Community structure of bacteria on different types of mineral particles in a sandy soil

Abstract

Various types of mineral particles in a soil probably provide different microenvironments for microorganisms. The purpose of this study is to investigate whether different types of mineral in a soil harbor different bacterial populations. DNA was extracted from five types (quartz, feldspar, pyroxene, magnetite, iron-coated reddish brown particles) of sand-size mineral particles separated from a sandy soil, and was amplified for partial 16 S rRNA gene by polymerase chain reaction (PCR). Twenty-nine to 69 amplicons per each type of mineral were cloned and sequenced, followed by phylogenetic affiliation of the sequences. As a result, some types of bacteria were detected on all of the types of mineral including the orders Rhizobiales, Bacillales, and Acidobacteriales. In the case of Acidobacteriales, higher percentages were found on magnetite and quartz. Some taxa were restricted to specific types of mineral; the class Actinobacteria was found on pyroxene but not on quartz, and rarely on magnetite and feldspar. Bacterial diversity at the order level estimated by Chao1 value was higher in feldspar and pyroxene than the other three types of mineral. The UniFrac Significance test indicated that the differences in bacterial communitiy structures among the particles were suggestive except that between feldspar and pyroxene. These results support the idea that different communities of bacteria were associated with each of the mineral types.

Keywords:

• bacterial community structure
• bacterial diversity
• microhabitat
• mineral particle
• 16 S rRNA gene.

Introduction

A large number of different microorganisms live in soil. The number of species in several grams of soil is estimated to be tens of thousands or more than one million (Gans et al. 2005; Roesch et al. 2007). It is still unclear which factors are responsible for the high microbial diversity in soil; however, like macroorganisms, high habitat heterogeneity in a soil is assumed to play a role (Horner-Devine et al. 2004). For example, in a paddy rice (Oryza sativa L.) field containing diverse habitats for microorganisms, the respective habitats such as floodwater, soil, percolating water, and rice straw have been shown to have distinct community structures of microorganisms (Kimura and Asakawa 2006; Kikuchi et al. 2007; Cahyani et al. 2007, 2008).

A variety of mineral particles contained within a soil would be one of the sources of microscale habitat heterogeneity for the resident soil microorganisms. Cahyani et al. (2007, 2008) showed that specific bacteria inhabit iron mottles and manganese nodules formed in a rice field. Carson et al. (2009) revealed that three types of minerals which were buried in and then taken from the soil after a 70-d incubation harbored different bacterial communities estimated by ribosomal intergenic spacer analysis. Gleeson et al. (2006) reported that the community structures of bacteria (not in soil, but on rock surfaces) associated with different minerals were different. In their review, Uroz et al. (2009) concluded that the chemical composition of mineral particles affects the structure of the associated bacterial communities.

In our previous paper (Kotani-Tanoi et al. 2007), we revealed that the community structure of bacteria associated with sand-size particles in a soil is semi-specific to the type of their inhabiting mineral, by comparing the polymerase chain reaction-denaturing gradient gel electrophoresis (PCR-DGGE) profiles of 16 S rRNA genes. In the present paper, more detailed community structures of the mineral-associated bacteria will be shown by phylogenetic affiliation of the sequences of their 16 S rRNA genes with the aim of investigating whether different types of minerals in a soil harbor different bacterial populations.

Materials and Methods

Mineral particles

Five types of sand-size particles with different appearances were isolated under a stereomicroscope from the same soil sample as used in the previous study (Kotani-Tanoi et al. 2007) (sand-dune Regosol, Entisol; pH (H₂O), 7.1; total carbon content, 6.8 g kg^–1; total nitrogen content, 0.68 g kg^–1; soil texture, sand 96.7%, silt 3.3%, clay 0.0%). The mineral species of the particles were estimated as quartz, magnetite, feldspar (orthoclase), and pyroxene (enstatite and/or ferrosilite) based on their appearance under a mineralogical microscope equipped with a polarizer and an analyzer (Olympus SZX12, Tokyo). As the remaining one type was reddish brown and lost the color after it was treated with hydrochloric acid (HCl) (pH 2.0) at 98°C for 15 min, it is possible that the reddish brown particles were those coated with ferric oxide, regardless of the mineral species of particle inside.

