Rice cultivars and endophytic bacteria towards the development of more effective nitrogen-fixing associations

Transcription

1 Rice cultivars and endophytic bacteria towards the development of more effective nitrogen-fixing associations A report for the Rural Industries Research and Development Corporation by Professor Barry G. Rolfe and Dr Jeremy J. Weinman December 2001 RIRDC Publication No 01/175 RIRDC Project No ANU-34A

2 2001 Rural Industries Research and Development Corporation. All rights reserved. ISBN ISSN Rice cultivars and endophytic bacteria towards the development of more effective nitrogen-fixing associations Publication No. 01/175 Project No. ANU-34A The views expressed and the conclusions reached in this publication are those of the authors and not necessarily those of persons consulted. RIRDC shall not be responsible in any way whatsoever to any person who relies in whole or in part on the contents of this report. This publication is copyright. However, RIRDC encourages wide dissemination of its research, providing the Corporation is clearly acknowledged. For any other enquires concerning reproduction, contact the Publications Manager on phone Researcher Contact Details Professor Barry G. Rolfe Genomic Interactions Group, Research School of Biological Sciences, ANU. PO Box 475, Canberra, ACT 2601 Phone: Fax: rolfe@rsbs.anu.edu.au Dr Jeremy Weinman Genomic Interactions Group, Research School of Biological Sciences, ANU. PO Box 475, Canberra, ACT 2601 Phone: Fax: weinman@rsbs.anu.edu.au RIRDC Contact Details Rural Industries Research and Development Corporation Level 1, AMA House 42 Macquarie Street BARTON ACT 2600 PO Box 4776 KINGSTON ACT 2604 Phone: Fax: rirdc@rirdc.gov.au. Website: Published in December 2001 Printed on environmentally friendly paper by Canprint ii

3 Foreword A marked increase in environmental awareness and a concern for sustainable agriculture has occurred over the last decade. This awareness has highlighted the need both to further promote plant growth and yields and also to find environmentally benign replacements for industrially produced nitrogen fertilisers. Agriculturalists have searched for various soil bacteria that might make reliable contributions to the growth of non-legume cereals and have found a group of bacteria which intimately associate with rice. Many of these rice-associating strains are also nitrogen-fixing. Recently Rhizobium bacteria, which normally nodulate legumes, have been shown to associate intimately with the roots of rice plants. The possibility of establishing a more effective type of Rhizobium-non-legume interaction is potentially available in rice because many of the plant compounds that could interact and stimulate rhizobia are also present in rice roots. However, the interaction of introduced bacterial strains with rice cultivars used in Australia had not been examined. This project set out to describe the types of interactions which occur between seedlings of Australian rice cultivars and a range of bacterial and Rhizobium strains. It details the extent and pattern of rice plant colonisation, as well as degree to which the growth of rice seedlings is influenced, and how environmental factors can influence the outcome. Finally, it identifies genetic regions within a model bacterial strain that are essential to the interaction of the bacteria with rice. This project was funded from industry revenue which is matched by funds provided by the Federal Government. This report, a new addition to RIRDC s diverse range of over 700 research publications, forms part of our Rice R&D program, which aims to improve the profitability and sustainability of the Australian rice industry. Most of our publications are available for viewing, downloading or purchasing online through our website: downloads at purchases at Peter Core Managing Director Rural Industries Research and Development Corporation iii

4 Contents Foreword...iii Executive Summary...v 1. Introduction Effect of the bacterial strain inoculation on rice growth Bacterial colonisation in root and lateral root tissues Multiplication of rice-associating strains within rice leaves Objectives Methodology Bacterial strains Bacterial growth media Plant growth media Plant growth studies Fluorescence microscopy Analysis of global changes in protein expression by proteome analysis Results Effect of the bacterial strains on rice growth Examination of inoculated rice roots Multiplication of rice-associating strains within rice leaves Seedlings grown in liquid media in Magenta jars Interaction of R. leguminosarum bv. trifolii strain ANU843 with rice seedlings Entry of the bacteria into rice root tissues Location and spread of the bacteria in the whole root system Better root growth provides more possible sites for colonisation, and increases the proportion of roots colonised by the bacteria Colonisation of rice roots and the bacterial effect on plant growth are not determined by the Sym-plasmid of Rhizobium or the Ti-plasmid of Agrobacterium Certain plasmids of Rhizobium leguminosarum bv. trifolii affect plant growth, but not the colonisation ability Colonisation of internal root tissues is common among bacteria Proteome analysis of strain ANU843 and its plasmid-cured derivatives Results Evaluation of rice plant growth 5.2 Interaction between bacterial strains and rice cultivars Two approaches to the analysis of this Plant Microbe Interaction Implications Recommendations Future Experiments References...17 iv

