Abstract
CK2 is known as a constitutively active, conserved serine/threonine kinase in eukaryotes and was investigated in the model fungus Magnaporthe oryzae. GFP fusions to CK2 subunits demonstrated nucleolar localization, although the catalytic subunit still localized to nucleoli in the absence of either regulatory subunit. In contrast, localization near septa, required all three subunits. Appressoria contain a filamentous form of CK2. The ~1300 proteins co-immunoprecipitating with the catalytic subunit were highly enriched for those known to reside at septa and nucleoli and of these, many were also found to be intrinsically unfolded proteins.
Furthermore, a large proportion of these proteins contain a CK2 phosphorylation motif that has been proposed to function to destabilize and unfold alpha helixes. This suggests a role for CK2 in the formation of protein aggregates through interaction with its substrates. Examining gene expression profiles, we find a correlation of CK2 expression with genes for protein disaggregation and autophagy. Our observations support the view that CK2 plays a general role in controlling formation of discrete regions (membraneless organelles), such as the nucleolus, through aggregation and disaggregation of many of its target proteins.
Introduction
Since its discovery (Meggio and Pinna, 2003), the constitutive serine/threonine (S/T) kinase activity of CK2 and the increasing number of proteins it has been shown to phosphorylate have puzzled scientists (Ahmad et al., 2008; Götz and Montenarh, 2016; Meggio and Pinna, 2003). Indeed CK2 has been implicated in a wide range of cellular processes (Götz and Montenarh, 2016). The typical CK2 holoenzyme is a heterotetrameric structure consisting of 2 catalytic α-units and 2 regulatory β-subunits (Ahmad et al., 2008). In mammals, there exist two different alpha subunits α (a1) and α’ (a2) and the enzyme can contain any combination of α-subunits (α1α1, α1α2, α2α2) combined with the β-subunits. The CK2 of Saccharomyces cerevisiae, also contains two different alpha- and two different β-subunits (b1 and b2) and deletion of both catalytic subunits is lethal (Padmanabha et al., 1990). CK2 has been extensively studied in the budding yeast S. cerevisiae (Padmanabha et al., 1990), however, functions of CK2 involved in multicellularity might be obscured in yeast. In comparison to yeast, filamentous fungi have many different cell types that allow detailed exploration of cellular differentiation and multicellular development (Shlezinger et al., 2012) and this, in combination with haploid life-cycles, well characterized genomes, and efficient methods for targeted gene replacement, makes fungi like M. oryzae and Fusarium graminearum good model systems for molecular studies of basic eukaryote functions including cell-cell communication (Cavinder et al., 2012; Ebbole, 2007). As plant pathogens, developmental processes needed for symbiosis can also be explored. We focused our study on M. oryzae one of the most important rice crop pathogens worldwide (Dean et al., 2012).
Our results show that M. oryzae CK2 holoenzyme (MoCK2) accumulates in the nucleolus, localizes in structures near septal pores, and assembles to form a large ring structure perpendicular to the appressorium penetration pore. The large-scale structures formed by CK2 protein kinase, combined with our finding of the interaction of CK2 with substrates associated with the location of CK2 enzyme aggregation, suggests that CK2 may control substrate stability and localization near their sites of action. Furthermore, CK2 interacts preferentially with proteins annotated as being intrinsically disordered carrying a phosphorylation motif that can destabilize alpha helix folding (Zetina, 2001). It has been shown for the specific case of CK2 in relation to the intrinsically highly disordered SRP40 protein (yeast), with Nopp140 and Nolc1 as common synonyms for homologues in other organisms, that SRP40 becomes phosphorylated to varying degrees by CK2 with effects on the SRP40 conformation, binding, and aggregation properties important for the diverse functions of the protein (Na et al., 2018; Tantos et al., 2013). Taken together, our work provides further evidence supporting the view (Zetina, 2001) that one of the main roles for the CK2 holoenzyme is as a general inducer of binding and conformational changes in intrinsically disordered proteins.
RESULTS
Deletion of MoCK2 components
We identified one CKa catalytic subunit ortholog (MoCka1, MGG_03696) and two MoCKb regulatory subunit orthologs (MoCKb1, MGG_00446 and MoCKb2, MGG_05651) (Figure 1 Supplement 1) based on BLASTp analysis using protein sequences for the CK2 subunits of S. cerevisiae, CKa1, CKa2, CKb1 and CKb2 (Padmanabha et al., 1990). Filamentous fungi have just one highly conserved catalytic subunit. In the case of F. graminearum, two genes with homology with CKa were identified previously (Wang et al., 2011), one that is highly conserved (we name FgCKa1), and one that is CKa-like (named FgCKa2). It remains to be determined if FgCKa2 is actually a CK2 subunit (Figure 1 Supplement 2–5). Targeted deletions of the two regulatory subunits succeeded (Figure 2a, b Figure 1 a, b and Table 1 for list of strains and their abbreviations). Attempts to delete the catalytic subunit, CKa, were unsuccessful, consistent with the essential role of CK2 activity (Hermosilla et al., 2005). The most obvious visible phenotype of the CKb mutants (strain b1 and b2) in culture was reduced growth rate (Figure 5 a, b and Figure 2 a, b) and conidial morphology in that they produced few conidia and those that were produced had fewer conidial compartments (Figure 2c).
