SEARCH

SEARCH BY CITATION

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

Sexual development in fungi is controlled by mating type loci that prevent self-fertilization. In the phytopathogenic fungus Ustilago maydis, the b mating type locus encodes two homeodomain proteins, termed bE and bW. After cell fusion, a heterodimeric bE/bW complex is formed if the proteins are derived from different alleles. The bE/bW complex is required and sufficient to initiate pathogenic development and sexual reproduction; for the stages of pathogenic development succeeding plant penetration, however, its role was unclear. To analyse b function during in planta development, we generated a temperature-sensitive bEts protein by exchange of a single amino acid. bEts strains are stalled in pathogenic development at restrictive temperature in planta, and hyphae develop enlarged, bulbous cells at their tips that contain multiple nuclei, indicating a severe defect in the control and synchronization of cell cycle and cytokinesis. DNA array analysis of bEts mutant strains in planta revealed a b-dependent regulation of genes encoding secreted proteins that were shown to influence fungal virulence. Our data demonstrate that in U. maydis the b heterodimer is not only essential to establish the heterodikaryon after mating of two compatible sporidia and to initiate fungal pathogenicity, but also to sustain in planta proliferation and ensure sexual reproduction.


Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

Mating is an essential step in the life cycle of all sexually reproducing organisms. In fungi, sexual compatibility is controlled by mating type genes, which function to prevent self-fertilization and ensure the genetic variability of the population. In the Basidiomycete Ustilago maydis, the causal agent of the smut disease on maize plants, mating is accompanied with a dramatic change of lifestyle. The haploid cells, called sporidia, grow by budding and are strictly saprophytic. Mating of two of such sporidia leads to the formation of a dikaryon that grows filamentous and requires the plant host for further propagation. The mating reaction is controlled by two independent mating type loci that are termed a and b (for review, see Kronstad, 1997). The biallelic a locus encodes a pheromone/pheromone receptor system mediating recognition and cell fusion events (Bölker et al., 1992; Spellig et al., 1994). Subsequently, the multiallelic b mating type locus controls filamentous growth, maintenance of the dikaryon and the initiation of the pathogenic program as a prerequisite for sexual development. b encodes two distinct homeodomain proteins, bE and bW, that can form a heterodimeric complex via their N-terminal dimerization domains. However, the formation of a transcriptional active bW/bE complex is only facilitated when the two proteins are derived from different alleles (Schulz et al., 1990; Gillissen et al., 1992; Kämper et al., 1995). The active b heterodimer is necessary and sufficient to initiate the pathogenic lifestyle of U. maydis, as shown by means of a haploid strain that carries compatible bE1 and bW2 alleles (Bölker et al., 1995). This so-called solopathogenic strain infects the plant without a compatible mating partner.

The homeodomains of the b heterodimer have been shown to mediate binding to a specific DNA sequence (b-binding site) in the promoter regions of b-responsive genes (Romeis et al., 2000; Brachmann et al., 2001). b-responsive genes have been identified in several attempts (Bohlmann et al., 1994; Schauwecker et al., 1995; Urban et al., 1996; Wösten et al., 1996; Brachmann et al., 2001, 2003), but with the exception of kpp6 (involved in appressoria formation, Brachmann et al., 2003) none of them has been linked to pathogenic development. Recently, DNA microarray analysis has led to the identification of 345 b-regulated genes (K. Heimel, M. Scherer and J. Kämper, unpublished), three of which were shown to have impact on filamentous growth or pathogenic development. clp1 encodes a protein with unknown function, which is involved in cell cycle progression and cell division after plant penetration (Scherer et al., 2006). rbf1 encodes a zinc-finger transcription factor, which is required for the regulation of the majority of b-dependent genes (Scherer et al., 2006; K. Heimel, M. Scherer and J. Kämper, unpublished). The third gene, biz1, encodes a zinc finger transcription factor that was shown to be involved in appressoria formation and cell cycle arrest (Flor-Parra et al., 2006). In addition, several b-regulated genes associated with cell cycle control, mitosis or DNA replication were identified, consistent with the observation that b induction leads to a cell cycle arrest, which is released after plant penetration (García-Muse et al., 2003; Scherer et al., 2006; Cánovas and Pérez-Martín, 2009).

The expression of a compatible b heterodimer is required for the initiation of pathogenic development. However, it is unclear whether this central regulator is also required at the developmental stages succeeding plant penetration. It is known that the bE and bW genes are expressed during biotrophic development of the fungus (Quadbeck-Seeger et al., 2000), but it is unclear which genes are expressed in a b-responsive manner during in planta development. To gain insights into the function of bE/bW during in planta development of U. maydis, we generated temperature-sensitive b alleles via random PCR mutagenesis. The temperature-sensitive bE allele (bEts) prevents fungal proliferation in planta at restrictive temperature, while fungal development is not altered at permissive temperature. At restrictive conditions in planta, fungal tip cells are enlarged and contain multiple nuclei, demonstrating the requirement of b to co-ordinate cell cycle and cell division. In planta DNA microarray expression analysis comparing the bon and the boff state revealed a b-dependent transcription network important for the expression of secreted proteins previously shown to influence pathogenicity (Kämper et al., 2006). Our data demonstrate that the b heterodimer is essential during in planta development of U. maydis, affecting fungal proliferation and the biotrophic interaction with the host plant.

