CoChat

Real-time in vivo imaging of p16Ink4a reveals cross talk with p53

Open access

The DOI matches its Crossref record; 68 of the 71 references it cites were found in a registry or an index.

Abstract

Expression of the p16Ink4a tumor suppressor gene, a sensor of oncogenic stress, is up-regulated by a variety of potentially oncogenic stimuli in cultured primary cells. However, because p16Ink4a expression is also induced by tissue culture stress, physiological mechanisms regulating p16Ink4a expression remain unclear. To eliminate any potential problems arising from tissue culture–imposed stress, we used bioluminescence imaging for noninvasive and real-time analysis of p16Ink4a expression under various physiological conditions in living mice. In this study, we show that oncogenic insults such as ras activation provoke epigenetic derepression of p16Ink4a expression through reduction of DNMT1 (DNA methyl transferase 1) levels as a DNA damage response in vivo. This pathway is accelerated in the absence of p53, indicating that p53 normally holds the p16Ink4a response in check. These results unveil a backup tumor suppressor role for p16Ink4a in the event of p53 inactivation, expanding our understanding of how p16Ink4a expression is regulated in vivo.

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 · 10.1083/jcb.200904105

Cites

68 of 68 references verified0 citation errors · 0 not checkable

Statistics

Figures

Figure 1. Generation of human p16Ink4a reporter mice. (A) A large genomic DNA segment (195.4 kb) of human chromosome that contains the entire INK4a/ARF gene locus and surrounding sequences, including a putative DNA replication origin (RD) known to regulate p16Ink4a gene expression (Gonzalez et al., 2006), was engineered to express luciferase-tagged p16Ink4a. (B) BAC vector containing p16-luc DNA or empty BAC vector was transfected into 293T cells. Expression of the p16Ink4a-luciferase fusion protein was analyzed by Western blotting after selection with antibiotics. (C) BAC vector containing p16-luc DNA was introduced into 293T cells with or without BMI-1 expression plasmid along with 0.2 mg of MMLV (Moloney murine leukaemia virus)-lacZ plasmid. Luciferase activities were normalized by lacZ activities. Error bars indicate SD. (D) The 6-wk-old p16-luc mice were subjected to noninvasive BLI. Representative images of five different experiments are shown (left). The same mice were incised through the mouth and anus under anesthesia. Representative BLI data of five different experiments are shown. The color bar indicates photons with minimum and maximum threshold values. (E) Bioluminescence intensity emitted from the organs was graphed (log10 scale). The mean ± SD of five independent experiments is shown. (F) The levels of exogenous (human) p16Ink4a gene expression and endogenous (mouse) p16Ink4a gene expression in p16-luc mice were analyzed by semiquantitative RT-PCR. β-Actin was used as a loading control. Representative data of five different experiments are shown.
Figure 1. Generation of human p16Ink4a reporter mice. (A) A large genomic DNA segment (195.4 kb) of human chromosome that contains the entire INK4a/ARF gene locus and surrounding sequences, including a putative DNA replication origin (RD) known to regulate p16Ink4a gene expression (Gonzalez et al., 2006), was engineered to express luciferase-tagged p16Ink4a. (B) BAC vector containing p16-luc DNA or empty BAC vector was transfected into 293T cells. Expression of the p16Ink4a-luciferase fusion protein was analyzed by Western blotting after selection with antibiotics. (C) BAC vector containing p16-luc DNA was introduced into 293T cells with or without BMI-1 expression plasmid along with 0.2 mg of MMLV (Moloney murine leukaemia virus)-lacZ plasmid. Luciferase activities were normalized by lacZ activities. Error bars indicate SD. (D) The 6-wk-old p16-luc mice were subjected to noninvasive BLI. Representative images of five different experiments are shown (left). The same mice were incised through the mouth and anus under anesthesia. Representative BLI data of five different experiments are shown. The color bar indicates photons with minimum and maximum threshold values. (E) Bioluminescence intensity emitted from the organs was graphed (log10 scale). The mean ± SD of five independent experiments is shown. (F) The levels of exogenous (human) p16Ink4a gene expression and endogenous (mouse) p16Ink4a gene expression in p16-luc mice were analyzed by semiquantitative RT-PCR. β-Actin was used as a loading control. Representative data of five different experiments are shown.CC BY-NC-SA · PubMed Central
