Student or educator? Sign up with your school email and get 6 months of Pro free.
CoChat
Sign Up
Cited by 0Glycosylation and Glycoproteins ResearchCarbohydrate Chemistry and SynthesisBlood groups and transfusion

The DOI matches its Crossref record; 52 of the 52 references it cites were found in a registry or an index.

An in vitro approach for simulating divergent Golgi O-glycosylation of tumor-associated MUC1 from normal MUC1

Open access
Nashed, Abdullateef, Dilsook, Kyllen, Senapathi, Tharindu, Naidoo, Kevin J.·2026·Nature Communications
DOI 10.1038/s41467-026-72151-yPMC PMC13103387

Published version of a preprint

Open this paper in CoChat — free accountOpen PDF

Abstract

Abstract Peptide O-glycosylation relies on the coordinated action of glycosyltransferases (GTs) within the endoplasmic reticulum (ER) and Golgi apparatus. An in vitro one-pot synthetic biology approach was developed to investigate the specificity and kinetics of GT O-GalNAc glycosylation that leads to tumor antigen glycoforms of mucin 1 (MUC1). The focus is to experimentally simulate the divergent glycosylation pathways that lead to the synthesis of cancer-associated antigens (Tn, T) and their sialylated derivatives. First, the biosynthetic details of the defining first step of GALNT relocation from the ER to the Golgi was modeled using the one-pot method. Our findings reveal that an ER enriched with GALNTs results in complete Galnac (Tn) MUC1 site occupancy. This comes about as a function of two processes that are i) extended GALNT reaction time and ii) prevention of inhibition by subsequent glycosylation enzymes such as C1GALT1. The modeling confirms that B3GNT6 has negligible specificity for MUC1 Tn, explaining the absence of core 3 and core 4 structures in MUC1 in both normal and cancerous breast cell lines. Moreover, ST6GALNAC1, and not ST6GALNAC2, is primarily responsible for α-2-6 sialylation of Tn and T antigens. Computer reaction dynamic simulations combined with kinetic experimental analysis show that ST6GALNAC1 prefers fully glycosylated MUC1 and more importantly that its preference is to sialyate the S9 and T13 sites in the SAPDTR motif. This is especially the case when the MUC1 concentration is high (i.e., high-level of expression), suggesting that sTn upregulation on MUC1 in cancer is linked to the occupancy status of S9 and T13 glycosylated sites, that were previously found to be cancer-associated. The results from the one-pot synthesis approach presented here demonstrate its ability to simulate cellular glycosylation within the Golgi-ER. This systems modelling unpacks the molecular details of enzyme localization and substrate glycan occupancy that is fundamental to the regulatory mechanisms that gives rise to tumor-associated MUC1 antigens.

