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



TBK1 and GABARAP family members suppress Coxsackievirus B infection by limiting viral production and promoting autophagic degradation of viral extracellular vesicles [1]

['Savannah Sawaged', 'The Smidt Heart Institute', 'Cedars-Sinai Medical Center', 'Los Angeles', 'California', 'United States Of America', 'Thomas Mota', 'The Center For Neural Science', 'Medicine', 'Regenerative Medicine Institute']

Date: 2022-11 

Host-pathogen dynamics are constantly at play during enteroviral infection. Coxsackievirus B (CVB) is a common juvenile enterovirus that infects multiple organs and drives inflammatory diseases including acute pancreatitis and myocarditis. Much like other enteroviruses, CVB is capable of manipulating host machinery to hijack and subvert autophagy for its benefit. We have previously reported that CVB triggers the release of infectious extracellular vesicles (EVs) which originate from autophagosomes. These EVs facilitate efficient dissemination of infectious virus. Here, we report that TBK1 (Tank-binding kinase 1) suppresses release of CVB-induced EVs. TBK1 is a multimeric kinase that directly activates autophagy adaptors for efficient cargo recruitment and induces type-1 interferons during viral-mediated STING recruitment. Positioning itself at the nexus of pathogen elimination, we hypothesized that loss of TBK1 could exacerbate CVB infection due to its specific role in autophagosome trafficking. Here we report that infection with CVB during genetic TBK1 knockdown significantly increases viral load and potentiates the bulk release of viral EVs. Similarly, suppressing TBK1 with small interfering RNA (siRNA) caused a marked increase in intracellular virus and EV release, while treatment in vivo with the TBK1-inhibitor Amlexanox exacerbated viral pancreatitis and EV spread. We further demonstrated that viral EV release is mediated by the autophagy modifier proteins GABARAPL1 and GABARAPL2 which facilitate autophagic flux. We observe that CVB infection stimulates autophagy and increases the release of GABARAPL1/2-positive EVs. We conclude that TBK1 plays additional antiviral roles by inducing autophagic flux during CVB infection independent of interferon signaling, and the loss of TBK1 better allows CVB-laden autophagosomes to circumvent lysosomal degradation, increasing the release of virus-laden EVs. This discovery sheds new light on the mechanisms involved in viral spread and EV propagation during acute enteroviral infection and highlights novel intracellular trafficking protein targets for antiviral therapy.

Coxsackievirus B (CVB) is a significant human enterovirus that can cause myocarditis, meningitis, and pancreatitis. The subversion of host immunity and mechanisms of viral dissemination are critical factors which promote pathogenesis. We had previously reported that following infection, CVB becomes engulfed by autophagosomes which evade lysosomal degradation and instead get released as infectious extracellular vesicles (EVs). In this current study, we report that in addition to its traditional role in interferon-mediated antiviral signaling, TANK-binding kinase (TBK1) is crucial in limiting viral production and EV-based viral egress through the autophagy pathway. Indeed, in the absence of TBK1, we observe (i) a disruption in autophagic flux, (ii) significant increases in intracellular viral burden and viral EV release, and (iii) elevated viral load in both in vitro and in vivo models of infection. EVs isolated from TBK1-deficient cells or mice treated with the TBK1-inhibitor Amlexanox were more infectious compared to controls. In all, the dual role TBK1 plays in suppressing viral escape in addition to mediating antiviral immunity makes it a promising therapeutic target for the treatment of CVB infection.

Funding: This research was funded in part by National Institutes of Health grant number R01DK125692 (J.S.), American Heart Association grant number 19CDA34770083 (J.S.), and the Dorothy and E. Phillip Lyon Chair in Molecular Cardiology (R.A.G.). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Jon Sin received percent effort salary from National Institutes of Health grant number R01DK125692 and American Heart Association grant number 19CDA34770083. Roberta A Gottlieb Received percent effort salary from the Dorothy and E. Phillip Lyon Chair in Molecular Cardiology.

