Lack of trehalose catabolism in Sinorhizobium species increases their nodulation competitiveness on certain host genotypes


  • Sequence data for the Sinorhizobium medicae thuB ortholog have been deposited in the GenBank (National Center for Biotechnology Information) database under accession number EF407855.

Author for correspondence:
T. V. Bhuvaneswari
Tel: +47 776 44433
Fax: +47 776 46333


  • • The role of host and bacterial genotypes in determining the competitiveness of trehalose utilization mutants of Sinorhizobium meliloti and Sinorhizobium medicae was investigated here.
  • • Trehalose utilization mutants of S. meliloti and S. medicae were obtained by mutagenesis of their trehalose utilization gene thuB. The mutant strains and the wild type were coinoculated on three cultivars of alfalfa (Medicago sativa) and two cultivars of Medicago truncatula and assessed for competitiveness in root colonization, and nodule occupancy.
  • • The thuB mutants formed more nodules than their parent strains on two of the three alfalfa lines tested and on one of the two M. truncatula lines tested. They were not more competitive on the other alfalfa and M. truncatula lines. Their competitiveness for nodule occupancy did not correlate positively with their ability to colonize these roots but correlated with the extent of thuB induction in the infection threads. Induction of thuB was shown to be dependent on the concentration of trehalose in the environment.
  • • These results suggest a direct role for host trehalose metabolism in early plant–symbiont interactions and show that the ability to manage host-induced stresses during infection, rather than the ability to colonize the root, is critical for competitive nodulation.


Sinorhizobium meliloti is a symbiont of both Medicago sativa (alfalfa) and the model legume Medicago truncatula. The symbiosis results in the formation of nodules wherein rhizobia convert atmospheric nitrogen to ammonia (Gage, 2004). Despite the final outcome being beneficial, early interactions between the plant and the microsymbiont bear similarities to plant–pathogen interactions (Djordjevic et al., 1987; Hirsch, 2004; Soto et al., 2006). As in pathogenesis, during infections, the bacterium is exposed to host-induced osmotic, oxidative and ethylene-related stresses (Oldroyd et al., 2001; Santos et al., 2001; Ampe et al., 2003; Wright & Beattie, 2004) which rhizobia suppress through their surface polysaccharides and through expression of catalases and superoxide dismutases during these interactions (Spaink, 2000). Sinorhizobium meliloti mutants with altered surface polysaccharides become defective in symbiosis (Leigh et al., 1985; Reuber et al., 1991; Campbell et al., 2002) and induce defense-related genes in M. truncatula (Jones et al., 2008).

In the legume–Rhizobium symbiosis, the determinants of interstrain competition for nodule occupancy are not fully understood. Adaptation to stress (Ma et al., 2004), the genotypes of the host and those of the competing rhizobium strains are known to influence the outcome (Cregan et al., 1989; Deoliveira & Graham, 1990; McDermott & Graham, 1990; Josephson et al., 1991; Vlassak & Vanderleyden, 1997). Biosynthesis and accumulation of trehalose modulate diverse stresses in bacteria (Miller & Wood, 1996; Benaroudj et al., 2001; Elbein et al., 2003; Makihara et al., 2005) and influence the virulence of some plant and animal pathogens (Van Dijck et al., 2002; Astua-Monge et al., 2005; Murphy et al., 2005). Trehalose is available in the root environment (Phillips & Streit, 1997; Jensen et al., 2005) and plays a role in stress adaptation in S. meliloti and in its interactions with alfalfa (Gouffi et al., 1999; Jensen et al., 2005). Sinorhizobium meliloti possesses one major pathway for trehalose utilization which is encoded by two genes, trehalose utilization genes A and B (thuA and thuB) (Jensen et al., 2002). The predicted biochemical functions for thuA and thuB are glutamine amido transferase and NAD(H) dependent dehydrogenase, respectively. Our preliminary results on functional analysis are consistent with these predictions (results will be presented elsewhere). Strains mutated in genes thuA (Sm7023) and/or thuB (Sm7024) can transport trehalose but fail to assimilate it and hence accumulate it in their cells (Jensen et al., 2005). When competed with the wild-type Rm1021, these mutants were less competitive in colonizing M. sativa cv. Moapa 69, but they formed a higher proportion of the infections and nodules than the wild type (Jensen et al., 2002, 2005).

