Defining basic rules for hardening influenza A virus liquid condensates

Abstract
Editor's evaluation
eLife digest
Introduction
Results
Discussion
Materials and methods
Data availability
References
Article and author information
Metrics

Abstract

In biological systems, liquid and solid-like biomolecular condensates may contain the same molecules but their behaviour, including movement, elasticity, and viscosity, is different on account of distinct physicochemical properties. As such, it is known that phase transitions affect the function of biological condensates and that material properties can be tuned by several factors including temperature, concentration, and valency. It is, however, unclear if some factors are more efficient than others at regulating their behaviour. Viral infections are good systems to address this question as they form condensates de novo as part of their replication programmes. Here, we used influenza A virus (IAV) liquid cytosolic condensates, AKA viral inclusions, to provide a proof of concept that liquid condensate hardening via changes in the valency of its components is more efficient than altering their concentration or the temperature of the cell. Liquid IAV inclusions may be hardened by targeting vRNP (viral ribonucleoprotein) interactions via the known NP (nucleoprotein) oligomerising molecule, nucleozin, both in vitro and in vivo without affecting host proteome abundance nor solubility. This study is a starting point for understanding how to pharmacologically modulate the material properties of IAV inclusions and may offer opportunities for alternative antiviral strategies.

Editor's evaluation

Etibor and collaborators have performed a series of well-thought and careful experiments to understand some of the physical and thermodynamic properties of liquid condensates produced by the infection with the influenza A virus. However, their approach and rules could be easily applied to any other cellular phenomena that involve the formation of intracellular liquid condensates. Finally, this article is setting up the basis for an in-depth theoretical analysis of the physical phenomena described here and their correlation with the biology of intracellular liquid condensates.

https://doi.org/10.7554/eLife.85182.sa0

eLife digest

Cells are organized into compartments that carry out specific functions. Envelope-like membranes enclose some of those compartments, while others remain unenclosed. The latter are called biomolecular condensates, and they can shift their physical states from a more liquid to a more solid form, which may affect how well they function. Temperature, molecular concentration and molecular interactions affect the physical state of condensates.

Understanding what causes physical shifts in biomolecular condensates could have important implications for human health. For example, many viruses, including influenza, HIV, rabies, measles and the virus that causes COVID-19, SARS-CoV-2, use biomolecular condensates to multiply in cells. Changing the physical state of biomolecular condensates to one that hampers viruses’ ability to multiply could be an innovative approach to treating viruses.

Etibor et al. show that it is possible to harden condensates produced by influenza A virus. In the experiments, the researchers manipulated the temperature, molecular concentration and strength of connections between molecules in condensates created by influenza A-infected cells. Then, they measured their effects on the condensate’s physical state. The experiments showed that using drugs that strengthen the bonds between molecules in condensates was the most effective strategy for hardening. Studies in both human cells and mice showed that using drugs to harden condensate in infected cells did not harm the cells or the animal and disabled the virus.

The experiments provide preliminary evidence that using drugs to harden biomolecular condensates may be a potential treatment strategy for influenza A. More studies are necessary to test this approach to treating influenza A or other viruses that use condensates. If they are successful, the drug could add a new tool to the antiviral treatment toolbox.

Introduction

Central to the spatiotemporal control of reactions in many viral infections is the formation of biomolecular condensates that facilitate key steps of viral lifecycles (Etibor et al., 2021). In influenza A virus (IAV) infection, this is key for assembling its segmented genome, a complex composed of eight different viral RNA segments (vRNA) (Pons, 1976). Each vRNA is encapsidated by molecules of nucleoprotein (NP) along its length, with one unit of the RNA-dependent RNA polymerase (RdRp, consisting of PB2, PB1, and PA) bound to the base-paired RNA termini, forming viral ribonucleoproteins (vRNPs) (Alenquer et al., 2019). How the eight vRNP complex self-assembles is unknown. It is known that genomic complex formation relies on RNA-RNA interactions between distinct vRNPs and is a selective process because most virions contain exactly eight different vRNPs, as reviewed elsewhere (Hutchinson et al., 2010). After export from the nucleus where vRNPs are synthesised, vRNPs reach the cytosol and induce the formation of cytosolic condensates, known as viral inclusions (Amorim et al., 2011; Avilov et al., 2012; Chou et al., 2013; Eisfeld et al., 2011; Lakdawala et al., 2014; Momose et al., 2011), which we postulated to be sites dedicated to IAV genome assembly (Alenquer et al., 2019). Interestingly, IAV cytosolic inclusions exhibit liquid properties (fuse and divide, dissolve upon shock and are dynamic) (Alenquer et al., 2019), providing the first indication that defined material properties are critical for the formation of influenza epidemic and pandemic genomes.

As the list of viruses utilising liquid biomolecular condensates rapidly increases, including reoviruses, human cytomegalovirus, HIV, rabies, measles, SARS-CoV-2 (reviewed in Etibor et al., 2021; Lopez et al., 2021), it becomes pertinent to ask whether targeting the material properties could constitute a novel antiviral approach. Recently, the Sonic hedgehog pathway antagonist cyclopamine and its analogue A3E were demonstrated to inhibit human respiratory syncytial virus (hRSV) replication by altering the material properties of viral condensates (Risso-Ballester et al., 2021). However, compounds targeting hRSV-related (Risso-Ballester et al., 2021) and cancer-associated condensates exhibited off-target effects (Klein et al., 2020). Therefore, a critical advance in condensate disease therapy, including in the case of viral infection, requires defining how to efficiently and specifically target selected biomolecular condensates. In several studies, it was demonstrated that the properties of biomolecular condensates respond to many factors in a system-dependent manner (Alberti et al., 2019; Falahati and Haji-Akbari, 2019; Hyman et al., 2014; Riback and Brangwynne, 2020; Mittag and Parker, 2018; Snead and Gladfelter, 2019; Milovanovic and De Camilli, 2017; Perdikari et al., 2020). Entropic free energy (Quiroz and Chilkoti, 2015), concentration (Riback et al., 2020), type, number and strength of interactions (Sanders et al., 2020) have all been demonstrated to affect the properties of biomolecular condensates. In vivo, it is unknown if these changes affect equally the material properties (and function) of biomolecular condensates. This knowledge is the basis to understand which pathways could be manipulated to modulate the material properties of selected condensates, ultimately targeting their behaviour and function. For example, pathways affecting local energy production, consumption, or metabolism will alter the free energy landscape of biomolecular condensates (Patel et al., 2017). Similarly, pathways that regulate the local density of condensate drivers could affect concentration (Banani et al., 2016; Riback et al., 2020). Finally, pathways regulating post-translational modifications (Rai et al., 2018), local pH (Kroschwald et al., 2018; Munder et al., 2016), or ionic strength (Yang et al., 2020), as well as strategies promoting aggregation or dissolution of condensate interactomes could affect the type, number (valency), and strength of interactions (Bracha et al., 2019a, Bracha et al., 2019b, Zhu et al., 2019).

Viral replication programmes rely on cellular pathways, and as such, their condensate biology is more complex than attainable with in vitro reconstituted systems. For this reason, studying them in their native intracellular environment will more accurately define how they respond to specific stimuli. Our study depends on introducing a disturbance and evaluating its effect on the material properties through two different approaches. The first is to map the intracellular biophysical traits of the perturbations on viral condensates measuring thermodynamic parameters, as done in Riback et al., 2020; Shimobayashi et al., 2021; Wei et al., 2020. As liquids tend to be spherical to minimise surface tension (Elbaum-Garfinkle et al., 2015; Lee et al., 2023), we assessed their morphology and topology, and as molecules move freely in liquids, we also measured the Gibbs free energy of partition (henceforth called free energy, ΔG) to define the degree of molecular stabilisation (Shin et al., 2017; Riback et al., 2020). The lower the value of ΔG, the more stabilised the system becomes. The calculation of ΔG requires assessing the ratio of concentration of material inside a condensate (C_dense), relative to the concentration dissolved in the cytosol (C_dilute), which provides the value of the partition coefficient (K), and allows the use of the formula ΔG = -RTInK and evaluate how the system progresses. In addition, the nucleation density determines how many viral condensates are formed per area of cytosol. Overall, the data will inform us if changing one parameter, e.g. the concentration, drives the system towards condensates of different sizes with the same or more stable properties (e.g. induced by phase transitions), or changes their abundance on account of additional available nucleation centres or dissolves the condensates (Riback et al., 2020; Snead et al., 2022). These are the most relevant parameters for our scientific question, which is to define how IAV liquid viral inclusions can be hardened. However, other values are retrieved from these analyses, revealing, for example, the types of interactions required to maintain the properties of the system. This can be illustrated by reports showing that with bulk concentration, C_dilute is constant in a binary mixture (Klosin et al., 2020) but increases in multi-component systems (Riback et al., 2020). This type of information has implications about the condensates formed during influenza infection. If the system is binary, the eight vRNPs could behave as a single component, on account of their similar structure, and interact with the host Ras-related protein in the brain 11 (Rab11), reported to be part of inclusions (Avilov et al., 2012; Eisfeld et al., 2011; Amorim et al., 2011). If IAV inclusions are multi-component systems, each vRNP could behave as an independent entity, on account of differences in length, RNA sequence, and valency or, as an alternative, viral inclusions could contain yet unidentified components other than Rab11 and vRNPs. The second type of approach to accurately define how IAV liquid inclusions respond to specific stimuli is to assess the effect of the perturbations in the dynamics and kinetics of viral condensates, thus inferring changes in elasticity and viscosity (Wang et al., 2022; Alberti et al., 2019). This relies on using live-cell imaging approaches to assess if two condensates can fuse, internally rearrange, or exchange material, which are all bona fide traits of liquids but not of solids (Wang et al., 2022; Alberti et al., 2019). The information collected will offer insight into how the material properties of IAV inclusions are regulated and maintained. In addition, understanding how IAV viral inclusions can be hardened can be therapeutically relevant if they rely on their liquid character for function.