5 Executive Summary The general aims of these studies were to (i) investigate whether the association between the roots of rice seedlings and bacteria, such as rhizobia, can enhance rice growth; (ii) establish that these riceassociating bacteria can intimately associate with rice roots and tissues; and (iii) determine whether a series of rice-rhizobium bacteria could be developed as an inoculum which might promote rice plant growth. In summary, the results obtained demonstrate that a number of the bacterial strains tested can significantly alter the growth of a rice seedling. In addition, many of these strains also have the ability to colonise onto and survive inside the rice seedling tissues without producing any gross disease effects of plant growth. As some strains possess the ability to colonise without altering seedling growth, it is possible that the genes responsible for the colonisation ability are separate from those genes responsible for the effects on seedling growth. By adapting and developing techniques to reliably assess the performance of rice-associating strains in their promotion of rice seedling growth, we can now begin to identify the genes in these bacteria which are responsible for rice colonisation and growth effects. Effect of the bacterial strain inoculation on rice growth The soil bacterium, Rhizobium, normally nodulates legumes and benefits the plant mostly through fixing atmospheric nitrogen but may also provide growth promotion through phytohormones and the provision of soil phosphates and other nutrients. Recently, Rhizobium strains have been shown to associate with the roots of rice plants and form potential beneficial associations. This project has developed simple Laboratory assays to study the basis of these microbe-rice interactions. The experiments with the inoculated rice seedlings revealed three sources of variation in the rates of growth for the inoculated plants. There were differences between rice cultivars in their ability to respond to bacterial inoculation, variation in the performance of the bacterial strains, and variations due to the growth medium used. Some strains stimulate the growth of rice, others have little effect, while others inhibit rice plant growth and development. The negative plant-microbe interactions could be altered by changing the growth medium conditions used. This was found to be a complex interaction between various calcium, potassium, phosphate ions and nitrogen addition. Moreover, it is thought that various growth media could exert their effect by altering bacterial production of phytohormones and hence affect plant growth. For example, slight additions of the phytohormone auxin induced increased lateral root formation, while excess levels caused stunting of root growth and development and hence overall reduction of rice plant growth. Bacterial colonisation in root and lateral root tissues Several new technologies were used for the visualisation and the evaluation of intercellular bacterial colonisation, entry, spread and the establishment of the internal colonisation. To study the timing and route of entry into roots of rice seedlings, the rhizobia were tagged with DNA sequences expressing the Green Fluorescent Protein (GFP) which causes the bacteria to emit a green glow. This method enables the use of a non-destructive assay to follow the bacterial association with the roots of plants grown under different growth conditions. Two clover nodulating bacteria, Rhizobium trifolii strain R4, which stimulates growth on a particular subset of rice cultivars, and strain E4, which inhibits growth, were used to inoculate cultivar Pelde seedlings to study the growth effects due to these different bacterial strains. If the seedlings were grown on the type of agar plates often used in studying plant-microbe interactions, the roots of the rice plants were colonised but this was not extensive colonisation. The bacteria were observed along the grooves of the main roots and at the junctions of a lateral roots. The bacteria attach to the root hairs and root surface often in a polar orientation which is similar to the attachment to legume roots. However, no distortion or curling of the root hairs, as observed in the legume root infections, were found. If the same experiments are done by growing rice plants in liquid medium, but in small containers which restricted extensive root growth, then strain R4 was unexpectedly found to enter into the lateral roots of the seedlings and form long lines of bacterial cells inside the root, between the plant cells v

6 (intercellularly), as the lateral root grew. Strikingly, strain E4 was not observed under these conditions to enter into rice roots but only to colonise the root surfaces. To mimic the responses of rice grown under flood, a procedure was developed for growing the seedlings in bigger containers with larger volumes of liquid medium. Cultivar Pelde seedlings were inoculated with strains R4 and E4 and then grown for three weeks and examined under a microscope. Individual plants were examined after 14 days and 21 days. The differences between strains R4 and E4 were now greatly reduced as they both associated with the roots of over 50% of the inoculated seedlings. Externally, the bacteria colonised the root tips and grew along the root. The rhizobia also entered the lateral roots and formed long lines of bacteria between the cells in these roots. Interestingly, strain E4 could also form infections at many of the junctions between lateral and main roots for most seedlings. In marked contrast, strain R4 was not observed to form these lateral root junction infections. Multiplication of rice-associating strains within rice leaves An additional plant assay was developed to analyse if there was any possible relationship between the ability of a Rhizobium strain to affect seedling growth and its ability to survive and multiply within rice tissues. As the environment provided by the rice leaf can be easily used to study internal colonisation by bacteria, this assay measured the multiplication, movement and compatibility of the Rhizobium strains within rice tissues. In addition, it enables the use of various bacterial strains to be used as biological probes of any induced responses or preformed systems of plant responses in the rice plants. In this bioassay, bacterial cells were pressure-infiltrated into sections of the rice seedling leaf and viable bacterial counts were recorded every two days for up to 15 days. These assays were based on the current bacterial plant pathology leaf assays which demonstrate that different bacterial cells can only grow and multiply within leaf tissues if they contain particular bacterial genes which are associated with specific nutrient uptake systems. Bacteria with mutations in these genes, or which do not contain these genes, do not grow in plant tissues. We have observed that the Rhizobium strains can survive and multiply within rice leaves. The results for cultivar Pelde show that after infiltration into the leaf the different Rhizobium strains could multiply for 8 to 10 generations over a 12 to 15 day period. Even strain E4, which normally inhibits rice seedling growth, multiplies very well within rice leaves. Interestingly, additional assays to examine the spread of these bacteria within the leaf tissues away from the site of application found that even Rhizobium strains which can form long lines of internal root colonisation do not move out of the 2 cm strip containing the sites of the original leaf inoculation. Therefore the bacterial growth within rice leaves is not equivalent to that observed in the lateral roots where the bacteria can also multiply. The possibility of establishing a more effective type of Rhizobium-non-legume interaction is potentially available in rice because rice roots contain many of the plant compounds that can stimulate rhizobia. This finding could be used to help develop a series of rice inoculum strains for the industry. We believe that by understanding what causes these different types of rice responses, stimulation or inhibition, we will be able to begin the development of a bacterial rice inoculum strain. Moreover, our results also suggest that because of their intimate association, rice-associating bacteria could be used to provide other valuable substances to the growing rice plant. vi

