Laser microdissection of Arabidopsis cells at the powdery mildew infection site reveals site-specific processes and regulators

Abstract

To elucidate host processes and components required for the sustained growth and reproduction of the obligate biotrophic fungus Golovinomyces orontii on Arabidopsis thaliana, laser microdissection was used to isolate cells at the site of infection at 5 days postinfection for downstream global Arabidopsis expression profiling. Site-specific profiling increased sensitivity dramatically, allowing us to identify specific host processes, process components, and their putative regulators hidden in previous whole-leaf global expression analyses. For example, 67 transcription factors exhibited altered expression at the powdery mildew (PM) infection site, with subsets of these playing known or inferred roles in photosynthesis, cold/dehydration responses, defense, auxin signaling, and the cell cycle. Using integrated informatics analyses, we constructed putative regulatory networks for a subset of these processes and provided strong support for host cell cycle modulation at the PM infection site. Further experimentation revealed induced host endoreduplication occurred exclusively at the infection site and led us to identify MYB3R4 as a transcriptional regulator of this process. Induced endoreduplication was abrogated in myb3r4 mutants, and G. orontii growth and reproduction were reduced. This suggests that, by increasing gene copy number, localized endoreduplication serves as a mechanism to meet the enhanced metabolic demands imposed by the fungus, which acquires all its nutrients from the plant host.

Powdery mildews (PMs) are significant agronomic pathogens that infect cereals (e.g., barley); tomato; grapevine; and ornamentals, including roses (1). Although PM resistance has been long studied in cereal species, the genetic/genomic tractability and resources of Arabidopsis (2) allow for rapid systems-level data acquisition, hypothesis generation, and experimental validation. The reference (sequenced) strain of Arabidopsis, Arabidopsis thaliana Columbia-0 (Col-0), is susceptible to the adapted PM Golovinomyces orontii (3). G. orontii exclusively infects epidermal cells of A. thaliana Col-0 with defined and microscopically visible stages of infection, including germination of the conidia, penetration of the epidermal cell, development of the haustorial complex (the feeding structure through which the fungus acquires its nutrients), and further growth and reproduction of the fungus on the leaf surface (3). The fungal (asexual) reproductive structures known as conidiophores contain the conidia (spores). In contrast, nonadapted PMs, such as the barley PM, Blumeria graminis f. sp. hordei, typically fail in the penetration phase of the interaction with the nonhost Arabidopsis (3). Instead of focusing on resistance mechanisms, we are interested in elucidating processes associated with the sustained growth and reproduction of an adapted PM. We would expect to uncover specific mechanisms associated with the sustained alteration of plant cellular processes and metabolism that allow for the continued accommodation of the haustorial complex and the associated fungal demand for nutrients and water, while limiting host defense responses, including cell death.

We previously performed whole-leaf expression profiling over the growth and reproduction phase of the G. orontii–Arabidopsis interaction (4). Here, we specifically profiled cells at the infection site to enhance our sensitivity at detecting responses localized to the infection site and to reduce complexity associated with differential infection site-specific and distal expression. We employed UV laser microdissection (LMD) to isolate epidermal and mesophyll cells of interest, because microcapillary methods do not readily allow one to examine mesophyll cells underlying the infected epidermal cell and protoplasting with cell sorting has an impact on cell wall integrity, which can affect defense responses. To use fragile mature Arabidopsis leaf tissue for LMD and downstream global RNA expression profiling, we developed/optimized methods of tissue preparation, RNA isolation, and amplification. This then allowed us to explore, at a global level, how the host transcriptional machinery is altered at the site of a compatible PM interaction and to identify known and previously unreported host responses associated with the sustained growth and reproduction of an adapted PM.

Results and Discussion

Optimized Experimental and Quality-Control Procedures Allowed for High-Quality and Highly Correlated Global Expression Profiling at the Site of Infection.

To identify specific host genes associated with the sustained growth and reproduction of G. orontii, we isolated groups of epidermal and mesophyll cells (∼20 cells) surrounding G. orontii-infected epidermal cells from fixed serial leaf sections of WT Arabidopsis Col-0 using LMD at 5 days postinfection (dpi) (Fig. 1). Samples of equivalent cell types from uninfected WT leaf sections were also isolated. With our experimental conditions, the growth and reproduction phase of G. orontii infection was well established at 5 dpi, yet individual colonies were well separated spatially, enabling quantitation of fungal growth and reproductive structures. To minimize cell type-associated variation, chosen infection sites did not include stomatal guard cells or vascular bundles. The phytohormone salicylic acid (SA) plays a significant role in limiting this stage of the G. orontii infection (4). Therefore, we also performed parallel experiments using the SA biosynthetic Arabidopsis null mutant ics1-2 (5). To compare infection site-specific and whole-leaf responses directly, whole leaves from the same batch of infected and uninfected WT plants used for LMD were also processed.

