Earliest Colonization Events of Rhizophagus ... - Oxford Journals

5 downloads 0 Views 2MB Size Report
impact of the degeneration stage of arbuscules in the colon- ization sequence. Keywords: Arbuscular mycorrhizal symbiosis Arbuscule degeneration Live ...
Earliest Colonization Events of Rhizophagus irregularis in Rice Roots Occur Preferentially in Previously Uncolonized Cells Yoshihiro Kobae* and Toru Fujiwara

Arbuscular mycorrhizal (AM) fungi form a symbiotic association with several plant species. An arbuscule, a finely branched structure of AM fungi, is formed in root cells and plays essential roles in resource exchange. Because arbuscules are ephemeral, host cells containing collapsed arbuscules can be recolonized, and a wide region of roots can be continuously colonized by AM fungi, suggesting that repetitive recolonization in root cells is required for continuous mycorrhization. However, recolonization frequency has not been quantified because of the lack of appropriate markers for visualization of the cellular processes after arbuscule collapse; therefore, the nature of the colonization sequence remains uncertain. Here we observed that a green fluorescent protein (GFP)-tagged secretory carrier membrane protein (SCAMP) of rice was expressed even in cells with collapsed arbuscules, allowing live imaging coupled with GFP–SCAMP to evaluate the colonization and recolonization sequences. The average lifetime of intact arbuscules was 1–2 d. Cells with collapsed arbuscules were rarely recolonized and formed a new arbuscule during the observation period of 5 d, whereas de novo colonization occurred even in close proximity to cells containing collapsed arbuscules and contributed to the expansion of the colonized region. Colonization spread into an uncolonized region of roots but sparsely into a previously colonized region having no metabolically active arbuscule but several intercellular hyphae. Therefore, we propose that a previously colonized region tends to be intolerant to new colonization in rice roots. Our observations highlight the overlooked negative impact of the degeneration stage of arbuscules in the colonization sequence. Keywords: Arbuscular mycorrhizal symbiosis  Arbuscule degeneration  Live imaging  Rice  Secretory carrier membrane protein. Abbreviations: AM, arbuscular mycorrhiza; CSLM, confocal laser scanning microscopy; DAPI, 40 ,6-diamidino-2-phenylindole; dpp, days post-planting; FITC, fluorescein isothiocyanate; GFP, green fluorescent protein; GUS, b-glucuronidase; NBT, nitroblue tetrazolium; PBS, phosphate-buffered saline; PT11, phosphate transporter 11; SCAMP, secretory carrier

membrane protein; SDH, succinate dehydrogenase; TGN, trans-Golgi network; UTR, untranslated region; WGA, wheat germ agglutinin; wpi, weeks post-infection.

Introduction

Regular Paper

Graduate School of Agricultural and Life Sciences, University of Tokyo, Tokyo, 113-8657 Japan *Corresponding author: E-mail, [email protected]; Fax, +81-3-5841-8032. (Received January 26, 2014; Accepted June 2, 2014)

Arbuscular mycorrhizal (AM) symbiosis is among the world’s most widespread mutualism and is observed in approximately 65% of land plant species studied to date (Wang and Qiu 2006). Host plants heavily rely on mycorrhizal fungi to absorb mineral nutrients from the soil, whereas fungal growth is completely dependent on the supply of photoassimilated carbon from the plant (Smith and Smith 2011). However, it is likely that this mutual association has an inherent instability at the microscopic level. Highly branched fungal structures, arbuscules, in which resource exchange occurs between the fungus and the host plant, grow in cortical cells and subsequently collapse to form clumps (Cox and Sanders 1974, Cox and Tinker 1976, Bonfante-Fasolo 1984). Early morphometric analyses have shown that arbuscules have short life spans and that arbuscule turnover is a very rapid process occurring in different plant species (Toth and Toth 1982, Toth and Miller 1984, Alexander et al. 1989). Our previous study has also demonstrated that rice AM symbiotic phosphate transporter PT11– green fluorescent protein (GFP) fusion proteins, which are localized on the periarbuscular membranes surrounding intact arbuscules, disappear after arbuscules begin to degenerate (Kobae and Hata 2010). The calculated life span of an intact arbuscule is 2.5 d in rice, which is in agreement with earlier estimates of a minimum of 2.5 d in several plant species (Alexander et al. 1988, Alexander et al. 1989). Although the morphology of arbuscules has been investigated in detail at the ultrastructural level (Cox and Sanders 1974, Cox and Tinker 1976, Bonfante-Fasolo 1984), the colonization sequences in the developing mycorrhizal roots have not been investigated. The fungus eventually disappears and the host cell regains its previous organization with a large central vacuole. The fate of reorganizing plant cells, however, is undetectable owing to the lack of a specific marker to track the degeneration stage. It is still unclear whether once the fungus disappears from the host cell

