Simple and efficient heterologous expression of necrosis‐inducing effectors using the model plant Nicotiana benthamiana

Abstract Plant fungal pathogens cause devastating diseases on cereal plants and threaten global food security. During infection, these pathogens secrete proteinaceous effectors that promote disease. Some of these effectors from necrotrophic plant pathogens induce a cell death response (necrosis), which facilitates pathogen growth in planta. Characterization of these effectors typically requires heterologous expression, and microbial expression systems such as bacteria and yeast are the predominantly used. However, microbial expression systems often require optimization for any given effector and are, in general, not suitable for effectors involving cysteine bridges and posttranslational modifications for activity. Here, we describe a simple and efficient method for expressing such effectors in the model plant Nicotiana benthamiana. Briefly, an effector protein is transiently expressed and secreted into the apoplast of N. benthamiana by Agrobacterium‐mediated infiltration. Two to three days subsequent to agroinfiltration, the apoplast from the infiltrated leaves is extracted and can be directly used for phenotyping on host plants. The efficacy of this approach was demonstrated by expressing the ToxA, Tox3, and Tox1 necrosis‐inducing effectors from Parastagonospora nodorum. All three effectors produced in N. benthamiana were capable of inducing necrosis in wheat lines, and two of three showed visible bands on Coomassie‐stained gel. These data suggest that N. benthamiana–agroinfiltration system is a feasible tool to obtain fungal effectors, especially those that require disulfide bonds and posttranslational modifications. Furthermore, due to the low number of proteins typically observed in the apoplast (compared with intracellular), this simple and high‐throughput approach circumvents the requirement to lyse cells and further purifies the target proteins that are required in other heterologous systems. Because of its simplicity and potential for high‐throughput, this method is highly amenable to the phenotyping of candidate protein effectors on host plants.


