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



A new class of antibodies that overcomes a steric barrier to cross-group neutralization of influenza viruses [1]

['Holly C. Simmons', 'Center For Vaccine Research', 'University Of Pittsburgh School Of Medicine', 'Pittsburgh', 'Pennsylvania', 'United States Of America', 'Department Of Microbiology', 'Molecular Genetics', 'Akiko Watanabe', 'Department Of Integrative Immunobiology']

Date: 2024-01

Abstract Antibody titers that inhibit the influenza virus hemagglutinin (HA) from engaging its receptor are the accepted correlate of protection from infection. Many potent antibodies with broad, intra-subtype specificity bind HA at the receptor binding site (RBS). One barrier to broad H1-H3 cross-subtype neutralization is an insertion (133a) between positions 133 and 134 on the rim of the H1 HA RBS. We describe here a class of antibodies that overcomes this barrier. These genetically unrestricted antibodies are abundant in the human B cell memory compartment. Analysis of the affinities of selected members of this class for historical H1 and H3 isolates suggest that they were elicited by H3 exposure and broadened or diverted by later exposure(s) to H1 HA. RBS mutations in egg-adapted vaccine strains cause the new H1 specificity of these antibodies to depend on the egg adaptation. The results suggest that suitable immunogens might elicit 133a-independent, H1-H3 cross neutralization by RBS-directed antibodies.

Citation: Simmons HC, Watanabe A, Oguin III TH, Van Itallie ES, Wiehe KJ, Sempowski GD, et al. (2023) A new class of antibodies that overcomes a steric barrier to cross-group neutralization of influenza viruses. PLoS Biol 21(12): e3002415. https://doi.org/10.1371/journal.pbio.3002415 Academic Editor: Ali H. Ellebedy, Washington University School of Medicine, UNITED STATES Received: April 25, 2023; Accepted: November 2, 2023; Published: December 21, 2023 Copyright: © 2023 Simmons 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 paper and its contained in the Supporting Information files or in the protein database under PDB accession identifiers 7TRH and 7TRI or NCBI GenBank, accession numbers OR669534-OR669571 and OR725558- OR725601 Funding: The research was supported by National Institute of Allergy and Infectious Diseases Program Project Grant P01 AI089618 (to G.H.K.) and funds from the University of Pittsburgh Center for Vaccine Research (to K.R.M).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 Abbreviations: BSA, bovine serum albumin; ELISA, enzyme-linked immunosorbent assay; HA, hemagglutinin; PEI, polyethylenimine; PBMC, peripheral blood mononuclear cell; PBS, phosphate-buffered saline; RBS, receptor-binding site; rHA, recombinant HA; SDS, sodium dodecyl sulfate; TLS, translation-liberation-screw; UCA, unmutated common ancestor

