Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Annals of Microbiology, 54 (4), 365-380 (2004) Localization of Azospirillum brasilense Cd in inoculated tomato (Lycopersicon esculentum Mill.) roots M. GRILLI CAIOLA1*, A. CANINI1, A.L. BOTTA2, M. DEL GALLO2 1Dipartimento di Biologia, Università di Roma Tor Vergata, Via della Ricerca Scientifica, 00133 Roma; 2Dipartimento di Biologia di Base e Applicata, Università dell’Aquila, 67010 L’Aquila, Italy Abstract - Tomato (Lycopersicon esculentum Mill.) seeds inoculated with Azospirillum brasilense Cd were cultured in vitro on solid medium with and without combined nitrogen. Inoculated and control roots from 30-day-old plants were examined by light microscopy (LM), scanning electron microscopy (SEM) and transmission electron microscopy (TEM), to monitor the adhesion, penetration, and localization of bacteria on and within the plant. We also studied the first colonisation steps of three-day-old plants under the microscope and bacteria in culture. Azospirillum brasilense Cd was identified by immunogold or immunofluorescence labelling. In culture, this bacterium has a thick polar flagellum and many thin peritrichous flagella. In aged dark cultures, small cells appeared within cyst-like forms. The bacteria used the polar flagellum to adhere to the epidermis, root hair and root cap areas of tomato roots. They also penetrated the epidermis, root hairs and outer cortex cells. Within the root, A. brasilense formed cyst-like cells that did not divide inside and were rich in poly-β-hydroxybutyrate granules and glycogen. Bacteria within the root had a thick capsule, many granules of different types, and high levels of SOD activity, suggesting that they can fix nitrogen in the intercellular spaces of tomato roots. Key words: Azospirillum, tomato, rhizobacteria, endophytic bacteria, ultrastructure, immunolocalisation. INTRODUCTION Recent studies (Whipps, 2001) have demonstrated that plants harbour an abundant, diverse microflora consisting primarily of bacteria and fungi. These microorganisms live both within the host plant as endophytes and on plant surfaces such as leaves and, in particular, in the rhizosphere. Many of these bacteria fix nitrogen and produce substances that directly or indirectly stimulate plant growth and protect the host plant from pathogens (Patriquin et al., 1983; Di Fiore and Del Gallo, 1995; Reinold-Hurek and Hurek, 1998). Many studies * Corresponding Author. Phone: +39 0672594344; Fax: +39 062023500; E-mail: [email protected] 365 involved the isolation of bacteria associated with plants, their culture and identification and reinoculation into the host plant (Levanony et al., 1989). These studies may facilitate the use of endophytes for agronomic purposes, reducing the need for fertilisers and pesticides during crop cultivation (Ceccherini et al., 2001; McCully, 2001). Demand for such techniques is currently increasing as they are seen as environmentally friendly, decreasing pollution levels. Most studies of nitrogen-fixing endophytes to date have dealt mostly with grasses and cereal crops (James and Olivares, 1997; Stenhout and Vanderleyden, 2000), with only a very small number of dicotyledonous plants such as Ipomoea batatas (Paula et al., 1991), Coffea arabica (McInroy and Kloepper, 1995), Gossypium (Jimenez-Salgado et al., 1997), and grapevine (Bell et al., 1995), considered. However, for some bacteria, studies have also been carried out on tomato (Lycopersicon esculentum Mill.) (Bashan et al., 1989; Bashan, 1998; Bashan and De-Bashan, 2002; Becker et al., 2002). Tomato is grown worldwide and is a crop of considerable economic, commercial and industrial interest. It is one of the most important crops in Italy, where particular efforts are made to select genotypes useful for cultivation in different soils and climates. Tomato crops require high inputs of nitrogen, phosphate and potassium. They are also frequently affected by numerous pathogens such as Rhizoctonia, Peronospora, Alternaria, and nematodes. This results in the application of large amounts of pesticides, considerably increasing the cost of tomato production, which is already expensive due to the cropping practices involved. The use of bacteria that fix nitrogen, promote plant growth, and protect against pathogens (Bashan and De-Bashan, 2002) and of mycorrhizal fungi capable of making phosphate available to plants, is therefore a highly attractive proposition, particularly as it would also reduce pollution levels in cultivated soils. We have therefore developed a research project investigating the association of tomato plants with a single bacterium, groups of bacteria and mycorrhizal fungi, with the aim of developing biological associations improving tomato crops in terms of nutritional, health and production qualities. This study dealt with tomato plants inoculated with Azospirillum brasilense Cd and grown in vitro. Bacteria of the genus Azospirillum are among the bestknown nitrogen-fixing and plant growth-promoting rhizobacteria (Okon and Kapulnik, 1986). Azospirillum brasilense is considered endophytic other than rhizospheric (Patriquin et al. 1983; de Oliveira Pinheiro et al., 2002). In this study, we followed bacterial adhesion, penetration and localization in the host plant. We demonstrated that the bacterium also establishes itself within the plant and that it reaches and penetrates the root early in the colonisation process. We also analysed morphological changes in the bacterium following colonisation of the plant, and interactions with the host. MATERIALS AND METHODS Bacterial strain. Azospirillum brasilense Cd, isolated from Cynodon dactylon (Eskew et al., 1977) was originally supplied by Yaacov Okon, Hebrew University Rehovot, Israel. It was grown on a modified Okon medium (OK) (MartinezDretz et al., 1984), solidified with 15 g l-1 agar (Oxoid), with or without 1 g l-1 combined nitrogen (NH4)2SO4. 