Evaluation of 23S rRNA PCR Primers for Use in Phylogenetic Studies of Bacterial Diversity

The availability of a diverse set of 23S rRNA gene sequencesenabled evaluation of the specificity of 39 previously publishedand 4 newly designed primers specific for bacteria. An extensiveclone library constructed using an optimized primer pair resultedin similar gene richness but slightly differing coverage ofsome phylogenetic groups, compared to a 16S rRNA gene libraryfrom the same environmental sample.

INTRODUCTION

Top

Introduction

There has been renewed interest in the use of the 23S rRNA genewith the decrease in sequencing costs and the growing popularityof techniques such as microarrays (3, 13), analysis of the 16S-23Sintergenetic region (7, 9), fluorescence in situ hybridization,and quantitative PCR. The 23S rRNA gene offers the same advantagesas the 16S rRNA gene (e.g., universal distribution, conservedfunction, and invariant and variable regions), yet it includesadditional diagnostic sequence stretches due to a greater length,characteristic insertions and/or deletions (12), and possiblybetter phylogenetic resolution because of greater sequence variation(4, 10-12, 20). However, use of the 23S rRNA gene for bacterialcommunity analysis is hampered by the lack of established broad-rangebacterial PCR amplification and sequencing primers.

This study incorporates data from large-scale sequencing effortsto develop new and evaluate existing bacterium-specific 23SrRNA PCR amplification primers. Additionally, this study includesthe first well-sampled environmental clone library of 23S rRNAsequences, greatly increasing the number of 23S rRNA gene sequences.

Evaluation of primers.

Top

Evaluation of primers.

To check the specificity of PCR primers, an alignment of 23SrRNA gene sequences was developed using the ARB software package(http://www.arb-home.de). Bacterial 23S rRNA sequences wereobtained from published sources: the European rRNA database(22), National Center for Biotechnology Information completebacterial genomes (as of 6 February 2005), the ARB LSU database,and environmental bacterial artificial chromosome clones (16,18). To ensure broad environmental representation of these primers,sequences were also retrieved using BLAST from the SargassoSea assembled database (21) with full-length query 23S rRNAsequences from the genomes of representative organisms (Shigellaflexneri 2a strain 301, Pirellula strain 1, Prochlorococcusmarinus CCMP 1986, Streptomyces coelicolor A3, Bradyrhozobiumjaponicum USDA110, and Bacteriodes fragilis YCH46). Using thismethod, 1,415 nonredundant 23S rRNA sequences of >400 bpeach were retrieved from the Sargasso Sea data set. Initialalignments of a total of 2,176 sequences were constructed usingthe ARB Fast Aligner with manual editing based on secondarystructure and the existing ARB alignment. This data set wasnot corrected for skewing, due to overrepresentation of commonlaboratory organisms, pathogens, and organisms abundant in theSargasso Sea.

The primers developed in this study (129f, 189r, 457r, and 2490r,with numbering based on Escherichia coli position) (6) showexcellent correspondence to sequences in the aligned database(Table 1); additionally, some mismatches may be the result ofPCR or sequencing error. Although some previously published"universal" bacterial primers display broad range, this extensivedatabase indicates that other suggested target regions are notsufficiently conserved to serve as bacterial PCR primers (Table2). Primers for ITS amplification show various degrees of specificity:the region corresponding to the position of primer 129f is highlyconserved (8, 15), but other primers are less conserved andexclude a large fraction of bacterial diversity (7, 19).

View this table:
[in this window]
[in a new window]
TABLE 1. Percentage mismatches to the 23S rRNA gene dataset for primers designed in this study^a

View this table:
[in this window]
[in a new window]
TABLE 2. Percentage mismatches to the 23S rRNA gene dataset at each position for previously described universal primers^a

On the basis of their broad specificity, length of amplifiedsequence, and good amplification properties, we propose usingthe primers 129f (modified in this study) and 2241r for studiesof bacterial 23S rRNA diversity. These primers amplify a largeportion of the 23S rRNA, consistently produce only a singleband of PCR product, and are highly conserved across the bacterialsequences currently available (Tables 1 and 2). Positive amplificationwas achieved with a diverse set of isolates under the followingconditions: 3 min at 94°C; then 30 cycles, each consistingof 1 min at 94°C, 1 min at 57°C, and 2 min at 72°C;and a final 5-min extension at 72°C. All isolates used totest the primers produced PCR product of the correct size; thephyla of bacteria are listed and the number of isolates testedis given in parentheses: ${alpha}$ -Proteobacteria (7), ß-Proteobacteria(2), ${delta}$ -Proteobacteria (1), ${varepsilon}$ -Proteobacteria (1), ${gamma}$ -Proteobacteria(22), Firmicutes (7), Bacteroidetes (8), and Cyanobacteria (2).

Analysis of 23S rRNA clone library.

Top

Analysis of 23s rrna...

The 129f-2241r primer set was subsequently used to constructa clone library to evaluate coverage and relative distributionof phyla in comparison with a 16S rRNA clone library constructedfrom a parallel sample (1). A surface seawater sample from themarine end of Plum Island Sound estuary (northeastern Massachusetts)was collected as previously described (1). Cells were lysedusing bead beading (5), and DNA was purified using phenol:chloroform:isoamylalcohol extraction, sodium acetate and ethanol precipitation,and RNase I treatment (17). DNA was amplified in 10 replicate20-µl PCRs, each reaction mixture containing 50 ng ofpurified DNA template. PCR conditions were as follows: 3 minat 94°C; then 15 cycles, each consisting of 1 min at 94°C,1 min at 57°C, and 2 min at 72°C; and a final 5-minextension at 72°C. PCR products were pooled, precipitatedwith ethanol, and gel extracted (QIAGEN gel extraction kit).Amplicons were cloned using the TOPO-TA kit (Invitrogen).

