Extended Data Fig. 2: AvrPm4 has an RNase-like N-terminus and an unstructured, highly repetitive, DNA binding MEA domain.

(A) Bgt-55142 has an RNase-like N terminus, structurally similar to previously described B. g. tritici Avr effectors, such as AvrPm2^72,73. α-helices and β-strands are depicted on the sequence of Bgt-55142, with indications whether the prediction had a low confidence (low C). TMalign⁶¹ was used to compare the effector-like part of Bgt-55142 with full length AvrPm2. The TM-alignment of the final part of both Avrs is shown here. In the alignment, a double point indicates high sequence identity and a single point low sequence identity. The intron position in the translated sequence is shown by a red triangle. Its conserved position indicates a common evolutionary origin of the RNase-like fold in both effectors. (B) The predicted aligned error (PAE) of the Alphafold prediction (shown in Fig. 1c) highlights a high quality of the prediction (low error) in the effector-like N terminus with an RNase-like fold (in pink), and a poor prediction of the MEA domain (in blue/green). (C) A dot plot (visual alignment) of AvrPm4⁹⁶²²⁴ against itself at protein level reveals multiple duplicated fragments in the second half of the MEA domain. The duplicated fragments are 17 (or 10) amino acids long, glycine and proline rich, and contain a nuclear localisation signal (NLS) predicted by NLStradamus (in yellow). (D) The domain prediction from NCBI conserved domain search reveals three overlapping hits (in blue), with the MED15 domain being the largest (aa 175-329), containing an EBNA-3B (aa 231-328) and a 104 kDa microneme/rhoptry antigen (aa 288-324) domains, with E-values of 4.23e⁻³, 2.83e⁻³ and 1.07e⁻³, respectively. The NLS (aa 243-267) is contained in both the EBNA-3B and the MED15 domain (in yellow). The first 129 amino acids have an RNase-like fold (in pink). (E) Density plot of DNA binding probability of AvrPm4 predicted by DP-bind. Binding (value = 1) and non-binding amino acid residues (value = 0) were binned in bins of 10 amino acids, and the average value of the bin is depicted in the plot. The black line indicates the majority consensus (at least 2 of the 3 DP-bind models predict the residue to bind DNA) and the blue line indicates the strict consensus (all 3 models predict the residue to be DNA binding). Raw data from DP-bind are in Supp. Table S6.