Yeast Two Hybrid screens were performed according to protocols outlined in the ProQuest^TM manual (Invitrogen). The Avrblb2 effectors genes were PCR-amplified using genomic DNA of P. infestans, P. parasitica and P. sojae using primers listed in Supplementary Table S2. The PCR products were cloned in the Gateway^TM pENTR/D-TOPO^® entry vector and verified by DNA sequencing. The PiAvrblb2 gene (PITG04090) without its signal peptide was cloned as Gal4-DNA-Binding-Domain (DBD) fusion in the bait pDEST32 vector using the LR clonase^TM II kit (invitrogen) for Y2H screens. This construct was transformed into MaV203 yeast cells using the PEG/LiAC method Invitrogen (2012)[ProQuest^TM manual www.invitrogen.com]

Cells expressing Gal4DBD-Avrblb2 fusion protein were selected on -Leu plates. Using an anti-Gal4-DBD-HRP antibody, correct sizes of all Gal4DBD-Avrblb2 fusion was verified using immunoblot analysis (Figure 1A). MaV203 cells carrying pDEST32-Avrlbl2 were transformed with 30 μg of a tomato bait cDNA library in the pDEST22 using the PEG/LiAC method. The Y2H cDNA library from tomato was made using the ProQuest^TM Two-Hybrid System in the Gateway^TM pDONR222 entry vector¹. It has a total of 1.04 × 10⁷ primary clones and an average > 1 Kb cDNA size. This library was transferred to the Y2H prey library destination vector (pDEST22) using the LR Clonase II^TM in vitro recombination kit (Invitrogen). MaV203 cells carrying pDEST32-Avrlbl2 and transformed with the Y2H library were plated on SC-Leu-Trp-His + 25 mM 3AT plates and positive colonies identified. DNA from several independent clones were sequenced, and putative positive clones were retransformed with pDEST32-Avrblb2 into MaV203 for verification. Three additional rounds of Y2H screens were performed.

For in vitro CaM-binding assays, all WT and mutant Avrblb2 constructs were expressed as GST fusions in the pDEST15 vector and all calmodulin constructs were expressed as 6X His fusions in pET32GW in E. coli BL21 (DE3) pLys S cells. Protein expression, isolation, in vitro Co-immunoprecipitation assays and CaM-HRP overlay assays were essentially performed as previously described (Reddy and Reddy, 2002). Briefly, Avrblb2 or calmodulins fusion proteins were induced in 50 mL E. coli BL21 (DE3) pLys S cells using 1 mM IPTG for 3 h at 30°C. Cells were pelleted and lysed in 5 mL cell lysis buffer (TBST:50 mM Tris–HCl pH 7.5, 150 mM NaCl, 0.1% Tween-20) containing 200 μg/mL lysozyme and EDTA-free complete protease inhibitors (Cat# 04693132001, Roche) on ice for 30 min. Lysate was centrifuged at 10,000 × g at 4°C for 15 min and the supernatant was filtered through 0.22 μm filters (Millipore). Protein concentrations were determined using the Bio-Rad Bradford protein assay, and were adjusted to approximately 1 mg/mL for each sample.

Co-immunoprecipitation assays were done using the Glutathion MagBeads (L00327, GenScript) according to the manufacturer’s protocol with modifications as follows. One hundred μL of Glutathione MagBeads were washed 3 × with TBST. One mL of GST-tagged Avrblb2 fusion protein extracts were added Glutathione MagBeads and incubated at 4°C for 2 h with gentle rotation. Beads were centrifuged at 200 × g for 2 min and washed five times with 1 ml TBST. 1 ml of HIS-CaM extracts in TBST containing either 2 mM CaCl₂ or 5 mM EGTA were added to the GST-Avrblb2-bound Glutathione beads and incubated at 4°C for 2 h with gentle rotation. Beads were washed five times with 1 ml TBST containing either 2 mM CaCl₂ or 5 mM EGTA. After the final wash, beads were resuspended and boiled in 100 μL SDS sample buffer (60 mM Tris–HCl pH 6.8, 2%SDS, 5% 2-mercaptoethanol, 10% glycerol, 0.005% bromophenol blue). Twenty μl were loaded per lane on 12% PAG in duplicate and run at 150 volts for 1 h in running buffer (25 mM Tris, 192 mM glycine and 0.1% SDS). Proteins were transferred to PVDF membranes in trans-blot buffer (25 mM Tris, 192 mM glycine, 20% methanol) at 100 volts for 1 h. One membrane was probed with anti-GST-HRP antibody (A00866, GenScript; 1:5000 dilution in TBST) and the other was probed with Anti-HIS-HRP antibody (#631210, clontech; 1:10,000 dilution in TBST) according to the manufacturer’s protocols.

