**2.1 From human microbiome to pathogenic bacterial communities**

At the early 21th century, studies on human-associated bacteria showed that there are at least 10 times as many bacterial cells as human cells, i.e. 1014 bacterial cells in the human gut (Turnbaugh et al., 2007). The current estimation of the number of genes in the human genome is about 23000 (Wei & Brent, 2006). Based on the diversity of gut microbes and the average number of genes contained in one bacterial genome, the diversity of bacterial genes in human gut was guessed to be 100 times greater than that of our human genome (Bäckhed et al., 2005). This number seems to be underestimated, since a more recent publication estimates to more than 9.000.000 the number of unique genes in human gut bacterial community (Yang et al., 2009). This huge community is named the human microbiome, a term coined by Joshua Lederberg in 2001 "to signify the ecological community of commensal, symbiotic, and pathogenic microorganisms that literally share our body space" (Lederberg & McCray, 2001). A new concept is to consider human organism as an assemblage of human and bacterial cells organized into organs, tissues, and cellular

Temporal Temperature Gel Electrophoresis to Survey

microbiome (Nossa et al., 2010).

yet described as members of skin microbiota (Dekio et al., 2005).

2007). They provide a "fingerprint" of the community diversity.

**2.3 16S rRNA gene PCR-TTGE: Advantages and limitations** 

Pathogenic Bacterial Communities: The Case of Surgical Site Infections 293

sample, while avoiding the bias of cultivability. The emergence of molecular ecology and metagenomics offers the potential of determining microbial diversity in an ecosystem without prior laboratory enrichment, isolation and growth on artificial media thanks to sophisticated methodological and computational tools. Generally, culture-independent approaches allow a precise description of an ecosystem by assessing its genetic diversity. As an example of molecular ecology efficiency, Dekio et al (2005) show that the total skin microflora explored by culture-independent molecular profiling is greater than previously believed, finding 22 potentially novel members, comprising 9 species and 13 phylotypes not

Metagenomics examine the complexity of the community by sequencing genomic libraries made from DNA extracted directly from the sample containing a complex mix of different bacteria. The complete metagenomic approach will give the total gene content of a community, thus providing data about biodiversity but also function and interactions (Tyson et al., 2004). For the purpose of biodiversity studies, metagenomics can focus on one common gene shared by all members of the community. The most commonly used cultureindependent method relies on amplification and analysis of the 16S rRNA genes in a

16S rRNA genes are widely used for documentation of the evolutionary history and taxonomic assignment of individual organisms because they have highly conserved regions for construction of universal primers and highly variable regions for identification of individual species (Woese, 1987). The 16S rRNA gene, in spite of some recognized pitfalls (von Wintzingerode et al., 1997), remains today the most popular marker for studying the specific diversity in a bacterial community. The different 16S rRNA genes representative of the community are amplified by PCR and then separated and identified either by cloning and Sanger sequencing or by direct pyro-sequencing (Nossa et al., 2010). Tools for sequencespecific separation after bulk PCR amplification, such as T-RFLP (Terminal-Restriction Fragment Lenght Polymorphism) (Kitts, 2001), D-HPLC (Denaturing High Performance Liquid Chromatography) (Penny et al., 2010), CDCE (Constant Denaturing Capillary Electrophoresis) (Thompson et al., 2004), SSCP (Single Stand Conformation Polymorphism) (Ege et al., 2011), DGGE (Denaturing Gradient Gel Electrophoresis) (Muyzer et al. 1993), TGGE (Temperature Gradient Gel Electrophoresis) (Zoetendal et al., 1998) and TTGE (Temporal Temperature Gradient Gel Electrophoresis) (Ogier et al., 2002), can also be used. Methods based upon separation in denaturing electrophoresis appear particularly suitable for the routine follow-up of microbiomes with low or medium diversity (Roudière et al.,

