New efficient intercellular spread mode of respiratory syncytial virus contributes to neutralization escape and persistence

There is no licensed vaccine or therapeutic antibody for respiratory syncytial virus (RSV). The induction of high-titer, potent neutralizing antibodies cannot completely inhibit breakthrough infection and enhanced respiratory disease (ERD), encouraging a focus on the relationship between virus intercellular spread and neutralizing antibodies. By blocking the known intercellular spread modes and with the aid of ultrahigh-resolution imaging, we observed a new efficient mode of intercellular spread in which RSV-infected cells directly transfer viral materials (including viral replication factories) to neighboring cells through protruding open-ended microfilament-rich intercellular nanotubes. The new mode is virion-independent and antibody-insensitive, beginning as early as 3 h post infection. Furthermore, replication-defective viral genomes (DVGs) might also utilize the new mode, facilitating the establishment of latent viral infections. Therefore, our data provide a new perspective on RSV cell-to-cell spread and might help to explain the immune escape and latent persistence of paramyxoviruses.


Introduction 28
Human respiratory syncytial virus (RSV) is the leading cause of hospitalization in children with 29 acute lower respiratory tract infection (ALRI) and the most important viral cause of ALRI mortality 30 in childhood (Nair et al., 2010). The RSV F and G proteins are the two main neutralizing antigens 31 and play key roles in virus entry and attachment, respectively (McLellan et al., 2013c). RSV infects 32 human respiratory tract epithelial cells (Hall, 2001). In vitro, cells infected with RSV form syncytia 33 and filamentous surface projections presenting F and G glycoproteins (Collins et al., 2013a). 34 Similar to those of other paramyxoviruses, the RSV replication process is characterized by the 35 formation of inclusion bodies (IBs) (Norrby et al., 1970), which consist of the nucleoprotein (N), 36 the phosphoprotein (P), the large polymerase subunit (L), and the transcription processivity factor 37 (M2-1), as well as viral genomic RNA and messenger RNA (mRNA) (Rincheval et al., 2017). IBs 38 are believed to regulate MDA5-related innate immune responses (Lifland et al., 2012) and are viral 39 factories where viral RNA synthesis occurs (Rincheval et al., 2017, Santangelo et al., 2006. interestingly, there is a hook effect regarding the severity of inflammation as the antigen dose 49 decreases and the level of viral replication in the lungs increases. It seems that once antigen-induced 50 immunity fails to prevent the virus from causing a breakthrough infection, viral replication and 51 pathological processes in the lungs run out of control despite the existence of immune memory. 52 The mechanism of this process remains to be elucidated. Therefore, to understand how RSV 53 breakthrough infection occurs in the presence of high titers of neutralizing antibodies, we focused 54 on the cell-to-cell transmission of RSV. 55 For a long time, our understanding of RSV cell-to-cell transmission has been limited to cell-free 56 virion dissemination and the formation of syncytia with low efficiency. In monolayer cell culture 57 systems, mature progeny virions germinate from the apical surfaces of polarized cells (Santangelo 58 and Bao, 2007), but the majority of them cannot be released from the host cells (Collins et al.,  Second, we tested whether the filopodia-driven virion intercellular delivery mode, which is virion-87 dependent, was completely blocked by a proper mAb concentration (as expected). In a simulation 88 model, RSV-A2-infected cells (preincubated with the high-titer mAb 5C4) were scraped with 89 preservation of their filopodia as much as possible and then coincubated with uninfected cells for 90 several hours. The ultrahigh-resolution microscopy data ( Figure 1C, 1D) showed that the virions 91 remained in the receptor cell membranes in the presence of antibodies, while the virions 92 successfully caused infection and produced filamentous progeny in the control sample without 93 antibodies. Therefore, the above two virion-based intercellular diffusion modes were completely 94 blocked by high titers of neutralizing antibodies. 95 Then, we utilized a coculture system to analyze the overall effect of the antibody on RSV 96 intercellular diffusion ( Figure 1E-1F). Infected cells (donor, red) were cocultured with uninfected 97 cells (receptor, green) in the presence or absence of the high-titer mAb 5C4. The double-positive 98 cells were then counted. We found that the high-titer mAb 5C4 significantly, but not completely, 99 suppressed intercellular diffusion, implying that there might be an unknown intercellular diffusion 100 mode that is insensitive to high-titer neutralizing antibodies.

119
The following source data is available for figure 1: 120 Source data 1. This spreadsheet contains data used to generate graphs in Figure 1B   Therefore, we analyze the differences between infected and uninfected HEp-2 cells from several 129 perspectives (Figure 2). At 24 h.p.i., the cells were fixed, permeabilized, and then stained with 130 WGA and 4',6-diamidino-2-phenylindole (DAPI). Our results showed that infected HEp-2 cells 131 (as well as A549 cells) developed more filopodia-like structures than uninfected cells, consistent 132 with a recent report (Mehedi et al., 2016). However, only open-ended intercellular nanotubes that 133 connected two cells were considered in this study. Aside from the number, the length and width of 134 these intercellular nanotubes were also greater than those of the nanotubes in uninfected cells 135 (

176
The following source data is available for figure 2: 177 Source data 1. This spreadsheet contains data used to generate graphs in Figure 2B, 2D, 2G, 2I, 2J.