For CaM-HRP overlay assays, different Avrblb2 proteins tagged with GST were run on duplicate 12% SDS-PAG and transblotted to PVDF membranes as described above. One membrane was probed with anti-GST-HRP antibody (A00866, GenScript; 1:5000 dilution in TBST) and the other two blots were used in CaM-HRP assays using an AtCaM2-HRP and a previously described standard protocol (Reddy and Reddy, 2002).

Deletion constructs and site-directed mutagenesis was performed using appropriate primers (Supplementary Table S1) using the In-Fusion^® HD^® Cloning Kit (cat# 638909, Clontech) according to the recommended instructions. All mutants were verified by DNA sequencing.

Using sequence-verified calmodulin and wild type and mutant Avrblb2 constructs cloned in pENTR/D-TOPO^® entry vector were used for making C-terminal YFPc-CaM3 and N-terminal YFPn-Avrblb2 constructs in plant binary BiFC vectors pGSA002-YFPc and pGSA002-YFPn, respectively. These constructs were transformed into Agrobacterium tumefaciens GV3101, and each pair of BiFC constructs were transiently expressed in N. benthamiana using the Agrobacterium-mediated transient transformation system. Confocal images were taken using the Olympus Spinning Disk (Ix81) confocal microscope.

HR response was analyzed in wild type N. benthamiana (Nb/WT) plants or N. benthamiana that were stably transformed with the Rpi-Blb2 gene (Nb/Rpi-blb2) using the Agrobacterium-mediated transient expression of Avrblb2 proteins. A. tumefaciens GV3101 carrying appropriate WT Avrblb2 or non-calmodulin-binding Avrblb2 constructs were syringe-infiltrated on at least 20 spots into fully expanded leaves of Nb/WT or Nb/Rpi-blb2, and visual HR (cell death) was recorded 2 and 15 days after inoculation. Each experiment was repeated three times.

To identify putative host targets of the P. infestans effector Avrblb2, extensive yeast two hybrid (Y2H) based screenings using mature PiAvrblb2 protein (aa 21–100) as a bait against a Solanum lycopersicum cDNA library were conducted. Expression of Gal4-DBD-Avrblb2 fusion protein in MAV203 cells was confirmed by immunoblot analyses (Figure 1A). Multiple Y2H screening resulted in the identification of several putative Avrblb2 targets (Supplementary Table S1 and Supplementary Figure S1). A subset of the putative Avrblb2-interacting proteins were calmodulins (SlCaM3 [Solyc10g077010], SlCaM4 [Solyc11g072240] and SlCaM5 [Solyc12g099990]) (Figure 1B). Due to the importance of calmodulin and calcium signaling in numerous plant microbe interactions, we conducted detailed investigations of the interaction of Avrblb2 with CAMs. Although calmodulins are highly conserved proteins in eukaryotes (Friedberg and Rhoads, 2001), there are variations at amino acid levels. Initially diverse CaMs; SlCaM3, NbCaM1, 3, 4, AtCaM2 and 6 were tested for their interaction with Avrblb2. For further Avrblb2- CaMs interaction studies, NbCaM3 (here after CaM3) and NbCaM1 (here after CaM1) were selected because these were the most divergent (19.2%) CaMs. After confirmation that Avrblb2 interacts with diverse CaMs, CaM3 was used for all other assays.

