Limited resolution of 16S rDNA DGGE caused by melting properties and closely related DNA sequences

https://doi.org/10.1016/S0167-7012(03)00038-1Get rights and content

Abstract

The phylogenetic affiliation of 91 operational taxonomic units, randomly sampled from three aquatic microcosm experiments, was investigated by two PCR based and one culture dependent method. The occurrence of multiple melting domains and poor coupling between Tm and DGGE retardation was demonstrated to cause poor resolution at the species level in PCR-DGGE analysis of microbial communities. We also showed that the problem of multiple melting domains was particularly prone for brackish water bacterioplankton in the Flavobacterium genus, providing characteristic band morphology for this genus. Banding patterns from DGGE analysis may therefore be misinterpreted in terms of the species richness in natural bacterial communities, when using commonly applied universal primers.

Introduction

Denaturing gradient gel electrophoresis of 16S rDNA fragments, generated by the polymerase chain reaction (PCR-DGGE), has become a popular method among microbial ecologists to study the diversity of natural microbial populations. This method constitutes direct extraction of the community DNA and amplification of typically 200–600 bp long 16S rDNA fragments. These fragments are separated according to their melting point on a denaturing gradient gel (e.g. Muyzer et al., 1998).

PCR-DGGE is considered a rapid and reliable method for the relative comparison of different bacterial communities. The method also provides a comparison of the true sequences if DGGE bands are excised and sequenced (Muyzer et al., 1998). The PCR-DGGE was originally developed to analyze fragments from single organisms where comprehensive knowledge of the nucleotide sequence was available (Muyzer et al., 1993). Natural samples, however, typically contain a number of species with unknown phylogenetic affiliation and abundance. A relatively lower sensitivity of PCR-DGGE, when applied to natural samples and ecological questions can therefore be anticipated, due to the lack of accurate knowledge of the nucleotide sequence obtained and the diversity of natural bacterial communities. In correspondence with this low resolution in DGGE profiles has been reported when there is a high diversity of bacterial taxa in the sample (Torsvik et al., 1998).

Potential causes to inaccurate phylogenetic information from PCR-DGGE can be found both prior and during the separation process. Pitfalls introduced during DNA extraction and artifacts caused by PCR amplification of mixed templates encompass preferential amplification of certain sequences, chimeric amplification products and erroneous nucleotides Speksnijder et al., 2001, Wintzingerode et al., 1997. These types of errors will not be further addressed here. The PCR-DGGE separation process itself, however, relies on that the nucleotide sequence (i.e. phylogenetic integrity) is directly proportional to the melting properties of the fragments. However, both the abundance of G–C and A–T pairs and their order in the fragment influence the melting temperature. For closely related organisms the relationship between nucleotide sequence, phylogentic affiliation and the melting point is not well established. Thereby the retardation of the fragment in the gel matrix may not properly indicate phylogenetic relatedness at high resolution, like the species level.

One potential source of low resolution due to the sequence of nucleotides is the occurrence of 16S rDNA with multiple melting domains (MMD), discussed first by Myers using two illustrative fragments (Myers et al., 1988). These MMDs typically result in an extended (fuzzy) band in the migration direction, hampering band resolution. More recently, the influence of MMD has been presented for a eukaryotic model gene (Wu et al., 1998) and probably the same phenomenon has been described for a soil bacterial community (“cloudy bands” in Wieland et al., 2001). However, the influence of MMD on the resolution of bacterial species in environmental samples has not been discussed thoroughly or affiliated with certain bacterial taxa.

A different type of inaccuracy in DGGE analysis is due to artificial or natural microheterogeneity in the DNA sequence (Speksnijder et al., 2001), resulting in that a single band may be composed of several species Van Hannen et al., 1998, Sekiguchi et al., 2001, or that several bands are generated from a single species. The application of DGGE is ideally dependent on a unique coupling between Tm (band retardation) and the phylogenetic identity. Due to the sensitivity of the DGGE this relationship must hold down to the level of single bases (i.e. at the microvariation level). Failure to meet this requirement may result in a misinterpretation of the richness in the examined sample. We are not aware of reports aiming to analyze the coupling of phylogenetic identity in terms of 16S RNA sequence and melting point in natural bacterial communities.

