Development of 16S rRNA-based probes for the identification of Gram-positive anaerobic cocci isolated from human clinical specimens


Corresponding author and reprint requests: A. C. M. Wildeboer-Veloo, Department of Medical Microbiology, UMCG, University of Groningen, PO Box 30001, 9700 RB Groningen, The Netherlands


Fluorescent probes targeted at 16S rRNA were designed for Peptostreptococcus anaerobius and Peptostreptococcus stomatis (Pana134), Parvimonas micra (Pamic1435), Finegoldia magna (Fmag1250), Peptoniphilus asaccharolyticus (Pnasa1254), Peptoniphilus ivorii (Pnivo731), Peptoniphilus harei (Pnhar1466), Anaerococcus vaginalis (Avag1280) and Anaerococcus lactolyticus (Alac1438), based on the 16S rRNA sequences of reference strains and 88 randomly chosen clinical isolates. These strains were also used for validation of the probes. Application of the probes to an additional group of 100 clinical isolates revealed that 87% of Gram-positive anaerobic cocci (GPAC) could be identified with this set of probes. The 16S rRNAs of 13 clinical isolates that could not be identified were sequenced. Most of these isolates were GPAC that were not targeted by the probes. No clinical isolates of Pn. asaccharolyticus were encountered. Near full-length sequences were obtained from 71 of 101 (n = 88 + 13) sequenced clinical isolates. Of these, 25 showed <98% similarity with the homologues of the closest established species. The Fmag1250, Pamic1435, Pnhar1466, Pana134, Pnasa1254 and Pnivo731 probes allowed reliable identification and hybridised with all corresponding isolates. The Avag1280 and Alac1438 probes failed to hybridise with two isolates and one isolate, respectively, because of intra-species variation. However, overall, the set of probes yielded fast and reliable identification for the majority of clinical isolates.


Gram-positive anaerobic cocci (GPAC) are part of the commensal microbiota of humans at different sites of the body, and are also known to be important in human disease. They account for about one-third of the anaerobic isolates recovered from clinical material [1]. During the last decade, the taxonomy of GPAC has changed substantially [2–4]. Peptostreptococcus productus was transferred to the genus Ruminococcus[5], and the genus Peptostreptococcus was divided into six new groups [6,7]. Murdoch and Shah [7] transferred the species Peptostreptococcus micros and Peptostreptococcus magnus to two new genera, Micromonas and Finegoldia, respectively, with each being the only species present in their respective genus. The genus Micromonas has now been replaced by Parvimonas, with Parvimonas micra (formerly Micromonas micros) being the only species in the genus [8]. Ezaki et al. [6] divided the remaining peptostreptococci into three phylogenetic groups, Peptoniphilus gen. nov., Anaerococcus gen. nov. and Gallicola gen. nov., with Gallicola barnesae being the only species present in the latter genus. The type species of the two other new genera are Peptoniphilus asaccharolyticus and Anaerococcus prevotii, respectively. The only species remaining in the genus Peptostreptococcus are Peptostreptococcus anaerobius and a recently described species, Peptostreptococcus stomatis, isolated from the human oral cavity [9].

Infections involving GPAC are often polymicrobial [10]. The GPAC isolated most commonly from infections are P. anaerobius, Pa. micra, Finegoldia magna and Pn. asaccharolyticus[11]. However, although these anaerobic cocci are isolated commonly from infections, little attention is paid to their detection and identification in diagnostic laboratories because of cumbersome and inadequate classification systems. The phenotypic identification of GPAC is based on morphological appearance, carbohydrate formation and the use of gas–liquid chromatography [12]. The availability of proteolytic enzyme profiles made identification of the acknowledged GPAC easier and reproducible [13]. This contributed to the development of several commercial identification kits, including the RapID ANA (Innovative Diagnostic Systems, Atlanta, GA, USA) and Rapid ID 32A (bioMérieux, Basingstoke, UK) systems. However, the newly described species have not yet been included in the databases of these kits. In addition, several new species, assigned originally to the genus Peptostreptococcus, have been described, e.g., Peptoniphilus harei, Peptoniphilus ivorii and Anaerococcus octavius[14]. The clinical relevance of these new species has not yet been assessed.

