The aim of this work was to evaluate the effect of the fermentation conditions, and in particular the pH and temperature, on the surface properties of L. rhamnosus GG and its adhesion to Caco-2 cells. The influence of the production process on the surface characteristics and adhesion abilities of probiotic lactobacilli is a relatively unexplored area. Recent studies from our research group and other published works have shown that the incubation time and the composition of the fermentation medium are important factors [7, 14, 23], whereas studies focusing on the downstream processing, e.g. drying of probiotics, have shown that drying affects the physicochemical surface properties of Lactobacillus cells [24, 25]. Identifying possible ways to control the surface properties of the bacterial cells is very important in order to produce cells with enhanced functionality. This is the first study of its kind that looks at the fermentation and processing part and tries to relate the physical and biochemical properties of the bacterial cells to find answers related to the functional aspects of the probiotics, such as adhesion to the human GI tract.
It can be deduced from the growth curves in Figure 1 that carrying out the fermentation at suboptimal temperatures (i.e. 25°C and 30°C) resulted in lower growth rates, and in the case of the 25°C to a lower final cell density too, compared to 37°C. In contrast, the pH of the fermentation, within the range studied, did not seem to affect considerably the growth rate, nor the final cell density. The literature has shown that the pH and temperature influence the growth behaviour of lactobacilli and the final yield obtained, although substantial differences can be observed, depending on the strain. It has also been reported that a lower pH of fermentation, for example pH 5 compared to pH of around 6, is likely to result to the production of physiologically more robust cells , which are able to survive better freezing and frozen storage , freeze drying  and acid stress . In the case of temperature, it was also suggested that suboptimal conditions of growth, for example 30°C compared to 37°C, produced cells that were better able to survive freezing and frozen storage. A link was also shown to exist between the membrane fatty acid composition and increased cryotolerance . The conditions used in this study for the control fermentation, i.e. 37°C and pH 6.5, are typical for Lactobacillus growth and reflect the optimal conditions for achieving high cell yields, while the fermentation medium used is typical of a growth medium used for large production, and was optimised in a previous study in terms of cell growth and cell survival upon freeze drying . The rationale behind selecting pH and temperature values lower than the optimal was the increased robustness and improved technological properties observed in previous studies, which were most likely associated with compositional and conformational changes taking place at the cell membrane. Thus, it could be likely that important changes were also taking place at the surface of the cells when growing the cells at such conditions, which would affect their adhesion properties.
The adhesion levels of the cells from the different fermentations ranged from 40 to 120 bacteria per Caco-2 cell (Figure 5), which is within the range of previous reports [14, 29, 30]. The main source of knowledge about the adhesive properties of lactobacilli is from studies using in vitro model systems, although in vivo/ex vivo studies have also been used to a much lesser extent . Various studies have shown a good correlation between in vitro adhesion, using for example Caco-2 and HT29 cell lines, and in vivo adhesion based on the results from human intervention studies [32, 33]. Tissue culture cells, such as Caco-2, HT-29 and HT-29 MXT cells, are commonly used for in vitro adhesion studies [34–36]. Caco-2 cells exhibit many properties of the small intestine as they form a polarised monolayer of differentiated columnar absorptive cells expressing a brush border . In the present study, with the exception of the cells from the fermentation carried out at pH 5, the rest of the cells were significantly less adhesive to Caco-2 cells compared to the cells from the control fermentation (p < 0.05).
