocyclic trichothecenes of M. roridum induce chlorotic and necrotic lesions on muskmelon leaves (Kuti et al., 1992). UV-induced mutants of F. sporotrichioides that lost trichothecene production were much less virulent than the wild type strain on parsnip root (Desjardins et al., 1989). However, molecular genetic studies generated unexpected results. When a critical gene (Tox5) encoding trichodiene synthase was disrupted in F. sporotrichioides, the virulence of toxin-nonproducing mutants on parsnip roots was significantly reduced when compared with the virulence of the parental strain; on potato tubers, in contrast, the virulence of mutants was unchanged (Desjardins et al., 1992). A plausible explanation for these results is that protein synthesis or some 22 other trichothecene target sites in parsnip and potato cells have different degree of sensitivity to trichothecenes. 4. T-toxin T—toxin is a group of linear long-chain (035 to C41) (Figure 1-2) polyketide molecules produced exclusively by Cochliobolus heterostrophus race T. Race 0 of C. heterostrophus does not produce T-toxin. Along the methylene backbone in T-toxin, there are repeated B-oxydioxo groups that are regularly spaced (Kono and Daly, 1979; Kono et al., 1981). Cochliobolus heterostrophus was first described in 1925 as a weak fungal pathogen of maize (Drechsler, 1925). An extremely vimlent race (named race T, in contrast to the original race 0) first appeared in the USA in 1969 and subsequently caused a severe epidemic on Texas male sterile (T) cytoplasm maize that resulted in a 15% crop loss in 1970 (Ullstrup, 1970). T-toxin is the disease determinant of race T and is regarded as a host-specific toxin (HST) due to its specific toxicity toward T-cytoplasm maize, not to other cytoplasmic types of maize. Extensive studies have demonstrated that T-toxin binds to a protein (URF- 13) unique to the inner mitochondrial membrane of T-cytoplasm maize (Dewey, et al., 1987, 1988). Expression of URF-13 in Escherichia coli, yeast, tobacco, or insect cells causes the cells to become sensitive to T-toxin, providing definitive evidence that URF-13 is responsible for T-toxin sensitivity (Dewey et al., 1988; Glab et al., 1990; von Allmen et al., 1991; Korth and Levings, 1993). Physical 23 biochemical studies and modeling indicate that URF-13 forms oligomeric pores in mitochondrial membranes in the presence of T-toxin (Korth et al., 1991; Levings et al., 1995). After 1971, T-cytoplasm maize was replaced by normal (N) cytoplasm maize, and C. heterostrophus race T virtually disappeared as a severe plant pathogen in the field. Classical genetic analysis showed that progeny from a cross between race T and race 0 of C. heterostrophus segregate in a 1:1 ratio with regard to T- toxin production and hence high virulence toward T—cytoplasm maize (Yoder and Gracen, 1975). The conclusion was then drawn that a single genetic locus (named TOX1) controls the production of T-toxin (Leach et al., 1982; Bronson, 1992). However, recent studies revealed that T-toxin production is actually governed by two loci (named TOX1A and TOX1B) that reside on two different chromosomes (Turgeon et al., 1995; Kodama et al., 1999). Additional genetic mapping indicated that a total of ~1.2 Mb of extra DNA is present in race T but not in race 0. This stretch of DNA is likely present at both the TOX1A and TOX1B loci (Rose, 1996). To clone the T-toxin biosynthesis genes, restriction enzyme mediated integration (REMI) was used to randomly mutagenize C. heterostrophus. Among approximately 1300 transformants recovered from the procedure, two no longer produced T-toxin (Lu et al., 1994). Subsequently, plasmid rescue was used to clone two genes that were mutated by insertions and were shown to be involved in T-toxin biosynthesis. The gene PK81 (TOX1A) encodes a polyketide synthase and is thought to synthesize the backbone of T- toxin (Yang et al., 1996) and a second gene DEC1 (TOX1B) encodes a putative 24 decarboxylase that probably decarboxylates the product of the polyketide synthase (Rose et al., 1996). Targeted inactivation of either gene resulted in loss of T-toxin production and reduced virulence on T-cytoplasm maize. PKS1 and 0501, and a putative reductase-encoding gene (RED1) that is linked to 0501 but has no apparent role in T-toxin biosynthesis, appear to be single-copy genes and are all found only in race T, but not in race 0, of C. heterostrophus. It is likely that these genes locate within the ~1.2 Mb of extra DNA (Kodama et al., 1999). 5. Victorin Victorin is a collection of closely related cyclic pentapeptides produced by Cochliobolus victoriae, which causes VIctoria blight of cats (Meehan and Murphy, 1946, 1947). The amino acids in victorin are extensively modified. The most abundant form of victorin is victorin C (Figure 1-2), which consists of 5,5- dichloroleucine, 3-hydroxylysine, chloroacrylic acid, and a cyclic alpha-amino acid derivative called victalanine. Additional minor forms of victorin have been identified and named as victorin B, D, E, and victoricine; they differ from victorin C in the degree of chlorination and hydroxyletion of various side chains (Macko et al., 1985; Wolpert et al., 1985, 1986). The toxicity of victorin is host-specific. Only oat lines containing Victoria- type resistance to crown rust (caused by Puccinie coronata) are susceptible to C. victoriee and sensitive to the toxin. Sensitivity to victorin and susceptibility to C. victoriae cosegregate and are controlled by a single dominant nuclear gene (Vb) (Litzenberger, 1949). Vb has been postulated to encode a receptor for victorin 25 (Scheffer and Livingston, 1984), but definitive proof of this hypothesis is still lacking. Great progress in elucidation of the mode of action of victorin has been achieved during the past decade. An in vivo assay identified a 100-kD protein that binds to 125l-labeled victorin C in a ligand-specific manner, and only protein extracts from oats carrying the Vb gene bind (Wolpert and Macko, 1989). This in vivo binding is competitive, covalent, and appears to be correlated with the biological activity of victorin. The gene encoding this 100-kD protein was cloned and the deduced protein sequence indicated that it encodes the P-protein component of the glycine decarboxylase complex (GDC), which is located in the mitochondrial matrix (Wolpert et al., 1994). VIctorin also binds to the 15-kD H- protein component of GDC in both susceptible and resistant oat protein extracts. The binding of victorin to two components of GDC inhibits the normal function of GDC during the photorespiratory cycle in susceptible oat tissues, which results in a series of senescence-like responses, such as the site-specific proteolytic cleavage of the large subunit of ribulose-1,5-bisphosphate carboxylase/oxygenase (Rubisco), loss of chlorophyll, and DNA leddering (Navarre and Wolpert, 1995, 1999). These results suggest that GDC is the target of victorin toxicity, but whether P- or H-protein or neither component of GDC is the product of the Vb gene still remains unclear. Genetic analysis indicated that victorin production in C. victoriae is controlled by a single genetic locus, TOX3 (Scheffer et al., 1967; Bronson, 1992). Considering that victorin is a cyclic pentapeptide, it is likely to be synthesized by a multifunctional cyclic peptide synthetase and many other enzymes responsible 26 for the chemical modifications. However, no protein has been purified or gene cloned for the biosynthesis of victorin. 6. HC-toxin HC-toxin is a small family of cyclic tetrapeptides produced by Cochliobilus carbonum race 1, the causal pathogen of Northern corn leaf spot. The major and most toxic component of HC-toxin, HC-toxin I (Figure 1-2), has the structure cyclo(D-Pro-L-Ala-D-Ala-L-Aeo), where Aeo stands for 2-amino-9,10-epoxi-8- oxo-decanoic acid (Pringle and Scheffer, 1967, 1971; Gross et al., 1982; Walton et al., 1982; Pope at al., 1983; Kawai et al., 1983). The biogenesis of the Aeo backbone as a fatty acid has been proved by 13C-acetate labeling and NMR analysis (Cheng et al., 1999; See Appendix). Three minor components (HC-toxin II, III, and IV) differ slightly in amino acid composition and have significant differences in toxicity (Rasmussen, 1987; Rasmussen and Scheffer, 1988). HC- toxin II is about 50% as toxic as HC-toxin l and has glycine in place of D-alanine; HC-toxin III has about 10% relative activity and contains trans-3-hydroxyproline instead of proline; HC-toxin IV has only 1% relative activity and has a hydroxyl group instead of a carbonyl group at position C8 in Aeo. Those HC-toxin components can be separated by TLC or reverse phase HPLC due to their differences in polarity. Northern corn leaf spot disease was first reported in 1938 on an inbred maize line, Pr, in Indiana. Disease symptoms include grayish tan necrotic spots with concentric rings on the leaves and ears. Longer development often results in 27 complete desiccation of leaves and plant death (Ullstrup 1941; Multani et al., 1998). Among four described races of C. carbonum that can be distinguished on the basis of their host range and lesion morphology, only race 1 produces HC- toxin and only race 1 is exceptionally virulent on maize lines that are homozygous recessive at the nuclear Hm1 locus (hm1lhm1). HC-toxin appears to be the sole determinant of fungal pathogenicity, although the various cell-wall- degrading enzymes secreted by the fungus might play a complementary role in fungal virulence. Genetic variants that do not produce HC-toxin are unable to colonize much beyond the site of initial penetration and, therefore, cause only small necrotic flecks on leaves (Comstock and Scheffer, 1973). The correlation between HC-toxin production and high virulence has been proved by gene- knockout experiments (Ahn and Walton, 1997, 1998; Panaccione et al., 1992; Chapters 3 and 4 in this dissertation). Physiologically, HC-toxin inhibits maize root growth and chlorophyll biosynthesis (Rasmussen and Scheffer, 1988), but does not inhibit protein, RNA, or DNA synthesis (Ciuffetti et al., 1995). Biochemically, HC-toxin inhibits histone deacetylase, perhaps thereby suppressing host defense responses. Acetylated isoforms of histone H4 and H3 (but not H2A and H28) accumulate in HC-toxin- treated maize tissues in a host-specific manner (Brosch et al., 1995; Ransom and Walton, 1997). Both the 8-carbonyl and the 9,10-epoxide groups are necessary for toxicity (Walton and Earle, 1983; Kim et al., 1987). HC-toxin IV with the carbonyl group hydrolyzed has only 1% toxicity compared with HC-toxin I (Rasmussen, 1987). 