Logo of aemPermissionsJournals.ASM.orgJournalAEM ArticleJournal InfoAuthorsReviewers
Appl Environ Microbiol. 2005 Dec; 71(12): 8042–8048.
PMCID: PMC1317383

Microarray-Based Detection and Typing of the Rhizobium Nodulation Gene nodC: Potential of DNA Arrays To Diagnose Biological Functions of Interest


Environmental screening of bacteria for the presence of genes of interest is a challenging problem, due to the high variability of the nucleotide sequence of a given gene between species. Here, we tackle this general issue using a particularly well-suited model system that consists of the nodulation gene nodC, which is shared by phylogenetically distant rhizobia. 41mer and 50mer oligonucleotides featuring the nucleotide diversity of two highly conserved regions of the NodC protein were spotted on glass slides and cross hybridized with the radioactive-labeled target genomic DNA under low-stringency conditions. Statistical analysis of the hybridization patterns allowed the detection of known, as well as new, nodC sequences and classified the rhizobial strains accordingly. The microarray was successfully used to type the nodC gene directly from legume nodules, thus eliminating the need of cultivation of the endosymbiont. This approach could be extended to a panel of diagnostic genes and constitute a powerful tool for studying the distribution of genes of interest in the environment, as well as for bacteria identification.

Many genes of interest or of concern (biodegradation, nitrogen fixation, molecule production, pathogenesis, antibiotic resistance, etc.) are carried by taxonomically diverse bacteria that live in different environments. Since many genes do not confer easily detectable phenotypes and since many bacteria are unculturable, we invested in the adaptation of the microarray technology to screen biological samples for particular genes with application in fields such as environment, health, and industry.

DNA microarrays are widely used for transcriptome analysis (16), single-nucleotide polymorphism (31) and mutation detection (28), resequencing (38), comparative genomics (14), and identification of bacterial species (10). However, although DNA microarray technology holds promise for microbial ecology and diagnosis, few microarray studies have been conducted to detect the presence of particular genes in biological or environmental samples. In most cases, strain or gene identification relies on specific hybridization of PCR-amplified or oligonucleotide probes with PCR-amplified genes or genomic DNA (4, 39). Genes frequently used are 16S rRNA (17, 37), virulence (2, 4), and nitrogen cycle genes (13, 30). A serious limitation to specific hybridization-based microarray is that both the amino acid and nucleotide sequence of a given gene vary considerably from one genome host to another, making it unfeasible to represent all possible alleles of a given gene on an array. For the same reason, the design of universal primers for amplification of a gene distributed in several taxonomic lineages is often impossible in practice. While our work was in progress, a pioneering use of microarrays for detection of known and new viruses using highly conserved sequences from virus families was published (35, 36).

Our incentive was to create a prototype DNA array that has the potential to detect and type known and unknown variants of a given gene. To achieve this goal, we focused on the nodulation gene nodC of rhizobia as a model system. Rhizobia is the generic name for phylogenetically diverse bacteria that induce the formation of nitrogen-fixing nodules on the root of legumes (21, 27). This property relies on the presence in their genome of a set of nodulation (nod) genes essential for symbiosis. nod genes are involved in the production of lipochito-oligosaccharides (Nod factors) that act as signaling molecules for the nodulation of specific legume hosts (23, 24). nodC is a key symbiotic gene which, together with nodA and nodB, is responsible for the synthesis of the core structure of Nod factors (12, 29). Since rhizobia exhibit widespread phylogenetic diversity, nodC is distributed in many genera of α- and β-proteobacteria (8, 21) but has no paralogs in nonrhizobia so far. Overall sequence similarity between the different nodC alleles ranges from 40% to 100% identity.

Here, we describe the construction and validation of a prototype nodC microarray with the potential to detect already described, as well as undescribed, nodC sequences in bacteria. A total of 130 probes consisting of 41mer and 50mer oligonucleotides derived from two conserved amino acid regions of the NodC protein were spotted on glass slides and hybridized with radioactively labeled genomic DNA of a collection of rhizobial and nonrhizobial strains under low-stringency conditions. Analysis of the hybridization patterns using hierarchical clustering allowed us to discriminate rhizobia from nonrhizobia and to group rhizobia on the basis of their nodC sequence. The method generated accurate results with genomic DNA from both bacterial cultures and bacteria-infected plant nodules and proved to be promising for the detection and typing of genes of interest.


Bacterial strains and growth conditions.

Strains used in these studies are listed in Table Table1.1. Strains were maintained and grown on yeast extract-mannitol medium (34) at 28°C.

Bacterial strains used in this studya

Microarray production.