DNA extraction and sequencing of partial 16 S rRNA genes

Each particle was transferred separately to a microtube and washed four times with 0.2 mL of sterilized water, using a vortex mixer for 10 s to remove microorganisms attached loosely to the particle and extraneous matter on the surface of the particle. As we considered that specific feature(s) in the structure of the bacterial community of a certain type of mineral appeared less evident if only one particle was analyzed, several particles of the same mineral type were combined for further analyses. Thus, DNA was extracted from 24 particles of quartz, 12 particles of magnetite, 10 particles of feldspar, 10 particles of pyroxene, and 10 of the reddish brown particles, as well as from the whole soil sample, by the same procedures described previously (Kotani-Tanoi et al. 2007). We consider that no significant bacterial contamination occurred during the processes above; no bacterial colony was detected after 70 d of incubation on 1/100 NB agar when particles were picked up from the autoclaved soil sample, washed, and poured onto the agar, while nearly a hundred or more colonies per particle were found when nonsterile particles were examined.

PCR amplification for the bacterial 16 S rRNA genes was achieved by the booster PCR method (Ruano et al. 1989) under the same reaction conditions as described previously (Kotani-Tanoi et al. 2007). The PCR primers used for the initial amplification were 5′-AGAGTTTGATCCTGGCTCAG-3′ (corresponding to positions 8–27 in Escherichia coli (Migula) Castellani and Chalmers numbering) and 5′-AAAGGAGGTGATCCAGCC-3′ (corresponding to positions 1542–1525). For the following amplification, the PCR primers 5′-AACGCGAAGAACCTTAC-3′ (corresponding to positions 952–968) and 5′-GGYTACCTTGTTACGACTT-3′ (corresponding to positions 1510–1492) were used. The amplicons of partial 16 S rRNA gene (positions 952–1510) were cloned and sequenced by the method described in Saito et al. (2008).

Analyses of the sequences

Based on the partial 16 S rRNA gene sequence (positions 969–1491), phylogenetic affiliation of the clones at the order level and higher was performed by the naïve Bayesian rRNA classifier program (Wang et al. 2007) provided by the Ribosomal Database Project (Cole et al. 2009). The taxonomic richness of bacteria on each type of particles was estimated using Chao1 estimator, which was calculated with EstimateS ver. 8.0 (Colwell 2006) based on the number of the order detected. The UniFrac (Lozupone et al. 2006) was used to determine whether two particle-associated bacterial communities differed significantly (significance analysis) and which bacterial group(s) had an excess of sequences in a particular type of particles (lineage-specific analysis). All the sequences were aligned with Clustal X 2.1 (Larkin et al. 2007) to create a single phylogenetic tree for input into UniFrac.

Accession numbers

Sequences of partial 16 S rRNA gene have been deposited in DDBJ/EMBL/GenBank with accession numbers AB497599 to AB497978.

Results and Discussion

A total of 380 amplicons of partial 16 S rRNA gene, i.e., 62 amplicons from quartz, 64 from magnetite, 29 from feldspar, 69 from pyroxene, 61 from reddish brown particles and 95 from the whole soil sample, were cloned and sequenced. The results of the sequence-based identification were shown as relative abundance of bacterial taxa on each mineral (Table 1). The orders Rhizobiales, Bacillales, and Acidobacteriales were detected from all six samples, but their relative abundance was different among the samples. Clones assigned to Verrucomicrobiales were found in quartz and magnetite, but not in feldspar, pyroxene, and reddish brown particles. Several clones from pyroxene, feldspar, and magnetite were assigned to the class Actinobacteria, but none from quartz and reddish brown particles. The UniFrac significance test indicated that except for the pair of feldspar and pyroxene, any two types of particle differed significantly in their bacterial communities (P < 0.01) although the differences were suggestive (P < 0.10) after correcting for multiple comparisons (Table 2). These results suggest that bacterial taxa were unevenly distributed among the various types of mineral particles in this soil.