Copyright: © 2022 Sawaged et al. This is an open access article distributed under the terms of the Creative Commons Attribution License , which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Given the importance of TBK1 in promoting autophagosome closure and fusion with the lysosome, we asked whether TBK1 plays a role in attenuating CVB-induced autophagy and viral extracellular release of EVs. It has been reported that CVB does not induce significant type I interferon responses [ 7 ], therefore we focused our studies on the effects of TBK1 on autophagic signaling. We found that genetic knockout or silencing of TBK1 increased intracellular viral content and promoted efficient release of extracellular virus. EVs isolated from TBK1-silenced cells not only contained more viral capsid protein VP1 but also more autophagosomal lipidated LC3 and GABARAP family members. Additionally, these EVs showed increased viral infection efficacy as revealed by plaque assay. Consistent with these findings, direct activation of TBK1 by Manassantin B (ManB) significantly reduced viral infection and viral dissemination via EVs [ 25 ]. We hypothesize that in addition to its canonical role in antiviral cytokine activation, TBK1 reduces viral production by increasing delivery of virus-laden autophagosomes to lysosomes for degradation.

TANK-binding kinase 1 (TBK1) is a serine-threonine kinase which belongs to the Iκκ family of kinases that play central roles in immunity and autophagy. TBK1 was initially described for its ability to stimulate NFκB [ 17 ]. Upon activation, TBK1 directly phosphorylates interferon regulatory factor 3 (IRF3) to induce its translocation to the nucleus and transcription of antiviral cytokines. Though TBK1’s major role as a protein kinase has classically been studied in the context of cellular innate immunity, more recent research highlights TBK1’s expanded role as a significant kinase in autophagy for pathogen clearance [ 18 , 19 ]. Defined as a highly conserved, catabolic process, autophagy maintains cellular homeostasis by directing damaged organelles, protein aggregates, and pathogens to the lysosome for degradation [ 20 ]. Autophagy is initiated by de novo formation of the double membrane phagophore by autophagy-related protein 5 (ATG5) and ATG12 [ 21 , 22 ]. During the elongation of the phagophore, autophagy adaptor proteins optineurin and p62/SQSTM1 are phosphorylated by TBK1 for efficient target selection and sequestration to the forming autophagosome [ 18 , 23 ]. ATG8 family members, including LC3, GABARAPL1, and GABARAPL2, are simultaneously phosphorylated by TBK1 to facilitate closure of the autophagosome and autophagosome fusion with the lysosome. Though TBK1 is not required for the initiation of autophagy, recent findings report its relevance in premature removal of ATG8 family members from autophagosomes, thereby disrupting fusion with the lysosome [ 24 ]. Through direct phosphorylation of the ATG8 family members LC3C and GABARAPL2, TBK1 prevents the ATG4 protease from prematurely cleaving the glycine-phosphatidylethanolamine bond and releasing them from the autophagosome.

Viruses require host cells for survival and replication. Coxsackievirus B (CVB) is a common enterovirus in the Picornaviridae family that exhibits multiorgan tropism [ 1 ]. CVB is known to cause autoinflammatory diseases such as myocarditis, meningitis, and pancreatitis [ 2 – 4 ]. It is also linked to multiple neurological pathologies including amyotrophic lateral sclerosis (ALS) and frontotemporal dementia (FTD) [ 5 , 6 ]. CVB is composed of four structural capsid proteins and seven non-structural proteins which are responsible for viral replication and host cell manipulation [ 7 – 9 ]. CVB activates host cell autophagy as a mode of viral spread via the release virus-laden extracellular vesicles (EVs) [ 10 – 12 ]. Our group observed that autophagic flux is blocked during later stages of infection when EVs are shed [ 10 ]. We saw that EVs released during CVB infection were enriched with LC3-II, suggesting their origin from autophagosomes. Other groups have also documented the ability for CVB to facilitate efficient infection through en bloc transmission of the virus via EVs [ 13 ]. MicroRNAs and protein aggregates are co-secreted with viral particles in EVs and may also facilitate viral infection [ 14 – 16 ]. Though EV-mediated viral spread has been described for some time now, the molecular mechanisms that contribute to EV release are poorly understood.

