Analysis of Stable Low-Molecular-Weight RNA Profiles of Members of the Family Rhizobiaceae

Abstract

Staircase electrophoresis in polyacrylamide gels was used to analyze the stable low-molecular-weight (LMW) RNA profiles of 24 type strains belonging to the family Rhizobiaceae. This new electrophoretic technique results in good separation of the molecules forming the LMW RNA profiles. Differences in the number and distribution of the RNA bands in these profiles allowed us to identify differences among the 24 strains assayed. Species assignments based on LMW RNAs proved to be consistent with the established taxonomic classification. Analysis of the data obtained and the corresponding dendrograms revealed relationships between genera and species; these relationships were essentially the same as those obtained with other techniques, such as DNA hybridization and 16S rRNA sequencing. Use of the technique described here, with which it is possible to analyze a large number of strains in a short time, permits rapid identification of species belonging to the family Rhizobiaceae and should in the future facilitate biodiversity studies and detection of new species.

From 1929, when Baldwin and Fred (1) proposed that rhizobia should be classified and identified on the basis of cross-inoculation groups, until 1974, when Bergey’s Manual of Determinative Bacteriology, 8th ed. (1b), was published, the number of species in the family Rhizobiaceae remained almost the same. The situation began to change in 1984, when Jordan (11) published a reclassification of Rhizobium species and added the new genus Bradyrhizobium, which includes slowly growing species. From 1988 to 1996 the number of nodule-forming species gradually rose, and two new genera, the genera Azorhizobium (7) and Sinorhizobium (4), were described during this period. The description of the genus Sinorhizobium in 1988 (4), in which Rhizobium fredii was reclassified as Sinorhizobium fredii, was the start of species reclassification. This reclassification affected the organisms described before 1984 as Rhizobium meliloti, for which the name Sinorhizobium meliloti was proposed (15), and was the beginning of the description of several new species in the genus Sinorhizobium, which today also includes Sinorhizobium xinjianensis (4), Sinorhizobium teranga (15), Sinorhizobium saheli (15), and Sinorhizobium medicae (21). Within the genus Rhizobium the new species described during or after 1984 include R. fredii (23), which was later reclassified as S. fredii, Rhizobium galegae (16), Rhizobium huakuii (3), Rhizobium etli (24), Rhizobium tropici (17), Rhizobium thiansanense (2), Rhizobium ciceri (19), and Rhizobium mediterraneum (18). Within the genus Bradyrhizobium the number of species increased from one to three with the descriptions of Bradyrhizobium elkanii (13) and Bradyrhizobium liaoningense (29). Also, in 1996 Young (31) discussed the proposal to create a new genus called Mesorhizobium, which encompasses some species previously considered members of the genus Rhizobium (Rhizobium loti, R. huakuii, R. thiansanense, R. ciceri, and R. mediterraneum). In view of this background, the number of species is expected to increase with analyses of more strains of rhizobia isolated from previously studied legumes and other legumes that have not been studied.

This situation contrasts with the situation found in the other two genera of the family Rhizobiaceae, the genera Phyllobacterium and Agrobacterium, whose species numbers have remained fairly stable even though a species reclassification has been proposed in the case of the genus Agrobacterium (12, 22) and a new species, Agrobacterium vitis, has been described (20).

Hybridization of nucleic acids has been used for describing new species together with rRNA sequencing, which has also been used in phylogenetic studies. However, these techniques are of little use for identification of microorganisms because they cannot be routinely applied to a large number of strains. This highlights the need for rapid methods of identification for these microorganisms and prompted a search for new techniques that can be applied to the identification of rhizobia (6, 14).

In 1990 Höfle proposed that tRNA profiles, which are part of the low-molecular-weight (LMW) RNA profiles, could be used as a way to fingerprint bacteria (10).

