Horizontal gene and chromosome transfer in plant pathogenic fungi affecting host range

Authors


  • Editor: Gerhard Braus

Correspondence: Pierre J.G.M. de Wit, Laboratory of Phytopathology, Wageningen University, Droevendaalsesteeg 1, 6708 PB Wageningen, The Netherlands. Tel.: +31 317 483 130; fax: +31 317 483 412; e-mail: pierre.dewit@wur.nl

Abstract

Plant pathogenic fungi adapt quickly to changing environments including overcoming plant disease resistance genes. This is usually achieved by mutations in single effector genes of the pathogens, enabling them to avoid recognition by the host plant. In addition, horizontal gene transfer (HGT) and horizontal chromosome transfer (HCT) provide a means for pathogens to broaden their host range. Recently, several reports have appeared in the literature on HGT, HCT and hybridization between plant pathogenic fungi that affect their host range, including species of Stagonospora/Pyrenophora, Fusarium and Alternaria. Evidence is given that HGT of the ToxA gene from Stagonospora nodorum to Pyrenophora tritici-repentis enabled the latter fungus to cause a serious disease in wheat. A nonpathogenic Fusarium species can become pathogenic on tomato by HCT of a pathogenicity chromosome from Fusarium oxysporum f.sp lycopersici, a well-known pathogen of tomato. Similarly, Alternaria species can broaden their host range by HCT of a single chromosome carrying a cluster of genes encoding host-specific toxins that enabled them to become pathogenic on new hosts such as apple, Japanese pear, strawberry and tomato, respectively. The mechanisms HGT and HCT and their impact on potential emergence of fungal plant pathogens adapted to new host plants will be discussed.

Introduction

Horizontal gene transfer (HGT) is defined as the stable integration of genetic material following transfer between individuals. HGT excludes transfer through meiotic or mitotic processes, which is also known as vertical gene transfer (Rosewich & Kistler, 2000). The involvement of HGT in prokaryotes and its impact on evolution have been studied extensively (Koonin et al., 2001; Kado, 2009; Kelly et al., 2009). HGT in bacteria occurs through conjugation, which requires physical contact between cells, transduction, involving DNA transfer by bacteriophages, or transformation, involving the uptake of free DNA from the environment (Natarajan et al., 2005; Kelly et al., 2009). In bacteria, HGT has contributed significantly to genome plasticity and is a driving force in their diversification and adaptation to new environments including new host plants (Juhas et al., 2009). Classic gene transfer within eukaryotic cells, also referred to as endosymbiotic gene transfer, including the transfer of genes from organelles like mitochondria and chloroplasts to the nucleus is well described (Kleine et al., 2009), but little is known about HGT between different individuals within the same species or unrelated species.

In fungi, there are several reports on suspected HGT, but clear experimental evidence indicating that the transferred genetic material originates from another related or unrelated fungal species is often lacking. This might be due to the limited availability of whole-genome sequence information from fungal species, and also due to the fact that many HGT events might have occurred a long time ago in evolutionary history and are difficult to prove due to the different rates of evolution that have subsequently occurred in donor and recipient organisms. With the availability of new and cheaper DNA sequencing technologies, more whole fungal genome sequences have become available. This allows a more precise identification of ‘foreign DNA’ acquired from different fungal species by HGT. In some cases, evidence for HGT could be provided as it occurred fairly recently in evolutionary history and conferred a selective advantage to the recipient species including gain of pathogenicity on a particular host. A good example of recent interspecific HGT between Stagonospora nodorum and Pyrenophora tritici-repentis has been provided by Friesen et al. (2006). They showed evidence for the transfer of a gene encoding a host-selective toxin (ToxA) from S. nodorum that conferred virulence to P. tritici-repentis on wheat and suggested that it occurred just before 1941, when P. tritici-repentis caused the first serious outbreak of tan spot on wheat. Recently, the occurrence of horizontal chromosome transfer (HCT) between a pathogenic and a nonpathogenic Fusarium species has also been reported, enabling the latter one to become pathogenic (Ma et al., 2010). HCT also provides a powerful mechanism of genetic exchange between fungi, enabling them to become pathogenic on new host plants. Recently, interspecific hybridization between different pathogenic species has also been reported, which has led to hybrid offspring with an expanded host range. One example involves the formation of hybrids between Phytophthora infestans and Phytophthora mirabilis, which are pathogens of potato/tomato and Mirabilis jalapa, respectively, while a fraction of the offspring can infect all three plant species (Kroon, 2010). These recent findings prompted us to review HGT, HCT and hybridization between plant pathogenic fungi and discuss the impact on the emergence of new pathogens adapted to new host plants.

