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



Identification of a novel lipoic acid biosynthesis pathway reveals the complex evolution of lipoate assembly in prokaryotes [1]

['Tomohisa Sebastian Tanabe', 'Institut Für Mikrobiologie', 'Biotechnologie', 'Rheinische Friedrich-Wilhelms-Universität Bonn', 'Bonn', 'Martina Grosser', 'Lea Hahn', 'Carolin Kümpel', 'Hanna Hartenfels', 'Evelyn Vtulkin']

Date: 2023-07

Lipoic acid is an essential biomolecule found in all domains of life and is involved in central carbon metabolism and dissimilatory sulfur oxidation. The machineries for lipoate assembly in mitochondria and chloroplasts of higher eukaryotes, as well as in the apicoplasts of some protozoa, are all of prokaryotic origin. Here, we provide experimental evidence for a novel lipoate assembly pathway in bacteria based on a sLpl(AB) lipoate:protein ligase, which attaches octanoate or lipoate to apo-proteins, and 2 radical SAM proteins, LipS1 and LipS2, which work together as lipoyl synthase and insert 2 sulfur atoms. Extensive homology searches combined with genomic context analyses allowed us to precisely distinguish between the new and established pathways and map them on the tree of life. This not only revealed a much wider distribution of lipoate biogenesis systems than expected, in particular, the novel sLpl(AB)–LipS1/S2 pathway, and indicated a highly modular nature of the enzymes involved, with unforeseen combinations, but also provided a new framework for the evolution of lipoate assembly. Our results show that dedicated machineries for both de novo lipoate biogenesis and scavenging from the environment were implemented early in evolution and that their distribution in the 2 prokaryotic domains was shaped by a complex network of horizontal gene transfers, acquisition of additional genes, fusions, and losses. Our large-scale phylogenetic analyses identify the bipartite archaeal LplAB ligase as the ancestor of the bacterial sLpl(AB) proteins, which were obtained by horizontal gene transfer. LipS1/S2 have a more complex evolutionary history with multiple of such events but probably also originated in the domain archaea.

Here, we provide conclusive experimental evidence for the existence of a novel sLpl(AB)-LipS1/S2-based lipoate assembly pathway not only in archaea but also in bacteria and raise the question of whether it is restricted to thermophilic archaea and sulfur-oxidizing bacteria or is of more general importance. To date, no studies have been published on the origin and evolution of lipoate assembly machineries despite the importance of lipoate for almost all living organisms. This prompted us to carry out an exhaustive large-scale analysis including a large fraction of prokaryotic diversity, and in particular, the archaea for which knowledge about lipoate assembly is scarce. We mapped the novel lipoate synthesis pathway on the tree of life revealing an enormously wide distribution. Finally, our analyses show that the novel lipoate synthesis pathway evolved in the archaeal domain.

Recently, proteins from the thermophilic archaeon Thermococcus kodakarensis similar to sLpl(AB) ligase (Tk-Lpl-N and Tk-Lpl-C) and RadSAM1 and RadSAM2 from sulfur oxidizers (now termed LipS1 and LipS2) were shown to exert octanoate/lipoate:protein ligase and LipA-like lipoate synthase functions, respectively, on chemically synthesized peptide substrates in vitro. Genetic analysis provided further evidence that these proteins are involved in archaeal lipoate biosynthesis [ 16 , 27 , 29 ] ( Fig 1B ). We have previously shown that the slpl(AB)-encoded protein:lipoate ligase from the gammaproteobacterial sulfur oxidizer Thioalkalivibrio sp. K90mix accepts only free precursors, i.e., octanoate or lipoate, and lacks octanoyltransferase activity [ 2 ]. The archaeal ligase is also thought to be restricted to free substrates, mainly because ACPs do not occur in archaea [ 30 ].

Initial evidence for an alternative pathway for lipoate assembly in prokaryotes emerged, when we showed that the GcvH-like LbpA proteins involved in sHdr-based sulfur oxidation are not modified by the canonical E. coli and Bacillus subtilis lipoyl attachment machineries [ 2 ]. Instead, the bacterial and archaeal shdr-lbpA clusters are accompanied by a set of genes encoding a specific lipoylation pathway ( Fig 1B and 1C ) that includes lipoate:protein ligases (sLpl(AB)) and 2 proteins of the radical SAM superfamily, originally termed RadSAM1 and RadSAM2. sLpl(AB) lipoate–protein ligases from sulfur oxidizers not only lipoylate LbpA acceptor proteins from the same organism in vitro but also show cross-species functionality among sulfur oxidizers while failing to recognize lipoyl domains/proteins from organisms lacking components of a sHdr-LbpA sulfur-oxidizing system [ 2 ].