The objective of this project is to determine the most effective and precise approaches for hardening IAV liquid inclusions. We find that the stabilisation of intersegment interactions is more efficient at hardening IAV inclusions than varying the temperature or the concentration of the drivers of IAV inclusions. Importantly, we show that the hardening topological phenotype is observed in the lungs of infected mice. We also report that it is possible to affect viral inclusions without imposing additional changes in host protein abundance and solubility using solubility proteome profiling (SPP) of infected cells (Sridharan et al., 2022). In sum, our data support the development of strategies targeting the material properties of cellular condensates in viral infections and provide a critical advance in how these structures may be regulated.

Results

Framework to identify perturbations that harden IAV liquid inclusions

We previously demonstrated that viral inclusions formed by IAV infection display a liquid profile in the sense that they drip, acquire a spherical shape upon fusion, and dissolve in response to hypotonic shock or brefeldin A treatment (Alenquer et al., 2019). Here, we seek to identify the best strategies to harden viral inclusions to investigate if altering their material properties may be a novel antiviral therapy. For this, we systematically probed and compared the impact of temperature, concentration, and number/strength of ligations on the material properties of liquid viral inclusions, as a proxy of entropic, molecular, and valency contributions, respectively. We selected these parameters given the deep understanding on how they regulate the interactions amongst components and the material properties of condensates (Riback et al., 2020; Sanders et al., 2020; Quiroz and Chilkoti, 2015; Figure 1A). Methodologically, we employed established protocols for imposing the selected thermodynamic perturbations. We quantified the impact of these perturbations on the number, nucleation density ( $ρ = \frac{n u m b e r o f i n c l u s i o n}{C y t o p l a s m A r e a}$ , µm^-2), size, shape, dynamics, supersaturation ( $S = I n \frac{C d i l u t e}{C s a t u r a t i o n}$ , in which C_saturation is the concentration above which molecules demix from an homogenous system), and the Gibbs free energy of partition (ΔG = -RTInK, in which $K = \frac{C d e n s e}{C d i l u t e}$ is the partition coefficient) to define how our system adapts to the perturbations. Material concentrations inside (C_dense) and outside (C_dilute) viral inclusions were measured using the analytical strategies described in Riback et al., 2020; Shimobayashi et al., 2021, and shown in Figure 1B (and validated as described in Materials and methods and in Figure 2—figure supplement 1). For this, we used the mean fluorescence intensity (MFI) of NP as proxy of vRNP concentration (Amorim et al., 2011; Vale-Costa et al., 2016), as it is well established that the majority of cytosolic NP is in the form of vRNPs (Avilov et al., 2012; Momose et al., 2011; Eisfeld et al., 2011; Amorim et al., 2011).

Figure 1

Download asset Open asset

Framework applied in this study to define the hierarchy in effectiveness in hardening influenza A virus (IAV) liquid inclusions or other condensates.

(A) To compare the contributions of entropy, concentration, and valency/strength/type of interactions, we subjected infected cells to the different perturbations: temperature, concentration of viral inclusion drivers (viral ribonucleoproteins [vRNPs] and Ras-related in brain 11a [Rab11a]) and number or strength of interactions between different vRNPs using the well-studied vRNP pharmacological modulator, nucleozin, that increases the number and strength of intersegment interactions. (B) Our aim is to determine which amongst these perturbations impact more dramatically the material properties of viral inclusions and for this we measured number, shape, size, and Gibbs free energy of partition (free energy, ΔG). Methodologically, we segmented circa 20 cells under the different conditions to measure the above-mentioned parameters and the amount of material inside (C_dense) and outside (C_dilute) viral condensates. With this, we calculated the partition coefficient K and extrapolated the ΔG. (C) When ΔG dramatically changed, we assessed how perturbations altered the material properties of IAV inclusions by comparing how fast and how much they moved (using coarsening assays, particle tracking, photobleaching to inspect internal rearrangements, and fluorescence loss after photoactivation [FLAPh]). (D) We also assessed whether the phenotype could be recapitulated *in vivo* using mice infected with influenza A virus reassortant X31. The overall goal of this framework is to determine, for IAV, how liquid inclusions may be efficiently hardened to prioritise research and development of strategies with that activity. Additionally, the framework may be applied to other systems, including other viruses, for informed decisions on how to harden condensates.

Our goal was to identify which perturbations translated into significant shifts in ΔG to further explore whether these resulted in dramatic alterations in the material properties of viral inclusions, by assessing their kinetics and dynamics (Figure 1C) and determine how they impact viral replication in vivo (Figure 1D).

Changes in temperature mildly perturb IAV inclusions

Cellular steady state is maintained at a narrow permissive physiological range, including of temperature. However, biomolecular condensates respond to fluctuations in temperature, and we took advantage of this to assess the entropic contribution of free energy and evaluate whether regulating host cell metabolism could offer future solutions to harden IAV liquid inclusions (Figure 2A). We quantitatively analysed the viral inclusions formed in cells incubated at 4°C, 37°C, and 42°C for 30 min at 8 hr post-infection (hpi) (representative images in Figure 2B). This short duration in temperature shift did not alter the levels of cytosolic vRNPs, as expected (Figure 2C). Increasing the temperature from 37°C to 42°C did not significantly change the size (Figure 2D), aspect ratio, or number of viral inclusions (Figure 2D–G), but decreased the concentration of vRNPs in condensates (C_dense), and increased the nucleation density despite not altering the concentration of vRNPs in the milieu (C_dilute) (Figure 2H–M, Supplementary file 1 (Sheet 1)). This means that increasing the temperature up to 42°C still maintains the system in a two-phase regime, but affects the nucleation capacity, increasing the number of inclusions. Of note, vRNPs become homogenously distributed in the cytosol at 43.5°C (data not shown), pointing that the entire system may undergo regulatory processes. Importantly, this increase in temperature modestly destabilised the structure, as observed by an increase in Gibbs free energy (–2167.3±2361 J/mol @ 37°C to –1477.9±228 J/mol @ 42°C, mean ± SD, Figure 2N–O, Supplementary file 1 (Sheet 1)). Conversely, decreasing the temperature until 4°C leads to an increase in the size of inclusions that is statistically significant considering 42°C to 4°C shifts only (shift in area from 0.2896±0.02 µm² at 42°C to 0.3474±0.05 at 4°C), rounds up liquid inclusions, and decreases their nucleation capacity and abundance (the latter significant only considering 42°C and 4°C, Figure 2D–H and Supplementary file 1 (Sheet 1)). A drop in temperature increases the concentration of vRNPs in inclusions (C_dense at 37°C of 3116.0±0.05 AU, mean ± SD, and at 4°C of 2144.5±0.04 AU, Supplementary file 1 (Sheet 1)), and does not significantly change the stability of IAV inclusions as determined by Gibbs free energy (–2415.6±273 J/mol @ 4°C, Figure 2N and O). Overall, the data indicate that the temperature increasing from 4°C to 42°C shifts our system to smaller inclusions that have less vRNPs. C_dilute did not change but there is an increase in nucleation density which indicates that heat disruption of weak molecular interactions leads to alterations in nucleation, fusion, and fission, as reported previously (Iserman et al., 2020). However, we did not observe significant alterations in the stabilisation of our system, supporting that alterations in the conditions tested for temperature do not affect the material properties of viral inclusions.

Figure 2 with 1 supplement see all

Download asset Open asset

Thermal changes mildly perturb the material properties of inclusions.

