Evaluation of Universal Probes and Primer Sets for Assessing Total Bacterial Load in Clinical Samples: General Implications and Practical Use in Endodontic Antimicrobial Therapy

Abstract

By reexamining 10 previously published “universal” PCR assays using the ARB phylogenetic software package and database with 41,000 16S rRNA gene sequences, we found that they differed considerably in their coverage of the domain Bacteria. We evaluated the broadest-range real-time quantitative PCR protocol for its efficacy in measuring the antimicrobial effects of endodontic treatments.

Bacteria in a tooth's root canal both initiate and perpetuate periapical inflammatory lesions (7). Thus, the principal goal of root canal treatment is to reduce the number of bacteria (1); conversely, the principal cause of treatment failure is considered to be the residual bacteria in the apical part of the root canal (18, 24). To evaluate the efficacy of any antimicrobial therapy, it is therefore important to determine the total number of bacteria in the root canal before and after treatment. For the enumeration of bacteria, the real-time quantitative PCR (RTQ-PCR) technology represents a promising alternative to the traditional but time-consuming and error-prone cultivation approach. However, the microbiota involved in endodontic infections is not a predefined group of pathogens but makes up a complex, dynamic, and varying consortium of many oral bacterial species (5, 22, 23). Thus, any RTQ-PCR-based evaluation of an endodontic treatment ideally should be able to measure the entire bacterial load without missing certain taxa. While several PCR-based pitfalls due to cell lysis techniques or PCR conditions have been reported (31), the universality of “universal” PCR primers has been less fully evaluated. Numerous broad-range 16S rRNA gene-directed primers and probes have been developed with the intention of targeting all bacteria present in clinical or environmental samples (10, 21, 26); to this end, the quality of PCR assays is usually confirmed by testing representative species of a wide range of bacterial taxa. However, since it is impossible to empirically test all bacterial strains it cannot be proven whether those PCR assays considered to be universal actually encompass the entire bacterial spectrum. For example, some evidence exists that even the prominent pair of primers, “27F” and “1492R” (32), which target highly conserved regions of the 16S rRNA gene, are not completely universal (3, 20, 28). For every PCR-based assay for the detection of bacteria in clinical samples in general and for the enumeration of endodontic bacteria in particular, it is therefore crucial to know the taxon coverage of the PCR system used. We addressed this issue by evaluating in silico 10 previously published broad-range 16S rRNA gene-directed PCR assays using the most updated ARB database (13). ARB is a graphically oriented software package that comprises various tools for database handling and sequence analysis. The special advantage of ARB is the development of a structured database of more than 41,000 validated 16S rRNA gene sequences in an aligned format that includes all recognized division-level lineages of the domain Bacteria. The high number of interacting software tools integrated in ARB permits not only a general probe evaluation against all sequences but also a disclosure of all phylogenetic groups (including clinically relevant taxa) that are not covered.

Using the function “probe match” in ARB, we determined for each assay the proportion of 16S rRNA gene sequences that perfectly matched with the primers and, if given, with the hybridization probe (Table 1). Although the site and the type of a mismatch are not equally critical for successful amplification in every case, the mere presence of a mismatch can lead to a biased retrieval of different 16S rRNA gene sequence types in a multitemplate PCR assay (25, 31) and, ultimately, to inaccurate quantification. Therefore, we considered only perfect matches (i.e., no mismatch between probe and target DNA) for estimating the “universal” capacity of the PCR assays. Using these criteria, we observed significant differences among protocols, as seen in Table 1. The PCR assays used by Siqueira et al. (23), Corless et al. (2), and Khan et al. (8) showed only a low incidence of perfect matches (5 to 17%); and protocols published by Labrenz et al. (12) and Yang et al. (33) indicated perfect matches with 27% and 35% of the sequences, respectively. The RTQ-PCR described by Klaschik et al. (9) had a 41% coverage; however, the gram-positive and gram-negative organism-specific hybridization probes matched only a very small proportion of the sequences included in ARB. In contrast, we observed a much higher percentage of perfect matches with the protocols used by Takai et al. (26), Tseng et al. (27), Maeda et al. (14), and Nadkarni et al. (17), with the last protocol having the highest scores for both PCR amplification and probe hybridization (i.e., 74% and 63%, respectively). However, as is evident from Table 2, no PCR protocol includes all taxonomic groups (i.e., phyla), and among themselves, the protocols vary strongly in their individual coverage. For example, although it is superior to all the other assays, the protocol of Nadkarni et al. (17) covers the phyla chlamydiae and spirochetes only poorly (Table 2), with the latter phylum including such clinically relevant genera as Treponema and Borrelia. These data reflect the difficulty of designing a broad-range protocol which would evenly cover all taxonomic groups. Because a “perfect” universal assay is lacking, we focused on the protocol of Nadkarni et al. (17) as the one that came the closest to being perfect (according to overall coverage) and whose general methodological characteristics (e.g., reproducibility and sensitivity) had been extensively validated (11, 15, 17).

