Teaching an old dog new tricks: Suppressing activation of specific mitogen-activated kinases as a potential virulence function of the bacterial AvrRpt2 effector protein

ABSTRACT

AvrRpt2 is one of the first Pseudomonas syringae effector proteins demonstrated to be delivered into host cells. It suppresses plant immunity by modulating auxin signaling and cleavage of the membrane-localized defense regulator RIN4. We recently uncovered a novel potential virulence function of AvrRpt2, where it specifically blocked activation of mitogen-activated protein kinases, MPK4 and MPK11, but not of MPK3 and MPK6. Putative AvrRpt2 homologs from different phytopathogens and plant-associated bacteria showed distinct activities with respect to MPK4/11 activation suppression and RIN4 cleavage. Apart from differences in sequence similarity, 3 of the analyzed homologs were apparently “truncated.” To examine the role of the AvrRpt2 N-terminus, we modeled the structures of these AvrRpt2 homologs and performed deletion and domain swap experiments. Our results strengthen the finding that RIN4 cleavage is irrelevant for the ability to suppress defense-related MPK4/11 activation and indicate that full protease activity or cleavage specificity is affected by the N-terminus.

Plants sense pathogens through highly conserved molecular microbial features (so-called pathogen-associated molecular patterns, PAMPs, e.g. bacterial flagellin).¹ This enables plants to react to potential infection by initiating a defense signaling program that leads to PAMP-triggered immunity (PTI). On the cellular level, early PTI signaling events include ion fluxes, an oxidative burst and activation of mitogen-activated protein kinase (MAPK/MPK) cascades.² MPK cascades consist of 3 hierarchically organized kinases, a MAP triple kinase that activates (phosphorylates) an MPK kinase (MKK), which then phosphorylates an MPK. MPKs, in turn, phosphorylate a multitude of substrate proteins to modify their function by changing protein activity, stability, localization or protein-protein-interactions.³ In Arabidopsis, 2 main branches of MPK cascades were described in pathogen defense: one involving MKK4/5-MPK3/6 was shown to positively regulate defense-related gene expression and production of defense metabolites,⁴ while the other branch (engaging the module MEKK1-MKK1/2-MPK4) was thought to negatively regulate defense.⁵ However, microarray profiling experiments painted a more complex picture of defense regulation with both positive and negative functions from these branches of MPK cascades.⁶ MPK11, the closest homolog to MPK4, is a fourth PAMP-activated MPK.^7,8 This was confirmed in an independent report, where MPK1 and MPK13 were also identified as additional PAMP-activated MPKs.⁹ Nevertheless, analysis using mpk mutants suggests that MPK3, MPK4, MPK6 and MPK11 are the major MPKs activated by PAMPs.^7,8

We recently reported that the Pseudomonas syringae effector protein AvrRpt2 (PstAvrRpt2) specifically suppresses PAMP-induced activation of MPK4 and MPK11, but not of MPK3 and MPK6. This presumably represents a novel virulence function of AvrRpt2 to manipulate and fine-tune MPK signaling.¹⁰ AvrRpt2 is a Cys-protease, which upon activation and self-cleavage within plant cells, targets the plant defense regulator RIN4 and Aux/IAA transcription factors through direct cleavage and promoting protein-turnover, respectively.^11-14 While the protease activity of AvrRpt2 is required to suppress MPK4/11 activation, MPK4/11 are not directly cleaved.¹⁰ We also showed that AvrRpt2 functions may be more widespread in bacteria-host interactions than previously anticipated. Several putative AvrRpt2 homologs from phytopathogens (e.g., Ralstonia solanacearum, Erwinia amylovora, Acidovorax citrulli) and plant-associated bacteria could block PAMP-induced MPK4/11 activation when transiently expressed in Arabidopsis mesophyll protoplasts. The homologs showing the suppressive effect on PAMP-induced MPK4/11 activation also cleaved RIN4, except for a homolog from Burkholderia pyrrocinia (a soil bacterium), which suppressed PAMP-induced MPK4/11 activation but did not cleave RIN4. Two of the tested homologs, one from Acidovorax avenae subsp. avenae (a pathogen of monocots, e.g. oat) and Collimonas fungivorans (a soil fungi pathogen), neither suppressed MPK4/11 activation nor cleaved RIN4. Overall, we concluded that the MPK4/11 suppression function is independent of RIN4 cleavage.¹⁰

