Staphylococcus aureus is a major threat to public health systems all over the globe. This second most cause of nosocomial infections is able to provoke a wide variety of different types of infection in humans and animals, ranging from superficial skin and skin structure infections to invasive disease like sepsis or pneumonia. But not enough, this pathogen is also notorious in acquiring and/or developing resistance to antimicrobial compounds, thus limiting available treatment options severely. Therefore, development of new compounds and strategies to fight S. aureus is of paramount importance. But since only 1 out of 5 compounds, which entered clinical trials, becomes a drug, the preclinical evaluation of promising compounds has to be reconsidered, too. The aim of this thesis was to address both sides of this problem: first, to improve preclinical testing by incorporating in vivo imaging technologies to the preclinical testing procedure in order to acquire additional and clearer data about efficacy of promising compounds and second, by evaluating lysostaphin, which is a promising, new option to fight S. aureus infections. The first aim of this thesis focused on the establishment of a dual modality in vivo imaging platform, consisting of Bioluminescence Imaging (BLI) and Magnetic Resonance Imaging (MRI), to offer detailed insights into the course and gravity of S. aureus infection in the murine thigh infection model. Since luciferase-expressing S. aureus strains were generated in former studies and enabled thus bioluminescence imaging of bacterial infection, this technology should be implemented into the compound evaluation platform in order to non-invasively track the bacterial burden over time. MRI, in contrast, was only rarely used in earlier studies to visualize and measure the course of infection or efficacy of anti-bacterial therapy. Thus, the first set of experiments was performed to identify benefits and drawbacks of visualizing S. aureus infections in the mouse model by different MR methods. Native, proton-based MR imaging showed in this regard increased T2 relaxation times in the infected thigh muscles, but it was not possible to define a clear border between infected and uninfected tissue. Iron oxide nanoparticles and perfluorocarbon emulsions, two MR contrast agents or tracer, in contrast, offered this distinction. Iron oxide particles were detected in this regard by their distortion of 1H signal in proton-based MRI, while perfluorocarbon emulsion was identified by 19F MRI. Mammals do not harbor sufficient intrinsic amounts of 19F to deliver specific signal and therefore, 19F MR imaging visualizes only the signal of administered perfluorocarbon emulsion. The in vivo accumulation of perfluorocarbon emulsion can be imaged by 19F MRI and overlayed on a simultaneously acquired 1H MR image, which shows the anatomical context in clear detail. Since this is advantageous compared to contrast agent based MR methods like iron oxide particle-based MRI, further experiments were performed with perfluorocarbon emulsions and 19F MRI. Experimental studies to elucidate the accumulation of perfluorocarbon emulsion at the site of infection showed robust 19F MR signals after administration between day 2 and at least day 8 p.i.. Perfluorocarbon emulsion accumulated in all investigated mice in the shape of a ‘hollow sphere’ at the rim of the abscess area and the signal remained stable as long as the infection prevailed. In order to identify the mechanism of accumulation, flow cytometry, cell sorting and histology studies were performed. Flow cytometry and cell sorting analysis of immune cells at the site of infection showed that neutrophils, monocytes, macrophages and dendritic cells carried contrast media at the site of infection with neutrophils accounting for the overwhelming portion of perfluorocarbon signal. In general, most of the signal was associated with immune cells, thus indicating specific immune cell dependent accumulation. Histology supported this observation since perfluorocarbon emulsion related fluorescence could only be visualized in close proximity to immune cell nuclei. After establishing and testing of 19F MRI with perfluorocarbon emulsions as infection imaging modality, the effects of antibiotic therapy upon MR signal was investigated in order to evaluate the capability of this modality for preclinical testing procedure. Thus, the efficacy of vancomycin and linezolid, two clinically highly relevant anti - S. aureus compounds, were tested in the murine thigh infection model. Both of them showed reduction of the colony forming units and bioluminescence signal, but also of perfluorocarbon emulsion accumulation strength and volume at the site of infection, which was visualized and quantified by 