Virulence strategies of an insect herbivore and oomycete plant pathogen converge on host E3 SUMO ligase SIZ1

Summary Pathogens and pests secrete proteins (effectors) to interfere with plant immunity through modification of host target functions and disruption of immune signalling networks. The extent of convergence between pathogen and herbivorous insect virulence strategies is largely unexplored. We found that effectors from the oomycete pathogen, Phytophthora capsici, and the major aphid pest, Myzus persicae target the host immune regulator SIZ1, an E3 SUMO ligase. We used transient expression assays in Nicotiana benthamiana as well as Arabidopsis mutants to further characterize biological role of effector–SIZ1 interactions in planta. We show that the oomycete and aphid effector, which both contribute to virulence, feature different activities towards SIZ1. While M. persicae effector Mp64 increases SIZ1 protein levels in transient assays, P. capsici effector CRN83_152 enhances SIZ1‐E3 SUMO ligase activity in vivo. SIZ1 contributes to host susceptibility to aphids and an oomycete pathogen. Knockout of SIZ1 in Arabidopsis decreased susceptibility to aphids, independent of SNC1, PAD4 and EDS1. Similarly SIZ1 knockdown in N. benthamiana led to reduced P. capsici infection. Our results suggest convergence of distinct pathogen and pest virulence strategies on an E3 SUMO ligase to enhance host susceptibility.


Introduction
The plant immune system is complex, featuring different classes of receptors to detect pathogens and pests and initiate multilayered defence responses. Pattern recognition receptors (PRRs) recognize conserved pest and pathogen molecules, called pathogen-associated molecular patterns (PAMPs), to activate immune responses and fight off the intruder (Jones & Dangl, 2006;Monaghan & Zipfel, 2012). Pathogens and pests deliver an arsenal of effector proteins inside their host to counter these and other plant defence pathways to promote effector-triggered susceptibility (ETS) through modulation of host protein activities. In addition, these effectors likely contribute to effective infection or infestation strategies by promoting the release of nutrients to support pathogen or pest growth. Another layer of plant immunity may be activated upon recognition of these effectors, or their activities, by nucleotide-binding leucine-rich repeat (NLR) proteins, which usually is associated with the activation of a hypersensitive response (HR). Given that plants carefully balance energy allocation between growth, development and reproduction, any effective immune responses need to be appropriate and controlled (Huot et al., 2014).
The identification of effector host targets and their effectorinduced modification(s) can reveal the mechanistic basis of virulence and the biological processes that lead to susceptibility. Moreover, the identification of effector host targets for a range of pathogens pointed to convergence on key host proteins. For example, Avr2 from the fungal pathogen Cladosporium fulvum, EPIC1 and EPIC2B from the oomycete Phytophthora infestans, and Gr-VAP1 from the plant-parasitic nematode Globodera rostochiensis target the same defence protease Rcr3 pim in tomato (Song et al., 2009;Lozano-Torres et al., 2012). In addition, the effector repertoires of distinct plant pathogens, such as the bacterium Pseudomonas syringae, oomycete Hyaloperonospora arabidopsidis, and the ascomycete Golovinomyces orontii disrupt key components of immune signalling networks (Mukhtar et al., 2011;Weßling et al., 2014). Specifically, transcription factor TCP14 is targeted by effectors from Pseudomonas syringae, H. arabidopsidis and Phytophthora capsici, and contributes to plant immunity (Weßling et al., 2014;Stam et al., 2021). These findings suggest that molecular virulence strategies have evolved independently in distinct pathogens and converged on a small set of regulators with central roles in immunity.
While over the past decades our understanding of pathogen virulence strategies and susceptibility has increased dramatically, the extent with which host targets of plant-herbivorous insects overlap with other pathogens remains to be investigated. Effector biology has recently emerged as a new area in plant-herbivorous insect interactions research, leading to the identification of effector repertoires in several species (Carolan et al., 2009;Bos et al., 2010b;Kaloshian & Walling, 2016;Thorpe et al., 2018;Rao et al., 2019), several host targets, and insights into their contribution to the infestation process (Rodriguez et al., 2017;Chaudhary et al., 2019;Wang et al., 2019;Xu et al., 2019). These studies support an extension of the effector paradigm in plantmicrobe interactions to plant-herbivorous insect interactions. Whether plant pathogenic microbes and insects adopt similar strategies to attack and reprogram their host to redirect immune responses is yet to be determined.
Here, we show that Myzus persicae (aphid) effector, Mp64, and P. capsici (oomycete) effector CRN83_152 (also called PcCRN4) (Stam et al., 2013a;Mafurah et al., 2015), associate with the immune regulator SIZ1 in the plant nucleus. SIZ1 stability and cell death activation in Nicotiana benthamiana are differentially affected by these effectors, suggesting these proteins feature distinct activities via this immune regulator. SIZ1 is an E3 SUMO ligase involved in abiotic and biotic stress responses, including salicylic acid (SA)-mediated innate immunity and EDS1/PAD4mediated resistance gene signalling (Miura et al., 2005(Miura et al., , 2009Catala et al., 2007;Lee et al., 2007;Jin et al., 2008;Ishida et al., 2012;Lin et al., 2016). Additionally, SIZ1 regulates plant immunity partially through the immune receptor SNC1 (Gou et al., 2017) and controls the trade-off between SNC1dependent immunity and growth in Arabidopsis at elevated temperature (Hammoudi et al., 2018). By using Arabidopsis knockout lines and gene silencing in N. benthamiana we show that SIZ1 negatively regulates plant immunity to aphids and an oomycete pathogen, and is required for pathogen infection and pest infestation. Moreover, Arabidopsis siz1-2 displayed reduced susceptibility to aphids and an oomycete pathogen. Critically, the observed immunity phenotype is independent of SNC1signalling suggesting that immunity is not specified by previously characterized SIZ1-immune signalling pathways. Our results suggest that the effector target convergence principle can be extended to herbivorous insects and raise important questions about mechanisms of action.

