Herbaspirillum colonization increases growth and nitrogen accumulation in aluminium-tolerant rice varieties

Authors

  • Prasad Gyaneshwar,

    1. Crop, Soil and Water Sciences Division, International Rice Research Institute, MCPO Box 3127, 1271 Makati City, Philippines;
    2. Present address: Department of Plant and Microbial Biology, University of California, Berkeley, CA 94720, USA
    Search for more papers by this author
  • Euan K. James,

    Corresponding author
    1. Centre for High Resolution Imaging and Processing, MSI/WTB Complex, School of Life Sciences, University of Dundee, Dundee DD1 5EH, UK;
    Search for more papers by this author
  • Pallavolu M. Reddy,

    1. Crop, Soil and Water Sciences Division, International Rice Research Institute, MCPO Box 3127, 1271 Makati City, Philippines;
    Search for more papers by this author
  • Jagdish K. Ladha

    1. Crop, Soil and Water Sciences Division, International Rice Research Institute, MCPO Box 3127, 1271 Makati City, Philippines;
    Search for more papers by this author

Author for correspondence: Euan K. James Tel: +44 (0)138 234 4741 Fax: +44 (0)138 234 5893 Email: e.k.james@dundee.ac.uk

Summary

  •  Varieties of rice (Oryza sativa) differing in tolerance to aluminium (Al) were evaluated for their N-fixation ability after inoculation with a gusA-marked strain of Herbaspirillum seropedicae Z67.
  •  Under axenic conditions, by 30 d, inoculation resulted in enhanced nitrogenase activity, d. wt, total N and total C content only in the Al-tolerant varieties, and one (cv. Moroberekan) showed significantly more 15N2 incorporation than an Al-sensitive variety (‘IR45’). There were no differences in the number of the bacteria colonizing the different varieties, but the Al-tolerant ones secreted larger amounts of C in their root exudates, and bacteria colonizing the roots of cv. Moroberekan strongly expressed gusA and NifH proteins.
  •  Under glasshouse conditions, by 30 d, inoculation resulted in increased growth of both cvs IR45 and Moroberekan, but the latter showed significantly greater nitrogenase activity and 15N dilution. In a long-term experiment, by 120 d, cv. Moroberekan showed a significant increase in N content after inoculation.
  •  Herbaspirilla were localized on and within roots and aerial parts of cvs Moroberekan and IR45 under both growth conditions. The role of N fixation in growth promotion of rice by H. seropedicae is discussed in terms of availability of C.

Introduction

Associative biological nitrogen fixation is a significant feature of the lowland rice systems that provide more than 80% of the world’s total rice production. However, rice yields in these systems are very low and must be increased by about 50% in order to meet the projected demands in 2020 (Ladha & Reddy, 2000). This could necessitate a doubling in the use of N fertilizers, which is neither desirable nor sustainable (Ladha & Reddy, 2000). A potential alternative is to increase the contribution made by N fixation.

Diverse diazotrophic bacteria abound in the rice rhizosphere (Ueda et al., 1995a,b), and many endophytic N2-fixing bacteria have also been isolated from surface-sterilized tissues of rice (Barraquio et al., 1997; Engelhard et al., 2000; Stoltzfus et al., 1997; Gyaneshwar et al., 2001). The possibility that all or some of these bacteria could be contributing to the N-balance of wetland rice has been given credence by recent 15N dilution studies, both with uninoculated (Boddey et al., 1995; Wu et al., 1995; Shrestha & Ladha, 1996; Malarvizhi & Ladha, 1999) and inoculated (Baldani et al., 2000; Mehnaz et al., 2001; van Nieuwenhove et al., 2001) plants. This associative N fixation is, however, highly variable, ranging from 0 to 36% of total N derived from air, depending on the rice variety (Shrestha & Ladha, 1996; Malarvizhi & Ladha, 1999).

The reasons for the high variability in biological N fixation (BNF) associated with grasses are not well understood. It has been suggested that associative N2 fixation, unlike symbiotic N fixation in legumes, is more likely to be affected by genotype and environment (G × E) interactions, since the diazotrophs are only loosely associated with the plant and thus are more vulnerable to changes in the environment (Roger & Ladha, 1992; Malarvizhi & Ladha, 1999). Another possibility is that the low levels of N fixed by both associative and endophytic diazotrophic bacteria result from a limitation of C and energy in the rhizosphere and/or inside the plant. This proposition is supported by several observations. Egener et al. (1999) showed, in rice, that nifH expression by Azoarcus was enhanced in the presence of 5 mg l–1 malate in the assay medium, and rice seedlings inoculated with Azorhizobium caulinodans showed acetylene reduction activity only in the presence of externally supplemented C (van Nieuwenhove et al., 2000). Similarly, studies with wheat and maize inoculated with Azospirillum brasilense (Vande Broek et al., 1993) and Klebsiella pneumoniae (Chelius & Triplett, 2000), respectively, have shown that expression of nifH and/or nitrogenase Fe protein is dependent on additional C being added to the rooting medium.

Using acetylene reduction and 15N dilution assays, Christiansen-Weniger et al. (1992) demonstrated that aluminium (Al) tolerant wheat varieties had significantly higher nitrogenase activity as compared with Al-sensitive varieties under axenic conditions and when inoculated with A. brasilense. As tolerance to Al-toxicity in many plants is linked with the enhanced biosynthesis and secretion of organic acids by roots (Pellet et al., 1995; Ma et al., 2000), the study of Christiansen-Weniger et al. (1992) supports the suggestion that grass-associated N fixation could be increased by enhancing the availability of C to the associated and/or endophytic bacteria. In consideration of this, the present study was aimed at determining if availability of C limits N fixation in rice under both axenic and glasshouse conditions, and how this may be overcome by using varieties that are Al-tolerant. Each variety was inoculated with a gusA-marked strain of Herbaspirillum seropedicae Z67 (Barraquio et al., 1997), a diazotroph that was originally isolated from the rhizosphere of rice by Baldani et al. (1986), and which has since been shown to be both an aggressive colonizer of its interior (James et al., 2000; James, 2000) and to promote its growth (Baldani et al., 2000).

Materials and Methods

Inoculation of rice seedlings with Herbaspirillum seropedicae Z67-gusA

Herbaspirillum seropedicae Z67 marked with a transposon based gusA (Barraquio et al., 1997) was maintained on JNFb medium (Olivares et al., 1996) containing spectinomycin (100 µg ml–1) and naladixic acid (10 µg ml–1). Resistance to spectinomycin is encoded by transposon mTn5ssgusA21 (Wilson et al., 1995) and resistance to naladixic acid is a natural property of this bacterium. The herbaspirilla were grown in Luria broth (LB) (10 g−1 l tryptone, 5 g−1 l yeast extract, 10 g−1 l NaCl supplemented with spectinomycin (100 µg ml–1) until they reached an optical density of 0.6. Cells were then harvested by centrifugation (5000 g, 5 min), washed twice with normal saline (0.9% w : v), and resuspended in N-free Fahraeus medium (FM) (Fahraeus, 1957) for use as an inoculum.

