Introduction

Nitrogen fixation in legumes takes place in highly specialized root nodules that result from the association between Rhizobia and legume plants 1, 2, 3. The bacteria enter the nodule via an infection thread, and then differentiate into nitrogen-fixation bacteroids, which are capable of fixing atmospheric nitrogen into ammonia that provided to the plant. In return, the host plant provides carbon and energy to the bacteroids in the form of C4-dicarboxylic acids 4, 5. Nodules are unique among plant organs as their development arises as a result of interactions among many cooperating plant and bacterial genes. However, at the molecular level, only a few events in the communication between bacteria and plants are understood. In plants, a set of early nodulin genes encode products that are involved in infection and nodule development. The products of late nodulin genes participate in the contribution of the endosymbiont to the specialized metabolic activity of the nodule 6, 7. Within bacteroids, nitrogenase synthesis (nif) and microoxic respiration (fix) genes and an oxygen-limited condition are required for nitrogen fixation 2, 8.

In Sinorhizobium meliloti, the nifA gene as well as nifHDKE, fixABCX and fixLJ are located in a large plasmid, pSymA 9. The fixL and fixJ genes encode a two-component regulatory system in which the oxygen sensor FixL transfers phosphate to the response regulator FixJ. Phosphorylated FixJ positively controls transcription of fixK and nifA 10, 11. FixK induces expression of fixNOQP and negatively affects expression of nifA, whereas NifA is required for transcription of fixABCX, nifN and nifB, as well as for transcription of the nifHDK operon that encodes the subunits of the nitrogenase 12. However, the FixL/FixJ system is not required for nifA gene expression in Bradyrhizobium japonicum 13. In addition to nif and fix genes, NifA controls some other genes not directly involved in nitrogen fixation, such as genes related to nodulation competitiveness, rhizopine synthesis, nodule development and bacteroid persistence in S. meliloti 14.

Previous studies inferred that nifA mutants not only failed to fix nitrogen but also elicited numerous small nodules whose necrotic interior was reminiscent of a hypersensitive response characteristic of non-compatible host-pathogen interactions 3. The nodules induced by the S. meliloti nifA mutant differed from those induced by wild type in that the fixation zone was less extensive and many of the cells interspersed within the fixation zone contained senescent bacteroids. In addition, a significantly larger senescent zone was observed in S. meliloti nifA mutant nodules 15. We previously found that small, white and ineffective nodules were also formed in plants infected with S. meliloti that had been transformed with a multi-copy plasmid carrying nifHDK genes driven by a P1 promoter 16. This indicated that the redundant nifHDK promoter brought about a shortage of NifA protein necessary for normal nodulation in the host plant. Thus, it is speculated that nifA not only regulates nif/fix genes but also genes involved in nodule formation and maintenance. Recently, our unpublished data indicated that nifA affects the nodulation on other lateral roots in split root system. However, little is known about how host genes act in response to the expression of S. meliloti nifA. In the present study, we compared gene expression profiles in an S. meliloti nifA mutant nodules with wild-type. We aimed to characterize the molecular events taking place during symbiotic association, and to provide new clues concerning the relationship between signaling molecules and the S. meliloti nifA gene.

Materials and Methods

Bacterial strains

The wild-type S. meliloti strain Rm1021 and its nifA mutant Sm1354 (nifA::Tn5) strain were used in this study 15. Escherichia coli DH5α was used for plasmid DNA transformation.

Plant cultivation

Seeds from Medicago sativa were surface sterilized, germinated and then grown in plastic barrels on nitrogen-free vermiculite. Wild-type Rm1021 and nifA mutant Sm1354 were inoculated on plants as described previously 17. The nodules induced by Rm1021 or Sm1354 were harvested 15 and 30 days after inoculation and stored at −70 °C.