Competitiveness in nodule formation is an outcome of host–symbiont interactions. Therefore, to elucidate the contribution of the host and symbiont genotypes in this outcome, we have examined the phenotype of Sm7024 on two other genotypes of M. sativa and M. truncatula. We have also mutated the ortholog of thuB in Sinorhizobium medicae, a relative of S. meliloti (Rome et al., 1996a,b), and assessed the competitiveness of this thuB mutant (Smd1306) on the same host genotypes.

Materials and Methods

Bacterial strains, plasmids and growth media

The bacterial strains and plasmids used in this work are listed in Table 1. Rhizobium strains were grown at 28°C on tryptone yeast (TY) medium or M9 minimal medium with various carbon sources at 0.4% (weight/volume (w/v)). Escherichia coli was grown at 37°C on Luria broth (LB). Selective media were supplemented with antibiotics to the final concentration as stated: streptomycin, 500 µg ml−1; kanamycin, 100 µg ml−1; chloramphenicol, 30 µg ml−1; tetracycline, 5 µg ml−1.

Table 1.  Bacterial strains and plasmids used in this study
Strain or plasmidRelevant characteristicsaReference
  • a

    Cm, chloramphenicol; GFP, green fluorescent protein; IacZY, β-galactosidase gene ZY of Escherichia coli; Km, kanamycin; r, resistant; Str, streptomycin; thuB, trehalose utilization gene B.

  • *

    Department of Applied Chemistry and Microbiology, University of Helsinki, Finland.

Escherichia coli
 S17-1/λpirRK2 tra regulon, λpir, host for pir-dependent plasmidsSimon et al. (1983)
Sinorhizobium meliloti
 Rm1021SU47 str-21, StrrMeade et al. (1982)
 Sm7024Rm1021 thuB::pThuB582, F(thuB′-lacZ) transcriptional fusion, Strr, KmrJensen et al. (2005)
Sinorhizobium medicae
 HAMBI 1838CmrKristina Lindstrom*
 Smd1306HAMBI 1838 thuB::pThuB582, F(thuB′- lacZ) transcriptional fusion, Cmr, KmrThis study
 pThuB582pVIK112, 582-bp thuB internal fragment, F(thuB′-lacZ) transcriptional fusion, KmrJensen et al. (2005)
 pPROGREENPproUGFP fusions, KmrAxtell & Beattie (2002)
 pVIK112lacZY for transcriptional fusions, KmrKalogeraki & Winans (1997)

Screening for thuA and thuB genes in S. medicae and construction of the thuB mutant

Primers to amplify the thuA and thuB genes of S. medicae were designed using the primer 3 program (LabCollector, Broadway, NY, USA), based on the published sequence information for S. meliloti 1021 (Finan et al., 2001). Primers for thuA were 5′-GTCACGGACAGTCGGCATC-3′ (forward) and 5′-GAGAATTAGCAAACGCATGG-3′ (reverse) and those for thuB were 5′-AACGTCCCGGTGGAGAAGGC-3′ (forward) and 5′-AGTCCTGCGACCCTACGGCCTT-3′ (reverse). Genomic DNA from S. medicae was isolated by the method of Wilson (1994) and amplified by polymerase chain reaction (PCR). The amplified products were purified using the QIAquick Gel purification kit (Qiagen Nordic, Oslo, Norway) and sequenced using the Big Dye® Terminator V3.1 cycle sequencing kit (Applied Biosystems, Foster City, CA, USA) according to the manufacturers’ instructions. The generated sequences were compared with that in the GenBank database using the BLAST program.

Insertional mutagenesis of S. medicae thuB was carried out as described by Jensen et al. (2005). Briefly, E. coli S17-1/λpir containing plasmid pThuB582, constructed by ligating an internal thuB fragment of S. meliloti (nucleotides +197 to +779 with respect to the thuB start codon) into vector pVIK112 (Kalogeraki & Winans, 1997), was cojugated into S. medicae. Transconjugants were tested for their ability to grow on M9 minimal medium containing 0.4% glucose, maltose, sucrose or trehalose (w/v). Genomic DNA was also isolated from these transconjugants and tested for the disruption of the thuB locus by PCR, with primers 5′-CCCATGCGTTT GCTAATTCT-3′ (sequences within the thuB gene but upstream of the fragment cloned into plasmid pVIK112) and 5′-GGCGATTAAGTTGGGTAACG-3′ (sequences within pVIK112). Products obtained were sequenced to determine the recombination site.

