II. Important tools for exprimentation with Ustilago myadis
III. Cell fusion requres a complex signalling network
IV. Development of the dikaryon: the bE/bW complex at work
V. A connection between cell cycle, morphogenesis and virulence
VI. The early infection stages
VII. Proliferation and differentiaton in the plant host
VIII. The Ustilago maydis genome
The smut fungus Ustilago maydis is a ubiquitous pathogen of corn. Although of minor economical importance, U. maydis has become the most attractive model among the plant pathogenic basidiomycetes under study. This fungus undergoes a number of morphological transitions throughout its life-cycle, the most prominent being the dimorphic switch from budding to filamentous growth that is prerequisite for entry into the biotrophic phase. The morphological transition is controlled by the tetrapolar mating system. Understanding the mating system has allowed connections to signalling cascades operating during pathogenic development. Here, we will review the status and recent insights into understanding pathogenic development of U. maydis and emphasize areas and directions of future research.
The fungus Ustilago maydis is the causative agent of smut disease on corn. In recent years U. maydis has emerged as an important model for plant pathogenic basidiomycetes, a large group of pathogens that causes smut and rust diseases of plants. U. maydis is a dimorphic fungus, and the yeast like, saprophytic form can easily be propagated on artificial media. However, these so-called sporidia are unable to cause disease symptoms when applied to the host plant in pure culture. To cause disease, two compatible sporidia first need to mate. This process is regulated by a tetrapolar mating system consisting of the a and b mating type loci. The biallelic a locus (a1 and a2) encodes components of a cell/cell recognition system consisting of precursors (mfa1 and mfa2) and receptors (pra1 and pra2) of lipopeptide pheromones (for review, see Banuett, 1995). The multiallelic b locus encodes a pair of homeodomain proteins, bE and bW, that are functional only as heterodimeric transcription factors with subunits derived from different alleles (for review, see Casselton, 2002). Thus, mating compatibility is regulated by gene products of a and b loci on the pre- and postfusion level, respectively.
The filamentous dikaryon resulting from mating depends on the host plant for further development. Upon contact with the plant, characteristic infection structures, so called appressoria, are produced that allow penetration and subsequent proliferation within the infected plant organ. The host plant reacts with the formation of tumors in which the fungus proliferates and eventually produces diploid teliospores (Fig. 1). After teliospore germination, meiosis occurs and haploid sporidia are generated by successive budding from probasidia. The entire life-cycle can be completed within two to three weeks in the greenhouse under controlled conditions, allowing classical genetic segregation analysis. In this review we will concentrate on recent developments that have made U. maydis an attractive system for studying concepts and mechanisms of pathogenesis. Our goal has not been a comparison with other pathogenic fungi, since this has been done in a number of excellent reviews published recently (D'Souza & Heitman, 2001; Sanchez-Martinez & Perez-Martin, 2001; Martinez-Espinoza et al., 2002; Fraser & Heitman, 2003; Kraus & Heitman, 2003; Lee et al., 2003).
II. Important tools for experimentation with Ustilago maydis
During the last decades, a large number of molecular techniques has been developed for U. maydis which allow to study gene function in detail. The sporidia can be efficiently transformed using a variety of dominant marker genes, and integrative vectors facilitate ectopic insertion as well as insertion by homologous recombination (Wang et al., 1988; Kronstad et al., 1989). The latter makes reverse genetics possible and permits the efficient replacement of endogenous genes with modified genes (Kämper, 2004). Autonomously replicating plasmids have been developed (Tsukuda et al., 1988), and their high transformation efficiency (> 104 µg−1 DNA) can be used for efficient cloning of genes by complementation. Inducible promoter systems based on available carbon and nitrogen sources (Bottin et al., 1996; Brachmann et al., 2001) are used to follow the consequences of protein depletion and allow the study of essential genes (Straube et al., 2001). The search for genes that are required for pathogenic development has become possible through the generation of haploid strains that express an active bE/bW heterodimer. Such so-called solopathogenic strains enter the disease cycle without having to fuse with a mating partner (Bölker et al., 1995b). Mutants can be generated by insertion mutagenesis, using either restriction enzyme mediated integration (REMI) or REMI combined with enhancer trapping (Bölker et al., 1995a; Aichinger et al., 2003). However, as also observed in other systems, many of such REMI mutants contain translocations or carry untagged second site mutations, limiting the use of such mutants (R. Kahmann, unpublished). Recently, a transposon mutagenesis system based on the heterologous transposon Tc1 from Caenorhabditis elegans has been established (Ladendorf et al., 2003) and protocols for insertional mutagenesis by Agrobacterium mediated integration have been developed (J. Kämper, unpublished). Since these techniques have not yet been used in large scale mutagenesis projects, it is presently unclear whether these tools provide clean insertions and increased tagging frequencies. The cell biology of U. maydis has been advanced by the adaptation of numerous staining techniques for visualization of subcellular structures (Steinberg et al., 2001; Banuett & Herskowitz, 2002; Wedlich-Söldner et al., 2002a; Straube et al., 2003). Life imaging of cellular processes was facilitated with proteins fused to different green fluorescent protein (GFP) variants (Spellig et al., 1996; Steinberg et al., 2001; Wedlich-Söldner et al., 2002b; Straube et al., 2003). With these tools available the activity of specific genes as well as the localisation of proteins can be followed throughout the entire life-cycle (Spellig et al., 1996; Basse et al., 2000; Huber et al., 2002).
