Phenotypic and molecular differences among rhizobia that nodulate Phaseolus lunatus in the Supe valley in Peru

Abstract

A collection of 16 bacterial strains isolated from root nodules of Lima bean (Phaseolus lunatus L.) in the Supe valley of Peru were characterised using phenotypic and molecular methods. The isolates were clustered into fast-growing, alkalinising and extra-alkalinising slow-growing isolates with marked morphological differences, according to the rate of growth and alkalinisation of yeast extract mannitol (YEM) medium. Fast-growing isolates were salt tolerant and sensitive to 40 °C, while alkalinising slow-growing isolates behaved oppositely. Extra-alkalinising slow-growing isolates were sensitive to 8 °C and 40 °C and tolerated 1 % NaCl. Fast-growing isolates also showed higher indole-3-acetic acid (IAA) production and tri-calcium phosphate solubilisation than the alkalinising slow-growing isolates. Half of the isolates were able to nodulate Phaseolus vulgaris and Vigna unguiculata. Individually, BOX-PCR, ERIC-PCR, and REP-PCR fingerprints patterns resulted in ten profiles from 16 isolates and they were clustered into three profile groups that correspond to the clusters obtained by YEM medium alkalinisation. Analysis of 16S rRNA gene sequences revealed that fast-growing isolates showed 99.7 % sequence identity with Rhizobium mesosinicum CCBAU 25010^T and Rhizobium alamii GBV016^T. Alkalinising slow-growing isolates were related to both Bradyrhizobium yuanmingense CCBAU10071^T and Bradyrhizobium liaoningense 2281^T, with 99.8 % sequence identity, and extra-alkalinising slow-growing isolates had 100 % sequence identity to both Bradyrhizobium paxllaeri LMTR 21^T and Bradyrhizobium icense LMTR 13^T.

Findings

Lima bean or pallar (Phaseolus lunatus L.) is the second most socioeconomically important crop among the Phaseolus species. Peru is one of the centers of Andean origin and diversity of the Lima bean, where this legume has been domesticated and cultivated since pre-Hispanic times. Cultivated Peruvian seed varieties are large, flat and white and belong to the Big Lima culti group (Fofana et al. 1997). It is known that Lima bean can also fix nitrogen in symbiosis with slow-growing rhizobial species from the Bradyrhizobium genus, like B. yuanmingense, B. paxllaeri and B. icense (Vinuesa et al. 2005; Ormeño-Orrillo et al. 2006; Durán et al. 2014), while Ensifer meliloti LMTR32 (formerly Sinorhizobium meliloti) is the only fast-growing rhizobia strain reported to nodulate P. lunatus (Ormeño et al. 2007).

Given the importance of P. lunatus in sustainable agriculture and the development of Peru, and the scarce knowledge about rhizobial specificity for this legume, we decided to study rhizobia naturally associated with this legume in local farms of the Supe valley, a recognized P. lunatus farming area with no records of rhizobia isolations.

Sixteen bacterial strains were used in this work. They were isolated from independent nodules collected in three fields of P. lunatus culti group Big Lima var. Criollo, in the Supe valley at the north-central coast of Peru. The sampling sites were Tutumo-1 (S 10° 51’ 13.4” WO 77° 38’ 27.0”), Venturosa-1 (S 10° 51’ 08.9” WO 77° 38’ 20.4”), and Venturosa-3 (S 10° 51’ 33.1” WO 77° 38’ 20.6”) (Table 1). Soils at collection sites had low macronutrient content and were sandy, slightly saline and moderately alkaline. The typical weather was subtropical arid, with temperatures between 15-26 °C and precipitation levels between 0 and 3.2 mm. Reference strains of Bradyrhizobium were also included for analysis. All strains were maintained in 25 % (v/v) glycerol in yeast extract mannitol (YEM) broth at −80 °C.

