ACTAUNIVERSITATIS

UPSALIENSISUPPSALA

2019

Digital Comprehensive Summaries of Uppsala Dissertationsfrom the Faculty of Science and Technology 1847

Structural Studies of Large dsDNAViruses using Single ParticleMethods

HEMANTH KUMAR NARAYANA REDDY

ISSN 1651-6214ISBN 978-91-513-0732-9urn:nbn:se:uu:diva-391671

Dissertation presented at Uppsala University to be publicly examined in Room C2:301,BMC, Husargatan 3, Uppsala, Friday, 11 October 2019 at 13:00 for the degree of Doctor ofPhilosophy. The examination will be conducted in English. Faculty examiner: Professor SarahButcher (University of Helsinki).

AbstractNarayana Reddy, H. K. 2019. Structural Studies of Large dsDNA Viruses using SingleParticle Methods. Digital Comprehensive Summaries of Uppsala Dissertations from theFaculty of Science and Technology 1847. 72 pp. Uppsala: Acta Universitatis Upsaliensis.ISBN 978-91-513-0732-9.

Structural studies of large biological assemblies pose a unique problem due to their size,complexity and heterogeneity. Conventional methods like x-ray crystallography, NMR, etc. arelimited in their ability to address these issues. To overcome some of these limitations, singleparticle methods were used. In these methods, each particle image is manipulated individuallyto find the best possible set of images to reconstruct the 3D structure. The structural studies inthis thesis, exploit the advantages of single particle methods.

The large data set generated by the SPI study of PR772 provides better statistics about thesample quality due to the use of GDVN, a container-free sample delivery method. By analyzingthe diffusion map, we see that the use of GDVNs as a sample delivery method produces widerange of particle sizes owing to the large droplet that are created.

The high-resolution structure of bacteriophage PR772 confirmed the speculation about theheteropentameric nature of the penton and revealed the new architecture of the vertex complexconsisting of a hetero-pentameric penton formed with three copies of P5 and two copies of P31.The beta propeller region of P2, formed by domains I and II is bound to the N-terminal domainof P5. The structure also reveals new conformations of N-terminal and C-terminal region of P3which play an important role in particle assembly and structural stability.

The study of Melbournevirus revealed the protein composition in a packed particle. TheCryoEM structure of Melbournevirus reveals a T=309 capsid with an inner lipid membrane.A dense body was found in the viral particle, a feature not observed in other viruses of theMarseilleviridae family. The density of this body is similar to a nucleic acid-protein complex.This observation, along with the histone-like protein identified during study, suggest genomeorganization in the viral particle, similar to higher organisms.

The soft X-ray microscope operated in the water-window shows the progression of theCedratvirus lurbo infection in the host cell without the use of chemical fixation, staining, sampledehydration or polymer embedding. The study revealed a significant bioconversion from thehost cell to the viral particle at later stages of infection.

Keywords: PR772, phage, PRD1, Bacteriophage, coliphage, Melbournevirus, Cedratvirus,lurbo, Pithovirus, CryoEM, Single particle imaging, Coherent, Diffractive, Imaging, Soft X-ray, Microscopy, Microscope, GDVN, High resolution, XFEL, aerosol, Injection, electrospray,gas dynamic virtual nozzle, CDI, CXI, FEL

Hemanth Kumar Narayana Reddy, Department of Cell and Molecular Biology, Molecularbiophysics, Box 596, Uppsala University, SE-75124 Uppsala, Sweden.

© Hemanth Kumar Narayana Reddy 2019

ISSN 1651-6214ISBN 978-91-513-0732-9urn:nbn:se:uu:diva-391671 (http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-391671)

Dedicated to Vanajakshi, Narayana Reddy, Shruthi

List of Papers

This thesis is based on the following papers, which are referred to in the text by their Roman numerals.

I Reddy HKN et al. Coherent soft X-ray diffraction imaging of coliphage PR772 at the Linac coherent light source. Scientific Data 4 170079 (2017).

II Reddy HKN et al. Electron cryo-microscopy of Bacteriophage PR772 reveals the elusive vertex complex and the capsid architecture. elife (submitted)

III Okamoto K*, Miyazaki N*, Reddy HKN et al. Cryo-EM structure of a Marseilleviridae virus particle reveals a large internal microassembly. Virology 516 239-245 (2018).

IV Kördel M, Svenda M, Reddy HKN et al. Giant DNA virus infection dynamics investigated by quantitative soft x-ray microscopy. Manuscript in preparation

Reprints were made with permission from the respective publishers.

* Authors have contributed equally.

List of Additional Papers

V Kurta RP et al. Correlations in Scattered X-Ray Laser Pulses Reveal Nanoscale Structural Features of Viruses. Physics Re-view Letters. 119 (15):158102 (2017).

VI Okamoto K et al. Structural variability and complexity of the giant Pithovirus sibericum particle revealed by high-voltage electron cryo-tomography and energy-filtered electron cryo-mi-croscopy. Scientific Reports. 7(1):13291 (2017).

VII Lundholm I V, Sellberg JA et al. Considerations for three-di-mensional image reconstruction from experimental data in co-herent diffractive imaging. IUCrJ. 5(5):531-541 (2018).

VIII Munke A et al. Coherent diffraction of single Rice Dwarf virus particles using hard X-rays at the Linac Coherent Light Source. Scientific Data. 3(1):160064 (2016).

IX Hantke MF, Bielecki J et al. Rayleigh-scattering microscopy for tracking and sizing nanoparticles in focused aerosol beams. IU-CrJ. 5(6):673–680 (2018).

X Bielecki J, Hantke MF et al. Electrospray sample injection for

single-particle imaging with x-ray lasers. Science Advanced. 5(5):8801 (2019).

Contents

1 Introduction ............................................................................................11

2 Viruses ...................................................................................................122.1 Classification of Viruses ..................................................................122.2 Bacteriophages ................................................................................13

2.2.1 Tectiviridae ..............................................................................132.3 Giant Viruses ..................................................................................15

2.3.1 Marseilleviridae .......................................................................162.3.2 Pithoviridae .............................................................................17

3 Structure Determination of Biomolecules ................................................193.1 Conventional Methods.....................................................................23

3.1.1 X-ray Crystallography ..............................................................233.2 Single Particle Methods ...................................................................25

3.2.1 Electron Cryo-Microscopy .......................................................253.2.2 Single Particle Coherent X-ray Diffractive Imaging ..................263.2.3 X-ray Cryo-Microscopy ...........................................................28

4 Sample Preparation and Characterization ................................................304.1 Bacteriophage PR772 ......................................................................304.2 Melbournevirus ...............................................................................324.3 Cedratvirus lurbo ............................................................................354.4 Sample Optimization .......................................................................36

4.4.1 Electron cryo-microscopy ........................................................364.4.2 Single Particle - Coherent X-ray Diffractive Imaging ...............374.4.3 X-ray Cryo-Microscopy ...........................................................39

5 Structural Studies of Bacteriophage PR772 from Tectiviridae .................405.1 Introduction .....................................................................................405.2 SP-CDI of PR772 ............................................................................42

5.2.1 Experimental setup ...................................................................425.2.2 Data Collection and Analysis ...................................................425.2.3 Results .....................................................................................435.2.4 Data Deposition .......................................................................44

5.3 Electron Cryo-Microscopy of PR772 ...............................................455.3.1 Data Analysis ...........................................................................455.3.2 Model Building ........................................................................455.3.3 Results and Discussion .............................................................46

6 Structural Studies of Giant Viruses .........................................................506.1 Structural studies of Melbournevirus ...............................................50

6.1.1 Identification of Viral Proteins in the Melbournevirus ..............506.1.2 Results .....................................................................................506.1.3 Electron Cryo-Microscopy .......................................................516.1.4 Data Analysis ...........................................................................516.1.5 Results .....................................................................................51

6.2 Structural studies of Cedratvirus lurbo ............................................526.2.1 Isolation ...................................................................................536.2.2 X-ray Cryo-Microscopy ...........................................................546.2.3 Results and Discussion .............................................................55

7 Summary and Outlook ............................................................................577.1 Achievements ..................................................................................577.2 Challenges and Opportunities ..........................................................58

Summary in Swedish .................................................................................61

Author Contribution ..................................................................................64

Acknowledgments .....................................................................................65

Reference ..................................................................................................67

List of Abbreviations

ADUs Analogue to Digital Units

AMO Atomic, Molecular and Optical beamline

APMV Acantamoeba Polypghega Mimi Virus

CCD Charged Coupled Device

CryoEM Electron Cryo-Microscopy

DLS Dynamic Light Scattering

DMA Differential Mobility Analyzer

DNA Deoxyribonucleic acid

dsDNA double stranded deoxyribonucleic acid

dsDNA-RT double stranded deoxyribonucleic acid - reverse tran-scriptase

dsRNA double stranded ribonucleic acid

EDTA Ethylenediaminetetraacetic acid

EM Electron Microscope

EMC Expand, Maximize and Compress algorithm

FWHM Full-Width Half-Maximum

GDVN Gas Dynamic Virtual Nozzle

kbp Kilo-base pair

LCLS Linac Coherent Light Source

LDB Large Dense Body

Mbp Mega-base pair

MCP Major Capsid Protein

MLM Multi-Layer Mirror

MOI Multiplicity of Infection

mRNA Messenger ribonucleic acid

NMR Nuclear Magnetic Resonance

NTA Nano-particle Tracking Analysis

ORF Open Reading Frames

PBS Phosphate Buffered Saline

PDB Protein Data Bank

PEG Polyethylene glycol

PFU Plaque forming units

RAAR Relaxed Averaged Alternating Reflections algorithm

RCF Relative Centrifugal Force

rRNA Ribosomal ribonucleic acid

SASE Self-amplified spontaneous emission

SAXS Small angle X-ray scattering

SP-CDI Single Particle Coherent Diffraction Imaging

SPA Single Particle Analysis

ssDNA single-stranded deoxyribonucleic acid

ssRNA single-stranded ribonucleic acid

ssRNA-RT single-stranded ribonucleic acid - reverse transcriptase

T-number Triangulation Number

tRNA transfer ribonucleic acid

XFEL X-ray Free-Electron Laser

11

1 Introduction

This thesis is about structural studies of large non-enveloped double stranded DNA (dsDNA) viruses using single particle structure determination methods. The work included here, encompasses structural studies of bacteriophage PR772 from the Tectiviridae, Melbournevirus from the Marseilleviridae and Cedratvirus lubro from the Pithoviridae. These viruses have a protein shell that encapsulates an inner lipid membrane and the genome. Most of the large dsDNA viruses share certain structural similarities with one another and the variations observed in their structures are usually acquired over time as these viruses evolve/adapt to new conditions to gain functions like resistance to varying temperature, salinity, pH or to overcome host de-fenses, develop the ability to infect new hosts and overcome host tropism, etc. Understanding the structural changes needed to acquire a new function would help us in engineering novel biomolecules that perform a defined function like targeting a specific host. Phages also have the potential to be used as an alter-native to antibiotics in the form of phage cocktails in phage therapy. Structural studies of large biological assemblies pose a unique problem due to their size, complexity and heterogeneity. Conventional methods like X-ray crystallography, nuclear magnetic resonance spectroscopy, etc. are somewhat limited in their ability to address disorder and heterogeneity. To overcome some of these limitations, structure determination by single particle methods can be used. In these methods, each particle image can be treated/manipulated individually to select the best possible set of images to assemble them to gen-erate a 3D structure. The structural studies in this thesis, use these advantages of the single particle methods. In chapters 2 and 3, the family of viruses studied and the single particle meth-ods used to study their structure as part of this thesis are introduced. Chapter 4 describes the sample preparation, characterization and sample optimization methods needed by different structure determination techniques. Chapters 5 and 6 summarize the structural studies of bacteriophage PR772, Melbournevi-rus and Cedratvirus lubro. In chapter 7, some of the prominent achievements, challenges and future opportunities for improvements are discussed.

12

2 Viruses

Viruses are parasites that depend on a host for biosynthesis and propagation. Viruses are usually assumed to be small particles with minimal genomes that code for a few essential proteins that are critically required for propagation in their specific host. They are simple, ubiquitous in the environment and infect all forms of life from single-cell organisms like bacteria to large mammals. There are about 1031 virions on earth [1]. The viruses in the newly discovered family of giant viruses have diverged from the simplistic nature that is generally attributed to viruses. These giant viruses can be visualized using a basic compound light microscope. These vi-ruses have genomes containing more than 500 kilo-base pairs (kbp) and en-code many proteins. Unlike typical viruses, the giant viruses are known to encode genes for DNA-modifying proteins and chaperons. In a typical virus, these proteins are absent and probably not essential for the propagation of the viral particle in the host. Up to two thirds of the identified open reading frames (ORFs) in the genomes of these viruses do not have a known homologue in the current databases of proteins and their functions are yet to be determined. There are also virophages, virus-parasitizing viruses, which cannot replicate in the host organism by themselves but hijack the replication machinery pro-duced by another co-infecting virus. These findings have compelled research-ers to reconsider the assumption that viruses have to be simplistic by nature [2,3].

2.1 Classification of Viruses Viruses can broadly be classified based on the nature of their genome as dou-ble stranded DNA (dsDNA) viruses, single stranded DNA (ssDNA) viruses, double stranded RNA (dsRNA) viruses, plus-stranded single stranded RNA ((+)ssRNA) viruses, minus-stranded single stranded RNA ((−)ssRNA) vi-ruses, single stranded RNA - reverse transcriptase (ssRNA-RT) viruses and double stranded DNA – reverse transcriptase (dsDNA-RT) viruses [4]. This method of classification is known as the Baltimore classification of viruses, named after the Nobel laureate in Physiology or Medicine, David Baltimore.

13

The thesis will focus on group I: dsDNA viruses. These viruses have a double stranded DNA as their genome. The genome is transcribed to mRNAs that act as templates for protein synthesis.

2.2 Bacteriophages Bacteriophages are viruses that infect bacteria. They were discovered inde-pendently by Frederick W. Twort in 1915 and Felix d’Herelle in 1917. The life cycle of bacteriophages is similar to other typical viruses, with some phages displaying a lysogenic cycle as seen in lambda phage [5] and other like the T4 phage with a lytic cycle [6]. Phages with a lysogenic life cycle enter a dormant phase during which there is minimal phage related activity in the host cell. Many of these lysogenic phages can revert to an active lytic phase (tem-perate life cycle). The phages with a lytic life cycle, hijacks the molecular machinery of the host cell and multiply rapidly resulting in quick disruption of the host cell. A “typical” phage has a polyhedral head with a tail that is used for injecting the viral genetic material into the host, but phage particles can also adopt other shapes such as helical or pleomorphic.