19F MRI. The efficacy pattern with linezolid being more efficient in clearing bacterial infection was shown similarly by all three methods. In consequence, 19F MRI with perfluorocarbon emulsion as MR tracer proved to be capable to visualize antibacterial therapy in preclinical testing models. The next step was consequently to evaluate a promising new compound against S. aureus infections. Thus, lysostaphin, an endo-peptidase that cleaves the cell wall of S. aureus, was tested in different concentrations alone or in combination with oxacillin for efficacy in murine thigh and catheter associated infection models. Lysostaphin only in the concentration of 5 mg/kg body weight or combined with oxacillin in the concentration of 2 mg/kg showed strong reduction of bacterial burden by colony forming unit determination and bioluminescence imaging in both models. The perfluorocarbon accumulation was investigated in the thigh infection model by 19F MRI and was strongly reduced in terms of volume and signal strength in both above-mentioned groups. In general, lysostaphin showed comparable or superior efficacy than vancomycin or oxacillin alone. Therefore, further development of lysostaphin for the treatment of S. aureus infections is recommended by these experiments. Overall, the antibiotic efficacy pattern of all applied antibiotic regimens was similar with all three applied methods, demonstrating the usefulness of MRI for antibiotic efficacy testing. Importantly, treatment with oxacillin either alone or in combination with lysostaphin resulted in stronger perfluorocarbon emulsion accumulation at the site of infection than expected compared to the results from bioluminescence imaging and colony forming unit determination. This might be an indication for immunomodulatory properties of oxacillin. Further murine infection experiments demonstrated in this context a differential release of cytokine and chemokines in the infected thigh muscle in dependence of the applied antibacterial therapy. Especially treatment with oxacillin, but to a less degree with minocycline or linezolid, too, exhibited high levels of various cytokines and chemokines, although they reduced the bacterial burden efficiently. In consequence, possible immunomodulatory effects of antibacterial compounds have to be taken into account for future applications of imaging platforms relying on the visualization of the immune response. However, this observation opens a new field for these imaging modalities since it might be extraordinary interesting to study the immunomodulatory effects of compounds or even bacterial factors in vivo. And finally, a two modality imaging platform which combines methods to visualize on the one hand the bacterial burden and on the other hand the immune response offers an innovative, new platform to study host-pathogen interaction in vivo in a non-invasive fashion. In summary, it could be shown that perfluorocarbon emulsions accumulate in immune cells at the site of infection in the murine S. aureus thigh infection model. The accumulation pattern shapes a ‘hollow sphere’ at the rim of the abscess area and its size and perfluorocarbon content is dependent on the severity of disease and/or efficacy of antibiotic therapy. Thus, 19F MRI with perfluorocarbon emulsions is a useful imaging modality to visualize sites and course of infection as well as to evaluate promising antibacterial drug candidates. Furthermore, since the accumulation of tracer depends on immune cells, it might be additionally interesting for studies regarding the immune response to infections, auto-immune diseases or cancer, but also to investigate the efficacy of immunomodulatory compounds and immunization.

Staphylococcus aureus is both a human commensal and a pathogen. 20%-30% of all individuals are permanently or occasionally carriers of S. aureus without any symptoms. In contrast to this, S. aureus can cause life-threatening diseases e.g. endocarditis, osteomyelitis or sepsis. Here, the increase in antibiotic resistances makes it more and more difficult to treat these infections and hence the number of fatalities rises constantly. Since the pharmaceutical industry has no fundamentally new antibiotics in their pipeline, it is essential to better understand the interplay between S. aureus and the human host cell in order to find new, innovative treatment options. In this study, a RNA interference based whole genome pool screen was performed to identify human proteins, which play a role during S. aureus infections. Since 1,600 invasion and 2,271 cell death linked factors were enriched at least 2 fold, the big challenge was to filter out the important ones. Here, a STRING pathway analysis proved to be the best option. Subsequently, the identified hits were validated