Rhinovirus C replication is associated with the endoplasmic reticulum and triggers cytopathic effects in an in vitro model of human airway epithelium

The clinical impact of rhinovirus C (RV-C) is well-documented; yet, the viral life cycle remains poorly defined. Thus, we characterized RV-C15 replication at the single-cell level and its impact on the human airway epithelium (HAE) using a physiologically-relevant in vitro model. RV-C15 replication was restricted to ciliated cells where viral RNA levels peaked at 12 hours post-infection (hpi), correlating with elevated titers in the apical compartment at 24hpi. Notably, infection was associated with a loss of polarized expression of the RV-C receptor, cadherin-related family member 3. Visualization of double-stranded RNA (dsRNA) during RV-C15 replication revealed two distinct replication complex arrangements within the cell, likely corresponding to different time points in infection. To further define RV-C15 replication sites, we analyzed the expression and colocalization of giantin, phosphatidylinositol-4-phosphate, and calnexin with dsRNA. Despite observing Golgi fragmentation by immunofluorescence during RV-C15 infection as previously reported for other RVs, a high ratio of calnexin-dsRNA colocalization implicated the endoplasmic reticulum as the primary site for RV-C15 replication in HAE. RV-C15 infection was also associated with elevated stimulator of interferon genes (STING) expression and the induction of incomplete autophagy, a mechanism used by other RVs to facilitate non-lytic release of progeny virions. Notably, genetic depletion of STING in HAE attenuated RV-C15 and -A16 (but not -B14) replication, corroborating a previously proposed proviral role for STING in some RV infections. Finally, RV-C15 infection resulted in a temporary loss in epithelial barrier integrity and the translocation of tight junction proteins while a reduction in mucociliary clearance indicated cytopathic effects on epithelial function. Together, our findings identify both shared and unique features of RV-C replication compared to related rhinoviruses and define the impact of RV-C on both epithelial cell organization and tissue functionality–aspects of infection that may contribute to pathogenesis in vivo.

Introduction Rhinoviruses (RVs) are responsible for over 40% of respiratory virus infections in the human population [1][2][3][4]. Although well known as etiologic agents of the common cold, rhinoviruses can also infect the lower respiratory tract causing bronchiolitis or pneumonia and are a leading cause of virus-induced exacerbations in acute and chronic lung disease [5][6][7][8]. No vaccine or direct-acting antiviral is currently available due in part to the diversity of RVs in circulation, with over 160 genotypes identified [9][10]. These genotypes comprise three species (RV-A, RV-B, and RV-C) where RV-A and RV-C are the most prevalent and RV-C is associated with more severe clinical disease, especially in children [4,11,12]. Indeed, RV infection during the first year of life has been associated with wheezing episodes and is considered a risk factor for the development of asthma [8,13,14].
RV-C was discovered in 2006 [15] and compared to previously defined rhinovirus species, is unique at the genetic [16], structural [17], and biological level [18]. While all RVs (-A, -B, and -C) infect airway epithelial cells, RV-C uses a different host protein, cadherin-related family member 3 (CDHR3), to mediate particle uptake [19]. The restricted expression of CDHR3 to ciliated cells in the upper and lower airway epithelium [20][21][22][23] limits the cellular tropism of RV-C, compared to other RVs that utilize low-density lipoprotein receptor (LDLR) or intercellular adhesion molecule (ICAM)-1 as receptors [18]. Notably, a non-synonymous single nucleotide polymorphism (SNP; rs6967330 [A]) that yields stabilized CDHR3 protein expression at the cell surface is a causal variant for early childhood asthma with severe exacerbations [20]. Subsequent investigation has associated the CDHR3 asthma risk allele with heightened risk of respiratory tract illness with RV-C, but not other viruses [21]. More recently, stimulator of interferon genes (STING), a key adapter protein for cytosolic DNA-sensing pathways, was found to play a proviral role in RV-A and -C, but not -B, replication [24]. Thus, mechanisms of infection and replication are not always conserved between rhinovirus species.
Despite these advances, details of the RV-C life cycle and underlying mechanisms that contribute to pathogenesis remain scarce. While such studies remain hampered by the absence of immortalized cell lines or an in vivo mouse model that is naturally susceptible and highly permissive for RV-C replication, previous reports demonstrate that ex vivo tissue and primary airway cultures support infection [22,25]. Nonetheless, aside from receptor usage and cellular tropism, little is known about RV-C interactions with primary airway epithelial cells, the principal target for infection. Here we utilized both single-cell, microscopy-based analyses, and culture-wide measurements, to investigate the details of RV-C15 replication in an in vitro model of human airway epithelium (HAE). Our data identify the endoplasmic reticulum (ER) as the primary site for RV-C15 replication and demonstrate the impact of infection on the structural integrity of ciliated cells and epithelial barrier function, thereby identifying both unique and shared features of RV-C amongst related viruses.

RV-C15 replicates in ciliated cells, yielding changes in CDHR3 expression
Pseudostratified models of HAE at air-liquid interface are permissive for RV-C replication [25,26]. To detail the kinetics of RV-C15 replication in this model at higher-resolution, we inoculated HAE at 34˚C with 10 10 RV-C15 RNA copies and quantified viral RNA intracellularly as well as in both the apical and basolateral compartments over time (Fig 1A and 1B). The dynamics of RV-C15 replication were similar in HAE from two different donors, where cell-associated RV-C15 RNA levels increased during the first 12 hours post-infection (hpi). This correlated with the detection of double-stranded (ds) RNA, a marker of viral replication, by immunofluorescence (IF; S1A Fig) and was in line with peak viral release into the apical chamber at 24hpi (Fig 1A and 1B). The lack of RV-C15 RNA in basolateral supernatants confirmed polarized release of RV-C15 to the airway lumen (Fig 1A and 1B), consistent with a previous report in nasal cells [27] and the clinical manifestation of RV-C-mediated disease.
Given that HAE cultures were permissive for RV-C15 replication, we next sought to investigate cell tropism in our system. Visualization of the RV-C receptor, CDHR3, in non-infected HAE by IF revealed expression at the apical surface in cells that were also positive for Forkhead box protein J1 (FoxJ1; known to promote ciliogenesis) or acetylated alpha-tubulin (a marker of mature ciliated cells; Fig 1C, 1D and 1E). Notably, the corrected total cellular fluorescence (CTCF) levels of CDHR3 had a limited relationship with FoxJ1 but were strongly correlated with acetylated alpha-tubulin levels, in line with increased expression of CDHR3 during differentiation (S1B and S1C Fig) [21]. Corroborating these data and previous research reporting ciliated cell tropism for RV-C [22,25], we detected dsRNA primarily in FoxJ1(+) cells, with some cells also staining positive for acetylated alpha-tubulin (Figs 1F and S1D). Notably, the loss of apical acetylated alpha-tubulin localization in dsRNA(+) cells (Figs 1D and S1D) may explain the lack of cilia visualized in these cells by en face IF (Fig 1F). The 3D views also suggested CDHR3 was internalized in some dsRNA(+) cells (Fig 1D and 1E).
To better understand the dynamics of FoxJ1, acetylated alpha-tubulin, and CDHR3 during infection, we analyzed global protein expression by Western blot (WB) and quantified fluorescence levels following immunostaining specifically in infected cells. Beyond the intrinsic variation expected across cultures in this model system, neither FoxJ1 nor acetylated alpha-tubulin protein levels were dramatically altered over the course of infection (S1E Fig). In contrast, CDHR3 protein expression decreased at 12 and 48hpi, coincident with robust detection of the viral capsid protein VP1 (S1F Fig). Subsequent single-cell analysis of acetylated alpha-tubulin and FoxJ1 fluorescence levels corroborated our WB data (S1G and S1H Fig). This was also . Data shown are the mean +/-standard deviation across n = 3 cultures/time point. C-E: Immunofluorescence detection of CDHR3 (red) in non-infected or RV-C15-infected (dsRNA+, gray) ciliated cells identified by either acetylated α-tubulin (green; C and D) or FoxJ1 (green; E) at 12hpi. C: scale bar = 10μm. D and E: 3D visualization; z-stacks were at 1μm of thickness). F: Immunofluorescence detection of RV-C15 replication (dsRNA; red) in ciliated cells (FoxJ1, gray) with or without motile cilia (acetylated α-tubulin, green) at 12hpi (scale bar = 10μm). G: Quantification of dsRNA and CDHR3 fluorescence levels (CTCF) in non-infected (N) and RV-C15-infected (C15; dsRNA+) ciliated cells following immunofluorescence staining at 12hpi. Graphs show dsRNA and CDHR3 total CTCF (line = mean) quantified in HAE using FoxJ1 (N(1) and C15(1)) or acetylated α-tubulin (C15 (2) and N(2)) as a marker of ciliated cells. Dots represent the sum of all fluorescence intensity (Total CTCF) quantified per cell. Statistical analysis was done using the Mann-Whitney U test (Two-tailed; 0.95% confidence interval; � p<0.05, ��� p<0.001, ���� p<0.0001). true for CDHR3, in which fluorescence levels decreased significantly at 12hpi in cells with active viral replication (Fig 1G).