Introduction Influenza virus hemagglutinin protein (HA) mediates cellular attachment by engaging terminal sialic acid groups on glycoproteins and glycolipids. The titer of circulating antibodies that inhibit this interaction correlates with protection from infection [1]. Immunity to influenza viruses is limited by large antigenic divisions between the 18 influenza A HA serotypes, which are classified as either group 1 (H1, 2, 5, 6, 8, 9, 11, 12, 13, 16, 17, 18) or group 2 (H3, 4, 7, 10, 14, 15). As a result, humans do not mount broadly protective responses to influenza. Moreover, continuous antigenic evolution erodes immunity elicited by prior strains of any specific serotype. The strain composition of vaccines to human H1 and H3 viruses must therefore be reformulated on a nearly annual basis. Broadly protective antibodies that engage conserved epitopes on HA can nonetheless be identified in human repertoires [2], and immunogens that selectively elicit these antibodies might confer broader and longer-lasting immunity than do current vaccines. One group of broadly neutralizing influenza antibodies engage the HA receptor-binding site (RBS) with sialic acid-like contacts [2–7]. RBS-directed antibodies that neutralize decades of antigenic variation within a single serotype appear to be more common than those that cross-neutralize H1N1 and H3N2 viruses [2–10]. An insertion at the RBS periphery in H1 HAs (at a position designated 133a to adhere to H3 amino acid numbering) creates a steric block to H1-H3 neutralization by RBS-directed antibodies [5,8,10–12]. The insertion was lost between 1995 and 2009, and cross-neutralizing antibodies have been described that engage H1 HAs that circulated during this time interval [5,8,9]. The 2009 H1N1 pandemic reintroduced an HA with a 133a insertion. Potent H1-H3 cross-neutralizing antibodies to the pandemic virus or its descendants have not been reported. The RBS coordinates the terminal sialic acid of a glycan chain. Human influenza viruses have a strong preference for α2,6-linked sialic acids, while avian viruses prefer α2,3-linkages [13,14]. Propagation of human viruses in chicken eggs, the major source of vaccine material, selects for mutants that more efficiently engage α2,3 receptors. These mutations occur within the RBS and can impact RBS-directed antibody binding [15–19]. One common substitution alters the conformation of the RBS [16], such that immunization with egg-adapted HAs may elicit immunity specific for the vaccine component, and the resulting antibodies then fail to engage the circulating virus [15–21]. These mutations limit vaccine efficacy and specifically interfere with the development of RBS-directed antibody responses. We define here a previously unrecognized class of RBS-directed antibodies that are abundant in circulating human memory B cells (Bmem). Members of this class have a common motif in HCDR3 that mimics many of the authentic sialic acid receptor contacts. The antibodies we examined neutralized certain H3 and H1 strains, including some H1 HAs with the K133a insertion and some without. Thus, they illustrate that the human immune system can surmount the steric barrier to cross-neutralization generated by the 133a insertion. Their widespread distribution suggests that the barrier may be relatively low. For the 2 antibodies we characterized in detail, from donors of different ages and different geographic locations, reactivity with historical H1 isolates appears to have extended from the early years of the 21st century through about 2015; the reactivity with historical H3 isolates spanned the late 1980s to the late 1990s. For one of the antibodies, H1 binding and neutralization depended on the mutation Q226R, which was present only in the vaccine strains as a result of adaptation to growth in eggs. These data suggest that an H3-specific B cell can evolve somatically, upon exposure to H1 by infection or vaccination, to recognize an H1 HA and that the H1 adaptation need not depend on the presence or absence of a K133a insertion.

