Programmed cell death ligand-1-mediated enhancement of hexokinase 2 expression is inversely related to T-cell effector gene expression in non-small-cell lung cancer
- 329 Downloads
We investigated the role of PD-L1 in the metabolic reprogramming of non-small cell lung cancer (NSCLC).
Changes in glycolysis-related molecules and glycolytic activity were evaluated in PD-L1low and PD-L1high NSCLC cells after transfection or knockdown of PD-L1, respectively. Jurkat T-cell activation was assessed after co-culture with NSCLC cells. The association between PD-L1 and immune response-related molecules or glycolysis were analyzed in patients with NSCLC and The Cancer Genome Atlas (TCGA).
Transfecting PD-L1 in PD-L1low cells enhanced hexokinase-2 (HK2) expression, lactate production, and extracellular acidification rates, but minimally altered GLUT1 and PKM2 expression and oxygen consumption rates. By contrast, knocking-down PD-L1 in PD-L1high cells decreased HK2 expression and glycolysis by suppressing PI3K/Akt and Erk pathways. Interferon-γ (IFNγ) secretion and activation marker expression was decreased in stimulated Jurkat T-cells when co-cultured with HK2-overexpressing vector-transfected tumor cells rather than empty vector-transfected tumor cells. Immunohistochemistry revealed that PD-L1 expression was positively correlated with HK2 expression in NSCLC (p < 0.001). In TCGA, HK2 exhibited a positive linear association with CD274 (PD-L1) expression (p < 0.001) but an inverse correlation with the expression of CD4, CD8A, and T-cell effector function-related genes in the CD274high rather than CD274low group. Consistently, there were fewer CD8+ T-cells in PD-L1positive/HK2high tumors compared to PD-L1positive/HK2low tumors in squamous cell carcinoma.
PD-L1 enhances glycolysis in NSCLC by upregulating HK2, which might dampen anti-tumor immunity. PD-L1 may contribute to NSCLC oncogenesis by inducing metabolic reprogramming and immune checkpoint.
KeywordsProgrammed cell death-ligand-1 Hexokinase 2 Glycolysis Non-small cell lung cancer Tumor microenvironment
Extracellular acidification rate
Non-small cell lung cancer
Oxygen consumption rate
Programmed cell death protein 1
Programmed death-ligand 1
Pulmonary squamous cell carcinoma
Small interfering RNA
The Cancer Genome Atlas
Tumor microenvironment immune type
The programmed cell death-1 (PD-1)/ programmed cell death-ligand (PD-L) pathway acts as an immune checkpoint for tumor cells to evade host immune surveillance [1, 2]. Recently, PD-1/PD-L1 blockade has emerged as a therapeutic strategy for patients with cancer , and has been approved for the first-line and second-line treatment of patients with non-small-cell lung cancer (NSCLC) . However, some NSCLC patients respond to PD-1/PD-L1 blockade; thus, predictive biomarkers have been identified including PD-L1 expression, tumor mutational burden, and preexisting adaptive immune responses . Although immunohistochemistry for PD-L1 has been approved as a companion/complementary diagnostics for PD-1/PD-L1 blockade, the response rate to PD-1/PD-L1 blockades is approximately 30% even in patients with PD-L1positive NSCLC . Thus, more studies are needed to understand the biology and the mechanism of action of PD-1/PD-L1 blockade.
Most research into the PD-1/PD-L1 pathway has focused on the PD-1/PD-L1 interaction between target cells and immune cells. However, some studies have suggested other functions for PD-1 and PD-L1 beyond immune suppression. Intrinsic PD-L1 signaling promotes tumor cell proliferation and growth in melanoma and ovarian cancer cells by regulating autophagy and the mTOR pathway, and protects tumor cells from interferon (IFN)-mediated cytotoxicity [6, 7, 8]. In addition, PD-1 signaling alters T-cell metabolism by inhibiting glycolysis and amino acid metabolism and promoting lipolysis and fatty acid oxidation, thereby inhibiting effector T-cell differentiation .