Overview

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

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

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

Integrity signals

Registry

Crossref · 10.1038/s41467-026-72151-y

Cites

52 of 52 references verified0 citation errors · 0 not checkable

Statistics

Figures

Fig. 1 Mucin O-GalNAc glycosylation. A O-GalNAc glycosylation pathway showing the reactions that lead to the formation of the cancer-associated Tn and T antigens and their sialylated forms sTn and sT, respectively. B Differential localization of GALNTs between normal and tumor epithelial cells is attributed to COP-1-mediated retrograde activation in cancer, shown alongside the corresponding in vitro synthesis design. Deregulation of glycosyltransferase expression is highlighted by tumor-associated upregulation of sialyltransferases. Created in BioRender. Naidoo, K. (2026) https://BioRender.com/kwimrmw.
Fig. 1 Mucin O-GalNAc glycosylation. A O-GalNAc glycosylation pathway showing the reactions that lead to the formation of the cancer-associated Tn and T antigens and their sialylated forms sTn and sT, respectively. B Differential localization of GALNTs between normal and tumor epithelial cells is attributed to COP-1-mediated retrograde activation in cancer, shown alongside the corresponding in vitro synthesis design. Deregulation of glycosyltransferase expression is highlighted by tumor-associated upregulation of sialyltransferases. Created in BioRender. Naidoo, K. (2026) https://BioRender.com/kwimrmw.CC BY-NC-ND · PubMed Central
Fig. 2 The one-pot biosynthesis method. A An AlphaFold structure of the fusion protein for peptide expression. The fusion protein design connected to the MUC1 frame was selected from the peptide tandem repeat. B Initial reaction rates for the tag-fused 27-mer MUC1 peptide, the unfused peptide, and the tag-only control as acceptors for GalNAc, catalyzed by various GALNTs. Rates were derived from the slopes of progress curves using the UGC assay. Data are presented as mean ± SD of n = 3 technical replicates, with individual data points overlaid. All values are normalized to their respective no-enzyme control reactions. Source data are provided as a Source Data file. C The pipeline of expression and enzymatic glycosylation of the MUC1 peptide used to construct the pathway model. It starts with cloning the fusion protein DNA into an expression vector, which is then used to transform E. coli. The expressed fusion protein is purified and used in a one-pot glycosylation, performed either stepwise or via multiple glycosylation reactions in a single mixture. The His-tag on the fusion protein makes it possible for a one-step IMAC purification of the glycosylated peptide. The glycosylated peptide is finally cleaved from the fusion protein by TEV protease and collected in the IMAC flowthrough. Created in BioRender. Naidoo, K. (2026) https://BioRender.com/7kjmpbt.
Fig. 2 The one-pot biosynthesis method. A An AlphaFold structure of the fusion protein for peptide expression. The fusion protein design connected to the MUC1 frame was selected from the peptide tandem repeat. B Initial reaction rates for the tag-fused 27-mer MUC1 peptide, the unfused peptide, and the tag-only control as acceptors for GalNAc, catalyzed by various GALNTs. Rates were derived from the slopes of progress curves using the UGC assay. Data are presented as mean ± SD of n = 3 technical replicates, with individual data points overlaid. All values are normalized to their respective no-enzyme control reactions. Source data are provided as a Source Data file. C The pipeline of expression and enzymatic glycosylation of the MUC1 peptide used to construct the pathway model. It starts with cloning the fusion protein DNA into an expression vector, which is then used to transform E. coli. The expressed fusion protein is purified and used in a one-pot glycosylation, performed either stepwise or via multiple glycosylation reactions in a single mixture. The His-tag on the fusion protein makes it possible for a one-step IMAC purification of the glycosylated peptide. The glycosylated peptide is finally cleaved from the fusion protein by TEV protease and collected in the IMAC flowthrough. Created in BioRender. Naidoo, K. (2026) https://BioRender.com/7kjmpbt.CC BY-NC-ND · PubMed Central
