**3. Results**

### *3.1. Nucleocapsid Movement Is Accompanied by Pulsative Actin Tails*

Here, we employ a novel live cell imaging system to monitor the long-distance movement of EBOV NCLS, in which only three viral proteins are expressed, namely, NP, VP24 and VP35 (Figure 1A, left panel) [5]. Using VP30-GFP, we monitor the intracellular transport of NCLS for two minutes, and the corresponding maximum-intensity projection reveals that a robust number of NCLS (6.9% ± 2.7%) are transported in a highly directed fashion over long distances (>10μm–20μm), now referred to as long-distance trajectories (Figure 1A, right panel). However, due to defective assembly or saturation of the cellular transport machinery, we also detect many VP30-GFP-positive NCLS that do not move, or travel only short distances (Figure S1A).

Actin tail formation at EBOV nucleocapsids was detected in an earlier study [3]. Therefore, we aimed here to determine whether or not the established transfection system can recapitulate the detection of actin tails at NCLS as well. In cells that co-express the actin marker LifeAct to visualize the actin cytoskeleton (Figure 1B), we observed that NCLS induces actin tails during their movement through the cell (movie S1), thereby further highlighting the relevance of this transfection system to the study of the transport of EBOV nucleocapsids. Here, we further assessed the movement of an individual NCLS-track over time, starting from its origin and following it through the cell (inset of Figure 1B is magnified in Figure 1C, white arrows). In Figure 1D, the red arrows indicate actin pulses

during movement, which appear to correlate with the altered movement of NCLS, such as changes in direction (Figure 1C, Movie S1). While with the live cell imaging setting we used approximately 80% of the tracks that moved in a directed fashion revealed detectable actin tails at their rear end (Figure 3E, control siRNA), we cannot exclude the possibility that smaller NCLS induce actin polymerization below the detection limit, due to the high intracellular background level of LifeAct.

**Figure 3.** NCLS induction of actin tails depends on Arp2/3 activity. (**A**) Huh7 cells transfected with NP, VP24 and VP35 and LifeAct-CLIP were monitored over time. The left panels show an overlay image of the tracks as calculated by Imaris (color coded for length) and LifeAct-CLIP (still image from movie). Note, CK666 incubation does not interfere with filopodia (asterisk) or stress fibers (arrow head) (see Movie S2). The zoomed image (right panels) show still images of the movie, revealing that CK666 treatment abolishes actin tail formation (white arrows). (**B**) Western Blot showing the depletion of Arp3 after siRNA treatment. (**C**) Cartoon showing different movements of trajectories and their corresponding straightness value. (**D**) Left panel shows maximum intensity projections of time lapse imaging of Huh7 cells transfected either with control or Arp3 siRNA. The right panel shows the graphical analysis of live cell imaging comparing all tracks > 10 μm in cells transfected either with control or Arp3 siRNA. Note, the relative amount of tracks with a straightness > 0.7, representing directed long-distance transport, is reduced in cells transfected with Arp3 siRNA (taken from *n* = 3 experiments, \* *p* = 0.0027, Mann–Whitney Test). (**E**) Graph shows the percentage of trajectories with pulsative actin tails in live cell imaging (*n* = 3, error bars indicate mean with standard deviation).

While live cell imaging was applied to capture rapid and transient phenomena, the image resolution remained low. To precisely determine the localization of actin tails at filoviral nucleocapsids, we decided to use a high resolution, single molecule detection technique (STORM). To this end, we either transfected cells with NP, VP35 and VP24, or infected cells with EBOV and fixed them after 24 h. This was followed by the staining of the actin cytoskeleton and immunolabelling for the viral NP protein. Intracellular NCLS or EBOV nucleocapsids appear as elongated NP-positive structures (Figure 1E,F). However, compared to NCLS in transfected cells, filamentous nucleocapsids in cells

infected with EBOV were commonly detected along the plasma membrane and within filopodia, likely representing virus particles prior to or during budding, thereby supporting previous observations from live-cell imaging and electron microscopy (Figure 1D, left panel) [3]. Co-staining with Phalloidin revealed actin tail structures in the vicinity of the nucleocapsids in infected cells and transfected cells (Figure 1D,E), which were less frequent in infected cells. This finding is likely a consequence of the very high spatial-temporal dynamics of actin tail formation, which do not allow a quantifiable detection in fixed cells. Further, in the minimalistic transfection system, NCLS remain in the cytosol and are not released, thus the detection of NCLS with actin tails may occur more frequently. Additionally, it cannot be excluded that in infected cells, the long fixation procedure required for the handling of BSL4 samples might interfere with the fixation of these transient actin structures.

