Since at least 30% of the oral cavity bacteria are uncultivable (Dewhirst et al., 2010), methods other than culture-based identification must be used to investigate complex bacterial samples such as oral samples. Culture-independent methods using DNA-probes (Tsai et al., 2003) and checkerboard DNA-DNA hybridization method (Socransky et al., 2004) as well as novel sequencing methods (Pace et al., 1985) have been developed to detect oral bacteria.
PCR-based methods have been used extensively for qualitative analysis of the presence of bacteria in oral bacterial samples over the years. Meurman and co-workers showed that by using PCR methods, it was possible to detect T. forsythia in 52 of 58 patients whereas when using cultures, only 22 of 58 samples were assessed positive (Meurman et al., 1997). Basic PCR methods do have limitations: they assess bacterial presence more or less qualitatively. A real-time (RT-) PCR is needed when assessing quantitative information. In one comparison study between the data obtained from culturing and from RT-PCR, it was shown that the prevalence of various bacterial species was lower in culture-based data than that demonstrated with the use of molecular methods (van Winkelhoff et al., 2005). Nowadays, “Next generation sequencing” technology including PCR amplification, cloning and sequencing of 16S rRNA segments is in wide use (Paster et al., 2001).
2.5.1 Next generation sequencing
A culture-independent framework for sequencing 16S rRNA gene sequences was introduced by Pace and colleagues in 1985 (Pace et al., 1985), which has allowed characterization of the microbial diversity from human and environmental microbiomes with an unprecedented depth and coverage (Turnbaugh et al., 2007;
Nelson et al., 2010). Along with modern sequencing technology, the number of
intracranial aneurysm rupture (Inenaga et al., 2018). Cagli et al. failed to detect Chlamydia pneumoniae from IA sac tissue samples (Cagli et al., 2003). In a recent French study, bacterial DNA was not detected from intracranial aneurysm samples (Aboukais et al., 2019). In an animal study (mice), bacterial DNA was assessed, but not detected, in the cerebral arteries using universal bacterial DNA primers.
Interestingly, depletion of the gut microbiota of the mice significantly reduced the incidence of aneurysms. (Fumiaki et al., 2019)
2.4 LOW-GRADE SYSTEMIC INFLAMMATION IN
PERIODONTITIS AND CARDIOVASCULAR DISEASES
In addition to direct bacterial invasion, another mechanism behind systemic complications caused, or worsened, by periodontitis is low-grade systemic inflammation. Systemic low-grade inflammation can be induced by the binding of circulating PAMPs to TLRs, which are expressed on immune cells, adipocytes, and endothelial cells (de Punder and Pruimboom 2015).
Systemic low-grade inflammation is a known risk factor of coronary heart disease (Danesh et al., 2000). It has been shown that periodontitis can cause low-grade systemic inflammation: A German 5- and 11-year follow-up study with 2622 subjects showed periodontitis affecting systemic inflammation (measured as white blood cell counts and fibrinogen) in a significant dose-dependent manner (Gocke et al., 2014).
Periodontitis has been shown to be associated with increased platelet activation in a severity dependent manner, which may partly explain the epidemiological association between periodontitis and cardiovascular disease (Papapanagiotou et al., 2009). In a classic human study by Nibali and co-workers, periodontitis subjects exhibited a low-grade systemic inflammation (increased white cell counts), dyslipidemia (lower HDL cholesterol, and higher LDL cholesterol) and increased non-fasting serum glucose levels when compared with controls (after correcting for differences in smoking, gender, age and ethnicity). A dose-dependent trend effect of the amount of periodontal pockets on the studied inflammatory and metabolic markers was observed (Nibali et al., 2007). A study by Wu et al. found a significant relation between indicators of poor periodontal status and increased C-reactive protein and fibrinogen. The association between periodontal status and total cholesterol level was shown to be much weaker. No association between periodontal status and HDL cholesterol was detectable in the study. (Wu et al., 2000) Anti-infective periodontal treatment results in short-term modest reductions in systemic CRP (Demmer et al., 2013; Mysak et al., 2017).