cDNA-AFLP analysis

cDNA-AFLP was carried out according to standard procedures with little modification 18. Briefly, total RNA was extracted from the nodules 30 days after inoculation (Total RNA isolation Kit IV, Watson, China) according to the manufacturer's instructions. mRNA was purified using an mRNA isolation system (Watson, China). Double-strand cDNA was synthesized using an anchored oligo-(dT)18 primer and Superscript II RNase H reverse transcriptase (Invitrogen) according to the manufacturer's protocol. The pre-amplification step for 16 cycles of polymerase chain reaction (PCR) was performed with an EcoRI+0/MseI+0 adaptor primer combination. Pre-amplified productions were diluted 50-fold and 5 μl was used as template for final amplifications using γ33-labeled EcoRI primer with two selective nucleotides and an MseI primer with two selective nucleotides. The obtained radioactive amplification products were separated on 5% polyacrylamide gels and run at 90 Watts in Sequi-Gen GT nucleic acid sequencing cell (Bio-Rad) until the xylene-cyanol dye reached the 1/3 position from the bottom. Gels were covered with plastic film and positionally marked before being exposed to Kodak film for 3 days.

Fragment characterization

The differentially expressed bands were individually collected as template for re-amplification with the same primers used in AFLP analysis. Re-amplified products were analyzed by agarose electrophoresis, cloned into pMD18-T vector (Takara) and sequenced. The obtained sequences were compared to nucleotide and protein sequences in publicly available databases (http://www.medicago.org/genome/blast.php).

Reverse Northern hybridization

Reverse Northern hybridization was used to confirm the differential gene expression results obtained from cDNA-AFLP. To prepare hybridization templates more conveniently, a pair of specific primers were designed according to pMD18-T vector sequence at cloning site (F: 5′GCG GAT AAC AAT TTC ACA CAG3′; R: 5′CCA GGG TTT TCC CAG TCAC3′). Equal volumes of PCR products (about 1 μg) were loaded into 1.5% agarose gels. DNA samples were then transferred to nylon membranes (Amersham Biosciences). Total RNA was extracted from 15- and 30-day-old nodules. Radiolabeled cDNA probes were synthesized by reverse transcription of 10 μg of total RNA for 1 h in the presence of 100 μCi [32P]dCTP with the Superscript reverse transcriptase. Probes derived from each RNA sample were hybridized against one set of transcript-derived fragments (TDFs) obtained above. Hybridization was performed in 0.5 M Na-phosphate (pH 7.2), 1 mM EDTA, 7% SDS buffer for 16 h at 65 °C, washed twice for 5 min with 40 mM Na-phosphate/1 mM EDTA/5% SDS and then washed four times for 10 min with 40 mM Na-phosphate/1 mM EDTA/1% SDS. Hybridization signals were observed first on Kodak X-ray film and then quantified by Dots hybridization analysis system (Tanon 2.20, China).

RT-PCR

To control for equal amounts and quality of RNA template, an additional RT-PCR was performed under the same conditions as PCR with specific primers based on the sequence of Ribosome large subunit gene (F: 5′ACC AGA GCC GCT AAG GTT3′; R: 5′CAG CTT ACA TAA CCG GTC3′). The primers used for 24 cDNA TDFs are listed in Table 1.

Table 1 Primers used in RT-PCR experiments for 24 TDFs
Full size table

Results

Characterization of differentially expressed TDFs

cDNA-AFLP reactions were performed with 64 combinations of EcoRI and MseI primers having two selective 3′ terminal nucleotides. This reduced the number of fragments per fingerprint to ∼50. Although the majority of the cDNA-AFLP fragments did not show any significant change, 44 differentially expressed TDFs were detected. Thirty-seven differentially expressed fragments were excised from the gels, re-amplified by PCR and sequenced. There were technical problems with sequencing the other seven fragments and thus they were not studied. Figure 1 shows an example of a typical cDNA-AFLP banding pattern. Medicago sativa is a tetrasomic plant and its sequence information was limited. However, the database of the diploid plant Medicago truncatula provided reliable and sufficient information for sequence comparison. TDF fragments that were only 50–300 bp long were subsequently extended to ∼500 bp for Northern and RT-PCR experiments, if homologous sequences were available in public databases (Table 2). The extended cDNA fragments were obtained by RT-PCR, and then cloned into pMD18-T for sequence analysis. Through BLAST analysis, TDFs were organized into several categories according to their putative functions, including primary and secondary metabolism, signal communication, protein degradation, cell growth and development (Table 3).