Fig. 1.

LMD of Arabidopsis leaf epidermal and mesophyll cells at the PM infection site. (A) Representative transverse section of leaf is shown with LMD-targeted area boxed (magnification ×200). Targeted epidermal and mesophyll cells at PM infection site (magnification ×400) before (B) and after (C) LMD. (D) Captured laser microdissected cells. (E) PM-induced expression of PR1::GUS (blue) at 5 dpi. Arrow indicates G. orontii haustorium in infected epidermal cell.

We optimized/developed tissue preparation (6), RNA isolation, and amplification methods to obtain high-quality RNA from pooled groups of LMD-isolated cells for global expression profiling. Two-round amplification of RNA extracted from 7,500 LMD-isolated cells (∼1 ng) yielded >20 μg of amplified RNA for reproducible expression profiling using the Affymetrix Arabidopsis ATH1 GeneChip, with highly correlated biological replicates (r ≥ 0.98; SI Appendix). To ensure that our LMD results were an accurate readout of the PM response, we performed parallel experiments to assess the separate impact of tissue preparation, LMD, and RNA amplification on RNA quality and gene expression. We found LMD and RNA amplification procedures did not appreciably affect our gene expression output; however, a very small set of genes (0.5% of total) was affected by the tissue preparation procedure and was removed from subsequent analyses (SI Appendix).

Arabidopsis Exhibits an Extensive Site-Specific Response to PM.

Using a conservative selection cutoff (≥2-fold change in infected vs. uninfected samples, p ≤ 0.05, false discovery rate <5%), we found that 6.5% (1,483 transcripts) of genes on the array exhibited PM infection-specific expression* in LMD-isolated cells. Quantitative RT-PCR (qPCR) performed for a set of 11 genes spanning the range of observed expression levels verified the array results (SI Appendix). Known PM-affected genes exhibited 2- to 21-fold enriched induction at the site of infection compared with parallel whole-leaf samples (SI Appendix). A promoter–reporter line for one of these genes (the pathogenesis-related protein PR1) confirmed that PR1 induction was site-specific (Fig. 1E), consistent with its 137-fold induction at the site of infection and dramatically enriched expression by qPCR.

Site-Specific Profiling Identifies Previously Hidden Localized PM Responses and Associated Transcriptional Regulators.

To determine whether site-specific profiling at 5 dpi uncovered previously hidden localized G. orontii responses, we compared PM-affected genes from our 5-dpi LMD dataset with the parallel 5-dpi whole-leaf dataset and our previous whole-leaf PM time series from 1 to 7 dpi (4). One thousand eighty-nine genes were exclusively identified in our site-specific analysis (Fig. 2A and SI Appendix). This suggests these genes are expressed specifically at the PM infection site at levels too low to permit detection in whole-leaf samples. Alternatively, these genes may exhibit more complex patterns of expression that are cell type- and/or distance from infection-dependent and are merged in a whole-leaf sample, thereby masking detection of an altered response.

Fig. 2.

Arabidopsis genes with differential expression at the G. orontii infection site. (A) Arabidopsis genes with differential expression at the G. orontii infection site (LMD PM) compared with parallel whole-leaf samples and whole-leaf PM time series data from the article by Chandran et al. (4). (B) Pie chart showing TFs with altered expression at the site of infection classified by functional process. Bold loci represent genes with established roles in associated functional processes; nonbold loci represent genes identified through correlation analysis.

Very few transcription factors (TFs)^† with a functional impact on a PM interaction have been identified. Of the 6 TFs with established roles mediating PM resistance/susceptibility (3, 4), only one, PUX2, exhibited altered expression at the site of PM infection. TFs often exhibit low-level and highly localized expression changes; therefore, we expected our site-specific expression analysis to identify unique TFs with important functional roles in the PM interaction. We found 67 TFs exhibited altered expression at the site of infection, with 79% not previously associated with a compatible PM interaction (4, 7–9). Although many of these PM site-specific TFs have no defined function, we identified sets of TFs associated with the following processes: photosynthesis, drought/cold tolerance, pathogen defense and cell death, auxin signaling/response, and the cell cycle (Fig. 2B). This suggests these host processes are modulated during the sustained growth and reproduction of this adapted PM.

