EVs can be isolated from Z. tritici cultures
To determine if Z. tritici secretes EVs under in vitro culture conditions, as described for other fungi, we isolated EVs from Z. tritici broth cultures using a differential ultracentrifugation (DUC) method adapted from Thery et al., (2006) and Rodrigues et al., (2007). We observed particles resembling fungal and mammalian EVs, based on their comparable morphology and size distribution, using transmission electron microscopy (TEM) and nanoparticle tracking analysis (NTA) (Fig. 1). TEM analysis revealed a heterogeneous population of EVs ranging from < 50 nm to > 300 nm in size, with most particles falling between 50—150 nm and median and mean particle size of 84.0 nm and 91.8 nm (sd 39.9 nm), respectively. Some particles were > 300 nm but these were less frequently observed (Fig. 1a, b). NTA corroborated the TEM-derived EV population size distributions, revealing a dominant peak in the particle size distribution between approximately 100—250 nm (Fig. 1c). The apparent lack of smaller particles represented in the NTA size distribution when compared to the TEM measurements, may be artefactual of NTA systems’ known underreporting of particles < 50 nm and/or the dehydration and shrinkage caused by negative staining of EVs for TEM [21, 33]. Z. tritici EVs samples typically had a concentration of 0.12—0.15 µg of protein and 1.45—2.43 ×108 particles per mL of culture filtrate. While Z. tritici isolate WAI332 grown in the semi-defined medium Fries 3 was typically used for EV isolations, we found that EVs could also be isolated from the same isolate cultured in potato dextrose broth and minimal media and from another Australian isolate, WAI321 (Additional file 1: Fig. S1).
EVs are described as membrane-bound, cargo-containing particles released by cells. We wanted to confirm the EVs observed in Z. tritici cultures were the product of viable cells rather than membranous artefacts of senescing cells or cell debris. Z. tritici is a heat-labile fungus so we compared the presence of EVs in heat-treated cultures with that of untreated cultures. Z. tritici cells were heat-treated after 72 hours of growth and incubated in fresh culture media alongside viable cultures for a further 72 hours, until the latter were of equivalent biomass (Fig. 2, Additional file 2: Table S1). Heat-treated cells were not viable when plated on YSA (data not shown). TEM suggested heat-treated cultures produced samples rich in irregularly-shaped debris, with some EV-like particles observed (Fig. 2b, c). Comparatively, EVs were abundant in the non-heat-treated culture. Nanoparticle tracking analysis indicated particles in the 100—250 nm range were less concentrated in heat-treated cultures. EVs from untreated cultures had mean and modal sizes ranging from 135.7—168.0 nm and 109.7—134.1 nm from two biological replicates, respectively. Particles from heat-treated cultures ranged from 125.1—129.3 nm in average size and 93.0—102.5 nm in modal size (Fig. 2d; Additional file 1: Fig. S2). We also found EV-like particles were absent from the medium only control, confirming EVs were not an artefact of the semi-defined growth medium or isolation procedure (Fig. 2a, d).
Z. tritici EVs have a protein profile distinct from soluble secreted proteins in the broth culture filtrate
It is well established that EVs from mammals, fungi, plants and bacteria contain protein cargo so we aimed to profile the protein content of Z. tritici EVs released under in vitro growth conditions using bottom-up proteomics. From five biological replicate EV samples we identified 240 high-confidence (HC) proteins consistently detected using LC-MS/MS – these proteins were identified using a minimum of two unique peptides in at least 4/5 EV replicates (Table 1). The full list of proteins is provided in Additional file 2, Table S2. We made qualitative presence/absence comparisons of protein content in EV samples with soluble secreted proteins (SS); the SS fraction was prepared from the supernatant separated from EVs by UC at 100 000 x g. This supernatant was again centrifuged at 120 000 x g to remove remaining EVs and insoluble debris. LC-MS/MS analysis of the SS fraction identified 79 HC proteins (detected in at least 2/3 biological replicates).
Of the 240 HC EV proteins, 210 were present only in the EV samples, while 49 SS proteins were unique to SS samples and 30 proteins were shared between the fractions (Fig. 3a, Additional file 2: Table S3). Whether the presence of proteins in both fractions is due to insufficient purification of EVs, contamination from disrupted EVs, or the secretion of proteins via EVs and canonical secretion pathways will require further investigation. Given this, we have focused on HC proteins unique to either the EV or SS fractions; values reported will be in reference to these HC, fraction-specific protein groups.
