Turgor-dependent and coronin-mediated F-actin dynamics drive septin disc-to-ring remodeling in the blast fungus Magnaporthe oryzae

The fungus Magnaporthe oryzae causes a devastating disease of cultivated rice called blast (Yan and Talbot, 2016), which threatens food security across the globe (Skamnioti and Gurr, 2009). M. oryzae uses a specialized pressure-generating infection cell called an appressorium to physically break into rice leaves (Talbot, 2019). Importantly, normal appressorium function is dependent on the formation of a septin ring, which assembles in its base (Dagdas et al., 2012). Septins are a broadly conserved family of GTP-binding cytoskeletal proteins, though absent in higher plants (Pan et al., 2007), which self-assemble into linear hetero-oligomeric rods and filaments through diffusion-driven annealing on membranes (Bridges et al., 2014). These filaments are organized into higher-order structures, including rings and gauzes at the cell cortex, by mechanisms that remain incompletely understood (Bridges and Gladfelter, 2015; Marquardt et al., 2019). Septins function in diverse cellular processes in eukaryotic cells and tissues including membrane remodeling (Beber et al., 2019), cell polarity (Berepiki and Read, 2013) and exocytosis (Tokhtaeva et al., 2015), and septin dysfunction is linked to numerous diseases in humans including cancer (Angelis and Spiliotis, 2016) and neurodegenerative disorders (Peterson and Petty, 2010). In M. oryzae, the appressorium septin ring provides cortical rigidity during pressure production and serves as a diffusion barrier to organize polarity determinants at the appressorium base (Dagdas et al., 2012; Ryder et al., 2013, 2019). Despite its critical role in role in plant infection, little is understood about how the appressorium septin ring forms in time and subcellular space. Importantly, improved understanding of this process in M. oryzae could provide new clues into the regulation of septin assembly in diverse systems (Bridges and Gladfelter, 2015). Previously, we showed that septin structures undergo remodeling from an incipient disc to a toroidal ring during appressorium development (Dulal et al., 2020). Here, we provide evidence that this septin disc-to-ring transitions is driven, in part, by dynamic remodeling of the F-actin cytoskeleton in M. oryzae.

To learn more about how septin rings form during appressorium development, and to better understand the role of the F-actin cytoskeleton in this process, we simultaneously imaged the septin (Sep5­–GFP) and F-actin (Lifeact–RFP) cytoskeletons over a period of 12 h during appressorium development (Fig. 1A,B). Imaging revealed that an F-actin ring forms around the circumference of an incipient septin disc soon after septation in the appressorium neck (Saunders et al., 2010b; Dulal et al., 2020) (Fig. 1A; Movie 1). Following contraction of the septin disc, the F-actin cytoskeleton was reorganized around the emerging septin ring (Fig. 1A–C). Septin ring formation is dependent on the appressorium reaching a minimum pressure threshold (Ryder et al., 2019), which requires the biosynthesis and deposition of melanin within the appressorium cell wall (Chumley and Valent, 1990). Melanin-deficient mutants cannot generate high pressure and are non-pathogenic (Chumley and Valent, 1990). We wondered at what point septin ring formation aborts in the absence of turgor, and what this might reveal about the progression of septin ring development. We therefore deleted BUF1, encoding for a tetrahydroxynaphthalene reductase that catalyzes the second-to-last step in melanin biosynthesis (Vidal-Cros et al., 1994), and imaged the F-actin and septin cytoskeletons during appressorium morphogenesis. Interestingly, Δbuf1 mutants formed normal septin discs, but these stalled and did not undergo contraction and remodeling (Fig. 1D,E). Strikingly, F-actin rings failed to form in Δbuf1 mutants and instead, Lifeact–RFP-labeled puncta decorated the periphery of the appressorium (Fig. 1E, lower panels). Similar outcomes were observed when melanin biosynthesis was inhibited chemically or when intracellular turgor production was perturbed by the application of exogenous glycerol to developing appressoria (Fig. S1) (Ryder et al., 2019). Taken together, our data support the idea that pressure-dependent F-actin ring formation and subsequent ring contraction initiates the remodeling of septin discs into toroidal rings during appressorium development.