In the present study we investigated the cause of a wide, slowly migrating DGGE band and tight clusters of distinct bands in our environmental samples. We exemplify with a few cases, although the conclusions reflect the observations for all 91 OTUs analyzed. All OTUs were identified by sequencing 16S rDNA fragments extracted from the gels and are reported elsewhere. Marked fuzzy bands, due to multiple melting domains, were only found to occur in aquatic isolates and clones in the Flavobacterium group. Cross-taxa similarities in Tm were demonstrated for some selected β-proteobacteria clones.

Section snippets

Origin of isolates and clones

Picoplankton from 1-l samples from enrichment experiments (Kisand et al., 2002) were collected by filtration onto 0.2 μm pore size polysulphone filters (Supor-200, Gelman). Total cellular DNA was extracted from these filters by lysing the cells with lysozyme and proteinase K, followed by CTAB/NaCl treatment and chloroform-phenol extraction to extract chromosomal DNA Kisand et al., 2002, Wilson, 1994.

The pure isolates and the 16S rDNA gene clone library were obtained from the same samples as

Results and discussion

All bacterial isolates in the Flavobacterium genus in our collection (12 species) had multiple non-cooperative melting domains at a discrete temperature based on an analysis of the 16S RNA sequence. By band morphology (thin and sharp vs. fuzzy) and Tm profiles calculated from the base sequence we could show that multiple melting domains are found in regions from ∼341 to ∼928 bp of the 16S rDNA gene of our isolates in the Flavobacterium genus Fig. 1, Fig. 2. Additionally several clones from our

References (23)

  • R. Myers et al.

    Detection of single base changes in DNA: ribonuclease cleavage and denaturing gradient gel electrophoresis

  • Cited by (92)

    • Basagran<sup>®</sup> induces developmental malformations and changes the bacterial community of zebrafish embryos

      2017, Environmental Pollution
      Citation Excerpt :

      The constrains of DGGE are well known: phylotypes are only represented as a band in the profile if their abundance reaches 1% of the total bacterial cells in the community, otherwise they will not be detected (Muyzer et al., 1993); different sequences may show similar migratory behaviour in the gel, resulting in bands that coincide. Moreover, the bands may appear fuzzy on gel due to multiple melting domains and microheterogeneity (Kisand and Wikner, 2003). This implies that caution must be taken in data interpretation.

    • Molecular approaches for the detection and monitoring of microbial communities in bioaerosols: A review

      2017, Journal of Environmental Sciences (China)
      Citation Excerpt :

      T-RFLP may misidentify polymorphisms of the same sequence as novel and tends to return many unknown sequences from samples because only a small percentage of 16S rRNA gene sequences have been recorded in databases. In addition, T-RFLP estimates of diversity can be influenced by sequence composition (Blackwood and Buyer, 2007; Gelsomino et al., 1999; Kisand and Wikner, 2003; Ronaghi, 2001). DGGE is inadequate to evaluate the prevalence and diversity of rare bacteria because the detection limits for specific bacterial groups are fairly high.

    • Parent material and vegetation influence bacterial community structure and nitrogen functional genes along deep tropical soil profiles at the Luquillo Critical Zone Observatory

      2015, Soil Biology and Biochemistry
      Citation Excerpt :

      “Universal” primers are supposed to be able to amplify most of the phylogroups of bacteria from natural environments, however, none of these primers are able to cover all bacterial groups (Hong et al., 2009). Minor groups may be beyond the detection limit and thus the richness of bacterial communities is likely underestimated (Kisand and Wikner, 2003). Indeed, we have only just begun to understand the diversity and distribution of bacterial communities in tropical soils and subsoils.

    • Dynamic changes in the structure of microbial communities in Baltic Sea coastal seawater microcosms modified by crude oil, shale oil or diesel fuel

      2013, Microbiological Research
      Citation Excerpt :

      Because so much oil is used, transported and stored in this region, oil and oil spills are considered a major threat to the Baltic Sea ecosystem (HELCOM 2009). Hence, a number of studies about the total microbial community composition and dynamics of the Baltic Sea and sediments have been conducted (Pinhassi et al. 1997; Hagström et al. 2000; Kisand and Wikner 2003; Kisand et al. 2005; Edlund et al. 2006; Riemann et al. 2008; Edlund and Jansson 2008; Andersson et al. 2010; Koskinen et al. 2011). However, data about the presence and diversity of the alkB genes in the Baltic Sea bacterioplankton have not been published and also little research has been performed regarding microbial community dynamics in response to hydrocarbons, especially in the water column.

    View all citing articles on Scopus
    View full text