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



Mechanism of agonist-induced activation of the human itch receptor MRGPRX1 [1]

['Bing Gan', 'Shanghai Key Laboratory Of Metabolic Remodeling', 'Health', 'Institute Of Metabolism', 'Integrative Biology', 'Fudan University', 'Shanghai', 'The Kobilka Institute Of Innovative Drug Discovery', 'School Of Medicine', 'The Chinese University Of Hong Kong']

Date: 2023-07

Mas-related G-protein-coupled receptors X1-X4 (MRGPRX1-X4) are 4 primate-specific receptors that are recently reported to be responsible for many biological processes, including itch sensation, pain transmission, and inflammatory reactions. MRGPRX1 is the first identified human MRGPR, and its expression is restricted to primary sensory neurons. Due to its dual roles in itch and pain signaling pathways, MRGPRX1 has been regarded as a promising target for itch remission and pain inhibition. Here, we reported a cryo-electron microscopy (cryo-EM) structure of G q -coupled MRGPRX1 in complex with a synthetic agonist compound 16 in an active conformation at an overall resolution of 3.0 Å via a NanoBiT tethering strategy. Compound 16 is a new pain-relieving compound with high potency and selectivity to MRGPRX1 over other MRGPRXs and opioid receptor. MRGPRX1 was revealed to share common structural features of the G q -mediated receptor activation mechanism of MRGPRX family members, but the variable residues in orthosteric pocket of MRGPRX1 exhibit the unique agonist recognition pattern, potentially facilitating to design MRGPRX1-specific modulators. Together with receptor activation and itch behavior evaluation assays, our study provides a structural snapshot to modify therapeutic molecules for itch relieving and analgesia targeting MRGPRX1.

Funding: This work was supported by funds from National Natural Science Foundation of China project 32070525 (to Z.J.C.); Shenzhen Science and Technology Program project JCYJ20220530140800001 (to Z.J.C.). B. G. were supported by Ganghong Youth Scholarship at the Chinese University of Hong Kong, Shenzhen. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

In this study, we reported the cryo-EM structure of the active MRGPRX1-G q complex bound to compound 16 at an overall resolution of 3.0 Å. Compound 16 is a new synthetic MRGPRX1 agonist with high potency and selectivity over other MRGPRXs and opioid receptor [ 17 ]. Our complex structure reveals the conserved mechanism of small molecule-induced receptor activation among MRGPRX receptors. The structure also clearly presents a highly conserved orthosteric pocket for natural agonist recognition, such as bovine adrenal medulla 8–22 peptide (BAM8-22) [ 18 ], γ2-MSH [ 19 , 20 ], and conopeptide (CNF-Tx2) [ 21 ]. Notably, a few variable residues in orthosteric pocket of MRGPRX1 exhibit the unique agonist recognition pattern for compound 16. These findings will give us clues to the modification of small molecule scaffolds targeting MRGPRX1 specifically, potentially accelerating the development of novel drugs for the modulation of itch and pain.

Extensive studies of MRGPRX1 were conducted in itch and pain sensations, and inflammation [ 7 ]. A series of natural and synthetic agonists, antagonists, and allosteric modulators of MRGPRX1 have been developed [ 12 ]. However, there is currently no drug targeting MRGPRX1 commercialized. The structure determination of GPCR may provide the detailed molecular basis of ligand interaction to facilitate modulator development [ 13 ]. Recently, 2 groups reported the agonist-stabilized cryo-electron microscopy (cryo-EM) structures of MRGPRX2 and MRGPRX4 in complexes with trimeric G proteins [ 14 , 15 ]. The structural characteristics of orthosteric pockets and modulator specificities are examined thoroughly. The critical acidic residues D184 5.38 and E164 4.60 in MRGPRX2 and the entirely positive orthosteric pocket in MRGPRX4 mainly determine the chemical property of varies modulators. The active structures of MRGPRX1 with varies of modulators are also reported [ 16 ].