In 2003 the U. maydis community has witnessed a major breakthrough when the genome sequence compiled independently by the Broad Institute and Bayer CropScience was released to the public (http://www.broad.mit.edu/annotation/fungi/ustilago_maydis/). Based on the sequence derived by Bayer CropScience, Affymetrix DNA arrays have been designed for the parallel detection of > 6200 U maydis genes (J. Kämper & R. Kahmann, unpublished). These arrays are used to study changes in global gene expression in response to expression of the b heterodimer and activation of various signalling cascades. These studies will allow us to visualize transcriptional networks and how they change during development. The challenge will be to extend these studies to the stages where the fungus develops inside the plant tissue.
III. Cell fusion requires a complex signalling network
Fusion of compatible sporidia is the first event that is required to generate the pathogenic form. This step is controlled by mating type specific lipopeptide pheromones that are secreted and perceived by cells of opposite mating type. Upon pheromone stimulation, cells arrest budding growth and start the formation of conjugation tubes, usually at one cell pole (Snetselaar & Mims, 1992; Spellig et al., 1994). Concomitantly, cells transiently arrest their cell cycle in the G2 phase (Garcia-Muse et al., 2003), most likely to coordinate their cell cycle prior to fusion. On solid support, like the leaf surface, these mating projections show directed growth towards each other following a pheromone gradient (Snetselaar et al., 1996).
Pheromone binding to cognate 7-transmembrane receptors (Pra1/2) activates downstream signalling cascades that lead to induction of a large number of genes. Pheromone induced gene expression is mediated by pheromone response elements (PREs) present in regulatory regions of a and b mating type genes (Urban et al., 1996b). PREs are recognized by the HMG-domain transcription factor Prf1 (Fig. 2), whose activity is regulated transcriptionally as well as post-transcriptionally (Hartmann et al., 1996, 1999). On the transcriptional level an upstream activating sequence in the prf1 promoter and two PRE elements are involved (Hartmann et al., 1999). The PRE sites are most likely used for autoregulation while the UAS responds to environmental factors like carbon source availability (Hartmann et al., 1999). The latter regulation involves another signalling pathway via a novel type of mitogen-activated protein kinase (MAPK), Crk1, and requires the MAPK kinase Fuz7/Ubc5 (Fig. 2, also see below, E. Garrido et al., unpublished). The post-transcriptional regulation of Prf1 involves phosphorylation: the protein contains six putative MAPK phosphorylation sites of which the central sites are important for function (Kaffarnik et al., 2003). Recently, it has been shown that a MAPK cascade consisting of MAPKKK Kpp4/Ubc4 (Andrews et al., 2000; Müller et al., 2003b), MAPKK Fuz7/Ubc5 (Banuett & Herskowitz, 1994a; Andrews et al., 2000) and MAPK Kpp2/Ubc3 (Mayorga & Gold, 1999; Müller et al., 1999) acts upstream of Prf1 (Müller et al., 2003b, Fig. 2). In addition, Ubc2 has been identified as a novel adaptor protein that, based on its domain structure, is proposed to act directly upstream of the MAP kinase module (Mayorga & Gold, 2001, Fig. 2). The contribution of all these genes to mating is summarized in Table 1.