Seventy-four (26 DMEY-E and 48 DMEY-P) 50mer oligonucleotide probes were designed that target nodC sequences corresponding to the IDMEYWL amino acid conserved region. Fifty-six (24 MCCCb-E and 32 MCCCb-P) 41mer oligonucleotide probes were designed that target the nodC sequence corresponding to the VMCCCGP conserved region. 41mer and 50mer oligonucleotides specific to purα and lecRK1 genes from Arabidopsis thaliana were spotted as positive and negative control spots, respectively. Probe sequences are given in Table S1 in the supplemental material. Oligonucleotide probes were synthesized by Sigma (France) with a 5′ amino linker modification. The amino linker modification permitted the covalent attachment of oligonucleotide probes to dendrimer-activated glass slides (dendrislides) available at the Toulouse Génopole (http://biopuce.insa-toulouse.fr/) (15). Oligonucleotide probes were solubilized at 20 μM in 0.15 M Na2HPO4 (pH 9). Each spot consisted of 2 nl of oligonucleotide, and each nucleotide was spotted in duplicate on each array with the VersArray ChipWriter Pro Bio-Rad Spotter at the Jouy-en-Josas Génopole.

After being spotted, the slides were allowed to dry overnight at room temperature. The reduction of the imine functions formed between probes and dendrimers was carried out by immersion of the slides into 3.5 mg/ml NaBH4 for 3 h at room temperature under agitation. The DNA slides were washed twice in room temperature milliQ water for 2 min and then dried under a stream of N2. The DNA arrays were then stored in a dry container at room temperature until used.

Target sequence generation and labeling.

Genomic DNA was isolated from bacterial cultures as previously described (20). Biological repeat samples represent independently grown strains from the same collection source, processed identically on the array. Bacteroids from 6- to 8-week-old nodules were isolated as previously described (5) except that extraction was not performed in liquid nitrogen. DNA was isolated from nodule bacterial pellets as previously described (9). A total of 200 to 600 ng of DNA enriched in bacterial DNA per nodule was routinely obtained.

A 300-bp purα fragment was amplified from pAD-purα (32) using the primer pairs TGGAAGCTAATTCAGGCGG and CTTGGAATCAAGCTGC. To check the quality of hybridizations, 0.2 ng of the purα PCR product was added to 1 μg genomic DNA before being labeled. Genomic DNA was degraded into 200-to 800-bp fragments by sonication in ice (22 cycles, each cycle consisting of 30 s on and 30 s off; Branson sonifier 250), precipitated, and resuspended in milliQ H2O. The quality and size of genomic DNA were checked on agarose gels (1%). A total of 0.2 ng purα PCR product and 1 μg genomic DNA were labeled by random priming (Biokit labeling; Invitrogen), following the supplier's recommendations, except that 0.012 mM each dATP and dCTP, 1.2 mM each dGTP and dTTP, and 40 μCi each of α-33P-labeled dATP and dCTP (Perkin-Elmer) were used instead of the provided deoxynucleoside triphosphates. Probes were purified with microspin S-200 HR columns (Amersham), dried under a vacuum, and resuspended in 50 μl of hybridization buffer (5× SSC [1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate], 5× Denhardt's solution, 100-μg/ml sheared salmon sperm, and 0.5% sodium dodecyl sulfate [SDS]).

Hybridization and scanning.

Prehybridizations were carried out using 100 μl of hybridization buffer under a coverslip at 65°C for 2 to 3 h at high humidity. Coverslips were removed by immersion in 0.1× SSC. The 33P-labeled genomic DNA was heated for 5 min at 95°C and rapidly cooled on ice. Hybridizations were carried out under a coverslip for 16 to 18 h at 65°C. Coverslips were removed by immersion in 3× SSC-0.5% SDS at room temperature. Arrays were routinely washed twice by immersion in 3× SSC-SDS 0.5% at 65°C for 15 min, followed by washing in 3× SSC at room temperature for 5 s. SSC (1× or 2×) was used instead of 3× SSC for the establishment of optimal stringency conditions. Arrays were then centrifuged (400 × g for 5 min), immersed in 95% ethanol, and dried by centrifugation. Slides were scanned on a Fuji BAS-5000. Hybridization signals (designated S) were quantified and analyzed with Array Gauge software.

Microarray analysis.

Background signal (B) was estimated as the mean of the signals of the 108 spots containing only buffer. Spots were considered positive where S/B was >3. The percentage of positive spots (N) observed in each experiment was used to differentiate between positive and negative experiments. Eight nonrhizobia experiments (two hybridizations from each of the strains GMI1000, C58, ATCC 33913, and K12) were used to compute a threshold (T) as follows: T = μ(N) + 3σ(N), where μ(N) corresponds to the average number of positive signals among the eight nonrhizobia experiments and σ(N) to its standard deviation. Positive and negative experiments were, respectively, defined by N > T and N < T.

Hierarchical clustering for global visualization and for the selection of positive experiments (S hierarchical clustering).