Mas-related G-protein-coupled receptors (MRGPRs) have been recently identified as pruritogenic receptors mediating the nonhistaminergic itch [ 7 ]. The Mrgpr gene family encodes MRGPRs, a large family which comprises 27 and 8 members in mice and humans, respectively [ 7 , 8 ]. MRGPRX1-X4 are 4 primate-specific receptors, suggesting that the X subfamily may be a simplified alteration in human evolution [ 4 ]. MRGPRX1 is the first identified human MRGPR that expresses in dorsal root ganglia (DRG) and trigeminal ganglia (TG) specifically [ 4 ]. Compared with other MRGPRX members, MRGPRX1 stands out for its dual roles in mediating itch [ 9 ] and inhibiting persistent pain [ 10 ]. Persistent pain is a severe health problem worldwide, and ordinary analgesics like opioids targeting opioid receptors may lead to several side effects such as drug addiction [ 10 , 11 ]. Notably, MRGPRX1 is insensitive to the classical opioid receptor antagonists, indicating that MRGPRX1 could be a new target for treating chronic pain [ 10 ].

Itch is defined as the sensation that causes the desire to scratch the skin [ 1 ]. It is a common and frequently occurring symptom associated with many skin diseases among humans [ 2 ]. Numerous factors can induce itches, such as chemicals, insect bites, and even self-generated substances resulting from varied diseases [ 3 ]. Unfortunately, due to diverse inducements and complicated pathogenesis, treating itch in the clinic is still challenging, especially the chronic itch, which will devastate people and cause much suffering [ 4 ]. The itch can be generally divided into histaminergic and nonhistaminergic [ 5 ]. Usually, most histaminergic itch results in acute itch, whereas chronic itch is more probable to be nonhistaminergic [ 6 ]. Therefore, the well-developed antihistamine drugs are inefficient in chronic itch relieving, which suggests the significance of finding novel drug targets for chronic itch treatment [ 6 ].

The coupling of G αq to MRGPRX1 is mainly maintained by interacting with residues on TM2, TM3, TM5, TM6, and ICL2 ( S9A Fig ). The interface between G αq and TMs comprises a series of hydrophobic residues, including F61 2.39 , V124 3.54 , I202 5.61 , L211 6.30 , and L214 6.33 on the receptor and L352, L356, and L361 on the α5-helix of G αq ( S9B Fig ). However, alanine substitutions of these hydrophobic residues have little effect on G αq coupling activity ( S9C Fig and S1 Table ). Only I202 5.61 A partially reduces G αq activation ( S9C Fig and S1 Table ), suggesting that the TM bundles are less critical for G αq activation in MRGPRX1. Moreover, in most class A GPCRs, ICL2 does not interact with G protein directly. However, extensive interactions of ICL2 with the αN-helix and α5-helix of G αq are observed in the MRGPRX1 structure ( S9D Fig ). Alanine substitutions of I128 ICL2 and H133 ICL2 nearly impair G αq coupling activity ( S9E and S9F Fig , S1 Table ), indicating that ICL2 plays a crucial role in G αq coupling. Similar G protein coupling interfaces are also observed in the complex structures of MRGPRX2 and MRGPRX4 previously reported ( S10A and S10B Fig ).

In addition to the unique twist structure in TM6, MRGPRX1 also shows significant differences in some classic motifs for class A GPCR activation. Firstly, the conserved D(E) 3.49 R 3.50 Y 3.51 motif on TM3 of most class A GPCRs forms an ionic lock in an inactive conformation and is broken upon activation [ 33 ]. In MRGPRX1, Y 3.51 is replaced by C121 3.51 , E119 3.49 interacts with R134 ICL2 via a strong salt bridge, and R120 3.50 interacts with T217 6.36 to limit the movement of TM6 and Y359 on G αq to stabilize the complex via polar interactions ( Fig 3E ). Secondly, the conserved motif P 5.50 I 3.40 F 6.44 [ 37 ] is substituted by an L187 5.46 L191 5.50 L110 3.40 F225 6.44 motif, constraining the conformation of TM3/5/6 ( Fig 3F ). However, the conserved N 7.49 P 7.50 XXY 7.53 motif, which does not interact directly with G proteins but is essential for receptor activation [ 33 ], is conserved in MRGPRX1 ( Fig 3G ). The above structure features are highly conserved among MRGPRXs, suggesting the common activation mechanism despite distinct agonist recognition features [ 14 , 15 ].