PCR-TTGE is a PCR-denaturing gradient gel electrophoresis that allows separation of DNA fragments in a temporal gradient of temperature (Yoshino et al., 1991; Ogier et al., 2002). PCR amplicons of the same size but with different sequences are separated in the gel. In a denaturing acrylamide gel, DNA denatures in discrete regions called melting domains, each of them displaying a sequence specific melting temperature. When the melting temperature (*Tm*) of the whole amplicon is reached, the DNA is denatured creating branched molecules. This branching reduces DNA mobility in the gel. Therefore, amplicons of the same size but with different nucleotide compositions can be separated based on differences in the

communities amounting to a super- or a hetero-organism. Metagenomics show clear differences between microbiomes in various body sites and together with metatranscriptomics and metaproteomics reveal how microbiomes contribute to organ and tissue functions. Consequently, human biology can no longer concern itself only with human cells: microbiomes at different body sites and functional metagenomics must be considered part of systems biology (Pflughoeft & Versalovic, 2011). The biological concept of hetero-organism further evolves into new medical conception on health and disease, involving not only human genes and cells but also related genomes from environment around and inside human body. For instance, the change in gut microbiome is associated with local pathology such as Crohn's disease but also with systemic diseases such as obesity (Turnbaugh et al., 2009), diabetes (Cani et al., 2007) and hypertension (Holmes et al., 2008), among other chronic diseases (Proal et al., 2009; Lampe, 2008). The microbiome of various human body sites has already been described for the digestive tract, the mouth, the skin, the genital tract and it will continue with the « Human Microbiome Project » (Turnbaugh et al., 2007).

Microbiome and its disturbances are now a major area of microbiology research, as attested by the leading publications produced in this field. However, these advances in 'microbiomology' have not yet change the patterns of medical reflection. In the field of infectious diseases, bacteriologists still strive to obtain a pure culture preferentially isolating known pathogens and focus on the morphology, physiology, and genetics of the small subset of the microorganisms in the body that are known to cause disease. Hence, most bacteria present in a community are neglected and so are the complex relationships among members of the community. It is well-known that a large range of endogenous bacteria can cause opportunistic infections such healthcare associated infections (HAI), consecutive to microbiome disequilibrium, and that strict pathogens must cross the cutaneous or mucosal barriers covered with local microbiota before developing their virulence. Therefore, considering the whole human microbiota is required to better understand the physiopathology of infections and improve their prevention and treatment of bacterial infectious diseases that remain major challenges in public health. Confronting microbiomes analyses to clinical issues will warrant these analyses to be good predictors and markers in infectious disease. This implies that microbiomes analyses become routine, which is not yet the case.

#### **2.2 How to explore bacterial communities?**

Bacterial communities are classically assessed through culture-dependent methods based on colony isolation on solid medium, sometimes after enrichment by growth in liquid medium. This way, the description of complex communities inevitably involved time-consuming steps of growth on multiple media under varying conditions, while the full description of the community is not insured. Indeed, it is now obvious that the real microbial diversity is poorly represented by the cultured fraction, and conventional culture techniques have been shown to explore less than 1% of the whole bacterial diversity in environment samples, such as soil samples (Riesenfeld et al., 2004). It is also estimated that as much as 20% to 60% of the human-associated microbiome, depending on body site, is unculturable (NIH HMP Working Group et al., 2009). A renewal of studies on complex bacterial communities has been made possible by molecular methods that allow direct analysis of bacteria present in a

communities amounting to a super- or a hetero-organism. Metagenomics show clear differences between microbiomes in various body sites and together with metatranscriptomics and metaproteomics reveal how microbiomes contribute to organ and tissue functions. Consequently, human biology can no longer concern itself only with human cells: microbiomes at different body sites and functional metagenomics must be considered part of systems biology (Pflughoeft & Versalovic, 2011). The biological concept of hetero-organism further evolves into new medical conception on health and disease, involving not only human genes and cells but also related genomes from environment around and inside human body. For instance, the change in gut microbiome is associated with local pathology such as Crohn's disease but also with systemic diseases such as obesity (Turnbaugh et al., 2009), diabetes (Cani et al., 2007) and hypertension (Holmes et al., 2008), among other chronic diseases (Proal et al., 2009; Lampe, 2008). The microbiome of various human body sites has already been described for the digestive tract, the mouth, the skin, the genital tract and it will continue with the « Human Microbiome Project »