7 1. Introduction The general aims of these studies were to investigate (i) whether the association between the roots of rice seedlings and bacteria, such as rhizobia, can enhance rice growth; (ii) establish that these riceassociating bacteria can intimately associate with rice roots and tissues; (iii) whether a series of rice- Rhizobium bacteria could be developed as an inoculum which might promote rice plant growth, In summary, the results obtained demonstrate that a number of the bacterial strains tested can significantly alter the growth of a rice seedling. In addition, many of these strains also have the ability to colonise onto and survive inside the rice seedling tissues without producing any gross disease effects of plant growth. As some strains possess the ability to colonise without altering seedling growth, it is possible that the genes responsible for the colonisation ability are separate from those genes responsible for the effects on seedling growth. By adapting and developing techniques to reliably assess the performance of rice-associating strains in their promotion of rice seedling growth, we can now begin to identify the genes in these bacteria which are responsible for rice colonisation and growth effects. 1.1 Effect of the bacterial strain inoculation on rice growth The soil bacterium, Rhizobium, normally nodulates legumes and benefits the plant mostly through fixing atmospheric nitrogen but may also provide growth promotion through phytohormones and the provision of soil phosphates and other nutrients. Recently, Rhizobium strains have been shown to associate with the roots of rice plants and form potential beneficial associations. This project has developed simple laboratory assays to study the basis of these microbe-rice interactions. The experiments with the inoculated rice seedlings revealed three sources of variation in the rates of growth for the inoculated plants. There were differences between rice cultivars in their ability to respond to bacterial inoculation, variation in the performance of the bacterial strains, and variations due to the growth medium used. Some strains stimulate the growth of rice, others have little effect, while others inhibit rice plant growth and development. The negative plant-microbe interactions could be altered by changing the growth medium conditions used. This was found to be a complex interaction between various calcium, potassium, phosphate ions and nitrogen addition. Moreover, it is thought that various growth media could exert their effect by altering bacterial production of phytohormones and hence affect plant growth. For example, slight additions of the phytohormone auxin induced increased lateral root formation, while excess levels caused stunting of root growth and development and hence overall reduction of rice plant growth. 1.2 Bacterial colonisation in root and lateral root tissues Several new technologies were used for the visualisation and the evaluation of intercellular bacterial colonisation, entry, spread and the establishment of the internal colonisation. To study the timing and route of entry into roots of rice seedlings, the rhizobia were tagged with DNA sequences expressing the Green Fluorescent Protein (GFP) which causes the bacteria to emit a green glow. This method enables the use of a non-destructive assay to follow the bacterial association with the roots of plants grown under different growth conditions. Two clover nodulating bacteria, Rhizobium trifolii strain R4, which stimulates growth on a particular subset of rice cultivars, and strain E4, which inhibits growth, were used to inoculate cultivar Pelde seedlings to study the growth effects due to these different bacterial strains. If the seedlings were grown on the type of agar plates often used in studying plant-microbe interactions, the roots of the rice plants were colonised but this was not extensive colonisation. The bacteria were observed along the grooves of the main roots and at the junctions of a lateral roots. The bacteria attach to the root hairs and root surface often in a polar orientation which is similar to the attachment to legume roots. However, no distortion or curling of the root hairs as observed in the legume root infections were found. If the same experiments are done by growing rice plants in liquid medium, but in small containers which restricted extensive root growth, then strain R4 was unexpectedly found to enter into the lateral 1

8 roots of the seedlings and form long lines of bacterial cells inside the root, between the plant cells (intercellularly), as the lateral root grew. Strikingly, strain E4 was not observed under these conditions to enter into rice roots but only to colonise the root surfaces. To mimic the responses of rice grown under flood, a procedure was developed for growing the seedlings in bigger containers with larger volumes of liquid medium. Cultivar Pelde seedlings were inoculated with strains R4 and E4 and then grown for three weeks and examined under a microscope. Individual plants were examined after 14 days and 21 days. The differences between strains R4 and E4 were now greatly reduced as they both associated with the roots of over 50% of the inoculated seedlings. Externally, the bacteria colonised the root tips and grew along the root. The rhizobia also entered the lateral roots and formed long lines of bacteria between the cells in these roots. Interestingly, strain E4 could also form infections at many of the junctions between lateral and main roots for most seedlings. In marked contrast, strain R4 was not observed to form these lateral root junction infections. 1.3 Multiplication of rice-associating strains within rice leaves An additional plant assay was developed to analyse if there was any possible relationship between the ability of a Rhizobium strain to affect seedling growth and its ability to survive and multiply within rice tissues. As the environment provided by the rice leaf can be easily used to study internal colonisation by bacteria, this assay measured the multiplication, movement and compatibility of the Rhizobium strains within rice tissues. In addition, it enables the use of various bacterial strains to be used as biological probes of any induced responses or preformed systems of plant responses in the rice plants. In this bioassay, bacterial cells were pressure-infiltrated into sections of the rice seedling leaf and viable bacterial counts were recorded every two days for up to 15 days. These assays were based on the current bacterial plant pathology leaf assays which demonstrate that different bacterial cells can only grow and multiply within leaf tissues if they contain particular bacterial genes which are associated with specific nutrient uptake systems. Bacteria with mutations in these genes, or which do not contain these genes, do not grow in plant tissues. We have observed that the Rhizobium strains can survive and multiply within rice leaves. The results for cultivar Pelde show that after infiltration into the leaf the different Rhizobium strains could multiply for 8 to 10 generations over a 12 to 15 day period. Even strain E4, which normally inhibits rice seedling growth, multiplies very well within rice leaves. Interestingly, additional assays to examine the spread of these bacteria within the leaf tissues away from the site of application found that even Rhizobium strains which can form long lines of internal root colonisation do not move out of the 2 cm strip containing the sites of the original leaf inoculation. Therefore the bacterial growth within rice leaves is not equivalent to that observed in the lateral roots where the bacteria can also multiply. The possibility of establishing a more effective type of Rhizobium-non-legume interaction is potentially available in rice because rice roots contain many of the plant compounds that can stimulate rhizobia. This finding could be used to help develop a series of rice inoculum strains for the industry. We believe that by understanding what causes these different types of rice responses, stimulation or inhibition, we will be able to begin the development of a bacterial rice inoculum strain. Moreover, our results also suggest that because of their intimate association, rice-associating bacteria could be used to provide other valuable substances to the growing rice plant. 2