The Y2H results were verified in vivo using BiFC analyses in the Arabidopsis mesophyll protoplasts using YFPn-CaM3 and YFPc-Avrblb2 fusion proteins (Figure 1C). These analyses also revealed that Avrblb2 associates with CaM primarily at the plasma membrane with some cytoplasmic and nuclear localization (Figure 1C). The Avrblb2-CaM BiFC interactions, were also verified in N benthamiana leaves using Agrobacterium-mediated transient expression assays (Figure 1D).

CaMs bind their target proteins either in a Ca²⁺-dependent or independent manner (Kaleka et al., 2012). To determine if Avrblb2 interacts with diverse CaMs in Ca²⁺-dependent manner and to further validate the Y2H and BiFC analyses, we conducted in vitro pulldown assays using bacterially expressed GST tagged Avrblb2 and 6×His-tagged CaM1 and CaM3 using anti-GST beads (Figure 2A). These analyses showed that Avrblb2 interacted with both CaMs in the presence of Ca²⁺ but not EGTA, a Ca²⁺ chelator (Figure 2B,C). Negative controls consisting of GST-tagged GUS and a truncated version of GST-Avrblb2 (Δ56-100) did not interact with CaMs, showing that the interaction of Avrblb2 with calmodulins is specific (Figure 2B,C).

The P. infestans genome encodes at least 11 Avrblb2 homologs. Amino acid sequence comparison shows that Avrblb2 homologs are highly similar at the N-terminus but are divergent at the C-terminus (Supplementary Figure S2). BLAST search analyses of sequenced Phytophthora spp. genomes showed that P. parasitica and P. sojae also carry at least eight and two Avrblb2 homologs, respectively (Figure 3C and Supplementary Figure S2). To test if divergent Avrblb2 homologs also interact with CaM3, we first tested the interaction of the P. parasitica Avrblb2 homolog (PpAvrblb2) with CaM3. BiFC analyses in N. benthamiana revealed that the PpAvrblb2 also interacted with CaM3 at plasma membrane (Figure 1D). Pull down assay of bacterially expressed GST tagged PpAvrblb2 and 6xHis-tagged CaM3 using anti-GST beads in the presence and absence of Ca2⁺ revealed that the interaction of PpAvrblb2 with CaM3 is also Ca²⁺-dependent (Figure 2C).

Phylogenetic analyses of the C-terminal effector domain of the P. infestans Avrblb2 homologs resulted in four major groups (Figure 3C). One representative member from each of these groups, PITG20300, PITG18683, PITG04085, and PITG20303, were also tested for binding to calmodulin in a CaM-HRP overlay assay. As is shown in Figure 3A, all Avrblb2 homologs also bound calmodulin. Similarly, the P. parasitica homolog (PPTG18954) also bound calmodulin (Figure 3A). The P. sojae genome appears to harbor six hypothetical Avrblb2 genes, which code for two Avrblb2 homologous proteins². These two P. sojae Avrblb2s are highly divergent from the P. infestans and P. parasitica Avrblb2s (Supplementary Figure S2). In vitro assays, each of these two P. sojae Avrblb2 homologs also bound calmodulin (Figure 3A). Using BiFC analyses, the interaction of all these Avrblb2 homologs to calmodulin was also verified in vivo in N. benthamiana leaves (Figure 3B). Overall, these analyses suggest that calmodulin interact with divergent Avrblb2 homologs.

To identify CaM binding domain (CBD) in Avrblb2, we made a series of C- and N-terminal deletion mutants of the PiAvrblb2 protein (Figure 4C). These deletion mutants were tested in vitro for interaction with calmodulin using calmodulin binding overlay assays. These analyses revealed that deletion mutants lacking C-terminal aa 77–100 (Δ77–100) or N-terminal aa 1–87 (Δ1–87) lost calmodulin binding activity, whereas constructs longer than these two these two mutants retained calmodulin-binding activity (Figure 4A,C). These results were verified in vivo BiFC analyses (Figure 4B). To make sure that only this region is sufficient for calmodulin binding, we mutated all five amino acids (aa 78–82) in this region to alanine, which completely abolished its interaction with calmodulin (Figure 4A,B). All together, these analyses suggest that amino acids 78–82 in the C-terminal region of Avrblb2 are important for binding to CaM (Figure 4C).