A549 were infected at a multiplicity of infection (MOI) of 3 with PR8 virus for 8 hr, incubated at different temperatures (4°C, 37°C, 42°C) for 30 min, fixed, and analysed by immunofluorescence using antibody staining against Rab11 and nucleoprotein (NP) as a proxy for viral ribonucleoprotein (vRNP). The biophysical parameters were extracted from immunofluorescence images (n=15–20), adapting the method published by Riback et al., 2020; Shimobayashi et al., 2021, to determine concentration C_dense as the mean fluorescence intensity of vRNPs in the segmented influenza A virus (IAV) inclusions, while concentration C_dilute was extrapolated from the cytoplasmic vRNP intensity outside the inclusions. Each dot is the average value of a measured parameter within or outside IAV inclusions per cell. Also, size and shape of inclusion were extracted from inclusions after image segmentation. Parameters that were normalised to an infection state without IAV inclusions (3 hr post-infection [hpi]) are indicated by a dashed horizontal line. Above each boxplot, same letters indicate no significant difference between them, while different letters indicate a statistical significance at α=0.05. All data are displayed in Supplementary file 1 (Sheet 1). Abbreviations: AU, arbitrary unit. (A) Representative depiction of the experimental analysis workflow. (B) Representative images of fixed A549 cells infected with PR8 virus showing alterations in viral inclusions at different temperatures. (C). Boxplot depicting the fold change in cytoplasmic to nuclear vRNP concentration; p=0.684 by one-way ANOVA followed by Tukey multiple comparisons of means. (D) Boxplot of mean viral inclusion area (µm²) per cell; p=0.00234 by Kruskal Wallis Bonferroni treatment. (E) Boxplot of aspect ratio of inclusion; p<0.001 by one-way ANOVA followed by Tukey multiple comparisons of means. (F) Scatter plot of inclusions circularity versus roundness. (G) Boxplot showing number of viral inclusions per cell; p<0.001 by one-way ANOVA, followed by Tukey multiple comparisons of means. (H) Scatter plot of nucleation density (ρ = $\frac{n u m b e r o f i n c l u s i o n}{c y t o p l a s m A r e a}$ µm^–2) versus degree of supersaturation (S=In $\frac{C d i l u t e}{C s a t}$ ), as a measure of propensity to remain dispersed in the cytoplasm. (I) Boxplot of of vRNP concentration within inclusions (C_dense (AU)) p<0.001 by one-way ANOVA, followed by Tukey multiple comparisons of means. (J) Scatter plot of C_dense (AU) versus surrounding cytoplasm (C_dilute, AU). (K) Scatter plot C_dense (AU) versus its total cytoplasmic vRNP concentration (C_cytoplasm, AU). (L) Boxplot showing C_dilute (AU); p=0.203 by one-way ANOVA followed by Tukey multiple comparisons of means. (M) Scatter plot of C_dilute (AU) versus C_cytoplasm (AU). (N) Scatter plot of fold change in free energy of partition (ΔΔG, J/mol) where ΔG = -RTInK, and K = ( $\frac{C d e n s e}{C d i l u t e}$ ), and ΔΔG = ΔG – ΔG_{3 hpi}, versus C_dilute (AU). (O) Boxplot of ΔΔG (J/mol); p<0.001 by one-way ANOVA followed by Tukey multiple comparisons of means.

Changes in concentration of viral inclusions’ drivers do not impact their liquid profile

Two factors were shown to drive the formation of IAV inclusions – vRNPs and Rab11a (Amorim et al., 2011; Eisfeld et al., 2011; Lakdawala et al., 2014; Vale-Costa et al., 2016; Alenquer et al., 2019; Veler et al., 2022). In fact, vRNP accumulation in liquid viral inclusions requires its association with Rab11a directly via the viral polymerase PB2 (Veler et al., 2022; Amorim et al., 2011), and the liquid character may be maintained by an incompletely understood network of intersegment interactions bridging several cognate vRNP-Rab11 units on flexible membranes (Vale-Costa et al., 2016). As the concentration of material is a key determinant for the physical properties of condensates (Riback et al., 2020; Weber and Brangwynne, 2015; Hernández-Vega et al., 2017), we evaluated how concentration of these two drivers impacts the behaviour of IAV inclusions.

For this, we took advantage of the fact that vRNP levels increase during infection (Kawakami et al., 2011), and analysed viral inclusions over a time course, in two conditions: with endogenous levels of Rab11a (using cells expressing GFP, as in Alenquer et al., 2019), and overexpressing Rab11a (in the form of GFP-Rab11a, as in Alenquer et al., 2019; Figure 3A–B, Figure 3—figure supplement 2). With this approach, we aimed at analysing whether the material properties of viral inclusions changed over time and whether increasing the levels of Rab11 would alter these properties. This strategy would reveal if regulating Rab11a activity could harden IAV liquid inclusions.

Figure 3 with 2 supplements see all

Download asset Open asset

Changes in concentration of viral ribonucleoproteins (vRNPs) and Ras-related in brain 11a (Rab11a) modestly alter the material properties of viral inclusions.

(A–H) A549 cells stably expressing GFP, or GFP-Rab11a-WT were infected at an MOI of 3 with PR8 virus and, at the indicated timepoints, were fixed, and analysed by immunofluorescence using an antibody against nucleoprotein (NP) (as a proxy for vRNPs). (Number of cells (n)=14–18 for GFP and 19–25 for GFP-Rab11-WT). (C–H) Each dot is the average value of measured parameters per cell. Above each boxplot, same letters indicate no significant difference between them, while different letters indicate a statistical significance at α=0.05 using one-way ANOVA, followed by Tukey multiple comparisons of means for parametric analysis, or Kruskal Wallis Bonferroni treatment for non-parametric analysis. All thermodynamic related values are displayed in Supplementary file 1 (Sheets 2 and 3). Abbreviations: AU, arbitrary unit. (A) Representative depiction of the experimental analysis workflow. (B) Immunofluorescence images of infected cells at different hours post-infection (hpi) in cells overexpressing GFP (left) or GFP-Rab11 (right) (both in green); NP (red, as a proxy of vRNPs), and nucleus (blue). Scale bar = 10 µm. (C) Boxplot depicting the fold change in the ratio of cytoplasmic to nuclear vRNPs concentration at different times of infection, with endogenous or overexpressed Rab11a; p<0.001; Kruskal Wallis Bonferroni treatment. (D) Boxplot of mean inclusion area (µm²) per cell; p<0.001 by one-way ANOVA, followed by Tukey multiple comparisons of means. (E) Scatter plot showing nucleation density (ρ, µm^–2) versus degree of supersaturation (S). (F) Boxplot of C_dense (AU); p<0.001 by Kruskal Wallis Bonferroni treatment. (G) Boxplot of C_dilute (AU); p<0.001 by Kruskal Wallis Bonferroni treatment. (H) Boxplot of ΔΔG (J/mol); p<0.001 by Kruskal Wallis Bonferroni treatment. Conditions were normalised to an infection state without IAV inclusions (3 hpi) that is indicated by the dashed black line.

In GFP expressing cells, as the progeny vRNP pool reaches the cytosol (Figure 3B and C), viral inclusions augment in size (from 0.172±0.04 to 0.289±0.06 µm², mean ± SD, Figure 3D), with similar aspect ratio (Figure 3—figure supplement 2A and B). There is a mild reduction in the number of inclusions from 8 hpi onwards, as measured by the nucleation density (ρ) (Figure 3E, Figure 3—figure supplement 2C, all topological data in Supplementary file 1 (Sheet 2)). As infection progresses, the concentration of vRNPs inside condensates increases until 8 hpi (Figure 3F and Figure 3—figure supplement 2D and E), accompanied by an increase in the diluted cytosolic phase (Figure 3G and Figure 3—figure supplement 2D and F, Supplementary file 1 (Sheet 2)), and both parameters stabilise thereafter, indicating that the critical concentration occurs around 8 hpi. This indicates that the liquid inclusions behave as a multi-component system and allow us to speculate that the differences in length, RNA sequence, and valency that each vRNP may be key for the integrity and behaviour of condensates. Importantly, Gibbs free energy (normalised to 3 hpi) is lowest at 6 hpi (–1799.0±623 J/mol) and destabilises mildly onwards (–1139.8±382, –1131.2±444, and –833.8±342 J/mol @ 8, 12, and 16 hpi, respectively) (Figure 3H, Figure 3—figure supplement 2G, H, Supplementary file 1 (Sheet 2)). These results are consistent with the increase in cytosolic vRNP leading to bigger sized inclusions that overall maintain the same concentration although becoming modestly destabilised, suggesting that the material properties are also modestly affected. When overexpressing Rab11a (right side of each graph), cytosolic vRNPs also accumulated in viral inclusions that increased with infection (Figure 3C–D, from 0.243±0.03 to 0.385±0.04 µm²), but were significantly bigger than viral inclusions in GFP expressing cells. In addition, the nucleation density was higher (Figure 3E and Figure 3—figure supplement 2C), despite having similar aspect ratio (Figure 3—figure supplement 2A and B), C_dense (Figure 3F and Figure 3—figure supplement 2D and E) and C_dilute (Figure 3G, Figure 3—figure supplement 2D and F). The lowest value of Gibbs free energy occurs at 8 hpi (–1337.7±331 J/mol) and destabilises from then onwards (–1145.3±443 and –895.3±394 J/mol @ 12 and 16 hpi, respectively, Figure 3H, Figure 3—figure supplement 2G and H, all thermodynamic data in Supplementary file 1 (Sheet 3)). This is consistent with Rab11a overexpression giving rise to bigger viral inclusions than cells expressing native levels of Rab11a that overall contained the same vRNP concentration and destabilise slightly later in the time course of infection, confirming that Rab11a is involved in nucleating and/or maturating viral inclusions. Therefore, these data indicate that altering the concentration of vRNPs and/or Rab11a affects the size but modestly impact IAV inclusions’ material properties.