Hybridization signal distributions deduced from each experiment were independently log2 transformed and normalized by a centering and reducing method to obtain values with a null mean and a unit standard deviation. Normalized values corresponding to negative spots (S/B < 3) were standardized to the lowest normalized value observed in the whole data set (all probes and all experiments). Hierarchical clustering was applied to a normalized hybridization signal data matrix, where each row corresponds to an experiment and each column corresponds to a probe. Distances computation (euclidean distance) and dendrogram reconstruction (complete linkage method) were made under the R software environment (http://cran.r-project.org) using the dist and hclust functions, respectively.

Hierarchical clustering of selected positive experiments (C hierarchical clustering).

A normalization procedure was conducted as described above except for negative spots where normalized hybridization signals were not standardized. Hierarchical clustering (euclidean distance and complete linkage) was applied to data matrices from which negative experiments were excluded. Bootstrap analyses were performed for >100 randomly generated artificial datasets containing 33% of the normalized hybridization values found in the original data matrix. Those analyses were conducted using the programs of the PHYLIP package, NEIGHBOR (unweighted-pair group method using average linkages tree reconstruction), to generate the 100 corresponding dendrograms and CONSENSE, to deduce consensus trees.

Sequencing and sequence analysis.

Partial nodC sequences (300- to 683-bp PCR products) were amplified and sequenced using the following primer pairs: H1-GTNGGNAARMGNAARGC and CANGGVCCRCARCARCACA, TAC GACCCTAGCGCCCGATGTC and B4-CGNGCCCANCKNARYTGYTG, GCGGAACTGATACTTAACGTCG and CCACAGCTTGTCTGGTACAAC GG, GGAGATCTGGTTCTCAACGTCG and B4-CGNGCCCANCKNARY TGYTG, H1-GAYATGGARTAYTGGYT and GTRCANARNGCRTA, H1-GTNGGNAARMGNAARGC and CTCAATGTACACARNGCRTA. The sequences of the H1 and B4 tags are GGTTCCACGTAAGCTTCC and GCGATTACCCTGTACACC, respectively. nodC partial sequences obtained in this study can be found as Table S2 in the supplemental material.

Plant tests.

Nodulation tests were performed as previously described (20) except that Macroptilium atropurpureum, Vigna unguiculata, Mimosa pudica, and Sesbania grandiflora seeds were superficially sterilized with concentrated sulfuric acid for a duration of 10 min, 30 min, 10 min, and 60 min, respectively, and that Medicago sativa seeds were sterilized with sodium hypochlorite for a duration of 2 min. Nodules were harvested after 4 to 8 weeks for bacterial DNA extraction.


CODEHMOP oligonucleotide design strategy.

Multiple alignment of available NodC protein sequences (18 full-length and 26 partial sequences) highlighted short regions with a high level of conservation (Fig. (Fig.1A).1A). Two of these regions were selected for oligonucleotide design. Oligomers (each, 50 bp) were derived from the 17-amino-acid region containing the IDMEYWL central motif (DMEY oligonucleotides) and oligomers (each, 41 bp) were derived from the 14-amino-acid region containing the MCCCGP central motif (MCCC oligonucleotides). To maximize the spectrum of detectable nodC sequences, two classes of oligonucleotides were synthesized for each conserved motif: (i) E-oligonucleotides (E-probes) corresponding to rhizobial sequences available in databanks (26 DMEY-E and 24 MCCC-E) and (ii) P-oligonucleotides (P-probes) that did not correspond to any described sequence but were derived from the conserved amino acid sequence by a strategy that provides a limited number of nucleotides encoding the chosen amino acid motif. We called this probe design strategy CODEHMOP (for consensus degenerate hybrid motif oligonucleotide probe), since it resembles the CODEHOP (for consensus-degenerate hybrid oligonucleotide primer) PCR primer design strategy (26). P-oligonucleotide sequences (48 DMEY-P and 32 MCCC-P) each contained a specific sequence in the central core, flanked by a common 5′ end sequence and a common 3′ end sequence (Fig. 1B and C). Each core corresponded to one of the possible combinations for encoding a highly conserved five- to seven-amino-acid sequence within the chosen region. Left and right end sequences, added to increase probe length, were derived from the most frequent nucleotide at each position as identified from the nodC sequence alignment. All nodC sequences available in databanks exhibited >75% identity with at least one of the P-oligonucleotides.

FIG. 1.
(A) Consensus 154-amino-acid protein sequence obtained from the ClustalX alignment of 44 NodC partial and complete sequences (from amino acid 138 to amino acid 293 in the S. meliloti sequence). Database accession numbers for the sequences used are as ...

Optimization of experimental conditions and data processing using reference bacteria.

Determination of hybridization conditions and data processing were accomplished using genomic DNA of reference strains. Reference strains included seven rhizobial strains whose nodC sequence was available and represented on the microarray, as well as four completely sequenced nonrhizobial strains known not to contain a nodC gene (Table (Table11).