To meet the metabolic requirements for growth and proliferation, tumor cells predominantly use glucose through aerobic glycolysis rather than oxidative phosphorylation (metabolic reprogramming known as the “Warburg effect”) . Metabolic reprogramming is also important for proliferation and effector function of T-cells [11, 12]. Upon activation, lymphocytes undergo a metabolic transition from oxidative phosphorylation to aerobic glycolysis [12, 13, 14], which indicates that immune and tumor cells utilize similar metabolic reprogramming for their survival and function. Thus, they could compete for restricted nutrients within the tumor microenvironment (TME) and increased nutrient consumption by tumor cells could lead to an immunosuppressive TME by dampening the T-cell metabolism.
Using a murine sarcoma model, Chang et al. recently reported that PD-L1 elevated glycolysis in tumor cells, which promoted tumor progression by reducing glycolytic capacity and IFN-γ production in T-cells . However, the role of PD-L1 in glucose metabolism and its clinical implication in human cancer cells is unclear. Therefore, we investigated the functional relationship between PD-L1 expression and glycolysis and its regulatory mechanisms in human lung cancer.
Materials and methods
Cell culture and reagents
NSCLC cell lines, including A549 and H460, were purchased from the American Type Culture Collection (Manassas, VA, USA). Jurkat cells were purchased from the Korean Cell Line Bank (Seoul, Republic of Korea). Cells were cultured in DMEM (A549 cells) and RPMI 1640 (H460 and Jurkat cells) supplemented with 10% FBS, 100 U/mL penicillin and 100 μg/mL streptomycin in a humidified atmosphere of 5% CO2 at 37 °C. 2-deoxy-D-glucose, LY294002, U0126, and SB203580 were purchased from Sigma-Aldrich (St. Louis, MO, USA).
Patients and samples
Tissue microarrays were constructed from formalin-fixed paraffin-embedded tumor tissues from 393 patients who underwent surgery for NSCLC at Seoul National University Hospital (SNUH, Seoul, Republic of Korea), including 228 pulmonary adenocarcinomas (pADCs) and 165 pulmonary squamous cell carcinomas (pSqCCs). An additional cohort of 80 patients with NSCLCs who received PD-1 blockade immunotherapy was established. Response to PD-1 blockade was assessed based on RECIST v1.1 . Clinicopathological features of patients were analyzed as described in Additional file 1: Supplementary Methods. This study followed the World Medical Association Declaration of Helsinki recommendations and was approved by the Institutional Review Board (IRB) of SNUH (No.: H-1404-100-572).
Transfection of plasmid vector and siRNAs
A human PD-L1-expressing plasmid (Catalog No. HG10084-UT; Beijing, China) and a human HK2-expressing plasmid Catalog No. HG17967-UT; Beijing, China) were purchased from Sino Biological Inc.. PD-L1-specific siRNA-1 and 2 complementary to PD-L1 (GeneBank accession No.: NM 014143.2) were designed and synthesized by Bioneer (Daejeon, Republic of Korea). The sequences of siRNAs were as follows: PD-L1 siRNA-1, sense 5′-CAGCAUUGGAACUUCUGAU(dTdT)-3′ and antisense 5′-AUCAGAAGUUCCAAUGCUG(dTdT)-3′; PD-L1 siRNA-2, sense 5′-GAAUCAACACAACAACUAA(dTdT)-3′ and antisense 5′-UUAGUUGUUGUGUUGAUUC(dTdT)-3′. For transfection, cells were plated in either 12-well plates or 24-well plates and allowed to adhere for 24 h. Then the cells were transfected with plasmid or siRNA using Lipofectamine 2000 (Invitrogen, Carlsbad, CA, USA) in Opti-MEM media (Qiagen, Germantown, MD, USA). After 6 h, the media were replaced with fresh complement media and cells were harvested 24–48 h after transfection.