2.2.1 Tectiviridae The family of Tectiviridae contain bacteriophages that infect a wide range of hosts [7]. The members of the family have a ~14 -16 kbp long linear dsDNA as their genome. They have a pseudo-icosahedral protein capsid that encapsu-lates a protein-rich lipid membrane and the genomic DNA. These phages are tail-less in their dormant state and produce a protein-assisted membranous

Figure 2.1. Baltimore classification of viruses

14

tube during the infection process. These membranous tubes deliver the viral dsDNA to the host during the infection [8]. Tectiviridae is currently classified into three genera, Alphatectivirus, Betatectivirus and Gammatectivirus [9]. Al-phatectivirus like bacteriophage PRD1 are lytic viruses that infect Gram-neg-ative bacteria that produce a plasmid-encoded receptor [9,10]. Betatectivirus like Bam35 follow a lysogenic cycle. They infect Gram-positive bacteria [9]. Gammatectivirus is the most recently proposed genus of the Tectiviridae. They infect alpha proteobacterial hosts. Phage GC1 from this genus infects a Gram-negative bacterium but shows a temperate lysogenic cycle like the Be-tatectivirus [11]. Entero-bacteriophage PRD1 is the most extensively studied species from the Tectiviridae. It has a pseudo-icosahedral protein capsid that encapsulates a protein rich internal membrane [12] and it is tail-less in the dormant state, displaying the typical features of the Tectiviridae. The diameter of the particle is about 65nm with a pseudo triangulation number (T-number) of 25 and it has a ~14 kbp linear dsDNA genome. The linear genome has inverted terminal repeats and the 5’ termini on either side are bound by a replication-priming protein, P8 [10]. The genome is packaged into the viral particle through a unique packaging vertex [13]. The genome encodes for 25 proteins with functions that are, at least partly, well understood [14]. The most abundant protein found in the viral particle is the major capsid protein (MCP) P3, which forms trimers that appear hexago-nal in shape. The overall structure of the P3 subunits that are present in the different regions of the capsid are similar but conformational variations have been reported in the C-terminal and the N-terminal regions of the protein. These conformational variations are known to be important for the assembly of the viral particle [15]. The N-terminal region of P3 subunits show two conformations; one is a long helix which contacts the membrane, the other is a helix turn helix where the short helix is twisted away from the membrane and interacts with the neigh-boring P3 subunit of the trimer [15]. The C-terminal region of the P3 subunits show four different conformations that play an important role in stabilizing the particle [15]. The P3 along with the tape protein P30, form most of the viral capsid. P30 dimers span between adjacent vertices and forms the skele-ton that holds the MCPs together in the icosahedral facet and controls the size of the particle [16]. The viral capsid is completed by the vertex complex on 11 of the 12 vertices of the icosahedral particle and the unique packaging portal on the 12th vertex through which the genomic dsDNA is actively transported into the procapsid with the help of the P9 ATPase [15,17]. The genomic dsDNA passes through

15

the internal membrane by the pore formed by proteins P20, P22 and facilitated by protein P6. The unique packaging portal lacks the proteins that are present in the other 11 vertices [13]. The vertex complex is formed by P31 (penton protein), P5 (spike protein), P2 (receptor binding protein) and P16 (infectivity protein). Mutational studies showed that five copies of P31 form the penton base at the vertices. P31 is essential for the binding of P5 and P2 proteins to the vertex complex. Particles that, through deletion mutation lacked P31, failed to form the vertex complex and these particles were also missing the P3 subunits that are adjacent to the penton. Mutants that lacked P5 produce intact particles containing the ge-nomic DNA but failed to incorporate P2. The P5 mutants failed to deliver the packed DNA. P2 mutants, lacking P2, produced intact particles in which both P31 and P5 were incorporated but the receptor binding activity was lost [18,19]. Based on these mutation studies, it was proposed that P31 forms the penton that binds P2 and the P5 protein was bound in-between P31 and P3. This was the first model that was proposed for the vertex complex. This model was later updated with information from in-vitro studies of the vertex proteins P31, P5 and P2. It was hypothesized that the pentameric P31 complex interacts with the trimeric P5 N-terminal domain. The C-terminal domain of P5 forms a trimer that binds P2 [20]. Later studies with a P16 knock-out mutant and a 4.2 Å X-ray crystallography structure of the PRD1 Sus539 mutant (lacks P2), showed that P16, an integral membrane protein found below the penton, stabilized the vertex complex and also plays an important role in the infection process [15,21]. With a small an-gle X-ray scattering (SAXS) model of the P5 protein and a low-resolution cryoem map, it was shown that P2 and P5 form two different spikes at the vertex rather than a single spike as previously hypothesized and that the N-terminal domain of P5 was embedded into P31 pentamers but the interaction site of P2 was not resolved [22].

2.3 Giant Viruses Typically referred to as nucleocytoplasmic large DNA viruses (NCLDV), these viruses have a size varying from 140 – 750 nm [23]. NCLDVs were earlier thought to consist of only five lineages based on monophyly; namely Poxviridae, Asfarviridae, Iridoviridae, Phycodnaviridae, Mimiviridae. NCLDVs were grouped together based on a small set of 9 core genes that are shared by all members of NCLDVs and other set of genes that were commonly found in most of the members [2,24]. Recently, new families like Mar-seilleviridae, Pithoviridae, Pandoraviridae, etc have been shown to have sim-ilarities to NCLDVs and share the 9 core genes. Most NCLDVs were also

16

believed to have a pseudo-icosahedral shape with the exception of the pox-viruses which are ovoid or brick-like [25]. The discovery of even larger vi-ruses like Pandoravirus spp, Pithovirus spp, Tupanvirus, etc having sizes ranging from 0.9 – 1.65 µm and also deviating from the pseudo-icosahedral shape. Pandoravirus shows a distinct amphora-shaped capsid and the Pithovi-rus has an ovoid shape with a cork-like feature at one of the apexes, to which the dsDNA is bound (Paper VI). Tupanvirus, is a tailed virus from the Mim-iviridae family [26–29].

Giant viruses also have significantly larger genomes compared to “traditional” viruses, varying between 0.3 – 2.8 mega base pairs (Mbp) [30]. There is no direct correlation between the size of the particle and the size of the ge-nome. Pithovirus sibericum, the largest among the currently known giant vi-ruses with a size of about 1.7 µm, has a relatively small genome size of 0.6 Mbp, whereas CroV, a 0.3 µm diameter particle has a genome size of 0.73 Mbp [30]. The other interesting feature of their genomes is that they have a very high percentage of ORFans; genes or ORFs which do not have a known homologue in our current databases and thus, the function of the correspond-ing proteins is not known [25]. Among the larger of the known giant viruses, the percentages of ORFans can be as high as 70% - 85% of all the ORFs in their genome [30]. Most of the giant viruses also show the presence of DNA modifying histone-like proteins, which is very interesting and this observation could suggest possible genome organization and also contradicts the currently accepted idea of a minimalistic nature of viruses [29,31].

2.3.1 Marseilleviridae Marseillevirus, the prototype of the Marseilleviridae family was first reported in 2009, as a large virus that infects Acantamoeba spp [32]. In recent years, many new species that belong to the Marseilleviridae family have been re-ported from samples that were collected from different part of the world, but very little is known about these viruses [33,34]. The viruses of this family are comparatively smaller than Acanthamoeba polyphaga mimivirus (APMV) or Mamavirus [33]. They have a pseudo-icosahedral viral capsid with a diameter ranging from 180 – 260 nm [35]. Beneath the viral capsid they have an internal membrane that encloses the genome. Some of the members of this family also produce fibers on the viral capsid surface [32]. The genome is usually a circu-lar dsDNA with a size of 340 – 390 kbp, but Lausannevirus, a member of the Marseilleviridae exhibits the genome as either a circular molecule or a linear molecule with terminal repeats [35,36]. The members of Marseilleviridae, like other large viruses, replicate by form-ing viral factories in the cytoplasm [37]. Once the viral particles are internal-ized by the host, the virus hijacks some of the host machinery and

17

compartmentalizes it to produce viral proteins. The viral factories in Marseil-liviridae are found in the cytoplasm of the host but close to the nucleus. The morphology of the host nucleus is modified, but it is not disrupted. It appears that the virus, during the viral propagation, shows a cooperative lifestyle with the amoeba. The time taken from ingestion of the virus by the host to host lysis, releasing the mature viral particles, is about 16 - 18 hrs [33]. The genomic analysis of the prototype virus from the Marseilleviridae family, the Marseillevirus, shows significant genomic mosaicism due to lateral gene transfer from the host or other organisms that were co-infected along with the virus [38]. The genome has a GC content of 44.7% and ORFs form 89 – 94 % of the genome. The genome contains all the NCLDV core genes along with most of the conserved genes that are typically seen in NCLDVs [23,35]. In Marseillevirus, 11% of the genes seem to have a NCLDV origin, 9% a bacte-riophage or bacterial origin, 19 % with an eukaryotic origin and the rest showed no clear origin [32]. The most interesting finding in the genomic anal-ysis is the presence of core histone-like doublets and DNA topo II genes that are usually seen only in higher organisms that could enable some form of ge-nome organization. These genes are proto-eukaryotic-like but homology stud-ies showed that these were not acquired by horizontal gene transfer [2,33,34].

2.3.2 Pithoviridae The family of Pithoviridae contain the largest known virus particles to date. The family consist of DNA viruses that infect Acanthamoebal hosts. They have a distinct ovoid shape with a diameter of 0.5 µm – 0.7 µm and length of 0.6 µm – 1.7 µm. Their genome size is relatively small, about 600 kbp, when compared to the size of the viral particle. Other viruses that are smaller in size possess larger genomes [27,28,30,39]. The type virus of this family is Pithovirus sibericum. It was isolated from a 30,000 year-old Siberian permafrost sample. Pithovirus appears ovoid with a cork-like feature at the apex and a dense outer envelope called the tegument. The cork-like structure at the apex appears tubular and is arranged like a hon-eycomb. The tegument is 60 nm thick and appears striated. The tegument co-vers an inner lipid membrane that encapsulates the genome but most of the inner space appears to have a lower density compared to the tegument and is devoid of any prominent features [27]. Recently, a new genus of viruses, called the Cedratvirus, were reported from this family, which has a similar overall morphology to the Pithovirus spp but show the presence of two cork-like features on both the apices. Cedratvirus spp shares about 103-113 genes with the previously described viruses of the Pithoviridae family [39,40].

18

The viruses from the Pithoviridae family, like other NCLDVs, are assembled in the cytoplasm of Acantamoeba spp [30]. The viral particle enters the host by active phagocytosis by the host. The apical cork is removed by low pH in the phagocytic vacuole, which in turn enables the inner membrane of the viral particle to fuse with the lipid membrane of the host vacuole. After 4 hrs of infection, the formation of viral factories in the cytoplasm can be seen, around which electron dense vesicles appear [27]. The most prominent phases during replication of the virus can be distinguished by the formation of the corked particle followed by the addition of the thick tegument, resulting in a mature ovoid particle [30]. As the viral particles are produced, they accumulate in the host cell. They are released from the host when the host membrane ruptures. A complete lysis of the host is typically seen 15-20 hrs after the initial infec-tion [27,30]. The Pithovirus genome is an AT-rich 610 kbp long dsDNA encoding 467 predicated proteins. The recognizable protein homologs from the protein da-tabases to the predicated proteins in Pithovirus is low [30]. Only 152 predi-cated proteins had homologs in the databases, of which only 125 proteins had functional attributes. Most of these proteins are involved in DNA transcrip-tion, replication, repair and nucleotide synthesis. Based on the best hits to the predicated proteins, the Pithovirus is most similar to Marseillieviridae fol-lowed by Megaviridae [27].

19

3 Structure Determination of Biomolecules

In the early 1900s, it was shown that the one could use X-ray crystallography to determine the structure of biological molecules. In 1935, tobacco mosaic virus was crystallized [41] and the diffraction data were collected using X-rays in 1941 [42]. With the improvements in technology and better under-standing of the crystallization techniques and X-ray crystallographic analysis, the structure of DNA and the protein myoglobin was elucidated in 1953 and 1958 respectively [43,44]. These structures shed light on how DNA performed its role as a genetic material and how myoglobin could store oxygen in the muscle. These discoveries showed the importance of understanding the struc-ture of biomolecules and laid the foundation for the field of modern structural biology. With over 100 thousand of known protein structures determined to date, we can define the organization of any protein into four different organization lev-els: the primary, secondary, tertiary and quaternary arrangement. The primary structural organization is a long chain of amino acids, the basic building blocks, which are bound together by peptide bonds. This is typically called a polypeptide chain. There are about 20 standard amino acids that are encoded by the triple codons of the gene. There are also a few non-standard amino acids like selenocysteine, formylmethonine, etc.; that are not typically found in living organisms. The amino acids have different side chains which give them different chemical properties. They can be loosely classified based on the properties of their side chains as polar-charged (ex. Arginine, Lysine), po-lar-uncharged (ex. Tyrosine), non-polar (ex. Leucine) and Glycine, with no side chain. These amino acid side chains play an important role in the for-mation of other higher order of arrangements [45]. The amino acid side chains can interact with other amino acid residues, other biological molecules or ions to form hydrogen bonds, disulphide bonds, etc. The backbone of the polypeptide chain has limited rotation angles that are allowed for the bonds due to steric clashes. The side chains also show certain preferred orientations in proteins. These preferred conformations seen in amino acids side chains are called rotamers. These geometric limitations along with the chemical properties of the amino acids give rise to a wide range of structural features in proteins. In most folded proteins, one can identify regular features called secondary structures. These secondary structures typically

20

include structures like alpha helices, beta strands, etc., which provide an effi-cient method to satisfy the restriction/limitations mentioned above by forming hydrogen bonds. Some of the regions might not be organized; these regions are called coils and loops. They could act as linkers that connect two second-ary structures or could form the flexible part of a protein. These secondary structures give rise to folds/motifs/topologies like jellyrolls, TIM barrels, etc. Different arrangements of secondary structures form a tertiary structure that are further stabilized by ionic bonds, salt bridges and hydrophobic interactions between the residues in the adjacent polypeptide chains. Many polypeptide chains with the aforementioned features, sometimes along with other biomol-ecules can be arranged to form larger complexes called a quaternary structure. Most functional proteins exhibit a quaternary structure [46].

Nucleic acids, like proteins also exhibit different levels of organizations. Nu-cleic acids are polymers of nucleotides. A nucleotide has a 5-carbon sugar (ribose or deoxyribose) to which a nitrogenous base and a phosphate group are bound. Depending on the type of 5-carbon sugar, we can have a ribonu-cleic acid (RNA) and deoxyribonucleic acid (DNA). The nitrogenous bases can be purine (adenine, guanine) or pyrimidines (cytosine, thymine, uracil). Adenine, guanosine and cytosine are common for both DNA and RNA whereas thymine is found in DNA and uracil is found in RNA. DNA typically forms a double helix as a secondary structure where the double helix can ex-hibit three forms. The most common form seen in-vivo in regular cells is the

Figure 3.1. Levels of structural organization seen in protein.

21

B form, which is a right-handed double helix. In higher organisms, the DNA along with DNA structure regulating proteins such as histone proteins can form complexes called nucleosomes which are the building blocks of chroma-tin. RNA is structurally more versatile compared to DNA and exhibits a wide range of conformations or folds. The primary structure is again a chain of nu-cleotides as seen in mRNA. They can form simple secondary structures like hairpin loops, stem loop arrangements. A complex arrangement of the second-ary RNA structure gives the tertiary structure. These tertiary structures of RNA are seen in some of the most important biological molecules like tRNA, rRNA, etc [46]. In some cases, nucleic acids can also form quaternary-like structure in combination with proteins.