RV replication complex distribution is associated with vesicle formation
Picornaviruses, including RV-A and -B, replicate in association with cellular membranes, leading to the formation of double-membrane vesicles known as replicative complexes [28][29][30][31]. Thus, we investigated the distribution of replication complexes in RV-C15-infected HAE cultures by visualization of viral dsRNA and the impact of infection on intracellular membrane organization by transmission electron microscopy (TEM). While non-infected HAE were negative for dsRNA, as expected (Fig 2A), dsRNA in RV-C15-infected cultures was found in either a perinuclear (Fig 2B and S1 video) or ring-like disposition closer to apical and basolateral membranes (Fig 2C and S2 video). Neither phenotype was specific for RV-C15, however, as HAE infected with either RV-A16 or RV-A2 revealed similar dsRNA profiles (S2A- S2D  Fig). Additionally, the ring-like distribution pattern was determined to be the minority profile across all dsRNA(+) cells visualized at 12hpi in cultures infected with RV-C15 (73 of 324; 22.53%), RV-A16 (20 of 148; 13.51%), or RV-A2 (12 of 34; 35.29%).  [30]. These vesicles were detected in close proximity to mitochondria (Mt) and multi-vesicular bodies (MVB; E-upper-right panel; scale bar = 2μm), and were found to have double-membranes (DMV; E-lower-left panel; scale bar = 0.5μm). Clusters of electron-dense structures in the cytoplasm of RV-C15-infected HAE were seen, reminiscent of viral crystals (VC; E-lower-right panel; scale bar = 0.5μm) described for other picornaviruses [32,33]. Glyc = glycogen. https://doi.org/10.1371/journal.ppat.1010159.g002

PLOS PATHOGENS
Rhinovirus C replication in human airway epithelium Notably, the nuclei in cells with the ring-like dsRNA pattern were found near the apical cell surface instead of the usual basolateral location in polarized columnar cells (Figs 2C, S2B and S2D, possibly a result of large intracellular vesicle formation (S2E and S2F Fig). Subsequent TEM analysis in non-infected HAE revealed a differentiated epithelium with no overt cytopathic effects (Fig 2D-left panel) and a limited number of vacuoles in the cytoplasm (Fig 2Dright panel). In contrast, in RV-C15-infected HAE, we identified ciliated cells with many small vesicles clustered near multivesicular bodies, mitochondria, and electron-dense structures similar to β-particles of glycogen (Fig 2E-upper-left and upper-right panels, respectively). Under higher magnification, we were able to confirm these small vesicles had a double-membrane (Fig 2E-lower-left panel) similar to replicative complexes previously described for RV-A and RV-B [30]. Interestingly, some ciliated cells in RV-C15-infected HAE also contained electron-dense structures in their perinuclear region similar to the "viral crystals" observed in cells infected by other picornaviruses (Fig 2E-lower-right panel) [32,33].

RV-C15 infection triggers fragmentation of Golgi stacks and induces PI4P
Different cellular membranes can contribute to the formation of replication organelles during viral infection; however, the Golgi is the main source reported for many picornaviruses, including rhinoviruses [31,34]. Consequently, viral replication is associated with fragmentation of the Golgi stacks and changes in expression of phosphatidylinositol-4-phosphate (PI4P), a Golgi resident lipid [31,34,35]. To determine if RV-C replication induced similar effects, we inoculated HAE with RV-C15 -or RV-A16 and RV-A2 as positive controls [35,36]-and analyzed the Golgi by detection of giantin expression at 12hpi. While giantin visualization revealed a compact structure close to the nucleus in non-infected HAE (Fig 3A), giantin signal was spread throughout the cytoplasm in cells with evidence of active viral replication for all RVs tested (Fig 3A). The fragmentation of Golgi structures inferred from our IF data was further supported by TEM, where, compared to the non-infected control, Golgi stacks were barely visible in cells from RV-C15-infected HAE cultures (Fig 3B).
In addition to a loss of Golgi integrity, we also observed enhanced detection of PI4P in cells with active RV replication compared to non-infected controls. Similar to giantin, PI4P was dispersed throughout the cytoplasm in infected cells, reflecting the same general distribution (perinuclear or ring-like) as observed for dsRNA (Figs 3C, S3A and S3B). To further assess the changes in giantin and PI4P levels, we compared the CTCF for these cellular targets in both non-infected and infected cells. Our data revealed both giantin and PI4P levels were elevated and strongly correlated in HAE infected with all RVs, though the change in PI4P was not significant for RV-A2 (Figs 3D, 3E, S3C and S3D). A direct relationship was also observed between giantin and dsRNA CTCF in RV-C15-and RV-A16-infected cells while PI4P and viral dsRNA CTCF were only correlated in RV-C15 HAE (Figs 3E, S3C and S3D).

PLOS PATHOGENS
Rhinovirus C replication in human airway epithelium confirmed these observations, with less than 9% of the total dsRNA detected at the same location as giantin or PI4P in RV-C15-infected cells (Fig 3H and 3I, and S4-S9 Tables). Thus, despite the fragmentation of the Golgi stacks and increase of PI4P in airway epithelial cells with active RV-C15 replication, our data indicate the Golgi is not the main site for viral genome replication and PI4P is not a marker for sites of RV-C15 replication in HAE as observed for other RVs [31,34].

