Bacterial adhesion is a complex process involving several factors, including: (i) surface-related properties (hydrophobicity, electrical charge, roughness, and topology); (ii) cell morphological properties (size, volume, dimension, and shape); (iii) the cell surface (chemical properties, envelope type, exposed proteins, and exopolysaccharide (EPS)); and (iv) the cell’s ability to move [99]. Adhesion is a key step for bacteria (pathogenic or not), allowing colonization and growth at a host-specific site [100]. The bacterial adhesion process consists of two phases: a non-specific phase, involving physicochemical bonds, and a specific phase involving molecular factors exposed on both host and bacterial cell surface [101]. The STEC–MFG association can be seen as a host–bacteria adhesion facilitated by the origin of the MFGM and its similarities with the membrane of intestinal cells [29,68]. In this context, both biological membranes will interact together through surface components. Various glycoconjugates are anchored on the MFGM surface (Figure 2) and can act as ligands.

Non-specific interactions have been described as the first step of adhesion, which is reversible and occurs rapidly (in the order of ~1 min) [102]. The process of initial bacterial adhesion is still not clearly understood, and physicians and microbiologists are working together to clarify the mechanisms. It is widely accepted that bacterial interaction is conducted according to the Derjaguin, Landau, Verwey, and Overbeek (DLVO) theory [103] and the extended DLVO theory [104]. The DLVO theory describes the force between charged surfaces interacting through a liquid. However, this theory may not be appropriate for modeling bacterial adhesion owing to the numerous processes involved and the influence of both biological and environmental factors (pH, ionic strength, and temperature) [105,106]. The non-specific phase of adhesion is a consequence of the balance between attractive and repulsive forces that are set up between the bacterium and the surface where it could adhere [107,108]. These forces include non-covalent interactions such as electrostatic interactions or surface charges, van der Waals forces, and Lewis acid/base interactions, as well as hydrophobic interactions [109,110]. Hydrophobic interactions and surface charges are the primary forces influencing bacterial adhesion [111].

Bacterial molecules involved in adhesion, called adhesins, recognize specific oligosaccharide moieties or peptide residues on the surface of target cells. There are many different adhesins, including porins, complex protein structures, glycoproteins, and glycolipids. Three main types of adhesin–receptor interactions have been described: lectin–glycan; protein–protein; and hydrophobin–protein [131]. Lectins are key factors in bacterial adhesion mechanisms [132,133,134]. Lectins are adhesins that recognize glycoconjugates, the sugar epitopes generally associated with proteins or lipids. Glycoconjugates are polymeric carbohydrates composed of monosaccharides arranged in chains and preferentially present on the external leaflet either attached to lipids or proteins [135]. Commonly, the polysaccharides of glycoconjugates are referred to as the ‘glycan layer’ or ‘glycocalyx’ [136]. The glycocalyx is directly exposed to the environment, allowing interactions with other cells to facilitate cell communication, immune regulation, and adhesion [137].

A wide range of STEC isolates can be responsible for human infections, and these can be genetically different [138]. However, regardless of the strain or serogroup, STEC possess virulence factors (Figure 1) that allow attachment to intestinal epithelial cells (IECs), and these adhesion factors are generally considered essential for infection. A large range of polysaccharides exists, but only a subset is exposed at the cell surface where they can be recognized by complementary receptors. Adhesins can be found at the distal end of bacterial pili (or fimbriae). These are bacterial extracellular appendages approximately 1 to 20 μm long and <2 to 10 nm in diameter [139]. Other adhesins are anchored directly in the biological membrane of bacteria and are referred to as afimbrial adhesins [59,63].

STEC detection in food matrices classically relies on four different steps: sample preparation; enrichment; detection; and confirmation by bacterial isolation. The enrichment step consists of adding an enrichment broth to the matrix to enable growth of the target bacteria. In the detection step, a genetic method is implemented to detect the presence of target bacteria by PCR screening. Finally, the confirmation step is carried out. This confirmation is based on isolation of target bacteria grown on selective media. Immunoconcentration tools using magnetic beads spiked with antibodies can also be used in this step. The ISO TS13136:2012 is the standard currently used to detect and isolate STEC belonging to O157, O26, O103, O111, and O145 serogroups and carrying eae and stx genes in food samples.

STEC detection in raw milk cheeses is particularly challenging. First, bacterial DNA is extracted from a specific volume of enrichment broth. Then, STEC target genes (eae, stx, and genes encoding one of the five somatic antigens) are detected. However, the microflora of cheese contains bacteria that carry some of the genes used to screen for STEC. For example, some non-STEC strains of E. coli (such as enteropathogenic E. coli (EPEC)) carry the eae gene or contain phages carrying the stx gene. Some other bacterial strains can also carry the stx gene, including Citrobacter freundi, Shigella spp., Acinetobacter, Aeromonas spp., Hafnia alvei, Escherichia albertii, Escherichia cloacae, and enterotoxigenic E. coli (ETEC) [44].The presence of these bacteria in raw milk cheese can lead to positive PCR results even though STEC isolates are not present in the enrichment broth (false positives). Furthermore, the performance of the methods (LOD) varies depending on the methods used and the matrices analyzed.