Subcellular localization of CK2 subunits
To assess the localization of the three CK2 subunits, we constructed N-terminal GFP fusions of all three proteins (Filhol et al., 2003) GFP-MoCKa, GFP-MoCKb1 and GFP-MoCKb2 (strains GfpA, GfpB1 and GfpB2 respectively Table 1). All three strains showed the same growth rate, morphology (data not shown) and pathogenicity (Figure 3) as the control strain Ku80. The CKa and CKb1&2 fusion proteins localized to nuclei and prominently to nucleoli and, interestingly, to both sides of septal pores in hyphae (Figure 4 a-e) and conidia (Figure 4 Supplement 1). The RNA levels for the GFP fusion genes in these strains were elevated ~10 to 15-fold over the control level (Figure 5 Supplement 1 and Figure 4 Supplement 2).
We then tested if the localization to septa and nucleoli were dependent on the association with the other subunits of the holoenzyme. We had measured MoCka expression in the two Mockb mutants (b1 and b2) using qPCR and noted it was downregulated two-to three-fold compared to the background control strain Ku80 (Figure 5 a). We constructed strains that over-express GFP-CKa in deletion strains b1 and b2 (strains b1GfpA and b2GfpA). Expression of GFP-CKa was elevated 25-fold and 15-fold in the b1GfpA and b2GfpA strains, respectively (Figure 5b). Localization to GFP-MoCKa to septa was not observed (Figure 4 f, g), however, nucleolar localization of GFP-MoCKa was clear in the b1GfpA and b2GfpA strains. To test if over-expression of any one of the CKb proteins could rescue the effect of the deletion of the other CKb, we constructed GFP-CKb overexpression strains in both CKb mutants (strains b1GfpB2 and b2GfpB1 Table 1). As noted above, the overexpression of either of the two CKbs in the control strain Ku80 showed normal localization to septa and nucleolus (Figure 4 c, e, h, i) but the overexpression in the CKb deletion strains could not rescue normal localization (Figure 4 h,i). Furthermore, both GFP-MoCKb1 and GFP-MoCKb2 appeared to localize to nuclei but were excluded from nucleoli in the b2GfpB1 and b1GfpB2 strains, respectively. A limited restoration of conidial production and morphology defect of ΔMockb1&2 deletions (strains b1 and b2) was observed in b1GfpA and b2GfpA (Figure 5 c, d, e). In addition, a significant restoration of growth rate was detected (Figure 6 a, b, d, e).
Infection phenotypes of CKb deletions
Deletion of CK2 genes has been shown to have effects on both growth and infection in F. graminearum (Wang et al., 2011) and we also found this to be the case for M. oryzae (Figure 6 c). Conidiation was virtually absent in both ΔMockb1 and ΔMockb2 deletion mutants (strains b1 and b2), thus, we used mycelial plugs to test for infection (Liu et al., 2010; Talbot et al., 1996). Compared to the background strain Ku80, mutants lacking one of the MoCK2b components had severely reduced or complete lost pathogenicity on intact leaves. However, wound inoculated leaves were impacted by the mutants (Fig 6c). Overexpression of MoCka in the ΔMockb1 and ΔMockb2 lines (strains b1GfpA and b2GfpA) allowed sufficient conidia production to perform conidial inoculations. Small lesions were observed in both cases, indicating that Ckb subunits are not required for pathogenesis (Fig 6d).
Overexpression of the MoCkb1 subunit in strain b2, strain b2GfpB1, restored growth rate and improved conidiation (Figure 6 Supplement 1). In contrast, overexpression of MoCkb2 in strain b1, strain b1GfpB2 did not restore growth or conidiation. There was a limited but detectable restoration of pathogenicity in spore inoculation of strain b2GfpB1 compared to strain b2 and in mycelial inoculation of unwounded plants of strain b1GfpB2 compared to strain b1. This suggests that overexpression of MoCkb2 in the ΔMockb1 mutant was unable to compensate for vegetative functions but could partially suppress the pathogenicity defect.