The properties of proteins, like flexibility, size and shape, charge, etc varies depending on the local environment in the organism where they are present and the kind of function they perform. Some proteins are very rigid, like col-lagen, keratin, etc seen in connective tissue and hair. Their rigidity makes them biologically inactive. Some proteins with biological activity can be fairly rigid, as seen in case of viral capsid proteins, where their function is to main-tain structural integrity of the complex, but they should still be flexible enough to assemble into a viral capsid. Most biologically active proteins are also dy-namic as seen in most enzyme, receptors, etc. In case of enzymes, they have a

Figure 3.2. Levels of structural organization seen in nucleic acids. Generated from PDB 1BNA and 4TNA

22

rigid overall structure but local regions that are important for catalysis or mo-bility of the substrate and product in and out of the active site respectively, could be flexible. This flexibility of proteins could result in a protein with different size and shape depending on the environment and the stage of catal-ysis. These different size and shapes that a protein exhibit are called the dif-ferent conformations of that protein. These conformations are critical for proper functioning of proteins [46]. Most of the unbound, soluble proteins are globular where the hydrophilic sur-face is exposed and hydrophobic regions of the protein are embedded within the protein. In case of membrane bound proteins, they have their hydrophobic surfaces exposed to the hydrophobic region of the membrane in which they are embedded. These proteins are insoluble in charged or polar solvents and tend to aggregate when they are not in a suitable solvent or bound in the mem-brane. Many such properties of proteins and nucleic acids pose a challenge to any technique that is used for determination of their structure. In the last 100 years, the importance of structure determination of biomole-cules and the need for understanding the organization of biological systems have been realized. Many different techniques have been developed to eluci-date these structures, with every method having its own advantages and dis-advantages. Usually, several methods are used to overcome the shortcomings of one single method

23

3.1 Conventional Methods There are some well-established traditional techniques used to determine the structure of purified biological molecules. The common techniques are crys-tallography and nuclear magnetic resonance (NMR) spectroscopy. In crystal-lography, diffraction data from crystallized biological molecules, exposed to electromagnetic radiation like X-rays or sub-atomic particles like electrons or neutrons are collected and processed. The quality of the structure determined by crystallography depends of the uniformity of biological molecules that form the crystal and the extent of radiation damage during exposure. In NMR spectroscopy, the behavior of the nuclei of atoms of a biological sample in solution when subjected to a strong magnetic field, is harnessed. It provides information about the flexibility and dynamics of the biomolecule but with larger sample the structure analysis is complex. For the structural studies of large viral assemblies, NMR is seldom used. In both these conventional meth-ods, the data collected reflect the average of all the information about the sam-ple in a certain volume.

3.1.1 X-ray Crystallography X-ray crystallography is the most popularly used structure determination method for biological macromolecules. There are about 137 thousand struc-tures deposited to the Protein Data Bank (PDB) that were determined using X-ray crystallography (as of Aug 5th, 2019) [47]. The roots of x-ray crystal-lography can be traced back to the discovery of Laue diffraction in 1912. Laue showed that when X-rays pass-through salt crystal they produced a specific pattern on a photographic film depending on the nature of the specimen. In 1913, the crystal structure of NaCl was reported by Bragg using the Bragg’s condition [48]. It was shown that at certain wavelengths and angle of inci-dence of the x-rays on the crystal they produced intense spots on the photo-graphic film. Under these conditions, the X-rays scattered by the crystal lattice interfered constructively and the dark spots seen on the photographic film were Bragg peaks. For constructive interference to occur, the interfering waves have to be in phase with each other and this is true when the path dif-ference between the waves is equal to an integer multiple of the wavelength of the incident wave. Bragg’s condition is given mathematically as:

where d is the distance between the lattice planes of the crystal, θ is the scat-tering angle of the wave with respect to the lattice plane and λ is the wave-length of the incident wave. As all moving particles exhibit wave properties, this law can also be applied to particles like electrons, neutrons, etc. The wave length of these particles is given by the de Broglie wavelength, λ:

24

where h is the Planck’s constant, p is the momentum, m is the mass of the

particle and v is the velocity of the particle.

Bragg peaks that can be seen on the detector due to diffraction, give infor-mation about the intensity but the relative phase information is lost. This phase information is required to reconstruct the electron density of the sample build-ing up the crystal and it has to be retrieved by other methods. This is typically referred to as the phase problem. The phases can be retrieved by using exper-imental phasing methods or by molecular replacement. Since many structural conformations have already been determined, the most commonly used tech-nique is the molecular replacement method where a model of a similar struc-ture is used to calculate initial phases which then are iteratively improved and the initial model bias can be removed. When there are no structurally similar models available, experimental phasing methods like anomalous dispersion or isomorphous replacement are used. The advent of synchrotrons and faster digital detectors has enabled rapid data collection as compared to the previous in-house X-ray sources and photo-graphic film. For a good crystal, a complete data set can be collected in a mi-nute. Recently, X-ray Free-Electron Lasers (XFELs) with higher peak bril-liance and femtosecond pulses have significantly reduced the size of the pro-tein crystal that is required to collect a complete data set. The X-ray pulses are diffracted when they interact with nanocrystals. These diffraction patterns are Bragg’s peaks from a small set of well-arranged individual particles and can be indexed separately. The indexed images are then sorted and merged to get a complete dataset for structural analysis [49]. Thus, reducing the size of the crystal needed and also the effect of radiation damage. Ultimately, improving the quality and resolution of the final reconstruction.

25

3.2 Single Particle Methods The advent of high-resolution single particle methods is fairly recent. In these methods, the data collected represents the information about a single individ-ual biological particle of the sample being measured. The single particle data can be individually assessed and manipulated to pick the best particles that contribute to the final assembled structure.

3.2.1 Electron Cryo-Microscopy Electron Cryo-Microscopy (CryoEM) has been around for a few decades and it was extensively used to study low-resolution structures of large protein do-mains and complexes. As the name suggests, the sample is at a cryogenic tem-perature when the data/images are collected. The sample is vitrified by rapid cooling using liquid ethane and stored in liquid nitrogen to avoid the formation of crystalline ice. The recent advancement in detector technology and the in-troduction of fast direct electron detectors and new microscopes with auto-mated data collection ability has enabled this method to reach resolutions high enough where building de-novo models into the density map is becoming more common. The two common 3D reconstruction methods are the single particle reconstruction method and electron cryo-tomography [50]. During the process of imaging, the biological sample is affected by the elec-tron beam that interacts with it. This is typically referred to as radiation dam-age. The extent of sample damage by the electrons is proportional to the time of exposure of the sample to the electron beam (electron dose). The vitrified condition under which the sample is imaged helps to mitigate the damage but the damage is not completely eliminated. The new direct electron detectors have also significantly reduced the exposure time and the electron dose needed to image biological samples [51]. 3D reconstruction of single particle images of biological molecules in differ-ent orientation that are embedded in vitreous ice is the most commonly used high resolution CryoEM structure determination method. The popularity of this single particle method has made it synonymous to the abbreviation “Cry-oEM”, even though it is a misnomer. The single particle images are collected as movies and a very low electron dose exposure of the sample to the electron beam contribute to the individual frames of this movie. The individual frames are noisy but they can be averaged after motion correction to improve the overall signal. This helps in controlling the effect of radiation damage on the final 3D reconstruction. The images collected by a given microscope are af-fected by the defocus used and blurring by the point spread function. This is corrected by contrast transfer function (CTF) correction [52].

26

The basic CTF corrected images can now be used to extract the individual particles for 3D reconstruction. To improve the signal - noise ratio (contrast from the background), the extracted particles with the same conformation and orientation are aligned, centered and averaged. In some cases, the biological particle can also be stained to increase the contrast of the particles but staining could introduce artifacts. The averaged particle images are used to reconstruct the 3D structure of the particle. The averaged images represent a 2D projection of a 3D object and the relative orientation of these 2D projections with respect to each other has to be determined to generate the 3D structure. This can be done with methods like random conical tilt, common lines, projection match-ing, etc. Once the approximate orientations of the 2D projections are deter-mined they can be iteratively refined. Using these refined orientations of the projections, a 3D structure is reconstructed using real-space methods like back projection, convolution or Fourier methods like Fourier inversion. For a good 3D reconstruction, one has to have good 2D projections that represent well distributed orientations covering the entire 3D volume of the particle [52]. The second CryoEM method is electron cryo-tomography, where a single vit-rified biological sample is imaged from different angles to generate 2D pro-jections that represent different orientations of the sample. In cryo-tomogra-phy, the radiation damage to the sample is higher compared to the other single particle CryoEM method. The total electron dose is higher since the same par-ticle is imaged from different angles. The stage that holds the sample grid is tilted, typically from -60o to 60o and images are collected for every 1o - 10o of tilt. The number of images that can be collected are limited by the maximum possible tilt angles of the stage. The lack of images representing the orienta-tion of the sample beyond the possible tilt angle results in missing sampling data. The missing wedge of data and significant radiation damage reduces the resolution of the final 3D reconstruction. The images of the sample from dif-ferent angles are aligned to the common tilt axis by tracking small electron dense particles called fiducials, that are frozen in vitreous ice along with the sample and optimizing the overall image cross-correlation. Once the images are aligned, the 3D structure is reconstructed by methods like back projection, simultaneous iterative reconstruction (SIRT), etc [53].

3.2.2 Single Particle Coherent X-ray Diffractive Imaging In conventional X-ray crystallography, the X-rays are produced from synchro-tron sources. They are stable, monochromatic and continuous. The quality of the Bragg’s peaks that are formed depends on the quality and size of the bio-logical crystal and the intensity of the X-ray beam. If the size of the crystal is reduced, then the intensity of the X-ray beam and the time of exposure used for imaging has to be increased to get Bragg’s peaks with a counting statistics similar to Bragg’s peaks produced if they were produced from a larger crystal.

27

This would lead to increase in radiation damage to the crystal. Thus, reducing the quality and the resolution of the diffraction that is produced [54]. In 1971, John Madey showed that the radiation emitted by an electron beam moving through an alternating magnetic field at relativistic speeds can have laser-type amplification and coherence [55]. In a FEL, electron bunches are accelerated to relativistic speed and passed through an undulator with an array of magnets that produce a standing magnetic field that alternates along the length of the undulator. As the electron bunches wiggle through the alternating magnetic field of undulator, they produce electro-magnetic radiation. This ra-diation interacts with the electrons and produces electron microbunching. The electron in the micro bunch, as they pass through the rest of the undulator, produce monochromatic radiation pulses with high coherence and intensity. This high peak brilliance and the short femtosecond photon pulses along with the tunable property of FEL energies makes FEL sources better suited for stud-ying small particles and they can also deliver better temporal resolution. The coherent femtosecond pulses can traverse a biological sample and produce a diffraction pattern before the effects of radiation damage could propagate through the biological sample, thus capturing an unaltered native structure of the sample [56].

Single particle coherent X-ray diffraction imaging (SP-CDI) is one such tech-nique that uses an X-ray pulse from a FEL to illuminate a single particle and

Figure 3.3. A schematic of a typical experimental setup for single particle imaging using FELs.

28

collect the diffraction pattern on a digital camera. The single particles could be layered on a surface or introduced directly into the X-ray pulse by substrate-free sample delivery methods for imaging. In most SP-CDI experiments, the sample is aerosolized using a gas dynamic virtual nozzle (GDVN) [57] and these beams of particles are focused and introduced into the interaction region in the vacuum chamber using an aerodynamic lens stack (Paper IX). Imaging experiment are typically performed in a vacuum chamber at 10-6 mBar. The vacuum is critical to avoid any interaction between the X-ray pulses and the atoms of the gas molecules in the chamber. The Uppsala injector (Paper IX), with differential pumping, helps in introducing the sample particles from at-mospheric pressure to the interaction region in vacuum. The biological sample interacts with an X-ray pulse in a random orientation, thus capturing diffrac-tion patterns that represents a random orientation of the sample. Like X-ray crystal diffraction, the diffraction pattern produced by SP-CDI represent the intensity and the phase information is lost. The phase information is retrieved iteratively during the image processing and 3D reconstructions phase.

The 3D reconstruction of a biological sample using SP-CDI is similar to the single particle reconstruction methods in CryoEM. Typically, the orientations of individual images are determined by the expand, maximize and compress algorithm (EMC algorithm) [58,59] and the iterative phase retrieval is per-formed by RAAR algorithm [60] with a constraint.

3.2.3 X-ray Cryo-Microscopy X-ray Cryo-Microscopy uses soft X-rays to image samples. It is similar to a light or an electron cryo-microscope but uses X-rays for imaging. In 1960, the

Figure 3.4. Shows the explosion of lysozyme induced by radiation damage [56]. For a FEL pulse with a FWHM of 2 fs, the structure of the protein before the interaction with the FEL pulse and after the interaction with the FEL pulse appears to be identi-cal. Reused with permission.

29

X-ray microscope was demonstrated by Cosslett and Nixon using an in-house point projection X-ray microscopy setup. Later, the development of spe-cial optics for focusing X-rays like the Kirkpatrick-Baez mirrors, compound refractive lenses, multilayer Laue lenses and the zone plates, helped to im-prove the signal-noise ratio and the resolution of the images [61]. The transi-tion from inhouse X-ray sources to highly monochromatic synchrotron-based X-ray sources also played an important role in improving the capability of these soft X-ray microscopes [61]. The exposure of biological samples to X-rays can result in significant radiation damage. Similar to electron cryo-mi-croscopy, the effects of radiation on the samples can be reduced by vitrifica-tion if the samples are imaged in a cryo-frozen condition.

In a modern in-house setup, X-rays are produced using a liquid-jet laser-plasma source and focused on to the sample using a normal-incidence multi-layer condenser mirror (MLM) and a zone plate acts as an objective lens to produce a real image on a CCD [62]. The sample preparation is similar to CryoEM and the grids that are prepared can be used interchangeably. Cur-rently, cryo-soft X-ray tomography has been demonstrated successfully in im-aging large intact cells and subcellular assemblies [62–64]. The use of soft X-rays in the water window (λ=2.3-4.4 nm; E=284-540 eV) provides certain ad-vantages like increased penetration depth which enables imaging of large sam-ples, good absorption contrast from carbon rich biological structures, whilst the absorption by oxygen in the water is minimal, also with an option to use phase contrast during imaging [65].

Figure 3.5. A schematic of a typical experimental setup for single particle imaging using a soft X-ray cryo-microscope.

30

4 Sample Preparation and Characterization

4.1 Bacteriophage PR772 Bacteriophage PR772 is a lytic phage from the Tectiviridae family that infects Gram-negative bacteria. The agar overlay method was used for the propaga-tion of Bacteriophage PR772 on an E.coli host. This method yielded signifi-cantly higher and consistent concentrations of the viral particles compared to viral propagation in broth. A 108 plaque forming unit (PFU) mL-1 stock of PR772 (ATCC BAA-769-B1) was prepared by following the procedure rec-ommended by ATCC, from where the viral sample was procured. The sample was prepared as described in Paper I and Paper II. In brief, Tryptic Soy Agar (Difco Tryptic Soy Agar) of 4% w/v was prepared and poured into petri dishes to form the bottom hard agar layer. The viral stock was mixed with host seed culture and the melted soft agar that was cooled to 45 oC and poured over bottom hard agar layer. The plates were incubated overnight at 37 °C.

The soft agar overlay was scraped off and 100 mL of the phage storage buffer (TRIS 50 mM, NaCl 100 mM, MgSO4 1 mM, EDTA 1 mM, pH 8.0) is added to it. The mixture is emulsified thoroughly to break any lumps of soft agar and mixed overnight at 4 °C to allow the viral particles to diffuse into the storage buffer. The agar and cell debris are pelleted by centrifugation at 8,000 g for 30 min. The pelleted material is discarded and the supernatant is filtered through a 0.45 μ filter (Filtropur S 0.45, Sarstedt) to remove any remaining debris

The viral particles are purified and concentrated in a two-step process. The viral particles are separated from the storage buffer solution by PEG precipi-tation. 9 % w/v PEG 8000 and 5.8 % w/v NaCl is added to the permeate and incubated overnight on a shaker at 4 °C. The solution with the viral precipitate is centrifuged for 90 min at 8,000 g. The supernatant is discarded and the viral pellet is suspended in a 1 ml storage buffer by gently mixing. In the second step, the viral particles are purified by ion exchange chromatography at 4 °C. The particle suspension is applied on to a Capto-Q anion exchange column

31

and the viral particles are eluted by increasing the concentration of NaCl (100 mM–1.5 M).