182
To further study possible viral material transfer, we needed to separate it from virion-based spread. 183 Fortunately, material transfer, a virion-independent dissemination mode, should theoretically start 184 before virion assembly begins. Hence, we monitored activities in the early period of RSV infection 185 (from 2 h.p.i. to 24 h.p.i.) by staining the N protein to detect IBs and the F protein to detect progeny 186 virion assembly. As expectedly, in the early stage of RSV infection (3 h.p.i.) before virion assembly, 187 viral materials were already detected in the intercellular nanotubes ( Figure 3, white notched arrows). 188 Obvious virion assembly on the plasma membrane did not occur until 11 h.p.i. (Figure 3, yellow 189 filled arrows), which was the basis of filopodia-driven virion intercellular delivery. Thus, we 190 determined that the transfer of viral materials through intercellular nanotubes might be the major 191 intercellular spread mechanism in the early stage.

201
To clarify the viral materials in the intercellular nanotubes, we first costained four 202 nucleocapsid/polymerase proteins of RSV (N, P, L and M2-1). Unexpectedly, we found that these 203 four proteins colocalized in the nanotubes ( Figure 4A-4C, white notched arrows), which indicated 204 that these proteins might be transported through the tunnels in the form of IBs. We also explored 205 the relative localization of the N protein and RSV RNA in the nanotubes, and the results showed 206 that genomic RNA was always localized with the N protein, while mRNA was probably not ( Figure  207 4D-4E, white notched arrows). The composition of these intercellular nanotubes was determined 208 by costaining the N protein, F-actin and β-tubulin as well as the membrane. Polymerized F-actin 209 (microfilaments) was considered the primary skeleton of these nanotubes; however, β-tubulin was 210 only sporadically present in the tunnels ( Figure 4F). Therefore, the intercellular nanotubes referred 211 to above were tunneling nanotube (TNT)-like. In addition, the TNT-like nanotube membranes had

227
The following source data is available for figure 4: 228 Source data 1. This spreadsheet contains data used to generate graphs in Figure 4H. induced replication and expression. To answer this question, we utilized another coculture system 258 like that described above; the only difference was in the detection step. At six hours post mixing, 259 mKate-DIO+ cells were sorted and further cultured for another twenty-four or forty-eight hours 260 ( Figure 6A). According to the results of statistical analysis, the number of mKate+DIO+ cells 261 significantly increased, and RSV genomic RNA was amplified in these cells ( Figure 6B-6D). In 262 summary, the migration of viral materials to receptor cells successfully induced infection, which 263 could not be completely inhibited by neutralizing antibodies. 264 We next sought to evaluate the relative contribution of VMTIN to RSV spread in monolayer culture. 265 In the coculture group, both mKate-DIO+ cells and mKate+DIO+ cells were collected after several 266 hours of incubation. In the control group, a Transwell system was utilized to mimic cell-free virion 267 dissemination. We first seeded donor cells on the lower surface of the Transwell insert and then 268 inserted the insert into a Transwell chamber that was preseeded with receptor cells ( Figure 6E). 269 After a period of incubation, the insert was removed, and the receptor cells in the bottom of the 270 chamber were washed three times and collected for further detection. From the results ( Figure 6F-271 6H), we found that the start of N protein transcription and expression induced by VMTIN in the 272 presence of 5C4 occurred more quickly than that of cell-free virions but was slightly slower than 273 total spread.

295
The results are presented as the geometric mean ± SD.

296
The following source data is available for figure 6: 297 Source data 1. This spreadsheet contains data used to generate graphs in Figure 6C-D, Figure 6G-I, 6M.

298
As mentioned above, VMTIN was not completely inhibited by the pre-F-specific neutralizing 299 antibody, but whether viral proteins indeed participated in this process remained unclear. We 300 examined the influences of several drugs on VMTIN ( Figure 6J-6N), including AM22 (another 301 high-titer neutralizing antibody that specifically targets pre-F), ab27758 (an antibody targeting the 302 receptor of F, nucleolin) and sG (the secreted form of G protein). First, A-cells were infected with 303 RSV-A2 for 5 h and pretreated with AM22 at a concentration of 5 μg/ml, while B-cells were 304 prestained with the dye DIO and pretreated with different drugs. Thereafter, A-cells and B-cells 305 were digested and resuspended in drug-containing medium and then mixed at a ratio of 5 to 1. The 306 mixture was incubated for another 4 h. After that, part of the mixture was examined for the 307 existence of virions by immunofluorescence (IF) (Figure 6K), and then DIO+ cells were sorted 308 from the rest of the mixture and further incubated for another 24 h. Finally, the sorted DIO+ cells 309 were examined, and the newly infected cells were counted. The results showed that these three 310 drugs suppressed VMTIN, while their synergistic effect was not significant (Figure 6L-6M). The 311 results suggested that the F protein and G protein might play roles in VMTIN. 312