366 M. GRILLI CAIOLA et al. Preparation of bacterial cultures for SEM, LM and TEM. Azospirillum cultures in exponential growth phase and old brown cultures of this bacterium in late stationary phase were fixed by incubation overnight at 4 °C in 2.5% glutaraldehyde (v/v) in 0.01 M phosphate buffer (PB). For SEM observations, the fixed bacteria were rinsed in PB, postfixed by incubation in 1% osmium tetroxide for 2 h at room temperature, rinsed in PB, dehydrated in a graded series of alcohol solutions, critical point dried, mounted on the stub, sputter coated with gold, and observed in a Zeiss DSM 950 microscope. For TEM observations, fixed bacteria were washed three times in 0.004% sucrose, collected by centrifugation and resuspended in 0.004% sucrose as described by Forni and Grilli Caiola (1992). A drop of this bacterial suspension was applied to a grid, dried and stained with 2% phosphotungstic acid (PTA), pH 6.5, for 10 min. Grids were washed in double-distilled water, dried and observed in a Zeiss CEM 902 microscope, operating at 80 kV. We also observed samples of these cultures by TEM, following PTA staining in the absence of fixation. For TEM observations of ultrathin sections, cultures fixed in glutaraldehyde as described above were postfixed in 1% osmium tetroxide, dehydrated and embedded in Epon resin. Sections (60 nm) were cut and stained with lead citrate as described by Reynolds (1963). Ultrathin sections were also stained with periodic acid-thyosemicarbazide-silver proteinate (PA-TCH-SP) for the detection of glycogen (Thiéry, 1967). Silver proteinate (SP) was applied to sections treated with osmium tetroxide or in the absence of osmium tetroxide, for the detection of phenolic compounds (Marinozzi et al., 1977). Inoculation of tomato and in vitro observation. Tomato (Lycopersicon esculentum Mill.) seeds supplied by Peto Seed, Seminis (Vegetable Seeds Italia srl, Parma, Italy), were sterilised in 5% NaClO for 3 min in a laminar flow hood. They were then rinsed in sterile tap water six times for 5 min each and four times for 20 min each, placed on sterile Whatman paper in Petri dishes, wetted with sterile distilled water and incubated for three days in the dark at 24 °C to ensure that there was no microbial contamination. Sterile germinated seeds were either transferred directly on Magenta bottles containing solid McKnight (Bergersen, 1980) medium supplemented with combined nitrogen (controls) or inoculated as follows. The seedlings were incubated for 30 min in a Petri dish containing a 0.5 cm layer of a 108 CFU (colony-forming units) of a washed culture of A. brasilense resuspended in phosphate-buffered saline (PBS). Inoculated seedlings were transferred to Magenta bottles containing solid McKnight medium without combined nitrogen and incubated in a growth chamber with a 16/8 light/dark cycle at 27 °C. Inoculated and control plants were harvested and analysed after 30 days. Some of the three-day-old sterile seedlings were also inoculated with Azospirillum under a phase-contrast microscope, to check that the bacterium displayed chemotactical behaviour towards the plant root. Preparation of roots for LM, SEM and TEM. For chemotactic observations, we cut the seedling rootlets, placed them on a slide, covered them with a coverslip and filled the space between the slide and the coverslip with PBS. We added 10 µl of a washed overnight culture of Azospirillum via the side of the Ann. Microbiol., 54 (4), 365-380 (2004) 367 coverslip, under a light microscope. We then followed the behaviour of the Azospirillum over a one-hour period. We also carried out observations on 30-day-old plants. Macroscopically, the areas of the root invaded by the bacterium appeared pink. We selected areas of the main and lateral roots displaying this pink coloration for investigation. Five roots from each of three plants (for both control and inoculated plants) were cut transversely into four segments: 1) from root tip to elongation zone; 2) hair zone; 3) mature zone; 4) upper part near the basal portion of the stem. Inoculated and control root segments were prepared for LM, SEM and TEM observations as described above for bacterial cultures. Thin (80-100 nm) and ultrathin (60 nm) sections were cut with a diamond knife. Uranyl acetate-lead citrate (Reynolds, 1963) staining was used for routine TEM observations, and PATCH-SP and SP staining was performed as described above (Thiery, 1967; Marinozzi et al., 1977). Preparation of antisera and immunoidentification of bacteria by LM and TEM. Azospirillum brasilense Cd was cultured to late exponential phase in OK liquid medium, in the presence of nitrogen. The culture was centrifuged, the cells were resuspended in PBS and the resulting suspension was autoclaved twice for 