( a ) Main known steps of established lipoate assembly pathways. Enzymes and steps not occurring in E. coli but described for other organisms are printed in gray. LipM and LipL have been demonstrated in Firmicutes, B. subtilis [ 24 ], Staphylococcus aureus [ 25 ], and Listeria monocytogenes [ 26 ], as well as in Tenericutes, Mycoplasma hyopneumoniae [ 15 , 16 ]. ( b ) Predicted novel lipoate assembly pathway. The pathway is substantiated by experiments reported here as well as by published work on proteins from the 3 model organisms depicted in c [ 2 , 9 ] and by genetic and biochemical work on LipS1 and LipS2 from the archaeon Thermococcus kodakarensis [ 27 ]. Lipoate:protein ligases from sulfur oxidizers were originally reported not to contain a carboxy-terminal LplB domain based on superposition of the structure modeled for Thioalkalivibrio sp. K90 by using the automated mode of SWISS_Model on E. coli Lpl(AB). We challenged this view and indeed, modeling by Alphafold [ 28 ] as well as sequence alignments yielded clear proof for the presence of the LplB domain ( S2 Fig ). ( c ) Genetic arrangement of 3 novel systems for lipoate assembly in Proteobacteria. Colors correspond to the biochemical roles as depicted in b . For Ts. sibirica locus tags are given according to JGI-IMG. LipT is an FAD-binding NAD(P)H-dependent oxidoreductase possibly delivering electrons for the LipS1/S2-catalyzed sulfur insertion step. The genes lipY and lipX encode a putative fatty acid transporter and a putative glutamine amidotransferase, respectively. ACP, acyl carrier protein; GcvH, glycine cleavage system protein H; LbpA, lipoate-binding protein; LD, lipoyl domains of the 2-oxoacid dehydrogenases.

Two posttranslational machineries are known to construct lipoyl moieties [ 1 , 3 ]: The first requires an acyl carrier protein (ACP)-bound octanoyl residue from endogenous fatty acid biosynthesis to be transferred to the ε-amino groups of conserved lysine residues in the accepting apo-proteins. In the second, free lipoate or octanoate are hooked up to the target lysine. Irrespective of the initial step, a sulfur atom must be added to each of the octanoyl C 6 and C 8 atoms to complete lipoate biosynthesis ( Fig 1A ). Using free precursors involves the enzyme lipoate:protein ligase that activates the precursors to lipoyl- or octanoyl-AMP at the expense of ATP before transfer to the target protein. In many bacteria, including Escherichia coli, the ligase consists of 2 fused domains, the catalytic domain LplA and the accessory domain LplB [ 10 , 11 ]. We denote these enzymes Lpl(AB) or in the circularly permutated case [ 12 ], Lpl(BA). Ligases with tight substrate specificity have been described that transfer free precursors only to GcvH and, in one case, also to the E2 subunit of 2-oxoglutarate dehydrogenase [ 13 – 15 ]. Additional amidotransferases or ligases are then necessary for modification of other lipoyl domains [ 13 , 14 , 16 ]. Bipartite lipoate–protein ligases forming a functional LplA-LplB heterodimer (denoted LplAB here) have so far been found primarily in archaea [ 10 ]. In the absence of free precursors, an octanoyl residue derived from ACP is attached to the target protein by an octanoyltransferase, LipB or LipM [ 1 , 16 – 18 ]. LipB has been found mainly in Proteobacteria, serves as an all-purpose transferase and provides octanoate or lipoate to all known lipoate-requiring pathways except the sHdr-system [ 2 , 19 ]. LipM has been proposed to occur predominantly in Firmicutes and to transfer octanoyl residues exclusively to GcvH. An amidotransferase, LipL, is required for the transfer of octanoyl or lipoyl moieties from GcvH to the E2-subunits of pyruvate and branched-chain α-ketoacid dehydrogenases [ 13 , 14 , 20 ]. Despite poor sequence conservation, an evolutionary relationship has been detected between lipoate:protein ligases and octanoyltransferases as well as biotin:protein ligases (BirA) [ 6 ]. Once the octanoyl residues arrive at their target proteins, they become substrates for lipoate synthase LipA, a member of the radical S-adenosylmethionine (SAM) superfamily [ 21 ], which sequentially adds 2 sulfur atoms in a single reaction, first at position C 6 and then at C 8 [ 1 , 22 , 23 ].