Plants and growth conditions
Nicotiana benthamiana plants were grown in a glasshouse with 16 h of light, at 25°C during daytime.

Aphid rearing and Phytophthora capsici growth conditions
Myzus persicae (JHI_genotype O; Thorpe et al., 2018) was maintained on oil seed rape (Brassica napus) plants in a Perspex growth chamber, with 12 h light, at 17°C and 50% relative humidity.
Phytophthora capsici isolate LT1534 (obtained from Kurt Lamour, University of Tennessee, Knoxville, TN, USA) was maintained on V8 agar cubes at room temperature. For zoospore collection, P. capsici LT1534 was grown on V8 agar plates at 25°C.
An entry clone carrying AtSUMO1 was kindly provided by Dr H.A. van den Burg, The University of Amsterdam (van den Burg et al., 2010). For in planta SUMOylation assays, AtSUMO1 was cloned into destination vector pK7WGR2 (N-terminal mRFP tag) (Karimi et al., 2002) by LR reactions using Gateway technology (Invitrogen). Vector specific primer p35s-F used in plasmid construction are listed in Table S1.
All plasmids generated in this study are listed in Table S2.
Yeast-two-hybrid assays Y2H screening of effectors against a N. benthamiana library was based on the Dualsystems Y2H system (Dual Systems Biotech) following manufacturer's instructions. Bait vectors (pLex-N) carrying effector sequences (lacking the signal peptide encoding sequence) were transformed into yeast strain NMY51. The prey library was generated in pGAD-HA from cDNA obtained from a combination of healthy leaves, leaves infected with P. capsici, and leaves infested with aphids. Transformants were selected on media plates lacking leucine, tryptophan, and histidine (-LWH) with addition of 2.5 mM 3-amino-1,2,4-triazole (3-AT). Yeast colonies were subjected to the β-galactosidase reporter assays according to manufacturer's instructions. The inserts of selected yeast colonies were sequenced and analysed. The Mp64/ CRN83_152-SIZ1 interaction was validated in yeast by independent co-transformation experiments and reporter assays.
For co-expression assays, mixtures of Agrobacterium cultures carrying N-terminal GFP tagged Mp64, CRN83_152_6D10 or GUS with cultures carrying C-terminal red fluorescent protein (RFP) tagged AtSIZ1, NbSIZ1 or GUS, respectively, were infiltrated into N. benthamiana leaves (OD 600 = 0.3 for each construct; for p19, OD 600 = 0.1). Cell death was scored 4-7 d postinoculation using a scale of 0-3 based on the severity of the phenotype. Infiltration sites were scored for no symptoms (score 0), chlorosis with localized cell death (score 1), < 50% of the site showing visible cell death (score 2), over 50% of the infiltration site showing cell death (score 3).
Statistical analyses were conducted by using RSTUDIO v.1.2.5001 running R-3.6.1. Differences between treatments were analysed using the Kruskal-Wallis test with post hoc Dunn's test for multiple comparisons.

Pathogen and pest infection/infestation assays on Arabidopsis
Two 2-d-old M. persicae nymphs (age-synchronized) were placed on 4-6-wk-old Arabidopsis plants. The plants were placed in a large plastic tube sealed with a mesh lid and placed in a growth cabinet (8 h of light, 22°C : 20°C for day : night, 60% humidity). Aphids were counted 10 d post-infestation.
Phytophthora capsici isolate LT1534 was grown in V8 agar plate for 3 d in the dark at 25°C and exposed to continuous light for 2 d to stimulate sporulation. Sporangia were collected in icecold water and incubated under light for 30-45 min to promote zoospore release. For Arabidopsis infection, 4-6-wk-old plants were spray-inoculated with 100 000 spores ml −1 . The percentage of infected leaves was scored 8 d after inoculation.
Statistical analyses were carried out using RSTUDIO v.1.2.5001 running R-3.6.1. A linear mixed effects model, with experimental block and biological replicate incorporated as random factors, was used for aphid fecundity assays. A linear mixed effects model, with biological replicates as a random factor, was used for P. capsici infection assays. ANOVA was used to analyse the final models, by using EMMEANS package calculating the least squares means as a post hoc test.
Infection assays on N. benthamiana leaves transiently expressing effectors Phytophthora capsici infection assays were performed on N. benthamiana leaves expressing CRN83_152_6D10, Mp64 or the vector control upon agroinfiltration (OD 600 = 0.3 each). Two days after infiltration, leaves were drop inoculated with 5 µl of zoospore solution (50 000 spores ml −1 ) from strain LT1534. Lesion diameters were measured at 2 d post-inoculation.

Virus-induced gene silencing assays
Tobacco rattle virus (TRV)-based virus-induced gene silencing (VIGS) was used to silence NbSIZ1 in N. benthamiana. The VIGS construct was generated by cloning a 249-bp fragment of NbSIZ1, amplified with primers Sumo_Vigs_Phusion_Frag3_F and Sumo_Vigs_Phusion_Frag3_R (Table S1). To generate a TRV control, a GFP fragment was amplified using the primers eGFP_Fw and eGFP_Rv (Table S1). Amplified fragments were cloned into the TRV vector (pTRV2) (Lu et al., 2003) using the In-Fusion HD cloning kit (Clontech, Mountain View, CA, USA). Agrobacterium strains containing desired pTRV2 constructs were coinfiltrated with strains carrying pTRV1 at OD 600 = 0.5 into N. benthamiana plants. Three weeks post infiltration, leaves at the same position of different plants were detached for quantification of NbSIZ1 transcripts by quantitative reverse transcription-polymerase chain reaction (qRT-PCR) and P. capsici infection assays. Six independent plants were used for each VIGS construct in each replicated experiment, with a total of three replicated experiments. For infection assays, leaves were drop-inoculated with 5 μl of zoospore suspension (50 000 spores ml −1 ) of P. capsici strain LT1534, or for data corresponding to Fig. S7 with 10 µl of zoospore suspension (100 000). Lesion diameter was recorded 2-3 d post-inoculation. Data analyses was carried out by using RSTUDIO v.1.2.5001 running R-3.6.1. Group comparison was conducted by Mann-Whitney U test for nonnormally distributed data.