We hypothesized that remodeling of the appressorium septin disc is driven by F-actin ring contractility (Fig. 2A) and reasoned that disruption of F-actin would prevent septin disc contraction and remodeling. To test this idea, we added the actin-depolymerizing agent latrunculin A (LatA), or DMSO (carrier control), to conidia undergoing appressorium development shortly after the emergence of incipient septin discs, but before the formation of F-actin rings (Fig. 2B,C). Strikingly, LatA-mediated disruption of F-actin caused septin discs to fragment (Fig. 2B,H; Movie 2) and, consistent with studies in the fungus Ashbya gossypii (DeMay et al., 2009), resulted in the emergence of numerous septin rings and puncta within conidia (Fig. 2C). In contrast, LatA-mediated disruption of F-actin in mature appressoria had little effect on septin structure (Fig. 2D). Thus, an intact F-actin cytoskeleton is essential for the integrity of transient disc-like septin assemblies but is dispensable for the maintenance of fully formed rings.

Microtubules play a pivotal role in positioning the actomyosin contractile ring during cytokinesis in animal cells (Pollard and O'Shaughnessy, 2019). We wondered whether they also promote F-actin ring formation, or positioning, during septin ring genesis in M. oryzae. Simultaneous imaging of microtubules (β-tubulin–GFP) and F-actin (Lifeact–RFP) revealed the presence of microtubule loops at the base of appressoria prior to F-actin ring formation and thickening (Fig. 2E). Furthermore, benomyl-mediated microtubule disruption impaired F-actin ring formation and contraction (Fig. 2F upper panels; Movie 3). Interestingly, microtubules have been shown to contribute to the assembly and closure of actomyosin contractile rings during wound closure in frog oocytes (Mandato and Bement, 2003), and might play a similar role during appressorium morphogenesis.

Septins organize microtubule arrays in polarizing cell types (Bowen et al., 2011). We were curious as to the spatial relationship and interdependency between microtubules and septins during appressorium repolarization. Fluorescence imaging revealed that appressoria contained vertically orientated polarized microtubule arrays, with their plus ends in close proximity to cortical septin structures (Fig. 2G upper panels and Fig. 2I). We speculate that septins might stabilize appressorium microtubules through modulation of their plus end dynamics (Bowen et al., 2011; Spiliotis, 2018; Nakos et al., 2019). Benomyl-mediated microtubule disruption resulted in fragmentation of the septin disc, which dispersed over the course of 12 h (Fig. 2G lower panels). Interestingly, intact microtubules are similarly required for the stability of cortical septin discs during interphase in human non-adherent K562 cells (Sellin et al., 2011) but are dispensable for septin assembly and organization in A. gossypii (DeMay et al., 2009). Importantly, deletion of SEP3 (also known as CDC3), resulted in mutants that produced structurally distorted appressoria that, consistent with previous studies (Dagdas et al., 2012), were unable to assemble cortical septin structures (Fig. 2J). In the absence of a basal septin structure, Δsep3 mutants retained the ability to generate vertical microtubule arrays, although these appeared less organized than those in the wild-type strain (Fig. 2J). Thus, there is a high degree of functional co-operation between the microtubule, actin and septin cytoskeletons during appressorium morphogenesis by M. oryzae.

To gain new insight into the control of septin ring assembly we genetically tagged Sep3 with TurboID (TbID; Branon et al., 2018), and identified putative proximal interacting proteins through streptavidin-mediated pulldown and mass spectrometry (Fig. 3A,B). Using this approach, we identified 220 proteins, 63 of which were identified exclusively in the Sep3–TbID sample including each of the core septins, as well as α-tubulin isoforms and the F-actin-modulating proteins cofilin and coronin (Fig. 3B; Table S1). Coronins can crosslink both microtubules and actin (Goode et al., 1999; Rothenberg et al., 2003), and facilitate the remodeling of F-actin networks for contractility (Michael et al., 2016). We were therefore curious as to whether M. oryzae coronin (Crn1) plays a role in microtubule and actin-dependent septin ring formation during appressorium morphogenesis. Additionally, Crn1 has previously been implicated in the organization of F-actin within appressoria and is required for pathogenesis (Li et al., 2019).