The development of a new electrophoretic technique, staircase electrophoresis, and its use to separate the molecules making up LMW RNA profiles have resulted in optimal separation of tRNA molecules, making them more useful in taxonomic studies (5). The strains used in this study were the reference strains of the species currently accepted as members of the family Rhizobiaceae in the American Type Culture Collection and the U. S. Department of Agriculture collections. These strains included Agrobacterium radiobacter ATCC 19358, Agrobacterium rhizogenes ATCC 13332, Agrobacterium rubi ATCC 13335, Agrobacterium tumefaciens ATCC 23308, Azorhizobium caulinodans ATCC 43989, B. elkanii ATCC 49852, Bradyrhizobium japonicum ATCC 10324, B. liaoningense USDA 3622, Phyllobacterium myrsinacearum ATCC 43590, R. ciceri USDA 3383, R. etli ATCC 51251, R. galegae ATCC 43677, R. huakuii USDA 4779, Rhizobium leguminosarum biovar viceae ATCC 10004, R. leguminosarum biovar trifolii ATCC 14480, R. leguminosarum biovar phaseoli ATCC 14482, R. loti ATCC 3669, R. mediterraneum USDA 3392, R. meliloti USDA 1002, R. thiansanense USDA 3592, R. tropici USDA 9030, S. fredii ATCC 35423, S. saheli USDA 4102, and S. teranga USDA 4101.

The following commercially prepared molecules obtained from Boehringer (Mannheim, Germany) and Sigma Chemical Co. (St. Louis, Mo.) were used as reference molecules: 5S rRNA from Escherichia coli MRE 600 (120 and 115 nucleotides) (1a), tRNA specific for tyrosine from E. coli (85 nucleotides), and tRNA specific for valine from E. coli (77 nucleotides) (25). Samples were prepared as reported elsewhere (5).

The RNAs of the strains studied were extracted by the method described by Höfle (9). LMW RNA profiles were obtained by staircase electrophoresis with a 14% polyacrylamide gel under denaturing conditions, with the voltage increasing in 50-V steps (10 min each) from 100 to 2,300 V, as described previously (5).

After electrophoresis, the gels were silver stained by the method described by Haas et al. (8).

The bands present in each profile obtained were coded for input into a database that included all of the strains studied, and Jaccard’s similarity coefficients were calculated to construct a distance matrix. A dendrogram was constructed from the distance matrix by using the unweighted pair group method with arithmetic means (UPGMA).

The LMW RNA profiles of the strains analyzed are shown in Fig. 1. All of these profiles contain the three zones expected, 5S rRNA, class 2 tRNA, and class 1 tRNA (5). The number of bands present in the tRNA zone was the expected number based on previous results obtained for E. coli CECT99 (= ATCC 9637) with staircase electrophoresis (5).

FIG. 1.

LMW RNA profiles of the reference strains of species belonging to the family Rhizobiaceae. (A) Lane 1, B. japonicum ATCC 10324; lane 2, B. elkanii ATCC 49852; lane 3, B. liaoningense USDA 3622; lane 4, Azorhizobium caulinodans ATCC 43989. (B) Lane 5, Agrobacterium rhizogenes ATCC 13332; lane 6, Agrobacterium radiobacter ATCC 19358; lane 7, Agrobacterium tumefaciens ATCC 23308; lane 8, Agrobacterium rubi ATCC 13335; lane 9, P. myrsinacearum ATCC 43590. (C) Lane 10, S. fredii ATCC 35423; lane 11, S. saheli USDA 4102; lane 12, S. teranga USDA 4101. (D) Lane 13, R. galegae ATCC 43677; lane 14, R. tropici USDA 9030; lane 15, R. etli ATCC 51251; lane 16, R. leguminosarum biovar phaseoli ATCC 14482; lane 17, R. leguminosarum biovar trifolii ATCC 14480; lane 18, R. leguminosarum biovar viceae ATCC 10004; lane 19, R. thiansanense USDA 3592. (E) Lane 20, R. huakuii USDA 4779; lane 21, R. meliloti USDA 1002; lane 22, R. loti ATCC 3669; lane 23, R. ciceri USDA 3383; lane 24, R. mediterraneum USDA 3392. Lanes MW contained molecular weight markers. nt, nucleotides.

