**1. Introduction**

The dynamic actin cytoskeleton is commonly utilized by pathogens for entry, exit and their intracellular assembly [1]. While for some intracellular bacteria and DNA viruses, the mechanism of hijacking the actin cytoskeleton is described in detail, the remodeling of actin through highly pathogenic RNA viruses such as filoviruses remains poorly understood [1,2]. Recently, it has been demonstrated that the intracytoplasmic transport of the highly pathogenic Marburg virus (MARV) and the Ebola virus (EBOV) depends on actin polymerization, but the detailed mechanisms and cellular interaction partners remain largely unknown [3–5].

MARV and EBOV belong to the family of *Filoviridae*, which are filamentous, enveloped viruses containing a single strand, negative-sense RNA genome, and which cause severe fevers in humans with very high fatality rates [6]. Strikingly, filoviral RNA encodes for only seven structural proteins, which are multifunctional, and diversely hijack, disable or reorganize cellular pathways [7]. For example, EBOV VP24 and VP35 are able to block the cell's interferon system, resulting in the viruses' escape from the innate immune response [8].

A hallmark of the filoviral life cycle is the formation of perinuclear inclusion bodies, where transcription and replication occur and de novo viral nucleocapsids are formed [9]. These nucleocapsids are mainly composed of the nucleoprotein (NP) that encapsidates the viral RNA in highly organized helical structures (approximately 1000 nm long and 50 nm in diameter) [10]. Furthermore, nucleocapsids contain the NP-binding proteins VP24 and VP35, as well as polymerase L and the transcription initiation factor VP30 [10–12].

Following their assembly in the inclusion bodies, individual nucleocapsids have to be transported over long distances through the cytoplasm to reach the cell periphery [3,5,9]. It has been shown that they accumulate in filopodia structures. Finally, nucleocapsids acquire their envelope at the plasma membrane, containing the viral trans-membrane glycoprotein (GP) and the viral matrix protein VP40, triggering the budding of the filamentous infectious virus particles. [8]. The directed transport is required for the rapid distancing away from the inclusion bodies, and to transfer functional nucleocapsids to the plasma membrane. It has been shown that individual single mutations within the nucleocapsid proteins critically a ffect long-distance transport, thereby leading to a significant delay and reduction of the release of filamentous nucleocapsids [13–15].

While other viruses utilize the microtubule network for intracellular distribution, [16] the transport of nucleocapsids of MARV as well as EBOV is entirely blocked by the pharmacological inhibition of actin polymerization [3,4]. However, the role of actin nucleators and the regulation of their upstream e ffectors in this process remains elusive. One major obstacle in studying filoviruses is the requirement of high containment laboratories, restricting cell biological experiments that could accelerate our understanding of the critical steps in the virus life cycle. To overcome this problem, we recently established a novel transfection-based system that only requires three viral proteins to produce nucleocapsid-like structures (NCLS), which highly resemble EBOV nucleocapsids in their structure as well as their intracellular dynamics [5]. Through the co-expression of GFP-tagged VP30, which faithfully labels NCLS, it is possible to monitor intracellular transport mechanisms with real time imaging under normal laboratory conditions (biosafety level 1) conditions (Figure 1A,B and Figure 2A).

The host cell contains a highly dynamic actin cytoskeleton that is required for many essential cellular processes, such as cytokinesis, contractility and motility [17,18]. Filamentous actin (called F-actin) is assembled from monomeric globular G-actin, and this polymerization, as well as depolymerization, is highly regulated through a plethora of di fferent cellular factors [19–21]. With this high power of regulation, the cell is able to assemble and coordinate diverse structures, such as the strong cortical actin network stabilizing the cell cortex, short and highly dynamic filaments that are involved in vesicular tra fficking, and filament networks that form membrane protrusions such as the filopodia and lamellipodia required for cell motility or cell–cell contact formation [22–25]. The Arp2/3 (actin-related protein) complex e fficiently nucleates the actin filaments typically attached to mother filaments to form the highly branched networks that (amongst other things) enable lamellipodia to rapidly adapt during cell migration [26]. This active protein complex has to be tightly regulated, and requires activation for polymerization [27]. So-called nucleation promoting factors (NFPs), such as WAVE1 and WASP proteins, directly induce Arp2/3 complex activity [27,28], and are themselves activated downstream of RhoGTPase signaling in a highly spatial–temporal manner [29].