with the help of inhibitors and a second, individualised small interfering RNA-based screen. In the course of this work two important steps were identified, that are critical for host cell death: the first is bacterial invasion, the second phagosomal escape. The second step is obligatory for intracellular bacterial replication and subsequent host cell death. Invasion in turn is determining for all following events. Accordingly, the effect of the identified factors towards these two crucial steps was determined. Under screening conditions, escape was indirectly measured via intracellular replication. Three inhibitors (JNKII, Methyl-beta-cyclodeytrin, 9-Phenantrol) could be identified for the invasion process. In addition, siRNAs targeted against 16 different genes (including CAPN2, CAPN4 and PIK3CG), could significantly reduce bacterial invasion. Seven siRNAs (FPR2, CAPN4, JUN, LYN, HRAS, AKT1, ITGAM) were able to inhibit intracellular replication significantly. Further studies showed that the IP3 receptor inhibitor 2-APB, the calpain inhibitor calpeptin and the proteasome inhibitor MG-132 are able to prevent phagosomal escape and as a consequence intracellular replication and host cell death. In this context the role of calpains, calcium, the proteasome and the mitochondrial membrane potential was further investigated in cell culture. Here, an antagonistic behaviour of calpain 1 and 2 during bacterial invasion was observed. Intracellular calcium signalling plays a major role, since its inhibition protects host cells from death. Beside this, the loss of mitochondrial membrane potential is characteristic for S. aureus infection but not responsible for host cell death. The reduction of membrane potential can be significantly diminished by the inhibition of the mitochondrial Na+/Ca2+ exchanger. All together, this work shows that human host cells massively contribute to different steps in S. aureus infection rather than being simply killed by bacterial pore-forming toxins. Various individual host cell factors were identified, which contribute either to invasion or to phagosomal escape and therefore to S. aureus induced cytotoxicity. Finally, several inhibitors of S. aureus infection were identified. One of them, 2-APB, was already tested in a sepsis mouse model and reduced bacterial load of kidneys. Thus, this study shows valuable evidence for novel treatment options against S. aureus infections, based on the manipulation of host cell signalling cascades.

The Gram-positive bacterium Staphylococcus aureus is the leading cause of nosocomial infections. In particular, diseases caused by methicillin-resistant S. aureus (MRSA) are associated with higher morbidity, mortality and medical costs due to showing resistance to several classes of established antibiotics and their ability to develop resistance mechanisms against new antibiotics rapidly. Therefore, strategies based on immunotherapy approaches have the potential to close the gap for an efficient treatment of MRSA. In this thesis, a humanized antibody specific for the immunodominant staphylococcal antigen A (IsaA) was generated and thoroughly characterized as potential candidate for an antibody based therapy. A murine monoclonal antibody was selected for humanization based on its binding characteristics and the ability of efficient staphylococcal killing in mouse infection models. The murine antibody was humanized by CDR grafting and mouse and humanized scFv as well as scFv-Fc fragments were constructed for comparative binding studies to analyse the successful humanization. After these studies, the full antibody with the complete Fc region was constructed as isotype IgG1, IgG2 and IgG4, respectively to assess effector functions, including antibody-dependent killing of S. aureus. The biological activity of the humanized antibody designated hUK-66 was analysed in vitro with purified human PMNs and whole blood samples taken from healthy donors and patients at high risk of S. aureus infections, such as those with diabetes, end-stage renal disease, or artery occlusive disease (AOD). Results of the in vitro studies show, that hUK-66 was effective in antibody-dependent killing of S. aureus in blood from both healthy controls and patients vulnerable to S. aureus infections. Moreover, the biological activity of hUK-66 and hUK-66 combined with a humanized anti-alpha-toxin antibody (hUK-tox) was investigated in vivo using a mouse pneumonia model. The in vivo results revealed the therapeutic efficacy of hUK-66 and the antibody combination of hUK-66 and hUK-tox to prevent staphylococcal induced pneumonia in a prophylactic set up. Based on the experimental data, hUK-66 represents a promising candidate for an antibody-based therapy against antibiotic resistant MRSA.