(A) Structural representation of TMs arrangement between TM3 and TM6. The distance between the C-terminus of TM3 and TM6 is shown as a dashed line. (B) Structural representation of the interactions near the substituted toggle switch G229 6.48 . (C) Structural superposition of active MRGPRX1, active μOR (PDB 7U2L) [ 34 ] and inactive μOR (PDB 7UL4) [ 35 ] from the side, cytoplasmic, and magnified views. The movement directions of TM6, TM7, and residues in MRGPRX1 relative to inactive μOR are highlighted as red arrows. MRGPRX1, active μOR, and inactive μOR are colored in slate, light pink, and gray, respectively. (D) BRET validation of essential residues in the extracellular half of TM6. Data are presented as mean ± SEM. n = 3; Emax, maximum effect; WT, wild type. Magnified view of D/ERY motif (E), LLLF motif (F), and NPXXY motif (G). Polar interactions are shown as yellow dashed lines. The underlying data for Fig 3D can be found in S1 Data . BRET, bioluminescence resonance energy transfer; μOR, μ opioid receptor.

The compound 16-MRGPRX1-G αq complex structure exhibited TM rearrangement in the cytoplasmic half. The cytoplasmic ends of TM3 and TM6 are about 16 angstroms apart, consistent with other class A G protein-engaged GPCRs in an active conformation [ 33 ] ( Fig 3A ). Notably, the conserved toggle switch W 6.48 in other GPCRs is replaced by G229 6.48 in MRGPRX1. This vital substitution results in an inward movement of the extracellular half of TM6, narrowing the gap between TM3 and TM6 and initiating the formation of a shallow orthosteric pocket ( Fig 3B ). Briefly, Y106 3.36 in TM3 engages with G229 6.48 in TM6 to form a twist. This twist is then stabilized by the hydrophobic interactions network among Y106 3.36 , F232 6.51 , and F237 6.56 , which prevents the ligands from entering the deeper location and exhibits a shallow pocket to accommodate ligands. Additionally, we conducted a structural comparison of MRGPRX1 complex to its functional closely related μ opioid receptor (μOR) in the active state (PDB 7U2L) [ 34 ] and inactive state (PDB 7UL4) [ 35 ] ( Fig 3C ). The structural comparison demonstrates that the MRGPRX1 complex shows a similar structure as the active μOR. Moreover, the structure superposition of G αq -coupled MRGPRX1 with G αq -coupled 5-HT 2A R (PDB 6WHA) [ 36 ] and G αq -coupled B1R (PDB 7EIB) [ 30 ] by receptors also exhibits similar conformations ( S8A and S8B Fig ), suggesting a common activation mechanism among these receptors. Significantly, except for the extracellular half of TM6, G αq -coupled MRGPRX1 shows nearly identical conformations of TM3, TM6, and TM7 with these receptors in active state (Figs 3C and S8 ). MRGPRX1 possesses several unique residues in its extracellular half of TM6, which are essential for receptor activation. Furthermore, consistent with our speculation, alanine mutations of these residues dramatically affected MRGPRX1 activation induced by compound 16 ( Fig 3D ). Hence, the initiation of MRGPRX1 activation is likely triggered by touching F237 6.56 at the bottom of the pocket upon agonist binding, pushing a series of residues in TM6 to move towards TM3 and resulting in the conformational change of G229 6.48 . G229 6.48 shifts to get close to Y106 3.36 , triggers the rotation of conserved F 6.44 , and further facilitates the intracellular half of TM6 moving outward to accommodate the downstream G protein (Figs 3C and S8 ).

The orthosteric pocket of MRGPRX1 accommodating compound 16 is composed of residues majorly located on TM3/4/5/6 ( Fig 2C ). C161 4.64 and C173 5.32 form a disulfide bond, which is conserved among MRGPRXs [ 14 ] ( S5 Fig ). Moreover, alanine substitutions of C161 4.64 and C173 5.32 nearly abolish the G q coupling activity ( Fig 2D ). Interestingly, MRGPRX1 lacks the canonical disulfide bond between TM3 and ECL2 in other class A family GPCRs [ 33 ]. Taken together, the disulfide bond substitution in MRGPRX1 may help to reorganize the extracellular loops and maintain the wide-open orthosteric pocket.