60 min each at 60 °C. Polyclonal antibodies against the resulting heatinactivated A. brasilense Cd cells were raised in rabbit as described by Leonardi et al. (1993). ELISA was carried out according to method previously described in Leonardi et al. (1993). Assay has been performed with A. brasilense cells and, as a control to determine the specificity of the polyclonal antibody, with other endophitic bacteria: Gluconoacetobacter diazotrophicus strain PAL 5 (Yamada et al., 1997), Burkholderia ambifaria strain PHP7 isolated from maize roots (Balandreau et al., 2001; Coenye et al., 2001) and Herbaspirillum seropedicae strain Z67 (Baldani et al., 1986). Azospirillum brasilense was identified on thin sections as follows. Slides with thin sections of tomato roots were incubated for 2 h at 25 °C with 5% BSA in normal goat serum diluted 1:20. The primary antibody (rabbit anti-Azospirillum), diluted to 1:300, was applied to the slides, which were then incubated overnight at 4 °C. Slides were washed three times with 0.1% Tween in PBS and incubated with the secondary antibody, a fluorescein isothiocyanate (FITC)conjugated goat anti-rabbit (GAR, SIGMA) antibody diluted 1:20. Finally, slides were washed in PBS and observed with a Leitz Aristoplan fluorescence microscope, using a PB 450-490 nm filter. Controls were performed by using preimmune serum diluted to 1:300. In order to evaluate autofluorescence due to fixative, the sections without primary antibody were observed. Thin sections were also observed at light microscopy after staining with toluidine blue. Bacteria from brown and light pure cultures and of inoculated roots were identified by TEM with a Zeiss CEM 902 microscope after immunohistological staining, as described elsewhere (Leonardi et al., 1993; Canini et al., 1996). Superoxide dismutase (SOD) detection. Immunogold labelling of Fe-SOD was achieved using a polyclonal antibody raised against Fe-SOD purified from the cyanobacterium Anabaena cylindrica. The antibody was applied at a dilution of 1:400 to ultrathin sections of bacteria or inoculated roots (Canini et al., 1992, 1996). As a control, preimmune serum was used in place of anti-Fe-SOD 368 M. GRILLI CAIOLA et al. antibody. The secondary antibody was a goat anti-rabbit (GAR, SIGMA) antibody conjugated with 10 nm gold particles diluted to 1:20. The density of gold particles was estimated based on the analysis of 10-20 micrographs for each sample. The values were represented by mean ± SD of data and analysed statistically with Student’s t-test. A value of p < 0.05 was considered statistically significant. RESULTS Azospirillum brasilense in culture Early in culture, A. brasilense formed slightly curved rods, about 0.9 µm in diameter, 2.1-3.8 µm in length. These cells had a thin cell wall with a single thick polar flagellum (Fig. 1A) and numerous shorter and thinner peritrichous flagella (Fig. 1B). Mature cultures were brown and contained larger, electrondense and roundish forms enclosed within a thick capsule housing two or more smaller cells (Fig. 1C) that were then released into the medium. The dark colour of these mature cultures probably results from the presence of melanin, as reported by Sadasivan and Neyra (1987) and Faure et al. (1995). PA-TCH-SP treatment showed that there was no glycogen or other polysaccharide with vicinal groups within the bacteria (Fig. 1D). Immunogold labelling, with an antibody against A. brasilense, was very intense on the bacterial cell wall (Fig. 1E). No labelling of the control, to which no antibody was added, was detected (Fig. 1F). We observed no difference in immunogold labelling between actively growing light-coloured cells and mature, brown colonies. ELISA ELISA was carried out with a polyclonal antibody against A. brasilense used at dilutions of 1:500, 1:1000 and 1:2000. Absorbance at 450 nm decreased with increasing dilution (1.524 for 1:500, 1.103 for 1:1000, and 0.985 for 1:2000). Weak absorbance was observed with PBS used as a control and with preimmune serum. The value obtained for preimmune serum, 0.302, was very similar to that obtained for the PBS control, 0.215. As a control, the serum was tested on Gloconoacetobacter diazotrophicus, Burkholderia cepacea, and Herbaspirillum seropedicae endophytic bacteria giving values similar to PBS. Inoculation and localisation of Azospirillum brasilense on tomato rootlets Light microscopic examination of the inoculation of three-day-old rootlets with A. brasilense showed strong chemotactic attraction of the bacteria for the roots. Within a few seconds, most of the bacteria had moved towards the root, showing a strict preference for the root cap (Fig. 2A) and in particular for the detached cells of the growing area (Fig. 2B). Two to three minutes later, the bacteria seemed to be moving preferentially towards the root hairs (Fig. 2C), where they began to form clumps after about 20 min. Infection thread-like structures were observed penetrating the root hairs (Fig. 2D). After 1 h, the bacteria were established in the root cap mucilage and large clumps of bacteria were present in the growing area of the root. These data were confirmed by SEM observations on roots of inoculated plants, which indicated that the bacteria were able to colonise and adhere to the Ann. Microbiol., 54 (4), 365-380 (2004) 369 FIG. 1 – TEM of Azospirillum