342
The following source data is available for figure 7: 343 Source data 1. This spreadsheet contains data used to generate graphs in Figure 7D. be observed in tissue (Neilson and Yunis, 1990), and most of the progeny virions produced were 388 not spontaneously released from the cell membrane (Collins et al., 2013b). Only approximately 5% 389 of the virions were released into the extracellular space as free virions. In addition, due to the 390 fragility of virions or the randomness of their movements, the propagation efficiency of cell-free 391 virions was low. In a recent study, it was found that intercellular filopodia directly delivered virions 392 that had not been released from RSV-infected cells to neighboring cells (Mehedi et al., 2016), and 393 this multicopy delivery was certainly efficient. 394 However, the aforementioned RSV transmission modes depended on the F protein or virions and 395 could theoretically be blocked by neutralizing antibodies. Interestingly, we found that intercellular 396 spread of RSV still occurred in the presence of high concentrations of neutralizing antibodies after 397 blockade of cell-free virion and filopodia-driven intercellular virion delivery (Figure 1). In addition, 398 we found that viral proteins were located in the open-ended intercellular nanotubes between 399 infected and uninfected cells, which pointed to a possible new mechanism of intercellular diffusion 400

in which infected cells transferred viral materials to neighboring cells via intercellular nanotubes. 401
We further observed this transfer process directly through ultrahigh-resolution live cell imaging. 402 We also found that this transmission occurred in the early stages of viral infection, before the 403 formation of virions, and was the major transmission event in the early stages of infection. The well plates were infected with RSV-A2 at an MOI of 4. At 24 h.p.i., the culture medium was 536 removed, and 2 ml of MEM with 0 μg/ml or 1 μg/ml 5C4 was added. After incubation and PBS 537 washing, the cells were scraped in 1 ml of MEM with 0 μg/ml or 1 μg/ml 5C4. Cells that had been 538 plated onto coverslips at a density of 2.5×105 cells per well in six-well plates and grown overnight 539 at 37 °C were incubated with 0.5 ml of liquid containing the scraped cells for 1 hour on a shaking 540 table at RT. After washing with PBS 3 times, the cells on coverslips were incubated with 2 ml of 541 complete medium with 0 μg/ml or 1 μg/ml 5C4. At 2 h and 16 h after the liquid containing the 542 scraped cells was added, the cells were fixed, stained and imaged. 543 Coculture assay for direct migration of viral materials from cell to cell. HEp-2 cells (5×105 cells 544 per well) in six-well plates were infected with RSV-A2-mKate (MOI=4). At 24 h.p.i., the culture 545 medium was removed, and 2 ml of MEM with 1 μg/ml 5C4 was added. After incubation and PBS 546 washing, the cells were digested with MEM with 1 μg/ml 5C4. A total of 5×105 naïve HEp-2 cells 547 were stained with Vybran DIO Cell-Labeling Solution (0.005 μg/ml, Thermo fisher) for 20 min at 548 37 °C and washed with PBS. Then, the infected cells were mixed with naï ve cells stained with DIO 549 in a 1:1 ratio in MEM with 1 μg/ml 5C4 (every 5×105 infected cells needed 2 μg of 5C4). The cells 550 were cultured, and the culture medium was changed every 6 h. 551 Flow cytometry analysis. Cells were washed with PBS, trypsinized with 0.25% trypsin-EDTA 552 (Gibco), incubated with culture medium to neutralize trypsin activity, and centrifuged at 162 ×g 553 for 3 min. The cells were washed and resuspended in PBS. The cell suspension was passed through 554 a 70-μm cell strainer, and the cells were subjected to analysis and sorting using a FACSAria III 555 flow cytometer (BD Biosciences, USA). HEp-2 cells infected with RSV-A2-mKate at an MOI of 556 4 and naï ve DIO-stained cells were used for compensation, and a fluorescence minus one control 557 was included to aid in setting gates. Acquisition was performed until 50,000 cells were recorded. with 0.2% Tween-20) and then incubated with secondary antibodies for 1 hour. After washing, the 574 membranes were imaged using an ImageQuant LAS4000 mini (GE Healthcare). 575

Preparation of low-DVG virus and high-DVG virus. Viruses containing different levels of 576
DVGs were produced according to methods in recent studies (Sun et al., 2015, Xu et al., 2017. 577 Statistical analysis. The results were plotted and statistical analysis was performed using 578 GraphPad Prism 9.00 (GraphPad, USA). Two populations data was performed with Mann-Whitney 579 test. Three or more populations data (one independent variable) was performed Kruskal-Wallis test 580 and Dunn's multiple comparisons test. Three or more populations data (two independent variables) 581 was performed Oridinary two-way ANOVA and Šídák's multiple comparisons test. Only a p value 582 <0.05 was considered to indicate statistical significance. Confidence interval was 95%.