High-throughput sequencing (HTS) has revolutionized bacterial genomics. Its unparalleled sensitivity has opened the door to analyzing bacterial evolution and population genomics, dispersion of mobile genetic elements (MGEs), and within-host adaptation of pathogens, such as Escherichia coli. One of the defining characteristics of intestinal pathogenic E. coli (IPEC) pathotypes is a specific repertoire of virulence factors (VFs). Many of these IPEC VFs are used as typing markers in public health laboratories to monitor outbreaks and guide treatment options. Instead, extraintestinal pathogenic E. coli (ExPEC) isolates are genotypically diverse and harbor a varied set of VFs -- the majority of which also function as fitness factors (FFs) for gastrointestinal colonization. The aim of this thesis was the genomic characterization of pathogenic and commensal E. coli with respect to their virulence- and antibiotic resistance-associated gene content as well as phylogenetic background. In order to conduct the comparative analyses, I created a database of E. coli VFs, ecoli_VF_collection, with a focus on ExPEC virulence-associated proteins (Leimbach, 2016b). Furthermore, I wrote a suite of scripts and pipelines, bac-genomics-scripts, that are useful for bacterial genomics (Leimbach, 2016a). This compilation includes tools for assembly and annotation as well as comparative genomics analyses, like multi-locus sequence typing (MLST), assignment of Clusters of Orthologous Groups (COG) categories, searching for protein homologs, detection of genomic regions of difference (RODs), and calculating pan-genome-wide association statistics. Using these tools we were able to determine the prevalence of 18 autotransporters (ATs) in a large, phylogenetically heterogeneous strain panel and demonstrate that many AT proteins are not associated with E. coli pathotypes. According to multivariate analyses and statistics the distribution of AT variants is instead significantly dependent on phylogenetic lineages. As a consequence, ATs are not suitable to serve as pathotype markers (Zude et al., 2014). During the German Shiga toxin-producing E. coli (STEC) outbreak in 2011, the largest to date, we were one of the teams capable of analyzing the genomic features of two isolates. Based on MLST and detection of orthologous proteins to known E. coli reference genomes the close phylogenetic relationship and overall genome similarity to enteroaggregative E. coli (EAEC) 55989 was revealed. In particular, we identified VFs of both STEC and EAEC pathotypes, most importantly the prophage-encoded Shiga toxin (Stx) and the pAA-type plasmid harboring aggregative adherence fimbriae. As a result, we could show that the epidemic was caused by an unusual hybrid pathotype of the O104:H4 serotype. Moreover, we detected the basis of the antibiotic multi-resistant phenotype on an extended-spectrum beta-lactamase (ESBL) plasmid through comparisons to reference plasmids. With this information we proposed an evolutionary horizontal gene transfer (HGT) model for the possible emergence of the pathogen (Brzuszkiewicz et al., 2011). Similarly to ExPEC, E. coli isolates of bovine mastitis are genotypically and phenotypically highly diverse and many studies struggled to determine a positive association of putative VFs. Instead the general E. coli pathogen-associated molecular pattern (PAMP), lipopolysaccharide (LPS), is implicated as a deciding factor for intramammary inflammation. Nevertheless, a mammary pathogenic E. coli (MPEC) pathotype was proposed presumably encompassing strains more adapted to elicit bovine mastitis with virulence traits differentiating them from commensals. We sequenced eight E. coli isolates from udder serous exudate and six fecal commensals (Leimbach et al., 2016). Two mastitis isolate genomes were closed to a finished-grade quality (Leimbach et al., 2015). The genomic sequence of mastitis-associated E. coli (MAEC) strain 1303 was used to elucidate the biosynthesis gene cluster of its O70 LPS O-antigen. We analyzed the phylogenetic genealogy of our strain panel plus eleven bovine-associated E. coli reference strains and found that commensal or MAEC could not be unambiguously allocated to specific phylogroups within a core genome tree of reference E. coli. A thorough gene content analysis could not identify functional convergence of either commensal or MAEC, instead both have only very few gene families enriched in either pathotype. Most importantly, gene content and ecoli_VF_collection analyses showed that no virulence determinants are significantly associated with MAEC in comparison to bovine fecal commensals, disproving the MPEC hypothesis. The genetic repertoire of bovine-associated E. coli, again, is dominated by phylogenetic background. This is also mostly the case for large virulence-associated E. coli gene cluster previously associated with mastitis. Correspondingly, MAEC are facultative and opportunistic pathogens recruited from the bovine commensal gastrointestinal microbiota (Leimbach et al., 2017). Thus, E. coli mastitis should be prevented rather than treated, as antibiotics and vaccines have not proven effective. Although traditional E. coli pathotypes serve a purpose for diagnostics and treatment, it is clear that the current typing system is an oversimplification of E. coli's genomic plasticity. Whole genome sequencing (WGS) revealed many nuances of pathogenic E. coli, including emerging hybrid or heteropathogenic pathotypes. Diagnostic and public health microbiology need to embrace the future by implementing HTS techniques to target patient care and infection control more efficiently.