To identify functional processes associated with the PM infection site further, we examined the genes and promoters of genes in our LMD dataset to determine altered host processes and enriched cis-acting regulatory elements (SI Appendix). For a subset of processes with known process-associated TFs, we then constructed putative process networks incorporating target information, when known, and coregulation analysis (SI Appendix).

PM Site-Specific Profiling Supports Diminished Photosynthesis and a Carbon Source to Sink Transition and Elucidates a Photosynthesis Regulatory Network.

Decreased photosynthetic rates have long been observed in response to compatible PM infections (10). However, the regulatory network associated with this response had not been defined. We found expression of genes involved in photosynthesis and related processes [e.g., chlorophyll (Chl) in the tetrapyrrole synthesis category] was significantly decreased at the site of infection (Fig. 3A and SI Appendix). Although a decrease in photosynthesis-associated gene expression had previously been observed in whole leaves in response to PM (4, 7, 11), our focus at the site of infection allowed us to identify a more extensive set of altered photosynthesis-related genes. It also eliminated potential complications associated with differential PM-induced impacts on photosynthesis at the site of infection compared with nearby or distal sites. In biotrophic interactions, repression of photosynthetic gene expression has been attributed to increased hexose accumulation at the site of infection (12, 13). Consistent with an increased demand for hexoses by the PM at the infection site, we observed a site-specific increase in expression of a sucrose phosphate synthase and the cell wall invertase BFRUCT1 that converts sucrose to the hexoses glucose and fructose (SI Appendix). We also observed a PM infection site-specific enrichment in the hexose transporters STP4 and AtPLT5 (SI Appendix). Because BFRUCT1, STP4, and AtPLT5 expression is associated with sink organs, our data suggest the PM carbon demand has converted the infection site from a carbon source to sink (13). In Fig. 3C, we present a simplified photosynthesis process network that links localized enhanced hexose to altered photosynthetic gene expression (detailed in SI Appendix). A 20-fold induction in the previously undetected MYB-like TF At3g11280 was observed at the PM infection site. This MYB-like protein has been shown to interact in vitro with the hexokinase I nuclear complex involved in glucose-mediated repression of Chl a/b binding genes (14), which we found had reduced expression at the infection site. Thirty-eight of the top 300 genes correlated with At3g11280 expression exhibited altered (almost exclusively induced) expression at the PM infection site, including the bHLH TF At4g14410. We also identified a known positive transcriptional regulator of Chl synthesis, sigma factor B (SIGB), with reduced expression at the PM infection site. Of the top 300 genes correlated with SIGB, 63 exhibited reduced expression at the site of infection, including numerous genes encoding Chl biosynthetic proteins and plastidic ribosomal proteins (RPs).

Fig. 3.

PM infection site-specific processes and predicted response networks. Functional classification of WT infected vs. uninfected LMD (A) and the ics1 mutant vs. WT-infected LMD (B) samples. MAPMAN categories or subcategories (P ≤ 0.05 when category was not significant) with bin numbers in parentheses. ABC, ATP-binding cassette; CHO, carbohydrate. Predicted site-specific response networks associated with photosynthesis (C) and cold/dehydration response (D). TFs with established roles (filled boxes) are connected to other TFs and genes in functional categories by lines with arrows (positive regulation) or bars (negative regulation) at the end. TFs in open boxes were identified as highly coexpressed, as indicated by dashed lines. Enhanced (red) and decreased (green) expression.

Site-Specific Up-Regulation of Respiration May Be Required to Accommodate Sustained Metabolic Demand.

Enhanced respiration has been measured in a PM interaction with susceptible barley, with approximately half of this enhanced respiration associated with the alternative pathway (15). In highly resistant cultivars, enhanced respiration was not observed, whereas in moderately resistant cultivars, slightly enhanced respiration was observed but the alternative pathway was not active (15). In our study, genes involved in respiration, including glycolysis, the tricarboxylic acid cycle, and the mitochondrial electron transport chain, exhibited increased expression in response to PM at the infection site (Fig. 3A and SI Appendix). These included genes associated with the alternative pathway and the γ-aminobutyrate shunt, which could allow for enhanced respiration under stress (15, 16). Taken together with previous findings, our data suggest that enhanced respiration primarily functions to provide the energy required to meet the elevated metabolic demands imposed by the fungus.