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

Generation of temperature-sensitive b alleles

In order to generate temperature sensitive (ts) mutant alleles of bE and bW, we applied in vitro mutagenesis to alter DNA fragments that encompass the N-terminal dimerization domain of bW1 and the N-terminal dimerization domain and homeodomain of bE2 respectively (Fig. 1A; see Experimental procedures). Alterations in the dimerization domains could possibly interfere with protein–protein interactions of bW and bE, resulting in an instable complex at higher temperatures unable to accomplish its regulatory function. Mutations within the homeodomain could affect or prevent DNA binding. The mutagenized fragments were cloned into autonomously replicating U. maydis plasmids to restore the complete open reading frames (ORFs) under control of the native promoter regions. Mutant libraries of bW1- and bE2-plasmids were transformed into U. maydis strain FBD11-21 (a1a2/b2b2) or FBD12-3 (a1a2/b1b1) respectively. Transformation of these strains with the bW1 or bE2 plasmids leads to the formation of an active bW1/bE2 complex, which is indicated through the induction of filaments on charcoal containing media plates (PDC, see Experimental procedures). A total of approximately 30 000 bE2- and 20 000 bW1 mutants were scored for filamentous growth at 22°C and budding growth at 31°C by means of replica plating. Two strains harbouring mutant bE2 alleles (FBD12-3 + pFSbE2ts52 and FBD12-3 + pFSbE2ts98) were identified that displayed a temperature-dependent growth phenotype. In both cases, filaments were induced at permissive temperature, while cells grew by budding at restrictive temperature (Fig. 1B). Sequence analysis of the two mutant bE2ts fragments revealed three missense mutations in each of the alleles. Both mutant-alleles shared one mutation leading to an exchange of serine to proline at position 183, located at the border of the third α-helix of the homeodomain of bE2 (Fig. 1C). To test whether this mutation causes the ts-phenotype, directed PCR mutagenesis was performed to alter serine to proline at position 183 in the wild-type bE2 allele. The resulting bE2ts183P allele revealed the same temperature sensitivity as observed for the alleles bE2ts52 and bE2ts98 (Fig. 1B).

image

Figure 1. Temperature-sensitive b alleles generated by random PCR mutagenesis. A. Schematic presentation of the bE2 and bW1 genes. Regions chosen for random PCR mutagenesis (flanked by restriction sites) are displayed as bidirectional arrows. Positions of the variable dimerization domain (DD) and homeodomain (HD) are indicated as ovals. B. b-induced growth on PDC plates at permissive (22°C) and restrictive temperature (31°C). Strain FBD12-3 + pNEBbW1UH harbours a non-compatible bE1/bW1 combination and grows non-filamentous at both temperatures (grey spots). Strain FBD12-3 + pFSbE2 harbours an active bW1/bE2 complex and growth filamentous at both temperatures (white spots). Strains carrying plasmids with temperature sensitive bE2 alleles grow filamentously only at permissive temperature. C. Alignment of the mutated regions of the temperature sensitive bE variants bE2ts52 and bE2ts98 with the wild-type bE2. Dimerization- and homeodomains are indicated in light grey and dark grey boxes respectively. Position and nature of generated amino acid exchanges are highlighted in black. The exchange of the wild-type serine to proline at the border of the third helix of the homeodomain (position 183) leads to the temperature sensitivity of both bE2ts52 and bE2ts98.

Download figure to PowerPoint

The serine to proline mutation leads to instability of bEts under restrictive conditions

To analyse the bEts alleles in more detail, we introduced the S183P-mutation into the bE1 gene of strain AB31. This strain allows the regulated expression of the compatible bW2 and bE1 genes, which are both under the control of the arabinose inducible crg1 promoter (Brachmann et al., 2001). In addition, we fused the bE1 and bE1ts genes to the triple-myc tag to facilitate protein expression analysis, yielding strains ABE1 and ABE1ts respectively. When spotted on glucose containing minimal medium (MM) with charcoal or grown in liquid array medium containing glucose, we observed no filament formation in ABE1 and ABE1ts; the crg1 promoter is inactive under these conditions, thus, bE1 and bW2 genes are not expressed (Fig. 2A and B, left panel). When arabinose was used as carbon source, the bE1 and bW2 genes were induced. The resulting filamentous growth of ABE1 and ABE1ts was comparable on charcoal containing media plates and in liquid culture, respectively, when grown at 22°C (Fig. 2A and B, middle panel). However, when the temperature was shifted to 31°C, filaments were only observed in ABE1, while ABE1ts continued to grow by budding (Fig. 2A and B, left panel). Thus, the S183P-mutation identified in bE2 leads to temperature-dependent functionality when introduced into bE1. Quantitative real-time qRT-PCR analysis revealed that the expression of bE1, bE1ts and bW2 is not altered as a result of the temperature shift (Fig. 2D and E). However, when bE1 protein levels were assessed by Western blot analysis using antibodies against the triple-myc tag, we detected significantly reduced bE1ts levels at 31°C, while protein levels of bE1 and bE1ts were comparable at permissive temperature. Thus, the observed temperature sensitivity of the mutant bEts is caused most likely by drastically reduced protein abundance at restrictive temperature.

image

Figure 2. bEts is functional at permissive temperature and instable at restrictive temperature. (A–F) ABE1 and ABE1ts harbour a triple-myc tagged bE1 and bE1ts allele controlled by the crg1 promoter, respectively. Strains are grown (A) on charcoal-containing MM plates and (B) in liquid array medium, under non-inducing conditions (glucose) and inducing conditions (arabinose) at permissive (22°C) and restrictive temperature (31°C) respectively. Induction of a functional b heterodimer is visible (A) as white filamentous spots or (B) as long filamentous cells. The white scale bar corresponds to 20 μm. (C) Western blot analysis on bE1-Myc and bE1ts-Myc obtained from the b induction experiment displayed in (B). Both proteins are about 54 kDa in size. (D–F) Real-time PCR analysis on RNA-transcripts obtained in the b induction experiment displayed in (B). (D) Relative expression of bE1-myc and bE1ts-myc controlled by the crg1 promoter. (E) Relative expression of bW2 controlled by the crg1 promoter. (F) Relative expression of the b-dependent genes clp1 and rbf1. (D–F) Gene expression is normalized to the constitutively expressed eIF2 gene. Mean expression values are displayed relative to the condition of lowest expression. Error bars indicate standard deviation of mean expression values.

Download figure to PowerPoint

To analyse if bE1ts function is sufficient to induce gene expression at permissive temperature, we investigated the expression of the b-dependently expressed genes clp1 and rbf1 by qRT-PCR. Both genes have b-binding sites located within their promoter regions and are directly regulated by the b heterodimer (Scherer et al., 2006; D. Schuler, A. Zahiri and J. Kämper, unpublished). clp1 expression levels were comparable at permissive temperature in ABE1 and ABE1ts; the change to restrictive temperature did not alter the expression in ABE1, while in ABE1ts clp1 expression decreased about 400-fold (Fig. 2F). Likewise, rbf1 expression was about 900-fold decreased under restrictive conditions in ABE1ts when compared with the control strain ABE1 (Fig. 2F). The rbf1 expression level was found to be threefold reduced in ABE1ts compared with ABE1 also under permissive conditions (Fig. 2F), which may argue that the bE1ts protein is less functional than the native bE1 protein. However, despite this potentially reduced functionality, the bE1ts and bE2ts proteins are sufficient to trigger filamentous growth (Fig. 2A and B), and to complete the pathogenic lifestyle of U. maydis up to the formation of fungal teliospores (see below).