HGT

Pyrenophora tritici-repentis

The fungus P. tritici-repentis causes the tan spot disease on wheat, but it was not considered a serious pathogen before 1941. Surprisingly, after 1941, P. tritici-repentis spread rapidly and became one of the major wheat pathogens in the United States and gradually also in other parts of the world (Friesen et al., 2008; Oliver & Solomon, 2008). Molecular and genetic analysis of P. tritici-repentis strains showed that the genome contained a host-selective toxin-producing gene, ToxA, which is highly similar to ToxA from the wheat pathogen S. nodorum (Ciuffetti et al., 1997; Friesen et al., 2006) (Table 1). The ToxA gene is present in both organisms, but is absent from all other cereal pathogens in Dothideomycete fungi, including closely related species in both genera. Comparison of ToxA and both flanking sequences in these two pathogens revealed that both fungi contain an 11-kb region surrounding ToxA that is highly similar, suggesting that it has been transferred between the two fungi (Friesen et al., 2006). ToxA sequences analyzed in 57 isolates of P. tritici-repentis collected in a large geographical area showed that they were almost identical, whereas considerable polymorphisms were identified in ToxA sequences analyzed in 600 S. nodorum strains (Faris et al., 2010). This suggests that the ToxA gene originated and evolved in S. nodorum and that the 11-kb region was transferred recently from S. nodorum to P. tritici-repentis by HGT. The transfer of ToxA has likely caused the strong increase of the virulence of this fungus on wheat, which carries the Tsn1 gene that encodes a protein that determines sensitivity to ToxA. Tsn1 encodes a disease resistance gene-like protein with a serine/threonine protein kinase, a nucleotide-binding site and leucine-rich repeat domains (Faris et al., 2010). Tsn1 is necessary to mediate ToxA recognition, which suggests that necrotrophic pathogens may thrive on host plants by subverting the resistance mechanisms acquired by plants to combat other pathogens. Additional experimental data showed that disruption of the ToxA gene in S. nodorum strains resulted in the loss of ToxA production in the mutants that were no longer virulent on normally susceptible wheat cultivars (Faris et al., 2010). Complementation with the P. tritici-repentis ToxA gene of the ToxA mutants of S. nodorum restored their virulence towards normally susceptible wheat cultivars. Thus, obtaining the ToxA gene by HGT allowed P. tritici-repentis to cause a serious disease in wheat. This event likely occurred shortly before 1941, after which the ToxA-expressing P. tritici-repentis isolate was globally distributed by shipments of infected grain, causing the emergence of a highly pathogenic form of P. tritici-repentis in wheat fields worldwide (Friesen et al., 2006).

Table 1.   Effector genes present in fungi including genes encoding ribosomal and nonribosomal toxins that are involved in pathogenicity and are supposed to be transferred by HGT or HCT
FungusEffector gene(s)Gene productsHGT/HCTRole in pathogenicity
  1. For further details, see text.

Pyrenophora tritici-repentisToxANecrogenic ribosomal peptideHGTEssential to cause new disease on wheat
Nectria haematococcaPEP genesSeveral enzymes, including  
 pisatine demethylaseHGTEssential to cause disease on pea
Cochliobolus heterostrophus
(race T)Tox1 genesT-toxin; long chainpolyketidesHGTEssential to cause disease on corn
Cochliobolus carbonum
(race 1)TOX2 genesHC-toxin; cyclic peptideHGTEssential to cause disease on corn
Alternaria alternate
apple strainAMT genesAM-toxin; cyclic peptideHCTRole in pathogenicity and host specificity
strawberry strainAFT genesAF-toxin; unsat. fatty acidHCTRole in pathogenicity and host specificity
Japanese pear strainAKT genesAK-toxin; unsat. fatty acidHCTRole in pathogenicity and host specificity
tomato strainALT geneAAL-toxin; polyketideHCTRole in pathogenicity and host specificity
Fusarium oxysporumSIX genesRibosomal peptidesHCTSeveral have a role in pathogenicity

Cochliobolus heterostrophus and Cochliobolus carbonum

Cochliobolus heterostrophus is the causal agent of Southern corn leaf blight and was first described in 1925 as a pathogen of corn. Before 1970, race O of this pathogen caused a common, but insignificant disease of corn. In 1970, a new form, race T, was identified that was extremely virulent on corn, causing a major epidemic, resulting in 20–30% yield loss in the United States’ corn production. (Ullstrup, 1972). Race T produces the T-toxin, a family of linear long-chain (C33–C45) polyketides (Chang & Bronson, 1996). Analyses of many C. heterostrophus strains have shown that the essential difference between race T and race O is the presence of a 1.2-Mb fragment of DNA carrying the T-toxin-producing gene, Tox1, in the T race (Yang et al., 1996) (Table 1). Presently, nine Tox1 genes are known to be involved in T-toxin production, including two polyketide synthases (including PKS1), one decarboxylase, five dehydrogenases and a protein with an unknown function (Yang et al., 1996; Rose et al., 2002; Baker et al., 2006). Tox1 genes are not only absent in race O, but they are also absent in closely related Cochliobolus species and other genera (Rose et al., 2002). Based on these data as well as codon usage, the A+T content and the genetic location of the Tox1 locus, an origin by HGT was suggested (Yang et al., 1996), but could not be established with certainty. To trace the evolutionary history of the PKS1 gene, DNA from >100 Dothideomycete species was screened for homologs (Inderbitzin et al., 2010). An ortholog (60% identity) was found in Didymella zeae-maydis, which produces the PM-toxin, a polyketide with a structure and biological specificity similar to that of the T-toxin. Only one additional Dothideomycete species, Delitschia winteri, harbored a paralog, but this could not represent the donor of PKS1 as it was more homologous to PKS1 of the Eurotiomycetes. The unresolved evolutionary history and distinctive gene signature of PKS1 (fast-evolving, discontinuous taxonomic distribution) strongly suggest HGT, but final evidence is lacking.