Genes Associated with Early PM Nonhost Penetration Resistance Exhibited Altered Expression at the Site of a Compatible PM Interaction at 5 dpi.

The expression of genes involved in localized SNARE-mediated responses important for penetration resistance of nonhost plants (17) exhibited enhanced expression at our adapted PM infection site (SI Appendix). These proteins have been shown to aggregate specifically at the site of PM infection at early time points (≤1 dpi) associated with fungal penetration and/or haustorial complex formation (17). Our observed enriched expression of these genes at the PM infection site at 5 dpi supports a more general role for these SNARE-mediated processes in defense-associated secretion, as does the enhanced growth and reproduction of G. orontii on plants with a lowered gene dosage of VAMP721 and VAMP722 (17). Nonhost penetration resistance also requires specific activated indole-3-glucosinolates (I3Gs) (18) that may function directly as antifungal agents and/or in defense signaling, resulting in and associated with callose formation (18, 19). We observed a dramatic up-regulation and enrichment in expression of genes involved in the synthesis of I3Gs from Trp at the infection site (SI Appendix), suggesting compatibility could result from the delayed formation of these defense compounds. Consistent with this possibility, a study focused on haustorial development (8–24 hpi) found enhanced AtCYP83B1 expression in whole leaves was delayed when infected with an adapted vs. nonadapted PM (9); CYP83B1 directs flux from a shared precursor to I3G synthesis. Alternatively, a specific adaptation of G. orontii to its cruciferous I3G-producing hosts could allow it to tolerate I3Gs, whereas nonadapted PMs do not.

Repression of abscisic acid (ABA) biosynthesis and dehydration-responsive gene expression is associated with the PM penetration resistance of nonhost plants (20), and we found cold/dehydration-responsive gene expression decreased at the PM infection site at 5 dpi. Five TFs associated with cold/drought tolerance but with no previous reported functional role in a PM interaction exhibited altered expression at the site of PM infection (Fig. 2B). We integrated four of these into a model depicting the down-regulation of cold/dehydration-responsive genes at the site of infection (Fig. 3D and SI Appendix). DREB1A/CBF3 is a cold inducible-specific TF that regulates expression of target genes by binding ABRE and/or DRE cis-acting promoter elements, resulting in enhanced tolerance of cold and dehydration stress (21). DREB1A direct targets include KIN1, ERD10, HVA22d, and COR47 (21), all of which display reduced expression at the site of PM infection, and ABRE and DRE/C-repeat cis-acting elements were enriched in the promoters of these cold/dehydration-associated genes. Two known negative regulators of DREB1A, MYB15 and ZAT12/RHL41, were induced at the PM infection site, consistent with the observed decreased expression of DREB1A. Furthermore, expression of NFXL1, a positive regulator of osmotic stress tolerance, was reduced at the PM infection site with coregulated genes with altered PM expression associated with the response to cold. Coregulation analysis identified seven additional TFs, including ERF2, a known positive regulator of jasmonic acid (JA)-mediated defensive responses, as well as a subset of genes associated with JA-mediated signaling. Perhaps the observed down-regulation of cold/dehydration-responsive genes and concomitant up-regulation of a subset of JA-responsive genes limit the extent of the compatible PM interaction. Alternatively, they may permit the continued fungal acquisition of water from host cells at the infection site by allowing for enhanced cellular water content. These latter responses would be integrated with other regulators of stomatal opening and cellular water content not discussed here.

Genes Involved in Calcium Signaling and Redox Regulation Largely Exhibited Increased Expression at the Site of PM Infection in an SA-Dependent Manner.

Calcium signaling plays an important role in PM penetration resistance (22) and, more generally, in abiotic/biotic stress response. Our site-specific analysis allowed us to identify specific components of the calcium response likely to mediate this phase of a compatible interaction (Fig. 3 A and B and SI Appendix). For example, calmodulin AtCML9 was induced 19-fold at the site of PM infection. AtCML9 acts as a negative regulator of ABA-regulated dehydration responses, with the cml9 mutant exhibiting hypersensitivity to ABA and enhanced dehydration tolerance (23). Thus, because CML9 expression is SA-dependent, it may integrate SA- and ABA-associated responses.