Five rice varieties (‘Moroberekan’, ‘IRAT104’, ‘Azucena’, ‘IR43’ and ‘IR45’) differing in Al-tolerance were selected for axenic experiments: ‘Moroberekan’, ‘IRAT104’ and ‘Azucena’ are considered to be Al-tolerant; ‘IR43’ and ‘IR45’ are Al-intolerant (Khatiwada et al., 1996). Dehulled seeds were surface-sterilized in 70% ethanol for 5 min followed by 0.2% mercuric chloride for 30 s, and were washed three times with sterile water. Seeds were germinated on 0.1% (w : v) tryptic soy agar plates, and uncontaminated seedlings transferred to 80 ml glass tubes with 20 ml N-free Fahreus medium and inoculated with 1 ml of suspension containing c. 107 herbaspirilla. Seedlings were placed in a growth chamber for 30 d after inoculation (DAI), but at 7 DAI the plastic top was removed from the tubes and replaced with laboratory film containing a small hole allowing shoot emergence. The plants were maintained in a growth chamber (14 h light/10 h dark cycle, irradiance level 50 µmol m–2 s–1, day/night temperatures 27°C/25°C). Uninoculated plants served as controls.

For the glasshouse experiments, 5-d-old seedlings of cvs. Moroberekan and IR45 were treated with a suspension of herbaspirilla cells containing 108 cfu ml–1. After coating for 15 min the seeds were allowed to germinate for 5 d before transplanting into soil amended with or without 15N-labelled urea. At this stage the seedlings had approximately 106 cfu g–1 d. wt of herbaspirilla. The glasshouse experiments were conducted during the dry season, between February and May.

Plant growth-promoting activity of Herbaspirillum seropedicae Z67-gusA

Plant growth promotion by H. seropedicae Z67 under both axenic and glasshouse conditions was determined by comparing the dry weights, total N and total C of the inoculated plants with uninoculated control plants. All plants were harvested at 30 d after inoculation (DAI) or transplantation (DAT), except for the final glasshouse experiment which was harvested at maturity (120 DAT). The roots and aerial parts (and grain in the long-term experiment) were dried to a constant weight in an oven and ground to a fine powder for estimation of N and C content using a Perkin-Elmer 2400 CHN analyser (Perkin-Elmer, Norwalk, CT, USA) (Jimenez & Ladha, 1993).

Collection and analysis of root exudates

In a separate experiment, seedlings germinated from surface-sterilized seeds were transferred to glass tubes containing N-free liquid Fahreus medium and were allowed to grow for 10 d in a growth chamber with a 14 h light/10 h dark cycle and day/night temperatures of 27°C/25°C, with an irradiance of 50 µmol m–2 s–1. The C content of the root exudates was measured by collecting the exudates in the rooting medium, which was then freeze-dried and analysed using a Perkin-Elmer 2400 CHN analyser (Jimenez & Ladha, 1993).

Nitrogenase (acetylene reduction) activity

The acetylene reduction activity of inoculated plants was determined according to Ladha et al. (1986). Ten seedlings from each axenic treatment were taken at 10 DAI and washed twice with sterile distilled water to remove loosely associated bacteria. The seedlings were then transferred to fresh, N-free, liquid Fahreus medium. With the Al-sensitive varieties the Fahreus medium of an additional set of seedlings was supplemented with 10 mm sodium malate as a C source. The tubes containing the plants were sealed with a rubber seal and 10% of the headspace volume was replaced with acetylene. They were then returned to the growth chamber and incubated in the dark for 12 h at 30°C, after which any ethylene produced was determined using a Hitachi 164F gas chromatograph (Hitachi Instruments Service Co, Tokyo, Japan). Uninoculated plants and tubes not injected with acetylene served as controls. Glasshouse-grown plants were removed from the soil, washed with tap water to remove the soil, and transferred to glass tubes within which their acetylene reduction activity was determined as above.

Incorporation of 15N2 into inoculated rice seedlings

In a separate experiment, seedlings of the varieties ‘Moroberekan’ and ‘IR45’ derived from surface-sterilized seeds were inoculated with H. seropedicae Z67-gusA in 80 ml glass tubes and allowed to grow for 7 d. The tubes were then sealed with Suba seals and 5% of the headspace volume of half the number of tubes was replaced with 15N2 (99.5%, Monsanto Research Corp, Miamisburg, OH, USA), while the other set of tubes contained normal air. Tubes were returned to the growth chamber and after 3 d of incubation the plants from both sets of tubes were harvested and dried at 70°C to a constant weight in an oven before being ground to a fine powder that was analysed for its 15N content with a mass spectrometer (VG-Model 903) equipped with a Dumas elemental analyser (Roboprep-CN 7001, Europa Scientific Ltd, Crewe, UK). Uninoculated seedlings served as controls. The total amount of 15N fixed was calculated according to Nayak et al. (1986) as (15N atom percentage excess of sample)/(15N atom per cent excess of gas) × total N, and the percentage N derived from air (%Ndfa) was calculated as (15N atom per cent excess (tissue) – 15N atom percentage excess (background))/(15N atom per cent excess (atmosphere) –15N atom per cent excess (background)) × 100, where 15N atom per cent (background) = 0.010 ± 0.006 and 15N atom per cent excess of atmosphere = 4.9.

Measurement of N fixation under glasshouse conditions using soil amended with 15N-labelled urea

The 15N-labelled soil in the glasshouse experiments was the Bulacan soil used previously in the studies of Shrestha & Ladha (1996) and Malarvizhi & Ladha (1999). It had originally been amended with 15N-labelled urea that had been stabilized under flooded conditions for 7 wk before planting (Malarvizhi & Ladha, 1999). The soil had the following characteristics: pH 6.3, organic C = 1.3 g kg–1, Kjeldhal N = 0.11 g kg–1, available Olsen P = 33 mg kg–1, 15N atom per cent excess = 0.158. Twelve kilograms of the wet soil was placed into each plastic tray and supplemented with 250 mg P and 230 mg K as KH2PO4 before transplanting four 7-d-old rice seedlings, with or without inoculation with H. seropedicae Z67-gusA. The plants were sampled at 30 DAT (10 plants per treatment), washed with tap water and dried in an oven to a constant weight. The roots and shoots were separated and ground to a fine powder, which was then used for analysis of total N and 15N content as described earlier. The percentage Ndfa was calculated according to Shrestha & Ladha (1996) as (15N atom percentage excess test genotype)/(15N atom percentage excess of genotype with highest 15N enrichment) × 100.

Long-term glasshouse experiment

Seedlings of cvs Moroberekan and IR45 were inoculated with either live or heat-killed H. seropedicae Z67 and grown in a glasshouse as described previously, except that the Maahas soil used in this experiment was not amended with 15N-labelled urea.