Another suspected case for HGT is reported in C. carbonum, which is the causal agent of Northern corn leaf spot. Susceptibility to C. carbonum race 1 is based on the production of the HC-toxin, a cyclic tetrapeptide (Walton, 1996). At least five genes are required for HC-toxin biosynthesis (collectively known as the TOX2 locus) (Table 1). These genes are loosely clustered over a region of >500 kb in C. carbonum. Interestingly, all of the known TOX2 genes are absent in natural nonpathogenic strains of C. carbonum (Walton, 2006). The absence of all genes in non-HC-producing isolates may suggest that an HGT event for the TOX2 locus has occurred in the past from an unknown organism to C. carbonum (Walton, 2000). Alternatively, it is possible that extensive chromosomal rearrangements observed in this fungus have led to loss of the TOX2 chromosomal region, resulting in isolates that are no longer able to produce the HC-toxin. However, the present experimental data cannot exclude either of the two hypotheses.

Nectria haematococca

The ascomycetous fungus N. haematococca (anamorph: Fusarium solani) is a member of a large group of 50 species known as the F. solani species complex. This species contains several mating populations (MPs) that are closely related (O'Donnell, 2000). An MP defines a group of isolates that is sexually active and produces sexual progeny, implying that it is a biological species (Coleman et al., 2009). Nectria haematococca (MPVI) contains a cluster of pea pathogenicity (PEP) genes (the so-called PEP cluster), which is located on supernumerary chromosome 14, that is required to cause disease on pea, but is dispensable for growth in vitro (Temporini & VanEtten, 2004). The PEP cluster genes encode, among others, enzymes that detoxify the phytoalexin pisatin, which is an important antimicrobial compound produced by pea after infection by fungi including N. haematococca (Funnell et al., 2002). There are several lines of evidence indicating that the PEP cluster has been introduced into N. haematococca via HGT. The genes present in the PEP cluster have a distinctly different G+C content and codon usage compared with those of other genes present in the genome of this fungus. More importantly, the PEP cluster is absent in the genomes of members of the same species that are very closely related, including N. haematococca MPI, MPIII, MPV and MPVII. Interestingly, the PEP cluster is also present in another pea pathogen, Fusarium oxysporum f. sp. pisi. In addition, the sequences of the PEP cluster in N. haematococca MPVI have an unusually high degree of similarity to the sequences present in F. oxysporum f. sp. pisi, which is phylogenetically distant by other criteria. The discontinuous phylogenetic distribution of the PEP cluster among Fusarium species, different DNA composition and feature of the PEP cluster compared with the genome and the unusually high level of similarity of the PEP cluster between the two distantly related fungi suggest that the PEP cluster may have been acquired by N. haematococca MPVI through HGT (Temporini & VanEtten, 2004).

Recently, it was shown that in addition to chromosome 14, chromosomes 15 and 17 of MPVI are also dispensable (Coleman et al., 2009). Whether these two chromosomes can be achieved by HGT enabling the fungus to occupy more niches and infect additional plant species needs to be determined.

HGT between fungi and oomycetes

Fungi and oomycetes represent some of the most distantly related eukaryotes, but their morphology and life style are very similar (Latijnhouwers et al., 2003). The fungi group with the animals, whereas the oomycetes group with photosynthetic algae, but both show filamentous growth and do infect plants and other hosts and cause many economically important diseases. Because of their distant relationships, it is expected that the similarities between genes present in both organisms could have resulted from HGT. Richards et al. (2006) explored the evolutionary history of over 11 000 predicted genes present in the genome of the filamentous ascomycete fungus Magnaporthe grisea and investigated whether HGT could have occurred between both organisms. They found that 11 M. grisea genes showed a significantly higher level of similarity to sequences present in the oomycete genus Phytophthora than to any fungal sequences. When more stringent criteria were applied in various phylogenetic analyses, four oomycete genes clearly grouped within the fungal clade in the phylogenetic trees with very high confidence, indicating that these genes most likely have originated from a fungal source possibly by HGT. Of these four genes, the first encodes a sugar transporter that could potentially increase the accessibility of sugar substrates for an osmotrophic oomycete. The second gene putatively encodes a permease protein that potentially facilitates the access of oomycetes to nucleotide substrates. The third gene putatively encodes an enzyme involved in the degradation of aromatic compounds and is likely part of the β-ketoadipate pathway. The fourth gene encodes an aldose-1-epimerase with broad substrate specificity and possibly is a key enzyme in lactose metabolism. Although no experimental proof has been provided as yet, this in silico analysis supports the occurrence of HGT between fungi and oomycetes. The mechanism of HGT between fungi and oomycetes is not yet clear, but anastomosis between mycelia, transduction via mycoviruses or propagation of retrotransposons could have facilitated HGT (Richards et al., 2006).