brasilense in culture. A: bacteria (b) from an actively growing culture, after PTA staining, showing a thick, long flagellum (f). B: bacteria (b) from the above culture, after PTA staining, with thin peritrichous flagella (pf). C: ultrathin sections of A. brasilense from a mature, brown culture with electrondense cyst-like structures (arrows) containing smaller cells. D: A. brasilense from a mature, brown culture after PA-TCH-SP staining, with no positive inclusions (arrows). Many small cells (s) are dispersed throughout the medium. E: gold particles on bacterial (b) capsule (arrows) after incubation with a gold-conjugated antibody against A. brasilense. F: control treated with preimmune serum. No gold granules are visible on bacteria (b). Bars: 1 µm. 370 M. GRILLI CAIOLA et al. FIG. 2 – Microscopy of a tomato rootlet inoculated under the light microscope with Azospirillum brasilense. A: bacteria (b), the small green spots, added at the point indicated by the arrows, reached within a few seconds the root cap area (r) surrounded by a cloud of mucilage and the detached cap cells (dcc). B: after a few minutes, the bacteria (b), small coloured spots, started to penetrate the root cap mucilage (m) and to form clumps (bc) as soon as they reached the root surface, close to the detached cells (dcc) of the growing area. C: the bacteria adhered to the root hairs (rh) and formed clumps (bc). D: a few minutes after inoculation, the bacterial colonies (bc) began to penetrate the root hairs (rh). Bars: 10 µm. Ann. Microbiol., 54 (4), 365-380 (2004) 371 root surface individually (Fig. 3A), by means of the polar flagellum whereas this behaviour cannot be detected in colonies. The favoured points on the root for adhesion were the sites of emergence of lateral roots, root hairs and the root cap. The presence of bacteria on the root surface seems to increase in correspondence of the abundance of exudates Localisation of Azospirillum brasilense within tomato roots Fluorescence microscopy of thin transverse sections of 30-day-old inoculated roots treated with a polyclonal antibody against A. brasilense (Fig. 3B) revealed that bacteria were present both on and in the epidermis, outer cortex cells, and root hairs. Bacteria were also detected in a mucigel substance exuded from the root surface and root hair. In roots deprived of the epidermis, SEM showed the presence of bacteria in large colonies in a cavity without separating walls (Fig. 3C). However, in this case, the bacteria were roundish, without flagella, and larger than those in culture and in root hairs. Ultrathin sections of root hairs observed by TEM after PA-TCH-SP staining (Fig. 3D) showed electrondense exudate outside the root hair wall and bacteria grouped within the curled root hair tip. Furthermore, each bacterium was surrounded by a thick electrontransparent envelope. These features suggest that immediately after their penetration into the root hairs, they are in a vegetative dividing stage but that their morphology and ultrastructure then rapidly change. Indeed, on TEM, the bacteria in the inner root were found to form large colonies in the intercellular spaces between the outer cortical cells (Fig. 3E). The intercellular spaces appeared to be enlarged, as the colonies increased in size and the host cell wall lysed as a result, ending up as a very thin layer similar to a membrane (Fig. 3E). The intercellular material disappeared and the space was totally occupied by bacteria. Tomato cells adjacent to the invaded spaces appeared to lack contents and to be dead, whereas controls showed normally structured cells (Fig. 3F). Tomato roots were not colonised over their entire surface and internal structures, but only in some portions, as demonstrated by the comparison of transverse sections of the root taken at various distances from the apex to the stem base. In none of the sections were bacteria present all over the epidermis and throughout the peripheral cortex. Similarly, longitudinal sections demonstrated discontinuous bacterial colonisation, with some parts of the root invaded and others not. Bacteria ultrastructure in the root Ultrathin sections of areas containing large amounts of bacteria, as shown by LM and SEM, were examined by TEM. The bacteria within the roots were large, roundish (Fig. 3C, 4A) or rod-shaped (Fig. 4B) with a thick capsule (Fig. 4A, 4B), similar to the cyst-like forms reported in other Azospirillum strains (UmaliGarcia et al., 1980; Lamm and Neyra, 1981). The bacterial cells contained numerous large, intracellular granules of poly-β-hydroxybutyrate (PHB) (Fig. 4B) and glycogen, as shown by PA-TCH-SP staining (Fig. 4C). No bacteria were observed in the endodermis cells, inner cortex layers or central root cylinder. No bacteria were visible in the control plant roots. Only in the first and second segments of root was it possible to detect, among the cyst-like forms, some bacteria still equipped with thin flagella. These forms were very similar to those observed in culture and may correspond to bacteria that had recently penetrated into areas that had already been invaded. 