Confocal microscopy
Agrobacterium strains carrying desired constructs were infiltrated individually or in combination in N. benthamiana plants with an OD 600 of 0.1. Cells were imaged at c. 36 h post-infiltration using Leica TCS SP2 AOBS (Leica Microsystems, Wetzlar, Germany) and Zeiss 710 confocal microscopes with HC PL FLUOTAR 63X0.9 and HCX APO L U-V 40X0.8 water-dipping lenses. GFP was excited with 488 nm from an argon laser, and emissions were detected between 500 and 530 nm. The excitation wavelength for mRFP was 561 nm and emissions were collected from 600 to 630 nm. Nicotinana benthamiana Histone (H2B) fused to mRFP was used as a nuclear marker (Goodin et al., 2007). Single optical section images or z-stacks images were collected from leaf cells those have relatively low expression level to minimize the potential artefacts. Images were projected and processed using the IMAGEJ 1.52p-FIJI (Wayne Rasband, National Institute of Health, Bethesda, MD, USA).

Detection of SUMO conjugates and altered SIZ1 stability by Western blotting
To determine whether ectopic/overexpression of SIZ1 alone or in combination with effectors altered SUMOylation, we made use of a plant expression construct expressing RFP-AtSUMO1 (van den Burg et al., 2010). Agrobacterium GV3101 strains carrying constructs to express AtSIZ1-myc/NbSIZ1-myc or myc-GUS (control) with or without GFP-Mp64/GPF-CRN83_152_6D10 or GFP (control) were combined with strains expressing RFP-AtSUMO1 for infiltration of N. benthamiana leaves. An OD 600 of 0.3 was used for each construct, with the addition of an Agrobacterium strain expressing the silencing suppressor p19 (OD 600 = 0.1). Forty-eight hours post infiltration, the N. benthamiana plants were exposed to heat stress by placing them in a 37°C incubator for 1 h.
For detection of SIZ1 protein levels in the absence/presence of effectors, side-by-side infiltrations were performed as mentioned earlier but without the presence of RFP-AtSUMO1 and without heat stress.
Protein was extracted from two leaf discs (1.5 cm diameter) in 200 µl GTEN buffer as described earlier. The protein lysate was mixed with 4× protein loading buffer (928-40004; Li-Cor, Lincoln, NE, USA) (with addition of 100 mM DTT) by a ratio of 3 : 1. For each sample, 5 µl was loaded into 4-20% Mini-Protean ® TGX Gels (4561096; Bio-Rad), followed by SDS-PAGE (20 mA per each gel, run for c. 1 h) after denaturation at 65°C for 5 min. The protein was subsequently transferred to polyvinylidene fluoride (PVDF) membranes for 90 min at 90 V using a wet transfer system. After transfer, the membranes were stained using RevertÔ 700 Total Protein Stain (926-11021; Li-Cor) following the manufacturer's manual for detection of total amounts of protein. The membranes were immediately imaged in the 700 nm channel using an Odyssey ® CLx Imaging System (Li-Cor).
For subsequent detection of epitope tagged proteins, membranes were blocked with Intercept ® (phosphate-buffered saline (PBS)) Blocking Buffer (927-70001; LI-COR) for an hour at room temperature, following an hour of primary antibody incubation in Intercept ® (PBS) Blocking Buffer with 0.2% Tween ® 20 (P1379; Sigma-Aldrich) at room temperature. After washing with PBS buffer (three times, 5 min per each), the membranes were incubated with IRDye secondary antibody in Intercept ® (PBS) Blocking Buffer with addition of 0.02% SDS and 0.2% Tween ® 20 (P1379; Sigma-Aldrich) for an hour at room temperature. After three washes with PBS buffer, the target proteins were detected in the 800 nm channel with an Odyssey ® CLx Imaging System.
Protein quantification was done by normalizing the band intensity of SIZ1 against the total protein amounts for the SIZ1 stability assays using EMPIRIA STUDIO v.2.1. Quantification of SUMO conjugates was done by normalizing the signal intensity of the selected molecular weight (MW) area of the blot corresponding to SUMO conjugates against the total protein amounts. Relative ratios of signal intensity within experimental set-ups we calculated based on comparisons to relevant control samples (e.g. GFP, myc-GUS).

RNA extraction and qRT-PCR
Total RNA was extracted by using RNeasy Mini Kit (Qiagen) and DNase I treated (Invitrogen). Briefly, 1 µg RNA was reversetranscribed using SuperScript III reverse transcriptase (Sigma-Aldrich) following the manufacturer's protocol. RT-qPCR was designed following the MIQE guidelines (Bustin et al., 2009) with gene specific primers (Table S1). EF1α (accession no. TC19582 (At5g60390)) and PP2A (accession no. TC21939 (At1g13320)) (Liu et al., 2012) were used as reference genes in N. benthamiana and PEX4 (or UBC; accession no. AT5G25760) in Arabidopsis thaliana (Dekkers et al., 2012). Each 12.5 μl reaction contained 1× GoTaq ® qPCR Master Mix, 1 μM of each primer, 1.4 mM magnesium chloride (MgCl 2 ), 2.4 μM CXR reference dye and a cDNA quantity of c. 25 ng. The PCR program was set on a StepOneÔ Real-Time PCR Machine (Applied Biosystems, Thermo Fisher Scientific, Waltham, MA, USA) as follows: 95°C for 15 min followed by 40 cycles of 15 s at 95°C, 30 s at 60°C, and 30 s at 72°C. A melting curve was generated at the end of the PCR program and 2 ÀΔΔC t value (Livak & Schmittgen, 2001) was calculated to determine the relative expression of NbSIZ1. Three technical replicates were performed in each run and three biological replicates were carried out.