Enumeration of bacteria

Axenically grown plants were sampled at 10 DAI. Loosely attached bacteria were removed by washing the roots in sterile water. The roots were then immersed in 5 ml sterile distilled water and vortexed for 30 s. The resulting solution was serially diluted and plated on JNFb (without bromothymol blue) agar plates containing spectinomycin (100 µg ml–1), naladixic acid (10 µg ml–1) and 5-bromo-4-chloro-3-indolyl-β-glucoronide (X-gluc) (40 µg ml–1). Counts of bacterial colonies were then determined, and these counts were assumed to be those bacteria that were closely associated with the root surface. In another set, the roots were surface-sterilized by immersion in 95% ethanol for 5 min, followed by treatment with 3% calcium hypochlorite containing 0.1% sodium dodecyl sulphate (SDS) for 1 min. After three washes with sterile distilled water, followed by maceration in saline, the homogenate was serially diluted and plated on JNFb agar as described above.

In the case of the glasshouse experiment, plants were sampled at 0, 10 and 30 DAT and examined for the presence of bacteria on and within the roots and aerial parts as for the axenically grown plants. To enumerate the total population of heterotrophic bacteria in washed, but not surface-sterilized plants, the solution was plated onto Luria agar plates.

Identification of bacteria isolated from glasshouse-grown plants

Bacteria isolated from roots and aerial parts at 10 and 30 DAT were analysed by fingerprinting using BOX-polymerase chain reaction (PCR) amplification fragment length polymorphism as described by Verslovic et al. (1994). The BOX A1R primer (5′-CTACGGCAAGGCGACGCTGACG-3′) was used at 50 pmol with 100 ng template DNA in a 25-µl PCR reaction mixture containing 1.25 mm of each dNTP, 2 U AmpliTaq DNA polymerase (Perkin Elmer) in a reaction buffer with 10% dimethyl sulphoxide (DMSO) (v/v). The reaction buffer stock (5×) contained 83 mm ammonium acetate, 335 mm Tris-HCl, 33.5 mm MgCl2, 33.5 µm EDTA, 150 mmβ-mercaptoethanol, 850 µg ml–1 bovine serum albumin (BSA), pH 8.8. Amplification for PCR was performed in a BIOMETRA Uno-Thermocycler (Biometra, Göttingen, Germany) with an initial denaturation (95°C, 7 min) followed by 30 cycles of denaturation (95°C, 30 s); annealing (52°C, 1 min) and extension (65°C, 8 min), with a single final extension step (65°C, 16 min). After the reaction, the samples were separated on 1.5% agarose gels, stained with ethidium bromide and visualized on a UV transilluminator.

Microscopy

At least three seedlings from three independent inoculations were collected at 10 DAI, washed with sterile distilled water and stained for GUS activity in 50 mm potassium phosphate buffer (pH 7.0) containing 400 µg ml–1 X-gluc for 4–6 h (Jefferson et al., 1987). After staining, the leaves were cleared by immersion in 90% ethanol for 15 min. For glasshouse grown plants, the material was sampled at both 10 and 30 DAT, and then washed and incubated in the same assay system as for axenic plants, but in this case it was supplemented with chloramphenicol (200 µg ml–1) to inhibit the induction of β-glucuronidase by native plant-associated bacteria (Wilson et al., 1995). With both laboratory and glasshouse materials, roots and shoots showing blue colour were cut into small pieces (1–2 mm) and fixed in 4% glutaraldehyde (in 50 mm phosphate buffer, pH 7.0, containing 0.1% (v : v) Triton-X-100 (Sigma Chemical Co., St. Louis, MO, USA). The samples in fixative were then immediately placed under vacuum for 30 min, followed by overnight storage at atmospheric pressure. Inoculated and uninoculated roots were then prepared for examination on the cold stage of a Hitachi 4700 field emission-scanning electron microscope (FE-SEM). This involved plunging them into liquid N2 slush, etching at –95°C for 15 min to remove surface water, and then sputter-coating with 5 nm of gold-palladium in an Oxford Alto cryo-preparation chamber (Gatan, Oxford, UK). The samples were then examined at –130°C at an accelerating voltage of 5 kV. Further samples were incubated for 2 h in a polyclonal antibody raised against H. seropedicae Z67 (James et al., 1997) diluted 1 : 400 in IGL buffer (James et al., 1994), followed by 1 h in a 1 : 50 dilution (in immunogold labelling (IGL) buffer) of 15 nm gold particles conjugated to goat antirabbit antibodies (Amersham, Aylesbury, UK). The gold labelled roots were then dehydrated in an ethanol–acetone series, and critical point dried in a Bal-Tec CPD 030 critical point dryer (Balzers, Fürstentum, Liechtenstein). Before being viewed under the FE-SEM, the samples were coated with 2 nm of chromium in a Cressington 208 sputter coater (Cressington Scientific Instruments Ltd, Watford, UK), followed by 5 nm of carbon in an Agar Turbo carbon coater. The samples were viewed in the FE-SEM at an accelerating voltage of 30 kV, with or without a yttrium aluminium garnet (YAG) back scattered electron (BSE) detector inserted.

Other X-gluc-stained samples were prepared for light and transmission electron microscopy (TEM) according to James et al. (1997). The identity of the bacteria in the sections was confirmed by immunogold labelling (IGL) with a polyclonal antibody raised against H. seropedicae Z67. The cross-reaction of the antibody against various bacteria commonly isolated from the rhizosphere/interior of rice was determined by enzyme-linked immunosorbent assay (ELISA), and no cross-reactions with bacteria other than H. seropedicae Z67 were observed at any dilution. To detect in situ expression of nitrogenase, an antibody raised against the Fe (nifH)-protein of the nitrogenase enzyme complex of Rhodospirillum rubrum (a gift from Dr P. Ludden, Madison, WI, USA) was also used for IGL (diluted 1 : 100 in IGL buffer). Controls for IGL with the anti-H. seropedicae or anti-nitrogenase Fe-protein antibodies were: omission of the primary antibody; and substitution of the primary antibody with non-immune serum, diluted appropriately in IGL buffer (1 : 400 or 1 : 100).

Data analysis

The data were compared by analysis of variance or t-tests.

Results

Effect of inoculation on growth and N fixation of rice seedlings under axenic conditions

Inoculation with H. seropedicae Z67 resulted in enhancement of the d. wt of all the Al-tolerant varieties, ranging from 38 to 54% for roots and from 22 to 50% for shoots (Table 1). These d. wt increases were accompanied by significant enhancements in total C (35–50% for roots and 13–35% for shoots) and total N (29–61% for roots, 37–85% for shoots). By contrast, inoculation had no significant effect on the growth, total C and total N contents of the Al-sensitive varieties, IR43 and IR45 (Table 1).

Table 1.  Effect of inoculation with Herbaspirillum seropedicae Z67-gusA on growth, C and N contents of five rice varieties grown under axenic conditions for 30 d
VarietyInoculationD. wt (mg per plant)Total C (mg per plant)Total N (mg per plant)
RootShootRootShootRootShoot
  1. *, **For each variety, values for inoculated plants are significantly different from controls at P = 0.05 and P = 0.01, respectively. Values are means of two experiments with 10 replicates. Standard deviations are in parentheses.

