This story was originally published in Plos One Journal:
URL: plosone.org. The content has not been altered
Licensed under Creative Commons Attribution (CC BY) license .
url:https://journals.plos.org/plosone/s/licenses-and-copyright
(C) Plos One [1]
--------------------



SARS-CoV-2-specific circulating T follicular helper cells correlate with neutralizing antibodies and increase during early convalescence

['Sushma Boppana', 'Division Of Infectious Diseases', 'Department Of Medicine', 'University Of Alabama At Birmingham', 'Birmingham', 'Alabama', 'United States Of America', 'Kai Qin', 'Jacob K. Files', 'Ronnie M. Russell']

Date: None

Abstract T-cell immunity is likely to play a role in protection against SARS-CoV-2 by helping generate neutralizing antibodies. We longitudinally studied CD4 T-cell responses to the M, N, and S structural proteins of SARS-CoV-2 in 26 convalescent individuals. Within the first two months following symptom onset, a majority of individuals (81%) mounted at least one CD4 T-cell response, and 48% of individuals mounted detectable SARS-CoV-2-specific circulating T follicular helper cells (cTfh, defined as CXCR5+PD1+ CD4 T cells). SARS-CoV-2-specific cTfh responses across all three protein specificities correlated with antibody neutralization with the strongest correlation observed for S protein-specific responses. When examined over time, cTfh responses, particularly to the M protein, increased in convalescence, and robust cTfh responses with magnitudes greater than 5% were detected at the second convalescent visit, a median of 38 days post-symptom onset. CD4 T-cell responses declined but persisted at low magnitudes three months and six months after symptom onset. These data deepen our understanding of antigen-specific cTfh responses in SARS-CoV-2 infection, suggesting that in addition to S protein, M and N protein-specific cTfh may also assist in the development of neutralizing antibodies and that cTfh response formation may be delayed in SARS-CoV-2 infection.

Author summary Since December 2019, the Coronavirus Disease 2019 (COVID-19) pandemic has caused significant morbidity and mortality worldwide. Recently approved vaccines against SARS-CoV-2 are understood to protect against infection by inducing neutralizing antibodies. However, the underlying immune responses necessary for protection remain unclear. It is well established that T follicular helper cells (Tfh), a subset of CD4 T cells, are essential to the development of neutralizing antibodies and that some of these cells, called circulating T follicular helper cells (cTfh), can be studied in the blood. Not much is known about Tfh responses mounted in SARS-CoV-2 infection. Here, we studied cTfh responses to three major structural proteins in individuals recovered from COVID-19. We find that SARS-CoV-2-specific cTfh frequencies correlate with neutralizing antibody responses. We also find that cTfh responses to SARS-CoV-2 increase well into convalescence before contracting. Our results suggest that cTfh responses against proteins other than the spike protein may contribute to the development of neutralizing antibodies and that the formation of cTfh responses in SARS-CoV-2 infection may be delayed.

Citation: Boppana S, Qin K, Files JK, Russell RM, Stoltz R, Bibollet-Ruche F, et al. (2021) SARS-CoV-2-specific circulating T follicular helper cells correlate with neutralizing antibodies and increase during early convalescence. PLoS Pathog 17(7): e1009761. https://doi.org/10.1371/journal.ppat.1009761 Editor: Paul G. Thomas, St. Jude Children’s Research Hospital, UNITED STATES Received: June 18, 2021; Accepted: June 27, 2021; Published: July 16, 2021 Copyright: © 2021 Boppana 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. Data Availability: All relevant data are within the manuscript and its Supporting Information files. Funding: We would like to acknowledge the following funding sources: internal funding from the University of Alabama at Birmingham School and Department of Medicine and F30AI140829 (SB) and F30AI155295 (JF). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing interests: The authors have declared that no competing interests exist.