The results obtained in this study reveal differences at the 5S rRNA level among the genera included in the family Rhizobiaceae. Thus, the genera Azorhizobium (Fig. 1A, lane 4), Bradyrhizobium (Fig. 1A, lanes 1 to 3), Sinorhizobium (Fig. 1C, lanes 10 to 12), Rhizobium (Fig. 1D, lanes 13 to 24), Agrobacterium (Fig. 1B, lanes 5 to 8), and Phyllobacterium (Fig. 1B, lane 9) produced characteristic profiles in this zone that allowed them to be differentiated from one another. This is consistent with the findings reported by other authors who studied other microbial groups (9, 10) and who, after analyzing the LMW RNA profiles of strains belonging to different genera, found that either the numbers of bands in the 5S rRNA migration zone differed or the bands differed in molecular weight (9, 10). This indicates that at this level (5S rRNA) it is possible to establish differences among genera. We found that the species B. liaoningense (Fig. 1, lane 3) has a 5S rRNA profile different from that of other bradyrhizobia. This means that assignment of this species to the genus Bradyrhizobium should be revised.

Moreover, the results obtained for fast-growing rhizobia seem to be consistent with the proposal supported by Young (31) that there is a new genus different from the genera Rhizobium and Sinorhizobium, namely, the genus Mesorhizobium. As indicated by Young, this genus should include R. loti, R. huakuii, R. thiansanense, R. ciceri, and R. mediterraneum. All of the species included in this genus except R. thiansanense produce identical bands in the 5S rRNA zone, but these bands are different from the bands produced by species belonging to the genera Rhizobium and Sinorhizobium (Fig. 1D). Based on our results, the reference strain of R. meliloti used in this study (USDA 1002) should be included in the genus Mesorhizobium and not in the genus Sinorhizobium, a conclusion that is clearly supported by the data obtained for the 5S rRNA zone (Fig. 1D). Nevertheless, other isolates obtained from Melilotus, Medicago, or Trigonella species can be included in the genus Sinorhizobium or other genera (27). This is true for the strains used in other studies (4).

According to our results, there are species level differences in the tRNA profiles, specifically the class 2 tRNA profiles (Fig. 1). Our findings are consistent with those reported by other authors for other microbial groups (9, 10).

According to our findings, class 1 tRNAs are not necessary to establish differences among the strains studied at the species level. The relative amounts of these tRNAs are greater than the relative amounts of class 2 tRNAs, which leads to overlapping of some bands. However, we believe that class 1 tRNAs could be useful in biodiversity studies when it is necessary to analyze a larger number of strains belonging to each species.

To analyze all of the results obtained, a database was compiled by coding the bands present in the LMW RNA profiles obtained and using Jaccard’s coefficient and the UPGMA method of dendrogram construction to obtain a distance matrix for the species in the study. The results are shown in Fig. 2. As Fig. 2 shows, two groups with a similarity coefficient of only 0.13 were distinguished; one of these groups contains Bradyrhizobium, and the other contains the rest of the genera studied.

FIG. 2.

UPGMA dendrogram based on Jaccard’s coefficients derived from LMW RNA profile characteristics for reference strains of species belonging to the family Rhizobiaceae. R. leg., R. leguminosarum.

In agreement with previous studies (7, 14, 29–31), the group that is most distant from the rest of the organisms is composed of the Bradyrhizobium species that exhibit interspecies variability in the 5S rRNA zone such that assignment to the same genus is not unambiguous; this is especially true for B. liaoningense, whose special characteristics as an extraslow grower have been described previously (29). This species can also be distinguished on the basis of 5S rRNA data.

The second group is separated into two subgroups with a similarity coefficient of 0.19; one subgroup contains Azorhizobium, and the other subgroup encompasses two large clusters with a similarity coefficient of 0.23. One of these clusters contains Agrobacterium and Phyllobacterium, and the other contains Rhizobium and Sinorhizobium. These results, obtained by using LMW RNA profiles, are essentially in agreement with previous observations (28, 30, 31).

According to our results, Azorhizobium should also be separated from the other species of the family and is more closely related to Bradyrhizobium than to the faster-growing species. This is consistent with the reports of other authors (7, 28).