Aphid effector Mp64 and oomycete effector CRN83_152 interact with AtSIZ1 and NbSIZ1
To gain novel insight into pathogen and pest effectors function towards virulence, we successfully applied Y2H screens to identify candidate host targets (Rodriguez et al., 2017). We identified the E3 SUMO ligase SIZ1 in screens against a N. benthamiana library (generated from aphid infested and P. capsici infected leaves) with M. persicae (aphid) effector Mp64 and P. capsici (oomycete) effector CRN83_152 as baits. Mp64 was screened against an estimated 5 × 10 6 cDNAs and revealed two independent prey clones with an insert showing similarity to SIZ1, whilst the effector CRN83_152 screen of 4 × 10 6 yeast transformants, identified three independent prey clones with an insert similar to SIZ1. All putative interactors identified in the two effector screens are summarized in Table S3. Since all NbSIZ1 (N. benthamiana SIZ1) prey clones from the Mp64 and CRN83_152 screens were partial-length, we designed primers to amplify and clone the full-length NbSIZ1. Although we were unable to amplify NbSIZ1 based on the two best BLAST hits against the N. benthamiana genome (Niben101Scf15836g01010.1 and Niben101Scf04549g09015.1) due to poor/no primer annealing at the 3 0 end, we successfully amplified NbSIZ1 sequences based on the 3 0 end of SIZ1 sequences from Nicotiana attenuata (XP_ 019237903) and Nicotiana tomentosiformis (XP_018631066). The full-length NbSIZ1 sequence we cloned was identical to our partial yeast prey clones and NbSIZ1 database sequences, except for a 27 amino acid insertion at position 225-252. A direct comparison between NbSIZ1 and the well characterized AtSIZ1 showed 60% identity between proteins (Fig. S1). Given that AtSIZ1 is well characterized and helps regulate plant immunity (Miura et al., 2005;Catala et al., 2007;Lee et al., 2007;Hammoudi et al., 2018), we included AtSIZ1 in our efforts to further validate effector-SIZ1 interactions and characterize the role of SIZ1 in plant-aphid/oomycete interactions using both Arabidopsis and N. benthamiana resources. It should be noted that whilst Arabidopsis is a host for M. persicae, this plant species is not a natural host for P. capsici. We first tested whether Mp64 and CRN83_152 interact with full-length NbSIZ1 and AtSIZ1 in yeast. Whilst yeast reporter assays showed interaction of Mp64 with both full-length SIZ1 versions (Fig. S2), we were unable to obtain yeast co-transformants expressing both CRN83_152 and full-length SIZ1 in repeated transformation experiments that included transformation controls. We also included a mutant of CRN83_152, called CRN83_152_6D10 (Amaro et al., 2018), which does not trigger CRN-cell death activation but retains virulence activity, in co-transformation experiments with similar results. Based on further data presented later, we hypothesize that the lack of CNR83_152/SIZ1 yeast co-transformants is due to enhanced E3 SUMO ligase activity of SIZ1 in the presence of this effector which may affect yeast cell viability.
To test for in planta effector-SIZ1 interactions, we coexpresssed GFP-Mp64 and GFP-CRN83_152_6D10 with either AtSIZ1-myc or NbSIZ1-myc in N. benthamiana (Fig. 1a,b). Immunoprecipitation of both effectors resulted in the copurification of NbSIZ1, suggestive of an association in planta (Fig. 1b). Co-IP of the effectors with AtSIZ1 gave similar results, however, with CRN83_152 and CRN83_152_6D10 only showing a weak band corresponding to AtSIZ1 upon co-purification (Fig. 1a). Altogether our data demonstrate that effectors from two distinct plant parasites associate with the same host protein, SIZ1, in planta, prompting us to further investigate the contribution of SIZ1 and the effectors to susceptibility.
Silencing of NbSIZ1 reduces N. benthamiana host susceptibility to P. capsici To assess the contribution of SIZ1 to immunity in a P. capsici host species, we made use of VIGS in N. benthamiana (Ratcliff et al., 2001;Lu et al., 2003). Our TRV-NbSIZ1 construct, designed to silence NbSIZ1, reduced transcripts levels by around 60% compared with plants expressing the TRV-GFPfrag control (a fragment of GFP) (Fig. S3). Silenced plants showed a slight reduction in growth compared with the TRV-GFPfrag control and cell death in older leaves (Fig. S3). In our hands, VIGS assays based on TRV in N. benthamiana are incompatible with aphid assays (TRV infection causes aphids to die), therefore, we only performed infection assays with P. capsici on NbSIZ1 silenced plants. Detached leaves were used for P. capsici infection assays based on zoospore droplet inoculations, followed by lesion size diameter measurements. P. capsici lesion size on NbSIZ1 silenced leaves was significantly reduced 2-4 d after inoculation when compared to control plants (Mann-Whitney U test, P < 0.001; Figs 2, S3). These results indicate that NbSIZ1 contributes to host susceptibility to this oomycete plant pathogen.
Loss-of-function mutation siz1-2 in Arabidopsis leads to reduced susceptibility to M. persicae and P. capsici Since AtSIZ1 negatively regulates plant innate immunity in Arabidopsis to the bacterial plant pathogen Pseudomonas syringae pv Tomato DC3000 (Pst) , we tested whether this also applies to interactions with M. persicae and P. capsici. We performed aphid performance assays, based on fecundity measurements, as well as P. capsici infection assays on the Arabidopsis loss-of-function mutant siz1-2. Given that siz1-2 mutants have a dwarf phenotype, associated with SA hyper-accumulation, we included Arabidopsis line siz1-2/NahG, in which this phenotype is (partially) abolished . While Arabidopsis is a host for the aphid M. persicae, only few P. capsici isolates infect Arabidopsis under controlled environmental conditions and high levels of inoculum (Wang et al., 2013), suggesting that Arabidopsis is not a natural host. Aphid performance assays showed a significant reduction in fecundity on the siz1-2 and siz1-2/ NahG lines compared to the Col-0 (ANOVA, P < 0.0001; Fig.  3a) and NahG controls (ANOVA, P < 0.01; Fig. 3a), respectively, with only few aphids surviving on the siz1-2 line. The siz1-2 reduced susceptibility to aphids is largely maintained in the NahG background, implying that this phenotype is largely independent of SA accumulation. For P. capsici infection assays, plants were spray inoculated with a zoospore solution and the percentage of symptomatic leaves was counted 10 d later. The percentage of symptomatic siz1-2 leaves was reduced by 83% compared with the Col-0 control (ANOVA, P < 0.0001; Figs 3b, S4). We did not observe a difference in P. capsici infection levels between the NahG line and Col-0 but did note a slight increase in infection on siz1/NahG compared to the NahG background (ANOVA, P < 0.01; Figs 3b, S4).