| INTRODUCTION
Cereal production is susceptible to severe yield losses due to plant pathogens. During infection, plant pathogens secrete a repertoire of proteins called effectors into the apoplast or cytoplasm of the host cells to modulate plant processes and promote infection (Dodds & Rathjen, 2010). Although these effectors underpin key disease mechanisms, many can be recognized by the plant, and the outcome of this interaction is broadly dependent on the lifecycle of the specific pathogen. In biotrophic pathogens, effectors can be recognized by specific host receptors, triggering programmed cell death at the site of infection that results in plant resistance against these pathogens (Dodds & Rathjen, 2010). In contrast, recent findings suggest that some necrotrophic pathogens, such as Parastagonospora nodorum, hijack these host resistance pathways for their own benefit (McDonald & Solomon, 2018). Necrotrophic effectors (NEs) induce cell death only in the presence of host dominant susceptibility genes that strikingly resemble typical defense receptors (Faris et al., 2010;Lorang et al., 2012;Shi et al., 2016). Recognition, and the subsequent host cell death, is favored by necrotrophic pathogens leading to plant susceptibility instead of resistance (Faris & Friesen, 2020). To counter this, researchers have often turned to expressing their effectors of interest in model plant systems, such as N. benthamiana, as many effectors and their host cognate receptors are active in such systems (Chen et al., 2017;Ma et al., 2012;Van Der Hoorn et al., 2000;Vleeshouwers et al., 2008). In these cases, the transient expression of the effector with the corresponding host receptor is often sufficient to produce a cell death phenotype that allows functional characterization of their interaction (e.g., determining required amino acids for recognition, etc.). However, this approach does not appear to work for NEs (particularly from monocotyledon cereal pathogens), and their dominant susceptibility genes and co-expression in model systems such as N. benthamiana does not lead to a visible cell death. Therefore, this lack of a visible phenotype precludes this coexpression approach from being used to study these interactions (as it can be with many biotrophic effectors). Consequently, as an alternative approach, NEs are often expressed using heterologous systems for characterization and subsequent phenotyping on host plants (Liu et al., 2009(Liu et al., , 2012Outram et al., 2021;Sung et al., 2021;Zhang et al., 2017). Almost exclusively, NEs have been expressed by using microbial expression systems such as Escherichia coli and Pichia pastoris.
Many known and predicted effectors are small cysteine-rich (SCR) proteins (Duplessis et al., 2011;Saunders et al., 2012;Sperschneider et al., 2015;Stergiopoulos & De Wit, 2009). It has been demonstrated that the cysteines in many of these effectors form disulfide bonds providing stability for these proteins in the apoplast (Saunders et al., 2012). For instance, the Tox3 and Tox1 effectors from P. nodorum contain six and 16 cysteines respectively, and at least one disulfide bond is required for their necrosis-inducing activity (Liu et al., 2009(Liu et al., , 2012Outram et al., 2021). Similarly, Cladosporium fulvum effector Avr2 has four disulfide bonds that are essential for its function of inhibiting cysteine protease Rcr3 in tomato plants (Rooney et al., 2005). Moreover, some effectors such as ToxA and Tox3 appear to contain pro-domains subsequent to their secretion signals that are thought to be important for folding and are processed prior to secretion from the fungi (Liu et al., 2009;Outram et al., 2021;Tuori et al., 2000). Recently, Outram et al. (2021) reported that the prodomain of Tox3 was cleaved by KEX2 protease that lead to a processed protein with demonstrably higher activity indicating that some NEs require posttranslational modification and processing to become fully active. However, because of these requirements, such proteins are difficult to express using a bacterial system due to the absence of eukaryotic cell organelles necessary for disulfide bond formation and posttranslational modifications. Lobstein et al. (2012) developed a new bacterial strain, E. coli SHuffle, which favors cytoplasmic disulfide bond formations. This new strain allowed researchers to express effectors harboring disulfide bonds in soluble forms Zhang et al., 2017). For example, one study demonstrated that the E. coli SHuffle system was successful in producing four out of five SCR effectors that were trialed . Subsequent functional studies on three of the effectors showed they were biologically active . However, yields were often low limiting subsequent functional characterization.
Another approach to express SCR effectors has been using secretion pathways in eukaryotic cells such as yeast. Using this method, the SCR effectors SnTox1 and SnTox3 were successfully expressed at low levels in secreted forms and showed activity in wheat lines harboring corresponding susceptibility genes (Liu et al., 2009(Liu et al., , 2012. However, in microbial expression systems, various physical parameters need to be adjusted to achieve optimal expression conditions that can vary among different strains and target proteins. For example, in the E. coli SHuffle system, the effect of temperature and inducer concentration were specific to a given protein when tested on seven different proteins requiring disulfide bonds (Lobstein et al., 2012). Likewise, in the P. pastoris system, methanol is the primary carbon source, and its concentration must be closely monitored depending on the rate of utilization by the cells; less methanol decreases productivity, whereas too much is toxic to cells (Karbalaei et al., 2020). In addition to these examples, many other parameters in both E. coli and P. pastoris most likely need optimization for every new target protein (Karbalaei et al., 2020;Lobstein et al., 2012), which can be time-consuming and laborious. This can be a significant bottleneck for laboratories who do not have substantial protein expertise or are not experienced with microbial expression systems.
Here, we describe a simple and high-throughput method for heterologous expression of cysteine-rich effectors using the secretion pathway of model plant N. benthamiana. We demonstrate the efficacy of this method by expressing three unrelated necrosis-inducing effectors (that differ in their cysteine content) from P. nodorum and subsequent investigation of their activity on the host plant. Our data show that this plant expression approach can be used for rapid on-host functional screening of candidate NEs, especially those that require disulfide bonds and posttranslational modifications. N. benthamiana plants were grown on 1/2 Debco seed-raising mix and 1/2 Martins Mix plus 2-gr 3-to 4-month mini osmocote per liter pot.

| Agroinfiltration assays
Agroinfiltration assays were carried out as described previously (Dagvadorj et al., 2017). Briefly, A. tumefaciens (GV3101, pMP90) carrying the effector constructs was streaked on LB-agar plates with antibiotics (Spectinomycin 100 μg/ml, Gentamycin 20 μg/ml and Rifampicin 100 μg/ml) and grown for 2-3 days at 28 C. The bacterial cells on the plate were collected by scraping with 1-ml pipette tip, transferred to new 1.5-ml microcentrifuge tube and resuspended in 1-ml distilled water. The cells were centrifuged at 4000 rpm for 5 min at room temperature. The cell pellet was washed in 1-ml fresh Agro-induction buffer (AI; 10-mM MES pH 5.6, 10-mM MgCl 2 ) and centrifuged again under the same conditions. Finally, the pellet was suspended in 1-ml AI buffer, and the cell suspension was diluted with AI buffer until the optical density (OD) of the cell suspension was adjusted to 0.2 for infiltration experiments. To enhance transient expression level, the cell suspensions were co-infiltrated with A. tumefaciens containing p19 RNA silencing suppressor construct (OD of .1). N. benthamiana 4-6 weeks of age was agro-infiltrated, and the leaves were harvested at 2-day post-agroinfiltration for apoplast wash fluid extraction.

