Probing the bacterial cell wall with chemical biology tools

After DNA and proteins, carbohydrates are the third language of life. Chapter 1 introduces the reader to this class of biomolecules, also called sugars or glycans, that can be found on the outer surface of almost all cells and plays a critical role as the social messengers of a cell. Although our knowledge about the role of glycans in eukaryotic cells has increased considerably in recent decades, our understanding of the glycan layer on bacterial cells is still very limited. Besides the carbohydrates that are present in both eukaryotes and prokaryotes an additional wide range of unique (e.g. microbial sialic acid), often very complex (e.g. pseudaminic acid), carbohydrates is present in prokaryotes. This chapter briefly introduces two research fields, carbohydrate chemistry and chemical biology, that when combined provide a powerful way to investigate the biological role of these unique bacterial carbohydrates at the molecular level. This chemistry-based approach, termed chemical microbiology, often starts with the development of a chemical synthesis for a target bacterial carbohydrate. Subsequently, the synthetic route towards this target allows for the introduction of unnatural functional groups, like chemical reporters, that result in the molecular tools needed to study their biological function. The studies described in this thesis, focus on developing such molecular tools to study the role of glycans and glycoconjugates in human gut bacteria and human-associated bacteria. Chapter 2 provides an overview of metabolic oligosaccharide engineering (MOE) a popular chemical biology technique to label glycans in living cells. In MOE, carbohydrates derivatives are synthesised with unnatural chemical reporters and used to study their incorporation in glycans of eukaryote to prokaryote species. The progress in this field over the last 6 years is reviewed in detail with a special emphasis on the synthesis of the unnatural carbohydrates from commercially available sources. The principle behind MOE is that these unnatural carbohydrates with e.g. azide, alkyne, cyclopropene, or isonitrile chemical reporter groups, are still recognised by the endogenous enzymes in the cell that salvage this new carbohydrate. In this way they can enter the associated biochemical pathways and end up in newly biosynthesised cellular glycans. Subsequent labelling techniques, such as strain promoted azide alkyne cycloaddition or tetrazine ligation, enable the visualisation of these incorporated unnatural carbohydrates with for instance fluorescence microscopy. Metabolic labelling is further explored in chapter 3. Key cell envelope glycoconjugates in the mucin-degrading gut microbiota member, Akkermansia muciniphila, were subjected to chemistry-based functional analysis, with Escherichia coli being used as a control species. Two novel non-toxic peptidoglycan (PG) probes were designed and synthesised to investigate the presence of PG in this species. Their design was based on the natural d-alanine dipeptide motif found in PG. Inspired by the fact that d-alanine dipeptide-derivatives were previously reported to be incorporated in newly synthesised PG, we synthesised a cyclopropene and isonitrile d-alanine dipeptide. Our probes proved to be non-toxic, as shown by growth and viable count analysis, and were therefore superior over existing PG probes. Another beneficial property was that the probes also did not influence the specific growth rate of A. muciniphila or E. coli. The PG probes were successfully incorporated into the peptidoglycan layer of A. muciniphila and visualised using a tetrazine click-ligation with a fluorophore. Our analysis proved for the first time that A. muciniphila has a PG layer. Besides PG labelling, we also investigated metabolic labelling of other glycoconjugates on the outer surface of A. muciniphila. This part of the study showed that azido-monosaccharide derivatives of N-acetylglucosamine, N-acetylgalactosamine, and fucose are successfully processed by A. muciniphila salvage pathways and incorporated into its surface glycoconjugates. Especially 6-azido-fucose was readily processed by the recently discovered l-fucose salvage pathway of A. muciniphila. The two compatible labelling techniques were next combined in a dual labelling experiment. Our isonitrile dipeptide peptidoglycan probe and 6-azido-fucose were successfully incorporated into A. muciniphila. Subsequent fluorescent labelling with bio-orthogonal techniques resulted in dual labelling of peptidoglycan and fucose-containing glycans in live A. muciniphila cells. With the positive results of MOE in A. muciniphila in