TABLE 1.

Overview of broad-range bacterial PCR assays and the proportion of perfect matches between primers (and the probe, if given) with the target molecule, determined by in silico analysis of all 16S rRNA gene sequences included in the ARB database^a

Assay no.	Primer or probe	Sequence (5′-3′)	Escherichia coli position	Reference	Intended application as published	% Perfect matchesb	Detection system
1	Forward	TCCTACGGGAGGCAGCAGT	331-349	Nadkarni et al. (17)	RTQ-PCR detection of bacteria in carious dentine	74, 63	ABI PRISM 7700, TaqMan (ABIc)
	Reverse	GGACTACCAGGGTATCTAATCCTGTT	772-797
	Probe	CGTATTACCGCGGCTGCTGGCAC	506-528
2	Forward	GTGSTGCAYGGYTGTCGTCA	1048-1067	Maeda et al. (14)	RTQ-PCR detection of bacteria in dental plaque	73	GeneAmp 5700, SYBR green (ABI)
	Reverse	ACGTCRTCCMCACCTTCCTC	1175-1194
3	Forward “p201”	GAGGAAGGIGIGGAIGACGT	1175-1194	Tseng et al. (27)	RTQ-PCR detection of bacterial pathogens	71	GeneAmp 5700, Sybr green (ABI)
	Reverse “p1370”	AGICCCGIGAACGTATTCAC	1371-1390
4	Forward “Bac349F”	AGGCAGCAGTDRGGAAT	349-365	Takai and Horikoshi (26)	RTQ-PCR detection of bacteria in various environments	64, 55	GeneAmp 5700, TaqMan (ABI)
	Reverse “Bac806R”	GGACTACYVGGGTATCTAAT	787-806
	Probe “Bac516F”	TGCCAGCAGCCGCGGTAATACRDAG	516-540
5	Forward “PLK1”	TACGGGAGGCAGCAGT	343-358	Klaschik et al. (9)	Gram-specific RTQ-PCR detection of 17 intensive care unit relevant pathogens	41, 3 (gram-negative bacteria), 0.1 (gram-positive bacteria)	LightCycler (Roche)
	Reverse “PLK2”	TATTACCGCGGCTGCT	520-535
	Probe “ISN2”	CCGCAGAATAAGCACCGGCTAACTCCGT	489-516
	Probe “ISP2”	CCTAACCAGAAAGCCACGGCTAACTACGTG	488-517
6	Forward “p891F”	TGGAGCATGTGGTTTAATTCGA	943-964	Yang et al. (33)	RTQ-PCR detection of bacterial pathogens	39, 35	ABI PRISM 7700, TaqMan (ABI)
	Reverse “p1033R”	TGCGGGACTTAACCCAACA	1083-1101
	Probe “UniProbe”	CACGAGCTGACGACARCCATGCA	1052-1074
7	Forward “Com1”	CAGCAGCCGCGGTAATAC	519-536	Labrenz et al. (12)	RTQ-PCR analysis of bacteria in aquatic systems	27	iCycler, SYBR green (Bio-Rad)
	Reverse “Com2”	CCGTCAATTCCTTTGAGTTT	907-926
8	Forward “ANA1F”	GCCTAACACATGCAAGTCGA	46-65	Khan et al. (8)	T-RFLPd analysis of intestinal microflora	17
	Reverse “K2R”	GTATTACCGCGGCTGCTGG	518-536
9	Forward	CCATGAAGTCGGAATCGCTAG	1327-1347	Corless et al. (2)	RTQ-PCR detection of bacterial pathogens	8, 7	ABI PRISM 7700, TaqMan (ABI)
	Reverse	ACTCCCATGGTGTGACGG	1403-1420
	Probe	CGGTGAATACGTTCCCGGGCCTTGTAC	1369-1395
10	Forward “968f”	AACGCGAAGAACCTTAC	968-984	Siqueira et al. (23)	DGGEe analysis of endodontic bacteria	5
	Reverse “1401r”	CGGTGTGTACAAGACCC	1385-1401