Plant materials and growth conditions

Plant genotypes used were Medicago sativa L. cvs Dorlah, Highline and Moapa 69 and Medicago truncatula Gaertn cvs A17 Jemalong and DZA 315.26. Alfalfa seeds were obtained from Larry Teuber and Donald Phillips (University of California, Davis, CA, USA). Medicago truncatula seeds were obtained from J. M. Prosperi (INRA, Montpellier, France) and multiplied locally in our controlled environment facility (Klimalab, Tromsø, Norway). Alfalfa seeds were surface-sterilized and germinated as described in Jensen et al. (2002). Seeds of M. truncatula were surface-sterilized in concentrated H2SO4 for 6 min, washed in sterile water and soaked in 5% bleach for 15 min. They were rinsed again and soaked in water for 4–6 h at room temperature with shaking. Seeds were then placed on tryptic soy agar (1/5 TSA) plates to screen for microbial contamination and germinated at 27°C after incubation at 4°C in the dark for 4–7 d. (For further details on Medicago spp., see

Competitive root colonization and nodule occupancy studies

Competitive root colonization and nodule occupancy experiments were carried out as described in Jensen et al. (2002, 2005). In brief, five sterile seedlings were planted in a Magenta jar (Sigma, St Louis, MO, USA) filled with sterile vermiculite. Each treatment had three replicates with three jars per replicate. Rhizobia at nonsaturating density (102 cells) and saturating density (106 cells) were inoculated onto sterile roots in colonization and nodule occupancy experiments, respectively. The inoculum density of 106 cells was chosen to avoid the occurrence of mixed infections as reported by Gage (2002). Our observations of 194 infection threads and 83 nodules on alfalfa cv. Moapa 69 with fluorescent tagged competing strains revealed no mixed infections or nodules at this inoculum (Jensen et al., 2005).

Plating on selective TY media was used to identify and evaluate numbers of each strain in the initial inoculum, to estimate their numbers colonizing the host root and to identify their occupancy in the root nodules. For nodules produced by S. medicae, selective B+ Man medium was used (Gresshoff & Rolfe, 1978) because we could not re-isolate nodule bacteria on TY. Lack of mixed occupancy in nodules under our experimental conditions was also confirmed in one experiment with alfalfa cv. Moapa 69 where nodule bacteria were plated on M9 selective media containing trehalose as the sole source of carbon in addition to TY media containing the selective antibiotics.

thuB::lacZ expression and the presence of osmotic and oxidative stresses in the infection threads of the host genotypes

Assessment of thuB::lacZ (thuB gene transcriptionally fused to E. coliβ-galactosidase gene Z) expression and oxidative stress in the infection threads of all the host genotypes was carried out with sterile seedlings planted in growth pouches (Lab-Line Instruments Inc, Merose Park, IL, USA) and inoculated with 106 cells of Sm7024 or Smd1306. Infection events were observed 1 wk after inoculation. To detect oxidative stress, roots were stained as described by Santos et al. (2001). For expression of thuB::lacZ, infected roots were observed under a light microscope after staining for β-galactosidase activity as described in Boivin et al. (1990). These experiments were repeated at least three times for each cultivar with a minimum of 20 seedlings per experiment. Moapa seedlings were included in each of these experiments as a positive control. All traceable infection threads on a root (0–5) were then observed with a Leica light microscope and some were photographed. Thus, a minimum of 60 infection threads were observed for each cultivar for assessment. The images were analysed using the software ImageJ available at Osmotic stress encountered by bacteria in the host genotypes was assessed by inoculating roots in pouches with 106 cells of Rm1021 transformed with plasmid pPROGREEN which we confirmed is induced only under osmotic stress in Sinorhizobium spp., as reported earlier for E. coli and Pseudomonas sp. (Axtell & Beattie, 2002). Microscopic observation of green fluorescent protein (GFP) expression in the infection threads was carried out using a Zeiss Axiovert 200M inverted microscope with BP470/40 excitation and BP525/50 emission filters.

thuB::lacZ induction assay in vitro

Induction assays were performed as described in Jensen et al. (2002). β-galactosidase activity was assayed in three replicate cultures according to the method of Miller (1972).