Results

Knockdown of TBK1 impairs autophagic flux Due to the reported role of TBK1 in autophagosome formation and autolysosomal fusion, we next sought to determine the consequences of TBK1 silencing on autophagy. Following TBK1 silencing, we infected cells at MOI 0.01 over a time course of 0, 6, and 24 h to observe the dynamics of viral infection over time. With TBK1 silencing, we observed an increase in autophagosome formation as indicated by elevated ATG5-ATG12 conjugation at 0 and 6 h p.i. [21,22]. Interestingly, LC3-II was maintained at a high level in siTBK1 cells compared to siSCRAMBLE controls indicating a potential block in autophagic flux (Fig 3A). To confirm this, we performed an autophagic flux assay to distinguish if the accumulation in LC3-II was due to reduced autophagosome-lysosome fusion. We treated siSCRAMBLE or siTBK1 cells with bafilomycin and measured the lipidation of LC3. Bafilomycin is a lysosomal proton pump inhibitor that is known to block autophagic flux [28]. As expected, in siSCRAMBLE cells, bafilomycin induced a marked elevation of LC3-II (60%) consistent with autophagosome accumulation (Fig 3B). However, in siTBK1 cells, the starting level of LC3-II was already elevated, and the addition of bafilomycin caused only a modest increase (12%), suggesting impaired flux with silencing of TBK1. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 3. Silencing TBK1 impairs autophagic flux. HeLa cells were treated with siRNA targeting TBK1 (siTBK1) or scrambled RNA (siSCRAMBLE). (A) Western blot of infected cells at 0 h, 6 h, and 24 h p.i. at MOI 0.01. Densitometric quantification of Western blots. *, p < 0.05, ***, p < 0.001, two-way ANOVA; n = 4. (B) Western blot of cells treated with vehicle or bafilomycin (100 nM) for 2 h. Densitometric quantification of Western blots. **, p < 0.01, ***, p < 0.001, two-way ANOVA; n = 3. Data are representative of 3 experiments. WL = whole lysate. https://doi.org/10.1371/journal.ppat.1010350.g003