Table 1. The composition of signalling pathways with roles in mating, morphogenesis and pathogenicity
Besides MAPK signalling, a conserved cAMP signalling pathway is necessary for pheromone response. It consists of the heterotrimeric G protein α subunit Gpa3, the adenylate cyclase Uac1, and the cyclic AMP dependent protein kinase A composed of regulatory (Ubc1) and catalytic subunits (Adr1) (Gold et al., 1994, 1997; Regenfelder et al., 1997, Fig. 2; Dürrenberger et al., 1998). Under high cellular cAMP levels a drastic increase in mfa1 expression is observed that depends on Prf1 (Krüger et al., 1998). Recent experiments have shown that Prf1 is not only phosphorylated by the MAP kinase Kpp2/Ubc3 but also by Adr1. While PKA phosphorylation sites in Prf1 are essential for induced expression of both a and b mating type genes, MAPK phosphorylation sites are needed only for b gene expression (Kaffarnik et al., 2003). Thus it is the difference in the Prf1 phosphorylation status that dictates which of the downstream genes are expressed. It follows that Prf1 serves to integrate PKA and MAPK signalling during mating. With respect to mating it is likely that this complex scheme (Fig. 2, Table 1) is used to fine tune the expression of a and b genes.
The MAP kinase module consisting of Kpp4/Ubc4, Fuz7/Ubc5 and Kpp2/Ubc3 is necessary both for the transcriptional response to pheromone as well as for the morphological transition, i.e. conjugation tube formation (Müller et al., 2003b). Unexpectedly, however, conjugation tube formation does not require Prf1 (Müller et al., 2003b). This illustrates that the signalling pathway branches at the level of Prf1 with a separate branch affecting the morphological response (Fig. 2). How this second branch mediates changes in the cytoskeleton and the switch to polar growth is presently unclear.
Ras proteins belong to a family of small GTPases that function as molecular switches controlling a wide variety of cellular processes including signal transduction, cell polarity, the cytoskeleton, and the identity and dynamics of membranous compartments (reviewed in Takai et al., 2001). U. maydis possesses two Ras proteins which are proposed to affect the MAP kinase and the cAMP pathway, respectively (Fig. 2). A constitutively active form of Ras1 was shown to elevate mfa1 gene transcription and therefore the gene has been tentatively placed to affect cAMP signalling (Müller et al., 2003a). Originally, ras2 was isolated as a suppressor for the filamentous phenotype of an adr1 mutant, and was subsequently shown to affect mating and formation of dikaryotic filaments (Lee & Kronstad, 2002). An activated ras2 allele promotes filamentation in a Fuz7/Ubc5, Kpp2/Ubc3 dependent manner. This places ras2 in the MAPK cascade upstream of Fuz7/Ubc5 (Lee & Kronstad, 2002; Müller et al., 2003a, Fig. 2, Table 1). Interestingly, the effects of the activated allele are independent of prf1 which could indicate that the same route as the one leading to conjugation tube formation becomes activated (Lee & Kronstad, 2002, Fig. 2).
The Cdc25-like guanine nucleotide exchange factor Sql2 may be an activator of Ras2, since its overexpression induces filamentous growth that cannot be suppressed by cAMP (Müller et al., 2003a, Fig. 2, Table 1). The connection of Sql2 and Ras2 to upstream signalling components is currently not known. Another kinase affecting mating is Smu1, a homologue of Ste20 from S. cerevisiae. The deletion of smu1 reduces basal as well as induced levels of pheromone gene expression (Smith et al., 2004). Whether this occurs via the cAMP or the MAP kinase pathway has not yet been established (Fig. 2).
While the influence of cAMP signalling and MAP kinase signalling on pheromone responsive gene expression is quite clear, it remains an unresolved question how the connections between these pathways and the control of morphology are made. It is also clear that critical upstream components are still missing (Fig. 2). This holds particularly true for the upstream components between the 7-transmembrane receptor Pra1/2 that recognizes the pheromone and the MAP kinase module. So far it has not been possible to link this receptor with any of the four Gα subunits encoded in the U. maydis genome (Regenfelder et al., 1997; Krüger et al., 1998). Recent experiments place the single Gβ subunit present in U. maydis (Bpp1) in the cAMP signalling cascade (Fig. 2, Table 1) and rule out Bpp1 as an effector for the pheromone MAP kinase module (Müller et al., 2004). Similarly, the signals that activate the cAMP pathway have remained elusive so far.
IV. Development of the dikaryon: the bE/bW complex at work
Once cells have fused the fate of the resulting dikaryon depends on the products of the b mating type locus. Only cells harbouring different alleles of the b locus are capable to form a stable dikaryon (Banuett & Herskowitz, 1994b). Since for all subsequent steps in development the dikaryon is dependent on the host plant, the b locus can be considered as molecular switch for the change from saprophytic to biotrophic life style.
