The complexity of carbohydrate metabolism is in part explained by the extensive assortment of carbohydrate compounds and their diverse stereochemistry. Catalysis of the biochemical reactions involving these compounds, therefore, requires a vast array of enzymes, now referred to as "carbohydrate-active enzymes" (CAZymes). Four CAZyme superfamilies of structurally related enzymes that degrade, modify, or create glycosidic bonds are known: Carbohydrate esterases (CE), Glycoside hydrolases (GH), Glycosyl transferases (GT), and Polysaccharide lyases (PL). Since most of these enzymes target carbohydrates that are part of the plant cell wall, they are also referred to as cell wall-degrading enzymes (CWDE). Analysis of the genomes of P. infestans and two other Phytophthora spp. has revealed that these organisms contain a large multiplicity of CAZymes (Table 2). Most of these enzymes are unequivocally involved in the biochemical pathways aimed at maintaining Phytophthora metabolism. The fact that a significant number of these CAZymes contain carbohydrate-binding modules, which allows them to recognize and bind carbohydrate compounds , would support this assessment. Moreover, the presence of relatively large complement of CAZy genes with predicted canonical or non-canonical secretion signals, which would enable them to effectively function as CWDE, would seem to indicate that these oomycetes also rely on enzymatic activities to successfully infect and colonize their hosts. Surprisingly, despite its greater number of CAZy homologs, P. infestans contains a smaller set of extracellular proteins than P. sojae or P. ramorum. The GH superfamily, with 115 families currently recognized based on their amino acid sequence [all of which catalyze the hydrolysis of a glycosidic bond between two or more carbohydrates or between a carbohydrate and a non-carbohydrate moiety [44–46]], is the most highly represented in the P. infestans, P. sojae, and P. ramorum genomes. Given the complexity of carbohydrate biochemistry and the broad range of hydrolytic activities it involves, is not unexpected that all genomes examined exhibit a considerable number of GH members. A number of these genes lack a cellulose-binding domain (CBD), a characteristic that was originally noted in the first GH family 5 gene cloned from a phytopathogenic fungus [47, 48]. The CBD anchors the enzyme to crystalline cellulose substrates and it has been suggested that its absence would, therefore, facilitate diffusion of the enzyme through the host cell wall . Within the GH group, the family with the largest number of members was family 28, which is comprised of enzymes with multiple, but related, functions (E.C. 3.2.1.*). These are mostly polygalacturonases (PGs) associated with the hydrolysis of galacturonic acid-based compounds, which are usually found as part of pectate and other galacturonans. PGs are believed to play a major role in the degradation of the plant cell wall by fungi through the hydrolysis of the pectin layer, which facilitates tissue invasion and maceration. Although large families of PG-coding gene have been characterized in P. cinnamomi  and P. parasitica  and individual PG-coding genes have been cloned and characterized from P. infestans , this is the first comprehensive report on this group of genes in P. infestans. PG genes are expressed in planta, and at least in the case of P. parasitica, their expression has been clearly linked to pathogenicity . GH families 3 and 95 had 20 members; giving the biochemical activities of the former (mainly β-glycosidase, which chiefly targets the hydrolysis of terminal, non-reducing β-D-glycosyl residues, releasing β-D-glucose) the number of gene copies is not unusual. In contrast, the relatively high extent of the latter family is puzzling, as its key role is the hydrolysis of fucose derivatives, which in comparison with other carbohydrates are less abundant in the plant or oomycete cell. Intriguingly, fucosterol (a sterol with a fucose moiety) is the most prominent sterol found in oomycetes capable of synthesizing sterols de novo; however, so far there is no evidence that Phytophthora is among such organisms. Therefore, the presence of this large complement of putative proteins with α-fucosidase activity remains difficult to explain. Finally, GH families 76 and 81, which are involved in the random hydrolysis of (1→6)-α-D-mannosidic linkages in unbranched (1→6)-mannans and the hydrolysis of (1→3)-β-D-glycosidic linkages in (1→3)-β-D-glucans, respectively, were also highly represented. Mannans are found in the fungal cell wall, but more importantly, are present in all lineages of land plants analyzed to date, where they are key constituents of the cell wall and play major roles as carbohydrate storage compounds and in metabolic networks devoted to other cellular processes  and β-1,3- and β-1,6-glucans compose the bulk of the oomycete cell wall but are also components of the plant cell. Hence, the presence of multiple members of these families could confer a significant evolutionary advantage to P. infestans.