28 Although race 1 can be extremely virulent, it is normally not a serious economic problem because modern maize breeding programs have nearly eliminated the recessive allele hm1. Hm1 encodes HC-toxin reductase (HCTR), which reduces the 8-carbonyl group of Aeo to the alcohol (i.e., 8-hydroxyl as in HC-toxin IV) thereby detoxifies HC-toxin (Meeley and Walton, 1991; Meeley et al., 1992; Johal and Briggs, 1992). Homologs of Hm1 are present in many other plant species, such as wheat, barley, rice, sorghum, and oats, and all monocots tested except the susceptible maize inbreds possess HCTR activity. It has been postulated that monocots might share a common ancestor Hm1 gene for protection against fungal pathogens that used HC-toxin or related cyclic tetrapeptides as pathogenicity factors. However, it is unknown how dicots defend themselves against HC-toxin since no HCTR activity has been detected in them so far (Multani et al., 1998; Meeley et al., 1992). Pathogenicity of C. carbonum race 1 is determined by a single Mendelian locus, TOX2, which also confers the ability to produce HC-toxin (Nelson and Ullstrup, 1961; Scheffer et al., 1967; Ahn and Walton, 1996). A list of biochemical activities required for HC-toxin production has been proposed (Table 2). Prior to the work presented in this dissertation, there were four genes (collectively called TOX2) that had been cloned and demonstrated to be involved in HC-toxin biosynthesis, export, or regulation. The first TOX2 gene, cloned by a reverse genetic approach, was HTS1, which encodes a 570-kDa tetrapartite non- ribosomal peptide synthetase (NRPS) called HC-toxin synthetase (HTS) (Panaccione et al., 1992; Scott-Craig et al., 1992). HTS is the central enzyme in 29 Table 2. Biochemical Activities Required for HC-toxin Production. Activity Gene Reference Cyclic peptide synthetase (CPS) H TS1 Scott-Craig et al., 1992; and Proline racemase HTS1 Panaccione et al., 1992 Alanine racemase TOXG“ Cheng and Walton, submitted Aeo synthesis - FAS** a subunit ? - FAS“ [3 subunit TOXC Ahn and Walton, 1996 - Amino group TOXF* Cheng et al., submitted - Epoxide group ? - Carbonyl group ? Efflux pump TOXA Pitkin et al., 1996 Regulatory protein 1 TOXE Ahn and Walton, 1998 Other regulatory proteins 7 * Research presented in this thesis. See Chapters 3 and 4. ** FAS, fatty acid synthase. HC-toxin biosynthesis. It activates and cyclizes the four amino acid constituents of. HC-toxin, and epimerizes L-proline to D-proline; TOXA, which is clustered with HTSl but transcribed divergently, encodes a putative HC-toxin transporter of the major facilitator superfamily (Pitkin et al., 1996). The product of TOXC is predicted to be a fatty acid synthase [3 subunit that works together with an unknown a subunit to synthesize the decanoic acid backbone of Aeo (Ahn and Walton, 1997). TOXE encodes a regulatory protein required for the expression of TOXA and TOXC but not HTS1 (Ahn and Walton, 1998). All of the known TOX2 genes are present only in HC-toxin-producing (Tox2‘) isolates of C. carbonum in multiple functional copies, are dedicated to HC-toxin production, and, with one 30 exception, are linked over a ~540 kb region on one chromosome (Ahn and Walton, 1996, 1998). No homologous sequences to those genes have been detected in HC-toxin-non-producing (Tox2') strains of the fungus. When all copies of an individual gene are mutated (except TOXA, which would be lethal), the mutants grow and develop normally but lose the ability to produce HC-toxin and no longer cause disease on susceptible maize. The study of HC-toxin also has medical implications. There are five classes of HC-toxin analogs: chlamydocin, Cyl-2, WF-3161, trapoxin and apicidin, all of which are produced by fungi and share striking similarity in structure (Walton, 1990; ltazaki et al., 1990; Derkin-Rattray et al., 1996). Both chlamydocin and HC-toxin have demonstrated cytostatic and antimitotic activities against transformed mammalian cells (Walton et al., 1985; Shute et al., 1987). Cyl-2 and WF-3161 have some degree of antitumor activity (Hirota et al., 1973; Umehara et al., 1983). Apicidin exibits potent antiprotozoal activity against Apicomplexan parasites (Darkin-Rattray et al., 1996). Specially, trapoxin induces morphological reversion of v-sis oncogene-transformed NIH3T3 fibroblast cells, as well as human tumor cells such as HeLa and T24 cells (ltazaki et al., 1990; Yoshida et al., 1992; Kijima et al., 1993). Mammalian cell lines treated with trapoxin accumulate highly acetylated core histones (H3 and H4). Nanomolar concentrations of trapoxin irreversibly inhibit deecetylation of acetylated histones. Later, Taunton et al. (1996) successfully used trapoxin as an affinity ligand to purify human histone deacetylase and subsequently cloned the gene. Since then, studies of histone deacetylase and acetylase in mediating chromatin structure 31 and gene expression have surged (Davie, 1996; Pazin and Kadonaga, 1997; Tsukiyama and Wu, 1997; Kouzarides, 1999). The genetically development of novel non-ribosomal peptide drugs for cancer chemotherapy is in the spotlight. As NRPSs are organized into distinctive functional domains, rational design of hybrid genes by the exchange of domain- coding regions among NRPSs leads to novel enzymes and the production of novel products (Stachelhaus et al., 1995; Jung et al., 1997; Cane et al., 1998). However, there is still much to learn about NRPSs. Objectives of this dissertation This century has witnessed the emergence and disappearance of three severe plant diseases all caused by Cochliobolus species (Scheffer and Livingston, 1984; Walton, 1996). The constant and frequent evolution of fungal genomes within and among species deserves close attention from fungal geneticists and plant pathologists, as well as from crop breeders (Dobinson and Hamer, 1993; Walton and Panaccione, 1993). The leading study of the interaction between C. carbonum and maize, in regard to aspects of host-specific toxin biosynthesis, molecular basis of toxicity and host resistance, has set a model for studying other plant fungal pathogens (Walton and Panaccione, 1993; Walton, 1996). However, there are still many unknown details with regard to the HC-toxin biosynthesis pathway. Based on the accumulated knowledge prior to the studies covered in this dissertation, we did not know how D-alanine was synthesized, what route led to Aeo biogenesis, or how many extra genes are 32 needed for HC-toxin production and hence pathogenicity. The objectives of this study were to contribute my effort to bring us one step closer to completely elucidating the biochemical and molecular nature of HC-toxin biosynthesis. 33 CHAPTER 2 CLONING AND ANALYSIS OF TOXD, A GENE THAT BELONGS TO THE TOX2 LOCUS BUT HAS NO APPARENT ROLE IN HC-TOXIN PRODUCTION The gene sequence reported in this Chapter has been deposited in GenBank with accession number X92391. 34 ABSTRACT Due to the production of a host-specific toxin, HC-toxin, race 1 isolates of the phytopathogenic fungus Cochliobolus carbonum are exceptionally virulent on maize lines that are recessive at the Hm1 locus (genotype hm1lhm1). Genetic analysis showed that a single Mendelian locus, TOX2, controls I-lC-toxin production. The TOX2 locus is complex in that it spans more than 540 kb on a single chromosome, and contains multiple genes, each with multiple copies. Previous studies identified four genes (HTS1, TOXA, TOXC, and TOXE) that are dedicated to HC-toxin production. All of them are present only in HC-toxin-producing (Tex?) isolates of C. carbonum. When all copies of any individual gene were mutated by targeted-gene disruption or replacement, the mutants lost HC- toxin production and no longer caused disease on susceptible maize. A new gene, named TOXD, was cloned based on its unique presence in Tox2“ isolates of C. carbonum. TOXD is present in three copies in the standard lab strain SB111, also but only one copy in strain 164R10. TOXD is regulated by TOXE, a regulatory gene that also controls TOXA and TOXC expression. TOXD is predicted to encode a 297-amino acid-protein that has sequence homology to a protein involved in lovastatin biosynthesis in Aspergillus terreus. TOXD was disrupted in strain 164R10 to create a null mutant. The null mutant had normal phenotypes in growth and development. Disruption of TOXD had no apparent effect on HC-toxin 35 production or fungal pathogenicity. The biochemical function of TOXD is unknown. INTRODUCTION Interactions between plants and their fungal pathogens are complicated by the fact that pathogens possess multiple ways to infect and colonize their host plants, and because plants can mount multifaceted defense strategies against pathogens. As a result, the occurrences of plant diseases in nature are relatively low, compared to the frequent encounters between plants and potential pathogens such as viruses, bacteria, fungi, and nematodes (Heath, 1991). However, when a severe plant disease does occur, the consequences can be serious. In human history, the outburst of late blight of potato in the 1840s in Northern Europe caused the death of hundreds of thousands of human lives and forced millions to desert their homes. In 1970, the sudden appearance of Southern corn leaf blight caused the loss of 15% of that year’s maize production in the United States (Agrios, 1988). In general, fungal pathogens may use one or more of the following strategies for pathogenesis: mechanical penetration of plant cell epidermis with special infection structures such as appressoria and penetration pegs, secretion of cell wall-degrading enzymes to chemically dissolve plant cell walls, production of toxic chemicals to condition plant tissues for colonization, and detoxification of 36 plant defensive compounds (Yoder, 1980). If a single strategy employed by a pathogen is necessary and sufficient to cause disease on a certain plant host, that strategy can be called a “pathogenicity factor"; otherwise, collectively such factors are called “virulence factors” because their actions are important but not decisive (Yoder, 1980). Cell wall-degrading enzymes are typically virulence factors, whereas host-specific toxins (HSTs) are generally regarded as pathogenicity factors. HSTs are low molecular weight compounds produced exclusively by fungi, mainly the two genera Cochliobolus and AIternan'a (Walton and Panaccione, 1993). They are the determinants of host range and disease specificity. A fungus that produces a HST causes more disease on its host plant than one that is otherwise identical but does not produce the HST. A good example is victorin, the most phytotoxic and most selective HST known, which is produced by C. victoriae. Victorin inhibits the growth of sensitive cats at picomolar concentrations, but does not affect resistant cats or any other plants even at a one million-fold higher concentration (Walton and Earle, 1984). Well-studied HSTs also includes T-toxin and HC-toxin, produced by C. heterostrophus and C. carbonum, respectively (Walton, 1996). Classical genetic analysis indicated that the production of HSTs by species of Cochliobolus is controlled by single Mendelian loci, called TOX1 for T-toxin in C. heterostrophus, TOX2 for HC-toxin in C. carbonum, and TOX3 