CBD domains usually contain hydrophobic anchor amino acids, which occur in different configurations, usually 1–8–14, 1–5–10 in different CaM binding proteins (Tidow and Nissen, 2013). Based on this information and our results from deletion analyses, we identified amino acids at position 64, 65, 74, and 79 as potential anchor sites. To determine if any of these amino acids are responsible for interaction with CaM, we changed each one of them to alanine (A) one by one and tested them for their interaction to CaM. Calmodulin-binding overlay assays showed, that none of these amino acids affected calmodulin binding suggesting that these amino acids are not solely responsible for interaction with calmodulin (Figure 5A, right panel). Absence of interaction was not due to lack of protein expression as all these mutants were expressed very well (Figure 5A, upper panels). Since deletion analyses delineated CBD to amino acids 78 – 82, using alanine scanning, we mutated single amino acids in this region to alanine and tested their interaction with calmodulin. All mutants bound CaM (Figure 5A) but when all five amino acids from position 78 to 82 were changed to alanine the interaction was broken (Figure 4A, (NCB) marked by ^∗), suggesting that different amino combinations between amino acids 78 – 82 contribute to calmodulin binding.

The amino acid position 69 in Avrblb2 is highly polymorphic and affects its recognition by Rpi-blb2 (Oh et al., 2009). To investigate whether this position affects CaM binding with Avrblb2 we tested different variants and found that regardless of amino acid type at position 69, all Avrblb2 homologs (PITG20300 with A69, PITG18683 with I69, PITG04085 with I69, and PITG20303 with F69) bound equally well to calmodulin (Figure 5A). To further verify we changed A69 to F69 in PITG04090, which is recognized by Rpi-blb2 and which we have used as template for constructing deletion and single amino acid mutants. Calmodulin-binding assays showed that the PITG04090-A69F mutant also retained calmodulin binding (Figure 5A arrow) suggesting that aa 69-dependent Rpi-blb2 activation by Avrblb2 is independent of calmodulin binding.

To determine whether CaM binding to Avrblb2 is essential for Rpi-blb2 mediated HR or not, we used Rpi-blb2 transgenic N. benthamiana to test our non-calmodulin binding mutant of Avrblb2 (NCB-Avrblb2) with mutated CBD (78–82aa). Agrobacterium-mediated infiltration of YFP tagged NCB-Avrblb2, WT-Avrblb2 and A69F-Avrblb2 (BiFC construct with nYFP) was done in both wild type and Rpi-blb2 transgenic N. benthamiana plants and HR symptoms were followed from 1 to 15 days post inoculation (dpi). WT-Avrblb2 was used as positive control and Avrblb2-A69F, which is not recognized by Rpiblb2 (Oh et al., 2009) was used as a negative control. Protein expression of these constructs was confirmed by looking for YFP under confocal microscope at 1 dpi (Figure 6D) and through western blot by using anti-GFP-HRP antibody (Figure 6C). As expected, none of the Avrblb2 variants caused any HR symptoms on WT N. benthamiana (Figure 6A). Consistent with previous results, WT Avrblb2 (PITG04090) did cause HR on transgenic N. benthamiana expressing Rpi-blb2 gene. HR symptoms were obvious at around 45% of spots infiltrated with WT-Avrblb2 only in Rpi-blb2 containing transgenic N. benthamiana (NB/Rpiblb2) leaves at 2 dpi and almost 100% HR spots were observed at 15 dpi (Figure 6B). Mild HR symptoms started to appear as early as 1 dpi and obvious cell degradation and disrupted YFP signals were observed under confocal microscope (Figure 6D). In contrast, NCB-Avrblb2 did not cause any HR just like A69F-Avrblb2 either 1 to 5 (Figure 6A,C) or even 15 dpi (Supplementary Figure S4). Moreover, NCB-Avrblb2 expression and plasma membrane localization in both WT and transgenic plants were exactly like WT-Avrblb2 in WT plants (Figure 6C). These findings suggest that calmodulin binding to Avrblb2 does play a role in its recognition by Rpi-blb2 receptor to induce cell death.