Of all the analyzed homologs, the putative B. pyrrocinia, A. avenae and C. fungivorans sequences were “truncated” when compared to the PstAvrRpt2 pre-pro-protein (see Fig. 1A). For brevity, we will designate these as BpAvrRpt2, AaAvrRpt2 or CfAvrRpt2, respectively. In particular, the ORF of CfAvrRpt2 begins 2 amino acids upstream of the first catalytic triad amino acid (i.e. corresponding to C122 of PstAvrRpt2). As there might be genome mis-annotation or possibly presence of ORF-disrupting insertion elements, we cannot exclude that, besides overall sequence divergence, the differential activities of the shorter homologs might not be caused by the “missing” N-terminus (Fig. 1A). To address this possibility, we created (i) a PstAvrRpt2 N-terminus deletion mutant (Δ1–119) that is analogous to the shortest CfAvrRpt2 variant, or (ii) replaced the N-termini of the 3 “truncated” homologs with the 120-amino-acid N-terminal region of PstAvrRpt2 (designated as Pst-Aa, Pst-Bp or Pst-Cf, respectively; see Fig. 1). These variants were expressed in protoplasts and assayed for PAMP-induced MPK4/11 activation and RIN4 cleavage. We also computed models to predict the impact of the N-terminus on the protein structure, where PstAvrRpt2 is modeled as containing 2 sub-domains (see Fig. 1 legend for details).

Figure 1.

Comparison of primary sequences and structural models of AvrRpt2 from Pseudomonas syringae pv. tomato and 3 “truncated” putative homologs. (A) Boxshade representation of the multiple sequence alignment performed with CLUSTAL O(1.2.1) (Boxshade: http://www.ch.embnet.org/software/BOX_form.html; Clustal Omega: http://www.ebi.ac.uk/Tools/msa/clustalo/; ¹⁸). Red boxes mark the residues forming the ligand pocket predicted in C below (Fpocket program ¹⁹) of the native PstAvrRpt2. (Abbreviations: Aa = Acidovorax avenae subsp avenae; Bp = Burkholderia pyrrocinia; Cf = Collimonas fungivorans; Pst= Pseudomonas syringae pv tomato). (B) Structural models based on the crystal structure of the peptidase domain of Streptococcus mutans comA protein (template: c3k8uA). Models were generated with Phyre2 ²⁰ (>99 % confidence scores) and presented in rainbow colors (blue N-terminus to red C-terminus). Most of the disordered N-terminus (marked with “?” in A) is omitted in the PstAvrRpt2 model (i.e. covering H109-A245). Note the residual N-terminal H109-M120 region (i.e., till the fusion site for the truncated variants) is predicted as a random coil between the grove of the N-terminal α1–3 subdomain and C-terminal αβ subdomain. The first residue included in the model of each truncated homologs is indicated. Yellow arrows (upper panels) mark the major structural differences compared to PstAvrRpt2 and white arrowheads (lower panels) indicate any changes upon replacement with the PstAvrRpt2 N-terminus. (C) Models as in B but with the ligand binding pocket highlighted in red (see corresponding red boxes in A).

Like the catalytic triad mutant (H208A), the truncated Pst-Δ1–119 mutant (i.e., the equivalent of the CfAvrRpt2) failed to suppress MPK4/11 activation and did not cleave RIN4 (Fig. 2). The PstAvrRpt2 N-terminus is thus necessary for both effects, presumably because Pst-Δ1–119 is no longer an active protease. Conversely, fusion of the PstAvrRpt2 N-terminus did not confer both of these activities to the “truncated” AaAvrRpt2 or CfAvrRpt2 (Fig. 2). However, the Pst-Aa fusion protein failed to auto-process despite carrying a PstAvrRpt2 self-cleavage site (see α-HA blot in Fig. 2). This suggests that the A. avenae homolog is probably not a functional protease (which is supported by the lower sequence homology to PstAvrRpt2, see Fig. 1) or alternatively, it has a different cleavage specificity. In agreement, an additional helix in the N-terminal sub-domain is predicted in the structural model of AaAvrRpt2 (Fig. 1B). Furthermore, unlike all the other AvrRpt2 homologs, the predicted ligand binding pocket resides in the C-terminal sub-domain (Fig. 1C). BpAvrRpt2 suppressed PAMP-induced MPK4/11 activation, while interestingly, the Pst-Bp fusion protein lost this ability. Both versions did not cleave RIN4, although the B. pyrrocinia homolog is a functional protease, as is evident from the partial auto-cleavage of the Pst-Bp fusion protein (see α-HA blot in Fig. 2). The CfAvrRpt2 homolog and the Pst-Cf fusion both failed to suppress PAMP-induced MPK4/11 activation. Notably, the Pst-Cf fusion protein weakly cleaved RIN4, although the original CfAvrRpt2 apparently failed to do so. The partial auto-cleavage of the Pst-Cf fusion protein (Fig. 2 lower panel) confirms that the shorter CfAvrRpt2 can be a functional protease when its N-terminus is extended. This is in agreement with the overall higher sequence identity of CfAvrRpt2 (61%) to PstAvrRpt2 compared to AaAvrRpt2 (27%) or BpAvrRpt2 (∼30%) and the overall best fit in the predicted 3-D structure (Fig. 1B, C). Altogether, these data corroborate our previous result that suppression of PAMP-induced MPK4/11 activation is independent of RIN4 cleavage. It also suggests that the Cys-protease activity or specificity can be modified through presence of a functional or auto-processed N-terminus of PstAvrRpt2. Replacement of a predicted extra helix^G15-S23 of BpAvrRpt2 (top yellow arrow and white arrowhead in Fig 1B) by the PstAvrRpt2 N-terminus did not result in any conformational change in the overall structure. However, this appears to modify (enlarge?) the Pst-Bp ligand binding pocket (Fig 1C) and may explain the change in substrate specificity. Finally, the substantial sequence variances of the homologs within the predicted PstAvrRpt2 ligand pocket (red boxes in Fig. 1A) suggests different substrates are being targeted by these various Cys-proteases. While the current work cannot clarify the roles or relevance of these putatively truncated AvrRpt2 homologs in bacteria-host interactions, it expands the molecular toolbox for studying AvrRpt2 functions. In particular, the distinct activities of BpAvrRpt2 and Pst-Bp fusion in blocking MPK4/11 activation (Fig. 2) can aid in identification of the protease substrate(s) relevant for the specific inhibition of PAMP-induced MPK4/11 activation. While AvrRpt2 is a well-studied bacterial effector that contributes to virulence on Arabidopsis^15,16 and pears,¹⁷ it probably has further undiscovered virulence functions and suppression of MPK4/11 may be one of them.