Microbiome and its disturbances are now a major area of microbiology research, as attested by the leading publications produced in this field. However, these advances in 'microbiomology' have not yet change the patterns of medical reflection. In the field of infectious diseases, bacteriologists still strive to obtain a pure culture preferentially isolating known pathogens and focus on the morphology, physiology, and genetics of the small subset of the microorganisms in the body that are known to cause disease. Hence, most bacteria present in a community are neglected and so are the complex relationships among members of the community. It is well-known that a large range of endogenous bacteria can cause opportunistic infections such healthcare associated infections (HAI), consecutive to microbiome disequilibrium, and that strict pathogens must cross the cutaneous or mucosal barriers covered with local microbiota before developing their virulence. Therefore, considering the whole human microbiota is required to better understand the physiopathology of infections and improve their prevention and treatment of bacterial infectious diseases that remain major challenges in public health. Confronting microbiomes analyses to clinical issues will warrant these analyses to be good predictors and markers in infectious disease. This implies that microbiomes analyses become routine, which is not yet

Bacterial communities are classically assessed through culture-dependent methods based on colony isolation on solid medium, sometimes after enrichment by growth in liquid medium. This way, the description of complex communities inevitably involved time-consuming steps of growth on multiple media under varying conditions, while the full description of the community is not insured. Indeed, it is now obvious that the real microbial diversity is poorly represented by the cultured fraction, and conventional culture techniques have been shown to explore less than 1% of the whole bacterial diversity in environment samples, such as soil samples (Riesenfeld et al., 2004). It is also estimated that as much as 20% to 60% of the human-associated microbiome, depending on body site, is unculturable (NIH HMP Working Group et al., 2009). A renewal of studies on complex bacterial communities has been made possible by molecular methods that allow direct analysis of bacteria present in a

(Turnbaugh et al., 2007).

the case.

**2.2 How to explore bacterial communities?** 

sample, while avoiding the bias of cultivability. The emergence of molecular ecology and metagenomics offers the potential of determining microbial diversity in an ecosystem without prior laboratory enrichment, isolation and growth on artificial media thanks to sophisticated methodological and computational tools. Generally, culture-independent approaches allow a precise description of an ecosystem by assessing its genetic diversity. As an example of molecular ecology efficiency, Dekio et al (2005) show that the total skin microflora explored by culture-independent molecular profiling is greater than previously believed, finding 22 potentially novel members, comprising 9 species and 13 phylotypes not yet described as members of skin microbiota (Dekio et al., 2005).

Metagenomics examine the complexity of the community by sequencing genomic libraries made from DNA extracted directly from the sample containing a complex mix of different bacteria. The complete metagenomic approach will give the total gene content of a community, thus providing data about biodiversity but also function and interactions (Tyson et al., 2004). For the purpose of biodiversity studies, metagenomics can focus on one common gene shared by all members of the community. The most commonly used cultureindependent method relies on amplification and analysis of the 16S rRNA genes in a microbiome (Nossa et al., 2010).

16S rRNA genes are widely used for documentation of the evolutionary history and taxonomic assignment of individual organisms because they have highly conserved regions for construction of universal primers and highly variable regions for identification of individual species (Woese, 1987). The 16S rRNA gene, in spite of some recognized pitfalls (von Wintzingerode et al., 1997), remains today the most popular marker for studying the specific diversity in a bacterial community. The different 16S rRNA genes representative of the community are amplified by PCR and then separated and identified either by cloning and Sanger sequencing or by direct pyro-sequencing (Nossa et al., 2010). Tools for sequencespecific separation after bulk PCR amplification, such as T-RFLP (Terminal-Restriction Fragment Lenght Polymorphism) (Kitts, 2001), D-HPLC (Denaturing High Performance Liquid Chromatography) (Penny et al., 2010), CDCE (Constant Denaturing Capillary Electrophoresis) (Thompson et al., 2004), SSCP (Single Stand Conformation Polymorphism) (Ege et al., 2011), DGGE (Denaturing Gradient Gel Electrophoresis) (Muyzer et al. 1993), TGGE (Temperature Gradient Gel Electrophoresis) (Zoetendal et al., 1998) and TTGE (Temporal Temperature Gradient Gel Electrophoresis) (Ogier et al., 2002), can also be used. Methods based upon separation in denaturing electrophoresis appear particularly suitable for the routine follow-up of microbiomes with low or medium diversity (Roudière et al., 2007). They provide a "fingerprint" of the community diversity.