To validate Crn1 as a septin-proximal protein, we generated a strain co-expressing Sep5–RFP and Crn1–GFP and determined the relative localization of these fluorescent fusion proteins during appressorium development (Fig. 3C,D). During the isotropic expansion of the appressorium, Crn1–GFP localized to discrete puncta, representing endocytic actin patches (Li et al., 2019), which were uniformly distributed around the periphery of the cell, but largely excluded from the septin disc at the base (Fig. 3C). Strikingly, as appressorium development progressed, Crn1–GFP transitioned from the appressorium periphery to the base, in close proximity to Sep5–RFP-labeled septin structures (Fig. 3C,D; Movie 4). We propose that this shift in Crn1–GFP localization reflects both the initiation of polarized endocytosis from the appressorium base and the association of Crn1–GFP with the F-actin cytoskeleton (Fig. 3E). Interestingly, Crn1–GFP only sporadically decorated contractile F-actin rings. However, at later stages of appressorium development, when F-actin filaments were more densely bundled and associated with the incipient septin ring, colocalization with Crn1–GFP was more evident (Fig. 3E; Movie 5). Consistent with these observations, coronin similarly localizes to the F-actin contractile ring during cytokinesis in fission yeast and Dictyostelium (Fukui et al., 1999; Pelham and Chang, 2002).

Septins act as diffusion barriers that spatially organize membrane-associated proteins and compartmentalize the plasma membrane (Bridges and Gladfelter, 2015). In M. oryzae, proteins involved in generating membrane curvature and actin nucleation mislocalize at the base of the appressorium in septin deletion mutants (Dagdas et al., 2012). We wondered whether septin rings organize Crn1–GFP-labeled actin patches in mature appressoria and whether we could detect quantitative changes in the diffusive behavior of these endocytic complexes in their absence. We therefore tracked Crn1–GFP puncta at the base of mature appressoria in the wild-type strain and in Δbuf1 mutants, in which septin discs are largely dispersed by 24 h post inoculation (hpi; Fig. 3F,G). Interestingly, we found the mean square displacement of Crn1–GFP puncta in Δbuf1 mutants to be significantly larger than in the wild-type strain, consistent with the idea that septin rings modulate the diffusion of Crn1–GFP-labeled actin patches (Fig. 3H,I). Thus, Crn1 is transiently in close proximity to both the septin and F-actin cytoskeletons during disc-to-ring remodeling during appressorium development.

To test the role of Crn1 in appressorium morphogenesis we deleted the Crn1-encoding gene and imaged F-actin and septin dynamics in Δcrn1 mutants and in the wild-type control. The Δcrn1 mutants formed F-actin rings around incipient septin discs in a manner indistinguishable from wild-type appressoria (Fig. 4A). Furthermore, septin discs underwent similar rates of contractions to those of the wild-type control in the genetic absence of Crn1 (Fig. 4B). Strikingly, however, we found that Δcrn1 mutants were more likely to undergo ‘catastrophes’ during the reorganization of F-actin onto the surface of incipient septin rings (wild-type control, 9.1%±2.0; Δcrn1, 24.4%±4.0; mean±s.d.) (Fig. 4C). During these catastrophes, Lifeact–RFP-labeled F-actin rapidly fragmented, resulting in structural disruption to the septin cytoskeleton (Fig. 4D,E; Movie 6). Consistent with this observation, when we imaged mature appressoria (24 hpi) we found that in Δcrn1 mutants, F-actin was more often disorganized and only loosely associated with septin rings compared with that in the wild-type strain (wild-type control, 9.1%±2.1; Δcrn1, 33.3%±5.2; mean±s.d.) (Fig. 4F,G). Furthermore, Sep5-labeled septin structures in Δcrn1 mutants were more heterogeneous in their morphology and often lacked a well-formed central pore (Fig. 4G,H). Thus, in the genetic absence Crn1, septin-scaffolded F-actin remodeling is more prone to failure, resulting in concomitant disruption of septin ring formation. Taken together, our data support a model in which F-actin rings are recruited, in a pressure-dependent manner, to incipient septin discs to initiate their contraction. Following contraction, F-actin is remodeled in a coronin-mediated manner along the surface of the septin disc to promote ring formation. In conclusion, our study highlights the utility of M. oryzae as a model for understanding septin biology and provides evidence that regulated F-actin dynamics drive septin remodeling during appressorium development.