The bE and bW homeodomain proteins encoded by the b locus dimerize when they are derived from different alleles and it is this heterodimer that controls all subsequent steps of development (Kämper et al., 1995). The bE/bW complex binds to a conserved DNA motif termed bbs in the upstream region of b responsive genes (Romeis et al., 2000; Brachmann et al., 2001). By using differential techniques and candidate gene approaches 20 b regulated genes have been identified (Bohlmann et al., 1994; Schauwecker et al., 1995; Urban et al., 1996b; Brachmann et al., 2001, Fig. 3). Of these only three were shown to be direct targets (Romeis et al., 2000; Brachmann et al., 2001, G. Weinzierl & J. Kämper, unpublished, Fig. 3) suggesting that the majority of b controlled genes are indirectly regulated. On these grounds it has been proposed that the bE/bW heterodimer triggers a regulatory cascade with one or more regulatory genes being direct targets (Fig. 3). From the differential analysis the number of b regulated genes had been estimated to be around 140, while new analyses making use of the U. maydis DNA arrays have led to the identification of 246 b regulated genes (M. Scherer and J. Kämper, unpublished). With the bE/bW heterodimer being the central control locus for disease one might have expected many of these genes to play an important role during pathogenicity. This expectation proved wrong for the 11 b regulated genes analyzed so far with the MAP kinase encoding gene kpp6 being the only exception (Brachmann et al., 2003, Fig. 3). One explanation is that b regulated genes required for cell wall structure like the repellent Rep1 (Wösten et al., 1996) and the hydrophobin Hum2 (Bohlmann et al., 1994) or cell wall modifying enzymes like the potential exochitinase Exc1 (Brachmann et al., 2001) and the endoglucanase EG1 (Schauwecker et al., 1995) are members of gene families which could have redundant functions (Fig. 3). In addition, the complex program initiated by the b heterodimer probably needs to be subdivided into different branches (Fig. 3) that must not all relate to pathogenicity. The challenge right now is to identify the nodes in the presumed regulatory cascade as this promises insights into separate pathways and associated functions operating in b induced filaments.
V. A connection between cell cycle, morphogenesis and virulence
U. maydis switches from budding to filamentous growth after a successful mating interaction in the dikaryon. This transition critically depends on an active bE/bW heterodimer, and it has been shown convincingly that the expression of elevated levels of the bE/bW complex in haploid cells alone is sufficient for the induction of filamentous growth (Hartmann et al., 1996; Brachmann et al., 2001, Fig. 2). Specific environmental cues like low ammonium or acidic pH (Kernkamp, 1939; Ruiz-Herrera et al., 1995) can also induce filamentation, but this is independent of the presence of an active bE/bW heterodimer. Recent findings show that the methylammonium permease Ump2 is involved in filamentous growth on low ammonium medium. The ability to mate or cause disease is unaffected in ump2 mutant strains (Smith et al., 2003). It has been speculated that ump2 could influence the cellular cAMP level. All conditions that lower the cAMP content, like mutations in gpa3 or uac1, or conditions that reflect this situation (mutation of adr1, encoding the catalytic subunit of PKA) grow in the filamentous form (Gold et al., 1994; Regenfelder et al., 1997; Dürrenberger et al., 1998). Contrary to this, mutants that lead to constitutive activation of the pathway, like those deleted for the regulatory subunit of the PKA, display a multiple budding phenotype (Gold et al., 1997). It should be noted here that filamentation induced by low cAMP levels does not render haploid strains pathogenic, thus there must be a fundamental functional difference in filaments induced under these conditions and in filaments induced after a successful mating reaction. There appears to be an intimate connection between cAMP signalling and MAP kinase signalling: screens have been conducted in which the filamentous phenotype of uac1 or adr1 mutants was suppressed, and when the suppressors were isolated, they turned out to be mutant versions of all four components of the pheromone MAP kinase module, as well as of ras2 (Mayorga & Gold, 1998, Fig. 2; Mayorga & Gold, 1999; Andrews et al., 2000; Mayorga & Gold, 2001; Lee & Kronstad, 2002). For these reasons, the components of the MAP kinase module have also been designated ubc for Ustilago bypass of cyclase. It can be inferred from these data that components from the MAP kinase pathway induce filamentation, while the PKA pathway represses this morphological transition. At present, it is not entirely clear at which level these opposing effects operate. Since the filamentous phenotype of uac1 and adr1 mutants is observed in haploid cells which have not been stimulated by pheromone, it is obvious that there must be additional signals that lead to activation of the MAP kinase module. This can also be deduced from the observation that pheromones and their receptors, in contrast to the MAPK module, are not needed for filamentous growth during the biotrophic phase. On these grounds it has been argued that signals from the plant environment (Fig. 2) act as alternative inducers of this signalling pathway (Müller et al., 2003b). Presently, the nature of the signal, its perception, as well as its transmission to the MAP kinase module is unknown. As a more downstream component, the crk1 gene encoding an unusual MAP kinase, has been shown to be transcriptionally repressed via the cAMP pathway and transcriptionally activated via the MAP kinase pathway. On these grounds it has been postulated that Crk1 could serve as an integrator of environmental signals (Fig. 2). Appropriate expression levels of crk1 are required for morphogenesis as well as for cell cycle adjustments to changing environmental conditions (E. Garrido et al., unpublished). This could explain why PKA activity needs to be tightly regulated during pathogenic development (Gold et al., 1997). Somehow linked to this interplay between cell cycle, signalling and morphogenesis must be the control of growth of the dikaryotic filament. U. maydis has the peculiarity that the dikaryon can be formed on artificial media, but it is dependent on the plant for further proliferation. We expect that finding the key players for this block will prove instrumental for an understanding of biotrophy not only in U. maydis.
