Open Access

Rabies virus (RABV) is an ancient, highly neurotropic rhabdovirus that causes lethal encephalitis. Most RABV pathogenesis determinants have been identified with laboratory-adapted or attenuated RABVs, but details of natural RABV pathogenesis and attenuation mechanisms are still poorly understood. To provide a deeper insight in the cellular mechanism of pathogenies of field RABV, this work was performed to assess virus strain specific differences in intra-neuronal virus transport, to identify cell culture adaptive mutations in recombinant field viruses and to explore shRNA-expressing RABVs as research tools for targeted host manipulation in infected cells. Comparison of chimeric RABVs with glycoprotein (G) ecto-domains of different lyssaviruses, together with field RABVs from dog and fox in dorsal root ganglion (DRG) neurons revealed no detectable differences in the axonal accumulation of the viruses. This indicates that previously described G-dependent transport of newly formed RABV in axons can occur both in laboratory-adapted and field RABV. Moreover, partial overlap of nucleoprotein (N) and G protein particles in field virus infected DRG axons supported the hypothesis of the “separate model” for anterograde RABV transport. Serial passages of recombinant dog and fox field clones in different cell lines led to the identification of general (D266N) and cell line specific (K444N) adaptive mutations in the G ecto-domain of both viruses. In BHK cells, synergistic effects of D226N, K444N and A417T on field dog virus G protein surface localization led to the loss of endoplasmic reticulum (ER) retention of G and increased virus titers in the supernatant, indicating that limited virus release by ER retention is a major bottleneck in cell culture adaptation. In addition, selection of mutations within the C-terminus of the RABV phosphoprotein (P) (R293H and R293C in fox and dog viruses, respectively) led to the hypothesis of altered binding affinities to nucleoprotein and RNP complexes. Identification of the above mentioned amino acid substitutions together with alterations in a suboptimal transcription stop signal in the P/M gene border indicated that adaptation to cell culture replication occurs on both levels, RNA transcription/replication and virus release. To evaluate the possibility of an expression of a functional microRNA-adapted short-hairpin RNAs (miR-shRNA) expressing RABV, recombinant RABVs encoding miR-shRNAs against cellular Dynein Light Chain 1 (DYNLL1) and Acidic Nuclear Phosphoprotein 32 family member B (ANP32B) were generated. In spite of cytoplasmic transcription of the respective mRNAs, downregulation of DYNLL1 and ANP32B mRNA and respective protein levels in infected cells revealed correct processing to functional shRNAs. Specific downregulation of the cellular genes at 2, 3 and 4 days post infection further demonstrated feasibility of the approach in standard cell lines. However, it remained open whether miR-shRNA expressing RABV can be used to study neuro-infection in vivo. Since first attempts in primary rat neuron cultures failed, it has to be clarified in further experiments whether this strategy can be used in mature, non-dividing neurons or whether breakdown of the nucleus in the course of cell division is a requirement for the processing of cytoplasmically expressed miR-RNA by nuclear RNases. By providing novel insights in axonal RABV transport and cell culture adaptive mutations this work extends the current understanding of RABV pathogenesis in natural and non-natural cell environments. Moreover, it provides a basis for further pathogenicity studies in which the impact of cell culture adaptation through increased virus release on RABV virulence can be investigated. With successful expression of functional miR-shRNAs from RABV vectors, this work also provides a tool for RABV gene targeting in infected cell lines and thus may contribute to the further investigation of RABV-host-cell-interactions.

