PCRs performed on mixed community DNA result in a mixture of sequences, potentially including amplicons from each targeted organism in the sample. The pool of amplicons can then be cloned and sequenced to obtain detailed information about the makeup of the community or can be profiled using a variety of DNA fingerprinting methods, such as terminal restriction fragment length polymorphism or denaturing gradient gel electrophoresis (23, 36). The most frequently used primers for studying microbial communities target entire domains (Bacteria, Eucarya, or Archaea). This implies that assays using these primers will detect any major shift in community structure. In practice, however, sampling issues make rare sequences difficult to detect, and there can be PCR bias toward organisms that are dominant or have preferentially amplified sequences (8, 27). In genetic profiling methods, the ability to distinguish between taxa is lowered by broad primers because phylogenetically disparate organisms can have similar genetic markers (5, 12). This is not as much of a problem when sequences are examined directly, but cloning and sequencing require a greatly increased effort to obtain a statistically robust picture of the community (11). With broad domain-specific primers, this problem is exacerbated because of the increased diversity of the PCR amplicons being characterized. One method of reducing these problems is to use sets of PCR primers which are each more narrow phylogenetically. This can result in enhanced sensitivity and more detailed information. The tradeoff in terms of phylogenetic breadth is offset if multiple assays are performed using primers which together cover the breadth of domain-specific primers or if hypotheses concern only a particular taxonomic group.

Group-specific PCR primers can also be used for rapid screening of PCR amplicon or metagenomic clone libraries for ribosomal sequences of a particular taxonomic affiliation (for example, see reference 28). The taxonomy of cultured organisms could also be screened using PCR assays with group-specific primers (39). Finally, the abundance of a targeted group can be monitored through quantitative or real-time PCR. The variety of applications for group-specific primers demonstrates their importance in the field of molecular microbial ecology.

Many primers have been previously developed for various groups. However, as sequence databases continue to grow, these primers must be continually reevaluated for their specificity and breadth (4). The purpose of the present study was to develop several primer sets targeting commonly occurring and important groups, some well represented by cultures and sequences and others not well represented. Primers specific for six groups important in soil, including new and previously published primers, were tested by comparison to database sequences, and chosen primers were further tested by cloning and sequencing from soil community DNA.

Primer design and PCR optimization. PCR primers specific for the ribosomal gene of several major groups of microorganisms (Table 1) were gathered from the literature or designed using sequences from the 2001 and 2003 releases of the ARB ribosomal small subunit database (http://www.arb-home.de/) (30). Basidiomycota ITS and large subunit sequences were also downloaded from GenBank and examined using BioEdit software (18). Bacterial groups were related to the taxonomy outlined by Garrity et al. (15) and fungal groups to the one outlined by Kirk et al. (25). Primers that were rejected at this stage missed a large proportion of target sequences, lacked specificity, or appeared to lack discriminatory bases at the 3′ end of the oligonucleotide compared to the primers that were chosen for further testing.

TABLE 1. Ribosomal primers used in this study

PCR conditions for each primer pair were optimized using genomic DNA of a small bank of cultured organisms (Table 2). Annealing temperature gradients were used to determine the lowest annealing temperature where PCR product of the appropriate size was obtained from members of the target taxa but not from other organisms. PCR product was visualized by electrophoresis on a 1% agarose gel stained with ethidium bromide. For primers targeting Planctomycetes (not present in our culture collection) we used DNA extracted from soil to optimize annealing temperature. Genomic DNA was extracted from soils and cultures using the UltraClean soil DNA extraction kit (Mo Bio Laboratories, Solana Beach, CA). PCRs were performed in a Touchgene gradient thermocycler (Techne, Cambridge, United Kingdom) using 0.025 U/μl Taq DNA polymerase, 1× PCR buffer accompanying Taq polymerase, a 160 μM concentration of each deoxynucleoside triphosphate, 3 mM MgCl₂ (all from Continental Lab Products, San Diego, CA), a 0.1 μM concentration of each primer (Integrated DNA Technologies, Coralville, IA), and 0.1 μg/μl bovine serum albumin (Promega Corporation, Madison, WI). During the PCR optimization stage, we found that reactions using the primer Bas1005F performed better using 120 μM concentrations of each deoxynucleoside triphosphate and 4 mM MgCl₂ instead of our standard conditions, and subsequent Bas1005F reactions were performed using this alteration. PCR conditions consisted of initial denaturation for 3 min at 95°C, followed by 35 cycles (30 s at 94°C, 30 s at annealing temperature [Table 1], 90 s at 72°C), followed by 7 min of final extension at 72°C.