Figure 2. Induction of human p16Ink4a gene expression during the onset of cellular senescence in cultured MEFs. (A and B) Primary MEFs derived from the p16-luc mice were rendered senescent by either serial passage (A) or oncogenic Ras expression (retrovirus infection; B). The levels of endogenous p16Ink4a expression and luciferase activity were measured. Representative data of three independent experiments are shown. As confirmation of senescence, representative photographs of the cells stained for SA β-gal activity are shown. β-Actin was used as a loading control. The means ± SD of three independent experiments are shown. WB, Western blot.
Figure 2. Induction of human p16Ink4a gene expression during the onset of cellular senescence in cultured MEFs. (A and B) Primary MEFs derived from the p16-luc mice were rendered senescent by either serial passage (A) or oncogenic Ras expression (retrovirus infection; B). The levels of endogenous p16Ink4a expression and luciferase activity were measured. Representative data of three independent experiments are shown. As confirmation of senescence, representative photographs of the cells stained for SA β-gal activity are shown. β-Actin was used as a loading control. The means ± SD of three independent experiments are shown. WB, Western blot.CC BY-NC-SA · PubMed Central
Figure 3. The dynamics of p16Ink4a gene expression during the aging process in vivo. (A) The same p16-luc mice were subjected to noninvasive BLI every 2 wk throughout their entire life span. Representative images of 10 independent mice are shown. (B) The p16-luc mice (young and old) were subjected to noninvasive BLI. The same mice were then incised through mouth and anus under anesthesia. (A and B) The color bars indicate photons with minimum and maximum threshold values. (C) Bioluminescence intensity emitted from the organs from young (1.5 mo) and old (22.5 mo) mice were graphed (log10 scale). The mean ± SD of five independent experiments is shown. (D) The levels of exogenous (human) p16Ink4a gene expression and endogenous (mouse) p16Ink4a gene expression from young (Y; 1.5 mo) and old (O; 22.5 mo) mice were analyzed by semiquantitative RT-PCR. β-Actin was used as a loading control. Representative data of five independent experiments are shown.
Figure 3. The dynamics of p16Ink4a gene expression during the aging process in vivo. (A) The same p16-luc mice were subjected to noninvasive BLI every 2 wk throughout their entire life span. Representative images of 10 independent mice are shown. (B) The p16-luc mice (young and old) were subjected to noninvasive BLI. The same mice were then incised through mouth and anus under anesthesia. (A and B) The color bars indicate photons with minimum and maximum threshold values. (C) Bioluminescence intensity emitted from the organs from young (1.5 mo) and old (22.5 mo) mice were graphed (log10 scale). The mean ± SD of five independent experiments is shown. (D) The levels of exogenous (human) p16Ink4a gene expression and endogenous (mouse) p16Ink4a gene expression from young (Y; 1.5 mo) and old (O; 22.5 mo) mice were analyzed by semiquantitative RT-PCR. β-Actin was used as a loading control. Representative data of five independent experiments are shown.CC BY-NC-SA · PubMed Central
Figure 4. The dynamics of p16Ink4a gene expression during skin papilloma formation. (A) The p16-luc mice treated with DMBA/TPA were subjected to noninvasive BLI at the indicated time points after initiation of TPA treatment. Representative images of 10 independent experiments are shown (top). These papillomas were photographed in dimmed light (bottom). The color bar indicates photons with minimum and maximum threshold values. (B) Representative Western blots of biopsy samples of skin papillomas or control normal skin are shown using the antibodies indicated on the left. Vinculin was used as a loading control. (C) Hematoxylin and eosin (HE) staining, SA β-gal staining, and immunohistochemistry for endogenous p16Ink4a expression, phosphorylation of pRb at Ser807, Ki67 expression, phosphorylation of histone H2AX (γ-H2AX), and phosphorylation of p53 at Ser18 were performed by using biopsy samples of skin papillomas or control normal skin. The boxes denote regions shown below at higher magnification (p16 staining). (D) The levels of ROS were measured by using biopsy samples of skin papillomas or control normal skin. The means ± SD of three independent experiments are shown.