Quantitative reverse transcription PCR (qRT-PCR)
Total RNA was extracted from cells using TRIzol reagent (Life Technologies, Carlsbad, CA, USA) and subjected to reverse transcription using a PrimeScript First Strand cDNA Synthesis Kit (Takara Bio, Otsu, Japan). PCR was performed using SYBR® qRT-PCR Kit (Clontech Laboratories, Mountain View, CA, USA) and a Step One Plus thermal cycler (Applied Biosystems, Foster City, CA, USA) in triplicate. GAPDH and β-actin were used as the internal control. The following primer sequences were used for qRT-PCR: GAPDH forward 5′-CCCTTCATTGACCTACCTCAACTACAT-3′ and reverse 5′-ACGATACCAAAGTTGTCATGGAT-3′; β-actin forward 5′- CTGGAACGGTGAAGGTGAC-3′ and reverse 5′-AAGGGACTTCCTGTAACAATGCA -3′; CD274 (PD-L1) forward 5′-TATGGTGGTGCCGACTACAA-3′ and reverse 5′-TGGCTCCCAGAATTACCAAG-3′; SLC2A1 (GLUT1) forward 5′-GATTGGCTCCTTCTCTGTGG-3′ and reverse 5′-TCAAAGGACTTGCCCAGTTT-3′; HK2 forward 5′-CAAAGTGACAGTGGGTGTGG-3′ and reverse 5′-GCCAGGTCCTTCACTGTCTC-3′; PKM2 forward 5′-CCACTTGCAATTATTTGAGGAA-3′ and reverse 5′-GTGAGCAGACCTGCCAGACT-3′; PFKP forward 5′-GGGCCAAGGTGTACTTCATC-3′ and reverse 5′-TGGAGACACTCTCCCAGTCG-3′; PFKL forward 5′-GGTGGACCTGGAGAAGCTG-3′ and reverse 5′- GGCACCCACATAAATGCC-3′; PFKM forward 5′-GCCATCAGCCTTTGACAGA-3′ and reverse 5′-CTCCAAAAGTGCCATCACTG-3′; GPI forward 5′- GGAGACCATCACGAATGCAGA − 3′ and reverse 5′-TAGACAGGGCAACAAAGTGCT-3′; PGK forward 5′-AAGTCGGTAGTCCTTATGAGC-3′ and reverse 5′- CACATGAAAGCGGAGGTTCT-3′.
Total cellular proteins were extracted using lysis buffer (5 mM EDTA, 300 mM NaCl, 0.1% NP-40, 0.5 mM NaF, 0.5 mM Na3VO4, 0.5 mM PMSF, and 10 μg/mL each of aprotinin, pepstatin, and leupeptin; Sigma-Aldrich). A total of 30–50 μg protein was separated using 10% SDS-PAGE and transferred to polyvinylidene difluoride membranes (Millipore, Bedford, MA, USA). Then immunoblotting was performed using antibodies against PD-L1 (clone E1L3N), GLUT1, HK2, PKM2, P-Akt, Akt (Cell Signaling Technology, Danvers, MA, USA), P-Erk, Erk, P-p38MAPK, p38MAPK, and β-actin (Santa Cruz Biotechnology, Dallas, TX, USA). Most images of western blots were from parallel gels and actin images were obtained from the stripped and re-probed blots. The immunoblots were visualized using an enhanced chemiluminescence detection system (Amersham Pharmacia Biotech, Uppsala, Sweden).
Glycolysis assessment: lactate production, hexokinase activity, and extracellular acidification rate (ECAR) assays
Glycolysis was evaluated using lactate production, hexokinase activity, and ECAR assays, as detailed in the Additional file 1: Supplementary Material and Methods.
Direct co-culture and transwell co-culture system were performed. Co-culture experiments were performed in 24-well plates without or with pore size 0.4 μm insets (Corning Costar, Corning, NY, USA). A549 cell (5 × 104) were seeded and cultured in the outer wells of 24-well plates in DMEM supplemented with 10% FBS for 24 h. A549 cells were transfected with empty or HK2-expressing vectors, as mentioned above. After 24 h, when fully upregulated HK2 expression, medium was changed to RPMI supplemented with 10% FBS and 1% penicillin/streptomycin. Incubating tumor cells for another 24 h, Jurkat cells (4 × 105) were added to directly to tumor cells or added to inner wells in transwell system. After 1 h of stabilization time, final concentration of 2 μg/ml soluble anti-CD3 (eBioscience, San Diego, CA, USA), 1 μg/ml soluble anti-CD28 (eBioscience) and 5 μg/ml anti-mouse Ig (SouthernBiotech, Birmingham, AL, USA) were added. 24 h later, media was harvested for IFN-γ ELISA assay and Jurkat cells were harvested for flow cytometry.