| AWF extraction
Apoplast washing fluid (AWF) was extracted as previously described with minor modifications (O'Leary et al., 2014). N. benthamiana leaves expressing the effector constructs were detached and submerged in a beaker with Milli-Q water. The beaker was placed into a desiccator and a vacuum of À80 kPa applied for 1 min. By gradually releasing the vacuum, the air within the apoplastic space was infiltrated with the water, and the infiltrated area appeared dark translucent by naked eye. This process was repeated until the whole leaf was infiltrated.
Then, the leaves were wiped gently with clean tissue to remove the surface water, and then three to four leaves sandwiched together in parafilm sheets, rolled and placed into a 20-ml syringe, in a 50-ml Falcon tube. To isolate AWF, the Falcon tube was centrifuged at 500g for 10 min at 4 C. The collected AWF (400-500 μl/per leaf) was transferred to microcentrifuge tubes and centrifuged again at 17,000g for 5 min at 4 C to remove cell debris and other insoluble materials.
The supernatant was transferred to a fresh tube and used immediately or stored at À80 C.

| MS for ToxA and Tox3 protein identification
For peptide mapping of the ToxA and Tox3 proteins, AWF samples were separated on a 15% sodium dodecylsulfate-polyacrylamide gel electrophoresis (SDS-PAGE) gel and stained with Coomassie blue. The proteins of interest were excised and subjected to in-gel trypsin digestion as previously described . Subsequent mass spectrometry data were analyzed using Max-Quant software, v.1.6..8 (Tyanova et al., 2016). Tryptic peptides were mapped to the sequence of ToxA and Tox3 by using the trypsin enzyme digestion in semispecific mode.

| AWF infiltration into wheat leaves
Undiluted AWF samples (each $500 μl) were infiltrated using a needleless syringe into the second leaf of 2-week-old wheat cultivars until the infiltration zone reached 3-5 cm. The AWF infiltrations were repeated in at least five independent plants from either wheat cultivar with or without corresponding susceptibility gene. The wheat leaves were checked for the induction of necrosis from 1-day postinfiltration (dpi) and recorded at 3 dpi.

| RESULTS
To trial the expression of SCR proteins in N. benthamiana, the ToxA, Tox3 and Tox1 effector proteins from P. nodorum were chosen (Table 1). ToxA, Tox3 and Tox1 induce necrosis on wheat lines harboring the susceptibility genes Tsn1, Snn3 and Snn1, respectively. The mature forms of ToxA, Tox3 and Tox1 each contain two, six and 16 cysteine residues respectively, and their ability to induce necrosis is reliant on disulfide bond formation (Liu et al., 2012;Sarma et al., 2005;Zhang et al., 2017). In addition, ToxA and Tox3 require posttranslational modifications such as processing of pro-domain to become fully active Tuori et al., 2000).
To secrete the effector proteins to the apoplast of N. benthamiana, all three effectors were cloned with the NtPR1sp (Ohshima et al., 1987) as shown in Figure 1a. This modification promotes the proteins to follow correct secretion pathway in N. benthamiana (Breen et al., 2016;Grosse-Holz et al., 2018;Ma et al., 2015). The effectors were expressed under control of CaMV To test biological activity of the effectors, the AWFs containing each of ToxA, Tox3 and Tox1 were infiltrated into wheat cultivars carrying Tsn1, Snn3 and Snn1 susceptibility genes (Grandin, Corack and Calingari), respectively (Figure 2). Our results showed that the ToxA, Tox3 and Tox1 AWFs were able to induce necrosis in a genotype specific manner. Although we could not detect Tox1 band in protein gel analysis, we did observe chlorosis and necrosis development in the Snn1 wheat cultivar (Calingari) infiltrated with AWF with Tox1 ( Figure 2). A slight chlorotic response was also evident on the snn1 cultivar Corack upon infiltration of the Tox1 AWF although a minor nonspecific reaction for this effector has been previously noted when purified from E. coli . Collectively, these results suggest that N. benthamiana expressed and secreted ToxA, Tox3 and Tox1 are biologically active.