These purified viral particles were visualized with an electron microscope (EM) (Quanta FEG 650, FEI) (Figure 4.1A). Further, the sample was

Figure 4.1. Characterization of purified bacteriophage PR772 particles. A) shows a micrograph of bacteriophage PR772 particles after purification. B) Show the results of NTA. The particle concentration was 108 and the single peak suggested that the sample was pure without significant aggregation. The sample was diluted by a factor of 105 before the analysis. adapted from Paper I

32

enumerated and characterized for structural studies. The concentration of the viral particles was estimated using plaque assay and Nano-particle Tracking Analysis (NTA). The size and monodispersity of the was tested using Dy-namic Light Scattering (DLS) and NTA.

4.2 Melbournevirus Melbournevirus belongs to the Marseilleviridae family and infects Acan-tamoeba spp. Melbournevirus was propagated using Acanthamoeba castel-lanii cells as described in Paper III. The host cells were cultured in T25 flasks using the enriched PPYG medium (2.0% w/v Proteose peptone, 0.1% w/v Yeast extract, 4mM MgSO4, 0.4 mM CaCl2, 0.05 mM Fe(NH4)2.(SO4)2, 2.5mM Na2HPO4, 2.5 mM KH2PO4, and 100mM Sucrose, pH 6.5). The cul-ture flasks were incubated in a CO2 incubator at 32 °C and monitored regu-larly. When the culture was 90% confluent, the spent media was discarded and a fresh 8 mL enriched PPYG media along with the virus was added. The host cells were infected at a multiplicity of infection (MOI) of 4. The host cells were monitored for cytopathic effects.

The viral particles were harvested when most of the host cells were disrupted. The infected media from the culture flasks was collected and thoroughly mixed to reduce any aggregation. This media was centrifuged at 500 g for 10 min to remove the cell debris. The collected supernatant was centrifuged at 8000 g for 30 mins to pellet the viral particles. The pellet was washed and

Figure 4.2. Characterization of purified bacteriophage PR772 particles using DLS. The purified sample produces a single peak and does not show aggregation. The mean di-ameter of the particle was 75.99 nm with a polydispersity index (PDI) of 0.012. adapted from Paper I

33

resuspended in PBS. The viral suspension was layered on a 10% - 40% w/v sucrose step gradient and centrifuged for 90 min at 6500 g. The viral particles accumulate as white disks at the interface of 20% and 30% w/v sucrose bands. The centrifuge tube was punctured using a needle-syringe and the white disk of viral particles was collected. These particles were washed five times using PBS by pelleting and resuspending in a fresh batch of PBS after each wash. The virus solution was dialyzed overnight in PBS to reduce any trace quanti-ties of sucrose that could be present.

The purified viral particles were visualized with an electron microscope (EM) (Quanta FEG 650, FEI). The sample was enumerated and characterized for structural studies. The concentration of the viral particles was estimated using TCID50 (Tissue Culture Infectious Dose) and Nano-particle Tracking Analysis (NTA).

Figure 4.3. Micrograph of showing Melbournevirus.

34

Figure 4.4. Characterization of purified Melbournevirus particles. A) Show the results of NTA, the particle concentration and sample purity. The sample was diluted by a factor of 105 before the analysis. B) Shows the cytopathic effect of the viral infection. The healthy amoebae after infection begin to float and shrink in size as the infection progress.

35

4.3 Cedratvirus lurbo Cedratvirus lubro was propagated using Acanthamoeba castellanii cells as described in Paper IV. The host cells were cultured in T 175 flasks using the enriched PPYG medium (PPYG + 100mM Sucrose) (see Section 4.2 for me-dia description). The culture flasks with the host were incubated in a CO2 in-cubator at 32 °C and monitored regularly. When the culture was 70 - 80% confluent, the spent media was discarded and a fresh 8 mL enriched PPYG media along with the virus at MOI 10 was added. The host cells were moni-tored for cytopathic effects.

The viral particles were harvested when most of the host cells were disrupted. The infected media from the culture flasks was collected and spun at 100 g for 10 min to remove cellular debris. The supernatant was spun at 15000 g for 30 min and viral pellet was collected. The pellet was washed and resuspended in PBS. The viral suspension was layered on a 30 % - 70 % w/v sucrose step gradient and centrifuged for 60 min at 5000 g. The viral particles accumulated as white disks in the tube. The centrifuge tube was punctured using a needle-syringe and the white disk of viral particles was collected. These particles were thoroughly washed with PBS buffer by pelleting and resuspending in fresh buffer.

The viral particles were visualized with an electron microscope (EM)

(Quanta FEG 650, FEI). The concentration of the viral particles was estimated using standard limited dilution titration.

Figure 4.5. High angle annular dark field image of the Cedratvirus lurbo after purifica-tion.

36

4.4 Sample Optimization The purified samples had to be further optimized for imaging, depending on the single particle method that was used to determine the structure. These methods need different sample concentrations, buffers and buffer conditions. In case of CryoEM and X-Ray Cryo-Microscopy, the sample is frozen in vit-reous ice on holey grids whereas with SP-CDI, a container-free delivery method is used.

4.4.1 Electron cryo-microscopy In CryoEM, one needs good sample coverage in the holey grids with no pre-ferred particle orientation or sample aggregation to collect a good data set. A FEI Vitrobot Mark 4 was used for the preparation and optimization of the sample grids for imaging. Multiple CryoEM grid types like Quantifoil, C-Flat, etc with various hole:space sizes were tried. These CryoEM grids were cleaned by glow discharging before the sample application. The appropriate sample concentration required for optimal sample coverage for collecting large particle image data set was determined empirically by screening experi-ments. During the sample application onto the grid and blotting of the excess sample; different temperatures, humidity, blotting times and blotting pressures were tried. To test various sample concentrations, the sample had to be ini-tially concentrated.

Bacteriophage PR772 was concentrated to 20 mg/mL by using ultracentrifu-gation as mentioned in Paper II. A solution of cesium chloride prepared with the phage storage buffer at a density of 1.34 g/mL (g/cm-3) was gently layered on top of the sample and centrifuged for 30 min at 100,000 g. The intact viral particles migrated to the top as a fine band. The concentrated viral particles were extracted using a needle and syringe. The concentrated sample was dia-lyzed overnight in the storage buffer to remove cesium chloride. Melbournevi-rus was concentrated by pelleting the particles as mentioned above and resus-pending in a lower volume of PBS buffer. A final concentration of 7 mg/mL was used during grid preparation

Previously for PRD1, a phage from the same family as PR772, it was sug-gested that when the host receptor binding protein, P2 was lost, the phage par-ticles became unstable and spontaneously released the packed DNA when stored in a buffer containing TRIS [66]. So, during the initial grid preparation trail with PR772, two batches of viral particles were prepared. One batch of viral particles stored in a buffer containing TRIS and other with a buffer con-taining HEPES instead of TRIS. The CryoEM particle images from screening these batches showed that no difference in the quality of the intact viral parti-cles (filled particles). Both batches had very high quantity of intact particles.

37

4.4.2 Single Particle - Coherent X-ray Diffractive Imaging The SP-CDI experiment uses a substrate-free sample delivery method. This requires the sample to be delivered as an aerosol to the interaction region free of buffer. So, before the sample is injected into the aerodynamic lens stack, it has to be transferred into a volatile buffer. The volatile buffer commonly used is an ammonium acetate solution. Depending on the injection method, the con-centration of the viral sample and ammonium acetate buffer varies. 250 mM ammonium acetate solution was used as the volatile buffer during the aerosol-ization process by a GDVN. A typical concentration of 1012 particles per mL were used [67]. The sample stability and the size distribution of PR772 in the aerosolized state were analyzed by performing sample injection studies. For the sample stability study, a GDVN was used to aerosolize the sample and for the sample size distribution study, a differential mobility analyzer (DMA) was used to deter-mine the size of the aerosolized particles in flight. The particles from the GDVN were introduced into the aerodynamic lens stack of the injector and collected at the interaction region in the vacuum chamber on a GelPak and formvar/carbon grids (#01754F, F/C 400 mesh Cu, Ted Pella Inc.) for imaging in an EM. The size distribution studies using a DMA was performed using an aerosol generator and a particle classifier as descried in Paper I. The classifier analyses the size distribution of particles in the aerosol.

38

Figure 4.6. PR772 sample optimization for SP-CDI. A) Show the micrograph of PR772 particles collected on an EM grid from the Uppsala injector after the aerosoli-zation process in the GDVN. B) The size distribution of the aerosolized PR772 parti-cles from DMA analysis. The particles appear to be 60 nm, as the actual gas flow during the DMA measurement was lower than the set value. This resulted in a lower estimation of the diameter. adapted from Paper I.

39

4.4.3 X-ray Cryo-Microscopy The amoebal cell infection was carried out as a two-step process to improve synchronous infection by Cedratvirus lurbo while reducing the number of ex-cess virions that were not involved in the infection and available freely float-ing in the media. Viral particles at a high multiplicity of infection (MOI) of 500 were added to 106 cell/mL in a 2 mL Eppendorf tube and incubated for 4 hours on a gentle shaker to allow phagocytosis of the viral particle by amoeba. These infected cells were seeded on to multiple gold grids. These grids were washed with incomplete PPYG medium (lacking glucose) to remove any ex-cess viral particles. Grids were picked at different time intervals, blotted and plunge-frozen in liquid ethane and stored in a custom sample holder at -165oC.

40

5 Structural Studies of Bacteriophage PR772 from Tectiviridae

5.1 Introduction Bacteriophage PR772 was first reported in 1978 as a P-1 incompatibility group R772 plasmid-dependent phage, isolated from a strain of Proteus mira-bilis, a Gram-negative bacterium. The phage also showed lytic activity in E.coli that carried a N or W incompatibility plasmid. These incompatibility plasmids are unstable but they can impart special functions to bacteria like anti-bacterial resistance [68]. Bacteriophage PR772 belongs to the genus Al-phatectiviridae of the Tectiviridae family and exhibits traits similar to other members of the genus. The pseudo T=25 icosahedral viral capsid of PR772 is about 70 nm in diameter with a lipid membrane beneath it. The dormant phage particles are tailless, but produce a membranous tube during the infection of the host. The lipid membrane encapsulates the 14.9 kbp linear dsDNA genome containing 32 recognized ORFs of at least a 40 codon length [69,70]. The genome of PR772 has a high similarity to bacteriophage PRD1(genome sequence identity of 97.2%), a well-studied member of the Tectiviridae fam-ily. On comparing the gene products of PR772 with the previously annotated proteins of PRD1, the probable functions of proteins in PR772 were infered (Refer Table 1) [10,69]. The general architecture of PR772 is similar to PRD1. The icosahedral viral capsid is mostly formed by the major capsid protein, P3. The P3 subunits are arranged as trimers and appear as hexons. Twelve of these hexons form one icosahedral triangular facet bound by the capsid associated protein, P30. Typical pentons are replaced by a vertex complex in 11 of the 12 vertices to complete the pseudo-icosahedral capsid. The vertex complex includes the P31 penton protein, the P5 host-recognition protein and the P2 receptor-binding proteins. Beneath the vertex complex, the infectivity protein, P16, cements the vertex complex to the internal membrane and the surround-ing P3 protein subunits. In PRD1, one of the twelve vertices was shown to contain a unique genome packaging vertex, where proteins P6, P9, P20 and P22 were shown to be involved in DNA packaging and proteins P7, P14, P11, P18, P32 and P34 are responsible for DNA delivery to the host [10,69,71–73].

41

Tabl

e 1:

OR

Fs a

nd th

eir

resp

ectiv

e ge

ne p

rodu

cts i

n Ba

cter

ioph

age

PR77

2 %

Iden

tity

to P

RD

1 10

0 99

99

95

91

10

0 97

99

98

96

91

95

10

0 10

0 10

0 10

0 96

10

0 10

0 10

0 10

0 10

0 97

71

94

96

96

99

77

99

98

99

Ta

ble

adap

ted

from

[69]

Des

crip

tion

Gen

ome

term

inal

pro

tein

D

NA

Pol

ymer

ase

Mur

amid

ase

Rec

epto

r bin

ding

Pent

on p

rote

in

Sp

ike

prot

ein

Ass

embl

y pr

otei

n A

ssem

bly

prot

ein

Min

or c

apsi

d pr

otei

n A

ssem

bly

prot

ein

DN

A p

acka

ging

ATP

ase

D

NA

pac

kagi

ng

Maj

or c

apsid

pro

tein

DN

A p

acka

ging

DN

A d

eliv

ery

DN

A d

eliv

ery

M

inor

cap

sid

prot

ein

DN

A d

eliv

ery

Infe

ctiv

ity

Tran

sglu

cosy

lase

D

NA

del

iver

y

Sing

le-s

trand

ed D

NA

bin

ding

pro

tein

Si

ngle

-stra

nded

DN

A b

indi

ng p

rote

in

Prot

ein

P8

P1

P15

P2

P31 P5

P17

P33

P6

P10

P9

P20

P3

P22

P18

P32

P34

P30

P11

P16

P7

P14

P19

P12

Mol

ecul

ar W

eigh

t (~k

bp)

29.5

63

.3

18.8

63

.66

5.2

13.7

7.

3 34

.5

9.5

7.4

17.4

20

.7

25.8

4.

5 4.

7 43

.5

9.3

5.5

4.4

9.8

5.4

6.7

9.3

22.3

12

.6

27.4

15

.4

12.7

13

.6

10.3

10

.4

16.7

Len

gth

(am

ino

acid

s)

259

553

164

591

44

126

64

340

86

68

166

203

227

42

42

395

84

47

40

90

54

68

84

207

117

269

154

117

128

90

94

160

OR

F 1 2 3 4 5 6 7 8 9 10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

42

5.2 SP-CDI of PR772 The study descried in Paper I was part of the single particle imaging initiative at the Linac Coherent Light Source (LCLS) in Stanford, a concerted effort by the CDI community to understand the challenges and improve the single par-ticle structure determination fundamentals [74]. Bacteriophage PR772 was se-lected as one of the two test sample for imaging studies. Bacteriophage PR772 was aerosolized and delivered into the focus of the LAMP instrument at the Atomic Molecular Optics (AMO) beamline of the LCLS X-ray laser [75,76].

5.2.1 Experimental setup The photon energy was tuned to 1.6 keV delivering 4 mJ into 70 fs duration pulses. Using a pair of Kirkpatrick-Baez (KB) mirrors [77], these pulses were focused into a nominal 1.5 μm2 FWHM region resulting in a nominal power density of ~3.8E18 Wcm‒2 or 1013 photons per pulse. Due to beamline losses and the fluctuations in the SASE process, the actual power density varied by a factor of 10. The photon energy selected was determined by the highest achievable resolution, while remaining well below the silicon absorption edge at 1.8 keV, thus reducing the background caused due to fluorescence of the sil-icon beam-conditioning apertures.