Although the number of putative GT members was considerably large, this was not surprising as the transfer of sugar moieties resulting in the formation of glycosidic bonds during the biosynthesis of disaccharides, oligosaccharides and polysaccharides involves the action of hundreds of different glycosyl transferases . Members of family 41 (one of the two most highly represented GT families in P. infestans) catalyze the glycosylation of proteins at asparagine residues while members of family 71 (E.C. 2.4.1*) have, in general, α-mannosyltransferase activity (but more than 200 specific activities can be found within this enzyme class). Perhaps, the most interesting members found within the GT superfamily are the four cellulose synthase genes (GT family 2), which represent a novel class and whose function has been recently characterized in depth . These genes are required for pathogenicity as evidenced by P. infestans inability to form functional appressoria when the genes were silenced through RNA interference . Even more intriguing is the presence of a fifth member of the GT family 2 that matches very strongly (E value = 0) the putative chitin synthase gene from Magnaporthe grisea. Although no direct evidence exists favoring the presence of chitin in P. infestans cell wall, the existence of chitin synthase genes has been demonstrated in other oomycetes [54, 55], suggesting that chitin is indeed produced in these species. We have cloned this gene from both P. infestans and P. sojae and preliminary experiments indicate that the gene is expressed in cultures grown in vitro. Results from functional characterization assays will be published elsewhere.
For Phytophthora pathogenesis, however, the presence of multiple PL and CE putative members is highly significant, as both types of enzymes are involved in the degradation of cell wall components either by cleavage of the polysaccharide chains, which leads to the formation of a double bond at the resulting non-reducing end, or by catalysis of the acyl group removal from substituted saccharides, respectively . All but a very few of the PL genes found belong to families 1 and 3 (18 and 39 putative genes in P. infestans, respectively), and almost 40% of these putative genes have canonical or non-canonical secretion signals. Both of these families are involved in the degradation of the plant cell's middle lamella, either by hydrolysis of pectate (families 1 and 3) or its methyl ester, pectin (family 1). Families 4 and 10 of the CE superfamily have the largest numbers of members in P. infestans. The putative CE family 4 genes found (14) act on deacetylation of xylans, chitin, and peptidoglycans while the family 10 members (15) are esterases acting on non-carbohydrate substrates. Previous reports indicate that the cutinase gene family has undergone a notable expansion in P. sojae and P. ramorum . Members of this family (CE family 5) hydrolyze cutin, a polymer of hydroxy fatty acids that are usually C16 or C18 and contain up to three hydroxy groups. In P. infestans, the actual set of genes with true cutin hydrolase activity (E.C. 22.214.171.124) is equal to P. ramorum's (4) but relatively small compared to the number of these genes found in P. sojae (16). It is worth noting that, so far, no conclusive evidence has been found linking Phytophthora pathogenicity with cutinase activity.
Phylogenetic analysis for each superfamily suggests a very active evolutionary history characterized by constant duplications (Figures 1, 2 and 3). Two major clades, in which most PL genes are contained, can be seen in the phylogenetic tree for the PL superfamily in P. sojae and P. ramorum. An even closer relationship among the PL members can be seen in the phylogenetic analysis results obtained for P. infestans. In this species, essentially all PL genes have evolved from a single common ancestor. Genomic comparisons conducted with other oomycete genomes including P. sojae, P. ramorum, Hyaloperonospora arabidopsidis and Pythium ultimum indicate that homologs for a large number of CAZyme-coding genes exist in all oomycete species studied; however, it is plausible that the fast evolutionary pace shown by the P. infestans genome has led to the appearance of a few unique genes for which no homologs have been found elsewhere (Table 3). We used a cut-off value of 10-5 to determine homology by BLASTP; however, even when a smaller cut-ff (10-10) was used a large set of potential homologs was found (in P. infestans this equals to only 90 less sequences than with the higher cut-off). This would support the validity of the results obtained using this method.
The nature of the plant cell wall, and the fact that cell walls constitute the fundamental tier where plant-pathogen interactions take place would help explain the need for a multiplicity of CAZymes in Phytophthora. In the primary cell wall, the cellulose microfibrils and the hemicellulose [this term applies to all glycans extracted from the cell wall, to which the cellulose microfibrils are non-covalently bound, including xyloglucans (XyGs) and glucuronoarabinoxylans (GAXs)], are embedded in a pectin matrix. Both, XyGs and GAXs are composed of an extensive variety of modified and non-modified carbohydrate monomers . As the cell wall is the first barrier that must be breached in order to penetrate and successfully colonize the host, it is plausible that an abundant assortment of enzymes targeting the glycosidic bonds be produced by the pathogen. In addition, these enzymes could also be associated with the necrotrophic phase . The structure and composition of P. infestans cell wall is still ill defined; however, it is clear that cellulose is one of its major components (as opposed to chitin, which is the major component of fungal cell walls ). Therefore, the overall carbohydrate metabolism and the specific chemical activities needed for pathogenicity would help explain the vast array of genes coding for CAZymes found in these three genomes.