The Arp2/3 complex is highly conserved, and di fferent pathogens, including *Listeria monocytogenes* and vaccinia virus, utilize its activity for viral intracellular transport steps [30]. For example, the membrane-integrated *Listeria* protein ActA mimics the NPF WASP, thereby recruiting and activating the Arp2/3 complex [31–33]. The Arp2/3 complex in turn induces local actin polymerization, resulting in so-called actin comet tails that e fficiently propel the bacterium through the cytoplasm, pushing it into neighboring cells. Furthermore, actin comet tails have been previously observed at EBOV nucleocapsids [3]; however, the mechanism by which viral nucleocapsids use actin dynamics for their transport has not been described in detail.

**Figure 1.** Ebola virus (EBOV) nucleocapsid-like structures induce polar actin tails during transport. (**A**) Exemplary still image from a movie of a Huh7 cell expressing NP, VP24, VP35 and VP30-GFP (left panel). During imaging, NCLS are tracked for 2 min producing long-distance trajectories as depicted in the maximum intensity projection of the movie (right panel). The white arrow highlights an individual track. IB labels inclusion bodies. (**B**) Co-imaging of actin using LifeAct-CLIP (stained with Alexa-657 dye) reveals a dense and dynamic actin network. White inset is magnified in (**C**) showing the time course of an individual NCLS (white arrow, Figure 1**B**), revealing a pulsative actin tail located on one site of the subviral particle during movement (see also Movie S1). (**D**) Graph showing the movement of this NCLS over time. The red arrows indicate actin pulses observed. (**E**,**F**) STORM images showing NCLS or EBOV nucleocapsids immunolabelled with NP (green) and stained with Phalloidin (magenta). (**D**) The left panel shows a filamentous virus particle likely prior to budding (white arrow). Individual intracellular nucleocapsids reveal a preserved actin tail (white arrow). (**E**) Huh7 cells were transfected with NP, VP24 and VP35, and fixed after 24 h. The samples were stained with anti-NP and Phalloidin, and prepared for STORM microscopy. The zoomed image shows an individual NCLS with a preserved actin tail (white arrow).

In this study, we employ our recently established live cell imaging approach to delineate the cellular pathways by which EBOV exploits host actin signaling, and extend the previously applied manual quantification approach to a semi-automatic high throughput method. Using small inhibitory compounds and siRNA-mediated knockdown, we demonstrate that Arp2/3 complex activity downstream of Rac1 is critically involved in the directed long-distance transport of EBOV nucleocapsid structures. Furthermore, through co-visualization of NCLS transport with the actin marker LifeAct, we detected pulsative actin tails accompanying the movement of NCLS through the cytoplasm, which requires Arp2/3 activity.

**Figure 2.** Analysis of NCLS transport using actin-modulating drugs. (**A**) Exemplary still image from a movie of a Huh7 cell expressing NP, VP24, VP35 and VP30-GFP (left panel). For live cell imaging, NCLS are tracked through the cell over 2 min with images captured every 900 ms. While the middle panel shows the maximum-intensity projection of this movie, the right panel depicts the same movie quantified using the spot algorithm (Imaris). The tracks are color-coded for length. To exclude artefacts, we semi-manually deleted areas with strong accumulation of VP30-GFP, such as the inclusion bodies (asterisk). (**B**–**E**) Huh7 cells transfected with NCLS (VP30-GFP) were recorded for 2 min. This was followed by a short incubation with cytoskeletal-modulating drugs after which the cells were re-recorded. Per experiment three cells were recorded and all tracks > 20 s were quantified. The pictures show a magnified area after quantification with Imaris. Cells were incubated with (**B**) 100 μM Cytochalasin D, (**C**) 50 μM Jasplakinolide, (**D**) 100 μM para-nitro-Blebbistatin or (**E**) 100 μM CK666 following the first imaging. The graphs depict the normalized number of tracks >10 μm, error bars show the SD of *n* = 3 experiments.

