Open Access

Herpesviruses are enveloped DNA viruses which are dependent on two fusion steps for efficient replication in the host cell. First, they have to fuse their envelope with the cellular plasma membrane or with the vesicle membrane after endocytic uptake to enter the host cell and second, they have to export the newly generated nucleocapsids from the site of assembly to the cytoplasm by fusion of the primary virion envelope with the outer nuclear membrane (ONM). The main goal of this project was to provide a better understanding of how herpesvirus capsids exit the nucleus. On the one hand this thesis aimed at finding cellular proteins involved in nuclear egress (Paper I), while on the other the focus was on further characterization of the viral nuclear egress complex (NEC, Paper II) and its interaction with the capsid (Paper III). It is the hallmark of viruses, including herpesviruses, to hijack host cell proteins for their efficient replication. Some of those interactions are well characterized, while others might not yet have been discovered. In the last step of the nuclear egress, where the primary virion membrane fuses with the ONM, most likely a cellular machinery is involved. The presented work focused on Torsin, the only known AAA+ ATPase localizing in the endoplasmic reticulum and the perinuclear space (PNS). For this, the effect of overexpression of WT and mutant proteins, as well as CRISPR/Cas9 generated knock-out cell lines, on PrV replication was analyzed. Neither single overexpression nor single knockouts of TorA or TorB had any significant effects on virus titers. However, infection of TorA/B double knockout cells revealed reduced viral titers and an accumulation of primary virions in the PNS at early infection times, indicating a delay in nuclear egress. The process of nuclear egress has been intensively investigated without revealing all its details. To address some of the missing aspects we generated monoclonal antibodies (mAbs) against the NEC and its components (pUL31 and pUL34) for a better visualization of the process in transfected as well as infected cells. These mAbs provide a useful tool for future analyses. The publication of the NEC crystal structure formed the basis for intensive research on the molecular details of the NEC formation and its interaction with the nucleocapsid. Recently, our lab showed that lysine (K) at position 242 in the membrane-distal part of pUL31 is crucial for incorporation of the nucleocapsid into budding vesicles. Replacing K by alanine (A) resulted in accumulations of vesicles in the PNS, while mature capsids were not incorporated. To test whether this is due to electrostatic interference or structural restrictions we substituted K242 by different aa to determine the requirements for nucleocapsid uptake into the nascent primary particles. To analyze whether the defect of pUL31-K242A can be compensated by second-site mutations, PrV-UL31-K242A was passaged and mutations in revertants were analyzed. Different mutations have been identified compensating for the K242A defect. A considerable number of mutations indicates that the NEC is much more flexible than previously thought. Further, we gained information that the K at position 242 is not directly involved in capsid interaction, while it is more likely involved in rearrangements within the NEC coat.

To enable control of African swine fever (ASF) in Eastern and Southern Africa, prototype live vaccine candidates were generated by targeted gene deletions from a Kenyan genotype IX ASF virus (ASFV). It was attempted to delete known nonessential genes involved in virulence (encoding TK, dUTPase, CD2v, 9GL), possibly essential genes (p12, pA104R, ribonucleotide reductase), and genes with widely unknown functions (pK145R). Isolation of the desired virus recombinants by plaque assays or limiting dilutions on a wild boar lung cell line (WSL-HP) was facilitated by substitutive reporter gene insertions encoding fluorescent proteins (GFP, DsRed), or the human membrane protein CD4. The latter protein permitted enrichment of recombinant virus particles by magnetic activated cell sorting (MACS). The isolated ASFV recombinants were characterized by PCR and sequencing of the mutated genome parts, and replication kinetics and virus spread in cell culture were investigated. Deletion of TK, CD2v, or pK145R had no detectable effect on in vitro growth of ASFV Kenya. Interestingly, virus mutants lacking the DNA binding protein pA104R which has been considered to be essential for DNA replication, also exhibited almost wild type-like growth properties. In contrast, ASFV mutants lacking ribonucleotide reductase or p12 could not be purified to homogeneity on WSL-HP cells, indicating these proteins are essential for virus replication in cell