Statistical analysis

Statistical analysis was carried out using minitab statistical program (Minitab, Coventry, UK).


Phenotype of Sm7024 on M. sativa and M. truncatula genotypes

The colonization and nodulation phenotypes of the S. meliloti thuB mutant were tested on two other cultivars of M. sativa in addition to cv. Moapa 69, and two M. truncatula genotypes. At all inoculum ratios tested, Sm7024 colonized M. sativa cvs Dorlah and Highline, and M. truncatula cvs A17 Jemalong and DZA315 26 roots as well as the wild-type Rm1021 (Table 2).

Table 2.  Competitive colonization of roots of Medicago sativa and Medicago truncatula genotypes by Sinorhizobium meliloti Sm7024 and Rm1021 at different ratios of nonsaturating inoculum density (102 cells per root)
Host genotypeInoculum ratio Rm1021:Sm7024Total cells per root (105) ± SERm1021 cells per root (105) ± SESm7024 cells per root (105) ± SE% ratio of Sm7024 in total*
  • *

    χ2 test performed.

  • a

    Numbers with same letters are not statistically significant at P < 0.05 from the expected values.

M. sativa cv. Dorlah 1 : 14.27 ± 1.292.14 ± 0.652.13 ± 0.685049.9a
 1 : 34.10 ± 0.401.11 ± 0.472.99 ± 0.517572.9a
 6 : 15.72 ± 0.344.54 ± 0.511.18 ± 0.4214.320.6a
12 : 13.89 ± 0.523.63 ± 0.510.26 ± 0.02 7.7 6.7a
M. sativa cv. Highline 1 : 14.04 ± 0.382.10 ± 0.181.94 ± 0.25048a
 6 : 15.88 ± 2.284.80 ± 2.021.08 ± 0.2714.318.4a
12 : 15.89 ± 2.365.45 ± 1.890.44 ± 0.33 7.7 7.5a
M. truncatula cv. A17 Jemalong 1 : 16.98 ± 3.503.37 ± 1.723.61 ± 1.825051.7a
 3 : 17.32 ± 0.135.01 ± 0.452.31 ± 0.572531.5a
 1 : 310.6 ± 0.252.96 ± 0.97.64 ± 0.317572.1a
M. truncatula cv. DZA 315.26 1 : 13.12 ± 1.171.49 ± 0.71.63 ± 0.55052.2a

On all the host genotypes tested, Sm7024 formed effective nitrogen-fixing nodules and, when inoculated singly, there was no difference between the wild type and Sm7024 in the average number of nodules formed on plants (data not presented). However, when coinoculated with the wild type, regardless of the inoculum composition, Sm7024 occupied a higher than expected proportion of nodules on M. sativa cv. Highline and M. truncatula cv. A17 Jemalong. The magnitude of the difference between the expected and observed values varied with the composition of the inocula (Table 3). On M. sativa cv. Dorlah and M. truncatula cv. DZA 315.26, in contrast, at all inoculum ratios tested, Sm7024 occupied only the expected proportion of nodules (Table 3). These results show that the competitiveness of Sm7024 compared with Rm1021 is dependent on the interacting host genotype.

Table 3.  Competition for nodule occupancy between Sinorhizobium meliloti Rm1021 and Sm7024 on Medicago sativa and Medicago truncatula genotypes at saturating inoculum density (106)
Host genotypeInoculum ratioa Rm1021:Sm7024Total nodules screenedExpected Sm7024 nodules (%)Observed Sm7024* nodules (%)
  • *

    χ2 test performed.

  • a

    Each treatment consists of three replicates.

  • b

    Not significant and

  • c

    significant at P < 0.05 compared with the expected values.

