Genome-wide annotation, comparison and functional genomics of carbohydrate-active enzymes in legumes infecting Fusarium oxysporum formae speciales

ABSTRACT Fusarium wilt caused by soil borne ascomycetes fungi Fusarium oxysporum which has host-specific forms known as formae speciales (ff. spp.), apparently requires plant cell wall degrading enzymes (PCWDE) for successful invasion. In this study, 12 F. oxysporum ff. spp. were taken for genome-wide annotation and comparative analysis of CAZymes, with an assessment of secretory PCWDE and orthologues identification in the three legumes infecting ff. spp. Further, transcriptomic analysis in two legumes infecting ff. spp. using publically available data was also done. The comparative studies showed Glycoside hydrolase (GH) families to be abundant and Principle Component Analysis (PCA) formed two distinct clusters of ff. spp. based on the CAZymes modules and families. Nearly half of the CAZymes in the legumes infecting ff. spp. coded for signal peptides. The orthologue clusters of secretory CAZymes common in all the three legume infecting ff. spp. mostly belonged to families of AA9, GH28, CE5 and PL1 and the expression analysis revealed the abundant PCWDE were differentially expressed in these legumes infecting ff. spp. Therefore, this study gave an insight into the distribution of CAZymes especially extracellular PCWDE in legumes infecting ff. spp. with further shedding light onto some of the key PCWDE families through differential expression analysis.