It is well known that low-grade systemic inflammation generates unfavourable alteration in serum lipid levels, such as total cholesterol and low-density lipoprotein (LDL) levels (Hotamisligil, 2017). On the other hand, opposite effect appears to exist as well: low-grade systemic inflammation (measured as white blood cell counts and fibrinogen) at baseline predicted destruction of periodontal tissues and development of clinical periodontitis at follow-up in one prospective study (Josey and Merchant,
2016). One animal study showed that oral administration of P. gingivalis induced a change in gut microbiota composition and in metabolic pathways related to amino acid metabolism and in tyrosine, phenylalanine and tryptophan biosynthesis. In addition to altering immune modulation and gut barrier function (intestinal epithelium works as a selectively permeable barrier permitting the absorption of nutrients, electrolytes and water, while maintaining an effective defence against intraluminal toxins, antigens and enteric flora), oral administration of P. gingivalis affects the host’s metabolic profile. (Groschwitz and Hogan, 2009; Kato et al., 2018) Dental infectious burden during the lifetime appears to associate with systemic low-grade inflammation: An association of the number of remaining teeth with the level of immunoglobulin (Ig) G against periodontal bacteria and with CRP levels has been demonstrated (Aoyama et al., 2018). In one Finnish study having teeth missing was associated with increased hazard for incident coronary heart disease events, acute myocardial infarction, incident cardiovascular disease, diabetes and death of any cause (Liljestrand et al., 2015).
2.5 MOLECULAR ANALYSIS OF BACTERIA
Since at least 30% of the oral cavity bacteria are uncultivable (Dewhirst et al., 2010), methods other than culture-based identification must be used to investigate complex bacterial samples such as oral samples. Culture-independent methods using DNA-probes (Tsai et al., 2003) and checkerboard DNA-DNA hybridization method (Socransky et al., 2004) as well as novel sequencing methods (Pace et al., 1985) have been developed to detect oral bacteria.
PCR-based methods have been used extensively for qualitative analysis of the presence of bacteria in oral bacterial samples over the years. Meurman and co-workers showed that by using PCR methods, it was possible to detect T. forsythia in 52 of 58 patients whereas when using cultures, only 22 of 58 samples were assessed positive (Meurman et al., 1997). Basic PCR methods do have limitations: they assess bacterial presence more or less qualitatively. A real-time (RT-) PCR is needed when assessing quantitative information. In one comparison study between the data obtained from culturing and from RT-PCR, it was shown that the prevalence of various bacterial species was lower in culture-based data than that demonstrated with the use of molecular methods (van Winkelhoff et al., 2005). Nowadays, “Next generation sequencing” technology including PCR amplification, cloning and sequencing of 16S rRNA segments is in wide use (Paster et al., 2001).
2.5.1 Next generation sequencing
A culture-independent framework for sequencing 16S rRNA gene sequences was introduced by Pace and colleagues in 1985 (Pace et al., 1985), which has allowed characterization of the microbial diversity from human and environmental microbiomes with an unprecedented depth and coverage (Turnbaugh et al., 2007;
Nelson et al., 2010). Along with modern sequencing technology, the number of
reference sequences has increased dramatically. Nowadays there are over 2 million rRNA reference gene sequences stored in public repositories such as Silva (Quast et al., 2013).
Processing the samples
After collecting the samples for analyses under sterile conditions, DNA extraction is a critical step towards the success of microbiome sequencing (Willner et al., 2012;
Abusleme et al., 2014). Since DNA extraction is shown to distort microbial profiles in simulated and clinical oral samples, the careful selection of a DNA extraction protocol is important to improve species recovery and make data comparison between oral microbiology studies possible(Abusleme et al., 2014).
The weightened unifrac distances between DNA extraction techniques are significantly bigger than between technical replicates (Willner et al., 2012). Cell lysis is an initial and critical step in DNA extraction procedures, which normally includes enzymatic, chemical and physical, disruption (Yuan et al., 2012; Bag et al., 2016) Protocols that include bead beating and/or mutanolysin produce significantly better microbial community structure description than methods without these two techniques (Yuan et al., 2012). Based on the evaluations made, it seems that DNA extraction procedures for microbial community analysis of samples of human origin should include bead beating and/or mutanolysin to successfully lyse cells (Yuan et al., 2012).
Variable regions of the 16S gene
Bacterial 16S ribosomal RNA (rRNA) genes contain nine “hypervariable regions” (V1 – V9) that show substantial sequence diversity between different bacteria (Chakravorty et al., 2007). The choice of a hyper-variable region(s) targeted for sequencing is an important decision (Teng et al., 2018).