Figure 4. The dynamics of p16Ink4a gene expression during skin papilloma formation. (A) The p16-luc mice treated with DMBA/TPA were subjected to noninvasive BLI at the indicated time points after initiation of TPA treatment. Representative images of 10 independent experiments are shown (top). These papillomas were photographed in dimmed light (bottom). The color bar indicates photons with minimum and maximum threshold values. (B) Representative Western blots of biopsy samples of skin papillomas or control normal skin are shown using the antibodies indicated on the left. Vinculin was used as a loading control. (C) Hematoxylin and eosin (HE) staining, SA β-gal staining, and immunohistochemistry for endogenous p16Ink4a expression, phosphorylation of pRb at Ser807, Ki67 expression, phosphorylation of histone H2AX (γ-H2AX), and phosphorylation of p53 at Ser18 were performed by using biopsy samples of skin papillomas or control normal skin. The boxes denote regions shown below at higher magnification (p16 staining). (D) The levels of ROS were measured by using biopsy samples of skin papillomas or control normal skin. The means ± SD of three independent experiments are shown.CC BY-NC-SA · PubMed Central
Figure 5. Bisulfite sequence analysis of the mouse p16Ink4a gene promoter. A schematic illustration of the mouse p16Ink4a gene promoter is shown at the top, with the sequence numbered backwards from immediately before the initiating Met codon (the positions of the numbers are indicated by vertical dashes). The transcriptional start site is indicated by a thick vertical line, with the direction of transcription shown by the associated arrow. The coding region of the mouse p16Ink4a gene is annotated and indicated with a closed box. The clustering of CpG dinucleotides is shown with thin vertical lines through the promoter region. The positions of the primers for methyl-specific PCR are indicated below with two small arrows. Bisulfite-treated DNA was prepared from normal skin or early or late papillomas (Fig. 4). Methylation-specific PCR products were subcloned into the pGEM-T vector, and 10 clones for each sample were sequenced. Representative results of three different experiments are shown.
Figure 5. Bisulfite sequence analysis of the mouse p16Ink4a gene promoter. A schematic illustration of the mouse p16Ink4a gene promoter is shown at the top, with the sequence numbered backwards from immediately before the initiating Met codon (the positions of the numbers are indicated by vertical dashes). The transcriptional start site is indicated by a thick vertical line, with the direction of transcription shown by the associated arrow. The coding region of the mouse p16Ink4a gene is annotated and indicated with a closed box. The clustering of CpG dinucleotides is shown with thin vertical lines through the promoter region. The positions of the primers for methyl-specific PCR are indicated below with two small arrows. Bisulfite-treated DNA was prepared from normal skin or early or late papillomas (Fig. 4). Methylation-specific PCR products were subcloned into the pGEM-T vector, and 10 clones for each sample were sequenced. Representative results of three different experiments are shown.CC BY-NC-SA · PubMed Central
Figure 6. Correlation between DNMT1 level and H3K9me2 level around the p16Ink4a gene promoter. (A) Early passage TIG-3 cells were infected with retrovirus encoding oncogenic Ras (+H-RasV12). Cells were then subjected to Western blot (WB) analysis at the indicated times with the antibodies shown on the left and to analysis of intracellular levels of ROS. (B) Early passage TIG-3 cells were infected with retrovirus encoding oncogenic Ras (RasV12) or control empty vector for 10 d and were subjected to Western blotting with the antibodies shown, analysis of intracellular levels of ROS, and to ChIP analysis using the antibodies indicated (IP). The precipitated DNA was amplified by real-time quantitative PCR (qPCR) using primers specific for the p16Ink4a gene promoter described previously (Bracken et al., 2007). (C and D) Early passage TIG-3 cells were infected with retrovirus encoding shRNA against DNMT1 or control scramble shRNA. Cell extracts were prepared from cells at 7 d after selection with puromycin and were subjected to Western blotting and quantitative real-time RT-PCR (RT-qPCR) analysis for p16Ink4a gene expression (C) and subjected to ChIP analysis using antibodies against H3K9me2 (D). (D) The precipitated DNA was amplified by real-time quantitative PCR using primers specific for the p16Ink4a gene promoter described previously (Bracken et al., 2007). (A–D) The means ± SD of three independent experiments are shown.
