(C) PLOS One
This story was originally published by PLOS One and is unaltered.
. . . . . . . . . .



ATG9A regulates the dissociation of recycling endosomes from microtubules to form liquid influenza A virus inclusions [1]

['Sílvia Vale-Costa', 'Cell Biology Of Viral Infection Lab', 'Cbv', 'Instituto Gulbenkian De Ciência', 'Igc', 'Fundação Calouste Gulbenkian', 'Oeiras', 'Temitope Akhigbe Etibor', 'Daniela Brás', 'Ana Laura Sousa']

Date: 2023-11

It is now established that many viruses that threaten public health establish condensates via phase transitions to complete their lifecycles, and knowledge on such processes may offer new strategies for antiviral therapy. In the case of influenza A virus (IAV), liquid condensates known as viral inclusions, concentrate the 8 distinct viral ribonucleoproteins (vRNPs) that form IAV genome and are viewed as sites dedicated to the assembly of the 8-partite genomic complex. Despite not being delimited by host membranes, IAV liquid inclusions accumulate host membranes inside as a result of vRNP binding to the recycling endocytic marker Rab11a, a driver of the biogenesis of these structures. We lack molecular understanding on how Rab11a-recycling endosomes condensate specifically near the endoplasmic reticulum (ER) exit sites upon IAV infection. We show here that liquid viral inclusions interact with the ER to fuse, divide, and slide. We uncover that, contrary to previous indications, the reported reduction in recycling endocytic activity is a regulated process rather than a competition for cellular resources involving a novel role for the host factor ATG9A. In infection, ATG9A mediates the removal of Rab11a-recycling endosomes carrying vRNPs from microtubules. We observe that the recycling endocytic usage of microtubules is rescued when ATG9A is depleted, which prevents condensation of Rab11a endosomes near the ER. The failure to produce viral inclusions accumulates vRNPs in the cytosol and reduces genome assembly and the release of infectious virions. We propose that the ER supports the dynamics of liquid IAV inclusions, with ATG9A facilitating their formation. This work advances our understanding on how epidemic and pandemic influenza genomes are formed. It also reveals the plasticity of recycling pathway endosomes to undergo condensation in response to infection, disclosing new roles for ATG9A beyond its classical involvement in autophagy.

Funding: This project has received funding from the European Research Council (ERC) under the European Union's Horizon 2020 research and innovation programme (grant agreement No. 101001521 to MJA) and by National Funds from the Portuguese Fundação para a Ciência e a Tecnologia (2022.02716.PTDC_EXPL WP1 to SVC). This work was also supported by the Instituto Gulbenkian de Ciência Advanced Imaging Facility, which is funded by PPBI-POCI-01-0145-FEDER-022122 (Lisboa 2020/FEDER/FCT; Portugal to GB), and by the Electron Microscopy Facility and Flow Cytometry Facility, which are funded by Fundação Calouste Gulbenkian (Portugal, Lisbon) to the facilities. Salary support from FCT: TAE, DB, VM are funded by PhD fellowships (PD/BD/128436/2017, PD/BD/148391/2019 and UI/BD/152254/2021, respectively) and SVC by D.L. 57. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

In this study, we sought to better define the interplay between the ER and IAV liquid inclusions. We observed that the ER supports viral inclusion fusion, fission, and sliding movements as reported for other RNP condensates [ 21 ]. From an siRNA screen of host factors involved in early steps of autophagy, we identified ATG9A (autophagy related gene 9A) as a host factor that impacted IAV liquid inclusion biogenesis and viral replication cycle. We found that ATG9A regulated trafficking of liquid viral inclusions between the ER and microtubules, removing recycling endosomes from microtubules and leading to their condensation close to ERES. ATG9A was initially identified as a core member of the autophagic machinery, mechanistically flipping phospholipids between the 2 membrane leaflets of the autophagosomal membrane to promote its growth [ 26 ]. However, we find that key initial players in autophagy (ULK1/2, TBC1D14, or ATG2A) did not regulate ER–microtubule trafficking, even though ULK2 affected virion production, suggesting that this function of ATG9A is novel. In fact, ATG9A was reported to display other roles unrelated to autophagy, including plasma membrane repair [ 27 ], lipid mobilization between organelles [ 28 ], and regulation of innate immunity [ 29 ]. Here, we show that ATG9A modulates liquid–liquid phase separation on or near the ER in mammalian cells. Interestingly, it was reported that ATG9A was able to modulate FIP200 phase separation adjacent to the ER during autophagy [ 22 ]. In this paper, we further contribute to understanding this mechanism by establishing a link between ATG9A and microtubules that has never been reported. It also contributes to how biomolecular condensates form by reprogramming preexisting pathways. The formation of numerous liquid condensates in the cell is initiated in response to specific stimuli. Therefore, our study has broader implications for biological systems by demonstrating the flexibility of unforeseen cellular machinery to change its function giving rise to biomolecular condensates.