TABLE 2. Cultures used in PCR optimizations

Clone library testing of primers. The specificity of primers was tested by sequencing PCR amplicons from a soil sample collected from either the University of Maryland Agricultural Experiment Station in Salisbury, MD, or the USDA Farming Systems Project in Beltsville, MD (Table 1). Salisbury soil was a loamy sand planted to sweet corn. Beltsville soil was a silt loam ultisol under either forest or soybean (Table 1). For each PCR, the cycle number was adjusted so that strong PCR product was obtained without amplification of nonspecific bands or background DNA smears (25 to 35 cycles, depending on primer and sample). PCRs for cloning were run in triplicate and pooled. PCR product was purified with PCR Preps Wizard kits (Promega), eluting with 25 μl H₂O heated to 55 to 65°C. Twenty nanograms of PCR product was cloned into Escherichia coli JM109 using the pGEM-T vector system (Promega). White colonies were picked with sterile toothpicks. For most primers, picked colonies were inoculated into 50 μl H₂O, vortexed vigorously, and frozen. White colonies from the Beta680F and BLS342F plates were inoculated into liquid LB medium (50 μg/ml ampicillin) and incubated at 37°C on a tube rotator for 20 h. Plasmid DNA from these cultures was isolated using the UltraClean 6-minute mini plasmid prep kit (Mo Bio).

PCR amplification from purified plasmid or transformed colonies was used to check the size of inserts. PCR-amplified inserts were also digested with the restriction enzymes MspI and RsaI (New England Biolabs, Beverly, MA) to examine the diversity of clone libraries. Restricted PCR amplicons were run on 4% NuSieve 3:1 agarose gels (BioWhittaker, Rockland, ME) in Tris-borate-EDTA buffer. Gels were stained for 4 to 5 h with Gelstar (BioWhittaker) and visualized with 400- to 500-nm illumination on a DarkReader transilluminator (Clare Chemical, Dolores, CO). Gel images were analyzed using Gel Compar II (Applied Maths, Austin, TX).

PCR product from a clone representing each restriction pattern was submitted for sequencing at the University of Maryland DNA Sequencing Facility (College Park, MD) or at the USDA Environmental Microbial Safety Laboratory (Beltsville, MD). Partial small subunit sequences were examined with the “Chimera check” program at the Ribosomal Database Project (RDP version 8.1) (7) and aligned to sequences in the ARB database using the integrated aligner. Alignments were manually adjusted if needed. Sequences were placed in the ARB phylogenetic tree using the parsimony insertion function. Nearest matches were also found using BLASTn in GenBank (1) and the RDP to confirm results from ARB analyses.

Nucleotide sequence accession numbers. Nucleotide sequences determined in this study have been deposited in the GenBank database under accession numbers AY555587 to AY555633 and AY802783 to AY802983.

Analysis of Alphaproteobacteria primers. The primer ADF681F was designed to target predominantly the organisms within the class Alphaproteobacteria (Table 1). This primer exploits the same discriminatory position as αδP688r (39) but is aforward primer which amplifies a larger DNA fragment. In the 2003 ARB database, ADF681F matches 1,895 of 2,106 Alphaproteobacteria sequences, 44 of 65 Fusobacteria sequences, 12 of 500 Deltaproteobacteria (Table 3) sequences, and 5 other sequences. The only Alphaproteobacteria groups which have sequences that consistently do not match ADF681F are the mitochondria, the genera Bartonella, Devosia, and Orientia, and some Wolbachia sequences. The sequences from Deltaproteobacteria orders Desulfobacterales, Myxococcales, and Syntrophobacterales have a single mismatch with this primer at the 5′ end and so may also amplify. In soil, amplicons from ADF681F should be predominantly from Alphaproteobacteria and Deltaproteobacteria, since Fusobacteria are not found in that environment (22). Previously published Alphaproteobacteria primers or probes that were compared to the ARB database included F203α (19), Alf-1 and Alf-2 (20), RB1 (35), Alf1b (31), and ALF968 (34).