hand, chapter 4 describes the further investigation of MOE. After successful validation of our Ac4FucAz probe for MOE in Bacteroides fragilis we continued their application in other human gut microbiota members, including the butyrate-producing Anaerostipes rhamnosivorans, Intestimonas butyriciproducens, and Eubacterium hallii. Labelling of these human gut microbes proved to be rather challenging with a-specific cellular labelling with the fluorophore being the major problem. Initial results, however, did show that a 6-azido-l-rhamnose probe resulted in fluorescent labelling of A. rhamnosivorans, which provides initial evidence for the existence of an as of yet undocumented salvage pathway. In this species the 6-azido-fucose probe was not salvaged. Via confocal microscopy and flow cytometry analysis we observed that the 6-azido-rhamnose probe was selective for A. rhamnosivorans in the presence of A. muciniphila. Such a co-culture experiment is a first step in mimicking the complex human gut microbiome. For E. hallii Ac4GalNAz gave clear metabolic labelling and the majority of the cell population could be labelled with the fluorescent dye after a strain-promoted azide alkyne cycloaddition. Other glycan probes (Ac4GlcNAz, Ac4FucAz, and Neu5Az) also resulted in labelling, but not as prominent as Ac4GalNAz. Surprisingly, MOE has never been reported for the common lab strain Escherichia coli MG1655. Curious to investigate this in more detail we started MOE in E. coli. However, no labelling was obtained when Ac4GlcNAz probe was added to E. coli, most likely due to the fast growth, metabolism and turnover. Only, when fresh Ac4GlcNAz probe was added every 30 minutes, metabolic labelling in E. coli was observed. To further investigate the influence of GlcNAc metabolism in E. coli on MOE, single-gene knock-outs of E. coli GlcNAc metabolism from the Keio collection were investigated. Labelling was observed for NagA (N-acetyl glucosamine 6 P deacetylase) and NagK (N-acetyl-d-glucosamine kinase) E. coli mutants. Both enzymes are involved in the last step of the biosynthesis towards UDP-N-acetylglucosamine. When the overall E. coli metabolism was inhibited, after addition of the respiration inhibitor sodium azide, no metabolic labelling was observed. These results indicate that MOE in E. coli is possible, but challenging and can only be performed under specific circumstances. An investigation into the total synthesis of pseudaminic acid, a sialic acid produced by specific human-associated prokaryotes, is described in chapter 5. Sialic acids are typically found at the terminal positions of surface glycoconjugates in both eukaryotes and prokaryotes. Other related microbial sialic acids are legionaminic and acinetaminic acid. The total synthesis of these microbial sialic acids is notoriously difficult, as exemplified by the fact that only a few chemical synthesis routes towards them are currently known. Our total synthesis of pseudaminic acid started from the readily available amino acid l-threonine that was transformed into a key versatile Garner aldehyde derivative intermediate. With this aldehyde in hand, the Henry nitro-aldol condensation reaction was investigated. After studying numerous conditions, such as asymmetric catalysis or elongated reaction times, and extensive optimisation efforts we were never able to obtain the Henry reaction product to continue with this route. As an alternative, a tethered aminohydroxylation was investigated for its ability to introduce the key functional group and stereochemistry onto an intermediate obtained from the Garner aldehyde derivative. This reaction indeed gave the desired amino-alcohol motif in the correct stereochemistry, but another diastereomer proved very difficult to separate from the desired product. After some additional transformations and protection steps we obtained a derivative in which the primary alcohol could be oxidised to provide a hexose intermediate that resembles the hexose intermediate present in pseudaminic acid biosynthesis. This key hexose intermediate will likely enable a subsequent Barbier reaction, a chain elongation step, in future studies. With most of the key transformations accomplished, the completion of a pseudaminic total synthesis based on l-threonine should soon be possible. Besides finishing the total synthesis, future work should also focus on adapting this synthesis route to allow installation of chemical reporter groups on pseudaminic acid for its application in MOE. Chapter 6 is the general discussion about all the work mentioned in the other chapters. It also contains additional information and suggestions for further research in the field of chemical microbiology.