CoChat

A Structural Model for Binding of the Serine-Rich Repeat Adhesin GspB to Host Carbohydrate Receptors

Open access

The DOI matches its Crossref record; 75 of the 75 references it cites were found in a registry or an index.

Abstract

GspB is a serine-rich repeat (SRR) adhesin of Streptococcus gordonii that mediates binding of this organism to human platelets via its interaction with sialyl-T antigen on the receptor GPIbα. This interaction appears to be a major virulence determinant in the pathogenesis of infective endocarditis. To address the mechanism by which GspB recognizes its carbohydrate ligand, we determined the high-resolution x-ray crystal structure of the GspB binding region (GspB(BR)), both alone and in complex with a disaccharide precursor to sialyl-T antigen. Analysis of the GspB(BR) structure revealed that it is comprised of three independently folded subdomains or modules: 1) an Ig-fold resembling a CnaA domain from prokaryotic pathogens; 2) a second Ig-fold resembling the binding region of mammalian Siglecs; 3) a subdomain of unique fold. The disaccharide was found to bind in a pocket within the Siglec subdomain, but at a site distinct from that observed in mammalian Siglecs. Confirming the biological relevance of this binding pocket, we produced three isogenic variants of S. gordonii, each containing a single point mutation of a residue lining this binding pocket. These variants have reduced binding to carbohydrates of GPIbα. Further examination of purified GspB(BR)-R484E showed reduced binding to sialyl-T antigen while S. gordonii harboring this mutation did not efficiently bind platelets and showed a significant reduction in virulence, as measured by an animal model of endocarditis. Analysis of other SRR proteins revealed that the predicted binding regions of these adhesins also had a modular organization, with those known to bind carbohydrate receptors having modules homologous to the Siglec and Unique subdomains of GspB(BR). This suggests that the binding specificity of the SRR family of adhesins is determined by the type and organization of discrete modules within the binding domains, which may affect the tropism of organisms for different tissues.

