Identification of early microbial colonizers in human dental biofilm


E.J. Helmerhorst, 700 Albany Street, W201, Boston, MA 02118, USA (e-mail:


Aims:  To elucidate the first colonizers within in vivo dental biofilm and to establish potential population shifts that occur during the early phases of biofilm formation.

Methods and Results:  A ‘checkerboard’ DNA–DNA hybridization assay was employed to identify 40 different bacterial strains. Dental biofilm samples were collected from 15 healthy subjects, 0, 2, 4 and 6 h after tooth cleaning and the composition of these samples was compared with that of whole saliva collected from the same individuals. The bacterial distribution in biofilm samples was distinct from that in saliva, confirming the selectivity of the adhesion process. In the very early stages, the predominant tooth colonizers were found to be Actinomyces species. The relative proportion of streptococci, in particular Streptococcus mitis and S. oralis, increased at the expense of Actinomyces species between 2 and 6 h while the absolute level of Actinomyces remained unaltered. Periodontal pathogens such as Tannerella forsythensis(Bacteroides forsythus), Porphyromonas gingivalis and Treponema denticola as well as Actinobacillus actinomycetemcomitans were present in extremely low levels at all the examined time intervals in this healthy group of subjects.

Conclusion:  The data provide a detailed insight into the bacterial population shifts occurring within the first few hours of biofilm formation and show that the early colonizers of the tooth surface predominantly consist of beneficial micro-organisms.

Significance and Impact of the Study:  The early colonizers of dental plaque are of great importance in the succession stages of biofilm formation and its overall effect on the oral health of the host.


Most bacteria colonizing higher organisms grow into populations by adhering to solid surfaces and ultimately forming mixed culture biofilms. Dental plaque is such a complex microbial community growing as a biofilm on enamel surfaces. The aetiology of both dental caries and various forms of periodontal disease has long been recognized to be related to bacterial accumulations and plaque composition. Despite extensive analysis of plaque samples from healthy and diseased subjects as well as data derived from gnotobiotic and germ-free animal experiments, no single microbe has been identified which satisfies Koch's postulates for an infectious agent in either caries or periodontitis (Marsh and Martin 1999). Most recent evidence suggests that both diseases have a multi-bacterial aetiology and therefore it is important to gain insight into the total bacterial composition of dental plaque.

To identify individual species in dental plaque, samples have been harvested from tooth surfaces and cultured on a variety of (semi)-selective agars. This has yielded considerable insights into the colonizing micro-organisms that can be identified by means of cultivation (Socransky et al. 1977; Nyvad and Kilian 1987). The ‘checkerboard’ DNA–DNA hybridization technique, developed in 1994 by Socransky et al., provides not only information on these species, but also enables the detection of unusual micro-organisms that are present in low numbers or those that are difficult to enumerate by culture. So far, the DNA–DNA hybridization assay has been used to study ecological relationships between bacterial species from plaque samples formed either supra-gingivally or subgingivally (Socransky et al. 1998; Ximenez-Fyvie et al. 2000a,b; Ramberg et al. 2003) and have focused on the bacterial distribution in mature biofilm. Very little is known about the complexity of microbial colonization of tooth surfaces in the very early phases of biofilm formation especially with regard to uncommon species. The initial process of adhesion comprises the specific adsorption of certain microbial species present in the oral cavity to the proteinaceous film covering tooth surfaces which is also known as acquired enamel pellicle (Yao et al. 2003). Early colonizers are of great importance, because after adhering to the tooth surface, they provide attachment substrates for the subsequent colonizers and ultimately influence the succeeding stages of biofilm formation. Thus, the identification of early colonizers will provide access to studying the molecular interactions between bacterial surfaces and constituents of pellicle. The purpose of the present investigation was to use the checkerboard DNA–DNA hybridization technique to identify the very early colonizers attaching within the first hours concurrent with pellicle formation using DNA probes prepared from the 40 most prevalent dental plaque species.