Shiga toxin producing E. coli strains (STEC) are a great concern to human health. Upon an infection with as few as 100 bacteria, humans can develop disease symptoms ranging from watery to bloody diarrhea or even develop the hemolytic uremic syndrome (HUS). The major factor contributing to the disease symptoms is Shiga toxin (Stx) which can bind to the eukaryotic cells in the intestine of the human and induce cell death via apoptosis. Based, among other things, on the microbiota composition, the impact of STEC can vary. Some bacteria of the microbiota can interfere with the colonization of STEC strains in the first place. Others cannot impair the colonization but interfere with the toxin production and there are still others which are even infected by stx encoding phages, being released from STEC strains. Those previously harmless bacteria subsequently contribute to the toxin increase and worsen the disease progression. Since the genetic information of Stx is encoded on a prophage, antibiotic treatment of patients can lead to an increased toxin and stx-phage release and is therefore not recommended. Several STEC epidemics in different countries, which even resulted in the death of some patients, demonstrated that there is an urgent need for alternative treatment strategies. The E. coli strain Nissle 1917 (EcN) has been used as a probiotic to treat gastrointestinal infections for more than 100 years. It harbors several fitness factors which contribute to the establishment of an intact intestinal barrier in the human gut. Moreover, studies with EcN unraveled that the probiotic E. coli can interfere with the colonization of STEC strains and their toxin production. This study aimed to investigate if EcN could be a possible alternative or supplementary treatment strategy for STEC infected patients, or a preventive treatment for the patient’s close contact persons. Therefore, EcN was firstly investigated for a possible stx-prophage integration into its’s genome which would eliminate it from being a potential treatment due to the possibility of disease worsening. Despite the presence of the stx-phage surface receptor YaeT, EcN demonstrated a complete resistance towards the lysis and the lysogeny by stx-phages, which was proven by PCR, phage-plaque assays and phage enrichment approaches. Transcriptome data could unravel that a lambdoid prophage in the genome of EcN is involved in the resistance towards the phage infection. Other commensal E. coli tested presented a stx-phage resistance as well and in silico analysis revealed that all of them harbor a complete lambdoid prophage besides the stx-phage susceptible K-12 strain MG1655. We assume that the resistance of EcN towards a stx-phage infection is connected to the presence of an intact lambdoid prophage which interferes with superinfection. Further experiments regarding the impact of the microcin negative EcN mutant SK22D towards STEC strains depicted that SK22D did not only interfere with the toxin production but also negatively regulated the transcription of the entire stx-prophage in coculture with all STEC strains tested (O157:H7, O26:H11, O145:H25, O103:H2, O111:H- and two O104:H4 isolates from the 2011 outbreak in Germany). This influence on the pathogenic factor production was evinced to be cell contact independent as SK22D could even interfere with the pathogenic factor production when being separated from the STEC strain EDL933 by a Transwell membrane with the pore size of 0.4 µm. From this data we concluded, that factor(s) released by SK22D interfere with the lysis of STEC strains by stabilizing the lysogenic state. Another positive aspect of EcN towards the pathogenicity of STEC strains was encountered when EcN was incubated with isolated stx-phages. The probiotic strain could reduce the infectivity of the phages towards a MG1655 lysis from ~ 1e7 pfus/ml to 0 after 44 h of incubation. Various approaches to determine the characteristics of the factor(s) of EcN which are involved in the phage inactivation depicted it to be a heat resistant stationary phase protein on the surface of EcN, which could be a component of its biofilm. Regarding the protective role of EcN we could further evince that SK22D was capable of interfering with the lysogenic K 12 mediated increase of Stx and stx phages. Lysogenic K-12 strains were characterized by a huge increase of Stx and stx-phage production. The presence of SK22D anyhow, could interfere with this K-12 mediated pathogenic factor increase. Transwell and stx phage infection kinetics led to the proposal that SK22D interfered with the stx-phage infection of K-12 strains in the first place rather than disturbing the lysis of lysogenic K 12. The protection from the phage infection could be due to the growth of K 12 strains within the SK22D culture, whereby the phage susceptible strains are masked from phage detection. Summarizing, this work could underline the beneficial attributes of EcN towards the STEC pathogenicity in vitro. These results should be considered as pioneers for future in vivo studies to enable EcN medication as a supportive STEC infection treatment strategy.