α-Lipoic acid is a cofactor found in all domains of life and is involved in key reactions of central carbon metabolism and dissimilatory sulfur oxidation [ 1 – 4 ]. In this eight-carbon saturated fatty acid, sulfur atoms replace the hydrogen atoms of carbons 6 and 8 of the acyl chain [ 5 ]. Only a few, but particularly important, lipoic acid-dependent enzyme systems have been described [ 2 , 3 , 6 ]. These include 3 α-ketoacid dehydrogenases, such as pyruvate dehydrogenase, whose E2 subunits bind lipoic acid. In the glycine cleavage complex, lipoate is bound to the glycine cleavage H protein (GcvH) [ 3 ]. Lipoylated proteins also play an important role in combatting reactive oxygen species [ 7 , 8 ]. In 2018, we discovered another lipoate-binding protein (LbpA) homologous to GcvH ( S1 Fig ) and demonstrated that it is an essential component of the sulfur-oxidizing heterodisulfide reductase-like (sHdr) system present in a wide range of bacterial and archaeal dissimilatory sulfur oxidizers [ 2 , 9 ].

Results

Biochemical and genetic proof for an sLpl(AB)-LipS1/S2-based lipoate assembly pathway in bacteria The initial proposal of a novel route for maturation of lipoate-binding proteins in bacteria relied on the detection of conspicuous lipS1-sLpl(AB)-lipT-lipS2 gene clusters, in vitro assays with sLpl(AB) lipoate:protein ligase from a model sulfur oxidizer and genetic complementation studies in E. coli and B. subtilis [2]. Here, we set out to collect conclusive experimental evidence for the functionality of the pathway in bacteria. A focus was kept on the biochemically characterized sLpl(AB) ligases from the sulfur oxidizers Thiorhodospira sibirica and Thioalkalivibrio sp. K90mix [2]. First, 3 Strep-tagged LbpA lipoate acceptor proteins from these 2 bacteria were recombinantly produced in E. coli, with or without a helper plasmid from which the Thioalkalivibrio assembly genes lipS1-slpl(AB)-lipT-lipS2-lipY were expressed under control of the pACYC184 tet promoter (Fig 2A and 2B). Native gel electrophoresis showed the faster mobility expected for holo-LpbAs only when produced in the presence of the helper plasmid. This is due to the lack of the positive charge when the lipoate-binding lysine is modified by covalent attachment of lipoate or octanoate (Fig 2A). Mass spectrometric analyses confirmed posttranslational modification of all 3 LbpA acceptor proteins by a 157-Da mercaptooctanyol moiety in the presence of the helper plasmid (S3 Fig), fully consistent with in vitro results for the archaeal system where LipS2 first catalyzes sulfur attachment at C 8 of an artificial octanoyllysyl peptide substrate and LipS1 then inserts the second sulfur at C 6 [29]. Although this last step was not efficiently catalyzed in the heterologous environment, our experiments clearly confirm specific modification of sulfur oxidizer LbpA acceptor proteins by lipoate assembly proteins from a sulfur oxidizer. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 2. Biochemical and genetic evidence for a novel lipoate assembly pathway in bacteria. (a) LbpAs from Thioalkalivibrio sp. K90mix and Ts. sibirica were produced in E. coli BL21(DE3) ΔiscR, a strain designed for improved synthesis of iron-sulfur proteins [32], either with or without a helper plasmid (pACYC-Tklpm) carrying genes lipS1-slpl(AB)-lipT-lipS2-lipY from Thioalkalivibrio sp. K90mix (shown in b) under control of the constitutive pACYC184 tet promoter. Holo-LbpAs migrate faster in native PAGE due to loss of the positive lysine charge upon modification. In the heterologous host, TsLbpA proteins are—albeit not fully—modified by the assembly proteins stemming from a different species, i.e., Thioalkalivibrio. (c) Thiosulfate (triangles) and sulfite (boxes) concentrations for 4 different H. denitrificans strains during growth on methanol (24.4 mM) as a carbon source in the presence of 2 mM thiosulfate. (d) Growth of H. denitrificans