In general, STEC detection is more difficult in cheese than in meat. The LODs of the different detection methods are approximately 5–10 cells/g and 10–50 cells/g for bovine meat and raw milk cheese, respectively [220,221]. The presence of a richer flora and a higher amount of fat in cheese compared with meat may be an explanation. As discussed above, bacteria are preferentially found in contact with MFGs in raw milk products; however, the available detection kits perform DNA extraction on the pelleted fraction. Moreover, MFGs can interfere with DNA extraction methods by blocking spin column filters and acting as a PCR inhibitor [222,223]. Lower efficiency of bacterial DNA extraction can lead to false negative PCR results from enrichment broth samples. Several authors have described this phenomenon for various milk origins and suggest performing the extraction on both the cream and pelleted fraction. Sun et al. showed that cream harbors bacterial species that may be underestimated when skimmed milk, rather than whole milk, is used for DNA extraction [96]. Stinson et al. showed that a significant amount of human and bacterial cells remains with the cream and that bacterial DNA profiles can vary between fractions, especially for staphylococcal species [223]. The authors suggested that high-speed centrifugation may be insufficient to pellet bacterial or eukaryotic cells from milk. Furthermore, MFGs and proteins such as caseins can disrupt the interaction between immunomagnetic beads and STEC during the confirmation step. Tween 20 can be added at this stage to improve sample homogenization and block non-specific interactions [224].

Finally, STEC isolation to confirm the presence of the bacteria is also very difficult in raw milk cheese because the cheese microbiota limits STEC growth on agar plates. In addition, the different challenges encountered during cheese processing as well as the stresses suffered by STEC during the detection protocol can lead to viable but nonculturable (VBNC) isolates [225].

These studies emphasize the importance of using whole milk instead of skimmed milk for DNA extraction, but MFGs can perturb downstream applications. To improve the recovery rate of STEC in raw milk products, it seems essential to identify the nature of the STEC–MFG association in order to dissociate the two before performing the detection process. The identification of milk components involved in PCR inhibition and the improvement of DNA purification methods would allow the development of new kits to extract bacterial DNA from milk and cream. It should be noted that, despite these limitations, available DNA kits are still very effective. Quigley et al. showed that commercial kits provided very pure DNA suitable for PCR amplification from raw milk and raw milk cheese [226]. Furthermore, ISO 6887 recommends the addition of Tween 80 when enriching high-fat matrix to improve detection and bacterial isolation. However, no formal study has shown a significant effect due to the addition of Tween 80. Finally, at the enrichment temperature (37 °C or 41.5 °C), the lipids form a surface layer that may contain the desired bacterial cells and contribute to the reduction of available oxygen and thus modify the physiological state of the bacteria.

Figure 2 brings together the different concepts discussed in this section.

One of the industry’s goals should be a non-invasive method to eliminate pathogenic bacteria in raw milk without affecting the nutritional qualities and raw milk microbiota of the final product. Many techniques exist, such as bactofugation and microfiltration, [227], but these techniques affect MFG structure [82] and also remove the raw milk microflora. In our lab, we performed a raw milk skimming assay by electric centrifuge, and E. coli were not found in significant numbers in the cream fraction [228]. Stronger centrifugal forces are applied by the centrifuge; therefore, the STEC–MFG association is probably too weak to overcome the centrifugal forces. However, it was reported that natural creaming produces reduced-fat milk with a lower bacterial count and fewer somatic cells [229]. MFGs spontaneously rise to the surface due to the difference in density between MFGs and the aqueous phase (Stokes’ law) [230]. As previously discussed, STEC were predominantly found in the cream layer after raw milk creaming [88]. Therefore, performing natural creaming methods before cheese transformation could decrease the level of STEC in the final product. However, no study has been conducted in experimental field conditions (with low STEC contamination levels).

As bacterial adhesion is the first step of infection, inhibiting this step is a key strategy for infection control. Competition for the natural binding sites of pathogenic bacteria by mimetic receptors could inhibit pathogen attachment. Several natural food components could act as efficient inhibitors of pathogen adherence [30,100,231,232,233], especially milk components [29,234]. Moreover, numerous experimental studies have shown that the association of bacteria with MFGs could prevent the adhesion of several enteropathogens to enterocytes through mimetic receptors [142,172,235,236,237,238,239,240]. To avoid STEC adhesion to the epithelium of the intestinal mucosa, the STEC–MFG complex must be maintained at the site of STEC adhesion. Therefore, the expression of STEC genes involved in adhesion must be able to occur in product and during the human digestive process.

MFGM glycoconjugates are the main macromolecules involved in the anti-adhesive properties of milk against enteropathogens [29,73]. The MFGM protein fraction shows similarities to intestinal cells. Major MFGM proteins are conserved between species, although there are variations in protein expression levels and molecular functions between species and stages of lactation [241]. The MFGM was recently recognized as a high value-added ingredient, and the valorization of this by-product is constantly increasing. The MFGM, or some of its components, are added to infant milk formulas because of the MFGM’s beneficial properties [241,242,243,244,245]. More studies should be conducted to identify the MFGM surface glyco-epitopes recognized by STEC. These data could lead to pharmaceutical development of a specific drug to treat STEC infections. Currently available therapeutic solutions for STEC are controversial.