Fungi have three tetraspanin families with distinct functions

Background Tetraspanins are small membrane proteins that belong to a superfamily encompassing 33 members in human and mouse. These proteins act as organizers of membrane-signalling complexes. So far only two tetraspanin families have been identified in fungi. These are Pls1, which is required for pathogenicity of the plant pathogenic ascomycetes, Magnaporthe grisea, Botrytis cinerea and Colletotrichum lindemuthianum, and Tsp2, whose function is unknown. In this report, we describe a third family of tetraspanins (Tsp3) and a new family of tetraspanin-like proteins (Tpl1) in fungi. We also describe expression of some of these genes in M. grisea and a basidiomycete, Laccaria bicolor, and also their functional analysis in M. grisea. Results The exhaustive search for tetraspanins in fungal genomes reveals that higher fungi (basidiomycetes and ascomycetes) contain three families of tetraspanins (Pls1, Tsp2 and Tsp3) with different distribution amongst phyla. Pls1 is found in ascomycetes and basidiomycetes, whereas Tsp2 is restricted to basidiomycetes and Tsp3 to ascomycetes. A unique copy of each of PLS1 and TSP3 was found in ascomycetes in contrast to TSP2, which has several paralogs in the basidiomycetes, Coprinus cinereus and Laccaria bicolor. A tetraspanin-like family (Tpl1) was also identified in ascomycetes. Transcriptional analyses in various tissues of L. bicolor and M. grisea showed that PLS1 and TSP2 are expressed in all tissues in L. bicolor and that TSP3 and TPL1 are overexpressed in the sexual fruiting bodies (perithecia) and mycelia of M. grisea, suggesting that these genes are not pseudogenes. Phenotypic analysis of gene replacementmutants Δtsp3 and Δtpl1 of M. grisea revealed a reduction of the pathogenicity only on rice, in contrast to Δpls1 mutants, which are completely non-pathogenic on barley and rice. Conclusion A new tetraspanin family (Tsp3) and a tetraspanin-like protein family (Tpl1) have been identified in fungi. Functional analysis by gene replacement showed that these proteins, as well as Pls1, are involved in the infection process of the plant pathogenic fungus M. grisea. The next challenge will be to decipher the role(s) of tetraspanins in a range of symbiotic, saprophytic and human pathogenic fungi.


Background
Tetraspanins are a superfamily of small integral membrane proteins that were first identified in mammals as cell-specific antigens [1] and, since then, in fishes, insects, worms, sponges [2] and fungi [2,3]. They are not found in plants, although tetraspanin-like proteins were identified in Arabidopsis thaliana [4,5]. To date, at least 33 distinct tetraspanins have been identified in humans, 37 in Drosophila melanogaster and 20 in Caenorhabditis elegans [2]. Animal tetraspanins interact tightly and specifically with other membrane proteins and have been proposed to act as molecular facilitators of membrane protein complexes [1,6]. Animal tetraspanins are also able to interact with other tetraspanins to form a scaffold that promotes the recruitment of tetraspanin-associated proteins, hence creating a web [7]. As a consequence of this complexity, tetraspanin webs from different cell types differ markedly in their protein composition and probably also in their cellular function. This may explain the diversity of phenotypes observed among tetraspanin knockout mutants in animals. Indeed, tetraspanin null mutants are altered in diverse unrelated processes such as sperm-egg fusion, susceptibility to parasites or viruses, neuronal or lymphocyte cell-cell interactions, cell adhesion, motility and polarity and protein trafficking [1,6,8].