### **2.3 16S rRNA gene PCR-TTGE: Advantages and limitations**

PCR-TTGE is a PCR-denaturing gradient gel electrophoresis that allows separation of DNA fragments in a temporal gradient of temperature (Yoshino et al., 1991; Ogier et al., 2002). PCR amplicons of the same size but with different sequences are separated in the gel. In a denaturing acrylamide gel, DNA denatures in discrete regions called melting domains, each of them displaying a sequence specific melting temperature. When the melting temperature (*Tm*) of the whole amplicon is reached, the DNA is denatured creating branched molecules. This branching reduces DNA mobility in the gel. Therefore, amplicons of the same size but with different nucleotide compositions can be separated based on differences in the

Temporal Temperature Gel Electrophoresis to Survey

of heteroduplexes and chimeric bands.

microbial ecology (Case et al., 2007).

lower taxonomic level being accurately identified.

Pathogenic Bacterial Communities: The Case of Surgical Site Infections 295

PCR can also be affected by preferential PCR amplification that may hinder the detection of some genotypes when a complex mix of DNA molecules is used as template. Preferential PCR amplification can be caused by primer mismatches at the annealing sites for some genotypes or by a lower rate of primer hybridization to certain templates due to a low local denaturation (Kanagawa, 2003). PCR carried out on complex bulk DNA can produce heteroduplexes particularly in later cycles when primer concentration decreases and the concentration of PCR products is high (Kanagawa, 2003). Chimeric amplicons can also be formed in later PCR cycles when template concentration is high enough to allow the reannealing of templates before primer extension (von Wintzergerode et al., 1997; Kanagawa, 2003). All these artifacts can generate additional signals that do not correspond to genotypes in the sample. Heteroduplexes and chimera produce additional bands in the TTGE pattern that lead to an overestimation of the diversity. These artifactual bands can be detected either on the basis of their very short migration distance or by sequencing. Consequently, the crude diversity index determined by simple band count should be optimized after exclusion

Last but not least, a major parameter in community studies is the choice of a molecular marker allowing the genotyping of the whole community. The notion developed by Woese that rRNA genes could identify living organisms by reconstructing phylogenies resulted in the adoption of 16S rRNA gene in microbiology (Woese, 1987). Its universality and the huge number of sequences stored in databases have established 16S rRNA gene as the "gold standard" not only in microbial phylogeny, systematics, and identification but also

For the purpose of identifying an isolated bacterial strain, the complete 16S rRNA gene (1500 bp) is generally used, giving accurate affiliation to a species in most cases. In PCR-TTGE experiments, the amplified fragments are short (200 to 400 bp) to allow migration in polyacrylamide gel. The fragment amplified should contain hypervariable regions of the 16S rRNA gene in order to compensate for the lack of information due to the small sequence size by a high rate of mutation. Bacterial 16S rRNA genes comprise nine hypervariable regions, V1-V9, exhibiting sequence diversity among species (Van de Peer et al., 1996). In most studies, the V3 region located in the 5' part of the gene is chosen (Jany & Barbier, 2008). However, the phylogenetic information is sometimes insufficient to achieve species identification. Depending on the bacterium, sequences provide identification to the genus or family level only. Consequently, the diversity of the community is not described by a list of bacterial species but by a list of operational taxonomic units (OTUs) corresponding to the