Because the ability to grow filamentously is prerequisite, though not sufficient on its own, to make U. maydis a successful pathogen, a lot of effort has been directed to the filamentous stage and to genes that are required for the morphological transition. Underlying all morphological transitions in fungi are changes in growth that depend on the localized delivery of membranes and enzymes to the sites of active growth. For their directed delivery to growth zones these molecules are actively transported by motor molecules that move along F-actin or microtubules. Some of these motor proteins proved essential which presently makes an in depth analysis of their function during the pathogenic phase challenging. However, the isolation of conditional mutants has already provided detailed information on their role in haploid cells as well as in dikaryotic hyphae. These studies revealed that individual motor molecules participate in numerous cellular processes to organize and shape the cell. For example, cytoplasmic dynein supports motility of the endoplasmatic reticulum (Wedlich-Söldner et al., 2002a), cooperates with a kinesin to organize the early endosomal compartment (Wedlich-Söldner et al., 2002b), moves nuclei (Straube et al., 2001) and organizes ordered microtuble arrays (Adamikova et al., 2004).
Mutations in the gene for the actin based myosin motor myo5 lead to irregularly shaped, aggregating cells that form septa, but do not separate. At permissive temperature, myo5ts cells are able to mate but are unable to develop conjugation tubes at restrictive temperature. Since pheromone secretion is not affected it has been suggested that myo5 could be involved in F-actin promoted transport of pheromone receptors (Weber et al., 2003). Infection experiments with myo5 mutants revealed a gross reduction in infection structures even at permissive temperature. The few intracellular hyphae detected were highly branched and swollen, and later infection stages were completely absent (Weber et al., 2003). It has been speculated that Myo5 could have roles in vesicle transport and polar exocytosis during tip growth and at sites of cell separation. Another nonessential motor protein, Kin2, encodes a conventional kinesin, a plus end directed microtubule motor. While kin2 mutants are largely unaffected during budding growth, a rather dramatic phenotype was observed in the dikaryon. In particular, mutants produced only short hyphae and failed to move the cytoplasm to the tip compartment, i.e they were unable to produce the empty sections typical for wild type hyphae. This phenotype was attributed to a secretion defect and the inability to enlarge basal vacuoles. The latter is thought to be prerequisite for moving the cytoplasm towards the hyphal tip and for laying down the next septum. kin2 mutants were significantly reduced in mating as well as in virulence, illustrating the critical role of this motor during the infection process (Lehmler et al., 1997; Steinberg et al., 1998). While these studies have identified several key players that affect cell shape, it remains to be established how these motor molecules mediate the morphological transitions seen during the U. maydis life-cycle. A major challenge will be to elucidate how the connections between morphology, signals and signalling pathways are made.