Overview

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1371/journal.ppat.1002112

Cites

75 of 75 references verified0 citation errors · 0 not checkable

Statistics

Figures

Figure 1 Schematic of the SRR adhesins. Selected SRR adhesins are aligned based on the N-terminus of their binding regions, with the largest binding region at the top and the smallest binding region at the bottom. The length of each domain and the entire protein is drawn to-scale. SP = signal peptide (dark gray), SRR1 = first serine rich repeat (light gray), BR = binding region (red), SRR2 = second serine rich repeat (dark gray), CWA = cell wall anchoring motif (light gray). The SRR adhesins used in this figure are S. aureus SraP (SraP); S. epidermidis SRR1 (seSRR), S. agalactiae SRR1 (SRR1GBS), S. gordonii strain M99 GspB (GspB), S. parasanguinis Fap1 (Fap1), S. pneumoniae PsrP (PsrP), S. gordonii strain Challis Hsa (Hsa), and S. sanguinis SrpA (SrpA).
Figure 1 Schematic of the SRR adhesins. Selected SRR adhesins are aligned based on the N-terminus of their binding regions, with the largest binding region at the top and the smallest binding region at the bottom. The length of each domain and the entire protein is drawn to-scale. SP = signal peptide (dark gray), SRR1 = first serine rich repeat (light gray), BR = binding region (red), SRR2 = second serine rich repeat (dark gray), CWA = cell wall anchoring motif (light gray). The SRR adhesins used in this figure are S. aureus SraP (SraP); S. epidermidis SRR1 (seSRR), S. agalactiae SRR1 (SRR1GBS), S. gordonii strain M99 GspB (GspB), S. parasanguinis Fap1 (Fap1), S. pneumoniae PsrP (PsrP), S. gordonii strain Challis Hsa (Hsa), and S. sanguinis SrpA (SrpA).CC BY · PubMed Central
Figure 2 Structure of GspBBR. A. Two views of the structure of GspBBR separated by a 90° rotation. The N-terminus of the binding region is located at the top of the figure and the C-terminus is located at the bottom. β-Strands are colored magenta, α-helices are colored blue, and loops are colored black. A cation bound within the Siglec domain is shown as a purple sphere. B. Representative experimental electron density (blue mesh) at 2.0 Å resolution, contoured at 1.0 σ, and depicted superpositioned onto residues 515–520 of the final model. C. A m|Fo|−d|Fc| omit electron density map at 1.4 Å resolution (green mesh) calculated in REFMAC5 [65] after the removal of the model contoured at 3.0 σ and depicted superpositioned onto residues 515–520 of the final model. D. Folding diagram of the Unique subdomain. β-Strands are colored magenta and the α-helix is colored blue.
Figure 2 Structure of GspBBR. A. Two views of the structure of GspBBR separated by a 90° rotation. The N-terminus of the binding region is located at the top of the figure and the C-terminus is located at the bottom. β-Strands are colored magenta, α-helices are colored blue, and loops are colored black. A cation bound within the Siglec domain is shown as a purple sphere. B. Representative experimental electron density (blue mesh) at 2.0 Å resolution, contoured at 1.0 σ, and depicted superpositioned onto residues 515–520 of the final model. C. A m|Fo|−d|Fc| omit electron density map at 1.4 Å resolution (green mesh) calculated in REFMAC5 [65] after the removal of the model contoured at 3.0 σ and depicted superpositioned onto residues 515–520 of the final model. D. Folding diagram of the Unique subdomain. β-Strands are colored magenta and the α-helix is colored blue.CC BY · PubMed Central
Figure 3 Ig-fold topologies. A. Topology diagram of a C-set Ig-fold. B. Topology diagram of the DE-variant of a C-set Ig-fold identified in MSCRAMMs [36]. C. Topology diagram of the GspBBR CnaA subdomain. D. Topology diagram of a V-set Ig-fold. E. Topology diagram of a Siglec [45] showing the binding location of the sialylated carbohydrate in pink. F. Topology diagram of the GspBBR Siglec subdomain showing the binding location of the sialyl-T antigen in pink.
Figure 3 Ig-fold topologies. A. Topology diagram of a C-set Ig-fold. B. Topology diagram of the DE-variant of a C-set Ig-fold identified in MSCRAMMs [36]. C. Topology diagram of the GspBBR CnaA subdomain. D. Topology diagram of a V-set Ig-fold. E. Topology diagram of a Siglec [45] showing the binding location of the sialylated carbohydrate in pink. F. Topology diagram of the GspBBR Siglec subdomain showing the binding location of the sialyl-T antigen in pink.CC BY · PubMed Central