‘Moroberekan’2.766.911.123.510.100.16
  (0.37)(1.24)(0.16)(0.28)(0.03)(0.08)
 +4.18**10.37**1.69*4.37*0.14**0.29**
  (0.62)(1.45)(0.14)(0.32)(0.03)(0.10)
Increase (%) 51.450.050.424.545.985.4
‘IRAT104’5.8812.652.174.610.070.36
  (0.95)(1.05)(0.28)(0.20)(0.02)(0.03)
 +9.04**15.47**3.16*5.20*0.13**0.49*
  (1.26)(1.22)(0.25)(0.24)(0.04)(0.06)
Increase (%) 53.722.345.212.761.237.0
‘Azucena’4.049.171.603.620.050.24
  (0.52)(1.22)(0.16)(0.35)(0.01)(0.03)
 +5.56*12.94*2.16*4.87*0.07*0.36**
  (0.46)(1.45)(0.28)(0.41)(0.01)(0.04)
Increase (%) 37.641.134.735.028.848.3
‘IR43’3.0011.230.882.240.040.27
  (0.68)(1.14)(0.18)(0.28)(0.01)(0.05)
 +4.0411.891.042.570.050.28
  (0.52)(1.06)(0.26)(0.24)(0.01)(0.02)
‘IR45’3.8011.850.651.920.030.26
  (0.55)(1.12)(0.15)(0.32)(0.01)(0.05)
 +4.0512.230.922.110.050.30
  (0.61)(1.28)(0.22)(0.47)(0.01)(0.06)

The Al-tolerant varieties ‘Moroberekan’, ‘IRAT104’ and ‘Azucena’ showed substantial acetylene reduction activity when inoculated with H. seropedicae Z67, ranging from 0.8 to 1.3 µmol C2H4 g–1 d. wt h–1 (Table 2). By contrast, no activity was detected with the Al-sensitive varieties, IR43 and IR45, except when carbon (10 mm sodium malate) was supplemented in the growth medium (Table 2). Addition of carbon to the Al-tolerant varieties did not significantly affect their nitrogenase activity (Table 2). With all the varieties, both Al-tolerant and sensitive, activity was abolished after surface-sterilization.

Table 2.  Nitrogenase (acetylene reduction) activity of five rice varieties after inoculation with Herbaspirillum seropedicae Z67-gusA under axenic conditions
VarietyTolerance to AlNitrogenase activity (µmol C2H4 g–1 d. wt h–1)1
UninoculatedInoculatedInoculated + 10 mm sodium malateInoculated and surface sterilized
  1. Values are means of two separate experiments with 10 replicates. Standard deviations are in parentheses. 1Assays were performed at 10 d after inoculation with Herbaspirillum seropedicae Z67-gusA.

‘Moroberekan’TolerantNil1.28 (0.45)1.35 (0.65)Nil
‘IRAT104’TolerantNil0.83 (0.50)1.23 (0.55)Nil
‘Azucena’TolerantNil1.04 (0.62)1.24 (0.48)Nil
‘IR43’SensitiveNilNil0.91 (0.35)Nil
‘IR45’SensitiveNilNil0.79 (0.42)Nil

The Al-tolerant variety ‘Moroberekan’ showed significantly higher incorporation of 15N2 compared with the Al-sensitive variety ‘IR45’ when inoculated with H. seropedicae Z67 (Table 3). The total N fixed was also significantly higher in ‘Moroberekan’ than in ‘IR45’. Moreover, upon inoculation, the percentage N derived from air was only 0.42 for ‘IR45’, whereas ‘Moroberekan’ could derive 4.66% of its N from air (Table 3).

Table 3.  Effect of inoculation with Herbaspirillum seropedicae Z67-gusA on incorporation of 15N2 into two rice varieties grown under axenic conditions
VarietyInoculation15N atom % excess1Total 15N fixed (µg per plant)% N derived from air
  • *

    For each variety, values for inoculated plants are significantly different from controls at P = 0.05.

  • 1

    Assays were performed at 10 d after inoculation with Herbaspirillum seropedicae Z67-gusA and 3 d after injection with 15N2. Values are means of two separate experiments with three replicates each.

‘Moroberekan’0.028 1.4060.038
‘Moroberekan’+0.129*15.993*4.659*
‘IR45’0.006 1.0330.041
‘IR45’+0.059* 3.585*0.418*
LSD (5%) 0.009 0.9350.360

Enumeration and localization of Herbaspirillum seropedicae Z67-gusA under axenic conditions

There were no significant differences between Al-tolerant and Al-sensitive varieties in the number of herbaspirilla that could be isolated from either the surface or from the internal tissues of inoculated plants (Table 4). In all cases, the bacterial numbers on the root surface were approximately log 5 cfu g–1 d. wt whereas only log 2–3 cfu g–1 d. wt could be isolated from the surface of the shoots. In addition to the surface bacteria, log 3 cfu g–1 and log 1 cfu g–1 d. wt of herbaspirilla could be re-isolated from surface-sterilized roots and shoots, respectively. No bacteria could be isolated from the uninoculated control plants under the test conditions.

Table 4.  Enumeration of Herbaspirillum seropedicae Z67-gusA colonizing five rice varieties grown under axenic conditions
Rice varietyTotal number of bacteria (log cfu g–1 d. wt)
Root surfaceRoot internalShoot surfaceShoot internal
  1. Values are means of two separate experiments with three replicates. Counts were performed at 10 d after inoculation with Herbaspirillum seropedicae Z67-gusA.

‘Moroberekan’5.43.23.01.5
‘IRAT104’5.23.12.81.8
‘Azucena’5.03.42.51.9
‘IR43’5.63.02.81.2
‘IR45’5.03.12.51.3
Standard deviation1.20.60.91.0

The physiological status of the inoculated bacteria associated with the seedlings was determined at 10 DAI by staining for glucuronidase (GUS) activity. Roots of Al-tolerant varieties inoculated with H. seropedicae Z67-gusA showed more a intense blue colour of GUS than the roots of corresponding Al-sensitive varieties (Fig. 1a,b). This was also true under nonsterile conditions (Fig. 1c,d; data not shown). Uninoculated control plants showed no blue coloration (not shown). Ultrastructural studies with material frozen for cryo-scanning electron microscopy showed that the deeply stained roots of cv. Moroberekan had a much denser surface population than the lighter-stained roots of cv. IR45 (Fig. 2a,b); the identity of the bacteria viewed under the SEM being confirmed by immunogold labelling with an antibody raised against H. seropedicae Z67 (Fig. 2c,d). The difference between the varieties in terms of surface bacterial populations was further confirmed when GUS-stained roots were sectioned for light microscopy and transmission electron microscopy (Fig. 3a–d), with the roots of cv. Moroberekan having a dense coating of bacterial colonies and those of cv. IR45 having relatively few (Fig. 3b). Interestingly, only a few bacteria were observed within the root interior of either variety (Fig. 3a–c). In cv. Moroberekan, single bacteria were occasionally seen within exodermal cells (Fig. 3a,c), whereas cv. IR45 tended to have slightly more internal bacteria, and these were located mainly within cortical/aerenchyma cells (Fig. 3b).

Figure 1.