9 2. Objectives The stated objectives at the outset of this project were to: (A) assess a range of endophytic nitrogen-fixing bacteria for their ability to colonise the roots, stems and leaves of rice plants and their ability to stimulate plant growth; (B) develop biological and molecular probes to prove that a new endophyte strain unequivocally establishes an intimate association with the tissues of rice seedlings; (C) examine how the rice endophytic strains establish an intimate association with the rice tissues of seedlings (D) assess in the laboratory the potential for Rhizobium endophytes to beneficially associate with Australian rice cultivars. (E) initiate the physiological and molecular genetic characterisation of Rhizobium trifolii endophytic strains of rice. 3

10 3. Methodology 3.1 Bacterial strains Rice-associating Rhizobium leguminosarum bv. trifolii strains were isolated from either the surfacesterilised root tissues (E strains, E4, E11) or the root rhizosphere (R strains, R4) of rice plants grown at Sakha Kafr El Sheikh in the middle Nile delta region of Egypt (Yanni et al 1997). The archetypal R. leguminosarum bv. trifolii strain ANU843 (Rolfe et al 1980) is a non-rice-associated bacterium was used as a control. Strain ANU843 has five plasmids, ranging from 180 kbp to ~700 kbp in size, and these, together with the chromosome, constitute a fully integrated genomic complex (Guerreiro et al 1998). Three derivative strains, ANU845 (pa the Sym plasmid cured), CFNS152 (pc cured) and CFNS309 (pe cured), were used. Broad host-range strain NGR234 can nodulate a large group of legumes and the non-legume Parasponia (Bender et al 1988). Agrobacterium tumefaciens strain C58 and its plasmid pti-negative derivative strain A136 were used to represent A. tumefaciens. The unidentified endophytic strains of rice were isolated by IRRI scientists in the Philippines from surfacesterilised rice root tissues (Barraquio et al. 1997) 3.2 Bacterial growth media Bacterial strains were grown on Bergersen's modified medium (BMM) at 29 o C for 3 d before inoculation of plants (Rolfe et al 1980). Bacteria for inoculation were suspended in sterile water. 3.3 Plant growth media Nitrogen-free modified Fahraeus medium (NFM) has been described previously (Rolfe et al 1980, Rolfe and McIver 1996). Where specified, 10 mm KNO 3 was added to this medium. Hoagland #2 (Sigma Chemical Co.) medium contains 15 mm nitrogen as 6 mm KNO 3, 4 mm Ca(NO3) 2, and 1 mm NH 4 H 2 PO 4. Colonisation studies used liquid media of NFM plus 10 mm KNO 3 in Magenta jars (containing 250 ml liquid medium). 3.4 Plant growth studies Three rice cultivars, Calrose, Pelde and IR-28 were used to examine the growth effects of several Rhizobium leguminosarum bv. trifolii isolates and a number of unidentified endophytic strains from IRRI. Rice cultivars Calrose and Pelde were used (seeds were obtained from Dr. L. Lewin, NSW Department of Agriculture) and the procedures were as described elsewhere (Prayitno et al 1999). Experimental design and help in the analysis of data was obtained from Dr. Ross B. Cunningham of the Statistical Support Unit of the ANU Graduate School. Data were analysed and the least significant difference (LSD) at p = 0.05 calculated. Fluorescent labelling of endophytic strains In order to study the timing and route of entry on and into the rice tissues we incorporated DNA sequences encoding the green fluorescent protein (gfp) into the bacterial strains. To do this, the plasmid ptb93fa (Gage et al. 1996) which expresses the sequences encoding gfp behind a constitutive promoter was transferred into each of the Rhizobium leguminosarum bv. trifolii strains isolated from rice tissues and also into the archetypal R. l. bv. trifolii strain ANU843. This plasmid enables non-destructive assays to be used to locate the bacteria and follow their association with the roots of young rice seedlings by using fluorescence microscopy. The presence of the plasmid ptb93fa did not alter the biological properties such as growth in different media, colony formation and morphology, or interactions with rice and clover plants of the original strains (data not shown). One isolate was then selected and used as the gfp-labelled representative of the particular strain. 4