Introduction Cases of COVID-19, caused by the novel severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), were first reported in Wuhan, China at the end of 2019 [1]. Since then, the COVID-19 pandemic has caused significant morbidity, mortality, and economic disruption worldwide [2]. In SARS-CoV-2 infection, initial studies reported significant lymphopenia in hospitalized patients [3]. Data on antigen-specific T-cell responses in individuals recovered from SARS-CoV-2 infection have reported CD4 T-cell responses to SARS-CoV-2 in 80–100% of convalescent individuals, with most publications focusing on the spike (S) protein [4–7]. Other groups have shown that T cells also target the nucleocapsid (N) and membrane (M) proteins [8]. The approval of effective SARS-CoV-2 vaccines has been heralded as a critical step to curtailing the COVID-19 pandemic. Studies in non-human primates have found that neutralizing antibodies (nAb) are a correlate of protection in infection and vaccination [9,10]. Phase I/II trial data from the Pfizer and Moderna vaccines, as well other candidate vaccines, have highlighted neutralizing antibodies as the driving force of vaccine efficacy [11–14]. While the Pfizer vaccine has been shown to elicit CD4 and CD8 T-cell responses in trial participants and that vaccine-induced CD4 T-cell responses correlated with antibody titers [15], our understanding of exact immune correlates of protection and the durability of these immune responses remains incomplete. Additionally, the factors driving the formation of nAb in the context of SARS-CoV-2 natural infection are important to study, particularly in the context of emerging viral variants that could be differentially infectious [16,17]. In both mild and severe SARS-CoV-2 disease, many groups have described an elevation of both activation and exhaustion markers on T cells [18–21]. Our group has demonstrated that SARS-CoV-2-induced immune dysregulation persists into convalescence [20]. Because direct study of lymphoid tissues in humans is difficult, circulating T follicular cells (cTfh), or T follicular helper cells (Tfh) circulating in the blood, serve as an important surrogate for understanding Tfh responses within germinal centers. While there is some controversy regarding how to best identify these cells, there is general consensus that these cells express CXCR5, a lymph node homing receptor, and many groups use PD1 expression in conjunction with CXCR5 to define cTfh [22–24]. While frequencies of circulating CXCR5+PD1+ CD4 T cells are typically low, these cells are closely linked to Tfh in lymphoid tissue [25] and have been shown to support humoral responses [26,27]. Antigen-specific cTfh have been shown to correlate with neutralizing antibodies in the context of infection and vaccination of several pathogens [23,28–32]. Although cTfh responses have not been described in the context of SARS-CoV or MERS-CoV infection, CD4 T-cell responses have been shown to be important in controlling SARS-CoV in mouse models [33], and a recent study of a MERS-CoV vaccine in mice found that Tfh frequencies in draining lymph nodes correlated with neutralizing antibodies [34]. There is an increasing amount of data on SARS-CoV-2-specific T follicular helper cell responses. Thevarajan et al. were the first to report cTfh frequencies in SARS-CoV-2 and found that frequencies of total cTfh increased during acute infection [35]. Since then, a few studies have drawn a correlation between total CD4 T-cell or total Tfh-like cell frequencies and antibody levels [36,37]. One study found increased