(A) Top view of the compound 16-binding pocket from the extracellular side (surface mode). Pocket is colored gray, and compound 16 is shown as cyan sticks. (B) Cut-away view of the compound 16-binding pocket. The distance between the ligand and the traditional toggle switch position is shown as dashed lines. (C) Top view of the compound 16-binding pocket from the extracellular side (cartoon mode). TMs and ECLs are colored slate. Compound 16 is shown as cyan sticks. C161 4.64 and C173 5.32 are shown as sticks and colored slate. (D) BRET validation of residues C161 4.64 and C173 5.32 . (E, F) Interaction between compound 16 and MRGPRX1 from 2 views. The key residues are shown as sticks and colored slate. The polar interactions are shown as yellow dashed lines. (G, H) BRET validation of residues in compound 16-binding pocket. Data are presented as mean ± SEM. n = 3; Emax, maximum effect; WT, wild type. The underlying data for Fig 2D, 2G and 2H can be found in S1 Data . BRET, bioluminescence resonance energy transfer.

The MRGPRX1 exhibits a shallow, broad, and wide-open ligand-binding pocket ( Fig 2A ). The distance between compound 16 and the critical toggle switch residue G229 6.48 is about 16.8 Å ( Fig 2B ), indicating that compound 16 is positioned near the extracellular surface but not buried deep in the receptor. The shallow pockets are also observed in the MRGPRX2-G αq [ 14 , 15 ] and MRGPRX4-G αq complex [ 14 ], suggesting the common pocket features among all MRGPRX receptors. Compound 16 occupies only about one-third of the pocket ( Fig 2A ) but is sufficient for the receptor activation. Similarly, the agonist (R)-ZINC-3573 and MS47134 take up only a small part of the pocket in MRGPRX2 and MRGPRX4, respectively ( S4A and S4B Fig ). The structure alignment shows that 3 agonists bind to different regions of the orthosteric pocket in 3 receptors, indicating distinct recognition mechanisms ( S4C Fig ). The electrostatic potential of the MRGPRX1 pocket is partially negative, partially positive, and partially hydrophobic ( S4D Fig ). In contrast, the electrostatic potential of the MRGPRX2 pocket is partially negative (sub-pocket 1) and partially hydrophobic (sub-pocket 2), and the electrostatic potential of MRGPRX4 pocket is positive ( S4E and S4F Fig ). These results suggest that MRGPRXs may prefer agonist scaffolds with distinct electro-properties.

(A) Cryo-EM density map of MRGPRX1-G αq in complex with compound 16. Receptor, compound 16, G αq , G β , G γ , Nb35, and scFv16 are colored slate, cyan, marine, forest, pink, salmon, and yellow-orange, respectively. The density of compound 16 is shown in mesh at the top-right corner. Compound 16 is fitted and shown as sticks. (B) Structural model for the compound 16-MRGPRX1-G αq complex. Receptor, ligand, G αq , G β , G γ , Nb35, and scFv16 are colored the same as (A). Compound 16 is shown as sticks. The 2D chemical structure of compound 16 is shown in the top-right corner. cryo-EM, cryo-electron microscopy.

The compound 16-MRGPRX1-G αq complex structure was determined by cryo-EM to yield a final map at an overall resolution of 3.0 Å (Figs 1A and S3A–S3F and S2 Table ). In the map, the densities for the receptor, G αq , G βγ , Nb35, scFv16, and compound 16 could be well distinguished, and the interface residues between MRGPRX1 and G αq (α5-helix) were clearly defined ( S3G Fig ). Thus, we built a reliable atomic model based on the well-traced α-helices and aromatic side chains ( Fig 1B ). Due to the flexibility, the N-terminus (M1-K25), part of the extracellular loops (I90) and long C-terminal residues (R279-Q322) of the receptor are invisible.