To probe potential differences in the characteristics of proteins in the EV and SS fractions we identified proteins with canonical secretion signals (SPs) and transmembrane domains (TMDs) or glycosylphosphatidylinositol- (GPI) anchors (Table 1; Additional file 2: Table S2, S3). The proportion of proteins with canonical SPs was greater (77.6%) in the HC SS fraction than in EVs (6.7%). The presence of proteins without a SP in yeast and filamentous fungal culture filtrates has been reported and may result from non-canonical secretion mechanisms [15, 34, 35]. Proteins lacking SPs have been identified in proteomic analyses of both in planta and in vitro Z. tritici secretomes [19, 20, 36]. EV proteins largely lacked SPs, aligning with previous analyses of fungal EV protein cargo [37]. EVs had more predicted membrane-associated proteins (7.1%) compared to the secreted fraction (2.0%).
We also aimed to identify proteins potentially important for virulence, such as hydrolytic enzymes like cell wall degrading enzymes (CWDEs), proteases and effector proteins, in EV and SS fractions. In vitro conditions may not completely reflect the gene and protein expression profile of Z. tritici in planta infection and the genes required for virulence. However, some in vitro growth conditions, such as Fries medium, can induce the expression of virulence-associated genes, including effectors and secondary metabolites, of plant pathogenic fungi [38, 39]. We found that hydrolytic enzymes – including proteins with carbohydrate active enzymatic domains (CAzymes) or peptidase protein domains – constituted a larger proportion of SS proteins (12.2% and 6.1%, respectively) compared to EV proteins (3.3% and 2.5%). Similarly, more SS proteins were predicted to be effectors by EffectorP (32.7% vs 8.1% in EVs) (Table 1, Additional file 2: Table S4). Putative effectors in SS samples were largely uncharacterised proteins. In comparison, EV-associated predicted effectors often had high sequence similarity to characterised proteins in the NCBI/Uniprot database, for example, several ribosomal subunit proteins. It remains to be seen if these are spurious effector predictions, or a true reflection of their importance for virulence in wheat.
Table 1. Characteristics of proteins identified in Z. tritici EV samples and the soluble secreted fraction
Proteins identified using LC-MS/MS
|
EV proteins
|
Soluble secreted proteins (SS)
|
Biological replicates
|
5
|
3
|
Total proteins identified
|
771
|
99
|
High confidence (HC) proteins1
|
240
|
79
|
HC proteins unique to sample type2
|
210 (100%)
|
49 (100%)
|
Proteins with signal peptides
|
14 (6.7%)
|
38 (77.6%)
|
Proteins with transmembrane domains/GPI-anchors3
|
15 (7.1%)
|
1 (2.0%)
|
Putative Cazymes
|
7 (3.3%)
|
6 (12.2%)
|
Putative proteolytic enzymes
|
6 (2.5%)
|
3 (6.1%)
|
Predicted effectors4
|
17 (8.1%)
|
16 (32.7%)
|
Median protein size
|
426 aa
|
257 aa
|
1Proteins were high-confidence if present in ≥ 4/5 EV biological replicates or ≥ 2/3 SS replicates, with ≥ 2 unique peptides detected. 2Protein counts and percentages reported refer to the subset of HC proteins detected only in EVs or SS fractions; proteins detected in both fractions were excluded from these counts. 3Transmembrane domains predictions by TMHMM were included when more than one domain predicted, or if only one was predicted, this was not in the first 60 amino acids. 4Effector count includes proteins predicted as effector by both v1.0 and v2.0 of EffectorP.
To gauge an approximate idea of the proportion of EV proteins involved in different cellular processes, proteins were assigned putative functional annotations based on predicted domains and sequence identity to characterised proteins in the Uniprot and NCBI non-redundant protein databases (Fig. 3, Additional file 2: Table S2). Based on this categorisation, EVs were distinct from the SS fraction in their protein cargo. Over 50% of SS proteins were uncharacterised proteins compared to 7% of EV proteins (Additional file 2: Table S2, S3). Approximately a quarter of EV proteins assigned a functional description were putatively associated with sugar, lipid and other metabolic processes (24.3%) and 18.6% involved in amino acid metabolism and protein homeostasis/metabolism. This included putative heat-shock proteins (Hsp60, Hsp82, Hsp90 co-chaperone), a protein disulphide isomerase, and translational machinery such as translation elongation (EF-2, EF-1-gamma, EF-3) and initiation factors. Similarly, ribosomal subunit proteins constituted 15.7% of EV proteins. The proportion of proteins in these categories were comparatively low in SS samples. Interestingly, of the 30 proteins identified in both SS and EV fractions, most were associated with carbohydrate and lipid metabolism (23.3%) or carbohydrate hydrolytic activity (20%), or were uncharacterised (36.7%) (Fig. 3, Additional file 2: Table S3).