Table 1 Geographical origin, symbiotic and main phenotypic characteristics of strains analysed in this study

Strains were Gram stained. Colony morphology and acid/alkaline reaction were evaluated on plates of yeast extract mannitol (YEM) solid medium containing 0.025 % (v/v) bromothymol blue (BTB) at 28 °C (Vincent 1970). Five microliters of bacterial cultures (10⁸cell/ml) were inoculated onto plates of YEM solid medium with 0.5, 1 and 2 % (w/v) NaCl, or with pH adjusted to 4, 5, 8 and 8.8, respectively (Matos et al. 1998). In addition, cultures were inoculated onto YEM plates that were kept at 8, 37 and 40 °C, respectively. Most of the plates were incubated from 2 to7 days, and those that were kept at 8 °C were evaluated up to 30 days. Additionally, all the strains that were used were grown on YEM solid medium with pH 7 and 0.02 % (w/v) NaCl at 28 °C, and this condition was considered as a standard control (Vincent 1970).

Evaluation of phosphate solubilisation was conducted according to Nautiyal (1999) on NBRIP plates with bi-calcium or tri-calcium phosphate. Plates were inoculated with 5 μl of YEM bacterial cultures (10⁸ cell/ml) and incubated at 28 °C. The production of a halo of phosphate solubilisation around the colonies was recorded for up to 30 days, and halo size was calculated according to Nautiyal (1999).

For production of indole-3-acetic acid (IAA), strains were grown in YEM broth supplemented with 5 mM L-tryptophan at 28 °C for 2 to 7 days (Glickmann and Dessaux 1995). Then they were centrifuged (1,200 rpm, 5 min) and supernatants were mixed in a 1:5 ratio with Salkowski reagent (10 mM FeCl₃ in 36 % H₂SO₄) and left in the dark for 15 min at room temperature (Gordon and Weber 1951). Absorbance was determined on a spectrophotometer at 530 nm and concentration was calculated with an IAA standard curve. Non-inoculated broth was used as a negative control.

Seeds of P. lunatus cv-gr. Sieva var. UNALM-1, Vigna unguiculata var. Vaina-Blanca and P. vulgaris var. Canario-Centenario were used for experiments. Axenically grown surface-sterilized seedlings were transferred to 50-ml glass containers filled with polypropylene pellets and watered with N-free nutrient solution (Broughton and Dilworth 1971). Each plant was inoculated with one milliliter of a bacterial suspension (10⁸ cell/ml) of each strain. Inoculated plants were placed in a growth chamber with controlled temperature (22–25 °C) and light cycle (13 h/day) for 30 days (P. lunatus) and 45 days (P. vulgaris and V. unguiculata) . Experiments were performed in triplicates per bacterial strain per plant. Results were recorded as presence or absence of nodulation per plant, and a control with no inoculation was used.

DNA was extracted from bacterial colonies. Colonies were suspended in 20 μl lysis buffer (0.05 M NaOH, 0.25 % SDS) and incubated for 10 min at 80 °C and 10 min in an ice bath. The lysate was diluted in 200 μl Milli-Q sterile water and spun down in a microcentrifuge for 5 minutes. Five microliters of the supernatant were used for PCR amplification of BOX-PCR, ERIC-PCR, and REP-PCR genomic fingerprints (Versalovic et al. 1991, 1994). DNA fragments were separated in a 1.5 % agarose gel. Genomic fingerprints were visually recorded, relationships were analysed by the UPGMA method, and distances were calculated with the Jaccard coeficcient using NTSYS-PC 2.0 software (Rohlf 1998).

Primers fD1 and rD1 were used to amplified 16S rRNA (Weisburg et al. 1991). Phylogenetic neighbors of the isolates were identified using the EzTaxon server (http://www.ezbiocloud.net/eztaxon; Kim et al. 2012) on the basis of 16S rRNA sequence data. Both studied and selected sequences were aligned by the Clustal X2 software (Larkin et al. 2007). Phylogenetic analysis was performed by the Neighbor-Joining (NJ) method and distances were calculated according to the Kimura-2 method using the MEGA 6 package (Tamura et al. 2013).