Discussion Engagement of cell-surface sialic acid moieties is an obligate step in the influenza virus replicative cycle [13]. Broadly neutralizing antibodies have been identified that mimic its contacts, while minimizing those with poorly conserved residues [2,3,5–8,11,23–26]. The antibodies described here engage sialic acid coordinating residues with a novel sequence motif that is abundant in the human B memory cell compartment. Antibodies belonging to this class cross-neutralize H1 and H3 viruses, agnostic to the presence or absence of the 133a insertion. Surmounting this steric barrier distinguishes these antibodies from previously described H1-H3 neutralizing antibodies [5,8,11]. Members of this antibody class have a common pattern of HA reactivity, defined by avid engagement of H3 HA isolates from the 1990s and H1 HAs from the 2009 H1N1 pandemic. Over a decade separates circulation of these viruses, and most of the antibodies do not engage HAs from intervening years. The pattern suggests that sequential H3 and H1 exposures gave rise to the observed lineages, especially in view of the differences among the donors and the absence of any common IGHV or IGVK/L gene usage. The response was particularly robust in S5 and S8, both of whom were infants or small children in the early 1990s. Infection with an early 1990s strain was therefore probably their first influenza exposure and hence the source of their initial immune imprint, as pediatric vaccination was not yet recommended at that time. KEL03, born in 1975, could plausibly have experienced an infection with an H3N2 virus during the 1990s and probably received an H3N2 vaccination. All 3 donors were likely susceptible to infection with 2009 H1N1 pandemic viruses, and each did indeed receive H1N1 vaccinations after 2009. In contrast, antibodies with the E-G-W motif were not recovered from donors S1 and S9 who were born in the 1960s and likely have distinctly different influenza exposure histories, particularly for H3N2 viruses, from S5 and S8. While humans have the capacity to readily produce naive progenitors of these antibodies, abundance in some donors may be a product of a specific series of exposures. When confident inference of the UCA of a lineage is possible, then the reactivity of that UCA with a panel of historical strains can identify the likely date of the primary exposure that gave rise to that lineage [24]. The uniquely determined UCA of the lineage from S5 and S8 both bind HA from A/Johannesburg/33/1994 (H3N2), but not much earlier or much later H3 HAs, as expected for a primary exposure in the early 1990s. Although K03.28 has no clonal lineage siblings in our data set, its reactivity also suggests that it derives from an early 1990s primary response. Since the K03 donor was born almost 20 years earlier, the properties of the antibody suggest that even if recall of memory from a first exposure appears to dominate later responses (i.e., “immune imprinting” [27,28]), primary responses to later exposures can contribute new specificities to the Bmem repertoire. Most administered influenza vaccines deliver HA immunogens derived from viruses propagated in embryonated eggs. Egg growth can yield RBS mutant viruses with tropism for cells bearing α2,3 sialic acids (required for avian transmission) instead of or in addition to α2,6-linked sialic acids (required for human transmission) [13,14]. The Q226R mutation is a characteristic H1 substitution and frequently occurs in new pandemic vaccines [15,18,29]. While not absolute, many antibodies described here depend upon R226 for H1 binding. Immunization with H1 R226 HAs has been shown to misdirect H1 antibody responses [15,18]. Our observations suggest that R226 can also influence H3 and H1-H3 responses. Whether there is some relationship between the RBS antigenic surfaces of mid-1990s H3 HAs and the initial 2009 pandemic vaccine A/California/07/2009(H1N1)(X-181) HA, as the observations from S5 and S8 suggest, will need additional examples. Nonetheless, in view of the observation that a single residue can govern H1-H3 neutralization, barriers to cross-serotype neutralization by RBS-directed antibodies may be relatively low. These and previous results also illustrate the need to transition away from egg-grown components (or even egg-selected reassortants) in influenza vaccine production [15–21]. Conserved sialic acid-coordinating HA residues provide a target for broadly neutralizing antibodies [2,3,5–8,11,23–26]. Broadly protective antibodies directed to the HA-stem or HA-head interface generally require antibody Fc-mediated killing of already infected cells [22,30,31], while genuinely neutralizing antibodies block the apparent establishment of infection. Although individual RBS directed antibodies lack universal breadth due to rapid antigenic evolution around the RBS periphery, their abundance and lack of genetic restrictions suggest that polyclonality can achieve broad protection. Antibodies K03.28 and those from the K03.12 lineage were isolated from donor KEL03 [8]. Together, they engage H1 HAs from 1995 to 2015 and nearly all H3 HAs from 1994 to at least 2014. This breadth includes pre- and post-2009 pandemic H1N1 isolates. Achieving broadly protective immunity will likely require polyclonal responses, directed to multiple conserved epitopes and that protect by multiple mechanisms. The antibodies described here expand the potential repertoire of broadly neutralizing antibodies that can contribute to such broad protection.

