Evolution of the EKA family of powdery mildew avirulence-effector genes from the ORF 1 of a LINE retrotransposon

Abstract

Background: The Avrk1 and Avra10 avirulence (AVR) genes encode effectors that increase the pathogenicity of the fungus Blumeria graminis f.sp. hordei (Bgh), the powdery mildew pathogen, in susceptible barley plants. In resistant barley, MLK1 and MLA10 resistance proteins recognize the presence of AVRK1 and AVRA10, eliciting the hypersensitive response typical of gene for gene interactions. Avrk1 and Avra10 have more than 1350 homologues in Bgh genome, forming the EKA (Effectors homologous to Avr k 1 and Avr a 10) gene family.

Results: We tested the hypothesis that the EKA family originated from degenerate copies of Class I LINE retrotransposons by analysing the EKA family in the genome of Bgh isolate DH14 with bioinformatic tools specially developed for the analysis of Transposable Elements (TE) in genomes. The Class I LINE retrotransposon copies homologous to Avrk1 and Avra10 represent 6.5 % of the Bgh annotated genome and, among them, we identified 293 AVR/effector candidate genes. We also experimentally identified peptides that indicated the translation of several predicted proteins from EKA family members, which had higher relative abundance in haustoria than in hyphae.

Conclusions: Our analyses indicate that Avrk1 and Avra10 have evolved from part of the ORF1 gene of Class I LINE retrotransposons. The co-option of Avra10 and Avrk1 as effectors from truncated copies of retrotransposons explains the huge number of homologues in Bgh genome that could act as dynamic reservoirs from which new effector genes may evolve. These data provide further evidence for recruitment of retrotransposons in the evolution of new biological functions.

Fig. 1

Consensus model Satine, which represents a LINE retrotransposon with two ORFs. ORF1 is homologous to Avra10 (99 % nucleotide identity of Avra10 to the corresponding portion of ORF1) and contains, in the 3’ region, a sequence coding for a cysteine-rich nucleotide binding domain conserved between different TEs (NB). ORF2 is homologous to the reverse transcriptase and RNase H (RT-RH) of the retrotransposon CgT1 identified in the fungal plant parasite Glomerella cingulata. UTR: Untranslated region

Fig. 2

Genome TE copies plotted on the consensus sequences Satine (a), Satine-like Bgh_RIX_G5642 (b) and Satine-like Bgh_RIX_G5622 (c). Brown lines represent the part of the consensus sequence that aligns with each annotated copy. Copies are ordered according to their coordinates (Start, End). The black curve represents the depth of coverage along the reference sequence

Fig. 3

Kryze potential genome copy containing an ORF1 homologous to AVRk1 containing a cysteine-rich NB domain (PF00098.16_zf-CCHC_GAG) and a putative ORF2p with RT (PFAM: PF00078.20) and RH (PFAM: PF00075.17) domains. a Blgr_v3_contig_001018.fa:18231..24230 bp corresponding to Bgh_RIX_G4472 full-length copy with 3000 bp downstream region. AVRK1 alignment (in red) and domains annotated by PASTEC classifier (green) are represented. b Blue lines represent the two TE genomic copies mapped in the context of all the copies (brown lines) plotted to the reference TE consensus sequences Bgh_RIX_G4472 and Bgh_RIX_G5646

Fig. 4

Phylogenetic tree of full length and truncated sequences homologous to AVRK1 or AVRA10 calculated using Bayesian method. Circles on nodes represent posterior probability (*significant nodes with a posterior probability >75 %). Colours represent homology to either AVRK1, AVRA10 or both, and differentiate between sequences corresponding to full length ORF1 or truncated sequences (AVR/effector candidates). Red and black dots indicate proteins expressed in haustoria or/and hyphae, respectively. Yellow dots indicate proteins putatively expressed in haustoria, with one significant peptide and a peptide just below the identity score threshold

Fig. 5

Identification of sites under positive selection of AVRK1 homologues as identified under both M3 (Naive Empirical Bayes, NEB) and M8 (Bayes Empirical Bayes, BEB) models. a truncated sequences and b full-length sequences. The vertical axis represents posterior probabilities for sites with different ω ratios (dN/dS) along the sequence. Positively selected sites (ω > 1) are highlighted on top of the graph. *: p > 0.95, **:p > 0.99 (as reported in the M3 model). w0, w1 and w2 are the three ω estimated values in the M3 model

Fig. 6

Model for the evolution of the EKA family by co-option as effectors by Bgh genome. 1. Genome rearrangements after retrotransposon activity produce degenerate copies of LINE retrotransposons in the Bgh genome. Some of these copies are truncated ORF1s. 2. One truncated ORF1 is co-opted as effector E1 that enhances pathogenicity. 3. Plant host evolves a resistance gene R1 that recognizes the presence of E1. 4. E1 becomes an avirulence gene (Avr1) if it is recognized by R1 in a gene-for-gene interaction. 5. Avr1 mutates to avoid recognition by R1. 6. Bgh co-opts other degenerate copies as effectors that contribute to enhanced pathogenicity if they are not recognized by host R genes. During this process, Bgh genome and EKA copies are evolving, and the copy co-opted at point 6 may not have existed at point 2 and could have come from a different EKA ancestor