for victorin in C. victoriae (Scheffer et al., 1967; Bronson, 1992). More recent studies revealed that T-toxin production is actually governed by two loci (TOX1A and TOX1B) that reside on two different 37 chromosomes (Turgeon et al., 1995; Kodama et al., 1999), and TOX2 is a complex locus containing duplication of multiple genes dedicated to HC-toxin production (Ahn and Walton, 1996, 1998). At the molecular level, two genes (PKS1, DEC1) have been identified and proven to participate in T-toxin biosynthesis (Rose et al., 1996; Yang et al., 1996), and four genes (HTS1, TOXA, TOXC, TOXE) for HC-toxin biosynthesis, export and regulation (Panaccione et al., 1992; Scott-Craig et al., 1992; Pitkin et al., 1996; Ahn and Walton, 1997, 1998). All of these genes are dedicated to toxin production and are present only in toxin-producing (Tox*) strains of the fungal species. Toxin-non- producing (Tox’) strains do not possess these genes or homologous DNA. A total of ~1.2 Mb DNA is present only in race T (T-toxin-producing) but not in race 0 (T-toxin-non-producing) isolates of C. heterostrophus (Rose, 1996), and the TOX2 locus of C. carbonum encompasses at least ~540 kb on a single chromosome (Ahn and Walton, 1996). These facts lead to some interesting questions. On the one hand, are genes devoted to toxin production all unique to toxin-producing strains? On the other hand, does every gene unique to toxin- producing strains have a role in toxin production? In this Chapter, I describe the cloning and characterization of a gene, named TOXD, which belongs to the TOX2 locus and is unique to HC-toxin-producing strains of C. carbonum, yet has no apparent role in HC-toxin production or fungal pathogenicity. 38 MATERIALS AND METHODS Fungal Strains. Strains of C. carbonum used in this study were maintained as glycerol stocks at —80 °C. Growth conditions for fungal cultures on solid V8- plates or in modified Fries’ liquid medium were previously described (Van Hoof et al., 1991). SB111 (ATCC 90305) and SB114 are standard Tox2“ (HC-toxin- producing) and Tox2' (HC-toxin-non-producing) laboratory strains of C. carbonum, respectively. 164R5 though 164R18 are random progeny from a cross between $8111 and SB114 (Ahn and Walton, 1997). Field isolates 81-64 (race 1), 1368 (race 2) and 1309 (race 3) were originally obtained from K. Leonard (University of Minnesota, St. Paul, MN). Field isolates 141 R, 151, 161, 171, 181, 1101 and 1111 were generous gifts from L. D. Dunkle (Purdue University, West Lafayette, IN; Jones and Dunkle, 1993). Nucleic Acid Manipulations and Sequence Analysis. C. carbonum DNA and RNA isolation was previously described (Pitkin et al., 1996). DNA and RNA blotting, probe labeling, hybridization, cDNA and genomic library screening, and DNA subcloning were done following standard procedures (Sambrook et al., 1989). Oligonucleotides were synthesized at the Michigan State University (MSU) Macromolecular Structure Facility. Automated DNA sequencing was done at the MSU DNA Sequencing Facility. Sequences were assembled and analyzed with the DNASter Software package (DNASter Inc., Madison, WI). Protein sequence alignments were generated with the CIusta/W Program (Thompson et 39 al., 1994). The transcriptional start site of TOXD was determined by 5’ RACE (Rapid Amplification of cDNA Ends) using a kit from Gibco-BRL (Frohman et al., 1988). The primer for reverse transcription (GSP1) was 5’- GTCGTCGGGTATGGCTGG-3’ and the nested primer for PCR amplification (GSP2) was 5’-GAATATCACCCTTGACGG-3’. Pulsed-field Gel Electrophoresis. Agarose gel-embedded intact chromosomal DNA was prepared as described (Ahn, 1996). Contour-clamped homogenous electric field (CHEF) electrophoresis was performed in a CHEF-DR II apparatus (Bio-Red, Richmond, CA). Running conditions were: 0.8% chromosomal-grade agarose gel (Bio-Red), 50 V with a 15 to 30 min switching interval for 72 hr, and a 10 to 20 min switching interval for 72 hr, in 0.5 x TBE buffer (22.5 mM Tris, 22.5 mM boric acid, 0.5 mM EDTA, pH 7.6) at 14 °C. The gel was stained with ethidium bromide for 30 min and destained in distilled water for 30 min before being photographed under UV light and blotted to a nylon membrane. Disruption of TOXD in Strain 164R10. Fungal protoplast preparation, transformation, selection, and single-spore isolation of transformants were done as described (Panaccione et al., 1992; Pitkin et al., 1996). To make the gene disruption vector pTOXDD4, a 611-bp internal fragment (from 540 bp to 1151 bp in the ORF) was amplified by PCR, filled-in, and digested with Sec I to generate a 320-bp fragment (from 540-bp to 860-bp in the ORF) that contains a unique Msc I site in the middle. This 320-bp DNA was subcloned into the Sec llSma I 40 (filled-in) sites of pGEM-72 (Promege, Madison, WI) to make an intermediate construct pGEMTOXD. A 350-bp Xbe llSac I fragment was released from pGEMTOXD and subcloned into the Xbe llSac l sites of pHYG1 (Sposato et al., 1995) to make the final disruption vector pTOXDD4 (5.8 kb). pTOXDD4 was linearized with Msc l before transformation. Strain 164R10 was transformed with linearized pTOXDD4. Tramsformants (named T492) were selected from medium containing 120 jig/ITII hygromycin. Analysis of HC-toxin and Pathogenicity Test. HC-toxin was extracted with chloroform from 28 d-old fungal culture filtrates as described (Walton et al., 1982). Extracts were separated on thin layer chromatography (TLC) plates (Si250-PA, J. T. Baker, NJ) and detected using an epoxide-specific spray as described (Meeley and Walton, 1991). Pathogenicity was assayed by spraying 2- week-old maize (Zea mays L.) inbred Pr (hm1/hm1) with a spore suspension (1 x 10‘Iml in 0.1% Tween-20). Plants were observed daily for a week and representive leaves were photographed at 4-d post-inoculation. RESULTS Discovery of TOXD. A study of restriction fragment length polymorphisms (RFLP) of TOXC revealed that when two subclones of TOXC (pA4-1 containing the 5’-region of TOXC and pA6-4 containing the 3’-region of TOXC) were used as probes against genomic blots of $81 11 DNA digested with EcoRV, Sal I, Xbe 4| I, or Xho I, an extra band appeared to the hybridization with pA4-1 (Figure 2-1; Ahn and Walton, 1997). The 6.5-kb band from the EcoRV digestion was subsequently subcloned and used as a probe to obtain a 1.1-kb cDNA clone (pTOXDC4). The corresponding gene was later named as TOXD. A genomic clone, pTOXDG5, containing a 3.1-kb Xho I insert (Figure 2-6D, lane 1), was also obtained. Characteristics of TOXD. A survey by Southern hybridizations showed that TOXD was present only in Tox2+ (race 1) isolates of C. carbonum and not in any tested Tox2‘ (race 2 or 3) strains (Figure 2-2). TOXD cosegregated with HC-toxin production. TOXD has three copies in most strains surveyed, but only one copy in 164R10 (Figure 2-2A, lane 8). A 1.1-kb message was detected by Northern hybridization (data not shown). Chromosomal mapping determined that the three copies of TOXD are physically tightly linked to the known TOX2 genes in strain SB111 within TOX2 locus on a 3.5-Mb chromosome (Figure 2-3; Ahn and Walton, 1996). Like TOXA and TOXC, the expression of TOXD is regulated by TOXE (Ahn and Walton, 1998). Near-full-length cDNA (pTOXDC4) and genomic (pTOXDG5) clones were sequenced on both strands. The 5’-end of the transcript was amplified by RACE and also sequenced. Sequences were assembled using the DNASter program (Figure 2-4) and deposited in GenBank with accession number X92391. There are no introns in TOXD, and the experimentally determined transcriptional start site is only 16 bp upstream of the translational start site. The TOXD open reading frame encodes a predicted 32.5-kDa protein 42 Q 23.1- __ __ __ m , nun- .... ~, 66— .... :3 " an 4.4— A B Figure 2-1. Restriction fragment length polymorphism (RFLP) of TOXC. Genomic DNA of SB111 was digested with (1) EcoRV, (2) Sal I, (3) Xbe l, or (4) Xho I. Two identical blots were hybridized with probes of (A) pA4-1 containing the 5’-region of TOXC and (B) pA6-4 containing the 3’-region of TOXC. EcoRV and Xbal digestions could not distinguish between the three copies of TOXC in SB111. 43 OFNCOVLOCOQ Ffl'mohwmw—Fv—Fv—w—Fwfi- :zsssssssmssmseoee Strain mmcocococccccocoEccmEcocoJ—cooo (DCDFT—v—w—v—Y-v-v—v—v—v—T—w—wv—v— Tox2 +—+—+——+——++—+—+—— 15.0 A .. 10.0 ,— T- 0: 1- 1- PFY—FY-i—OF mV'LDCDNwFT- (01-1-1-1-1-1-1— L0 , , ...... J22.0 we magmas...» -200 Figure 2-2. Presence of TOXD only in HC-toxin-producing (T ox2+) isolates of C. carbonum. Genomic DNA was digested with (A) BamHl, or (B) EcoRI and blots were probed with TOXD cDNA. Strain 164R5 though 164R18 are random progeny from a cross between SB111 (Tox2+) and SB114 (Tox2-). Strain 81- 64, 1368 and 1309 are independent race 1 (Tox2+), race 2 (Tox2-) and race 3 (Tox2-) isolates, respectively. Strains used in blot (B) are all independent isolates of race 1 (Tox2+) of C. carbonum. 44 V 1— 90 5:52 82865 9.6me {mm .6395 use as 338.39 98.1 .o .558...“ 8538 £22.. a-“ 9:2". BaumquoEqEEOMHunheeogfiEfi—quodgch I IE8 mag hmm can no" ovN «an em." ow." ONH Md." Hm mm UFOs—"U4 GNOBOU Chan—4 anOBOU 2304 £200 UFO-mud GNOBUU UPSU‘ anon—.00 06.304 Dung—.00 Durand anon—.00 48 compared to the wild type 164R10. Together with the fact that TOXD is absent in all Tox2‘ isolates, we postulate that TOXD has no house-keeping function. Analysis of Null Mutant: HC-toxin preparations from culture filtrates of the wild type and T492 were analyzed by TLC (Figure 2-7A). No reproducible difference was observed. Although the toxin spot in T492 appeared smaller than that of wild type, this was likely due to experimental variation. The disease phenotype of T492 was the same as wild type with regard to lesion morphology and disease development (However, no quantitative analysis was performed) (Figure 2-7B). This is consistent with the unaltered HC-toxin production in T492. Therefore, TOXD appears to have no, or at least no apparent, role in HC-toxin production or fungal pathogenicity. DISCUSSION TOXD was cloned based on its properties of being unique to Tox2+ isolates of C. carbonum. TOXD is also within the TOX2 locus region and is regulated by TOXE. However, the loss-of-function experiment described in this Chapter failed to establish a role for TOXD in HC-toxin production or fungal pathogenicity. Although the sequence similarity between TOXD and lovC (Figure 2-5) suggests that the product of TOXD is likely an enzyme, no signature or motif has been detected in the deduced protein sequence. The protein encoded by lovC 49 ..9Eoo 9.68. me new: we; 153... 98.1 Lo 882% 9: a 8:899 65 86:55 E .223: 2203 5:598 .6 598.1 .o 5:935 84. use". 50 m> 6 2502026 3635 Am: Mamet Ease :8 6x9 5 Em E5 8 .225: we: 2:5 5 5x96: Lo 22.84.. 02. 2V .3: 2.2:... ..2. 98.1 .o 8505;“. .3 2:9". < «me... ._.