SA-mediated changes in redox status regulate the conformation of NPR1, a master regulator of SA-mediated defense genes, through thioredoxins such as TRXh5, allowing the active NPR1 monomer to translocate to the nucleus and activate the expression of pathogenesis-related genes such as PR1 (24). We observed an SA-dependent enriched induction of both TRXh5 and PR genes at the infection site at 5 dpi, emphasizing the localized nature of these redox-mediated responses (Fig. 3 A and B and SI Appendix). However, these SA-dependent responses are not sufficient to result in resistance. Although enhanced SA synthesis and signaling have been associated with resistance to compatible PMs for a number of mutants, elevated SA and resistance were also correlated with cell death in these cases (3).

Site-Specific Profiling Suggests Modulation of the Cell Cycle at the PM Infection Site.

The promoters of genes with altered expression at the site of infection are enriched in cis-acting regulatory elements associated with S- and M-phases of the cell cycle (SI Appendix). The site II element, an S-phase-associated element bound by TCP domain TFs, was enriched in the promoters of genes with altered expression at the PM infection site. TCP TFs are known to regulate the expression of RPs required for protein synthesis and to couple cell growth and division (25). Although we do not detect TCP TFs with altered expression at the infection site, nonplastidic RP expression is dramatically enhanced. We also observed enrichment in the cell cycle-1b cis-acting regulatory element. These elements are also associated with the S-phase of the cell cycle and are found in the promoters of histones, a number of which exhibited enhanced expression at the site of infection. In addition, we identified six TFs associated with the cell cycle-related processes transcription, translation, and cell elongation (Fig. 2B). For example, the transcription initiation factor IIF β-subunit and transcription elongation factor SPT42 exhibited enhanced expression at the site of infection, consistent with enhanced mRNA synthesis. Finally, the mitosis-specific activator (MSA) element was enriched in promoters of genes with altered expression at the PM infection site, as was expression of MYB3R4, a TF that binds the MSA element and activates G₂/M progression (26).

G. orontii–Induced Host Endoreduplication Is Observed at the PM Infection Site.

Although modulation of the cell cycle had not been previously associated with the sustained growth of an adapted PM, our site-specific global expression profiling suggested this possibility. Therefore, we performed microscopic analyses to determine whether enhanced cell division or endoreduplication (progression through the growth and replication phases of the cell cycle without mitosis) occurred at the infection site. We did not observe microscopic evidence of enhanced cell division at the PM infection site or enhanced expression of cell division reporters (SI Appendix). However, we observed a clear visual difference in DAPI staining of DNA at the site of PM infection for both epidermal and underlying mesophyll cells (Fig. 4A). We further focused on quantifying these differences for the mesophyll cells directly underlying the G. orontii-infected (haustoria-containing) epidermal cell, because there is less natural variation in ploidy of mesophyll cells compared with leaf epidermal cells (27), and the observed differences were more dramatic for these underlying mesophyll cells. We found these underlying mesophyll cells were significantly larger, with increased ploidy and nuclear size compared with replicate uninfected samples of equivalent cell types and locale (Fig. 4 A and B). In contrast, mesophyll cells distal to the infection site exhibited no increase in ploidy, nuclear size, or cell size compared with uninfected cells (SI Appendix). The mesophyll cells underlying the infection site exhibited increased DNA content up to 64C at 5 dpi with a median of 32C, whereas DNA content in mesophyll cells in parallel uninfected leaves or distal to the infection site had a median of 8C. In all cases, DNA content quantified by DAPI staining was proportional to nuclear size (SI Appendix).

Fig. 4.

