CD73 complexes with emmprin to regulate MMP-2 production from co-cultured sarcoma cells and fibroblasts
- 120 Downloads
Interaction between cancer cells and fibroblasts mediated by extracellular matrix metalloproteinase inducer (emmprin, CD147) is important in the invasion and proliferation of cancer cells. However, the exact mechanism of emmprin mediated stimulation of matrix metalloprotease-2 (MMP-2) production from fibroblasts has not been elucidated. Our previous studies using an inhibitory peptide against emmprin suggested the presence of a molecule on the cell membrane which forms a complex with emmprin. Here we show that CD73 expressed on fibroblasts interacts with emmprin and is a required factor for MMP-2 production in co-cultures of sarcoma cells with fibroblasts.
CD73 along with CD99 was identified by mass spectrometry analysis as an emmprin interacting molecule from a co-culture of cancer cells (epithelioid sarcoma cell line FU-EPS-1) and fibroblasts (immortalized fibroblasts cell line ST353i). MMP-2 production was measured by immunoblot and ELISA. The formation of complexes of CD73 with emmprin was confirmed by immunoprecipitation, and their co-localization in tumor cells and fibroblasts was shown by fluorescent immunostaining and proximity ligation assays.
Stimulated MMP-2 production in co-culture of cancer cells and fibroblasts was completely suppressed by siRNA knockdown of CD73, but not by CD99 knockdown. MMP-2 production was not suppressed by CD73-specific enzyme inhibitor (APCP). However, MMP-2 production was decreased by CD73 neutralizing antibodies, suggesting that CD73-mediated suppression of MMP-2 production is non-enzymatic. In human epithelioid sarcoma tissues, emmprin was immunohistochemically detected to be mainly expressed in tumor cells, and CD73 was expressed in fibroblasts and tumor cells: emmprin and CD73 were co-localized predominantly on tumor cells.
This study provides a novel insight into the role of CD73 in emmprin-mediated regulation of MMP-2 production.
KeywordsEmmprin CD73 MMP-2 Sarcoma Fibroblasts
- CD147; MMP
Extracellular matrix metalloproteinase inducer
In situ proximity ligation assay
Small interference RNA
Interaction between cancer cells and stromal cells is essential for cancer invasion and metastasis, however, the mechanism of interaction remains unclear. Invasion and metastasis of cancer cells require remodeling of the matrix surrounding the cancer cells. Fibroblasts present in the stroma are mainly responsible for producing the matrix metalloproteinases (MMPs), which play a central role in remodeling of the matrix . Extracellular matrix metalloproteinase inducer (emmprin), a protein expressed in most tumors, plays an important role in regulating production of MMPs from fibroblasts [1, 2, 3, 4, 5, 6]. Emmprin is a multifunctional protein that is expressed in reproductive cells, brain, eyes, and muscle, and play roles in immunity and reproduction in addition to its role in cancer [6, 7, 8, 9]. Emmprin is coded by a gene present on p13.3 of chromosome 19, and it has 10 exons. Emmprin is a member of the immunoglobulin (Ig) superfamily, with two extracellular immunoglobulin-like domains and three conserved asparagine (N)-glycosylation sites and a molecular weight of 31–65 kDa, depending on the extent of the glycosylation [1, 6, 7, 10].
The first extracellular Ig domain of emmprin is required for inducing MMPs production from fibroblasts; N-glycosylation is also essential for activity [1, 3, 11, 12]. Moreover, we have previously reported that MMP-2 production from fibroblasts is induced by emmprin’s first Ig domain (ECI) peptide, when substituted with chitobiose (the disaccharide with which N-glycosylation starts) . Gelatinase A (MMP-2) is the most abundant MMP in the tumor stroma and contributes to tumor invasion [14, 15]. However, to date, emmprin receptors on fibroblasts have not been confirmed, and the detailed mechanism of regulation of MMP-2 production is unknown. Previously, we reported that synthetic peptides carrying a partial ECI construct of emmprin sequence without chitobiose could inhibit emmprin activity. Four peptides were synthesized, each consisting of 20–23 amino acids and corresponding to about one fourth of the full length ECI sequence of the emmprin molecule. Only the second peptide (emp#2), which contains a putative N-glycosylation site sequence, inhibited emmprin-stimulated production of MMP-2 in co-cultures of fibroblasts . As a hypothetical mechanism of emp#2 peptide inhibition of MMP-2 production from fibroblasts, it was proposed that binding of other proteins with emmprin is inhibited in the presence of emp#2 peptide. In other words, our findings suggested the formation of a complex with homophilic or heterophilic cis- or trans- interactions between emmprin molecules within the plasma membrane or between emmprin and an unidentified cell surface molecule(s).