At the genomic level, rRNA genes are generally organized in multigene families (Acinas et al., 2004). The members of a rRNA multigene family are subject to a homogenization process allowing the multiple gene copies to evolve in concert. In a concerted evolution mode, mutation occurring in one copy will be fixed in all of them or lost from all. Thus, rRNA sequences show low variability within species, subspecies or genome (Liao, 2000). However, intra-genomic heterogeneity in the form of nucleotide differences between 16S rRNA gene copies are often described. For examples micro-heterogeneity has been identified in *Escherichia coli*, *Mycobacterium terrae*, *Paenibacillus polymyxa*, members of the classes *Mollicutes*, and *Actinomycetales* (Teyssier et al., 2003). Analyzing of complete genome sequences has recently assessed intra-genomic heterogeneity. For genomes with more than

behavior of their melting domains. When DNA is extracted and amplified from a complex community, TTGE lead to the separation of the different amplicons and produce a banding pattern characteristic of the community. The gradient obtained by varying the temperature over time in TTGE generally produces more clear and reproducible profiles than does the chemical gradient in DGGE.

Direct observation and counting bands on the TTGE profile provides a diversity score that roughly corresponds to the number of molecular species in the sample. However, it must be remembered that two amplicons with different sequences can give identical migration distances when their *Tm* are identical. The banding profile can be further analyzed by the affiliation of each band to a species or other taxon. Affiliation can be realized by comparing the migration distance of each band to a molecular ladder, named diversity ladder (Roudière et al., 2009; Ogier et al., 2002), constructed by using amplicons corresponding to species known to be representative of the community under study. A more accurate analysis of each band can be achieved by cutting bands from the gel, extracting DNA from bands and sequencing. This way the diversity ladder will be completed and updated. A method associating the use of diversity ladder with sequencing has shown its efficiency in describing bacterial communities of low complexity such as the gut microflora of neonates (Roudière et al., 2009). Such an approach is simple enough to survey dynamics of bacterial communities on a wide range of samples, particularly in health and disease (Jacquot et al., 2011).

PCR-TTGE appears not suitable for investigating highly diverse communities. This limit is due to number of bands that can be separated within the length of the gel. Optimization of TTGE conditions allows separation of bands by a minimum of 0.1 mm over all the gel length. We showed that the number of specific bands separable by TTGE could not exceed about 50 for the migration of an artificial diversity ladder. For instance, a ladder contained 53 different bacterial species found in stool of neonates can be efficiently separated in the optimized TTGE conditions (Roudière et al., 2009). For samples obtained from natural ecosystem, TTGE would be difficult to interprete if the diversity exceeds 25 to 30 bands.

Prior to TTGE migration, other technical steps are limitative and should be carefully considered and optimized. Particularly, DNA should be recovered and amplified from all the genotypes in the community, i.e. extraction and PCR should be as universal as possible. Special attention should be given to *Firmicutes* and *Actinobacteria* because they display thick and resistant cell wall. The extraction efficiency should be tested on a wide panel of bacteria to scan a large range of bacterial types. Extraction is generally improved by the use of largespectrum lytic enzymes and/or by a mechanical grinding (Roudière et al., 2009; Le Bourhis et al., 2007).

The PCR itself is a cause of limitations in the PCR-TTGE approach. Molecular methods are often praised for their sensitivity. However, this detection sensitivity can fail when complex samples are analyzed. For example, detection thresholds of 103-104 CFU/mL are currently described for universal PCR-TTGE or PCR-DGGE (Le Bourhis et al., 2007; Temmerman et al., 2003; Roudière at al., 2009). The detection limit in PCR-TTGE cannot be easily assessed as it depends on both CFU/g count of each species and the relative representation of species in the community. Minor populations of less than 1% of total population are undetectable by PCR-TTGE. This breakpoint is commonly reported for denaturing-gel-based methods used in microbial ecology (Ogier et al., 2002; Zoetendal et al., 1998; Roudière et al, 2009).

behavior of their melting domains. When DNA is extracted and amplified from a complex community, TTGE lead to the separation of the different amplicons and produce a banding pattern characteristic of the community. The gradient obtained by varying the temperature over time in TTGE generally produces more clear and reproducible profiles than does the