Plant Cell Physiol. 55(8): 1497–1510 (2014) doi:10.1093/pcp/pcu081, available online at www.pcp.oxfordjournals.org ! The Author 2014. Published by Oxford University Press on behalf of Japanese Society of Plant Physiologists. All rights reserved. For permissions, please email: [email protected]

Plant Cell Physiol. 55(8): 1497–1510 (2014) doi:10.1093/pcp/pcu081 ! The Author 2014.

1497

Y. Kobae and T. Fujiwara

it can then recolonize the same cells, or whether an arbuscule can be stably formed over a long period. Studies have noted that recolonization is ‘sometimes’ seen (Cox and Sanders 1974, Cox and Tinker 1976), but quantitative information about the frequency of occurrence is unavailable. Considering the continuous accommodation of a fungus in roots and the significant contribution to nutrient transfer through this symbiosis (Gutjahr and Parniske 2013), the mechanism of colonization sequences should be investigated. The purpose of this study is to understand the colonization sequences that underlie the development of mycorrhizal roots. Rice PT11–GFP fusion proteins can track the beginning of arbuscule collapse, coinciding with shrinking of arbuscule branches and aggregation of GFP at the tips of branches (Kobae and Hata 2010). Most of the degeneration phase, however, loses detectable PT11–GFP signals. These AM-specific phosphate transporter proteins are probably degraded promptly during the turnover of periarbuscular membranes (Harrison et al. 2002, Pumplin and Harrison 2009, Kobae and Hata 2010). In order to investigate the colonization sequences, a new molecular marker to track the cells of the degeneration phase should be developed. Intracellular accommodation of a fungus involves the dynamic internal reorganization of plant cells (Pumplin and Harrison 2009). Transmission electron microscopy studies have revealed the accumulation of endoplasmic reticulum, Golgi stacks and trans-Golgi networks (TGNs) in cytoplasmic aggregation of a pre-penetration apparatus that directs hyphae through the epidermal and cortical cells (Genre et al. 2008) and in the area around young arbuscules (Cox and Sanders 1974, Bonfante-Fasolo 1984). Live-cell imaging studies using fluorescent protein-tagged VAPYRIN, EXO-84 subunit, R-SNARE or Qb-SNARE proteins located on endomembrane vesicles also point to substantial secretory activity in the formation of periarbuscular membranes (Pumplin and Harrison 2009, Genre et al. 2012, Ivanov et al. 2012, Lota et al. 2013). Given that arbuscule degeneration probably consists of recycling of periarbuscular membranes, reconstruction of the central vacuole and reorganization of the endomembrane system, it is expected that substantial membrane traffic also occurs during arbuscule degeneration. An AM-specific marker gene of rice, AM42, which encodes a secretory carrier membrane protein (SCAMP), is expressed in mycorrhizal roots but not detected in other tissues or roots infected by the pathogen (Gu¨imil et al. 2005, Gutjahr et al. 2008). The tissue localization of the gene expression or protein function of SCAMP remains unclear. SCAMP is a family of integral membrane proteins, predicted to have four transmembrane domains, which are involved in mediating exocytosis in animal cells (Law et al. 2012). Plant SCAMPs have multiple roles including exocytosis (Lam et al. 2007, Cai et al. 2011), endocytosis (Lam et al. 2009), vacuolar traffic (Law et al. 2012), cell plate biogenesis (Lam et al. 2008, Toyooka et al. 2009) and a possible role in arbusculated cells.

1498

Here, we developed a molecular marker allowing visualization of the life cycle of arbuscules in vivo. This is the first study that quantifies the recolonization to form arbuscules, establishing that the life span of intact arbuscules is only a few days and repetitive ‘de novo’ colonization is required for the consecutive intracellular accommodation of AM fungus in rice roots. This study proposes that a senescent colonized region tends to be temporarily intolerant to new colonization during the colonization sequence.