strains. Symbols and lines in c and d correspond to H. denitrificans strains as follows: filled black symbols, solid lines: H. denitrificans ΔtsdA; symbols filled gray, solid lines: H. denitrificans ΔtsdA lbpA2-His; open symbols, solid lines: H. denitrificans ΔtsdA ΔlbpA2; open symbols, dotted lines: H. denitrificans ΔtsdA lbpA2-His Δslpl(AB). For all measurements, standard deviations based on 3 technical replicates are indicated, but too small to be visible for determination of biomass and sulfite. (e) SDS-PAGE of HdLbpA2-His enriched from H. denitrificans ΔtsdA lbpA2-His (left lane, 2.5 μg protein) and ΔtsdA Δslpl(AB) lbpA2-His (right lane, 1.5 μg protein). (f) Native gel mobility shift assay for HdLbpA2-His enriched from H. denitrificans ΔtsdA lbpA2-His (left lane, 2.5 μg protein) and ΔtsdA Δslpl(AB) lbpA2-His (right lane, 1.3 μg protein). HdLbpA2-His proteins were visualized after Western blotting using an Anti-His peroxidase conjugate. The data underlying panels c and d is provided as S1_data.xlsx. https://doi.org/10.1371/journal.pbio.3002177.g002 In a second approach, 4 strains of the Alphaproteobacterium Hyphomicrobium denitrificans were studied. The organism is accessible for manipulative genetics, the necessity of its LbpA2 protein for the oxidation of thiosulfate is documented [2,9] and the respective genes are located in immediate vicinity of a lipS1-lipT-lipS2-slpl(AB) cluster (Fig 1C). H. denitrificans ΔtsdA served as the reference strain. It lacks thiosulfate dehydrogenase (TsdA) that catalyzes the formation of the dead-end product tetrathionate. Thus, the strain can oxidize thiosulfate exclusively via the sHdr-LbpA pathway [9,31] that substantially simplifies its elucidation by reverse genetics. When grown in the presence of methanol as a carbon source and thiosulfate as an additional electron source, it excretes toxic sulfite, which causes growth retardation [31]. Functionality of the sHdr-LbpA pathway is thus easily detectable by sulfite formation and diminished growth rate (Fig 2C and 2D). The second strain studied carries a ΔlbpA2 deletion in a ΔtsdA background, is unable to oxidize thiosulfate, and served as control. In the third strain, lbpA2-His, encoding carboxy-terminally His-tagged LbpA2, replaces the original lbpA gene in the chromosome of H. denitrificans ΔtsdA, so that LbpA2 can be purified from this strain. The fourth strain carries an in frame deletion of slpl(AB) in a ΔtsdA lbpA2-His background. Thus, the modification of the lipoate acceptor LbpA2 can be compared in presence or absence of the sLpl(AB) ligase. H. denitrificans ΔtsdA and ΔtsdA lbpA2-His oxidized thiosulfate completely and excreted up to 0.6 mM sulfite (Fig 2C and 2D). This demonstrates that the addition of the carboxy-terminal His-tag does not prevent proper function of the LbpA2 protein. In contrast, in both, the ΔtsdA ΔlbpA2 and the ΔtsdA lbpA2-His Δslpl(AB) strains, thiosulfate degradation was very slow, sulfite formation was not observed (Fig 2C), and the growth rates were higher than those for strains ΔtsdA and ΔtsdA lbpA2-His (Fig 2D). This confirmed the crucial function of LbpA2 in the cytoplasmic sHdr-LpbA sulfur oxidation pathway and more importantly, showed that the absence of sLpl(AB) lipoate:protein ligase had the same effect as the complete absence of the LbpA2 protein, strongly suggesting that sLpl(AB) ligase is essential for the modification and thus the proper function of LbpA2. We found evidence in support of this hypothesis by enriching the LbpA2 acceptor proteins from the H. denitrificans strains ΔtsdA lbpA2-His and ΔtsdA lbpA2-His Δslpl(AB) producing them with a His-Tag and comparing their behavior in SDS and native PAGE. While the H. denitrificans strain lacking sLpl(AB) ligase produced only apo-LbpA2, the holo-protein was produced in the strain containing the complete assembly pathway as evident from the native gel mobility shift (Fig 2E and 2F).