The availability of genotypic data now makes it possible to develop molecular techniques for the detection and identification of GPAC. Song et al. [15] evaluated the use of 16S rRNA sequences to identify GPAC. These authors were able to identify 84% of clinical isolates of GPAC by comparison with their own sequence database of type strains. These sequences revealed ambiguous data in public databases. Song et al. [16] developed a multiplex PCR assay for the rapid identification of GPAC, using genus- and species-specific primers. Riggio et al. [17–19] developed additional molecular detection assays for GPAC, including a PCR method for the detection of F. magna and Pa. micra in oral clinical specimens, and a PCR–restriction fragment length polymorphism assay of 16S rRNA genes for the identification of oral Peptostreptococcus isolates. DNA probes have also been used for the identification of Pa. micra[20] and P. anaerobius[21]. Amplification of the 16S−23S intergenic spacer region was used by Hill et al. [22] to differentiate among different species of the former peptostreptococci. When the banding patterns of 38 test strains were compared with those of reference strains, fewer than half of the strains were identified. The remaining strains could not be identified, either because of intra-species variation or because they differed significantly from the type strain.

GPAC have also been detected using PCR in faecal samples [23], and have been quantified using fluorescent in-situ hybridisation [24]. The latter technique makes it possible to detect and identify GPAC in pure culture, and perhaps also directly in clinical specimens. This technique is relatively inexpensive to perform, and can be implemented easily in routine diagnostic laboratories. The present report describes the design and validation of 16S rRNA-based probes for the detection and identification of clinically relevant GPAC. Probes were designed for a selection of GPAC isolated in our laboratory, namely, P. anaerobius/stomatis, Pa. micra, F. magna, Pn. asaccharolyticus, Anaerococcus vaginalis, Anaerococcus lactolyticus, Pn. ivorii and Pn. harei.

Materials and methods


Reference strains and clinical isolates were cultured on Brucella blood agar anaerobic plates and incubated in an anaerobic atmosphere for 48 h at 37°C. Cells were harvested and fixed for fluorescent in-situ hybridisation analyses in 1:1 phosphate-buffered saline (8 g of NaCl, 0.2 g of KCl, 1.44 g of Na2HPO4 and 0.24 g of KH2PO4/L) and ethanol 96% v/v. Fixed cells were stored at −20°C. In total, 188 random clinical isolates of GPAC were obtained from the diagnostic laboratory of the University Medical Center, Groningen, and the Regional Public Health Laboratory, Enschede, The Netherlands. The original clinical samples were obtained from a variety of anatomical sites, e.g., abdomen, head/neck area, leg, arm and groin. Only one isolate of each species from a single patient was included in the study. The isolates were identified phenotypically using the Rapid ID 32A system and the Wadsworth manual [12]. Isolates that could not be identified, or only ambiguously, were designated as GPAC. Phenotypically, Pn. asaccharolyticus can be distinguished from Pn. harei only by colony form and cell morphology [12]. Since this is difficult to achieve, these organisms were designated Pn. asaccharolyticus/harei.

16S rRNA gene sequencing

The 16S rRNA genes of 88 randomly chosen clinical isolates were sequenced. DNA was isolated as described previously [25] and the 16S rRNA genes were amplified and sequenced using universal 16S rRNA-specific primers [26]. The resulting sequences were aligned with sequences of reference strains derived from the EMBL database and the Ribosomal Database Project [27], using ARB software [28]. Similarities of the sequences were calculated using a DNA distance matrix embedded in BioEdit ( The Escherichia coli sequence for positions 82–1371 was included to ensure accurate determination of species similarity. A phylogenetic tree of sequences that had <98% similarity with their closest relative was constructed using the same alignment by the neighbour-joining method with Jukes Cantor correction, based on a distance matrix that only included positions with >50% conservation and parsimony, implemented in the ARB software. As a control, the phylogenetic tree was also calculated using the maximum-parsimony and maximum-likelihood methods. The topology of the tree was calculated following bootstrap analyses of 1000 replicate trees. Sequences with <98% similarity were deposited in the EMBL database and assigned accession numbers.