Fig. 3 O-GalNAc glycosylation pathway initiation (synthesis of Tn antigens). A Reactivity of the 23-mer and 27-mer peptides with various GALNT enzymes. Initial rates were derived from the linear range of the reaction progress curves. Data are presented as mean ± SD of n = 3 technical replicates, with individual data points overlaid. All rates are normalized to their respective no-enzyme control reactions. Attention is drawn to the only instance where GALNT4 in combination with GALNT2 exhibited significantly greater activity on the 23-mer peptide than the 27-mer peptide, through a red emboldened framed bar. Source data are provided as a Source Data file. B A schematic illustration of the direct and lectin-dependent mechanisms of the glycosylation reactions of GALNT1, GALNT2 and GALNT4 with the 23-mer and the 27-mer peptides. C In vitro synthesis of GalNAc-glycosylated 23mer: I. Brief description of synthesis designs and steps. II. Abundance of the products of each reaction as measured by liquid chromatography. Source data are provided as a Source Data file. III. Structures of the products of each reaction, as determined from the LC-MS analysis and the synthesis pathway, lead to these structures, as concluded from the preceding Data. Details of the LC-MS data and analysis are given in the Supportive Information (Supplementary Figs. 4–9). Created in BioRender. Naidoo, K. (2026) https://BioRender.com/d5imyhz.
Fig. 3 O-GalNAc glycosylation pathway initiation (synthesis of Tn antigens). A Reactivity of the 23-mer and 27-mer peptides with various GALNT enzymes. Initial rates were derived from the linear range of the reaction progress curves. Data are presented as mean ± SD of n = 3 technical replicates, with individual data points overlaid. All rates are normalized to their respective no-enzyme control reactions. Attention is drawn to the only instance where GALNT4 in combination with GALNT2 exhibited significantly greater activity on the 23-mer peptide than the 27-mer peptide, through a red emboldened framed bar. Source data are provided as a Source Data file. B A schematic illustration of the direct and lectin-dependent mechanisms of the glycosylation reactions of GALNT1, GALNT2 and GALNT4 with the 23-mer and the 27-mer peptides. C In vitro synthesis of GalNAc-glycosylated 23mer: I. Brief description of synthesis designs and steps. II. Abundance of the products of each reaction as measured by liquid chromatography. Source data are provided as a Source Data file. III. Structures of the products of each reaction, as determined from the LC-MS analysis and the synthesis pathway, lead to these structures, as concluded from the preceding Data. Details of the LC-MS data and analysis are given in the Supportive Information (Supplementary Figs. 4–9). Created in BioRender. Naidoo, K. (2026) https://BioRender.com/d5imyhz.CC BY-NC-ND · PubMed Central
Fig. 4 Assessment of the first layer of the structure extension after Tn synthesis. A Effect of GalNAc site occupancy on the specificity of ST6GALNAC1, C1GALT1 and B3GNT6. Serial dilution of concentrations of GalNAc3-23mer and GalNAc5-23mer was calculated per GalNAc site occupancy. Data is presented as mean ± SD, of n = 3 technical replicates, with individual data points overlaid. Kinetics parameters are reported as apparent values, due to the presence of multiple reactive sites on the substrates; therefore, they reflect the combined contribution of all the sites52. Vmax values are presented as a percentage relative to GalNAc3-23mer for each enzyme. Kinetics values in the table represent the best-fit parameter ± SE. Source data are provided as a Source Data file. B Core 1 Synthesis: Effect of GALNTs vs. C1GALT1 competition on site occupancy. I. Synthesis designs of different combinations of sequential synthesis reflecting C1GALT1 competition with GALNTs in isolation or in different combinations. II. shows the LC peaks and structures deduced from analyzing mass spectrometry results (Supplementary Figs. 10–14, and source data are provided as a Source Data file).** Peak vi corresponds to the incorporation of two GalNAc molecules, as indicated from the LC-MS analysis. C Sialylation of core 1 via ST3GAL1, ST6GALNAC1, and ST6GALNAC2. Data is presented as mean ± SD, of n = 3 technical replicates, with individual data points overlaid. Source data are provided as a Source Data file. For panels (A, C), statistical significance was determined using a two-tailed, two-sample equal variance Student’s t-test. Only P-values < 0.01 are displayed on the graph; comparisons not labeled were not statistically significant. Created in BioRender. Naidoo, K. (2026) https://BioRender.com/ps0u8gp.
Fig. 4 Assessment of the first layer of the structure extension after Tn synthesis. A Effect of GalNAc site occupancy on the specificity of ST6GALNAC1, C1GALT1 and B3GNT6. Serial dilution of concentrations of GalNAc3-23mer and GalNAc5-23mer was calculated per GalNAc site occupancy. Data is presented as mean ± SD, of n = 3 technical replicates, with individual data points overlaid. Kinetics parameters are reported as apparent values, due to the presence of multiple reactive sites on the substrates; therefore, they reflect the combined contribution of all the sites52. Vmax values are presented as a percentage relative to GalNAc3-23mer for each enzyme. Kinetics values in the table represent the best-fit parameter ± SE. Source data are provided as a Source Data file. B Core 1 Synthesis: Effect of GALNTs vs. C1GALT1 competition on site occupancy. I. Synthesis designs of different combinations of sequential synthesis reflecting C1GALT1 competition with GALNTs in isolation or in different combinations. II. shows the LC peaks and structures deduced from analyzing mass spectrometry results (Supplementary Figs. 10–14, and source data are provided as a Source Data file).