Results Cell specificity of AM42 promoter activity To develop a molecular marker allowing visualization of the arbuscule life cycle, we investigated the expression of the candidate gene AM42 (Os03g0582300) using transgenic rice expressing an AM42 promoter–GUS (-glucuronidase) fusion (AM42pro-GUS). The 2 kb genomic fragment of AM42 including the promoter and 50 -untranslated region (UTR) was used to express the GUS gene. We generated five independent T0 transgenic lines and examined all lines. Because we observed very similar expression patterns in all five lines, we show here the data from the T1 generation of only one line. The initial day of the growth of germinated seeds in the soil with the fungal inoculum (Rhizophagus irregularis: formerly Glomus intraradices DAOM 197198) is considered as 0 days post-planting (dpp). AM42 was expressed in exodermal cells beneath hyphopodia (Fig. 1A, C), where hyphae pass through intracellularly (Bonfante-Fasolo 1984). Some roots had strong GUS staining sparsely distributed in a few cells (Fig. 1B). Double staining of GUS and the fungal cell wall using wheat germ agglutinin (WGA)–Alexa Fluor594 revealed that the strong signals were from arbusculated cells (Fig. 1D). These results suggest that AM42 is expressed in both the early colonization stage and during later arbuscule formation. Double staining of GUS and fungal succinate dehydrogenase (SDH) activity (vital staining for fungus) suggested that although possible false positives owing to residual GUS protein are expected, AM42 was probably expressed even in cells containing collapsed and SDHnegative arbuscules (Fig. 1E). Taken together, AM42 could be a useful marker for visualizing the broad developmental stages of arbuscules. In our attempt to determine the molecular function of AM42, we employed a knockout mutant of a retrotransposon Tos17-induced insertion line (NE0517), in which Tos17 is inserted in the third exon of the AM42 genomic sequence (Miyao et al. 2003). Three knockout and three wild-type seedlings were isolated using PCR genotyping with combinations of AM42-specific primers and Tos17 border primers. These seedlings were all inoculated with R. irregularis, and fungal cell walls were stained. All seedlings formed virtually normal arbuscules with regard to numbers and morphology (Supplementary Fig. S1); however, we detected no abnormal phenotype in this mutation.

Plant Cell Physiol. 55(8): 1497–1510 (2014) doi:10.1093/pcp/pcu081 ! The Author 2014.

Live imaging of colonization sequences in AM roots

Fig. 1 Expression and localization of AM42 in roots of AM42pro-GUS transgenic rice inoculated with Rhizophagus irregularis. (A) Double detection of AM42 promoter activity and fungal cell wall stained with WGA–AlexaFluor594 (red) at the early colonization stage. The hyphopodium (hp) is detected on the root surface. (B) Condensed GUS signals (arrowheads) are observed in roots. (C) An enlarged image of the root surface of GUS and WGA double staining at the early colonization stage. (D) An enlarged image of a root cell containing a condensed GUS signal. WGA staining indicates the presence of an arbuscule in the cell. (E) Double staining for GUS activity (blue) and fungal SDH activity (purple). Some GUS-positive cells contain SDH-negative and degenerating arbuscules (da). An asterisk indicates an SDH-positive arbuscule. Images are from roots at 10 days post-planting (dpp) (A) or 12 dpp. (B–E). Ih, intercellular hyphae; epi, epidermal cells; exo, exodermal cells; ab, arbuscule. Scale bar = 50 mm (A and B), 20 mm (C–E).