### *3.2. Characterization of NCLS Transport Dynamics*

To measure and characterize the long-distance movement of NCLS in a semi-automated manner, we analyzed our time-lapse movies with a tracking tool within the software Imaris. To this end, we set up an algorithm to follow NCLS, and filtered for trajectories > 10 μm that show mean velocities comparable to those measured in previous reports (here, 167 ± 58 nm/s compared to 187 ± 67 nm/s as in Takamatsu et al., 2018 [5]) (Figure 2A).

These studies demonstrated that the long-distance movement of EBOV nucleocapsids, as well as NCLS, is sensitive to the inhibition of actin polymerization [3,5]. Here, we recapitulated these experiments by monitoring NCLS movement followed by an incubation with an actin-modulating drug, and then re-recorded the very same cell. Using this system, we ensured that the imaged cells were capable of showing long-distance trajectories, and would consequently enable us to analyze alterations in NCLS movement. As observed previously, the application of the inhibitor of actin polymerization, Cytochalasin D, results in a rapid and strong decrease in any long-distance trajectories of NCLS, quantified as a highly reduced number of tracks over 10 μm (Figure 2B) [3,5]. Expanding on previous results, we further incubated the cells with the actin polymerization and stabilization agen<sup>t</sup> Jasplakinolide, resulting in a similar phenotype, thereby indicating that indeed not only actin polymerization, but also depolymerization, is required for the long-distance transport of NCLS (Figure 2C). Treatment with Blebbistatin (here we used para-nitro-Blebbistatin to avoid phototoxicity in live cell experiments) did not interfere with long-distance movement, suggesting that myosin II activity is not required for the long-distance transport of NCLS (Figure 2D). Taken together, these findings reinforce the importance of actin regulation and dynamics in the long-distance transport of NCLS.

### *3.3. Arp2*/*3 Complex Activity Is Required for Actin Tail Formation and Directed Long-Distance Transport*

Previous experiments showed that inhibitors of Arp2/3 a ffected the long-distance movement of EBOV nucleocapsids [3]. To confirm and further elaborate on these results, we used the minimalistic transfection system to analyze how treatment with an Arp2/3 inhibitor (CK666) a ffects NCLS movement [36]. Movie S2 shows that blocking Arp2/3 activity rapidly abolishes any directed long-distance movement of NCLS, and that minimal movement likely derives from Brownian motion (Figure 2E, Movie S2). When dissecting di fferent pools of tracks, we observed that treatment with CK666 strongly decreases tracks >10 μm, subsequently resulting in increases in tracks < 5 μm, which also show altered mean speeds (Figure S1A,B). In cells treated with CK666, we observed a slight reduction in the overall NCLS (−19 ± 7%), which might have resulted from the decreased NCLS transport from the inclusion bodies into the cytoplasm (Figure S1C). Furthermore, Movie S2 reveals that the abolishment of NCLS transport, using CK666, coincides with the loss of actin tail formation (Figure 3A, tracks are color-coded for length and actin is visualized with stills from the confocal movie). Time-lapse imaging further demonstrates that Arp2/3-independent structures, such as filopodia or stress fibers, are not compromised during the observed time frame, thereby highlighting a specific role for Arp2/3-dependent actin polymerization in the directed long-distance transport of NCLS (Movie S2).