culture. Therefore, trans-complementing cells lines stably expressing ASFV p12 have been prepared which can now be used for mutant virus purification. If this approach is successful the resulting defective mutant ASFV Kenya-p12 might be suitable as a safe “disabled in second cycle” (DISC) live vaccine in swine. In a novel approach to improve reverse genetics of ASFV the CRISPR/Cas9 cell line WSL-gRp30 (Hübner et al., 2018a) was co-transfected with genomic DNA of ASFV-KenyaCD2vDsRed, sgRNA plasmids targeting K145R or 9GL, and GFP-expressing recombination plasmids for homology-directed repair. For booting up of the noninfectious virus genome the cells were infected with phylogenetically distant helper virus (genotype II ASFV Armenia, 84% identity) which was selectively inhibited on the used cell line. The desired double-fluorescent double-deletion mutants could be isolated after few plaque purification steps on selective WSL-gRp30 cells. Next generation sequence (NGS) analyses of reconstituted ASFV Kenya genomes showed that no unwanted recombination with the helper virus occurred, indicating that the method might be also suitable for booting of synthetic ASFV genomes cloned and mutagenized in E. coli or yeast. The modified CRISPR/Cas9 system of S. pyogenes might be also usable for generation of ASFV resistant pigs. To evaluate this alternative control measure WSL cell clones stably expressing Cas9 nuclease and single or multiple sgRNAs against essential ASFV proteins were prepared and tested for their susceptibility to infection. Strain specific sgRNAs targeting the p30 gene of ASFV Kenya or Armenia selectively inhibited the respective viruses, and a p12 gene-specific sgRNA abrogated replication of both genotypes almost completely. Interestingly, coexpression of four ASFV-specific sgRNAs did not enhance virus inhibition, but might help to reduce the frequency of escape mutants which were occasionally isolated from the single sgRNA-expressing cells, and exhibited silent base substitutions or in-frame deletions within the target genes. First attempts to express the in vitro tested CRISPR/Cas9 constructs in transgenic pigs are in progress. CRISPR/Cas9 supported rescue of a defective BAC clone of pseudorabies virus (PrV) vaccine strain Bartha (Hübner et al., 2018b) was used to develop putative vectored vaccines against ASFV. In the present study expression cassettes for the codon-optimized p12 and p54 genes of ASFV were successfully inserted into the PrV genome. The insertions did not significantly affect PrV recombination in cell culture, and the transgenes were expressed at similar levels as in ASFV-infected cells. It has to be tested whether coinfection with vector constructs for these and other immunogenic ASFV proteins is able to protect pigs against a lethal challenge. For characterization of the generated ASFV mutants and PrV vector constructs, monospecific antisera against several ASFV gene products (p11.5, p12, p54, pK145R, p285L) were prepared by immunization of rabbits with bacterial GST fusion proteins. The anti-p12 serum showed only weak and strain-specific reactions with the ASFV Kenya protein, but was nevertheless useful for identification of p12-expressing PrV recombinants and WSL cell lines. All other sera showed satisfying reactions in Western blot and mostly immunofluorescence analyses, and allowed i.a. precise localization of the pK145R and p285L proteins in ASFV-infected cells and virions (Hübner et al., 2019).

Herpesviruses are a fascinating group of enveloped DNA viruses, which rely on membrane fusion for infectious entry and direct cell-to-cell spread. Compared with many other enveloped viruses, they utilize a remarkably complex fusion machinery. Three conserved virion proteins, the bona fide fusion protein gB, and the presumably gB activating gH/gL heterodimer constitute the conserved core fusion machinery and are believed to drive membrane fusion in a cascade-like fashion. Activation of this cascade in most alphaherpesviruses is proposed to be triggered by binding of gD to specific host cell receptors. The molecular details of this fusion process, however, remain largely elusive. Yet, a detailed mechanistic knowledge of this process would be greatly beneficial for the development of efficient countermeasures against a variety of diseases. In this thesis, the functional relevance of individual components of the essential gH/gL complex of the alphaherpesvirus PrV has been assessed by two different approaches: by reversion analysis (paper II) and site-directed mutagenesis (papers III-V). In contrast to other herpesviruses, gL-deleted PrV is able to perform limited cell-to-cell spread, providing the unique opportunity to passage the entry-deficient virus in cell culture to select for PrV revertants capable of infecting cells gL-independently. This