VI. The early infection stages
On the plant surface, the initial mating reaction and the growth of the dikaryon follows the sequence of events that has already been described in detail (see Sections III and IV). The next discrete stage is the formation of appressoria. Fibrous material occurs between the appressorium and the host cell wall which could indicate the production of some kind of adhesin (Snetselaar & Mims, 1993). So far it has not been possible to generate infection structures in vitro which may indicate the need for certain plant signals or surface cues that need to be recognized to trigger this differentiation. The appressoria formed by U. maydis are not very prominent structures but appear only as a swelling of the hyphal tips. The appressoria are not sealed off from the hyphae by a septum prior to penetration, and there is no evidence for melanization. In these respects U. maydis appressoria differ substantially from appressoria generated by Magnaporthe grisea and Colletotrichium graminicola where entry occurs by mechanical force after build-up of enormous turgor pressure (de Jong et al., 1997; Bechinger et al., 1999). Recent evidence using GFP-tagged U. maydis strains indicate that penetration has already occurred at a stage where cytoplasm is still found in the hyphal parts at the plant surface as well as in the appressorium (H. Böhnert et al., unpublished). This could suggest an entry mode primarily based on lytic enzymes. The wide penetration peg seen in EM-cross sections (Snetselaar & Mims, 1992; Snetselaar & Mims, 1993) also supports an enzyme based entry mechanism. The question that remains to be addressed is how the fungus manages to arrest and redirect its growth in a 90 degree angle at this stage and how cell wall degrading enzymes are delivered precisely to the point of entry.
Although the host cell walls are ruptured by invading hyphae, the host cell plasma membrane remains intact around the intracellular growing hyphae (Snetselaar & Mims, 1992). Mutants lacking an ER-localised alpha glucosidase (Gas1) that is presumably involved in N-glycosylation of proteins, are arrested immediately after penetration. In these mutants transfer of cytoplasm to the tip compartment appears blocked specifically in invading hyphae. gas1 mutants show characteristic cell wall alterations; in addition, changes have been observed in the interphase between hyphae and host, emphazising the importance of fungal cell wall structure in interaction with the host plant (H. Böhnert et al., unpublished). In line with this assertion it has been observed that disruption of the chitin synthase gene chs6 leads to morphological alterations which are most prominent in the dikaryon. Mutants show a severe reduction in disease symptoms presumably due to an early block before or during entry (Garcera-Teruel et al., 2004).
In addition, an active cAMP pathway appears crucial for the entry into the plant, as both Δgpa3 and Δadr1 mutations generated in solopathogenic strain backgrounds fail to produce any disease symptoms (Regenfelder et al., 1997; Dürrenberger et al., 1998). Appressoria differentiation has not specifically been addressed in these mutants, but the failure to detect anthocyanine induction indicates a block prior to entry. This is reminiscent to M. grisea and other plant pathogenic fungi where the integrity of the cAMP pathway plays a critical role for infection related development (Talbot, 2003).
While deletion mutants of the MAP kinases kpp2/ubc3 and kpp6 reduce virulence (Müller et al., 1999; Brachmann et al., 2003), mutants expressing inactive versions that cannot be phosphorylated have more dramatic effects on the ability to cause disease. kpp2/ubc3 mutants have problems to develop the filamentous stage on the leaf surface and do not develop appressoria (Müller et al., 2003b). On the other hand, kpp6 mutants develop appressoria, but these are unable to penetrate. kpp6 codes for an unusual MAP kinase that is characterized by a large N-terminal extension. The gene was originally identified among the genes up-regulated by the bE/bW heterodimer (Brachmann et al., 2003) (Fig. 3). Except for the N-terminal extension, Kpp6 is highly similar to Kpp2/Ubc3 which may explain why the two genes can partially substitute for each other and why deletion mutants give a less severe phenotype than mutants that express the inactive form. Kpp6 is thought to regulate genes required for penetration, most likely a set of hydrolytic enzymes (Brachmann et al., 2003). From their partially redundant function it has been inferred that Kpp6 may also lie in a module with Kpp4/Ubc4 and Fuz7/Ubc5 as upstream components (Müller et al., 2003b, Fig. 2). However, at present it cannot be ruled out that Kpp6 is activated through a pathway involving the other MAPKKs and MAPKKKs present in U. maydis. While this question needs to be settled, it is clear that the signalling input that leads to an activation of Kpp6 comes into play only after cells have fused and have developed the dikaryotic hyphae (Brachmann et al., 2003; Müller et al., 2003b). Elucidating the signals that lead to an activation of Kpp6 as well as finding the downstream targets of this pathway promises new insights into the function of appressorial structures.
VII. Proliferation and differentiation in the plant host
In the initial stages after penetration U. maydis follows the growth mode whereby the dikaryotic tip compartment continues rapid, unbranched growth and older compartments become devoid of cytoplasm, are sealed off and collapse. Growth in this stage is mostly intracellularly. Between 3 and 4 d post infection (p.i), hyphae start branching and are filled with cytoplasm. The trigger for this switch which is likely to involve mitotic divisions is presently unknown. The dramatic change in growth mode coincides with the beginning of tumor development around day 5 p.i. In subsequent stages, branching increases profusely and occurs at closer intervals, signalling the beginning of teliospore formation. At 9 d p.i., the fungal hyphae are surrounded by mucilaginous material, and tend to twist around each other. Fragmentation of hyphae into segments of one to several elongated cells also occurs at this time point giving a ‘sac full of worms’ appearance. Karyogamy probably occurs at this stage, and is followed by maturation of individual rounded cells to yield the diploid ornamented teliospores (Snetselaar & Mims, 1992; Snetselaar & Mims, 1993; Banuett & Herskowitz, 1994b; Snetselaar & Mims, 1994).