Probe design

Probes (Table S1, see Supplementary material) were designed using the sequences of clinical isolates and reference strains present in the Ribosomal Database Project [27] and EMBL database. ARB software was used for alignment and probe design [28]. When necessary, unlabelled helper nucleotides (Table S2, see Supplementary material) were designed to increase the in-situ accessibility of 16S rRNA, as described by Fuchs et al. [29]. For practical purposes, probes were designed to have a similar hybridisation temperature. All probes were labelled with fluorescein-5-isothiocyanate at the 3′ and 5′ ends and were synthesised by Eurogentec (Seraing, Belgium).


Ethanol-fixed cells were spotted on a slide and fixed for a further 10 min using ethanol 96% v/v. Hybridisations were performed at 50°C in hybridisation buffer (0.9 M NaCl, 20 mM Tris-HCl, pH 7.2, SDS 0.1% w/v) containing 10 ng of probe, as described previously [30]. For the Pana134, Pivo731 and Avag1280 probes (Table S1), 5 ng of the appropriate helper nucleotides (Table S2) were added to the hybridisation mixture. Hybridisations were performed overnight. The bacterial probe EUB338 [31] served as a positive control, and its complement non-EUB338 [32] as the negative control. The specificities of the Pana134 and Pnasa1254 probes were increased by adding formamide 20% v/v, which is a duplex-destabilising agent, to the hybridisation buffer. The hybridisation signal obtained with the species-specific probes was compared visually with the EUB338 and non-EUB338 signals, using an Epifluorescence BH2 microscope (Olympus, Hamburg, Germany), and scored negative or positive. Strains of F. magna, A. vaginalis and Pn. ivorii were permeabilised with proteinase K (500 mg/L; 50 mM Tris-HCl, pH 7.6). The cell spots were covered with the proteinase K solution and incubated for 10 min at room temperature. The enzymic reaction was stopped by incubating the slides in ethanol 96% v/v for 2 min. All probes were validated against the reference strains and sequenced clinical isolates listed in Table 1. In order to assess the proportion of GPAC that could be identified, the set of species-specific probes was also applied to an additional group of 100 clinical isolates.

Table 1.   Reference strains and sequenced clinical isolates used for validation of the species-specific probes for Gram-positive anaerobic cocci
  • a

    To increase specificity of the hybridisation, formamide 20% v/v was added to the hybridisation buffer.

  • b

    Unlabelled helper nucleotides were added to the hybridisation mixture to increase the in-situ accessibility of the 16S rRNA.

  • c

    Before hybridisation, cells were permeabilised with proteinase K 500 mg/L for 10 min at room temperature.

  • d

    All MMB strains are clinical isolates identified by routine procedures.

  • e

    All gpac strains are sequenced clinical isolates used in this study.

  • DSM, Deutsche Sammlung von Mikrooganismen und Zellkulturen (Braunschweig, Germany); ND, not determined.

Finegoldia magna DSM 20470Tc+
Parvimonas micra DSM 20468T+
Peptoniphilus harei DSM 10020T+
Peptostreptococcus anaerobius DSM 2949T+
Anaerococcus vaginalis DSM 7457Tc+
Anaerococcus lactolyticus DSM 7456T+
Peptoniphilus ivorii DSM 10022Tc+
Peptoniphilus asaccharolyticus DSM 20463T+
Anaerococcus prevotii DSM 20548T
Anaerococcus tetradius MMBd
Peptoniphilus indolicus DSM 20464T
Clostridium sporosphaeroides DSM 1294TNDNDNDNDNDNDND
Dermabacter hominis DSM 7083TNDNDNDNDNDNDND
Clostridium colinum DSM 6011TNDNDNDNDNDNDND
Eubacterium tenue DSM 20695TNDNDNDNDNDNDND
Clostridium glycolicum DSM 1288TNDNDNDNDNDNDND
Clostridium difficile DSM 1296TNDNDNDNDNDNDND
Enterococcus moraviensis DSM 15919TNDNDNDNDNDNDND
Clostridium beijerinckii MMBNDNDNDNDNDNDND


The set of probes used in this study demonstrated 100% specificity when validated against the reference strains listed in Table 1. F. magna, A. vaginalis and Pn. ivorii required permeabilisation with proteinase K before hybridisation. All other strains hybridised successfully without additional treatment. The probes were also validated against a further set of 88 sequenced clinical isolates (Table 2) and were shown to be 100% specific.