Enzyme-linked immunosorbent assay (ELISA) for IFN-γ
IFN-γ level in cell-free media was estimated using Human IFN-γ ELISA kit (R&D system, #DY285–05, Minneapolis, MN, USA) according to the manufacturer’s protocol.
Cells were harvested and washed with FACS buffer (0.5% BSA and 0.05% sodium azide in PBS). For discriminating dead cells, cells were first stained with zombie aqua (#423101, Biolegend, San Diego, CA, USA) in PBS for 10 min at room temperature in dark. Cells were stained with the PE-cy7 conjugated anti-CD69 antibody (#310911, Biolegend), PE-conjugated PD-L1 antibody (#329705, Biolegend) in FACS buffer for at least 30 min at 4 °C in dark. Flow cytometry was carried out on the LSRFortessa X-20 (BD Biosciences). Data were analyzed using FlowJo v10.1 (Treestar).
IHC was performed for PD-L1 (clone E1L3N), HK2, GLUT1, PKM2, and CD8, as detailed in Additional file 1: Supplementary Material and Methods. PD-L1 positivity was defined as ≥10% of tumor cells with moderate-to-strong membranous intensity. H-scores for GLUT1, HK2 and PKM2 were estimated by integrating the intensity and proportion of staining . CD8+ TILs per mm2 were automatically enumerated, as described in Additional file 1: Supplementary Material and Methods.
RNA seq and gene set enrichment analyses (GSEAs)
A549 cells transfected with PD-L1-expressing plasmid and H460 cells transfected with PD-L1 siRNA were submitted to RNA seq and GSEAs, as detailed in Additional file 1: Supplementary Material and Methods.
The Cancer Genome Atlas (TCGA) data analyses
The level 3 data of TCGA were downloaded from the UCSC Cancer Browser (https://genome-cancer.ucsc.edu) including pADC (N = 513) and pSqCC (N = 502) datasets and genome analysis was performed, as described in Additional file 1: Supplementary Material and Methods.
Statistical analyses were performed using SPSS v.21 (IBM Corp., New York, NY, USA), as detailed in Additional file 1: Supplementary Material and Methods. Two-sided p values < 0.05 were considered statistically significant in all analyses.
PD-L1 upregulates HK2 expression and aerobic glycolysis in lung cancer cells
GLUT1 is the most common glucose transporter in humans. HK2 and PKM2 catalyze the first and final step of glycolysis, respectively [10, 18, 19]. To investigate whether PD-L1 affects the glycolysis in human NSCLC, we screened basal expression of PD-L1 and glycolysis-related molecules and glycolytic activity in several NSCLC cell lines (Additional file 3: Figure S1A-E). PD-L1low A549 cells showed lower HK2 expression and decreased aerobic glycolysis, as determined using lactate production, hexokinase activity, and ECAR assays, whereas PD-L1high H460 cells showed higher HK2 expression and increased glycolysis (Additional file 3: Figure S1F-J). Thus, we overexpressed PD-L1 in PD-L1low A549 cells and knocked-down PD-L1 in PD-L1high H460 cells for further experiments.
PD-L1 expression is positively correlated with HK2 expression in NSCLCs from patients
Increased tumoral HK2 expression impairs T-cell effector function
HK2 expression and glycolytic signature are inversely related to the expression of T-cell effector molecules in PD-L1high human NSCLCs
High HK2 expression in NSCLC tended to be associated with poor response to anti-PD-1 immunotherapy in patients
Finally, to evaluate the effects of HK2 overexpression on the response to PD-1/PD-L1-targeted immunotherapy, we analyzed 80 NSCLC patients who were treated with nivolumab (N = 63) or pembrolizumab (N = 17). The characteristics of the patients are detailed in Additional file 2: Table S2. The median duration of follow-up was 5.2 months (1.0–49.9 months) and the overall response rate (ORR) was 16.3% (13/80). Tumors that exhibited the lowest and highest quartile of HK2 expression were designated HK2low and HK2high cases, respectively. The ORR was higher in the PD-L1positive group (26.5% [9/34]) than the PD-L1negative group (8.7% [4/46]). In addition, the ORR was higher in the PD-L1positive/HK2low group than the PD-L1positive/HK2high group, although statistically insignificant (Additional file 3: Figure S9). There was no significant difference in patient survival according to PD-L1 and HK2 status (data not shown). These findings suggest that patients with HK2high NSCLC might show poor response to anti-PD-1 immunotherapy despite PD-L1 expression. However, further validation is required.