AM42 protein visualizes the developmental stages of arbuscules To determine AM42 protein localization and to verify its usefulness as a protein marker for visualizing the degeneration stage of arbuscule, we generated transgenic rice expressing a GFP–AM42 fusion protein under the control of the AM42 promoter (AM42pro-GFP-AM42). After the generation and examination of five independent T0 transgenic lines, we observed very similar localization patterns in all. We accordingly show here the data of the T1 generation of only one line. Consistent with the results of AM42pro-GUS plants, strong fluorescence of GFP–AM42 proteins was observed in arbusculated cells (Fig. 2A). These arbusculated cells can be identified by a decrease in transparency of transmitted light under a light microscope (Supplementary Fig. S2A, B). In addition, double detection of GFP–AM42 and fungal cell wall using WGA–fluorescein isothiocyanate (FITC) indicated that all GFP signals observed were from arbusculated cells (Supplementary Fig. S3). GFP signals in exodermal cells were below the detection limit. The GFP localization in cells containing developing arbuscules comprising branches with low density was similar to that of PT11–GFP (PT11pro-PT11-GFP), which is specifically localized on periarbuscular membranes (Fig. 2B; Supplementary Fig. S4A, B). However, unlike PT11–GFP,

GFP–AM42 protein localization in cortical cells that contain well-developed arbuscules was observed as a patchy membrane system around mature arbuscule branches (Fig. 2C, D), whereas PT11–GFP protein is evenly localized on periarbuscular membranes which were appressed to one another at this stage (Supplementary Fig. S4C). GFP–AM42 protein was localized even on the membrane surrounding the arbuscule trunk or intracellular thick hyphae (Fig. 2E, F) where PT11–GFP signal was not detected (Supplementary Fig. S4). In addition, aggregated fluorescence was detected around collapsed arbuscules (Fig. 2G, H), but not detected before their collapse. This labelling method does not actually track the arbuscules, but indicates what is indirectly happening to arbuscules due to the plant endomembrane system surrounding arbuscules. Nevertheless, the fluorescent image and the corresponding bright-field image of arbuscule trunk, young arbuscule, mature arbuscule and justcollapsed arbuscule confirmed that GFP–AM42 reveals the outlines of arbuscules (Fig. 2; Supplementary Fig. S2C, D). These various GFP–AM42 localizations associated with different developmental stages of arbuscules make it a molecular marker potentially useful for evaluating the life cycle of arbuscules. To assess the relationship between the GFP–AM42 localization and the arbuscule life cycle in vivo, we examined

Plant Cell Physiol. 55(8): 1497–1510 (2014) doi:10.1093/pcp/pcu081 ! The Author 2014.

1499

Y. Kobae and T. Fujiwara

Fig. 2 Intracellular localization of GFP–AM42 protein in roots of AM42pro-GFP-AM42 transgenic rice inoculated with R. irregularis. (A) A merged image of a bright-field image and a GFP fluorescent image of a colonized root. (B) An enlarged image of a cell containing a young arbuscule. (C) An enlarged image of cells containing densely branched arbuscules. The area outlined is magnified in (D). GFP–AM42 signals are localized in a patchy pattern. (E and F) An enlarged image of a cell containing a young arbuscule which is mostly trunk (t). (G and H) An enlarged image of a cell containing a collapsed arbuscule (ca). Granular signals of GFP–AM42 were detected around collapsed arbuscules. (E) and (G) are the brightfield images of (F) and (H), respectively. Images were collected by epifluorescence microscopy (A, E–H) or CLSM (B–D) at 12 dpp (A–F) or 14 dpp (G and H). Scale bar = 50 mm (A), 10 mm (B, C, E–H), 1 mm (D).

arbusculated cells of AM42pro-GFP-AM42 from the early colonization to the degeneration stages using a live imaging system (Kobae and Hata 2010), modified for long-term observation by increasing the amount of soil and improving the watering method (Supplementary Fig. S5). In brief, a rice seedling was grown in a 90 mm Petri dish with a 50 20 mm rectangular glass window at the bottom. The window was covered with R. irregularis inoculants and soil. Roots were effectively infected with R. irregularis just above the glass window, and illuminated and observed from the underside by inverted confocal laser scanning microscopy (CLSM) or epifluorescence microscopy. Water was supplied from the bottom by capillary action through a 20 mm wide unwoven cloth. Live imaging of an infection front in a small or developing infection unit, each of which comprises an internal mycelium arising from a single hyphal infection (Cox and Sanders 1974, Walker and Smith