expression of CXCR5 and ICOS, two Tfh markers, on SARS-CoV-2-specific CD4 T-cells but did not examine cTfh responses directly [38]. In deceased donors with COVID-19, Kaneko et al. found that BCL6-expression in germinal center Tfh was lost within thoracic lymph nodes, suggesting that Tfh response formation may be impaired in severe SARS-CoV-2 infection [39], but the formation of antigen-specific Tfh responses, particularly in milder cases of COVID-19, remains unclear. Juno et al examined circulating Tfh, defined as CD45RA-CXCR5+ CD4 T cells, in the blood of SARS-CoV-2 infected individuals. They demonstrated a correlation between S protein-specific cTfh and nAb, suggesting that functional Tfh responses are formed in mild SARS-CoV-2 infection [40]. However, these data leave several questions unanswered, including at what point in convalescence these responses evolve. While this study was a useful first glimpse at antigen-specific Tfh responses and included PD1, a canonical Tfh marker, in their panel, PD1 expression was not used to define the Tfh population and was not reported. Additionally, the study used Ox40 and CD25 as activation markers to identify antigen-specific responses, which have previously been shown to include a high percentage of T regulatory cells [41]. Another recent study examined the persistence of S protein-specific CD4 T-cell responses in convalescence [42]. They examined frequencies of circulating Tfh (ICOS+CXCR5+ CD4 T cells) roughly one month and three months after symptom onset, and although they saw responses greater than background at three months, they did not see a difference over time. A more detailed examination of CD4 T-cell and cTfh responses in convalescence would help to establish how these responses develop over time. More recent studies have investigated cTfh populations up to 6 months after initial symptom onset [43,44]. However, these longitudinal studies did not investigate correlations between antigen-specific cTfh and SARS-CoV-2-specfic antibodies. It is also important to note that all of these studies primarily focused on S protein-specific responses. cTfh specificity does not necessarily correspond with neutralizing antibody specificity. For example, in HIV infection and vaccination, intrastructural help occurs, where CD4 T-cell responses to internal, structural proteins correlate with neutralizing antibodies against the exterior, envelope protein [45,46], underscoring the importance of examining cTfh responses across the SARS-CoV-2 proteome. Here, we report on SARS-CoV-2-specific CD4 T-cell responses to the membrane (M), nucleocapsid (N), and spike (S) proteins studied longitudinally in 26 convalescent individuals. We directly examined antigen-specific cTfh (CXCR5+PD1+ CD4 T cells) and observed correlations between antigen-specific cTfh responses across all protein specificities and antibody neutralization at the first convalescent visit. We find the M protein-specific cTfh responses increase in magnitude from Visit 1 to Visit 2. High magnitude SARS-CoV-2-specific cTfh responses (>5% activation of total cTfh population) were only detected at the second convalescent visit, more than 30 days following symptom onset, but these responses do not correlate with antibody neutralization. These data are the first to examine the kinetics of cTfh responses that arise after SARS-CoV-2 infection as well as the relationship between neutralizing antibodies and cTfh responses to the SARS-CoV-2 M and N proteins. Our findings also suggest that cTfh formation may be delayed in SARS-CoV-2 infection.