Direct observation and counting bands on the TTGE profile provides a diversity score that roughly corresponds to the number of molecular species in the sample. However, it must be remembered that two amplicons with different sequences can give identical migration distances when their *Tm* are identical. The banding profile can be further analyzed by the affiliation of each band to a species or other taxon. Affiliation can be realized by comparing the migration distance of each band to a molecular ladder, named diversity ladder (Roudière et al., 2009; Ogier et al., 2002), constructed by using amplicons corresponding to species known to be representative of the community under study. A more accurate analysis of each band can be achieved by cutting bands from the gel, extracting DNA from bands and sequencing. This way the diversity ladder will be completed and updated. A method associating the use of diversity ladder with sequencing has shown its efficiency in describing bacterial communities of low complexity such as the gut microflora of neonates (Roudière et al., 2009). Such an approach is simple enough to survey dynamics of bacterial communities on a wide range of samples, particularly in health and disease (Jacquot et

PCR-TTGE appears not suitable for investigating highly diverse communities. This limit is due to number of bands that can be separated within the length of the gel. Optimization of TTGE conditions allows separation of bands by a minimum of 0.1 mm over all the gel length. We showed that the number of specific bands separable by TTGE could not exceed about 50 for the migration of an artificial diversity ladder. For instance, a ladder contained 53 different bacterial species found in stool of neonates can be efficiently separated in the optimized TTGE conditions (Roudière et al., 2009). For samples obtained from natural ecosystem, TTGE would be difficult to interprete if the diversity exceeds 25 to 30 bands.

Prior to TTGE migration, other technical steps are limitative and should be carefully considered and optimized. Particularly, DNA should be recovered and amplified from all the genotypes in the community, i.e. extraction and PCR should be as universal as possible. Special attention should be given to *Firmicutes* and *Actinobacteria* because they display thick and resistant cell wall. The extraction efficiency should be tested on a wide panel of bacteria to scan a large range of bacterial types. Extraction is generally improved by the use of largespectrum lytic enzymes and/or by a mechanical grinding (Roudière et al., 2009; Le Bourhis

The PCR itself is a cause of limitations in the PCR-TTGE approach. Molecular methods are often praised for their sensitivity. However, this detection sensitivity can fail when complex samples are analyzed. For example, detection thresholds of 103-104 CFU/mL are currently described for universal PCR-TTGE or PCR-DGGE (Le Bourhis et al., 2007; Temmerman et al., 2003; Roudière at al., 2009). The detection limit in PCR-TTGE cannot be easily assessed as it depends on both CFU/g count of each species and the relative representation of species in the community. Minor populations of less than 1% of total population are undetectable by PCR-TTGE. This breakpoint is commonly reported for denaturing-gel-based methods used

in microbial ecology (Ogier et al., 2002; Zoetendal et al., 1998; Roudière et al, 2009).

chemical gradient in DGGE.

al., 2011).

et al., 2007).

PCR can also be affected by preferential PCR amplification that may hinder the detection of some genotypes when a complex mix of DNA molecules is used as template. Preferential PCR amplification can be caused by primer mismatches at the annealing sites for some genotypes or by a lower rate of primer hybridization to certain templates due to a low local denaturation (Kanagawa, 2003). PCR carried out on complex bulk DNA can produce heteroduplexes particularly in later cycles when primer concentration decreases and the concentration of PCR products is high (Kanagawa, 2003). Chimeric amplicons can also be formed in later PCR cycles when template concentration is high enough to allow the reannealing of templates before primer extension (von Wintzergerode et al., 1997; Kanagawa, 2003). All these artifacts can generate additional signals that do not correspond to genotypes in the sample. Heteroduplexes and chimera produce additional bands in the TTGE pattern that lead to an overestimation of the diversity. These artifactual bands can be detected either on the basis of their very short migration distance or by sequencing. Consequently, the crude diversity index determined by simple band count should be optimized after exclusion of heteroduplexes and chimeric bands.

Last but not least, a major parameter in community studies is the choice of a molecular marker allowing the genotyping of the whole community. The notion developed by Woese that rRNA genes could identify living organisms by reconstructing phylogenies resulted in the adoption of 16S rRNA gene in microbiology (Woese, 1987). Its universality and the huge number of sequences stored in databases have established 16S rRNA gene as the "gold standard" not only in microbial phylogeny, systematics, and identification but also microbial ecology (Case et al., 2007).