M. oryzae strains were cultured and stored using standard procedures (Crawford et al., 1986) and media prepared as previously described (Talbot et al., 1993). M. oryzae plate cultures were maintained on complete medium (CM) and incubated at 25°C under a 12 h:12 h photoperiod for 10–12 days. Desiccated filter stocks, stored at −20°C, were used to regenerate strains after no more than two rounds of subculture. For fluorescence microscopy-based experiments, conidia were isolated from 12-day-old plate cultures in sterile water, passed through two-layers of Miracloth (EMD Millipore) and washed twice by centrifugation (5000 g for 5 min) to remove hyphal debris and traces of medium. Conidia were enumerated on a hemocytometer and resuspended to a concentration of 5×10⁴ conidia/ml in a total volume of 350 μl. Conidial suspensions were pipetted into 8-well Nunc Lab-Tek chambers (Thermo Scientific) and left undisturbed for ∼30 min to allow the adherence of conidia to the borosilicate coverglass.

DNA constructs for targeted gene replacements and protein tagging were assembled using either In-Fusion cloning (Clontech Laboratories) or yeast gap repair (Orr-Weaver et al., 1983) from linear PCR products amplified using high-fidelity Phusion polymerase (New England Biolabs). Oligonucleotides were designed in SnapGene (version 4.3.10, GSL Biotech) and genomic DNA sequences were retrieved from the M. oryzae database (http://fungi.ensembl.org/Magnaporthe_oryzae/Info/Index).

Sep5–GFP:BAR (Dagdas et al., 2012), Lifeact–RFP:ILV1 (Berepiki et al., 2010; Dagdas et al., 2012) and β-tubulin–GFP:ILV1 (Saunders et al., 2010a) plasmids were obtained from Nick Talbot (The Sainsbury Laboratory, Norwich, UK). The Sep5–RFP:BAR plasmid, in which the RFP-encoding gene is fused in-frame at the 3′ end of SEP5 (MGG_03087), was generated by replacing the GFP-encoding gene of the Sep5–GFP plasmid (Dagdas et al., 2012) with the RFP-encoding gene derived from the Lifeact–RFP:ILV1 plasmid (Berepiki et al., 2010; Dagdas et al., 2012). Similarly, to generate the Lifeact–RFP:BAR plasmid, a fragment encompassing the Neurospora crassa ccg-1 promotor, the Lifeact peptide sequence and the RFP-encoding sequence, was amplified from the Lifeact–RFP:ILV1 plasmid (Berepiki et al., 2010; Dagdas et al., 2012) and cloned into plasmid pCB1265 containing the BAR cassette (Sweigard et al., 1997). The Crn1–GFP plasmid harboring a bleomycin-resistance cassette (Li et al., 2019), was obtained from Zengguang Zhang (Nanjing Agricultural University, Nanjing, China).

To generate the EB1–RFP construct an ∼3.4 kb fragment encompassing the EB1 open reading frame (ORF) (MGG_05427) and 2 kb of putative upstream promoter sequence was fused in-frame with the RFP-encoding gene derived from the Lifeact-RFP:ILV1 plasmid (Berepiki et al., 2010; Dagdas et al., 2012), followed by ∼0.3 kb of sequence corresponding to the EB1 3′ untranslated region, in a yeast gap repair plasmid harboring a bleomycin-resistance cassette.

To generate the Sep3–TurboID-V5 construct, a ∼1.5 kb fragment of the SEP3 ORF (MGG_01521) was fused in-frame with the TurboID-V5-encoding sequence (Branon et al., 2018), amplified from plasmid pRS415 (Addgene), followed by a hygromycin resistance cassette amplified from pCB1004 (Sweigard et al., 1997), and an additional ∼1 kb of targeting sequence downstream of the SEP3 ORF, in a yeast gap-repair plasmid.

To generate the CRN1 gene replacement construct, two ∼1 kb fragments flanking the CRN1 ORF (MGG_06389) were amplified from gDNA and assembled at either end of a hygromycin resistance cassette (HYG), amplified from plasmid pCB1004 (Sweigard et al., 1997), in a yeast gap-repair plasmid.

To generate the BUF1 gene replacement construct, two ∼1 kb fragments flanking the BUF1 ORF (MGG_02252) were amplified from gDNA and assembled at either end of a hygromycin resistance cassette (HYG), amplified from plasmid pCB1004 (Sweigard et al., 1997), in a yeast gap-repair plasmid.