Introduction
Most of the fungal pathogens of ascomycetes are necrotrophic including Fusarium oxysporum, a soilborne wilt causing pathogen which affects the host plant mainly through root by colonising the treachery elements and subsequently results in wilting of aerial part, yellowing of leaves, necrosis and death (Leslie and Summerell 2006;Nene et al. 2012). This fungal pathogen exists in host-specific forms known as formae speciales (ff. spp.) singular formae specialis (f. sp.) which are also believed to have evolved into many races and within these formae speciales, F. oxysporum f. sp. lycopersici with its host, i.e. tomato, is considered as model system for molecular studies of resistance and susceptibility of plant (Takken and Rep 2010;Essarioui et al. 2016). Leguminous plants including chickpea, pea and alfalfa are of high nutritional and economic values where the former two are considered as second and third largest important legume crops worldwide (FAO: www.fao.org) and are prone to be affected by f. sp.
ciceris (Haware et al. 1986), f. sp. pisi (Hepple 1963) and f. sp. medicaginis (Rhodes 2015) respectively. These formae speciales exhibit different races on the basis of their virulence or differential reaction to host cultivars (Correll 1991); for instance f. sp. pisi has four races (race 1,2,5 and 6) where race 1 and 2 have been reported frequently in almost every country (Infantino et al. 2006) and for f. sp. ciceris, eight races have been reported worldwide, i.e. race 0, 1A, 1B/C, 2, 3, 4, 5 and 6 (Del Mar Jiménez-Gasco et al. 2001), where race 1 is highly pathogenic in India (Singh et al. 2010) and devastating, if unchecked. Although no races have been observed for the f. sp. medicaginis, still it is used as a model legume pathosystem (Williams et al. 2016) due to Medicago being model species in the fabaceae family.
All the phytopathogens synthesise Plant Cell Wall Degrading Enzymes (PCWDE) (Blackman et al. 2014;Kubicek et al. 2014) and are identified as the Carbohydrate-active Enzymes (CAZymes) (Lombard et al. 2014). The Carbohydrate-active Enzymes (CAZymes) present as different modules deposited in CAZy database (www.cazy.org) are the classes of enzymes required for the synthesis, modification and degradation of polysaccharides and are further classified as glycoside hydrolase (GH) (Henrissat 1991;Henrissat and Bairoch 1993;Henrissat and Davies 1997), glycosyltransferase (GT) (Campbell et al. 1997;Coutinho et al. 2003), Carbohydrate esterase (CE) (Lombard et al. 2010), polysaccharide lyase (PL) (Garron and Cygler 2010;Lombard et al. 2010), auxillary activities (AA) (Levasseur et al. 2013) and Carbohydrate-binding modules (CBM) (Boraston et al. 2004). Currently, glycoside hydrolase (GH) consisting of 161 families, plays a role in hydrolysing glycosidic bonds between two or more carbohydrates or between a carbohydrate and a non-carbohydrate component while glycosyltransferase (GT) with 106 families, helps in the formation of glycosidic bond and biosynthesis of di-, oligo-and polysaccharides whereas Carbohydrate esterase (CE) containing 16 families, catalyzes and splits the ester into respective acid and alcohol. Polysaccharide lyase (PL) consisting of 36 enzyme families cleaves the polysaccharide containing uronic acid via β-elimination mechanism. Meanwhile, enzymes identified to show Auxiliary activities (AA) are distributed to16 families and are involved in redox reaction in conjunction with other CAZymes, where 9 families of it are involved in lignolytic activity and 7 families in lytic polysaccharide mono-oxygenase (LPMO) activities. Some of the CAZymes function as Carbohydrate-Binding Modules (CBM) which have a carbohydrate-binding activity and are reported to constitute approximately 84 families. Thus, these CAZymes modules are required for both growth (Caracuel et al. 2005) and pathogenesis (Jorge et al. 2006;Gibson et al. 2011) while the families of GH, CE, PL and AA involved in cellulose, hemicelluloses, pectin and lignin degradation are reported to be crucial for invasion and successful pathogenesis (Kubicek et al. 2014;Berlemont 2017;Sista Kameshwar and Qin 2017). Besides these, some families of AA also assist GH for degradation of polysaccharides and are known as Lytic Polysaccharide Mono-oxygenase (LPMO) which are also crucial in plant cell wall degradation (Agger et al. 2014;Beeson et al. 2015;Courtade et al. 2016).
Therefore, the PCWDE produced by ascomycetes phytopathogens (Glass et al. 2013;Rytioja et al. 2014) are revealed to be secretory in nature and involved in pathogenicity to establish their own survival (Gibson et al. 2011;Zhao et al. 2013;Kubicek et al. 2014).
Perhaps, these PCWDE help in depolymerisation of plant cell wall components composed of cellulose, hemicellulose and pectin (Kubicek et al. 2014) when the pathogens penetrate and invade and also to sustain in the host tissue (Gibson et al. 2011).
Though the genomic sequence for the abovementioned legumes infecting ff. spp. have been accomplished by Williams et al. (2016) who further revealed GH3 and GH43 alone to be representative of dispensable scaffolds in these legumes infecting ff. spp. and in the meanwhile, Joint Genome Institute (JGI) Mycocosm database (genome.jgi.doe.gov/mycocosm/home) (Grigoriev et al. 2014) provides CAZyomes information only on F. oxysporum f. sp. lycopersici and f. sp. pisi till date. This necessitates the exploration of these CAZymes based on genome-wide distribution of plant cell wall degrading enzymes in different F.oxysporum ff. spp. Thus, the present study was focused on 12 F. oxysporum ff. spp. subjected to annotation and comparative study of CAZymes distributed among their genome with further exploration of PCWDE and secretory CAZymes in the three legumes infecting ff. spp. (f. sp. ciceris, f. sp. medicaginis and f. sp. pisi). However, comparative expressional studies using the publically available RNA-seq data based on different nutritional medium for legumes infecting ff. spp with respect to the f. sp. lycopersici were performed to assess the expressional changes of CAZymes. Accordingly, extracellular secretory enzymes having a role in plant cell wall degradation were assessed to shed the light on the distribution and identification of common orthologues among the legumes infecting ff. spp. as well the key CAZymes which showed significant differential expression in two of the legumes infecting formae speciales. To understand the effect of nutrition on CAZymes expression, RNA-seq data for F. oxysporum f. sp. lycopersici grown in minimal medium at 28ºC with three replicates (PRJNA450835), F. oxysporum f. sp. medicaginis grown in potato dextrose broth (PDB) at 22ºC with three replicates (PRJNA294248) and F. oxysporum f. sp. pisi grown in both rich and minimal medium pooled with one replicate (PRJNA538191) were taken from EBI ENA database (https://www.ebi.ac.uk/ena).