Silencing TBK1 increases EV release Our findings thus far have shown that loss of TBK1 increases CVB infection and disrupts autophagic flux. It is unclear if these two observations occur independently of each other or work in concert to increase viral infection and consequently, viral propagation. Our group previously described the release of viral EVs displaying both autophagosomal and mitophagosomal markers [10]. These EVs appeared to derive from fragmented autophagosome-enveloped mitochondria and contained infectious virions and pro-viral microRNAs [14]. Therefore, we investigated whether EV release may be impacted during TBK1 knockdown. We first isolated EVs from cell culture supernatant clarified of dead cells and debris using ExoQuick-TC and measured viral titers in free virus versus EV virus. We confirmed that EVs isolated with Exoquick-TC expressed common EV markers including ALIX, flotillin-1, and CD63 (S3A Fig). Further, to confirm that the virus is inside EVs, we performed a co-IP on EV isolates to pulldown flotillin-1 (S3B Fig). In addition to CD63, VP1 also co-precipitated with flotillin-1 in EVs from CVB-infected cells, showing that indeed the virus was associated with EVs. We next immunoprecipitated reticulon-3, an intracellular membrane protein located on the endoplasmic reticulum, with VP1 (S3C Fig). VP1 only co-precipitated with reticulon-3 in the infected cell lysate, indicating that the viral protein actively secreted in EVs was not associated with lysed intracellular remnants. Next, we isolated EVs from TBK1-knockdown cells. Plaque assays revealed that EV virus was increased by ~200% when TBK1 was silenced compared to controls (Fig 4A). Strikingly, the ratio of virus in EVs vs free virus remaining from the supernatant was significantly increased during TBK1-silencing which supports our hypothesis that TBK1 limits viral-EV spread (Fig 4B). We further interrogated the protein content of these EVs by western blot. By comparing equivalent protein concentrations for each EV sample, we observed that EVs shed from mock or infected cells contained similar amounts of CD63 (Fig 4C). However, we found that EVs shed from siTBK1 cells were much more enriched with LC3-II compared to EVs from siSCRAMBLE cells during infection (Fig 4C). EVs isolated from siTBK1 cells also contained significantly more VP1 which is consistent with the increased viral titers (Fig 4D). As mentioned, we had reported that CVB-induced EVs could also derive from virus-containing mitophagosomes, and indeed we observe that EVs from infected siTBK1 cells also contained increased amounts of mitochondrial outer membrane protein TOM70. These results suggest that loss of TBK1 in the setting of CVB infection leads to increased release of virus-laden EVs that may be originating from mitophagosomes. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 4. Loss of TBK1 increases the release of infectious extracellular vesicles. (A) HeLa cells were infected with eGFP-CVB at MOI 0.1. Plaque assays on free virus (free, isolated from Exoquick-TC supernatant) and EV virus (EV) isolated from cells 24 h p.i. ****, p < 0.0001, two-way ANOVA; n = 3. (B) Ratio of free virus to EV virus plaque forming units shown in (A). *, p < 0.05, Student t test; n = 3. (C) Western blots of EV lysates isolated from mock-infected and CVB-infected cells at 24 h p.i. including densitometry. ***, p < 0.001, two-way ANOVA; n = 3. (D) Western blots of EV lysates from CVB-infected cells at 24 h p.i. including densitometry. **, p < 0.01, ***, p < 0.001 two-way ANOVA; n = 3. (E) ATG5 and TBK1 were silenced in HeLa cells then infected at MOI 0.01 for 24 h. Fluorescence microscopy images of cells at 24 h p.i. Phase contrast images show similar cell density. Scale bars represent 100 μm. (G) Extracellular viral titers of cell supernatants as measured by plaque assay. **, p < 0.01, Student t test; n-3. (G) Western blots on WL from mock verus infected cells at 24 h p.i. *, p < 0.05, Student t test; n = 3. Data are representative of 3 experiments. WL = whole lysate. https://doi.org/10.1371/journal.ppat.1010350.g004 To conclude that CVB is degraded via autophagy and thereby prevents CVB release through EVs, we performed a double silencing of ATG5 and TBK1. ATG5 is an indispensable factor for autophagic vesicle formation, whereby knockdown of ATG5 inhibits autophagy [22,29]. It has been shown that CVB utilizes autophagy machinery to replicate on the autophagosome membrane, therefore it is expected that knockdown of ATG5 would inhibit viral replication [10,30–33]. We silenced ATG5 and TBK1 (siATG5/TBK1) in HeLa cells then infected them at MOI 0.01 for 24 h. By fluorescence microscopy, we observed a reduction in eGFP-positive cells in siATG5/TBK1 cells (60% decrease; **, p < 0.01) (Fig 4E) and a significant reduction in extracellular viral titers (Fig 4F). Finally, Western blots confirmed a significant decrease in cellular VP1 (Fig 4G). In all, these data depict the complex relationship between CVB replication, egress, and elimination.

CVB infection is unaltered during IFNAR knockdown To understand if the observed increase in viral infection during TBK1-deficiency is a consequence of type I interferon response inhibition, we examined viral infection in interferon-alpha/beta receptor (IFNAR)-silenced cells. Because TBK1 mediates the phosphorylation and nuclear transcription of IRF3 during infection [34], we also examined activation of IRF3. We silenced IFNAR in HeLa cells and infected them at MOI 0.01 for 24 h. No differences were observed in regard to viral eGFP expression (7% decrease; not significant) (Fig 5A) and levels of both extracellular free virus and EV virus was similar between siSCRAMBLE controls and siIFNAR cells (Fig 5B). Although Western blot revealed knockdown of IFNAR and reduction in IRF3 phosphorylation at 24 h p.i., no remarkable differences in VP1 were seen (Fig 5C). Though not significant, there was a small trend towards increased infection to in siIFNAR cells. Thus, this data supports our hypothesis that TBK1 preferentially exerts its antiviral role through autophagy. These data correspond with previous reports that CVB does not produce a significant interferon response [7]. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 5. Silencing IFNAR does not alter CVB infection. HeLa cells were treated with siRNA targeting IFNAR (siIFNAR) or scrambled RNA (siSCRAMBLE) and subsequently infected with eGFP-CVB at MOI 0.01. (A) Fluorescence microscopy of infected HeLas at 24 h postinfection (p.i.). Phase contrast images show similar cell density at 24 h p.i. Scale bars represent 100 μm. (B) Extracellular viral titers of infected cells at 24 h as measured by plaque assay. *, p < 0.05, **, p < 0.01; two-way ANOVA; n = 3. (C) Western blots of infected cells at 24 h p.i. including densitometry. **, p < 0.01, ***, p < 0.001, Student t test or two-way ANOVA; n = 3. Data are representative of 2 experiments. WL = whole lysate. https://doi.org/10.1371/journal.ppat.1010350.g005