1500

1984, Javot et al. 2007), clearly revealed developing young arbuscules (Fig. 3A; Supplementary Fig. S6A–C). This localization pattern was clearly different from that of PT11pro-PT11-GFP plants, in which PT11–GFP was primary localized on periarbuscular membranes (Supplementary Fig. S6D). A time series of observations of arbuscules showed that the states of intact arbuscules are represented by the uniform localization of GFP–AM42 protein in the endomembrane system around the arbuscule at low magnification (10 objective) (Fig. 3B, 0 h). The beginning of arbuscule degeneration was identified by shrinkage of the arbuscule in the longitudinal direction and enhanced fluorescence of GFP–AM42, probably due to the condensation of membranes around the arbuscule (Fig. 3B, 8– 19 h). Live imaging of PT11–GFP showed that arbuscule degeneration was accompanied by similar morphological changes (e.g. the appearance of GFP aggregations and longitudinal

Plant Cell Physiol. 55(8): 1497–1510 (2014) doi:10.1093/pcp/pcu081 ! The Author 2014.

Live imaging of colonization sequences in AM roots

Fig. 3 Live imaging of GFP–AM42 protein; GFP–AM42 indirectly visualizes the sequences of arbuscule development and degeneration. (A) Timelapse images of GFP–AM42 expression at five points (0, 3.7, 7, 23 and 48 h after time-lapse started) in an infection front. AM42pro-GFP-AM42 seedlings were grown in a ø90 mm live imaging system, inoculated with R. irregularis, and observed using an inverted fluorescence microscope. (B) Time-lapse images at six points (0, 0.8, 8, 10, 14 and 19 h after time-lapse started) for a single cell that contains a senescing arbuscule observed by inverted CLSM. The beginning of arbuscule degeneration was coincident with a decrease in the longitudinal length of the arbuscule (double arrow with the length) and enhanced GFP–AM42 fluorescence (arrowheads). The setting of the detector was constant during the observation. Three-dimensional graphs of the intensities of pixels at each time point were created (inset). The collapse of this arbuscule was recognized at 8 h and the GFP–AM42 signal was visible over 10 h from the beginning of the degeneration. The imaging started at 10 dpp (A) or 12 dpp (B). t, trunk; ab, arbuscule; ca, collapsed arbuscule. Scale bar = 20 mm (A).

shrinking) and that periarbuscular membranes are visible for only 2.5–5.5 h during the course of arbuscular collapse (Kobae and Hata 2010). The modified live imaging system used in the present study also verified that PT11–GFP signal disappeared in cells with collapsed arbuscules at any time (Supplementary Fig. S7). However, the arbuscule images represented by GFP– AM42 were visible over 10 h from the beginning of arbuscular

degeneration, indicating that GFP–AM42 is able to reveal the internal structure of host cells where symbiotic interaction is already discontinued. A just-collapsed arbuscule was observed as a clump with aggregation of GFP–AM42 fluorescence (Fig. 3A, 23 h; Fig. 3B, 10 h). Although just-collapsed arbuscules exhibited weak autofluorescence, which is regarded as an indicator of a collapsed arbuscule (Vierheilig et al. 2001), the clump

Plant Cell Physiol. 55(8): 1497–1510 (2014) doi:10.1093/pcp/pcu081 ! The Author 2014.

1501

Y. Kobae and T. Fujiwara

began exhibiting strong autofluorescence accompanied by condensation of GFP–AM42 signals into a single or several mobile granules (Fig. 3A, 48 h; Supplementary Fig. S8A, B; Supplementary Movies S1, S2). Merged images of GFP– AM42 signals and 40 ,6-diamidino-2-phenylindole (DAPI) fluorescence indicated that the mobile granules were not nuclei (Supplementary Fig. S8C). Such GFP fluorescence was absent from the same developmental stage of the infection unit of AM42pro-GUS root (Supplementary Fig. S9), suggesting that the fluorescence in GFP–AM42 plants comes from the tagged AM42 protein. The colonization levels and patterns (e.g. the number of colonized regions, arbuscule density and arbuscule morphology) in roots of AM42pro-GFP-AM42 plants examined using fungal cell wall staining were indistinguishable from those in wild-type plants or AM42pro-GUS plants, thus giving confidence in the investigation of the degeneration stage of arbuscules based on the pattern of GFP–AM42 expression.