Although tetraspanins display limited amino acid sequence similarity, they share conserved structural hallmarks consisting of four transmembrane domains (TM1 to TM4), a small extracellular loop (ECL1), an intracellular loop (ICL) and a large extracellular loop (ECL2) containing a characteristic cysteine-based pattern [9,10]. In fungi, the first tetraspanin identified was in the plant pathogenic fungus M. grisea and was named Pls1 ( [3]; Figure  1). Since then, genes homologous to PLS1 have been identified in other ascomycetes [11] and basidiomycetes [2]. The functional role of Pls1 in the fungal cell has been investigated in three plant pathogenic ascomycetes [12], namely M. grisea (MgPls1, [3]), Botrytis cinerea (BcPls1, [13]) and Colletotrichum lindemuthianum (ClPls1, [14]). These fungi are responsible for major diseases of agriculturally important plants, M. grisea causing the most devastating fungal disease of rice (blast) worldwide, B. cinerea being a necrotrophic pathogen affecting more than 200 different host plants and C. lindemuthianum causing anthracnose diseases on a wide range of crops and ornamental plants. During the infection process, these fungi differentiate a specialized cell called the appressorium that is able to perforate the plant cuticle and cell wall, allowing the fungus to penetrate into host tissues. The ΔMgpls1, ΔBcpls1 and ΔClpls1 null mutants were all nonpathogenic on intact host plant tissues. Although the mutants could differentiate appressoria, these infection structures were unable to direct the penetration of the fungus into host plant tissues [3,13,14]. These results demon-strate that Pls1 tetraspanins are involved in a function conserved among fungi that is essential for appressoriummediated penetration.
Previous studies suggested that fungal genomes possess a single tetraspanin gene [11]. However, analysis of recently sequenced fungal genomes revealed that besides Pls1 orthologs, basidiomycetes have other tetraspanins called Tsp2 [2]. This prompted us to systematically search for novel tetraspanins in fungal genomes and EST databases to better characterize this protein family in fungi. Using low stringency BlastP and HMM (Hidden Markov Model) searches using the PFAM tetraspanin profile (PF00335), we identified a novel tetraspanin (Tsp3) and a tetraspanin-like protein (Tpl1) in M. grisea and their orthologs in other ascomycetes. Furthermore, genes from the TSP2 family were identified in newly available basidiomycete genomes and they displayed a complex evolutionary history with numerous paralogs. The expression profiles of PLS1, TSP2, TSP3 and TPL1 were determined by quantitative RT-PCR in M. grisea or by microarray analysisin the symbiotic L. bicolor. Expression of these genes was also estimated in silico by compiling their occurrence in EST or cDNA libraries. Finally, a functional study of TSP3 and TPL1 was performed in M. grisea, whereby deletion mutants were constructed by targeted gene replace-Schematic structure of Pls1 tetraspanins Figure 1 Schematic structure of Pls1 tetraspanins. Tetraspanins contain four transmembrane domains (TM1 to TM4) with conserved polar/charged residues (pink circles), a small extracellular loop (ECL1), a small intracellular loop (ICL) and a large extracellular loop (ECL2). ECL2 contains a CCG motif and further two conserved cysteine residues (yellow circles). The N-terminal and C-terminal tails are intracellular. These cysteine residues allow formation of two disulphide bridges (red lines), crucial for the folding of ECL2. One putative palmitoylation site is located proximal to TM4 (green circle).