Research
New Phytologist the reduced susceptibility of siz1-2 to aphids is independent of EDS1. In addition, aphid fecundity was reduced on siz1-2/pad4-1 by around 65% compared with the pad4-1 mutant (ANOVA, P < 0.0001; Fig. 3c), and was comparable to siz1-2. These data suggest that siz1-2 reduced susceptibility to aphids is also independent of PAD4.
The siz1-2/snc1-11 double mutant also showed enhanced resistance to P. capsici, with 55% less symptomatic leaves compared to the snc1-11 mutant (ANOVA, P < 0.0001; Figs 3b, S4). The percentage of symptomatic leaves on siz1-2/snc1-11 was slightly higher compared with the siz1-2 mutant (ANOVA, P < 0.05; Figs 3b, S4). With the siz1-2 enhanced resistance to P. capsici largely maintained in the snc1-11 background, this phenotype is likely independent of the immune receptor SNC1. Overall, siz1-2 reduced susceptibility to both M. persicae and P. capsici is independent of defence signalling components previously implicated in SIZ1 immune functions. These data are in line with a model wherein SIZ1 contributes to host susceptibility to certain pests and pathogens, perhaps upon effector-mediated modulation.
Nuclear aphid effector Mp64 enhances Arabidopsis susceptibility to M. persicae While the nuclear PcCRN83_152 effector from P. capsici was previously shown to be essential for pathogen virulence and promotes plant susceptibility (Stam et al., 2013a;Mafurah et al., 2015), the role of aphid effector Mp64, which was identified as a candidate effector in Acyrthosiphon pisum and M. persicae through bioinformatics pipelines (Carolan et al., 2011;Thorpe et al., 2016;Boulain et al., 2018), is unknown. Mp64 is a protein of unknown function with a predicted nuclear localization based on PROTEINPREDICT (Yachdav et al., 2014) and NLSTRADAMUS (Nguyen Ba et al., 2009), and Mp64 homologues are present in other aphid species (Fig. S5). We investigated the subcellular localization of Mp64 by confocal microscopy of N. benthamiana leaves transiently expressing GFP-Mp64 (lacking the predicted signal peptide). Imaging of epidermal cells expressing Mp64 revealed accumulation of GFP-Mp64 in the nucleus and nucleolus, with no signal detectable in the cytoplasm (Fig. 4a). Nuclear localization of GFP-Mp64 was confirmed upon co-localization with the nuclear marker Histone 2B (H2B) (Fig. 4b). In addition, we observed dots within the nucleoplasm corresponding to GFP-Mp64.
To confirm that Mp64 contributes to aphid virulence, we generated Arabidopsis transgenic lines expressing the mature Mp64 protein driven by the 35S promoter. Arabidopsis lines expressing Mp64 showed no developmental or growth phenotypes (Fig. S6) and were subjected to aphid fecundity assays. Two agesynchronized M. persicae aphids were placed on transgenic Mp64 lines and Col-0 control plants, and progeny was counted after 10 d. The average number of aphids on two independent transgenic Mp64 lines (25.6 and 29.5) was around 30% higher than on the Col-0 control (Kruskal-Wallis test with multiple comparisons Benjamini-Hochberg correction based on false discovery rate (FDR); Fig. 4c), indicating that Mp64 enhances Arabidopsis host susceptibility to M. persicae. While our two Mp64 transgenic lines showed different levels of Mp64 expression, with line 29.5 showing lower expression than line 25.6. We did not find a correlation between higher expression and virulence effect, with both lines showing a similar increase in aphid numbers.
Distinct nuclear localization patterns of Mp64, CRN83_152 and SIZ1 Since both Mp64 and CRN83_152 are nuclear effectors, and their host interacting protein SIZ1 is reported to localize and function in the plant nucleus (Miura et al., 2005), we determined whether the effectors co-localize with SIZ1 in this

Research
New Phytologist subcellular compartment. We performed confocal imaging of N. benthamiana leaves transiently co-expressing SIZ1-mRFP and GFP-effector fusions. We did not detect signal corresponding to the effectors or SIZ1 outside the nucleus (Figs 4, 5, S8). In line with previous reports, mRFP signal corresponding to both AtSIZ1 and NbSIZ1 was visible in the plant nucleoplasm, along with distinct speckles in the nucleolus (Fig. 5, S8). Expression of full-length SIZ1-RFP was confirmed by Western blotting (Fig. S9). GFP signal corresponding to GFP-Mp64 and CRN83_152 was detectable in the nucleus, with GFP-Mp64 more localized within the nucleolus and in speckles within the nucleoplasm, and GFP-CRN83_152 present only in the nucleoplasm (Figs 4, 5, S8). Indeed, GFP-CRN83_152 colocalizes with AtSIZ1-RFP and NbSIZ1-RFP in the nucleoplasm, whereas GPF-Mp64 shows a more distinct localization in the nucleus that only partially overlaps with nucleoplasm localization of SIZ1. Interestingly, nucleolar speckles corresponding to SIZ1-RFP occasionally coincided with loss of GFP-Mp64 signal, and GFP-Mp64 speckles within the nucleoplasm coincided with loss of SIZ1-RFP signal. However, we did not find evidence for altered localization of SIZ1 in the presence of the effectors, and vice versa (Figs 5, S8). With the effectors and SIZ1 only detected within the nucleus, this likely is the compartment where interactions take place.