Gene prediction, CAZymes annotation and data analysis
Since the gene and protein information for f. sp. ciceris and f. sp. medicaginis were not available on NCBI database, genes and their corresponding proteins were predicted using webAUGUSTUS (Hoff and Stanke 2013) gene prediction server (http://bioinf.unigreifswald.de/webaugustus/) which has a web-based interface by keeping Fusarium oxysporum f. sp. lycopersici reference sequences data as a training dataset.
The protein sequences from 12 ff. spp. were used for CAZYmes annotation on dbCAN2 ) (http://cys.bios.niu.edu/dbCAN2/) which has three integrated automated annotation tools namely -HMMER (Finn et al. 2011) to search against the dbCAN HMM (hidden Markov model) database, DIAMOND (Buchfink et al. 2015) to search against CAZy pre-annotated CAZymes sequence database and HotPep (Busk et al. 2017) to search against the conserved CAZymes short peptide database with their default parameters and suggests that CAZymes annotated from at least two or more tools should be considered for higher confidence in the prediction. Thereafter, hierarchical clustering of the CAZymes families copy numbers was done for these 12 ff. spp. using Cluster3 (de Hoon et al. 2004) and visualised through Java Treeview software. Principal component analysis of these ff. spp. based on CAZymes modules and the copy numbers of each CAZymes families were also done using an online tool ClustVis (Metsalu and Vilo 2015) (https://biit.cs.ut.ee/clustvis/) with unit variance scaling applied to the row and singular value decomposition (SVD) imputation for principal components (PC). PC1 and PC2 were considered for explained variation and represented as 2D plot. Also, the copy number variations in PCWDE were manually assessed for three legume infecting formae speciales.

Identification of secretory CAZymes
All the predicted CAZymes in F. oxysporum f. sp. ciceris affecting chickpea and other two legumes infecting ff.

Orthologues identification
Orthologous studies of the protein sequences identified as extracellular soluble CAZymes in three of the legumes infecting ff. spp. were done using Orthovenn (Wang et al. 2015) (http://www.bioinfogenome.net/ OrthoVenn/), which is a web-based tool for annotation and prediction of orthologues in the organisms by keeping the default parameters. The results were generated as cluster Venn diagram with list of orthologues and their GO annotations.

Differential expression analysis
To find out the expressional changes in CAZYmes among legumes infecting ff. spp. and their comparison with the abundant CAZYmes annotated in genomic assemblies, three publically available RNA-seq data (as mentioned in data retrieval section) were taken and subjected to differential gene expression analysis by comparing two legumes infecting ff. spp against f. sp. lycopersici using Galaxy suite (https://usegalaxy.org/) (Giardine et al. 2005), where initial FastQC (http://www. bioinformatics.babraham.ac.uk/projects/fastqc/) was done followed by HISAT2 (Kim et al. 2015) for reads alignment, FeatureCount (Liao et al. 2014) for reads quantification and DESEQ2 (Love et al. 2014) for differential expression analysis. The differentially expressed genes (DEGs) were filtered using FDR cut-off <0.05 and logFC ≥ |2| and after comparison of DEGs in f. sp. medicaginis vs. f. sp. lycopersici and f. sp. pisi vs. f. sp. lycopersici, the common accessions were annotated for CAZymes and identified for the presence of secretory CAZymes within these two legumes infecting ff. spp.

Gene prediction
Genome and protein sequences of all the 12 F. oxysporum ff. spp. were retrieved from NCBI database and summed up in details containing strain, family, genome size, host and total protein content (Supplementary Table S1). In this study, a total of 16,414 genes were predicted in the f. sp. ciceris genomic assembly while 15,741 genes were predicted in f. sp. medicaginis genomic assembly and all these predicted sequence information can be accessed through figshare (10.6084/m9. figshare.8124359). Protein sequences of the predicted genes from f. sp. ciceris and f. sp. medicaginis along with other 10 ff. spp. were taken further for CAZymes annotation.