** Peak vi corresponds to the incorporation of two GalNAc molecules, as indicated from the LC-MS analysis. C Sialylation of core 1 via ST3GAL1, ST6GALNAC1, and ST6GALNAC2. Data is presented as mean ± SD, of n = 3 technical replicates, with individual data points overlaid. Source data are provided as a Source Data file. For panels (A, C), statistical significance was determined using a two-tailed, two-sample equal variance Student’s t-test. Only P-values < 0.01 are displayed on the graph; comparisons not labeled were not statistically significant. Created in BioRender. Naidoo, K. (2026) https://BioRender.com/ps0u8gp.CC BY-NC-ND · PubMed Central
Fig. 5 Computational modeling of the reaction site specificity of ST6GALNAC1. A Proposed SN1-like reaction mechanism of ST6GALNAC1 elucidated from QM/MM reaction dynamic simulation. The position of link atoms added to prepare the system for QM/MM simulations is marked in green for Hydrogen link atoms and in red for Simple Link Atom Saccharide Hybrid (SLASH) link atoms (see section “Computational methods”). Enzyme amino acids are drawn in blue with GalNAc and Neu5Ac rings filled in yellow and purple, respectively. B 2D schematic indicating key distances and angles that influence reactivity. Average C6-C2, C6-NE2, O2P-HE2 distances, and NE2–C6–C2 angle values across the QM/MM simulation for each system can be found tabulated below. C 1D representation of the minimum energy pathway determined from the free energy surface showing the free energy associated with the bond forming and bond breaking in going from the reactants (R) to the products (P) via two transition states (TS1 & TS2) and an oxocarbenium intermediate (OC). Transition state barrier heights are summarized in a table below the 1D plots. 3D Molecular representations of key chemical species along the reaction pathway (R, TS1, OC, TS and P) are provided in Supplementary Fig. 24. Source data are provided as a Source Data file. Created in BioRender. Naidoo, K. (2026) https://BioRender.com/v8q23hm.
Fig. 5 Computational modeling of the reaction site specificity of ST6GALNAC1. A Proposed SN1-like reaction mechanism of ST6GALNAC1 elucidated from QM/MM reaction dynamic simulation. The position of link atoms added to prepare the system for QM/MM simulations is marked in green for Hydrogen link atoms and in red for Simple Link Atom Saccharide Hybrid (SLASH) link atoms (see section “Computational methods”). Enzyme amino acids are drawn in blue with GalNAc and Neu5Ac rings filled in yellow and purple, respectively. B 2D schematic indicating key distances and angles that influence reactivity. Average C6-C2, C6-NE2, O2P-HE2 distances, and NE2–C6–C2 angle values across the QM/MM simulation for each system can be found tabulated below. C 1D representation of the minimum energy pathway determined from the free energy surface showing the free energy associated with the bond forming and bond breaking in going from the reactants (R) to the products (P) via two transition states (TS1 & TS2) and an oxocarbenium intermediate (OC). Transition state barrier heights are summarized in a table below the 1D plots. 3D Molecular representations of key chemical species along the reaction pathway (R, TS1, OC, TS and P) are provided in Supplementary Fig. 24. Source data are provided as a Source Data file. Created in BioRender. Naidoo, K. (2026) https://BioRender.com/v8q23hm.CC BY-NC-ND · PubMed Central
Fig. 6 A proposed model of the effect of the colocalization of GALNTs with C1GALT1 and St6GALNAC1 on site occupancy and sialylation of the Tn antigen. The alignment of the Double Tandem Repeat (DTR)-MUC1 peptide with the Single Tandem Repeat (STR)-MUC1 23-mer (23-mer) peptide is shown in panel A. The DTR-MUC1 was expressed in fusion form using the same tag elements employed for the 23- and 27-mer peptides. A MS1 spectra obtained from peptide-related LC peaks are shown alongside the corresponding sample preparation schematic and the digestion enzyme used to release the peptide from the fusion protein (TEV protease or IMPa). Details of the LC–MS analyses are provided in the Supplementary Information (Supplementary Figs. 14–16). Sites bearing the T antigen are highlighted in orange, whereas sites carrying the sialyl-6-T antigen are indicated in red. Note: IMPa digestion was performed under mild conditions (see Supplementary Information figure captions); therefore, partially digested fragments are also observed. Because the peptide contains identical repeats, ambiguous assignments are shown at equivalent sites on both N- and C-terminal repeats. Source data are provided as a Source Data file. B Schematic representation of the proposed effect of GALNTs colocalization with C1GALT1 and ST6GALNAC1 on glycosylation site occupancy. The model shows that colocalization limits the elongation and sialylation of the Tn antigen to GALNT_D sites. In contrast, glycosylation is reduced at the S19 site or absent at GALNT_L sites. Created in BioRender. Naidoo, K. (2026) https://BioRender.com/qban85z.