Methods Cell lines Human 293F cells were maintained at 37°C with 5% CO2 in FreeStyle 293 Expression Medium (Thermo Fisher) supplemented with penicillin and streptomycin. High Five Cells (BTI-TN-5B1-4) (Trichoplusia ni) were maintained at 28°C in EX-CELL 405 medium (Sigma) supplemented with penicillin and streptomycin. Recombinant Fab expression and purification Synthetic heavy- and light-chain variable domain genes for Fabs were cloned into a modified pVRC8400 expression vector, as previously described [32]. Fab fragments used in crystallization were produced with a C-terminal, noncleavable 6xhistidine (6xHis) tag. Fab fragments were produced by polyethylenimine (PEI) facilitated, transient transfection of 293F cells that were maintained in FreeStyle 293 Expression Medium. Transfection complexes were prepared in Opti-MEM and added to cells. Supernatants were harvested 4 to 5 days post transfection and clarified by low-speed centrifugation. Fabs were purified by passage over Co-NTA agarose (Clontech) followed by gel filtration chromatography on Superdex 200 (GE Healthcare) in 10 mM Tris-HCl, 150 mM NaCl at pH 7.5 (buffer A). Single B cell Nojima cultures Nojima cultures were previously performed [22]. Briefly, PBMCs were obtained from 4 human subjects: S1 (female, age range 51 to 55), S5 (male, age 21 to 25), S8 (female, age 26 to 30), and S9 (female, age 51 to 55). Single human Bmem cells were directly sorted into each well of 96-well plates and cultured with MS40L-low feeder cells in RPMI1640 (Invitrogen) containing 10% HyClone FBS (Thermo Scientific), 2-mercaptoethanol (55 μm), penicillin (100 units/ml), streptomycin (100 μg/ml), HEPES (10 mM), sodium pyruvate (1 mM), and MEM nonessential amino acid (1×; all Invitrogen). Exogenous recombinant human IL-2 (50 ng/ml), IL-4 (10 ng/ml), IL-21 (10 ng/ml), and BAFF (10 ng/ml; all Peprotech) were added to cultures. Cultures were maintained at 37° Celsius with 5% CO 2 . Half of the culture medium was replaced twice weekly with fresh medium (with fresh cytokines). Rearranged V(D)J gene sequences for human Bmem cells from single-cell cultures were obtained as described [22,23,33]. Specificity of clonal IgG antibodies in culture supernatants and of rIgG antibodies was determined in a multiplex bead Luminex assay (Luminex Corp.). Culture supernatants and rIgGs were serially diluted in 1 × phosphate-buffered saline (PBS) containing 1% BSA, 0.05% NaN 3 , and 0.05% Tween20 (assay buffer) with 1% milk and incubated for 2 h with the mixture of antigen-coupled microsphere beads in 96-well filter bottom plates (Millipore). After washing 3 times with assay buffer, beads were incubated for 1 h with Phycoerythrin-conjugated goat anti-human IgG antibody (Southern Biotech). After 3 washes, the beads were re-suspended in assay buffer and the plates read on a Bio-Plex 3D Suspension Array System (Bio-Rad). Recombinant IgG expression and purification The heavy chain variable domains of selected antibodies were cloned into a modified pVRC8400 expression vector to produce a full-length human IgG1 heavy chain [22,32,34]. IgGs were produced by transient transfection of 293F cells as specified above. Five days post-transfection supernatants were harvested, clarified by low-speed centrifugation, and incubated overnight with Protein A Agarose Resin (GoldBio). The resin was collected in a chromatography column, washed with a column volume of buffer A, and eluted in 0.1M Glycine (pH 2.5) which was immediately neutralized by 1M tris(hydroxymethyl)aminomethane (pH 8.5). Antibodies were then dialyzed against PBS (pH 7.4). Recombinant HA expression and purification Recombinant HA (rHA) head domain constructs were expressed by infection of insect cells with recombinant baculovirus as previously described [32,34]. In brief, a synthetic DNA corresponding to the globular HA-head was subcloned into a pFastBac vector modified to encode a C-terminal rhinovirus 3C protease site and a 6xHis tag. Supernatant from recombinant baculovirus infected High Five Cells (Trichoplusia ni) was harvested 72 h post infection and clarified by centrifugation. Proteins were purified by adsorption to Co-NTA agarose resin, followed by a wash in buffer A, a second wash (trimers only) with buffer A plus 5–7 mM imidazole, elution in buffer A plus 350 mM imidazole (pH 8) and gel filtration chromatography on a Superdex 200 column (GE Healthcare) in buffer A. Full-length HA ectodomain (FLsE) were produced by PEI facilitated, transient transfection of 293F cells maintained in FreeStyle 293 Expression Medium. Synthetic DNA corresponding