M. sativa cv. Dorlah1 : 11595046.4b
3 : 11682525.4b
1 : 31687577.7b
M. sativa cv. Highline1 : 11825069.4c
3 : 11802546.5c
1 : 31837585.6c
M. truncatula cv. A17 Jemalong1 : 11805067c
3 : 11802549.3c
1 : 31807589.7c
M. truncatula cv. DZA 315.261 : 11725053.5b
3 : 11712528b
1 : 31827570.9b

Genes thuA and thuB are present in S. medicae and they code for similar physiological and symbiotic functions

As the outcome of competition between S. meliloti wild-type Rm1021 and its mutant Sm7024 was dependent on the interacting host genotypes, we tested the effect of the thuB mutation in a different genetic background. Sinorhizobium medicae was chosen because of its ability to nodulate both M. sativa and M. truncatula (Rome et al., 1996a,b; Roumiantseva et al., 1999) and its similar genome organization to S. meliloti (Roumiantseva et al., 1999; Bailly et al., 2006). PCR-amplified thuA and thuB genes of S. medicae revealed 100% sequence identity to those of S. meliloti 1021. Sequencing of the mutated thuB in S. medicae confirmed the insertion of plasmid pThuB582 (Table 1) at nucleotide +197 (data not presented). We refer to this mutant strain as Smd1306. The phenotype of Smd1306 is identical to that reported for Sm7024 (Jensen et al. 2005). It grew on M9 minimal media containing glucose, maltose and sucrose but not on media containing trehalose as the sole source of carbon (data not shown).

Symbiotic properties and competitiveness of S. medicae Smd1306

When it was not in competition with the wild type, Smd1306 colonized and formed effective nodules as well as the wild type on all the tested host genotypes, including M. sativa cv. Moapa 69 (data not presented). Even when it was competed with the wild-type strain, Smd1306 colonized the roots as well as the wild type on all hosts tested except M. sativa cv. Moapa 69. On this cultivar, Smd1306 was less competitive than the wild type for root colonization (Table 4). The colonization phenotype of Smd1306 on all the hosts tested is thus similar to that of Sm7024.

Table 4.  Competitive colonization of roots of the Medicago sativa and Medicago truncatula genotypes by Sinorhizobium medicae wild type and Smd1306 at a 1 : 1 ratio of nonsaturating density (102 cells per root)
Host genotypeTotal cells per root (104) ± SDWild-type cells per root (104) ± SDMutant cells per root (104) ± SD% ratio of mutant in total*
  • *

    χ2 test performed.

  • a

    Significant and

  • b

    not significant at P < 0.05 compared with the expected values.

M. sativa cv. Moapa 692.75 ± 0.11.99 ± 0.60.76 ± 0.145027.6a
M. sativa cv. Dorlah1.7 ± 0.30.91 ± 0.20.79 ± 0.125046.5b
M. sativa cv. Highline1.27 ± 0.20.67 ± 0.10.6 ± 0.115047.2b
M. truncatula cv. A17 Jemalong5.5 ± 0.232.89 ± 0.212.61 ± 0.025047.4b
M. truncatula cv. DZA 315.261.85 ± 0.51.0 ± 0.20.86 ± 0.25046.2b

The results on nodule occupancy are also in complete agreement with those of Sm7024 on all host cultivars. Smd1306 occupied a significantly higher proportion of the nodules formed than wild type on M. sativa cvs Moapa 69 and Highline and M. truncatula cv. A17 Jemalong. On M. sativa cvs Dorlah and M. truncatula DZA315.26, the nodule occupancy ratio did not deviate significantly from the expected values (Table 5). These results also confirm that host genotype is the critical factor influencing the outcome of these interactions.

Table 5.  Competition for nodule occupancy between Sinorhizobium medicae wild type and Smd1306 on Medicago sativa and Medicago truncatula genotypes at a 1 : 1 ratio of saturating inoculum (106)
Host genotypeTotal nodules screenedExpected Smd1306 nodules (%)Observed Smd1306 nodules (%)*
  • *

    χ2 test performed.

  • a

    Significant and

  • b

    not significant at P < 0.05 compared with the expected values.