Microbial diversities assessed from the same sample differ significantly depending on the choice of variable regions, even in the era of further developed sequencing technologies, higher coverage and longer variable regions (Claesson et al., 2010). Therefore, the critical evaluation of the choice of hyper-variable regions is crucial in order to minimize distortion and dissensions in sequence-based analysis and comparison of oral microbiota (Teng et al., 2018).
In 16S rRNA gene-based metagenomics the V1–V3, V3-V4 and V4-V5 regions are in common use (Fouhy et al., 2016). The use of V1-V3 region has hitherto been limited to the Roche/454 pyrosequencing platform, and V3-V4 is commonly used in the MiSeq platform, allowing more accurate and cost-effective characterizations of microbiome samples (Fouhy et al., 2016).
The V4-V5 region is also a traditionally used hypervariable region in 16S rRNA gene-based microbial biodiversity studies (Claesson et al., 2010).
The observed oral microbiota structure is greatly affected by the choice of DNA extraction technique (R2=0.764), whereas the influence of 16S rRNA hypervariable regions is somewhat minor (R2=0.210). Together the DNA extraction method and the choice of hypervariable region collectively explained up to 97.4% of the variation of microbiota structure. Enzymatic-mechanical-lysis based DNA extraction techniques performed best in the characterization of oral microbiota diversity and both the V3-V4 and V3-V4-V5 hypervariable regions seemed to lead to more accurate oral-microbiota structures than the V1-V3 hypervariable regions (Teng et al., 2018). In an older study by Claeson and colleaques, the V4/V5 region demonstrated the best performance across the two sequencing technologies, based on classification efficiencies, simulation accuracies and consistency between two different classification approaches (RDP-Classifier and MEGAN based on BLAST searches) (Claesson et al., 2010). Surprisingly, V3-V4 region performed the poorest, probably due to the primer selection: V3 forward (F338/19) and V4 reverse (R802/18) (Claesson et al., 2010).
The choice of the sequencer is made according to the need: The Illumina HiSeq-sequencer produces >50 Gb per day, and in the course of a 10.8 day run, it produces 1.6 billion 100-base paired-end reads. By contrast, the MiSeq is for smaller single-day experiments, and generates 1.5 Gb per day from 5 million 150-base paired-end reads (Caporaso et al., 2012). Within a few years the efficiency of sequencing technologies has evolved rapidly to produce bigger scale output data: according to Illumina(https://emea.illumina.com/systems/sequencingplatforms.html?langsel=/fi/
), Hiseq platform maximum output for <3days run time is 1.8Tb (6 billion reads) and for Miseq, the maximum output for 4 to 55h run time is 15Gb (25 million reads). This study was conducted using an Illumina MiSeq sequencer. Other sequencer technologies commonly used are 454 pyrosequencing (Tremblay et al., 2015) and Ion Torrent (Lahens et al., 2017).
A comparative study by Allali and colleaques concluded that as long as the data is collected and analyzed coherently during the experiment, the same biological conclusions could be made from the data. It is crucial for researchers to take into account the limitations of each sequencing platform, and to choose a system which is suitable for their experimental design (Allali et al., 2017). All three platforms (MiSeq, Ion Torrent PGM, and Roche 454 GS FLX Titanium) compared in the study were capable of discriminating samples by treatment, despite differences in diversity and abundance, leading to similar biological conclusions (Allali et al., 2017). A limitation of this study is that the hypervariable region of choice was V1-V2, which is hitherto limited to the Roche/454 pyrosequencing platform (Fouhy et al., 2016). However, as described earlier, the results in microbiota structure are greatly affected by the choice of DNA extraction method, whereas the influence of 16S rRNA hypervariable regions is somewhat minor (Teng et al., 2018).
reference sequences has increased dramatically. Nowadays there are over 2 million rRNA reference gene sequences stored in public repositories such as Silva (Quast et al., 2013).
Processing the samples
After collecting the samples for analyses under sterile conditions, DNA extraction is a critical step towards the success of microbiome sequencing (Willner et al., 2012;
Abusleme et al., 2014). Since DNA extraction is shown to distort microbial profiles in simulated and clinical oral samples, the careful selection of a DNA extraction protocol is important to improve species recovery and make data comparison between oral microbiology studies possible(Abusleme et al., 2014).