### **2. Materials and Methods**

### *2.1. Cells and Viruses*

Huh7 (human hepatoma) cells were cultured in DMEM (Life Technologies, Carlsbad, CA, USA) supplemented with 10% (*v*/*v*) fetal calf serum (FCS) (PAN Biotech), 5 mM l-glutamine, 50 U/mL penicillin and 50 μg/mL streptomycin (Life Technologies) and grown at 37 ◦C with 5% CO2. For live cell imaging experiments, cells were kept in phenol-free Leibovitz's medium (Life Technologies) with PS/Q, non-essential amino acid solution and 20% (*v*/*v*) FCS.

The virus used in this study was based on EBOV Zaire (Strain Mayinga; GenBank accession no AF27200 (National Center for Biotechnology Information, Bethesda, MD USA)). The experiments with infectious EBOV virus were performed in the BSL-4 facility at the University of Marburg.

### *2.2. Transfections, Plasmids, siRNA and Inhibitors*

The plasmids coding for EBOV proteins (pCAGGS-NP, -VP35, -VP24) and pCAGGS-VP30-GFP were described previously [34]. Transient transfections of Huh7 cells were carried out using TransIT-LT1 (Mirus, Madison, WI, USA) according to the manufacturers' instructions with 3 μL reagen<sup>t</sup> per 1 μg plasmid DNA [5]. Transfection of siRNAs was performed using DharmaFECT (Horizon, Waterbeach, UK) using 25 nM siRNA and 5 μL transfection reagen<sup>t</sup> in Opti-MEM (Thermo Fisher, Waltham, MA, USA). Huh7 cells were transfected with siRNA in a 6-well μ-slide, seeded to a 4-well μ-slide (IBIDI) and then transfected with the plasmids encoding for the viral proteins (200 ng/μ<sup>L</sup> VP30-GFP, 30 ng/μ<sup>L</sup> VP24, 200 ng/μ<sup>L</sup> VP35 and 200 ng/μ<sup>L</sup> NP). Then, NCLS movement was monitored 24 h later through detection of VP30-GFP. The FlexiTube siRNA used were purchased from Qiagen and diluted to 10 μM. The siRNAs used in the study were HS-ACTR3\_5 (AAAGTGGGTGATCAAGCTCAA), HS\_RAC1\_6 (ATGCATTTCCTGGAGAATATA), HS\_WASF1\_3 (CAAGAACGTGTGGACCGTTTA) and HS-CDC42\_16 (5-CATCAGATTTTGAAAATATTTAA 3). For treatment with inhibitors, individual transfected cells (NP, VP24, VP35 and VP30-GFP) were monitored, then the inhibitor was added to the cells in the appropriate dilution and after a short incubation the very same cell was imaged. Cytochalasin D, Jasplakinolide and NSC 23766 were obtained from Sigma Aldrich, and CK666 was purchased from EMD Millipore.

### *2.3. Antibodies and Reagents for Microscopy*

The following primary antibodies were used in this study: polyclonal anti-chicken EBOV NP [35], polyclonal rabbit anti-Wave1 (Sigma Aldrich, St. Louis, MO, USA), mouse anti-Arp3 (Sigma, A5979), mouse anti-Rac1 (Cytoskeleton) and mouse anti-Tubulin (Sigma). The corresponding secondary antibodies were anti-mouse-HRP and anti-rabbit-HRP (DAKO, Jena, Germany), and anti-chicken-Alexa555 (Invitrogen, Carlsbad, CA, USA). Western Blot analyses were performed as described previously [5]. SNAP/CLIP technology was used to visualize actin by using LifeAct-CLIP. LifeAct-CLIP was cloned using standard cloning procedures and then transfected with the NCLS system. For co-visualization of actin and NCLS transport, transfected cells cells were incubated with the dye CLIP-647 (1:500 diluted in media, NEB) for 30 min prior to the experiment, then washed and replaced with Leibovitz medium for live cell imaging.