The lowest adhesion to Caco-2 cells was observed for the cells from the 25°C and uncontrolled pH fermentation, which had a higher total N content and N/C ratio as compared to the control sample. The cells from the 25°C fermentation showed also a slightly lower total O content and O/C ratio compared to the control fermentation. In the same way, the cells from the 30°C were characterised by a lower adhesion value and a higher total N content and N/C ratio (0.10) than the control fermentation. The above suggest that the exposure of proteins at the surface of the cells was higher than in the case of the control fermentation, whereas the exposure of carbohydrates and phosphate-containing compounds was probably lower. Consequently, it could be deduced that a cell surface richer in proteins resulted in a lower adhesion to Caco-2 cells, a conclusion that has also been suggested by our previous published studies . No differences were observed between the concentrations of various functional groups, especially the C-(O,N), C=O and O-C-O groups, which could correspond to carbohydrates. On the other hand, a considerable increase in the case of the O = (C,P) group for the uncontrolled pH, pH 5.5, 25°C and 30°C fermentations was observed compared to the control fermentation. This could potentially suggest an increase in the concentration of amide bonds, and thus of proteins. The results obtained from the XPS analysis coincided with the FT-IR results, although it is difficult to quantitatively correlate these. The FT-IR data showed a peak shift in the carbohydrate region of the spectra (1200 – 950 cm-1), as well as a number of peaks that emerged in the amide I and II regions, which could suggest that conformational changes took place in the cell surface components. Principle Component Analysis (PCA) was performed on the FTIR data. Though the subtle differences observed in the peak shift were not picked in the PCA analysis and were not statistically significant. These changes along with other compositional and conformational modifications of the bacterial surface might have an effect on the net surface properties such as adhesion to Caco-2 cells. Interestingly, the hydrophobicity of the cells from the 25°C and 30°C fermentations was drastically reduced compared to the control, which is difficult to explain, as the general consensus is that hydrophilic cells are most likely covered by carbohydrates [7, 38, 39]. However, this depends on the strain, and it most likely applies for Lactobacillus strains that contain the highly hydrophobic S-layer proteins, meaning that fermentation conditions that do not favour the expression of the S-layer proteins on the surface result in more hydrophilic cells. However, L. rhamnosus GG does not contain an S-layer. In addition, calculation of the GRAVY index for the key surface proteins that were identified by proteomic analysis, and which are listed in Table 4, indicated that they were all hydrophilic. This could thus explain the observed decrease in the hydrophobicity for these particular samples.
The cells from the fermentations carried out at pH 6, pH 5.5, and with no pH control, exhibited significantly (p < 0.05) lower adhesion to Caco-2 cells compared to the cells from the control fermentation. The atomic ratios and the concentrations of the functional groups obtained from XPS analysis, were similar to those for the 25°C and 30°C fermentations, suggesting that the cells had a protein-rich surface with very low amounts of phosphate-containing compounds. These observations were also in agreement with the hydrophobicity values, which were considerably lower than in the case of the control fermentation. Regarding the fermentation carried out at pH 5, the cells showed higher adhesion than the cells from the control fermentation, and very high hydrophobicity as compared to the control (p < 0.05). It is interesting to note that the XPS data in this case were more similar to the data from the control fermentation rather than to the other fermentations, and more specifically the total N and O content. Also, the concentrations of the C-(O,N), C=O and O-C-O groups were all similar, suggesting a similar surface composition to the cells from the control fermentation. The differences in adhesion between the cells from the pH 5 fermentation and the control fermentation could be either due to differences in the surface composition, in particular carbohydrates, which were not picked up by the XPS analysis, or to differences in the levels or profiles of the surface proteins. However, proteomic analysis indicated no differences in the protein profiles between all the fermentation samples, although quantitative analysis was not conducted; such work is planned for the future. In summary, all the samples contained proteins that have been associated with the adhesion of various lactobacilli. Table 4 enlists the proteins identified on the bacterial cell surface, their target site and molecular mass, as reported in previous studies. Among these, α-enolase, elongation factor Tu (EF-Tu), glyceraldehyde-3-phosphate dehydrogenase (GAPDH), and GroES chaperonin stand out in terms of likely importance. More specifically, α-enolase has been shown to be involved in the adhesion of L. johnsonii and L. crispatus to extracellular matrix components [40, 41], EF-Tu in the adhesion of L. plantarum and L. johnsonii to intestinal cells, GroEL in the adhesion of probiotics and pathogens to the gastrointestinal mucosa , and GAPDH in the adhesion of L. plantarum to Caco-2 cells . It is interesting however to note that the association of enolase and GAPDH with the cell wall of L. crispatus has been shown to be pH dependent, with the strongest association being at pH 5 . In accordance with this, the high levels of adhesion observed for the grown at pH 5, could be attributed to the stronger association of these two proteins with the cell wall of L. rhamnosus GG, although this is a hypothesis that needs to be investigated further.
A pili containing human mucus binding protein has been identified in L. rhamnosus GG, which could explain how this strain may persist in the host and compete with pathogens for residence sites in the human intestinal tract . The study reported that the pili were required for the adhesion of the bacteria to the host and suggested a possible role of pili in other probiotic effects as well. Another study from the same group reported a number of pilin subunits in the same strain and showed that they play a role in the adhesion of bacteria to the intestinal mucus . In the present study the pili protein was detected, however not at significant levels, therefore it was not included in the analysis. A similar result was also reported in another study on the surface proteins of L. rhamnosus GG . The reasons for this are likely to be differences in the extraction process or the growth conditions.