4
. Endoplasmic reticulum appears to be the RV-C15, but not RV-A16 or RV-A2, genome replication site Given the low ratio of colocalization between RV-C15 dsRNA and Golgi markers, we next evaluated ER membranes as a potential site for viral genome replication. WB analysis of the ER-associated protein calnexin in RV-C15-infected HAE did not indicate a dramatic change compared to non-infected cultures at 12, 24, or 48hpi (S4A Fig). However, at the single-cell level, calnexin fluorescence was not only elevated in cells infected with RV-C15, RV-A16, and RV-A2 compared to the negative control at 12hpi (Fig 4A and 4B) but also directly correlated with fluorescence levels of viral dsRNA (Fig 4C). Interestingly, the 3D view of HAE infected with RV-C15 at 12hpi showed calnexin spread throughout the cytoplasm, and there was a strong indication of calnexin and viral dsRNA colocalization (Fig 4A and 4D), which was further supported by 3D surface plot data (Fig 4E). Adding to this, the perinuclear and ring-like distributions of dsRNA were also observed for calnexin in RV-infected cells (S4B Fig).
Pixel intensity-based analysis confirmed the high level of colocalization between viral dsRNA and calnexin in cells infected not only with RV-C15 (56.3%), but also RV-A16 (28%) and RV-A2 (49.1%) at 12hpi (S10-S12 Tables). However, this high colocalization ratio was only confirmed for RV-C15-infected cells at the 3D level (spatial colocalization analysis; Fig  4F and 4G), suggesting that, unlike RV-A16 and RV-A2, the ER is the main site for RV-C15 genome replication.

RV-C15 infection induces incomplete autophagy
Previous studies have shown picornaviruses, including several rhinoviruses, manipulate the autophagy pathway to mediate non-lytic release of progeny following replication [37]. Since the induction of autophagy is rhinovirus genotype-specific [37][38][39], we sought to determine if RV-C15 triggered the induction of autophagy in HAE. Towards this goal, we inoculated cultures with RV-C15 alongside RV-A16 and RV-A2 as controls, and probed for Lysosome-associated membrane glycoprotein 1 (Lamp-1; a lysosome marker) and LC3b (an autophagosome marker) by WB. LC3b protein levels were minimal in non-infected HAE, suggesting a low basal rate of autophagy while Lamp-1, LC3b-I, and LC3b-II expression varied across donors and time points during RV-C15 infection. Still, LC3b-I and LC3b-II levels were clearly elevated at 12 and 48hpi in cultures with robust infection evidenced by RV-C15 VP1 protein levels ( Fig  5A). Therefore, we further evaluated levels of both Lamp-1 and LC3b in HAE by IF 12hpi. Lamp-1 and LC3b detection in HAE was significantly stronger in RV-C15-and RV-A2-infected cells while the level of detection in RV-A16-infected HAE was similar to the negative control (Fig 5B and 5C). Curiously, the 3D view of non-infected cells showed an apical localization of Lamp-1 while this protein was spread through the cytoplasm in RV-infected cells (Fig 5B). A high correlation ratio between Lamp-1 fluorescence levels and dsRNA was obtained in RV-infected cells; however, LC3b levels strongly correlated with dsRNA only in cultures infected with RV-C15 and RV-A2 (Figs 5D and S5). Together, these data suggest RV-C15 and RV-A2 (but not RV-A16) induce autophagy to significant levels over the baseline in HAE.

PLOS PATHOGENS
Rhinovirus C replication in human airway epithelium ER is a site for RV-C15 genome replication in HAE. A: Detection of calnexin (ER marker, green; nuclei, blue) in non-infected or rhinovirus-infected (dsRNA+, red) HAE at 12hpi (z-stacks at 1μm of thickness; scale bar = 5μm). B: Quantification of fluorescence levels (CTCF; line = mean) of dsRNA and calnexin in non-infected cells and RV-infected (dsRNA+) HAE detected by immunofluorescence at 12hpi. Dots represent the total CTCF per cell. Statistical analysis was done using Kruskal-Wallis test following Dunn's multi-comparison test ( �� p<0.01, ���� p<0.0001). C: Spearman correlation analysis (two-tailed; 0.95% confidence interval) between calnexin and dsRNA fluorescence levels (CTCF) in RV-C15, -A16, or -A2-infected HAE at 12hpi. Each dot represents the fluorescence levels of dsRNA and calnexin quantified in the same Z-slice. D: 3D view of RV-C15-infected HAE shows the immunodetection of calnexin (green), viral dsRNA (red), and calnexin/dsRNA colocalization (yellow) at 12hpi (z-stacks at 1μm of thickness). E: 3D surface plot of RV-C15-infected HAE shows the detection of calnexin (green), dsRNA (red), and calnexin/dsRNA colocalization (yellow) by immunofluorescence at 12hpi. F-G: Spatial colocalization analysis Since the inhibition of autolysosome formation (the last step in the autophagy pathway) is characteristic of picornaviruses that induce this pathway to promote virus release [40], we further evaluated autolysosome formation defined by Lamp-1 and LC3b colocalization. Results from spatial colocalization analysis in cultures infected with RVs indicated greater detection of autolysosomes in RV-A16-infected cells compared to RV-C15-and RV-A2-infected cells (Fig  5E and S13-S15 Tables). Thus, these data suggest that, similar to RV-A2 but unlike RV-A16, RV-C15 induces incomplete autophagy in HAE.

STING expression promotes replication of RV-C15 in HAE
In this study, we observed RV-C15 genome replication in HAE to be associated with the ER (Fig 4) and the induction of incomplete autophagy at 12hpi (Fig 5). Interestingly, STING, which localizes to the ER, was recently shown to induce the autophagy pathway during RV-A16 infection [41] and was identified as a proviral factor for RV-A and RV-C, but not RV-B, where STING overexpression facilitated viral genome replication in Huh-7 cells [24]. Based on these data, we investigated the expression of STING in HAE infected with RV-C15. Interestingly, we detected an increase in STING expression overtime, in line with increasing viral VP1 levels (Fig 6A) as well as robust detection of STING by IF in dsRNA(+) ciliated cells at 12hpi (Fig 6B). Notably, 3D view analysis revealed that dsRNA/STING-double positive cells also exhibited dispersed distribution of acetylated alpha-tubulin, suggesting STING levels increase as replication progresses (Fig 6C). Indeed, the CTCF of STING not only increased significantly in HAE infected with RV-C15 (Fig 6D), but also strongly correlated with dsRNA levels at 12hpi (Fig 6E). However, despite the maximum Z-projection, 3D surface plot, and pixel intensity colocalization analysis suggesting high level of dsRNA/STING colocalization in RV-C15-infected HAE (Fig 6F and 6G, and S16 Table), spatial colocalization analysis indicated the median colocalization ratio between both targets was lower than 5% (Fig 6H).
To better understand the importance of STING to RV-C15 genome replication, we genetically knocked-out STING using a CRISPR/Cas9 approach in immortalized human airway epithelial cells (BCi-NS1.1 cells [42]), differentiated them into HAE cultures (Fig 6I), and infected them with RV-C15, RV-A16, or RV-B14. Immunofluorescence detection of dsRNA at 12hpi confirmed the susceptibility of control cultures expressing a non-targeting guide RNA to RV-C15, RV-A16, and RV-B14 infection (Fig 6J). However, the frequency of dsRNA(+) cells was lower in STING-depleted HAE infected with RV-C15 or RV-A16 (Fig 6J). While infection with RV-B14 was less efficient overall, a similar frequency of dsRNA(+) cells were observed in both control and knockout HAE (Fig 6J). STING-knockout cultures also had significantly lower intracellular viral RNA copy numbers for RV-C15 and RV-A16 but not RV-B14, further validating our results (Fig 6K) and suggesting that the increase of STING expression in RV-C15-infected HAE has an advantageous impact on viral genome replication.