CKa localization in appressoria
Since we found large effects in infection of the deletion of the CKb components we decided to investigate localization of GFP-CKa in the appressoria. As in hyphae and conidia, GFP-CKa (strain GfpA) localizes to nuclei (Figure 7a top row and Figure 7 Supplement 1b), to the septa between the appressorium and the germ tube (Figure 7 Supplement 1) and also assembles a large ring structure perpendicular to the penetration pore (Figure 7 b-d, Figure 7 Supplement 1, Figure 7 Supplement 2 for ring size measurements, and movies associated with the images of Figure 7b-d showing 3D rotations to visualize the ring and the appressorium). MoCKa nuclear localization was present in appressoria formed by the two CK2b deletion mutants (strains b1GfpA and b2GfpA) (Figure 7a middle and bottom row), however ring structures were not observed. Concentration of GFP-MoCKa in nucleoli is clear in conidia, however, we could not clearly observe preferential nucleolar localization in appressoria. As can be seen in Figure 7d, the CK2 large ring structure is positioned perpendicular to the penetration pore where the F-actin-septin ring has been shown to form around the pore opening (Dagdas et al., 2012) (Figure 7d and 8 schematic drawing).
Identification of potential septal and nucleolar substrates for MoCK2 by GFP-CKa pulldown
The localization pattern suggested that CK2 may have substrates associated with septa and nucleolar function. To explore this, we performed co-immunoprecipitation to identify proteins interacting with CK2 using GFP-CKa as a “bait”, and in addition to the bait, identified 1505 proteins (Supplementary File 1). We also searched the M. oryzae proteome for proteins containing the CK2 phosphorylation helix unfolding motif identified by Zetina (Zetina, 2001) using the FIMO tool at the MEMEsuit website (http://meme-suite.org/) and found 1465 proteins (Supplementary File 2) with the motif, out of a total of 12827 proteins annotated for M. oryzae.
There is the risk of false positives in the pulldown. We estimated the number of false positives and removed 155 (~10%) of the lower abundance proteins to arrive at a list of 1350 CKa interacting proteins (see Methods). We found 275 of these proteins contain at least one unfolding motif for alpha helixes. Thus, there is an overrepresentation of the motif among the pulldown proteins (Supplementary File 2) (P-value for the null hypothesis of same frequency as in the whole proteome = 4 E-19, Fisher’s Exact test) lending support for the proposed role for this motif as a target for CK2 phosphorylation and protein unfolding. As also expected, the pulldown caught both CKb proteins.
Since CK2 localizes to septa we looked for known septal proteins in the pulldown. All previously identified proteins by Dagas et al. (Dagdas et al., 2012) that are involved in appressorium pore development, were found in the pulldown as was a protein annotated as the main Woronin body protein, Hex1 (MGG_02696). Since the Woronin body in Ascomycetes is tethered to the septal rim by Lah protein (Han et al., 2014; Ng et al., 2009; Plamann, 2009) we searched for a homologue in M. oryzae and found a putative MoLah (MGG_01625) with a similar structure as in Aspergillus oryzae (Han et al., 2014) that is also present in the pulldown. In addition to the Lah, 18 other intrinsically disordered septal pore associated proteins (Spa) were described for Neurospora crassa (Lai et al., 2012). We identified putative orthologs for 15 of the 18 Spa proteins in M. oryzae (Supplementary File 3). Of these putative MoSpa proteins, six were present in the CKa pulldown, Spa3 (MGG_02701), Spa5 (MGG_13498), Spa7 (MGG_15285), Spa11 (MGG_16445), Spa14 (MGG_03714) and Spa 15 (MGG_15226). Spa3, Spa5 and Spa15 also contain the CK2 phosphorylation alpha helix unfolding motif (Supplementary File 1).
To further explore the hypothesis that CK2 could interact with and possibly phosphorylate intrinsically unfolded proteins we used the FuncatDB (http://mips.helmholtz-muenchen.de/funcatDB/) to make a functional classification of the pulldown proteins including those containing the alpha helix unfolding motif (Zetina, 2001). We found strong overrepresentation for proteins involved in rRNA processing among the pulldown proteins containing the alpha helix unfolding motif as well as for proteins that, themselves, are known to interact with other proteins, DNA, and RNA (Supplementary File 4). These classes of proteins are enriched for intrinsically disordered proteins. Such intrinsically disordered proteins can interact with each other to form ordered subregions that have been described as membraneless organelles, such as nucleoli (Wright and Dyson, 2015). Since CK2 localizes to the nucleolus we were especially interested in the interaction of CK2 with nucleolar localized proteins. We identified homologues to the well described S. cerevisiae nucleolar proteins and found a total of 192 proteins in M. oryzae homologous to yeast nucleolar proteins (Supplementary File 5). We found 120 (63%) of the nucleolar proteins in the pulldown and 60 of these (50% of the ones found) had the alpha helix unfolding motif (Supplementary File 1 and 4). The nucleolar proteins were highly overrepresented in the pulldown (P-value for the null hypothesis of same frequency as in the whole proteome 9E-43 Fisher’s Exact test) (Supplementary File 5) compared to the whole proteome as was also nucleolar proteins having the unfolding motif (P-value for the null hypothesis of same frequency as in the whole proteome 2E-13 Fisher’s Exact test) (Supplementary File 5).