True synteny is loosely defined as a correspondence in the actual chromosomal locations of two gene homologs from two related species . Despite the gross and small-scale chromosomal rearrangements typical of fungal genomes, which can vary by more than an order of magnitude both within and between kingdoms , P. infestans, P. sojae, and P. ramorum appear to have retained many structural similarities in their chromosomes (Figure 4). Previously, extensive collinearity between orthologs from P. sojae and P. ramorum had been reported . Whereas it is not possible to ascertain the actual level of synteny among the three Phytophthora genomes given that individual chromosomes were not sequenced, the presence of multiple, collinear homologs in every scaffold or supercontig from each genome would indicate that, in most cases, they all share a very similar chromosomal arrangement (Figure 5). A few CE and PL genes appear to be the exceptions (Figure 6). Determining the true ortholog of a gene is a challenging task given that sequences evolve at different rates and duplications and losses are fairly common; in some cases, orthologs are 100% identical and in other cases there is no detectable sequence similarity. This makes the use of distance measuring methods insufficient to determine orthology (Jeffrey Boore, pers. comm.). For this reason, in addition to using the gene mappings between genomes, we also looked at the evolutionary trees generated by PHRINGE, which provide the actual evolutionary history of the genes and facilitate the accurate determination of orthology by analyzing gene duplications and losses. Most instances of orthology were validated by both methods; however, in several cases, potential orthologs found in the gene mappings were not validated by PHRINGE and vice versa. Even within PHRINGE, while numerical values (low seed score) could suggest orthology, the phylogenetic tree did not support such relationship. Moreover, there were cases in which seed score was high but the phylogenetic tree appeared to support orthology. Interestingly, in the GH and GT superfamilies, the number of potential orthologs in the target genomes (when present) was usually one, and only in a few situations there appeared to be more than one ortholog for a gene. In contrast, most P. infestans CE and PL genes had more than one (usually up to four) orthologs in both P. sojae and P. ramorum genomes. The overall expansion of these gene families in each species is evidenced by the number of paralogs found, especially within the GH and PL superfamilies, with the latter showing an extremely high percentage of members (81.4%) with paralogous genes (Table 3 and Figures 1, 2 and 3).
For a large number of homologs identified there is EST evidence that suggests they are expressed in vitro. However, a considerable group of gene models in all genomes still lacks any evidence of expression. Therefore, we designed specific primers targeting distinctive regions of these genes using a Clustal-based sequence alignment, as a starting point to select the most dissimilar regions, and Primer-Blast. We attempted to design primers that would span an intron, but frequently this was not possible due to the fact that many P. infestans genes do not contain introns. In addition, these primers were intended for both RT-PCR and qPCR use, but the more stringent primer design constraints of the latter technique became a limiting factor, making it difficult to design primers that would successfully work in both types of assays. Using RT-PCR, we were able to detect seven of the 16 genes targeted. Because of its greater sensitivity and the need to quantify the differences in expression rates, we evaluated the same genes using qPCR. When determining the mean threshold cycle (Tc) only trials in which individual replicate values had standard deviation (SD) < 0.5 were used; this would minimize the effect of potential pipetting errors. Hence, on several occasions multiple trials were run in order to obtain consistent values that were reliable for further analyses. Because we used equal amounts of starting material for each qPCR experiment, we were able to use the Relative Quantity (ΔCT) method of analysis, in which no modification of the data is needed to obtain normalized data. Clearly, in comparison to actin A, all CE genes were expressed at much lower levels than actin A, and the relative quantity and fold expression of the majority of genes was below 0.005. Although all genes could be detected, there were apparent differences in expression as suggested by the RFE and the ΔCT method of analysis. Two of the genes that were expressed at the highest rate (PITG_02545 and PITG_08912) belong to CE family 8, whose known activity is pectin methylesterase. The third one (PITG_03543) has pectin acetylesterase activity (family 13). Interestingly, in all genomes studied, there is a considerably high number of CE family 8 gene copies. This result matches our expectations very well as pea seeds constitute the main nutritional component of growth medium and these combined activities would be required for its utilization. The other genes analyzed have various esterase activities but may not be essential for in vitro growth. We are in the process of evaluating by qPCR the expression of all CAZyme gene models for which there is no EST evidence or that contain questionable intron size or a number of introns exceeding the usual number typically found in oomycetes.