Figure 4 Identification of the receptor binding site. A. Inhibition of S. gordonii strain M99 binding to immobilized glycocalicin by the α-2,3-sialyl (1-thioethyl)galactose disaccharide. Binding was assessed in the absence or presence of 44 mM disaccharide, and is expressed as the percent of input bacteria that remained bound to glycocalicin after repeated washing of the microtiter wells (mean ± standard deviation). B. Stereo view of the disaccharide binding site superpositioned with m|Fo|−d|Fc| omit electron density (green mesh) calculated in REFMAC5 [65] and contoured to 2.5 σ after the removal of disaccharide from the model. C. A space filling model of GspBBR with bound disaccharide (stick model) demonstrates that the carbohydrate receptor binds within a pocket on the surface of the protein. The protein surface is colored by subdomain with the Siglec subdomain in yellow and the Unique subdomain in blue. The stick model of the disaccharide has carbon atoms colored magenta, oxygen atoms colored red, the nitrogen atom colored blue, and the sulfur atom colored yellow. The model is rotated 50° along the y-axis and 40° along the x-axis as compared to panel B. D. Overview of the GspBBR structure highlighting the location of the disaccharide binding site in the Siglec subdomain. The model is rotated 90° along the x-axis as compared to panel B. The disaccharide is shown with magenta bonds and the cation is highlighted as a purple sphere.
Figure 4 Identification of the receptor binding site. A. Inhibition of S. gordonii strain M99 binding to immobilized glycocalicin by the α-2,3-sialyl (1-thioethyl)galactose disaccharide. Binding was assessed in the absence or presence of 44 mM disaccharide, and is expressed as the percent of input bacteria that remained bound to glycocalicin after repeated washing of the microtiter wells (mean ± standard deviation). B. Stereo view of the disaccharide binding site superpositioned with m|Fo|−d|Fc| omit electron density (green mesh) calculated in REFMAC5 [65] and contoured to 2.5 σ after the removal of disaccharide from the model. C. A space filling model of GspBBR with bound disaccharide (stick model) demonstrates that the carbohydrate receptor binds within a pocket on the surface of the protein. The protein surface is colored by subdomain with the Siglec subdomain in yellow and the Unique subdomain in blue. The stick model of the disaccharide has carbon atoms colored magenta, oxygen atoms colored red, the nitrogen atom colored blue, and the sulfur atom colored yellow. The model is rotated 50° along the y-axis and 40° along the x-axis as compared to panel B. D. Overview of the GspBBR structure highlighting the location of the disaccharide binding site in the Siglec subdomain. The model is rotated 90° along the x-axis as compared to panel B. The disaccharide is shown with magenta bonds and the cation is highlighted as a purple sphere.CC BY · PubMed Central
Figure 5 Comparison of carbohydrate binding in the Siglec subdomain of GspBBR and Siglec-5. A. Overlay of the Siglec subdomain of GspBBR and Siglec-5. Siglec-5 is colored teal with the carbons of 3′ sialyllactose colored purple. The Siglec subdomain of GspBBR is colored yellow with the carbons of α-2,3-sialyl (1-thioethyl)galactose colored in magenta. In this panel, the Siglec carbohydrate binding site is occupied by a helix in the GspBBR Siglec subdomain therefore eliminating the possibility that this could be a second binding site for GspBBR. B & C. Comparison of the location and details of the carbohydrate binding site in (B) GspBBR with that of (C) Siglec-5 (PDB entry 2ZG3; [45]). The Siglec subdomain of GspBBR is colored yellow with the carbons of α-2,3-sialyl (1-thioethyl)galactose colored in magenta and Siglec-5 is colored teal with the carbons of 3′ sialyllactose colored purple. While Siglec-5 is used as an example, it is noted that for all structurally characterized Siglecs (Siglec-1 (aka sialoadhesin); PDB entry 1QFO; [44]), Siglec-5 (2ZG3; [45]), and Siglec-7 (PDB entry 2HRL; [74])), the sialic acid binding pocket is found at the same location.