Array hybridizations were performed with radioactively labeled genomic DNA under low-stringency conditions (3× SSC wash buffer; see Materials and Methods) in at least two independent biological repeats. For the validation of the microarray, only hybridizations with E-oligonucleotide probes were considered. Sequence identities between E-probes and targets ranged from 70% to 100%.

Typically, each reference rhizobial genomic DNA strongly hybridized with the perfect-match probe, as well as with a panel of closely related probes, while nonrhizobial strains hybridized poorly with all probes. Hybridizations were statistically analyzed in two steps.

In the first step, we defined criteria to filter out nonrhizobia. Both the hybridization pattern and the percentage of positive spots were found to be discriminating. The patterns of hybridization of all bacteria were compared using a modified hierarchical clustering method, which will be referred to in the text as S hierarchical clustering. This method standardized and normalized values corresponding to negative spots (see Materials and Methods). All nonrhizobia grouped to the same clade on the dendrogram (Fig. (Fig.2A).2A). A threshold value (T) of positive signals was then calculated based on hybridizations with the nonrhizobial strains (see Materials and Methods) (Fig. (Fig.2A).2A). For each rhizobial DNA, the percentage of positive signals (N) was found above this value. The only exception was Azorhizobium caulinodans, for which the value of N was above T in one experiment and below T in the other experiment and which globally gave the lowest response. Such a result was not surprising, since the A. caulinodans nodC gene is the most distant nodC gene known, with only a maximum of 78% and 84% identity with the MCCC and DMEY nucleotide regions from other rhizobia, respectively. In further studies, bacteria that are clustered using S hierarchical clustering and that exhibit an N value below that of T were filtered out. The remaining bacteria, where N was >T, were considered as nodC-harboring strains and selected for subsequent analysis.

FIG. 2.
Microarray analysis of reference bacteria. (A) nodC detection. The dendrogram was constructed from 11 reference bacteria using S hierarchical clustering. A gray horizontal bar beside each bacterial name corresponds to the percentage of positive spots ...

In a second step, a classical hierarchical clustering, which will be referred to as C hierarchical clustering (see Materials and Methods), was applied to selected bacteria, i.e., rhizobia. Identical sequences from biologically independent experiments were found to group in a same clade on the dendrogram, confirming the consistency of the results (Fig. (Fig.2B2B).

Experiments performed using other stringency conditions (washing in 1× SSC or 2× SSC) (see Materials and Methods) established that washing in 3× SSC gave the best results for both nodC detection and typing (data not shown). This condition was thus used for further experiments.

Use of the microarray to screen a collection of rhizobia and nodule isolates.

Total genomic DNA, purified from 14 rhizobial strains whose nodC sequences were unknown except in one case, was hybridized to the nodC microarray (Table (Table1).1). These rhizobia belonged to nine different genera and 12 different species from both α- and β-proteobacteria and were isolated from many different legume species. We also tested two recent nodule isolates belonging to the γ-subclass of proteobacteria, Pseudomonas sp. Hs1 and Escherichia sp. Hp22a, that were proposed to be legume endosymbionts (3). Two independent biological repeats were done in most cases. For most of the strains, the highest signal intensity was obtained with P-oligonucleotides rather than with E-oligonucleotides, thus emphasizing the power of P-oligonucleotides. The 11 P-probes that gave the best hit with at least one strain were added to E-probes for data processing.

All strains except four were unambiguously identified as nodC-harboring strains by the criteria established previously, i.e., the percentage of positive signals and S hierarchical clustering analysis (Fig. (Fig.3A).3A). nodC partial amplification and sequencing was performed for 9 strains out of the 10 positive strains tested. Four of the strains exhibited <80% identity with any known nodC gene over a 370-bp length. Among them, USDA3001 exhibited only a best identity of 74% with a known nodC sequence over a 690-bp length. The new nodC sequences were found to harbor the chosen amino acid motifs, except for strains ORS1073 and USDA3001, which possessed an LCCCGP motif instead of MCCCGP. No nodC amplification could be obtained for the rhizobial strain Devosia neptuniae J1, which was however selected by microarray analysis. The two stem-nodulating and photosynthetic rhizobia Bradyrhizobium sp. ORS278 and Bradyrhizobium denitrificans Btai1 were not selected by microarray (Fig. (Fig.3A).3A). Their nodC could not be amplified, despite numerous attempts, indicating that these strains possessed highly divergent nodC genes. The nodule isolates Pseudomonas sp. Hs1 and Escherichia sp. Hp22a also gave a clear negative response (Fig. (Fig.3A),3A), indicating that they either did not possess a nodC gene or harbored a very divergent one. The absence of the highly conserved nitrogen fixation nifH gene, as indicated by PCR amplification (data not shown), suggests that these strains may not be genuine rhizobia.