Arbuscules are temporary structures Although arbuscules are thought to be ephemeral, the life spans of intact arbuscules remain unclear. To evaluate the life span of arbuscules, we examined the intactness of arbuscules in each colonized region of AM42pro-GFP-AM42 roots for 5 d. Here, we use a term ‘infection segment’, each of which harbors one or more infection unit (Buwalda et al. 1984). A single infection segment is bounded by two infection fronts, both of which can spread into uncolonized roots. Infection segments were randomly selected from roots, and the presence of all intact arbuscules noted at the time of first observation was tracked (Fig. 4A). New arbuscules that appeared during the observations were not included in this analysis. When infection fronts from the two infection segments met to form a single infection segment, the data from these segments were not included. Amijee et al. (1986) reported that when the two fronts coalesced to form a single infection segment, there is no increase in hyphal density, suggesting that separate infection segments may compete for root occupancy and thus the life span of arbuscule may be underestimated in these regions. All infection segments were detected in large lateral roots, consistent with the characteristics of rice mycorrhizas (Gutjahr et al. 2009, Vallino et al. 2013). Plants in live-imaging 90 mm dishes were grown under a 16 h light/8 h dark regime and immediately returned to the same growth condition after microscopic observation. Some colonized regions could not be observed up to the end because other roots crawled under the target roots or the target roots were out of the field of view (these data were omitted). The intactness of 466 arbuscules in total from 12 infection segments was tracked. Although the age (days after infection) of each infection segment and the developmental stage of each arbuscule formed in colonized regions varied, 98% of intact arbuscules collapsed within 48 h of the beginning of the observation period (Fig. 4B), indicating that arbuscules are essentially temporary structures.

1502

Fig. 4 All observed arbuscules are temporary structures. (A) Timelapse images of GFP–AM42 expression at six points (0, 17, 26, 41, 49 and 66 h after time-lapse started) in a root of AM42pro-GFP-AM42 seedlings inoculated with R. irregularis in a ø90 mm live imaging system. Images were obtained by inverted fluorescence microscopy. Intact arbuscules are indicated by white-colored numbering and collapsed arbuscules are indicated by purple-colored numbering. A white arrow indicates a newly formed arbuscule; and a purple arrow indicates the collapsed arbuscule. (B) The intactness of arbuscules in each infection segment of AM42pro-GFP-AM42 roots examined by live imaging. All intact arbuscules in randomly selected infection segments were tracked. The numbers of intact arbuscules at the beginning of the observation was defined as 100%. A total of 466 arbuscules from 12 infection segments (lines with various colors) were tracked. Newly formed arbuscules which appeared during the observations in each infection segment were not included in this analysis. Scale bar = 50 mm (A).

A small fraction of cells that contain degenerating arbuscules are recolonized to form arbuscules We sought to determine whether plant cells in which an arbuscule had once degenerated could be recolonized by the AM fungus. Recolonization with new arbuscule formation occurred in only three of 466 arbusculated cells (0.6%) that were analyzed in Fig. 4B. Time-lapse imaging indicated that recolonization could occur within 24 h after the collapse of most arbuscule branches (Fig. 5). However, we could not identify a predictor of recolonization. Two recolonizations (Fig. 5A) showed longitudinal shrinking and reduction of GFP–AM42 signal during the collapse, whereas aggregations of GFP–AM42 were not observed. One recolonization (Fig. 5B) showed aggregation of GFP–AM42, followed by the development and collapse of the arbuscule. We found no specific pattern of GFP–AM42 localization in the recolonization. Most arbuscules observed (Fig. 5A) collapsed during imaging over 143 h, but one arbuscule was

Plant Cell Physiol. 55(8): 1497–1510 (2014) doi:10.1093/pcp/pcu081 ! The Author 2014.

Live imaging of colonization sequences in AM roots

Fig. 5 A small fraction of colonized cells was recolonized and formed a new arbuscule. (A) Time-lapse images of GFP–AM42 at six points (0, 48, 72, 96 128, and 143 h after time-lapse started) in a root of AM42pro-GFP-AM42 seedlings inoculated with R. irregularis in a live imaging system. Images were obtained by inverted epifluorescence microscopy. Two arbuscules indicated by asterisks at 0 h were intact or beginning to collapse. These two cells are recolonized later. They were collapsed at 48 and 72 h, but new arbuscules were formed at 96 h. Imaging started at 12 dpp. (B) Time-lapse images at 10 points (0, 16, 23, 40, 64, 97, 112, 119, 136 and 145 h after time-lapse started) for a single arbuscule are shown. Arrowheads indicate an aggregate of GFP–AM42. Double arrows with numbers are the longitudinal lengths of arbuscules. ab, arbuscule. Imaging started at 12 dpp. Scale bar = 20 mm (A).

long-lived for at least 4 d, indicating that one infection segment harbored arbuscules with various life spans.