Ck2 is known to interact with the disordered nucleolar protein SRP40 that has a multitude of CK2 phosphorylation sites and in addition CK2 is known to change the protein binding activity towards other proteins depending the SRP40 phosphorylation status. When SRP40 becomes highly phosphorylated it binds to and inhibits CK2 activity in a negative feedback loop ensuring that CK2 phosphorylation level will balance (Tantos et al., 2013; Na et al., 2018). We identified a putative MoSRP40 (MGG_00613) and it is disordered in a similar way as other SRP40 proteins and it is well conserved in filamentous fungi (Supplementary File S8). The MoSRP40 protein was highly represented in the CKa pulldown indicating that it interacts with CKa (Supplementary File S1).
Interestingly, proteins that are imported into mitochondria and involved in oxidative phosphorylation (“02.11 electron transport and membrane-associated energy conservation” category from Funcat) (Supplementary File 4) were enriched in the pulldown proteins (60 of 130 in the whole proteome, P-value for the null hypothesis of same frequency as in the whole proteome 1.0E-29). In contrast with septal and nucleolar interacting proteins, the mitochondrial proteins were not enriched for the known unfolding motif.
There was no enrichment for specific pathogenicity related proteins (Funcat category 32.05 disease, virulence and defence) (Supplementary File 4). This is generally true within the whole Funcat category related to stress and defence (32 CELL RESCUE, DEFENSE AND VIRULENCE) with the exception of proteins involved in the unfolded protein response (32.01.07 unfolded protein response) (e.g. ER quality control), which were overrepresented. This is notable since an involvement of CK2 in protein import into the ER has be established (Wang and Johnsson, 2005). An association of pathogenicity related proteins with CK2 was not expected because of the in vitro growth conditions of the experiment.
Interestingly, five putative S/T phosphatases (MGG_03154, MGG_10195, MGG_00149, MGG_03838, MGG_06099) were in the pulldown set of proteins (Supplementary File 1). Conceivably these might de-phosphorylate CKa substrates as well as substrates of other kinases to expand the reach of CK2 in regulating the phosphoproteome. To examine the relationship between the expression of CK2 and these phosphatases, we downloaded expression data from a range of experiments with M. oryzae and plotted the expression of the five phosphatases found in the pulldown, and an S/T phosphatase not found in the pulldown, as a function of the CKa expression. We found that two of the S/T phosphatases present in the pulldown were strongly correlated with CKa expression and the others were less strongly correlated (Figure 9).
CKa expression correlates with expression of genes associated with disaggregation and autophagy
Since CK2 activity has the potential to favour protein-protein binding between intrinsically disordered proteins it consequently also has the potential to enhance protein aggregation. Some of these unfolded proteins may trigger the unfolded protein response involved in disaggregation. Hsp104 is a disaggregase that cooperates with Yjdg1 and SSa1 to refold and reactivate previously denatured and aggregated proteins (Glover and Lindquist, 1998). Alternatively, accumulated aggregates may be degraded through autophagy since these kinds of aggregates are too big for proteasome degradation (Wong and Cuervo, 2010). If this is the case, CK2 upregulation should be accompanied by higher autophagy flux or at least there should not be low expression of key autophagy genes when CK2 expression is high (Wong and Cuervo, 2010). Atg8 is a key autophagy protein for which its turnover rate can reflect autophagy flux (Klionsky et al., 2016). To test this hypothesis, we used the expression data we downloaded for plant infection experiments with M. oryzae and also for another fungal plant pathogen, F. graminearum, that has rich transcriptomic data available (see methods), to examine expression of HSP104, YDJ1, SSA1, and ATG8 relative to CKa.
For M. oryzae, we found an approximately 60-fold increase in MoHSP104 expression associated with a doubling of MoCka transcript levels. With increasing expression of MoCka the MoHSP104 levels did not increase further. MoSSA1 expression had a similar pattern to MoHSP104 with a 16-fold increase across the initial 2-fold increase in MoCka expression. For MoYDJ1, expression increased with MoCKa expression, but not as dramatically (Fig 10). For M. oryzae, we find a log-log linear relationship between the MoCKa expression and MoAtg8 expression (Figure 10).
In the case of F. graminearum, we also found increased expression of all of the genes correlated with FgCKa expression across a large range of experiments (Figure 10 Supplement 1a). Within the F. graminearum experiments a time course experiment was selected to examine expression of these genes during the course of infection. Once again, the relationship could be observed (Figure 10 Supplement 1b), furthermore, FgCK2a expression increased during the course of infection (Figure 10 Supplement 2). Overall, these correlations support the hypothesis that protein disaggregation and autophagy are increasingly needed to remove protein aggregates stimulated to form by increasing levels of CKa and its activity in the cell.