>> «2:. ts 52 was proposed to have enoyl reductase activity, and to be an accessory component for the lovastatin polyketide synthase complex found in Aspergillus terreus (Kennedy et al., 1999). IovC mutant accumulated a lovastatin intermediate with double bonds in the polyketide backbone of a lovastatin precursor. Could the TOXD product possess a reductase activity that has a minor role in HC-toxin production? One possibility for this speculation is that the TOXD protein reduces the 8-carbonyl group of Aeo. HC-toxin with a reduced 8- carbonyl group (HC-toxin IV) has only 1% relative activity compared to HC-toxin l (Rasmussen, 1987). Maize lines that are resistant to HC-toxin produce a HC- toxin reductase (HCTR) encoded by Hm1. HCTR specifically reduces the same 8-carbonyl group in order to detoxify HC-toxin (Meeley and Walton, 1991; Johal and Briggs, 1992; Meeley et al., 1992). However, no protein sequence similarities were detected between that of TOXD and Hm1. It is not unusual that a gene physically belongs to a gene cluster but has no apparent role in the metabolic pathway. In C. heterostrophus race T, a putative reductase- encoding gene (RED1) was cloned and shown to be linked to PKS1, but it has nothing to do with T-toxin biosynthesis (Kodama et al., 1999). A plausible explanation for the unalteration of toxin production in TOXD or RED1 null mutants is gene redundancy. There are likely many enzymes possessing reductase activity in fungal cells. When TOXD or RED1 is mutated, other reductases could process the normal toxin biosynthetic steps. Therefore no phenotype change would be observed. Otherwise, TOXD (and maybe also 53 RED1) could be regarded as remnant genes that happen to have similarities to known TOX genes, but no longer are needed for HC-toxin or T-toxin production, modification, or fungal pathogenicity. Nevertheless, TOXD has been useful as a genetic marker for mapping the TOX2 locus in C. carbonum (Ahn and Walton, 1996) and for successfully cloning two more genes (TOXF and TOXG) that have proven roles in HC-toxin production (Chapters 3 and 4). 54 CHAPTER 3 A PUTATIVE BRANCHED-CHAIN AMINO ACID AMINOTRANSFERASE GENE REQUIRED FOR HC-TOXIN BIOSYNTHESIS AND PATHOGENICITY IN COCHLIOBOLUS CARBONUM The gene sequence reported in this Chapter has been deposited in GenBank with accession number AF157629. The content of this Chapter is a re-formatted version of a manuscript submitted to Microbiology (Cheng et al., 1999). 55 ABSTRACT The cyclic tetrapeptide HC-toxin is required for pathogenicity of the filamentous fungus Cochliobolus carbonum on maize. HC-toxin production is controlled by a complex locus, TOX2. Here we report the isolation and characterization of a new gene in the TOX2 locus, TOXF, and show that it is specifically required for HC-toxin production. and pathogenicity. TOXF is present as two or three copies in all HC-toxin-producing (Tox2+) isolates ‘ and Is absent in toxin non-producing strains. The deduced amino acid sequence of TOXF has moderate homology to many known or putative branched-chain amino acid aminotransferases from various species. A strain of C. carbonum with all copies of TOXF disrupted grew normally but lost HC-toxin production and pathogenicity. We propose that TOXF has a biosynthetic role in HC-toxin synthesis, perhaps to aminate a precursor of Aeo (2-amino-9,10-epoxi-B-oxo-decanoic acid). INTRODUCTION Host specific toxins, which are produced exclusively by fungi, are generally low molecular weight secondary metabolites with diverse structures. They are critical determinants of virulence or pathogenicity in many plant disease interactions (Walton, 1996). The host-selective toxin made by race 1 isolates of Cochliobolus carbonum, called HC-toxin, selectively affects maize (Zea mays L.) 56 lines of genotype hm1/hm1. Structurally, HC-toxin is a cyclic tetrapeptide, cyclo(D-Pro-L-Ala-D-Ala-L-Aeo), where Aeo stands for 2-amino-9,10-epoxi-8- oxo-decanoic acid. Previous studies on the Mendelian and molecular genetics of HC-toxin production have shown that HC-toxin production is controlled by a Mendelian locus, TOX2, but that this locus has a complex molecular structure (Nelson and Ullstrup, 1961; Scheffer et al., 1967; Ahn and Walton, 1996). All of the known genes necessary for HC-toxin production are present only in HC-toxin-producing (Tox2‘) isolates in multiple functional copies, are dedicated to HC-toxin production, and, with one exception, are linked over a ~540 kb region of one chromosome (Ahn and Walton, 1996, 1998). TOX2 genes include HTS1, which encodes a 570-kDa tetrapartite non-ribosomal peptide synthetase (NRPS) called HC-toxin synthetase (HTS) (Panaccione et al., 1992; Scott-Craig et al., 1992); TOXA, encoding a putative HC-toxin transporter of the major facilitator superfamily (Pitkin et al., 1996); and TOXC, encoding a fatty acid synthase B subunit (Ahn and Walton, 1997). TOXE encodes a regulatory protein required for expression of TOXA and TOXC (Ahn and Walton, 1998). Another gene, TOXD, is also Tox2*-unique, linked to the other TOX2 genes, and co-regulated by TOXE, but TOXD has no defined role in HC-toxin biosynthesis (Chapter 2. Cheng and Walton, unpublished results). A homolog of TOXD was recently shown to be necessary for correct processing of the growing polyketide chain by the lovastatin nonaketide synthase in Aspergillus terreus (Kennedy et al., 1999). 57 The known genes of TOX2 can account for the synthesis and assembly of the components of HC-toxin other than the unusual amino acid Aeo. Biogenetically, Aeo is a fatty acid or polyketide (Cheng et al., 1999. See Appendix), and the specific requirement of TOXC for toxin production argues that the decanoic acid backbone of Aeo is a fatty acid. Nothing is known about the other steps in Aeo biosynthesis, but at least two oxidations (to produce the 8- carbonyl and the 9,10-epoxide) and an aminetion (at the 2- position) must also occur. We describe here a strategy using bacterial artificial chromosomes (BACs) to search for additional genes of TOX2. It is based on the assumption that new TOX2 genes would be physically linked, but not clustered, to the known TOX2 genes, and would also be restricted in their taxonomic distribution to Tox2“ isolates of C. carbonum. Here we describe the identification of TOXF, a new TOX2 gene. METHODS Fungal Strains. SB111 (ATCC 90305) and SB114 are standard Tox2+ and Tox2‘ laboratory strains, respectively, and 164R10 is a Tox2+ progeny of a cross between them (Ahn and Walton, 1997). Bacterial Artificial Chromosome (BAC) Library Construction. The BAC vector pBACwich, a generous gift of the Clemson University Genomics Institute 58 (Zhu et al., 1997), was modified by the insertion of restriction sites for Ascl and Pmel (Figure 2-1). Two Oligonucleotides, 5’-AGCTGTTTAAACTGGCGCGCC-3’ and 5’-AGCTGGCGCGCCAGTTTAAAC-3’, were mixed at equimolar concentration, heated at 95 ° C for 3 min, and annealed slowly to form a complementary DNA linker. This linker was subcloned in-frame into the unique Hindlll site of pBACwich to make a new vector called pBACocta. Fungal chromosomal DNA was embedded and digested in low-melting agarose as described (Ahn and Walton, 1996). Proteinase K was removed from the agarose by washing with 50 mM EDTA, followed by 10 mM Tris/1 mM EDTA (TE) (pH 8.0), and then with the appropriate buffer for the particular restriction enzyme. DNA was digested with 10 U Asc l per gel block for 24 hr at 37 °C, and the gel blocks were then digested with B-agarase I (New England BioLabs, Boston, MA) following the manufacture’s instructions. DNA fragments were gently precipitated with ethanol, dried under vacuum, and dissolved in TE. Ligation reactions contained 5.0 pg of pBACocta linearized with Asc l and dephosphorylated and 1 ug of digested DNA in a volume of 20 III, and were incubated for 15 hr at 16 °C. Each aliquot (2 ul) of ligation mixture was transformed into 40 pl of EIectroMAX DH10B E. coli cells (Gibco-BRL, Bethesda, MD) by electroporetion (2.5 kV, 25 F, 100 D, 0.1 cm cuvette) using a GenePulser apparatus (Bio-Rad, Richmond, CA). Cells were transferred immediately to a new tube, diluted with 1 ml SOC medium (2% tryptone, 0.5% yeast extract, 10 mM NaCl, 2.5 mM KCI, 10 mM MgCIz, 10 mM M9804, and 20 mM glucose, pH 7.0), incubated at 37 °C for 1 hr, and spread on LB plates with 12.5 119! ml chloremphenicol, 0.75 uglml 5-bromo-4- 59 .mEchEoom. .2 .929: 288.3 me now: was 6:6 20:82 .686: 60559 memo Neg on... .:o_30__0:U_>_n:_ 0009.0 0:0 0:05.209. :02»: 0 2 00:05 ..00 009000 o\o 0.0 0 :0 000.0000 ..IEmm 5:5 0020090 002, <20 2580.0 .2. E 09865 .6801me Eu 2+ 3 Races west :50 0020.00 .005: :_ 00:00.0 .9 5:000an <20 E 00603 002, 000.0 :000 6 0>=0E00069 0:0 00000.0 NF 95 53052520090 3 00:8 002, m E Cm2:6.0kb . ' 2.9kb ' DE * -—- I g ‘3 E 18% E fij —> Wt-.- D Rpm Wt D _ R D,R A 23.1- 9.4- , ,‘ 1 6'6" < “" . “I!" 4.4- r, w “*"‘" , ” was- , 22 ° ' ' ? T fl _ ~ 2.0-, " > Try . 75 The conclusions drawn from probing the transformants with the 383-bp internal fragment were confirmed by re-probing with the entire TOXF cDNA. In the replacement strain R and the double mutant DIR, a new band of 3.9 kb was visible. This 3.9-kb signal corresponds to the original 2.9-kb Pst I fragment plus the 1.4-kb hyg gene cassette minus the 383-bp fragment of TOXF that had been deleted (Figure 3-6c). Phenotypes of TOXF Mutants. The single (D and R) and double TOXF mutants (DIR) showed normal growth and development on V8-agar or in modified Fries’ liquid medium (Walton et al., 1982). As expected, the TOXF mRNA was absent in the DIR strain (Figure 3-7a). Strain DIR also failed to make detectable HC- toxin in culture (Figure 3-7b). Mutation of one or the other copy of TOXF (D or R strains) did not affect pathogenicity of C. carbonum on maize of genotype hm 1/hm1, but mutation of both copies completely abolished pathogenicity (Figure 3—8). Therefore, both copies of TOXF are functional. These experiments establish . that TOXF is not required for normal growth and development but is specifically required for HC-toxin biosynthesis and hence pathogenicity. DISCUSSION TOX2 is a complex locus consisting of multiple genes each present in multiple functional copies (Ahn and Walton, 1997). Unlike the genes for other secondary metabolite pathways in fungi, such as those for aflatoxins 76 4 HC-toxin GPD- 1 (a) (b) Figure 3-7. Analyses of TOXF mutants. (a) RNA blot showing the TOXF mRNA in 164R10 (WT) and its disappearance in the double mutant (D/R). Thirty ug of total RNA were loaded per lane, and the blot was probed with the TOXF cDNA (pAATC1). The blot was subsequently stripped and re-probed with C. carbonum GPD-1 encoding glyceraldehyde-3-phosphate dehydrogenase as loading control. (b) Thin layer chromatography (TLC) of extracts of culture filtrates of C. carbonum 164R10 (WT) and the double mutant (D/R) showing that the double mutant does not make HC-toxin. HC-toxin was visualized with an epoxide-specific reagent. 