Accumulating evidence shows that membrane-bound organelles and liquid biomolecular condensates may intimately interact in physiological contexts (reviewed in [ 19 , 20 ]). In line with this, the ER has occupied a central role [ 21 – 23 ]. The ER has critical and numerous roles in the cell, from protein and lipid synthesis, to carbohydrate metabolism, and calcium storage and signaling [ 24 ]. It has an expansive membrane able to easily rearrange and to connect with other intracellular organelles in response to specific stimuli [ 24 ]. Interestingly, the ER was shown to act as a platform for the phase separation of Tiger and Whi3 ribonucleoproteins, TIS (TPA-induced sequence) granules, Sec bodies, and autophagosome nucleation sites (reviewed in [ 19 , 20 ]), and it was shown to regulate the fission of liquid ribonucleoprotein granules to maintain their size [ 23 ]. Further examples on the interplay between membrane-bound organelles and biomolecular condensates include the demonstration that phase separated synaptic vesicles form as a mechanism for ready deployment for neurotransmission release [ 25 ].

The only confirmed cellular driver of viral inclusion formation is Rab11a, which has a role well established in recycling slow cargo to the plasma membrane in uninfected cells, binding directly to Rab11 family interacting proteins (FIPs) that, in turn, recruit molecular motors [ 12 ]. During IAV infection, the initial view that Rab11a transported vRNPs to the plasma membrane [ 7 – 10 , 13 – 15 ] was challenged by reports demonstrating that Rab11a-mediated recycling was hindered [ 16 , 17 ] resulting in the formation of liquid viral inclusions [ 4 , 6 ]. Binding of vRNPs and FIPs at the same domains in Rab11a [ 6 , 18 ] suggested that vRNPs outcompeted FIPs and molecular motors impairing cytoskeletal-based transport [ 6 , 9 ]. Interestingly, recent research indicates FIP binding to Rab11a remains unaffected during infection, hinting at a regulated process involving dynein in viral inclusion formation [ 16 ]. This, coupled with Rab11a’s association with modified endoplasmic reticulum (ER) and proximity of IAV liquid inclusions to ER exit sites (ERES) [ 4 ], strongly suggests a connection between recycling endosomes, the ER, and IAV genome assembly. However, which cellular factors regulate the biogenesis and dynamics of viral inclusions near the ER are yet to be defined.

We have recently proposed an appealing model to explain IAV genome assembly [ 4 ], which involves the formation of biomolecular condensates designated viral inclusions. We found that IAV viral inclusions share properties with bona fide liquid condensates formed by liquid–liquid phase separation–based processes [ 4 , 5 ]. They are not delimited by a membrane, are highly dynamic, react to stimuli, and internally rearrange [ 4 , 5 ]. Interestingly, despite not being delimited by membranes, IAV inclusions result from the accumulation of Rab11a recycling endosomes interacting with the different vRNP types, which are embedded as part of condensates [ 4 , 6 ]. In our model, the liquid-like character results from a network of weakly interacting vRNPs that bridge multiple cognate vRNP-Rab11a units on flexible membranes resulting presumably in percolation and condensation [ 5 ], which is currently being validated in our lab using in vitro reconstitution systems. More than just a confined space wherein IAV genome assembly may be efficiently orchestrated, viral inclusions with liquid properties constitute a change in paradigm that offer new hypotheses to test how IAV genomic complexes form. In fact, the flexibility of movement within liquid structures combined with critical recent advances in understanding the rules governing the formation of cellular biomolecular condensates [ 5 ] raises the possibility that complete genomes may have different affinities for condensates. Viral inclusions with liquid properties are important for IAV replication. This is supported by evidence that abrogating the formation [ 4 , 7 – 11 ] or forcing viral inclusions to transition from a liquid into a hardened state efficiently blocks viral production in cellular and animal infection models [ 5 ]. It also illustrates that modulating the material state of viral inclusions could become an innovative strategy to control influenza infections.