β-Glucuronidase (GUS) staining of rice seedlings at 10 d after inoculation with Herbaspirillum seropedicae Z67-gusA. (a) Intense GUS activity (arrows) on roots of an axenically grown seedling of cv. Moroberekan. Bar, 2 mm. (b) Rather less intense GUS activity (arrows) on roots of an axenically grown seedling of cv. IR45. Bar, 2 mm. (c) Root of a glasshouse grown plant of cv. Moroberekan with an emerging lateral root, which is stained deep blue (large arrow). Note the blue patches on the main root (small arrows). Bar, 3 mm. (d) Surface of a leaf on a glasshouse grown plant of cv. Moroberekan showing patches of blue coloration (arrows). Bar, 2 mm.

Figure 2.

Scanning electron microscope (SEM) images of axenically grown rice seedlings 10 d after inoculation with Herbaspirillum seropedicae Z67-gusA. (a) Cryo-SEM image of a frozen hydrated root of cv. Moroberekan covered in bacteria (asterisk). Bar, 10 µm. (b) Cryo-SEM image of a frozen hydrated root of cv. IR45. Note that there are considerably less bacteria than on cv. Moroberekan, and they are mainly located in small clumps (arrow). Bar, 10 µm. (c) Backscattered electron SEM image of a bacterium on a root of cv. Moroberekan. The bacterium was immunogold labelled with an antibody raised against H. seropedicae-Z67 followed by 15 nm goat anti-rabbit gold. The labelling can be seen as white spots on the surface of the bacterium (arrow). Bar, 100 nm. (d) Secondary electron image of the same bacterium, and an arrow indicates the same region indicated in (d) Bar, 100 nm.

Figure 3.

Transverse sections of primary roots of axenically grown rice seedlings at 10 d after inoculation by Herbaspirillum seropedicae Z67-gusA. (a) Light micrograph of bacteria covering the surface of cv. Moroberekan (large arrow). A few bacteria can also be seen within an exodermal cell (small arrow). A, aerenchyma; Ep, epidermis; Ex, exodermis. Bar, 5 µm. (b) Light micrograph showing considerably fewer bacteria (small arrow) on the surface of cv. IR45, but note that some bacteria can be seen within cortical (aerenchyma) cells (large arrow). A, aerenchyma; Ep, epidermis; Ex, exodermis. Bar, 5 µm. (c) Transmission electron micrograph of bacteria on the surface of cv. Moroberekan (large arrows). There is also a bacterium within an exodermal cell (small arrow). Note that the surface bacteria are surrounded by material (asterisk). Ep, epidermis; Ex, exodermis. Bar, 2 µm. (d) Higher magnification transmission electron micrograph of a bacterial colony on the surface of cv. Moroberekan. This section was immunogold labelled with an antibody raised against H.seropedicae Z67 followed by 15 nm goat anti-rabbit gold. The bacterial surfaces are gold labelled (arrows), as is the material that surrounds some of them (asterisk). Bar, 200 nm. (e) Herbaspirillum cells on the surface of cv. Moroberekan. This section was immunogold labelled with an antibody raised against the Fe-protein of nitrogenase followed by 15 nm goat anti-rabbit gold. Most of the labelling is confined to electron-dense regions within the bacteria (arrows). Bar, 100 nm. (f) Serial section to (e). This section was treated as for (e), except that the anti-nitrogenase antibody was replaced by non-immune serum. Note that there is no immunogold labelling on the electron-dense regions in the bacteria (arrows). Bar, 100 nm.

As with the bacteria seen under the SEM (Fig. 2c), under the TEM, those on the root surfaces of cv. Moroberekan (Fig. 3a,c) were immunogold labelled with an antibody raised against H. seropedicae Z67. Most of the gold labelling was located on the bacterial surfaces and on electron-dense material that had apparently been released by the bacteria (Fig. 3d). The herbaspirilla on the surfaces of cv. Moroberekan roots also labelled strongly with an antibody raised against the Fe-protein of nitrogenase (Fig. 3e), with most of the labelling being located on electron-dense regions within the bacteria. No such labelling was observed on the few bacteria colonizing the root surfaces of cv. IR45 (not shown), or on any of the internal bacteria within either variety (not shown). Bacteria in sections incubated in non-immune serum substituted for the nitrogenase Fe-protein antibody showed no immunogold labelling (Fig. 3f). No bacteria were seen on or in uninoculated, control plants (not shown).

Herbaspirilla were also localized within the aerial parts of both varieties (Fig. 4a–f). In the case of cv. Moroberekan, although the bacteria were mostly observed in fairly large numbers within the stem and leaf aerenchyma (Fig. 4a), the upper parts of the stems of most of the plants had regions that were particularly heavily infected with the bacteria (Fig. 4b,c). This dense colonization appeared to have caused some localized damage to the surrounding plant tissue, causing small, but visible, GUS-stained lesions about 1–2 mm in length (not shown, but similar to Fig. 1d). In these cases, the bacteria had formed dense colonies within the central tissues and were especially localized within intercellular spaces and xylem vessels (Fig. 4b,c). The bacteria within the colonies were immunogold labelled with the anti-H. seropedicae antibody (Fig. 4d). The cells surrounding the heavily infected regions appeared to have shrunken or degraded cytoplasm. This level of infection was never observed within aerial parts of cv. IR45, in which the bacteria were usually discretely localized only within a few metaxylem vessels in the leaves (Fig. 4e,f). Low numbers of bacteria were also occasionally seen in intercellular spaces within the central regions of the stems of cv. IR45 (not shown).

Figure 4.

Transverse sections of aerial parts of axenically grown rice seedlings at 10 d after inoculation by Herbaspirillum seropedicae Z67-gusA. (a) Light micrograph of bacteria (small arrows) within the aerenchyma (A) of a leaf of cv. Moroberekan. Bar, 20 µm. (b) Light micrograph of the inner part of the stem of cv. Moroberekan at the site of a β-glucuronidase (GUS)-stained lesion. Much of the tissue is densely infected by bacteria, particularly the intercellular spaces, some of which have been greatly enlarged by the presence of the bacteria (large arrows). Note that a xylem vessel is also heavily colonized (small arrow). Bar, 20 µm. (c) Light micrograph of a heavily infected xylem vessel (X) within the upper stem of cv. Moroberekan. Note that the cytoplasm of the surrounding parenchyma cells (asterisk) is degraded and has pulled away from the cell walls. Bar, 5 µm. (d) Transmission electron micrograph of a bacterial colony within a xylem vessel in the stem of cv. Moroberekan. The bacteria were immunogold labelled with an antibody raised against H. seropedicae Z67 followed by 15 nm goat anti-rabbit gold (small arrows). Bar, 500 nm. (e) Light micrograph of a leaf of cv. IR45. Note that one of the metaxylem vessels is completely full with bacteria (arrow), but no bacteria can be seen within the nearby aerenchyma (A). Bar, 20 µm. (f) Transmission electron micrograph of Fig. 3e. The bacteria in the xylem vessel (large asterisk) appear to be healthy, and they are not surrounded by host-derived material. However, the cytoplasm in some of the nearby xylem parenchyma cells (small asterisk) is showing signs of degradation and it has shrunk away from the cell walls (cf. c). Bar, 1 µm.

