In this paper, we have investigated the evolutionary processes which give rise to the biodiversity of the fungal class II hydrophobin genes and proteins. This class of hydrophobins has so far been reported to be restricted to Ascomycetes only. Results from this study, however, suggest that the distribution of these genes may be even more restricted, i.e. the majority of the members of this class was actually found in the Sordariomycetes, and only few were found in Leotiomycetes and Eurotiomycetes. This picture may however be biased by the fact that Sordariomycetes are overrepresented sequenced genomes, and the six genes which we retrieved from the two species of Dothidiomycetes (Passalora, Mycosphaerella) suggest that this subphylum may also be rich in class II hydrophobins. We cannot completely rule out that the low number retrieved for Leotio- and Eurotiomycetes could be due to a failure to identify these genes by BLAST search. However, our approach also identified several class I hydrophobin genes from all these fungi (data not shown), and we would therefore assume that our screening was broad enough to identify all class II genes. Also, our results are in agreement with the results of manual annotation of several fungal genomes (Aspergillus spp., M. grisea, N. crassa, G. zeae, N. haematococcae). Therefore, while it is possible that a potential HFB encoding gene has been overlooked, our data indicate that while most species contained only 1 or 2 genes (e.g. Gibberella, Nectria, Botryotinia, Aspergillus spp.), species of Trichoderma/Hypocrea clearly exceed this with their gene number (i.e. 6 genes in H. jecorina, 9 in H. virens and 10 in H. atroviridis). The reason for this remains obscure: neither the morphology of the hyphae, the conidia or of the perithecium of Hypocrea/Trichoderma show microscopic differences which may necessitate new or multiple hydrophobins to support these structures. What differentiates this fungal genus from others, however, is its mycoparasitic and necrotrophic lifestyle . While completely speculative at this moment, it is nevertheless possible that a versatile arsenal of class II hydrophobins may help the fungus to attach to the hyphae of a broad range of asco- and basidiomycetes. An amplified spectrum of genes has also been found for the chitinases of H. jecorina, which undoubtedly also aid to its mycoparasitic abilities . With the availability of the hydrophobin gene sequences now in hand for two strongly mycoparasitic species – H. atroviridis and H. virens – this work lays a strong phylogenetic foundation to investigate this possibility by means of respective knock-out strains and expression analysis.
The results from this paper show that the class II hydrophobin genes of Trichoderm/Hypocrea contain a high number of duplicated genes, and at least two cases of pseudogenes. This suggests that the class II hydrophobins evolve by a death-and-birth mechanism , a term which has been created for a process in which genes undergo gene duplications, resulting in the maintenance of some of the copies for a considerable period of time whereas other copies are rapidly lost or converted to pseudogenes. Our data render the operation of concerted evolution unlikely, because of the high sequence divergence and also by the absence of recombination at the hydrophobin loci. (in concerted evolution, member genes evolve together as a unit by mechanisms such as gene conversion or unequal crossing-over). The fact that most of the duplicated genes occupy terminal branches in phylogenetic trees and that only few obvious pseudogenes were found, indicates that the rate of evolution of the class II hydrophobins in Trichoderma is relatively fast. This rapid evolution, and equally rapid loss of some genes is also reflected in the findings that the clades leading to the hfb genes in Trichoderma seldom contain members of other fungi, and their evolution thus took place after formation of the genus Trichoderma.
In addition, the numbers of synonymous differences of nucleotide sequences between genes from the same species are very large and frequently close to the saturation level. This high level of synonymous differences further supports the claim of a birth-and-death evolution at the DNA level, and supports the long time persistence of these genes in the genome. On the other hand, genes from different species (e.g. H. atroviridis and H. virens) but belonging to the same phylogenetic clade are highly similar (cf. Fig. 3). Such a long-term conservation of amino acid sequence is best explained by strong purifying selection. Interestingly, and in contrast to Rajashekar et al. , we found only a few individual cases where the K
s ratio was >1 and would reveal a history of accelerated evolution. If such a period of accelerated evolution occurred, most of the gene duplicates from this time apparently have not been maintained and the Trichoderma/Hypocrea hydrophobin genes characterized in this study are therefore mostly of recent origin.