77 .cozmiooclmoa 9v umcafimofloca 295 886.. .6992: $38 58 EB 899:5 "RE Lo 9.2 >88 .m 69855 1x9 Io 8:0 >88 .o 5 Ex: 25 2:5 as 2.5:me2 .6 >38 £28852. .3 2:9". ma m .0 <5 78 [sterigmatocystins, trichothecenes, and gibberellins, those involved in HC-toxin production spread over more than 540 kb and show only limited clustering (Yu et al., 1995; Ahn and Walton, 1996; Brown et al., 1996; Trapp et al., 1998; Tudzynski and Holter, 1998). Therefore, the molecular genetic analysis of HC- toxin biosynthesis cannot rely on sequencing and short-range chromosome walking. The functional redundancy of the HC-toxin biosynthetic genes excludes the use of mutagenesis approaches such as restriction enzyme mediated integration (REMI) (Lu et al., 1994; Sweigard et al., 1998). Here we have described the use of an alternate strategy, based on bacterial artificial chromosomes (BACs), to find HC-toxin biosynthetic genes. This approach assumes that new HC-toxin biosynthetic genes will be within 50-150 kb of each other and present only in Tox2” isolates. The BAC method is being extended by analysis of regions adjacent to other known TOX2 genes (e.g., HTS1, TOXC, and TOXE) and by constructing a BAC library with larger inserts (up to 400 kb) (Monaco and Larin, 1994). The TOXF product, TOXFp, shares moderate homology to a large group of branched-chain amino acid aminotransferases (BCATs) (Figure 3-5). All of the essential amino acid residues necessary for transaminase function are conserved in TOXFp. Considering that three of the four amino acids of HC-toxin (D-Pro, L-Ala, and D-Ala) are directly derived from primary metabolism, a role for TOXF in the biosynthesis of Aeo (2-amino-8-oxo-9,10-epoxidecanoic acid) seems most reasonable. A plausible reaction catalyzed by TOXFp would be to aminate a precursor of Aeo. Insofar as the decanoic acid backbone of Aeo is 79 biogenically a fatty acid (Ahn and Walton, 1997; Cheng et al., 1999), a transamination reaction seems essential. The amino-acceptor substrate for the reaction catalyzed by TOXFp could be 2-oxo—decanoic acid, which itself could be produced by oxidation of decanoic acid by an unidentified enzyme. Alternative substrates for TOXFp could be derivatives of 2-oxo-decanoic acid that already contained the 8-carbonyl and/or the 9,10-epoxide. All known BCATs are involved in primary catabolism of branched-chain amino acids. Typically, they catalyze the transfer of an amino group from Leu, Val, or lie to a-ketoglutaric acid. TOXFp appears to be the first BCAT dedicated to a non-essential secondary metabolic pathway. The failure of TOXF to cross- hybridize with any other genes in C. carbonum at low stringency (data not shown) suggests that TOXF and the sequences of the house-keeping BCATs of C. carbonum are not closely related. It may be significant that BCATs are structurally and functionally related to bacterial D-amino acid aminotransferases and that TOXFp also shows limited sequence similarity to this sub-family of enzymes (Tanizawa et al., 1989; Mehta et al., 1993; Alexander et al., 1994; Sugio et al., 1995). D-amino acids are common in nonribosomal peptides, and HC-toxin contains two D-amino acids. From its similarity to BCATs, TOXFp probably uses a branched—chain amino acid as amino donor, but the possibility that the donor is a D-amino acid should also be considered. 80 CHAPTER 4 A NOVEL EUKARYOTIC ALANINE RACEMASE GENE INVOLVED IN CYCLIC PEPTIDE BIOSYNTHESIS IN THE FUNGUS COCHUOBOLUS CARBONUM The gene sequence reported in this Chapter has been deposited in GenBank with accession number AF169478. The content of this Chapter is a re-formatted version of a manuscript submitted to J. Biol. Chem. (Cheng and Walton, 1999). 81 ABSTRACT Non-ribosomally synthesized peptides and other secondary metabolites mediate many interactions between organisms and also have major importance in human medicine. HC-toxin, a cyclic tetrapeptide containing D-Ala and D-Pro, is an essential virulence determinant for the plant pathogenic fungus Cochliobolus carbonum. HC-toxin biosynthesis is controlled by a complex genetic locus, TOX2, that contains multiple functional copies of genes involved in the biosynthesis, export, and regulation of HC-toxin. The central enzyme in HC-toxin biosynthesis is HC- toxin synthetase (HTS), a 570-kDa non-ribosomal peptide synthetase that activates all four amino acids and epimerizes L-proline. We present here the cloning and characterization of a new gene of the TOX2 locus, TOXG. TOXG is present as two or three copies in all HC-toxin-producing (Tax?) isolates and is absent from toxin-non-producing (Tox2‘) isolates. The deduced amino acid sequence of TOXG has significant similarity to that of an alanine racemase involved in cyclosporin biosynthesis by the fungus Tolypocladium inflatum. Despite no significant similarity of its deduced amino acid sequence to bacterial alanine racemases, TOXG was able to complement a bacterial strain defective in D-alanine synthesis. C. carbonum with all copies of TOXG mutated by homologous recombination grew normally but had an altered HC-toxin profile. Specifically, it could no longer make the three forms of HC-toxin that contain D-alanine but could 82 still make a minor form of HC-toxin that contains glycine in place of D- alanine. The mutant has reduced virulence, as manifested by delayed disease development, probably due to Its inability to make the most predominant and most biologically active form of HC-toxin. Feeding D- alanine to the mutant in culture restored production of the forms of HC- toxin containing D-alanine. We conclude that TOXG encodes a novel alanine racemase that converts L-alanine to D-alanine specifically for incorporation into HC-toxin by HTS. INTRODUCTION Cyclic peptides are a large class of medically and biologically important secondary metabolites produced by bacteria and fungi. Cyclic and some linear peptides are synthesized by a special class of enzyme called non-ribosomal peptide synthetases (NRPS) (Kleinkauf and von Dohren, 1996). NRPS enzymes are organized into domains, each of which activates and binds one amino acid. Some NRPS domains can catalyze amino acid modifications, especially N- methylation and epimerization of L to D amino acids. The filamentous fungus Cochliobolus carbonum (anamorph Helminthosporium carbonum) makes a cyclic tetrapeptide known as HC-toxin. The major form of HC-toxin (HC-toxin I) has the structure cyclo(D-Pro-L-Ala-D- Ala-L—Aeo), where Aeo stands for 2-amino-9,10—epoxi-8—oxo-decanoic acid. Three minor forms (HC-toxin ll, Ill and IV) slightly differ in amino acid 83 composition (Rasmussen, 1987; Rasmussen and Scheffer, 1988). HC-toxin is required for C. carbonum to pathogenize maize of genotype hm1lhm1. Most maize is resistant because it contains the dominant allele of Hm1, which encodes an enzyme, HC-toxin reductase, that detoxifies HC-toxin by reducing the 8- carbonyl of the Aeo side chain (Meeley and Walton, 1991; Meeley et al., 1992; Johal and Briggs, 1992). Like other Aeo-containing cyclic tetrapeptides, HC-toxin inhibits histone deacetylase in maize and other organisms (Brosch et al., 1995; Ransom and Walton, 1997). Previous studies on the Mendelian and molecular genetics of HC-toxin production have shown that HC-toxin production is controlled by a single Mendelian locus, TOX2, but that this locus has a complex molecular structure (Scheffer et al., 1967; Ahn and Walton, 1996). All of the known genes (HTS1, TOXA, TOXC, TOXE, TOXF) that are specifically required for HC-toxin production are present only in HC—toxin-producing (Tox2") isolates in multiple functional copies, have no housekeeping roles, and, with one exception in some isolates, are linked over a ~540 kb region on one chromosome (Panaccione et al., 1992; Scott-Craig et al., 1992; Pitkin et al., 1996; Ahn and Walton, 1996, 1997, 1998; Chapter 3, Cheng et al., submitted). The central enzyme in HC-toxin synthesis is HC-toxin synthetase (HTS), a 570-kDa NRPS with four amino acid activating domains. HTS is encoded by a 15.7-kb gene called HTSf (Panaccione et al., 1992; Scott-Craig et al., 1992). Biochemical assays of partially purified HTS indicated that it activates by ATP/PPj-exchange L-proline, L-alanine, and D-alanine, but not D-proline, and 84 epimerizes L-proline and L-alanine to the corresponding D-amino acids (Walton and Holden, 1988). Subsequent analysis of the sequence of HTS revealed a motif in domain A of HTS that was also present only in other epimerizing domains, very few of which were known at that time (i.e., the single-domain enzymes gramicidin synthetase l and tyrocidine synthetase I, and domain C of the three-domain enzyme ACV synthetase) (Scott-Craig et al., 1992). However, the putative epimerase signature motif was found only once in HTS, in domain A, whereas there should have been another copy in domain C, on the basis of the ability of partially purified HTS to epimerize L-alanine as well as L-proline. The subsequent cloning and sequencing of numerous additional NRPS genes, including many more domains that have epimerizing capacity, has resulted in the conclusion that there are characteristic signature motifs for epimerization domains, and that HTS clearly has only one, in domain A (Marahiel et al., 1997). This left in doubt the actual source of D-alanine for HC-toxin. Cyclosporin is an undecapeptide containing D-alanine. Analysis of the primary sequence of cyclosporin synthetase also indicated a lack of an epimerization signature motif for domain A, which activates D-alanine (Marahiel et al., 1997). This led to the discovery of a separate enzyme, alanine racemase, that is responsible for the synthesis of D-alanine for incorporation into cyclosporin by cyclosporin synthetase (Hoffmann et al., 1994). The gene from Tolypocladium inflatum that encodes the alanine racemase, cssB, has been submitted to GenBank (accession number A40406) as a patent submission (patent number WO9425606) but not otherwise published. 