For the purpose of identifying an isolated bacterial strain, the complete 16S rRNA gene (1500 bp) is generally used, giving accurate affiliation to a species in most cases. In PCR-TTGE experiments, the amplified fragments are short (200 to 400 bp) to allow migration in polyacrylamide gel. The fragment amplified should contain hypervariable regions of the 16S rRNA gene in order to compensate for the lack of information due to the small sequence size by a high rate of mutation. Bacterial 16S rRNA genes comprise nine hypervariable regions, V1-V9, exhibiting sequence diversity among species (Van de Peer et al., 1996). In most studies, the V3 region located in the 5' part of the gene is chosen (Jany & Barbier, 2008). However, the phylogenetic information is sometimes insufficient to achieve species identification. Depending on the bacterium, sequences provide identification to the genus or family level only. Consequently, the diversity of the community is not described by a list of bacterial species but by a list of operational taxonomic units (OTUs) corresponding to the lower taxonomic level being accurately identified.

At the genomic level, rRNA genes are generally organized in multigene families (Acinas et al., 2004). The members of a rRNA multigene family are subject to a homogenization process allowing the multiple gene copies to evolve in concert. In a concerted evolution mode, mutation occurring in one copy will be fixed in all of them or lost from all. Thus, rRNA sequences show low variability within species, subspecies or genome (Liao, 2000). However, intra-genomic heterogeneity in the form of nucleotide differences between 16S rRNA gene copies are often described. For examples micro-heterogeneity has been identified in *Escherichia coli*, *Mycobacterium terrae*, *Paenibacillus polymyxa*, members of the classes *Mollicutes*, and *Actinomycetales* (Teyssier et al., 2003). Analyzing of complete genome sequences has recently assessed intra-genomic heterogeneity. For genomes with more than

Temporal Temperature Gel Electrophoresis to Survey

**of surgical site infection in cardiac surgery** 

coronary bypass procedures… (Filsoufi et al., 2009).

**3.1 Surgical site infections: State of the art 3.1.1 Epidemiology of surgical site infection** 

Pathogenic Bacterial Communities: The Case of Surgical Site Infections 297

**3. An original method of PCR-TTGE to learn more about the physiopathology** 

Surgical site infections (SSIs) are among the most frequent healthcare associated infections (HAIs), along with urinary tract and pulmonary infections, and remain an unresolved problem for modern medicine, their occurrence having significant impact on patient morbidity, length of stay and cost of care. Data from longitudinal surveillance studies show SSI rates of 1 to 5% (Klevens et al., 2007; Astagneau et al., 2009; de Lissovoy et al., 2009), whereas higher rates are reported from interventional studies, where control groups can have up to 8.5% infection rates (Bode et al., 2010; Perl et al., 2002), owing to different population case mixes. Large-scale epidemiological studies have identified risk factors for SSI that can be grossly classified as related to patient condition, surgical procedure and environment. The American National Nosocomial Infection Surveillance system (NNIS) developed an easily calculated risk index that combines the patient-related risk assessment of the American Society of Anesthesiologists (ASA) score, and 2 surgical procedure-related factors: type of surgery as defined by pre- or per-operative microbial contamination (Altemeier classification, from I – clean surgery- to IV – septic surgery), and duration of operation exceeding 75th percentile for a given procedure. For all categories of surgery, there is a linear increase in the incidence of SSI when the NNIS risk index increases (Coello et al., 2005). Beside these surveillance-derived risk assessment scores, prospective controlled studies identified many more factors associated with a higher risk of SSI. Most important are poorly controlled diabetes mellitus, malignant diseases, smoking, advanced age, per operative hypothermia, emergency surgery (Coello et al., 2005). Specific risk factors have been identified for cardiac surgery: obesity, pre operative myocardial infarction, chronic obstructive broncho-pulmonary disease, duration of extra corporeal circulation, early post operative bleeding, combined valve and