M. sativa cv. Moapa 692085076.4a
M. sativa cv. Dorlah1985056.6b
M. sativa cv. Highline1975066.5a
M. truncatula cv. A17 Jemalong2035070.4a
M. truncatula cv. DZA 315.262085056.7b

Evidence of osmotic and oxidative stress during early infection events in the different host genotypes

Early in the nodulation process, rhizobia are exposed to diverse stress(es) including osmotic stress and oxidative bursts (Santos et al., 2001; Jensen et al., 2005). Kiss et al. (2004) proposed that stresses associated with symbiotic infections might vary among host genotypes and physiological conditions within each host. We therefore examined osmotic and oxidative stresses encountered by S. meliloti during infection of all the M. sativa and M. truncatula genotypes used in this study to determine if the stress levels could have influenced competitiveness on these cultivars. Roots were inoculated with wild-type strain Rm1021 transformed with plasmid pPROGREEN in which expression of the cloned GFP is modulated by an E. coli promoter responsive to osmotic shock (Axtell & Beattie, 2002) and examined 1 wk after inoculation for bacteria showing expression of GFP. Strong GFP fluorescence was observed in the infection thread on all the host genotypes, indicating that the bacteria in the infection threads were osmotically stressed in all hosts tested (Fig. 1).

Figure 1.

Green fluorescent protein (GFP) expression in the infection threads (*) formed in roots of (a) Medicago truncatula cv. A17 Jemalong and (b, c) alfalfa (Medicago sativa) cvs (b) Moapa 69 and (c) Dorlah inoculated with Sinorhizobium meliloti Rm1021 containing the plasmid pPROGREEN. Bars, 20 µm.

Roots inoculated with Rm1021 were also stained for oxidative stress response as described by Santos et al. (2001). We observed a dark purple reaction in the infection thread on all the tested host genotypes (data not presented).

Expression of thuB::lacZ on the root surface and in the infection threads

We reported previously that thuB::lacZ is expressed in bacteria colonizing the root surface and infection threads of M. sativa cv. Moapa 69 (Jensen et al., 2005). As the competitive nodulation phenotypes of Sm7024 and Smd1306 were influenced by the host genotypes, we compared levels of thuB::lacZ expression in all these host genotypes to see if thuB expression levels correlated with the competitiveness of the strains. Cultivar-dependent variations in thuB::lacZ expression were evident in the infection thread bacteria but not in bacteria colonizing the root surfaces (data not presented). Observation of at least 60 infections in each cultivar in multiple experiments showed that, in all instances, the lacZ staining was most intense in the infection threads of M. sativa cv. Moapa 69 (Fig. 2a,b,i). It was less intense in M. truncatula cv. A17 Jemalong (Fig. 2c,d,i) and M. sativa cv. Highline (data not presented), and least intense in M. truncatula cv. DZA 315.26 and M. sativa cv. Dorlah (Fig. 2e–i). The expression levels of thuB::lacZ thus correlated with competitiveness of Sm7024 and Smd1306 on these various hosts.

Figure 2.

Expression of β-galactosidase gene Z fused to thuB (thuB::lacZ) expression in the infection threads (*) formed (a, b) in alfalfa (Medicago sativa) cv. Moapa 69 by Sinorhizobium meliloti (a) 7024 and (b) Smd1306; (c, d) in Medicago truncatula cv. A17 Jemalong by (c) Sm7024 and (d) Smd1306; (e, f) in M. truncatula DZA315.26 by (e) Smd1306 and (f) Sm7024, and (g, h) in alfalfa cv. Dorlah by (g) Sm7024 and (h) Smd1306. Bars, 20 µm. RH, root hair; EP, epidermis. (i) LacZ color intensity in infection threads (IT) over background were quantified in frames (a–h) using ImageJ software. Error bars represent 95% confidence intervals of mean blue color intensity in each frame.

The effect of trehalose concentration on the induction of thuB::lacZ

As the intensity of thuB::lacZ expression within the infection threads varied with the host genotype, we tested whether these differences could be indicative of trehalose concentration available to bacteria in the infection thread. lacZ expression was quantified over a period of 3 h in mannitol-grown Sm7024 cells induced with 1 nm to 100 µm trehalose. Results (Fig. 3) show a lack of thuB::lacZ induction up to a concentration of 100 nm trehalose and an almost linear induction response between concentrations of 100 nm and 100 µm trehalose. thuB::lacZ was not induced in control treatments with no trehalose or induction medium containing 1000 µm glucose.