PLOS PATHOGENS
Rhinovirus C replication in human airway epithelium

PLOS PATHOGENS
Rhinovirus C replication in human airway epithelium was restricted to the apical compartment and did not exceed 20% of the maximum levels obtained from lysed control cultures; however, LDH levels increased over time (Fig 7A) in line with RV-C15 replication kinetics (Fig 1A and 1B). Thus, we sought to further evaluate the global impact of RV-C15 infection on epithelial integrity. Interrogation of HAE permeability during RV-C15 infection revealed a significant, albeit transient, decrease in transepithelial electrical resistance (TEER) at 12hpi (Fig 7B). Since the transepithelial transport of ions is mediated by pores formed by integral membrane proteins termed claudins [43], we next investigated the expression and localization of claudin-1 during RV-C15 infection in HAE. While claudin-1 protein levels in cultures infected with RV-C15 did not vary (Fig 7C), we observed a change in claudin-1 distribution in RV-C15-infected cells at 12hpi by IF (Fig 7D). To determine if this translocation was specific to claudin-1, we characterized Zona occludens 1 (ZO-1), a cytoplasmic tight junction-associated protein. Similar to claudin-1, with the exception of one donor at 48hpi, ZO-1 protein levels for both subunits (ZO-1 +α and ZO-1 -α) did not vary in RV-C15-infected HAE compared to the negative control (Fig 7C), while ZO-1 disposition was more widespread in dsRNA(+) cells (Fig 7E). To quantify the effect of RV-C15 infection on ZO-1 disposition over time, we utilized the Junction Analyzer Program (JAnaP) [44][45][46], which allowed us to assess the profile of ZO-1 detected in non-infected and RV-C15-infected cells (Fig 7F). The increase of discontinuous ZO-1 quantified in this assay (Fig 7G) indicates the ZO-1 translocation in HAE (Fig 7E) parallels the progression of RV-C15-infection. Additionally, two profiles of discontinuous ZO-1 were evaluated, and ZO-1 was found in a more perpendicular than punctual disposition in RV-C15-infected HAE (Fig 7H and 7I).
Given the ciliated cell tropism RV-C15 and the important role of these cells in promoting mucociliary clearance (MCC), we also assessed the global impact of RV-C15 infection on mucus transport. HAE cultures were equilibrated at 34˚C and the transport rate of red-fluorescent microspheres in the extracellular mucus gel was calculated immediately prior to inoculation, and up to 48hpi. IF confirmed viral replication at 12hpi in HAE inoculated with viable, but not UV-inactivated, RV-C15 (S6 Fig). The transport of microspheres indicated a slight increase in MCC in HAE inoculated with PBS and a similar, albeit significant, elevation in microsphere transport rate in UV-inactivated RV-C15-inoculated cultures at 12hpi (Fig 7J), likely attributable to the addition of fluid on the apical surface after the T = 0 time point. Despite this, RV-C15 infection resulted in a significant decrease in MCC at both 12 and 24hpi (Figs 7J and S6B) indicating viral replication impairs this innate host defense mechanism. showing STING and VP1 expression at 12 and 48hpi in non-infected or RV-C15-infected HAE. Fold change graph represents STING protein levels normalized to the endogenous control (actin) and compared to non-infected cultures. Data shown are from two independent donors, represented by circles and triangles. Blot above is from the donor represented by triangles. B: Immunofluorescence detection of STING (green) in non-infected or RV-C15-infected (dsRNA+; red) ciliated cells (acetylated α-tubulin, magenta) at 12hpi (scale bar = 50μm). C: XZ orthogonal view (upper) and 3D view (below) of RV-C15-infected HAE shows the detection of STING (green) and viral dsRNA (red) in ciliated cells (acetylated αtubulin, gray) at 12hpi (z-stacks at 1μm of thickness). D: Quantification of dsRNA and STING fluorescence levels (CTCF) in non-infected and RV-C15-infected (dsRNA+) HAE at 12hpi. Dots represent the total CTCF per cell and the line indicates the mean. Statistical analyses were done by Mann-Whitney test (two-tailed; 0.95% confidence interval; ���� p<0.0001). E: Spearman correlation analysis (two-tailed; 0.95% confidence interval) between fluorescence levels for dsRNA and STING in RV-C15-infected HAE at 12hpi. Each dot represents the fluorescence levels of dsRNA and STING quantified in the same Z-slice. F: 2D split channel views of RV-C15-infected HAE showing the detection of STING (green), dsRNA (red), and STING/dsRNA colocalization (yellow) by immunofluorescence at 12hpi (scale bar = 5μm). G: 3D surface plot of RV-C15-infected HAE shows the detection of STING (green), dsRNA (red), and STING/dsRNA colocalization (yellow) by immunofluorescence at 12hpi. H: Spatial colocalization analysis shows the ratio of dsRNA/STING colocalization in RV-C15-infected HAE at 12hpi (bar = median). I: Western blot showing STING expression, quantified below, in HAE derived from BCi-NS1.1 cells transduced with non-targeting control (NTC), sgRNA1 (sg1), or sgRNA2 (sg2). J: Immunofluorescence detection of dsRNA (red) in HAE derived from BCi-NS1.1 cells transduced with NTC, sgRNA1 (sg1), or sgRNA2 (sg2) infected with RV-C15, RV-A16, or RV-B14 at 12hpi (scale bar = 100μm). K: qPCR to detect RV RNA in HAE derived from BCi-NS1.1 cells transduced with NTC, sgRNA1 (sg1), or sgRNA2 (sg2) and infected with RV-C15, RV-A16, or RV-B14 at 4 and 12hpi. Each bar represents the mean value obtained across n = 3 cultures/condition (individual dots); the black line represents the limit of detection (RV-C15 = 10 3.5 ; RV-A16 = 10 5.9 (not visible on the graph); RV-B14 = 10 4.8 ); statistical analysis was done by Ordinary One-way ANOVA and Tukey's multiple comparisons test ( �� p<0.01). https://doi.org/10.1371/journal.ppat.1010159.g006