5.2.2 Data Collection and Analysis Diffraction images were recorded using two pairs of pnCCD [78] detectors at a 120 Hz repetition rate. One pair was located 10 cm downstream of the inter-action region (front), and the other pair was located further downstream at 58.1 cm from the interaction region (back). Each pair of pnCCD detectors has two adjacent segments, each with 512 × 1,024 pixels for a total of 1,024 × 1,024 pixels and with a pixel size of 75 μm. The effective resolution of the downstream detector was 11.6 nm at the edges and 8.3 nm in the corners at the wavelength used. The online analysis of the data was performed using Hummingbird [79] and psana [80]. The data were preprocessed using the psana framework and the raw analog pixel data were converted to digital units (analog-digital-units; ADUs) and saved as CXI files. Pixel based gain calibration was done using the silicon k-edge fluorescence flat-field run data set. These gain-calibrated ADUs were thresholded based on an average of 128 ADUs/photon to obtain photon counts using the relation:

43

where is the photon count at pixel i, is the preprocessed, gain-calibrated ADUs measured at pixel i, and γ is the average ADUs per photon for the de-tector calculated from a flat-field run.

The hits were identified using a chi-squared metric [81].

for the jth image Ij was calculated by subtracting a running median back-

ground Bj and normalizing it by the variance of Bj. The diffraction images with the χ2 calculated within the annular area of radii 150 – 400 pixels from the beam center, having a value above 10 were consid-ered as hits. Other alternative hit finding methods like manifold embedding [82] and principle component analysis [83] were also tested.

5.2.3 Results The diffusion map analysis of all the hits helped the identification of single particle hits, multi-particle hits, spherical hits from salt contaminants, back-ground scattering and frames with defective detector readouts. From a total of about 2.9 million frames, we found thousands of good single particle hits. The single particle hits from different hit-finding methods like Diffusion Map Em-bedding, Principle Component Analysis and Manifold Embedding were 12678, 7992 and 37550 respectively.

44

5.2.4 Data Deposition The data were deposited in the Coherent X-ray Imaging Data Bank; CXIDB ID58 in the CXIDB format [84]. In addition to the CXI file, the conversion script, additional metadata files, detector panel calibration files mapping data to real space, configuration files for Hummingbird and psana were also pro-vided along with usage instructions. The deposited data have been used by the CDI community to develop newer and improved techniques for structure stud-ies.

Figure 5.1. Diffusion map analysis of all the hits reduced to two dominant dimen-sions φ1 and φ2. Diffraction patterns seen in the diffusion map of hits: a) single hits and b) multi-particle hits, c) the spherical particles, classified as ‘water’ may also contain contaminant residue from the sample and buffer, d) average background scat-tering of the beamline and e) artifacts in detector readout. reused with permission from Paper I

45

5.3 Electron Cryo-Microscopy of PR772 For the study descried in Paper II, 3 μL of a 7 mg/mL concentrated bacterio-phage PR772 sample solution were added on to a glow-discharged C-Flat grid CF-2/2-2C at 100 % humidity in room temperature. This sample was vitrified in liquid ethane using a Vitrobot Mark IV (ThermoFisher). The data were col-lected on a Titan KRIOS (ThermoFisher) equipped with a K2 Summit (Gatan) direct electron detector and a GIF Quantum LS (Gatan) energy filter. About 3220 movies with 40 frames each and a total dose per movie of 40 e-/Å2 were collected. A magnification of 130k in EFTEM mode was used, resulting in an effective apix of 1.06Å.

5.3.1 Data Analysis The movie files were motion corrected and averaged using MotionCor2 [85] to produce dose weighted and non-dose weighted micrographs. The non-dose weighted micrographs were used for defocus estimation and CTF correction using CTFFIND4 [86]. Defective micrographs with poor CTF fits were dis-carded. The 3D reconstruction was performed using RELION [87]. 2x binned parti-cles images were used for 2D and 3D classification. The best classes from the classification were selected for further processing. The icosahedrally averaged 3D refinement with 2x binned particles images reached Nyquist resolution, so the images were un-binned and the refinement process was continued. The converged map was post-processed to generate a sharpened map. ResMap [88] was used to estimate the local resolution using the two unfiltered half maps. For model building, the map was sectioned and optimized by local sharpening. The asymmetric features of the vertex region were resolved using localized reconstruction [89]. The maps for the penton base were also reproduced using focused classification.

5.3.2 Model Building The icosahedrally averaged map was sectioned in Chimera [90] and this sec-tioned map was optimized by local sharpening using PHENIX: Autosharpen [91]. An atomic model of the asymmetric unit of the PR772 was built using Coot [92] and refined using PHENIX: Real space refinement [93]. The model-map fit was validated with MolProbity [94], Mtriage [95] and EMRinger [96].

46

5.3.3 Results and Discussion The overall resolution of the post-processed map, determined by fourier shell correlation (FSC)@0.143 was 2.75 Å. The local resolution determined by ResMap for the capsid region, vertex region and the regions interacting with the inner lipid membrane was 2.3 Å, 2.3 Å – 3.0 Å and 3.2 Å respectively.

The overall architecture of the capsid in PR772 is similar to the type species of Tectiviridae, bacteriophage PRD1 [15] but with subtle variations. The ma-jor capsid protein, one of the P3 subunits in the asymmetric unit of PR772 shows a new long helix with a kink at the N-terminal region that was not

Figure 5.2. Local resolution at different region of the map. A) The local resolution estimate of the CryoEM map from ResMap. The map shows the distribution of reso-lution in different regions. Most of the capsid that was used for model building is resolved at 2.3 Å. B) Quality of the map at the core of the capsid protein P3 (Chain B, residues 162-173) where the local resolution estimate is 2.3 Å. C) Quality of the map close to the membrane (P3 Chain B, residues 18-35) where the resolution is estimated at 3.2 Å. D) Quality of the map close to the five-fold vertex of the icosahedral viral particle. adapted from PAPER II

47

previously described. This conformation could play an important role in ac-commodating the P30 “tape protein” and thus enabling the viral capsid assem-bly. The C-terminal region of P3 also shows an interlocking mechanism with the P30 tape protein. P30 holds the adjacent facet together and maintain the size and the shape of the capsid. This could explain the previous observations in PRD1 with a P30 deletion mutant, where incomplete particles were pro-duced and the membrane vesicles formed were smaller compared to the wild type [16].

The structure and composition of the vertex complex was resolved by local-ized reconstruction and focused classification. The penton base was revealed to be a heteropentamer with three copies of P5 and two copies of P31. This explains some of the previous observations made in studies of PRD1 [18,20,66,97]. The beta-propeller region of P2 interacts with the N-terminal domain of P5 and is stabilized by the P5 stalk region. The electrostatic surface potential and hydrophobicity analysis around the binding region were found to be favorable for and promoted the interaction. At low occupancy, the P5 N-terminal region with the 8 residues closest to the N-terminal showed an alter-nate conformation in which these residues are wedged between the two adja-cent P3 subunits. In the other, more prominent conformation the N-terminal would hug the adjacent subunit of the penton base. The P2 occupancy at the vertex was also found to be lower when compared to PRD1, but this observa-tion was in agreement with previous results from quantitative western blots [14]. The lower occupancy of P2 could be related to the alternate conformation of P5 N-terminal residues.

Figure 5.3. Shows the different N-terminal conformations of the P3 monomer. The N-terminal conformations of P3 in A), B) and C) represent the helix turn helix, the long helix and the long helix with a kink respectively. The newly discovered confor-mation of the N-terminal conformations of P3 can be seen in (C). adapted from PA-PER II

48

The infectivity protein, P16 forms a clamp-like complex and locks the adja-cent P3 subunit from the neighboring facet. P16 along with the adjacent P3 subunits also form a complex with P30. These interactions could be helpful in stabilizing the capsid and the vertex region as previously predicted using PRD1 P16 deletion mutants [21].

Figure 5.4. Vertex Assembly. A) top view of the penton map generated from the lo-calized asymmetric reconstruction the yellow arrows show the lack of stalk in 2 of the 3 subunits of the penton. B) The structure of P31 (shown in yellow) and P5 (shown in pink, residues 1-124) are superimposed and yellow arrow points to the region where P31 terminates and P5 (residues 121-124) continues upward. C) the map-model fit with two copies of P31 and three copies of P5. D) Model of P2 (PDB:1N7U) fitted into the isolated CryoEM density from the localized reconstruction. E) The vertex complex with the heteropentameric penton and the P2 protein. adapted from PAPER II

49

Based on these observations, we propose a model of the vertex complex as shown in Figure 5.6.

Figure 5.5. Shows the P3-P30-P16 complex, P16 (shown in blue) locks the two adja-cent P3 trimers close to the penton (shown in green) and the P30 protein (shown in orange). adapted from PAPER II

Figure 5.6. Proposed model for the vertex region. Shows the structural arrangement of P5 (P5 N-terminal base shown in bright pink, Collagen-like motif of P5 shown in bright green and P5 Knob is shown in pale blue), P31 (yellow) and P2 (orange) that form the vertex. The region below the vertex complex (shown as a grey density) is the unorganised/unmodeled region that links the surrounding P16 (blue) and the P30 (black). Capsid is shown in pale green (raw CryoEM map is labelled as Capsid and the modelled region is labelled as P3 Trimer). The lipid membrane is shown in pur-ple. adapted from PAPER II

50

6 Structural Studies of Giant Viruses

6.1 Structural studies of Melbournevirus Melbournevirus belonging to the Marseilleviridae, has a genome size of 346kbp. It has an icosahedral particle with a diameter of 230 nm. Unlike other viruses of the Marseilleviridae family, Melbournevirus does not show fibers on the protein capsid. During the viral replication the host nucleus appears to remain intact, suggesting that the replication of the virus happens in the cyto-plasm of the host. Viral assembly is performed in multiple viral factories. About 403 ORFs have been identified, occupying about 86.7% of the genome. Similar to Tectiviridae, Melbournevirus also has an internal membrane under the capsid [34].

In Paper III, we study the protein composition of the packed viral particle and the basic structure of the virus.

6.1.1 Identification of Viral Proteins in the Melbournevirus A typical SDS-PAGE was performed to assess the prominently available pro-tein in the virion. To identify structural proteins, tandem mass spectrometry was performed. About 500 μL of the viral sample with a concentration of 1012 particles/mL, was pelleted and the supernatant was discarded. The pellet was treated with 9 M urea buffer and sonicated to extract the proteins. The sample was sent for tandem mass spectrometry at Mass spectrometry-based prote-omics facility at Uppsala University. The database searches were performed by SEQUEST [98] against proteins from Melbournevirus and Acanthamoeba castellanii. The protein identification criterion was set to include only matches where at least two matching peptides of 95% confidence level per protein.

6.1.2 Results The SDS page revealed 10 prominently expressed protein where we could po-tentially identify the conserved MCP. Using tandem mass spectrometry, about 121 proteins were identified based on the protein identification criterion. The prominent proteins were the major capsid protein, histone-like protein, puta-tive viral enzymes, some membrane proteins and functionally uncharacterized proteins.

51

6.1.3 Electron Cryo-Microscopy For the study descried in Paper III, about 5 μL of the 1E12 particles/mL Mel-bournevirus sample was added on to a glow-discharged Quantifoil grid R1.2/1.3 at 100 % humidity in room temperature. This sample was vitri-fied in liquid ethane using a Vitrobot Mark IV (ThermoFisher). For tomogra-phy, the grid was prepared similarly to what has been described above but the sample was mixed with equal volumes of 15nm gold fiducials. The data were collected on a JEM2200FS (JEOL Ltd. Japan) equipped with a TemCam-F415MP (TVIPS GmbH) charged-coupled device and an omega-type energy filter. A magnification of 45k, resulting in an effective apix of 3.31Å was used for single particle analysis. For tomography, a lower magnification of 24K with apix of 6.62Å was used. The sample was tilted from 70o to -70o with 2o angular steps.

6.1.4 Data Analysis EMAN2 [99] was used for the 3d reconstructing of Melbournevirus by single particle analysis. EMAN2: e2boxer.py was used to pick 7005 particle images from 800 micrographs and ctf correction was performed by using e2ctf.py. The particles were 2x binned for initial processing. 5 random seeded initial models were produced by imposing icosahedral symmetry. A total of 7005 good particle images were used to reconstruct a 35 Å FSC@0.5 resolution map. This map was 2x scaled and served as an initial model for further refine-ment. The un-binned images of the 7005 particles were used to reconstruct a 26.3 Å FSC@0.5 resolution map.

IMOD [100] was used for tomography reconstruction of Melbournevirus. A total of 147 reconstructed tomograms from IMOD were loaded into EMAN2 and post processed. The tomograms were aligned to the SPA model and sym-metrically rotated so that the large dense bodies seen in the tomograms were also aligned with one another. By averaging all the aligned sub-tomograms an averaged model was generated.

6.1.5 Results The electron micrographs revealed the presence of a large dense body (LDB) in the viral particle. The LDB was about 30 nm in diameter with an approxi-mate density of 1.6 g/cm3 and the rest of the interior regions had a density varying between 1.10–1.30 g/cm3. The density of LDB is similar to the density of a nucleic acid-protein complex. With the presence of histone-like proteins in the virus, we hypothesize that the LDB could be a complex of nucleic acids and protein.

52

The 3D reconstruction from CryoEM revealed a capsid with T=309 (h=7 and k=13) and about 84 Å in thickness. The internal membrane was about 30Å thick but when present close to the vertex region the membrane appeared to be layered. The tomographic analysis showed that most of the LDBs were present close to the vertex region below the membrane. In a few of the cases, the LDB was also present at the center of the viral particle.

6.2 Structural studies of Cedratvirus lurbo Cedratvirus lubro is a newly discovered giant virus, from a soil sample col-lected from the banks of a stream southwest of Uppsala, Sweden. Very little is known about the virus. Based on the size and morphology, we assume that the newly isolated virus particle belongs to the Pithoviridae family. The viral particle shows striking resemblance to Cedratvirus [39,40]. In Paper IV, we investigate the infection dynamics of a giant DNA virus using quantitative soft X-ray microscopy.

Figure 6.1. A) A micrograph showing the Melbournevirus. The white arrows point to the newly discovered LDB in the viral particle. The presence of histone-like protein and the density analysis suggests that LDB could be a nucleic acid – protein complex. B) The T=309 (h,k:7,13) capsid lattices seen in Melbournevirus. The black scale bars indicate 50nm. adapted from PAPER III

53

6.2.1 Isolation About 30 g of the soil sample was suspended in 100 mL PBS supplemented with 100 ug/mL of Amphotericin B. The mixture was incubated for 24 hrs at 4 oC with gentle stirring. After incubation, the suspension was allowed to settle for 1 hr to remove the large soil particulates. About 10 mL of the supernatant was pipetted and sequentially spun for 10 mins at increasing RCF of 500 g, 1000 g, 4000 g, 10000 g and 15000 g. After each step of spinning the pellet was collected and the supernatant was subjected to the next higher RCF. The pellets collected after different RCF runs were resuspended in PBS supple-mented with 25 mg/mL amphotericin B, 50 µg/mL Ampicillin, 100 units Pen-icillin and 100 µg/mL of Streptomycin. Each of these pellet suspensions was serially diluted with PBS supplemented with above mentioned supplements.