In the mid 1980s, 30 to 35% of nosocomial (i.e. hospital-acquired) infections were deemed evitable (Haley et al., 1985). A recent analysis of infection control interventional studies estimates that as many as 26 to 54% of SSIs could be avoided by comprehensive implementation of evidence-based prevention strategies, foremost of which are pre operative cutaneous antisepsis, no pre operative shaving of surgical site, timely antibiotic

Still, even when all known preventive measures are implemented, even in low-risk (i.e. clean, non urgent) surgery for low risk patients, SSI can occur. This seemingly irreducible rate of "inevitable" infections raises the problem of how surgical site infections develop. Success of preventive measures based on antisepsis and optimization of patient status compounds the hypothesis that infection results from disequilibrium between host defense mechanisms and microbial infectiveness. In a 10-to-1 inequity between human cells and colonizing microbes, the balance of power in ensured by integrity of skin and mucous membranes, both obviously disrupted by surgery (Wenzel, 2010). However, the intimate

prophylaxis and strict per operative glycemic control (Umsheid et al., 2011).

mechanisms of infection are not known, starting with the origin of germs involved.

one rRNA operon, 62% display some degree of sequence divergence between 16S rRNA loci in a same genome (Case et al., 2007). In PCR-TTGE, the intra-genomic 16S rDNA heterogeneity can lead to multiple bands for a single OTU and then to an overestimation of OTU diversity. This pitfall inherent to the 16S rRNA gene marker will be avoided by band sequencing. However, as a pre-requisite of diversity analysis by PCR-TTGE, the major known species expected in a particular ecosystem should be individually studied by TTGE in order to explore heterogeneity in 16S rRNA gene copies (Roudière et al, 2007; Michon et al., 2011).

Alternative markers can also be proposed such as *rpoB* (Case et al., 2007) but universal *rpoB* PCR primers allowing the exploration of the whole bacterial diversity can not be designed (personal data) and the databases remain poor in *rpo* sequences.

Several authors remarked that culture-independent methods regularly fail to identify species obtained using culture-dependent methods (Jany & Barbier, 2008). By contrast, culture-dependent methods have yielded information on the structure of microbial populations but they are limited by the in vitro growth capacity of most bacteria in a community. Such a discrepancy is not observed for all the communities studied. For instance, culture dependant and independent approaches of the premature neonate gut microbiome give globally congruent results (Roudière et al., 2009). However, it is accepted that culture-independent methods remain the only approach for monitoring the rapid dynamics of microbial communities. Nevertheless, the two types of methods reveal different images of the same community and combining culture-dependent and culture-independent methods may be worthwhile to obtain a more accurate view of the structure of the microbial community (Case et al., 2007).

In spite of the limited growth capacity of most bacteria, culture-dependent methods remain the sole approach available for monitoring sub-populations selected on the basis of phenotypic traits such as dependence to metabolites or resistance to antimicrobial agents. In this context, PCR-TTGE can be used after culture in specific conditions in order to describe the diversity of cultivable population. The colonies growing in diverse conditions can be bulked and further analyzed by PCR-TTGE as described before. This culture- and geneticbased mixed approach is particularly suitable to describe dynamics of populations according to their level of resistance to antimicrobial drugs in natural environments (Vanhove et al., 2011).

Culture-independent approaches have previously shown their interest in cardiology to detected new or atypical infectious agents (Marchandin et al., 2009; Daïen et al., 2010). Considering the interest and limitations of 16S rRNA PCR-TTGE, we will proposed a protocol for describing the diversity and following the dynamics of the bacterial community that colonize surgical wound of the patient during hospitalization for cardiac surgery. We will show how 16S rRNA PCR-TTGE is particularly suited to the low bacterial diversity encountered in aseptic surgical settings, where antibiotic prophylaxis and cutaneous antisepsis effectively reduce the bacterial load of patients.

In addition, one example of the use of 16S PCR-TTGE in a culture-dependent analysis will be detailed. This approach associates determination of Minimal Inhibitory Concentration (MIC) at the community level and determination of the diversity by 16S PCR-TTGE in the resistant sub-population at each concentration of antimicrobial agent.