Figure 2.

Suppression of PAMP-induced MPK activation by the different AvrRpt2 variants. Arabidopsis protoplasts expressing the indicated constructs were treated with 100 nM flg22 (flagellin peptide) for 15 min. Samples were split and run on 2 gels. Activated forms of MPKs were visualized by an α-pTEpY (i.e. recognizing dual phosphorylated MPKs) antibody. Accumulation of RIN4 and expression of the AvrRpt2 variants (and also their auto-cleavage) were monitored by α-RIN4 and α-HA antibodies, respectively. The weak expression of the BpAvrRpt2 is highlighted with an asterisk. Note the double bands in Pst-Bp and Pst-Cf represent the partially auto-cleaved proteins and precursors. Amido black staining of the Rubisco large subunit served as a loading control. Similar results were obtained in 3 independent experiments. Identities of the MPKs and molecular weight marker sizes are indicated. Generation of deletion/fusion constructs was performed using typeIIs restriction enzyme-based cloning. Gateway®-compatible Aa/Bp/CfAvrRpt2 Entry clones and the PstAvrRpt2 N-terminus as a ligation-compatible fragment were PCR-amplified and then used in a combined digestion-ligation reaction.^21,22 (Primers with underlined typeIIs restriction enzyme recognition sites [BsaI]: PstΔ1–119_fwd 5′-ATATGGTCTCAATGGGATGTTGGTATGCCTG-3′ and PstΔ1–119_rev 5′-TTAAGGTCTCTCCATGGTGAAGGGGGCGGCCGC-3′; PstN-terminus_fwd 5′-ATATGGTCTCCCATGAAAATTGCTCCAGTTG-3′ and PstN-terminus_rev 5′-TTAAGGTCTCGAGCATCCCATTCGCTCATTACCTT-3′; Pst-Aa_fwd 5′-ATATGGTCTCTTGCTGGGAGGCTACCATGAACATG-3′ and Pst-Aa_rev 5′-TTAAGGTCTCCCATGGAGCCTGCTTTTTTGTACAAAG-3′; Pst-Bp_fwd 5′-ATATGGTCTCTTGCTGGTACGCTGCTGCTTGC-3′ and Pst-Bp_rev identical to Pst-Aa rev; Pst-Cf_fwd 5′-ATATGGTCTCTTGCTGGTACGCTTGCGCTAGAATG-3′ and Pst-Cf_rev identical to Pst-Aa_rev). Sequence-verified clones were transferred by Gateway® LR-recombination into the pUGW14 vector ²³ to express C-terminally 3xHA-tagged proteins in protoplasts.

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.

Acknowledgments

This work was supported by the BMBF and the DFG under the programs “ProNET-T3” (03ISO2211B) and “Molecular mechanisms of information processing in plants” (SFB648/TP-B1), respectively. We thank Nicole Bauer for excellent technical assistance.