Several genes have been isolated by virtue of their induced expression during fungal growth within the plant tissue. The mig1 gene and the five genes of the mig2 cluster all encode secreted peptides, however, their function during the biotrophic phase remains obscure as deletion mutants do not have a discernible phenotype (Basse et al., 2000, 2002). This holds also true for another set of plant regulated genes encoded by the 24 kb p-locus (Aichinger et al., 2003), and for ssp1, a gene encoding a potential dioxygenase expressed highly in teliospores (Huber et al., 2002). Preliminary data from DNA array analyses indicate that a set of > 500 genes is plant regulated. This set includes various genes encoding potential transporters for C-compounds and amino acids, which is likely to reflect the adaptation of U. maydis to the conditions in the plant apoplast (M. Vranes and J. Kämper, unpublished). The analysis of these genes is expected to provide important insights into nutrient acquisition during the pathogenic phase.
From the analysis of mutants affected in cAMP signalling it became obvious that a regulated cAMP pathway is essential for proliferation inside the plant tissue: mutants lacking the regulatory subunit Ubc1 of PKA are able to penetrate but fail to proliferate and neither produce tumors nor spores (Gold et al., 1997). Mutants with a constitutively active gpa3 allele, which, in comparison to the ubc1 mutation, is presumed to lead to a lower activation of the cAMP pathway, also show reduced proliferation and do not produce teliospores, but still induce tumors (Krüger et al., 1998). Three other mutants, hgl1, rum1, and hda1 have a similar phenotype in that tumors are formed, but spore formation is abolished (Quadbeck-Seeger et al., 2000; Dürrenberger et al., 2001; Reichmann et al., 2002). Hgl1, a protein with unknown function, is a potential target of the cAMP signalling cascade and is hypothesized to act as a repressor for budding growth and pigment production as well as an activator for filamentous growth and teliospore formation (Dürrenberger et al., 2001). rum1 and hda1 encode proteins with similarities to the human retinoblastoma binding protein 2 and to a histone deacetyase of the Rpd3 family, respectively. Both proteins are thought to act as components of a putative corepressor complex that was proposed to function in the repression of b regulated genes in sporidia (Quadbeck-Seeger et al., 2000; Reichmann et al., 2002). We are presently entertaining the possibility that the genes regulated by this complex might be those which are normally induced during the biotrophic phase (M. Treutlein, M. Vranes, J. Kämper, unpublished). Whether there is a more direct link between these genes causing similar phenotypes when mutated, is currently unclear. The fact that tumors can be generated that contain only minute amounts of fungal material indicates that tumor formation can be initiated by only few fungal cells. This raises the intriguing question on the nature of the tumor inducing signal, how it is perceived and transmitted to induce cell division and cell expansion of infected plant tissue.
VIII. The Ustilago maydis genome
Based on pulse field gel electrophoresis, U. maydis has an estimated genome size of 20.5 Mb. The combination of sequence assemblies based on a physical map (Bayer CropScience) and a shotgun sequencing approach followed by the Broad Institute (formerly the Whitehead Institute Center for Genome Research, WICGR) has provided one of the most complete sequenced genomes to date. U. maydis strain 521 from the collection of Robin Holliday, one of the pioneers in U. maydis genetics, was sequenced in both cases. In the BayerCropScience genome project, similar to other genome projects, as a first step an ordered bacterial artificial chromosomes (BAC) library was generated. However, it was exceptional that prior to sequencing plasmid subclones of the BACs were again arranged by a high troughput mapping approach, leading to a set of 46.485 ordered plasmid clones. This mapping appproach allowed to minimize the sequencing effort and to complete the sequence with coverage of only 2.92 fold (compared with > 8 fold for other genome projects). Finally, 17.4 Mb in 28 physically linked ‘contigs’ covering the 23 chromosomes of strain 521 were generated. However, the low coverage resulted in a low sequence accuracy of c. 99.6%. The Broad assembly of strain 521 consists of 48 supercontigs covering 19.75 Mb at a more than 10 fold sequence coverage of the genome. Exelixis has provided 5 × whole-genome shotgun sequences of the U. maydis strain FB1 (Banuett & Herskowitz, 1989) that were assembled by Broad. Missing in all assemblies are parts of the ribosomal RNA gene cluster on chromosome IV which is estimated to comprise 650 kb. In addition, the presence of subtelomeric sequences (Sanchez-Alonso & Guzman, 1998) on only 16 of the 46 chromosome ends (J. Kämper, unpublished) illustrates that some parts of the genome are still missing.