Polythene glycol-mediated transformation of M. oryzae protoplasts was performed using an established protocol (Talbot and Talbot, 2001). A split marker strategy was used to generate the Δcrn1 and Δbuf1 mutants (Catlett et al., 2003). Briefly, protoplasts of the appropriate entry strains were simultaneously transformed with two separate PCR products, amplified from the respective full-length gene replacement constructs, representing 1 kb of either upstream or downstream targeting sequence fused to complementary halves of the hygromycin resistance cassette sequence (HYG). A similar split HYG approach was used to target the Sep3–TbID-V5 construct to the endogenous Sep3 locus. Homologous integration of the gene-deletion and fusion constructs was confirmed by diagnostic PCR using primers upstream and downstream of the expected integration sites and outside of the constructs themselves. For the generation of strains expressing fluorescent fusion proteins, protoplasts were transformed with plasmids that integrated into the genome ectopically.

Fluorescence images were acquired on an inverted Nikon Ti-E Eclipse microscope equipped with a 100×1.49 N.A oil immersion Apo TIRF Nikon objective, a Perfect Focus System (Nikon), a iXon Ultra 897 electron multiplier CCD Camera (Andor Technology) and an AURA II solid-state triggered illuminator with a four-channel light source (395 nm, 485 nm, 560 nm and 640 nm), all controlled by NIS-Elements AR (version 4.60). For comparative timecourse experiments, treatments were imaged simultaneously using multi-well Labtek chambers (Thermo Fisher, Pittsburgh, PA). Acquired 3D and 4D data sets were deconvolved in NIS-Elements AR with spherical aberration correction and background subtraction, using the ‘Automatic’ 3D deconvolution option. Brightness and contrast adjustments to maximum intensity projections of deconvolved 3D images were made in Imaris (9.5.1; Bitplane), ImageJ (version 2.0; National Institutes of Health) and Photoshop CC (2017.1.4; Adobe), and figures were compiled in Illustrator CC (22.1; Adobe). Unless otherwise stated in the figure legend, micrographs represent maximum intensity projections of z series acquired at 0.2 μm intervals spanning the entire depth of the appressorium. Movies were generated in Imaris (9.5.1; Bitplane) and annotated in Premiere Pro 2020 (Version 14.0.0; Adobe).

Appressoria forming in multi-well Labtek chambers (Thermo Fisher, Pittsburgh, PA) were monitored by fluorescence microscopy, and when Sep5–GFP-labeled septa emerged at the germ tube–appressorium interface, either dimethyl sulfoxide (DMSO; 1%, Sigma-Aldrich), latrunculin A (LatA; 100 μM, MiliporeSigma), benomyl (Ben; 5 μg/ml, Sigma-Aldrich) or glycerol (1.5 M, Chem-Impex) were added and carefully mixed by micropipette. Tricyclazole (100 μg/ml, Sigma-Aldrich) was added to conidial suspensions at 0 hpi. Drug and chemical effects were monitored and imaged by 4D fluorescence microscopy.

Line scans were performed in Fiji (ImageJ) using the line tool and ‘plot profile’ function. Briefly, 10 μm lines were drawn across the diameter of Sep–GFP-labeled septin structures, perpendicular to the position of the germ tube, in single focal planes of fluorescence micrographs. Fluorescence intensity values were exported to Excel (Microsoft, version 15.24) for normalization, and values were plotted in Prism 8 (GraphPad, version 8.2.1).

Genomic DNA was extracted from fungal mycelia using the Cetyl Trimethyl Ammonium Bromide (CTAB) method, as described previously (Talbot and Talbot, 2001). Polymerase chain reaction, restriction enzyme digests and gel electrophoresis were carried out under standard procedures (Sambrook and Russell, 2001). Purification of PCR-amplified DNA was performed using Wizard SV Gel and PCR Clean-up system (Promega), and plasmids were isolated using Wizard Plus SV Miniprep DNA purification System (Promega).