Discussion
The analysis of the MoCKb mutants and the localization of the GFP-labelled MoCK2 proteins showed that all identified MoCK2 components are needed for normal function and also normal localization. Localization to septa requires all three subunits, presumably as the holoenzyme. Mutation of either CKb subunit blocks nucleolar localization of the other CKb subunit. Surprisingly, nucleolar localization of CKa was observed in the CKb mutants. This shows that the holoenzyme is not required for CKa localization to the nucleolus. It seems likely that CKb1 and CKb2 must interact with each other in order to interact with CKa, and that CKa is required for movement of CKb subunits into the nucleolus as the holoenzyme.
The pattern of localization to septa (Figures 4) observed is remarkably similar to that displayed by the Woronin body tethering protein AoLah from A. oryzae (Figure 4b in (Han et al., 2014)). The pulldown experiments demonstrate that CK2 interacts with proteins that function in septum formation and function, including the MoLah ortholog, supporting the view that localization of the GFP-fusion proteins gives a proper representation of CK2 localization. Our results thus demonstrate that the MoCK2-holoenzyme assembles as a large complex near, and is perhaps tethered to, septa, possibly through binding to MoLah. Since septal pores in fungi are gated (Shen K-F. et al., 2014), as are gap junctions and plasmodesmata in animal and plant tissue, respectively (Ariazi et al., 2017; Kragler, 2013; Neijssen et al., 2005), CK2 has a potential to play a general role in this gating.
The crystal structure suggested that CK2 can form filaments and higher-order interactions between CK2 holoenzyme tetramer units, and based on this it has been predicted that autophosphorylation between the units could occur to down-regulate activity (Litchfield, 2003; Poole et al., 2005). Filament formation has been shown to occur in vitro (Glover, 1986; Seetoh et al., 2016; Valero et al., 1995) and in vivo (Hübner et al., 2014). Several forms of higher order interactions have been predicted, and it has been demonstrated that at least one of these has reduced kinase activity (Poole et al., 2005; Valero et al., 1995). However, in our localization experiments focused on septa, we cannot distinguish if the large structure is due to co-localization of the CK2 with another protein, such as the MoLah ortholog, or if CK2 is in an aggregated form near septa. Since MoLah has the characteristics of an intrinsically disordered protein (Han et al., 2014), and CK2 has been proposed to interact with proteins to promote their disordered state (Zetina, 2001; Tantos et al., 2013; Na et al., 2018), we favour the view that CK2 interacts with MoLah and other proteins to form a complex near septa.
The large ring observed in appressoria may be a true filament of CK2 in a relatively inactive state that is a store for CK2 so that upon infection, it can facilitate rapid ribosome biogenesis, appressorial pore function, and other pathogenesis-specific functions.
Our pulldown experiment with GFP-CKa further showed that there was a strong overrepresentation of proteins interacting with CKa that contain known phosphorylation motifs for unfolding of alpha-helixes and this is what would be expected for intrinsically disordered proteins (Uversky, 2015; Zetina, 2001). The finding of overrepresentation of this signal in the set of CK2 interacting proteins corroborates the previous suggestion that CK2 is involved in the destabilization/binding of intrinsically disordered proteins (Zetina, 2001; Tantos et al., 2013; Na et al., 2018) and is consistent with the strong accumulation of both CK2 and intrinsically disordered proteins in the nucleolus (Fig. 4a and b) (Frege and Uversky, 2015) and also at pores between cell compartments (Lai et al., 2012) (Figure 4d). In addition, and further supporting this conclusion, the six septal pore associated proteins (SPA) that we find in the CKa pulldown are homologues for intrinsically disordered proteins that are expected to form temporary gels that are used to reversibly plug septal pores and regulate traffic through septa (Lai et al., 2012). CK2 could actively be involved in the gelling/un-gelling of the regions near septa to create a membraneless organelle controlling the flow through septa. As a counterpart to CK2 in gelling/un-gelling, disaggregase activity involving the MoHSP104 complex, may be critical for control of this. The observation of transcriptional co-regulation between CKa and HSP104 supports this notion.
Previous studies of subcytosolic localization reveals that this enzyme is also associated with import into organelles. CK2 promotes protein import into endoplasmic reticulum (Wang & Johnsson, 2005) and into mitochondria during mitochondrial biogenesis and maintenance (Rao et al., 2011). CK2 phosphorylation has been shown to activate Tom22 precursors to assemble a functional mitochondrial import machinery (Rao et al., 2011). Although CK2 has been implicated to be located in mitochondria in earlier studies in other organisms, no proteomic study of yeast mitochondria has detected the presence of CK2 (Rao et al., 2011). Hence, we do not expect MoCK2 to be present in mitochondria, and we saw no evidence for mitochondrial localization. Of special interest was however the strong overrepresentation of mitochondrial proteins among the CKa pulldown proteins without the alpha helix phosphorylation unfolding motif (Supplementary File 4). Since these proteins need to be imported into mitochondria in an unfolded state, this may point to the existence of CKa phosphorylation and unfolding motifs other than the one identified by Zetina (Zetina, 2001) that help keep these proteins unfolded until they reach their destination inside the mitochondria.
