Molecular characterization of Legionella populations present within slow sand filters used for fungal plant pathogen suppression in horticultural crops

Abstract

The total bacterial community of an experimental slow sand filter (SSF) was analyzed by denaturing gradient gel electrophoresis (DGGE) of partial 16S rRNA gene PCR products. One dominant band had sequence homology to Legionella species, indicating that these bacteria were a large component of the SSF bacterial community. Populations within experimental and commercial SSF units were studied by using Legionella-specific PCR primers, and products were studied by DGGE and quantitative PCR analyses. In the experimental SSF unit, the DGGE profiles for sand column, reservoir, storage tank, and headwater tank samples each contained at least one intense band, indicating that a single Legionella strain was predominant in each sample. Greater numbers of DGGE bands of equal intensity were detected in the outflow water sample. Sequence analysis of these PCR products showed that several Legionella species were present and that the organisms exhibited similarity to strains isolated from environmental and clinical samples. Quantitative PCR analysis of the SSF samples showed that from the headwater sample through the sand column, the number of Legionella cells decreased, resulting in a lower number of cells in the outflow water. In the commercial SSF, legionellae were also detected in the sand column samples. Storing prefilter water or locating SSF units within greenhouses, which are often maintained at temperatures that are higher than the ambient temperature, increases the risk of growth of Legionella and should be avoided. Care should also be taken when used filter sand is handled or replaced, and regular monitoring of outflow water would be useful, especially if the water is used for misting or overhead irrigation.

FIG. 1.

Ethidium bromide-stained DGGE gel (negative image) showing the 194-bp PCR-amplified fragment of the 16S rRNA genes (V3 region) from the total bacterial communities from the SSF samples. Lane 1, reservoir; lane 2, storage tank; lanes 3 to 5, top layer of three replicate sand columns; lane 6, L. cherrii; lane 7, headwater; lane 8, outflow water; lane M, bacterial marker containing P. fluorescens (a), S. yanoikuyae (b), B. subtilis (c), B. phenazinium (d), P. amylolyticus (e), A. rhizogenes (f), and A. polychromogenes (g). The arrowhead indicates the DGGE band that was initially excised from the total bacterial community, cloned, and sequenced (homology to L. cherrii).

FIG. 2.

Ethidium bromide-stained DGGE gel (negative image) of PCR products amplified by using the Legionella-specific LEG-448 GC clamp and LEG-858 primers for the 16S rRNA genes. Samples were taken from experimental and commercial SSF units. Lanes 2 to 8 show the changes within the Legionella populations in the sand column and water samples in the experimental SSF unit (SSF-1). Lanes 9 to 18 show comparative DGGE banding patterns obtained from the top layer sand column samples from the experimental system (SSF-1) and six commercial farms (SSF-2 to SSF-18). Lanes 1 and 9, L. pneumophila type strain; lane 2, reservoir; lane 3, storage tank; lane 4, top layer; lane 5, middle layer; lane 6, bottom layer; lane 7, headwater; lane 8, outflow water; lane 10, SSF-1; lane 11, SSF-2; lane 12, SSF-3; lane 13, SSF-4; lane 14, SSF-5; lanes 15 and 16, SSF-6 samples a and b; and lanes 17 and 18, SSF-7 samples a and b. Sample a was a sand sample taken from an SSF when it was not in use (winter period), and sample b was a sand sample taken from an SSF when it was in use. The arrowhead indicates the DGGE band from L. pneumophila.

FIG. 3.

Ethidium bromide-stained DGGE gel (negative image) obtained by using PCR products from 19 different clones (lanes 1 to 19) and L. pneumophila type strain (lane 20) amplified with the Legionella-specific LEG-448 GC clamp and LEG-858 primers for the 16S rRNA genes. Samples were obtained from the outflow water from the experimental SSF unit. The arrowhead indicates the DGGE band from L. pneumophila.

FIG. 4.

Phylogenetic tree showing distances based on the 16S rRNA sequences obtained from 27 clones from samples from the reservoir, the storage tank, the headwater tank, the top, middle, and bottom sand column layers, and outflow water from the experimental SSF unit. Sequences of the 16S rRNA genes of Legionella strains used for comparison in this study were retrieved from the EMBL DNA database (52). The 16S rRNA sequences obtained in this work are partial sequences (430 bp), and therefore their positions are indicated by dashed lines. The dendrogram was generated by using the programs DNASIST (maximum likelihood) and NEIGHBOR (neighbor joining). Scale bar = 0.1% estimated sequence divergence. Resampling (bootstrap analysis) was applied to the data, and the values obtained (percentages) are indicated at the nodes.