The acquired movies were cropped using ImageJ such that each cropped movie contained a single appressorium. The cropped movies were then analyzed by the trackpy Python package (Trackpy v0.4.2), from which the positions of coronin puncta were obtained. The positions were linked, using trackpy, into trajectories ⁠, following standard algorithms (Crocker and Grier, 1996; Manley et al., 2008; Sadoon and Wang, 2018; Sadoon et al., 2020) with a search range of three pixels and a memory of one frame. From the trajectories, the ensemble mean square displacements (MSD) were calculated using built-in functions in trackpy and the following equation: ⁠, where τ is the lag time. For each sample, the MSD data were averaged over different appressoria from multiple movies. The averaged MSD data were fitted with a line, MSD=4Dτ, where D is the apparent diffusion coefficient, using the curve_fit function of the scipy.optimize Python package. The standard deviation errors of the fitted D values were estimated by the square root of the diagonal elements of the covariance matrix of the linear regression.

Cell lysates were separated by SDS–PAGE (NuPAGE 4–12% Bis-Tris gels, Thermo Fisher Scientific) and transferred to pre-cut nitrocellulose blotting membranes using the XCell II blot module (Thermo Fisher Scientific). Membranes were blocked in either TBS+5% bovine serum albumin (for streptavidin–HRP) or TBST (TBS containing 0.05% Tween 20) with 5% milk for 30 min. Membranes were probed with either streptavidin–HRP conjugate (Thermo Fisher Scientific, catalog no. SA1001; 1:5000 dilution), or anti-V5 primary antibody (Thermo Fisher Scientific, R960-25; 1:5000 dilution) followed by HRP-conjugated goat anti-mouse IgG (H+L) secondary (Thermo Fisher Scientific, 62-6520; 1:10,000 dilution). Western blots were imaged using SuperSignal West Pico chemiluminescence substrate (Thermo Fisher Scientific) and an ImageQuant LAS 500 gel documentation system (GE Healthcare).

Mycelia from the outer edge of 10-day-old plate cultures of the wild-type (Guy11) and Sep3–TbID-V5-expressing strains were homogenized in 50 ml of liquid CM using a laboratory blender (Waring Commerical). Liquid cultures were supplemented or not with 100 μM biotin (BIO-500, Avidity, USA) and incubated for 48 h at 28°C with shaking (250 rpm). Mycelia were harvested by filtering through Miracloth (MiliporeSigma), washed thoroughly with sterile distilled water, and lyophilized overnight in a freeze dryer (Labcoco). For cell lysis, 1 ml of RIPA lysis and extraction buffer (25 mM Tris-HCl pH7.6, 150 mM NaCl, 1% NP-40, 1% sodium deoxycholate and 0.1% SDS; Thermo Scientific), supplemented with Halt protease inhibitor cocktail and EDTA (Thermo Scientific), was added per 40 mg of freeze-dried mycelium. Samples were briefly vortexed with acid-washed glass beads (425–600 µm; MiliporeSigma) then incubated at 4°C with gentle rocking for 10–20 min. Cell lysates were clarified by centrifugation at 20,010 g for 1 h in an HB-6 rotor (Thermo Scientifc Sorvall) at 4°C. Seventy microliters of streptavidin-conjugated beads (Dynabeads MyOne Streptavidin T1, Thermo Fisher Scientific) were added to 400 µl of clarified cell lysates, and samples were incubated overnight at 4°C with gentle rocking. Streptavidin-conjugated beads were immobilized on a magnetic stand and washed three times with 2 µl of RIPA buffer. Bead-bound proteins were eluted by boiling in 70 µl of elution buffer (50 mM Tris-HCl, pH 6.8, 2% SDS, 20 mM DTT and 12.5 mM EDTA). Eluted proteins were separated by SDS PAGE (NuPAGE 4-12% Bis-Tris gels, Thermo Fisher Scientific) and stained with Coomassie Brilliant Blue.