372 M. GRILLI CAIOLA et al. FIG. 3 – Inoculated tomato roots viewed by LM, SEM and TEM. A: tomato root surface viewed by SEM, with Azospirillum brasilense (b) adhering to the root surface by means of its polar flagellum (arrow); bar: 1 µm. B: thin section of tomato root inoculated with A. brasilense, treated with the antibody against A. brasilense and observed by fluorescence microscopy. The root hairs (rh) and epidermis (arrow) are filled with fluorescent bacterial colonies (bc); bar: 25 µm. C: root of tomato with colonies of A. brasilense (b) after removal epidermis, viewed by SEM; bar: 3 µm. D: ultrathin section of a root hair containing bacteria (b) within the enlarged tip and covered with electrondense mucigel (arrow) stained with PA-TCH-SP; bar: 1 µm. E: ultrathin section of tomato root containing bacterial colonies (bc) in the intercellular spaces; bar: 1 µm. F: cells from an uninoculated tomato root. The intercellular spaces (is) are filled with electrondense material and the cell wall (cw) appears linear. Most of the cell is occupied by a large vacuole (v), whereas the cytoplasm (cy) is reduced to a thin layer. Pb-uranyl acetate staining; bar: 1 µm. Ann. Microbiol., 54 (4), 365-380 (2004) 373 FIG. 4 – Azospirillum brasilense within tomato roots, as viewed by TEM. A: bacterial cells (b) with a very thick capsule (c) and numerous large PHB granules dispersed in abundant fibrillae-like (fl) structures adhering to the bacterial wall; bar: 0.5 µm. B: bacteria showing a thick capsule (c), large PHB and cytoplasmic electrondense granules (g). The tomato cell wall (cw) is visible towards the top of the micrograph (cw); bar: 0.5 µm. C: tomato cell containing bacterial colonies. Root cell wall (cw) and cytoplasm (cy) appear lysed, whereas bacteria (b) show PHB and electrondense granules (g) after PA-TCH-SP staining; bar: 1 µm. D: bacteria after PA-TCH-SP staining show many electrondense glycogen (g) granules, PHB, capsule (c), and cytoplasm membrane (cm); bar: 1.8 µm. 374 M. GRILLI CAIOLA et al. FIG. 5 – Immunogold labelling of bacteria within tomato roots. A: heavy immunogold labelling (arrow) of Azospirillum brasilense by an antibody against A. brasilense. B: immunogold labelling of A. brasilense (arrow) by an antibody against Fe-SOD. Βars: 1 µm. Immunogold detection of bacteria within tomato roots Bacteria within tomato roots were detected with an antibody raised against A. brasilense cultures. Immunogold labelling was very evident on the bacterial wall (Fig. 5A), as observed in culture. No differences were observed between the bacteria present in the inoculated root sections. No immunolabelling was detected in uninoculated tomato roots. Immunogold detection of SOD within the bacteria in tomato roots Immunogold labelling with an anti-Fe-SOD antibody was much more intense in bacteria within the root (12 ± 1 gold particles µm-2) than in cells in culture (4 ± 0.2 gold particles µm-2). In bacteria within the roots, labelling was intense and associated with the bacterial plasma membrane, just beneath the cell wall (Fig. 5B). DISCUSSION Azospirillum brasilense in culture had morphological features similar to those reported by other authors for this species (Tarrand et al., 1978). In solid culture, it had a thick, long flagellum and numerous other thinner peritrichous flagella. However, pure actively growing cultures contained no PHB, whereas PHB and glycogen granules were abundant in bacteria within tomato roots. Actively growing and mature stationary cultures displayed different cell organisations. In the brown mature cultures, many colonies were very electronAnn. Microbiol., 54 (4), 365-380 (2004) 375 dense, contained smaller cells and had a thicker envelope than observed in younger cultures. These structures could be identified as the cyst-like forms described by Lamm and Neyra (1981), Sadasivan and Neyra (1985 and 1987) and Biondi et al. (personal communication) both in the wild type and in a mutant of A. brasilense. The brown colour of the cultures may be attributed to the melanin present in these (Sadasivan and Neyra, 1985, 1987; Givaudan et al., 1993; Faure et al., 1995) and other bacteria (Cubo et al., 1988; Castro-Sowinski et al., 2002). In this study, we followed the colonisation of tomato roots by Azospirillum and considered the distribution of these bacteria within the root. In inoculated tomato roots, A. brasilense occurs along the root axis from the apical zone to the base of the stem. Externally, bacteria tended to concentrate at the sites of lateral root emergence, the root hairs, and root cap. The root cap is a preferred site for bacterial colonisation right from the start, as shown by microscopic analysis of the inoculation of three-day-old roots. We observed the behaviour of this bacterium during the first few minutes of contact with the root and after it has established itself within the plant. The bacteria rapidly attach to the roots by means of polar and lateral flagella and adhesion seems to be mediated by attractants present in the mucigel excreted from root cells and in the bacterial capsule (Burdman et al., 2000). The bacteria adhere firmly to the root surface and were not detached by the numerous preparation treatments for electron microscopy. Bacteria often displayed a polar disposition. Both the bacterial surface and the polar flagellum were found to be involved in attachment, as suggested by Croes et al. (1993), Burdman et al. (2000), and Steenhoudt and Vanderleyden (2000). The bacteria penetrate the host plant via the root hairs and epidermis cells. However, they