Figure 5 Comparison of carbohydrate binding in the Siglec subdomain of GspBBR and Siglec-5. A. Overlay of the Siglec subdomain of GspBBR and Siglec-5. Siglec-5 is colored teal with the carbons of 3′ sialyllactose colored purple. The Siglec subdomain of GspBBR is colored yellow with the carbons of α-2,3-sialyl (1-thioethyl)galactose colored in magenta. In this panel, the Siglec carbohydrate binding site is occupied by a helix in the GspBBR Siglec subdomain therefore eliminating the possibility that this could be a second binding site for GspBBR. B & C. Comparison of the location and details of the carbohydrate binding site in (B) GspBBR with that of (C) Siglec-5 (PDB entry 2ZG3; [45]). The Siglec subdomain of GspBBR is colored yellow with the carbons of α-2,3-sialyl (1-thioethyl)galactose colored in magenta and Siglec-5 is colored teal with the carbons of 3′ sialyllactose colored purple. While Siglec-5 is used as an example, it is noted that for all structurally characterized Siglecs (Siglec-1 (aka sialoadhesin); PDB entry 1QFO; [44]), Siglec-5 (2ZG3; [45]), and Siglec-7 (PDB entry 2HRL; [74])), the sialic acid binding pocket is found at the same location.CC BY · PubMed Central
Figure 6 Interdomain Angle. Differences in interdomain angles between structures of GspBBR. The structures of the Siglec and Unique subdomains of unbound and disaccharide-bound GspBBR were aligned using Dyndom [69]. The unbound structure is colored in silver, and the structure in complex with α-2,3-sialyl (1-thioethyl) galactose is colored by subdomain with the CnaA subdomain in red, the Siglec subdomain in yellow and the Unique subdomain colored in blue, and cation colored purple. The side chains of hinge residues Lys398, Asp399, and Thr400 are highlighted in the boxed inset. The difference in interdomain angle between the two structures is a 40°. A movie morphing these states is seen in Video S1 .
Figure 6 Interdomain Angle. Differences in interdomain angles between structures of GspBBR. The structures of the Siglec and Unique subdomains of unbound and disaccharide-bound GspBBR were aligned using Dyndom [69]. The unbound structure is colored in silver, and the structure in complex with α-2,3-sialyl (1-thioethyl) galactose is colored by subdomain with the CnaA subdomain in red, the Siglec subdomain in yellow and the Unique subdomain colored in blue, and cation colored purple. The side chains of hinge residues Lys398, Asp399, and Thr400 are highlighted in the boxed inset. The difference in interdomain angle between the two structures is a 40°. A movie morphing these states is seen in Video S1 .CC BY · PubMed Central
Figure 7 Amino acid substitutions in the carbohydrate binding site of GspBBR. A. Expression of GspB and variant proteins on the S. gordonii cell surface. Peptidoglycan-linked proteins were extracted from the bacterial cell surface and analyzed by western blotting using a polyclonal anti-GspB serum (upper panel). The nitrocellulose membrane was then re-probed using a polyclonal anti-GspA serum (lower panel) in order to assess the relative efficiency of protein extraction and protein loading (GspA is another LPXTG-linked protein that is expressed by S. gordonii M99). B. Binding of S. gordonii strain M99 and derivative strains to glycocalicin immobilized in microtiter wells. Binding is expressed as the percent of input bacteria that remained bound to glycocalicin after repeated washing of the wells (mean ± standard deviation). The labels of each column indicate the sequence of the protein expressed in S. gordonii, with M99 indicating wild type, ΔgspB indicating a strain where the gene encoding GspB has been deleted, and remaining columns indicating the amino acid substitution present in the protein. E401 is a control residue not located near the binding pocket. C. Binding of sialyl-T antigen to purified GST-GspBBR. The indicated amounts of GST or GST-GspBBR wild-type (wt) and variant (R484E) proteins were immobilized in microtiter wells, and the binding of biotinylated sialyl-T antigen to each protein was detected by using peroxidase-conjugated streptavidin along with a chromogenic peroxidase substrate. Binding is expressed as the mean ± standard deviation (n = 3). The o represents purified GST-GspBBR-R484E, the x represents purified GST, and the represents purified GST-GspBBR. D. Binding of M99 and derivative strains to human platelets. Binding is expressed as the percent of input bacteria that remained bound to platelets after repeated washing of the wells (mean ± standard deviation).