Next, we continued with the depletion of one protein within the Arp2/3 complex, namely Arp3, using siRNA transfection to further evaluate its role in actin tail-dependent NCLS transport (Figure 3B). The siRNA depletion of Arp3 likely results in the destabilization or reduction of functional Arp2/3 complexes, and does not entirely block all long-distance movement of NCLS, as observed under treatment with CK666. Thus, we assessed whether the remaining tracks showed altered directionality in cells depleted for Arp3, through the calculation of trajectory straightness (ratio of distance to length) (Figure 3C). We determined that tracks with a straightness > 0.7 (Figure 3C), and indeed the depletion of Arp3, interferes with the total number of straight tracks, further highlighting a role for Arp2/3 activity in the long-distance transport of NCLS (Figure 3D,E). Next, we compared the formation of actin tails in control and Arp3-depleted cells. While approximately 80% of the NCLS that travel over long distances showed pulsative actin tails, cells depleted for Arp3 showed a decreased level of detectable actin polymerization at their rear end (Figure 3E,F). Taken together, these results indicate that Arp2/3 activity is required for the efficient directed long-distance transport of NCLS, and that the Arp2/3 complex is directly involved in the formation of actin tails in Huh7 cells.

### *3.4. Identification of Rac1*/*WAVE1*/*Arp2*/*3 Signaling Network that Is Involved in Long-Distance Transport of NCLS*

The Arp2/3 complex resides, in inactive status, within the cell, and requires activation by NFPs, such as N-WASP or Wave proteins, that bind and activate the Arp2/3 complex via their conserved VCA-domain (Figure 4A) [29]. As it was recently demonstrated that the inhibition of *N*-WASP using Wiskostatin did not interfere with the long-distance transport of EBOV nucleocapsids [3], we focused on Wave1, an alternative activator of the Arp2/3 complex. Interestingly, siRNA directed against WASF1 (Wave1) faithfully recapitulated the phenotype of Arp3 siRNA treatment, leading to the reduced straightness of the NCLS (Figure 4B). This indicates that activation of the Arp2/3 complex downstream of Wave1 is involved in the directionality of long-distance trajectories. Wave1 activity itself can be controlled by the RhoGTPase Rac1 (Figure 4A–C) [37,38]. Again, the treatment of Huh7 cells with siRNA against Rac1 resulted in a decreased level of directed long-distance trajectories (Figure 4D,F,G). In contrast, the depletion of Cdc42, another member of the RhoGTPase family, did not affect the straightness of trajectories (Figure S1D), thereby further reinforcing the inference that Rac1 activity is involved in NCLS transport (Figure 4F,H). For further confirmation, we applied the Rac1 inhibitor NSC 23766, which also interfered with long-distance transport, confirming the relevance of Rac1 activity in the regulation of the long-distance transport of EBOV NCLS (Figure 4H).

Taken together, these findings indicate that EBOV nucleocapsids exploit the canonical Rac1/WAVE1/Arp2/3 signaling pathway for long-distance transport of NCLS inside the cell. This signaling pathway classically induces lamellipodial protrusions, in which Rac1 acts coordinately with other upstream signals to activate actin regulators. Future studies shall reveal whether other actin-regulatory mechanisms are also involved in the viral life cycle, and how alterations in long-distance transport affect virus propagation.

**Figure 4.** Identification of a Rac1/Wave1/Arp2/3 signaling. ( **A**) Cartoon depicting canonical Arp2/3 signaling downstream of Rac1 [39]. (**B**–**D**) Images show maximum intensity projection of time-lapse images of cells recorded for 2 min; images were captured every 900 ms. Cells were transfected with control siRNA (**B**), siRNA against ( **C**) Wave1 (*WASF1*) or ( **D**) Rac1. (**E,F**) Analysis of live cell imaging comparing relative straightness of tracks > 10 μm and a straightness > 0.7 after either Wave1 siRNA (**E**) or Rac1 siRNA (**F**) transfection. Cells were transfected with siRNA against (**B**) Wave1 (*WASF1*) or ( **C**) Rac1. Note that siRNA treatment against Wave1 and Rac1 results in a decrease in straight long-distance trajectories (taken from three independent experiments, at least 5 cells per experiments, ns = non-significant, \*\* *p* > 0.001, \* *p* > 0.01). ( **G**) Western Blots showing e ffective siRNA depletion of Wave1 and Rac1. ( **H**) Incubation of Huh7 cells transfected with NCLS (VP30-GFP) was first recorded for 2 min, then they were incubated with 100 nM Rac1 inhibitor (NSC23766) for 1 h and then reimaged. The overview images reveal that NCS23766 also inhibits long-distance transport. Tracks are color-coded for mean speed (6 cells in *n* = 3, *p* < 0.0079, *t*-test).