HCT between fungi

Colletotrichum gloeosporioides

Colletotrichum gloeosporioides is a facultative parasite with two different biotypes (A and B) that are indistinguishable in culture, but are asexual and vegetatively incompatible. Biotypes A and B have been shown to be genetically distinct, while isolates within each biotype appear to be nearly monomorphic (Braithwaite et al., 1990). Unlike biotypes A, most isolates of biotype B lack a supernumerary 2-Mb chromosome (Masel et al., 1996). Interestingly, some field isolates of biotype B contain this supernumerary chromosome that is thought to originate from biotype A by HCT (He et al., 1998). Experiments were performed under laboratory conditions to find experimental support regarding whether HCT could occur between these vegetatively incompatible biotypes. A transformant of biotype A, carrying the hygromycin resistance gene integrated in the 2-Mb chromosome, was co-cultivated with a biotype B transformed with the phleomycin resistance gene. Interestingly, double antibiotic-resistant colonies were obtained from conidia harvested from these co-cultivated mixed cultures that contained the 2-Mb chromosome in a biotype B genetic background (He et al., 1998). These results demonstrate that HCT between two vegetative incompatible biotypes of C. gloeosporioides can occur during co-cultivation under laboratory conditions and this most likely also occurs in nature (Masel et al., 1996; He et al., 1998).

Alternaria alternata

Alternaria alternata is an asexual filamentous fungus, of which both pathogenic and nonpathogenic pathotypes do occur. Pathogenic pathotypes cause leaf spots and blights on different hosts plants and usually harbor small supernumerary or conditionally dispensable chromosomes (CDCs) of <1.7 Mb, which are absent in nonpathogenic isolates (Akamatsu et al., 1999). All host-specific toxin-encoding genes are located on supernumerary chromosomes. For example, apple, strawberry, Japanese pear and the tomato pathotypes produce AM, AF, AK and AAL-toxins, respectively, which define the host specificity of A. alternata pathotypes (Lee et al., 1992; Tanaka et al., 1999; Harimoto et al., 2007) (Box 1). Indeed, the cyclic peptide synthetase gene, AMT, which is involved in AM-toxin biosynthesis of the apple pathotype of A. alternata, was located on a small chromosome of 1.1–1.7 Mb, depending on the strain. At least four genes AMT1, AMT2, AMT3 and AMT4 are involved in AM-toxin biosynthesis (Harimoto et al., 2008). In the apple pathotype strains, these genes were found to be clustered and reside on small CDCs, which are not required for growth, but do confer an advantage for colonizing certain ecological niches. Indeed, in A. alternata strain, IFO8984, the AMT genes reside on a 1.4-Mb chromosome and are clustered within about a 120-kb region (Harimoto et al., 2007, 2008). Interestingly, multiple copies of the AMT gene cluster are also found in the genome of A. alternata. Strain IFO8984 seems to contain three sets of the AMT gene cluster in the genome, which all have a similar structure on the 1.4-Mb chromosome (Harimoto et al., 2007, 2008). The AF-toxin biosynthesis gene cluster (TOX cluster or AFT cluster) was also reported to be present on a single small chromosome of 1.05 Mb in the strawberry pathotype of A. alternata (Hatta et al., 2002). The AKT1, AKT2, AKT3 and AKTR genes (AKT cluster) are involved in AK-toxin biosynthesis and DNA gel blot analysis with pulsed-field gel electrophoresis (PFGE) showed that the AKT gene cluster is located on a single 4.1-Mb chromosome (Tanaka & Tsuge, 2000). Finally, the AAL-toxin is produced by all tomato pathotypes and is synthesized through a polyketide biosynthetic gene, ALT1. PFGE analysis showed that all strains of the tomato pathotype contained a 1.0-Mb CDC, indicating that also in this pathogen ALT1 is encoded on a 1-Mb CDC (Akamatsu, 2004).

The origin and evolution of CDCs is unknown, but it has been suggested that CDCs are horizontally transferred across A. alternata pathotypes and can explain their host range and host-range adaptation. This hypothesis is supported by the fact that the sequences of genes located on the CDC of a worldwide tomato pathotype collection were highly identical, whereas the sequences of genes located on chromosomes other than the CDC showed high levels of polymorphisms, indicating that the origin of the CDC might be different from that of the core chromosomes present in the tomato pathotype (Akagi et al., 2009). In addition, in protoplast fusion experiments, it was shown that the CDC from a tomato pathotype could be transferred into a strawberry pathotype, which made the resulting fusant pathogenic on both tomato and strawberry plants (Akagi et al., 2009). The high identity between supernumerary chromosomes as compared with core chromosomes in A. alternata strains strongly suggests that supernumerary chromosomes have recently been transferred into nonpathogenic strains of A. alternata possibly by HCT. Thus, in A. alternata, HCT could provide a possible mechanism by which pathogens with novel host specificities might have arisen in nature (Akagi et al., 2009). However, the origin of the supernumerary chromosomes in A. alternata pathotypes is as yet unknown.