In-gel trypsin digestion and mass spectrometry were performed as previously described (Singh et al., 2020). Each lane was cut into five pieces, each covering a range of molecular weights. To increase their surface area and aid in trypsin digestion, each of these sections was further sliced into ∼1 mm² pieces. Once the gel pieces were destained with 50% acetonitrile (ACN) in 25 mM ammonium bicarbonate, they were dehydrated with pure HPLC-grade acetonitrile and completely dried with a SpeedVac. Proteins on the gel were reduced by incubating the dried gel slices with 10 mM DTT in 25 mM ammonium bicarbonate (pH 7.8) at 60°C. After 1 h, excess DTT was pipetted off. To alkylate the reduced proteins, 20 mM iodoacetamide in 25 mM ammonium bicarbonate was added to the gel pieces and incubated at room temperature for 1 h in the dark. Gel pieces were then thoroughly washed with 25 mM ammonium bicarbonate and dehydrated and dried again, as above. The digestion process was started by adding ∼100 μl MS-grade (Thermo Fisher Scientific) trypsin, enough to cover the dried gel pieces, which was at 10 nM concentration constituted in 25 mM ammonium bicarbonate. To maximize trypsin absorption by the gel pieces, samples were kept at 4°C for 30 min. After adding an additional 100 μl of 25 mM ammonium bicarbonate to each sample, they were incubated at 37°C for 24 h. During the digestion, tryptic peptides diffused out into the solution. Gel pieces were then extracted three times using 50% ACN in 5% formic acid (FA) solution, and all extracts were pooled. Pooled extracts were evaporated to dryness and reconstituted in 50 μl of 0.1% FA before being subjected to liquid chromatography-tandem mass spectrometry (LC-MS/MS).

After trypsin digestion, LC-MS/MS was used to analyze 5 μl of the in-gel tryptic peptide extracts from each gel section, using an Agilent 1200 series microflow high-performance liquid chromatography (HPLC) system coupled to a Bruker amaZon SL quadrupole ion trap mass spectrometer with a captive spray ionization source. Separation of the tryptic peptides in samples corresponding to each gel section was achieved by reverse-phase high-performance liquid chromatography (RP-HPLC). A capillary column with a Zorbax SB C18 column (150×0.3 mm, 3.5 μm particle size, 300 Å pore size, Agilent Technologies) was used as the HPLC column. HPLC solvent gradient of 5 to 38% solvent B in 320 min at a flow rate of 4 μl/min was established using 0.1% FA in water (solvent A) and 0.1% FA in ACN (solvent B). MS analyses were performed in a positive ion mode using a Bruker captive electrospray source. Dry nitrogen gas temperature of 200°C and a nitrogen flow rate of 3 liters/min were used during the LC-MS/MS run. LC-MS/MS data acquisition was carried out in the Auto MS(n) mode. The optimized trapping condition for the ions at m/z 1000 was set. While regular MS level scans were performed using the enhanced scanning mode (8100 m/z/second), MS/MS fragmentation (collision induced fragmentation) scans were performed automatically for the top ten precursor ions with a set threshold for 1 min using UltraScan mode (32,500 m/z/second) (Karash et al., 2017). Bruker DataAnalysis 4.0 software was used to pick peaks from the LC-MS/MS chromatogram using a default setting, as recommended by the manufacturer, to create Protein Analysis Results.xml files. These files were then used to search proteins in the UniProtKB/trEMBL_Magnaporthe_oryzae database, made up of 12755 protein entries, using MASCOT v 2.2 search engine (Matrix Science, London, United Kingdom). The parent ion and fragment ion mass tolerances were both set at 0.6 Da with cysteine carbamidomethylation and methionine oxidation as fixed and variable modifications, respectively, in MASCOT search. Mascot.dat files were then exported into Scaffold Proteome Software version 4.8 (http://www.proteomesoftware.com). Mascot.dat files from sections of the wild-type control and Sep3–TbID gels were combined separately to create two groups for the comparison of identified proteins. To be accepted, protein identifications had to be represented by two or more unique peptides with 99.9% or greater protein identification probability threshold and a peptide identification probability greater than 99.9%. The Protein Prophet algorithm was used to assign protein probabilities (Keller et al., 2002).

No statistical methods were used to pre-determine sample size. All statistical significance testing was performed in Prism 8 (GraphPad, version 8.2.1), and for all tests *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001.

We thank Zhengguang Zhang (Nanjing Agricultural University) for sharing the Crn1:GFP plasmid, Nick Talbot (The Sainsbury Laboratory) for helpful discussion, and the University of Arkansas Statewide Mass Spectrometry Facility for LC-MS/MS work.

The peer review history is available online at https://jcs.biologists.org/lookup/doi/10.1242/jcs.251298.reviewer-comments.pdf