seem to make use of the natural intercellular spaces or artificial breaks between the epidermis cells. In our observations, the bacteria concentrated, 30 min after inoculation, in the growing area of the root tip, which contains detached cells and breaks. Indeed, within the roots, we observed bacteria between the host cells, aligned in rows or aggregated into large colonies in the intercellular spaces, and within cells that appeared to be lysed and dead. Similar observations were reported by Sprent and de Faria (1988) for endophytic bacteria whereas according to de Oliveira Pinheiro et al. (2002) A. brasilense CD should be unable to infect the interior of wheat roots. However, cell penetration may be favoured by pectinolytic enzymes or other polymer-degrading enzymes, as suggested by Umali-Garcia et al. (1980) and demonstrated by Vande Broeck et al. (1998) for Azospirillum irakense. Whereas bacteria seemed to adhere and to penetrate mostly in the apical and root hair zones, isolated bacterial colonies were visible below the epidermis, right up the root axis to the stem base. Within the root, the morphology and activity of the bacteria change from those observed in culture. They appeared quite similar to rhizobium bacteroid forms and to azospirillum cyst-like forms, with a very thick capsule, and numerous large PHB and glycogen granules. Within the tomato root, cyst-like bacterial forms may provide a means of defence against the host cell, but some aspects of these cysts suggest that they may instead be active forms, with high levels of oxidative metabolism, as revealed by the large amounts of SOD present and the accumulation of reserves (PHB, glycogen). Other nitrogen-fixing mi- 376 M. GRILLI CAIOLA et al. croorganisms display similar features, which may help to generate microaerobic conditions, protecting the nitrogenase from oxygen (Clara and Knowles, 1984). However, these cyst-like forms are different from those observed in cultures rich in melanin, which are frequently divided into smaller cells, and lack glycogen. Changes in cell organisation may enable the bacterium to change function. We cannot exclude the possibility that intermediate bacterial forms are produced between penetration and the end of our observation period. Bacteria within root hairs do not present a cyst-like form, but are instead more like rhizobia in the infection thread. However, the numerous observations carried out on inoculated tomato roots systematically showed the presence of Azospirillum in cystic forms. The number and sizes of PHB and glycogen granules observed and the thick capsule are typical of endophytic bacteria in tomato roots. This study is the first to report the presence of glycogen storage bodies in Azospirillum. PHB has been detected in Azospirillum brasilense in other studies (Tal and Okon, 1985; Kadouri et al., 2002), but was interpreted differently. Similar inclusions have been detected in rhizobia (Kim and Copeland, 1996) and in various other nitrogen-fixing bacteria, both free-living and associated with plants. They may correspond to substrate accumulated in the presence of large amounts of carbon, used to supply hydrogen for nitrogen fixation. Activity of this type has been detected in Azospirillum cultures and in inoculated tomato plants (Becker et al., 2002), and by means of photoacoustic laser in our material (unpublished results). Nitrogen fixation in A. brasilense occurs at very low oxygen concentrations (Clara and Knowles, 1984; Hartman and Burris, 1987) and, within the plant, structural and biochemical protection is probably required to enable this process to occur. The location of the capsule and SOD at the cell periphery, as observed in mature Chroococcidiopsis cells (Grilli Caiola and Canini, 1996) may provide a barrier, protecting the cell against oxygen radicals in the cytoplasm and in the extracytoplasmic area. The adhesion to, penetration into and distribution of bacteria on and in the inner root tissues, forming colonies in intercellular spaces rather than in tomato cells, are similar to those reported by Schlother and Hartmann (1998) in wheat. Lysis of the walls of tomato cells surrounding the bacterial colonies requires pectinolytic and cellulolytic enzymes, which have yet to be identified and which seem to be limited to the penetration zone in that the bacteria seem to be unable to invade the inner cortex and endodermis, thereby limiting systemic diffusion in the plant. The association of tomato with A. brasilense may protect the bacterium against competition with other rhizosphere and soil bacteria, and provide the tomato plant with additional nitrogen and auxins. The root hairs and epidermis cells lost due to bacterial penetration may be replaced by new roots and root hairs, the production of which is induced by the bacterium. This capacity of A. brasilense has been reported by Bashan (1998) and Steenhoudt and Vanderleyden (2000) in tomato and other plants. Experiments with antibodies demonstrated that A. brasilense CD was present within tomato roots, but we were unable to detect bacteria in the endodermis and central cylinder despite the presence of large numbers of bacteria on plate counts of bacteria within the plant after surface sterilisation: 107 per g plant dry weight in the stem and leaves, and 108 per g root dry weight. Ann. Microbiol., 