Table 1. Relative abundance (%) of bacterial taxa of 16 S rRNA gene fragment in those cloned from each type of mineral and whole soil

Phylum-Class-Order	Quartz	Magnetite	Feldspar	Pyroxene	‘Reddish’	Whole soil
	(n = 62)†	(n = 64)	(n = 29)	(n = 69)	(n = 61)	(n = 95)
Chloroflexi-Anaerolineae-Caldilineales						2
Chloroflexi-Chloroflexi-Chloroflexales		2	3			1
Nitrospira-Nitrospira-Nitrospirales		5				2
Proteobacteria-Alphaproteobacteria-Rhizobiales	2	16	10	6	5	12
Proteobacteria-Alphaproteobacteria-Rhodospirillales		3	3	6		1
Proteobacteria-Alphaproteobacteria-Sphingomonadales				1	2
Proteobacteria-Alphaproteobacteria-Unknown	3			1		1
Proteobacteria-Betaproteobacteria-Burkholderiales	2		3	7	31
Proteobacteria-Betaproteobacteria-Unknown	8	3	3	3	2	9
Proteobacteria-Gammaproteobacteria-Legionellales						1
Proteobacteria-Gammaproteobacteria-Pseudomonadales		2				1
Proteobacteria-Gammaproteobacteria-Xanthomonadales				3
Proteobacteria-Gammaproteobacteria-Unknown		2
Proteobacteria-Deltaproteobacteria-Myxococcales				1		1
Proteobacteria-Unknown-Unknown	6			1	3	2
Firmicutes-Bacilli-Bacillales	8	8	24	16	28	13
Firmicutes-Bacilli-Lactobacillales	2		3	1	5
Firmicutes-Bacilli-Unknown			7	1	5
Firmicutes-Clostridia-Clostridiales	6	2		1	2	5
Firmicutes-Unknown-Unknown	3		3	1	2	5
Actinobacteria-Actinobacteria-Actinomycetales			3	4		7
Actinobacteria-Actinobacteria-Rubrobacterales			3	3		1
Actinobacteria-Actinobacteria-Unknown		3		10		1
Planctomycetes-Planctomycetacia-Planctomycetales	5	2		1
Acidobacteria-Acidobacteria-Acidobacteriales	35	42	10	20	7	6
Bacteroidetes-Sphingobacteria-Sphingobacteriales	2
Verrucomicrobia-Verrucomicrobiae-Verrucomicrobiales	6	2				1
Gemmatimonadetes-Gemmatimonadetes-Gemmatimonadales				1
(incertae sedis)-TM7						3
Unknown-Unknown-Unknown	11	11	21	7	10	23
†: number of clones

Table 2. P-value of UniFrac significance test for each pair of particle-associated bacterial communities with the number of permutations set to 100

	Quartz	Magnetite	Feldspar	Pyroxene	‘Reddish’
Quartz	–	<0.10	<0.10	<0.10	<0.10
Magnetite	0.00	–	<0.10	<0.10	<0.10
Feldspar	0.00	0.00	–	1.00	0.10
Pyroxene	0.00	0.00	0.34	–	<0.1
‘Reddish’	0.00	0.00	0.01	0.00	–
Note: Values in normal letters are P-values corrected for multiple comparisons (upper right) and those in italics are raw P-values (lower left).

The bacterial communities of quartz and magnetite were characterized by the dominance of the phylum Acidobacteria (Table 1) and the low diversity at the order level estimated by Chao1 (Table 3). An excess of Acidobacteria in magnetite was indicated also by the UniFrac's lineage-specific analysis (data not shown). Recent studies on the bacterial communities based on the environmental DNA analyses indicate that the phylum Acidobacteria is one of the most abundant bacterial groups in the environment including soil (Janssen 2006; Ward et al. 2009). Indeed, in the present study, acidobacterial sequences were recovered from all types of mineral and from the whole soil. According to Janssen (2006), acidobacterial sequences make up 5 to 46%, with an average of 20%, of the total sequences in libraries of 16 S rRNA and 16 S rRNA genes prepared from soil bacterial communities. The high percentages of acidobacterial sequences in the present study, i.e., 35% for quartz's community and 42% for magnetite's, both higher than the average reported by Janssen (2006), suggest that these minerals provide such microsites as the phylum Acidobacteria favors. The abilities to adhere to ferric iron-rich substrates (Johnson et al. 2008) and to catalyze iron redox reactions (Ward et al. 2009) reported for acidobacterial strains may contribute to the dominance of this phylum in the bacterial community of magnetite. Their ability to exhibit slow metabolic rates in nutrient-limited conditions (Ward et al. 2009) may be suitable for survival on quartz. The bacterial community of quartz was also characterized by the presence of Verrucomicrobia and the absence of Actinobacteria although the reasons are unclear.