11 Colonisation of rice by rice-associating strains Different methods of incubating the inoculated seedlings with tagged bacterial strains were investigated and the procedures described elsewhere (Prayitno et al., 1999). The Rhizobium strains containing the gfp marker were screened for their colonisation ability by inoculating them onto axenically grown rice seedlings and following their fate with fluorescence microscopy. The initial 3 day old seedlings of cv. Pelde that were placed into the bacterial suspensions had one developing shoot and one developing root and were suspended in approximately 4 x106 Rhizobium (gfp) cells for 90 min. The inoculated seedlings were transferred to growth medium plates and were grown in a growth cabinet which had similar conditions to those used in the plant growth studies. The procedures for the gfp assays were those described by Chalfie et al. (1994) and Gage et al. (1996). The intact root(s) of rice seedlings were cut from the seedling and placed on a microscope slide and examined for bacterial colonisation by observing the green fluorescence of the gfp-tagged rhizobia under fluorescence microscopy. To obtain cross sections of the root samples, fresh root segments were fixed in 3.6% formaldehyde solution in phosphate buffer (0.5M) (ph 7.0) for 2 days, then rinsed three times with phosphate buffer. Fixed root segments were embedded in 5% agarose and then sectioned using Lancer a Vibratome Series Attachment of Rhizobium strains to rice root hairs Newly inoculated seedlings with one root and measurements of bacterial attachment were made of root hairs from and the same 3 cm region of the root in all experiments were examined microscopically. Initial recordings were made of 30 randomly chosen root hairs at 2 h after inoculation. 3.5 Fluorescence microscopy Fluorescence microscopy was done on a Nikon Optiphot inverted microscope stand, and images were taken on Fujichrome Sensia 400 film. For fluorescence microscopy, a 495 nm excitation filter, a dichroic DM505 filter and eyepiece-side absorption 515W filter slides were used. 3.6 Analysis of global changes in protein expression by proteome analysis Proteome analysis was used to resolve the complex mixtures of Rhizobium strain ANU843 proteins on a 2-dimensional polyacrylamide electrophoresis (2-DE) gel. The growth of strains and the procedures used were as previously published (Guerreiro et al 1997, 1998, 1999). 5

12 4. Results Many of the results described here have already been published in Prayitno et al., (1999) and Rolfe et al. (2000). However, the reader should find that the findings from these publications are sufficiently detailed below. These publications have been lodged with the RIRDC; however, reprints can be requested from the authors. 4.1 Effect of the bacterial strains on rice growth Cultivar IR-28 inoculated with strain R4 and grown in test tubes gave the greatest growth stimulation but this was significant at the p = 0.05 level (data not shown). The growth of all other rice-rhizobium combinations was not significantly stimulated. Rice plants from each cultivar inoculated with strain E12 gave the poorest growth results. However, this procedure gives use to rice plants that are stressed and developed poorly over the 30 day period. The average shoot dry weight of the uninoculated plants ranged from 70 to 180 mg. Hence, we trialed alternative methods for growing rice plants that would be less labour intensive and would optimise the plant growth over the 30 day period. The cylindrical specimen jar method was used for growth-assessment in medium-term (30 day) experiments. This procedure using Hoagland s growth medium routinely gave average shoot dry weight of the uninoculated plants in the range of mg depending on the rice cultivar. Growth of the rice cultivars Calrose, Pelde and IR-28 was examined by the cylindrical specimen jar method following inoculation with a variety of the rice-associating bacterial isolates. The strains tested were Rhizobium strains R2, R4, and E4, and the IRRI isolates R38-0, R38-T, R53, and R58. Each of the three cultivars gave different growth responses to bacterial inoculation. On NFM the IRRI strain R38-T significantly enhances the growth of rice seedlings of cultivars Pelde and Calrose. The growth of cv. IR-28 seedlings inoculated with strain R38-T was enhanced but inhibited by strain R53 inoculation. However, these growth effects were not statistically significant. When the same experiments were done with 10 mm KNO3 added to the NFM growth medium, then the Rhizobium strain R4 significantly stimulated the seedling growth of cv. Pelde and IR-28, and both the IRRI strains R38-0 and R38-T significantly stimulated the growth of cv. Pelde seedlings (Prayitno et al., 1999). The IRRI strain R58 significantly stimulated the growth of cv. IR-28 seedlings. Under these conditions most of the bacterial inoculations had little effect on the growth of cv. Calrose seedlings. However, the Rhizobium strain E4 and the IRRI strain R53 significantly inhibited seedling growth and development in all three cultivars. On Hoagland s medium all the bacterial inoculum strains had minimal growth effects on the development of the different rice cultivars. The use of this plant growth medium greatly reduced the inhibition of seedling growth caused by strains E4 and R53. The basis of the modulation of the growth inhibition by Hoagland's medium was investigated by changing the various media compositions in an attempt to define the crucial constituents involved. As twelve different media combinations were compared in this series of experiments 2 plants were grown per specimen jar. The consequence is that the growth of the individual seedlings by 30 days is reduced but the original differences between F10 and Hoagland's media were still very apparent. Our studies indicate a complex interaction occurs between the levels of Ca 2+, K + and phosphate ions and nitrogen addition. Rice seedlings grown in medium F10#6 combination were similar to that found in Hoagland's medium (Prayitno et al., 1999). Seedlings grown in all other medium combinations were significantly poorer in growth. 6