Materials and methods

Subject population

Human biofilm samples and whole saliva specimens were collected from 15 subjects (seven males, eight females; mean age 31·5 ± 5·5 years). All subjects presented with good oral hygiene, firm gingiva and no overt signs of gingivitis or dental caries. The exclusion criteria were: current smokers, pregnancy, lactation, periodontal or antibiotic therapy in the previous 3 months, and systemic conditions which would require premedication. The protocol for this investigation was approved by the Institutional Review Board of Boston University Medical Center and informed consent was obtained from each subject.

Clinical monitoring

Clinical evaluation included the presence/absence of gingival redness, bleeding on probing, suppuration and determination of pocket depth and attachment loss by a periodontist. These measurements were carried out on the mesial-buccal, middle-buccal and distal-buccal parts of teeth in the upper and lower arches comprising first molars, premolars and anterior teeth.

Biological sample collection and processing

Dental biofilm.  All biofilm samples were collected by one investigator from 15 subjects who had at least 20 restoration-free and caries-free buccal/labial surfaces excluding the second and third molars. After pumicing tooth surfaces, the teeth were isolated from buccal/labial mucosa with cotton rolls during the entire sampling procedure to avoid contact between tooth surfaces and oral mucosa. The collection area was rinsed twice with water and dried with air. Biofilm was collected at 0, 2, 4 or 6 h after pumicing, with no sampling in between. Subjects were asked to refrain from eating, drinking (except water) or brushing during the biofilm formation phase. Biofilms formed for 2 or 4 h were collected on one day and biofilm formed for 0 or 6 h were collected on the following day. To collect biofilm, polyvinylidene fluoride (PVDF) membranes (45 μm pore size, 13 mm diameter; Durapore; Millipore, Bedford, MA, USA) soaked in 0·5 mol l−1 sodium bicarbonate, pH 8·4, were held with cotton pliers and the coronal two-thirds of buccal and labial dental surfaces were swabbed applying mild pressure using a single PVDF membrane per quadrant. After collection, the PVDF membranes containing the bacteria were immediately placed in 300 μl of Tris–EDTA buffer (10 mmol l−1 Tris, 1 mmol l−1 EDTA, pH 7·6) followed by vortexing for 30 s and sonication for 5 min. Subsequently, 200 μl of freshly prepared NaOH (0·5 mol l−1) solution was added to lyse the bacteria. Samples treated in this manner can be stored at room temperature for up to 2 months allowing for batchwise processing (unpublished observations).

Whole saliva.  Immediately before biofilm collection, 1 ml of resting whole saliva was obtained from each subject by having the donor expectorate into a prechilled tube. In contrast to biofilms, saliva was only collected at one time point (t = 0), because it can be assumed that the bacterial composition of saliva does not change over a 6-h time interval. Immediately after saliva collection, 200 μl aliquots were mixed with 150 μl of Tris–EDTA buffer and vortexed for 10 s. From this mixture, 200 μl were combined with 100 μl of freshly prepared NaOH (0·5 mol l−1) and samples were stored at room temperature.

Checkerboard DNA–DNA hybridization

A checkerboard DNA–DNA hybridization technique was conducted as described previously (Socransky et al. 1994). The processed dental biofilm or saliva samples were boiled to generate single-stranded DNA and samples were placed into the extended slots of a Minislot 30 apparatus (MiniSlot; Immunetics, Cambridge, MA, USA), concentrated onto a 15 × 15 cm positively charged nylon membrane (Boehringer Mannheim, Indianapolis, IN, USA) and fixed to the membrane by cross-linking under ultraviolet light. After fixation of the DNA to the membrane, the membrane was placed in a Miniblotter 45 (Immunetics) with the lanes containing DNA at an angle of 90° to the lanes of the device. Digoxigenin-labelled whole genomic DNA probes to 40 oral bacterial strains were hybridized in individual lanes of the Miniblotter. The preparation of DNA probes and their source strains were described in Ximenez-Fyvie et al. (2000a). After hybridization, the membranes were washed at high stringency and the DNA probes detected using antibody to digoxigenin conjugated with alkaline phosphatase (Boehringer Mannheim). Signals were detected using AttoPhos substrate (Amersham Life Science, Arlington Heights, IL, USA) and a Storm Fluorimager (Molecular Dynamics, Sunnyvale, CA, USA). Two positive control lanes on each membrane contained standards that consisted of a mixture of 105 and 106 cells of each bacterial strain tested. The sensitivity of this assay was adjusted to permit detection of 104 cells of a given strain by adjusting the concentration of each DNA probe. This procedure was carried out in order to provide the same sensitivity of detection for each strain. Signals were quantified by comparing signal intensities of samples with those of standards on the same membrane and were converted to bacterial counts of each strain. The specificity of the probes has been demonstrated in a previous study (Socransky et al. 1998). Cross-reaction among probes are uncommon and limited to low-level cross-hybridization of species within the same genus.