Influenza A virus (IAV) is a major causative agent of yearly flu epidemics responsible for high mortality and morbidity, despite worldwide surveillance of circulating viruses, yearly vaccination programs, and availability of antivirals. This zoonotic virus has presented occasional host-species jumps from other animals (birds, pigs) that have led to pandemics of serious consequences (reviewed in [ 1 ]). Underlying factors contributing to the perpetuation of IAV circulation in humans (and other animals) combine viral mutation rate and genomic mixing between different IAV strains. Genomic mixing accelerates viral evolution and is feasible as the IAV genome is segmented, composed by 8 distinct RNA segments arranged into viral ribonucleoproteins (vRNPs). Despite the advantage for fast viral evolution, genomic segmentation poses an interesting challenge for genome assembly, as it is known that most IAV virions contain exactly 8 vRNPs and one of each kind (reviewed in [ 2 ]). Decades of seminal research have convincingly demonstrated that IAV genome assembly is a selective process, involving intersegment RNA–RNA interactions (reviewed in [ 3 ]). However, to date, the molecular mechanism governing the assembly of influenza genomes remains unclear.

Results

Rab11a-regulated recycling is impaired by IAV infection We have recently shown that liquid viral inclusions, condensates enriched in Rab11a endosomes and vRNPs, develop in the vicinity of the ER subdomain ERES [4]. Hence, both Rab11a and the proteins associated with vRNPs (viral RNA polymerase subunits and nucleoprotein (NP)) can be used as a proxy to visualize viral inclusions. How Rab11a endosomes accumulate near the ER to form viral inclusions and how the recycling function is consequently affected during IAV infection is unclear. We hypothesize that, upon nuclear export, progeny vRNPs bind to Rab11a endosomes, which, together, are rerouted to the ER to form viral inclusions. As a consequence, Rab11a recycling capacity is expected to be impaired during IAV infection (Fig 1A, steps 1 to 2). This hypothesis is supported by several pieces of evidence. First, given that the cytoplasmic content of vRNPs increases as infection progresses and that vRNPs bind Rab11a (via PB2 viral protein) [7,8,18], we have shown that vRNPs outcompete Rab11a adaptors/molecular motors for Rab11a binding [6]. Second, as a consequence of such competition, we have shown that transferrin (Tf) recycling is reduced throughout infection [6]. Third, another group has detected the presence of Rab11a and vRNPs close to membranes of a remodeled ER during IAV infection [30]. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 1. IAV infection impairs Rab11a-regulated recycling and sustains the dynamics of viral inclusions. (A) Schematic representation of Rab11a-regulated recycling in mock- and IAV-infected cells. In mock-infected cells, Rab11a endosomes are involved in recycling material from several organelles to the plasma membrane. Upon IAV infection, progeny vRNPs bind to Rab11a endosomes and start concentrating at ERES (steps 1–2) to form liquid viral inclusions by a mechanism ill-defined (step 3). The ER likely facilitates formation of viral inclusions to promote assembly of the 8-vRNP genome (step 3). How assembled genomes reach the plasma membrane is unknown (step 4). (B) Cells (GFP-Rab11a WTlow and DNlow) were infected or mock-infected with PR8 virus for 12 h at an MOI of 3. The levels of Tf-Alexa647-fluorescent conjugates were quantified inside cells and at the cell surface by flow cytometry upon 5, 10, and 15 min of incubation at 37°C. Results were plotted as the percentage (%) of recycled Tf as a function of time of incubation. Values were normalized to the mock condition at 15 min of incubation. Three pooled independent experiments are shown. Statistical analysis was done by two-way ANOVA, followed by a Sidak’s multiple comparisons test (**p < 0.01). (C) The levels of Tf were quantified inside cells and at the cell surface by flow cytometry at 15 min of incubation at 37°C. Results were plotted as the percentage (%) of recycled Tf as a function of cell type. Four pooled independent experiments are shown. Statistical analysis was done by one-way ANOVA, followed by a Tukey’s multiple comparisons test (**p < 0.01, ***p < 0.001). (D) Cells (GFP-Rab11a WTlow, green) were simultaneously transfected with a