Fig. 6 A proposed model of the effect of the colocalization of GALNTs with C1GALT1 and St6GALNAC1 on site occupancy and sialylation of the Tn antigen. The alignment of the Double Tandem Repeat (DTR)-MUC1 peptide with the Single Tandem Repeat (STR)-MUC1 23-mer (23-mer) peptide is shown in panel A. The DTR-MUC1 was expressed in fusion form using the same tag elements employed for the 23- and 27-mer peptides. A MS1 spectra obtained from peptide-related LC peaks are shown alongside the corresponding sample preparation schematic and the digestion enzyme used to release the peptide from the fusion protein (TEV protease or IMPa). Details of the LC–MS analyses are provided in the Supplementary Information (Supplementary Figs. 14–16). Sites bearing the T antigen are highlighted in orange, whereas sites carrying the sialyl-6-T antigen are indicated in red. Note: IMPa digestion was performed under mild conditions (see Supplementary Information figure captions); therefore, partially digested fragments are also observed. Because the peptide contains identical repeats, ambiguous assignments are shown at equivalent sites on both N- and C-terminal repeats. Source data are provided as a Source Data file. B Schematic representation of the proposed effect of GALNTs colocalization with C1GALT1 and ST6GALNAC1 on glycosylation site occupancy. The model shows that colocalization limits the elongation and sialylation of the Tn antigen to GALNT_D sites. In contrast, glycosylation is reduced at the S19 site or absent at GALNT_L sites. Created in BioRender. Naidoo, K. (2026) https://BioRender.com/qban85z.CC BY-NC-ND · PubMed Central
Fig. 7 A proposed model of the effect of the ER localization of GALNTs on site occupancy and the subsequent glycosylation of the Tn antigen. A MS1 spectra obtained from integrated peptide-related LC peaks is shown beneath the corresponding sample preparation schematic and the digestion enzyme used to release the peptide from the fusion protein (IMPa). Details of the LC–MS analyses are provided in the Supplementary Information (Supplementary Figs. 17–20). Sites bearing the Tn antigen are highlighted in purple, whereas the numbers of sites carrying the T antigen or the sialyl-6-T antigen are indicated in orange and red, respectively. When a glycosylation site is confidently assigned, it is indicated directly in the sequence. Note: IMPa digestion of C1GALT1 and ST6GALNAC1 reaction products was performed under extensive conditions (see Supplementary Information figure captions), which may have contributed to non-specific C-terminal cleavage observed in some peaks. Because the peptide contains identical repeats, ambiguous assignments are shown at equivalent sites on both N- and C-terminal repeats. Source data are provided as a Source Data file. B Schematic representation of the proposed effect of GALNT localization in the ER on glycosylation site occupancy. The model shows that ER localization enhances occupancy at all GALNT-D and GALNT_L sites. In support of the kinetic and computational analyses, the LC–MS data suggest that ST6GALNAC1 exhibits preferential activity toward the T13 site, followed by T20 and T8. Created in BioRender. Naidoo, K. (2026) https://BioRender.com/d7iikxz.
Fig. 7 A proposed model of the effect of the ER localization of GALNTs on site occupancy and the subsequent glycosylation of the Tn antigen. A MS1 spectra obtained from integrated peptide-related LC peaks is shown beneath the corresponding sample preparation schematic and the digestion enzyme used to release the peptide from the fusion protein (IMPa). Details of the LC–MS analyses are provided in the Supplementary Information (Supplementary Figs. 17–20). Sites bearing the Tn antigen are highlighted in purple, whereas the numbers of sites carrying the T antigen or the sialyl-6-T antigen are indicated in orange and red, respectively. When a glycosylation site is confidently assigned, it is indicated directly in the sequence. Note: IMPa digestion of C1GALT1 and ST6GALNAC1 reaction products was performed under extensive conditions (see Supplementary Information figure captions), which may have contributed to non-specific C-terminal cleavage observed in some peaks. Because the peptide contains identical repeats, ambiguous assignments are shown at equivalent sites on both N- and C-terminal repeats. Source data are provided as a Source Data file. B Schematic representation of the proposed effect of GALNT localization in the ER on glycosylation site occupancy. The model shows that ER localization enhances occupancy at all GALNT-D and GALNT_L sites. In support of the kinetic and computational analyses, the LC–MS data suggest that ST6GALNAC1 exhibits preferential activity toward the T13 site, followed by T20 and T8. Created in BioRender. Naidoo, K. (2026) https://BioRender.com/d7iikxz.CC BY-NC-ND · PubMed Central