54 (4), 365-380 (2004) 377 Studies dealing with the metabolic activity of these bacteria within the plant, focusing particularly on the regulation of nitrogen fixation and IAA production, may provide significant insight into the development of plant growth-promoting rhizobacteria. Acknowledgments This research was funded by grants from the Ministry of University and Technological Research (COFIN 2000). REFERENCES Balandreau J., Viallard V., Cournoyer B., Coenye T., Laevens S., Vandamme P. (2001). Burkholderia cepacia genomovar III is a common plant associated bacterium. Appl. Environ. Microbiol., 67: 982-985. Baldani J.I., Baldani V.L.D., Soldin L., Dobereiner J. (1986). Characterization of Herbaspirillum seropedicae gen. nov., a root associated nitrogen fixing bacterium. Int. J. Syst. Bacteriol., 36: 86-93. Bashan Y., Singh M., and Levanony H. (1989). Contribution of Azospirillum brasilense Cd to growth of tomato seedlings is not through nitrogen fixation. Can. J. Bot., 67: 2429-2434. Bashan Y. (1998). Azospirillum plant-growth-promoting strains are non-pathogenic on tomato, pepper, cotton, and wheat. Can. J. Microbiol., 44: 168-174. BashanY., De-Bashan L.E. (2002). Protection of tomato seedlings against infection by Pseudomonas syringae pv. tomato by using the plant growth-promoting bacterium Azospirillum brasilense. Appl. Environ. Microbiol., 68: 2637-2643. Becker D., Stanke R., Fendrik I., Frommer W.B., Vanderleyden J., Kaiser W.M., Hedrich R. (2002). Expression of the NH (+)(4)-transporter gene LEAMT1; 2 is induced in tomato roots upon association with N(2)-fixing bacteria. Planta, 215: 424-429. Bergersen F.J. (1980). Methods for evaluating biological nitrogen fixation. John Wiley and Sons, Chichester. Bell C.R., Dickie G.A., Harvey W.L.G., Chan. J.W.Y.F. (1995). Endophytic bacteria in grapevine. Can. J. Microbiol., 41: 46-53. Burdman S., Okon Y., Jurkevitch E. (2000). Surface characteristics of Azospirillum brasilense in relation to cell aggregation and attachment to plant roots. Critical Review Microbiol., 26: 91-110. Canini A., Civitareale P., Marini S., Grilli Caiola M., Rotilio G. (1992). Purification of iron-superoxide dismutase from the cyanobacterium Anabaena cylindrica Lemm. and localization of the enzyme in heterocysts by immunogold labeling. Planta, 187: 438-444. Canini A., Albertano P., Leonardi D., Di Somma D., Grilli Caiola M. (1996). Superoxide dismutase in cyanobacteria of the Baltic Sea. Arch. Hydrobiol. Algological Studies, 83: 129-143. Clara R.W., Knowles R. (1984). Superoxide dismutase, catalase, and peroxidase in ammonium-grown and nitrogen-fixing Azospirillum brasilense. Can. J. Microbiol., 30: 1222-1228. Castro-Sowinski S., Martinez-Drets G., Okon Y. (2002). Laccase activity in melaninproducing strains of Sinorhizobium meliloti. FEMS Microbiol. Lett., 209: 119-125. Ceccherini M.T., Castaldini M., Azzini A., Fancelli S., Bazzicalupo M., Miclaus N. (2001). Occurrence of Azospirillum brasilense in soils amended with swine manure. Ann. Microbiol., 51: 29-38. Coenye T., Mahenthiralingan E., Henry D., Hoste B., Vandemeulebroukcke K., Gillis M., Speert D.P., Vandamme P. (2001). Burkholderia ambifaria sp. nov. a novel 378 M. GRILLI CAIOLA et al. member of the Burkholderia cepacia complex comprising biocontrol and cystic fibrosis related isolates. Int J. Syst. Evol. Microbiol., 51: 1481-1490. Croes C.L., Moens S., Van Bastelaere E., Vanderleyden J., Michiels K.W. (1993). The polar flagellum mediates Azospirillum brasilense adsorption to wheat roots. J. Gen. Microbiol., 139: 2261-2266. Cubo M.T., Buendia-Claveria A.M., Beringer J.E., Ruiz-Sainz J.E. (1988). Melanin production by Rhizobium strains. Appl. Environ. Microbiol., 54: 1812-1817. De Oliveira Pinheiro R., Boddey L.H., James E.K., Sprent J.I., Boddey R.M. (2002). Adsorption and anchoring of Azospirillum strains to roots of wheat seedlings. Plant Soil, 246: 151-166. Di Fiore S., Del Gallo M. (1995). Endophytic bacteria: their possible role in the host plant. In: Fendrik I., Del Gallo M., Vanderleyden J., de Zamaroczy M., Eds, Azospirillum VI and Related Microorganisms. NATO ASI. Series G Ecological Sciences, Springer Verlag, Berlin, pp. 169-187. Eskew D.L., Focht D.D., Ting I.P. (1977). Nitrogen fixation, denitrification and pleiomorphic growth in a highly pigmented Spirillum lipoferum. Appl. Environ. Microbiol., 34: 582-585. Faure D., Jacoud C., Bouillant M.L., Bally R. (1995). Laccase activity in Azospirillum lipoferum: characteristics, melanization and Tn5 mutants. In: Fendrik I., Del Gallo M., Vanderleyden J., de Zamaroczy M., Eds, Azospirillum VI and Related Microorganisms. NATO ASI. Series G Ecological Sciences, Springer Verlag, Berlin, pp. 169-187. Forni C., Grilli Caiola M. (1992). The morphogenesis of Arthrobacter species isolated from the fern Azolla filiculoides Lam. Microbiologica, 15: 271-280. Givaudan A., Effosse A., Faure D., Potier P., Bouillant M-L., Bally R. (1993). Polyphenol oxidase in Azospirillum lipoferum isolated from rice rhizosphere: evidence for laccase activity in non motile strains of Azospirillum lipoferum. FEMS Microbiol. Lett., 108: 205-210. Grilli Caiola M., Canini A. (1996). Iron superoxide dismutase (Fe-SOD) localization in Chroococcidiopsis sp. (Chroococcales, Cyanobacteria). Phycologia, 35: 90-94. Hartmann A., Burris R.H. (1987). Regulation of