85 Our search for new genes involved in HC-toxin biosynthesis has proceeded on the assumption that they will be linked, but not tightly clustered, to the known TOX2 genes, and that they will be restricted in distribution to Tox2+ isolates. Using a baCterial artificial chromosome (BAC) library, a new gene involved in HC-toxin biosynthesis, TOXF, was recently cloned and described (Chapter 4. Cheng et al., submitted). Immediately adjacent to TOXF is another gene, which we call TOXG. This paper describes the characterization of TOXG and biochemical and genetic evidence that it encodes an alanine racemase dedicated to HC-toxin production. MATERIALS AND METHODS Fungal Strains and Growth Conditions. SB111 (ATCC 90305) and SB114 are standard Tox2+ (HC-toxin-producing) and Tox2‘ (HC-toxin-non-producing) laboratory strains of C. carbonum, respectively, and 164R10 is a Tox2+ progeny of a cross between them (Ahn and Walton, 1997). Growth conditions for fungal culture on solid V8-plate or in modified Fries’ liquid medium were previously described (Van Hoof et al., 1991). Nucleic Acid Manipulations and Sequence Analysis. C. carbonum DNA and RNA isolation was previously described (Pitkin et al., 1996). DNA and RNA blotting, probe labeling, hybridization, cDNA and genomic library screening, and DNA subcloning were done following standard procedures (Sambrook et al., 86 1989). Oligonucleotides were synthesized by the Michigan State University (MSU) Macromolecular Facility. Automated DNA sequencing was done at the Yale University Keck Foundation Biotechnology Resource Laboratory and at the MSU DNA Sequencing Facility. Sequences were assembled and analyzed with the DNASter Software package (DNAStar Inc., Madison, WI). Protein sequence alignment was generated with the CIusta/W Program (Thompson et al., 1994). The transcriptional start site of TOXG was determined by 5’ RACE using a kit from Gibco-BRL (Frohman et al., 1988). The primer for reverse transcription (GSP1) was 5’—CGATTCAT'ITTAGGGTGTGCCAGAT—3’ and the nested primer for PCR amplification (GSP2) was 5’-TG'ITTCGACTAACCGGTAGCAGGG-S’. Bacterial Complementation and Enzyme Assay. E. coli strain TKL10 (dadX- air-ts, CGSC’5466) stman, 1972) was obtained from the E. coli Genetic Stock Center at Yale University and maintained in Luria-Bertani (LB) medium supplemented with 200 uglml D-alanine. The TOXG expression vector pARE was constructed as follows: a 1.2 kb TOXG coding region (lacks 14 amino acids at the 5’-end of ORF) was released from pGC1 by BamHl/Kpn l digestion and subcloned into pQE31 (Qiagen, Valencia, CA). pARE was verified by sequencing and transformed into TKL10 by electroporation (2.5 kV, 25 uF, 200 (2, 0.1 cm cuvette) using a Bio-Rad Gene Pulser (Bio-Rad, Richmond, CA). Transformants were selected on LB plates with 100 uglml ampicillin at 42 °C. Two independent transformants were designated as TKL10T1 and TKL10T2. 87 Bacterial crude protein extract was prepared as follows. Cells were allowed to grow in LB medium with 100 uglml ampicillin at 37 °C for 12 hr and were collected by centrifugation (20 min, 5000 g) at 4 °C. The pellet was resuspended in extraction buffer of 50 mM Tris, pH 8.7 with 10% glycerol, 4 mM EDTA, 20 mM DTT, and 30 pM pyridoxal-5’-phosphate (PLP). Lysozyme (Sigma, St. Louis, MO) was added to 1 mg/ml and incubated on ice for 30 min. Cells were broken by sonication (3 min, 50% cycle, maximum output for microtip on a Model 450 Sonifier, Branson, CT). Cellular suspension was centrifuged at 20,000 g for 30 min at 4 °C. The supernatant was then saved as crude protein extract. The alanine racemase assay was adopted from Hoffmann et al. (1994). Step 1 reactions contained 50 mM Tris (pH 8.7), 50 mM L-alanine, 20 mM DTT, 30 M PLP and an appropriate amount of protein extract in a total volume of 1 ml. Reactions were allowed to proceed for up to 4 hr at 42 °C, then terminated by heating at 95 °C for 10 min. The mixture was centrifuged for 10 min and 500 pl of supernatant was transferred to a new tube. Five hundred pl of reaction 2 mixture (50 mM Tris, pH 8.7 with 0.5 U of D-amino acid oxidase, 25 U of lactate dehydrogenase, 2 mM FAD, 0.2 mM NADH) (enzymes from Sigma, St. Louis, MO) was added to the tube and incubated at 37 °C for 2 hr. The decrease of absorbance at 340 nm was used to calculate the relative racemase activity. For each sample, a control with boiled protein extract was used as blank. Creation of TOXG Null Mutant. Fungal protoplast preparation, transformation, selection, and single-spore isolation of transformants were done as described 88 (Panaccione et al., 1992; Pitkin et al., 1996). The two copies of TOXG in strain 164R10 were sequentially mutated by targeted gene inactivation. One was mutated by gene replacement mediated by double-crossover homologous recombination, and the other by gene disruption mediated by single-crossover integration. To make the replacement vector pTOXGR1, a genomic clone pAATG1 was trimmed with EcoRV and Hpa I to eliminate the TOXF coding region and re- ligated to make the intermediate construct pARM2. A 420-bp Apa llPsf I fragment (containing the 3’ end of the TOXG cDNA) from pGC1 was subcloned into the BamHl/Pst I sites of pARM2 to make another intermediate construct pARM3. The 512-bp internal Sph llPsf I fragment (corresponding to +362 bp to +874 bp in the genomic sequence) of pARM3 was replaced by a hygromycin-resistance cassette (composed of the E. coli hph gene encoding hygromycin phosphotransferase driven by the Aspergillus nidulans trpC promoter from plasmid pCB1003) (Carroll et al., 1994) to make the final replacement vector pTOXGR1 (5.9 kb). The fragment (2.5 kb) containing the hph cassette plus flanking TOXG DNA was released from pTOXGR1 by digestion with Not I and Hindlll and used for transformation. To make the disruption vector pTOXGD1, an 8-bp BamHl linker (Boehringer, Colone, Germany) was inserted into the unique Sac I site of pARM3 to make pARM4. The central part of the coding region from pARM4 (a 512-bp Sph llPst l fragment plus 8-bp linker) was then subcloned into the Hindlll and Xba I sites of the pBC-phleo vector (CAYLA, Toulouse, France), which contains 89 the A. nidulans ZEO gene cassette for phleomycin resistance, to obtain the final construct pTOXGD1 (6.6 kb). Vector pTOXGD1 was linearized with BamHI before transformation. Strain 164R10 was first transformed with the 2.5-kb fragment from pTOXGR1. Strain T697 with one copy of TOXG replaced with the hph cassette was selected in medium containing 120 ug/ml hygromycin. Strain T697 was subsequently transformed with linearized pTOXGD1. Strain T698 with the second copy of TOXG disrupted was selected in medium containing 50 uglml phleomycin. Analysis of Mutants. HC-toxin was extracted from fungal culture filtrates, separated by silica-gel thin layer chromatography (TLC) (Si250-PA, J. T. Baker, NJ) with a solvent of acetone:dichloromethane (50:50 by volume), and detected with an epoxide-specific spray as described (Meeley and Walton, 1991). Compounds of interest were scraped from the TLC plate, suspended in ethanol, and briefly centrifuged. The supernatant was dried under vacuum and redissolved in ethanol for mass spectrometry analysis by the MSU Mass Spectrometry Laboratory. Fungal pathogenicity was assayed on maize inbred Pr (genotype hm1lhm1) by spraying with conidia (Panaccione et al., 1992; Pitkin et al., 1996). Infected maize leaves were photographed every day from 2- through 6-d post-inoculation. 9O Feeding Studies. Modified Fries’ liquid medium (Van Hoof et al., 1991) was inoculated with conidia and grown at room temperature (25 °C) in still culture for 3 d before addition of D-alanine to a final concentration of 1, 10, 25, or 50 mM. Cultures were allowed to grow for an additional 12 d before collecting the culture filtrates for HC-toxin extraction and analysis. RESULTS Characterization of TOXG. We previously reported the cloning of TOXF on the basis of its linkage to known TOX2 genes using bacterial artificial chromosomes (Chapter 3; Cheng et al., submitted). Within a 2.9-kb genomic DNA fragment (cloned in pAATG1) containing TOXF, we detected a partial open reading frame that starts immediately in the 5’ direction of TOXF and is transcribed in the opposite direction (Figure 4-1). The deduced partial protein sequence of this new open reading frame, whose corresponding gene has been named TOXG, showed strong similarity to many threonine aldolases (e.g., GLY1p of Saccharomyces cerevisiae, GenBank P30831) and to an unnamed patent sequence in the data base (GenBank A40406) from the fungus Tolypocladium inflatum. A40406 is annotated as encoding an alanine racemase (EC 5.1.1.1) involved in cyclosporin biosynthesis. This gene has now been named cssB (K Schorgendorfer, Novartis-Biochemie, Austria, personal communication). 91 Copy 1: 5.2 kb 2.9 kb 1!- EcoRV EcoRV Pst I _j_ :3 $3 h fl TOX G TOXF Xba I Pst l Sph l Copy 2: 4.8 kb 2.9 kb -— - d— - EcoRV EcoRV :3 h g TOX G TOXF Xba I Pst l Sph l Pst l Xba I Figure 4-1. Restriction map of the two copies of the TOXF/G region in 164R10. The 2.9-kb Pstl fragment (cloned in pAATG1) was sequenced. 92 Full-length cDNA and genomic sequences of TOXG were obtained. A 1.1-kb Pst Iala I fragment from pAATG1 was used as probe to obtain a full-length cDNA clone of TOXG (pGC1). The insert of pGC1 was used to obtain the 3’-end of a genomic copy of TOXG (pGGZ). The inserts of plasmids pGC1 and pGG2 were sequenced on both strands. TOXG encodes a protein (TOXGp) of 389 amino acids with a calculated molecular weight of 42.7 kDa and a pl of 6.5. TOXGp contains no predicted signal peptide or glycosylation sites. TOXG has one intron of 52 bp. The transcriptional start site is at —46 bp. The distance between the translational and transcriptional start sites of TOXF and TOXG is 299 bp and 195 bp, respectively (Figure 4—2). The deduced protein sequence of TOXG has an overall 42% identity to the product of 0338 and 32% identity to GLY1p (Figure 4- 3). A conserved lysine residue for binding of the cofactor pyridoxal-5’-phosphate (PLP) was identified at position 235 in TOXGp (Liu et al., 1997; Monschau et al, 1997) TOXG is present in three copies in all tested Tox2+ isolates except 164R10, which has only two copies, and is completely absent in all Tox2‘ strains tested (data not shown). Like TOXF, TOXG expression requires the regulatory gene TOXE (Figure 4-4) (Ahn and Walton, 1998). Complementation of Bacterial Mutant. To test the putative function of TOXGp as an alanine racemase, we constructed a TOXG expression vector pARE and transformed it into bacterial strain TKL10, which is defective in D-alanine biosynthesis and therefore requires D-alanine in the medium for survival 93 Figure 4-2. Sequence of TOXG and Its relationship to TOXF. Transcriptional and translational start sites are indicated above the DNA sequence. Deduced amino acid sequence and the polyadenylation site (1) are placed under the DNA sequence. The sole intron is typed in italic lower case. 94 TOXF translational start site (—| TAAAATGCAGGCAA8GGTATCGCCATGNTGTCGATGGCTGAGAACCTTTAG TOXF transcriptional start site e—l AGATTAOTTAACCTTTTCGCTGA'I'PTGITCTTGAATTAAMATAAGAOGGT TTACCATAAGTTATAGAAGAACAGCTCCTTGTCGAAGATAGATAGAGTCAT CGCTTACGTGATTGGATACGCGAGATGAACTCAGNTTACGTAAGATTGTTO TGGTCTTCCATCCTGCACTATTACTTAATCTACGTGTTATTCCTTCACCCG |—> TOXG transcriptional start site TAGCCTTAAAGCATGTGCTTCCCAATTTGAAAAATCCATCTGTTGCAACGO |-—> TOXG translational start site ATATTTTCGAAATTTCGAGATGTCGAACATGGTTTTGAATGGAAACATTGA N 8 N N V L N G N I D CAAATCCGATAGGAATTCCATACTAGACATACTTCAATCGCTTGAGAATAT R 8 D R N 8 I L D I L Q 8 L 8 N I TGCATGGGGACAACCAGGCTCTGCAAOGTGCGATTTTAGGAgtRRttale. A W O Q P G 8 A R C D F R tyctgaaccgtacuagaag‘tccgatactuccactytgcagGPGATGI'TA 8 D V TTACACGCCCTAGCCTTAGGATGTTGTCCOCGGTCCTCAAGACTACACTCG I T R P 8 L R N L 8 A V L R T T L GCGATGACGTNTTTCGAGAAGACCTAACGACTGCCCACTTCGAAGCGCATG G D D V r R I D L T T A N F I A H TGGCTGAAATCAGCGGCCGAGAAGAAGGCATGTTCGTGATCACAOOGACGA V A I I 8 G R I I G N P V' I T G T