Immune regulation by human colonic bacteria and short-chain fatty acids

Winaris, Nuning


The intestinal microbiota plays a crucial role in the homeostasis of the human gastrointestinal tract by maintaining an anti-inflammatory status. Microbial imbalance in the gut, which is often referred to as ‘dysbiosis’, is known to be one of the major contributors to many human diseases, including inflammatory bowel disease (IBD). IBD is characterized by a chronic inflammation, but the current available treatment with anti-inflammatory drugs is often not effective. Therefore, the overarching aim of this thesis was to try to find bacterial strains or bacterial metabolites that have an immunomodulatory (i.e. an anti-inflammatory) function and may therefore be used as a therapy or treatment of IBD. In collaboration with the Rowett Institute, University of Aberdeen in Scotland, UK and the Department of Medical Microbiology, at the University Medical Centre Groningen, The Netherlands, a large number of colonic anaerobic bacteria were isolated from healthy patients. In our lab, we cultured more than 100 of these different strains of bacteria under strictly anaerobic conditions. We observed their growth characteristics and investigated their immunomodulatory properties. A high ratio between secreted anti-and pro-inflammatory cytokines (IL-10/IL-12 ratio) has been reported to be an indicator of positive correlation between in vitro trials and the attenuation of clinical symptoms in in vivo mouse models of colitis. Our initial screening therefore started with stimulation of peripheral mononuclear blood lymphocytes (PBMCs) with standardized concentrations of the bacteria or their culture supernatant. We determined the cytokine secretion of these PBMC after bacterial stimulation and determined the IL-10/IL-12 ratio. Additionally we checked if the viability of the PBMC was not affected by the bacterial strains or their metabolites which were secreted in the supernatant. As IBD is characterized by periods of remission and occasional flare-ups (periods of higher inflammation), we investigated both healthy and disease situations. To achieve this, we added heat inactivated bacteria (HIB) to our PBMC culture. These HIB triggered a strong induction of both pro-and anti- inflammatory cytokines, which is reminiscent of actual inflammation. Co-stimulation with both HIB and our bacterial strains enabled us to investigate the effect of the bacterial strains ‘during inflammation’. Additionally, we checked whether the bacterial strains were able to trigger NF-κB signalling via Toll like receptor (TLR) activation, which is one of the most known mechanisms to modulate immune response of the host. Finally, to assess oxidative stress, which is known to occur during flare-ups and damages the epithelial cells, we investigated whether the different bacterial strains were able to modulate nitric oxide (NO) secretion by a mouse macrophage (RAW 264.7) cell line. Again we were able to mimic an inflammatory situation by addition of bacterial lipopolysaccharides (LPS) to these cells and could thereby also investigate the effect of bacteria in an already inflamed gut.

In Chapter 2 we summarize the characterisation of 68 different colonic anaerobic bacteria tested. Most importantly, we found that there is a large variation among the tested strains in their immunomodulatory properties. The variation in induction of cytokine and NO secretion as well as in the ability to trigger NF-κB signalling between strains was significant. Interestingly, the bacterially induced immune profiles were highly strain dependent and not characteristic for a specific bacterial species. We must therefore conclude that generalizations cannot be made easily and any newly discovered strain needs to be individually investigated to determine its immunomodulatory properties. However, we could identify three different immune profiles, resulting from bacterial stimulation. The first ‘immunostimulatory’ profile was characterized by strongly inducing cytokine secretion in PBMCs. Most of these strains also elicited relatively high concentrations of NO secretion and strong NF-κB signalling after TLR activation. The second ‘immunomodulatory’ profile was characterized by induction of only moderate amounts of cytokine secretion. However, when an inflammatory status was mimicked (addition of HIB), these strains were able to attenuate the secretion of pro- inflammatory cytokines. The final ‘immuno suppressive’ or ‘silent’ profile was characterized by a low capacity to induce cytokine or NO secretion. More importantly, when HIB was added as an inflammatory stimulus, these strains attenuated the resulting pro-inflammatory cytokine response.