Fusarium oxysporum

Fusarium is a large genus with over 20 species and contains saprophytic, opportunistic and pathogenic species. Plant pathogenic Fusarium species are very diverse, widely dispersed and attack different parts of plants, causing many economically important diseases (Summerell & Leslie, 2004). Among the Fusarium species, F. oxysporum is a ubiquitous soil inhabitant and one of the most important plant pathogenic species in the Fusarium genus (Michielse & Rep, 2009). Although they are predominantly harmless as soil saprophytes, many formae speciales are found within the F. oxysporum complex that cause disease in only a narrow range of plant species. Host adaptation and specificity within formae speciales have been studied extensively (Lievens et al., 2008), but the evolutionary origin of the host specificity genes is unknown. Comparison of the genomes of Fusarium graminearum, Fusarium verticillioides, F. solani and F. oxysporum f.sp. lycopersici revealed four lineage-specific (LS) chromosomes in F. oxysporum that are rich in transposons, containing genes encoding proteins involved in signal transduction, and effector proteins involved in pathogenicity and virulence (Ma et al., 2010). Each of the four Fusarium species carries a core set of chromosomes with a high level of synteny. Fusarium oxysporum f.sp. lycopersici and F. solani each have LS chromosomes that are distinct with regard to repetitive sequences and genes involved in pathogenicity, indicating that LS chromosomes may have a distinct evolutionary origin compared with the core chromosomes. Interestingly, among the LS chromosomes, the 2-Mb chromosome 14 of F. oxysporum f.sp. lycopersici is enriched in genes encoding secreted effectors such as SIX1, SIX2, SIX3, SIX5, SIX6 and SIX7, of which some have proven to be virulence factors (Table 1). This suggests that chromosome 14 might carry the main determinants for adaptation of F. oxysporum towards tomato (Ma et al., 2010). Chromosome 14 and another smaller strain-specific chromosome can undergo transfer between pathogenic and nonpathogenic strains during co-cultivation, resulting in a new pathogenic lineage. The effector genes are all conserved in strains causing tomato wilt, but are generally not present in other strains. LS regions are highly enriched in transposable elements as they contain >74% of the identifiable transposable elements present in the genome, including 95% of all DNA transposons. Only 20% of the predicted genes in the LS regions could be functionally classified on the basis of homology to known proteins. In addition to effector genes, these regions are enriched for a variety of cell wall-degrading enzymes, genes for lipid metabolism, transcription factors and proteins involved in signal transduction, but are deficient in genes for housekeeping functions. Notably, many of these genes are expressed during the early stages of tomato root infection. Codon usage and codon adaptation index analysis indicated that the LS-encoding genes exhibit distinct codon usage and have a higher G+C content compared with the conserved genes on core chromosomes, supporting distinct evolutionary origins (Ma et al., 2010).

It is also hypothesized that horizontal transfer of chromosome 14 from F. oxysporum f.sp. lycopersici to nonpathogenic F. oxysporum strains confers pathogenicity of those strains towards tomato. To prove this hypothesis, chromosome 14 of the tomato strain was marked with the zeocin resistance gene, whereas a nonpathogenic F. oxysporum strain was marked with the hygromycin-resistant gene. Microconidia of the two transgenic strains were mixed on agar plates and after 6–8 days of incubation, spores emerging on these plates were selected for resistance to both zeocin and hygromycin. Interestingly, double drug-resistant colonies were recovered and further molecular analyses proved that the tomato strain exclusively behaved as the donor of chromosome 14 to the recipient nonpathogenic strain. Subsequent pathogenicity assays demonstrated that the nonpathogenic strains had now gained the ability to infect tomato plants, indicating that the pathogenicity of nonpathogenic F. oxysporum strain towards tomato can be specifically attributed to the acquisition of F. oxysporum f.sp. lycopersici chromosome 14 by HCT (Ma et al., 2010). These experiments also demonstrated that simple co-cultivation of genetically distinct strains can easily generate new pathogenic genotypes and these events might have also occurred in nature in the past. This finding may also explain the rapid emergence of new pathogenic lineages in distinct nonpathogenic genetic backgrounds (Ma et al., 2010).

Interspecies hybridization leading to the broadening of host range

The possibility of hybridization between two closely related species of plant pathogenic fungi or oomycetes has been considered for many years, but conclusive proof of interspecies hybridization has been provided only recently with the arrival of modern molecular genetic tools.

There is increasing evidence that hybridization between different Phytophthora species plays a major role in the generation of new species that can potentially infect new host plants. The incidence of interspecific hybridization may have increased due to intensifying global trade and traffic of host plant species infected by different plant pathogens.

The occurrence of natural hybrids between Phytophthora nicotianae and Phytophthora cactorum has been reported in different countries (Man In't Veld et al., 1998; Hurtado-Gonzales et al., 2009). Hybrids between Phytophthora porri and Phytophthora primulae have also been reported (Declercq et al., 2010). Whether these hybrids have an extended host range has not been tested as yet.

Interspecies hybrids between P. infestans and P. mirabilis could also be generated in the laboratory (Kroon, 2010). The resulting F1 isolates were intercrossed and backcrossed to P. infestans and P. mirabilis isolates. The genetic makeup of the hybrids was verified by DNA fingerprinting, and the pathogenicity of the hybrids was tested in detached leaf assays on M. jalapa (a host of P. mirabilis) and potato and tomato (hosts of P. infestans). Hybrids were shown to share the host range of both the parental isolates. Not much is known about the detailed genetic constitution of the reported hybrids and we will not discuss the phenomenon of interspecies hybridization in further detail. However, interspecies hybridization could be considered as a potential mechanism to broaden the host range of fungi.

Gain and loss of chromosomes by vertical chromosome transfer between fungi

In contrast to HCT, vertical chromosome transfer occurs when an organism receives genetic material from its ancestor, for example its parent or a species from which it evolved. Vertical chromosome transfer has contributed significantly to genome plasticity. In some cases, non-disjunction during meiosis may occur and hence has generated copy number polymorphisms (CNPs). Strains with CNPs in haploid filamentous fungi such as Neurospora crassa are generally lethal or seriously impaired in the sexual phase (Perkins, 1997), while in other species, CNP does not seriously affect pathogenicity or fitness. The best example to illustrate this phenomenon has been reported for Mycosphaerella graminicola (Wittenberg et al., 2009).