Statistical analysis

Significance of differences between the average values of the bacterial counts or the percentage of DNA in the total DNA counts for each strain were assessed in saliva and in pellicle samples collected at different time points using one-way anova.


The 15 selected subjects from whom pellicle was collected had at least 20 caries-free and restoration-free buccal/labial surfaces, no pocket depth >4 mm and no periodontal attachment loss. The average percentage of sites that exhibited gingival redness, bleeding on probing and suppuration were 6 ± 7, 15 ± 12 and 0%, respectively. The clinical parameters of these subjects resembled those observed in orally healthy populations (Haffajee et al. 1983; Armitage 1996; Ximenez-Fyvie et al. 2000b).

In a study to test the reproducibility of samples collected at different days, biofilm samples formed for 4 h were collected at three sequential days from the same three subjects. Results showed a remarkable consistency within subjects in the bacterial numbers and distribution patterns between days (data not shown). These results suggested that day-to-day variation was minimal and justified the collection of biofilm samples formed after 2 and 4 h and those formed at time 0 and after 6 h on two sequential days.

A typical ‘checkerboard’ bacterial analysis is shown in Fig. 1. Samples of whole saliva and of biofilm specimens collected at four different time points from three healthy subjects reflect results obtained with 40 bacterial strains. All strains were detectable to different degrees at different time points, indicating that in the early phases of pellicle formation, significant numbers of micro-organisms attach to tooth surfaces. The data also showed that the ‘checkerboard’ DNA–DNA hybridization technique was a suitable approach to characterize and enumerate early colonizing bacteria. Furthermore, the data indicate that the bacterial populations found in saliva are not representative of those recovered from teeth in terms of their species distribution. For example, in all subjects, levels of Streptococcus mitis and S. oralis were low in saliva, but were present in higher levels in dental biofilm, in particular in the 6-h pellicle samples. In contrast, other bacterial strains such as Actinomyces naeslundii genospecies 1, Fusobacterium nucleatum subsp. vincentii, Micromonas micros, Veillonella parvula, Act. israelii, Campylobacter showae, Tannerella forsythensis (Bacteroides forsythus) and Prevotella melaninogenica that were found to be present in significantly high numbers in saliva, were almost absent in the biofilm samples.

Figure 1.

Analysis of the bacterial composition of saliva and dental biofilm samples using the checkerboard DNA–DNA hybridization assay. Saliva samples and dental biofilm samples collected 0, 2, 4 or 6 h after prophylaxis treatment of the teeth were subjected to checkerboard DNA–DNA hybridization analysis. Samples were analysed for the presence of 40 prevalent oral bacterial strains (one lane per strain), using specific DNA probes. The image shows the reactions of 40 DNA probes to samples collected from three different healthy subjects. Mixed standards containing 105 and 106 cells are shown in the lower two lanes. (Act. naeslundii 1: Actinomyces naeslundii genospecies 1; A. actinomyc.: Actinobacillus actinomycetemcomitans; Fus. nuc. vincentii: Fusobacterium nucleatum subsp. vincentii; Act. naeslundii 2: Act. naeslundii genospecies 2; Act. odontolyticus 1: Act. odontolyticus serotype 1; Fus. nuc. polymorph.: Fus. nucleatum subsp. polymorphum; Fus. nuc. nuc.: Fus. nucleatum subsp. nucleatum)