In this study, we identified new molecules which formed a complex with emmprin in co-cultures of tumor cells and fibroblasts and analyzed whether these molecules were involved in the regulation of MMP-2 production.
Epithelioid sarcoma cell line FU-EPS-1 and immortalized human dermal fibroblast ST353i was established in our laboratory [17, 18]. These cell lines were maintained in D-MEM/Ham’s F-12 (Wako, Tokyo, Japan) growth medium, supplemented with 10% fetal calf serum (FCS), streptomycin (50 μg/ml), and penicillin (50 U/ml).
Co-culture experiments were performed as previously described [9, 13, 19] using fibroblasts (ST353i) and tumor cells (FU-EPS-1). Inhibition experiments in co-cultures using anti-CD73 blocking antibody (7G2, Abcam, Cambridge, UK, 1100) were performed as described previously [16, 20]. For the co-culture experiments employing CD73 specific inhibitor adenosine 5′-(α,β-methylene) diphosphate (APCP), co-cultured cells were incubated at 37 °C for 24 h to 7 d with 20, 40 and 80 μM of APCP.
FU-EPS-1 and ST353i were co-cultured at a ratio of 1:1. The conditioned medium was replaced with 0.2% lactalbumin enzymatic hydrolysate containing serum-free medium after three washes with serum-free medium on day 2 and day 5. The cells were incubated with 3 mM BS3 (bis (sulfosuccinimidyl)suberate) solution for 30 min at room temperature. After removing the BS3 solution, free primary amine (10 mM tris pH 7.4, 2 x protein inhibitor) was added for 5 min to stop the cross-linking reaction.
SDS-PAGE and immunoblotting were performed using 4–15% gradient gel (Bio-Rad, Hercules, CA) and antibodies against emmprin (mouse monoclonal, R&D System, Flanders, NJ), anti-CD73 (rabbit monoclonal, Cell Signaling, Danvers, MA), MMP-2 (monoclonal antibody, Daiichi Fine Chemical, Toyama, Japan) and MT1-MMP (Millipore, Bedford, MA). After electrophoresis, the proteins were transferred to an immobilon membrane (Millipore). Non-specific sites were blocked using 5% dry fat milk in Tris Buffered Saline (TBS) at 37 °C for 1 h and the membrane was incubated overnight at 4 °C with the primary antibody. After washing with TBS-T (TBS containing 0.05% tween 20), the membrane was incubated for 1 h with peroxidase-conjugated secondary antibody. Color was developed with chemiluminescence reagents according to the manufacturer’s instructions (PerkinElmer,Waltham,MA).
Cell lysates (1 ml) were incubated with 2 μg anti-emmprin antibody (goat polyclonal antibody, R&D System), or anti-CD73 antibody (rabbit monoclonal, Cell Signaling) for 3 h. Protein G Sepharose (GE Healthcare, Chicago, Illinois) was used to immunoprecipitate proteins linked to the primary antibody overnight at 4 °C. After five washes with lysis buffer, bound proteins were eluted with 2 x sample buffer and subjected to immunoblotting.