Mycosphaerella graminicola (asexual stage: Septoria tritici) is a haploid hemibiotrophic ascomycete that causes the Septoria tritici blotch disease of wheat. This disease occurs in almost all wheat-growing areas (Eyal, 1999). Mycosphaerella graminicola is a model species for the Dothideomycetes that comprise over 3000 plant pathogens (Mehrabi et al., 2006). The fungus has a very active sexual lifestyle, which explains the presence of extremely diverse natural populations, where virtually all collected isolates are genotypically different. Based on detailed analyses provided by high-density genetic linkage maps (Wittenberg et al., 2009), and comparative genomic hybridizations (Goodwin et al., 2011), several CNPs were found with no apparent effect on fitness or virulence, suggesting that the genome of M. graminicola is highly plastic and that much of the genome plasticity is generated during meiosis. This extreme plasticity might explain the high adaptability observed in field populations of this pathogen.

The sequenced genome of M. graminicola is composed of a core set of 13 chromosomes and eight dispensable chromosomes that are distinct in structure, gene and repeat content. The dispensable chromosomes are smaller (ranging from 0.41 to 0.77 Mb); they comprise 38% of the DNA content, but have significantly lower gene densities than the core chromosomes (only 6% of total number of genes). This so-called dispensome is composed of a higher density of transposons and shows a different codon usage. Most of the genes are duplicated on the core set of chromosomes and little synteny is observed between the M. graminicola dispensome and those of other Dothideomycetes.

Dispensable chromosomes have been discussed before to occur in other fungi, but there they typically represent only a single or a few chromosomes. They have also been shown to contain genes involved in pathogenicity or virulence (Han et al., 2001; Hatta et al., 2002; Garmaroodi & Taga, 2007), which has not yet been shown experimentally for M. graminicola. In contrast, Ware (2006) showed that all mapped loci of M. graminicola that are involved in pathogenicity reside on the core set of chromosomes.

A dispensome of eight is the highest number of dispensable chromosomes reported so far in fungi. Their function is still unclear, but they should confer a selective advantage to the pathogen; otherwise, they would be lost from the population. For their origin, two hypotheses have been put forward. They could either represent degenerated copies of core chromosomes that have once been obtained by hybridization or they could have been obtained by HCT. The latter hypothesis is favored, but conclusive evidence for HCT is lacking (Goodwin et al., 2011).

Potential mechanisms of HGT and HCT

As discussed before, intra- and interspecies HGT and HCT occur, but it is not clear whether they require different mechanisms. DNA exchange through physical contact requires the passage of DNA through the cell wall and the cell membrane of the donor and the recipient fungal cells. Gram-negative bacteria have evolved pili as a structure required for conjugation and subsequent interbacterial DNA transfer. However, no biological bridges like pili in bacteria have been identified in fungi.

Little is known about the mechanisms of HGT and HCT between the plant pathogenic fungi that have been discussed in previous paragraphs. In this paragraph, we will discuss the mechanisms of anastomosis in fungi that precede HGT and HCT. Fusion between fungal cells has been observed to occur through conidial, germ tube or hyphal anastomosis, but it is not clear whether these processes are regulated through the same mechanisms. However, all hyphal fusion mutants that have been analyzed so far are blocked in both hyphal fusion and conidial anastomosis, indicating that both phenomena might be governed by similar genetic mechanisms (Roca et al., 2003, 2005a, b). Conidial anastomosis occurs through specific tubes called conidial anastomosis tubes (CATs), which are different in shape and size compared with germ tubes (Ishikawa et al., 2010b) (Fig. 1). They are thin, usually unbranched and arise from conidia. CAT fusions have been observed in the majority of ascomycete fungi, covering 73 species of filamentous fungi in 21 genera, and they do occur both in vitro as well as during plant infection (Roca et al., 2005a, b; Ishikawa et al., 2010a). Their formation depends on conidial density, conidial age and nutritional and environmental factors (Ishikawa et al., 2010b). The three major phases of CAT-mediated cell–cell fusion are defined as CAT induction, CAT homing (CAT chemo-attraction) and CAT fusion (Fig. 1) (Read et al., 2009). It has been postulated that CATs not only improve the chances of colony establishment by distribution of nutrients or water within the environment, but CAT fusion may also facilitate parasexual recombination and may contribute to the high level of genetic variation found in fungi that lack sexual reproduction and expedite HGT between pathogens (Roca et al., 2005a, b; Read & Roca, 2006; Ishikawa et al., 2010b). Therefore, it has been suggested that CATs might act as biological bridges for genetic exchange between individuals of the same or different species (Proft & Baker, 2009).

Figure 1.

 Three types of anastomosis occurring in fungi. (a) Scanning electron micrograph of CATs in Neurospora crassa. Arrows show CATs grown toward each other and fused (adapted from Roca et al., 2005a). (b) Light microscopy image of germ tube anastomosis in Fusarium oxysporum f. sp. lycopersici (adapted from Ruiz-Roldán et al., 2010). (c). Light microscopy images of hyphal anastomosis in Mycosphaerella graminicola. (d) A higher magnification of the box shown in (c). Three distinct phases are depicted during the anastomosis process, including the (i) attraction, (ii) contact and (iii) fusion phase (adapted from Mehrabi et al., 2009).