Given the membranous nature of EVs we expected a proportion of proteins to be membrane associated: 6.7% of proteins were putatively membrane associated and included ABC transporters and other ATPase domain-containing proteins. Proteins involved with endo- and exocytosis or the endomembrane system (3.3% of proteins) included clathrin, a putative t-SNARE protein ADP-ribsoylation factor and a Sar1 GTPase-like protein. Actin, tubulin and fimbrin were among 7 proteins (3.3%) associated with the cytoskeleton. Proteins in these categories were largely absent from the SS fraction, with the exception of a GPI-anchored uncharacterised protein.
Z. tritici EV proteins overlap with proteins identified in other fungal EVs
We looked for similar proteins in the cargos of other fungal EVs to determine the overlap of Z. tritici EVs with other fungi. We used protein sequences of fungal EV proteins from previously published datasets for the human pathogenic fungi C. albicans, C. neoformans, H. capsulatum and Paracoccidioides brasiliensis; the cotton pathogen Fov; and S. cerevisiae (total n = 6648); the datasets and articles referred to are summarised in the additional files (Additional file 2: table S5). 77.8% of these proteins were assigned to 1453 orthogroups (OGs), with the majority of OGs (88.9%) containing proteins from more than two species (Fig. 4a). Only three OGs included similar EV proteins from all 7 species (Fig. 4a, c). These consisted of peptidyl-prolyl cis-trans isomerase proteins (OG9), plasma membrane ATPases (OG12) and ADP-ribosylation factors (ARFs) (OG18), respectively, the latter including three ARFs characterised in yeast as proteins associated with vesicle trafficking (Additional file: table S6). 95.5% of Z. tritici EV proteins were assigned to 177 OGs (12.2%), of which 2 were species-specific. Z. tritici EV proteins were most frequently grouped with similar proteins from S. cerevisiae, C. albicans and Histoplasma, though this is likely a reflection of the larger size of the EV proteomes currently defined for these fungi, compared to the P. brasiliensis or Fov proteomes (Fig. 4b, c). Orthogroups shared by Z. tritici and Fov exclusively were of interest, given their shared plant pathogenic lifestyles. Two groups of orthologous proteins were exclusive to these pathogens and consisted of subtilisin-like peptidases (OG1244) and peroxidase-catalase like proteins (OG765). Only 14 Z. tritici EV proteins were not assigned to an OG or grouped in a species-specific OG, suggesting significant similarity in the proteins identified in Z. tritici EVs and other fungal EVs characterised to date.
The proposed fungal EV protein marker Sur7 is consistently detected in Z. tritici EVs
A recent comprehensive study of potential marker proteins for Candida albicans EVs showed that the Sur7-family members CaSur7 and Evp1 were promising markers for C. albicans EVs [40]. We found a putative CaSur7 homologue, here referred to as ZtSur7, was consistently detected in Z. tritici EV samples from in vitro cultures (Additional file 2: Table S2). ZtSur7 has a 28.3% pairwise sequence identity with CaSur7, a shared Sur7 protein domain (L12—I207) and four predicted TMDs with a similar topology to the C. albicans protein (Additional file 1: Figure S2a). This protein was not detected in secreted samples and was only detected in 2/5 cell lysate samples (CL), which we analysed alongside the Z. tritici EV samples (a list of proteins identified in CL samples is available in additional file 2, Table S7). This aligns with the criteria Dawson et al., (2020) used to identify this protein as a fungal EV protein marker in C. albicans, which was based on recommendations for marker proteins by the International Society of Extracellular Vesicles [40, 41]. Analysis of RNA-seq evidence from a previous study describing Z. tritici WAI332 gene expression in planta (Wang et al., under review) found the ZtSur7 gene is expressed during infection of wheat at 9- and 14-days post infection (Additional file 1: Fig. S2b).
Dawson et al., (2020) also proposed 7 negative marker proteins for C. albicans EVs. Negative markers are needed for confirming EV cargoes are distinct from the general contents of the cell or plasma membrane, as outlined by Thery et al., (2018) [40, 41]. We identified putative orthologous proteins in Z. tritici with a reciprocal best-hit blastp search. Orthologues of five of these C. albicans proteins –ABP1, APR1, Cyp1, LAP41 and LPD1 – were identified in 80% of Z. tritici cell lysate (CL) samples (Additional file 2: Table S7, S8). These proteins were not consistently identified in Z. tritici EVs. Together, this suggests Sur7 is an EV-associated protein and potential marker in Z. tritici as well as C. albicans, while Z. tritici proteins orthologous to the aforementioned negative protein markers, may also be applicable for studies of Z. tritici EVs in vitro.