Figure 1
figure 1

cDNA-AFLP display of transcripts in nodules elicited by S. meliloti wild-type strain Rm1021 and nifA mutant Sm1354. RNA was extracted from 30-day-old nodules induced by wild-type Rm1021 (A+) and nifA mutant Sm1354 (A−). Lanes are in groups of two, and each group was amplified using one combination of EcoRI and MseI primers having two selective 3′ terminal nucleotides. (A) A selection of cDNA-AFLP display. (B) Enlarged view of the box in (A). The arrow shows the differentially expressed TDFs.

Full size image
Table 2 The EST database in GenBank showing homology with differentially expressed TDFs
Full size table
Table 3 AFLP, Northern blot and RT-PCR analyses of the differentially expressed TDFs in nifA mutant Sm1354 nodules at 30 days after inoculation
Full size table

Reverse Northern and RT-PCR verification

To assess the results of cDNA-AFLP analysis, gene expression patterns were first verified by reverse Northern blot. The results from hybridization-based methods were not consistent with cDNA-AFLP, as only 28 TDFs (75.6%) showed coincident expression patterns with those in AFLP experiments (Figure 2, Table 3). In order to confirm our results, RT-PCR was carried out on 24 TDFs, including all of the nine TDFs showing conflicting expression patterns between cDNA-AFLP and Northern blot results. All of the RT-PCR products were sequenced directly for verification. The results of 23 TDFs (95.8%) in RT-PCR experiments were consistent with those from Northern blot analyses, whereas the remaining G12 was consistent with the result from cDNA-AFLP. This indicated that reverse Northern blot is more reliable than cDNA-AFLP (Figure 3, Table 3).

Figure 2
figure 2figure 2

Reverse Northern blot analysis of TDFs. (A) Electropherogram of methanol denaturing gel electrophoresis of total RNA from 30-day-old nodules induced by wild-type Rm1021 (A+) and nifA mutant Sm1354 (A−). (B) Autoradiograms of reverse Northern hybridization. The duplicate membranes were hybridized against total cDNA probes resulting from wild-type Rm1021 and Sm1354 nodules, separately. Upper part show agarose gel electrophoresis of the 37 amplified cDNA fragments stained with EtBr. (C) Quantification of mRNA expression levels. Bands in their intensities were quantified by scanning and normalized with ubiquitin control. The differences in expressions were determined by Student's t-test. N=3, p<0.01.

Full size image
Figure 3
figure 3figure 3

RT-PCR analysis of differentially expressed TDFs in 30-day-old nodules induced by wild-type Rm1021 (A+) and nifA mutant Sm1354 (A−). (A) Agarose gel electrophoresis of PCR productions. The expressions of 24 TDFs were checked at 20, 25 and 30 cycles. RsL5, ribosome large subunit gene, was used as control. (B) Quantification mRNA levels. Bands in their intensities were quantified by scanning. The differences in expression rates were determined by Student's t-test. N=3, p<0.01.

Full size image

Analysis of gene expression time series in nifA mutation nodules

The expression patterns of TDFs are dynamic in nodules. In Northern blots of the nodules 15 days after inoculation, the expression patterns of some TDFs differed from those 30 days after inoculation. Compared with wild-type nodules, the phospholipaseA2 (G36), receptor-like kinase (G25) and three unknown functional cDNA fragments (G3, G17, G18) showed no distinct difference in expression level in 15-day-old nifA mutant nodules. A senescence-related protein (G30) was down-regulated in 15-day nifA mutant nodules but up-regulated in 30-day nodules. In contrast, 13 TDFs (35%) including one nodulin (G22) and the G-protein α-subunit (G16) were up-regulated in 15-day nodules, and yet they were down-regulated at 30 days (Figure 4).

Figure 4
figure 4figure 4

Same as Figure 2, except for 15-day-old nodules.

Full size image

Discussion

The cDNA-AFLP technique is a reliable method for revealing small differences between close individuals, especially in non-sequenced organisms. Since its first application on potato tuber development, several modifications of this technique have improved its validity and advantages over other fingerprinting techniques and DNA chip-based approaches 18, 19, 20. In most cases, it has been employed for surveying transcriptional changes in host plants during interactions with fungi, bacteria, nematode or environmental stimuli 21, 22, 23, 24, 25, 26, 27, 28, 29. In this study, cDNA-AFLP was successfully used to identify 37 differentially expressed host plant genes in Alfalfa nodules.