The b heterodimer is essential for biotrophic development of Ustilago maydis

To test the effect of the bE2ts allele at restrictive temperature during in planta development of U. maydis, we introduced the bE2ts allele and, as a control, the wild-type bE2 gene into U. maydis stain FB1otef:pra2 (a1pra2/b1). This strain expresses an active pheromone/pheromone receptor system (mfa1, pra2), which is required for increased expression of the bE and bW genes (Bölker et al., 1995), and the bE1 and bW1 genes of which the latter is needed to form the complex with bE2ts. Since the resulting strains, RAbE2ts and RAbE2, harbour the active bE1/bW2 combination, they should be able to infect plants without a mating partner. To test bE2 function in the two strains, they were spotted on charcoal containing media plates. In RAbE2ts, filaments were formed only at permissive temperature, while RAbE2 grew filamentous at both temperatures (Fig. 3A). We then infected maize plants with the solopathogenic strains RAbE2ts and RAbE2 at 22°C, and shifted the plants to restrictive temperature (31°C) at different time points (0, 2, 4 days post infection, dpi). Plants were scored for infection symptoms at 7 dpi (Fig. 3B). All plants infected with RAbE2 developed severe disease symptoms, irrespective of the time-point shifted to the higher temperature (Fig. 3B). The plants infected with RAbE2ts that were shifted to the restrictive temperature, however, developed drastically attenuated symptoms. Most plants showed only chlorosis, and, pending on the length they were exposed to the restrictive temperature, either no (0 and 2 dpi) or drastically reduced tumour development (4 dpi). When continuously grown at permissive temperature, plants infected with RAbE2ts developed symptoms that were indistinguishable from those infected with RAbE2 (Fig. 3B). These results clearly demonstrate that the function of the b heterodimer is indispensable for biotrophic development of U. maydis.

image

Figure 3. Temperature-dependent inactivation of b function in planta abolishes tumour formation and fungal proliferation. A. Filamentous growth of RAbE2 harbouring the wild-type bE2 gene and RAbE2ts harbouring the temperature sensitive bE2tsP183 allele on PDC plates at permissive (22°C) and restrictive temperature (31°C). The progenitor strain FB1otef:pra2 was used as negative control. B. Pathogenic development of the temperature-sensitive solopathogenic strains RAbE2 and RAbE2ts at permissive and restrictive conditions. Plants were infected with both strains, respectively, and either kept constantly at permissive, constantly at restrictive temperature, or were shifted from permissive to restrictive temperature 2 and 4 dpi. Pictures were taken and tumour formation was analysed 7 dpi. Percentage and range of tumour formation of infected plants are colour-coded (n = total number of plants analysed). C. Clorazol Black E staining of RAbE2 and RAbE2ts hyphae in planta at restrictive and permissive temperature 3 dpi (kept at 22°C or shifted to 31°C at 2 dpi). Hyphal morphology of RAbE2ts is indistinguishable from that of RAbE2 at permissive temperature, whereas at restrictive temperatures the tip cells of hyphae are severely enlarged (white arrows). D. Visualization of nuclei by expression of a nuclear localized 3xGFP after switching off b activity 2.5 dpi. After 2 dpi, infected plants were shifted from permissive (22°C) to restrictive (31°C) conditions for 12 h or kept at 22°C as control. RAbE2ts hyphae contain one nucleus per cell at permissive temperatures as observed for RAbE2. At restrictive temperature (31°C) the enlarged hyphal tip cells contain multiple nuclei (white arrows). White scale bars correspond to 20 μm (C and D).

Download figure to PowerPoint

To study the role of the b heterodimer during in planta proliferation of U. maydis in more detail, we examined the growth of RAbE2ts-filaments microscopically after infection at permissive and restrictive temperatures. Chlorazol Black E staining of fungal hyphae revealed that proliferation of RAbE2ts at 22°C was comparable to that of RAbE2 (Fig. 3C). However, at 31°C, the RAbE2ts cells were found to be enlarged, especially the cells of the hyphal tip (Fig. 3C). To visualize nuclei, RAbE2ts and RAbE2 were transformed with a triple GFP gene that was N-terminally fused to a nuclear localization sequence (NLS) and controlled by the strong plant inducible mig2–5 promoter (Zheng et al., 2008). In planta, RAbE2 hyphae were composed of cells that each contained a single nucleus, as observed previously in the solopathogenic strain SG200 (Fig. 3B; Scherer et al., 2006). Similarly, the hyphae of RAbE2ts contained single nuclei at permissive temperature. When plants were grown at restrictive temperature, however, the enlarged cells were found to contain multiple nuclei (Fig. 3D), indicating that inactivation of the b heterodimer leads to abnormal cell growth, while cell cycle and nuclear divisions persist.

Temperature inactivation of the bEts affects the expression of pathogenicity factors in planta

To investigate the influence of the b heterodimer on the U. maydis transcriptome during in planta proliferation, we performed DNA array analyses using the Affymetrix U. maydis gene chip. After infections with the solopathogenic strains fungal biomass in whole leaf samples was insufficient to resolve the transcriptome of U. maydis due to the high background of plant-derived mRNA (data not shown). To yield sufficient fungal biomass for DNA-microarray analysis, strains RAb1ts (a1, bW1bE1ts) and RAb2ts (a2, bW2bE2ts) were constructed. These compatible strains were able to develop an infectious dikaryon harbouring the compatible, temperature-sensitive b heterodimers bW1/bE2ts and bW2/bE1ts (Fig. S1A). At permissive temperature, infections with a mixture of RAb1ts and RAb2ts gave symptoms comparable to that of infections with the compatible wild-type strains FB1 (a1b1) and FB2 (a2b2), leading to tumours that were filled with spores 14 dpi (Fig. S1B–D). As expected, pathogenic development of the RAb1ts/RAb2ts dikaryon was stalled at restrictive temperature (Fig. 4A and Fig. S1B, C).