Several studies showed that there is a negative correlation between the relative abundance of F. prausnitzii and the disease severity of IBD, therefore a causal connection has been suggested. Indeed, several in vivo studies have shown that addition of certain strains of F. prausnitzii could attenuate clincal symptoms of colitis in mice. We therefore focused on another 28 bacterial strains that all belonged to the species of F. prausnitzii in Chapter 3. After thorough in vitro investigation we have to conclude that the immunomodulatory properties are really strain specific, as the tested properties of the strains (cytokine, NO secretion and NF-κB signaling via TLR activation) do not correlate with genomic phylogenetic clusters. Among the different strains tested, we found all three different immune profiles that were observed in Chapter 2. Moreover, the general assumption that all F. prausnitzii strains induce strong IL-10 secretion was found to not be universally true, as there were some ‘silent’ strains that hardly induced any IL-10 secretion. Interestingly, one of the strains that was the strongest activator of NF-κB signaling, hardly induced any cytokine secretion, which suggest some specific mechanism to prevent downstream effects. Indeed this strain manifested a ‘silent’ profile and as such was considered to be of interest for further in vivo trials.

During the in vitro screening procedures described above we observed that the culture supernatant of all strains tested was able to attenuate HIB induced cytokine secretion. Further investigations showed that the original growth medium (without any bacteria present), and especially the short chain fatty acids (SCFA) in the medium also triggered this attenuation effect. SCFA were reported to have immunomodulatory effects, but the studies were not  all consistent. Most studies showed that butyrate induced  IL-10  secretion  by  immune cells and thereby triggered regulatory T cell differentiation. In Chapter 4 we compared the effects of acetate and butyrate on different immune cell mechanisms. We found that butyrate decreased cell viability when administered at higher concentrations, whereas similar, or even higher concentrations of acetate did not affect cell viability. Interestingly, although both acetate and butyrate were able to attenuate HIB induced secretion of pro- inflammatory cytokines, only acetate was able to increase the anti-inflammatory cytokine IL-10. More importantly, butyrate actually decreased the secretion of the anti-inflammatory IL-10 in vitro in both PBMC as well as CD14+ monocytes. This decreased IL-10 secretion could result in an overall more inflammatory response (lower IL-10/IL-12 ratio) compared to the response triggered by acetate. To investigate how butyrate and acetate elicited their effects, we tried to determine the mechanism by which they affected cytokine secretion. As G protein-coupled receptors (GPCR) are known to mediate the effect of SCFA, we blocked the expression of GPCR in monocyte-derived dendritic cells (MDDC) with specific inhibitors. The attenuating effects of SCFA on HIB induced cytokine secretion were still observed, suggesting that modulation of cytokine secretion used an GPCR-independent mechanism. Butyrate (and to a lesser extent acetate) are also known to affect histone acetylation and could thereby modulate gene transcription. We found that butyrate indeed increased histone acetylation, which may point to a mechanism used in modulation of cytokines secretion. Interestingly, NF- κB activation was also found to be differentially modulated by acetate compared to butyrate, although the underlying mechanism has not yet been elucidated.

As we found clear effects of both acetate and butyrate on cytokine secretion by PBMC and CD14+   monocyte, we wondered if other biological pathways would be affected as well. We therefore stimulated CD14+ monocytes with acetate and butyrate (with or without co- stimulation with HIB) and investigated the resulting transcription profile in Chapter 5. Our monocytes showed the same attenuation of pro-inflammatory cytokine secretion in HIB induced samples as was seen in the PBMC and CD14+ monocyte. Acetate did not cause major effects in the number of regulated genes, but butyrate significantly affected the regulation of many different (immune) pathways and genes therein.

To compare the effects that SCFA might have on epithelial cells with the effects we found on immune cells, we investigated the effect of SCFA on an ex vivo 3D porcine ileum organoid model in Chapter 6. We exposed the organoid cells to acetate and butyrate and performed a transcriptomic analysis. Similar to the results of the transcriptomic analysis of CD14+ monocyte, butyrate proved to elicit greater changes in gene expression compared to acetate and substantially affected apoptosis and cell-cycle related pathways. In contrast, acetate mainly affected cellular metabolism process- related pathways, suggesting a less damaging effect on gut epithelial function compared to butyrate.

To conclude our studies, we tested several bacterial strains that appeared promising in the in vitro trials, in an in vivo mouse DSS-induced colitis model in Chapter 7. We observed attenuation of the clinical symptoms after addition of these ‘silent’ strains, which confirm our hypothesis that the bacterial strains that induced an ‘silent’ profile would be able to reduce colonic inflammation.

In the last chapter, Chapter 8, we summarize and discuss the combined results from this thesis in the context of other studies regarding host- microbe interactions, probiotics and the effects of SCFA. We explain how these findings contribute to a better understanding of the immunomodulatory properties of colonic anaerobic bacterial strains and their metabolites and provide suggestions for future research, and reflect on the overall aim of this thesis.