Acanthamoeba castellanii host cells that were adapted to 2.5 mg/mL Ampho-tericin B were grown in 24 well plates in PPYG medium (2.0 % w/v Proteose peptone, 0.1 % w/v Yeast extract, 4 mM MgSO4, 0.4 mM CaCl2, 0.05 mM Fe(NH4)2.(SO4)2, 2.5 mM Na2HPO4, 2.5 mM KH2PO4, pH 6.5) supplemented

Figure 6.2. A micrograph of Cedratvirus lurbo showing a morphology similar to Pi-thovirus. The dual apical cork a unique feature of the genus Cedratvirus. It also shows the outer hair, the tegument and the lipid membrane similar to Pithovirus. The scale bar represents 500 nm. adapted from PAPER IV

54

with 25 mg/mL amphotericin B, 50 µg/mL Ampicillin, 100 units Penicillin and 100 µg/mL of Streptomycin. Each of the serially diluted suspensions were added into individual wells of the 24 well plate with the Acanthamoeba cas-tellanii host. These plates were incubated at 28 oC and observed for any cyto-pathic effects. About 200 µL of the culture media from the wells in which the host cells showed signs of infection was pipetted out and serially diluted. These were re-plated on a new set of 24 well plates with the host to confirm the infection. End point dilution was performed to clone the virus.

6.2.2 X-ray Cryo-Microscopy Imaging was performed using the X-ray Microscope at Stockholm Laboratory as described in Paper IV. It consists of a liquid-nitrogen-jet laser-plasma source, a normal-incidence multilayer condenser mirror (58 mm diameter Cr/V), circular central stop before the sample, a cryo sample stage (modified TEM goniometer stage), a Ni zone plate, and a back-illuminated CCD detector (iKon-L, Andor). It operates in the water window. The liquid- jet source is powered by a 40-60 W 2 kHz 600 ps diode-pumped Nd:YAG slab laser to produce narrow-bandwidth λ = 2.48 nm line emission. A 200 μm diameter Ni zone plate with an outer zone width of 30 nm was used for focusing on to the CCD. The data were collected from the grids frozen at different time intervals after infection along with controls. A 1s short exposure scan was performed on the grids to find a region of interest; then a long 60s exposure image of the region of interest was collected. These images were coupled to the control experi-ment.

6.2.2.1 Data Analysis The images collected were Gaussian filtered to smoothen the hard edges in the image. About 51 images were selected representing different time intervals after infection of which about 4-6 images were acquired at each of the consid-ered intervals. As an initial measure, the number of viral particles present in the cells at each time point were individually counted using minimum length, maximum length and minimum ellipticity parameters. Also, all high contrast objects were se-lected and analyzed using Matlab. Virions were sorted from other high con-trast objects by using the minimum length, maximum length and minimum ellipticity parameters. During error analysis, the incorrectly sorted other high contrast objects and the viral particles that do not fit the defined parameters due to their orientation were considered.

55

By calculating the absorbance of X-rays by the viral particles present in the cell to the absorbance of X-rays by the whole cell at a given instance, the carbon conversion from the host to the virus can be estimated. As the illumi-nation varied from image to image, the illumination background was esti-mated using X-ray images with a large open area. This was used to normalize the images for estimating the X-ray absorption. Two estimates for X-ray ab-sorption by viral particle were calculated. One estimate assumes that all the viruses have similar absorption, which can be calculated by imaging an indi-vidual virus particle. So, by counting the number of viral particles present in the cell and multiplying with the X-ray absorption, the value from a single particle will yield the total absorption by the enumerated viral particles. In the second estimate, actual X-ray absorption in the regions marked as viral parti-cles is calculated. Here, the calculated X-ray absorption also includes the ab-sorption by the cell in which the viral particles are multiplying. Both these estimates are compared to the absorption by the whole cell.

6.2.3 Results and Discussion By comparing the images of a typical healthy Acanthamoeba to a Lurbovirus-infected amoeba at different time intervals, the progression of viral infection and viral replication in the host cell could be assessed. Uninfected amoebae are attached to the surface of the culture flask/grid and appear flat with multi-ple vacuoles and a prominent nucleus with a nuclear envelope. As the viral infection progresses, many changes were observed. From 6 h post-infection (PI) to 15 h PI, there is an increase in the dense circular objects ob-served in the cytoplasm. These circular objects could be mitochondria or car-bon-rich storage granules. The viral infection could alter/direct the host cell into producing these carbon-rich objects as a preparation for viral replication. At 18 h PI, virions start to appear in the cell and accumulate as the time pro-gresses. The accumulation of the viral particles is localized but there appear to be multiple regions of the host cell where the localization could occur. This could possibly indicate multiple viral factories. Beyond 18 h PI, the number of vacuoles in the host cell is also reduced but the nucleus appears intact till the 54 h PI. By comparing the bioconversion from two different estimates, the studies show a significant bioconversion by 8% - 18% from the host cell to the virus during later stages of infection.

56

Figure 6.3. Virus infection series visualized using laboratory cryo X-ray microscopy from non-infected to 70 h PI. The non-infected amoeba (A) is included for comparison and has been treated identically as the infected ones aside from not adding the virus. The infection times in the displayed images (B-I) are 6 h, 12 h, 15 h, 18 h, 24 h, 30 h, 54 h and 70 h, respectively. In the images up to 15 h PI (A-D) no distinguishable virions were detected. (E) At 18 h PI, a few virions can be seen. At longer infection times (F-I) the virions take up an increasing part of the cell volume, as well as the surrounding medium, until they appear densely packed throughout the cell. (E-G). The vacuolated structure of the healthy cells seems to disappear as the production of virions takes off, while the cell nucleus can be seen in all images up to 54 h PI (A-H). Scale bar is 10 µm and valid for all images. reproduced from PAPER IV

57

7 Summary and Outlook

7.1 Achievements SP-CDI is a new and rapidly developing method for structural studies of bio-logical entities. The use of femto-second X-ray pulses provides an opportunity to collect diffraction image of biological samples before they are affected by significant damage and improves the temporal resolution. The container free sample delivery system together with the quick feedback from the online mon-itoring data collection software developed, enables us to manipulate the bio-logical sample in real time and observe the structural changes that occur. The data generated in Paper I has enabled the CDI community to better understand the requirements of single particle imaging, as a method for structural studies of biological samples. The large data set generated provides better statistic about the sample quality after using the GDVN container-free sample delivery method, the number and the quality of diffraction images that are needed to produce a 3D structure of the sample to a given resolution. The deposited data has been used to develop and test new methods for particle size determination, hit finding, particle classification/sorting, 3D structure determination, etc (Pa-per V, [101]). For a method that was developed in the last 10 years, the results from SP-CDI show the potential for break-through science and developments are underway to fulfill this promise. The recent “Resolution Revolution” in CryoEM as propelled the technique as the most preferred single particle method for structure determination of bio-logical sample. In Paper II, the high-resolution structure of bacteriophage PR772 confirmed the speculation about the heteropentameric nature of the penton and revealed the new architecture of the vertex complex consisting of a heteropentameric penton formed with three copies of P5 and two copies of P31. The P2 protein is bound to the N-terminal domain of P5 and the orienta-tion is different compared to what was previously described in the case of PRD1. The beta propeller region of P2, formed by domains I and II is bound to the N-terminal domain of P5. The structure also reveals new conformations of N-terminal and C-terminal region of P3 which play an important role in particle assembly and structural stability. With a low copy number of P2 in PR772 and the high infectious titer, we propose that the membranous tube that is used for infection is formed at the vertex that binds to the host and that these tubes are used for injecting the viral DNA during infection.

58

Paper III describes the proteins present in the packed Melbournevirus and the first 3D CryoEM structure of a viral particle from the Marseilleviridae family. The structure shows a T=309 capsid with an inner lipid membrane similar to phages from the Tectiviridae family. The inner membrane shows a unique bulge formed by layering, at the vertex which was not previously known. A dense body was found in the viral particle, a feature not observed in other viruses of the Marseilleviridae family. The density of this body in the viral particle is similar to a nucleic acid-protein complex. This observation, along with the identification of histone-like protein during the protein identification study, could suggest some genome organization in the viral particle, similar to higher organisms. The tomographic reconstruction also reveals that the dense body is found close to the capsid near the vertex. In Paper IV, we could study the progression of a viral infection and the rep-lication process in the host cell close to their native state without using sample modifying processes like chemical fixation, staining, sample dehydration or resin/polymer embedding. The study also showed that there is a significant bioconversion from the host cell to the viral particle during later stages of in-fection.

7.2 Challenges and Opportunities Electron microscopy is limited in scattered flux per exposure and in temporal resolution. Electrons have relatively short penetration depth, and thick sam-ples pose problems. Electron cryo-microscopy has made huge advances since 2012 by improving the signal to noise ratio of the images by using fast direct detection cameras. The frame rates of the cameras allow computational cor-rection of vibrations, diffusion and beam-induced specimen movement, and have enabled determination of structures around 2-4 Å resolution for particles from 100 KDa to several MDa mass. The very small field of view at high-resolution reduces the throughput in studies on big objects. Less than 10 par-ticles will fill the field with 200 nm objects, and collecting a million images on such particles takes significant time. However, the throughput increases significantly with smaller objects where electron cryo-microscopy shows its real strength. The time resolution of cryo-EM is limited to milliseconds due to the speed of data collection and time needed to freeze samples. Freezing is important but the vitreous ice adds to background. Space charge is detrimental to imaging with electrons, and as a consequence, "diffraction before destruc-tion" experiments are not feasible in an electron microscope. Ultra-fast diffractive imaging with coherent X-ray pulses from FELs is new and has a short history. The pulses can give femtosecond/attosecond time

59

resolution. The penetration depth of X-rays permits studies on very big ob-jects. Phases need to be determined from the data and this is not always trivial. Damage can be more or less eliminated in the image by "diffraction before destruction". X-ray lasers can produce millions of exposures in a single day, and can do so on any object that fits into the focus.

Soft X-rays used in cryo-microscopy provides better penetration depth com-pared to electrons accelerated by a typical 200-300kV microscope. The longer wavelength of X-rays from an in-house microscope and the X-ray focusing optics severely limit the resolution. The current detector technology to gener-ate real space image from an X-ray cryo-microscopy are not as well developed as fast direct detection cameras used in electron cryo-microscopy. Similar to electron cryo-microscopy, we see radiation damage accumulate over time dur-ing imaging, leading to reducing in resolution. These single particle methods have tremendous potentials but none are per-fect. A synthesis of ideas and data could bring significant benefits to structural sciences. There is room for more than one avenue to be explored, and different techniques will be appropriate for different applications. Such flexibility is intrinsic to the nature of research and essential for scientific progress.

SP-CDI is a new and rapidly developing method for structural studies of bio-logical samples. As a new method, there are several areas where SP-CDI could improve significantly. Currently, there are very few FEL beamlines around the world. This with the immense competition for access to these beamlines reduces the opportunity to perform an experiment. The resolution attainable for 3D reconstruction is low due to the low signal to noise ratio in the images produced. The low repetition rate of X-ray pulses at LCLS, where the data for Paper I were collected, also reduces the number of diffraction images that could be collected. From the diffusion maps of hits shown in Paper I, we see that the use of GDVNs as a sample delivery method produces wide range of droplet sizes, resulting in a broad size distribution of objects at the interaction region. This also increases the sample volume need when compared other sin-gle particle methods. Some of these issues have been alleviated with the advent of electrospray sam-ple injection method (Paper X), another container free sample delivery method, that produces smaller droplets and also reduces sample consumption. The high-intensity pulses of the recently commissioned European XFEL [102], could improve the signal to noise ratio in the diffraction images and the higher repetition rate of X-ray pulses generated by the beam line helps in the collection of larger data sets for analysis.

60

In Paper II, bacteriophage PR772 was resolved to a high resolution. The abil-ity to manipulate individual images for 3D reconstruction also helped us in resolving the structure and composition of the heteropentameric penton and the vertex complex. The low copy number of P2 and flexible C-terminal do-mains of P5 limited the resolution of vertex complex but the use of an un-masked localized reconstruction method ascertains the correct orientation of P2 with respect to the penton. In PRD1, a unique packaging vertex is found in all the packed virions. It appears as a cylinder with a 20-23 nm diameter and extending 25 nm into the virion from the outer surface of the capsid. In PR772, this prominent feature was not visible in the 2D or 3D classes. It was also not found in the localized asymmetric reconstruction of the vertex sub-particles or the whole particle symmetry relaxed reconstruction. A symmetry expanded approach along with focused classification could shed more light onto resolv-ing this feature but this also entails more computational resources. In Paper III, the size of the Melbournevirus made it significantly challenging to use CryoEM for structure determination. The grid preparation for the large sample was also difficult. The need for particle images at higher magnifica-tion, to reach a higher resolution 3D reconstruction, significantly reduced the number of particles per micrograph to 3-4. The large particles also exponen-tially increase the computational resources needed for data analysis, 3D re-construction and structure validation. Due to the large size, one would also need an electron cryo-microscope with higher electron acceleration voltage and specific corrections to the defocus and ctf estimation during image pro-cessing will be needed to compensate for the thickness of the particle. In Paper IV, the high penetration energy of the X-ray in the water window, along with better contrast due to X-ray absorption by the carbon-rich sample, helped visualize large viral particle and follow the viral propagation in the cell without the need of staining or sectioning. Currently, the main drawback is that the resolution is significantly lower compared to other single particle methods. It was hard to resolve viral assembly intermediates. The 200 µm diameter Ni zone plate objective with a 30 nm outer zone width that was used during the imaging process limits the resolution together with other parame-ters. Newer and improved fabrication methods for zone plates could improve the resolution in future [103].

61

Summary in Swedish

Redan i början av 1900-talet visste man att DNA och proteiner var essentiella be-ståndsdelar hos levande organismer och även deras funktioner var till viss del kända. Det som däremot inte var känt var hur dessa biomolekyler utförde sina funktioner i organismerna. Det blev därför nödvändigt att försöka under-söka de biologiska molekylernas strukturer för att förstå hur de kunde utföra sina funktioner och hur de interagerade med varandra. Som ordspråket säger: "En bild säger mer än tusen ord". Denna strävan har lett till en utveckling av många olika tekniker för att avbilda biologiska molekyler, där de olika meto-derna har sina för och nackdelar. Tidigt under strukturbestämningens historia trodde man att dessa biomoleky-ler hade en rigid struktur precis som många andra mer välstuderade kemiska föreningar. Grunden för strukturbestämning av biologiska molekyler finns nämligen i tidigare studier av saltkristaller med hjälp röntgendiffraktion. När röntgenstrålar interagerar med olika saltkristaller produceras olika diffrakt-ionsmönster och de mönster som produceras är relaterade till hur atomerna är arrangerade i saltmolekylen och i sin tur hur dessa saltmolekyler är arrange-rade i kristallen. Det visade sig senare att även proteiner kunde kristalliseras på samma sätt som salter. Vetskapen om att proteiner kunde kristalliseras, till-sammans med tidigare kunskap om röntgenkristalldiffraktion ligger till grund för en av de mest framgångsrika och mest använda strukturbestäm-ningsme-toderna, proteinröntgenkristallografi. När allt fler proteinstrukturer studerades visade det sig att det verkligen fanns en stark korrelation mellan strukturen hos proteinerna och den funktion de hade i organ-ismen. Till skillnad från den tidigare uppfattningen att biomole-kylerna har en rigid struktur, indikerade dessa strukturstudier att de flesta bi-ologiskt aktiva molekyler hade strukturer som var dynamiska och uppvisade olika konformationer i olika stadier av deras funktion. För att kunna studera detta krävdes en utveckling av nya strukturbestämningsmetoder för att fånga så många viktiga konformationer som möjligt. Ett exempel är magnetisk re-sonansspektroskopi (NMR), som har har utvecklats för att studera strukturerna hos biologiska molekyler i lösning. Denna me-tod har varit mycket fram-gångsrik för studier av dynamiken hos små biologiska molekyler, men den blir allt mer beräkningsintensiv när storleken på den biologiska molekylen ökar.