image

Figure 4. b inactivation in planta alters the expression of genes important for fungal virulence. A. Time series of enlarged tip cell formation starting 4 dpi after infection with RAb1ts/RAb2ts crossings to determine the shortest duration of b inactivation in planta, which is suitable for expression analysis. Black arrows indicate enlarged tip cells. Scale bar corresponds to 50 μm. B. Pie chart of differentially expressed genes after b inactivation comparing bon and boff states at 31°C in planta. Categories classifying individual genes are indicated. Numbers of differentially expressed genes belonging to the individual categories are displayed in the chart. The number of genes within the two subgroups of secreted proteins (green) is significantly enriched (P-value of 7.6 × 10−12; only the secreted proteins identified by Target-P were considered (34; see Experimental procedures). C. Correlation matrix of U. maydis microarrays performed with RNA from plant tumours obtained from infections with crossings of FB1/FB2 and RAb1ts/RAb2ts; compared are the bon and boff states at permissive and restrictive conditions 5 dpi respectively. The correlation matrix was calculated with expression values of all genes represented on the DNA chip to display similarities in gene expression of a single array compared with all others. Expression profiles of FB1/FB2 and RAbE1ts/RAbE2ts are similar at permissive conditions, whereas their profiles differ significantly at restrictive conditions. Furthermore, the effect of temperature on gene expression is more significant than the effect of b inactivation, as arrays with plants infected at 31°C are more similar to each other than to those infected at 22°C. The colour scale bar indicates the similarity of one array compared with another (red = similar, green = different).

Download figure to PowerPoint

For DNA array expression analysis, plants were infected with either a mixture of RAb1ts and RAb2ts or of FB1 and FB2. Infected plants were then kept for 5 days at 22°C, or shifted from 22°C to the restrictive temperature 31°C at 111 h post infection for a period of 9 h. This period was sufficient to stall proliferation for the majority of RAb1ts/RAb2ts hyphae (Fig. 4A).

In a first set of experiment we compared the expression profiles of FB1/FB2 and RAb1ts/RAb2ts at restrictive and permissive temperature by performing one single array experiment each. The RAb1ts/RAb2ts and FB1/FB2 infections at permissive temperature (22°C) revealed comparable transcription profiles, demonstrating that the bEts function is sufficient for the gene regulation during biotrophic growth under permissive conditions (Fig. 4C). Comparative analysis of all four samples revealed that the temperature effect had a more prominent effect on global gene expression than the effect of b inactivation (data not shown, Fig. 4C). Thus, the comparison of RAbE1ts/RAbE2ts at permissive and restrictive temperatures was not suitable to analyse b-dependent gene expression in planta. Therefore, to eliminate changes in gene expression induced by the increase of the temperature, the bon and boff states in planta were analysed by comparison of FB1/FB2 to RAbE1ts/RAbE2ts under equal conditions at 31°C. To this end, we performed two additional biological replicates each. Correlation analysis comparing all experiments revealed three main clusters with related expression profiles (Fig. 4C). The first cluster is formed by the three independent RAbE1ts/RAbE2ts infections at 31°C, the second by the three FB1/FB2 infections at 31°C and the third by the two infections with wild-type and mutant strains at permissive conditions (Fig. 4C).

The comparison of FB1/FB2 to RAbE1ts/RAbE2ts at 31°C revealed 81 genes to be differentially expressed in response to the inactivation of the bE/bW heterodimer (for filter criteria see Experimental procedures, Table S1). The expression of six genes identified by the array analysis was exemplarily assessed by qRT-PCR, which showed comparable results (Table S4). We did not observe a significant alteration of specific functional categories among the 81 deregulated genes. However, a group of 42 genes (52%) coding for predicted secreted proteins was significantly enriched, including the most downregulated gene um05027 (Fig. 4B, Table S2, criteria for prediction of secreted proteins see Experimental procedures). Twelve of these genes have been described previously as part of in planta induced gene clusters for secreted proteins, several of which were shown to be important for fungal virulence (Kämper et al., 2006; Table S2). Ten of the genes organized in the clusters 2A (two genes – increased virulence), 6A (two genes – reduced virulence), 9A (one gene – unaffected) and 19A (five genes – markedly reduced virulence) were downregulated, whereas only two genes affecting cluster 2B (unaltered virulence) were induced (Kämper et al., 2006; Table S2). Only 10 of the 42 genes coding for potentially secreted proteins have a functional annotation, three of which encode for cell wall degrading enzymes (endoglucanase, um06332; endochitinase, um06190; pectine lyase, um10671; Tables S1 and S2).

Of the 345 b-dependently regulated genes that have been recently identified by DNA microarray analysis monitoring b induction in axenic culture (K. Heimel, M. Scherer and J. Kämper, unpublished), only 14 were differentially expressed after temperature restriction of b activity in planta (Table S3). One of these genes is clp1, which is a direct target gene of the b heterodimer (Scherer et al., 2006; D. Schuler, A. Zahiri and J. Kämper, unpublished) and the third most downregulated gene (11.3-fold) after switching off b activity in planta. Furthermore, we observed an induction of the pheromone and the pheromone receptor genes (mfa1, mfa2, pra1 and pra2; Table S3) known to be downregulated by an active b complex in axenic culture (Urban et al., 1996). In accordance, 18 pheromone-dependent genes were upregulated (Tables S3 and S4; Urban et al., 1996; Zarnack et al., 2008). Amoung these genes is rbf1, which encodes a transcription factor with central function in the b-dependent regulatory cascade. The rbf1 gene is directly regulated by the bE/bW heterodimer, but in addition it is also induced via the pheromone-dependent signalling cascade (K. Heimel, M. Scherer and J. Kämper, unpublished; Zarnack et al., 2008). Although the expression level of rbf1 was below the detection level of the DNA-microarrays, qRT-PCR revealed that the gene was 3.5-fold induced upon b inactivation (Table S4).