Table 3. Estimated number (Chao1) of the orders in bacterial community of each type of mineral and whole soil

	Quartz	Magnetite	Feldspar	Pyroxene	‘Reddish’	Whole soil
No. of sequences	62	64	29	69	61	95
No. of orders detected	14	14	13	21	12	21
Chao1 value	16.0	17.8	27.0	32.3	15.0	30.0

The bacterial community of reddish brown particles was characterized by the dominance of the orders Burkholderiales (31%) and Bacillales (28%). Burkholderiales was suggested to be enriched in reddish brown particles by the UniFrac's lineage-specific analysis (data not shown). The expected presence of amorphous ferric oxide/oxyhydroxide coatings on the reddish brown particles attracts organic compounds (Violante et al. 2002; Wagai et al. 2009). Members of the order Burkholderiales have a variety of physiological traits and lifestyles (Vial et al. 2011); they are able to use a wide array of compounds as carbon sources (e.g., Chiarini et al. 2006; Morimoto et al. 2008; Saito et al. 2008) and are revealed to have efficient mineral weathering potentials (Calvaruso et al. 2010). This metabolic versatility may help them survive on the reddish brown particles where organic compounds accumulate.

In contrast to their dominance on reddish brown particles, no Burkholderiales clone was found in the whole soil in this study. It seems to be in accordance with the previous report by Fulthorpe et al. (2008), in which Burkholderia was absent from the ten most abundant genera in their four soils analyzed by culture-independent, 16 S rRNA gene techniques.

Bacillales clones were recovered from all the types of mineral and the whole soil sample in this study. According to Nicholson (2002), several Bacillus spp. have been isolated from the interiors of such locations as manganese rock varnish, desert rocks, deep subsurface boreholes, and near-surface granite formations. Flores et al. (1997) found Bacillus spp. on granite buildings and monuments that were slowly deteriorating. Together with these previous studies, the present result may suggest that some members of the order Bacillales are common residents on mineral surfaces.

As compared to the other three minerals, high values of Chao1 were found for the bacterial community of pyroxene and feldspar (Table 3). Pyroxene and feldspar are known to be weathered more easily than quartz (Jackson and Sherman 1953), and thus could supply more bioavailable inorganic nutrients. Gleeson et al. (2006) suggested that mineral composition, especially of aluminum (Al), silicon (Si), and calcium (Ca), has a significant impact on the structure of the bacterial communities associated with minerals of pegmatitic granites. The present results may imply that such elements as become bioavailable more on pyroxene and feldspar than on quartz affect the degree of the diversity of the mineral-associated bacterial community in soil.

Several taxonomic groups such as incertae sedis TM7, the order Caldilineales, and the order Legionellales, appeared only in the whole soil, not in any of the five types of the particles examined. It is uncertain where in the soil the members of those taxa lived. As we picked up the sand mineral particles from the soil and washed them before extraction of DNA in this study, it is possible that they lived in the soil as free-living organisms, were associated with fine mineral particles or organic matter, or attached loosely rather than firmly to the sand particles. As members of these taxa have often been detected in biofilms and organic granules (Atlas 1999; Hugenholtz et al. 2001; Sekiguchi et al. 2003; Yamada and Sekiguchi 2009), the latter two might be more probable as sites for these groups.

In contrast, some taxa were not or were rarely detected in the whole soil but were detected in specific types of mineral. For instance, no Planctomycetes clone was found in the whole soil, whilst 3 out of 62 clones (5%) were found in quartz. Only 1 clone (1%) of Verrucomicrobia was found in the whole soil, whilst 4 clones (6%) were found in quartz. This suggests that investigation into specific type(s) of mineral is a good approach to find rare or unknown microorganisms in soil.

In summary, the results in this study support the idea that different communities of bacteria are associated with each of the mineral types. It would be necessary for a more clear characterization of mineral-associated bacterial community to investigate the same types of minerals in other soil samples.

Acknowledgments

This work was supported in part by the president's discretionary fund of Nagasaki University, Japan, to M.N.