LipS1/S2-type lipoate synthases and their cooperation partners Regarding the novel lipoate synthesis pathway, our analyses confirm that lipS1/S2 genes are often associated with genes for lipoate:protein ligases (Figs 1C and 3) [2], usually with a Lpl(AB) domain structure (S5 Table). A lipS2-lplA-lplB-lipS1 arrangement seems typical for Archaea but is also found with some rearrangement in the bacterial phyla Chloroflexota, Aerophobota, and Synergistota. Occasionally, direct linkage of genes for canonical lipoate synthesis with lipS1/S2 is observed, e.g., in several Sporomusa species (phylum Firmicutes). In some LipS1/S2-containing bacteria (e.g., members of the Schekmanbacteria, Synergistota, and Thermotogota) and archaea (members of the Asgardarchaeota, Thermoproteota, and Themoplasmatota), a gene for lipoate:protein ligase is absent (Fig 3 and S2 Table). Instead, lipS1/S2 co-occur with a gene for octanoyltransferase LipM. In the bacterial cases, this implies that LipS1/S2 insert sulfur into target proteins that have been octanoylated by a transferase reaction. For the archaea, as discussed above, the question of the substrate for the LipM homologs is unresolved. The lipT encoded FAD-containing oxidoreductase is a likely candidate to provide electrons, probably derived from NAD(P)H, for the reductive sulfur insertion catalyzed by LipS1/S2 (Fig 1). Indeed, lipT occurs almost exclusively in bacteria containing lipS1/S2 (91% of the cases), often in a lipS1-lpl(AB)-lipS2-lipT arrangement (Fig 1C and S5 Table). The picture is different for archaea, where only 22.3% of the lipT-containing genomes also contain lipS1 and lipS2 (S4 Table). Approximately 53% and 17% of the LipS1/S2-containing bacteria and archaea, respectively, also encode LipT (S2, S3 and S4 Tables). The lipS1/S2 genes were first detected in bacterial and archaeal sulfur oxidizers that use the sHdr pathway for sulfur oxidation [2]. LbpA proteins are essential components of this pathway [2] and are encoded in shdr-containing genomes with very few exceptions, probably due to incompleteness of the respective assemblies (Fig 3 and S2 and S3 Tables), raising the question of whether the assembly of LbpA proteins is strictly dependent on LipS1/S2. While the majority of sHdr-containing prokaryotes are indeed equipped with LipS1/S2 (74.3% and 88.6% for bacteria and archaea, respectively; Fig 1C and S2, S3 and S4 Tables), the reverse is not true, i.e., LipS1/S2 are not restricted to sulfur oxidizers (42% and 8% of LipS1/S2-containing bacteria and archaea, respectively, have sHdr; S2, S3 and S4 Tables).