ment and compared to the existing Δpls1 mutant. In marked contrast to this latter mutant, which was fully non-pathogenic on barley and rice, the Δtsp3 and Δtpl1 mutants were reduced in pathogenicity only on rice. Thus deletion of these genes has a host-specific effect on the pathogenicity of M. grisea which is different from that of the PLS1 deletion. The sequence conservation between the different Pls1 tetraspanins is essentially in the TM domains, some regions of the ECL2 including its cysteine-based pattern (CCGYx(13)-C-x(11/19)-C-x(14)-TM4) and the C-terminal tail (Figure 2, [Additional file 1] and [Additional file 2]). The ECL1 sequences are more variable, as those in basidiomycete Pls1 proteins (17 aa) are shorter than in ascomycetes (26 aa). Similarly, the conserved motif in ICL from Pls1 proteins of basidiomycetes (QRNHVTLGLV) is very different from that found in Pls1 from ascomycetes (RGWLK). Three TM domains contain charged/polar amino acids (D in TM1, S/T in TM2, T in TM4) conserved among all Pls1 proteins. Some Pls1 proteins from ascomycetes have additional charged/polar amino acids (N in TM2, S/T in TM3 and D in TM4; see Figure 1 and [Additional file 1]). In animal tetraspanins, these conserved polar/charged amino acids are thought to stabilize interactions between transmembrane domains [10]. A putative palmitoylation site (cysteine residue) is located at the end of TM4 in most Pls1 tetraspanins, except those from the Leotiomycetes, Botrytis cinerea and Sclerotinia sclerotiorum (Figures 1 and 2, [Additional file 1]). In animals, the palmitoylation of several conserved juxtamembrane cysteine residues influences the formation of tetraspaninenriched membrane microdomains [15][16][17]. A variable region is located in the ECL2 loop of fungal Pls1 tetraspanins (Figures 1 and 2, [Additional file 1]), between the last two cysteine residues whereas, it is located between the second and the last cysteine residues of the ECL2 in animal tetraspanins. Interestingly, in animals this variable region contains sites involved in specific tetraspanin protein-protein interactions [10]. The cysteinebased pattern of fungal Pls1 tetraspanins (CCGY-x(13)-Cx(11/19)-C-x(14)-TM4) contains a conserved tyrosine residue located after the CCG motif that is not in the cysteine-based pattern of animal tetraspanins (Figures 1  and 2, [Additional file 1] and [Additional file 2]). The cysteine-based pattern of ascomycete Pls1 EC2 (CCGYx(13)-C-x(10)-GC-x(14)-TM4) is slightly different from that found in Pls1 from basidiomycetes (CCGY-x(13)-Cx(19)-C-x(14)-TM4, [Additional file 1] and [Additional file 2]) suggesting their ECL2 tertiary structure slightly differs from those of ascomycetes. The amino acid sequences of the C-terminal tail of fungal Pls1 tetraspanins are strongly conserved and are rich in charged/polar amino acids (average 70%). Indeed, these C-terminal tails display a conserved RxExERF/YxxIDxK motif found in Pls1 from ascomycetes as well as from basidiomycetes [see Additional file 1]. This conservation suggests that the Cterminal tail of fungal Pls1 tetraspanins plays a functional role as does that of animal tetraspanins [18,19].

Pls1 tetraspanins are widespread in ascomycetes and basidiomycetes
Structural comparison of fungal tetraspanins (Pls1, Tsp2, Tsp3) and tetraspanin like proteins (Tpl1) Figure 2 Structural comparison of fungal tetraspanins (Pls1, Tsp2, Tsp3) and tetraspanin like proteins (Tpl1). The Pls1, Tsp2 and Tsp3 tetraspanin families display the typical structure of tetraspanin: four transmembrane domains (TM1, TM2, TM3, TM4, yellow boxes) and a small extracellular loop (ECL1, gray line), a small intracellular loop (ICL, purple line), a large extracellular loop (ECL2, orange line), intracellular Nterminal (black line) and C-terminal (brown line) tails. ECL2 contains a CCG motif and further two conserved cysteine residues (red triangles). A. The Pls1 tetraspanin family contains conserved polar/charged residues in TM1, TM2, TM3 and TM4 (blue lines) and one putative palmitoylation site is proximal to TM4 (green triangle). B. The Tsp2 tetraspanin family displays large intracellular N-terminal and C-terminal tails. One putative conserved palmitoylation site is located at the end of TM4 (green triangle). C. The Tsp3 tetraspanin family has a short ECL1 and a large C-terminal tail. D. The Tpl1 tetraspanin-like proteins lack the typical cysteine-based pattern in their ECL2 (CCG motif and further two conserved cysteine residues). Nevertheless, their ECL2 regions contain two conserved cysteine residues close to TM3 and TM4.