To have such dynamic function as an unfolder of proteins by phosphorylation, CK2 should be partnered with phosphatases as counterparts and their activity may track CK2 activity. Consistent with this possibility, we found that two of the five S/T phosphatases that are present in the pulldown are strongly co-regulated with CKa (Figure 9), further supporting the view that CKa-dependent phosphorylation/dephosphorylation plays a major role in shaping protein interactions. Together with the high expression of CK2 in cells, this suggests an important function of CK2 as a general temporary unfolder of intrinsically disordered proteins, that comprise roughly 30 % of eukaryotic proteins (Vucetic et al., 2003), in a similar way as it is known to interact with SRP40 and influence its activity (Tantos et al., 2013; Na et al., 2018).
As MoCK2 is present in the cytoplasm and nucleoplasm it would generally assist intrinsically disordered proteins in protein interactions (Uversky, 2015). It also seems to be essential for assembling ribosomes containing large numbers of intrinsically disordered proteins (Uversky, 2015). All these functions also explains why CK2 is needed constitutively (Meggio and Pinna, 2003).
In the absence of well-functioning autophagy removing incorrectly formed larger protein aggregates, like those formed in brain cells of Alzheimer’s patients (Zare-shahabadi et al., 2015), CK2 activity facilitates protein aggregate formation and hastens the progression of Alzheimer’s disease (Rosenberger et al., 2016). Using publicly available transcriptome datasets we could show that CKa expression in M. oryzae and F. graminearum is strongly correlated to disaggregase and Atg8 expression (Figure 10), and thus autophagy, giving further support for a relationship of CK2 in facilitating the formation of protein aggregates from intrinsically unfolded proteins that are then subjected to autophagy. As autophagy is important to appressorium development (Liu and Lin, 2008; Kershaw and Talbot, 2009), it will be of interest to further examine the role of the CK2 ring structure during appressorial development and infection.
Conclusion
We conclude that CK2 likely has an important general role in the correct assembly/disassembly of intrinsically disordered proteins as well as allowing these proteins to pass through narrow pores between cell compartments, in addition to its already suggested role in organelle biogenesis (Rao et al., 2011). Our results further point to one of the main functions of the CK2 holoenzyme as a general facilitator of protein-protein interactions important for a large range of cellular processes including a potential role for gel formation that creates membraneless organelles at septa through its likely interaction with, and modification of, intrinsically disordered proteins. We thus feel it appropriate to cite “Using basket terminology, one would say that CK2 looks like a “playmaker” not a “pivot”: hardly ever does it make scores; nevertheless, it is essential to the team game” (Meggio and Pinna, 2003).
Methods
Fungal strains, culture, and transformation
The M. oryzae Ku80 mutant (constructed from the wild type Guy11 strain was used as background strain since it lacks non-homologous end joining which facilitates gene targeting (Villalba et al., 2008). Ku80 and its derivative strains (Table 1) were all stored on dry sterile filter paper and cultured on complete medium (CM: 0.6% yeast extract, 0.6% casein hydrolysate, 1% sucrose, 1.5% agar) or starch yeast medium (SYM: 0.2% yeast extract, 1% starch, 0.3% sucrose, 1.5% agar) at 25°C. For conidia production, cultures were grown on rice bran medium (1.5% rice bran, 1.5% agar) with constant light at 25°C. Needed genome and proteome FASTA files was downloaded from an FTP-server at the Broad Institute (ftp://ftp.broadinstitute.org/pub/annotation/fungi/magnaporthe/genomes/magnaporthe_oryzae_70-15_8/). Fungal transformants were selected for the appropriate markers inserted by the plasmid vectors. The selective medium contained either 600 µg/ml of hygromycin B or 600 µg/ml of G418 or 50 µg/ml chlorimuron ethyl.
MoCKb gene replacement and complementation
Gene replacement mutants of MoCKb1 encoding protein MoCKb1 were generated by homologous recombination. Briefly, a fragment about 0.9 Kb just upstream of Mockb1 ORF was amplified with the primers 446AF and 446AR (Table 2), so was the 0.7Kb fragment just downstream of Mockb1 ORF amplified with the primers 446BF and 446BR (Table 2). Both fragments were linked with the hygromycin phosphotransferase (hph) gene amplified from pCX62 (containing the fragment of TrpC promoter and hygromycin phosphotransferase (hph) gene, HPH resistance). Then the fusion fragments were transformed into protoplasts of the background strain Ku80. The positive transformant ΔMockb1 (strain b1, Table 1) was picked from a selective agar medium supplemented with 600 µg/ml of hygromycin B and verified by Southern blot.