FIG. 3.
Microarray analysis of a collection of rhizobia and nodule bacteria. (A) nodC detection. Dendrogram constructed from 27 bacteria using S hierarchical clustering. A gray horizontal bar beside each bacterial name corresponds to the number of positive spots ...

Analysis of the 12 selected strains together with the reference rhizobia using C hierarchical clustering resulted in the dendrogram shown in Fig. Fig.3B.3B. Sequences were divided in 13 groups, most of them strongly supported by bootstrap analysis, and three individual sequences corresponding to unique experiments. Clusters 1 to 8, 10, 11, and 13 grouped biologically independent repeats of a same strain. Cluster 9 grouped two strains, TJ182 and STM815, that exhibited identical nodC sequences for both IDMEYWL and VMCCCGP motifs. Cluster 12 grouped sequences from three different strains, which exhibited very conserved nodC sequences (>93% identity). The use of either DMEY or MCCC oligonucleotides for in silico analysis did not modify nodC detection but significantly altered the C hierarchical clustering dendrogram (data not shown), showing that the strain identification was more reliable using two protein motifs than one.

nodC detection and typing from legume nodule material.

To assess the performance of the microarray in detecting and typing nodC directly from plant samples without previous in vitro cultivation of the endosymbiont, we hybridized the microarray with bacterial DNA extracted from one or several nodules coming from different legumes inoculated by their rhizobial partners under axenic conditions: Macroptilium atropurpureum with Rhizobium sp. NGR234 or Bradyrhizobium japonicum USDA110, Vigna unguiculata with B. japonicum USDA110, Mimosa pudica with Ralstonia taiwanensis LMG19424, Sesbania grandiflora with Sinorhizobium terangae ORS604, and Medicago sativa with Sinorhizobium meliloti RCR2011. Bacterial DNA extracted from several nodules from one M. atropurpureum plant inoculated by a mixed inoculum of three of its symbionts, Rhizobium sp. NGR234, B. japonicum USDA110, and Rhizobium etli CFN42, was also used. All hybridization patterns from nodule samples could be grouped with the patterns obtained from corresponding free-living cultures on the dendrogram constructed using E-probe, the 11 selected P-probe signals, and C hierarchical clustering (Fig. (Fig.3B).3B). The hybridization pattern obtained from several nodules in the coinoculation experiment grouped with B. japonicum USDA110, revealing the prevalence of this strain in examined infected nodules.


Although molecular hybridization is a powerful method for gene detection in the environment, it faces the problem that nucleotide sequences encoding functionally equivalent proteins exhibit overall similarity commonly varying from 40% to 100% identity, depending on the host genome. Using the nodulation gene nodC as a model system, we developed a microarray approach to detect and type a given gene in various genomic backgrounds. Our approach resembles the strategy recently developed for virus detection (35, 36), insofar as it uses a pool of related oligonucleotide probes designed from conserved DNA regions for target identification. Our strategy however exhibits three features that make it more efficient for the detection and typing of genes: (i) an original probe design including oligonucleotide probes predicted ab initio from the protein sequence (P-probes), (ii) the use of low-stringency hybridization conditions, and (iii) the analysis of hybridization patterns by statistical methods.

To maximize the spectrum of detectable sequences and thus limit the problem of gene sequence variability, two highly conserved regions (17-residue DMEY motif and 14-residue MCCC motif) of the protein were represented on the microarray by multiple and closely related oligonucleotides corresponding to all available sequences (E-oligonucleotides). The 50 E-oligonucleotides spotted, however, only represented a small proportion of the total possible nodC nucleotide diversity, since >1 million different sequences could potentially encode the total 14- to 17-residue protein sequences. We thus enriched the microarray with sequences (P-oligonucleotides) that would encode the two motifs and were designed with a CODEHMOP strategy somewhat similar to that of the CODEHOP PCR primer design, i.e., a variable central region flanked by fixed ends (26). The anatomy of P-oligonucleotides allowed the addition of a limited number of sequences that constituted a reservoir of probes potentially hybridizing with homologous sequences not explicitly represented on the array. P-oligonucleotides that gave the best hit with at least one target were retained and added to E-oligonucleotides for subsequent in silico analysis. This strategy proved to be valuable, since the highest signal was obtained with a P-oligonucleotide for 9 of the 13 unknown nodC sequences. In the future, the design of the P-oligonucleotides could probably be improved by reducing the length of the spotted oligonucleotides and hence that of fixed ends.

Low-stringency conditions improved cross-hybridization of genomic DNA with multiple oligonucleotides, which allowed us to detect not only genes that were explicitly represented on the microarray but also many novel nodC genes, some of which exhibited <80% identity with any known sequence. The only two potential nodC genes undetected by the array could not be amplified by PCR, indicating that very high sequence divergence between target and probes is, however, a limitation of the technique. As for viral discovery (35), these results demonstrated that maximizing potential cross-hybridization is an appropriate strategy to detect novel variants of a gene of interest.