Mp64 but not CRN83_152_6D10 stabilizes AtSIZ1 in planta
In co-IP experiments, we consistently observed increased protein levels of AtSIZ1 in input samples upon co-expression with Mp64 but not CRN83_152_6D10 (Fig. 1a,b). To test whether Mp64 indeed stabilizes SIZ1 in planta, we performed co-expression assays of both effectors with SIZ1 in parallel in three independent experiments. Western blot analyses combined with quantitative analyses of band intensities, showed that AtSIZ1 levels were higher than NbSIZ1 levels among all replicates. In addition, we consistently observed an increase in SIZ1 protein levels in the presence of Mp64 compared to the GFP-GUS control (Fig. 6a, additional replicates in Fig. S10), indicating that Mp64 stabilizes SIZ1 in planta. Although not consistent across replicates, we did observe slightly decreased AtSIZ1 levels in the presence of CRN83_152_6D10, compared to the GFP control, but this observation was not consistent across replicated experiments (Fig.  6a, additional replicates in Fig. S10).
Overall, we noted that full length AtSIZ1 and NbSIZ1 are rather difficult to detect by Western blotting, suggesting low expression and/or low protein stability. Indeed, Lin et al. (2016) previously showed that the 26S proteasome inhibitor MG132 reduces degradation of GFP-AtSIZ1 mediated by the ubiquitin E3 ligase COP1 (CONSTITUTIVE PHOTOMORPHOGENIC 1), an ubiquitin E3 ligase. In line with this we show that both AtSIZ1 and NbSIZ1 are more readily detected by Western blotting upon MG132 treatment (Fig. 6b), suggesting that the levels of both these SIZ1 versions is tightly regulated in planta. When performing transient expression assays in N. benthamiana with AtSIZ1 and NbSIZ1, we observed the onset of cell death starting from 3 d after infiltration specifically upon expression of AtSIZ1. We investigated whether co-expression of the aphid and oomycete effectors with SIZ1 would either enhance or reduce this cell death activation. In the absence of any effectors, AtSIZ1 consistently activated cell death from 3 to 4 d after infiltration, whereas only occasional microscopic cell death was visible in infiltration sites expressing NbSIZ1. Both AtSIZ1 and NbSIZ1 fusion proteins, with a C-terminal RFP tag, were detectable in transient expression assays (Fig. S9). While co-expression of Mp64 with SIZ1 did not affect the cell death phenotype, co-expression of CRN83_152_6D10 with AtSIZ1 led to a stronger cell death response compared to the AtSIZ1 and CRN83_152_6D10 controls (Figs 7, S11). These data suggest that CRN83_152_6D10 but not Mp64 enhances AtSIZ1-triggered cell death.

CRN83_152 but not Mp64 enhances SUMOylation upon transient co-expression with SIZ1
To assess whether both AtSIZ1 and NbSIZ1 are active upon transient expression in N. benthamiana and whether effectors CRN83_152 and Mp64 alter any E3 SUMO ligase activity, we performed co-expression assays with RFP-tagged AtSUMO1. First, we co-infiltrated Agrobacterium strains carrying constructs for AtSIZ1-myc, NbSIZ1-myc and myc-GUS (control) with RFP-SUMO1, to assess whether ectopic/overexpression of SIZ1 increased detectable SUMO profiles upon heat treatment. Western blot analyses showed an increase in the presence of SUMO profiles, as detected with RFP-antibodies against RFP-SUMO1, upon expression of SIZ1 compared to the GUS control, with the strongest and most consistent increase upon expression with AtSIZ1 (Figs 8a, S12). We then performed similar co-expression assays and Western blot analyses in the presence and absence of effectors Mp64 and CRN83_152_6D10. As described earlier, we observed increased levels of SIZ1 in the presence of Mp64, as well as a slight decrease in the presence of CRN83_152_6D10 (Fig. 8b). Furthermore, in the presence of both SIZ1 and CRN83_152_6D10 we noted an increase in SUMO profile levels compared to the no effector (GFP) control (Figs 8b, S12). We detected no increase in SUMO profile levels in the presence of CRN83_152_6D10 in combination with the GUS (control) indicating that that the observed increase in SUMOylation mediated by this effector is dependent on SIZ1 ectopic/overexpression. Our data suggest that the P. capsici effector CRN83_152_6D10 enhances SIZ1 activity, most likely to enhance host susceptibility.

Discussion
Pathogen infection strategies involve extensive modification of host cell biology, which rely on the modulation of hubs that