Figure 3.

Induction of β-galactosidase gene Z fused to thuB (thuB::lacZ) expression in Sinorhizobium meliloti Sm7024 at increasing trehalose concentrations. Fold induction values were calculated from β-galactosidase activity according to Miller (1972).


We previously reported that S. meliloti strains Sm7023 and Sm7024, mutated in the trehalose utilization genes thuA and thuB, were more competitive for infection and nodule formation on alfalfa cv. Moapa 69, although they were less competitive in root colonization (Jensen et al., 2005). The results presented here extend these observations to show that S. medicae mutated in the thuB ortholog also possessed the same phenotype on alfalfa cv. Moapa 69. In addition, we found that thuB::lacZ expression and competition phenotypes of both S. meliloti (Sm7024) and S. medicae (Smd1306) mutants exhibited quantitative differences on different host cultivars tested (Tables 3, 5, Fig. 2). These observations agree with earlier reports that the outcome of interstrain competition is influenced by host genotype (Diatloff & Brockwell, 1976; Jones & Hardarson, 1979; Demezas & Bottomley, 1986; Hynes & Oconnell, 1990; Josephson et al., 1991; Leung et al., 1994; Lohrke et al., 1995).

It is at present unclear what regulates the induced expression of the thuA and thuB genes in the infection threads. In the S. meliloti genome, the thuA and thuB genes form part of the trehalose/sucrose transport operon thuEFGKAB, which is located on the pSymB megaplasmid. The thuA and thuB genes are induced in vitro by trehalose but not by sucrose (Jensen et al., 2005) and here we have shown that the extent of its induction depends on the concentration of trehalose available in the medium (Fig. 3). In planta, thuB is induced in the infection thread bacteria but not in bacteroids (Jensen et al., 2005). Recently, Dominguez-Ferreras et al. (2006) reported a 3.5-fold induction of the thuA and thuB genes in S. meliloti cells subjected to sudden salt or osmotic stress in vitro. Both salt stress and osmotic stress also induce trehalose biosynthesis in many bacteria (Giaever et al., 1988; Csonka & Hanson, 1991; Jepsen & Jensen, 2004; Makihara et al., 2005) and in S. meliloti Rm1021 (J. B. Jensen and T. V. Bhuvaneswari, unpublished observation). Sinorhizobium meliloti also possesses other mechanisms to cope with osmotic stress in vitro (Miller & Wood, 1996). Whether the upregulation of thuA and thuB seen by Dominguez-Ferreras et al. (2006) is caused directly by the stress or is caused indirectly by the trehalose synthesized by these stressed cells is unknown at present. Mutants of S. meliloti impaired in the trehalose biosynthesis pathway(s) are necessary to choose between these alternatives and these are not yet available. Reduced induction of thuB::lacZ seen in some host cultivars can therefore be interpreted as either reduced stress for these bacteria in the infection thread or reduced synthesis and/or availability of trehalose for transport in the infection threads of these cultivars. Expression of GFP (Fig. 1) and the positive staining reaction for oxidative stress in the infection threads suggest that both S. meliloti and S. medicae cells are subject to both osmotic and oxidative stresses in the infection threads of all the cultivars tested. In fact, the presence of H2O2 is essential for optimal growth of infection threads (Jamet et al., 2007). We therefore contend that the reduced induction of thuB::lacZ in some cultivars is suggestive of less trehalose being available for transport in the infection threads of these cultivars and not indicative of lack of osmotic or oxidative stress or stress-induced trehalose biosynthesis in bacteria.