For its own assembly, Broad has provided a prediction of 6522 genes; to the less complete Bayer CropScience assembly, a set of 6129 genes was assigned (J. Kämper, G. Mannhaupt, W. Mewes, unpublished). While the automatic gene prediction was less accurate in the latter case (due to the suboptimal sequence quality), an additional manual annotation was performed that integrated known proteins and EST sequences for the gene modeling. Based on the annotation of the Bayer CropScience assembly, for 30% of the predicted genes no significant similarities could be found (G. Friedrich, E. Koopmann, J. Kämper, unpublished). For a comprehensive analysis of gene functions and pathways in U. maydis, however, the community has to await a functional annotation of the more complete Broad assembly.
The comparisons of chromosomal sequences and searches for tracts of conserved gene order did not reveal evidence for large-scale genome duplications in U. maydis (D. Haase, J. Kämper, unpublished). However, as in other eukaryotic genomes, various types of repeats can be found in the U. maydis genome. Within the genome, 24 genes for the 5S rRNA are dispersed, while all genes for the 5.8S rRNA, 18S and 28S rRNA belong to the ribosomal DNA cluster on chromosome IV. A prominent tandem repeat is the heptameric sequence GATTCAC(4−7), several hundred copies are found predominantly within intergenic regions. This motif might play a role in the context of chromosome replication, which is supported by the finding that sequences resembling this consensus exist on a fragment that confers autonomous replication when present on plasmids (Tsukuda et al., 1988) in U. maydis.
The genomes of most organisms (with the exception of the ascomycetous yeasts) contain substantial amounts of various DNA transposons (active or inactive). In U. maydis only 17 partial sequences (Bayer CropScience assembly) with similarities to transposases, mainly of the TC1/mariner class were found, but in no case these sequences resemble an active DNA transposon. The only two potentially functional transposable elements identified belong to the retrotransposon class, i.e elements that transpose via a reverse transcribed RNA intermediate. TigR, related to a copia-like retrotransposon from Oryza sativa, has one copy in the U. maydis genome that could be a functional retrotransposon. In addition, three incomplete copies and 17 copies of the terminal repeats are found (J. Kämper, unpublished). The other element, HobS, shows the highest similarities to a copia-like plant retrotransposon from A. thaliana. Interestingly, each chromosome contains only one copy of HobS, either the complete element or only the direct terminal repeats. The retroelement is always located close to a region with repetitive sequences that does not code for genes (O. Ladendorf, J. Kämper, unpublished). We speculate that these regions coincide with the U. maydis centromeres. The observation that retroelements are associated with centromeric regions has also been made for example in Neurospora crassa (Borkovich et al., 2004).
In addition to these multicopy repeat families, a substantial amount of simple duplicated sequences was detected, ranging in size between several hundred bases to several kilobases. Among the larger duplications is one exact duplication of nearly 15 kb that is found both on chromosomes VII and XV.
Once the gene annotation has been completed it will be possible to do comparative genomics. It is probably naive to expect that the simple comparison of genes found in U. maydis and genes in other pathogenic and saprophytic fungi will shed light on the primary differences related to pathogenic development. However, we are convinced that the ability to combine this type of in silico analysis with genome wide expression data and the repertoire of molecular techniques available will make U. maydis one of the prime models to elucidate the key differences between saprophytic and biotrophic life styles.
Efforts during the past 15 years have made U. maydis the most tractable system among the smut fungi. At the same time these studies have led to the identification of some of the key players for mating, cell-cell signalling, morphology determination and pathogenesis. What is presently emerging is the immense complexity and crosstalk between the pathways involved. It is expected that the available genome sequence will help in finding missing links and sort out genetic redundancies. The challenge for the decade ahead will be to expand our knowledge with respect to the processes occurring once the fungus is in contact with the plant. In this area past efforts have been largely descriptive. This will need to change as we should embark on a molecular understanding of the processes involved, how they are triggered and how they bring about the morphological changes in both partners of this fascinating pathosystem.
We thank Gero Steinberg and Philip Müller for helpful comments and suggestions. Our work was supported by Bayer CropScience and the BMBF.