to the full-length ectodomain (FLsE) were cloned into a pVRC vector modified to encode a C-terminal thrombin cleavage site, a T4 fibritin (foldon) trimerization tag, and a 6xHis tag [32,35]. Transfection complexes were prepared in Opti-MEM and added to cells. Supernatants were harvested 4 to 5 days post transfection and clarified by low-speed centrifugation. HA trimers were purified by passage over Co-NTA agarose (Clontech) followed by gel filtration chromatography on Superdex 200 (GE Healthcare) in 10 mM Tris-HCl, 150 mM NaCl at pH 7.5 (buffer A). ELISA Five hundred nanograms of rHA FLsE were adhered to high-capacity binding, 96-well plates (Corning) overnight in PBS (pH 7.4) at 4°C. Plates were washed with a PBS-Tween-20 (0.05%v/v) buffer (PBS-T) and then blocked with PBS-T containing 2% bovine serum albumin (BSA) for 1 h at room temperature. Blocking solution was then removed, and 5-fold dilutions of IgGs (in blocking solution) were added to wells. Plates were then incubated for 1 h at room temperature followed by removal of IgG solution and 3 washes with PBS-T. Secondary, anti-human IgG-HRP (Abcam ab97225) diluted 1:10,000 in blocking solution was added to wells and incubated for 30 min at room temperature. Plates were then washed 3 times with PBS-T. Plates were developed using 150 μl 1-Step ABTS substrate (Thermo Fisher, Prod#37615). Following a brief incubation at room temperature, HRP reactions were stopped by the addition of 100 μl of 1% sodium dodecyl sulfate (SDS) solution. Plates were read on a Molecular Devices SpectraMax 340PC384 Microplate Reader at 405 nm. KD values for ELISA were obtained as follows. All measurements were performed in technical triplicate. The average background signal (no primary antibody) was subtracted from all absorbance values. Values from multiple plates were normalized to the S5V2-29 standard that was present on each ELISA plate. The average of the 3 measurements were then graphed using GraphPad Prism (v9.0). KD values were determined by applying a nonlinear fit (One site binding, hyperbola) to these data points. A Bmax constraint of Bmax must be greater than 0.1 absorbance units was applied to all KD analysis parameters. Standard error of the mean (SEM) were calculated in GraphPad Prism and plotted as error bars. Clonal antibody lineages and UCA inference For both subject 5 and subject 7, the heavy chains of the sequences from time points 1 and 2 were grouped into clones based on the same V and J gene segment usage, same CDR3 length, and CDR3 sequence identity, using the software package Cloanalyst [36] (https://www.bu.edu/computationalimmunology/research/software/). For clones S5V2-107 and S8V1-172, paired heavy and light chain sequences of each clone were used to infer UCA sequences and their lineages were reconstructed using Cloanalyst [36] and visualized as clonograms using FigTree v1.4.4 (https://github.com/rambaut/figtree/). Virus microneutralization assays Virus neutralization endpoint titers were determined using the influenza microneutralization assay as described [8,37–40]. Monoclonal antibodies were diluted to test concentration in 2-fold dilution series in virus diluent in a flat-bottomed 96-well tissue culture plate. Samples were then mixed equimolar with virus diluent containing 100 TCID50 of each influenza virus of interest. After virus addition, samples are incubated for 60 min at 37°C 5% CO2; 1.5e4 MDCK cells (London strain, IRR FR-58) were added to each well. Plates were incubated overnight at 37°C 5% CO2. Each well was aspirated, and cells were washed 1 time with PBS. The PBS was aspirated. Then, 250 μl of −20°C 80% acetone was added to each well, and plates were incubated at room temperature for 10 min. The acetone was removed and plates air-dried. Each well was washed 3 times with wash buffer. Primary antibody (Mouse Anti-Influenza A NP, Millipore MAB8251 or Mouse Anti-Influenza B NP, Millipore MAB8661) was diluted 1:4,000 in antibody diluent, and 50 μl was added to every well. Plates were incubated for 60 min at room temperature. Each well was washed 3 times with wash buffer. Secondary antibody (Goat anti-mouse + horseradish peroxidase, KPL 474–1802) was diluted 1:4,000 in antibody diluent, and 50 μl was added to every well. Plates were incubated for 60 min at room temperature. Each well was washed 5 times with wash buffer, and 100 μl substrate was added to every well and incubated at room temperature. The reaction was stopped with 100 μl 0.5 N sulfuric acid after apparent color change was observed in virus-only control wells. Absorbance was