PM-induced endoreduplication at the infection site is regulated by MYB3R4. Epidermal layer showing guard cell nuclei in uninfected (UI) leaves and germinated G. orontii conidium (c) in infected leaves at 5 dpi in WT Col-0 (A) and myb3r4 mutant (C) plants. Arrows point to a representative DAPI-stained nucleus in the guard cell or underlying mesophyll layer. (Scale bar, 10 μm.) Nuclear DNA quantitation by DAPI fluorescence of underlying mesophyll cells normalized to 2C guard cells for WT Col-0 (B) and myb3r4 mutant (D). Log₂ (RFU/RFU_{guard cell}) corresponds to C-values as follows: up to 0 is 2C, 0–1 is 4C; 1–2 is 8C; 2–3 is 16C; 3–4 is 32C, and 4–5 is 64C. RFU, relative fluorescence unit. Results are shown for myb3r4-3; similar results were obtained using myb3r4-1. (E) Reduced PM growth and reproduction on whole leaves was exhibited by myb3r4 mutants at 10 dpi. The number of plants with disease ratings from 0 (no visible symptoms of infection) to 4 (100% coverage of infected leaves) is shown. (F) Reduced reproduction (cp/colony) was exhibited by myb3r4 mutants at 5 dpi. c, initial germinated conidia; cp, conidiophore. (Scale bar, 100 μm.) Independent experiments for A–F showed similar results. (G) Putative targets of MYB3R4 at the infection site were identified as mitosis-associated and/or NtMYBA2-activated genes with MSA core element(s) in 500-bp upstream promoter. Red (enhanced) and green (decreased) expression. Values in black fell just below stringent selection criteria. (H) Model for MYB3R4-mediated induced endoreduplication at the site of an adapted PM infection. Black indicates functionality, as does filled lines. P, phosphorylated.

The Arabidopsis TF MYB3R4 Regulates G. orontii–Induced Host Endoreduplication and the Extent of Sustained PM Growth and Reproduction.

The enhanced expression of TFs involved in transcription, translation, and cell elongation at the site of infection is consistent with the observed endoreduplication. However, we did not observe statistically significant differences in expression for known TFs that mediate endoreduplication (28, 29). This could be because most of these TFs were identified in developmental studies of specialized cells with enhanced ploidy (e.g., trichomes) or because their activity is not regulated at the transcript level. The known transcriptional activator of mitosis, MYB3R4, was, however, differentially induced with 2.5-fold higher expression in localized PM-infected cells.

Using a genetic approach, we therefore examined the role of MYB3R4 in the adapted PM–Arabidopsis interaction. We assessed two independent transferred DNA (T-DNA) insertion mutants of MYB3R4 (see Materials and Methods) and WT plants for differences in PM growth and reproduction and in host endoreduplication at the site of infection. The myb3r4 mutants exhibited reduced visible PM symptoms (Fig. 4E and SI Appendix), with fewer conidiophores per germinated conidium compared with WT (Fig. 4F). These differences were not attributed to gross changes in plant or leaf size, morphology, or development (SI Appendix), and, similar to Haga et al. (26), we did not observe significant cytokinesis-associated defects in myb3r4 leaf cells. A number of Arabidopsis mutants conferring enhanced PM resistance exhibit constitutive or PM-induced cell death (3). We, however, did not observe either constitutive or induced cell death in the myb3r4 mutants (SI Appendix). We next determined whether PM-induced endoreduplication was altered in myb3r4 plants. We found host endoreduplication in the mesophyll cells underlying the infected epidermal cells was abrogated in myb3r4 plants with a median DNA content equivalent to 16C for both infected and uninfected samples (Fig. 4 C and D).

Model for MYB3R4 Function in PM-Induced Endoreduplication at the Site of Infection.

MYB3R4 is one of five Arabidopsis TFs containing three tandem imperfect repeat sequences of ∼50 amino acids in the MYB domain similar to mammalian c-MYB proteins. Like mammalian c-MYB proteins, MYB3R4 regulates the cell cycle. It is a known transcriptional activator of cytokinesis that binds the G₂/M-phase-specific MSA element (26). Orthologous MYBs, such as tobacco NtMYBA2 (30), contain a C-terminal domain that allows for posttranslational control of NtMYBA2 activity. For NtMYBA2, full activation requires phosphorylation of the C-terminal domain of the protein by specific cyclin (CYC)/cyclin-dependent kinase (CDK) complexes (30). This also appears to be the case for MYB3R4 because it contains this C-terminal domain with multiple CDK consensus phosphorylation sites, and coexpression of CYCB1 with MYB3R4 enhanced its transactivation of specific target genes (26).