Activating TBK1 limits CVB infection and propagation via EVs We have so far demonstrated that loss of TBK1 impairs autophagic flux and promotes the release of infectious EVs. This suggests that TBK1 plays a novel role in viral elimination. To further interrogate this, we treated HeLa cells with the specific TBK1 agonist Manassantin B (ManB) to test effects on CVB infection and viral EV shedding. ManB is a neolignan isolated from Saururus chinensis that exhibits antiviral and anti-inflammatory effects via activation of the STING-TBK1 pathway [25]. After infecting with eGFP-CVB for 24 h, we saw a marked reduction in eGFP cells following ManB treatment (14% decrease; *, p < 0.05) (Fig 6A) and plaque assays revealed a substantial reduction in EV-associated viral titers but interestingly, free viral titers were not significantly different (Fig 6B). This may indicate that TBK1 upregulation inhibits CVB egress primarily by blocking viral EV release. We also observed a reduction in cellular VP1; however, we also observed a significant increase in LC3-II (Fig 6C). To understand if this increase was due to increased autophagy or a block in autophagic flux, we incubated vehicle or ManB-treated cells with bafilomycin. We found that bafilomycin increased accumulation of lipidated LC3 to a greater extent in ManB-treated cells than vehicle-treated cells, suggesting activating TBK1 with ManB increases autophagic flux (Fig 6D). Importantly, western blots of isolated EVs from ManB-treated cells showed a dramatic reduction in VP1 compared to vehicle-treated cells (Fig 6E). These findings highlight a key role for TBK1 in degradation of CVB via autophagy, thereby preventing CVB release via EVs. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 6. Activating TBK1 with Manassantin B suppresses CVB infection and secretion into EVs. HeLa cells were treated with 1 μg ManB 24 h prior to infection with eGFP-CVB for an additional 24 h. (A) Fluorescence microscopy images of cells infected at MOI 0.01 for 24 h. Phase contrast images show similar cell density at 24 h p.i. Scale bars represent 100 μm. (B) Extracellular viral titers of cells infected at MOI 0.1 for 24 h as measured by plaque assay. ***, p < 0.001, ****, p < 0.0001, two-way ANOVA; n = 3. (C) Western blots of mock-infected and CVB-infected cells at MOI 0.01 for 24 h. Densitometric quantification of Western blots. ***, p < 0.001, ****, p < 0.0001, two-way ANOVA or Student t test; n = 3. (D) Western blot of cells treated with bafilomycin (100 nM) for 2 h following treatment with ManB (1 ug) or vehicle for 24 h. Densitometric quantification of LC3-II. *, p < 0.05, two-way ANOVA; n = 4. (E) Western blot of EV lysates from CVB-infected cells including densitometric quantification of VP1. **, p < 0.01, Student t test; n = 3. Data are representative of 3 experiments. WL = whole lysate. https://doi.org/10.1371/journal.ppat.1010350.g006