Technological advances in light microscopy have always gone hand in hand with unprecedented biological insight. For microbiology, light microscopy even played a founding role in the conception of the entire discipline. The ability to observe pathogens that would otherwise evade human observation makes it a critical necessity and an indispensable tool to infectious disease research. Thus, the aim of this thesis was to optimize, extend, and functionally apply advanced light microscopy techniques to elucidate spatio-temporal and spatio-morphological components of bacterial and viral infection in vitro and in vivo. Pathogens are in a constant arms race with the host’s immune system. By finding ways to circumvent host-mediated immune responses, they try to evade elimination and facilitate their own propagation. The first study (publication I) demonstrated that the obligate intracellular pathogen Coxiella burnetii is not just able to infect natural killer (NK) cells, but is actually capable of surviving the harsh degradative conditions in the cytotoxic lymphocyte’s granules. Using live-cell imaging of reporter-expressing Coxiella burnetii, the transient NK cell passage was closely monitored to provide detailed spatio-temporal information on this dynamic process in support of a range of static analyses. Bacterial release from NK cells was pinpointed to a time frame between 24 to 48 hours post-infection and the duration of release to about 15 minutes. The second approach (publications II-V) aimed at shedding light on the greater spatio-morphological context of virus infection. Thus far, most studies investigating the distribution or tropism of viruses in vivo have used conventional immunohistochemistry in thin sections. Omitting the native spatial context of the infection site in vivo inherently bears the risk of incomplete description. While the microscopic tools and sample preparation protocols needed for volumetric 3D immunofluorescence imaging have recently been made available, they had not gained a foothold in virus research yet. An integral part of this thesis was concerned with the assessment and optimization of available tissue optical clearing protocols to develop an immunofluorescence-compatible 3D imaging pipeline for the investigation of virus infection inside its intact spatio-morphological environment (publication II). This formed the basis for all subsequent volumetric analyses of virus infection in vivo presented here. Consequently, this thesis provided a valuable proof of concept and blueprints for future virus research on the mesoscopic scale of host-pathogen interactions in vivo (publications II-V), using rabies virus (RABV; publications II-IV) and the newly-emerged severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2; publication V) as infection models for the nervous system and the respiratory tract, respectively. Applying and further improving this volumetric 3D imaging workflow enabled unprecedented insights into the comprehensive in vivo cell tropism of RABV in the central (CNS) (publication III) and peripheral nervous system (PNS) (publication IV). Accordingly, differential infection of CNS-resident astrocytes by pathogenic and lab-attenuated RABV was demonstrated (publication III). While either virus variant showed equal capacity to infect neurons, as demonstrated by quantitative image analysis, only pathogenic field RABVs were able to establish non-abortive infection of astrocytes via the natural intramuscular inoculation route. A combined 3D LSFM-CLSM workflow further identified peripheral Schwann cells as a relevant target cell population of pathogenic RABV in the PNS (publication IV). This suggested that non-abortive infection of central and peripheral neuroglia by pathogenic RABV impairs their immunomodulatory function and thus represents a key step in RABV pathogenesis, which may contribute significantly to the establishment of lethal rabies disease. Finally, utilizing the full volumetric acquisition power of LSFM, a further refined version of the established 3D imaging pipeline facilitated a detailed mesoscopic investigation of the distribution of SARS-CoV-2 in the respiratory tract of the ferret animal model (publication V). Particularly for this newly-emerged pathogen of global concern, in-depth knowledge of host-pathogen interactions is critical. By preserving the complete spatio-morphological context of virus infection in the ferret respiratory tract, this thesis provided the first specific 3D reconstruction of SARS-CoV-2 infection and the first report of 3D visualization of respiratory virus infection in nasal turbinates altogether. 3D object segmentation of SARS-CoV-2 infection in large tissue volumes identified and emphasized a distinct oligofocal infection pattern in the upper respiratory tract (URT) of ferrets. Furthermore, it corroborated a preferential replication of SARS-CoV-2 in the ferret URT, as only debris-associated virus antigen was detected in the lower respiratory tract of ferrets, thus providing crucial information on the spatial distribution of SARS-CoV-2.

Lyssaviruses, the causative agents of rabies, are a long-known threat for animals and humans. To date, terrestrial rabies still accounts for tens of thousands of human deaths annually, notwithstanding ambitious vaccination campaigns targeting susceptible dog and wildlife populations that act as reservoirs for the prototypic rabies virus. Moreover, the continuing discovery of newly emerging virus species in hitherto unconcerned chiropteran hosts and geographic regions drive the expansion of the Lyssavirus genus by unveiling its actual variety, host range and distribution.In this work, the genetic diversity of three distinct lyssaviruses, namely EBLV-1, KBLV and RABV, was elucidated by in-depth genomic analyses to provide further insight into lyssavirus evolution. The generation of full-genome sequences from primarily bat-associated Danish EBLV-1 samples significantly increased the number of available Danish EBLV-1 genome sequences while phylogenetic and phylogeographic analysis revealed a stronger phylogeographic structure for the cluster A1 of the sublineage EBLV-1a than it was postulated in previous studies. In addition, the acquisition of a nearly complete genome sequence for the Kotalahti bat lyssavirus provided the basis for the classification of this putative new lyssavirus species as a recognized member of the genus. Furthermore, phylogenetic analysis revealed the affiliation of KBLV to a group of Myotis-associated lyssaviruses giving a deeper insight into the shared evolutionary history of lyssaviruses co-evolving with particular bat species. Moreover, a deep-sequencing approach was utilized to assess the high genetic diversity of vaccine virus populations, uncovering three independent patterns of single nucleotide variants (SNVs) that became selected in ERA-related vaccine-induced cases. However, no apparent influence of the genetic diversity of vaccine viruses on microevolutionary processes like a potential reversion to virulence or a species-specific adaptation of the vaccine virus strains could be detected, leaving the question for the cause of rabies induction in the affected animals unanswered. Lastly, the successful implementation of a hybridization capturing system for the generation of full-genome sequences and deep-sequencing variant analyses of RABV and KBLV samples was demonstrated for a diagnostic bait set, highlighting the versatility and consistency of this approach to assess the genetic spectrum of known and novel lyssavirus species while setting the basis for its application and optimization in upcoming projects.In conclusion, as shown by the studies in this work, the investigation of lyssavirus genomes at the sub-consensus, full-genome and population level remains crucial to assess the complexity of lyssavirus evolution, as it provides an indispensable source of information to cover the diversity of the genus and understand evolutionary dynamics on a long-term and microevolutionary scale.