The increase in type/strength of vRNP interactions dramatically stabilises IAV inclusions

Another critical regulator of condensate properties is the type and strength of interactions among its components (Alberti and Hyman, 2021). Therefore, we predict that oligomerising vRNPs to each other, or to Rab11a, will change the viscoelasticity of condensates in similar manner to iPOLYMER in intracellular hydrogels (Nakamura et al., 2018). For IAV, it was shown by many independent groups that the drug nucleozin operates as a pharmacological modulator that oligomerises (sticks together) all forms of NP (Amorim et al., 2013; Nakano et al., 2021; Kao et al., 2010). In fact, it was demonstrated that this drug has affinity for three different sites in NP (Kao et al., 2010) chemically polymerising NP either free or in vRNPs in a reversible manner (Amorim et al., 2013). Interestingly, nucleozin was described as a novel class of influenza antivirals targeting the viral protein NP, potently inhibiting IAV replication in cultured cells and in a mouse model of influenza infection (Cianci et al., 2012). However, it readily evolved escape mutant viruses carrying the single substitution tyrosine to histidine in position 289 of NP (NP-Y289H) (Kao et al., 2010). Despite its capacity to evolve resistance, our strategy is to take advantage of a well-known tool to probe the effects of increasing the number and type of intra- and inter-vRNP interactions in the material properties of IAV inclusions (Figure 4A).

Figure 4 with 2 supplements see all

Download asset Open asset

Increasing interaction number and strength stabilises influenza A virus (IAV) inclusions.

A549 cells were infected at an MOI of 3 with PR8 virus for 8 hr, then incubated with 5 µM of nucleozin (Ncz), a viral ribonucleoprotein (vRNP) pharmacological modulator, for different time periods from 5 min to 2 hr, before fixing. Cells were processed for immunofluorescence analysis, using antibodies against nucleoprotein (NP) and Ras-related in brain 11a (Rab11a) (n=20–21). Each dot is the average value of a measured parameter per cell. Conditions normalised to an infection state without IAV inclusions (3 hr post-infection [hpi]) are indicated by a dashed black horizontal line. Above each boxplot, same letters indicate no significant difference between them, while different letters indicate a statistical significance at α=0.05 using one-way ANOVA, followed by Tukey multiple comparisons of means for parametric analysis, or Kruskal Wallis Bonferroni treatment for non-parametric analysis. All the values calculated for the thermodynamics parameters have been included as Supplementary file 1 (Sheet 4). Abbreviations: AU, arbitrary unit, CM, complete media, and Ncz, nucleozin. (A) Representative depiction of the experimental and analysis workflow. (B) Representative images of infected A549 cells subjected (or not) to increasing periods of Ncz treatment. NP (green), Rab11a (red), and nucleus (blue). Scale bar = 10 µm. (C) Boxplot depicting the fold change in the ratio of cytoplasmic to nuclear vRNPs concentration before and after Ncz treatment at 8 hpi; p=6.16e-14 by Kruskal Wallis Bonferroni treatment. (D) Boxplot of mean inclusion area per cell; p<0.001 by Kruskal Wallis Bonferroni treatment. (E) Boxplot of inclusion aspect ratio; p<2e-16 by Kruskal Wallis Bonferroni treatment. (F) Scatter plot of inclusion circularity versus roundness. (G) Boxplot showing the number of inclusions per cell; p<0.001 by Kruskal Wallis Bonferroni treatment. (H) Scatter plot of nucleation density (ρ, µm^–2) versus degree of supersaturation (S). (I) Boxplot showing increasing inclusion C_dense (AU) with increasing Ncz incubation period; p<0.001 by Kruskal Wallis Bonferroni treatment. (J) Scatter plot of C_dense (AU) versus C_dilute (AU). (K) Scatter plot of C_dense (AU) and C_cytoplasm (AU). (L) Boxplot showing C_dilute (AU); p<0.001 by Kruskal Wallis Bonferroni treatment. (M) Scatter plot of C_dilute (AU) versus C_cytoplasm (AU). (N) Scatter plot of ΔΔG, J/mol versus C_dilute. (O) Boxplot of fold change in free energy of partition (ΔΔG, cal/mol); p<0.001; Kruskal Wallis Bonferroni treatment. (P) Representative images of A549 cells infected and mock infected with PR8 containing NP-WT and NP-Y289H treated and untreated with nucleozin for 1 hr (n=17–22). Scale bar = 10 µm. (Q) Boxplot of mean inclusion area per cell, per treatment p<0.001 by Kruskal Wallis Bonferroni treatment. (R) Boxplot of fold change in free energy of partition (ΔΔG, cal/mol) of viral inclusions arising in A549 cells infected and mock infected with PR8 containing NP-WT and NP-Y289H; p<0.001; Kruskal Wallis Bonferroni treatment.

With this reasoning, we evaluated the thermal stability of inclusions in the presence or absence of nucleozin in order to confirm its pharmacological modulator activity on liquid viral inclusions (Sridharan et al., 2019). It is well established that increasing temperature shifts a thermodynamic system to a homogeneous mix. In agreement, when we exposed IAV infected cells to a range of temperatures (4°C, 37°C, and 42°C), we found that higher temperatures yield smaller inclusions (Figure 2, Figure 4—figure supplement 2) during the two-phase regime, but tending towards its homogenous distribution in the cytoplasm that was achieved at 43.5°C. Interestingly, when infected cells were exposed to the same thermal conditions after nucleozin treatment, inclusions were irresponsive to thermal fluctuation, which indicates that nucleozin stabilises viral inclusions (Figure 4—figure supplement 2).

Next, we tracked how nucleozin affected IAV liquid inclusions, by imposing the infected cells to this drug for different periods ranging from from 5 min to 2 hr. We observed that nucleozin-treated inclusions form a multi-shaped meshwork unlike the rounded liquid droplets formed without nucleozin (Figure 4B) and confirmed that nucleozin does not affect Rab11a localisation in uninfected cells (Figure 4P). Nucleozin affected the concentration of vRNPs in the cytosol that decreased with the time of treatment (Figure 4C), presumably by blocking vRNP nuclear export and/or changes accessibility of antibodies to oligomerised NP. Conversely, nucleozin treatment increased the size of viral inclusions (from 0.284±0.04 without nucleozin to 1.02±0.18 µm² with 2 hr treatment, Figure 4D), which lost circularity (0.893±0.02 without nucleozin to 0.761±0.02 2 hr treatment) and roundness (0.734±0.01 without nucleozin to 0.672±0.02 with 2 hr treatment, Figure 4E–F) and decreased in number (from 310.5±133 to 38.1±34 after 2 hr treatment, Figure 4G and H), suggesting that they were stiffer. Interestingly, C_dense increased dramatically (from 2125.8±0.09 without nucleozin to 3650.0±0.03 with 2 hr nucleozin), Figure 4I–K) and C_dilute decreased and became stable after 20 min treatment (from 728.1±213 without nucleozin to 398.6±94 after 2 hr treatment, Supplementary file 1 (Sheet 4, total C_dilute), Figure 4J and L–M), suggesting a decrease in C_sat (saturation concentration), that is compatible with the formation of a stable complex rather than a liquid condensate. Importantly, these structures were energetically more stable, with lower free energy (from –1711.1±397 J/mol without nucleozin to –5388.4±808 J/mol 2 hr post-nucleozin addition (Figure 4N–O, all topological and thermodynamic values in Supplementary file 1 (Sheet 4)).

Using the knowledge that the viral protein NP is critical to vRNP-nucleozin binding, and that the NP-Y289H renders the virus resistant, we sought to validate our findings with the mutant after adding nucleozin for 1 hr at 8 hpi. Immunofluorescence microscopy confirmed that nucleozin did not affect Rab11a in uninfected cells (mock), aggregated NP, and Rab11a in cells infected with the WT PR8 virus and that the cytoplasmic vRNP aggregation did not occur in cells infected with the NP-Y289H virus mutant (Figure 4P). Furthermore, these observations were validated by calculating the area of viral inclusions (from 0.16 to 0.18 µm² in all mock conditions, to 0.26–0.28 µm² upon infection both with WT and NP-Y289H mutant virus). With nucleozin, the size of viral inclusions with WT virus increased to 0.81±0.22 µm² and this increase was not observed with the mutant and maintained 0.27±0.04 µm² (Figure 4Q). The same was observed for nucleozin stabilisation, as measured by Gibbs free energy that was only seen with the WT PR8 virus and not with NP-Y289H, demonstrating that viral inclusion stiffness related to the viral protein NP and its effect on vRNP-vRNP interactions (Figure 4R).

Together, the data suggest that stabilising vRNP interactions changes inclusions more efficiently than the other strategies tested above.

Modifiers of strength/type of interactions among vRNPs but not concentration or temperature harden liquid IAV inclusions

Changing the strength of interactions amongst vRNPs impacted viral inclusions’ thermodynamics the most and our assumption is that this translated into modification in the material properties.

To validate our assumption, we compared the material properties of viral inclusions arising in infected cells with and without overexpression of Rab11a, with those subjected to changes in strength/type of interactions using nucleozin. Our prediction is that in the first case, we preserved the liquid character whilst in the second, inclusions were stiffer. In addition, it would be possible to accurately determine if overexpressing Rab11a impacts in material properties.