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

Temperature inactivation of the b heterodimer during in planta development of U. maydis blocks fungal proliferation

We could show that the b heterodimer is not only essential to establish the heterodikaryon after mating of two compatible sporidia and to initiate fungal pathogenicity, but also to sustain fungal proliferation in planta and to facilitate sexual reproduction.

bE2ts183P, a temperature-sensitive allele of bE2, harbours a single mutation, which leads to a drastic reduction of protein abundance at restrictive temperature. The mutation leads to an amino acid exchange at the border of the third α-helix of the homeodomain. The 3D structural analysis of the homeodomains of Antennapedia from Drosophila melanogaster and MATa1 from Saccharomyces cerevisiae revealed that the positions corresponding to proline 183 in bE are part of the last loop of the DNA binding α-helices (Fig. S2A and B). One possible explanation for the observed low bEts protein level is that the mutation leads to temperature-dependent misfolding of the protein, which then triggers protein degradation (Goldberg, 2003; Medicherla and Goldberg, 2008). But irrespective from the underlying mechanism, the temperature sensitivity achieved by the mutation made the bEts allel a suitable tool to study b-dependent gene expression in planta.

Shutting off the activity of the b heterodimer during in planta development of the fungus leads to an enlargement of fungal tip cells that contain multiple nuclei, indicating a defect in cell division and/or an uncoupling of the cell cycle control from cytokinesis. One of the previously described functions of the b heterodimer is the control of the mitotic cell cycle. In axenic culture, induced expression of an active bE/bW heterodimer leads to a G2 cell cycle arrest, reminiscent to the G2 arrest observed after fusion of two sporidia in the resulting dikaryon (Snetselaar and Mims, 1992; García-Muse et al., 2003; Scherer et al., 2006; Cánovas and Pérez-Martín, 2009). However, after plant penetration, the cell cycle arrest must be released to allow hyphal proliferation. bE/bW expression can be detected during the entire biotrophic phase (Quadbeck-Seeger et al., 2000; M. Vranes and J. Kämper, unpubl. data). It is thought that the function of the heterodimer is modulated by action of Clp1, a protein that was shown to counteract b function (Scherer et al., 2006). Clp1 is expressed within the nucleus at the time point of cell cycle release upon plant penetration; however, the ‘plant signal’ triggering this event is still unknown (Scherer et al., 2006). Clp1 was also found to be essential for the formation of clamps, a specialized structure necessary for the distribution of nuclei and cell division in planta, and subsequently for cell division of the dikaryotic hyphae in planta (Scherer et al., 2006).

We have now identified the clp1 gene as the top-third most downregulated gene upon b inactivation (11.3-fold downregulated). It is well possible that the loss of b function in the bEts strains (controlling cell cycle) and of the b-dependently regulated gene clp1 (controlling b function and cell division) leads to an uncoupling of cell cycle and cytokinesis, which would account for the multinucleated, enlarged tip cells formed after inactivation of b.

Pheromone-dependent genes are derepressed due to b inactivation in planta

We observed the upregulation of a total of 23 genes upon b inactivation that were previously shown to be induced via the pheromone pathway (Table S3; Hartmann et al., 1996; Urban et al., 1996; Zarnack et al., 2008). It has been shown that formation of an active b heterodimer leads to the repression of the pheromone and pheromone-receptor genes (Urban et al., 1996; Hartmann et al., 1999). Since the infectious dikaryon harbours a compatible combination of both pheromone- (mfa1 and mfa2) and both receptor-genes (pra1 and pra2) it is conceivable that, upon b inactivation, the derepression of the pheromone/receptor genes leads to an activation of a-dependent genes. One of the a-dependently induced genes upregulated upon b inactivation is rbf1 (Tables S3 and S4; Zarnack et al., 2008). Rbf1 encodes a transcription factor that serves as a central node in the b-regulatory cascade, involved in the regulation of the majority of b-dependently regulated genes (Scherer and Kämper, unpubl. data; Scherer et al., 2006). Thus, the derepression of the a-pathway may maintain the expression of rbf1 in planta, resulting in expression of b-regulated genes even after b inactivation.

The b heterodimer is necessary for the regulation of secreted proteins important for fungal virulence during pathogenic development

Secreted proteins of plant pathogenic bacteria, oomycetes and fungi have been shown to play crucial roles for the establishment of the different pathogenic lifestyles (Birch et al., 2006; Catanzariti et al., 2006; Chisholm et al., 2006; Kamoun, 2006; 2007; Kämper et al., 2006; O'Connell and Panstruga, 2006; Ridout et al., 2006; Morgan and Kamoun, 2007). For U. maydis, up to 750 proteins have been predicted to be secreted, dependent on the stringency of the method used for secretion signal prediction (MUMDB; Kämper et al., 2006; Müller et al., 2008). Recently, we have shown that several of these predicted proteins were (i) specifically expressed during in planta development and (ii) organized as clusters within the U. maydis genome (Kämper et al., 2006). In total, 12 of these clusters of U. maydis-specific secreted proteins have been identified, of which 5 were found to be crucial for pathogenic development (Kämper et al., 2006).

Fifty-two per cent (n = 42) of the genes differentially expressed in the bEts strains encode proteins with a bioinformatic prediction to be secreted. We identified 12 genes organized in five of the described secreted clusters, among these five of the 26 genes of the largest identified ‘cluster 19’ (Table S2). Since not all genes were affected by the b heterodimer within these 5 clusters, we have to assume additional regulatory circuits that lead to the in planta expression of the entire cluster (Table S2). Intriguingly, b inactivation affects 3 out of the 5 clusters that were shown to be crucial for pathogenic development of U. maydis (clusters 2A, 6A and 19; Kämper et al., 2006; Table S2). Although it is currently not known which of the cluster genes are responsive for the observed pathogenicity phenotypes, our finding clearly demonstrates that the b heterodimer plays an essential role for the regulation of secreted proteins important for fungal virulence.

Most of the genes differentially regulated by b in planta, including the genes coding for secreted proteins, are not expressed after b induction in axenic culture (K. Heimel, M. Scherer and J. Kämper, unpublished), indicating that additional plant-specific signals and regulators modify b-mediated transcription. We have identified three genes coding for putative transcription factors (um06257, um10500, um01523) as downregulated after b inactivation during biotrophic growth (Table S1). It is well possible that these regulators integrate additional environmental cues into the b-dependent regulatory cascade to adapt the fungus to changing environmental conditions as the plant surface or different plant tissues.