Table 2.   Validation of the newly designed probes against 88 clinical isolates, the 16S rRNA gene sequences of which were used for the design of the probes
Species nameIdentification (n)
Finegoldia magna2626
Parvimonas micra1313
Peptoniphilus harei1616
Anaerococcus lactolyticus77
Anaerococcus vaginalis65
Peptoniphilus ivorii55
Peptostreptococcus anaerobius55
Peptoniphilus lacrimalis3 
Anaerococcus tetradius1 
Anaerococcus hydrogenalis1 
Peptococcus niger1 
Atopobium parvulum2 
Ruminococcus gnavus1 
Peptostreptococcus sp. E3_321 

The set of probes was then applied to an additional group of 100 clinical isolates that had been identified phenotypically (Table 3). The majority (29%) of the isolates was identified as F. magna on the basis of both genotypic and phenotypic characteristics. The Pamic1435 probe hybridised with 28% of the clinical isolates. Three of these isolates were misidentified phenotypically as F. magna; one could not be identified.

Table 3.   Application of the species-specific probes to 100 clinical isolates identified phenotypically
Probe identification (n)Phenotypic identification (n)Sequence identification (n)
  1. GPAC, Gram-positive anaerobic cocci.

Finegoldia magna (29)F. magna (29) 
Parvimonas micra (28)Pa. micra (24) 
F. magna (3) 
GPAC (1) 
Peptoniphilus harei (17)Peptoniphilus asaccharolyticus/ harei (13) 
Peptoniphilus indolicus (1) 
GPAC (3) 
Anaerococcus lactolyticus (1)GPAC (1) 
Anaerococcus vaginalis (2)A. vaginalis (2) 
Peptoniphilus ivorii (6)Pn. ivorii (4) 
GPAC (2) 
Peptostreptococcus anaerobius (4)P. anaerobius (4) 
Not identified (13)Peptococcus niger (2)Pc. niger (2)
A. vaginalis (1)A. vaginalis (1)
Pn. lacrimalis (1)A. vaginalis (1)
F. magna (1)A. lactolyticus (1)
Sarcina ventriculi (1)S. ventriculi (1)
GPAC (7)Atopobium parvulum (3)
Anaerococcus tetradius (1)
Peptoniphilus octavius (1)
Not identified (1)
Bacterium N14-24 (1)

The Pana134 probe hybridised with 4% of the clinical isolates, all identified phenotypically as P. anaerobius. The weak hybridisation signal with this probe was increased by adding the unlabelled helper nucleotides H115 and H155 (Table S2). The Pnhar1466 probe hybridised with 17% of the clinical isolates, four of which were misidentified phenotypically, one as Peptoniphilus indolicus, and three that could not be identified according to the current identification scheme. Of all isolates, 6% were identified genotypically as Pn. ivorii, with four isolates phenotyped correctly and two incorrectly. The Avag1280 probe hybridised with two (2%) clinical isolates, which was in agreement with the phenotypic identification. The hybridisation signal of the Pnivo731 and Avag1280 probes was increased using the helper nucleotides H716 + H750 and H1263 + H1299, respectively (Table S2). The Alac1438 probe hybridised with one of the clinical isolates (1%) that could not be identified phenotypically.