Verification

Claims the paper makes

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

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

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 52 references

Research works printed but not linked: 1

  1. Thrane, P. et al. Generalizing Michaelis–Menten theory to account for substrate heterogeneity. Preprint at https://www.biorxiv.org/content/biorxiv/early/2025/08/24/2025.08.21.671619.full.pdf (2025).

Metadata

Authors
Nashed, A., Dilsook, K., Senapathi, T., & Naidoo, K. J.
Affiliations
University of Cape Town (ZA)
Corresponding author
Naidoo, Kevin J.
ORCID
Venue
Nature Communications
Publisher
Springer Science and Business Media LLC
Published
2026-04-22
Volume
17
Issue
1
ISSN
2041-1723 (ISSN-L)
DOI
10.1038/s41467-026-72151-y
PMC
PMC13103387
OpenAlex
W7155175031
Cited by
0
References
52
FWCI
0.00
Type
journal article
Language
en
Open access
Licence
Locations
  • Nature Communications (publishedVersion, cc-by-nc-nd)
  • Research Square (acceptedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by-nc-nd)
  • PubMed (publishedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
Funders
  • Dane O. Kildsig Center for Pharmaceutical Processing Research
  • National Research Foundation (CPPR 466624)
  • South African Medical Research Council
  • University of Cape Town
  • Center for High Performance Computing
  • Medical Research Council
Topics
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Glycosylation and Glycoproteins Research
  • Physical Sciences › Chemistry › Organic Chemistry › Carbohydrate Chemistry and Synthesis
  • Health Sciences › Medicine › Hematology › Blood groups and transfusion
Keywords
MUC1, tumor-associated antigens, O-glycosylation, Golgi apparatus, endoplasmic reticulum, ST6GALNAC1, Tn antigen, enzyme localization, synthetic biology
MeSH terms
beta-D-Galactoside alpha 2-6-Sialyltransferase, Endoplasmic Reticulum, Glycosylation, Golgi Apparatus, Humans, Kinetics, Neoplasms, Sialyltransferases, Antigens, Tumor-Associated, Carbohydrate, Antigens, CD, N-Acetylgalactosaminyltransferases, Mucin-1

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

How to cite

Cite CoChat for the checks on this page, and the paper for its own findings.

This page
CoChat. (2026). Citation check: An in vitro approach for simulating divergent Golgi O-glycosylation of tumor-associated MUC1 from normal MUC1. https://app.cochat.ai/papers/10.1038/s41467-026-72151-y
The paper
Nashed, A., Dilsook, K., Senapathi, T., & Naidoo, K. J. (2026). An in vitro approach for simulating divergent Golgi O-glycosylation of tumor-associated MUC1 from normal MUC1. Nature Communications. https://doi.org/10.1038/s41467-026-72151-y

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.