Discussion
RV is becoming increasingly recognized as a cause of both upper and lower respiratory tract infection [4][5][6][7][8]. Furthermore, the risk for development of asthma in children after infection by RV, especially RV-C [8,13,14], highlights the clinical relevance of these viruses. Due to restricted receptor expression in traditional immortalized cell culture systems [26], we investigated RV-C15 replication using an in vitro model of HAE that supports the entire RV-C life cycle. Previous studies in similar systems have shown the impact of temperature and cell differentiation stage on RV-C replication [22,25,27]. Here, we detail a multi-step replication curve for RV-C15 (Fig 1A and 1B), indicating the peak of intracellular viral load at 12hpi by qPCR and IF (Figs 1A, 1B and S1A) and peak viral release at 24hpi (Fig 1A and 1B) in agreement with a previous report characterizing RV-C11 and RV-C15 infection in bronchial cells [26].
In well-differentiated HAE, we detected CDHR3 on the apical surface of ciliated cells (Fig  1C-1E). Interestingly, we did not observe any evidence of cytosolic CDHR3 in non-infected cells, differing from previous immunofluorescence data [21,23] but in agreement with CDHR3 detection on cilia by TEM [21]. However, CDHR3 distribution changed and protein levels decreased in RV-C15-infected cells (Figs 1C-1E, 1G and S1F) without altering levels of acetylated alpha-tubulin or FoxJ1 (S1E,S1G and S1H Fig). Unlike ICAM-1 (the receptor for majorgroup RVs) and LDLR (the receptor for minor-group RVs), the dynamics of CHDR3 during infection and associated pathway(s) of RV-C entry remain unknown. After endocytosis of RV-A and RV-B particles, ICAM-1 and LDLR are either degraded in the lysosome with empty capsids, or LDLR is trafficked from the endosome to be recycled [47]. The intracellular distribution of CDHR3 and the loss of protein expression after RV-C15 infection in HAE suggests CDHR3 follows the same path as other RV receptors.
After confirming the susceptibility of our HAE models to RV-C15 infection, we further investigated the localization and association of replication complexes with various organelle markers at the single-cell level. Using dsRNA as a marker for virus replication, we observed both perinuclear and ring-like distribution of replication centers closer to the basolateral membrane during RV-C15, RV-A16, and RV-A2 replication (Figs 2A-2C and S2A-S2D). To our knowledge, the latter profile has not been described during RV or any other picornavirus infection, and our TEM data indicates the formation of large intracellular vesicles filled with unknown electron-dense content may be the underlying cause of altered dsRNA disposition. Interestingly, other cellular markers such as giantin, PI4P, and calnexin (S3D and S4B Figs) showed a similar ring-like distribution in these cells, further suggesting the presence of a large vesicle that precipitates global changes in intracellular organization. Whether these large . Each bar represents the mean value obtained across n = 3 cultures/condition (individual dots); statistical analysis was done by Ordinary one-way ANOVA following Tukey's multiple comparisons test ( � p<0.05, ns = nonsignificant). C: Fold change graph represents claudin-1 and zona occludens-1 (ZO-1) α+ and α-protein levels normalized to the endogenous control (actin) and compared to non-infected cultures. Data shown are from three independent donors, represented by diamonds, triangles, and squares. Blot above is from the donor represented by triangles, which includes the detection of VP1 of RV-C15. D-E: Visualization of claudin-1 (D, gray) and ZO-1 (E, gray) in non-infected and RV-C15-infected (dsRNA+, red) HAE (nuclei, blue) detected by immunofluorescence at 12hpi (z-stacks at 1μm of thickness; scale bar = 5μm). F-I: Quantification of ZO-1, detected by immunofluorescence, in non-infected and RV-C15-infected using the JAnaP software [44][45][46]. Total ZO-1 detected per cell perimeter (percentage of coverage; F); detection of discontinuous ZO-1 (disrupted; G)-perpendicular (H) and punctate (I) profiles. Each bar represents the mean ZO-1 value obtained across different fields per culture/time-point (individual dots). Statistical analysis was done by Ordinary one-way ANOVA following Dunnett's multiple-comparison test ( � p<0.05, ���� p<0.0001). J: Mucociliary clearance (MCC) quantified in non-infected HAE (negative control) and HAE inoculated with either UV-RV-C15 or RV-C15 (10 10 RNA). Box represents the 10-90 percentile values of MCC quantified in the same culture/condition at different time-points. Statistical analysis was done by Ordinary one-way ANOVA following Dunnett's multiple-comparison test ( � p<0.05, ��� p<0.001, ���� p<0.0001). https://doi.org/10.1371/journal.ppat.1010159.g007