Overall, the set of probes allowed identification of 87% of the GPAC isolates (Table 3). The 16S rRNA genes of the remaining 13 isolates were sequenced to determine whether the new probes had hybridised correctly. The sequence results revealed two isolates of Peptococcus niger, two of A. vaginalis, one of A. lactolyticus and one of Sarcina ventriculi. Of these six isolates, one of the A. vaginalis isolates was misidentified phenotypically as Peptoniphilus lacrimalis, and the A. lactolyticus isolate was misidentified as F. magna. The four other isolates were identified correctly using the phenotypic system. Seven remaining isolates could not be identified phenotypically; sequence analysis revealed three Atopobium parvulum, one Anaerococcus tetradius, one Pn. octavius and one Bacterium N14-24. One isolate could not be identified because amplification of the 16S rRNA gene was unsuccessful. Sequence analysis of the 16S rRNA genes of the A. vaginalis and A. lactolyticus isolates revealed that the corresponding probes failed to hybridise with these isolates because of mismatches. The A. lactolyticus isolates had one mismatch with the Alac1438 probe and the two A. vaginalis isolates had one and three mismatches, respectively.

A distance matrix was calculated from all sequence results with a near complete full-length sequence. Partially determined sequences were suitable for probe design, but not for phylogenetic analysis. Near full-length sequences were obtained for 71 isolates, of which 25 showed <98% similarity with their closest established species (Table 4). Using these sequences, a phylogenetic tree was constructed. The calculation was performed using three different tree construction methods, each of which gave the same result for the significant branchings. The phylogenetic tree obtained using the neighbour-joining method is shown in Fig. 1. All sequences of F. magna, P. anaerobius, A. tetradius, At. parvulum, Pc. niger, Ruminococcus gnavus and Bacterium N14-24 showed ≥98% similarity to the corresponding reference strains, while more than half of the sequences of Pa. micra showed <98% similarity to the reference strain. In addition, some sequences of Pn. harei, A. vaginalis and Pn. lacrimalis showed <98% similarity to their respective closest established species. All sequences of Pn. ivorii, A. lactolyticus, Anaerococcus hydrogenalis and Peptostreptococcus sp. E3_32 showed <98% similarity to the corresponding reference strains.

Table 4.   Similarity of the 16S rRNA sequences of clinical isolates to those of reference strains, calculated using the DNA distance matrixa
≥98% (n)<98% (n)Total (n)
  • a

    Only sequencing experiments that yielded a near full-length sequence were included in this analysis.

Finegoldia magna19019
Parvimonas micra3710
Peptoniphilus harei7411
Peptoniphilus ivorii033
Peptostreptococcus anaerobius303
Anaerococcus lactolyticus077
Anaerococcus vaginalis415
Peptoniphilus lacrimalis112
Anaerococcus tetradius202
Atopobium parvulum404
Peptococcus niger101
Ruminococcus gnavus101
Bacterium N14-24101
Peptostreptococcus sp. E3_32011
Anaerococcus hydrogenalis011
Figure 1.

 Phylogenetic tree showing the relationship between Gram-positive anaerobic cocci and clinical isolates (shown in bold) with <98% similarity to their closest relative. The percentages of similarity are indicated. The neighbour-joining tree was constructed using an alignment corresponding to Escherichia coli base-pair positions 82–1371. Only bootstrap values >90% are shown. The bar indicates 10% sequence divergence. Note: Peptostreptococcus sp. E3_32 should be renamed Peptoniphilus sp. E3_32.


In total, 188 isolates of GPAC were analysed in this study, using 16S rRNA-targeted probes and/or sequencing of the 16S rRNA gene to obtain a genotypic identification. Cell morphology is used as an initial first step in the identification of GPAC; however, the results are not always unambiguous, as illustrated by the sequence results (Tables 2 and 3), which revealed the presence of At. parvulum, which is actually a coccobacillus similar to Slackia heliotrinreducens.