For complementation of the mutant, fragments of the native promoter and gene coding region were amplified using the primers 446comF and 446comR listed in Table 2. This fragment was inserted into the pCB1532 to construct the complementation vector using the XbaI and KpnI. Then this vector was transformed into the protoplasts of strain b1. The positive complementation transformant, strain b1B1, was picked up from the selective agar medium supplemented with 50µg/ml chlorimuron ethyl.
As for the ΔMoCKb1 deletion mutant, we constructed a knockout vector to delete the MoCKb2 from the background strain Ku80. All the primers are listed in the Table 2. The 1.0Kb fragment upstream of MoCKb2 ORF was amplified with the primers 5651AF and 5651AR, inserted into the plasmid pCX62 using the KpnI and EcoRI to get the pCX-5A vector. The 1.0Kb fragment downstream of Mockb2 ORF was amplified with the primers 5651BR and 5651BR, inserted into the vector pCX-5A using BamHI and XbaI to construct the knockout vector pCX-5D. Then this vector was transformed into the protoplasts of Ku80. The positive transformants were picked up from the selective medium supplemented with the 600 µg/ml hygromycin B. For complementation of the resulting mutant, strain b2 (Table 1), fragments of the native promoter and gene coding region were amplified using the primers 5651comF and 5651comR listed in the Table 2. This fragment was inserted into pCB1532 to construct the complementation vector using the XbaI and XmaI. Then this vector was transformed into protoplasts of the strain b2. The positive complementation transformant, strain b2B2, was picked up from the selective agar medium supplemented with 50 µg/ml chlorimuron ethyl.
The construction of localization vectors
In order to detect the localization of MoCK2, we constructed localization vectors. The vector pCB-3696OE containing the RP27 strong promoter was used to detect the localization of GFP-MoCKa (strain GfpA). The vector pCB-446OE expressed under RP27 strong promoter was used to detect the localization of GFP-MoCKb1 (strain GfpB1). The vector pCB-5651OE expressed by RP27 strong promoter was used to detect the localization of GFP-MoCKb2 (strain GfpB2).
Analysis of conidial morphology, conidial germination and appressoria formation
Conidia were prepared from cultures grown on 4% rice bran medium. Rice bran medium was prepared by boiling 40g rice bran (can be bought for example through Alibaba.com) in 1L DD-water for 30 minutes. After cooling pH was adjusted from to 6.5 using NaOH and 20 g agar (MDL No MFCD00081288) was added before sterilization by autoclaving (121°C for 20 minutes). Conidia morphology was observed using confocal microscopy (Nikon A1+). The Conidial germination and appressoria formation were incubated on hydrophobic microscope cover glass (Beckerman and Ebbole, 1996) (Fisherbrand) under 25℃ in the dark. Conidial germination and appressoria formation were examined at 24 h post-incubation (Beckerman and Ebbole, 1996; Ding et al., 2010).
Pathogenicity assay
Plant infection assays were performed on rice leaves. The rice cultivar used for infection assays was CO39. In short, mycelial plugs were put on detached intact leaves or leaves wounded by a syringe stabbing. These leaves were incubated in the dark for 24h and transferred into constant light and incubated for 5 days to assess pathogenicity (Talbot et al., 1996). For infections using conidial suspensions (1 × 105conidia/ml in sterile water with 0.02% Tween 20) were sprayed on the rice leaves of 2-week-old seedlings.
RNA extraction and real-time PCR analysis
RNA was extracted with the RNAiso Plus kit (TaKaRa). First strand cDNA was synthesized with the PrimeScript RT reagent Kit with gDNA Eraser (TaKaRa). For quantitative real-time PCR, MoCKa, MoCKb1, and MoCKb2 were amplified with the primers listed in Table 2. β-tubulin (XP_368640) was amplified as an endogenous control. Real-time PCR was performed with the TaKaRa SYBR Premix Ex Taq (Perfect Real Time) (Takara). The relative expression levels were calculated using the 2−ΔΔCt method (Livak and Schmittgen, 2001).
Pulldown and identification of CKa interacting proteins
Total protein samples were extracted from vegetative mycelia of strain GFP-MoCKa and incubated with anti-GFP beads (Chromotek, Hauppauge, NY, USA) 90 minutes at 4°C with gentle shaking. After a series of washing steps, proteins bound to anti-GFP beads were eluted following the manufacturer’s instruction. The eluted proteins were sent to BGI Tech (Shenzhen, Guangdong province, China) and analysed by mass spectrometry for analysis of sequence hits against the M. oryzae proteome. The transformant expressing GFP protein only was used as the negative control and the Ku80 was used as Blank control. Data from three biological replicates were analyzed against the background of proteins that were bound non-specifically to the anti-GFP beads in GFP transformant and in Ku80 to get the final gene list of genes that was pulldown with CKa (Supplementary File1).