We first assessed the material properties of viral inclusions upon changes in concentration of Rab11a. Initially, we tested whether, in the two conditions, viral inclusions could fuse and divide. We observed that both fusion and fission events were taking place at late timepoints (Figure 5A, Figure 5—video 1, Figure 5—video 2, Figure 5—video 3, Figure 5—video 4). In a second approach, we measured the internal rearrangement in viral inclusions. Because of the small size and highly dynamic nature of IAV inclusions, previous attempts to perform fluorescence recovery after photobleaching experiments resulted in highly variable recovery rates (Alenquer et al., 2019; Amorim et al., 2011) that were unable to accurately determine if internal rearrangements were taking place viral inclusions. As the microtubule depolymerising drug nocodazole largely blocks the movement of IAV inclusions, rendering them larger and more spherical (Amorim et al., 2011; Avilov et al., 2012), we opted for bleaching IAV inclusions upon treating them with nocodazole (Figure 5B, Figure 5—videos 5; 6). We observed that at late stages of infection, vRNPs were able to internally rearrange in normal conditions as well as when Rab11a was overexpressed (Figure 5C). These two experiments were important to validate that concentration did not impact the material properties of viral inclusions. However, to establish if Rab11a overexpression had any impact in the dynamics of viral inclusions, we tested the inclusion molecular dynamics by fluorescence loss after photoactivation (FLAPh, Figure 5D). In a live imaging experiment, a region of interest (ROI) was photoactivated (Figure 5E), its decay profile was monitored for 120 s and the plot fitted to a single exponential model. At late points of infection, native levels of Rab11a versus overexpressed Rab11 conditions exhibited distinct vRNP decay profiles (Figure 5F), half-lives (14.48±2.2 s [mean ± SEM] and 52.15±9.6 s, respectively) and rate constants (0.06±0.04 s^-1 [mean ± SEM] and 0.02±0.003 s^-1, respectively), despite not changing the mobile and immobile fractions (Figure 5G–I, Supplementary file 1 (Sheet 5) and Figure 5—videos 7–12). This indicates that although maintaining a liquid character, in conditions of overexpressing Rab11a, inclusions become less dynamic.

Figure 5 with 12 supplements see all

Download asset Open asset

Next, we sought to assess if nucleozin altered the material properties of IAV inclusions. We first checked if nucleozin-treated viral inclusions maintained the ability to dissolve upon shock treatments, as illustrated in Figure 6A. We observed that native inclusions responded to shock treatment as expected, however, nucleozin strongly held inclusions together that did not dissolve when exposed to either hypotonic or 1,6-hexanediol shock treatments (Figure 6B and C, Supplementary file 1 (Sheet 11)). This unresponsiveness to shock suggests that IAV inclusions undergo hardening when vRNP interactions are stronger.

Figure 6 with 10 supplements see all

Download asset Open asset

Increasing the strength/type of interactions between viral ribonucleoproteins (vRNPs) changes the material properties of liquid influenza A virus (IAV) inclusions.

(**A–C**) A549 cells were infected at an MOI of 3 with PR8 virus and treated with 5 µM nucleozin (Ncz) or dimethyl sulfoxide (DMSO) at 7 hr post-infection (hpi). An hour later, cells were treated for 30 min with 80% water (hypotonic shock, Hyp), with 1,6-hexanediol (Hex) or complete media (CM) as control, before allowing recovery from stress treatment in CM for 1 hr. Cells were fixed, stained for NP for analysis by immunofluorescence, and the percentage of cells with IAV inclusions was scored manually. (D–K, P–Q) A549 cells were infected with PR8 virus at an MOI of 10 and simultaneously transfected with plasmids encoding (D–G) GFP-NP, (H–K) mcherry-NP, or (L–O) paGFP-NP and mcherry-NP. Cells were then live imaged after 12 hpi. (A) Experimental schematics of inclusion shock assay. (B) Representative images showing the response of IAV inclusions (nucleoprotein [NP], as proxy) to shock treatments after incubation in Ncz or DMSO. Scale bar = 10 µm. (C) Boxplot showing percentage cells with inclusions, after DMSO or Ncz treatment, by manual scoring; p<0.001 by Kruskal Wallis Bonferroni treatment. Analyses is a representative of three biological replicates. (Number of areas counted =15–17 and number of cells per area counted =39–56, Supplementary file 1 (Sheet 11).) (D) Scheme showing how IAV inclusions were tracked over time. (E) Plot showing inclusion (GFP-NP, as proxy) particle trajectory when treated with DMSO (n=13) or Ncz (n=27). Data analysis was pooled from two biological replicates, Supplementary file 1 (Sheet 6). (F) Graph showing the mean square displacement (µm²) versus time (s) of IAV. (G) Boxplot depicting the resulting mean square displacement (µm²) after 100 s tracking of IAV inclusions; p<0.001 by Kruskal Wallis Bonferroni treatment. (H) Schematics of the coarsening assay model, in which liquid and hardened IAV inclusions are represented by orange and blue dots, respectively. Unlike hardened inclusions, native liquid inclusions would fuse and relax to a spherical droplet. (I) Aspect ratio (AR) was used as a measure of IAV inclusion coalescence into a sphere (analysis is a pool of two biological replicates, n=10 for DMSO and 29 for Ncz). Horizontal grey dash lines depict a perfect sphere (aspect ratio =1). (J) Pseudo-coloured time lapse images of coalescing viral inclusions (GFP-NP used as proxy; extracted from Figure 6—videos 1; 2) in the presence or absence of Ncz. Scale bar = 2 µm. (K) Boxplot of the fusion time (s) of IAV liquid inclusions. Dots represent fusion time of individual fusion event (Supplementary file 1 (Sheet 7)). (L) Schematic of a fluorescence loss after photoactivation (FLAPh) experiment. (M) Time lapse pseudo-colour images showing the fluorescence loss in photoactivated IAV inclusions (photoactivatable GFP-NP used as proxy) upon treatment with Ncz (n=16) or DMSO (n=17) (extracted from Figure 6—videos 3–8). The analysis is a representative of two biological replicates. Bar = 10µm. (N) Fluorescence intensity decay of photoactivated (paGFP-NP) normalised to the corresponding IAV inclusions expressing mcherry-NP. Coloured lines are single exponential model fitting (y₀ = (1-a) + ae^-kt) of the data point, dots are the mean of the data per second, and vertical lines denote the standard deviation (SD) per time (s), (Supplementary file 1 (Sheet 8)). (O) Half-life (t_1/2 = $\frac{I n (2)}{k}, k = r a t e c o n s t a n t$ ) of liquid and hardened IAV inclusions decay post-activation (s); p=1.386e-6 by Kruskal Wallis Bonferroni treatment (Supplementary file 1 (Sheet 8)). (P) Schematic depiction of an internal rearrangement of viral inclusion after an ROI within the inclusion is photobleached. (Q) A549 cells were transfected with plasmids encoding mcherry-NP and co-infected with PR8 virus at an MOI of 10. At 12 hr post-infection (hpi), cells were treated with nocodazole (10 µg/mL) for 2 hr to reduce the highly stochastic motion of liquid IAV inclusions and subsequently treated with DMSO or Ncz. Small regions inside IAV inclusions were photobleached to assess internal rearrangement of vRNPs (mCherry-NP as proxy). Time lapse pseudo-colour images show internal rearrangements after photobleaching (extracted from Figure 6—videos 9; 10). Scale bar = 10 µm.

Lastly, we measured the internal rearrangement in viral inclusions (Figure 6P). In native conditions, the photobleached region quickly disappeared, consistent with internal rearrangement of vRNPs inside IAV inclusions, whilst in nucleozin-treated inclusions, the photobleached area remained unaltered, revealing stiffness (several examples in Figure 6Q and Figure 6—videos 9 and 10).

To formally establish that IAV liquid inclusions can be hardened, we compared the dynamics of viral inclusions in the presence or absence of nucleozin using four different approaches. First, we assessed their movement and measured speed and displacement from their point of origin (Figure 6D). Native liquid inclusions (treated with sham vehicle – dimethyl sulfoxide [DMSO]) display a highly stochastic movement and long displacement, whilst nucleozin-hardened inclusions were less mobile with smaller displacement, as observed by analysing loss of movement in individual tracks (Figure 6E). There is an overall reduction in mean square displacement (MSD) with nucleozin (Figure 6F) that results in a lower MSD at 100 s (MSD_{100 s} = 0.838 ± 1.17 µm² without nucleozin shifting to 0.057±0.22 µm² with treatment, median ± SD, Figure 6F–G and Supplementary file 1 (Sheet 6)).

In a second approach, we measured the time that two droplets take to relax to a sphere upon fusion by coarsening assays (shifting the aspect ratio from 2 to 1, Figure 6H). DMSO-treated inclusions relax fast to a single sphere upon fusion (5.8±1.94 s; mean fusion time ± SEM), shifting the aspect ratio from 2 to 1. Nucleozin-treated inclusions retain a stable aspect ratio over time (Figure 6I), as they are unable to fuse (Figure 6I–K, Supplementary file 1 (Sheet 7), Figure 6—video 1 and Figure 6—video 2). The results demonstrate that nucleozin stiffens IAV inclusions.

In a third approach, inclusion molecular dynamics was tested by FLAPh (Figure 6L). In a live imaging experiment, an ROI was photoactivated (Figure 6M), its decay profile monitored for 120 s ,and the plot fitted to a single exponential model. DMSO- and nucleozin-treated inclusions exhibited distinct decay profiles (Figure 6N), with half-life of 14.41±0.9 s (mean ± SEM) and 85.02±19.8 s, respectively (Figure 6O, Supplementary file 1 (Sheet 8) and Figure 6—videos 3–8). This indicates that nucleozin-treated inclusions become more static.