Counts for each bacterial strain in different samples were deduced from the signal intensities of the sample compared with those of two control samples containing a known number of cells. These data provided the cell numbers of bacteria of each strain as well as the proportions of each with respect to the 40 strains assayed per specimen (Fig. 2a,b). The data shown represent the mean values obtained from 15 subjects. The 40 bacterial strains examined contained representative species of seven microbial ‘complexes’ defined by cluster analysis and community ordination techniques in 13 261 plaque samples (Socransky et al. 1998). In saliva, most examined bacterial species were detectable with a considerable variation in total counts and relative counts between the examined strains (Fig. 2a,b, first column). In biofilms collected immediately after prophylaxis treatment, the total counts of all bacterial species were extremely low compared with specimens collected after biofilms were allowed to form for 2, 4 or 6 h (Fig. 2a, second, third, fourth and fifth column). The distribution of the bacteria in biofilm collected at time 0 (Fig. 2b, second column) is very similar to that of saliva (Fig. 2b, first column) but the pattern clearly becomes distinct from that in saliva upon longer incubation times, pointing towards a selective adhesion process (Fig. 2b, third, fourth and fifth column). With respect to the seven bacterial groups illustrated in Fig. 2, the periodontopathic species of the red complex and most species of the orange complex were present in extremely low levels at all intervals examined. However, species of the yellow complex increased in total numbers as well as relative numbers with the progression of biofilm formation. Actinomyces and species of the purple and green complex were present in the early stages, and their numbers remained relatively stable over the 6-h time interval investigated. However, because more and more streptococci attached over time, this caused a reduction in the proportion of Actinomyces over time (Fig. 2a,b, third, fourth and fifth columns). For Leptotrichia buccalis, Neisseria mucosa and P. melaninogenica, no selective adhesion occurred because the levels were high both in saliva and in dental biofilm.

Figure 2.

Bacterial distribution in saliva and biofilm samples expressed in counts (a) or in proportions (b). The 40 bacterial strains examined are grouped into seven microbial ‘complexes’ (Socransky et al. 1998). Total counts per sample (×105) or relative counts (%) of each bacterial strain were averaged over the 15 subjects examined

As expected, the sum of bacterial counts of the 40 examined strains increased over time and were 4·6 × 105, 15·1 × 105, 18·5 × 105 and 25·0 × 105 after 0, 2, 4 and 6 h of biofilm formation, respectively. The differences in these total counts were statistically significant (P = 0·002). The major bacterial strains that contained more than 105 bacteria per sample or that comprised more than 5% of the total bacterial counts were listed separately (Tables 1 and 2). The data describing the predominant strains further support the results shown in Fig. 2 that the bacterial distribution on teeth was not identical to the bacterial distribution in saliva. For example, none of the bacterial strains which were predominant in dental biofilm collected after 6 h exhibited a proportion higher than 5% in saliva and vice versa.

Table 1.  Oral bacterial species in early biofilm samples with average counts of more than 1 × 105 bacteria per sample at one of the biofilm formation time points
Bacterial speciesBiofilm formation time
0 h2 h4 h6 h
  1. Data represent the mean counts (×105) of each bacterial species in biofilm samples collected from 15 subjects.

  2. <, Mean counts <1 × 105 bacteria per sample.

Act. naeslundii 2<1·791·441·47
S. gordonii<<<1·07
S. mitis<1·373·544·76
S. oralis<<1·592·78
E. corrodens<1·19<1·10
N. mucosa<<<1·31
Table 2.  Oral bacterial species (%) in saliva and early biofilm samples with average proportions of more than 5% per sample at one of the biofilm formation time points
Bacterial speciesSalivaBiofilm formation time
0 h2 h4 h6 h
  1. *Significance of differences between saliva and biofilm samples was assessed by one-way anova.