Discussion In this study, we longitudinally examined the CD4 T-cell responses targeting the major SARS-CoV-2 structural proteins, M, N, and S, in 26 convalescent individuals by measuring the expression of activation markers and the production of effector cytokines. We found that at the first convalescent visit, antigen-specific cTfh responses could be detected against all three proteins and that the frequency of antigen-specific cTfh in these individuals correlated with nAb, albeit to varying degrees. We also found that cTfh responses increase over time in convalescence, particularly M protein-specific ones, that truly robust cTfh responses (>5% frequency) were only detected at a second visit, and that these responses eventually undergo contraction late in convalescence. The relative weakness of the correlation between N protein-specific cTfh frequency and antibody neutralization compared to the M and S proteins may relate back to the structure of SARS-CoV-2. Both the spike and membrane proteins have portions that are located exteriorly, while the nucleocapsid protein is found exclusively internally. Collectively, these data suggest that cTfh responses induced against different SARS-CoV-2 proteins may not be equally effective in aiding B cells and bolsters the foundation for vaccine strategies recently approved by the FDA under emergency use authorization and those currently in testing which only include the Spike protein. In fact, many of these vaccines have reported levels of antibodies similar to those seen in natural SARS-CoV-2 infection and mild disease, which may be a result of focusing the cTfh response on the S protein [12,14]. However, as prior HIV studies have shown, CD4 T cells across different protein specificities may contribute to nAb induction [45,46]. Future studies should work to ascertain the level to which M and N protein-specific cTfh responses contribute to the formation of neutralizing antibodies. It is possible that cTfh responses across different protein specificities all play a synergistic role in the development of nAb. Interestingly, the robust cTfh responses detected at the 2nd timepoint were only found against the M and S proteins, not the N protein, and the correlation between cTfh response magnitude and antibody neutralization no longer existed at this later timepoint. Many groups have described significant T-cell dysfunction in acute SARS-CoV-2 infection [18,19,51], and our group has recently illustrated that this dysfunction is sustained during convalescence, even in nonhospitalized individuals [20]. It is possible that the robust cTfh responses detected in this subset of individuals is evidence of a dysregulated immune response and that these cTfh do not contribute to nAb formation. Future studies would ideally delve deeper by examining additional relevant cytokines, like IL4, IL13, and IL21, and combine activation marker and cytokine staining to allow for comprehensive functional analysis of the CD4 T cells arising later in convalescence. A more comprehensive assessment is also important as some SARS-CoV-2-specific CD4 T cells may not express the traditional surface markers or produce typical cytokines as a result of this immune dysfunction. Meanwhile, the observed increase in cTfh responses over time suggests that cTfh response formation may also be delayed in SARS-CoV-2 infection. A study of influenza vaccination showed that cTfh responses peaked seven days after vaccine administration [31]; meanwhile, a longitudinal study of cTfh in dengue virus infection found that the frequency of antigen-specific cTfh decreased from the time of acute infection [28]. In mouse models of influenza infection, cTfh responses peak in the first week after infection and have undergone contraction well within the first month of infection [52,53]. In comparison with these studies, it appears that cTfh response formation in SARS-CoV-2 infection continues farther into convalescence than expected—the second visit for all individuals assessed in this study occurred a median of 38 days following symptom onset. Although we did not observe cTfh responses in late convalescence, other groups have detected longitudinal cTfh responses up to 6 months after initial symptom onset [43,44]. These studies classified cTfh cells as CXCR5+ (no PD1) and investigated activation markers other than Ox40 and PDL1. Taken together, future investigation into longitudinal cTfh responses is warranted. A delay in cTfh response formation could be due to the T-cell dysfunction that occurs in SARS-CoV-2 infection. These high magnitude cTfh responses could also be the result of persistent antigen exposure, as several groups have reported prolonged detection of SARS-CoV-2 by PCR [50,54]. Prolonged cTfh responses could also go hand in hand with recent reports of B cell and antibody response evolution over time in natural infection [55]. However, our understanding of the natural kinetics of cTfh responses in viral infections is incomplete, and future studies should examine the frequency and magnitude of these responses over time in SARS-CoV-2 infection, but also in influenza and other common viral infections. These data further our understanding of CD4 T-cell responses, particularly cTfh responses, against SARS-CoV-2. Our study directly measures SARS-CoV-2-specific cTfh responses to three major structural proteins, M, N, and S. We clearly demonstrate that SARS-CoV-2-specific cTfh responses that arise early in convalescence strongly correlate with antibody neutralization and that S protein-specific responses most closely relate to antibody neutralization. However, we also show that cTfh responses against other SARS-CoV-2 proteins early in convalescence correlate with antibody neutralization, indicating a possible role for intrastructural help. Finally, in measuring these responses over time, we observe the emergence of several high magnitude responses more than a month following symptom onset, suggesting that cTfh response formation may be delayed in SARS-CoV-2 infection. These responses undergo contraction within the following two months and do not correlate with antibody neutralization, suggesting they may represent immune dysregulation.