TABLE 3. Percentage of sequences which match group-specific and general primersa

Analysis of Betaproteobacteria primers. The primer Beta680F (33) (Table 1) matched 1,285 of 1,411 sequences in the class Betaproteobacteria (Table 3) in the ARB 2003 database, with only four matches outside this group. Oxalobacter, Nitrosomonas, Thauera, and Hydrogenophilus had several mismatches at the 5′ end of the oligonucleotide. Sequences in the Xanthomonadales, outside Betaproteobacteria, had several mismatches, but they were clustered at the 5′ end of the oligonucleotide, similar to Betaproteobacteria without complete matches. F948β (19) also matched an acceptable group of sequences and could be used to construct a reverse primer as well, but we chose to use Beta680F because it amplifies a larger fragment of the 16S gene. Other Betaproteobacteria primers examined included βP774f (39) and BET27 (20).

Analysis of Bacilli primers. The primer BLS342F (Table 1) was designed to target the Bacilli, a broad class of aerobic Firmicutes, and matches 1,573 of 1,673 target sequences (Table 3) in the ARB 2003 database. Sequences from Bacillus vedderi, Oenococcus, Brevibacillus, Thermoactinomyces, and Marinococcus halophilus have mismatches 3 or 4 bases from the 3′ end. Initially we tried to locate a primer site which would be specific for all Firmicutes; however, this did not appear possible. BLS342F is in a position similar to LGC344 (32) but should be more discriminatory against non-gram-positive organisms and does not require degenerate oligonucleotides. LGC29 (20) was found to miss many target sequences compared to BLS342F.

Analysis of Actinobacteria primers. The primer Act1159R (Table 1) was designed to target all sequences in the phylum Actinobacteria. It matches 2,510 out of 2,563 Actinobacteria sequences (Table 3) in the ARB 2003 database. The orders (containing one family each) Rubrobacterales and Coriobacteriales and the genera Leucobacter and Thermobispora have multiple mismatches, including the position 5 bp from the 3′ end of the primer. Act1159R is in a position similar to HGC1153 (39) but greatly expands the number of sequences matched within the Actinobacteria due to the degeneracy introduced. HGC29 (20) was found to miss many target sequences compared to Act1159R.

Analysis of Planctomycetes primers. The primer Pln930R (Table 1) was designed to target all sequences in the phylum Planctomycetes. It matches 153 out of 185 Planctomycetes sequences (Table 3) in the ARB 2003 database. It has mismatches in the three terminal 3′ bases, with sequences clustered with “Candidatus Kuenenia stuttgartiensis” and “Candidatus Brocadia anammoxidans,” but also matches one uncultured bacterium sequence clustered with this group. Pln930R is similar to sequencing primer Pln926R (41) but is much more discriminatory against non-Planctomycetes sequences. Other Planctomycetes primers examined include 1230fplanc (41) and PLA40F (10). A modified version of the primer Eub338F was paired with Pln930R for amplification of Planctomycetes sequences (Table 1) (9).

Analysis of Basidiomycota primers. A forward primer specific for the phylum Basidiomycota was designed near the 3′ end of the SSU ribosomal sequence so that part of the ITS region could be amplified. The ITS region was chosen for amplification for Basidiomycota because computer simulations suggested that the SSU rRNA did not contain enough variability to be practical for molecular fingerprinting applications such as terminal restriction fragment length polymorphism (C. Blackwood, unpublished data). The primer Bas1005F matched 287 out of 350 Basidiomycota sequences (Table 3) in the ARB 2003 database. Out of 63 missed Basidiomycota, 52 were in the class Urediniomycetes (rusts), which was the only group of Basidiomycota not well covered. A reverse, nonspecific primer which targets the 5.8S sequence of the Basidiomycota ITS region was also designed (Table 1) to be paired with Bas1005F. These primers result in amplification of ITS1 and approximately 800 bases of the SSU, with amplicons ranging from 1,030 to 1,270 bp in size. The Basidiomycota primer ITS4-B (14) was not used, as it missed a large number of target sequences.

Sequencing clone libraries. Eighty-five percent or more of the sequences obtained from clone libraries, representing 91% or more of the clones, were placed in the ARB 2003 phylogenetic tree with the groups intended as targets (Table 4). Most of the sequences that were not placed with the correct groups were ribosomal sequences from nontarget organisms, although the four nontarget sequences from the Bas1005F-5.8SR library and one nontarget sequence from the BLS342F-1392R library did not match any entry in GenBank and may have been PCR artifacts. Also, two sequences from the Pln930R-Eub338F- 0-III library were clearly chimeric. In preliminary analyses, we found that amplicons that were not the correct size were often nontarget sequences or sequences that did not match any GenBank entries, even for Basidiomycota ITS reactions where the size is somewhat variable. This highlights the need to optimize PCRs carefully and obtain clean bands.