Figure 7 Amino acid substitutions in the carbohydrate binding site of GspBBR. A. Expression of GspB and variant proteins on the S. gordonii cell surface. Peptidoglycan-linked proteins were extracted from the bacterial cell surface and analyzed by western blotting using a polyclonal anti-GspB serum (upper panel). The nitrocellulose membrane was then re-probed using a polyclonal anti-GspA serum (lower panel) in order to assess the relative efficiency of protein extraction and protein loading (GspA is another LPXTG-linked protein that is expressed by S. gordonii M99). B. Binding of S. gordonii strain M99 and derivative strains to glycocalicin immobilized in microtiter wells. Binding is expressed as the percent of input bacteria that remained bound to glycocalicin after repeated washing of the wells (mean ± standard deviation). The labels of each column indicate the sequence of the protein expressed in S. gordonii, with M99 indicating wild type, ΔgspB indicating a strain where the gene encoding GspB has been deleted, and remaining columns indicating the amino acid substitution present in the protein. E401 is a control residue not located near the binding pocket. C. Binding of sialyl-T antigen to purified GST-GspBBR. The indicated amounts of GST or GST-GspBBR wild-type (wt) and variant (R484E) proteins were immobilized in microtiter wells, and the binding of biotinylated sialyl-T antigen to each protein was detected by using peroxidase-conjugated streptavidin along with a chromogenic peroxidase substrate. Binding is expressed as the mean ± standard deviation (n = 3). The o represents purified GST-GspBBR-R484E, the x represents purified GST, and the represents purified GST-GspBBR. D. Binding of M99 and derivative strains to human platelets. Binding is expressed as the percent of input bacteria that remained bound to platelets after repeated washing of the wells (mean ± standard deviation).CC BY · PubMed Central
Figure 8 Binding of wild type S. gordonii strain M99, PS2116 (strain M99 with gspb harboring the R484E substitution), and PS846 (ΔgspB) to platelets. Fixed platelets were adhered to poly-L-lysine coated cover slips, quenched, and bound to live bacteria stained with DAPI. A–C. Fluorescent images of each strain with DAPI stain. D–F. Merge between the DIC image and the DAPI image to show relative number of platelets.
Figure 8 Binding of wild type S. gordonii strain M99, PS2116 (strain M99 with gspb harboring the R484E substitution), and PS846 (ΔgspB) to platelets. Fixed platelets were adhered to poly-L-lysine coated cover slips, quenched, and bound to live bacteria stained with DAPI. A–C. Fluorescent images of each strain with DAPI stain. D–F. Merge between the DIC image and the DAPI image to show relative number of platelets.CC BY · PubMed Central
Figure 9 Modular organization of the binding regions within adhesins of the SRR superfamily. A. A summary of BLAST [50] or ClustalW [51] sequence alignments of the binding regions of SRR adhesins. The first number indicates sequence identity and the second indicates sequence similarity. GspBBR has been divided into its subdomains, with GspBBR-C indicating the CnaA subdomain and GspBBR-SU indicating both the Siglec and Unique subdomains. Only the C-terminal subdomain of SRR1GBS (SRR1GBS-β) and Fap1NR (Fap1NR-β) are used in this analysis since the N-terminal domains do not share detectable sequence identity with other sequences used in this analysis. Boxes in grey indicate SRR pairs that do not have detectable sequence identity within the binding region. B. A schematic summarizing (A) The colored rectangles represent similar regions of sequence, and are drawn to scale. If known, representative structures of each domain are illustrated along the top, and binding partners are indicated along the side.
Figure 9 Modular organization of the binding regions within adhesins of the SRR superfamily. A. A summary of BLAST [50] or ClustalW [51] sequence alignments of the binding regions of SRR adhesins. The first number indicates sequence identity and the second indicates sequence similarity. GspBBR has been divided into its subdomains, with GspBBR-C indicating the CnaA subdomain and GspBBR-SU indicating both the Siglec and Unique subdomains. Only the C-terminal subdomain of SRR1GBS (SRR1GBS-β) and Fap1NR (Fap1NR-β) are used in this analysis since the N-terminal domains do not share detectable sequence identity with other sequences used in this analysis. Boxes in grey indicate SRR pairs that do not have detectable sequence identity within the binding region. B. A schematic summarizing (A) The colored rectangles represent similar regions of sequence, and are drawn to scale. If known, representative structures of each domain are illustrated along the top, and binding partners are indicated along the side.CC BY · PubMed Central