Methods and materials Ethics statement All patients included in this study were adults and recruited from the University of Alabama at Birmingham (UAB) HIV care clinic, also known as the 1917 clinic, after obtaining written, informed consent and approval from the Institutional Review Board (IRB-160125005) at UAB. Patient samples Cryopreserved PBMC samples for T-cell assays and plasma samples for antibody assays were acquired through the UAB COVID Enterprise Biorepository. All samples were obtained with patient consent under the appropriate IRB guidelines. Patient demographic information is shown in Table 1. Paired Visit 1 and Visit 2 PBMC samples from 26 individuals who had recovered from COVID-19 were assessed in this study. 4 of these individuals were tested at additional later timepoints. CID2, CID11, CID12, and CID21 were tested at Visit 3, and CID11, CID12, and CID21 were tested at Visit 4. Clinical data from these individuals were retrieved from the Enterprise Biorepository REDCap database [56]. All tested individuals were symptomatic, but none were hospitalized during the course of their illness. Symptom severity was quantified using a self-reported severity score on a scale of 1 to 3, where 1 represented no interference in daily life, 2 a moderate impact on daily life, and 3 a significant decrease in quality of life due to symptoms. A majority of individuals reported moderate severity (62%, 16/26) and a minority reported mild severity of symptoms (38%, 10/26). None reported severe symptoms. Additionally, all but two had a positive SARS-CoV-2 PCR by nasopharyngeal swab. The two individuals who did not have a PCR test completed had a known COVID contact, were symptomatic, and had detectable T-cell responses. PBMCs from 10 healthy donors (all collected prior to the COVID-19 pandemic) were assessed for T-cell responses in parallel. Peptide pools Overlapping peptides spanning the SARS-CoV-2 M, N, and S proteins (NCBI reference number MN985325.1) were designed as 20mers overlapping by 10 amino acids which has previously been shown to effectively detect CD4 T-cell responses [57,58]. Peptides were synthesized by New England Peptide in a 96-well plate format and stored at -70°C after reconstitution in DMSO. Flow cytometry For activation-induced marker (AIM) staining, cells were thawed and stimulated with SARS-CoV-2 peptide pools for each of the M, N, and S proteins. An unstimulated, negative control and an SEB stimulated, positive control were included for each sample and were incubated and stained in parallel with experimental conditions. The unstimulated cells were cultured with 1% DMSO to best account for background activation levels. Co-stimulatory anti-CD28 and anti-CD49d were added (BD Pharmingen). After an 18 hour incubation at 37°C, cells were washed with FACS wash (2% FBS in PBS), stained with CCR7-PercpCy5.5 at 37°C for 20 min, washed, and then stained with the following antibodies: CD4-Pe610, CD3-A780, CD8-FITC, CD14-A700, CD19-A700, Ox40-PeCy7, PDL1-PE, CXCR5-BV421, PD1-BV785, CD45RABV510, CD137-BV650, CD69-BUV737, and Dead cell dye-UV. Cells were then washed and fixed in 2% formaldehyde. Events were collected on a BD FACSymphony A3 within 24 hours and analyzed using FlowJo software (v10). PD1 gating was set by a fluorescence minus one control (FMO). Intracellular staining (ICS) experiments were set up in parallel with the AIM staining experiments for 21 of the 26 individuals and performed similarly, with a few notable exceptions. CD107a-FITC was added with the co-stimulatory antibody mix; Monensin and Brefeldin A (BD Bioscience) were added after 1 hour. Cells were incubated for 12 hours in total, instead of 18. Staining was conducted in three steps: 1) Surface marker staining for 30 min at 4°C with Dead cell dye-UV, CD3-A780, CD4-BV785, CD8-V500, CD14-PercpCy5.5, and CD19-PercpCy5.5. 2) Permeabilization with CytoFix/CytoPerm solution (BD Biosciences) for 20min at 4°C. 3) Intracellular staining for 30 min at 4°C with IFNγ-A700, TNFα-PeCy7, and CD154-APC. CD154 was plotted against IFNγ. Additional details regarding the antibodies used in both the AIM and ICS assays can be found in S1 Table, and the gating strategies for AIM and ICS in an unstimulated, negative control are shown in S1 Fig. For both AIM and ICS, positive responses were determined as those at least three times higher than the unstimulated control and significantly higher than unstimulated control by Fisher’s Exact (p value < 0.0001). Antibody assays Plasma samples from the first and second time point were tested for SARS-CoV-2-specific antibodies. The Abbott Architect assay was used to detect immunoglobin G (IgG) reactivity to the SARS-CoV-2 nucleocapsid protein [59]. The IgG quantity is reported as a calculated index specimen/calibrator ratio, and values over 1.4 were considered positive for N protein IgG. Manufacturer-reported specificity of this assay is 99.6% (99.1%-99.9%). Antibody neutralization assays were conducted as previously described [60]. Briefly, the SARS-CoV-2 Spike (G614 variant, with a 19 amino acid cytoplasmic tail deletion) was pseudotyped onto an HIV-1 nanoluciferase reporter backbone by co-transfection in HEK 293T cells. Pseudovirus was incubated with five-fold serial dilutions of patient plasma and then used to infect 1.5x104 293T clone 22 cells expressing ACE2. Two days post-infection, cells were washed with PBS, lysed, and nanoluciferase activity was determined according to manufacturer’s instructions (Nano-Glo Luciferase Assay System). Luciferase activity in wells with virus and no patient plasma were set to 100%, and the dilution of plasma at which luminescence was reduced to 50% (ID50) was calculated. Abbott IgG titer was not conducted for the second timepoint for two individuals, CID2 and CID13. Otherwise, antibody measurements were collected for all 26 individuals. Statistical analysis Comparisons between paired visit 1 and visit 2 magnitudes were conducted by Wilcoxon signed-rank tests. Comparisons between responder frequency at visit 1 and visit 2 were conducted by Fisher’s exact tests. All correlations were determined by Spearman Rank tests, with the exception of S2A Fig, where multiple measurements were plotted for each individual (across the three proteins) and therefore a generalized linear mixed effect model accounting for multiple measurements per individual was employed. In Figs 3 and S4, axes were transformed using log10(x+1) to allow for visualization of zeros, and correlations were determined with untransformed data.