Evolution of lipoate:protein ligases and octanoyltransferases All lipoate:protein ligases and octanyoltransferases belong to the cofactor transferase family PF03099. Calculating rooted phylogenetic trees for these proteins was expected to bring new insights into the origin and evolution of these enzymes. If they initially originated in archaea, the tree should be a priori rooted in the archaeal domain and similarly for bacteria. The structurally related biotin ligase BirA, which is also a member of the cofactor transferase family, was chosen as a suitable outgroup to root the tree. The tree for the complete lipoate:protein ligase/octanoyltransferase dataset contains 3 clearly delineated clades with high bootstrap support (Fig 4). The first clade contains the bacterial and archaeal LipB octanoyltransferases and resides between the BirA root and all other analyzed proteins. The second clade harbors a group of lipoate:protein ligases derived exclusively from bacteria with LipA but usually without LipS1/S2 lipoate synthase. Broad substrate range E. coli Lpl(AB) as well as the narrow substrate range ligases LplJ from B. subtilis, and Mhp-LplJ and Mhp-Lpl from M. hyopneumoniae reside in this clade. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 4. Rooted phylogenetic tree for the complete lipoate:protein ligase/octanoyltransferase dataset. The tree was rooted with the structurally related biotin ligase BirA as an outgroup. Red or blue dots placed on each leaf identify the source organisms as archaea or bacteria, respectively. The ligase/transferase type is color-coded in the next circle. In the outermost rings, the presence of other lipoate synthesis enzymes occurring in the same genome is labeled. The data underlying this figure is provided in Supplementary S3 Data. https://doi.org/10.1371/journal.pbio.3002177.g004 For the proteins of clade 3, an archaeal origin is inferred, since bipartite LplABs ligases from the archaeal phyla Thermoproteota and Thermoplasmatota including the characterized Thermoplasma acidophilum enzyme [10] are the deepest branching sequences. Three well-supported subgroups (bootstrap ≥92) branch off from these, each again with deep-branching archaeal proteins. In the first subgroup, bipartite LplABs from Nano- and Haloarchaeota form the deepest branches, which are immediately adjacent to LplABs from Burkholderiales (class Gammaproteobacteria according to GTDB). The remaining sequences in the subgroup are circularly permuted Lpl(BA) proteins, nearly exclusively stemming from Actinomycetota and including the characterized S. coelicolor Lpl(BA) [12]. This topology suggests an evolutionary history with lateral transfer of LplAB from Archaea to Gammaproteobacteria followed or accompanied by rearrangement of the gene order and final fusion of the genes upon transfer to the phylum Actinomycetota, where the gene was then vertically transmitted. The second subgroup consists of archaeal and bacterial bipartite LplABs ligases and LipM-type octanoyltransferase and provides insights into the origin of LipM: (1) Archaeal LplAB ligases, mostly encoded near lipS1/lipS2 and including the characterized Thermococcus kodakarensis protein [30], are the most deeply branching sequences and appear to be the ancestors of a large number of bacterial LipMs, which thus arose from a single interdomain horizontal gene transfer event. A scenario is supported in which archaeal LplAB lost its accessory peptide LplB, developed into LipM and simultaneously or later moved into a bacterial host. (2) In the remaining part of the subgroup, many archaeal and some bacterial LplABs ligases are mixed with many archaeal and some bacterial LipM octanoyltransferases, indicating that the described horizontal gene transfer, loss of LplB and transformation of the remaining catalytic domain LplA into an octanyoltransferase was not a singular event but happened multiple times. The third subgroup of clade 3 is made up of even further archaeal and bacterial LplABs and Lpl(AB) lipoate protein ligases, the vast majority of which originate from organisms containing LipS1/S2. The genetically and biochemically characterized sLpl(AB) ligases from proteobacterial sulfur oxidizers fall into this group. We challenged the idea that clade 3 has an origin within the archaea by calculating 3 separate trees for this group. The trees were rooted by the most closely related, similarly sized and biochemically characterized bacterial lipoate:protein ligases from clade 2 and confidence levels were increased by not including LipM octanoyltransferases and/or circular permuted cpLplBA lipoate:protein ligases (Figs 5, S4 and S5). The trees have a high to very high bootstrap support especially for the higher order splits and all 3 indeed show a root in the archaeal domain with Thermoproteota and Thermoplasmatota proteins as the deepest branching sequences. Moreover, it is clear that bacterial Lpl(AB)s, which co-occur in the same organism with LipS1/S2, originate from an archaeal ancestor. Several horizontal gene transfer events are also evident. The 2 earliest were transfer of LplAB ligases from Hadarchaeota to Synergistota and from Altiarchaeota to Chloroflexota. On the other hand, several transfers from bacteria back to archaea can be delineated, e.g., into members of the Thorarchaeota, Baldrarchaeota, Jordarchaeota, Sifarchaeota, Thermoplasmatota, and Thermoproteota. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 5. Phylogeny for clade 3 lipoate:protein ligases without LipM and cpLpl(BA). LipMs do not have LplB domains and their sequences are consequently shorter. If a sequence is incomplete, parts of the information used to calculate the phylogenetic tree are missing. This can lead to erroneous estimates of the relationships between sequences and can bias the result and weakens statistical significance of the calculation. In addition, Lpl(BA) clearly shows an individual evolution and may also cause weakening of statistical support. The data underlying this figure is provided in S3 Data. https://doi.org/10.1371/journal.pbio.3002177.g005

[END]
---
[1] Url: https://journals.plos.org/plosbiology/article?id=10.1371/journal.pbio.3002177

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.