nitrogenase activity by ammonium chloride in Azospirillum spp. J. Bacteriol., 165: 864-870. James E.K., Olivares F.L. (1997). Infection and colonization of sugar cane and other graminaceous plants by endophytic diazotrophs. Crit. Rev. Plant Sci., 17: 77-119. Jimenez-Salgado T., Fuentes-Ramirez L.E.,Tapia-Hernandez A., Mascarua-Esparza M.A., Martinez-Romero E., Caballero-Mellado J. (1997). Coffea arabica L. a new host plant for Acetobacter diazotrophicus, and isolation of other nitrogen-fixing acetobacteria. Appl. Environ. Microbiol., 63: 3676-3683. Kadouri D., Burdman S., Jurkevitch E., Okon K. (2002). Identification and isolation of genes in poly (β-hydroxybutyrate) biosynthesis of Azospirillum brasilense and characterization of a phbC mutant. Appl. Environ. Microbiol., 68: 2943-2949. Kim S.A., Copeland L. (1996). Enzymes of poly-β-hydroxybutyrate metabolism in soybean and chickpea grasses growing in New Jersey and New York. Appl. Environ. Microbiol., 62: 4186-4190 Lamm R.B., Neyra C.A. (1981). Characterization and cyst production of azospirilla isolated from selected grasses growing in New Jersey and New York. Can. J. Microbiol., 27: 1320-1325. Leonardi D., Canini A., Forni C. (1993). Immunological comparison between Arthrobacter isolates and bacteria living in Azolla filiculoides Lam. Symbiosis, 15: 269-283. Levanony H., BashanY., Romano B., Klein E. (1989). Ultrastructural localization and identification of Azospirillum brasilense Cd on and within wheat root by immunogold labeling. Plant Soil, 117: 207-218. Marinozzi V., Ciofi Luzzatto A., Conti Devirgiliis L., Mazzatenta C. (1977). Osmium-dependent argentaffinity of elastic fibers: its relations with polyphenolic compounds. J. Submicr. Cytol., 9: 267-274. Ann. Microbiol., 54 (4), 365-380 (2004) 379 Martinez-Dretz G., Del Gallo M., Burpee C., Burris R.H. (1984). Catabolism of carbohydrates and organic acids and expression of nitrogenase by Azospirilla. J. Bacteriol., 115: 436-442. McCully M.E. (2001). Niches for bacterial endophytes in crop plants: a plant biologist’s view. Aust. J. Plant Physiol., 28: 983-990. McInroy J.A., Kloepper J.W. (1995). Survey of indigenous bacterial endophytes from cotton and sweet corn. Plant Soil, 173: 337-342. Okon Y., Kapulnik Y. (1986). Development and function of Azospirillum-inoculated roots. Plant Soil, 90: 3-16. Patriquin D.G., Dobereiner J., Jain D.K. (1983). Sites and processes of association between diazotrophs and grasses. Can. J. Microbiol., 29: 900-915. Paula M.A., Reis V.M., Dobereiner J. (1991). Interactions of Glomus clarum with Acetobacter diazotrophicus in infection of sweet potato (Ipomoea batatas), sugarcane (Saccharum spp.), and sweet sorghum (Sorghum vulgare). Biol. Fertil. Soils, 11: 111-115. Reinhold-Hurek B., Hurek T. (1998). Life in grasses: diazotrophic endophytes. Trends Microbiol., 6: 139-144. Reynolds C.S. (1963). The use of lead citrate of high pH as electron opaque stain in electron microscopy. J. Cell Biol., 47: 195-200. Sadasivan L., Neyra C.A. (1985). Flocculation in Azospirillum brasilense and Azospirillum lipoferum: exopolysaccharides and cyst formation. J. Bacteriol., 163: 716-723. Sadasivan L., Neyra C.A. (1987). Cyst production and brown pigment formation in aging cultures of Azospirillum brasilense ATCC 29145. J. Bacteriol., 169: 1670-1677. Schloter M., Hartmann A. (1998). Endophytic and surface colonization of wheat roots (Triticum aestivum) by different Azospirillum brasilense strains studied with strainspecific monoclonal antibodies. Symbiosis, 25: 159-179. Sprent J.I., de Faria S.M. (1988). Mechanisms of infection of plants by nitrogen fixing organisms. Plant Soil, 110: 157-165. Steenhoudt O., Vanderleyden J. (2000). Azospirillum, a free-living nitrogen-fixing bacterium closely associated with grasses: genetic, biochemical and ecological aspects. FEMS Microbiol. Rev., 24: 487-506. Tal S., Okon Y. (1985). Production of the reserve material poly-β-hydroxybutyrate and its function in Azospirillum brasilense Cd. Can. J. Microbiol., 31: 608-613. Tarrand J.I., Krieg N.R., Dobereiner J. (1978). A taxonomy study of the Spirillum lipoferum group, with descriptions of a new genus, Azospirillum lipoferum (Baijerinck) comb. nov. and Azospirillum brasilense sp. nov. Can. J. Microbiol., 24: 967-980. Thiéry J.P. (1967). Mise en evidence des polysaccharides sur coupes fines en microscopie electronique. J. Microscop., 6: 987-1018. Umali-Garcia M., Hubbel D.H., Gaskins M.H., Dazzo F.B. (1980). Association of Azospirillum with grass roots. Appl. Environ. Microbiol., 39: 219-226. Vande Broeck A., Bekri A.M., Dosselaere F., Faure D., Lambrecht M., Okon Y., Costacurta A., Prinsen E., De Troch P., Desair J., Keijers V., Venderleyden J. (1998). Azospirillum-plant root associations: genetics of IAA biosynthesis and plant cell wall degradation. In: Elmerich C., Kondorosi A., Newton W.E., Eds, Biological Nitrogen Fixation for the 21st Century, Kluwer Academic Publishers, Boston, pp. 375-376. Whipps J.M. (2001). Microbial interactions and biocontrol in the rhizosphere. J. Exper. Botany, 52: 487-511. Yamada Y., Hoshino K., Ishikawa T. (1997). The phylogeny of Acetic Acid Bacteria Based on the Partial Sequences of 16S Ribosomal RNA: The elevation of the Subgenus Gluconoacetobacter to the generic level. Biosci. Biotech. Biochem., 61: 1244-1251. 380 M. GRILLI CAIOLA et al.