Identification of protein coding genes putatively involved in infection by combining metagenomics analysis and protein orthologue clustering.

Contributors

Christine Sambles and David Studholme. University of Exeter, Devon.

Introduction

In order to identify fungal protein-coding genes associated with Fraxinus:Hymenoschyphus in planta interactions, we took an orthologue clustering approach. By identifying fungal transcripts that are present in four samples taken from infected ash and removing transcripts that are also present in the KW1 isolate could reveal some infection-related transcripts from H. pseudoalbidus. Additionally, F. excelsior transcripts present in the infected material and absent from F. excelsior with no signs of infection could identify transcripts involved in the plants response to infection by H. pseudoalbidus.

Material

Transcriptome assemblies:

F. excelsior: ATU1

C. fraxinea: KW1

Mixed material: AT1 , AT2 , Upton , Holt

Output from BLASTX searches against GenBank:

F. excelsior: ATU1

C. fraxinea: KW1

Mixed material: AT1, AT2, Upton, Holt

Methods & Results

We used MEGAN as previously described (http://oadb.tsl.ac.uk/?p=704), to assign transcripts to taxonomic bins. These transcripts came from four transcript assemblies:

* 1 H. pseudoalbidus isolate (KW1) and

* 4 mixed material (AT1, AT2, Holt & Upton).

This resulted in 36,945 transcripts being allocated to the bin for order Helotiales.

The longest open reading frame for each Helotiales-binned transcript (Table 1) was translated into a predicted protein sequence. These protein sequences were clustered using OrthoMCL.

Table 1: Numbers of transcripts and percentages of all transcripts for each sample or isolate that were binned to the order Helotiales using MEGAN.

	AT1	AT2	Holt	Upton	KW1	ATU1
Helotiales	8,214	7,403	6,930	7,410	6,561	0
% all transcripts	15.61%	8.80%	6.44%	12.25%	31.75%	0.00%

OrthoMCL analysis

Between 4,548 and 5,551 proteins were clustered from each sample; the number of protein clusters was 6,505 in total. A Venn diagram of the clustered proteins can be seen in Figure 1.

Fig 1 : Venn diagram of Helotiales-binned proteins clustered with OrthoMCL for one H. pseudoalbidus isolate (KW1) and four mixed material samples from H. pseudoalbidus infected F. excelsior (AT1, AT2, Holt and Upton).

There was a core set of 3,118 protein clusters from detectable transcripts. A set of 113 protein clusters was identified only in H. pseudoalbidus samples that were from infected F. excelsior (AT1, AT2, Holt & Upton) and 33 only identified in KW1, a H. pseudoalbidus isolate. These will be referred to as the ‘in planta’ and ‘ex planta’ groups respectively.

The 113 protein clusters found only in H. pseudoalbidus infected F. excelsior (in planta) contained a total of 565 transcripts (459 excluding isoforms). We annotated the transcript sequences based on results of BLASTX searches. Additionally the GO, EC, KEGG, PFAM and CAZy (Carbohydrate-Active enzymes) databases were used to annotate the full set of 565 transcripts.

GO, EC and KEGG annotation were inferred using annot8r (Schmid and Blaxter 2008), PFAM domains were identified with Pfam scan (a wrapper script around hmmpfam) and CAZy-family members were annotated using the CAZYmes Analysis Toolkit (CAT) ( Park, Karpinets et al. 2010).

GO analysis revealed a reduction of growth-related and an increase of cell differentiation and proliferation proteins in infected material (Fig 2).

[Fig 2](http://figshare.com/articles/Gene_Ontology_GO_analysis_of_the_the_pan_proteome_KW1_AT1_AT2_Upton_Holt_compared_to_in_planta_proteins_/988717) : Gene Ontology (GO) analysis of the the pan-proteome (KW1, AT1, AT2, Upton, Holt) compared to in planta proteins. The in planta proteins were translated from Helotiales-binned transcripts (MEGAN) and were identified only in H. pseudoalbidus samples that were from infected F. excelsior (AT1, AT2, Holt & Upton). The pan-proteome proteins were also translated from Helotiales-binned transcripts (MEGAN) and include the isolate, KW1.