Estimation of non-specific binding of proteins in the pulldown
We developed two methods to estimate the number of non-specific binding proteins found in the CKa pulldown. The first approach is a chemistry-based reasoning and assumes that the degree of unspecific association to the protein per protein surface area is the same for GFP specific hits and for the CK2 holoenzyme pulled down. Using this technique, we estimate that 44-132 proteins are false positive in the CKa pulldown (all proteins pulled down by GFP-Beads or the Beads already removed from the list) (Supplementary File 1). The Second approach is statistical where we assume that binding of the true interacting proteins to CKa are log-normally distributed related to the abundance of each protein in the pulldown, since the median is low and close to zero and negative amounts are impossible. Using the deviation from the theoretical distribution, with higher than expected amounts of a specific protein, for the less abundant proteins we estimate that 46-81 proteins found in the CKa pulldown (with controls subtracted) were false positive. The higher number was used to set a conservative threshold for which proteins should be included in the analysis (See Supplementary File 1 for details of both methods).
Finding M. oryzae proteins containing the helix unfolding motif
The MEME motif LSDDDXE/SLEEEXD (Zetina, 2001) was used to search through the proteome of M. oryzae using the FIMO tool at the MEMEsuite website http://meme-suite.org/). Results were then downloaded and handled in MS Excel to produce a list of proteins with at least one motif hit (Supplementary File 2)
Analysis of CKa expression in relation to disaggregase related protein, Atg8 and Ser/Thr phosphatase expression
For M. oryzae, transcriptome experiment data was downloaded as sra/fastq files from Gene Expression Omnibus, https://www.ncbi.nlm.nih.gov/geo/ and mapped onto the genome found at http://fungi.ensembl.org/Magnaporthe_oryzae/Info/Index. The procedure was the following: Gene Expression Omnibus (Barrett et al., 2012) was queried for SRA files originating from M. oryzae and the files downloaded. The conversion from SRA to Fastq was done using the SRA toolkit (http://ncbi.github.io/sra-tools). The resulting samples were subjected to quality control using FASTQC (Andrews, 2010). Quantification of RNA was performed using Kallisto with default settings (Bray 2016), the data was then normalized using the VST algorithm implemented in DESeq2 (Love et al., 2014).
For F. graminearum transcriptomic data (FusariumPLEX) was directly downloaded for mainly in planta experiments from the PlexDB database (http://www.plexdb.org/modules/PD_general/download.php?species=Fusarium). For each fungus an expression matrix with the different experiments as columns and gene id using the FGSG codes according to BROAD (ftp.broadinstitute.org/distribution/annotation/fungi/fusarium/genomes/fusarium_graminearum_ph-1) as rows were prepared. From the resulting matrixes (Supplementary Files 6 and 7) we used the data needed to plot expression of Atg8 vs CKa for the two fungi. In M. oryzae data are expressed as log2 of RPKM values. Similarly, for F. graminearum, data were log2 of reported relative expression. Gene expression data used were from MoCKa, MoHSP104, FgHSP109, MoYDJ1, FgYDJ1, MoSSA1and Fg SSA1 homologues identified in this study as well as from MoAtg8 (MGG_01062) (Veneault-Fourrey, 2006), FgCKa (FGSG_00677) (Wang et al., 2011) and FgAtg8 (FGSG_10740) (Josefsen et al., 2012). Data from the M. oryzae expression matrix was also used for plotting MoCKa expression versus the expression of annotated serine/threonine phosphatases found in the CKa pulldown.
Data availability
The data that support the findings of this study are available from the corresponding authors upon request.
Authors’ contributions
Conceived and designed the experiments: G. L., S. O. and Z.
W. Performed the experiments: L. Z., D. Z., Y. C., W. Y. and Q. L. Analysed the data: L. Z., D. Z., D. J. E., S. O. and Z. W. Wrote the paper: L. Z., D. Z., D. J. E., S. O. and Z. W.
Competing interests statement
The authors declare that they have no competing financial interests.
Acknowledgements
We thank Dr. Guanghui Wang, Dr. Wenhui Zheng, Dr. Ya Li and Dr. Huawei Zheng (Fujian Agriculture and Forestry University, Fuzhou, China) for their helpful discussions. We thank Professor Jin-Rong Xu, Department of Botany and Plant Pathology Purdue University, U.S.A. for providing the Ku80 strain. We thank Dr. Bjoern Oest Hansen, Goettingen for help with mapping and constructing the used M. oryzae transcriptome datafile from downloaded data. This work was supported by the National Natural Science Foundation of China (U1305211), National Key Research and Development Program of China (2016YFD0300700) and National Natural Science Foundation for Young Scientists of China (Grant No.31500118 and No.31301630), the 100 Talent Program of Fujian Province, and USDA NIFA Hatch project 1013944.
Footnotes
§ These authors jointly supervised the work