Arbuscule formation occurs preferentially in previously uncolonized cells We next examined the spatiotemporal colonization sequence in roots. Live imaging of AM42pro-GFP-AM42 roots suggested that a single infection unit developed longitudinally rather than radially from the initially infected area and eventually occupied only up to three longitudinal columns of cortical cells (Fig. 6A). Throughout this experiment, >90% of arbuscules (n > 50) were

formed in a cortical monolayer of cells irrespective of the positions within the roots (Fig. 6B). Time-lapse imaging of a colonized area revealed that a new arbuscule cluster developed into an uncolonized region adjacent to a previously formed arbuscule cluster (Fig. 6C). Given that recolonization rarely occurred (Figs. 4, 5), the arbuscule clusters rarely coalesced in a solid three-dimensional cortical cell layer during the development of an infection segment (Fig. 6D). An infection segment extended in both the upper and lower direction of the roots (Fig. 7A). However, it was often observed that senescent regions in the infection segment were hardly colonized, at least during the

Plant Cell Physiol. 55(8): 1497–1510 (2014) doi:10.1093/pcp/pcu081 ! The Author 2014.

1503

Y. Kobae and T. Fujiwara

Fig. 6 Repetitive de novo colonization contributes to the development of an infection segment. (A) Time-lapse images of GFP–AM42 at four points (0, 12, 24 and 36 h after time-lapse started) in a root of AM42pro-GFP-AM42 seedlings inoculated with R. irregularis in a live imaging system. Images were obtained by inverted epifluorescence microscopy. Imaging started at 12 dpp. A single infection unit develops longitudinally and occupies only 1–3 longitudinal columns of cortical cells. Purple arrowheads indicate the positions of collapsed arbuscules. (B) A horizontal section of a colonized root. A GFP–AM42 image (green) and an autofluorescence image (blue to purple) are merged. Arrowheads indicate the position of arbuscules formed in a monolayer of cortical cells. (C) Time-lapse images of GFP–AM42 at five points (0, 11, 24, 35 and 47 h after time-lapse started) in a live imaging system. Images were obtained by inverted epifluorescence microscopy. Imaging started at 12 dpp. White arrowheads indicate the position of intact arbuscules, and purple arrowheads indicate the position of collapsed arbuscules. Numbers indicate the order of appearance of arbuscule clusters. Red fluorescence is from soil microbes. (D) The schematic representation of colonization sequences observed in (C). epi, epidermis; exo, exodermis; cor, cortex. Scale bar = 20 mm (A, B), 50 mm (C).

time when aggregations of GFP–AM42 were observed (Fig. 7A). To examine the colonization status of the fungus in the senescent region, the roots marked by only aggregated GFP–AM42 signals (Supplementary Fig. S10A) were dissected and stained with WGA–FITC (Supplementary Fig. 10B). We examined 12 senescent colonized regions and found that all of them

1504

harbored many intercellular hyphae. Double detection of GFP–AM42 and fungal SDH activity indicated that collapsed arbuscules, if any, in the senescent colonized region had little metabolic activity (Fig. 7C, E), whereas intact arbuscules in the young infection segment had strong SDH activity (Fig. 7B, D). These observations suggested that colonized regions with only

Plant Cell Physiol. 55(8): 1497–1510 (2014) doi:10.1093/pcp/pcu081 ! The Author 2014.