Our results support the function of the b mating type locus as the determinant for pathogenic development. The b heterodimer can be positioned as master regulator within a transcriptional network for the spatial and temporal control of cell cycle and cell division, but, as we show now, also for genes required to establish and maintain the biotrophic interaction with its host plant.

Experimental procedures

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

Strains and growth conditions

Escherichia coli strain TOP10 (Invitrogen) was used for cloning purposes. U. maydis cells were grown at 28°C in YEPS (Tsukuda et al., 1988) as precultures for mating assays and plant infections. Mating assays were performed at 22°C (permissive) or 31°C (restrictive) temperature as described by Gillissen et al. (1992) on 1% charcoal containing potato dextrose (PDC) medium (Difco) or complete medium (Holliday, 1974). For screening of the plasmid library used for bE or bW mutagenesis all U. maydis strains harbouring plasmids were grown in hygromycin-containing media (200 μg ml−1). U. maydis derivatives of AB31 (Brachmann et al., 2001) were grown in liquid array medium (Scherer et al., 2006) supplemented with either 1% glucose or 1% arabinose for b induction assays. U. maydis strains used in this study are listed in Table 1.

Table 1.  Strains used in this work.
StrainRelevant genotypeReference
FBD11-21a1a2/b2b2Banuett and Herskowitz (1989)
FBD11-21 + pNEBbW1UHa1a2/b2b2 + bW1This work
FBD11-21 + pNEBbW1UH mutant librarya1a2/b2b2 + bW1mutThis work
FBD12-3a1a2/b1b1Banuett and Herskowitz (1989)
FBD12-3 + pNEBbW1UHa1a2/b1b1 + bW1This work
FBD12-3 + pFSbE2a1a2/b1b1 + bE2mutThis work
FBD12-3 + pFSbE2 mutant librarya1a2/b1b1 + bE2This work
FBD12-3 + pFSbE2ts52a1a2/b1b1 + bE2ts52This work
FBD12-3 + pFSbE2ts98a1a2/b1b1 + bE2ts98This work
FBD12-3 + pFSbE2tsP183a1a2/b1b1 + bE2tsP183This work
FB1otef:pra2a1pra2/b1P. Müller, unpublished
RAbE2a1pra2/b1bE2This work
RAbE2_nls3gfpa1pra2/b1bE2; Pmig2–5:NLS-3xeGFPThis work
RAbE2tsa1pra2/b1bE2tsThis work
RAbE2ts_nls3gfpa1pra2/b1bE2ts; Pmig2–5:NLS-3xeGFPThis work
FB1a1/b1Banuett and Herskowitz (1989)
FB2a2/b2Banuett and Herskowitz (1989)
JB1a1/Δb1Scherer et al. (2006)
JB2a2/Δb2This work
RAb1tsa1/bW1bE1tsThis work
RAb2tsa2/bW2bE2tsThis work
AB31Pcrg:bw2, Pcrg:bE1Brachmann et al. (2001)
ABE1Pcrg:bw2, Pcrg:bE1-mycThis work
ABE1tsPcrg:bw2, Pcrg:bE1ts-mycThis work