plasmid encoding mCherry tagged to the ER (magenta) and infected or mock-infected with PR8 virus for 12 h at an MOI of 10. Cells were imaged under time-lapse conditions at 12 h postinfection. Representative cells are shown on the left. The respective individual frames with single moving particles are shown in the small panels on the right. The yellow arrowheads highlight fusion/fission events of viral inclusions (green), as well as their interaction with the ER (magenta). Bar = 10 μm. Images were extracted from S1 and S2 Videos. (E) A linescan was drawn as indicated to assess Rab11a dynamics associated with the ER. The fluorescence intensity of ER tubules (magenta) and Rab11a endosomes or viral inclusions (green) at indicated times was plotted against the distance (in μm). Representative analysis was performed using images from (D). Experiments were performed twice. For each condition, at least 10 cells were analyzed. All the values of individual and pooled experiments are provided in S1 Data File. ER, endoplasmic reticulum; ERES, ER exit site; IAV, influenza A virus; MOI, multiplicity of infection; Tf, transferrin; vRNP, viral ribonucleoprotein. https://doi.org/10.1371/journal.pbio.3002290.g001 Here, we extended our previous studies [4,6] to gain mechanistic insight into the fate of Rab11a during IAV infection. Our aim was to demonstrate that IAV infection impairs Rab11a-regulated recycling and to show that Rab11a endosomes accumulate near the ER to form viral inclusions (Fig 1A, steps 1 to 2). The combined visual inspection of Rab11 distribution in mock and infected cells with transferrin (Tf) recycling assays allows determining the fate of Rab11a endosomes in infection. We used A549 lung epithelial cells expressing low levels of Rab11a wild-type (GFP-Rab11a WTlow) and dominant-negative (GFP-Rab11a DNlow) fused to green fluorescent protein (GFP) and infected or mock infected with PR8 virus for 12 h. Cells expressing GFP-Rab11a WTlow produce significantly more viruses (2.5 log) than GFP-Rab11a DNlow at 12 h after infection (S1A Fig, mean plaque-forming units (PFUs).mL−1 ± standard error of the mean (SEM): WT– 141,875 ± 65,599 versus DN– 360 ± 142). Besides that, GFP-Rab11a WTlow cells are able to form large cytosolic puncta or viral inclusions, whereas GFP-Rab11a DNlow cells are unable to mount these condensates near the ER (S1B Fig), as we have shown before [4,6]. These results indicate that both cell lines are adequate to analyze Rab11a-regulated recycling as well as Rab11a dynamics during IAV infection. To test if Rab11a-regulated recycling is altered by IAV infection, we quantified by flow cytometry the recycling capacity of both cell lines infected or mock infected with PR8 virus for 12 h (Fig 1B and 1C). Upon feeding with a Tf-Alexa647-fluorescent conjugate, a classical cargo protein shown to be recycled by Rab11a endosomes [31–33], cells were allowed to recycle Tf for 5, 10, and 15 min at 37°C. We observed that both cell types have a significantly decreased ability to recycle Tf upon infection, in comparison to the respective mock-infected cells (Fig 1B). The drop in Tf recycling (at 15 min) caused by infection in GFP-Rab11a WTlow cells is 53.5%, whereas in GFP-Rab11a DNlow cells, the reduction in recycling is more pronounced, being around 75.4% (Fig 1B). When both cell types were compared directly (at 15 min; Fig 1C), we observed a reduction in Tf recycling levels caused by infection (% mean Tf recycling ± SEM: WT Mock– 100.0 ± 0.0% versus WT PR8–47.7 ± 5.1%, DN Mock– 87.2 ± 2.3% versus DN PR8–27.1 ± 4.5%). Of note, both infected and mock-infected GFP-Rab11a DNlow cells have a small decrease in Tf recycling relative to GFP-Rab11a WTlow cells. This reduction slightly intensifies during infection (12.8% to 20.6%, respectively; Fig 1C). This points to Tf recycling being facilitated through redundant pathways (Rab4, Rab10, Rab11) [12] in uninfected cells, with a modest involvement of Rab11a. With PR8 infection, Tf recycling decreases significantly, and a modest increase on Rab11a dependence is observed. Together, our results demonstrate that all cellular recycling is impaired during IAV infection. Moreover, the observation of enlarged cytosolic Rab11a puncta (corresponding to the liquid viral inclusions) near the ER agrees with previous reports and confirms that Rab11a pathway is severely affected in infected cells [4,6,7].