Several studies have found that HIF-1α enhances glycolysis by upregulating glycolytic enzymes [22, 23], and upregulates PD-L1 , which suggests a potential link between PD-L1 and glycolysis in cancer. The current study demonstrates that PD-L1 upregulates aerobic glycolysis in NSCLC cells by enhancing HK2 expression. This is the first study to demonstrate the intrinsic effects of PD-L1 on HK2-mediated glycolysis in human lung cancer cells. Moreover, we found a significant positive correlation between PD-L1 and HK2 expression in human NSCLC tissues. Although oncogenic mutations affect tumor cell metabolism [10, 25], the relationship of PD-L1 and HK2 expression was not affected by EGFR mutation status in NSCLCs. However, it remains unclear whether mutations in other genes affect link between PD-L1 and HK2 in NSCLCs. Meanwhile, TCGA analyses showed a positive correlation between PD-L1 and HK2 expression in hepatocellular carcinoma (rho = 0.461, p < 0.001), sarcoma (rho = 0.188, p < 0.001), and stomach adenocarcinoma (rho = 0.225, p < 0.001), but not in renal cell carcinoma and urothelial carcinoma (data not shown). These findings suggest that a functional link between PD-L1 and HK2 might occur in various cancers in a histologic type-dependent manner.
In this study, we confirmed that the changes in surface PD-L1 as well as total PD-L1 expression after PD-L1 overexpression and knockdown (Figs. 1 and 2 and Additional file 3: Figure S2A). Therefore, there is a potential that surface PD-L1 expression might give an intrinsic signal to modulate metabolism of tumor cells. By treating murine tumor cells with anti-PD-L1 blocking antibodies, Chang et al. previously showed that surface PD-L1 might be involved in tumor cell glycolysis . In this study, we demonstrated that upregulation of total PD-L1 expression raised glycolysis/HK2 expression of lung cancer cells in the absence of PD-1 ligation. However, we did not compare the actual effect of surface and intracellular PD-L1 expressions on HK2 expression and glycolysis of human lung cancer, and thus it is unclear if both surface and cytoplasm PD-L1 or only surface PD-L1 might be involved in the PD-L1-mediated increase in HK2 overexpression and glycolysis in tumor cells. This is an intriguing issue and remains to be determined in further studies incorporating biochemical assays.
Of note, we also found that upregulated tumoral HK2 expression impaired an effector T-cell function (i.e. IFN-γ production), in a co-culture system. These findings were further validated using TCGA data and IHC for lung cancer tissues. HK2 expression was inversely correlated with the expression of T-cell effector response-related genes and the number of CD8+ TILs in PD-L1high NSCLCs but not PD-L1low NSCLCs. These findings suggest that the expression status of PD-L1 and HK2 in tumor cells might be involved in the regulation of immune response in the TME of NSCLCs. However, since the correlations between variables were often weak to moderate despite statistical significance in analysis of TCGA and patients tissue data, the cause-effect relationship between variables needs to be interpreted with caution. Despite of this limitation, our results regarding the relationship between PD-L1, HK2, and T-cell effector function suggests that combination analyses of HK2 and PD-L1 expression in tumor cells might provide useful information to predict the TME immune signature in NSCLC.
When classifying TME immune types (TMIT) based on PD-L1 expression and CD8+ T-cell, tumors can be divided into four groups: PD-L1high/CD8+ T-cellhigh (TMIT I), PD-L1low/CD8+ T-celllow (TMIT II), PD-L1high/CD8+ T-celllow (TMIT III), and PD-L1low/CD8+ T-cellhigh (TMIT IV). Patients with TMIT I tumors are considered the best candidates for PD-1/PD-L1-targeted therapy [21, 26]. By contrast, those with TMIT III tumors exhibit a poor response to PD-1/PD-L1 blockade despite high PD-L1 expression . The current study demonstrates that PD-L1high/CD8+ T-celllow TMIT III NSCLCs could be characterized by high HK2 expression and glycolytic signature.