  2. The proportion of each bacterial species was calculated based on the proportion of their DNA in the total DNA pool per sample. Data represent the mean of samples collected from 15 subjects.

  3. <, Mean percentage <5% of the total DNA probe count.

Act. naeslundii 1 (P < 0·001)*8·5710·207·345·31<
Act. naeslundii 2 (P < 0·01)<<11·877·245·97
V. parvula (P < 0·01)6·879·306·00<<
S. mitis (P < 0·001)<<6·2314·7218·53
S. oralis (P < 0·001)<<<7·7610·48
E. corrodens (P < 0·05)8·135·445·27<<
Camp. gracilis6·685·58<<<
Camp. showae (P < 0·001)5·18<<<<
Fus. nucleatum subsp. vincentii (P < 0·001)<6·15<<<
L. buccalis<6·70<<<
N. mucosa7·70<<5·27<
P. melaninogenica (P < 0·001)10·8812·746·10<<


The aetiology of dental caries and various forms of periodontal disease have long been recognized to be correlated with bacterial plaque accumulation as well as plaque composition. In view of the important role of bacteria in these oral diseases, we investigated the early events on the clean tooth surfaces with particular emphasis on the early bacterial colonizers. The importance of early colonizers lies in the fact that they provide surfaces for attachment, or modify the environment, for subsequent colonizers. Thus, they play a key role in the microbial succession occurring in dental biofilms that will ultimately determine oral health or disease.

The current study showed a temporal change of the bacterial adhesion pattern in the initial stages of dental biofilm formation. In samples collected immediately after pumicing, only low numbers of the tested bacterial species could be detected and the bacterial composition was essentially identical to that in saliva. This could be explained by brief exposure of teeth after pumicing to saliva leading to the immediate deposition of salivary micro-organisms onto tooth surfaces. These very early identified species therefore likely represent loosely bound micro-organisms rather than firmly attached bacteria. Within a few hours of tooth exposure to the oral environment, the bacterial pattern clearly became distinct from that in saliva, indicating that only specific organisms could achieve firm attachment to the tooth surface, showing the selectivity of the colonization process.

To achieve successful adherence, bacteria have developed strategies to evade the clearance forces that operate in the oral cavity. Such clearance forces include mechanical flushing of saliva as well as forces related to physiological movements such as swallowing, chewing and speaking. In addition, salivary components such as secretory IgA (sIgA), inhibit bacterial cell adhesion to tooth surfaces by promoting bacterial agglutination in saliva (Williams and Gibbons 1972). It has been found that many of the early colonizers of the human oral cavity produce IgA1 protease which destroys sIgA (Genco et al. 1975) which may provide these species with a specific advantage over other species in colonizing the tooth surface. Successful colonizers not only avoid clearance, but also exhibit mechanisms to achieve firm attachment. Such attachment involves specific interactions between bacterial surface adhesins and receptors on the tooth surface. Dental surfaces exposed to the oral environment are almost instantaneously covered by a proteinaceous film, designated the acquired enamel pellicle. This protein film is formed by a selective adsorption process comprising salivary proteins, peptides and other organic molecules. In vitro findings have shown that bacteria adhering to saliva-coated and uncoated hydroxyapatite are different (Clark et al. 1978) suggesting that pellicle somehow determines at least the initial stages of bacterial attachment to the tooth surface. Diverse bacterial adhesins have been identified which can recognize different salivary receptors present on tooth surfaces. Examples are molecules belonging to the Ag I/II family (Jenkinson and Demuth 1997), amylase-binding proteins (Scannapieco 1994) and surface lectins (Murray et al. 1986) that are found on S. mitis and S. oralis. The existence of multiple adhesins not only enables an individual cell to bind more avidly to one salivary receptor but also allows it to bind to different salivary receptors on the tooth surface, creating a firm attachment. Subsequently, bound bacteria adhering to the pellicle-coated tooth surfaces provide new binding sites through cell–cell recognition or by releasing extracellular molecules that promote the attachment of other micro-organisms (Kolenbrander and London 1993).