ATG9A impacts viral inclusion formation without affecting the binding of vRNPs to the recycling endosome Given that the recycling endosome could also be a putative source of ATG9A [42,43] during IAV infection and that both ATG9A and Rab11a could act in concert to allow the formation of viral inclusions, we tested the effect of depleting ATG9A in cells expressing a functionally active (WT) or inactive (DN) Rab11a. Cells expressing GFP-Rab11a WTlow or GFP-Rab11a DNlow were treated with siRNA non-targeting (siNT) or targeting ATG9A (siATG9A) for 48 h and then infected or mock infected with PR8 virus for 10 h. In this case, we explored the link between Rab11a and ATG9A at 10 h after infection, as the GFP-Rab11 DNlow cells produce low levels of viral particles before this period (by plaque assay), as we have shown before [4]. We observed that the drop in viral titers caused by ATG9A depletion was identical (approximately 0.6 log) in both cell lines, indicating that the effect of ATG9A in IAV infection is independent from Rab11a (Fig 5A, mean PFU.mL−1 ± SEM: siNT Rab11a WT 908,333 ± 177,678, siATG9A Rab11a WT 195,000 ± 18,394, siNT Rab11a DN 1,612 ± 333, siATG9A Rab11a DN 320 ± 85). We also confirmed that the efficiency of ATG9A depletion was above 80% for both cell lines (Fig 5B, mean relative expression ± SEM: siNT Rab11a WT 1.000 ± 0.000; siATG9A Rab11a WT 0.1067 ± 0.027; siNT Rab11a DN 1.000 ± 0.000; siATG9A Rab11a DN 0.180 ± 0.090). As observed before [4], introducing GFP-Rab11a DNlow exogenously in cells resulted in a 2.8 log difference (Fig 5A, mean PFU.mL−1 ± SEM: siNT Rab11a WT 908,333 ± 177,678 versus siNT Rab11a DN 1,612 ± 333) in viral titers relative to the introduction of GFP-Rab11a WTlow. By immunofluorescence, we verified that Rab11a-positive structures were also tubular upon ATG9A depletion in GFP-Rab11a WTlow cells (Fig 5C). On the contrary, GFP-Rab11a DNlow cells do not establish viral inclusions (S1B Fig) as previously shown [4], regardless of the presence of ATG9A (Fig 5C). PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 5. ATG9A impacts viral inclusions independently of Rab11a–vRNP interaction. (A–C) Cells (GFP-Rab11a WTlow or GFP-Rab11a DNlow) were treated with siRNA non-targeting (siNT) or targeting ATG9A (siATG9A) for 48 h and then infected or mock infected with PR8 virus for 10 h, at an MOI of 3. (A) Viral production was determined by plaque assay and plotted as PFU per milliliter (mL) ± SEM. Data represent 6 replicates from a single experiment. Two independent experiments were performed. Statistical analysis was done by one-way ANOVA, followed by a Kruskal–Wallis test (*p < 0.05; ***p < 0.001). (B) The protein level of ATG9A, lamin B, GFP, and Rab11a before infection were quantified by western blotting. The levels of ATG9A were plotted as the relative expression to lamin B level ± SEM. Expression was normalized to siNT from mock-infected cells. The data are a pool from 3 independent experiments. Statistical analysis was done by unpaired t test between siNT vs. siATG9A conditions of each condition (Rab11a WT vs. DN mock; ***p < 0.01). (C) Localisation of Rab11a (magenta) and PDI (gray) at 10 h postinfection was determined by immunofluorescence using antibody staining. Viral inclusions/Rab11a are highlighted by white boxes. Cell periphery and nuclei (blue, Hoechst staining) are delineated by yellow and white dashed lines, respectively. Mock-infected cells can be found in S1B Fig. Bar = 10 μm. (D–F) Cells (A549) were treated with siRNA non-targeting (siNT) or targeting ATG9A (siATG9A) for 48 h and then infected or mock infected with PR8 virus for 8 h, at an MOI of 3. (D) The localisation of host Rab11a (green) and viral NP (magenta) proteins at 8 h postinfection was determined by immunofluorescence using antibody staining. Viral inclusions/vRNPs are highlighted by white boxes. Cell periphery and nuclei (blue, Hoechst staining) are delineated by yellow and white dashed lines, respectively. Bar = 10 μm. Experiments were performed twice. (E) Colocalization between Rab11a and NP in the images acquired in (D) was determined using the Colocalization Threshold analysis tool (Image J, NIH) and plotted as the Pearson R value. At least 20 cells, pooled from 2 independent experiments, were analyzed per experimental condition. Red bar represents the median of values. Statistical analysis was done by Mann–Whitney test (n.s., not significant). (F) The roundness and circularity of Rab11a structures in the images acquired in (D) were determined using the Shape Descriptor tool (Image J, NIH) and plotted against each other. The maximum value of roundness and circularity (1) corresponds to a circular structure, whereas the minimum value represents a linear structure (0). Approximately 30 cells, from 2 independent experiments, were analyzed per condition. Statistical analysis was done by Mann–Whitney test (***p < 0.001). The frequency distribution of roundness and circularity of structures marked by Rab11a is shown in S4C and S4D Fig. All the values of individual and pooled experiments are provided in S1 Data File. GFP, green fluorescent protein; IAV, influenza A virus; MOI, multiplicity of infection; NP, nucleoprotein; PFU, plaque-forming unit; SEM, standard error of the mean; vRNP, viral ribonucleoprotein. https://doi.org/10.1371/journal.pbio.3002290.g005 We hypothesized that vRNP tubulation caused by ATG9A depletion could be due to the lack of vRNP association to Rab11a endosomes. To test this, the distribution of vRNPs and Rab11a vesicles was detected by immunofluorescence using antibodies against viral NP (magenta) and the host Rab11a (green), respectively. We observed that, although ATG9A depletion induced vRNP tubulation, it did not interfere with the association between vRNPs and Rab11a endosomes (Fig 5D), as NP and Rab11a colocalize in both siNT and siATG9A-treated cells (Fig 5E, mean Pearson R value ± SEM of: siNT 0.5855 ± 0.02015 versus siATG9A 0.6015 ± 0.0287). The quantification of the circularity versus roundness of structures marked by Rab11a showed that ATG9A depletion also caused their tubulation (Fig 5F), thus matching the previous quantification made using NP (Fig 3F). The viral inclusions in siNT-treated cells had circularity values ranging from 0.26 to 0.66 with 95% confidence interval of [0.50 to 0.57], whereas Rab11a structures in siATG9A-treated cells values ranged from 0.25 to 0.57 with 95% confidence interval of [0.37 to 0.43] (Fig 5F). The viral inclusions in siNT-treated cells had roundness values ranging from 0.62 to 0.72 with 95% confidence interval of [0.65 to 0.67], whereas Rab11a structures in siATG9A-treated cells values ranged from 0.57 to 0.66 with 95% confidence interval of [0.60 to 0.61] (Fig 5F). Calculation of the frequency distribution of circularity and roundness, using Rab11a as marker, also showed that viral inclusions in control cells were skewed toward a circular shape, whereas Rab11a structures in ATG9A-depleted cells were skewed toward a linear shape (S4C and S4D Fig). We conclude that ATG9A is critical for proper establishment of IAV inclusions and that, in its absence, these fail to form. This defect, however, is unlikely to be related to the association of vRNPs to Rab11a vesicles as ATG9A depletion did not interfere with the colocalization and spatially synchronized dynamic movement of vRNPs-Rab11.

[END]
---
[1] Url: https://journals.plos.org/plosbiology/article?id=10.1371/journal.pbio.3002290

Published and (C) by PLOS One
Content appears here under this condition or license: Creative Commons - Attribution BY 4.0.

via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/

You are viewing proxied material from magical.fish. The copyright of proxied material belongs to its original authors. Any comments or complaints in relation to proxied material should be directed to the original authors of the content concerned. Please see the disclaimer for more details.