Root exudates

The total C in the root exudates of the Al-tolerant varieties ‘Moroberekan’, ‘IRAT104’ and ‘Azucena’ was 38.4, 39.6 and 35.6 mg g–1 root d. wt, respectively, whereas for the Al-sensitive varieties ‘IR43’ and ‘IR45’ it was approximately half, being only 18.6 and 19.0 mg g–1 root d. wt, respectively. These differences were significant at the 5% level.

Effect of inoculation on N fixation and growth of rice seedlings under glasshouse conditions

Substantial acetylene reduction activity could be detected in plants of both uninoculated and inoculated cv. Moroberekan at 10 DAT, and this had increased nearly threefold by 30 DAT (Table 5). By contrast, no activity could be detected in the uninoculated Al-sensitive variety ‘IR45’ at 10 DAT and, even though it showed some activity at 30 DAT, it was still significantly lower than that of cv. Moroberekan. Inoculation with H. seropedicae Z67-gusA had no significant effect on the nitrogenase activity of either variety (Table 5).

Table 5.  Effect of inoculation with Herbaspirillum seropedicae Z67-gusA on the nitrogenase (acetylene reduction) activity of rice cvs ‘Moroberekan’ and ‘IR45’ grown in a glasshouse (under nonsterile conditions). Assays were performed at 10 d and 30 d after transplantation (DAT)
 Nitrogenase activity (µmol C2H4 g–1 d. wt h–1)
 ‘Moroberekan’‘IR45’
  • *

    The 30 DAT values are significantly different from 10 DAT values at P = 0.05. Values are means of three replicates.

Inoculation++
10 DAT0.450.62NilNil
30 DAT1.22*1.45*0.350.40
LSD (5%)0.280.350.260.25

Inoculated plants of the Al-tolerant variety ‘Moroberekan’ had significantly higher shoot and root d.wts and C contents, and higher root N contents than the inoculated Al-sensitive variety ‘IR45’ (Table 6). ‘Moroberekan’ also showed significant 15N dilution in both roots and aerial parts (Table 6), regardless of inoculation, and calculations using ‘IR45’ as a ‘nonfixing’ reference showed that ‘Moroberekan’ had approximately 27% more N derived from air than ‘IR45’. Inoculation with H. seropedicae Z67-gusA resulted in significant increases in d. wt, total C and total N content in both the varieties, but had no significant effects on 15N dilution (Table 6).

Table 6.  Effect of inoculation with Herbaspirillum seropedicae Z67-gusA on growth, C and N contents, and concentration of 15N of Al-tolerant (‘Moroberekan’) and Al-sensitive (‘IR45’) rice varieties grown in a glasshouse (under nonsterile conditions) for 30 d in a soil enriched in 15N
VarietyInoculationRootShoot
D. wt (mg)C (mg)N (mg)15N atom % excessD. wt (mg)C (mg)N (mg)15N atom % excess
  • *

    For each variety, values for inoculated plants are significantly different from controls at P = 0.05. Values are per plant and means of 10 replicates.

‘Moroberekan’3621073.1140.102134252645.6100.095
 +519*160*5.709*0.0852008*795*72.390*0.087
Increase (%) 43.349.583.0 49.651.058.7 
‘IR45’191532.3600.138137944851.790.144
 +260*70*3.462*0.1421705*510*66.89*0.148
‘Increase’ (%) 363146 2313.828 
LSD (5%) 0.1160.0030.2360.0080.2730.0183.450.006

There were no significant differences between inoculated and control plants in the long-term experiment, except that inoculation increased the N-content of the straw of ‘Moroberekan’ (Table 7).

Table 7.  Effect of inoculation with Herbaspirillum seropedicae Z67-gusA on the dry weight, grain yield and N content of rice cvs Moroberekan and IR45 grown in a glasshouse (under nonsterile conditions) for 120 d
 ‘Moroberekan’‘IR45’
  • *

    For each variety, values for inoculated plants are significantly different from controls at P = 0.05. Values are per plant and means of 10 replicates.

Inoculation++
Root d. wt (g) 6.6 7.0 5.7 6.0
Straw d. wt (g)14.615.213.013.2
Grain d. wt (g) 6.8 7.1 7.7 8.0
Root N (mg g–1 d. wt)25.225.822.821.8
Straw N (mg g–1 d. wt)49.757.5*58.658.9

Survival and localization of H. seropedicae Z67-gusA under glasshouse conditions

The inoculated bacterial strain, H. seropedicae Z67-gusA, could be reisolated from both varieties until the end of the experiment (30 DAT). The identity of bacteria reisolated from various plant parts using JNFb media, which is semi-specific for Herbaspirillum spp. (Olivares et al., 1996), vis à vis the inoculated strain, was confirmed by BOX-PCR fingerprinting. All the reisolated strains examined showed identical fingerprints to H. seropedicae Z67, regardless from which part of the plants they were reisolated (Fig. 5). Counts of H. seropedicae Z67-gusA over this 30-d period showed that the population of the bacteria associated with the root surfaces declined substantially from log 6 cfu g–1 d. wt at the time of transplanting to log 1.7 cfu g–1 d. wt at 30 DAT. Similarly, the number of surface bacteria associated with the aerial parts also decreased from log 2.5 cfu g–1 to log 1 cfu g–1 d. wt over the same sampling period. In parallel with the surface populations, those within the root and shoots also showed a decline from log 3.5 cfu g–1 to log 1 cfu g–1 d. wt by 30 DAT. At 30 DAT no bacteria could be isolated from the uninoculated plants using media selective for Herbaspirillum spp., but the ‘total heterotrophic’ bacterial counts in non-surface-sterilized plants (as enumerated on LB plates) were similar to those from the inoculated plants (log 6.6 ± 1.7 cfu g–1 and log 3.6 ± 1.2 cfu g–1 d. wt for roots and aerial parts, respectively).

Figure 5.

The BOX-polymerase chain reaction fingerprints of Herbaspirillum seropedicae Z67-gusA reisolated from inoculated glasshouse-grown rice plants at 10 d and 30 d after transplantation (DAT). Lane 1 and 26, 1 kb DNA ladder profiles of the strains isolated from cv. IR45 (lanes 2–13) and cv. Moroberekan (lanes 14–25); lanes 2–4, strains isolated from root surfaces at 10 DAT; lanes 5–7, strains isolated from surface-sterilized roots at 10 DAT; lanes 8–10, strains isolated from stem surfaces at 10 DAT; lanes 11–13, strains isolated from surface-sterilized stems at 10 DAT; lanes 14–16, strains isolated from root surfaces at 30 DAT; lanes 17–19, strains isolated from surface-sterilized roots at 30 DAT; lanes 20–22, strains isolated from stem surfaces at 30 DAT; lanes 23–24, strains isolated from surface-sterilized stems at 30 DAT; lane 25, H. seropedicae Z67-gusA.