Cross-hybridization among closely related gene sequences is often considered a limitation for gene identification (18). Here, we used the strain-specific pattern of hybridization generated by cross-hybridization as a signature to classify the target genes. A powerful addition was the use of hierarchical clustering that proved to reliably group identical sequences from biologically independent experiments and very close sequences, demonstrating the efficacy of the microarray for gene typing. As expected, the diagnosis was improved when two protein motifs were used instead of one, indicating that the more motifs—from one or several genes—that are present on the array, the stronger the diagnosis is. In the future, a library of hybridization data can be created for nodC or other genes of interest that can be enriched in the course of time and used for detection and typing.

Rhizobia are soil bacteria that intracellularly colonize the nodules they induce on the roots of their host plants. We showed that nodC can be typed directly from bacterial DNA extracted from a single infected legume nodule, thus avoiding bias and limitations in bacterial isolation and cultivation. Microarray-based nodC typing from both free-living bacteria and infected plant tissues obviates the need for DNA amplification that is required for the widely used PCR-restriction fragment length polymorphism technology. It will thus be of great value for studies of the molecular ecology of rhizobia, including geographical distribution of nodC alleles, plant-rhizobia coevolution studies (1), and coinoculation experiments (19). Moreover, direct gene detection in legume nodules would be of special interest for field experiments. The natural diversity of rhizobia, which has recently been extended from α- to β-proteobacteria (8, 21), is thought to be still largely underestimated, since the symbionts of <10% of the legumes have been investigated so far. Direct typing of plant intracellular bacteria will also help in confirming the rhizobium status of atypical bacteria by comparing hybridization patterns of an infected nodule with the corresponding isolate culture.

Bacterial genomes consist of a basic genome carrying all the essential genetic information and an additional genome necessary for niche adaptation (11). Most functions of interest or of concern belong to the accessory genome, such as virulence, toxin production, antibiotic resistance, nitrogen cycle, metabolite production, heavy metal resistance, biodegradation, methylotrophy, and symbiosis. While the basic genome is conserved between related species, the adaptative genome, which is often horizontally acquired and can be lost (22), is highly variable from one species to the other and sometimes from one strain to an other (6). Both taxonomical and functional diagnoses are crucial for full bacterial identification. The nodC prototype can be extended simultaneously to many different genes, improving the robustness of the bacterial microarray signature. A microarray that represents both housekeeping genes and genes encoding functions of special interest has the potential to be a powerful diagnostic tool for bacteria in many fields such as environment, health, and industry.

Supplementary Material

[Supplemental material]


We thank Jacques Batut for helpful discussions, Michelle Glew for English corrections, and Andreas Squartini for providing Pseudomonas sp. and Escherichia sp. strains. We thank the Toulouse and Jouy en Josas Génopoles for microarray facilities.

This research was supported by the Toulouse Génopole and the Bureau des Ressources Génétiques. C.B. was supported by a fellowship (MRT French government and Toulouse Génopole).


Supplemental material for this article may be found at http://aem.asm.org/.