^aOnly PCR assays that amplified fragments of less than 500 bp were considered in order to meet the ABI guidelines for RTQ-PCR. Only assays with primers located within E. coli positions 45 and 1430 (the testable range of the ARB sequences) were included in the study. The list of PCR protocols may not be complete. The ARB database currently consists of 41,019 16S rRNA gene sequences of at least 1,000 bp in length in an aligned format (last update, 2004).

^bNumbers indicate the proportion of 16S rRNA gene sequences that showed a perfect match with both the forward and the reverse primers; underlining indicates the proportion of perfect matches with both primers and the detection probe.

^cABI, Applied Biosystems.

^dT-RFLP, terminal restriction fragment length polymorphism.

^eDGGE, denaturing gradient gel electrophoresis.

TABLE 2.

Coverage of selected bacterial phyla by 10 different broad-range PCR assays^a

Reference	% Coverage of the following phyla (no. of sequences in ARB):
	Proteobacteria (18,920)	Actinobacteria (5,348)	Firmicutes (8,022)	Spirochetes (852)	Chlamydiae (168)	Bacteroidetes (2,677)	Further phyla (5,032)
Nadkarni et al. (17)	86, 74	82, 72	81, 69	1, <1	0	83, 76	16, 9
Maeda et al. (14)	79	85	72	44	1	61	54
Tseng et al. (27)	77	77	73	57	79	62	38
Takai and Horikoshi (26)	63, 56	82, 73	80, 69	<1, <1	0	87, 65	18, 8
Klaschik et al. (9)b	67, 4	1, 0	8, <1	<1, 0	0	84, <1	20, 0
Yang et al. (33)	47, 42	<1, <1	55, 51	88, 85	82, 81	16, 12	21, 17
Labrenz et al. (12)	33	1	3	<1	<1	83	40
Khan et al. (8)	29	1	9	<1	0	11	9
Corless et al. (2)	16, 16	0	<1, <1	0	0	<1, 0	<1, <1
Siqueira et al. (23)	<1	3	20	0	<1	0	1

^aNumbers indicate the proportion of 16S rRNA gene sequences within the particular phylum that showed a perfect match with both the forward and the reverse primers; underlining indicates the proportion of perfect matches with both primers and the detection probe. The assignment of sequences from uncultured bacteria to the indicated phylum was based on the topology of the universal tree implemented in ARB.

^bFor the protocol of Klaschik et al. (9), underlined values within the actinobacteria and the firmicutes were retrieved by testing with the gram-positive organism-specific probe; all other phyla were tested by use of the gram-negative organism-specific probe.

Since to our knowledge the RTQ-PCR technology has not so far been applied to the field of endodontics, our aim was to test the protocol of Nadkarni et al. (17) for its principal applicability in quantifying endodontic bacteria. To accomplish this, we measured the bacterial load before and after applying two different intracanal irrigating substances. Thirty-two patients who presented for root canal treatment at the Piracicaba Dental School and who were otherwise healthy and who had not received antibiotic treatment during the previous 3 months were selected for this study. Their ages ranged from 19 to 63 years. All teeth selected were uniradicular and asymptomatic, did not respond to sensitivity testing, had not received previous endodontic treatment, and showed radiographic evidence of periapical bone loss. The teeth were randomly divided into two treatment groups: the 2.5% NaOCl (group 1, n = 16) and the 2% chlorhexidine (CHX) gel (group 2, n = 16). The irrigating substances were prepared according to Vianna et al. (29).