Taken together, DMSO- and nucleozin-treated IAV inclusions exhibit distinct responses to shocks, dynamics, internal rearrangement, and coalescing properties, supporting that nucleozin hardens IAV liquid inclusions.

Modifiers of strength/type of interactions between vRNPs hardens IAV liquid inclusions in vivo

Recently, the condensate-hardening drugs steroidal alkaloid cyclopamine and its chemical analogue A3 were shown to reduce viral titres in respiratory syncytial virus (RSV) infected mice (Risso-Ballester et al., 2021). However, at the organismal level, it was not demonstrated that RSV inclusion bodies in infected cells retained hardened features. To test if we could phenocopy the in vitro function of nucleozin, we aimed at analysing vRNP morphology inside the lung cells of infected mice. For this, we challenged mice with the IAV strain X31 for 2 days. At 30 min, 1 or 2 hr before the collection of the lungs, each mouse was treated with PBS (sham vehicle) or nucleozin, administered intranasally (Figure 7A–C). Interestingly, when we analysed viral inclusions under control conditions in cells of lungs of infected mice, we observed a punctate-like NP distribution. Upon nucleozin treatment, these cytosolic inclusions grew larger (inclusions per cell mean ± SEM: nucleozin 30 min, 0.101±0.006 µm²; 2 hr, 0.226±0.012 µm², Figure 7B, Supplementary file 1 (Sheet 9)). This indicates that the pharmacological induced modulator activity of nucleozin on liquid inclusions (Kao et al., 2010; Amorim et al., 2013) was retained in vivo. Having seen an effect in vRNP cytosolic localisation in vivo, we aimed at confirming a nucleozin-dependent abrogation of IAV infection in our system as reported before (Kao et al., 2010). In fact, nucleozin was reported to affect viral titres by 1 log and increase survival of IAV (A/Vietnam/1194/04 H5N1) infected mice by 50%. For this, we challenged nucleozin pre-treated mice with X31 and treated them with a daily dose of PBS (sham vehicle) or nucleozin. We found that nucleozin-treated mice had a faster recovery from viral infection (Figure 7C, Supplementary file 1 (Sheet 10)). In sum, the data serves as proof of concept that the material properties of condensates may be targeted in vivo, in agreement with Risso-Ballester et al., 2021.

Figure 7

Download asset Open asset

Hardened inclusions emerge *in vivo* when infected mice are treated with nucleozin.

(**A–B**) Mice were intranasally infected with 4000 plaque forming units (PFU) of X31 virus, and after 2 days were intranasally administered PBS or 8.3 nmoles/g mice of nucleozin (Ncz) at 30 min, 1 or 2 hr before the collection of the lungs. Data were extracted from inclusions (nucleoprotein [NP], as proxy) from fixed immunofluorescence images of lung tissues (number of cells counted = 446–1694). (A) Representative immunofluorescence images show sections of lung tissue stained for NP (red) and nucleus (blue) after PBS or Ncz treatment. (B) Boxplot showing the mean area (µm²) of inclusions from cells in lung section; p=3.378e-8 by Kruskal Wallis Bonferroni treatment (Supplementary file 1 (Sheet 9)). (C) Mice were pre-treated intraperitoneally with 8.3 nmoles/g mice Ncz or PBS for 1 hr before being intranasally infected with 1000 PFU of X31 virus, injected with a daily dose of Ncz or PBS for 11 days and the weight loss monitored daily. Body weight analysis is from a pool of two biological replicates. The number of mice is, in total, 23 for mock infected mice and 33 for infected mice (Supplementary file 1 (Sheet 10)).

Nucleozin rescues formation of hardened IAV inclusions in the absence of Rab11a

Given the possibility to harden IAV inclusions, it is important to define the molecular mechanisms conferring the liquid material properties of these condensates, which remain elusive. As Rab11a drives the formation of IAV inclusions (Amorim et al., 2011; Eisfeld et al., 2011; Lakdawala et al., 2014; Vale-Costa et al., 2016; Alenquer et al., 2019; Veler et al., 2022), we asked if nucleozin could artificially reform viral inclusions and mimic its behaviour in the absence of Rab11a. Stable cell lines expressing Rab11a dominant negative (DN) (henceforward Rab11a-DN) did not form IAV inclusions, as expected, maintaining vRNPs dispersed throughout the cytosol (Figure 8A). Interestingly, both Rab11a-WT and Rab11a-DN cell lines, in the presence of nucleozin, exhibited cytosolic puncta (despite smaller in Rab11a-DN lines, Figure 8A–B). This indicates that nucleozin bypasses the need for Rab11a to concentrate vRNPs, forming aberrant inclusions as predicted. We next tested the fusion ability of nucleozin-induced IAV inclusions in Rab11a-DN cell lines. Unlike native inclusions in WT cells, nucleozin-induced IAV inclusions in Rab11a-DN are not able to fuse in coarsening assays (Figure 8C–E). In sum, the liquid properties of IAV inclusions derived from flexible intersegment interactions and interaction with Rab11a harden to form stiff aggregates upon nucleozin treatment even when active Rab11a is absent.

Figure 8 with 6 supplements see all

Download asset Open asset

Only hardened inclusions emerge in nucleozin-treated Rab11a-DN cell line.

(**A–B**) A549 cells constitutively expressing GFP-Rab11a-WT and GFP-Rab11a-DN were infected for 10 hr with PR8 at an MOI of 3 and treated with 5 µM nucleozin (Ncz) (n=10) or dimethyl sulfoxide (DMSO) (n=10) before fixing for analysis by immunofluorescence. The data are a representative analysis from biological duplicates. (A) Representative images of cells analysed by immunofluorescence staining using antibodies against viral protein nucleoprotein (NP) (magenta), host Rab11 (green), and ER (cyan). Nuclei and cell periphery delimited by yellow and white dashed line respectively, and white boxes are insets showing presence or absence of viral inclusions. Scale bar = 10 µm. (B) Scatter plot of circularity versus roundness of viral inclusions. (**C–E**) A549 cells constitutively expressing GFP-Rab11a-DN were transfected with mcherry-NP and co-infected with PR8 virus at an MOI of 3. At 12 hr post-infection (hpi), the cells were treated with 5 µM Ncz or DMSO for 10 min before imaging. (C) Schematic depicting the possible outcomes when Rab11a-DN cell lines are treated with Ncz. (D) Representative time lapse pseudo-colour images show fusion of IAV inclusions in a coarsening assay of PR8 infected Rab11a-DN cell line treated with Ncz or DMSO (extracted from Figure 8—videos 1–6). (E) Plot depicting the aspect ratio of fusing inclusions over time in infected Rab11a-DN cell line treated with Ncz.

Nucleozin affects vRNP solubility in the absence of Rab11a without altering host proteome profile

Next, to understand how both the viral and host proteomes remodel in response to nucleozin treatment, we used a recently developed quantitative mass spectrometry-based approach called SPP (Sridharan et al., 2019). This is a lysate centrifugation assay, which can distinguish the soluble (supernatant) from insoluble (dense assemblies) protein pools. The majority of proteins annotated to be part of membraneless organelles, as well as many cytoskeletal proteins, exhibit prominent insolubility. In SPP, two aliquots of cellular lysates are extracted with either a strong (SDS) or a mild (NP40) detergent. Protein extracted with SDS represent the total proteome, while the supernatant of NP40-extracted lysate represents the soluble sub-pool. The ratio of NP40- and SDS-derived protein abundance represents the solubility of a protein (Figure 9A). Protein solubility is a proxy to track phase transition events in different cellular states. However, this measurement cannot distinguish between different events, such as solidification, phase separation, percolation, and gelation (Alberti and Hyman, 2021) that may underlie the phase transition.

Figure 9

Download asset Open asset

Hardening of influenza A virus (IAV) inclusions changes its proteome solubility.

(A–E) A549 cells constitutively expressing GFP-Rab11a-WT or GFP-Rab11a-DN were infected for 12 hr with PR8 at an MOI of 5 and treated with 5 µM nucleozin (Ncz) or dimethyl sulfoxide (DMSO) for 1 hr. Thereafter, cells were lysed in mild (NP40) or strong detergent (SDS), while NP40 lysate was ultracentrifuged (100,000 × g) to pellet materials in condensates from the soluble fraction in the supernatant. Soluble and total host and viral proteome were identified by LC-MS/MS and solubility was determined as the ratio of soluble NP40- to SDS-derived total proteome abundances at the indicated timepoints. Data were from two biological replicates. (A) Schematic representation of solubility proteome profiling (SPP). (B) Volcano plot representing relative host protein abundance in Rab11a-WT and Rab11a-DN infected cell lines (at 12 hr post-infection [hpi]) after treatment with Ncz or DMSO. Differentially upregulated proteins in these conditions (statistical significance – see Materials and methods) are indicated in blue dots. (C) Bar graphic comparing abundances of viral proteins (in log₂ scale) in Rab11a-WT and Rab11a-DN cell lines PR8-infected (12 hpi) and treated with either Ncz or DMSO. (D) Volcano plot representing relative solubility of host and viral proteins in Rab11a-WT and Rab11a-DN infected cell lines (at 12 hpi) after treatment with Ncz. Differentially soluble proteins in these conditions (statistical significance – see Materials and methods) are indicated in pink and green dots. (E) Bar graph comparing solubility (in log₂ scale) of viral proteins when PR8 infected (12 hpi) Rab11a-WT and Rab11a-DN cell lines were treated with either Ncz or DMSO.