Regarding functional aspects of the immune responses, our results suggest that PD-L1-mediated HK2/glycolysis upregulation of tumor cells might dampen T-cell function in the TME of NSCLCs. In previous studies, tumor-derived lactate, a metabolite of glycolysis, inhibits T-cell function , and upregulates PD-L1 expression via the lactate receptor-TAZ pathway in lung cancer, thereby contributing to T-cell suppression . Taken together, these and the current findings suggest a potential positive feedback loop of the glycolysis/lactate-PD-L1-HK2-glycolysis axis in the TME. Thus, it is feasible that tumor cells might survive and outcompete surrounding TILs in the TME by expressing PD-L1 through metabolic reprogramming for survival/proliferation, dampening TIL functions by rendering a metabolically harmful microenvironment, and giving a direct inhibitory signal to PD-1 on adjacent immune cells, as schematically summarized in Additional file 3: Figure S10. Thus, the metabolic status of tumor cells could affect the responsiveness of patients to anti-PD-1/PD-L1 immunotherapy. To address this issue, we evaluated the effects of HK2 overexpression in tumor cells on the clinical response to PD-1 blockades. We observed a trend toward a better response in patients with PD-L1positive/HK2low NSCLC compared to those with PD-L1positive/HK2high NSCLCs, but no difference in the survival time between these groups. These results might be attributable to small number of patients, variable performance status and previous treatment history, laboratory-developed PD-L1 test, limitations of retrospective evaluation, and a possible occurrence of resistance mechanisms other than HK2 overexpression. Therefore, the clinical value of HK2 as a potential metabolic biomarker for anti-PD-1/PD-L1 therapy needs further validation.
Moreover, HK2 itself could be a therapeutic target in cancer therapy. HK2 participates in aerobic glycolysis and can affect many other metabolic processes through its product, glucose-6-phosphate. Besides metabolic process, HK2 has been demonstrated to prohibit apoptosis. HK2 binds to a voltage-dependent anion channel (VDAC) and the HK2/VDAC interaction stabilizes mitochondrial membrane and prevents binding pro-apoptotic factors to VDAC [29, 30, 31]. Therefore, HK2 inhibitors or in vivo deletion by CRISPR/Cas9  could affect tumor cells by inhibiting tumor metabolism and promoting apoptosis. Applying the concept of combinational therapy for cancer [33, 34], HK2 inhibition could enhance the response rate of PD-1/PD-L1 therapy by releasing immune cells from metabolic competition and directly increasing tumor cell apoptosis. This hypothesis needs to be validated by future studies.
In summary, this study demonstrates that PD-L1 increases glycolysis in NSCLC cells by upregulating HK2, which is associated with reduced expression of T-cell effector genes. These findings suggest that PD-L1 expression might enhance tumor cell survival by increasing glycolysis intrinsically and providing inhibitory signals to neighboring T-cells extrinsically, through interaction with PD-1 and metabolic competition. Thus, HK2 expression and glycolysis could be potential biomarkers or therapeutic targets in NSCLC patients in the era of cancer immunotherapy.
We’d like to thank Hyun Kyung Ahn for her technical assistance.
Designed the study and experiments: DHC, YKJ. Performed the experiments and analyzed the data: SK, JYJ, JK, DK, YAK, JCP. Performed statistical analysis: SK, CYO. Performed acquisition of clinical data: CYO, BK, MK, TMK, DSH. Wrote the paper: SK, JYJ, DHC, YKJ. All authors read and approved the final manuscript.
This work was supported by the Seoul National University Hospital Research Fund (grant No. 800–20160023), the Seoul National University (SNU) Invitation Program for Distinguished Scholar (2017. 3. to YKJ) (grant No. 0320170160 [2017–1312]), and the Basic Science Research Program (grant No.: NRF-2016R1D1A1B01015964) through the National Research Foundation (NRF) funded by the Ministry of Education, Science and Technology (MEST), Republic of Korea.
Ethics approval and consent to participate
This study was performed in accordance with the World Medical Association Declaration of Helsinki and was approved by the Institutional Review Board (IRB) of SNUH (H-1404-100-572). Informed consent for participation was waivered by the IRB of SNUH on the basis that this study was a retrospective study using archived material, and did not increase risk to the patients.
Consent for publication
The authors declare that they have no competing interests.
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.