Access to the pulp chamber and sample collection (before and after endodontic procedures) were performed by the protocol described by Jacinto et al. (6). The initial samples were collected and transported to the laboratory within 15 min. Aliquots (100 μl) were immediately processed for culture analysis, while 900 μl was frozen (−70°C) for later molecular analysis. The working length (1 mm from the radiographic apex) was established with a radiograph and was confirmed with an apical locator (Novapex; Forum Technologies, Israel). The apical preparation was performed by using K files (DYNA-FIDM, Bourges, France), followed by the use of Step-Back preparation. In the first group, the root canal was irrigated with 5 ml of 2.5% NaOCl after each filing; and in the second group, the root canal was irrigated with 1 ml of the CHX gel and immediately after with 4 ml of physiological saline solution. The working time for the chemomechanical procedure was 20 min for all cases. Before collection of the second sample, the root canal was rinsed for 1 min with 5 ml of irrigating neutralizers (for the NaOCl group, 0.5% sodium thiosulfate, 0.5% Tween 80, and 0.07% lecithin). The time that elapsed for the subsequent processing of the second samples was identical to the time required for the initial sample set (see above). Finally, all teeth were filled and the access cavities were restored with 2 mm of Cavit and resin (Z-250; 3M Dental Products, St. Paul, Minn.).

In order to determine the RTQ-PCR-measurable scale of bacterial reduction, we used both the SYBR green and the TaqMan formats (i.e., SYBR green PCR master mix and TaqMan PCR master mix, respectively; Applied Biosystems). The PCR conditions used were different for both assays: (i) for the TaqMan format, denaturation at 94°C for 10 min and 40 cycles of 94°C for 1 min and annealing at 60°C for 1 min and 45 s; (ii) for the SYBR green format, denaturation at 94°C for 10 min and 40 cycles of 94°C for 1 min, annealing at 60°C for 1 min, and elongation at 72°C for 1 min and 30 s, followed by a final elongation at 72°C for 5 min. Melting curve analysis was performed to assess reaction specificity. RTQ-PCR was performed with the aid of an ABI-PRISM 7000 sequence detection system (Applied Biosystems, Foster City, Calif.) by using optical-grade 96-well plates. Samples were run in duplicate in a total volume of 25 μl. Final reaction mixtures contained 100 nM of each primer and 2 μl of template DNA (approximately 50 ng of template DNA). Data acquisition and subsequent analysis were performed with ABI-PRISM 7000 SDS software (Applied Biosystems). DNA extracted from Prevotella nigrescens ATCC 33563 was used to establish the standard curve, based on a series of 10-fold dilutions. The bacterial load was quantified by determining the cycle threshold (C_T), i.e., the number of PCR cycles required for the fluorescence to exceed a value significantly higher than the background fluorescence. We assumed a threshold value of 0.2, which was approximately 10 times the background fluorescence, defined as the mean fluorescence values of the first 6 to 15 PCR cycles. Since there is an inverse linear relationship between the logarithm of the initial