At 10 DAT, H.seropedicae Z67-gusA could be localized on both the roots and shoots of glasshouse-grown rice plants using GUS staining (Fig. 1c,d). Light microscopy of sections of roots (both inoculated and uninoculated) showed that there were large numbers of unidentified bacteria on the surfaces of both varieties (Fig. 6a,b), as well as some within the roots, particularly where there were wounds or breaks in the epidermis (Fig. 6b). Some, but not all of these bacteria were immunogold labelled with an antibody raised against H.seropedicae (not shown). A consistent feature of the roots of cv. IR45 were fungal hyphae within the cortical cells (Fig. 6c) and more detailed studies using transmission electron microscopy combined with immunogold labelling showed H.seropedicae associated with these (Fig. 6d). As suggested by the GUS staining and plate counts, at 10 DAI there was a substantial population of herbaspirilla on the leaf surfaces. Light microscopy confirmed that there were microorganisms on the surfaces and within the interior of leaves of both varieties (Fig. 7a,b), including on uninoculated plants (Fig. 7a). Leaves of inoculated plants of cv. Moroberekan, however, often also contained numerous bacteria and fungi within intercellular spaces and epidermal cells (Fig. 7b,c). Although most of the bacteria within the leaves did not react with the H.seropedicae-specific antibody (not shown), there were still some within cells and intercellular spaces that were immunogold labelled (Fig. 7d).

Figure 6.

Transverse sections of roots of glasshouse-grown rice seedlings. (a) Light micrograph of cv. Moroberekan at 10 d after inoculation (DAI) by Herbaspirillum seropedicae Z67-gusA. Bacteria are visible on the root surface (arrows), but few are visible within the root tissues. A, aerenchyma; S, stele. Bar, 20 µm. (b) Light micrograph of an uninoculated root of cv. Moroberekan with a wound/break in the epidermis (asterisk). Note the bacteria within the wound (large arrows) and on the root surface (small arrows). A, Aerenchyma, S, stele. Bar, 20 µm. (c) Light micrograph of cv. IR45 at 10 DAI by H. seropedicae Z67-gusA. Few bacteria are visible, but fungal hyphae can be seen within many of the cortical and aerenchyma cells (arrows). A, aerenchyma; S, stele. Bar, 20 µm. (d) Transmission electron micrograph of cortical cell in (c). This section was immunogold labelled with an antibody raised against H. seropedicae Z67 followed by 15 nm goat anti-rabbit gold. Note the immunogold labelled bacteria (arrows) adjacent to a fungal hypha (asterisk). W, cell wall. Bar, 200 nm.

Figure 7.

Sections of leaves of glasshouse-grown rice seedlings. These were taken either from uninoculated plants (a), or from plants at 10 d after inoculation by Herbaspirillum seropedicae Z67-gusA (b–d). (a) Light micrograph of a transverse section of an uninoculated leaf of cv. IR45 showing bacteria on the leaf surface (arrows). No bacteria are visible within the leaf tissues. A, aerenchyma. Bar, 20 µm. (b) Light micrograph of a longitudinal section of a leaf of cv. Moroberekan showing colonization by microorganisms within intercellular spaces (arrows) and in a mesophyll cell (asterisk). Bar, 5 µm. (c) Transmission electron micrograph of mesophyll cells in a leaf of cv. Moroberekan densely colonized by bacteria (small arrows) and fungi (large arrows). The bacteria did not label with an antibody raised against H. seropedicae. Bar, 1 µm. (d) Transmission electron micrograph of bacteria within a mesophyll cell in a leaf of cv. Moroberekan. The bacteria were immunogold labelled with an antibody raised against H. seropedicae followed by 15 nm goat anti-rabbit gold (arrows). W, cell wall. Bar, 200 nm.

Discussion

Nitrogen fixation by axenically grown rice seedlings

The Al-tolerant varieties showed significantly higher acetylene reduction activity compared with the semitolerant and Al-sensitive varieties when inoculated with H.seropedicae Z67-gusA (Table 1). These results are similar to earlier studies showing higher nitrogenase activities associated with Al-tolerant varieties of wheat (Christiansen-Weniger et al., 1992). The differences in activity observed here may not be due to differences in the number of bacteria associated with the seedlings, as similar numbers were isolated from all the varieties (Table 4). This suggests that the bacteria associated with the Al-tolerant varieties were physiologically more active than those associated with the sensitive varieties. This is further supported by the more intense GUS staining on roots of the Al-tolerant cv. Moroberekan compared with the Al-sensitive cv. IR45 (Fig. 1), and also that the colonies of Herbaspirillum on the roots of cv. Moroberekan expressed the Fe-protein of nitrogenase. However, the roots of cv. Moroberekan examined by cryo-scanning electron microscopy, light microscopy and transmission electron microscopy had considerably more bacteria colonizing their surfaces than those of cv. IR45. Although it is not easy to reconcile the apparent discrepancy between plate counts and direct microscopical observations, particularly the cryo-SEM (since it involved minimal disturbance to the samples), inaccuracies inherent in the plate-counting technique (Boddey et al., 2000) may have masked a greater population of herbaspirilla associated with roots of cv. Moroberekan compared with cv. IR45.

Aluminium tolerance can, in part, result from the ability of the roots to exude organic acids (Christiansen-Weniger et al., 1992; Ma et al., 2000). Analyses of root exudates showed that there were significant quantitative differences in the total amount of C secreted, with the Al-tolerant varieties (‘Moroberekan’, ‘IRAT104’ and ‘Azucena’) secreting almost double the amount of C than the Al-sensitive varieties (‘IR43’ and ‘IR45’). Although the nature of the exuded C sources was not determined in the present study, Kirk et al. (1999) have shown that cv. Azucena secretes large amounts of citric and malic acids. Therefore, since both citric and malic acids are suitable C sources for growth and N fixation by Herbaspirillum spp. (Baldani et al., 1992, 1996), this increased amount of C in the root exudate may have allowed for higher nitrogenase activity (and possibly more growth) by the associated bacteria, as has been previously shown for Al-tolerant wheat varieties inoculated with Azospirillum (Christiansen-Weniger et al., 1992). In addition to higher acetylene reduction activity, inoculated plants of cv. Moroberekan also showed significantly greater incorporation of 15N2 than cv. IR45 (Table 2).

Most of the N fixation shown by cv. Moroberekan will have come from bacteria on the surface, as shown by the abolition of acetylene reduction activity by surface-sterilization (Table 1), and also by the fact that most bacteria, particularly those expressing nitrogenase Fe-protein, were observed only on the root surfaces (Figs 2 and 3). However, there were some endophytic herbaspirilla, particularly within the aerial parts (Fig. 4), and these may also have contributed to the observed N fixation. However, their numbers were relatively low compared with those associated with the roots (Table 4) and this contribution will probably have been minor. Indeed, as many of the bacteria within the aerial parts were showing some degree of saprophytic/pathogenic behaviour, and did not express the nitrogenase Fe-protein, they may actually have been a net drain on the plant’s resources. The pattern of localization shown by H. seropedicae within the rice cv. Moroberekan was actually redolent of the disease-forming colonies of its close phytopathogenic relative, Herbaspirillum rubrisubalbicans, within sugarcane and sorghum leaves (James et al., 1997; Olivares et al., 1997; James & Olivares, 1998). Unlike H. rubrisubalbicans in these latter studies, however, the occurrence of large H. seropedicae colonies within the rice cv. Moroberekan in the present study did not result in any ‘macro’ symptoms of disease on the aerial parts, except for small blue ‘lesions’ on the leaves and stems after they were stained with GUS.