13 4.2 Examination of inoculated rice roots We examined the roots in those cases where interactions were observed on the growth of the rice plants. When enhancement of plant growth takes place, it was always accompanied by extensive lateral root and root hair formation. Conversely, when rice seedling growth is severely inhibited, there was a marked reduction of the number of main roots, lateral roots and root hair formation along the main roots that did form. In the case of the IRRI strain R53, inhibition of rice seedling growth was accompanied by a major reduction in the total number of main roots formed, the lateral roots were very stunted and the formation of root hairs was greatly reduced and inhibited in their development. In contrast, the seedling growth inhibition induced by strain E4 was not accompanied with an obvious effect on the formation of these root structures (Prayitno et al., 1999). Having demonstrated enhanced rice growth, situations of growth inhibition and environmental effects on the responses of rice seedlings, we made use of a number of techniques to further characterise the bacterial colonisation of the rice seedling roots. 4.3 Multiplication of rice-associating strains within rice leaves An additional plant assay was developed to analyse if there was any possible relationship between the ability of a Rhizobium strain to affect seedling growth and its ability to survive and multiply within rice tissues (Prayitno et al., 1999). As the environment provided by the rice leaf can be easily used to study internal colonisation by bacteria, this assay measured the multiplication, movement and compatibility of the Rhizobium strains within rice tissues. In addition, it enables the use of various bacterial strains to be used as biological probes of any induced responses or preformed systems of plant responses in the rice plants. In this bioassay, bacterial cells were pressure-infiltrated into sections of the rice seedling leaf and viable bacterial counts were recorded every two days for up to 15 days. These assays were based on the current bacterial plant pathology leaf assays which demonstrate that different bacterial cells can only grow and multiply within leaf tissues if they contain particular bacterial genes which are associated with specific nutrient uptake systems. Bacteria with mutations in these genes, or which do not contain these genes, do not grow in plant tissues. We have observed that the Rhizobium strains can survive and multiply within rice leaves. The results for cultivar Pelde show that after infiltration into the leaf the different Rhizobium strains could multiply for 8 to 10 generations over a 12 to 15 day period (Prayitno et al., 1999). Even strain E4, which normally inhibits rice seedling growth, multiplies very well within rice leaves. Interestingly, additional assays to examine the spread of these bacteria within the leaf tissues away from the site of application found that even Rhizobium strains which can form long lines of internal root colonisation do not move out of the 2 cm strip containing the sites of the original leaf inoculation. 4.4 Seedlings grown in liquid media in Magenta jars The liquid medium systems were found to be the best for studying bacterial colonisation and the investigation of entry site of colonisation, location and the spreading of the bacteria in root tissues, and quantification of bacterial colonisation in roots. Therefore to mimic the responses of rice grown under flooded conditions, a procedure was developed using Magenta jars with liquid medium. When rice seedlings were inoculated with either strain R4(gfp) (rice growth stimulator), strain E4(gfp) or strain ANU843(gfp) (both inhibitors of rice growth) and grown in Magenta jars, a different phenomenon was observed with both strains E4(gfp) and ANU843(gfp). In McCartney bottles, the seedlings inoculated with strains E4(gfp) and ANU843(gfp) were not observed to colonise the intercellular spaces of the seedling. However, in the Magenta jar system at 21 days after inoculation, both strains E4(gfp) and ANU843(gfp) bacteria were observed along the surface grooves on main and lateral roots, at lateral root junction, and on root tips. In addition, both strain E4(gfp) and ANU843(gfp) bacteria were observed colonising intercellular spaces of lateral roots and forming short 7

14 lines of cells. Strain R4(gfp) behaved as in the experiments using liquid medium in McCartney bottles, forming long lines of bacteria in the lateral roots of inoculated seedlings. 4.5 Interaction of R. leguminosarum bv. trifolii strain ANU843 with rice seedlings Cultivar Pelde seedlings were inoculated with the clover strain ANU843 and the plants grown for 3 weeks. In comparison with the uninoculated plants, the strain ANU843-inoculated seedlings were very inhibited in their growth and were light yellow in colour. Although many of the individual plasmids of strain ANU843 can be cured, their presence does influence the phenotypic properties of the strain (Guerreiro et al 1998). To test whether information encoded on these plasmids could influence the interaction between strain ANU843 and rice, rice seedlings were inoculated with various derivative strains of ANU843, which had lost one of their plasmids. Strain ANU845, which had lost its Sym-plasmid, pa, behaved like the parental strain ANU843 inhibiting rice growth. Derivative strains, however, lacking either the pc plasmid or the pe plasmid or both plasmids pa and pc, did not inhibit rice seedling growth, and the plants were green at 3 weeks. These experiments were then extended to examine the effects of other strains on rice plant growth (Table 1) (Rolfe et al., 2000). Rice seedlings inoculated with Agrobacterium tumefaciens strain C58, its Ti-plasmid-cured derivative strain A136, or with Rhizobium strains E4, E11, or NGR234 all inhibited plant growth and produced yellow plants by 3 wk. In marked contrast was the finding that rice-associating strain R4 did not inhibit rice growth and the plants were green after 3 wk of incubation (Table 1). Table 1. Effect of bacterial inoculation on the growth of cultivar Pelde seedlings. Inoculant strain Uninoculated control Plant response a Plants grow, leaves green ANU843 Growth inhibited, leaves yellow ANU845pSym - Growth inhibited, leaves yellow CFNS152 pc - Plants grow, leaves green CFNS309pe - Plants grow, leaves green CFNS601pa - pc - Plants grow, leaves green R4 Plants grow, leaves green E11 Growth inhibited, leaves yellow E4 Growth inhibited, leaves yellow NGR234 Growth inhibited, leaves yellow C58 Growth inhibited, leaves yellow A136pTi - Growth inhibited, leaves yellow a Seedlings grown for 21 days in F10 medium. 4.6 Entry of the bacteria into rice root tissues To locate the entry site of the bacteria into rice root tissues, the GFP-labelled strains, R4(gfp), E4(gfp) and ANU843(gfp), were used. In this study, a slide system based on the work of Gage et al. (1996) and Magenta jars used to grow the inoculated plants. 8