GABARAPL1/2 facilitate viral degradation To further elucidate the formation of viral EVs, we examined additional autophagy-related targets of TBK1 including GABARAP family members [24]. The ATG8 proteins (including LC3, GABARAP, and their subfamily members) each have a unique role in orchestrating membrane-trafficking events. LC3 is widely used as an indication of autophagy; the lipidated form (LC3-II) is bound to the autophagosome by conjugation to membrane phosphatidylethanolamine (PE) and functions in autophagosome biogenesis and substrate selection. GABARAP and the closely related paralogs GABARAPL1 and GABARAPL2, are cytosolic proteins that participate in autophagosome initiation and interact with cargo adaptor molecules (such as p62/SQSTM1) [35]; they also participate in vesicle transport via microtubules [36]. Although GABARAP family members are less-studied than LC3, studies have demonstrated their active role in facilitating insulin secretion, GABA(A) receptor trafficking, and angiotensin II type 1 receptor trafficking to the plasma membrane [37–39]. Therefore, we chose to examine the effect of GABARAP family members on EV shedding because of their known role in autophagosome trafficking and potential role in EV release. First, we identified whether EVs isolated using the ExoQuick-TC system contained GABARAP family members. We analyzed EV lysates from siSCRAMBLE vs. siTBK1 cells. Strikingly, GABARAPL1 and GABARAPL2 were expressed at higher levels in siTBK1 EVs compared to siSCRAMBLE EVs in CVB-infected groups (S4 Fig). To definitively identify the role of GABARAP family members during viral infection, we performed individual silencing of GABARAPL1 or GABARAPL2 in HeLa cells (S5 Fig). We did not observe a difference in infection in either of the groups at 24 h p.i. This led us to perform a double-knockdown experiment using both siRNAs. Remarkably, infection was strongly reduced at 24 h p.i. as observed by eGFP expression (37% decrease; *, p < 0.05), and infectious EV release was also decreased (Fig 7A–7D). Interestingly, CVB reduced cellular levels of GABARAPL1 24 h p.i. (Fig 7C). Of note, silencing both GABARAPL1 and GABARAPL2 (GABA1/2) did not alter LC3 compared to siSCRAMBLE cells (Fig 7C) which strongly reveals the importance of GABARAP family members in maintaining autophagic flux. To confirm whether GABARAP family members are important to maintain autophagic flux, we treated GABA1/2-silenced cells with bafilomycin. Lipidation of LC3-II was significantly increased in siSCRAMBLE cells (~41%) compared to siGABA1/2 (~3%), leading us to conclude that GABARAP family members are critical in maintaining flux (Fig 7D). The observed increase in LC3-II despite limited viral replication in siGABA1/2 cells appears to be independent of productive infection as demonstrated by a slight increase in LC3-II during infection of HeLa cells with UV-inactivated or heat-inactivated CVB in both whole cell and EV lysates (S6 Fig). PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 7. Silencing GABARAPL1 and GABARAPL2 reduces CVB infection and secretion into EVs. (A) HeLa cells were treated with siRNA targeting GABARAPL1 and GABARAPL2 (siGABA1/2) and subsequently infected with eGFP-CVB for 24 h. Fluorescence microscopy of HeLa cells infected at 24 h p.i. at MOI 0.01. Phase contrast images show similar cell density at 24 h p.i. Scale bars represent 100 μm. (B) Extracellular viral titers of free virus (free) or EV virus (EV) at 24 h p.i. at MOI 0.1 as measured by plaque assay. ***, p < 0.001, ****, p < 0.0001, two-way ANOVA; n = 3. (C) Western blots of infected cells at 24 h p.i. at MOI 0.01. Densitometric quantification of Western blots. **, p < 0.01, ***, p < 0.001, ****, p < 0.0001, two-way ANOVA; n = 3. (D) siSCRAMBLE or siGABA1/2 cells were treated with vehicle or bafilomycin (100 nM) for 2 h. Cell lysates were analyzed by Western blot and densitometry quantification was performed. *, p < 0.05, two-way ANOVA; n = 3. Data are representative of 3 experiments. WL = whole lysate. https://doi.org/10.1371/journal.ppat.1010350.g007 We have shown that knocking down TBK1 enhances CVB release through EVs, and we hypothesized this is due to disruption in GABA1/2-mediated autophagic flux. Based on recent literature that shows that TBK1 directly activates GABARAPL2 through phosphorylation to control autophagosome shedding and that secretion of EVs is dependent on GABARAPL1 [24,40], we hypothesize that TBK1 activates GABARAPL1 and GABARAPL2 to coordinate viral degradation in the host cells by chauffeuring virus-laden autophagosomes to the lysosome to be degraded. Because GABARAPL1 and GABARAPL2 are also required for autophagosome closure, we further hypothesize that silencing GABA1/2 restricts CVB infection early on by limiting the completion of autophagosome structures used by CVB for replication [41–43]. For the first time, these data illuminate a novel and fundamental role for autophagy modifiers GABARAPL1 and GABARAPL2 in viral elimination.

[END]
---
[1] Url: https://journals.plos.org/plospathogens/article?id=10.1371/journal.ppat.1010350

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/