PLOS PATHOGENS
Rhinovirus C replication in human airway epithelium vesicles play a specific role in RV replication or if they are simply indicative of cellular changes that precede cell death is unknown. Indeed, we hypothesize that the perinuclear and ring-like patterns of dsRNA may be indicative of different stages of infection, with the ring-like formation occurring later during the viral life cycle, coincident with more advanced cytopathic effects.
Notably, our TEM analysis also identified smaller, double-membraned vesicles, similar to those identified during replication of other RVs and related picornaviruses [28][29][30][31] that likely represent sites of RV-C15 genome replication (Fig 2D and 2E). Prior reports identified the induction of replicative complexes derived from the ER by the poliovirus 2BC and 3A proteins [48,49] and the same proteins in RV-A16 were associated with the ER at an early stage of viral infection [35]. Still, work to date suggests the Golgi is the primary site for picornavirus genome replication [28,31,34]. Further, the formation of replicative complexes has been associated with cholesterol exchange driven by OSBP1 [31], accumulation of ER-vesicles resulting from the inhibition of ERGIC-to-Golgi transport [36], and fragmentation of Golgi stacks induced by the viral 3A protein [34,35]. In this study, we observed an increase of giantin and PI4P levels, dissolution of Golgi stacks, and spread of PI4P-positive vesicles in RV-C15-infected cells (Figs 3A-3D and S3A) suggesting RV-C impacts Golgi membranes similar to other picornaviruses. However, despite the strong correlation between dsRNA and fluorescence levels of both giantin and PI4P (Fig 3E), colocalization analysis indicated neither giantin-nor PI4P-positive vesicles are the main site for RV-C15 genome replication (Fig 3H and 3I). Similar observations were made in RV-A16-infected HAE (Fig 3) although dsRNA fluorescence levels only strongly correlated with giantin and not PI4P (S3B Fig). These data were unexpected, as prior work in HeLa and nasal epithelial cells noted PI4P enrichment of Golgi membranes during RV-A16 genome replication [31]. Different from the literature [31], Golgi and PI4P-positive vesicles were not the main site for RV-A2 genome replication in HAE either (Fig 3H and 3I), and neither the significant increase in levels of giantin nor insignificant increase in PI4P correlated strongly with viral dsRNA (Figs 3D and S3C). It is important to highlight that our study evaluated rhinovirus replication in HAE at a later time point than other reports (3-8hpi) [50][51][52]; thus, our data do not exclude the possibility that Golgi and PI4P-positive membranes support RV replication earlier in infection.
Most notably, our results demonstrated elevated levels of calnexin in RV-infected HAE (Fig  4A and 4B) and a high ratio of dsRNA/calnexin colocalization in cells infected with RV-C15, but not RV-A2 and RV-A16 (Fig 4E-4G), identifying the ER as the main site for RV-C15 replication. The increase in calnexin levels observed in this study could be the result of ER stress in infected cells, as reported for RV-A1B and RV-A16 [53,54]. The RV-A16 2B protein [54] forms pores on ER membranes, promoting calcium ion efflux, which increases membrane permeability and triggers stress. These effects on ER membranes have been associated with more efficient release of vesicles that may represent additional sites for picornavirus replication [55]. Notably, inhibition of ER stress resulted in a decrease in RV-A1B replication [53] corroborating the hypothesis that ER-derived vesicles are a site for viral replication [55] and supporting our conclusion that RV-C15 replication is associated with the ER in HAE (Fig 4E-4G).
In the final stages of the viral life cycle, progeny picornavirus virions are released through cell lysis or by usurping the autophagy pathway [37][38][39]. Exploitation of the autophagy pathway was originally shown for poliovirus [40] and is characterized by virus-mediated inhibition of autolysosome formation and release of autophagosomes full of nascent virions that likely protects particles from the immune response [38]. RV-A2 induced incomplete autophagy in HAE in contrast to RV-A16 (Fig 5B and 5C), as expected [39]. This conclusion was based on elevated Lamp-1 and LC3b levels, but a very low ratio of Lamp-1/LC3b colocalization indicating a failure to form autolysosomes (Fig 5E). The results obtained in RV-C15-infected HAE were similar to RV-A2, indicating the induction and manipulation of the autophagy pathway (Fig 5B, 5C and 5E). Use of the autophagy pathway to release progeny would explain the absence of cytopathic effects visible under light microscopy in HAE cultures infected with RV-C15. To our knowledge, this is the first time the autophagy pathway has been analyzed during RV-C infection.
The identification of ER as a site for genome replication and induction of autophagy during RV-C15 infection of HAE led us to investigate STING, which has been associated with the induction of ER stress and autophagy [41,56]. In this study, we detected an increase in STING expression in RV-C15-infected HAE (Fig 6A, 6B, 6D and 6F) which correlated with levels of dsRNA (Fig 6E) and VP1 protein (Fig 6A). This correlation, however, was not related to a close interaction between STING and dsRNA, as the median colocalization ratio at the spatial level was lower than 5% (Fig 6H). Additionally, viral genome replication was impacted in the absence of STING, resulting in significantly lower intracellular viral RNA levels at 12hpi for RV-C15 and RV-A16, but not RV-B14 (Fig 6J and 6K). These data, which assayed endogenous STING expression in RV-infected HAE, corroborate published data in undifferentiated cells indicating STING expression is important for RV-A and RV-C genome replication [24]. STING is an ER-resident transmembrane protein which can be phosphorylated by TANKbinding Kinase I (TBK1) or translocated to the ER-Golgi intermediate compartment (ERGIC) before phosphorylation [57]. The phosphorylation of STING in the ER is necessary for autophagy induction during RV-A16 infection [41], while the translocation to the ERGIC is required for ER stress [56]. The low ratio of dsRNA/STING spatial colocalization at 12hpi (Fig  6H) is surprising and indicates that either genome replication of RV-C15 occurs at different sites on ER membranes (STING-negative) or most of STING had already been translocated to ERGIC. RV-C15 replication also induced autophagy in HAE, though more experiments are needed to relate this phenomenon to STING activation.
Given the observed impacts of RV-C15 replication in infected cells, we further probed the effects of viral infection on the epithelium functionality, including barrier permeability and mucociliary clearance. The level of cytotoxicity of RV-C15 infection in HAE, measured by LDH release, increased overtime in the apical compartment only (Fig 7A), in agreement with viral cellular tropism for ciliated epithelial cells that face the luminal surface of the airway (Fig  1) [22,25]. The dissociation of tight junctions during viral infection, including RV-A and RV-B, has been associated with the loss of epithelial-selective permeability and an increase in bacteria translocation across the epithelium [58]. A previous analysis of RV-A16, RV-A1B, and RV-A39 found the reduction of TEER and increase in epithelium permeability for inulin were associated with both claudin-1 and ZO-1 dissociation [55]. Similar to these reports, RV-C15 temporarily increased epithelial permeability to small particles and ions (as indicated by a temporary loss in TEER, Fig 7B) which may be due, at least in part, to claudin-1 translocation (Fig 7D). Additional claudins (e.g., claudin-3, -4 and -5) that are also expressed in the airway epithelium and known to impact transcellular transport were not assayed here [59,60]. Still, the altered expression of tight junction proteins was not restricted to claudin-1 as we also detected ZO-1 throughout the cytoplasm in dsRNA(+) cells without any change in overall protein levels during infection (Fig 7C and 7E). Despite being a tight junction protein, ZO-1 interactions are more important to cell polarization and paracellular permeability [61]. Although we did not check paracellular transport of large particles, the observed disrupted states for ZO-1 indicated progressive translocation with ongoing viral infection (Fig 7F-7I), possibly impacting cellular organization. Nonetheless, the loss of TEER and changes in claudin-1 and ZO-1 observed in this study for RV-C15 were less dramatic than previous reports for RV-A and RV-B [37,39,58]. The lesser impact on epithelium permeability caused by RV-C15 infection may be associated with an earlier induction of the mechanism of repair, which could result from the shift in cell metabolism to a glycolysis state followed by lipid biosynthesis [62].
Beyond epithelial barrier integrity, we also identified an impact of RV-C15 on mucociliary clearance functionality. MCC decreased significantly after 12hpi in HAE infected with RV-C15 but not UV-inactivated virus, similar to a previous study that reported an impact on MCC 5 days post-infection [63], and indicating that active viral replication interferes with this cellular mechanism of pathogen removal (Fig 7J). We speculate that the temporary increase in MCC observed in cultures inoculated with UV-inactivated virus is due to the addition of fluid on the apical surface. However, as this increase was not significant for the negative control, it may also indicate that the host response to incoming viral particles stimulates MCC. The coordinated beating of ciliated cells drives the basal MCC rate of~5.5 mm/min, which can be altered by mucus composition, temperature, and humidity [64][65][66]. Indeed, mucus hypersecretion is a known symptom during RV infections [67], which has also been associated with changes in mucus composition [68]. In this study, we noted altered distribution of acetylated alpha-tubulin in cells with active viral replication (Fig 1D and 1F) which likely indicates impaired cilium structure and function in these cells. Whether the progressive loss of cilia function and eventual death of infected cells contributes to our observed change in MCC during RV-C15 infection is not clear.
In conclusion, our present study expands our current understanding of RV replication in a physiologically-relevant setting, substantiating previous observations related to virus-induced membrane reorganization, and demonstrating that RV-C15 displays many typical features of picornavirus infection. In addition, we identify RV-C15 replication in association with ER membranes unlike previously characterized RVs. While we speculate that these data identify a unique feature of RV-C replication, our study is limited by the use of only one RV-C genotype; thus, further analysis to confirm these observations using additional RV-C isolates is important. Notwithstanding, our findings underscore the fact that virus-host interactions critical for rhinovirus replication are not always conserved, possibly contributing to their different clinical profiles, and supporting further investigation into this group of pathogens.
For target cell transduction, lentivirus-containing supernatants were applied to BCi-NS1.1 cells (kindly provided by Drs. Matthew Walters and Ronald Crystal [42]), maintained as HAE at 40-60% confluence with a final concentration of 20mM HEPES (Gibco) and 4μg/mL Polybrene (Thermo Scientific). Cells were then centrifuged (1,000 x g for one hour at 37˚C) and incubated at 37˚C with 5% CO 2 for 6hr. The inoculum was removed and replaced with fresh Pneumacult-Ex Plus media (StemCell Technologies). At 60-80% confluence, eGFP-positive cells were enriched by fluorescence-activated cell sorting on a BD FACSAria-II Cell Sorter (BD Bioscience, San Jose, CA). These sorted BCi-NS1.1 cells were then expanded, seeded at 3.3x10 4 cells/well on 6.5mm rat-tail collagen type I-coated transwell membranes, and cultured at air-liquid interface as previously described for HAE.