approach already resulted in an infectious gL-negative PrV mutant (PrV-ΔgLPass), in which the function of gL was compensated by formation of a gDgH hybrid protein. Here, the requirements for gL-independent infectivity of a second independent revertant (PrV-ΔgLPassB4.1), were analyzed. Sequencing of the genes encoding for gB, gH and gD, revealed mutations in each of them. By means of a robust infection-free, transfection-based cell-cell fusion assay (paper I), we identified two amino acid substitutions in the gL-binding domain I of gHB4.1 (L70P, W103R) as sufficient to compensate for lack of gL. Two mutations in gB (G672R, ΔK883) were found to enhance fusogenicity, probably by lowering the energy, required for gB refolding from pre- to postfusion conformation. Coexpression of gHB4.1 and gBB4.1 led to an excess fusion, which was completely suppressed by gDB4.1 in the fusion assays. This was surprising since PrV gD is normally not required for in vitro fusion or direct viral cell-to-cell spread, clearly separating this process from fusion during entry, for which PrV gD is essential. The fusion inhibiting effect of gDB4.1 could be attributed to a single point mutation resulting in an amino acid substitution within the ectodomain (A106V). In conclusion, these results indicated that gL is not central to the fusion process, as its function can be compensated for. As found so far, gL-independent infectivity can be realized by compensatory mutations in gH (as in PrV-ΔgLPass) or in gH plus gB (as in PrV-ΔgLPassB4.1). Excessive fusion induced by gHB4.1 and gBB4.1 was counter-regulated by gDB4.1, indicating that the interplay between these proteins is precisely regulated and further implies that gL and gD, despite being not absolutely essential for the fusion process, have important regulatory functions on gH and/or gB. Both PrV-ΔgLPass mutants had acquired compensatory mutations in gH affecting the predicted gL-binding domain I in gH. By construction of an artificial gH32/98, which lacked the predicted gL-binding domain and was similar to the recently crystallized gH-core fragment present in the gDgH hybrid protein, we identified the N-terminal part of PrV gH as essential for gH function during fusion (paper III). gH32/98 was unable to promote fusion of wild-type gB in fusion assays and led to a total loss of function in the viral context. These results indicated that the gD moiety, present in gDgH, is critical for proper function of the gH-core fragment. We hypothesize that the gD moiety may adopt a stabilizing or modulating influence on the gH structure, which is normally executed by gL and important for interaction of gH with wild-type gB. Remarkably, substitution of wild-type gB by gBB4.1 rescued function of gH32/98 in the cellular and viral contexts. These findings suggest that gBB4.1 has been selected for interaction with “gL-less” gH. In conclusion, these results demonstrated that gL and the gL-binding domain are not strictly required for membrane fusion during virus entry and spread but that compensatory mutations must be present in gB to restore a fully functional fusion machinery. These results strongly support the notion of a functional gH-gB interaction as a prerequisite for membrane fusion. In addition to the N-terminal domain, we identified the transmembrane domain of PrV gH as an essential component of the fusion machinery, while the cytoplasmic domain was demonstrated to play a modulatory but nonessential role (paper IV). Whereas truncation or substitution of the PrV gH TMD by a gpi-anchor or the analogous sequence from PrV gD rendered gH non-functional, the HSV-1 gH TMD was found to functionally substitute for the PrV gH TMD in cell-cell fusion and complementation assays. Since residues in the TMD which are conserved between HSV and PrV gH but absent in PrV gD, are placed on one face of an α-helical wheel plot, we hypothesize that the gH TMD has an intrinsic property to interact with membrane components such as lipids or other molecules as a requirement for promoting membrane fusion. In a final study focusing on the function of gH, we identified the N-glycosylation sites utilized by PrV gH, and determined their individual role in viral infection (paper V). PrV gH was found to be modified by N-glycans at five potential glycosylation sites. N-glycans at PrV specific N77 and the highly conserved site N627 were found to be critical for efficient membrane fusion in the fusion assays, and during viral entry and cell-to-cell spread. N627 was further shown to be crucial for proper gH transport and maturation. In contrast, inactivation of N604, conserved in the Varicellovirus genus, enhanced in vitro fusion activity and viral cell-to-cell spread. These findings demonstrated a role of the N-glycans in proper localization and function of PrV gH.