Figure 6. Correlation between DNMT1 level and H3K9me2 level around the p16Ink4a gene promoter. (A) Early passage TIG-3 cells were infected with retrovirus encoding oncogenic Ras (+H-RasV12). Cells were then subjected to Western blot (WB) analysis at the indicated times with the antibodies shown on the left and to analysis of intracellular levels of ROS. (B) Early passage TIG-3 cells were infected with retrovirus encoding oncogenic Ras (RasV12) or control empty vector for 10 d and were subjected to Western blotting with the antibodies shown, analysis of intracellular levels of ROS, and to ChIP analysis using the antibodies indicated (IP). The precipitated DNA was amplified by real-time quantitative PCR (qPCR) using primers specific for the p16Ink4a gene promoter described previously (Bracken et al., 2007). (C and D) Early passage TIG-3 cells were infected with retrovirus encoding shRNA against DNMT1 or control scramble shRNA. Cell extracts were prepared from cells at 7 d after selection with puromycin and were subjected to Western blotting and quantitative real-time RT-PCR (RT-qPCR) analysis for p16Ink4a gene expression (C) and subjected to ChIP analysis using antibodies against H3K9me2 (D). (D) The precipitated DNA was amplified by real-time quantitative PCR using primers specific for the p16Ink4a gene promoter described previously (Bracken et al., 2007). (A–D) The means ± SD of three independent experiments are shown.CC BY-NC-SA · PubMed Central
Figure 7. Accelerated induction of p16Ink4a gene expression by DNA damage in mice lacking p53. (A) 8-wk-old p16-luc mice lacking the p53 gene (p53−/−) or their wild-type controls (p53+/+) were injected i.p. with 20 µg/g DXR and subjected to noninvasive BLI at the indicated time points. The same mice were then sacrificed, tissues were rapidly removed and placed in culture dishes, and ex vivo tissue BLI was performed. The tissues examined were as follows: 1, brain; 2, cervical LNs; 3, heart; 4, lung; 5, thymus; 6, small intestine; 7, liver; 8, kidney; 9, spleen. The color bars indicate photons with minimum and maximum threshold values. Representative images of five different experiments are shown. Schematic drawings illustrating the mouse bodies are shown (dotted lines). (B and C) Isolated tissues were subjected to analysis of quantitative real-time RT-PCR (RT-qPCR) for p16Ink4a gene expression (B) or to immunofluorescence analysis using antibody against γ-H2AX (red; C). DNA was stained with DAPI (blue). The histogram indicates the percentage of nuclei that were positive for γ-H2AX staining. The means ± SD of three independent experiments are shown. (C) Enlarged images of the boxed areas are shown below. Small intestinal crypts are marked by dashed lines. (D) Early passage TIG-3 cells were infected with retrovirus encoding shRNA against p53 or control scramble shRNA. Cell extracts were prepared after selection with puromycin and subjected to Western blotting using the antibodies shown on the left. β-Actin was used as a loading control.
Figure 7. Accelerated induction of p16Ink4a gene expression by DNA damage in mice lacking p53. (A) 8-wk-old p16-luc mice lacking the p53 gene (p53−/−) or their wild-type controls (p53+/+) were injected i.p. with 20 µg/g DXR and subjected to noninvasive BLI at the indicated time points. The same mice were then sacrificed, tissues were rapidly removed and placed in culture dishes, and ex vivo tissue BLI was performed. The tissues examined were as follows: 1, brain; 2, cervical LNs; 3, heart; 4, lung; 5, thymus; 6, small intestine; 7, liver; 8, kidney; 9, spleen. The color bars indicate photons with minimum and maximum threshold values. Representative images of five different experiments are shown. Schematic drawings illustrating the mouse bodies are shown (dotted lines). (B and C) Isolated tissues were subjected to analysis of quantitative real-time RT-PCR (RT-qPCR) for p16Ink4a gene expression (B) or to immunofluorescence analysis using antibody against γ-H2AX (red; C). DNA was stained with DAPI (blue). The histogram indicates the percentage of nuclei that were positive for γ-H2AX staining. The means ± SD of three independent experiments are shown. (C) Enlarged images of the boxed areas are shown below. Small intestinal crypts are marked by dashed lines. (D) Early passage TIG-3 cells were infected with retrovirus encoding shRNA against p53 or control scramble shRNA. Cell extracts were prepared after selection with puromycin and subjected to Western blotting using the antibodies shown on the left. β-Actin was used as a loading control.CC BY-NC-SA · PubMed Central
Figure 8. Induction of p16Ink4a gene expression in p53 knockout mice. (A) p16-luc mice lacking the p53 gene (p53−/−) or their wild-type controls (p53+/+) were subjected to noninvasive BLI at the indicated age. The same mice were then incised under anesthesia (photographed under regular light) and subjected to BLI again (invasive BLI). Representative images of five different experiments are shown. The color bars indicate photons with minimum and maximum threshold values. Schematic drawings illustrating the mouse bodies are shown (dotted lines). The arrows show bioluminescence signals derived from thymus. (B–F) Thymus tissue was isolated from both genotypes of 10-wk-old mice (A) and subjected to analysis of quantitative real-time RT-PCR (RT-qPCR) for p16Ink4a gene expression (B) or DNMT1 gene expression (F) or histochemistry (hematoxylin and eosin [H&E] staining and SA β-gal staining; E), γ-H2AX foci (C), or intracellular levels of ROS (D). The means ± SD of three independent experiments are shown.
Figure 8. Induction of p16Ink4a gene expression in p53 knockout mice. (A) p16-luc mice lacking the p53 gene (p53−/−) or their wild-type controls (p53+/+) were subjected to noninvasive BLI at the indicated age. The same mice were then incised under anesthesia (photographed under regular light) and subjected to BLI again (invasive BLI). Representative images of five different experiments are shown. The color bars indicate photons with minimum and maximum threshold values. Schematic drawings illustrating the mouse bodies are shown (dotted lines). The arrows show bioluminescence signals derived from thymus. (B–F) Thymus tissue was isolated from both genotypes of 10-wk-old mice (A) and subjected to analysis of quantitative real-time RT-PCR (RT-qPCR) for p16Ink4a gene expression (B) or DNMT1 gene expression (F) or histochemistry (hematoxylin and eosin [H&E] staining and SA β-gal staining; E), γ-H2AX foci (C), or intracellular levels of ROS (D). The means ± SD of three independent experiments are shown.CC BY-NC-SA · PubMed Central
Figure 9. Cross talk between the p53 and p16 pathways through DDR. Although oncogenic Ras signaling has a potential to activate p16Ink4a gene expression, this effect is initially counteracted by an elevation of DNMT1 level and thereby causes a strong proliferative burst, resulting in the accumulation of DNA damage. The accumulation of DNA damage activates ROS production, which in turn blocks DNMT1 gene expression, thereby causing epigenetic derepression of p16Ink4a gene expression and thus senescence cell cycle arrest. This pathway is counterbalanced by the p53 pathway because p53 is immediately activated by DNA damage and blocks proliferation of damaged cells that cause further accumulation of DNA damage. Thus, the DDR pathway activating p16Ink4a expression is accelerated in the event of p53 inactivation.
Figure 9. Cross talk between the p53 and p16 pathways through DDR. Although oncogenic Ras signaling has a potential to activate p16Ink4a gene expression, this effect is initially counteracted by an elevation of DNMT1 level and thereby causes a strong proliferative burst, resulting in the accumulation of DNA damage. The accumulation of DNA damage activates ROS production, which in turn blocks DNMT1 gene expression, thereby causing epigenetic derepression of p16Ink4a gene expression and thus senescence cell cycle arrest. This pathway is counterbalanced by the p53 pathway because p53 is immediately activated by DNA damage and blocks proliferation of damaged cells that cause further accumulation of DNA damage. Thus, the DDR pathway activating p16Ink4a expression is accelerated in the event of p53 inactivation.CC BY-NC-SA · 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