PLOS PATHOGENS
Rhinovirus C replication in human airway epithelium 24-hour incubation at RT and plaques visualized by staining the monolayer with 0.1% Crystal Violet (#C0775; Sigma-Aldrich) diluted in a 20% ethanol (200 proof) solution.

RV infection in HAE and H1 HeLa cells
One week before the experiment in HAE, cells were washed 3 x 30 minutes with 100μl of phosphate buffered saline (PBS; #P5119; Gibco) at 37˚C to remove excessive mucus. In CRISPR/ Cas9-modified BCi-NS1.1-derived HAE an additional wash was performed immediately before inoculation. HAE cultures were inoculated on the apical surface with 10μl of PBS (negative control), sucrose-purified RV-C15 (10 10 copies of RNA), RV-A16 (5x10 5 PFU), RV-B14 (5x10 5 PFU), or RV-A2 (5x10 5 PFU) and incubated at 34˚C with 5% CO 2 until sample collection. In the replication curve and cytotoxicity assays, the inoculum was removed, and the apical surface rinsed with PBS, at 4hpi.
For HAE culture sample collection, 100μl of PBS (Gibco) was added to the apical chamber and harvested after a 30-minute incubation at 34˚C; basolateral samples (500μl ALI media) were collected directly from underneath the culture. Cells were harvested from the transwell membranes using 350μl RLT buffer (#74104; Qiagen) for downstream RNA extraction, or 75μl of Radio Immuno-Precipitation Assay (RIPA) buffer (#89900S; Thermo) plus protease inhibitors (#88666; Thermo Scientific) for subsequent Western blot (WB) assays. The aliquots for RNA extraction and WB were stored at -80˚C and -20˚C, respectively. Alternatively, cultures were fixed with 4% (w/v) freshly-prepared paraformaldehyde (#157-8-100; Electron Microscopy Sciences (EMS), Hatfield, PA) for immunofluorescence (IF) assays or 2% (w/v) glutaraldehyde in 0.1M cacodylate buffer for downstream transmission electron microscopy analysis.

IF assays in HAE and H1 HeLa cells
HAE cultures were fixed with 100μl (apical) and 500μl (basolateral) 4% (w/v) freshly prepared paraformaldehyde (#157-8-100; EMS). After a 15-minute incubation at RT, cultures were washed once for 3 minutes with PBS and then incubated for 10 minutes at RT with a quenching solution (25mM NH 4 Cl diluted in PBS) before being washed twice for 3 minutes each with PBS. Permeabilization of cell membranes was done using 0.2% Triton-X diluted in PBS; after a 15-minute incubation at RT, the cells were washed 3 x 3 minutes with PBS. A blocking solution (10% normal donkey (#017-000-121) or normal goat serum (#005-000-121, Jackson Immu-noResearch Labs, West Grove, PA) diluted in PBS with 0.2% Tween-20) was added and only removed from the apical side after a 10-minute incubation at RT. The apical surface of the

PLOS PATHOGENS
Rhinovirus C replication in human airway epithelium culture was washed 3 x 3 minutes with PBS. Primary antibody diluted in PBS with 0.2% Tween-20 and 1% Bovine Serum Albumin (BSA; #BP9700100; Fisher Scientific, Hampton, NH) was added to the apical chamber and incubated overnight at 4˚C, protected from light. The apical surface was washed 3 x 5 minutes with PBS and incubated with the secondary antibody diluted in PBS with 1% BSA. After a 2-hour incubation at RT, protected from light, the secondary antibody solution was removed and the apical surface of HAE was washed 3 x 5 minutes with PBS. If necessary, additional stains were done by repeating this protocol; reducing the secondary antibody incubation to 20 minutes when the next primary antibody applied was generated in the same species as another primary antibody used in the same experiment. Fluorescent-conjugated primary antibodies were applied during the final round of staining, where applicable. Nuclei were visualized in the final step by adding 1ng/ml of Hoechst 33342 (#H3570; Invitrogen, Thermo) diluted in PBS with 0.2% Tween-20 to the apical surface. After a 5-minute incubation at RT, the cultures were washed 3 x 5 minutes with PBS, after which the transwell membrane was separated from the plastic holder and mounted on a glass slide using VectaShield Antifade mounting medium (#H-1000-10; Vector Laboratories, Burlingame, CA) and 1.5mm-thick cover glass.
Images were acquired with a Zeiss Axio Observer 3 Inverted fluorescence microscope (Cell Observer HS image system, Zeiss Axiocam 503 mono camera, optimal acquisition mode, AIM-Zen 2007 software) equipped with EC-Plan-NEOFLUOR 20x/0.5NA Ph2 and 40x/ 0.75NA Ph2 air objectives. Higher resolution images and z-series optical sections (at 1μM intervals) were acquired with a LSM710 Zeiss confocal microscope (Argon laser, pinhole of 1 airy unit (AU), zoom factor of two, optimal acquisition mode, AIM-Zen 2009 software) equipped with a 63x/1.4NA Oil DIC Plan Apo objective at the Imaging Core, University of Maryland, College Park. All images were analyzed using Fiji-ImageJ v.2.1.0/1.53c software [74]. Z-series from a pre-selected region of interest (ROI) are displayed as maximum intensity z-projection, orthogonal views (XY, XZ and YZ planes), and 3D surface plots; they were also used for 3D volume reconstruction (3D view) using Fiji [74].
The IF protocol described above was used in all assays with the following exception: for PI4P detection, permeabilization was done using a 20mM Digitonin solution (diluted in PBS). Antibodies used in this study are listed below (Tables 1 and 2).

Quantification of fluorescence levels and colocalization analysis
The CZI files from z-stacks collected with a step-size at 1μM were analyzed at region and single-cell levels using Fiji-ImageJ v.2.1.0/1.53c software [74]. For single-cell analysis, the

PLOS PATHOGENS
Rhinovirus C replication in human airway epithelium selection of the region of interest (ROI) was based on positivity for dsRNA(+) in RV-infected HAE. At least 5 cells per region were selected to increase rigor. Quantification of fluorescence levels was done in ROI single-cell images following the Corrected Total Cell Fluorescence (CTCF) method [75]. Total CTCF (equal to the sum of CTCF per slice/sample) was calculated per channel and statistically significant differences determined by Mann-Whitney U (Two-tailed; p<0.05 significance; 0.95% confidence interval) using Prism GraphPad v.9 software (GraphPad Software, San Diego, CA). Individual CTCF values were used in correlation analysis by the Spearman method (Two-tailed; p<0.05 significance; 0.95% confidence interval) using Prism GraphPad v.9 (GraphPad Software).
To evaluate the ratio of colocalization between two markers, the background was subtracted from ROI single-cell images following application of a threshold model and watershed filter to better identify the centroids. The final images were first used in colocalization analysis by pixel intensity-based methods (Pearson, threshold Manders, and Van Steensel's methods), followed by spatial analysis based on the distance between centers of mass [76]. All colocalization analysis was done using the Just Another Co-localization Plug-in (JACoP) plugin from Fiji [74].