bacterial DNA load and the corresponding C_T value, the change in the C_T value (ΔC_T) from before and after chemomechanical preparation of the root canal gives a first estimate of the bacterial reduction and, thus, of treatment efficacy. The mean ΔC_T determined by the SYBR green format in the NaOCl treatment group was 9.70, whereas it was 8.91 by the TaqMan assay (Table 3). In contrast, the mean ΔC_T determined by the SYBR green format in the CHX gel treatment group was only 3.45, whereas it was 4.61 by TaqMan analysis. These values could indicate a better bacterial clearance in the NaOCl group. It is important that determination of the precise cell number of a multispecies bacterial population is complicated by the wide range of rRNA operon numbers among different bacterial taxa (range, 1 to ≥10) (4, 17). The numbers (bacterial loads) calculated here are therefore referred to as “rRNA gene copy numbers,” since the ratio between rRNA genes and cells is unknown. The individual bacterial load differed considerably among samples, ranging from 3.2 × 10³ to 1.2 × 10⁸ rRNA gene copy numbers before treatment and from “negative” to 9.6 × 10⁶ rRNA gene copy numbers after endodontic treatment (Table 4). In the CHX gel treatment group, the SYBR green- and the TaqMan-based detection formats led, with a few exceptions (samples C6 and C7), to similar results, which is in accordance with previous findings (14). In the NaOCl treatment group, however, we observed a posttreatment trend toward lower gene copy numbers when we used the SYBR green format. This might largely be due to the SYBR green-specific effect on impairment of PCR efficiency (19), which becomes more relevant with low template concentrations. Irrespective of the individual bacterial load, the antimicrobial reduction within treatment groups was largely consistent in both the SYBR green and the TaqMan analyses. While the microbial reduction in the NaOCl treatment group was in most cases greater than 99% (for the SYBR green format, mean of 99.64% and median of 99.99%; for the TaqMan format, mean of 94.23% and median of 99.63%), we observed a much lower microbial reduction in the CHX gel treatment group (for the SYBR green format, mean of 82.91% and median of 96.62%; for the TaqMan format, mean of 86.62% and median of 96.60%). This difference between the two treatment groups was statistically significant by the nonparametric Mann-Whitney test (P < 0.01). We also determined the bacterial load by parallel plate counting. In the initial samples, the numbers of CFU ranged from 4 × 10² to 1 × 10⁶, with a median of 3.2 × 10⁵ CFU. In contrast, the CFU counts in the posttreatment samples declined drastically to a median of 0 (range, from 0 to 6.8 × 10² CFU). While a direct comparison between the cell numbers retrieved by counting the numbers of CFU and the gene copy numbers determined by RTQ-PCR is not possible, the scale of microbial reduction can be compared since it is proportional to the initial values measured. The bacterial reduction determined by parallel cell counting was similar in both treatment groups (for the CHX gel group, mean of 99.6% and median of 99.9%; for the NaOCl group, mean of 99.9% and median of 100%).