Despite the pathogenic colonization of some leaves under axenic conditions, inoculation of the Al-tolerant varieties with H. seropedicae Z67 gave generally positive effects, not observed with the Al-sensitive varieties (‘IR43’ and ‘IR45’), such as increases in d. wt, total N and total C in both roots and shoots (Table 3). The Al-tolerant varieties released substantial quantities of C from their roots (regardless of the presence of bacteria), and after inoculation with H. seropedicae fixed significant quantities of N, in parallel with significant growth promotion (Tables 1–3). These results show that rice plants that can provide additional C to diazotrophs may not necessarily result in a drain on their resources (Ladha & Reddy, 1995). Another possibility is that the growth increases shown by these varieties are due to a congruence between plant N demand and supply, resulting in an enhanced C sink and hence increased photosynthesis (as shown by the significant increases in C contents; Table 3).

Nitrogen fixation by glasshouse-grown rice seedlings

The results obtained under axenic conditions were further tested under glasshouse conditions using soil amended with 15N-labelled urea. ‘Moroberekan’ showed higher acetylene reduction activity, as well as greater dilution of 15N compared with the sensitive variety ‘IR45’ (Tables 5 and 6), and calculations using cv. IR45 as the ‘nonfixing’ reference showed that ‘Moroberekan’ was able to fix c. 27% more N. It is possible that these differences in the ability to support N fixation would become more pronounced in the later stages of growth, as rice generally shows maximum nitrogenase activity at the heading stage (Ladha et al., 1986). This is partly supported by the long-term experiment, which showed that the N content of the straw of inoculated cv. Moroberekan was significantly greater than controls at maturity (Table 7), although further studies are needed to determine if this was due to enhanced N fixation or to other factors.

Interestingly, and in contrast to the laboratory experiments, inoculation with H. seropedicae Z67 had no significant effect on N fixation over the 30-d period (Tables 5 and 6), thus indicating that the differences in the ability to support N fixation by these two varieties were due to the native populations of associative/endophytic diazotrophs and not to the inoculated bacteria. However, and in contrast to the lack of effects on N fixation, inoculation with H. seropedicae Z67 did result in enhancements of d. wt, total N and total C of both cvs Moroberekan and IR45 (Table 6), thus indicating that this bacterial strain can also enhance the growth of rice under nonsterile conditions. Similar results have been observed recently with a Brazilian upland rice cultivar, Guarani (Baldani et al., 2000). This variety, which has not yet been evaluated for Al-tolerance or C-exudation, gave relatively high values for percentage N derived from air after inoculation with H. seropedicae Z67 (32% under axenic conditions for 30 d, and 18.6% under glasshouse conditions for 130 d), and hence differed from the present study in suggesting that N fixation was strongly involved in the growth promotion by the bacteria, even under nonsterile conditions. This difference between the present study and that of Baldani et al. (2000), particularly with glasshouse-grown plants, may be due to the much longer growing period used in the latter study, as higher rates of nitrogenase activity are known to occur in the later growth stages of wetland rice (Ladha et al., 1986). Indeed, studies are currently being undertaken to determine if the higher straw N-content of mature inoculated cv. Moroberekan plants (Table 7) was due in any way to N fixation.

Since the growth promotion observed in the present study appears not to be due in any large part to N fixation (at least in the early stages up to 30 DAT), it must therefore involve other factors, such as the production of phytohormones, particularly indoleacetic acid (IAA). These are known to be released by many diazotrophic, plant growth-promoting rhizobacteria (PGPR) (Steenhoudt & Vanderleyden, 2000; Mehnaz et al., 2001), including H. seropedicae Z67 (Bastian et al., 1998).

Localization of plant-associated bacteria under nonsterile conditions

The gusA marker gene has been used with a variety of plant–microbe interactions, but mainly under sterile conditions (Van de Broek et al., 1993; Hurek et al., 1994; Saxena et al., 2000). With the exception of rhizobia (Wilson et al., 1995; Wilson et al., 1999), the present study with H. seropedicae Z67 is the first to use gusA to track and localize an associative diazotrophic bacterium under nonsterile conditions. However, the addition of chloramphenicol was essential in localizing the GUS activity of the inoculated bacteria, as it inhibits the induction of GUS activity by the indigenous bacteria (Wilson et al., 1995), which are abundant on and in glasshouse-grown plants. The GUS staining suggested that the herbaspirilla were localized on the surfaces of both roots and aerial parts, and this was confirmed using sections for light microscopy and TEM (Figs 6 and 7). The latter not only showed that the bacteria were present on root surfaces, but also that they were within leaf intercellular spaces and mesophyll cells. This pattern of localization was similar to that shown for the axenically grown plants, but the herbaspirilla were much more difficult to find within the glasshouse-grown plants, particularly as there were numerous unidentified microorganisms present (Figs 6 and 7). Therefore, immunogold labelling with an antibody specific to H. seropedicae was essential in recognizing these bacteria under the TEM. It was clear from both the microscopy studies and from plate counts that after only 10 d H. seropedicae, even though it had been inoculated, was not particularly abundant compared with the ‘background’ microflora. Certainly, there was no sign of the large ‘pathogenic’ colonies of herbaspirilla that were observed within the aerial parts of the axenically grown ‘Moroberekan’ plants (Fig. 3). The fact that the bacteria were difficult to find under the microscope, combined with the low population data (< 2 cfu g–1 d. wt at 30 DAT), suggests that the inoculated herbaspirilla were too few to have any significant effect on N fixation by either variety, and that the 15N dilution and acetylene reduction data shown in Tables 5 and 6, respectively, are most likely to be the result of activity from indigenous, uninoculated bacteria whose identities are so far unknown. This further supports the suggestion that growth in the present study was promoted by H. seropedicae through mechanisms other than N fixation.

Conclusion and further work

Inoculation with H. seropedicae under laboratory and glasshouse conditions resulted in significant growth promotion and N accumulation over the experimental periods (30 d and 120 d), and further studies are needed to determine more exactly the mechanisms underlying this. Although there was an apparent lack of N fixation by the inoculated bacteria (particularly under glasshouse conditions), the axenic study, as well as the study of Baldani et al. (2000), suggests it may still play a role in the interaction between rice and Herbaspirillum. Therefore, the breeding and selection of varieties for higher C exudation could lead to the identification of genotypes that would support higher BNF, and hence even greater growth promotion. Since quantitative trait loci (QTLs) for Al-tolerance in rice have been identified (Wu et al., 2000), it would be of interest to determine if these QTL are also linked with the ability for enhanced N fixation.

Acknowledgements

Doctor A. R. Prescott is thanked for use of electron microscope facilities at the University of Dundee, UK, and G. Baeta da Cruz (Embrapa-Agrobiologia, Seropédica, Rio de Janeiro, Brazil) for the H. seropedicae antibody. The Analytical Services Laboratory, IRRI, is thanked for the mass spectrometry and M. Alumaga, M. Gruber and M. Kierans for technical assistance. The financial assistance for this work was partly provided by grants from Gesellschaft fur Technishe Zusammenarbeit (GTZ), and partly by the Joint Infrastructure Fund (JIF) and the Carnegie Trust.

Ancillary