TGGCCAATCAGCTTTGTTTGCATGCTCTAGTCTCAACTAGACCGTGTOGGA M A N 0 L C L H A L V 8 T R P C G TTTTACTTAGCTCAGAGTCCCATGCCATACACTATGAGGCAGGTGGCTCTT I L L 8 8 I 8 H A. I N Y I A G O 8 CGATGCTTAGCGGCGCCATGCTTCAGCCTGTGCAGCCGTCGAATGGTAAAT 8 M L 8 G A N L Q P V' 0 P 8 N G R ATCTGAGAGTGGAAGATTTGGAGGAGCACGCAATTCTAACCGACGATGTCC Y L R V B D L I I R .A I L T D D V AcAAATGCCCCACAAGCATTGTTTCTATGGAAAATACAGCTGGTGGAGCAG H R C P T 8 I V 8 N I N T A G O A TCGTTCCTGTCCATGAGCTCCGCCGCATCCGGGATTGGGCAAAGCAAAATA V V P V R I L R R I R D N A R Q N ACGTGAGGACTCACCTGGACGGTGCTAGACTTTTCGAGGCTGTTGCTACCG N V R T R L D G .A R L F I A. V' A T GTGCTGGGACTCTCAAAGAATATTGCAGCCTTATTGACCTAGTCTCOGTGO G A G T L R I I C 8 L I D L V' 8 V ATTTTNGTAAGAATCTTGGAGCCCCAATGGGCGCAATGNTTCTTGGTGACA D P 8 R N L G A P N O .A N’ I L O D AAAAATTGATTCAGCAGATGCGAAGAACTCGAAAAGGGNTTGGAGGAGGAA R R L I Q Q N R R T R R O I G G G TGAGACAAGGAGGGGTANTCACTGCAGCTGCACGGGAGGCACTGTTTGAAA N R Q G G V I T A A A. R I A L F 8 ACTTTGGACTGGGAGCGGAAATCGAANGTCAAACCCTATTGCAAGTGCACA N I G L G A I I I 8 Q T L L 0 V N AAGTCGCAAAGCGTCTAGGGGAAGAATGGACTAGAAAAGGTGOGAAGTTGA R V A R R L G I I N T R R G G R L GCAAAGAGATCGAGACAAATATCATATGGCTCGATCTTGATGCGGTTGGAA 8 R B I I T N I I N L D L D A. V' G TTAAGAAAAGTCAATTTATTGATAAGGGAAGGGAATATGGGGTGATTCTAG I R R 8 Q r I D R G R I Y O V I L ACOGCTGTCGCATTGTTTGCCATCACCAGATTGATATATATGCAGTAGAGG D O C R I V C R B Q I D I I A. V’ I CTCTTATCGACGTCTTCCATGATATACTAAAAGCAGATCCTATCAAAAACA A L I D V r R D I L R A D P I R N AGAATAGCGNTAGATAGTCATGAATATACACACCGGGATAAATATATAAGT R N 8 D R * TTTAACTACTAGTTAAGTCTAGATAACTAGTTGTCTAOGTAGTTGGCATCC t AATATACTACGGCTGCAAAGTTAGGTGTCTACCTNGTGCTGAGAAGGTGTT 95 -275 -228 -173 -122 -71 32 11 83 28 138 81 185 88 236 61 287 78 338 95 389 112 880 129 891 186 582 163 593 180 688 197 695 218 786 231 797 288 888 265 899 282 950 299 1001 316 1052 333 1103 350 1158 367 1205 388 1256 389 1307 1358 CcTOXG TiCSSB ScGLYl CcTOXG 49 w VLI TLGDDVFIED HFE- HV -ISGREuGMFVITGT TiCSSB 35 e 0A -- TLEDD eE IQ TzFDvrmfikGKEh $911 ScGLY1 26‘ ~>u~uAflLn~ ::G-- . GED . d E0 VAJ' GREHGLF GTLS CcTOXG 99 TiCSSB 85 ScGLY1 76 LOILHAL S RP‘GIL' C‘SIL-S 'IHYEAG SMLSGAMLQP 0.PSNGKYL -'.Gv . -_._ snag-trams: , CLSGAaIQPV=PANGflYL :IEAAGI '” LS. PSNGIIYL CcTOXG 189 TiCSSB 138 ScGLY1 126 CcTOXG 198 TiCSSB 183 ScGLY1 176 nah CcTOXG 288 DKKLIOO) RRq IGGGMRQG REALaENFGa .Ils TiCSSB 233 St uLImR; o 4 IGGGV .-' . IENF I'Gu - - I QR 8cGLYl 226: ooGGGIRo: ---JUA I . NN----DWR8 CcTOXG 298 an 4LGEIMTiKGGt ~ rIETNIIWLDLu GIHKSO TiCSSB 283 Iv SIG LI" ‘TNEEWLDLS -G uKs ScGLY1 269 r :3 h K I- LI SP-DT fife LHA";DP CcTOXG 388 TiCSSB 333 ScGLY1 319 CcTOXG 389 TiCSSB 377 ScGLY1 369 Figure 4-3. Multiple sequence alignment of TOXGp and homologs. Alignment was performed with ClustaIW and decorated with BOXSHADE. Black shading indicates identical amino acids between TOXGp and at least one of the others; light shading indicates similar amino acids. The origin of amino acid sequences is: CcTOXG from C. carbonum (GenBank AF169478), TiCSSB (GenBank A40406) from 7'. inflatum and ScGLY1 (GenBank P30831) from S. cerevisiae. Asterisk indicates the cofactor pyridoxal-5’-phosphate binding site. 96 1m 1. .s (VVIjsman, 1972). Two random transformants, TKL10T1 and TKL10T2, selected on LB plates with 100 uglml ampicillin at 42°C, survived without D-alanine supplementation (Figure 4—5). The alanine racemase activity of TKL10T1 and TKL10T2 was two to three-fold above background (data not shown). These experiments indicated that TOXG encodes a functional alanine racemase. Creation of a TOXG Null Mutant and Analysis of its Phenotype. To investigate the role of TOXG in HC-toxin biosynthesis, both copies of TOXG in isolate 164R10 were mutated by homologous recombination-mediated gene replacement or disruption. The correct integration of constructs into strains T697 and T698 was confirmed by Southern analysis (Figure 4-6). The 5.2-kb Xba I band, corresponding to copy 1 of TOXG, was gone in both T697 and T698 (Figure 4-6A). Integration of pTOXGDl into the 4.8-kb Xba I band (corresponding to copy 2 of TOXG) in strain T697 resulted in the disappearance of this band and appearance of a new band of 18.0 kb in strain T698, presumably due to tandem integration of two copies of pTOXGDl (Panaccione et al., 1992) (Figure 4-6A). When the same blot was re-probed with the full-length cDNA clone pGC1, a 6.1- kb band was visible in both T697 and T698, resulting from the gene replacement event (Figure 4-63). Creation of a TOXG null mutant was tested by RNA blotting. Because TOXF and TOXG are tightly clustered, TOXF was included in the analysis to be sure that mutation of one gene did not affect the expression of the other. Total RNA from 164R10(wild type), the TOXF null mutant DIR (renamed as T696; 97 Figure 4-4. TOXF and TOXG expression requires TOXE. Twenty pg of total RNA from wild type (WT, 164R10) and a TOXE null mutant (toxE) was separated in a denaturing formaldehyde gel and blotted to nylon membrane. Three identical blots were probed separately with TOXF, TOXG or GPD1. 98 .mm 395 5:5 Emctoficm: _o._Eoo .N ”ofivfi EmSE 629.com .F .oocoucoamug 6553-0 new 8.5562 c_=_o_aEm .2 386.3 963 $525929... 603989696 3 ofivz. BE voozuobs mm; mm9 E838 own—coon. 0553 .aoo .m ho cozflcoEoEEoo .mé 959". QE< + m.. .m mum—n. m_<-n + m.. .< mum—n. 99 4.4 — 2.3 — 2.0 — A B Figure 4-6. Southern blot analysis of TOXG mutants. Genomic DNA of (1) wild type 164R10, (2) single mutant T697, and (3) double mutant T698 was digested with Xba I, separated in a 0.9% agarose gel, blotted to nylon membrane, and probed with (A) the 512-bp Sph I/Pstl fragment that was replaced, (B) the full-length cDNA clone (pGC1). 7». 7'"~."-- ,. .1': .7 Fly-3: ya: . .‘ ~13”? --f?3.{‘;..'.‘:‘:: ‘-L‘,;LI‘-.'-2“J'e.;'-i TOX F III‘J ad. or. IX; .‘-.‘.H\L’V; - V. ' , - . . . ._- . . TOX G firms . wax-Ft'rrxfr: x4 ' V. ‘ .‘1 < ~ i". , GPD1 :1: “It . mi fl ' 1' .’ ‘ ‘ ' . my;;12::mra2fss 3:9; Figure 4-7. Northern blot verification of the independent mutation of TOXF and TOXG. Twenty pg of total RNA from wild type 164R10 (WT), a TOXF null mutant (T696), and a TOXG null mutant (T698) was separated in a denaturing formaldehyde gel and blotted to nylon membrane. Three identical blots were probed separately with TOXF, TOXG or GPD1. 101 Cheng et al., submitted), and the TOXG null mutant T698 was analyzed on three identical blots and probed with TOXF, TOXG, or GPD1 (encoding glyceraldehyde-3-phosphate dehydrogenase, as loading control). The results indicate that the TOXF and TOXG mutants had no detectable TOXF or TOXG mRNA, respectively, and that disruption of one gene did not influence expression of the other (Figure 4-7). Both the single (T697) and double (T698) TOXG mutants showed normal growth and development on agar plates or in liquid culture. Along with the restricted distribution of TOXG to Tox2+ isolates, this indicates that TOXG has no essential housekeeping functions. HC-toxin was extracted from culture filtrates of the wild type and T698 null mutant with or without D-alanine supplementation. Wild type isolates of Tox2+ C. carbonum produce four forms of HC-toxin that can be detected on TLC plates using an epoxide-specific reagent (Figure 4—8) (Meeley and Walton, 1991). The three minor forms, HC-toxin II, III, and IV, contain glycine in place of D-alanine, D-trans-3-hydroxyproline in place of D-proline, and 8-hydroxy-Aeo in place of Aeo, respectively (Kim et al., 1985; Tanis et al., 1986; Rasmussen, 1987; Rasmussen and Scheffer, 1988). HC-toxin forms I, II, and III are active at 0.2, 0.4, and 2.0 pg/mi, respectively, whereas form IV is active only at ~20 pglml (Rasmussen and Scheffer, 1988). The wild type isolate 164R10 produced abundant HC-toxin forms I, II, and IV but HC-toxin Ill could not be reliably detected. The TOXG null strain did not produce forms I or IV but still produced form II. Therefore, disrupting TOXG did 102 not affect production of all forms of HC-toxin. Form ll differs from the others in having glycine in place of D-alanine (Kim et al., 1985). Supplementing the medium with D-alanine restored production of HC-toxins l and IV (Figure 4-8). The TLC results are consistent with TOXG having a specific role in the synthesis of forms of HC-toxin that contain D-alanine. Pathogenicity of the toxG Mutant. To test the affect of the TOXG mutation on fungal virulende, conidia of C. carbonum wild type 164R10, single-mutant T697 and double-mutant T698 were spray-inoculated onto the leaves of susceptible maize. On plants inoculated with the wild type, lesions became visible in 2 d and after 6 d the entire infected leaf became totally brown and dessicated. Mutation of one copy of TOXG had no obvious affect on virulence, as seen in T697. Mutation of both copies of TOXG in T698 caused a noticeable slowing of disease development (Figure 4-9). A large portion of initial infection sites of T698 failed to develop into full lesions. The lesions that developed were consistently smaller than those caused by the wild type within the first 6 d of disease development. This modest decrease in virulence of the TOXG mutant differs from the results obtained with mutants of other TOX2 genes such as HTS1, TOXC, TOXE, and TOXF, which are completely avirulent (Panaccione et al., 1992; Ahn and Walton, 1997, 1998; Cheng et al., submitted). The apparent explanation for this is that the TOXG mutant can still make HC-toxin iI, which is about half as biologically active as HC-toxin I (Rasmussen and Scheffer, 1988), but sufficient enough to cause disease symptoms. 103 Strain WT T698 D-A|a(mM) o 1 25 o 1 1o 25 so HC-toxin Figure 4-8. TLC Analysis of HC-toxin in Wild type (WT, 164R10) and in TOXG null Mutant (T698). .co_a_8o:_-aoa Rm Em .m .1. .m .m R 25o cocafimoaoca who; mo>wo_ o>=mEomoaom .88..» ESE: ciao—o ucm :85 2.3:... 29%. .55 Sax: .0 >33 £25958 .3 9:9... mach haw... run—.0 ED.N 105 DISCUSSION HC-toxin is the agent of compatibility in the interaction between the fungal pathogen C. carbonum and its host, maize (Walton, 1996). The complete biosynthesis pathway of HC-toxin has not yet been deciphered. HC-toxin, a cyclic tetrapeptide, contains two D-amino acids and two L-amino acids. In contract to the essential role of D-alanine in bacterial cell wall biogenesis (Faraci and Walsh, 1988), the existence of D-amino acids in eukaryotic cells is not common. Other than the occurrence of D-aspartate and D-serine as neuromodulators in mammalian tissues (Nagata et al., 1989; Hashimoto and Oka, 1997; Wolosker et al., 1999), D-amino acids or derivatives are restricted to secondary metabolites in fungi (Kleinkauf and von Dohren, 1996). Hoffmann et al. (1994) reported the purification and characterization of a eukaryotic alanine racemase as a key enzyme in cyclosporin biosynthesis in Tolypocladium niveum and demonstrated that the enzyme catalyzes the interconversion of L- and D-alanine. In this paper, we present the molecular cloning of a gene encoding an alanine racemase from a plant fungal pathogen C. carbonum and the genetic confirmation of its specific role in HC-toxin biosynthesis. Like five other genes proven to be involved in HC-toxin production, TOXG is present only in Tox2+ isolates of C. carbonum. Null mutants of TOXG failed to synthesize three forms of HC-toxin that contain D-alanine. D-ala supplementation restored the wild type toxin profile (Figure 4-8). 