The weightened unifrac distances between DNA extraction techniques are significantly bigger than between technical replicates (Willner et al., 2012). Cell lysis is an initial and critical step in DNA extraction procedures, which normally includes enzymatic, chemical and physical, disruption (Yuan et al., 2012; Bag et al., 2016) Protocols that include bead beating and/or mutanolysin produce significantly better microbial community structure description than methods without these two techniques (Yuan et al., 2012). Based on the evaluations made, it seems that DNA extraction procedures for microbial community analysis of samples of human origin should include bead beating and/or mutanolysin to successfully lyse cells (Yuan et al., 2012).
Variable regions of the 16S gene
Bacterial 16S ribosomal RNA (rRNA) genes contain nine “hypervariable regions” (V1 – V9) that show substantial sequence diversity between different bacteria (Chakravorty et al., 2007). The choice of a hyper-variable region(s) targeted for sequencing is an important decision (Teng et al., 2018).
Microbial diversities assessed from the same sample differ significantly depending on the choice of variable regions, even in the era of further developed sequencing technologies, higher coverage and longer variable regions (Claesson et al., 2010). Therefore, the critical evaluation of the choice of hyper-variable regions is crucial in order to minimize distortion and dissensions in sequence-based analysis and comparison of oral microbiota (Teng et al., 2018).
In 16S rRNA gene-based metagenomics the V1–V3, V3-V4 and V4-V5 regions are in common use (Fouhy et al., 2016). The use of V1-V3 region has hitherto been limited to the Roche/454 pyrosequencing platform, and V3-V4 is commonly used in the MiSeq platform, allowing more accurate and cost-effective characterizations of microbiome samples (Fouhy et al., 2016).
The V4-V5 region is also a traditionally used hypervariable region in 16S rRNA gene-based microbial biodiversity studies (Claesson et al., 2010).
The observed oral microbiota structure is greatly affected by the choice of DNA extraction technique (R2=0.764), whereas the influence of 16S rRNA hypervariable regions is somewhat minor (R2=0.210). Together the DNA extraction method and the choice of hypervariable region collectively explained up to 97.4% of the variation of microbiota structure. Enzymatic-mechanical-lysis based DNA extraction techniques performed best in the characterization of oral microbiota diversity and both the V3-V4 and V3-V4-V5 hypervariable regions seemed to lead to more accurate oral-microbiota structures than the V1-V3 hypervariable regions (Teng et al., 2018). In an older study by Claeson and colleaques, the V4/V5 region demonstrated the best performance across the two sequencing technologies, based on classification efficiencies, simulation accuracies and consistency between two different classification approaches (RDP-Classifier and MEGAN based on BLAST searches) (Claesson et al., 2010). Surprisingly, V3-V4 region performed the poorest, probably due to the primer selection: V3 forward (F338/19) and V4 reverse (R802/18) (Claesson et al., 2010).
The choice of the sequencer is made according to the need: The Illumina HiSeq-sequencer produces >50 Gb per day, and in the course of a 10.8 day run, it produces 1.6 billion 100-base paired-end reads. By contrast, the MiSeq is for smaller single-day experiments, and generates 1.5 Gb per day from 5 million 150-base paired-end reads (Caporaso et al., 2012). Within a few years the efficiency of sequencing technologies has evolved rapidly to produce bigger scale output data: according to Illumina(https://emea.illumina.com/systems/sequencingplatforms.html?langsel=/fi/
), Hiseq platform maximum output for <3days run time is 1.8Tb (6 billion reads) and for Miseq, the maximum output for 4 to 55h run time is 15Gb (25 million reads). This study was conducted using an Illumina MiSeq sequencer. Other sequencer technologies commonly used are 454 pyrosequencing (Tremblay et al., 2015) and Ion Torrent (Lahens et al., 2017).
A comparative study by Allali and colleaques concluded that as long as the data is collected and analyzed coherently during the experiment, the same biological conclusions could be made from the data. It is crucial for researchers to take into account the limitations of each sequencing platform, and to choose a system which is suitable for their experimental design (Allali et al., 2017). All three platforms (MiSeq, Ion Torrent PGM, and Roche 454 GS FLX Titanium) compared in the study were capable of discriminating samples by treatment, despite differences in diversity and abundance, leading to similar biological conclusions (Allali et al., 2017). A limitation of this study is that the hypervariable region of choice was V1-V2, which is hitherto limited to the Roche/454 pyrosequencing platform (Fouhy et al., 2016). However, as described earlier, the results in microbiota structure are greatly affected by the choice of DNA extraction method, whereas the influence of 16S rRNA hypervariable regions is somewhat minor (Teng et al., 2018).