Tsp2 tetraspanins are restricted to basidiomycetes
Prior to this study, the first set of Tsp2 tetraspanins was identified in basidiomycetes such as Coprinus cinereus (CcTsp2A, CcTsp2B, CcTsp2C), Phanerochaete chrysosporium (PcTsp2) and Cryptococcus neoformans (CnTsp2, [2]). Tsp2 proteins display a long N-terminal tail (83 to 200 aa) and a long C-terminal cytoplasmic tail (66 to 101 aa). These domain lengths are characteristic of the Tsp2 family as Pls1 tetraspanins contain only short N-terminal and Cterminal tails (3-5 and 17-22 aa, respectively, Figure 2 and [Additional file 3]). Furthermore, the C-terminal tails of Tsp2 proteins are not as rich in charged amino acids (30%) as those of Pls1 proteins (60-75%). We identified four homologues of TSP2 (LbTSP2A, LbTSP2B, LbTS2C and LbTSP2D) in the basidiomycete L. bicolor. The corre-spondingcDNAs reconstructed from ESTs were used to define introns and start/stop codons in the corresponding genes. This family displays all the structural hallmarks of tetraspanin secondary structure with four transmembrane domains, a small extracellular loop (ECL1, 19 aa), a small intracellular loop (ICL, 4 aa), a large extracellular loop (ECL2, 72 aa) with a typical cysteine-based pattern (CCGY/F-x(12)-CY/F-x(6)-GCK-x(13)-TM4) and a conserved C-terminal cytoplasmic tail ( Figure 2, [Additional file 2] and [Additional file 3]). Three TM domains contain one or two conserved charged/polar amino acids (N and Y in TM1, S and T in TM2 and Y in TM3) as observed for animal and fungal Pls1 tetraspanins [10,11]. One putative conserved palmitoylation site (a cysteine residue at the junction between a transmembrane domain and an intracellular domain) is located at the end of TM4, proximal to the inner side of the membrane according to the predicted fungal tetraspanin topology. Additionally, LbTsp2A and PcTsp2 display another putative palmitoylation site (cysteine residue) at the start of TM1, proximal to the inner side of the membrane. We were not able to identify proteins orthologous to Tsp2 in ascomycetes using blast analysis or the tetraspanin HMM profile.

Tsp3 is a novel tetraspanin family specific to ascomycetes
We identified a novel tetraspanin in M. grisea which we named MgTsp3, using the tetraspanin HMM profile (PF00335). MgTsp3 is a modified version of MGG_13913.5 and annotation errors were corrected using the TSP3 cDNA sequence. This novel tetraspanin displays the structural hallmarks of tetraspanins including a characteristic cysteine-based pattern (CCG-x(18/22)-Cx(9)-C-x(11)-TM4) and a C-terminal tail containing a high proportion of polar/charged amino acids (70%;

Tpl1 is a tetraspanin-like protein restricted to ascomycetes
In addition to the tetraspanins described above, the M. grisea genomecontains a tetraspanin-like gene, which we have named TPL1 (MGG_08113.5). This gene was identified by use of the tetraspanin HMM profile (PF00335) not only in M. grisea but also in other ascomycetes such as C. globosum, S. nodorum, N. crassa, P. anserina and A. nidulans ( Figure 4). In marked contrast, the Tpl1 protein is absent from basidiomycete genomes. The MgTPL1 cDNA was reconstructed from ESTs and used to define introns and start/stop codons. MgTPL1 has two introns at position 40 and 262 bp from ATG [see Additional file 2]. Tpl1 proteins display some structural hallmarks of tetraspanins such as the presence of four transmembrane domains, a small ECL1 loop (14 aa), a short ICL loop (6-12 aa), a large ECL2 loop (56-68 aa) and a C-terminal cytoplasmic tail (9-17 aa) similar in size to that of Pls1 tetraspanins (Figures 2 and 4, [Additional file 2]). However, Tpl1 proteins markedly differ from Pls1, Tsp2 and Tsp3 tetraspanins in that they lack the typical cysteine basedpattern in the ECL2. Instead, they have two conserved cysteine residues close to the TM3 and TM4, respectively (Figures 2 and 4). For this reason, we have classified these proteins as tetraspanin-like.