To define the effect of nucleozin in viral inclusions, we compared proteome abundance and solubility profiles of Rab11a-DN cell lines, where the formation of liquid inclusion is blocked, with that of Rab11a-WT cell lines at 12 hpi, in the absence or presence of nucleozin (1 hr treatment) (Figure 9A–E, Supplementary file 2 and Supplementary file 3). Nucleozin treatment did not induce significant alteration in host proteome abundance in both cell lines (Figure 9B). Similarly, no major changes in terms of protein solubility were observed for the host proteome during this treatment period (Figure 9B). Overall, our results suggest that nucleozin does not induce changes in cellular protein levels or their solubility.

In terms of the viral proteome, the abundance of all protein components of vRNPs (NP, PB1, PB2, PA) and M1 show a modest increase in Rab11a-DN cell lines (Figure 9C). On the solubility level, NP exhibited a prominent change. NP remains more soluble in Rab11a-DN lines compared to Rab11a-WT infected cells (fold change of 0.188, p=7.97e-6, Figure 9C). This corroborates the observation that vRNPs remain uniformly distributed in Rab11a-DN cells. Upon nucleozin treatment, SPP data reveal that the solubility of NP remains unaltered in Rab11a-WT cells, while increasing the proportion of NP in insoluble pool in Rab11a-DN cells (Figure 9D–E, red square). Although there were no changes in solubility by SPP, we observed IAV inclusions growing larger and hardening upon nucleozin treatment at the microscopic level in Rab11a-WT cells (Figure 8A). This can be explained, as vRNPs are already insoluble in viral inclusions before nucleozin treatment and the net increase in size of the inclusions does not result in higher insolubility of vRNPs. Both SPP and microscopy complement each other in the case of Rab11a-DN cells, as viral inclusions change from soluble to insoluble and become bigger upon nucleozin treatment. Overall, these data substantiate our finding that vRNPs form Rab11a-dependent insoluble and liquid inclusions that undergo a distinctive (aberrant) phase transition upon nucleozin treatment.

PR8-infected Rab11a-WT and Rab11a-DN cells were treated with either DMSO (vehicle) or 5 µM of nucleozin for 1 hr. The protein abundance changes upon nucleozin (or DMSO) treatment in RAB11a-WT and Rab11a-DN cells is listed in this table.

Discussion

In thermodynamics, the demixing from the surrounding media implies a preference of alike molecules to interact and self-sort, excluding the milieu. This is well understood for binary systems but deviate considerably for multi-component systems, even in vitro (Klosin et al., 2020; Riback et al., 2020; Snead et al., 2022). How living cells, that are complex multi-component systems at non-equilibrium, operate lacks understanding. Small alterations in the interactions, caused by changes in the environment or the interactome of the condensate, originate different self-assembled structures (Riback et al., 2020) that respond distinctly to thermodynamic variables such as concentration, temperature, and type/strength of interactions. For example, increasing the concentration in a system is mostly associated with more ordered, less flexible structures, however higher ordered structures were reported to arise in response to a concentration reduction (Helmich et al., 2010). Therefore, understanding how physical modulators of phase transitions impact the properties of condensates is key to comprehend how biological systems may be regulated (Hermans et al., 2009). IAV infection forms cytosolic liquid inclusions that are sites for genome assembly. Our study to address the fundamental question of whether the material properties of IAV inclusions may be modulated shows that IAV inclusions may be hardened by targeting vRNP interactions but not by lowering the temperature down to 4°C nor by altering the concentration of the factors that drive their formation. The data on temperature reveals that a decrease in the entropic contribution leads to a growth of condensates, as observed for other systems (Falahati and Haji-Akbari, 2019; Hyman et al., 2014; Riback et al., 2020), that is, however, mild and does not significantly impact the stability of the structures (Figure 10). Similarly, altering the concentration of drivers of IAV inclusions impact their size but not their material properties (Figure 10). This is unexpected because many studies have shown that changing the temperature or concentration of condensate drivers dramatically impacts their phase diagrams (Bracha et al., 2018; Riback et al., 2020; Zhu et al., 2019) and material properties (Shin et al., 2017). For influenza, these minor effects demonstrate that system is flexible, which may result from the necessity to maintain the liquid character over a wide range of vRNP concentration in the cytosol (low levels in the beginning and high at late stages of infection). The maintenance of the liquid character may be a regulated process involving fission and fusion events associated with the ER, as reported for other systems (Lee et al., 2020). In fact, IAV liquid inclusions develop in proximity to a particular part of a modified endoplasmic reticulum (ER) (de Castro Martin et al., 2017), the ER exit sites (Alenquer et al., 2019). In addition, the fusion and fission events of inclusions may be necessary to promote vRNP interactions, which is essential for genome assembly, as proposed before (Eisfeld et al., 2015; Lakdawala et al., 2014).

Figure 10

Download asset Open asset

Working model for the strategies to modulate influenza A virus (IAV) liquid inclusions.

Defining the rules for hardening the condensates is important for understanding how biological condensates may be manipulated in cells and has consequences for development of novel antiviral treatments. By demonstrating that targeting the type/strength of interactions modulates the material properties of liquid viral inclusions in in vitro and in vivo models, we show that the development of molecules that affect the interactions between two components (such as post-translational modifications, local pH, or ionic strength or pharmaceutical modulators) should be prioritised over those increasing their concentration or local entropy (Figure 9). Such targeting may prevent off-target effects, especially by developing compounds able to distinguish free vRNP components from those in the supramolecular complex. In fact, the SPP herein reported demonstrates that it is possible to harden a liquid condensate without imposing changes in the host proteome abundance and solubility, which is important to increase specificity. However, a cost of targeting conserved molecules is the evolution of escape mutants (Kao et al., 2010; Cheng et al., 2012; Hu et al., 2017). Therefore, a concern to address in the future is how to design suitable combinatory therapies able to reduce their emergence. Since single nucleotide mutations underpin numerous resistance mechanisms to antivirals (Lampejo, 2020), an alternative is to engineer condensate hardening drugs that require multiple amino acid changes for escaping.

In this work, we explored the rules for hardening IAV liquid condensates. Other alternatives to modulate the material properties tailored for function can be developed. For example, accumulating evidence shows that blocking viral inclusion formation hinders viral infection (Amorim et al., 2011; Eisfeld et al., 2011; Momose et al., 2011; Vale-Costa et al., 2016; de Castro Martin et al., 2017; Vale-Costa and Amorim, 2017; Alenquer et al., 2019; Han et al., 2021; Veler et al., 2022). Herein, we observe that increase in temperature biases the system to dissolving viral inclusions (that is complete at 43.5°C), therefore activating exothermic reactions close to IAV inclusions may lead to their dissolution. Furthermore, it has been previously demonstrated that blocking Rab11 pathway, directly or indirectly, hampers viral infection (Amorim et al., 2011; Eisfeld et al., 2011; Momose et al., 2011; Han et al., 2021). Future research could also explore this route. As Rab11a has emerged as a key factor for the replication of members of many unrelated viral families relevant for human health (Bunyaviridae, Filoviridae, Orthomyxoviridae, Paramyxoviridae, and Pneumoviridae), targeting its activity may serve as a pan-antiviral strategy (Amorim et al., 2011; Bruce et al., 2010; Nakatsu et al., 2013; Nanbo and Ohba, 2018; Cosentino et al., 2022).

Limitations of the study

Understanding condensate biology in living cells is physiological relevant but complex because the systems are heterotypic and away from equilibria. This is especially challenging for influenza A liquid inclusions that are formed by eight different vRNP complexes, which although sharing the same structure, vary in length, valency, and RNA sequence. In addition, liquid inclusions result from an incompletely understood interactome where vRNPs engage in multiple and distinct intersegment interactions bridging cognate vRNP-Rab11 units on flexible membranes (Chou et al., 2013; Gavazzi et al., 2013; Sugita et al., 2013; Shafiuddin and Boon, 2019; Haralampiev et al., 2020; Le Sage et al., 2020). At present, we lack an in vitro reconstitution system to understand the underlying mechanism governing demixing of vRNP-Rab11a-host membranes from the cytosol. This in vitro system would be useful to explore how the different segments independently modulate the material properties of inclusions, explore if condensates are sites of IAV genome assembly, determine thermodynamic values, thresholds accurately, perform rheological measurements for viscosity and elasticity, and validate our findings. The results could be compared to those obtained in cell systems to derive thermodynamic principles happening in a complex system away from equilibrium. Using cells to map how liquid inclusions respond to different perturbations provide the answer of how the system adapts in vivo, but has limitations. One of the constraints of using cells in this work relates to the range and precision of the concentrations we can vary in our system. Herein, we compared endogenous Rab11a cellular levels to a single pool of transduced cells that contained low, but still heterogeneous, levels of Rab11a as a way to avoid toxicity and/or uncharacterised effects of exceedingly high concentration of Rab11a in the cell. To minimise this limitation, we combined overexpressing Rab11a with a range of low and high levels of vRNPs (analysing the entire time course of infection) to understand if a combination of high levels of vRNPs and of Rab11a could synergistically change the material properties of IAV inclusions. Technically, we retrieved thermodynamic parameters (such as C_dense, C_dilute, shape, size) from images in which vRNPs were stained using antibodies. As mentioned above, antibodies may have some difficulty accessing the inner of the condensates, which could affect measurements of C_dense or the total concentration of vRNPs in the cytosol. This could, in turn, affect the calculation of some thermodynamic parameters, including nucleation density and Gibbs free energy. Importantly, differences in antibody access may depend on some condensate properties, which may even change during infection, inducing artefactual trends. Alternatives to using antibodies comprise viruses with fluorescently tagged vRNPs, like the reported virus with the viral polymerase PA tagged with GFP (Bhagwat et al., 2018). This virus could be the ideal approach to evaluate inclusion thermodynamics for the whole study, were it not for the fact that the virus is attenuated, exhibiting delayed infection, reduced levels of viral proteins and of accumulation of vRNPs in the cytosol, with viral inclusions forming later in infection. As such, the use of these viruses would also affect the thermodynamic analyses. Future assessment of when one approach is more suitable over the other is needed. A second technical limitation relates that we acquired data from images in z-stacks as the sum of slices at specific snapshots of infection. However, although requiring a very complex imaging analysis that we lack, in the ideal scenario, the analysis should have been done using the whole volumetry of each viral inclusion, and using live images quantified over time that is yet to be reported.