1. Aguilar, O. M., O. Riva, and E. Peltzer. 2004. Analysis of Rhizobium etli and of its symbiosis with wild Phaseolus vulgaris supports coevolution in centers of host diversification. Proc. Natl. Acad. Sci. USA 101:13548-13553. [PMC free article] [PubMed]
2. Bekal, S., R. Brousseau, L. Masson, G. Prefontaine, J. Fairbrother, and J. Harel. 2003. Rapid identification of Escherichia coli pathotypes by virulence gene detection with DNA microarrays. J. Clin. Microbiol. 41:2113-2125. [PMC free article] [PubMed]
3. Benhizia, Y., H. Benhizia, A. Benguedouar, R. Muresu, A. Giacomini, and A. Squartini. 2004. Gamma proteobacteria can nodulate legumes of the genus Hedysarum. Syst. Appl. Microbiol. 27:462-468. [PubMed]
4. Bodrossy, L., and A. Sessitsch. 2004. Oligonucleotide microarrays in microbial diagnostics. Curr. Opin. Microbiol. 7:245-254. [PubMed]
5. Cabanes, D., P. Boistard, and J. Batut. 2000. Identification of Sinorhizobium meliloti genes regulated during symbiosis. J. Bacteriol. 182:3632-3637. [PMC free article] [PubMed]
6. Chan, K., S. Baker, C. C. Kim, C. S. Detweiler, G. Dougan, and S. Falkow. 2003. Genomic comparison of Salmonella enterica serovars and Salmonella bongori by use of an S. enterica serovar Typhimurium DNA microarray. J. Bacteriol. 185:553-563. [PMC free article] [PubMed]
7. Chen, W. M., S. Laevens, T. M. Lee, T. Coenye, P. De Vos, M. Mergeay, and P. Vandamme. 2001. Ralstonia taiwanensis sp. nov., isolated from root nodules of Mimosa species and sputum of a cystic fibrosis patient. Int. J. Syst. Evol. Microbiol. 51:1729-1735. [PubMed]
8. Chen, W. M., L. Moulin, C. Bontemps, P. Vandamme, G. Bena, and C. Boivin-Masson. 2003. Legume symbiotic nitrogen fixation by β-proteobacteria is widespread in nature. J. Bacteriol. 185:7266-7272. [PMC free article] [PubMed]
9. Chen, W. P., and T. T. Kuo. 1993. A simple and rapid method for the preparation of gram-negative bacterial genomic DNA. Nucleic Acids Res. 21:2260. [PMC free article] [PubMed]
10. Cho, J. C., and J. M. Tiedje. 2001. Bacterial species determination from DNA-DNA hybridization by using genome fragments and DNA microarrays. Appl. Environ. Microbiol. 67:3677-3682. [PMC free article] [PubMed]
11. Dobrindt, U., and J. Hacker. 2001. Whole genome plasticity in pathogenic bacteria. Curr. Opin. Microbiol. 4:550-557. [PubMed]
12. Geremia, R. A., P. Mergaert, D. Geelen, M. Van Montagu, and M. Holsters. 1994. The NodC protein of Azorhizobium caulinodans is an N-acetylglucosaminyltransferase. Proc. Natl. Acad. Sci. USA 91:2669-2673. [PMC free article] [PubMed]
13. Jenkins, B. D., G. F. Steward, S. M. Short, B. B. Ward, and J. P. Zehr. 2004. Fingerprinting diazotroph communities in the Chesapeake Bay by using a DNA macroarray. Appl. Environ. Microbiol. 70:1767-1776. [PMC free article] [PubMed]
14. Koide, T., P. A. Zaini, L. M. Moreira, R. Z. Vencio, A. Y. Matsukuma, A. M. Durham, D. C. Teixeira, H. El-Dorry, P. B. Monteiro, A. C. da Silva, S. Verjovski-Almeida, A. M. da Silva, and S. L. Gomes. 2004. DNA microarray-based genome comparison of a pathogenic and a nonpathogenic strain of Xylella fastidiosa delineates genes important for bacterial virulence. J. Bacteriol. 186:5442-5449. [PMC free article] [PubMed]
15. Le Berre, V., E. Trevisiol, A. Dagkessamanskaia, S. Sokol, A.-M. Caminade, J. P. Majoral, B. Meunier, and J. François. 2003. Dendrimeric coating of glass slides for sensitive DNA microarrays analysis. Nucleic Acids Res. 31:e88. [PMC free article] [PubMed]
16. Lockhart, D. J., and E. A. Winzeler. 2000. Genomics, gene expression and DNA arrays. Nature 405:827-836. [PubMed]
17. Loy, A., C. Schulz, S. Lücker, A. Schöpfer-Wendels, K. Stoecker, C. Baranyi, A. Lehner, and M. Wagner. 2005. 16S rRNA gene-based oligonucleotide microarray for environmental monitoring of the betaproteobacterial Order “Rhodocyclales.” Appl. Environ. Microbiol. 71:1373-1386. [PMC free article] [PubMed]
18. Miller, N. A., Q. Gong, R. Bryan, M. Ruvolo, L. A. Turner, and S. T. LaBrie. 2002. Cross-hybridization of closely related genes on high-density macroarrays. BioTechniques 32:620-625. [PubMed]
19. Moawad, H., W. M. Abd El-Rahim, and D. Abd El-Haleem. 2004. Performance of phaseolus bean rhizobia in soils from the major production sites in the Nile Delta. C. R. Biol. 327:445-453. [PubMed]
20. Moulin, L., G. Bena, C. Boivin-Masson, and T. Stepkowski. 2004. Phylogenetic analyses of symbiotic nodulation genes support vertical and lateral gene co-transfer within the Bradyrhizobium genus. Mol. Phylogenet. Evol. 30:720-732. [PubMed]
21. Moulin, L., A. Munive, B. Dreyfus, and C. Boivin-Masson. 2001. Nodulation of legumes by members of the beta-subclass of Proteobacteria. Nature 411:948-950. [PubMed]