A total of 535 operational taxonomic units (OTUs) were identified,based on sequential digests with restriction enzymes HhaI andMspI of cloned inserts amplified using internal plasmid primers(M13). Inserts with restriction patterns adding up to >2,500nucleotides were excluded, as they were assumed to originatefrom more than one cloned 23S rRNA gene insert. To determinethe phylogenetic coverage, at least one member of each OTU wassequenced and grouped into higher taxonomic groups (subphylumor phylum). Both 129f and 457r were used as sequencing primerson plasmids extracted using a QIAprep Spin Miniprep kit (QIAGEN)and M13-amplified PCR products, respectively. A total of 614clone library sequences were edited using Sequencher, and phylum-levelidentification of the OTUs was made using discontinuous megaBLASTwith a scoring metric (match = 4; mismatch = –5) to allowidentification of sequences highly divergent from those presentin the database. The cutoff for categorization of a sequencewas a sequence length of 300 bp of at least 85% similarity toan organism of known phylogeny.

A comparison of 23S and 16S rRNA (1) gene clone libraries constructedfrom replicate water samples yielded gross similarities butalso some important differences (Fig. 1). The observed levelsof richness in the two libraries were comparable when the digestion-definedOTUs in the 23S library were approximated by 99% sequence identityclusters (1, 14) in the 16S rRNA library (535 versus 520 forthe 23S and 16S rRNA gene libraries, respectively). In bothlibraries, Bacteroidetes and ${alpha}$ -Proteobacteria were the most abundantgroups (2). However, the 23S library displayed a higher percentageof Bacteriodetes (42.8% versus 32.5%) and lower percentagesof ${gamma}$ -Proteobacteria (3.9% versus 22.8%), Actinobacteria, andminor groups. This comparison is of interest because it mayreflect the primer bias of either 23S or 16S rRNA primers, ashallower depth of sequence coverage in the 23S library maskingrare variants, or a limited 23S rRNA database preventing identificationof certain groups. Planctomycetales were probably excluded bythese 23S rRNA primers because the forward primer targets aregion not present in their 23S rRNA gene. Additionally, >5%mismatches were observed at position 10 of primer 129f to theset of aligned sequences (Table 1); these mismatches occurredprimarily in environmental sequences rather than in culturedisolates, confirming the value of incorporating environmentalshotgun sequences in the alignment. This mismatch may explainthe low level of abundance of ${gamma}$ -Proteobacteria in the clone library,as this alternate sequence is present in the SAR-86 group (16,18) and other ${gamma}$ -Proteobacteria in the database, as well as membersof other phyla. This problem can be remedied by adding an additionaldegeneracy to primer 129f with the final sequence as CYGAATGGGRVAACC;this modified primer paired with 2241r positively amplifieda subset of 14 isolates from diverse phyla. The large numberof unknowns is due to the difficulty of 23S rRNA sequence identificationbecause of the poor depth of sequence coverage, especially forless-well-studied phyla (i.e., ß-, ${delta}$ -, and ${varepsilon}$ -Proteobacteria).

View larger version (18K):
[in this window]
[in a new window]
FIG. 1. Relative frequency distribution of major phylogenetic groups detected among the environmental sequences from a 16S rRNA library (black) and a 23S rRNA library (gray).

Applications using 23S primers, especially techniques such asautomated rRNA intergenic spacer analysis that are highly sensitiveto the primers chosen (7), should be reevaluated and perhapsmodified in light of this data. Nonetheless, this comparisonof 16S and 23S rRNA gene sequences shows that reasonable coverageand agreement between broad-range primer pairs can be achieved.

The alignment used to check the primers is available online(see the supplemental material and the ARB database, availablefor download at http://web.mit.edu/polz/seq=align.html).

Nucleotide sequence accession numbers.

Top

Nucleotide sequence accession...

Sequences were submitted to GenBank with accession numbers DQ312516to DQ313129.

ACKNOWLEDGMENTS

This research was supported by grants from NSF, NIEHS, and DOE(Genomes to Life).

Thanks to Ivan Ceraj for programming assistance and PatriciaMcAndrew for help with initial clone library screening.

FOOTNOTES

* Corresponding author. Mailing address: Massachusetts Institute of Technology, 48-421, 77 Massachusetts Ave., Cambridge, MA 02139. Phone: (617) 253-7128. Fax: (617) 258-8850. E-mail: mpolz{at}mit.edu.

Supplemental material for this article may be found at http://aem.asm.org/.

Present address: Department of Microbiology and Molecular Genetics,Harvard Medical School, Boston, MA 02115.

Present address: Laboratory for Optics and Biosciences, InstitutNational de la Santé et de la Recherche Médicale,Ecole Polytechnique, 91128 Palaiseau Cedex, France.

^¶ Present address: Limnology/Department of Ecology and Evolution,Norbyvägen, Uppsala University, SE-75236 Uppsala, Sweden.

REFERENCES

Top