Materials and methods

Key resources table

Reagent type (species) or resource	Designation	Source or reference	Identifiers	Additional information
Cell line (Homo sapiens)	A549	ATCC	CCL-185	Human alvelolar basal cell
Cell line (Canis familiaris)	MDCK.1	ATCC	CRL-2935	Mardin-Darby canine kidney cell
Antibody	Anti-Rab11a (Rabbit polyclonal)	Proteintech	Cat# 15903-1-AP, RRID:AB_2173458	IF(1:100)
Antibody	Anti-Calnexin (Rabbit polyclonal)	Abcam	Cat#22595; RRID:AB_2069006	IF(1:1000)
Antibody	Anti-NP (Mouse monoclonal)	Abcam	Cat#20343; RRID: AB_445525	IF(1:1000)
Recombinant DNA reagent	GFP-NP (plasmid)	Amorim et al., 2011	N/A	GFP version of NP
Recombinant DNA reagent	mCherry-NP (plasmid)	Amorim et al., 2011	N/A	Cherry version of NP
Recombinant DNA reagent	paCherry-NP (plasmid)	This paper	N/A	Cherry photoactivatable version of NP
Recombinant DNA reagent	paGFP-NP (plasmid)	This paper	N/A	GFP photoactivatable version of NP
Sequence-based reagent	Gipc1_F	This paper	PCR primers	GGGAAAGGACAAAAGGAACCC
Sequence-based reagent	Gipc1_R	This paper	PCR primers	CAGGGCATTTGCACCCCATGCC
Sequence-based reagent	paGFP_L64F/T65S_Fw	This paper	PCR primers	CCCTCGTGACCACCTTCAGCTACGGCGTGCAGT
Sequence-based reagent	paGFP_T203H/A206K_Fw	This paper	PCR primers	GACAACCACTACCTGAGCCACCAGTCCAAGCTGA GCAAAGACCCCAAC
Sequence-based reagent	paGFP_V163A_Fw	This paper	PCR primers	GAAGAACGGCATCAAGGCGAACTTCAAGATCCGCC
Sequence-based reagent	paCherry_NP_NheI_Fw	This paper	PCR primers	GATCCGCTAGCGGTCGCCACCATGG
Sequence-based reagent	paCherry_NP_XhoI_Rv	This paper	PCR primers	GCGCCTCGAGGATCTGAGTCCGGACTTGTA
Chemical compound, drug	DMEM, high glucose, pyruvate, no glutamine (Gibco)	Thermo Fisher	Cat# 21969035
Chemical compound, drug	L-Glutamine	Thermo Fisher	Cat# 25030024
Chemical compound, drug	OPTIMEM-I W/GLUTAMAX-I (CE)	Thermo Fisher	Cat# 51985026
Chemical compound, drug	Leibovitz’s L-15 Medium, no phenol red (Gibco)	Life Technologies	Cat# 21083–027
Chemical compound, drug	Lipofectamine LTX Reagent with PLUS Reagent (Invitrogen)	Thermo Fisher	Cat# 15338100
Chemical compound, drug	Penicillin-Streptomycin Solution	Biowest	Cat# L0022-100
Chemical compound, drug	Dimethyl sulfoxide (DMSO)	BioLabs	Cat# B0515A
Chemical compound, drug	Formaldehyde, extra pure, solution 37–41%, AR grade (Fisher Chemical)	Acros	Cat# 10231622
Chemical compound, drug	Fetal Bovine Serum, qualified, heat inactivated, Brazil (Gibco)	Thermo Fisher	Cat# 10500064
Chemical compound, drug	Nucleozin	Target Mol	Cat# 282T7330
Chemical compound, drug	1,6-Hexanediol	Aldrich	Cat# 240117–50G
Chemical compound, drug	Triton X-100	Sigma	Cat# X100
Chemical compound, drug	Dako Faramount Aqueous Mounting Medium	Agilent Technologies	Cat# S3025
Chemical compound, drug	Complete protease inhibitor cocktail	Merck	Cat# 11836170001
Chemical compound, drug	PhosphoStop	Merck	Cat# 4906837001
Chemical compound, drug	RNasin Plus RNase Inhibitor	Promega	Cat# N2615
Chemical compound, drug	NP-40	Thermo Fisher Scientific	Cat# FNN0021
Chemical compound, drug	SDS	NZYTech	Cat# MB01501
Chemical compound, drug	Benzonase Nuclease HC	Merck	Cat# 71206-3
Chemical compound, drug	Ethanol	VWR Chemicals	Cat# 20821.330
Chemical compound, drug	Sequencing Grade Modified Trypsin	Promega	Cat# V5111
Chemical compound, drug	Lysyl Endopeptidase, Mass Spectrometry	Wako	Cat# 125-05061
Chemical compound, drug	HEPES	Alfa Aesar	Cat# A14777
Chemical compound, drug	TMT-16plex reagents	Thermo	Cat# A44522
Chemical compound, drug	PIERCE BCA protein assay	Thermo	Cat# 23225
Software, algorithm	FIJI	ImageJ	RRID:SCR_002285	https://imagej.net/software/fiji/
Software, algorithm	R Project for Statistical Computing	R	RRID:SCR_001905	https://www.r-project.org/
Software, algorithm	Trackmate plugin	Tinevez et al., 2017; Ershov et al., 2022	RRID:SCR_002285	https://imagej.net/plugins/trackmate/
Software, algorithm	limma	Ritchie et al., 2015.	RRID:SCR_010943	https://bioconductor.org/packages/release/bioc/html/limma.html
Software, algorithm	ClusterProfiler (R Bioconductor)	Yu et al., 2012.	RRID:SCR_016884	https://bioconductor.org/packages/release/bioc/html/clusterProfiler.html
Software, algorithm	isobarQuant	Quant	https://doi.org/doi:10.18129/B9.bioc.isobar	https://www.bioconductor.org/packages/release/bioc/html/isobar.html
Software, algorithm	Mascot 2.4 (Matrix Science)	Mascot	RRID:SCR_014322	http://www.matrixscience.com/server.html
Other	Hoechst stain	Thermo Fisher Scientific	H3570	(1 µg/mL)

Share this article

Cite this article

Framework applied in this study to define the hierarchy in effectiveness in hardening influenza A virus (IAV) liquid inclusions or other condensates.

Thermal changes mildly perturb the material properties of inclusions.

Changes in concentration of viral ribonucleoproteins (vRNPs) and Ras-related in brain 11a (Rab11a) modestly alter the material properties of viral inclusions.

Increasing interaction number and strength stabilises influenza A virus (IAV) inclusions.

Changes in concentration of viral ribonucleoproteins (vRNPs) and Ras-related in brain 11a (Rab11a) modestly alter the material properties of viral inclusions.

Increasing the strength/type of interactions between viral ribonucleoproteins (vRNPs) changes the material properties of liquid influenza A virus (IAV) inclusions.

Hardened inclusions emerge in vivo when infected mice are treated with nucleozin.

Only hardened inclusions emerge in nucleozin-treated Rab11a-DN cell line.

Hardening of influenza A virus (IAV) inclusions changes its proteome solubility.

Working model for the strategies to modulate influenza A virus (IAV) liquid inclusions.

Author details

Temitope Akhigbe Etibor

Contribution

Competing interests

Silvia Vale-Costa

Contribution

Competing interests

Sindhuja Sridharan

Contribution

Competing interests

Daniela Brás

Contribution

Competing interests

Isabelle Becher

Contribution

Competing interests

Victor Hugo Mello

Contribution

Competing interests

Filipe Ferreira

Contribution

Competing interests

Marta Alenquer

Contribution

Competing interests

Mikhail M Savitski

Contribution

Competing interests

Maria-João Amorim

Contribution

For correspondence

Competing interests

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

Categories and tags

Research organisms

Further reading