Avian influenza viruses (AIVs) have their natural reservoir in wild aquatic birds but occasionally spread to terrestrial poultry. While AIVs of subtypes H5 and H7 are well known to evolve highly pathogenic avian influenza viruses (HPAIVs) during circulation in domestic birds, non-H5/H7 subtypes exhibit only a low to moderate pathogenicity. Furthermore, spillover events to a broad range of mammalian hosts, including humans, with self-limiting to severe illness or even fatal outcomes, were reported for non-H5/H7 AIVs and pose a pandemic risk. The evolution of high virulent phenotypes in poultry and the adaptation of AIVs to mammalian hosts are predominantly linked to genetic determinants in the hemagglutinin (HA). The acquisition of a polybasic cleavage site (pCS) is a prerequisite for the evolution of HPAIVs in poultry, while changes in the receptor binding preference and virus stability are essential for adaptation of AIVs to mammals. In August 2012, an H4N2 virus with the pCS motif 322PEKRRTR/G329 but preserved trypsin dependend replication and low pathogenicity in chickens was isolated on a quail farm in California. In the first two publications, we followed different approaches to investigate virulence factors and the potential risk for the transition of H4N2 to high virulence in chickens. The loss of N-terminal glycosylations in the vicinity of the pCS resulted in decreased binding to avian-like receptors and dramatically decreased virus stability. On the other hand, one deglycosylation increased virus replication and tissue tropism in chicken embryos but did not alter virulence or excretion in chickens. Furthermore, additional basic amino acids in the natural pCS motif improved the trypsin-independent cleavage of HA and caused slightly increased tissue tropism in chickens. However, the engineered motifs alone did not affect virulence in chickens. Intriguingly, they even had a detrimental effect on virus fitness, which was restored after reassortment with segments of HPAIV H5N1. Together, the results show the importance of HA glycosylations on the stability of H4N2 and reveal the important role of non-HA segments in the transition of this virus to high virulence in poultry. The transmission of another non-H5/H7 AIV of subtype H10N7 from birds to seals resulted in mass deaths in harbor seals in 2014 in northern Europe. The third publication describes nine mutations in the HA1 subunit of seal isolates compared to avian H10Nx viruses. We found that some of these mutations conferred a dual specificity for avian and mammalian receptors and altered thermostability. Nevertheless, the H10N7seal remained more adapted to avian host cells, despite of the alteration in the receptor binding specificity. Altogether, this thesis demonstrates that naturally evolved AIVs beside H5 and H7 subtypes support a highly pathogenic phenotype in the appropriate viral background and alter virulence and host receptor specificity by few amino acid substitutions in the HA. These findings improve our knowledge of the potential of non-H5/H7 AIVs to shift to high virulence in birds and the adaptation in mammals.

Die Afrikanische Schweinepest (ASP) ist eine Viruserkrankung, die Mitglieder der Suidae-Familie wie Buschschweine, Warzenschweine, Hausschweine und Wildschweine befällt. Das Virus wird durch direkten Kontakt zwischen infizierten und naiven Tieren, durch Zecken der Gattung Ornithodoros oder durch Kontakt mit kontaminiertem Material übertragen. Während die Krankheit bei Warzenschweinen und Buschschweinen im Allgemeinen asymptomatisch verläuft, verursacht die ASP eine hohe Mortalität bei Hausschweinen und Wildschweinen. Daher ist die jüngste Ausbreitung von ASP in Europa eine ernste Bedrohung für die Schweinehaltung in der EU. Bis heute ist keine wirksame Behandlung oder Impfung verfügbar. Und es liegen nur wenige Informationen über Virus-Wirt-Wechselwirkungen vor, die als Grundlage für die Etablierung antiviraler Strategien verwendet werden könnten. Das Virus der afrikanischen Schweinepest (African swine fever virus, ASFV) ist der einzige bekannte Vertreter der Familie der Asfarviridae. Das DNA-Genom des ASFV kodiert für über 150 Gene. Über die Expressionsprodukte ist wenig bekannt, nur wenige virale Proteine sind bisher funktionell charakterisiert. Die Morphogenese von ASFV ist sehr komplex. So entstehen neben den zweifach umhüllten reifen extrazellulären Virionen auch einfach umhüllte intrazelluläre Partikel, die die die Präparation reiner extrazellulären Virionen erschweren. In früheren in vitro Studien wurde die Zusammensetzung der extrazellulären Viruspartikel mittels 2D-Gelelektrophorese