The GPAC encountered most frequently were F. magna (29%), Pa. micra (22%), Pn. harei (18%), Pn. ivorii (6%), A. vaginalis (5%), A. lactolyticus (5%) and P. anaerobius/stomatis (5%). Pn. asaccharolyticus was not observed, although previous studies have reported that this is one of the most frequently encountered GPAC in infections [11]. This is probably because Pn. asaccharolyticus can be phenotypically distinguished from Pn. harei only by its colony form and cell morphology [12]. The present data suggest that Pn. harei has probably been misidentified as Pn. asaccharolyticus in the past. Of all the GPAC analysed, 18% were identified genotypically as Pn. harei, indicating that this is clearly a clinically relevant species that can be isolated from a variety of sites, as also suggested by Song et al. [16], who identified 48 (25%) of 190 isolates as Pn. harei. In the present study, 17 isolates were identified as Pn. harei by phenotypic methods, most of them as Pn. asaccharolyticus/harei. One isolate was misidentified as Pn. indolicus and three isolates could not be identified. The four isolates identified with the Pana134 probe were identified phenotypically as P. anaerobius. All strains of F. magna were correctly identified on the basis of phenotypic characteristics. In contrast, A. lactolyticus strains were either misidentified as F. magna, or could not be identified. Most of the Pn. ivorii, Pa. micra and A. vaginalis strains were also correctly identified phenotypically.

Several technical problems required resolution. First, the Pana134, Pnivo731 and Avag1280 probes initially gave a weak hybridisation signal. Fuchs et al. [29] reported that the inclusion of unlabelled helper oligonucleotides during hybridisation could increase the in-situ accessibility of 16S rRNA. The use of unlabelled helper oligonucleotides that bind adjacent to the probe target site (Table S2) was successful in increasing the hybridisation signal. Second, the initial hybridisation results with F. magna, A. vaginalis and Pn. ivorii showed no signal for EUB338 and the specific probes. This was resolved by treating these large cocci with proteinase K to increase permeabilisation before hybridisation. Thus, consideration of cell morphology may facilitate the selection of probes for hybridisation: for large cocci, the Fmag1250, Avag1280 and Pnivo731 probes should be used; for middle-sized cocci, the probes Pnasa1254, Alac1438 and Pnhar1466; for small cocci, the Pamic1435 probe; and for cocci in chains, the Pana134 probe. However, it is also possible to use the complete set of probes.

The Avag1280 and Alac1438 probes failed to hybridise with some of the corresponding isolates, and analysis of the sequence data revealed mismatches of the probes with the clinical isolates. Hill et al. [22] revealed the heterogeneity of GPAC by analysis of 16S−23S intergenic rRNA polymorphisms, demonstrating a considerable intra-species variation for A. vaginalis, while all members of Pn. ivorii produced identical banding patterns. Analysis of the sequence data from the present study confirmed the intra-species variation demonstrated by Hill et al. [22], and revealed, in particular, that the A. lactolyticus group showed extensive heterogeneity in 16S rRNA sequences. When designing species-specific probes for species with such heterogeneity, the probability exists that probes will be restricted in specificity and may have mismatches with the corresponding isolates. Attempts were made to overcome this by including the sequences of clinical isolates in the probe design, and this was successful for most of the newly designed species-specific probes. Only the Avag1280 and Alac1438 probes showed mismatches with corresponding isolates. Among the 71 isolates for which a nearly full-length sequence was obtained, 25 had <98% similarity to the closest established species. This is probably because of the intra-species variation within GPAC, but such isolates might also be considered as belonging to new species.

Since the newly designed probes were validated and directed against Gram-positive bacteria, all hybridisations were performed overnight to obtain an optimal hybridisation signal. A shorter hybridisation time (2 h) is also possible, although the hybridisation signal for some GPAC will be less detectable than that after overnight hybridisation. With the current probe set, 87% of all GPAC isolates were identified quickly and reliably. These probes make it possible to differentiate Pn. asaccharolyticus from Pn. harei, which is difficult with phenotypic methods. The probe-based identification method is more reliable than phenotypic methods, especially for Pn. harei, Pn. ivorii and A. lactolyticus. Moreover, the cost of phenotypic methods is higher than that of fluorescent in-situ hybridisation. The probes also have the potential for direct application to clinical material in order to provide fast and direct detection without costly anaerobic cultivation.


We would like to thank the technicians of the anaerobic workgroup and the PCR group for their contribution to the practical work. We are indebted to E. Roelofsen for providing clinical isolates. We thank S. Smith for critical reading of the manuscript. None of the authors has commercial relationships. This study was presented, in part, at the 8th Biennial Congress of the Anaerobe Society of the Americas (Boise, ID, USA, 2006).