In many fungi, including the majority of ascomycete fungi, conidial anastomosis between genetically identical cells known as self-anastomosis is regularly observed. Likewise, in several cases, non-self anastomosis (intraspecies anastomosis) and even interspecies anastomosis have been described (Roca et al., 2004), but so far, direct evidence for intergenus anastomosis has not been reported, but this might be due to the fact that it only occurs under specific conditions and at a low frequency (Ishikawa et al., 2010b).

In N. crassa, a mitogen-activated protein (MAP) kinase pathway, but not the cAMP pathway is involved in CAT induction. The MAPKKK (NRC-1), MAPKK (MEK-2), MAPK (MAK-2), the transcription factor STE12, a transmembrane protein HAM-2 and PP-1 are involved in CAT formation (Xiang et al., 2002; Pandey et al., 2004). In addition, a WW domain protein called SO plays a role in CAT formation because mutants defective in so form CATs less efficiently than the wild type (Xiang et al., 2002; Fleissner et al., 2005). MAK-2 and SO are involved in chemo-attraction and MAK-2 seems to play a role in fusion pore formation, whereas the plasma membrane protein, PRM-1, is involved in membrane merger (Read et al., 2009). In M. graminicola, a G protein-encoding gene, MgGpb1, negatively regulates cell fusion and anastomosis (Mehrabi et al., 2009). The MgGpb1 mutant produced germ tubes that undergo extensive anastomosis and subsequently produce an extremely dense biomass, which resulted from uncontrolled fusion. The addition of exogenous cAMP rescued uncontrolled anastomosis, suggesting that the cAMP pathway could be involved in the anastomosis process (Mehrabi et al., 2009).

The genetic basis of heterokaryon anastomosis in fungi is not yet clear, but the incompatibility mechanisms allow this process to occur only when the fungi are genetically identical. The viability of heterokaryons is genetically controlled by Het loci (for heterokaryon incompatibility). Heterokaryotic cells formed between individuals of different het genotypes undergo a characteristic cell death reaction or are severely inhibited in their growth. Several studies have shown that heterokaryon anastomosis can form between genetically distinct isolates, albeit at a low frequency (Di Primo et al., 2001; Roca et al., 2003, 2004; Toda & Hyakumachi, 2006; Qu et al., 2008; Young, 2009). In Colletotrichum Lindemuthianum, CAT fusion has been shown to occur previously in culture and recently also on the host surface and within the acervulus (asexual fruiting body) in anthracnose lesions on the host (Ishikawa et al., 2010a), suggesting that this phenomenon may frequently occur in nature and might potentially provide an important tool for genetic exchange between fungi. A detailed genetic analysis of two different vegetatively incompatible biotypes of C. gloeosporioides was discussed before. Although no cytological or molecular experiments were performed to further unveil the possible mechanism that generated chromosome transfer between these normally incompatible genotypes, it is suggested that either CAT or hyphal anastomosis fusions could have occurred in nature to allow the introgression of genetic material and the acquisition of supernumerary chromosomes.

Heterokaryon anastomosis can also result in the parasexual cycle, during which diploid cells are formed temporally that subsequently undergo mitotic crossing over and haploidization by the loss of individual chromosomes (McGuire et al., 2005; Castro-Prado et al., 2007; Forgan et al., 2007; Milgroom et al., 2009). If heterokaryon anastomosis occurs, it may facilitate unusual HCT or HGT as a mechanism for the exchange of genetic material between individuals (Fig. 2).

Figure 2.

 Schematic presentation of heterokaryon anastomosis in filamentous fungi leading to HGT or HCT. (a) Three stages of heterokaryon fusion are shown that can eventually lead to (b) HGT or (c) HCT.

One of the important issues that remains to be solved is how DNA exchange takes place after anastomosis. In C. lindemuthianum, nuclear migration through fused CATs using nuclei labeled with green fluorescent protein was frequently observed (Ishikawa et al., 2010b). It is assumed that after intraspecies or interspecies anastomosis, DNA fragments (in case of HGT) or intact chromosomes (HCT) may pass through CATs from one individual to another (Fig. 1). However, how chromosomes are released from the nucleus of the donor cell, pass through CATs and move to the nucleus of the recipient cell, and the signals that regulate this process are still largely unknown. Dissecting the mechanisms of self and non-self signaling between conidia and between CATs of the same and different genotypes remains a major challenge, but will be important to understand the molecular regulation of CAT, HGT or HCT.

It is also still unknown why and how most of the CD, LS or supernumerary chromosomes are enriched in transposon elements or repeats and whether these transposons and repeats facilitate horizontal transfer. It is known that retrotransposons transcribe themselves into RNA, possibly along with their proximate genomic areas, and are reverse transcribed to DNA that is pasted back into the genome at multiple sites. Class II transposons usually move by a cut and paste, rather than copy and paste, mechanism facilitated by the transposase enzyme. If this process occurs actively during heterokaryon anastomosis, the DNA generated might be released from the nucleus of the donor through transposon activity and move and insert somewhere else in the genome of the recipient. During the process of insertion, the repeat-rich regions or transposon sequences may facilitate homologous recombination among different repeats, leading to successful HGT. The challenging question to be addressed in the future is to show to what extent repeats and transposons promote horizontal transfer and whether the association of repeats and transposons with CD chromosomes or virulence factors is accidental or not.