TABLE 3.

Mean RTQ-PCR C_T values determined for bacterial samples of 32 patients with periapical lesions before and after chemomechanical treatment with either NaOCl or CHX gel as the irrigating substance^a

Detection format	CT value (standard deviation)
	NaOCl group (n = 16)			CHX gel group (n = 16)
	Before	After	ΔCT	Before	After	ΔCT
SYBR green	21.38 (2.77)	31.08 (2.59)	9.70	21.92 (2.47)	25.37 (1.72)	3.45
TaqMan	25.76 (4.70)	34.67 (1.86)	8.91	26.99 (3.01)	31.60 (2.63)	4.61

^aData acquisition was based on a threshold value of 0.2, which was approximately 10 times the background fluorescence, defined as the mean fluorescence values for the first 6 to 15 PCR cycles. All samples were run in duplicate. The variation between duplicates was less than 3%.

TABLE 4.

Bacterial load and percent reduction determined for root canal samples of 32 patients with periapical lesions before and after chemomechanical treatment with either NaOCl or CHX gel as the irrigating substance^a

Sample	NaOCl group						CHX gel group
	SybrGreen			TaqMan			Sample	SYBR green			TaqMan
	Bacterial load (rRNA gene copy no.)		% Reduction	Bacterial load (rRNA gene copy no.)		% Reduction		Bacterial load (rRNA gene copy no.)	% Reduction	Bacterial load (rRNA gene copy no.)		% Reduction
	Before	After		Before	After			Before		After	Before		After
H1	19 × 106	2 × 102	99.99	32 × 106	64 × 104	98.00	C1	69 × 103	63 × 103	8.69	46 × 103	41 × 103	10.86
H2	32 × 102	1 × 102	96.87	43 × 103	12 × 103	72.09	C2	16 × 104	78 × 102	95.12	36 × 104	11 × 103	96.94
H3	19 × 104	Negative	100	37 × 104	25 × 103	93.24	C3	47 × 104	38 × 104	19.14	89 × 104	48 × 104	46.06
H4	15 × 106	2 × 102	99.99	11 × 106	76 × 102	99.93	C4	20 × 104	70 × 103	65.00	24 × 104	22 × 103	90.83
H5	30 × 105	2 × 102	99.99	43 × 105	42 × 102	99.90	C5	13 × 106	53 × 103	99.59	84 × 105	28 × 103	99.66
H6	20 × 105	14 × 103	99.30	31 × 105	16 × 103	99.48	C6	16 × 105	28 × 103	98.25	24 × 105	34 × 104	85.83
H7	22 × 106	22 × 102	99.99	11 × 106	12 × 103	99.89	C7	30 × 105	43 × 103	98.56	37 × 105	11 × 102	99.97
H8	53 × 106	1 × 102	99.99	36 × 106	17 × 103	99.95	C8	63 × 104	15 × 103	97.61	10 × 105	10 × 103	99.00
H9	71 × 106	21 × 103	99.97	59 × 106	36 × 103	99.93	C9	43 × 106	47 × 103	99.89	67 × 106	51 × 103	99.92
H10	12 × 107	16 × 103	99.98	87 × 106	31 × 104	99.64	C10	18 × 104	61 × 103	66.11	42 × 104	44 × 103	89.52
H11	33 × 103	3 × 102	99.09	33 × 103	17 × 103	48.48	C11	14 × 106	56 × 104	96.00	21 × 106	12 × 105	94.28
H12	58 × 104	2 × 102	99.96	10 × 105	17 × 103	98.30	C12	12 × 105	66 × 103	94.50	18 × 105	54 × 103	97.00
H13	19 × 105	1 × 102	99.99	24 × 105	12 × 103	99.30	C13	47 × 105	13 × 104	97.23	91 × 105	34 × 104	96.26
H14	25 × 105	3 × 102	99.98	21 × 105	81 × 102	99.61	C14	30 × 105	60 × 103	98.00	44 × 105	11 × 104	97.50
H15	19 × 106	1 × 102	99.99	17 × 106	10 × 103	99.94	C15	17 × 106	15 × 104	99.11	18 × 106	44 × 104	97.55
H16	22 × 105	18 × 103	99.18	35 × 106	22 × 103	99.93	C16	67 × 106	42 × 105	93.73	63 × 106	96 × 105	84.76
Mean	21 × 106	45 × 102	99.64	19 × 106	73 × 103	94.23	Mean	10 × 106	37 × 104	82.91	12 × 106	80 × 104	86.62
Median	28 × 105	2 × 102	99.99	76 × 105	16 × 103	99.63	Median	23 × 105	62 × 103	96.62	30 × 105	53 × 103	96.60

^aDNA from Prevotella nigrescens (ATCC 33563) was used to establish the standard curve for calculating the gene copy numbers. The linear scope of detection ranged from 10² to 10⁸.

Thus, the difference between the treatment groups was much more pronounced when the reduction was measured by RTQ-PCR. This assay detects not only noncultivable species but also, to a certain extent, dead cell debris, a risk factor for a successful clinical outcome (16). The most broad-range RTQ-PCR might therefore be a valuable, complementary tool for the monitoring of anti-infective therapies.

In conclusion, assessment of the total bacterial load in a sample by universal PCR will certainly have an increasing impact on future microbiology, and important formats will be RTQ-PCR and PCR-based microarrays for diagnostic purposes (30). We have shown that the universal PCR assays published previously might potentially detect only a small to medium proportion of the bacterial 16S rRNA gene sequences included in ARB. Therefore, every user of a PCR protocol should first ensure its relevance for its intended application by retesting the probes and primers for covering the most important or dominant species in a particular sample. Even then, the results should be interpreted carefully, since the problem of finding a true universal PCR assay that reliably and invariably detects all bacterial species present in complex samples remains unresolved.