| DISCUSSION
In this study, we showed that fungal SCR effectors can be expressed and processed using the N. benthamiana-agroinfiltration system.
Three unrelated SCR effectors, differing in their cysteine content, were successfully produced as active forms. In comparison with other methods currently used in the field, the approach we presented here is simple and efficient, which may facilitate the identification of putative NEs by screening for cell death phenotype on host plants. This method may also be an alternative for researchers who struggle with expressing SCR effectors using microbial expression systems. In line with this, a recent study showed that using agroinfiltration, the apoplast targeted NLP2 (necrosis and ethylene-inducing peptide-like protein) from Peyronellaea pinodes was successfully expressed and induced necrosis in legume plants including field pea, faba bean and lentil, but not in chickpea (Debler et al., 2021). This further corroborates that plants can serve as a heterologous expression system to produce biologically active fungal effectors in secreted form.
This approach has several advantages compared with existing methods used to express NEs. Firstly, the plant signal peptide allows proteins to be trafficked through secretion pathways such as the endoplasmic reticulum and Golgi facilitating the formation of disulfide bonds and other posttranslational modifications. This facilitates the correct folding of SCR proteins, which are therefore more likely to be soluble compared with cytoplasmic-expressed proteins. Secondly, the apoplast can be separated mechanically from the plant cells, which enables the resolution of the mature proteins in the apoplast from the unprocessed forms inside the cell. This is not the case in cytosolically expressed proteins if crude extract is used for the functional studies.
Thirdly, the apoplast of the plant cell contains significantly less proteins and molecules than in the cytosol. Therefore, isolated apoplast wash fluid with target protein has much less background or contaminating proteins compared with proteins expressed in the cytosol for functional studies. Furthermore, the reduced protein content in the  F I G U R E 2 Apoplast washing fluid (AWF) containing the secreted ToxA, Tox3 and Tox1 trigger necrosis in wheat cultivars carrying the susceptibilty genes Grandin (Tsn1), Corack (Snn3) and Calingari (Snn1), respectively. Wheat cultivars Corack (tsn1, snn1) and BG261 (snn3) were used for genotype specificity controls, and AWF from N. benthamiana expressing GFP was used for AWF background control. The leaves were harvested and recorded 3 days after AWF infiltration apoplast simplifies subsequent purification of the target protein if required. Although yeast eukaryotic systems can provide these advantages, various parameters influence the optimal expression, which can differ for any given protein. Thus, the optimization process is more challenging than compared with the N. benthamiana-agroinfiltration system.
However, the limitations of this method must be acknowledged.
This approach may not be suitable for certain effectors, specifically those that trigger rapid cell death in N. benthamiana. Several studies reported that effectors caused cell death in N. benthamiana when expressed with their respective signal peptide (Dagvadorj et al., 2017;Fang et al., 2016;Kettles et al., 2017;Ma et al., 2015). For instance, 11 out of 119 putative effectors from Ustilaginoidea virens induced cell death in N. benthamiana; however, the cause of the cell death was unknown. In other studies, a group of effectors from Zymoseptoria tritici (Kettles et al., 2017), XEG1 effector from Phytophthora sojae (Ma et al., 2015) and PstSCR1 effector from Puccinia striiformis (Dagvadorj et al., 2017) triggered cell death in N. benthamiana, which appear to be dependent on immune receptors such as NbSER3/ NbBAK1. These results suggest that some effectors can activate basal immunity resulting the plant cell death. For these types of effectors, nonplant systems may be more suitable.
Our study shows that N. benthamiana-agroinfiltration system can serve as an expression system for fungal cysteine-rich effectors.
The expression of three unrelated effectors using this system displays production of high and biologically active effectors. This method can be upscaled and combined with a typical protein purification procedure to purify the protein of interest with little efforts.
More importantly, this system applicable for expressing almost any protein (with some exceptions) resides in an extracellular space of the cell.