PFAM and CAZy analysis of the 565 transcripts of the pan-proteome resulted in 88 PFAM domains/families and the following CAZy families:

* Glycosyl hydrolases family 18 (Pfam: Glyco_hydro_18, PF00704)

* Alcohol dehydrogenase GroES-like domain (Pfam: ADH_N, PF08240) & Zinc-binding dehydrogenase (Pfam: ADH_zinc_N, PF00107)

* alpha/beta hydrolase fold (Pfam: Abhydrolase_3, PF07859)

* Protein of unknown function, a putative transmembrane protein from bacteria. It is likely to be conserved between Mycobacterium species (Pfam: DUF2029, PF09594) & PAP2 superfamily (Pfam: PAP2_3, PF14378)

* Regulator of chromosome condensation (RCC1) repeat (Pfam: RCC1, PF00415)

* Chalcone-flavanone isomerase (Pfam: Chalcone, PF02431)

* Myosin head (motor domain) (Pfam: Myosin_head, PF00063) & Chitin synthase (Pfam: Chitin_synth_2, PF03142)RhgB_N|fn3_3|CBM-like.

BLASTX hits from the in planta transcripts included putative CFEM domain-containing protein (Marssonina brunnea) and Galactose mutarotase-like protein (Glarea lozoyensis). The Galactose mutarotase-like protein is of interest as it is also similar to rhamnogalacturonate lyase found in Aspergillus spp. and is known to degrade plant cell walls by cleaving the pectin backbone ( de Vries and Visser 2001). Some CFEM-containing proteins are proposed to have important roles in fungal pathogenesis (Kulkarni, Kelkar et al. 2003).

Comparisons of Pfam domain content among samples

PFAM domains and families in the ‘pan-proteome’ of KW1, AT1, AT2, Holt & Upton were identified using the hmmpfam wrapper script, Pfam scan. These were compared to the PFAM annotation of the ‘in planta’ group to identify over-representation of specific domains within this group. The domains and families in which >80% annotations were present in the ‘in planta’ group when compared to the ‘pan-proteome’ are shown in Table 1.

Table 1: Pfam domains and families in which >80% ‘pan-proteome’ annotations were present in the ‘in planta’ group ( http://pfam.sanger.ac.uk/).

Domain/Family	Name	Pfam accession
ATP12	ATP12 chaperone protein	PF07542
BOP1NT	BOP1NT (NUC169) domain	PF08145
iPGM_N	BPG-independent PGAM N-terminus	PF06415
CDC37_M	Cdc37 Hsp90 binding domain	PF08565
CDC37_N	Cdc37 N terminal kinase binding domain	PF03234
CDC37_C	Cdc37 C terminal domain	PF08564
Chalcone	Chalcone-flavanone isomerase	PF02431
Copper-bind	Copper binding proteins plastocyanin/azurin family	PF00127
Sdh5	Flavinator of succinate dehydrogenase	PF03937
HD_3	HD domain	PF13023
Hpt	Hpt domain	PF01627
Metalloenzyme	Metalloenzyme superfamily	PF01676
CENP-I	Mis6	PF07778
Myosin_tail_1	Myosin tail	PF01576
TRM	N2 N2-dimethylguanosine tRNA methyltransferase	PF02005
Es2	Nuclear protein Es2	PF09751
Tom37	Outer mitochondrial membrane transport complex protein	PF10568
PAP2_3	PAP2 superfamily	PF14378
PMC2NT	PMC2NT (NUC016) domain	PF08066
Porphobil_deam	Porphobilinogen deaminase dipyromethane cofactor binding domain	PF01379
Porphobil_deam(C)	Porphobilinogen deaminase C-terminal domain	PF03900
DUF2012	Protein of unknown function	PF09430
DUF775	Protein of unknown function	PF05603
Prp31_C	Prp31 C terminal domain	PF09785
Ribosomal_L32p	Ribosomal L32p protein family	PF01783

Several of the Pfam hits struck us as interesting; these are described below. The pairs of numbers in brackets are the number found within the in planta group / number found in entire ‘pan-proteome’:

Porphobil_deam and Porphobil_deamC (6/6) were found in two AT1 isoforms, AT2, two Holt isoforms and Upton. There were no peptides with this domain in the Helotiales binned KW1 proteome. Heme-biosynthetic porphobilinogen deaminase protects Aspergillus nidulans from nitrosative stress. In A. nidulans, a novel NO-tolerant (nitric oxide-tolerant) protein PBG-D (the heme biosynthesis enzyme porphobilinogen deaminase) modulates the reduction of environmental NO and nitrite by flavohemoglobin (FHB, encoded by fhbA and fhbB)) and nitrite reductase (NiR, encoded by niiA) ( Zhou, Narukami et al. 2012). NO is part of the plant hypersensitive response, a localized programmed cell death and conﬁnes pathogen to site of attempted infection (Mur, Carver et al. 2006).