Among the 37 TDFs obtained in this study, six were found to be related to signal transduction. The ethylene response factor contains a core sequence of GCC box, which is recognized by ethylene-responsive element-binding proteins to regulate ethylene responses via regulating gene transcription and expression 30, 31, 32. Calmodulin is recognized as a major calcium sensor and orchestrator of regulatory events through its interaction with a diverse group of cellular proteins, including cytoskeletal elements, ion channels, kinases and phosphatases. 33, 34, 35. In Arabidopsis and rice plants, a wide range of processes including seed germination, shoot and root growth, and stomatal regulation are altered by mutations in G-protein components 36, 37, 38, 39, 40, 41. In Alfalfa, G proteins mediate the nod factor signaling pathway during symbiosis 42. However, expression patterns of up-regulated G-protein-coupled receptors (G7) and down-regulated G-protein α-subunit (G16) indicate a complex regulation pattern of G proteins in nodules. The retinoid-interferon-induced mortality-19 gene (GRIM-19) and cytokine IK are novel cell death regulators characterized only in animals 43, 44, 45. Further studies on their homologous genes in plants may provide new insights into apoptotic plant cell death.

The ubiquitin/26S proteasome pathway is a major route for the selective degradation of cytoplasmic and nuclear proteins in eucaryotes 46, 47. At least three enzymes are involved in this action – a ubiquitin-activating enzyme (E1), a ubiquitin-conjugating enzyme (E2) and a ubiquitin ligase (E3) – which work sequentially in a cascade 48, 49. In the present study, one ubiquitin carboxyl terminal hydrolase and one ubiquitin-conjugating enzyme (E2) were down-regulated in 30-day nifA mutant nodules. This indicates that the elaborate regulation of protein is damaged to some degree in nifA mutant nodules. The effects of protein degradation are not well understood in legume nodules and further study is needed.

The nutrient exchange between Rhizobia and legume plants is executed by a number of genes including those involved in primary and secondary metabolism. In nifA mutant bacteroids, a number of genes related to central intermediary metabolism are down-regulated in bacteroids 30 days after inoculation (unpublished data). In our study, seven host plant genes related to central intermediary metabolism were identified in S. meliloti nifA mutant nodules. Five genes were down-regulated in 30-day nifA mutant nodules, whereas the phenylalanine ammonia-lyase class II (G8) gene and arabinogalactan (G5) were up-regulated. PAL catalyses the non-oxidative deamination of L-phenylalanine to trans-cinnamic acid. This enzyme is ubiquitous in plants and is involved in the production of phenylpropanoids such as lignin and phytoalexins 50, 51. The amino-acid permease (G37) is an integral membrane protein involved in the transport of amino acids into the cell. A study of Saccharomyces cerevisiae found that the amino acid permease AGP1 is dependent on an ubiquitin ligase complex, suggesting that an ubiquitination step is required for amino acid signal transduction 52.

The nifA gene exerts its pleiotropic nature in controlling further nodulation by eliciting host plant defense reactions at early stages in development 3. In addition to the nif/fix genes, four nodulation-specific genes (nodH, nodL, nolF and noeB) were down-regulated in 30-day nodules induced by the S. meliloti nifA mutant (unpublished data). In our study, two late nodulin genes (G22, G35) were found to be down-regulated in nifA mutant nodules. In addition, the cellulose synthase-like protein D3 was also down-regulated. Its homologous gene in Arabidopsis is involved in biosynthesis of polysaccharides required during root hair elongation 53, 54, 55. As root hair deformation is a defined procedure for nodule initiation, cellulose synthase-like protein D3 is likely to influence nodulation development. Considering our results together with previous findings, the data suggest that nodule development is affected by mutations of the S. meliloti nifA gene.

Although several molecular events taking place in nifA mutants are illustrated in the present study, there are not enough data to fully understand the relationship between these molecular events and the nifA gene. Even though time-dependent gene expression was deduced from the expression patterns of 15- and 30-day nodules, more studies are needed to elucidate the roles of these genes in nodule development.