HGT and HCT as a mechanism to acquire pathogenicity towards plants or to broaden the host range of a plant pathogen

The ability of fungal pathogens to colonize host plants and retrieve nutrients is crucial for their reproduction, survival and competitive abilities. Many fungal pathogens use vertical transfer to enhance their genetic variability and dynamically share their virulence factors like toxins and effectors through an active sexual cycle. Maintenance of high genetic variability by recombination is important for biotrophic fungal pathogens that need to adapt quickly to changing environments such as overcoming a new resistance gene in a crop or sensitivity towards crop protection agents. As discussed above, there are some examples of the emergence of new fungal and oomycete pathogens through interspecies hybridization, but interspecies sexual recombination seems not to occur frequently in most biotrophic fungal plant pathogens. Thus, sexual recombination may enable adaptation to a new cultivar of a host, while the fittest individuals may be selected by subsequent asexual reproduction during the growing season. Most fungal and oomycete pathogens reproduce sexually once or a few times per season, whereas many cycles of asexual reproduction occur during the growing season.

HGT or HCT could be a more efficient mechanism to become a new pathogen or extend the host range of a pathogen than sexual recombination. Likely, in fungi, HGT or HCT can facilitate adaptation to new host plants. Gaining important virulence genes by HGT or HCT like those encoding host-selective toxins, gene clusters or pathogenicity chromosomes can alter the host range of a fungal pathogen or even create a new pathogen from a nonpathogenic fungus as has been discussed in the previous section. The transfer of the ToxA gene from S. nodorum is a good example where HGT has caused a serious new disease. In a similar way, the transfer of pathogenicity chromosome 14 from a pathogenic Fusarium species to a nonpathogenic one is a good example where HCT transformed a nonpathogen into a pathogen.

HGT vs. HCT

As discussed above, HCT can occur for large chromosome segments or whole chromosomes and seems to occur more frequently than HGT. In C. heterostrophus, a 1.2-Mb chromosome carrying nine genes involved in T-toxin production was transferred from race O to race 1 strain (Yang et al., 1996; Rose et al., 2002; Baker et al., 2006). In C. carbonum, the suggested horizontally transferred TOX2 locus (involved in HC-toxin production) contains five genes that are loosely clustered over a region of >500 kb (Walton, 2000, 2006). In N. haematococca, the PEP cluster is located on a supernumerary chromosome 14 (Temporini & VanEtten, 2004; Coleman et al., 2009). In C. gloeosporioides, a supernumerary 2-Mb chromosome was transferred from biotype A to biotype B (Masel et al., 1996; He et al., 1998). As discussed in the previous sections in A. alternata, genes required for the production of different toxins such as AM-toxin, AF-toxin, AK-toxin and AAL-toxin gene clusters reside on supernumerary chomosomes between 1 and 4 Mb that can move around in the A. alternata population by HCT generating pathotypes adapted to different host plants (Tanaka & Tsuge, 2000; Hatta et al., 2002; Akamatsu, 2004). In F. oxysporum, chromosome 14 from the tomato pathotype was transferred to a nonpathogenic F. oxysporum strain. In all cases, it seems that the transfer of a large DNA segment or whole chromosome is a rule rather than an exception. The only exception is the transfer of an 11-kb region containing the ToxA gene that was transferred from S. nodorum to P. tritici-repentis by HGT rather than by HCT. Indeed, keeping all pathogenicity genes clustered on one large DNA fragment or chromosome can explain gain of pathogenicity for a nonpathogen. This could not or less likely be achieved when all pathogenicity genes would be scattered over the whole genome as this would require many independent transfers of the different genes.

Therefore, clustering of genes or the location of pathogenicity genes on a linkage-specific chromosome or supernumerary chromosome would provide a high selective advantage for efficient one-step horizontal transfer to enable a fungus to become pathogenic on a new host plant.

Conclusions and prospects

The importance of HGT and HCT to extend the host range of pathogens has only been appreciated recently by the scientific community. HGT can cause disease outbreaks on new crops when the encoded protein is an important pathogenicity factor as is the case for the acquisition of the ToxA gene by P. tritici-repentis, which now causes the new tan spot disease on wheat. Similarly, for species of Fusarium and Alternaria, it has been shown that transfer of a chromosome carrying crucial pathogenicity genes to nonpathogenic strains can render them pathogenic on new host plants. Although the co-cultivation of different fungal species was performed in the laboratory, it is quite possible that HCT can also occur in nature, where it could threaten crop plant protection by biological control. In cases where crops are protected against pathogenic Fusarium strains by applying nonpathogenic strains HCT from pathogenic to nonpathogenic strains could cause serious problems.

The recent discovery of HGT and HCT between asexual fungi adds another level of plasticity to fungal genomes that impacts the fungal host range. It will be interesting to determine whether HCT is a widely occurring phenomenon in fungi or whether it is only limited to particular species. Molecular diagnostic tools and the availability of many new whole genome sequences of fungi will enable to answer this question. The process of HGT and HCT is poorly understood and more research is needed to understand the underlying mechanisms.

Ancillary