15 Crack entry of lateral root emergence Colonisation of the bacteria at the cracks of epidermal cells which resulted from lateral root emergence was detected as early as three days after inoculation. These cracks formed on the main root at sites where epidermal cells breached longitudinally. Thus, one lateral root emergence resulted in two crack sites, one is on the region toward the culm and another site is toward root tip of the main root. Colonisation of cracks at lateral root emergence in this study refers to these two sites, not at the region encircling the lateral root junctions. The bacteria colonised either one of these sites or both of these sites resulted in an eye shaped formation under fluorescence microscopy. In our studies, the bacteria did not surround the lateral root junction to form a ring of colonisation at the lateral root junction as observed by Reddy et al. (1997). However, on rare occasions, the bacteria colonised epidermal cells of the main root or at the lateral root junction close to crack sites. From these cracks, the bacteria spread into the neighbouring epidermal cells, or into intercellular spaces of the main roots. In intercellular spaces, the bacteria were spread further along the root. When the inoculated plants were grown for a longer time period (21 days), the sub-lateral root emerged from the lateral roots of the main roots which resulted in the formation of fissures at the site of root emergence. The bacteria colonise the sub-lateral root junction either at one site of the cracks or at both site of the cracks as observed on the main roots. The bacteria also colonised cracks at root junctions of young sub-lateral root. At these cracks, the bacteria spread further intercellularly along the lateral roots. Entry between epidermal cell walls The initial colonisation of the bacteria is in the grooves between epidermal cells. From those grooves, the bacteria enter root tissue between epidermal cell walls and then get into intercellular spaces. The bacteria did not colonise the epidermal cells adjacent to the intercellular spaces containing the bacteria, suggesting that those epidermal cells were intact. However, the bacteria can colonise a collapsed epidermal cell, and then spread deeper into the root cortex. Thus, there was more than one mode of bacterial entry into rice tissues occurring in this bacteria-rice interactions. 4.7 Location and spread of the bacteria in the whole root system Spreading of the bacteria is mainly through intercellular space After entering root tissue at the cracks or between epidermal cell walls, the bacteria spread in the root tissue intercellularly. Intercellular spaces of rice tissues act as tunnels for the bacteria to spread further. At certain sites, these intercellular spaces were joined together, thus forming a network of intercellular spaces in roots. Intercellular spreading of the bacteria could be connected to the root exterior, thus providing a continuum from internal root tissue to the environment. Experiments showed the bacteria released from the internal tissue of root formed a large mass of cells. When this large mass of cells was disrupted, individual bacteria were spread into the liquid, and the bacteria in intercellular space moved toward that disrupted mass of cells. This suggested that intercellular spaces contain fluids, which enabled the bacteria to move inside them. In this study, there was no evidence that the bacteria spread further into the vascular bundles of the roots. When the root tips of the seedlings were cut before inoculating with the bacterial suspension and the roots were examined at 21 days, the bacteria could be seen in intercellular space in the cortex region, but not in the xylem of the root. Colonisation in lateral and main roots In order to be able to compare the intercellular colonising ability of bacterial strains isolated by Yanni et al. (1997), it was necessary to develop a quantitative plant assay. Intercellular colonisation (IC) was defined as a colonisation in intercellular spaces of lateral roots, or in the main roots, which have formed lines of colonies. The colonisation of intercellular spaces was considered positive if the length of line of colonies was more than the length of two plant cells at the time of examination. This operational definition was made to distinguish from intracellular colonisation of epidermal cells and 9

16 the colonisation at cracks of (sub) lateral roots. The colonisation of these strains was compared for plants grown in F10 medium and in Hoagland s medium. Intercellular colonisation (IC) of long lines of bacteria was observed on lateral roots as well as main roots. However, intercellular colonisation of lateral roots only occurred in fine lateral roots, which had diameters less than 150 mm. Some lateral roots that formed on main roots can develop to form sublateral roots, which then were colonised by the bacteria at the sub-lateral roots cracks. However, others did not form sub-lateral roots, but the bacteria could also colonise intercellularly. This intercellular colonisation could be initiated at the lateral root junctions and penetrate into the main root, or could reach the region just behind the root tip. However, when examination was made at 21 days after inoculation, intercellular colonisation of the bacteria in some of the lateral roots ended at a certain point between lateral root junction and root tip. Colonisation occurred in the lateral roots that did not form sub-lateral roots during their emergence and elongation. Three-day-old seedlings had one main root (M1) at the time of inoculation. The inoculated seedling in the growth medium forms the second root (M2), third root(m3), fourth root (M4), and so forth from the culm region by the seventh day of the experiment. Colonisation of the strains E4(gfp) and ANU843(gfp) occurred mainly on M1 roots that were inoculated with the bacterial suspension. This first main (M1) root can be easily distinguished with other main roots because it has fine and extensive lateral root structures, and the root was brown compare to light yellow of other main roots. When the plants were grown for 21 days, the main roots other than the M1 root have been colonised by the bacteria. Strains E4(gfp) and ANU843(gfp) could colonise the new main roots when the plants were grown in F10 medium or Hoagland medium. The colonisation by these strains was only on the main root and not in the lateral root. However, strain R4(gfp) was only found associated with the M1 roots when the seedlings were grown in F10 liquid medium. This finding suggested that the bacteria have to survive and grow in the liquid medium to be able to colonise the new main roots. When the bacteria were isolated from the Hoagland liquid medium of the plant at 14 d after inoculation, there were a high number of viable cells in the medium. The number of viable bacteria in Hoagland s medium was not different among the strains. However, strain R4(gfp) had the highest number of viable cells in F10 medium compared to E4(gfp) or ANU843(gfp). This result implied that the plant growth and root development plays an important role in the success of colonisation of the bacteria in root tissue. 10