Phylogeny of fungal tetraspanins reveals paralogs only in the Tsp2 family
Protein sequences from Pls1, Tsp2 and Tsp3 families were aligned using ClustalX 1.8 [see Additional file 4]. A phylogenetic analysis was then conducted using the PHYML software. The Tpl1 tetraspanin-like family was excluded from this alignment as these proteins are too divergent from Pls1, Tsp2 and Tsp3 proteins to be aligned correctly.
The resulting phylogenetic tree ( Figure 5) shows that Pls1, Tsp2 and Tsp3 form three distinct families. This phylogenetic tree also shows that the PLS1 genes from ascomycetes and basidiomycetes are orthologs, as the tree is congruent to the corresponding species phylogeny [20]. Similarly, TSP3 genes are orthologs ( Figure 5). The TSP2 family consists of numerous paralogs, some being recent as their closest relative gene is in the same species (Tsp2B  : -----MPGILPLVVMGLLFLGVVGAAIYEIVHATQLSLPTYP-LTIPVLLLPLLALANTLFAVRSTALRLRARGDLATAKPF Nc_Tsp3
The number of introns in PLS1 genes varies from one to three in ascomycetes and from three to four in basidio-mycetes [see Additional file 2]. One intron position is conserved in all PLS1 genes from ascomycetes (position 403, [see Additional file 2]) except in the related BcPLS1 and SsPLS1 (Leotiomycetes). This intron conservation supports the orthology found between PLS1 genes from ascomycetes. In PLS1 from basidiomycetes, three of the four intron positions are conserved [see Additional file 2] confirming a common origin for these three genes. The number of introns in genes from the TSP2 family varies from zero to four. However closely related TSP2 genes such as CcTSP2A, CcTSP2B, LbTSP2B and LbTSP2D (60% identity at the protein level) that cluster as a single clade in the phylogenetic tree ( Figure 5) share the 2 among three introns, suggesting that they result from recent duplications in both C. cinereus and L. bicolor genomes. In Sordariaceae, TSP3 genes contain three introns [see Additional file 2] except N. crassa which has four, although this may be due to an incorrect exon/intron annotation in the N. crassa gene that could not been corrected since it has no ESTs. Other TSP3 genes have three conserved intron positions (positions 31-52, 60-81 and 565-601, [see Additional file 2]) except in the related BcPLS1 and SsPLS1 (Leotiomycetes), which have an additional specific intron. This conservation of intron number and position supports the orthology between TSP3 genes. ESTs corresponding to genes from the PLS1 family were identified [see Additional file 2] in M. grisea (peritheciasexual fruiting bodies), T. reesei,B. cinerea (mycelia), S. sclerotiorium (mycelia, sclerotia, and apothecia-sexual fruiting bodies), Gibberella moniliformis (mycelia), P. anserina (perithecia-sexual fruiting bodies, germinating ascospores, mycelia),C. cinerea (mycelia, sexual fruiting bodies), P. chrysosporium (tissue undetermined) and L. bicolor (free-living mycelia andmycorrhiza). In all these fungi PLS1 genes are expressed, ruling out the possibility that they correspond to pseudogenes. Indeed PLS1 cDNAs of L. maculans, S. sclerotiorium and C. posadasii were obtained by the screening cDNA libraries. qPCR expression profiling in M. grisea revealed that MgPLS1 mRNAis expressed at similar levels in mycelia and perithecia (0.9 × reference constitutive gene ILV5, Figure 6) and in spores, appressoria and infected barley leaves (0.5 × ILV5, Figure  6). In L. bicolor, PLS1 is constitutively expressed in all tissues analyzed (mycorrhizal symbiotic tissues, mycelia and fruiting bodies, Figure 7).