To improve receptor expression, we fused thermostabilized apocytochrome b 562 (BRIL) [ 22 ] at the N-terminus of MRGPRX1. NanoBit tethering strategy [ 23 ] was used for the complex formation, with the LgBit and HiBit fused to the C-terminus of the receptor and G β subunit, respectively ( S1A and S1B Fig ). We used bioluminescence resonance energy transfer (BRET) assay to evaluate the impact of receptor modification on G protein coupling capability. The fusion of BRIL and LgBit to receptor only marginally affected receptor activity ( S1C Fig and S1 Table ). To further stabilize the complex, we used an engineered G αq chimera in the complex assembly. The engineered G αq chimera was designed based on the mini-G αs/q71 [ 24 , 25 ] with several modifications ( S1D Fig ). Briefly, the N-terminal 1–18 residues of the mini-G αs/q71 [ 24 ] were replaced by corresponding N-terminal sequences of the human G αi1 , while the α-helical domain of G αi1 was subsequently inserted into the mini-G αs/q71 , thus providing possible binding sites for 2 antibody fragments scFv16 and Fab-G50 [ 26 , 27 ]. Additionally, 2 dominant-negative mutations (G203A and A326S) were introduced to decrease the affinity of nucleotide binding [ 28 ]. The same engineered G αq chimera had been successfully used in the structure determination of several G q -bound GPCRs, including the G q -bound ghrelin receptor [ 29 ] and bradykinin receptors [ 30 ]. The engineered G αq chimera used in the further structure study will be simplified as G αq . We co-expressed BRIL-MRGPRX1-Lgbit, G αq , and G βγ -HiBit to obtain the MRGPRX1-G αq complex. The complex was further stabilized by incubating with Nb35 [ 31 ] and scFv16 [ 32 ] in the presence of compound 16 ( S2 Fig ).

Discussion

In this study, we used the NanoBiT strategy to determine the structure of compound 16-bound MRGPRX1 in complex with G αq via cryo-EM. We compared our compound 16-MRGPRX1-G αq complex structure to the recently reported MRGPRX1-G αq complex structure contributed by Liu and colleagues [16]. The overall structures are similar (S11A Fig), but the significant difference is the orientation of the phenylmethyl group in compound 16 (S11B and S11C Fig). Due to the better ligand density in our structure, compound 16 could be accommodated well with our proposed conformation. In contrast, the ligand leaves a part of the phenyl group out of the density map when we use Liu’s structure to fit the density. Our structure reveals the common feature of shallow, broad, and wide-open orthosteric pockets in all MRGPRX members. We speculated that this shallow and broad pocket might easily accommodate various small compounds with distinct scaffolds. The less selectivity helps the receptor expand the ligand spectrum of itch sensation and facilitates the body’s quick response to the diverse exogenous stimulus.

Notably, the binding site of compound 16 is closed to TM3 and TM4 of MRGPRX1, while the binding site of (R)-ZINC-3573 is closed to TM5 and TM6 of MRGPRX2 (Figs 2C and S12A). The conformation of ECL2 in MRGPRX2 may prevent the ligand access to the corresponding position in MRGPRX1 (S12B Fig). Similarly, the binding site of MS47134 is closed to TM2 and TM3 of MRGPRX4 (S12C Fig). The inward movement of TM3, TM4, and ECL2 in MRGPRX4 may prevent the ligand access to the corresponding position in MRGPRX1 (S12D Fig). Due to the distinct binding regions and the orthosteric pocket differences of these receptors, we evaluated the activation of compound 16 on MRGPRXs. As a result, MRGPRX1 is the only receptor that can be significantly activated by compound 16 with high potency (Figs 4A and S13A–S13I). All the above further confirms that the MRGPRXs differ in ligand recognition. Together with the highly conserved G protein interfaces among MRGPRXs, it can be concluded that MRGPRXs are activated by different ligands but use a general approach to recruit G proteins. These structural differences provide clues to design agonists with improved specificity and potency.

PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 4. High selectivity of compound 16 to MRGPRX1 and its low itch risk. (A) Activation effect of compound 16 on MRGPRX1-X4. Data are presented as mean ± SEM. n = 3; Emax, maximum effect. (B) Itch behavior evaluation of compound 16. Scratching responses are induced by subcutaneous injection of vehicle (5% DMSO + 95% saline containing 20% SBE-β-CD, n = 6, 3.667 ± 1.647), compound 16 (100 μg, n = 10, 50.20 ± 13.93; 200 μg, n = 11, 44.82 ± 9.978) and CQ (200 μg, n = 9, 156.1 ± 18.24) in WT mice. Each dot represents an individual mouse. All data are presented as mean ± SEM. ns, not significant, P > 0.5; *, P < 0.05; ***, P < 0.001. (C) Representative calcium traces of MrgprA3 responding to compound 16 (10 μM). CQ (1 mM) was used as the positive control. (D) Representative calcium traces of MrgprC11 responding to compound 16 (10 μM). BAM8-22 (20 μM) was used as the positive control. The underlying data for Fig 4A–4D can be found in S1 Data. CQ, chloroquine; WT, wild type. https://doi.org/10.1371/journal.pbio.3001975.g004