In the present study, the genus Phyllobacterium clustered with the genus Agrobacterium. The results of other authors (30, 31) show that the genus Phyllobacterium is closely related to Agrobacterium, although it also groups with some fast-growing species of rhizobia.

In the genus Agrobacterium, it is possible to define two groups with a relatively low similarity coefficient (0.65); one group includes Agrobacterium rhizogenes, and the other includes Agrobacterium rubi, Agrobacterium tumefaciens, and Agrobacterium radiobacter (Fig. 2). This is consistent with the findings of other authors (12, 22) since Agrobacterium radiobacter and Agrobacterium tumefaciens are closely related.

It should be noted that some authors have shown that Agrobacterium rhizogenes clusters with species of the genus Rhizobium (30, 31). However, the strain used in the present study, Agrobacterium rhizogenes ATCC 13332, produces the typical LMW RNA profile of the genus Agrobacterium. In view of this, we also analyzed the profile of strain ATCC 11325 (data not shown). We found that the LMW RNA profile of this strain corresponds to the typical profile of the genus Rhizobium and not the profile of the genus Agrobacterium, both in the 5S rRNA zone and in the tRNA zone, and is similar to the LMW RNA profile of R. tropici. This result seems to agree with the results of other authors (30, 31), so that its importance at the phylogenetic level is debatable since the data suggest that a bacterium with a Rhizobium chromosomal background behaves like an Agrobacterium, probably because of the presence of the Ti plasmid. This is not surprising if it is remembered that this plasmid has been successfully transferred to R. leguminosarum biovar trifolii to produce strains able to produce the root proliferation syndrome and to form nodules in clovers (26).

Species of the genera Rhizobium and Sinorhizobium are distributed in four groups (Fig. 2). One of these groups includes most Rhizobium species, another includes the Sinorhizobium species, another includes R. galegae, and another includes the species included in the genus Mesorhizobium, in agreement with the proposal discussed by Young. As stated above, R. meliloti USDA 1002 is a member of the last group, while R. thiansanense is in the same group as the genus Sinorhizobium.

According to our results, the genus Rhizobium seems to include only species isolated from hosts belonging to the R. leguminosarum cross-inoculation group; there are differences among the biovars of R. leguminosarum, and this casts doubt on the classification of all of the strains belonging to the three biovars in a single species since the reference strains have a similarity coefficient of only 0.6. Despite this, there may be strains of the three biovars that do belong to a single species, R. leguminosarum. This would have to be confirmed in exhaustive studies of large numbers of strains isolated from the inoculation group hosts, which with the technique described here should prove to be relatively easy and rapid.

The profile of R. galegae in both the 5S rRNA zone and the tRNA zones differs considerably from the profiles of the rest of the reference strains used; the differences are great enough that this taxon could be included in a different genus because it has a low similarity coefficient with the rest of the species included in the genera containing fast-growing rhizobia. This is consistent with the results reported by other authors (31).

These findings emphasize the necessity of having safe criteria available for defining the species in the family Rhizobiaceae; until now symbiotic and plasmid-encoded pathogenicity characteristics have been used to define species, and in many cases these characteristics do not reflect the reality of the taxonomic structure of the group, as pointed out by other authors (12, 22).

In this sense, because of their characteristics and diversity, LMW RNAs may help define species in an easy, rapid, and safe manner. In light of the results obtained here it was concluded that staircase electrophoresis is useful for separating the molecules in LMW RNA profiles of species belonging to the family Rhizobiaceae. The results obtained show that the number and arrangement of the bands in these profiles can be used to distinguish all of the reference strains of the species employed. Species assignments in this group can be confirmed by LMW RNA profiles, and the results based on analyses of the data obtained are in agreement with the results obtained from DNA hybridization and 16S rRNA sequencing analyses.

Since the proposed technique is rapid and easy to perform, in the future studies of large numbers of samples should allow us to rapidly identify many strains, not only strains of species that form nodules in legumes but also strains of species that share the plant rhizosphere with them. This should enable workers to perform biodiversity and environmental impact studies, which are of great interest because of the ecological and economic importance of legumes.