Research
New Phytologist control plant immunity. We show that effectors from an herbivorous insect and oomycete plant pathogen target the host E3 SUMO ligase SIZ1. Our findings suggest that the virulence strategies of two plant parasites, with distinct evolutionary histories and lifestyles, convergence on an important host immune component.
We show that SIZ1 is a key target of distinct plant parasites, which is in line with a recent study on the cyst nematode Globodera pallida, which shows that effector GpRbp1 associates with potato SIZ1 in planta (Diaz-Granados et al., 2019). StSIZ1 emerged as a negative regulator of immunity in plantnematode interactions (Diaz-Granados et al., 2019), but the signalling requirements for this immunity have not yet been reported. We propose that SIZ1 is an important regulator of susceptibility to a broad range of plant parasites, including herbivorous insects. Indeed, Arabidopsis siz1-2 plants show reduced susceptibility not only upon pathogen infection as reported here (Fig. 3) and previously  but also upon aphid infestation. In contrast to siz1-2 enhanced resistance to Pseudomonas syringae pv tomato DC3000, which is dependent on SA, EDS1, PAD4 and SNC1, we find that resistance to the aphid M. persicae and the oomycete P. capsici is largely independent from these signalling components. These results point to (1) the involvement of a yet to be identified SIZ1-dependent signalling pathway that regulates plant immunity, and/or (2) a yet to be characterized role of SIZ1 in promoting pathogen and pest susceptibility. Although PAD4 has been reported to play an important role in plant defence against M. persicae (Pegadaraju et al., 2007), in line with Lei et al. (2014), we did not observe an enhanced susceptibility phenotype of Arabidopsis pad4-1 in our aphid performance assays. This may be due to differences in experimental design and conditions.
A reduction of SA levels in the NahG line did not enhance defence against the aphid M. persicae (this study and previous reports Pegadaraju et al., 2007;Lei et al., 2014), nor did this reduce nonhost resistance to the oomycete P. capsici, in contrast to an earlier report by Wang et al. (2013). However, we did observe a trend towards reduced resistance of the transgenic NahG line to P. capsici, but this reduction was not statistically significant and may be less pronounced due to differences in experimental set-up and infection conditions compared to Wang et al. (2013). Arabidopsis defence to insect herbivores is mediated predominantly through jasmonic acid (JA)-signalling, whereas defence against (hemi-)biotrophic pathogens tend to rely on SAsignalling (Howe & Jander, 2008;Pieterse et al., 2012). In the Arabidopsis-M. persicae interaction, siz1-2 reduced susceptibility is largely independent of SA accumulation, with the siz1-2/NahG line being more resistant to aphids than the NahG control and Col-0 (Fig. 3). Therefore, and in contrast to Lee et al. (2007), SIZ1-regulated immunity to aphids is independent of SAsignalling. Interestingly, the Arabidopsis siz1-2 mutant features changes in cell division, cell expansion and secondary cell wall formation, including reduced secondary cell wall thickening (Miura et al., 2010;Liu et al., 2019). Aphid feeding can trigger changes in cell wall composition that are associated with defences (Rasool et al., 2017), and therefore changes in cell wall formation can be responsible for altered susceptibility. However, reduced cell wall thickening most likely would lead to a reduction in defence against aphids rather than an increase as observed in the siz1-2 mutant. Leaf material was harvested 2 d post-inoculation. Total protein amount was detected using the RevertÔ 700 Total Protein Stain followed by imaging in the 700 nm channel using an Odyssey ® CLx Imaging System. The panel indicated by 'loading' shows a proportion of the membrane that includes Rubisco. Detection of GFP-fusion and SIZ1-myc fusion proteins upon antibody incubation was in the 800 nm channel using an Odyssey ® CLx Imaging System. Asterisk indicated bands corresponding to GFP-effectors/GFP. SIZ1 protein quantitation was done by normalizing the band intensity of SIZ1-myc to the total protein amounts using EMPIRIA STUDIO 2. With SIZ1 comprised of several conserved domain involved in different stress responses (Cheong et al., 2009), it is possible that Mp64 and CRN83_152 target different protein regions and functions. Arabidopsis and N. benthamiana SIZ1 domains include the SAP (scaffold attachment factor A/B/acinus/PIAS) domain, PINIT (proline-isoleucine-asparagine-isoleucinethreonine) domain, an SP-RING (SIZ/PIAS-RING) domain, SXS motif (serine-X-serine), and a PHD (plant homeodomain). Functional analyses, using a set of (deletion) mutants revealed that these domains contribute differently to the wide range of SIZ1 functions in both abiotic and biotic stress (Cheong et al., 2009). The SP-RING domain of AtSIZ1 contributes to the nuclear localization, SUMOylation activity, as well as the regulation of SA levels and associated plant defence responses. This domain is the suggested SIZ1 target site of the nematode effector GpRbp1 to interfere with SA-mediated defences (Diaz-Granados et al., 2019). Our Arabidopsis-M. persicae interaction assays though suggest that SIZ1 may also regulate immunity/susceptibility in a SA-independent manner where other domains may play an important role. Interestingly, SIZ1-mediated SUMOylation is involved in regulating sugar signalling independent of SA (Castro et al., 2015(Castro et al., , 2018, with the siz1-2 mutant showing reduced starch levels and increased expression of starch and sucrose catabolic genes. Aphid infestation affects sugar metabolism as reflected for example by an increase in sucrose and starch in infested Arabidopsis plants (Singh et al., 2011). With sugars in phloem sap also being the main aphid food source, it will be interesting to further explore a possible link between the role of SIZ1 in regulating sugar signalling and host susceptibility.
Our data support a key role for SIZ1 in host susceptibility to P. capsici and M. persicae that is targeted during infection and infestation, and point to potential different mechanisms by which effectors CRN83_152 and Mp64 modulate SIZ1 function. The presence of host SIZ1 is required for infestation/infection as knockout of AtSIZ1 and knockdown of NbSIZ1 result in reduced host susceptibility phenotypes. Therefore, we propose that Mp64 and CRN83_152 redirect and perhaps enhance SIZ1 function rather than inhibit its signalling activity. Indeed, we show that CRN83_152_6D10 increased SIZ1-mediated SUMOylation in planta, indicating that this effector modulates E3 SUMO ligase and an RFP control. Graph represents data from a combination of three biological replicates of 11-12 infiltration sites per experiment (n = 35). Data was collected 7 d after infiltration. (c) Bar graph showing the proportions of infiltration sites with different levels of cell death upon expression of SIZ1 (with C-terminal RFP tag) either alone or in combination with aphid effector Mp64 or Phytophthora capsici effector CRN83_152_6D10 (both effectors with green fluorescent protein (GFP) tag), or a GFP control. Data was collected 7 d after infiltration. Graph represents data from a combination of three biological replicates of 11-12 infiltration sites per experiment (n = 35). ***, P < 0.001 (Kruskal-Wallis test with post hoc Dunn's test for multiple comparison) and ns, no significant difference. (2022)