Expression of fungal genes encoding tetraspanins
Phylogenetic tree of fungal Pls1, Tsp2 and Tsp3 tetraspanins The expression patterns of the different TSP2 genes in L. bicolor were determined using genome wide long oligonucleotide microarrays (Figure 7). LbTSP2-A shows a barely detectable expression level in contrast to LbTSP2-B, LbTSP2-C and LbTSP2-D, which are expressed in all tissues of L. bicolor. LbTSP2-B is highly expressed in mycorrhiza and mycelia and at a lower level in fruiting bodies. Transcripts corresponding to LbTSP2-C are mainly found in fruiting bodies, whereas LbTSP2-D is over-expressed in mycorrhiza (Figure 7). ESTs corresponding to TSP2 were found in P. chrysosporium, C. neoformans, C. cinerea (mycelia, sexual fruiting bodies) and L. bicolor (mycelia) suggesting that these genes are expressed in the corresponding species.
ESTs for TPL1 were only identified in M. grisea and P. anserina [see Additional file 2]. Quantitative PCR expression profiling in M. grisea revealed that MgTPL1 is overexpressed in perithecia and mycelia (0.4 × ILV5 and 0.3 × ILV5, respectively), but its expression is not detected in spores and appressoria ( Figure 6). quency of homologous recombination [21]. Fifteen and 11 hygromycin-resistant transformants of TSP3 and TPL1, respectively, were isolated andanalyzed by PCR for the replacement of their wild-type alleles. Deletion mutants were obtained with an efficiency of 100% (15/15) and 82% (9/11) for TSP3 and TPL1, respectively. Their mycelial growth and sporulation rates were similar to those of P1.2-Δku80::bar (reference strain). These mutants were inoculated on detached barley leaves using droplets of conidial suspensions or on barley plants by spraying.
The up-regulation of TSP3 and TPL1 in perithecia suggested involvement of these genes in mating. To address this question, the differentiation of perithecia and ascospores was examined in crosses between wild-type TH12 isolate and Δtsp3::hyg or Δtpl1::hyg mutants. A cross between the P1.2-Δku80::bar mutant (phenotypically similar to the wild-type P12 isolate) and wild-type TH12 isolate was used as control. Production of perithecia was monitored and three weeks after mating and those ascospores were observed and then allowed to germinate. No significant differences were observed in the fertility in crosses between Δtsp3::hyg, Δtpl1::hyg and wild type M. grisea strains, indicating that these genes are not essential for sexual reproduction.

Discussion
We have performed an exhaustive search for tetraspanins in fungal genomes. In addition to expanding the previously described Pls1 [11] and Tsp2 tetraspanin families Δku80::bar/Δtsp3 and Δku80::bar/Δtpl1 mutants. Seedlings were incubated for 7 days under alternate day/night at 25°C for development of disease. The deletion mutants Δku80::bar/ Δtsp3 and Δku80::bar/Δtpl1 produced significantly fewer disease lesions than P1.2-Δku80::bar (reference strain).
and a tetraspanin-like family (Tpl1). These gene families display a different evolutionary history. An ancestral tetraspanin family specific to zygomycetes (RO3G_02583 and RO3G_14268 used as roots of the tree; Figure 5) appears to be lost in higher fungi (basidiomycetes and ascomycetes). The identification of a Tsp2 tetraspanin in the zygomycete Rhizopus oryzae (RO3G_08988/RoTsp2-A and RO3G_17009/RoTsp2-B) shows that this family is ancient and predates the divergence between this phylum and higher fungi. Pls1-encoding genes were identified in almost all ascomycete and basidiomycete species. Their absence in zygomycetes suggests that the Pls1 family is more recent than the Tsp2 one ( Figure 5). The Tsp3 and Tpl1 families are restricted to ascomycetes suggesting a more recent origin than the Pls1 and Tsp2 families. The absence of tetraspanins in all hemiascomycete species suggests that these genes were lost early after the divergence of these fungi from the other ascomycetes. In the Aspergilli, the unique tetraspanin identified belongs to the Tsp3 family. Interestingly no tetraspanin genes were found in the basidiomycete, U. maydis.