62

Under slutet av 1900-talet etablerades strukturbestämningstekniker som elektron kryo-mikroskopi, röntgen kryo-mikroskopi och koherent diffrakt-ionsavbildning av enstaka partiklar. Dessa tekniker har visat en stor potential i att hantera den biolo-giska molekylens dynamiska natur. Bilder av de en-skilda proteinerna kan sorteras med avseende på deras konformation och strukturen för de olika konformationer som ett protein kan anta kan studeras individuellt. Metoderna visar också stor potential för hantering av stora och komplexa biologiska molekyler som membranproteiner och virus, vilka är svåra att kristallisera. De ovan nämnda metoderna för att avbilda enstaka partiklar har även möjlig-gjort studier av stora virus som bakteriofag PR772, Melbournevirus och Ced-ratvirus lurbo med en diameter på respektive 70 nm, 230nm och 1,2um. Ge-mensamt för dessa virus är att de har ett dubbelsträngat DNA-genom och ett lipidmembran under proteinskalet. Bakteriofag PR772 och Melbournevirus har en liknande morfologi. De har en ikosahedral struktur och uppvisar en strukturell likhet i kapsiden där det huvudsakliga kapsidproteinet organiserar sig i ett hexonliknande arrangemang. Cerdatvirus spp har en helt annan mor-fologi jämfört med de tidigare nämnda virusen. Det är äggformat med en kork-liknande struktur i vardera kortända. Den har också ett tjockt tegument (skal) med hårliknande fibrer på ytan. Trots skillnaderna i morfologi så finns det ändå vissa likheter mellan Cedratvirus och Melbournevirus på genetisk nivå. Med den ökande förekomsten av antibiotikaresistens har intresset för bakteri-ofagterapi (fagterapi) fått en renässans. Fagterapi innebär att man administre-rar en cocktail av bakteriofager (bakterievirus, ofarliga för människor och djur) till den som drabbats av en bakterieinfektion. Det har visat sig att fagte-rapi kan vara ett bra alternativ till antibiotika för att kontrollera mikrobiella infektioner som lunginflammation, septisk chock, matförgiftning osv. Fagte-rapi har visats kunna bekämpa P. aeruginosa, S. aureus, A. baumannii, E. coli etc. som orsakar olika infektioner hos både människor och djur. Fagcocktails kan även användas som alternativ till starka kemiska steriliseringsmedel som vanligtvis används i kliniker och på sjukhus. Med detta i åtanke är det viktigt att försöka förstå bakteriofagernas livscykel, infektions-process och deras för-måga att infektera en specifik värd. PR772 och andra Tectiviridae är fager som infekterar flera olika värdar och som är lämpliga att ingå i potentiella fagcock-tails. Både Melbournevirus och Cedratvirus utmanar de för närvarande accepterade dogmerna för vad som kännetecknar ett typiskt virus. De motsäger de accep-terade principerna såsom virusens minimalistiska natur och brist på genom-komplexitet. Melbournevirus från Marseilleviridae uttrycker histonliknande proteiner och härbärgerar också en stor struktur med en densitet som indikerar att det är ett nukleinsyraprote-inkomplex. Detta indikerar i sin tur att det kan

63

finnas någon typ av genomorganisation i den virala partikeln. En sådan kom-plex organisation av ett virusgenom har aldrig tidigare rapporterats. Melbour-nevirus kodar också för ett stort antal proteiner med okänd funktion eftersom generna som kodar för dem saknar motsvarighet i någon annan studerad org-anisms genom. Cedratvirus lurbo, ett av de största virus som upptäckts i Nor-den, isolerades från strandbanken vid en å söder om Uppsala. Det liknar Pit-hovirus, ett 30 000 år gammalt zombievirus som återupplivats från den sibi-riska permafrosten. Även hos Cedratvirus är funktionerna för de flesta prote-iner som kodas av genomet okända. För närvarande finns det många fler frågor än svar. Varför är Cedratvirus så likt det 30 000 år gamla viruset? Vilken funktion har de proteiner som kodas av de gener som saknar motsvarighet i andra organismers genom? Hur effektiva är de i sin funktion? Endast vidare studier kan ge svar på dessa frågor.

64

Author Contribution

Paper I I analyzed and characterized various samples that were selected as potential candidates for the single particle imaging (SPI) initiative. This helped in se-lecting the two best candidates for the SPI experiments. I mainly prepared, characterized and optimized bacteriophage PR772 for all the experiments that were part of the single particle imaging initiative. I also supported the sample injection team before, during and after the experiments. Paper II I took part in planning the experiment. I prepared and optimized bacterio-phage PR772 for the CryoEM data collection. I analyzed the data collected and reconstructed the high-resolution map of PR772 along with the vertex re-gion. I also built the model of the asymmetric unit of the virus. I wrote the manuscript. Paper III I prepared the Melbournevirus sample and took part in the proteomics study. Paper IV I isolated the new Cedratvirus lurbo from the soil sample. I maintained the host cells and viral cultures for the experiment.

65

Acknowledgments

This work was possible, thanks to the multi-disciplinary teams of enthusiastic and inspired individuals who are constantly pushing the boundaries of knowledge. I have greatly benefitted from their guidance and support in car-rying out this work. In particular I thank: Martin Svenda and Janos Hajdu for their support and encouragement as supervisors. I am grateful for many opportunities they have provided and the knowledge they shared over the years to improve myself. I thank Inger Andersson for being available and sorting all the issues whether it was related to lab work or otherwise. You will always be the inspiration and the role model to whom I look up to. I thank Kenta Okamoto and Daniel Larsson for all the help you have provided, be it sample preparation or cry-oem. You have taught me hard work pays-off. I thank Dirk Hasse for intro-ducing me to a great team of people from molecular biophysics and always helping me with wet lab related queries. I thank Filipe Maia for the general guidance and all the help provided for working with Davinci and fixing related issues even on weekends. I thank Marvin Seibert for providing me with teaching opportunity. I thank Margareta Ingelman for helping us maintain the lab. I am thankful to Anna Munke for sharing the burden of sample han-dling and characterization for the precious beamtime at LCLS. I thank Karin Valegard, Gunilla Carlsson, Anna Larsson, and Laura Gunn for their in-sightful feedback and discussions about the wet lab. I am thankful to Max Hantke, Benedikt Daurer, Gijs van der Schot, Ida Lundholm, Jonas Sell-berg, Tomas Ekeberg, Carl Nettelblad and Alberto Pietrini for their dis-cussion about imaging analysis and the challenges with 3D reconstruction. I thank Nicusor Timneanu for the calm, joyful and lively lunch room discus-sions which helped me in keeping my spirits and morale high. Johan Bielecki, Daniel Westphal, Olena Kulyk, Kerstin Muehlig and Federico Benzi; I am thankful for the discussions about sample injection and sample optimizations methods for substrate free injection. I like to thank Anna Pabis, Ana Giraldo, Ricardo Ferreira and Saurav Halder for making lab life interesting.

I am grateful to Marta Carroni for all the help, support and feedback she provided with cryoem sample optimization and data collection.

66

I am thankful to Naoyuki Miyazaki and Kazuyoshi Murata for their collab-orations in many projects and providing the sample cryoem data set to test the feasibility of cryoem reconstruction of PR772. I thank Jean-Michel Claverie and Chantal Abergel for inspiring me to isolate new giant viruses. I thank Hans Hertz and Mikael Kördel for the fruitful collaboration and introducing me to soft X-ray microscopy.

I am grateful to my parents, Vanajakshi and Narayana Reddy and my sister, Shruthi for the encouragement and support they have provided. I am thankful to my friends Lisa, Sanket, Evelyn and Vimal who made Uppsala, a home away from home. This encouraged me to pursue a PhD in Uppsala. I am thank-ful to Sandesh, Yashraj, Pooja, Prashanth, Punya, Smrithi, Malavika, Phani for being my family in Uppsala. Lastly, I am thankful to the funding agencies ERC, VR, KAW and Röntgen-Ångström Cluster for their assistance, without which this work would not be possible.

67

Reference

1. Microbiology by numbers. Nat Rev Microbiol. 2011 Sep 12;9(9):628–628. 2. Aherfi S, Colson P, La Scola B, Raoult D. Giant Viruses of Amoebas: An

Update. Front Microbiol. 2016 Mar 22;7:349. 3. Fischer MG, Allen MJ, Wilson WH, Suttle CA. Giant virus with a remarkable

complement of genes infects marine zooplankton. Proc Natl Acad Sci U S A. 2010 Nov 9;107(45):19508–13.

4. Baltimore D. Expression of animal virus genomes. Bacteriol Rev. 1971 Sep;35(3):235–41.

5. Lederberg EM. Lysogenicity in Escherichia coli strain K-12. Vol. 1, Microb. Genet. Bull. 1950. 5–9 p.

6. Delbruck M. the Growth of Bacteriophage and Lysis of the Host. J Gen Physiol. 1940 May 20;23(5):643–60.

7. Jalasvuori M, Koskinen K. Extending the hosts of Tectiviridae into four additional genera of Gram-positive bacteria and more diverse Bacillus species. Virology. 2018 May;518:136–42.

8. Tectiviridae. In: Virus Taxonomy. Elsevier; 2012. p. 317–21. 9. ICTV. Poxviridae - dsDNA Viruses - dsDNA Viruses (2011) - International

Committee on Taxonomy of Viruses (ICTV) [Internet]. ICTV 9th Report. 2011. Available from: https://talk.ictvonline.org/ictv-reports/ictv_9th_ report/dsdna-viruses-2011/w/dsdna_viruses/74/poxviridae

10. Butcher SJ, Manole V, Karhu NJ. Lipid-containing viruses: Bacteriophage PRD1 assembly. Rossmann MG, Rao VB, editors. Adv Exp Med Biol. 2012;726:365–77.

11. Philippe C, Krupovic M, Jaomanjaka F, Claisse O, Petrel M, le Marrec C, et al. Bacteriophage GC1, a Novel Tectivirus Infecting Gluconobacter Cerinus, an Acetic Acid Bacterium Associated with Wine-Making. Viruses. 2018 Jan 16;10(1):39.

12. Mindich L, Bamford D, McGraw T, Mackenzie G. Assembly of bacteriophage PRD1: particle formation with wild-type and mutant viruses. J Virol. 1982 Dec;44(3):1021–30.

13. Gowen B, Bamford JKH, Bamford DH, Fuller SD. The tailless icosahedral membrane virus PRD1 localizes the proteins involved in genome packaging and injection at a unique vertex. J Virol. 2003 Jul;77(14):7863–71.

14. Saren A-M, Ravantti JJ, Benson SD, Burnett RM, Paulin L, Bamford DH, et al. A snapshot of viral evolution from genome analysis of the tectiviridae family. J Mol Biol. 2005 Jul 15;350(3):427–40.

15. Abrescia NGA, Cockburn JJB, Grimes JM, Sutton GC, Diprose JM, Butcher SJ, et al. Insights into assembly from structural analysis of bacteriophage PRD1. Nature. 2004;432(7013):68–74.

16. Rydman PS, Bamford JK., Bamford DH. A minor capsid protein P30 is essential for bacteriophage PRD1 capsid assembly. J Mol Biol. 2001 Nov 2;313(4):785–95.

68

17. Strömsten NJ, Bamford DH, Bamford JKH. In vitro DNA packaging of PRD1: a common mechanism for internal-membrane viruses. J Mol Biol. 2005 May 6;348(3):617–29.

18. Rydman PS, Caldentey J, Butcher SJ, Fuller SD, Rutten T, Bamford DH. Bacteriophage PRD1 contains a labile receptor-binding structure at each vertex. J Mol Biol. 1999 Aug;291(3):575–87.

19. Bamford JK, Bamford DH. A new mutant class, made by targeted mutagenesis, of phage PRD1 reveals that protein P5 connects the receptor binding protein to the vertex. J Virol. 2000 Sep 1;74(17):7781–6.

20. Javier Caldentey *, Roman Tuma and, Bamford DH. Assembly of Bacteriophage PRD1 Spike Complex:  Role of the Multidomain Protein P5†. 2000;

21. Jaatinen ST, Viitanen SJ, Bamford DH, Bamford JKH. Integral membrane protein P16 of bacteriophage PRD1 stabilizes the adsorption vertex structure. J Virol. 2004 Sep 15;78(18):9790–7.

22. Huiskonen JT, Manole V, Butcher SJ. Tale of two spikes in bacteriophage PRD1. Proc Natl Acad Sci. 2007 Apr 17;104(16):6666–71.

23. Claverie J-M, Abergel C. Mimiviridae: An Expanding Family of Highly Diverse Large dsDNA Viruses Infecting a Wide Phylogenetic Range of Aquatic Eukaryotes. Viruses. 2018;10(9).

24. Iyer LM, Aravind L, Koonin E V. Common origin of four diverse families of large eukaryotic DNA viruses. J Virol. 2001 Dec;75(23):11720–34.

25. Koonin E V, Senkevich TG, Dolja V V. The ancient Virus World and evolution of cells. Biol Direct. 2006 Sep 19;1(1):29.

26. Philippe N, Legendre M, Doutre G, Couté Y, Poirot O, Lescot M, et al. Pandoraviruses: amoeba viruses with genomes up to 2.5 Mb reaching that of parasitic eukaryotes. Science. 2013 Jul 19;341(6143):281–6.

27. Legendre M, Bartoli J, Shmakova L, Jeudy S, Labadie K, Adrait A, et al. Thirty-thousand-year-old distant relative of giant icosahedral DNA viruses with a pandoravirus morphology. Proc Natl Acad Sci. 2014 Mar 18;111(11):4274–9.

28. Okamoto K, Miyazaki N, Song C, Maia FRNC, Reddy HKN, Abergel C, et al. Structural variability and complexity of the giant Pithovirus sibericum particle revealed by high-voltage electron cryo-tomography and energy-filtered electron cryo-microscopy. Sci Rep. 2017 Dec 16;7(1):13291.

29. Abrahão J, Silva L, Silva LS, Khalil JYB, Rodrigues R, Arantes T, et al. Tailed giant Tupanvirus possesses the most complete translational apparatus of the known virosphere. Nat Commun. 2018 Dec 27;9(1):749.

30. Abergel C, Legendre M, Claverie J-M. The rapidly expanding universe of giant viruses: Mimivirus, Pandoravirus, Pithovirus and Mollivirus. Filloux A, editor. FEMS Microbiol Rev. 2015 Nov 1;39(6):779–96.

31. Erives AJ. Phylogenetic analysis of the core histone doublet and DNA topo II genes of Marseilleviridae: evidence of proto-eukaryotic provenance. Epigenetics Chromatin. 2017 Dec 28;10(1):55.

32. Boyer M, Yutin N, Pagnier I, Barrassi L, Fournous G, Espinosa L, et al. Giant Marseillevirus highlights the role of amoebae as a melting pot in emergence of chimeric microorganisms. Proc Natl Acad Sci U S A. 2009 Dec 22;106(51):21848–53.

33. Aherfi S, La Scola B, Pagnier I, Raoult D, Colson P. The expanding family Marseilleviridae. Virology. 2014 Oct 1;466–467:27–37.

69

34. Doutre G, Philippe N, Abergel C, Claverie J-M. Genome analysis of the first Marseilleviridae representative from Australia indicates that most of its genes contribute to virus fitness. J Virol. 2014 Dec 15;88(24):14340–9.

35. Colson P, Pagnier I, Yoosuf N, Fournous G, La Scola B, Raoult D. “Marseilleviridae”, a new family of giant viruses infecting amoebae. Arch Virol. 2013;158:915–20.