Supporting information S1 Fig. Flow cytometry gating strategies. (A) Gating strategy for CD4 T cell and cTfh by activation-induced marker (AIM). (B) Gating strategy for CD4 T cell staining by intracellular cytokine staining. https://doi.org/10.1371/journal.ppat.1009761.s001 (TIF) S2 Fig. SARS-CoV-2-reactive CD4 T cells are infrequently detected in COVID negative individuals. Representative examples of CD4 T-cell responses detected in COVID negative individuals by upregulation of activation-induced markers (A) and by intracellular cytokine staining (B) upon stimulation by SARS-CoV-2 N protein peptide pool. Responder frequency of CD4 responses to any SARS-CoV-2 protein and to the M, N, and S proteins individually by AIM (C) and ICS (D). Positive SARS-CoV-2-specific responses are indicated by gate frequencies in red. https://doi.org/10.1371/journal.ppat.1009761.s002 (TIF) S3 Fig. Upregulation of activation markers detected a broader range of SARS-CoV-2-specific CD4 T-cell responses. Correlation between response magnitude by AIM versus response magnitude by ICS. Statistics determined by mixed effect model accounting for multiple protein stimulations per individual, and correlation represented by linear regression line. Data transformed by log10(x+1) to allow for visualization of 0s. https://doi.org/10.1371/journal.ppat.1009761.s003 (TIF) S4 Fig. Nine SARS-CoV-2-specific cTfh responses were detected at the first convalescent visit. Each row shows responses from a different individual. From right to left, unstimulated, media control; M protein, N protein, S protein stimulations; and positive, SEB-stimulated control. Positive SARS-CoV-2-specific responses are indicated by gate frequencies in red. https://doi.org/10.1371/journal.ppat.1009761.s004 (TIF) S5 Fig. cTfh response frequency does not correlate with antibody neutralization at Visit 2. (A) Correlations between N protein IgG titers and cTfh frequencies towards the M, N, and S proteins. (B) Correlations between antibody neutralization (ID50, dilution of plasma at which luminescence was reduced to 50%) and cTfh frequencies. (C) Correlations between the total cTfh frequency and antibody titer and neutralization. (All correlations represented by a linear regression line. Y axis in A-B and both axes in C are transformed by log10(x+1) to allow for visualization of 0s. Statistics determined by a Spearman Correlation test. Points are colored for each individual.) https://doi.org/10.1371/journal.ppat.1009761.s005 (TIF) S6 Fig. Summary of all responses detected across the first two convalescent visits. (A-C) Response summary for CD4 T cells by activation-induced marker staining, for cTfh by activation-induced marker staining, and for CD4 T cells by intracellular cytokine staining, respectively. Blue-filled cells indicate a positive response; white cells indicate a negative response. (D) Responder frequency by AIM across Visit 1 and Visit 2 (positive at either visit) overall and to each protein. (E) Responder frequency by ICS across the first two visits (positive at either visit). https://doi.org/10.1371/journal.ppat.1009761.s006 (TIF) S7 Fig. CD4 T-cell and cTfh responses can be detected late in convalescence. Each row shows CD4 T-cell (A) and cTfh (B) responses from a different individual. From right to left, unstimulated, media control; M protein, N protein, S protein stimulations; and positive, stimulated control. Positive SARS-CoV-2-specific responses are indicated by gate frequencies in red. https://doi.org/10.1371/journal.ppat.1009761.s007 (TIF) S1 Table. Flow cytometry panels. Details of antibodies used for activation-induced marker flow cytometry and intracellular staining flow cytometry. https://doi.org/10.1371/journal.ppat.1009761.s008 (TIF)

Acknowledgments We would like to thank the Center for AIDS Research at UAB’s Flow Cytometry Core for their assistance and all those who helped process samples, especially Sarah Sterrett.

[1] Url: https://journals.plos.org/plospathogens/article?id=10.1371/journal.ppat.1009761

(C) GlobalVoices
Licensed under Creative Commons Attribution (CC BY 4.0)
URL: https://creativecommons.org/licenses/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.