The present study also expands our knowledge on the structure of class II hydrophobins. While most of them are small, compact proteins, which contain little other structures than the four beta-sheets and the single helix [5, 6], we have detected several proteins which display a long extended N-terminus (ENT). With respect to class II HFBs, such structures have so far only been found in H. jecorina HFB6 , and in the pseudohydrophobin QID3 . Interestingly, an ENT was recently also identified in the class I HFB Hum3 from Ustilago mayidis . Lora et al.  hypothesized that the ENT of QID3 mediates cell wall binding because it resembles a module which is also present in plant bimolecular proteins [47–49]. Interestingly, our work reveals that there are at least two types of these ENTs: a major one, typified by H. virens HV_21a, H. atroviridis HA_2c, H. lixii QID3, and also in P. fulva HCF6, and in the spacers between the hydrophobin units in the multipartite genes of C. paspali and C. fusii, which are characterized by a conserved repeat of glycine and asparagines; and second type, shown by e.g. H. atroviridis HA_6a, H. virens HV_1d, and H. jecorina HFB6, in which the repeated motif is replaced by several PG/PD repeats, a P-rich stretch or a D-rich stretch, respectively. These proteins did not cluster together, indicating that these proteins do not show a common ancestry of the cysteine-containing core domain. Among the proteins with this terminus, one (HV_13a) is intriguing as its ENT is very short, which gives rise to the speculation that this extended N-terminus may arise by segment duplication. Support for this hypothesis would also come from the multipartite hydrophobins found in Claviceps spp. [50, 51], wherein paralogous hydrophobin gene copies are connected by P, G and N-rich loops, and which may have been trapped in the stage of gene duplications at the extended N-terminus before recombining individual copies into new loci. It is thereby intriguing to observe the similarity of the nucleotide sequence of the "GN" repeat (GGTAAT) to that found to act as a recombination hot-spot in Penicillium chrysogenum (TGTAA [A/T]; ). Therefore, the occurrence of the Claviceps multipartite hydrophobins would be due to multiplication of some of the class II hydrophobins by tandem duplication [53, 54], for which these sequences could act as recombination targets.
Nevertheless, it may still be likely that these ENTs are not only evolutionary artefacts: the [GN] repeats are reminiscent of S. cerevisiae Ure2p, a regulator of nitrogen catabolism, which can become transformed into a prion form by polymerization into filaments . These filaments have an amyloid fibril backbone formed by an N-rich sequence which form a parallel superpleated beta-structure. The prion domain is thereby divided into nine seven-residue segments, each with a four-residue strand and a three-residue turn, that zig-zag in a planar serpentine arrangement, the interior of the filament being stabilized by H-bond networks generated by the stacking of N side chains. Interestingly, hydrophobins themselves are known to form amyloid-like structures [56–58], and we consider it therefore possible that the ENTs form defined structures which additionally contribute to the structural rigidity of the hydrophobins.
During phylogenetic analysis of the amino acid sequence, most hydrophobins from Trichoderma/Hypocrea did not group into strongly supported clades. However, a few exceptions were noted, notably the clades containing H. jecorina HFB1, HFB2 and HFB4, respectively. Clade "HFB4" is intriguing as its members – in contrast to HFB1 and HFB2  were not expressed in submerged culture but only found in surface cultivation. This clade may thus contain hydrophobins relevant for hyphal growth. Unfortunately, the differences in aa-sequence with that of the other Trichoderma hydrophobins do not provide a clear clue as to the understanding of its function. One notable change is the substitution of the phenylalanine residue in the middle of the single helix, which is otherwise conserved in the class II hydrophobins from all other fungi (with the exception of A. niger and V. lecanii which contain an L and M, respectively), by a leucine, which may give rise to weaker hydrophobic interaction within the protein (hydrophobicity index F = 2.24; L = 1.99). However, Linder et al.  speculated that the aromatic ring of F39 is inserted between two Pro-residues (P11 and P50) from the two β hairpin structures into the protein and may serve to stabilize the fold through hydrogen bonds. Interestingly, members of the "HFB 4 clade" consistently have P11 replaced by an A which may hydrophobically interact with this L. In addition, the first beta-strand contains a conserved motif of two asparagines which provide it with a positive charge. Hydropathy plots show that members of the "HFB4 clade" have almost no hydrophobicity in the area between aa20 and aa40, and their helix is in contrast positively charged. While the consequence of these changes on the structure and function is however unclear our phylogenetic analysis sets the stage for future functional studies that may include transcript and gene deletion analysis.