The Pls1 family is highly conserved in higher fungi and is present as a single copy in each genome, defining a family of orthologous genes. In phytopathogenic fungi, Pls1 is essential for infection and is required for appressoriummediated penetration into host plant in three species (M. grisea, B. cinerea,C. lindemuthianum). The corresponding null mutants have appressoria that are unable to form functional penetration pegs that direct the penetration of the fungus into host plant [3,13,14]. These results support the hypothesis that Pls1 tetraspanins are involved in a conserved cellular function essential for appressorium function. However, the existence of Pls1 in the human ascomycete pathogen Coccidioides posadasii, in saprotrophic fungiand in symbiotic basidiomycetes such as ectomycorrhiza suggests that Pls1 plays other cellular roles than appressorium-mediated penetration of host tissues since these fungi do not differentiate appressoria. At present, these functions of Pls1 remain unknown. In fungi, functional studies are just beginning to be applied to discover such functions. Furthermore, the presence of a single copy of PLS1 in these fungal genomes offers an advantage in deciphering its cellular functions, compared to animals that have numerous tetraspanins with possible functional redundancy [6].
The Tsp2 tetraspanins are highly conserved suggesting an important functional role in most basidiomycetes although it is curious that the Ustilago maydis genome does not contain tetraspanins. Interestingly, this Tsp2 family is multigenic in L. bicolor and C. cinerea, which contrasts with other fungal tetraspanins families. Indeed, L. bicolor displays four Tsp2 paralogs and C. cinerea contains three Tsp2 paralogs. The functional significance of this multigene family is unknown. The finding in L. bicolor that LbTSP2B, LbTSP2C and LbTSP2D were expressed in all analyzed tissues suggests that corresponding proteins could be involved in the formation of a tetraspanin web, as described in animals [6]. However, differential expression profiles were observed; LbTSP2B is over-expressed in mycelia, LbTSP2C is over-expressed in fruiting bodies and LbTSP2D is over-expressed in mycorrhiza. This tissue specific expression of Tsp2 tetraspanins may explain the redundancy of this family.
Although TSP3 and TPL1 are over-expressed in perithecia, no particular phenotype was observed in perithecia when either of these genes was mutated. Perithecial production after mating and ascospore germination were identical to the wild type strain. No significant differences in the fertility were observed in crosses between Δtsp3::hyg, Δtpl1::hyg and wild type M. grisea strains. In animals, analysis of mutants with deleted tetraspanins showed no effect on development, but often had slightly altered phenotypes. For example, deletions of murine tetraspanin genes led to a mild alteration of lymphocyte proliferation and motility [6].

Microarray experiments in Laccaria bicolor
The Laccaria whole-genome expression array manufactured by NimbleGen (Madison, WI) contains in duplicates eight independent, non-identical, 60-mer probes per whole gene model. Included in the microarray are 20, 614 annotated gene models (genome sequence v1.0), 1, 680 additional predicted gene models, 30, 000 random 60mer control probes and labeling controls. A manuscript fully describing the array is in preparation. Free-living mycelium of L. bicolor S238N was grown onto cellophanecovered agar plates containing Pachlewski medium, and was grown for three weeks before harvesting the proliferating hyphal tips at the periphery. Ectomycorrhizas of L. bicolor/Douglas fir and L. bicolor/Poplar were synthesized in greenhouse experiments. In addition, ectomycorrhizas of L. bicolor/Poplar were synthesized using an in vitro system. Fruiting bodies of L. bicolor S238N were collected below Douglas fir seedlings grown in a greenhouse and inoculated using L. bicolor S238N. Tissues were immediately frozen in liquid nitrogen and RNA extraction was carried out using the RNeasy Plant Mini Kit (Qiagen). Total RNA preparations (two biological replicates for each sample) were amplified using the SMART PCR cDNA Synthesis Kit (Clontech) according to the manufacterer's instructions. Single dye labeling of samples, hybridization procedures, data acquisition, background correction and normalization were performed at the NimbleGen facility (NimbleGen Systems, Reykjavik, Iceland) following their standard protocol. Average expression levels were calculated for each gene from the independent probes on the array and were used for further analysis. The complete expression dataset is available as series GSE9784 at the Gene Expression Omnibus at NCBI [45].