17 Table 2. Effect of plant growth medium on intercellular colonisation (IC) of the bacteria on rice cv. Pelde at 21 days after inoculation Strain media % plants having IC 1) number of. plants having IC infections in 2 nd main root no. of lateral roots per plant 2) no. of IC infections per plant 2) % of IC infections per plant 2) Uninoculated F ± 36 a 2) - - R4(gfp) F ± 52 a 1.4 ± ± 0.1 (12/42) E4(gfp) F ± 39 b 1.5 ± ± 0.2 (13/42) ANU843(gfp) F10 22 (9/42) ± 35 b 1.1 ± ± 0.3 uninoculated H ± 42 a - - R4(gfp) H ± 38 a 7.5 ± ± 0.3 (24/42) E4(gfp) H ± 32 c 5.8 ± ± 0.2 (22/42) ANU843(gfp) H 41 (17/42) ± 39 b 3.3 ± ± 2.3 1) Values are average of three independent experiment having the total of 42 plants. 2) Values are mean ± SD of three independent experiments. Each jar contained six plants. 2) Values in the column followed by the different letter are significantly different from the other according to ANOVA test followed by the Least Significant Difference at P = F10, F10 medium; H, Hoagland s medium. 4.8 Better root growth provides more possible sites for colonisation, and increases the proportion of roots colonised by the bacteria The GFP tagged cells of strains (R4(gfp), E4(gfp) and ANU843(gfp)) were shown to colonise intercellular spaces of main and lateral roots. In general, the plants grown in F10 liquid medium had less lateral roots than the plants grown in Hoagland s medium, either in inoculated plants or noninoculated plants. In F10 medium, the inhibition of plant growth by strain E4(gfp) and ANU843(gfp) can be detected as early as 7 d after inoculation. This inhibition effect resulted in a significant reduction of lateral root formation (Table 2). This reduction also was observed in plants grown in Hoagland s medium. The lateral root formation of the plant inoculated with strain R4(gfp), which stimulate plant growth in specimen jars, was similar to the uninoculated control. The number of IC by these strains on plants grown in F10 liquid medium was not different. By growing the plants in Hoagland s medium, the number of IC of strains R4(gfp), E4(gfp) and ANU843(gfp) were increased from three to five fold. This increase in the number of IC per plant was also followed by an increase of the proportion of the root colonised by the bacteria (Table 2). Therefore, more lateral root formation provides more potential sites to be colonised by the inoculated bacteria. 11

18 4.9 Colonisation of rice roots and the bacterial effect on plant growth are not determined by the Sym-plasmid of Rhizobium or the Ti-plasmid of Agrobacterium The presence of the nod and nif genes on the psym (symbiotic) plasmid are responsible for the formation of a successful nitrogen-fixing nodule in Rhizobium-legume symbiosis. The role of the psym plasmid in lateral root junction (LRJ) and intercellular colonisation (IC) of rice roots was investigated using ANU845, a Sym-plasmid cured mutant of strain ANU843 (Djordjevic et al., 1983). It was found that the psym plasmid cured strain (ANU845gfp) colonised rice root intercellularly and LRJ at the same frequency as the wild type strain. Furthermore, the plant growth was inhibited by this strain to the same extent with its wild type strain ANU843(gfp). These results suggest that psymplasmid did not determined the colonisation ability of the bacteria and did not play a role in the bacterial inhibition of plant growth. Agrobacterium are soil bacteria which can cause crown gall tumours on a number of dicotyledonous plants (Thomashow et al., 1986). Agrobacterium is attracted chemotatically to wound site by wound exudates. The Ti plasmid of the Agrobacterium is responsible for the induction of the tumorous growth of plant cells. Host range of Agrobacterium is determined by Ti plasmid (Thomashow et al., 1986). Strain A136, which is cured of its Ti-plasmid, was used to study bacterial colonisation and plant growth. The GFP plasmid was transferred to strain A136 and the ability of strain A136(gfp) to colonise roots intercellularly and at LRJ was the same as its wild type strain C58(gfp) when rice seedlings were examined after 21 days. Moreover, its effects on plant growth was also similar to that its wild type strain or like other Rhizobium strains (Table 1) Certain plasmids of Rhizobium leguminosarum bv. trifolii affect plant growth, but not the colonisation ability R. l. bv. trifolii strain ANU843 harbours five plasmids with a size range from 180 kb to about 700 kb, the smallest size plasmid a (or Sym plasmid), b, c, d and e the largest. Recently, each plasmid or combination of these plasmids have been cured to study plasmid-encoded functions (Guerreiro et al., 1998). The GFP plasmid was successfully transferred into the five single plasmid cured mutants, of plasmid a (pa - ), c (pc - ), e (pe - ) and one combination of a and c (pa - c - ). The bacteria were inoculated onto three-day-old rice seedlings and the roots were examined under fluorescence microscopy after 14 days to study the effects that each plasmid might have on bacterial colonisation and on rice growth. All the mutant strains tested can colonise the plant intercellularly, suggesting that at least the plasmids a, c and e do not determine the ability of the bacteria to colonise rice root tissue. Strain pc - (gfp) had the highest percentage of IC in roots compared with the other strains and their wild type strain ANU843(gfp). However, the extent of the colonisation per plant by these strains was not different from each other. The proportion of roots having IC per plant were the same. Interestingly, mutants of strain ANU843, cured of plasmid c or e, did not inhibit plant growth as observed with their wild type strain ANU843(gfp) and for the mutant of Sym plasmid cured (pa - (gfp)) (Table 1) (Rolfe et al., 2000). Furthermore, the shoot dry mass of rice seedlings inoculated with pc - (gfp), pe - (gfp) or pa - c - (gfp) were not different from uninoculated control plants. However, the root dry mass of the seedlings inoculated with the strains pc -, or pe - were significantly different to the controls. These findings suggest that inhibition of plant growth is determined by genes influenced by the presence of plasmid c, or e or both of these plasmids. 12