The 16 isolated strains were Gram-negative rods. Isolates grew in 2 days (fast-growing) and 5–7 days (slow-growing) (Vincent 1970). They were divided into two fast-growing isolates, ten alkalinising slow-growing (SG) isolates and four extra-alkalinising SG isolates, according to the colour change of the pH-indicator dye in YEM–BTB medium from green to yellow, green-blue and blue, respectively (Table 1). Additionally, fast-growing isolates had colonies of 4–6 mm with abundant and soft gum. Alkalinising SG isolates showed 1–2 mm colonies with regular and hard gum, while extra-alkalinising SG isolates had 0.5-1 mm colonies with no gum. The fast-growing and extra-alkalinising SG isolates shared the same sample site, in contrast to alkalinising SG isolates that were isolated from different sites.

All isolates grew at 28 °C and 37 °C, 0.5 % NaCl and at an acid pH (4 and 5) and an alkaline pH (8 and 8.8). All fast-growing and extra-alkalinising SG isolates failed to grow at 40 °C. No extra-alkalinising SG isolates grew at 8 °C and the others isolates grew weakly. All isolates grew with 0.5 % (w/v) NaCl, while fast-growing and extra-alkalinising SG isolates grew with 1 % (w/v) NaCl. In the presence of 2 % (w/v) NaCl, fast-growing isolates showed poor growth (Table 1). Salt tolerance varies among rhizobial species, and it is still a useful characteristic for distinguishing related species. Some bradyrhizobia related to P. lunatus, like B. yuanmingense, do not tolerate 1 % (w/v) NaCl (Yao et al. 2002), while others, such as B. paxllaeri and B. icense, can (Durán et al. 2014).

Fast-growing isolates produced 35–54 μg IAA/ml, which is approximately six times more IAA than the amount produced by alkalinising SG isolates (Table 1). Extra-alkalizing SG isolates produced less than 0.5 μg/ml of this auxine. IAA production has been observed in many strains of rhizobial species (Boiero et al. 2007). Solubilisation of phosphate was only detected when tri-calcium phosphate was used as substrate. Solubilisation halos were produced by fast-growing isolates after 10 days of incubation (Table 1), whilst some alkalinising SG isolates produced a halo only after 30 days of incubation. Nonetheless, phosphate solubilisation detection based on halo formation on agar plates is not an infallible technique, since many isolates that do not show any halo can solubilise insoluble phosphates in liquid medium (Nautiyal 1999). In addition, reference strain Bradyrhizobium elkanii USDA76 produced higher levels of IAA (20 μg/ml) and solubilised both bi-calcium and tri-calcium phosphate, but its halos were larger in bi-calcium-containing media, which has not been reported.

All the isolates nodulated P. lunatus as was expected, since this is the original host (Table 1). Both P. vulgaris and V. unguiculata were nodulated with just over the half of the P. lunatus isolates, but there was no evidence of nitrogen fixation on P. vulgaris nodules since they were small and did not show the typical leghaemoglobin reddish colour. Also, soybean-derived reference strains B. elkanii USDA76^T and Bradyrhizobium diazoefficiens USDA110^T (formerly B. japonicum) did not nodulate P. lunatus (Table 1). Bradyrhizobium yuanmingense CCBAU10071^T nodulated P. lunatus, despite that it was isolated from Lespedeza cuneata (Yao et al. 2002). This species was also reported to nodulate Pachyrhizus erosus (Rodríguez-Navarro et al. 2004), V. unguiculata and V. radiata (Zhang et al. 2008). Further studies of nodulation are required with P. lunatus, L. cuneata and Glycine max to confirm nodulation specificity, since B. yuanmingense CCBAU10071^T was not able to nodulate Glycine max (Yao et al. 2002). These differences could imply modifications to bradyrhizobia cross-inoculation groups and the existence of symbiovars among Bradyrhizobium spp.