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Citations

Cites

Show all 75 references

Research works printed but not linked: 1

  1. S Bailey, 1994, The Ccp4 Suite - Programs for Protein Crystallography., Acta Crystallogr D Biol Crystallogr, 50, 760, 10.1107/S0907444994003112 · DOI · Paper page

Cited by

Metadata

Authors
Pyburn, T. M., Bensing, B. A., Xiong, Y. Q., Melancon, B. J., Tomasiak, T. M., Ward, N. J., Yankovskaya, V., Oliver, K. M., Cecchini, G., Sulikowski, G. A., Tyska, M. J., Sullam, P. M., & Iverson, T. M.
Affiliations
  • Vanderbilt University (US)
  • San Francisco VA Medical Center (US)
  • University of California, Los Angeles (US)
  • Vanderbilt University Medical Center (US)
Corresponding author
Iverson, T. M.
Venue
PLoS Pathogens
Published
2011-07-07
Volume
7
Issue
7
Pages
e1002112
ISSN
  • 1553-7366 (ISSN-L)
  • 1553-7374
OpenAlex
W2067583120
Cited by
84
References
75
Citations by year
2026: 3 · 2025: 2 · 2024: 5 · 2023: 2 · 2022: 3 · 2021: 4 · 2020: 3 · 2019: 10 · 2018: 8 · 2017: 2 · 2016: 11 · 2015: 6 · 2014: 7 · 2013: 10 · 2012: 7
FWCI
3.23
Type
journal article
Language
en
Open access
Locations
  • PLoS Pathogens (publishedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
  • Figshare (submittedVersion, cc-by)
Funders
  • U.S. Department of Energy (DE‐ AC02‐06CH11357, Contract No. DE-AC02-06CH11357., AC02‐06CH11357, 085P1000817, 06CH11357, No. DE-AC02-06CH11357, DE-AC02-)
  • U.S. Department of Veterans Affairs (I01 BX001077)
  • American Heart Association (09GRNT2310188, UL1 RR024975)
  • Michigan Economic Development Corporation (085P1000817, DE-AC02-06CH11357, Grant 085P1000817)
  • Vanderbilt University (UL1 RR024975)
  • Vanderbilt Institute for Clinical and Translational Research (RR024975, UL1 RR024975)
  • Georgia Clinical and Translational Science Alliance (UL1 RR024975, RR024975)
  • National Institutes of Health (DE-AC02–06CH11357, 085P1000817, RR024975)
  • Office of Science (Contract No. DE-AC02-06CH11357, AC02‐06CH11357, DE-AC02, No. DE-AC02-06CH11357., 085P1000817, 06CH11357, DE-AC02-06CH11357)
  • Michigan Technology Tri-Corridor (DE-AC02-06CH11357, 085P1000817)
  • National Institute on Drug Abuse (T90DA022873)
  • National Institute of General Medical Sciences (T32GM065086, GM079419, GM61606, R01GM061606, GM65086, DE-AC02-06CH11357, R01GM079419)
  • National Institute of Allergy and Infectious Diseases (AI057433, AI079558, R21AI079558, AI041513, R01AI057433, R01AI041513)
  • National Institute of Diabetes and Digestive and Kidney Diseases (R01DK075555, DK075555)
  • National Center for Research Resources (RR024975, UL1RR024975)
  • Basic Energy Sciences (DE-AC02, DE-AC- 02-06CH11357, AC02‐06CH11357, Contract No. DE-AC02-06CH11357, 06CH11357)
  • Biological and Environmental Research (DEAC0206CH11357)
  • Vanderbilt Institute of Chemical Biology, School of Medicine, Vanderbilt University
Topics
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Biochemical and Structural Characterization
  • Health Sciences › Medicine › Cardiology and Cardiovascular Medicine › Infective Endocarditis Diagnosis and Management
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Glycosylation and Glycoproteins Research
Keywords
Streptococcus gordonii, sialyl-T antigen, GPIbα, disaccharide binding, infective endocarditis, platelet adhesion, virulence, binding specificity
MeSH terms
Animals, Bacterial Proteins, Binding Sites, Blood Platelets, Endocarditis, Bacterial, Female, Humans, Microscopy, Fluorescence, Mucins, Protein Binding, Serine, Mutagenesis, Site-Directed, Rats, Sprague-Dawley, Point Mutation, Sequence Analysis, DNA, Protein Structure, Secondary, Adhesins, Bacterial, Platelet Glycoprotein GPIb-IX Complex, Lectins, Rats, Streptococcus gordonii, Sialic Acid Binding Immunoglobulin-like Lectins

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.