CAZymes annotation and distribution
The overall distribution of CAZymes in each of the ff. spp. generally coded for approximately 3% of the total protein-coding sequences as already evidenced by Davies et al. (2005), according to whom in most of the prokaryotes and eukaryotes, the percentage of CAZymes ranged between 1% and 3% of the total protein-coding genes.
The sequences of CAZymes distributed among all the 12 ff. spp., revealed the highest of 835 and lowest of 573 CAZymes in f. sp. pisi and f. sp. ciceris respectively by considering at least two of the prediction tools of dbCAN2. CAZymes module CBM was feebly represented and encoded for just 0.5-2.2% in all the ff. spp., while GH was leading to the highest (52%) of the total CAZymes followed by GT, AA, CE, PL and CBM ( Figure 1). The copy numbers of GH were found to be highest (432) in f. sp. melonis and lowest (301) in f. sp. ciceris. Likewise, comparison of the CAZymes distribution in all ff. spp. also showed f. sp. pisi to outperform the others. For instances, the highest GT was observed in f. sp. pisi (195) while the lowest was in f. sp. ciceris (105). Similarly, modules AA (106) and CBM (19) were also highest in f. sp. pisi and the other modules such as PL (29) and CE (49) were observed to be higher in f. sp. conglutinans and f. sp. vasinfectum respectively (Supplementary Table S2).
The families within the modules were sorted and accordingly, module GH represented 79 different families where 12 of them were associated with CBM, while GT, AA, CE, PL and CBM each represented 31, 13, 9, 6 and 3 families, respectively. Within the GH module among all the ff. spp., f. sp. pisi (49), f. sp. raphani (40) and f. sp. melonis (34) over-represented the CAZymes families with their copy number ranging between 27 and 49 for GH3, 23 and 40 for GH16 and 24 and 34 for GH18, respectively. In the meanwhile, GH5 exhibited an average copy number of 22 in all the ff. spp. GH13 was higher in f. sp. melonis (18), GH32 in f. sp. lycopersici (13) and GH31 with an average copy number of 9 in all the ff. spp. Out of the total CAZymes coding CBM along with GH families, only GH78 associated with CBM67 exhibited higher copy number ranging from 8 to 14 among all the ff. spp.
In GT module, GT2 and GT1 over-represented with copy numbers ranging from 22 to 48 and 12 to 22 and high in f. sp. pisi (48) and f. sp. melonis (22), respectively, whereas GT4 and GT8 that ranged between 6-19 and 7-14 was revealed to be high in f. sp. conglutinans (19) and f. sp. lycopersici (14) respectively. In module AA, the family AA3 was ranging highest (25-36) followed by AA1 (14-21) and AA9 (13-20) each represented high copy number within f. sp. pisi (36), f. sp. radicislycopersici (21) and f. sp. vasinfectum (20), respectively. In CE module, CE4 and CE5 were found to be higher in copy number with an average of 9 in all these ff. spp. In PL module, PL1 with copy number ranging from 10 to 14 was highest with f. sp. cepae (14) having the highest copy number followed by PL3 with an average copy number of 7 in all ff. spp. Unlike the others, CBM21 of CBM module exhibited the highest copy number of 8 in two of the ff. spp. including f. sp. conglutinans and f. sp. pisi (Supplementary Table S3).
These formae speciales are often considered as polyphyletic due to horizontal gene transfer (van Dam et al. 2018) which make them difficult to identify. Although variations observed in accessory genome also known as conditionally dispensable chromosomes (CDC) probably harbours effector genes responsible for host specificity (van Dam et al. 2016;Williams et al. 2016) despite the fact, the core genomes of many of the formae speciales, roughly of the same size (van Dam et al. 2016) consist majority of CAZymes. Hence, from the above evidences, it could be suggested that the CAZymes modules annotated in genomic assemblies of the ff. spp. might have a correlation established with the host specificity which may contribute to the closeness of FOC and FOMED but surprisingly, keeping FOP segregated even though it belongs to legumes infecting ff. sp. Also, FOC was further separated out from FOMED based on CAZymes families and this is also evidenced by (Williams et al. 2016), that these legumes infecting ff. spp. were rather sharing conserved candidate effector proteins and not large genomic region of pathogenicity-related content.
Interestingly, FORC affecting a range of cucurbits shows rot symptoms (Vakalounakis 1996;Vakalounakis et al. 2004;Vakalounakis and Fragkiadakis 1999) and since the symptom is similar to FOCEP (Entwistle 1990), both were clustered together. Therefore, based on the CAZymes distribution, these two legumes infecting ff. spp. (i.e. FOC and FOMED) were much closer to rot causing as well as monocot affecting ff. spp. than FOP which was clustered with other ff. spp. affecting host from the families of malvaceae, brassicaceae, solanaceae and cucurbitaceae.
Based on the information retrieved from the CAZy database, we limited our scope to families of GH, AA, CE, PL that are involved in plant cell wall degradation while focussing intensively on the secretory CAZYmes from the legumes infecting formae speciales and subsequent identification of secretory PCWDE.