### *2.4. Confocal Microscopy, Live Cell Imaging and STORM Microscopy*

Huh7 cells were fixed using 4% PFA/DMEM, permeabilized with 0.3%Triton-X100/PBS and blocked with blocking buffer containing 3% glycerol, 2% BSA, 0.2% Tween and 0.05% NaN3. Primary antibodies as well as secondary antibodies were diluted in blocking buffer and both were incubated for 1 h at RT. The actin cytoskeleton was visualized with Phalloidin conjugated with Alexa647 (Cytoskeleton). For STORM analysis of transfected cells, cells were fixed with 4% PFA/PBS, permeabilized with 0.3% Triton-X100 and then stained with Phalloidin-Alexa647 (1:50) in PBS for 48 h at 4 ◦C, before being immunolabelled for NP using anti-chicken NP (1:1000) in blocking buffer followed by an incubation with anti-chicken-Alexa555 (Invitrogen, 1:250). Every instance of Phalloidin or antibody staining was followed by a post-fixation step using 4% PFA/PBS. Cells infected with EBOV were fixed with 4% PFA/PBS and permeabilized, then stained with Phalloidin for 1 h in the BSL4 lab, followed by 48 h incubation with 4% PFA/DMEM. Cells were then blocked and immunolabeled as described above and re-stained with Phalloidin for 48 h. For STORM analysis, the coverslip was mounted in switching buffer (1 M MEA-HCl in Glucose buffer containing catalase, TCEP and glucose oxidase).

Dual color live-cell imaging was performed on a SP8 confocal laser scanning microscope (Leica) equipped with a 64 × 1.4 oil objective, and movies for single color live cell imaging were recorded using a Nikon ECLIPSE TE2000-E microscope with a 64x-oil objective. The movies for quantification were acquired every 900 ms for 120 s. All live cell imaging was performed at 37 ◦C. STORM images were acquired using the Oxford Nanoimaging system (ONI). The light was collected by the 100× objective and imaged onto the EM-CCD camera at 30 ms per frame. Images were processed using the NimOS localization software (Oxford Nanoimaging).

### *2.5. Particle Tracking, Quantification and Statistics*

Image processing was performed with Imaris (Bitplane, Oxford Instruments, Abingdon, UK), FIJI (NIH) and Photoshop CS6 (Adobe, San Jose, CA, USA). For figure design, we used Adobe Illustrator. NCLS tracking was performed using Imaris (Bitplane). To this end, VP30-GFP signal was monitored every 900 ms over 2 min and then moving NCLS were analyzed using the "spots feature", a presetting that has been optimized for cell and particle tracking within the software. Within the "spots" feature, we determined an algorithm that automatically classifies objects of over 0.4-μm diameter appearing continuously for at least 20 s as NCLS. The positions of the structures were then calculated in each image of the 2-min movie. From these data, the software is able to calculate trajectories using an autoregressive motion algorithm. After the computational calculation and visualization of all tracks, we semi-manually filtered out particle aggregations, as well as GFP signals derived from inclusion bodies, by deleting spots within certain areas within the cell (see Figure 2A), resulting in approximately 400–800 tracks per cell. We filtered the data (e.g., for tracks >10 μm or straightness > 0.7) which were then used for quantification. This algorithm was then applied to all files of a set of experiments. Each experiment was performed in at least three independent repetitions. Statistical analyses were performed with Prism x (GraphPad, San Diego, CA, USA).