Genomic fingerprints analysis of the 16 root-nodule isolates resulted in ten different profiles for each primers set; nonetheless, ERIC-PCR and REP-PCR clustered the isolates in the same groups, differing slightly from BOX-PCR results (Fig. 1). The combined analysis of the profiles of the three techniques showed, at 9 % of similarity, three major clusters of profiles that correspond to the same classifications obtained with the YEM–BTB medium: fast-growing isolates (I and II), extra-alkalinising SG (III) and alkalinising SG (IV to XI). Most of REP-PCR groups were restricted to their nodular origin and sampling sites but REP-PCR group III isolates (extra-alkalinising SG) came from different nodules from the same sampling site. Despite the low sample size, the profiles from the three PCR amplifications revealed that there appears to be more genetic diversity within alkalinising SG isolates than within extra-alkalinising ones, although all the isolates were obtained from independent nodules collected from different plants.

Fig. 1

Dendrogram generated from the combined REP-PCR, ERIC-PCR and BOX-PCR fingerprints patterns of Lima bean isolates and B. yuanmingense LMTR28 (reference strain) grouped by UPGMA and Jaccard coefficient. The same pattern code within columns corresponds to isolates with similar banding patterns

16S rRNA gene sequencing revealed that, at 99.7 % sequence identity, the fast-growing isolate LMTR-R3 was phylogenetically related to Rhizobium mesosinicum CCBAU 25010^T, and LMTR-R6 was related to Rhizobium alamii GBV016^T (Fig. 2). Alkalinising SG isolates were related to both B. yuanmingense CCBAU10071^T and B. liaoningense 2281^T, with 99.8 % sequence identity, and extra-alkalinising SG isolates had 100 % sequence identity to both B. paxllaeri LMTR 21^T and B. icense LMTR 13^T. These results suggest that fast-growing and slow-growing isolates belong to the Rhizobium and Bradyrhizobium genus, respectively. The fact that most of the isolates are from the Supe valley and from others areas of the Peruvian coast, supports the idea that P. lunatus specifically selects bradyrhizobial genotypes as its predominant microsymbionts (Ormeño-Orrillo et al. 2006).

Fig. 2

Phylogenetic analysis of 16S rRNA sequences from isolates obtained in this study. Type strains were included for analysis with the Neighbor joining method. Strains having the same sequence are shown in the same terminal branch. Lima bean isolates are indicated in bold and sequence accession numbers are given within brackets. Only bootstrap values greater than 60 % are shown (1,000 pseudoreplicates)

Despite 16S rRNA gene sequence analysis providing important information about bacteria identity, it is necessary to conduct a sequence analysis of other house-keeping genes or ITS sequences for establishing an accurate identity of the isolate. Some species have identical 16S rRNA gene sequences but vary in their recA, atpD, glnII, dnaK and gyrB genes, e.g., B. paxllaeri LMTR21^T and B. icense LMTR13^T (Durán et al. 2014).

The results we obtained herein highlight the diversity of fast-growing and slow-growing rhizobial isolates that nodulate P. lunatus, and suggest phenotypic differences among P. lunatus strains as selection tools of bacterial candidates to be used as biofertilizer. Additional phenotypic data were provided in comparison to previous molecular-oriented studies, especially in the case of rhizobia and extra-alkalinising bradyrhizobia. Our study points out the need for further studies on the diversity and taxonomy of fast-growing rhizobia associated with P. lunatus, since it seems to be a numerically less abundant group than bradyrhizobia. This work offers basic information for further studies on the diversity of and complex interactions among the root-nodule bacteria, the host plant, and environmental factors. This research was supported by Consejo Nacional de Ciencia y Tecnología del Perú (CONCYTEC) PROCYT N°229-2005-CONCYTEC-P and FDA biol-111/UNALM.