Distribution of Plant Cell Wall Degrading Enzymes (PCWDE) in legumes infecting formae speciales
Since these pathogens are necrotrophic in nature (Vajna 1985), the high copy number of PCWD CAZYmes encoded in the genome of the presently investigated organisms could be attributed to their mode of nutrition which is relatively recognised to be inevitable for their synthesis (Zhao et al. 2013). Thus, the overall distribution of the CAZymes modules constituting GH, AA, CE, and PL by FOC, FOMED and FOP accounted for 384, 449 and 514, respectively. Since these three ff. spp. correspond to infect legumes, their ability to degrade plant cell wall were analysed by sorting the CAZymes manually which are evidenced to be involved in catabolic processes like cellulose degradation, hemicellulose degradation, pectin degradation, lignin degradation and LPMO activities (Supplementary Table S4).

Pectin degrading
Here the families of CAZymes having pectin degrading ability constituted 22 families in these legumes infecting ff. spp. which mostly belonged to arabinase, rhamnogalacturonase, mannanase and galactase (Choi et al. 2013;Rytioja et al. 2014) constituting GH, CE and PL families, respectively. GH 43, an endo-α-1, 5-L-arabinanase was found to be higher in copy number ranging from 25 to 34. Subsequently, GH28 (poly galactouranase), GH47 (α-mannosidase) and GH76 (αmannanase) with copy numbers varying from 9 to 13 were found in these legumes infecting ff. spp. PL1, a pectate lyase was higher in all of them and ranged between 11 and 12. In GH having CBM domain, GH78 (α-rhamnosidase) with CBM67 (L-rhamnose binding activity) was having copy number ranging from 8 to 14. The GH having single copy number in each of them was found to be GH38 (α-mannosidase). Whereas only carbohydrate esterase CE8 (pectin methylesterase) and CE12 (pectin acetyl esterase) families were present with copy number of four in each of them.
Here the PCWDE families identified in the three legumes infecting ff. spp. were also reported in the studies done by (Chang et al. 2016;Ferreira Filho et al. 2017;Sista Kameshwar and Qin 2017). The families representative of GH94, GH130, PL10, which were supposed to be present in ascomycetes (Zhao et al. 2013) were found to be absent in these ff. spp. and our study also led to the observation of some more CAZymes families to be present in these ff. spp. involved in the process of cell-wall degradation. The families of GH which are degrading cellulose and hemicellulose identified with high copy numbers in these ff. spp. were also found to be higher in other species of ascomycetes belonging to both dothideomycetes and sordariomycetes (Zhao et al. 2013;Morales-Cruz et al. 2015;Paolinelli-Alfonso et al. 2016).
Accordingly, CAZymes families that were abundantly present (GH16, GH43, CE3, CE4 and CE5) evidenced their involvement in pathogenesis of Fusarium sp. (Kikot et al. 2009) and PL1 seemed to significantly contribute to virulence in pathogenesis (Kikot et al. 2009) Although both GH28 and GH47 are found in moderate copy numbers but might play a critical role in pectin degradation and degradation of both the hemicellulose and pectin, respectively (Zhao et al. 2013).
The high copy numbers of AA1, AA3 and AA9 in all these legumes infecting ff. spp. corresponded to lignin degradation and LPMO activities and perhaps the association of AA3 with AA9 was suggested to be involved in the stimulation of the lignocelluloses degradation (Langston et al. 2011). Also, the abundant copy number of CAZYmes families known to have a role in metabolic processes like cellulose, hemicellulose and pectin degradation in these ff. spp. as evidenced in F. oxysporum species by (Sista Kameshwar and Qin 2017) are known to deconstruct the plant cell wall. Hence, the analyses on plant cell wall degrading ability of these legumes infecting ff. spp. led us further to focus on secretory CAZymes. Therefore, these CAZymes belonging to modules GH, AA, CE and PL were assessed for their secretory nature.