Live imaging of colonization sequences in AM roots

Fig. 7 A senescent colonized region tends to be intolerant to new colonization. (A) Time-lapse images at five points (0, 11, 24, 35 and 47 h after time-lapse started) for an infection segment. AM42pro-GFP-AM42 seedlings were grown in a live imaging system and observed by inverted epifluorescence microscopy. Imaging started at 12 dpp. White arrowheads indicate the position of intact arbuscules, and purple arrowheads indicate the position of collapsed arbuscules. The senescent region is intolerant to new colonization. (B–E) Double detection of GFP–AM42 and fungal SDH activities (vital staining). Root segments with GFP–AM42 fluorescence were dissected and were immediately immersed in NBT solution. (B and D) A colonized region that contains only active arbuscules (white arrowheads). (C and E) A colonized region in which most arbuscules are collapsed (purple arrowheads). (D) and (E) are the bright-field images of (B) and (C), respectively. Images were obtained at 14 dpp (B–E). Scale bar = 200 mm (A); 50 mm (B–E).

aggregated GFP–AM42 lacked active intracellular colonization even though many hyphal structures were maintained. To evaluate the preference of arbuscule formation for previously uncolonized cells, we observed the development of randomly selected infection segments using live imaging over 97 h, and found that in 37/48 (77%) of them, secondary (or more)

colonization preferentially occurred in uncolonized regions outside of them, but 11/48 (23%) of them showed interrupted growth at least during the observation period. Therefore, R. irregularis might prefer arbuscule formation in previously uncolonized cells rather than colonized cells. We occasionally observed a new colonization within senescent colonized

Plant Cell Physiol. 55(8): 1497–1510 (2014) doi:10.1093/pcp/pcu081 ! The Author 2014.

1505

Y. Kobae and T. Fujiwara

regions; however, such repeated colonization provided only a small number of arbuscules, which were not enough to assess the preference for colonization of previously uncolonized/colonized cells within the senescent regions. In rare cases, we observed a normal development of infection units within a senescent region (Supplementary Fig. S11), but it was a very exceptional case in the current live imaging irrespective of the molecular marker. Taking all these findings together, senescent colonized regions tended to be intolerant to repetitive colonization and the intracellular accommodation of arbuscules occurred preferentially in previously uncolonized cells in rice roots.

Discussion We developed a molecular marker for visualizing the life cycle of arbuscules. We found that GFP–AM42 proteins localized on the endomembrane surrounding arbuscule branches when arbuscules were developing, but became uneven with patchy localization in densely branched stages. GFP–AM42 also localized to the membrane surrounding the arbuscule trunk, which is functionally distinct from periarbuscular membranes that contain a phosphate transporter (Pumplin and Harrison 2009, Hata et al. 2010, Harrison 2012). These localizations suggest that the AM42 protein is a general element for organizing the accommodation of the fungus rather than specifically functioning in the development of periarbuscular membranes. In transgenic tobacco BY-2 cells expressing the rice SCAMP1–yellow fluorescent protein (YFP) fusion construct, fusion proteins were found to localize to the cell plate of dividing cells (Lam et al. 2008). Immunocytochemical and ultrastructural analysis in BY-2 cells expressing the tobacco SCAMP2–YFP fusion construct identified a secretory vesicle cluster that was generated from the TGN, and it moved to the cell plate in dividing BY-2 cells (Toyooka et al. 2009). Development of perifungal membranes has several parallels with cell plate formation during cytokinesis (Genre et al. 2012, Harrison 2012). VAMP721d/e are expressed in the inner cortical cells, and their knockdown via RNA interference (RNAi) resulted in aberrant arbuscules (Ivanov et al. 2012). GFP–VAMP721 fusion proteins localized to both the vicinity of the arbuscule and the cell plate of dividing meristematic root cells. Although the AM42 protein possibly functions in the secretion of structural materials to the periarbuscular space during arbuscule development, such delivery may be achieved by several SCAMP homologs or other protein families in rice, given that the T-DNA knockout line of AM42 formed virtually normal arbuscules. GFP–AM42 proteins were also expressed in cells that contained degenerating arbuscules. The long-lasting expression of GFP–AM42 enabled us to visualize the intracellular structure along with the arbuscule life cycle. Tracking degeneration of an arbuscule identified by the aggregation of GFP–AM42 fluorescence and the shrinking of arbuscules in vivo indicated that most arbuscules are destined to collapse within 2 d. Live

1506

imaging of PT11–GFP showed that the life span of an intact arbuscule is 2.5 d in rice (Kobae and Hata 2010). The life span of intact arbuscules indicated by GFP–AM42 is expected to be longer than that of PT11 because GFP–AM42 localizes even in the perifungal membrane surrounding the arbuscule trunk which could not be visualized with PT11–GFP. However, the life span of intact arbuscules examined in this study was