read at 490 nM in a Synergy H1 automated microplate reader (BioTek Instruments). Wells with absorbance values less than or equal to 50% of virus-only control wells were scored as neutralization positive. Data were expressed as the geometric mean of the reciprocal of the final dilution factor that was positive for neutralization. All samples were assayed in at least duplicates. Endpoint values are reported. Influenza viruses were propagated in embryonated-specific pathogen-free chicken hen eggs or MDCK (CCL-34) cells as described [8]. Reagents obtained through BEI Resources, NIAID, NIH include: Influenza A viruses A/Aichi/2/1968 (H3N2) NR-3177; Kilbourne F123: A/Victoria/3/1975 (HA, NA) x A/Puerto Rico/8/1934 (H3N2), Reassortant X-47 NR-3663; A/Philippines/2/1982 (H3N2) NR-28649; Kilbourne F178: A/Shanghai/11/1987 (HA, NA) x A/Puerto Rico/8/1934 (H3N2), High Yield, Reassortant X-99a NR-3505; Kilbourne F86: A/Johannesburg/33/1994 (HA, NA) x A/Puerto Rico/8/1934 (H3N2), Reassortant X-123a NR-3580; Kilbourne F97: A/Moscow/10/1999 (HA, NA) x A/Puerto Rico/8/1934 (H3N2), Reassortant X-137 NR-3587; A/Kansas/14/2017 Reassortant X-327 (H3N2) FR-1697; A/Michigan/45/2015(H1N1) FR-1483; A/California/07/2009 Reassortant NYMC X-181 (H1N1) NR-44004; A/California/04/2009(H1N1) NR-13659; polyclonal influenza virus, A/Aichi/2/1968 (H3N2) serum (guinea pig), NR-3126; NR-4282. A/USSR/90/1977, reassortant X-67 (H1N1) NR-3666; A/Chile/01/1983 reassortant X-83 (H1N1) NR-3585, A/Beijing/262/1995 reassortant X-127 (H3N2) NR-3571; A/Solomon Islands (H1N1) NR-41798. Influenza A virus A/Wisconsin/67/2005 (H3N2), FR-397; and MDCK London cells (FR-58) were obtained through the International Reagent Resource (formerly the Influenza Reagent Resource), Influenza Division, WHO Collaborating Center for Surveillance, Epidemiology and Control of Influenza, Centers for Disease Control and Prevention, Atlanta, Georgia, United States of America. Crystallization Fab fragments were co-concentrated with HA-head domains at a molar ratio of approximately 1:1.3 (Fab to HA-head) to a final concentration of approximately 20 mg/ml. Crystals of Fab-head complexes were grown in hanging drops over reservoir solutions Crystals of the K03.28- A/California/07/2009(H1N1)(X-181) HA head domain were grown in hanging drops over a reservoir of 25% (w/v) poly(ethylene glycol) 1500. Crystals of S8V1-172 complexed with the HA head domain of A/Sydney/05/1997(H3N2) were grown in drops over a reservoir of 0.1 M tris(hydroxymethyl)aminomethane (pH 8.5) and 25% (w/v) poly(ethylene glycol) 3350. Crystals were cryoprotected with glycerol at concentrations of 22% in cryoprotectant buffers that were 20% more concentrated than the well solution. Cryoprotectant was added directly to the drop, crystals were harvested, and flash cooled in liquid nitrogen. Structure determination and refinement We recorded diffraction data at the Advanced Photon Source on beamline 24-ID-C. Data were processed and scaled (XSCALE) with XDS [41]. Molecular replacement was carried out with PHASER [42], dividing each complex into 4 search models (HA-head, Vh, Vl, and constant domain). Search models were 3UBE, 6E56, 4HK0, 4WUK for the K03.28-HA head domain complex and 6XPZ, 6E4X, 6MHR, 6E4X for the S8V1-172-HA head domain complex. We carried out refinement calculations with PHENIX [43] and model modifications, with COOT [44]. Refinement of atomic positions and B factors was followed by translation-liberation-screw (TLS) parameterization and, if applicable, placement of water molecules. All placed residues were supported by electron density maps and subsequent rounds of refinement. Final coordinates were validated with the MolProbity server [45]. Data collection and refinement statistics are in S1 Table. Figs were made with PyMOL (Schrödinger, New York, New York, USA).

Acknowledgments We thank the many members of our Program Project Consortium for advice and discussion. We thank Lindsey R. Robinson-McCarthy for invaluable discussion and comments. X-ray diffraction data were recorded at beamline ID-24-C (operated by the Northeast Collaborative Access team: NE-CAT) at the Advanced Photon Source (APS, Argonne National Laboratory). We thank NE-CAT staff members for advice and assistance in data collection. NE-CAT is funded by NIH grant P30 GM124165. APS is operated for the DOE Office of Science by Argonne National Laboratory under contract DE-AC02-06CH11357.

[END]
---
[1] Url: https://journals.plos.org/plosbiology/article?id=10.1371/journal.pbio.3002415

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.