TABLE 4. Environmental clone libraries

Nontarget ADF681F-1392R sequences included one which was similar (BLAST scores of >600) to Acidobacteria sequences and another which matched metazoan sequences from the family Enchytraeidae. The nontarget Beta680F-1392R sequence was similar to Verrucomicrobia sequences. Two of the nontarget BLS342F-1392R sequences displayed low levels of similarity (BLAST scores of <400) to database sequences affiliated with the order Rubrobacterales in the Actinobacteria, and the third clustered with the Chloroflexi. One nontarget Pln930R-Eub338F-0-III sequence had low similarity to several uncultured bacterial sequences.

The placement of experimental sequences into phylogenetic groups by analyses in ARB (Table 5) was generally confirmed using BLAST searches in the GenBank database. The only exception was that the majority of the sequences from the Bas1005F-5.8SR library had Ustilago spp. as nearest neighbors in GenBank (subclass Ustilaginomycetidae within the phylum Basidiomycota), but sequences from this taxa were absent from the ARB 2003 database, and the experimental sequences were placed with the subclass Exobasidiomycetidae in ARB (Table 5). These sequences made up 74% of the Bas1005F-5.8SR clone library. Preliminary sequencing of several other Bas1005F clone libraries created during PCR optimization indicated that other samples were dominated by sequences placed in the class Basidiomycetes. These clone libraries included one generated using the reverse primer ITS4 (42) instead of 5.8SR (Salisbury cornfield soil, six sequences placed with the Tremellomycetidae, four with Agaricomycetidae, and two with the Ustilaginomycetidae) and two other clone libraries generated using 5.8SR (bromegrass, two sequences placed with Tremellomycetidae, four with Agaricomycetidae; deciduous forest soil, nine sequences placed with Agaricomycetidae).

TABLE 5. Distribution of sequences found in clone libraries in target groupsa

Coverage by the clone libraries of orders within the targeted phyla or classes was best for Alphaproteobacteria in the ADF681F-1392R library (Table 5). For other groups, many of the orders not detected in our clone libraries include organisms that are not normally present in soil or are present in low numbers compared to detected taxa or include very few species. According to our computer analyses, sequences from orders not detected in our clone libraries but listed in Table 5 should amplify. However, the primers should be further tested if they are to be used in environments where organisms from these orders are known to be abundant or if amplification from a particular species must be verified. Note that we have successfully obtained PCR product from target organisms from a variety of other soils for all primers tested (data not shown).

While only one order is currently defined within the Planctomycetes, the divergence of sequences within the order indicates that there is a large amount of uncultured diversity in this group which has yet to be explored. Clone library sequences were placed with each of the four accepted Planctomycetes genera (Pirellula, Isosphaera, Gemmata, and Planctomyces) as well as a separate group of sequences within the Planctomycetes from uncultured organisms in a variety of environments (6,16, 21, 26, 29, 34, 40) (listed as “unaffiliated” in Table 5). None of the Pln930R clone library sequences were placed with the candidate genera which contained sequences not matching Pln930R, “Ca. Kuenenia,” and “Ca. Brocadia” (38).

Conclusions. The goal of this study was to provide primers which could be used to obtain more phylogenetically informative molecular assays of microbial community structure but which also maintain a broad phylogenetic focus for studies where the structure of the general community is of interest. Compared to other group-specific primers, the primers chosen for testing were more specific or matched a greater percentage of the targeted group in computer analyses. Their breadth within each group also compared favorably with commonly used domain-specific or universal primers (Table 3). The environment of particular interest to us is soil, which has been shown to harbor a highly diverse and complex microbial community (for example, see reference 11). The six groups for which we successfully tested primers are important components of the soil community but do not provide a complete picture. To maintain the maximum phylogenetic breadth of analysis, primers will have to be similarly tested for additional groups of Bacteria (22), Archaea, and other microbial Eucarya.

It is important for the specificity and breadth of primers to be checked periodically (4, 9). We were surprised to find that many previously published phylum- or class-specific PCR primers or probes did not match a substantial number of target sequences or matched large numbers of nontarget sequences. This can be explained in part by the evolving taxonomy of microbial species and continual expansion of the diversity of sequences in public databases. Neither of these factors is likely to decrease, and we recognize that the primers tested in this study will probably be improved upon in the future. Any primers should at least be compared to target and nontarget database sequences before being adopted for a given study.

We acknowledge the help of Stan Tesch for technical assistance.

A.O. was supported by a Howard Hughes Medical Institute Undergraduate Research Fellowship through the University of Maryland College of Life Sciences.