analysiert. Die Reinigung erfolgte über ein im Jahre 1985 veröffentlichtes Reinigungsprotokoll, welches auf einer Percolldichtegradientenzentrifugation und einer Gelchromatographie basierte. Das Protokoll wurde für die Reinigung des auf Vero-zellen adaptierten Virusstamm Ba-71V etabliert. In einer frühen MS Studie wurden 54 Proteine in ASFV Partikeln detektiert, 15 davon Wirtsproteine. Der Einbau von Aktin, α-Tubulin und β-Tubulin ins Virion konnte ebenfalls bestätigt werden. Systematische massenspektrometrische Untersuchungen zur Charakterisierung des Proteoms der ASF Virionen lagen zu Beginn der vorliegenden Dissertation nicht vor, erst während der Anfertigung des Manuskripts wurde eine solche Studie durch Alejo et al. veröffentlicht. Im Rahmen dieser Arbeit wurde ein auf einer Dichtegradientenzentrifugation ohne nachfolgende Gelchromatographie beruhendes Reinigungsprotokoll entwickelt und die Zusammensetzung reifer ASF Viruspartikel mittels MALDI-TOF/TOF Massenspektrometrie analysiert. Zur Anzucht einer GFP-positiven ASFV OUR T88/3 Mutante wurde die vom Wildschwein abstammende Zelllinie WSL-HP verwendet. Wesentliche Schritte der Reinigung waren eine niedertourige Zentrifugation zur Entfernung zellulärer Verunreinigungen, gefolgt von einer Sedimentation des Virus durch ein Saccharosekissen und einem Proteaseverdau. Final wurde die Präparation über einen selbstgenerierenden Optiprep™ Dichtegradienten gereinigt. Die Titerausbeute lag zwischen 30 und 70 %, die spezifische Infektiosität bei 2,4 x 109 TCID50/mg. Elektronenmikroskopische Untersuchungen zeigten, dass die Präparation zwar Virionen enthielt, aber auch, dass die Fixierung mit Glutaraldehyd die Stabilität der Virionen beeinträchtigt. In der massenspektrometrischen Analyse wurden 29 der 33 bekannten ASFV Strukturproteine bestätigt. Von den neu identifizierten Strukturproteinen konnten vier (pK145R, pC129R, pE146L und pI73R) in allen drei Replikaten und sechs in zwei von drei Replikaten (p5, CP123L, CP312R, E184L, M1249L und M2248R) bestätigt werden. Ein weiteres bis dato nicht charakterisiertes Protein, p285L, konnte als mögliches neues Strukturprotein identifiziert werden. 152 Wirtsproteine wurden im Virion detektiert, darunter hauptsächlich Membranproteine oder Proteine des Zytoskeletts. Daneben wurde eine Reihe an phospholipidbindenden Proteine gefunden. Unter den identifizierten Proteinen waren fünf aus dem glatten ER und einige Vertreter der Hitzeschockproteine. Im zweiten Teil dieser Arbeit sollte das intrazelluläre Proteom des ASFV identifiziert werden. Für diese Untersuchungen wurden drei empfänglichen Zelllinien verwendet, die vom Wildschwein abstammenden Linie WSL-HP, Vero Zellen, die in der Vergangenheit für viele Studien herangezogen wurde und die menschliche Linie HEK-293, die aus einem weiteren nicht empfänglichen Wirt stammt. Der in dieser Studie verwendete Virusstamm ASFV OUR T88/3 besitzt 157 ORFs. In früheren Studien konnte die Existenz eines Proteins für 44 ORFs bestätigt werden. Für weitere 69 ORFs wurden Transkripte, nicht aber die korrespondierenden Proteine, beschrieben, sodass für 44 ORFs kein Nachweis der Expression vorlag. In der massenspektrometrischen Analyse wurden je Wirtszelle rund 1000 Proteine identifiziert. Insgesamt belief sich die Zahl der identifizierten ASFV Proteine auf 94, davon 88 in WSL-HP, 83 in Vero und 57 in HEK-293 Zellen. 54 ASFV Proteine wurden in allen drei Zelllinien detektiert. Für 34 der identifizierten ASFV Proteine war bisher nur die Existenz des Transkripts beschrieben, für 23 weitere weder die Existenz eines Proteins noch eines Transkripts. Für 44 der 94 identifizierten Proteine wurde das N-terminales Peptid detektiert. Bei fünf der MGF-110 Proteinen (1L, 2L, 4L, 5L und 14L) und den Proteinen pI329L und pCP123L wurde die Abspaltung der vorhergesagten Signalsequenz experimentell bestätigt. Die MS Analysen wurden unter Verwendung des emPAI auch quantitativ ausgewertet. Die geringe Zahl detektierter ASFV Proteine in HEK-293 Zellen korrelierte mit dem geringeren Anteil an ASFV Proteinen im Gesamtproteingehalt der Zelle (6,3 Mol%). Allerdings wurden einige Proteine in HEK-293 Zellen ähnlich stark oder sogar stärker exprimiert als in Vero bzw. WSL-HP Zellen. Die Abundanz einzelner ASFV Proteine variierte in den verschiedenen Zelllinien. Einige wurden jedoch durchgehend stark exprimiert wie z.B. das Strukturprotein p11.5. Einige bisher nicht charakterisierte Proteine, wie z.B. pK145R, pI73R und pC129R, wurden überraschenderweise ebenfalls in allen Zellen stark exprimiert und sind somit möglicherweise Träger wichtiger viraler Funktionen, die weiter untersucht werden sollten.