Moreover, MRGPRX1 reportedly involves in itch sensation [9] and pain inhibition [10]. Chloroquine (CQ), a drug widely used in malaria treatment, can cause itch in some people [38–40]. It is recently reported that MRGPRX1 mediates CQ-induced itch in humans [40]. Compound 16 was designed to inhibit chronic pain by targeting MRGPRX1. It is more abundant in the spinal cord than in the circulatory system, suggesting a lower risk of side effects caused by unexpected activation of MRGPRX1 [17]. Given its high potency, high selectivity, and restricted distribution, compound 16 is a viable candidate drug worthy of more attention and further study [17]. Accordingly, we tested the severity of itching that compound 16 might induce to evaluate whether the itch side effects would limit its application. Here, we used the scratching responses on the mouse model to evaluate the itch severity of compound 16 and CQ. Resultedly, compound 16 induces much less itch than a similar quantity of CQ (200 μg) (Fig 4B). To further investigate why compound 16 behaved differently than CQ, we tested the activation effect of compound 16 on MrgprA3 and MrgprC11, both the mouse orthologs of human MRGPRX1. Notably, MrgprA3 was found to be the main receptor mediating CQ-evoked responses in mice [40]. Our results showed that compound 16 could not activate MrgprA3 or MrgprC11 even at the concentration of 500 μM (Figs 4C and 4D and S13L and S13M). In other words, compound 16 failed to act as an agonist of mouse MrgprA3 and MrgprC11 but is a specific human MRGPRX1 agonist. Conversely, CQ shows higher potency to MrgprA3 (EC 50 : 27.55 μM) than MRGPRX1 (EC 50 : 297.68 μM) [40].

Determination of high-resolution CQ-bound MRGPRX1 and MrgprA3 structures may help us decipher the molecular basis of ligand recognition specificity. However, it is challenging to get a CQ-bound MRGPRX1 structure due to its low efficacy. Here, we performed molecular docking to analyze the recognition difference between MRGPRX1 and MrgprA3 for CQ (S14A and S14B Fig). The model of activated MrgprA3 was generated using MRGPRX1 as a template. Results show that CQ occupies a position away from TM6 in the binding pocket of MRGPRX1 compared with compound 16 in the MRGPRX1-compound 16 complex structure. It shows that the interactions between CQ and critical hydrophobic residues (F2366.55, F2376.56, and L2406.59) in TM6 are weaker than in compound 16. We speculate that CQ weakly activates MRGPRX1 due to its weak interactions with critical hydrophobic residues in TM6. Interestingly, the 7-chloroquinolin group of CQ in the pocket of MrgprA3 gets closer to TM6 when compared with that in MRGPRX1 structure. It is consistent with our speculation that stronger interactions with critical hydrophobic residues in TM6 will improve the ligand-binding affinity. Meanwhile, we also noticed several amino acids in the ligand-binding pocket that are unconserved between MRGPRX1 and MrgprA3. This may induce the different binding pose and affinity for compound 16 and CQ between MRGPRX1 and MrgprA3. Y993.29 in MRGPRX1, critical for compound 16 binding, is replaced by a histidine in MrgprA3, which may affect the activation of MrgprA3 by compound 16.

Furthermore, studies on the downstream effectors of MRGPRX1, especially transient receptor potential vanilloid 1 (TRPV1) and transient receptor potential ankyrin 1 (TRPA1), show some conflicts. TRPV1 is usually regarded as an ion channel involved in pain sensation [41]. Wilson and colleagues found that TRPA1 is required for CQ-induced mice itch mediated by MrgprA3 [42]. The tick salivary peptide IPDef1 has been reported to evoke mice itch via MrgprC11 and result in the activation of the downstream ion channel TRPV1 rather than TRPA1 [43]. These data are consistent with the overlap between itch-sensing pathways and pain-sensing pathways. Thus, the differences in downstream signaling of MRGPRX1-mediated pain and itch sensations still need further investigation.

[END]
---
[1] Url: https://journals.plos.org/plosbiology/article?id=10.1371/journal.pbio.3001975

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.