The fact that PM-induced endoreduplication was abrogated in myb3r4 plants indicates that the functional MYB3R4 protein is required for establishment of the endocycle. Furthermore, it suggests that unphosphorylated (or under-phosphorylated) MYB3R4 acts as a repressor of mitosis required for induced endoreduplication (Fig. 4H). To examine this possibility further, we identified putative targets of MYB3R4 with altered expression in our LMD dataset (Fig. 4G and SI Appendix). These genes exclusively exhibited decreased expression at the PM infection site and include the known MYB3R4 target CYCB1:2 (26). MYB3R4 activator vs. repressor function is likely controlled by localized concentrations of specific CYCs (e.g., CYCBs) and CDKs (CDKA or CDKB) required for the C-terminal phosphorylation and mitosis-activating function of MYB3R4 (26, 30). Therefore, localized regulated proteolysis and/or reduced expression of these CDKs and/or CYCs could modulate MYB3R4 function. CDKB1;2, CYCB1;2, and CYCB1;5 expression was reduced at the infection site, suggesting they might play this role (Fig. 4H). An alternative possibility is that a PM-induced inhibitor of the relevant CDK(s) could prevent phosphorylation of MYB3R4, thereby promoting endoreduplication. A number of known and putative CDK inhibitors are induced in response to pathogen, including SIAMESE, a small nuclear protein that interacts with CDKA to suppress mitosis as part of the switch to endoreduplication in trichomes (31).

The myb3r4 mutants exhibited abrogated PM-induced endoreduplication, with reduced growth and reproduction of G. orontii. This suggests that induced endoreduplication at the site of infection facilitates the sustained growth and reproduction of G. orontii on Arabidopsis, which acquires all its nutrients from the plant host (Fig. 4H). Localized endoreduplication increases gene copy number, and thus could enhance the metabolic capacity of host cells at the infection site. Although it is not always the case, endoreduplication has been associated with both elevated metabolic activity and larger cell sizes (28, 32). Consistent with this role at the site of PM infection at 5 dpi, we observed the enhanced expression of genes required to increase metabolic capacity (e.g., involved in transcription, translation, energy generation, and cofactor synthesis; SI Appendix). In addition to a general role in increasing metabolic capacity, endoreduplication at the site of PM infection might allow for the dramatically enhanced expression of genes specifically required to facilitate the growth and reproduction of the fungus such as those involved in localized nutrient (e.g., hexose) accumulation and transport. Indeed, we observed a dramatic increase in expression of these genes (SI Appendix). In the case of nutrient transporters, the combination of enhanced expression of the transporters and the larger surface area of the higher ploidy cells could allow for dramatically enhanced transport rates. Although defense-associated genes (e.g., PR genes) may also be more highly expressed as a consequence of endoreduplication, in our system, it appears this enhanced host defense is outweighed by the enhanced capacity to provide nutrients required by the fungus. This strategy, employing induced localized endoreduplication to meet enhanced metabolic demands, has also been observed for other intimately associated obligate plant biotrophs that establish a defined site of nutrient uptake/exchange. For example, root knot and cyst nematodes establish metabolically active specialized nematode feeding sites of enhanced ploidy (33). In addition, endoreduplication plays a critical role in the Sinorhizobium–Medicago symbiosis, with reduced endoploidy in nodules of Medicago sp. resulting in inefficient nitrogen fixation (34). Therefore, we propose that induced endoreduplication at the interaction site may be an important common strategy used in this type of intimate interaction and that targets identified in our system may have broader applications.

Materials and Methods

A. thaliana Col-0, ics1-2, and mybr3r4 mutants in the Col-0 background were grown in environmental chambers, with a subset infected with G. orontii at 4 weeks and assayed for PM phenotypes as in the article by Chandran et al. (4) and SI Appendix. Homozygous myb3r4-1 (SALK_059819c) and myb3r4-3 (SALK_116974c) T-DNA insertion lines were obtained from the Arabidopsis Biological Resource Center. The myb3r4-1 was previously characterized as a lack-of-function T-DNA mutant with an insertion in exon 2 (26); the myb3r4-3 T-DNA insertion is located in exon 1. For PM expression profiling, fully expanded mature leaves were harvested at 5 dpi. Sample preparation, RNA extraction and amplification, and microarray analysis are described in the text and further detailed in SI Appendix. TAIR9 gene functional descriptions were employed (2), with MAPMAN (35) used to identify statistically enriched processes (p ≤ 0.05) and the ATTED-II Arabidopsis expression database (36) used for coexpression analyses. In brief, for nuclear DNA quantitation, fluorescence values from 30 DAPI-stained mesophyll cell nuclei of WT and mutant leaves underlying PM-infected and uninfected sites were quantified and normalized to those of nuclei of reference diploid guard cells from the same image after subtraction of background from each measurement (SI Appendix). All experiments were repeated at least twice.