Research
New Phytologist activity (Fig. 8). However, Mp64 but not CRN83_152_6D10 enhanced stability of SIZ1 (Fig. 6). In line with these results, we found that Arabidopsis transgenic lines expressing Mp64 do not show a reduced growth phenotype similar to the siz1-2 mutant ( Fig. S4 and S6). However, expression of CRN83_152 but not Mp64 in N. benthamiana led to an increase in P. capsici infection. Based on these observations, we propose that while both virulence strategies have converged onto SIZ1, their mechanisms of action are distinct. In this context, we cannot rule out that additional candidate targets of Mp64 and CRN83_152 identified in our Y2H screens (Table S1) explain our observed differences in effector virulence activities.
As an E3 SUMO ligase, SIZ1 is required for SUMOylation of a range of substrates including chromatin modifiers, coactivators, repressors, and transcription factors that are associated with biotic and abiotic stress responses (Rytz et al., 2018). Similar to ubiquitination, SUMOylation involves three key steps (Verma et al., 2018). First, the SUMO precursor is cleaved and the SUMO moiety is linked to a SUMO-activating enzyme (E1). Activated SUMO is then transferred to the SUMO-conjugating enzyme (E2), after which it is linked to target substrates with the help of SUMO-ligases (E3). In the SUMO cycle, SUMO proteases are responsible for processing of the SUMO precursor and release of SUMO from target substrates. Given that CRN83_152 enhances SIZ1 E3 SUMO ligase activity, we hypothesize that the cell death triggered by AtSIZ1 upon transient expression in N. benthamiana is linked to its enzyme activity. Perhaps AtSIZ1 expression in a different plant species than Arabidopsis leads to mis-targeting of substrates, and subsequent activation of cell death. Although Mp64 did not enhance the cell death triggered by AtSIZ1, this effector did increase SIZ1 protein stability. Similarly, the effector AVR3a from Phytophthora infestans interacts with and stabilizes the E3 ubiquitin ligase CMPG1, likely by modifying its activity, to suppress plant immunity (Bos et al., 2010a). The mechanism underlying the stabilization of SIZ1 by Mp64 is yet unclear. However, we hypothesize that increased . Blots were prepared using total plant extracts of N. benthamiana infiltration sites expressing different SIZ1-myc versions or myc-GUS (control). Leaf material was harvested 2 d postinoculation after 1 h of heat stress in a 37°C incubator. Total protein amount was detected using the RevertÔ 700 Total Protein Stain followed by imaging in the 700 nm channel using an Odyssey ® CLx Imaging System. The panel indicated by 'loading' shows a proportion of the membrane that includes Rubisco. Detection of SIZ1-myc fusion proteins as well as RFP-AtSUMO1 upon antibody incubation was in the 800 nm channel using an Odyssey ® CLx Imaging System. Protein quantitation of SUMO-conjugates was done by normalizing the total band intensity of the area indicated to correspond to RFP-SUMO-conjugates to the total protein amounts using EMPIRIA STUDIO 2.1. RFP-SUMO1-conjugate levels in samples with SIZ1-myc were compared to myc-GUS (control, set at 1) to generate band intensity ratios, indicated by values below the Western blot incubated with RFP-antibodies. (b) Western blot showing levels of SUMO conjugates as in (a) in the presence of GFP-Mp64, GFP-CRN83_152_6D10 or GFP (control). Western blotting and protein detection as in (a) was used to detect the presence of SIZ1-myc and GFP-effector proteins and compare levels of RFP-SUMO1-conjugates. RFP-SUMO1-conjugate levels in samples with GFP-Mp64 or GFP-CRN83_152_6D10 were compared to GFP combined with myc-GUS (control, set at 1) to generate band intensity ratios, indicated by values below the Western blot incubated with RFP-antibodies.
Ó 2022 The Authors New Phytologist Ó 2022 New Phytologist Foundation New Phytologist (2022) 235: 1599-1614 www.newphytologist.com stability of SIZ1, which functions as an E3 SUMO ligase, leads to increased SUMOylation activity towards its substrates and will likely affect SIZ1 complex formation with other key regulators of plant immunity. SUMOylation indeed regulates key immune signalling components, as recently shown for NPR1 (nonexpressor of pathogenesis-related (PR) genes 1) (Saleh et al., 2015). Whether NPR1 is a substrate of SIZ1 remains to be investigated. SUMOylation of target proteins plays an important role in plant immunity and is known to be targeted as part of bacterial plant pathogen infection strategies (Verma et al., 2018). For example, effector XopD from Xanthomonas campestris pv vesicatoria (Xcv) functions as a SUMO protease inside host cells to modulate host defence signalling (Hotson et al., 2003;. SUMOylation sites are predicted in Mp64 (Fig. S5) and CRN83_152 (1 SUMO interaction motif: SVEKGANILSVEVPGCDVD; SUMOylation site: VKMLIE-VKREVKSAS) using prediction software GPS-SUMO (Zhao et al., 2014). However, using similar assays with RFP-SUMO1 described in this study, we have not detected any SUMOylation forms of Mp64 and CRN83_152, suggesting that these effectors are not SUMOylation targets themselves. Overall, our data suggest that modification of host SUMOylation is a common strategy of plant parasites to enable host colonization, and that the targeting strategies have evolved independently in distinct plantfeeding organisms including herbivorous insects. A detailed analyses of changes in the SIZ1-dependent host plant SUMOylome and their impact on susceptibility and immunity is needed to understand how distinct plant parasites promote virulence through SIZ1 targeting.

Supporting Information
Additional Supporting Information may be found online in the Supporting Information section at the end of the article.