Our data demonstrate that actinomyces are prominent colonizers in biofilm formed within 6 h on dental enamel surfaces. One explanation for the abundance of this species is that both Act. naeslundii genospecies 1 and 2 strains synthesize neuraminidases which remove terminal sialic acid residues in mucous glycoprotein 1 (MG1), which is one of the major salivary glycoproteins in saliva and in in vivo pellicle (Li et al. 2003). We have reported that sialic acid residues are indeed absent from pellicle associated MG1 (Li et al. 2003). As actinomyces are known to have galactosyl-binding adhesins (Ellen et al. 1980; Cisar 1986), the increased exposure of internal galactosyl residues as a result of the neuraminidase activity on MG1 would be expected to promote the interaction of actinomyces with this salivary molecule in the early stages of biofilm formation.

Consistent with previous observations (Theilade et al. 1966; Eastcott and Stallard 1973; Socransky et al. 1977; Nyvad and Fejerskov 1987; Nyvad and Kilian 1987), we found that actinomyces and streptococci are predominant in biofilm collected within the first hours of its formation. Although our data show that streptococci, in particular S. mitis and S. oralis, become more predominant than actinomyces within the first 6 h of biofilm formation, the reason as to why the abundance of the Actinomyces levelled off after the first 2 h while streptococci steadily increased for the first 6 h is not entirely clear. This characteristic may be related to different adhesion kinetics and/or differences in doubling times between these two species (Socransky et al. 1977; Beckers and van der Hoeven 1982, 1984; Beighton et al. 1986).

Of the 40 strains examined in this study, Porphyromonas gingivalis, Tannerella forsythensis and Treponema denticola belong to the ‘red complex’. Species in this complex have been found in high numbers in lesions of adult periodontitis (Umeda et al. 1996) and exhibit a strong relationship with inflammatory or other indicators of periodontitis such as pocket depth and bleeding on probing (Socransky et al. 1998). Bacteria belonging to the ‘orange complex’ have also been related to pocket depth (Socransky et al. 1998) but to a lesser degree than bacteria from the ‘red complex’. We demonstrate that bacteria belonging to the red and orange complexes as well as A. actinomycetemcomitans can be identified on tooth surfaces within the first 6 h, but at extremely low levels. However, species thought to be host compatible or even beneficial for oral health were relatively abundant in dental biofilm. It is of interest that actinomyces spp. have been suggested to neutralize the local pH by conversion of lactic acid into weaker acids, thereby lowering the risk of caries formation (van der Hoeven and van den Kieboom 1990; Takahashi and Yamada 1999). The absence of periodontopathic strains in early biofilm can be explained by the fact that the levels of oxygen in biofilms need to be reduced before these strains, for which oxygen is toxic, can increase in number to detectable limits. In later stages, this is usually brought about by the facultative strains scavenging the available oxygen. At the same time, some predominant members of this group, such as S. mitis, S. oralis and S. sanguinis, are known to produce hydrogen peroxide which may help to prevent the periodontopathic bacteria achieving levels at which they are capable of initiating disease (Vernazza and Melville 1979; Willcox and Drucker 1988). Thus, antagonistic relationships exist between host compatible and pathogenic bacteria in the initial microbiota to create an environment that favours the colonization of one group of organisms over the other. This could explain the low abundance of periodontopathic bacteria in the initial microbiota in dental biofilms collected from the healthy subjects in this study.

As pointed out earlier, the molecular basis for the initial stages of plaque formation is dictated by the nature of components comprising acquired enamel pellicle. It would appear that this selective pellicle formation process favours an early colonization by bacteria predominantly beneficial for the host with very low levels of bacteria promoting diseases. We are currently using proteomic approaches to identify the relationship between the composition of pellicle and bacterial profiles on a molecular level.


This study was supported by NIH/NIDCR grants DE 05672, DE 07652, DE 14950, DE 11691 and DE 14368.