Identification and distribution of secretory CAZymes
All the pathogenic fungi produce a diverse array of secretory CAZymes, which are extracellular in nature and prominently contributes to phytopathogenic activity in the necrotrophic organisms irrespective of the fungal families. Hence, the analysis of the present study revealed nearly half of the CAZymes constituting GH, CE, PL and AA to encode for signal peptides. The signalP and DeepLoc analysis enabled not only to identify CAZymes encoding for signal peptides but also their corresponding subcellular locations in all the three legumes infecting ff. spp. Comparing the CAZymes coding for signal peptide revealed FOC and FOMED to represent approximately 52% while it was 45% in FOP. Further, subcellular localisations of these signal peptide coding CAZymes differentiated as extracellular and soluble were predicted to be high in FOC (84%) whereas around 50% were recognised in the other two legumes infecting ff. spp. (Table 1). Thus, subsequent embodiment of the modules representing the secretory CAZymes implies the families of AA9, CE5, GH5, GH16 and PL1 to be abundantly represented and exhibiting moderate representation of families AA1, AA5, CE3, CE4, CE16, GH3, GH18, GH28, GH43 and PL3 in all the three legumes infecting ff. spp. However, there was an overrepresentation of the families AA3, AA9, CE5, GH3, GH16, GH18, GH28, GH43 and PL4 in FOC (Table 2).

Orthologues study of secretory CAZymes
The three legumes infecting ff. spp. formed 187 orthologous clusters with 49 clusters containing sequences from at least two of the ff. spp. and 121 clusters containing sequences from all the three ff. spp. (Supplementary Figure S1) in which 112 clusters contained single copy sequences from each organism.
The predicted secretory PCWDE protein sequences of FOC were involved in 162 clusters, FOP in 160 clusters and FOMED in 156 clusters with the number of sequences which did not form any clusters, i.e. singletons were 56, 10 and 8, respectively. Further analyses of the 121 clusters containing sequences from all the three ff. spp. revealed 90 clusters with GO annotations which were majorly involved in carbohydrate metabolic processes having hydrolase, lyase, carbohydrate and pectin binding activities (Supplementary Table S5). Out of these 90 clusters, 74 clusters were found to be extracellular according to assigned GO annotation process of which 12 clusters belonged to AA, 9 to CE, 16 to PL and 37 to GH and within the GH clusters, 4 clusters of GH were associated with CBM. Highest number of CAZYmes families involved in clusters were AA9 (6), CE5 (5), GH28 (6) and PL1 (10) (Table 3). Hence, these CAZYmes orthologues which were having maximum clusters might be suggested to play a role in survival and cell-wall degradation process as explored through evidences. Accordingly, AA9 family previously known as GH61 (Levasseur et al. 2013) present in high frequency of orthologue clusters and multiple copies in genome (Harris et al. 2010), assists GH and other families like PL and CE in the breakdown of cellulose (Monclaro and Filho 2017) whereas GH28 observed to play a significant role in pectin degradation is reported to involve in the process of plant cell wall degradation (García-Maceira et al., 2001;Sprockett et al. 2011). Moreover, the existence of CE5 corresponding to cutinase or acetyl xylan esterase (Kolattukudy 1981) and PL1, a pectate lyase (Herron et al. 2000), reported to be secreted by plant pathogens perhaps suggest to help in the process of cleaving cutin and pectin where both are crucial with one as protective covering and other as providing physical strength to the plants. Therefore, these abundant orthologue clusters are indeed crucial for the plant cell wall degradation process, identified in these legumes infecting ff. spp.