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

3 more claims about this paper

Create a free account to see every claim other papers make about this one.

See all claims — free account

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

1 retracted

Show all 68 references

Cited by

See all 172

Metadata

Authors
Yamakoshi, K., Takahashi, A., Hirota, F., Nakayama, R., Ishimaru, N., Kubo, Y., Mann, D. J., Ohmura, M., Hirao, A., Saya, H., Arase, S., Hayashi, Y., Nakao, K., Matsumoto, M., Ohtani, N., & Hara, E.
Affiliations
  • Japanese Foundation For Cancer Research (JP)
  • Tokushima University (JP)
  • RIKEN (JP)
  • Imperial College London (GB)
  • Kanazawa University (JP)
  • Keio University (JP)
Corresponding author
  • Ohtani, Naoko
  • Hara, Eiji
Venue
Journal of Cell Biology
Published
2009-08-10
Volume
186
Issue
3
Pages
393-407
ISSN
  • 0021-9525 (ISSN-L)
  • 1540-8140
OpenAlex
W2148949613
Cited by
172
References
71
Citations by year
2026: 11 · 2025: 6 · 2024: 14 · 2023: 14 · 2022: 19 · 2021: 13 · 2020: 14 · 2019: 11 · 2018: 5 · 2017: 12 · 2016: 7 · 2015: 11 · 2014: 7 · 2013: 8 · 2012: 7
FWCI
3.33
Type
journal article
Language
en
Open access
Licence
cc-by-nc-sa
Locations
  • The Journal of Cell Biology (publishedVersion)
  • PubMed Central (publishedVersion, cc-by-nc-sa)
  • PubMed (publishedVersion)
Funders
Japan Society for the Promotion of Science (21390290)
Topics
  • Health Sciences › Medicine › Oncology › Cancer-related Molecular Pathways
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Epigenetics and DNA Methylation
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Cell Biology › Telomeres, Telomerase, and Senescence
Keywords
p16INK4a, p53, tumor suppressor, epigenetic derepression, DNMT1, DNA methylation, DNA damage response, oncogenic stress, Ras activation, bioluminescence imaging, in vivo gene expression
MeSH terms
Animals, Cells, Cultured, Disease Models, Animal, Humans, Mice, Inbred C57BL, Mice, Transgenic, Time Factors, Tumor Suppressor Protein p53, Mice, Knockout, Cyclin-Dependent Kinase Inhibitor p16, Reverse Transcriptase Polymerase Chain Reaction, Gene Expression Profiling, Mice

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: Real-time in vivo imaging of p16Ink4a reveals cross talk with p53. https://app.cochat.ai/papers/10.1083/jcb.200904105
The paper
Yamakoshi, K., Takahashi, A., Hirota, F., Nakayama, R., Ishimaru, N., Kubo, Y., Mann, D. J., Ohmura, M., Hirao, A., Saya, H., Arase, S., Hayashi, Y., Nakao, K., Matsumoto, M., Ohtani, N., & Hara, E. (2009). Real-time in vivo imaging of p16Ink4a reveals cross talk with p53. Journal of Cell Biology. https://doi.org/10.1083/jcb.200904105

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.