106 TOXGp shares strong sequence homology with 0888p of T. inflatum, as well as threonine aldolases from many species (Figure 4-3). However, there is no obvious homology between the sequences of eukaryotic alanine racemases including TOXGp and CSSBp, and their bacterial counterparts (e.g. GenBank ABOZ1683, AF038438, AF081283), despite the fact that they are functionally compatible, as demonstrated in the TKL10 complementation experiment (Figure 4-5). Based on recent structural and evolutionary studies (Alexander et al., 1994; Jansonius, 1998), most PLP—dependent enzymes can be classified into three families (a,B, and y) by sequence analysis. Bacterial alanine racemase is among thefew PLP enzymes that does not fall into the above classification system. These exceptions may represent another family of PLP enzymes. VVIth the cloning of TOXG and 0833 as eukaryotic alanine racemases that have no evolutionary relationship with their bacterial counterparts, a novel family of PLP- dependent enzymes emerges. Biochemically, alanine racemase activity has been previously detected specifically in Tox2+ isolates of C. carbonum. Initial purification of HC-toxin biosynthetic activities resulted in the identification of two enzymes, HTS-1 and HTS-2 (Walton, 1987). HTS-1, with a molecular weight of ~220 kDa, epimerizes L-proline to D-proline and catalyzes ATP/PP. exchange in the presence of L- proiine; HTS-1, with a molecular weight of ~160 kDa, appeared to epimerize L- alanine to D-alanine and catalyze both L-alanine-depedent and D-alanine- dependent ATP/PP. exchange. Later studies identified and cloned a single 15.7- kb ORF that encodes the entire HTS protein; the appearance of two separate 107 proteins as HTS-1 and HTS-2 was, in fact, a purification artifact (Panaccione et al., 1992; Scott-Craig et al., 1992). Sequence analysis indeed found epimerization signatures in the region between domain A and B of HTS and is believed to be responsible for D-proline formation, but no similar signatures were found after domain C which is assumed to activate D-alanine (Scott-Craig et al., 1992; Kleinkauf and von Dohren, 1996). Together with the purification of a separate alanine racemase protein from the cyclosporin-producing fungus Tolypocladium niveum (Hoffmann et al., 1994), we speculated that the epimerization activity detected in previously named HTS-2 enzyme might result from co-purification of an independent alanine racemase. The cloning of TOXG together with genetic evidence presented in this paper proved this speculation. The clustering of TOXG with the recently reported TOXF bears both similarity and contrast with that of HTS1 and TOXA. There is a 386-bp space between the transcriptional start sites of HTS1 and TOXA (Pitkin et al., 1996), and 195 bp between TOXF and TOXG (Figure 4-2). Both pairs of genes are transcribed divergently. However, the evidence showed that intergenic regions of these clusters contain different elements with regard to response to the pathway- specific transcription factor encoded by TOXE. Both TOXF and TOXG are regulated by TOXE. In a TOXE null mutant, both TOXF and TOXG are silent (Figure 4-4). TOXA is also silent in the TOXE null mutant, but the expression of HTS1 remains unaffected (Ahn and Walton, 1998). A similar multifaceted regulation mechanism has been found in the penicillin biosynthesis pathway (Brakhage, 1998). 108 In the broad view, the clustering of genes involved in fungal secondary metabolism pathways is common, if not the rule. Gene clusters have been discovered in Penicillium for penicillin biosynthesis, in Aspergillus for aflatoxins/sterigmatocystins, in Fusarium for trichothecenes, and in Gibberella for gibberellins (Yu et al., 1995; Brown et al., 1996; Trapp et al., 1998; Tudzynski and Holter, 1998). In contrast, genes involved in HC-toxin production are spread over more than 540 kb and show only limited clustering (Ahn and Walton, 1996, 1998). The distance between TOX2 genes can be as short as a few hundred bp (between HTS1 and TOXA, and TOXF and TOXG), or can be as far as 220 kb between HTS1 copy 1 and TOXC copy 2. The molecular basis of interactions between C. carbonum and its host, maize, is one of the most thoroughly studied cases of plant fungal diseases. Together with TOXG, four genes (HTS1, TOXC, TOXF and TOXG) encode enzymes involved in HC-toxin biosynthesis, one gene (TOXA) encodes an efflux pump for toxin export, and TOXE encodes a transcription factor that regulates the expression of all known TOX2 genes other than HTS1. Taking the complexity of Aeo into account, at least four more biochemical activities are required for its biosynthesis. One is a putative fatty acid a subunit, which would work together with the fatty acid synthase (3 subunit encoded by TOXC, to synthesize decanoic acid from acetyl-CoA and malonyl-CoA; one enzyme is needed to oxidize decanoic acid to 2-oxo-decanoic acid; and two other enzymes are postulated to catalyze the formation of the 8-carbonyl group and the 9,10-epoxide group of Aeo. Whether the genes encoding these putative proteins are clustered with 109 known TOX2 genes, and whether they are regulated by TOXE, awaits future studies. 110 CHAPTER 5 SUMMARY AND PERSPECTIVE lll The studies presented in this dissertation focused on the molecular genetics of HC—toxin biosynthesis. The striking features shared by the known TOX2 genes (HTS1, TOXA, TOXC and TOXE) are: (1) Each has two or three functional copies per genome; (2) they are present only in HC-toxin-producing isolates (Tox2”) of C. carbonum; (3) they are scattered over ~540 kb on a special chromosome; (4) when all copies of an individual gene are mutated (except TOXA, which was unsuccessful due to lethality), the mutants lost the capability to produce HC-toxin and also fungal pathogenicity. These features guided my studies toward cloning additional genes that are involved in HC-toxin production. By taking a bacterial artificial chromosomal clone-mapping, transcript- screening and sequencing approach, I have identified two additional genes, TOXF and TOXG, that participate in HC-toxin biosynthesis. TOXF and TOXG are two tightly linked genes. Both are present in three copies in the standard lab strain SB111, and two functional copies in isolate 164R10; both are exclusively present in Tox2+ isolates of C. carbonum; both map to the TOX2 locus, and the expression of both is regulated by TOXE. TOXF encodes a branched-chain amino acid aminotransferase that is believed to aminate an a-keto acid on the Aeo biosynthesis pathway. A null mutant of TOXF no longer produces HC-toxin and causes severe leaf spot disease on maize leaves. Therefore, TOXF can be regraded as a pathogenicity gene. TOXG encodes a eukaryotic alanine racemase that catalyzes the interconversion of L-alanine and D-alanine. D-alanine is a critical constituent in 112 HC-toxin I, III, and IV. A null mutant of TOXG fails to make HC-toxin l, III and IV, but the biosynthesis of HC-toxin II is unaffected because HC-toxin II has a glycine in place of D-alanine. Compared to wild type, the TOXG mutant causes a delayed disease phenotype that eventually results in full symptoms due to the presence of HC-toxin Il. Therefore, TOXG can only be classified as a virulence gene. The biochemical function of TOXF is unproven due to the lack of substrate and assay methodology. In contrast, genetic and biochemical experiments were successfully adopted to confirm TOXG function. TOXG complemented a bacterial mutant defective in D-alanine biosynthesis. Alanine racemase activity (L to D) was also detected in a bacterial crude extract that expresses TOXG protein (TOXGp) under a constitutive promoter. Attempts to study the enzyme kinetics were not successful due to the insolubility of overexpressed TOXG gene product in an E. coli expression system (data not shown). Another TOX2 candidate gene, TOXD, was cloned many years ago, based on its indirect linkage to TOXC and its unique presence in Tox2“ isolates of C. carbonum. TOXD shares the first three features (previous page) of known TOX2 genes. However, HC-toxin production and fungal pathogenicity were unchanged in a TOXD null mutant. It appeared that TOXD has no obvious role in HC-toxin production. Recent reports showed that TOXD has strong similarity to the lovC gene of the lovastatin gene cluster in Aspergillus terreus. Knowledge from other systems may ultimately assist us in understanding the function(s) of TOXD in the future. 113 TOXF and TOXG along with the previously identified TOX2 genes (including TOXD) have been mapped to the TOX2 locus on the 3.5—Mb chromosome in strain SB111 (Figure 5-1) (Ahn and Walton, 1996; J. -H. Ahn and J. D. Walton, unpublished data). In essence, Pac l sites were introduced into each copy of TOXF/G by transformation-mediated homologous recombination. Chromosomes or chromosomal fragments were then separated by pulsed-field gel electrophoresis (CHEF), and Southern hybridizations were performed to decide the locations of the three copies of TOXF/G relative to other mapped genes. TOXF/G copy 1 is located about 20 kb away from HTS1 copy 1 toward HTS1 copy 2. TOXF/G copy 2 is about 20 kb away from TOXD copy 1 toward TOXD copy 3 and TOXF/G copy 3 is about 40 kb away from TOXC copy 1 (Figure 5-1). Up to now, a total of six genes (HTS1, TOXA, TOXC, TOXE, TOXF, and TOXG) have been cloned and shown to be involved in HC-toxin production. Among them, four genes (HTS1, TOXC, TOXF, and TOXG) encode biosynthetic enzymes, one gene (TOXE) encodes a pathway trancriptional factor that regulates the expression of TOXA, TOXC, TOXF and TOXG but not HTS1, and one (TOXA) encodes a membrane protein to export HC-toxin. The biosynthesis pathway for HC-toxin is tentatively summarized (Figure 5—2). As for the four amino acid constituents of HC-toxin, L-alanine comes from the cellular amino acid pool; D—proline is from L-proline by the action of an epimerization domain in HC-toxin synthetase (HTS) encoded by HTS1; the TOXG protein converts L- alanine into D-alanine, and L-Aeo comes from a long route. Aeo has a backbone 114 1r;— 3% 852335 ass; .0 ... as... cc< .I- .... Home 50:35 ucm cc< So: 838$ mcozficoto cozazomcm: m..: 2865 9592. comes: 38650 $9855 69E? ocom 9.56:2 69:5: 059.4 $62683. .GXOR ucm [XOR .QXOR .OXOR fxok ngI E8659 0 ucm .u. d .0 .< .I .Amv oEomoEoEo as.-m.n 05 0:13 mocom 86;. no no... 35035020 .Tm 959". n U [Rev 2.“ 3 9.5 m I I I n.- e. R .8 om. noun 8 .. e_ o: .30. .3...— 115 .EochmE £35585 5 Exofio: 65 Co EEmmE .~-m 959". w 3:093 + oo<... Eon 20:33 IOC_E