Differential gene expression analysis of secretory CAZymes
The differential expression analysis after FDR cut-off of <0.05 and LogFC of ≥ |2|, resulted in 5483 significant differentially expressed genes (DEGs) for FOMED vs. FOL and 5599 for FOP vs. FOL where 3525 DEG accessions were identified to be common in both the studies through Venn diagram (Supplementary Figure S2). These 3525 common DEGs when subjected to CAZymes annotation resulted in 109 CAZymes which were both down-and up-regulated. In order to identify   Table S6), from which 56 were observed to code for secretory peptides, out of which 53 were identified as extracellular CAZymes belonging to the families of AA, GH, CE and PL while the other 3 were found to be localised in cell membrane and lysozyme (Supplementary S7). The pattern of expression in both the legumes infecting ff. spp. were almost similar except for CE2, CE3 and one accession coding for AA1 which showed contrasting expression pattern in both the formae speciales. Most of the secretory CAZymes categorised as PCWDE were significantly down regulated with exceptional accessions coding for the same CAZymes families such as GH16, GH43, and PL1 which were found to be both up-and down-regulated in these ff. spp. and the contrasting difference might have been contributed by the nutritional media in which they were grown. The accessions coding for families like CE16 (acetyl esterase), GH28 (polygalacturonase) and GH78 (α-L-rhamnosidase) were found to be up regulated in both the ff. spp. and are involved in hemicellulose and pectin degradation (Figure 3).
The observation on the majority of secretory CAZymes annotated revealed GH5, GH28, AA9 and PL1 to be abundantly present in the genomic assemblies as well as in orthologue clusters in the three of the legumes infecting formae speciales and were also found to be differentially expressed. The observation conceived from the secretory CAZymes significantly expressed are in accordance with the study done by  who revealed a transcription factor was controlling the expression of genes coding for GH3, GH5, GH28, AA3, AA9 and PL1 in Verticillium dahlia, a soil-borne wilt causing pathogen of ascomycetes, although the families, GH5, an endoglucanase was having a mixed expression and GH28, a polygalactouranase was significantly up regulated in these two legumes infecting ff. spp. involving in degradation of cellulose and pectin, respectively. This could be further substantiated by a study done on Aspergillus niger (Daly et al. 2017), where genes coding for pectin lyase induced by galacturonic acid and endoglucanase were found to be significantly up-regulated when grown on wheat straw. In general, the secretory CAZymes identified in higher and moderate copy numbers were found to be differentially expressed in these formae speciales and interestingly, the accessions coding for pectin degrading CAZYmes were up-regulated implicating their importance in growth and pathogenesis. Hence, the family GH28 with other less abundant pectin degrading families are found to be up-regulated and could possibly play role in pathogenesis.

Conclusion
Genome-wide distribution of CAZymes families among the ff. spp. revealed the variations in the copy number of the families and also the abundance of some PCWDE in the legumes infecting ff. spp. Thus, the details generated through the present study significantly contributed to the distribution pattern of these important PCWDE families and led us to the identification and annotation of some of the key families which are involved in cell-wall degradation process among the legumes infecting ff. spp. PCWDE and their orthologues especially present in single copy in these legumes infecting ff. spp. and also differential expression analysis shed light on some of the secretory CAZymes with their expressional changes. Therefore, the important PCWDE orthologues identified through this study in these ff. spp. which were also abundantly present in the genomic assemblies and also found to be significantly expressed may further enable us to choose these candidate CAZymes to understand their co-ordination, mechanistic role and explore them as targets to attenuate these phytopathogens.