Plasmid and strain constructions

Plasmid pCR-Blunt-II-TOPO (Invitrogen) was used for cloning, subcloning and sequencing of genomic fragments and fragments generated by PCR. Plasmid pFSbE2 (Kämper et al., 1995) was used to construct the bE2 mutant libraries and the plasmids pFSbE2ts52, -ts98 and -tsP183, in all cases a 642 bp HindIII-XbaI bE2 fragment was cloned into the vector backbone [primer pairs: RW11 (5′-CAC TCC CAC CTT TAG CCT CTA ACA-3′) and RW12 (5′-CGC CAT ACT TGA TCC AGC TGA TC-3′)]. Plasmid pNEBUH (Weinzierl, 2001) was used for cloning of the bW1 gene including a 439 bp 5′ region upstream the ORF. The bW1 gene was amplified from pbW1-pcx (Kämper et al., 1995) in three individual reactions to insert the restriction sites needed for random PCR mutagenesis [359 bp KasI-KpnI fragment, primer pairs RW1 (5′-AGG CGC CTT TGC TGG ATC GTT TCG-3′) and RW2 (5′-GGG GAG ACA AAA GGG GTA CCT GAG-3′); 530 bp KpnI-EcoRI fragment, primer pairs RW3 (5′-CAT CCT CAG GTA CCC CTT TTG TCT-3′) and RW4 (5′-GCC TGC TCC AGA ATT CGG ACT GCT-3′); 1661 bp EcoRI-SphI fragment, primer pairs RW5 (5′-AGC AGT CCG AAT TCT GGA GCA GG-3′) and RW6 (5′-GGC ATG CGA GAA TTG TGA AAA GTA-3′)]. The three fragments were introduced into pNEBUH to yield plasmid pNEBbW1UH, which was used to construct the bW1 mutant libraries; for this purpose, the 530 bp KpnI-EcoRI fragment was mutagenized by misincorporation-PCR (primer pairs bW1rPCR1 5′-GGC GCA AGG AAA TGA ATG TGT GTG-3′ and bW1rPCR2 5′-TGC TTT GGC TTG AGT CCA GTG ACC-3′). The respective bE2 and bW1 plasmids were transformed in FB11-21 and FB12-3 for screening purposes (Banuett and Herskowitz, 1989). The strains RAbE2 and RAbE2ts were constructed by stable, ectopic integration of SspI-linearized plasmids pFSbE2 and pFSbE2P183 in the genome of FB1otef:pra2 respectively. FB1otef:pra2 is an FB1 derivative (Banuett and Herskowitz, 1989), in which the otef:pra2 construct was integrated into the ip-locus (Brachmann, 2001; P. Müller, unpubl. data). The plasmid pUThsp (Brachmann et al., 2001) was used as backbone to clone the 3376 bp NotI Pmig2–5:NLS-3xeGFP fragment from pMS76 (Scherer et al., 2006). The resulting plasmid pRWnlsGfp was linearized with SspI and ectopically integrated in RAbE2 and RAbE2ts respectively. The resulting strains were named RAbE2_nls3gfp and RAbE2ts_nls3gfp. For the generation of strains RAb1ts and RAb2ts pUmbE/bW was constructed, introducing an FseI-linker into the StuI site of pSL1180 (Pharmacia); subsequently, a 1533 bp NotI-XhoI bW2 fragment and a 1398 bp FseI/EcoRI fragment of bE2 were integrated into the respective sites. The EcoRI site was located 607 bp 3′ of the ORF of bE2, and the XhoI site was generated close to a HaeII site 573 bp 3′ of bW2. Recombinant PCR was used to integrate synonymous mutations to generate the FseI and NotI sites at amino acid position 234–235 in bE2 and at position 338–339 in bW2 respectively. Subsequently, the carboxin-resistance gene was inserted as EcoRV/SmaI fragment from pCBX122 (Keon et al., 1991), allowing targeted integration into the ip-locus of U. maydis (Loubradou et al., 2001). The resulting plasmid, pUmbE/bW, harbours the constant regions of bE2 and bW2 with unique NotI and FseI sites. The 2.2 kb fragments of the b1 and b2 locus comprising the N-terminal regions of bE (from amino acid 236) and bW (from amino acid 339) as well as the promoter region were amplified using primer pairs W2Not339 (5′-GCA CGC GGC CGC ATG TAA TCA AAG-3′) and E2Fse235 (5′-GAG TGG CCG GCC GAG GTT GTC TG-3′), generating synonymous mutations that introduce an FseI site at amino acid position 235/236 in bE and a NotI site at amino acid position 338/339 in bW. Digestion of the PCR products with NotI and FseI allows cloning into plasmid pUmbE/bW to reconstitute functional b alleles, resulting in the plasmids pTHEA2 (bW1bE1) and pTHEB5 (bW2bE2). Plasmids pTHEA2 and pTHEB5 were used for construction of bW1bE1ts and bW2bE2ts constructs respectively. The bE2ts mutation was integrated by cloning a 182 bp XhoI/XbaI fragment from pFSbE2tsP183 containing the bE2ts-mutation and parts of the homeodomain. The resulting plasmids pRAb1ts and pRAb2ts were linearized with AgeI for integration into the ip-locus (Brachmann, 2001) of strains JB1 (a1Δb1; Scherer et al., 2006) and JB2 (a2Δb2; FB2 derivative, in which the b2 locus was substituted as described for JB1 by Scherer et al., 2006). The resulting stains were named RAb1ts and RAb2ts respectively. AB31 was used as progenitor strain to construct ABE1 and ABE1ts (Brachmann et al., 2001). Primer pairs used to generate the flanks for homologous recombination were bE1_fw (5′-ACC AAC GAA TCC ACA GCA AGA G-3′) and bE1_rev_sfi (5′-GTT GGC CGC GTT GGC CGC GCC AAA CGC AGT AGA AAG G-3′) for the flank matching the C-terminus of bE1 and bE1ts, respectively, and RF_fw_sfi (5′-ACT GGC CTG AGT GGC CCA GTG ATA CGT TTA GTC CCT TTG-3′) and RF_rev (5′-CGA TCC GTA GTT GTG CGA GAG-3′) for the flank matching the 3′ UTR. Primer bE1-fw binds upstream of the ts mutation site in pRAb1ts; therefore, it is suitable to integrate the mutation into AB31. Primers bE1_rev_sfi and RF_fw_sfi carry the SfiI sites compatible to the triple-myc cassette of pUMA796 containing the hygromycin resistance gene (M. Feldbrügge, unpublished). The triple-myc construct was integrated into the b locus of AB31 by homologous recombination (Kämper, 2004). Sequence analyses of all fragments generated by PCR were performed with an automated sequencer (ABI 373 A; Applied Biosystems) and standard bioinformatic tools.

DNA procedures

Molecular methods followed described protocols (Sambrook et al., 1989). Random PCR mutagenesis was performed following the protocol from Spee et al. (1993). Transformation of U. maydis protoplasts with the indicated plasmids was performed as described previously (Tsukuda et al., 1988). DNA isolation from U. maydis and transformation procedures were performed as described (Schulz et al., 1990). Homologous recombination was verified by DNA gel blot analysis (Brachmann, 2001).

Western blot analysis

Strains were grown to an OD600 of 0.5 in liquid array medium with 1% glucose. For induction of the crg1 promoter, cells were shifted to array medium with 1% (w/v) arabinose and incubated for 6 h at 22°C and 31°C respectively. Pellets of 50 ml cultures were resuspendet in 2 ml lysis buffer [50 mM Tris-HCL, pH 7.5, 10% glycerol, 1 mM EDTA, 200 mM NaCl, 2 mM PMSF, 5 mM Benzimidine 1× Complete EDTA free (Roche)], frozen in liquid nitrogen and homogenized in a Retsch MM200 cell homogenizer at maximum frequency for 5 min. Homogenized material was spun down at 14 000  r.p.m. for 2 min at 4°C. Protein supernatant concentrations were equalized according to Bradford analysis. Fifty micrograms of total cell extracts was separated by SDS-PAGE and transferred on a PVDF membrane (Amersham) by electroblotting. The membrane was saturated with 3% non-fat dry milk in TBST (50 mM Tris-HCL, pH 7.5, 150 mM NaCl, 0.05% Tween20) for 30 min at room temperature. For detection of bE1-Myc and bE1ts-Myc a c-Myc monoclonal antibody (Clontech) was used (1:500, diluted in TBST containing 3% milk powder). As secondary antibody an anti-mouse IgG HRP conjugate (Promega) was used (1:4000, diluted in TBST containing 3% milk powder). For chemiluminescence detection, ECL Plus Western blot detection reagent (GE Healthcare) was used according to the manufacturers protocol.