22. Ochman, H., and N. A. Moran. 2001. Genes lost and genes found: evolution of bacterial pathogenesis and symbiosis. Science 292:1096-1099. [PubMed]
23. Parniske, M., and J. A. Downie. 2003. Plant biology: locks, keys and symbioses. Nature 425:569-570. [PubMed]
24. Perret, X., C. Staehelin, and W. J. Broughton. 2000. Molecular basis of symbiotic promiscuity. Microbiol. Mol. Biol. Rev. 64:180-201. [PMC free article] [PubMed]
25. Rivas, R., E. Velázquez, A. Willems, N. Vizcaíno, N. S. Subba-Rao, P. F. Mateos, M. Gillis, F. B. Dazzo, and E. Martínez-Molina. 2002. A new species of Devosia that forms a unique nitrogen-fixing root-nodule symbiosis with the aquatic legume Neptunia natans (L.f.) Druce. Appl. Environ. Microbiol. 68:5217-5222. [PMC free article] [PubMed]
26. Rose, T. M., J. G. Henikoff, and S. Henikoff. 2003. CODEHOP (COnsensus-DEgenerate Hybrid Oligonucleotide Primer) PCR primer design. Nucleic Acids Res. 31:3763-3766. [PMC free article] [PubMed]
27. Sawada, H., L. D. Kuykendall, and J. M. Young. 2003. Changing concepts in the systematics of bacterial nitrogen-fixing legume symbionts. J Gen. Appl. Microbiol. 49:155-179. [PubMed]
28. Sougakoff, W., M. Rodrigue, C. Truffot-Pernot, M. Renard, N. Durin, M. Szpytma, R. Vachon, A. Troesch, and V. Jarlier. 2004. Use of a high-density DNA probe array for detecting mutations involved in rifampicin resistance in Mycobacterium tuberculosis. Clin. Microbiol. Infect. 10:289-294. [PubMed]
29. Spaink, H. P., A. H. Wijfjes, K. M. van der Drift, J. Haverkamp, J. E. Thomas-Oates, and B. J. Lugtenberg. 1994. Structural identification of metabolites produced by the NodB and NodC proteins of Rhizobium leguminosarum. Mol. Microbiol. 13:821-831. [PubMed]
30. Taroncher-Oldenburg, G., E. M. Griner, C. A. Francis, and B. B. Ward. 2003. Oligonucleotide microarray for the study of functional gene diversity in the nitrogen cycle in the environment. Appl. Environ. Microbiol. 69:1159-1171. [PMC free article] [PubMed]
31. Tillib, S. V., and A. D. Mirzabekov. 2001. Advances in the analysis of DNA sequence variations using oligonucleotide microchip technology. Curr. Opin. Biotechnol. 12:53-58. [PubMed]
32. Tremousaygue, D., L. Garnier, C. Bardet, P. Dabos, C. Herve, and B. Lescure. 2003. Internal telomeric repeats and ‘TCP domain’ protein-binding sites co-operate to regulate gene expression in Arabidopsis thaliana cycling cells. Plant J. 33:957-966. [PubMed]
33. Van Berkhum, P., R. E. Tully, and D. L. Keister. 1995. Nonpigmented and bacteriochlorophyll-containing bradyrhizobia isolated from Aeschynomene indica. Appl. Environ. Microbiol. 61:623-629. [PMC free article] [PubMed]
34. Vincent, J. M. 1970. A manual for the practical study of root-nodule bacteria. Blackwell Scientific Publications, Oxford, United Kingdom.
35. Wang, D., L. Coscoy, M. Zylberberg, P. C. Avila, H. A. Boushey, D. Ganem, and J. L. DeRisi. 2002. Microarray-based detection and genotyping of viral pathogens. Proc. Natl. Acad. Sci. USA 99:15687-15692. [PMC free article] [PubMed]
36. Wang, D., A. Urisman, Y. T. Liu, M. Springer, T. G. Ksiazek, D. D. Erdman, E. R. Mardis, M. Hickenbotham, V. Magrini, J. Eldred, J. P. Latreille, R. K. Wilson, D. Ganem, and J. L. DeRisi. 2003. Viral discovery and sequence recovery using DNA microarrays. PLoS Biol. 1:E2. [PMC free article] [PubMed]
37. Wilson, K. H., W. J. Wilson, J. L. Radosevich, T. Z. DeSantis, V. S. Viswanathan, T. A. Kuczmarski, and G. L. Andersen. 2002. High-density microarray of small-subunit ribosomal DNA probes. Appl. Environ. Microbiol. 68:2535-2541. [PMC free article] [PubMed]
38. Wong, C. W., T. J. Albert, V. B. Vega, J. E. Norton, D. J. Cutler, T. A. Richmond, L. W. Stanton, E. T. Liu, and L. D. Miller. 2004. Tracking the evolution of the SARS coronavirus using high-throughput, high-density resequencing arrays. Genome Res. 14:398-405. [PMC free article] [PubMed]
39. Zhou, J. 2003. Microarrays for bacterial detection and microbial community analysis. Curr. Opin. Microbiol. 6:288-294. [PubMed]

Articles from Applied and Environmental Microbiology are provided here courtesy of American Society for Microbiology (ASM)
PubReader format: click here to try


Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...


  • Gene (nucleotide)
    Gene (nucleotide)
    Records in Gene identified from shared sequence links
  • MedGen
    Related information in MedGen
  • Nucleotide
    Published Nucleotide sequences
  • Protein
    Published protein sequences
  • PubMed
    PubMed citations for these articles

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...