Transmission Electron Microscopy (TEM) in HAE
The TEM protocol used in this study was based on published methods [29] with modifications, as follows: Cultures were fixed with 2% (w/v) glutaraldehyde in 0.1M cacodylate buffer (100ml on top and 500ml on the bottom) for 60 minutes at RT. The transwell membranes were then separated from the plastic holder and transferred to a new 24-well plate where a second fixation step was carried out with 1% osmium tetroxide (OsO 4 ) in 0.1M cacodylate buffer plus 1% potassium ferricyanide (K 3 Fe(CN) 6 ). Two percent uranyl acetate (diluted in distilled H 2 O) was used as post-fixative. The membrane was then incubated in propylene oxide before being embedded in Spurr's Resin. The samples were sectioned to 60-90nm with a diamond knife (DiATOME) and ultra microtome (Reichart-Jung) and two slices were placed per copper grid (EMS). The images were obtained using a Hitachi S-4700 Field Emission Scanning Electron Microscope with transmitted electron detector in the Laboratory for Biological Ultrastructure, University of Maryland, College Park.

Immunoblotting assays
Total protein in cell lysates (stored at -20˚C in RIPA buffer) were quantified by BCA Protein Assay (#23225; Pierce). A total of 20μg per sample (or 25μg and 30μg for STING and ZO-1 blots, respectively) was then separated on a NovexWedgeWell 4-20% Tris-Glycine gel (#XP04202BOX; Invitrogen-Thermo) and transferred to a PVDF membrane. Membranes were blocked with 5% non-fat milk solution (diluted in 0.1% Tween-20 in Tris-buffered saline (TBS-T)) or 3% BSA (diluted in TBS-T) for the detection of the tight junction proteins and

PLOS PATHOGENS
Rhinovirus C replication in human airway epithelium incubated overnight at 4˚C with the primary antibody ( Table 3). Following a series of washes in TBS-T, membranes were incubated for 1hr at RT with the appropriate peroxidase-conjugated secondary antibody. All antibodies were diluted in 5% non-fat milk TBS-T solution.

Lactate dehydrogenase (LDH) release cytotoxicity assay
Apical and basolateral samples were used to quantify LDH release in RV-C15-infected HAE with the CytoTox 96 Non-Radioactive Cytotoxicity Assay kit (#G1780; Promega, Madison, WI) following the manufacturer's protocol.

Transepithelial Electrical Resistance (TEER) assay
TEER was quantified in HAE using the Millicell Electrical Resistance System (ERS)-2 (Sigma) after adding 100μl of Pneumacult-ALI medium to the apical surface and incubating cultures for 30 minutes at 34˚C. Statistical analysis of resulting data was done by ordinary one-way ANOVA and Tukey's multiple comparison test methods using Prism GraphPad v.9 software (GraphPad).

Quantification of ZO-1 disposition using JAnaP
Junction coverage and characterization were quantified using the Junction Analyzer Program (JAnaP) as previously described [44][45][46]. In short, the perimeter of each cell was identified via waypoints in immunofluorescent images of ZO-1. The junctions were isolated from the background using a threshold value of 5-8. Threshold identification is described in the supplement of [44] and in the JAnaP User-Guide available at https://github.com/StrokaLab/JAnaP along with the JAnaP program in its entirety. Junction characterization was performed by calculating the length of each individual junction piece that coincides with the perimeter as well as the

PLOS PATHOGENS
Rhinovirus C replication in human airway epithelium relative aspect ratio (RAR) with respect to the cell perimeter. A junction was classified as continuous if its length was greater than 15 pixels, otherwise it was deemed discontinuous and further separated into perpendicular or punctate if the RAR was greater or less than 1.2, respectively. Statistical analysis was done by one-way ANOVA and Dunnett's multiple comparison test using Prism GraphPad v.9 software (GraphPad).

Measurement of Mucociliary Clearance (MCC)
Mucociliary transport was measured based on the transport of 2μm red-fluorescent polystyrene microspheres (Sigma-Aldrich). Five microliters of microsphere suspension (1:500 dilution in PBS; Sigma-Aldrich) was added on top of the native mucus; after a 24-hour incubation at 34˚C, videos of three regions were recorded using a Zeiss Axio Observer 3 Inverted fluorescence microscope (Cell Observer HS image system, Zeiss Axiocam 503 mono camera, optimal acquisition mode, Zen 2007 software) equipped with a 10x/0.25NA Ph1 air objective. Images were collected at a frame rate of 0.5 Hz for 10 seconds on the plane of the mucus gel. Images were acquired centrally within cultures and away from the edges, where mucus tends to accumulate. The microsphere tracking data analysis was based on an image processing algorithm that was custom written in MATLAB (The MathWorks). Briefly, the analysis software computes the XY-plane trajectories of each fluorescent microsphere in each frame. Using the trajectory data, displacement of microspheres was computed, and transport rate was calculated by dividing the displacement of microsphere by total time elapsed.

S1 Fig. RV-C15 replicates in ciliated epithelial cells leading to decreased CDHR3 levels. A:
Immunofluorescence to detect dsRNA (red) in RV-C15 (10 10 RNA)-infected HAE (nuclei, blue; scale bar = 10μm). B-C: Spearman correlation analysis (two-tailed; 0.95% confidence interval) between fluorescence levels of CDHR3 and FoxJ1 (B) or acetylated α-tubulin (C) in non-infected HAE at 12hpi. Each dot represents the fluorescence levels of CDHR3 and FoxJ1 or acetylated α-tubulin quantified in the same Z-slice. D: Orthogonal XY view from noninfected and RV-C15-infected (dsRNA+, red) HAE (nuclei, blue) stained by immunofluorescence for FoxJ1 (gray) and acetylated α-tubulin (green) at 12hpi (z-stacks at 1μm of thickness; scale bar = 10μm). E and F: Fold change graph represents FoxJ1 (E), acetylated α-tubulin (E), and CDHR3 (F) protein levels normalized to the endogenous control (actin) and compared to non-infected cultures. Data shown are from two independent donors, represented by circles and triangles. Blot above is from the donor represented by circles (E) and triangles (F); VP1 protein of RV-C15 was detected in cultures from the donor "triangle" used for both assays (E, F). G-H: Quantification of fluorescence levels (CTCF) for acetylated α-tubulin (G and FoxJ1 (H) in non-infected and RV-C15-infected (dsRNA+) HAE at 12hpi. Dots represent the total CTCF per cell and the line represents the mean. Statistical analysis was done using Mann-Whitney U test (Two-tailed; 0.95% confidence interval).

PLOS PATHOGENS
Rhinovirus C replication in human airway epithelium S1 Table. Pixel intensity-based and spatial (distance between center-mass) colocalization analysis between giantin and PI4P in RV-C15-infected HAE.

PLOS PATHOGENS
Rhinovirus C replication in human airway epithelium S15 Table. Pixel intensity-based and spatial (distance between center-mass) colocalization analysis between dsRNA and giantin in RV-A2-infected HAE.