Expression profiling by real-time quantitative PCR
Total RNA was extracted from M. grisea mycelia, conidia, mature appressoria (24 h), perithecia and infected barley leaves using the hot acid/phenol protocol [46]. Mycelia were derived from cultures grown in complete liquid medium. Conidia from P1.2 strain were obtained on agar rice medium. To differentiate appressoria, a conidial suspension (3.10 5 conidia/mL) supplemented with 10 μM 1,13-hexadecendiol was placed on Teflon membranes (FP301050 Goodfellow, Cambridge, U.K.) and incubated for 24 h at 26°C. Perithecia were obtained by crossing M. grisea strains P1.2 and TH12. Reverse transcriptasepolymerase chain reactions (RT-PCR) were performed with 5 μg total RNA using the ThermoScript™ RT-PCR system kit (Invitrogen, Carlsbad, CA) according to the manufacturer' s instructions. Real-time PCR experiments were performed in 96-well plates using ABI-7900 (Applied Biosystems, USA). Primer pairs were designed using Primer Express software (Applied Biosystems, Foster City, CA) with standard parameters (optimal melting point, 58°C; optimal primer length, 20 bp; amplicon length, 101 bp). cDNA dilutions of 10 -3 and 10 -2 were used respectively for cDNA obtained from fungal tissues (mycelia, conidia, appressoria, and perithecia) and infected barley leaves. Standard conditions were used with SYBR green PCR Master mix (Applied Biosystem, 95°C for 10 min and 40 cycles of 95°C for 30 s, 60°C for 1 min and 72°C for 30 s). Gene expression was calculated relative to the transcripts levels of the constitutively expressed gene ILV5 (MGG_01808.5) using the formulae 2 -ΔCt = 2 -(Ctgene-CtILV5) . The accession numbers of fungal tetraspanins are indicated in [Additional file 5] and the real-time PCR primers are listed in [Additional file 7].

Phenotypic analysis
Pathogenicity assays were performed using susceptible barley cultivar Plaisant (Hordeum vulgare L.) and the susceptible rice cultivars Maratelli, LS1, Azucena and CO-39 (Oriza sativa). Conidia harvested from 10-14 day old cultures of M. grisea growing on rice agar were inoculated on barley or rice plants. Barley seedlings were cultivated at 15°C with 60% humidity for 2-3 weeks. Segments (3 cm long) from the barley leaves were placed on water agar (1%) containing kinetin (2 mg.L-1), inoculated with 50 μL droplets of spore suspension (3 × 10 4 spores.mL -1 ) and incubated at 26°C under a photoperiod of 12 hours light. Leaf symptoms were recorded 7 days after inoculation. Rice seedlings were cultivated for 20-30 days at 70% relative humidity and 25°C day/20°C night in a growth chamber. Barley or rice (five and 160 plants per pot or tray, respectively) were sprayed with spore suspensions (20 mL of 3 × 10 4 and 2.5 × 10 4 spores.mL -1 , respectively) containing 0.5% gelatin (w/v). Inoculated rice plants were grown at 100% humidity for 24 h at 20-22°C in the dark and subsequently transferred to a greenhouse at 23°C.
Leaf symptoms were recorded 7 days after inoculation. Quantitative analysis of leaf blast infection was carried on 30 leaves in two biological replicates. Mean number of lesions per leaves were recorded. Statistical analyses were carried out with Statbox Pro 6.5 (Grimmersoft, Paris, France; Student and Mann-Whitney tests, threshold alpha/2 = 0,025). The M. grisea penetration assays into barley epidermal cells were performed as described in [3].