36. Thomas V, Bertelli C, Collyn F, Casson N, Telenti A, Goesmann A, et al. Lausannevirus, a giant amoebal virus encoding histone doublets. Environ Microbiol. 2011 Jun 1;13(6):1454–66.

37. Mutsafi Y, Fridmann-Sirkis Y, Milrot E, Hevroni L, Minsky A. Infection cycles of large DNA viruses: Emerging themes and underlying questions. Virology. 2014 Oct 1;466–467:3–14.

38. Raoult D, Boyer M. Amoebae as Genitors and Reservoirs of Giant Viruses. Intervirology. 2010;53(5):321–9.

39. Bertelli C, Mueller L, Thomas V, Pillonel T, Jacquier N, Greub G. Cedratvirus lausannensis - digging into Pithoviridae diversity. Environ Microbiol. 2017 Oct;19(10):4022–34.

40. Andreani J, Aherfi S, Khalil JYB, Di Pinto F, Bitam I, Raoult D, et al. Cedratvirus, a double-cork structured giant virus, is a distant relative of pithoviruses. Viruses [Internet]. 2016;

41. Stanley WM. Isolation of a crystalline protein possessing the properties of tobacco-mosaic virus. Science (80- ). 1935 Jun 28;81(2113):644–5.

42. Bernal JD. X-Ray and Crystallographic Studies of Plant Virus Preparations: I. Introduction and Preparation of Specimens Ii. Modes of Aggregation of the Virus Particles. J Gen Physiol. 1941 Sep 20;25(1):111–46.

43. WATSON JD, CRICK FHC. Molecular Structure of Nucleic Acids: A Structure for Deoxyribose Nucleic Acid. Nature. 1953 Apr;171(4356):737–8.

44. KENDREW JC, BODO G, DINTZIS HM, PARRISH RG, WYCKOFF H, PHILLIPS DC. A Three-Dimensional Model of the Myoglobin Molecule Obtained by X-Ray Analysis. Nature. 1958 Mar;181(4610):662–6.

45. Lodish HF. Molecular cell biology [Internet]. W.H. Freeman; 2000. 36 p. Available from: https://www.ncbi.nlm.nih.gov/books/NBK21475/

46. Liljas A. Textbook of structural biology [Internet]. World Scientific; 2009. 572 p. Available from: https://books.google.se/books/about/Textbook_of_Structural_Biology.html?id=aCvtLB0hX44C&redir_esc=y

47. PDB Statistics: Growth of Structures from X-ray Crystallography [Internet]. Available from: https://www.rcsb.org/stats/growth/xray

48. Bragg WL. The Structure of Some Crystals as Indicated by Their Diffraction of X-rays. Proc R Soc A Math Phys Eng Sci. 1913 Sep 22;89(610):248–77.

49. Chapman HN, Fromme P, Barty A, White TA, Kirian RA, Aquila A, et al. Femtosecond X-ray protein nanocrystallography. Nature. 2011 Feb 3;470(7332):73–7.

50. Quentin D, Raunser S. Electron cryomicroscopy as a powerful tool in biomedical research. J Mol Med. 2018 Jun 5;96(6):483–93.

51. Jensen GJ. Cryo-EM. Part A, Sample preparation and data collection [Internet]. Academic Press/Elsevier; 2010. 410 p. Available from: https://www.sciencedirect.com/bookseries/methods-in-enzymology/vol/481

52. Jensen GJ. Cryo-EM. Part B, 3-D reconstruction [Internet]. Academic Press/Elsevier; 2010. 410 p. Available from: https://www.sciencedirect.com/bookseries/methods-in-enzymology/vol/482

70

53. Wan W, Briggs JAG. Cryo-Electron Tomography and Subtomogram Averaging. Methods Enzymol. 2016 Jan 1;579:329–67.

54. Acharya KR, Lloyd MD. The advantages and limitations of protein crystal structures. Trends Pharmacol Sci. 2005 Jan 1;26(1):10–4.

55. Madey JMJ. Stimulated Emission of Bremsstrahlung in a Periodic Magnetic Field. J Appl Phys. 1971 Apr 1;42(5):1906.

56. Neutze R, Wouts R, van der Spoel D, Weckert E, Hajdu J. Potential for biomolecular imaging with femtosecond X-ray pulses. Nature. 2000 Aug 17;406(6797):752–7.

57. DePonte DP, Weierstall U, Schmidt K, Warner J, Starodub D, Spence JCH, et al. Gas dynamic virtual nozzle for generation of microscopic droplet streams. J Phys D Appl Phys. 2008 Oct 7;41(19):195505.

58. Loh NTD, Elser V. Reconstruction algorithm for single-particle diffraction imaging experiments. Phys Rev E - Stat Nonlinear, Soft Matter Phys. 2009 Aug 24;80(2):026705.

59. Loh ND, Bogan MJ, Elser V, Barty A, Boutet S, Bajt S, et al. Cryptotomography: Reconstructing 3D Fourier Intensities from Randomly Oriented Single-Shot Diffraction Patterns. Phys Rev Lett. 2010 Jun 2;104(22):225501.

60. Luke DR. Relaxed averaged alternating reflections for diffraction imaging. Inverse Probl [Internet]. 2005;

61. Kirz J, Jacobsen C. The history and future of X-ray microscopy. J Phys Conf Ser. 2009 Sep 1;186(1):012001.

62. Harkiolaki M, Darrow MC, Spink MC, Kosior E, Dent K, Duke E. Cryo-soft X-ray tomography: using soft X-rays to explore the ultrastructure of whole cells. Emerg Top Life Sci. 2018 Apr 20;2(1):81–92.

63. Hagen C, Guttmann P, Klupp B, Werner S, Rehbein S, Mettenleiter TC, et al. Correlative VIS-fluorescence and soft X-ray cryo-microscopy/tomography of adherent cells. J Struct Biol. 2012 Feb;177(2):193–201.

64. Uchida M, McDermott G, Wetzler M, Le Gros MA, Myllys M, Knoechel C, et al. Soft X-ray tomography of phenotypic switching and the cellular response to antifungal peptoids in Candida albicans. Proc Natl Acad Sci U S A. 2009 Nov 17;106(46):19375–80.

65. Berglund M, Rymell L, Peuker M, Wilhein T, Hertz HM. Compact water-window transmission X-ray microscopy. J Microsc. 2000 Mar;197(Pt 3):268–73.

66. Grahn AM, Caldentey J, Bamford JK, Bamford DH. Stable packaging of phage PRD1 DNA requires adsorption protein P2, which binds to the IncP plasmid-encoded conjugative transfer complex. J Bacteriol. 1999 Nov;181(21):6689–96.

67. Daurer BJ, Okamoto K, Bielecki J, Maia FRNC, Mühlig K, Seibert MM, et al. Experimental strategies for imaging bioparticles with femtosecond hard X-ray pulses. IUCrJ. 2017 May 1;4(3):251–62.

68. Coetzee JN, Lecatsas G, Coetzee WF, Hedges RW. Properties of R plasmid R772 and the corresponding pilus-specific phage PR772. J Gen Microbiol. 1979 Feb 1;110(2):263–73.

69. Lute S, Aranha H, Tremblay D, Liang D, Ackermann HW, Chu B, et al. Characterization of coliphage PR772 and evaluation of its use for virus filter performance testing. Appl Environ Microbiol. 2004 Aug 1;70(8):4864–71.

70. Reddy HKN, Yoon CH, Aquila A, Awel S, Ayyer K, Barty A, et al. Coherent soft X-ray diffraction imaging of coliphage PR772 at the Linac coherent light source. Sci data. 2017 Jun 27;4:170079.

71

71. Grahn AM, Daugelavicius R, Bamford DH. Sequential model of phage PRD1 DNA delivery: active involvement of the viral membrane. Mol Microbiol. 2002 Dec;46(5):1199–209.

72. Grahn AM, Daugelavicius R, Bamford DH. The small viral membrane-associated protein P32 is involved in bacteriophage PRD1 DNA entry. J Virol. 2002 May 15;76(10):4866–72.

73. Peralta B, Gil-Carton D, Castaño-Díez D, Bertin A, Boulogne C, Oksanen HM, et al. Mechanism of Membranous Tunnelling Nanotube Formation in Viral Genome Delivery. Rey FA, editor. PLoS Biol. 2013 Sep 24;11(9):e1001667.

74. Aquila A, Barty A, Bostedt C, Boutet S, Carini G, DePonte D, et al. The linac coherent light source single particle imaging road map. Struct Dyn [Internet]. 2015;

75. Ferguson KR, Bucher M, Bozek JD, Carron S, Castagna J-C, Coffee R, et al. The Atomic, Molecular and Optical Science instrument at the Linac Coherent Light Source. J Synchrotron Radiat [Internet]. 2015;

76. White WE, Robert A, Dunne M. The Linac Coherent Light Source. J Synchrotron Radiat [Internet]. 2015;

77. KIRKPATRICK P, BAEZ A V. Formation of optical images by X-rays. J Opt Soc Am. 1948 Sep;38(9):766–74.

78. Strüder L, Epp S, Rolles D, Hartmann R, Holl P, Lutz G, et al. Large-format, high-speed, X-ray pnCCDs combined with electron and ion imaging spectrometers in a multipurpose chamber for experiments at 4th generation light sources. Nucl Instruments Methods Phys Res Sect A Accel Spectrometers, Detect Assoc Equip [Internet]. 2010;

79. Daurer BJ, Hantke MF, Nettelblad C, Maia FRNC. Hummingbird : monitoring and analyzing flash X-ray imaging experiments in real time . J Appl Crystallogr [Internet]. 2016;

80. Damiani D, Dubrovin M, Gaponenko I, Kroeger W, Lane TJ, Mitra A, et al. Linac Coherent Light Source data analysis using psana. J Appl Crystallogr. 2016 Apr 1;49(2):672–9.

81. Loh ND. Effects of extraneous noise in cryptotomography. In: Image Reconstruction from Incomplete Data VII [Internet]. 2013.

82. Hosseinizadeh A, Dashti A, Schwander P, Fung R, Ourmazd A. Single-particle structure determination by X-ray free-electron lasers: Possibilities and challenges. Struct Dyn [Internet]. 2015;

83. Bobkov SA, Teslyuk AB, Kurta RP, Gorobtsov OY, Yefanov OM, Ilyin VA, et al. Sorting algorithms for single-particle imaging experiments at X-ray free-electron lasers. J Synchrotron Radiat [Internet]. 2015;

84. Maia FRNC. The Coherent X-ray Imaging Data Bank. Nat Methods [Internet]. 2012;

85. Zheng SQ, Palovcak E, Armache J-P, Verba KA, Cheng Y, Agard DA. MotionCor2: anisotropic correction of beam-induced motion for improved cryo-electron microscopy. Nat Methods. 2017 Apr 27;14(4):331–2.

86. Rohou A, Grigorieff N. CTFFIND4: Fast and accurate defocus estimation from electron micrographs. J Struct Biol [Internet]. 2015;192(2):216–21.

87. Scheres SHW. RELION: Implementation of a Bayesian approach to cryo-EM structure determination. J Struct Biol [Internet]. 2012;180(3):519–30.

88. Kucukelbir A, Sigworth FJ, Tagare HD. Quantifying the local resolution of cryo-EM density maps. Nat Methods. 2014;11(1):63–5.

89. Ilca SL, Kotecha A, Sun X, Poranen MM, Stuart DI, Huiskonen JT. Localized reconstruction of subunits from electron cryomicroscopy images of macromolecular complexes. Nat Commun. 2015 Nov 4;6:8843.

72

90. Goddard TD, Huang CC, Ferrin TE. Visualizing density maps with UCSF Chimera. J Struct Biol. 2007 Jan;157(1):281–7.

91. Terwilliger TC, Sobolev O V., Afonine P V., Adams PD. Automated map sharpening by maximization of detail and connectivity. Acta Crystallogr Sect D Struct Biol. 2018 Jun 1;74(6):545–59.

92. Emsley P, Lohkamp B, Scott WG, Cowtan K. Features and development of Coot. Acta Crystallogr D Biol Crystallogr. 2010 Apr;66(Pt 4):486–501.

93. Adams PD, Afonine P V., Bunkóczi G, Chen VB, Davis IW, Echols N, et al. PHENIX : a comprehensive Python-based system for macromolecular structure solution. Acta Crystallogr Sect D Biol Crystallogr. 2010 Feb 1;66(2):213–21.

94. Chen VB, Arendall WB, Headd JJ, Keedy DA, Immormino RM, Kapral GJ, et al. MolProbity: all-atom structure validation for macromolecular crystallography. Acta Crystallogr D Biol Crystallogr. 2010 Jan;66(Pt 1):12–21.

95. Afonine P V., Klaholz BP, Moriarty NW, Poon BK, Sobolev O V., Terwilliger TC, et al. New tools for the analysis and validation of cryo-EM maps and atomic models. Acta Crystallogr Sect D Struct Biol [Internet]. 2018;

96. Barad BA, Echols N, Wang RY-R, Cheng Y, DiMaio F, Adams PD, et al. EMRinger: side chain–directed model and map validation for 3D cryo-electron microscopy. Nat Methods. 2015 Oct 17;12(10):943–6.

97. Merckel MC, Huiskonen JT, Bamford DH, Goldman A, Tuma R. The structure of the bacteriophage PRD1 spike sheds light on the evolution of viral capsid architecture. Mol Cell [Internet]. 2005;18(2):161–70.

98. Eng JK, Fischer B, Grossmann J, MacCoss MJ. A Fast SEQUEST Cross Correlation Algorithm. J Proteome Res. 2008 Oct 3;7(10):4598–602.

99. Tang G, Peng L, Baldwin PR, Mann DS, Jiang W, Rees I, et al. EMAN2: An extensible image processing suite for electron microscopy. J Struct Biol. 2007 Jan;157(1):38–46.

100. Kremer JR, Mastronarde DN, McIntosh JR. Computer visualization of three-dimensional image data using IMOD. J Struct Biol [Internet]. 1996;

101. Hosseinizadeh A, Mashayekhi G, Copperman J, Schwander P, Dashti A, Sepehr R, et al. Conformational landscape of a virus by single-particle X-ray scattering. Nat Methods. 2017 Sep 14;14(9):877–81.

102. Schneidmiller EA, Yurkov M V. Photon beam properties at the European XFEL [Internet]. Hamburg. 2011: European X-Ray Free-Electron Laser Facility GmbH; 2011.

103. Mohacsi I, Vartiainen I, Rösner B, Guizar-Sicairos M, Guzenko VA, McNulty I, et al. Interlaced zone plate optics for hard X-ray imaging in the 10 nm range. Sci Rep. 2017 May 8;7(1):43624.

Acta Universitatis UpsaliensisDigital Comprehensive Summaries of Uppsala Dissertationsfrom the Faculty of Science and Technology 1847

Editor: The Dean of the Faculty of Science and Technology

A doctoral dissertation from the Faculty of Science andTechnology, Uppsala University, is usually a summary of anumber of papers. A few copies of the complete dissertationare kept at major Swedish research libraries, while thesummary alone is distributed internationally throughthe series Digital Comprehensive Summaries of UppsalaDissertations from the Faculty of Science and Technology.(Prior to January, 2005, the series was published under thetitle “Comprehensive Summaries of Uppsala Dissertationsfrom the Faculty of Science and Technology”.)

Distribution: publications.uu.seurn:nbn:se:uu:diva-391671

ACTAUNIVERSITATIS

UPSALIENSISUPPSALA

2019