Proteins matching the ‘copper binding proteins, plastocyanin/azurin’ family (Pfam: Copper-bind, PF00127) (3/3) domain were found in AT1, Holt & Upton. OrthoMCL clustered an AT2 protein with them, but the assembled transcript was incomplete at the 5’ end and the PF00127 was therefore not present. BLASTX searches indicated an amino acid sequence similarity to cupredoxin from Glarea lozoyensis and HHPred predicts similarity to cucumber stellacyanin. Due to the amino acid sequence similarity between the phytocyanins and fungal laccases, this may potentially be a laccase. White-rot fungi (e.g. Trametes cinnabarina, Trametes versicolor and Phlebia radiata) are reported to produce laccases which degrade lignin (Tuor, Winterhalter et al. 1995; Eggert, Temp et al. 1997) and laccase-mediated detoxification of phytoalexins generated by the plant defence systems has been observed in Botrytis cinerea (Pezet, Pont et al. 1991; Sbaghi, Jeandet et al. 1996; Adrian, Rajaei et al. 1998; Breuil, Jeandet et al. 1999).

The Hpt domain (Pfam: Hpt, PF01627) (5/5) was identified in two AT1 isoforms, AT2, Upton & Holt. The histidine-containing phosphotransfer (HPt) domain is a novel protein module with an active histidine residue that mediates phosphotransfer reactions in the two-component signalling systems (Catlett, Yoder et al. 2003).

Although below the threshold of 80%, 35.71% (5/14) of the CFEM domains identified in the ‘pan-proteome’ of KW1, AT1, AT2, Holt & Upton were present in the ‘in planta’ group and none were present in the ‘ex planta’ group. The CFEM domains were distributed across 4 clusters, only one of which is not present in KW1:

ClusterID : Clustered protein present in:

HELO2454 : AT1, AT2, HOLT, UPTON

HELO4337: AT1, AT2, HOLT, UPTON, KW1

HELO5213: AT1, HOLT, UPTON, KW1

HELO5952: AT2, UPTON, KW1

Fig 2: Phylogenetic tree of H. pseudoalbidus sequences from four OrthoMCL clusters where at least one sequence in the cluster contains a CFEM domain (Pfam: PF05730)). The names of full-length proteins are shown in black; in grey are names of shorter length proteins from incomplete transcript assembly that lack a CFEM domain but that cluster with CFEM domain sequences due to sequence similarity and inferred orthology. Orthologue clustering was performed on all translated transcripts binned to the Helotiales using MEGAN from the one H. pseudoalbidus isolate (KW1) and all four H. pseudoalbidus samples that were from infected F. excelsior (AT1, AT2, Holt & Upton).

The 33 clusters (representing 72 peptides) in the ex planta group which were only identified in the isolate KW1 were annotated with PFAM as previously described. This resulted in identification of 17 Pfam domains/families (Table 2).

Table 2: Pfam domains/families identified in the ex planta group

Domain/Family	Name	Pfam accession
COX1	Cytochrome C and Quinol oxidase polypeptide I	PF00115
DASH_Spc34	DASH complex subunit Spc34	PF08657
Pentapeptide_4	Pentapeptide repeats	PF13599
Vac7	Vacuolar segregation subunit 7 P	PF12751
DHQ_synthase	3-dehydroquinate synthase	PF01761
LtrA	Bacterial low temperature requirement A protein	PF06772
FSH1	Serine hydrolase	PF03959
Tyrosinase	Common central domain of tyrosinase	PF00264
Glyco_hydro_47	Glycosyl hydrolase family 47	PF01532
DUF202	Domain of unknown function	PF02656
SET	SET domain	PF00856
Abhydrolase_1	alpha/beta hydrolase fold	PF00561
adh_short_C2	Enoyl-(Acyl carrier protein) reductase	PF13561
Glyco_hydro_3	Glycosyl hydrolase family 3 N terminal domain	PF00933
ADH_zinc_N	Zinc-binding dehydrogenase	PF00107
AAA	ATPase family associated with various cellular activities	PF00004
adh_short	short chain dehydrogenase	PF00106

This low number of peptides not identified in any of the H. pseudoalbidus infected ash samples limits the ability to perform any comparative analysis.

Conclusions

Proteins putatively involved in plant-pathogen interactions have been identified from groups of translated transcripts exclusively found in planta and were not identified in isolate KW1. They included a copper binding protein within the plastocyanin/azurin family, porphobilinogen deaminase, a CFEM domain-containing protein and a Galactose mutarotase-like protein.