The quantitative differences observed in the competition phenotype correlated with the thuB::lacZ induction levels in the infection threads of five host genotypes tested (Fig. 2). This correlation provides additional support for our earlier hypothesis that the ability to transport and accumulate trehalose helped the thuAB mutants to better withstand infection-related stresses and became more competitive (Jensen et al., 2005). Even in organisms that possess multiple stress adaptation strategies, trehalose accumulation has been shown to be critical for stress adaptation, particularly under nutrient-limiting conditions (Herzog, 1990). Moreover, bacteria prefer to accumulate trehalose through transport from the environment to save energy (Oren, 1999). The ability to synthesize and accumulate trehalose during infections also regulates the pathogenicity and virulence of both animal and plant pathogens (Van Dijck et al., 2002; Astua-Monge et al., 2005; Murphy et al., 2005). We suggest that, in host interactions where expression of thuB was greatly reduced, mutants failed to outcompete their parents because the trehalose levels in the wild type and the catabolic mutants were similar as a result of reduced expression of catabolic genes in the wild type. In contrast, in the case of their interactions with alfalfa cv. Moapa 69 and alfalfa cv. Highline and M. truncatula cv. A17 Jemalong, where thuB::lacZ expression was high and moderate, respectively, mutants would have accumulated more trehalose than the wild type.

The source of trehalose present in the infection threads is not clear. Host genotype-dependent differences in the accumulation of trehalose have been reported in legume nodules (Muller et al., 1994; Farias-Rodriguez et al., 1998). In these and other reports, which discuss the significance of trehalose accumulation in legume nodules (Streeter, 1980; Streeter, 1985), bacteroids alone have been considered the source of the trehalose found in nodules. At present, however, we cannot rule out the possibility that the plant host can be the source of trehalose in the infection threads. All higher plants tested to date have been shown to synthesize trehalose via a two-step pathway (Goddijn & van Dun, 1999; Leyman et al., 2001; Grennan, 2007), involving the enzymes trehalose-6-phosphate synthase and trehalose phosphate phosphorylase. Moreover, in plants, the intermediate trehalose-6-phosphate has been shown to be a crucial regulatory molecule in the sugar signaling pathway, particularly in carbon source/sink allocations (Goddijn & van Dun, 1999; Grennan, 2007; Ramon & Rolland, 2007). Different concentrations of trehalose in the infection thread environment could thus signify differences in carbon allocation in different cultivars. Accumulation of trehalose in the infected tissues has been documented in Arabidopsis and cabbage (Brassica oloracea) roots infected with the root hair infecting pathogen Plasmidiophora brassicae (Keen & Williams, 1969; Brodmann et al., 2002). In Arabidopsis, regulation of the trehalose concentration has been suggested to be part of the defense response of the plant to prevent excess accumulation of trehalose in plant cells, where it could interfere with the regulation of carbon metabolism (Brodmann et al., 2002).

The expression of thuB::lacZ in bacteria colonizing the root surface of all cultivars suggests the availability of trehalose in root exudates of all these cultivars. However, trehalose utilization provided a competitive advantage in root colonization only on alfalfa cv. Moapa 69. The mutants and their parents colonized the roots of other cultivars equally well, confirming our earlier report that higher nodule occupancy need not always correlate with competitiveness in root colonization (Duodu et al., 2005; Jensen et al., 2005). These results also suggest that trehalose is not important as a carbon source in the root exudates of these cultivars. The presence of other carbon sources (Boivin et al. 1990; Phillips et al., 1992; Phillips & Streit, 1997; de Rudder et al., 1999; Bringhurst et al., 2001; Jimenez-Zurdo et al., 1997) in adequate amounts may have compensated for their inability to utilize trehalose. Root exudate compositions do vary in different genotypes of the same species (Cieslinski et al., 1997).

In conclusion, our results show that the ability to accumulate trehalose in the infection threads may play an important role in the cultivar-dependent nodulation competitiveness exhibited by both S. meliloti and S. medicae strains. The results reported here also open up the possibility of identifying the host gene(s) involved during infection events that determine the outcome of interstrain competition for nodule occupancy in M. sativa and M. truncatula. Further, in the rhizobial inoculum industry, the common strategy to improve nodulation by inoculated strains has been to use strains with superior survival and colonization potential. Our results show that, for nodule occupancy, the ‘infectivity’ of inoculum strains is more important than their relative numbers in the root environment. Thus, in the future it may be feasible to produce inocula that occupy more nodules even if they colonize the roots poorly.


We thank the Klima Laboratory, University of Tromsø for multiplying M. truncatula seeds, Professors Larry Teuber and Donald Phillips for supplying alfalfa seeds, Dr G. A. Beattie for the pPROGREEN plasmid and Dr K. A. Bråthen for help with statistical analysis of images in Fig. 2.