Microscopy

Microscopic analysis was performed using a Zeiss Axioplan 2 microscope. Photomicrographs were obtained with an Axiocam HrM camera, and the images were processed with Axiovision (Zeiss) and Photoshop (Adobe). Chlorazole Black E staining of fungal cells in planta was performed as described (Brachmann et al., 2003). GFP was observed by fluorescence microscopy (excitation/emission for eGFP: 450–490 nm/515–565 nm)

RNA isolation

Ustilago maydis infected maize plants (Early Golden Bantam) were grown in a phytochamber in a 15 h/9 h light-dark cycle; light period started/ended with 1 h ramping of light intensity. Prior to infection with U. maydis temperature was 28°C (light) and 20°C (dark). Plantlets were individually sown in pots with potting soil (Fruhstorfer Pikiererde) and infected 7 days after sowing, 1 h before end of the light period, as described (Brachmann et al., 2001). After infection the plants were kept at 22°C (permissive temperature) or 31°C (restrictive temperature) depending on the requirements. Samples used for RNA preparation were collected 1 h before the end of the light period and directly frozen in liquid nitrogen for three independently replicates. For each experiment, 10 plants were sampled. Total RNA was extracted using Trizol reagent (Invitrogen) according to the manufacturer's instructions. RNA samples to be used for microarray analyses or real-time RT-PCR were further column purified (RNeasy; Qiagen) and the quality checked using a Bioanalyzer with an RNA 6000 Nano LabChip kit (Agilent).

RNA extraction from liquid culture was performed as described by Scherer et al. (2006).

DNA microarray

Affymetrix Gene ChipRUstilago genome arrays were done in three biological replicates, using standard Affymetrix protocols (staining: EukGe2V4 protocol on GeneChip Fluidics Station 400; scanning on Affymetrix GSC3000). Expression data were submitted to GeneExpressionOmnibus (http://www.ncbi.nlm.nih.gov/geo/), Accession GSE16501. Data analysis was performed using Affymetrix Micro Array Suite 5.1, the R bioconductor package (http://www.bioconductor.org/) and dChip1.3 (http://biosun1.harvard.edu/complab/dchip/), as described by Eichhorn et al. (2006). Probe sets present in at least two of the replicates were defined as ‘expressed’, resulting in 48% (3263 out of 6795) present calls in both wild-type and temperature sensitive strains. We considered changes > twofold with a difference between expression values >50 and a corrected P-value of < 0.01 as significant. For genes displayed by more than one probe set, the probe set giving the strongest signal intensity was chosen. Functional enrichment analyses were performed with the functional distribution tool integrated in the Ustilago maydis genome database (http://mips.helmholtz-muenchen.de/genre/proj/ustilago/Search/index.html). Enrichment analysis of secreted proteins was performed calculating the total number of secreted proteins expressed in the wild-type background 5 dpi (only considering the 754 genes containing Target-P signals, see below). A total of 395 probe sets (about 12% of the expressed probe sets) representing secreted proteins were identified to be present under these conditions. Hypergeometrical distribution analysis was performed and P-values < 0.01 were considered to be significant.

Quantitative real-time PCR analysis

For cDNA synthesis, the SuperScript III first-strand synthesis SuperMix assay (Invitrogen) was employed, using 1 μg of total RNA. qRT-PCR was performed on a Bio-Rad iCycler using the Platinum SYBR Green qPCR SuperMix-UDG (Invitrogen). Cycling conditions were 2 min 95°C, followed by 45 cycles of 30 s 95°C/30 s 65°C/30 s 72°C. To verify the data obtained by DNA-microarray analysis, real-time analyses were performed on um05027, um11413, clp1, pra1, mfa1, lga2 and also rbf1 using the identical RNA probes as for the DNA-array analysis. The U. maydis eIF2B (um04869) gene was used as reference. Primer sequences for rbf1, clp1 and lga2 are described in Scherer et al. (2006). Primer sequences were rt-eIF-2B-F (5′-ATC CCG AAC AGC CCA AAC-3′) and rt-eIF-2B-R (5′-ATC GTC AAC CGC AAC CAC-3′) for eIF2B, rt-5027-for (5′-CCA AAT TCA CCG TCT TCG CCT CTC-3′) and rt-5027-rev (5′-GTC GAG CTT GGT GTT GAG CGT GAG-3′) for um05027, rt-11413-for (5′-TAT GAG CCA AGA CCC CAC TCG ACT-3′) and rt-11413-rev (5′-AAA GAT GCG GCC TAA AAA GTT GGC-3′) for um11413, rt-pra1-for (5′-AAC CGA AGG CAT CTG CAC TGC-3′) and rt-pra1-rev (5′-CCC GCA TGT CGA TGT CAG ACT-3′) for pra1, rt-mfa1-for (5′-ATG CTT TCG ATC TTC GCT CAG AC-3′) and rt-mfa1-rev (5′-TAG CCG ATG GGA GAA CCG TTG-3′) for mfa1. bE, bW, clp1 and rbf1 were analysed using primer pairs described by Scherer et al. (2006).

Prediction of secreted proteins

Classical secretion signals (TP) were predicted with Target-P (http://www.cbs.dtu.dk/services/TargetP/). RC-values indicate the reliability classes (1–5), with class 1 having the highest probability to be secreted (Table S2; Emanuelsson et al., 2000). RC = 3 was used as cut-off. A complete list of 754 Ustilago maydis proteins that encompass this criteron can be obtained at the MIPS Ustilago maydis Database (http://mips.helmholtz-muenchen.de/genre/proj/ustilago/Search/index.html). Non-classical secretion (SP) was predicted with Secretome-P (http://www.cbs.dtu.dk/services/SecretomeP/). Obtained NN-scores indicate the prediction reliability (0–1), with 1 having the highest probability to be secreted (Table S2; Bendtsen et al., 2004). A NN-score >0.5 was used as cut-off.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

We would like to thank Regine Kahmann and the Max-Planck-Institute for terrestrial Microbiology, Marburg, Germany, for generous support, Philipp Müller (MPI for terrestrial Microbiology) for the strain FB1otef:pra2, David Schuler for strain ABE1, Volker Vincon (MPI for terrestrial Microbiology) for technical assistance and Miroslav Vranes and Kai Heimel (Institute for Applied Biosciences, Karlsruhe Institute of Technology, Germany) for critical comments on the manuscript. This work was supported by a grant from the International Graduate School GRK767, from the Research Group FOR666, both funded by the German Research Foundation, and by grants from the German Ministry of Science and Education for the U. maydis DNA array set-up.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information
FilenameFormatSizeDescription
MMI_6984_sm_Figures_S1-S2_and_Tables_S1-S4.pdf2872KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.