Mass spectrometric analysis
Following complexation of emmprin with membrane proteins using cross-linker (BS3) in FU-EPS-1 cells, the lysates were subjected to immunoprecipitation and immunoblotting. Three bands detected in co-culture of tumor cells and fibroblasts (75–100 kDa, 100–140 kDa, and 220 kDa) and a single band (220 kDa) detected in tumor cells alone were analyzed (Additional file 1: Figure. S1A and B). Mass spectrometry (MS) analysis was performed using the excised sample (identical molecular weight bands which reacted with anti-emmprin antibody in BS3 treated and BS3 non-treated samples). Shotgun proteomic analyses were performed using a linear ion trap-orbitrap mass spectrometer (LTQ-Orbitrap Velos, Thermo Fisher Scientific, Waltham MA) coupled with a nanoflow LC system (Dina-2A, KYA Technologies, Tokyo, Japan) as previously described . Proteins were identified by querying MS and MS/MS data against the RefSeq (National Center for Biotechnology Information) human protein database using Mascot (Matrix Science, London, UK).
Transient knockdown using RNA interference
Small interference RNA (siRNA) sequences were used for transient knockdown of CD99, CD73 and emmprin mRNA. siRNAs targeting the CD99 (Invitrogen, Carlsbad, CA), CD73 (Invitrogen) and emmprin (Invitrogen, Carlsbad, CA) were transfected in FU-EPS cells using Lipofectamine transfection reagent in Opti-MEM media (Thermo Fisher Scientific) in the absence of serum and antibiotics according to the manufacturer’s instructions. Knockdown of CD99, CD73 and emmprin expression in cells was analyzed by immunoblotting. Each siRNA was transfected into fibroblasts and tumor cells separately, and transfected cells were used for co-culture experiments.
Enzyme-linked immunosorbent assay (ELISA)
After 48 h of incubation, conditioned media were collected from ST353i alone or FU-EPS-1 and ST353i co-cultures and the expression of MMP-2 was measured using the Total MMP-2 Quantikine ELISA Kit (sensitivity: 0.082 ng/ml) according to the manufacturer’s specifications (R&D Systems, Minneapolis, MN). All assays were performed in triplicate and statistical analyses were performed using Student’s t-test.
Gelatinolytic activities in conditioned media were demonstrated using gelatin as a substrate, as described previously [19, 20]. We measured the enzymatic activity of pro-MMP-2 using the commercially available Gelatin-Zymography kit (Cat.No.AK47-COS, Cosmo Bio, Tokyo) by following the protocols supplied by the manufacturer. The enzyme activity was detected as a clear band on the resulting blue background of undigested gelatin.
In situ proximity ligation assay (PLA)
In situ PLA was used to assess protein-protein interactions. Cells grown on 8-well culture slides (Nunc Lab-Tek chamber Slide System, ThermoFisher scientific), were immediately fixed and subjected to in situ PLA using the Duolink detection kit (Sigma-Aldrich, St. Louis, MO) according to the manufacturer’s instructions and described previously . After blocking, slides were incubated with mouse anti-Emmprin (1:100; R&D) or rabbit anti-CD73 (1:100; GeneTex, Irvine, CA) primary antibodies. For isotype controls, the primary antibody was substituted with either mouse (emmprin) or rabbit (CD73) IgG. To analyze the results, we used a Duolink ImageTool to obtain objective quantification of PLA signals.
Four-micrometer thick paraffin-embedded sections were deparaffinized and heated in a microwave oven (700 watts) for 10 min to expose antigens in a 10 mM Na-citrate buffer, followed by blocking in 3% hydrogen peroxide solution. The sections were incubated with anti-emmprin (mouse monoclonal, R&D, 1:200) and CD73 (rabbit polyclonal, Abcam, 1:100) antibodies at 4 °C overnight. The exposed antigen was detected using EnVision reagent conjugated horseradish peroxidase (Agilent, Santa Clara, CA). The reaction was identified with DAB and counterstained with Mayer’s hematoxylin. The staining results were evaluated semi-quantitatively by two independent observers. Immunostaining was considered negative if the percentage of stained tumor cells or stromal cells was < 10%. In specimens considered positive, staining of the cells was quantified on a scale of 1–4 based on the percentage of positive cells. The scale was structured as follows: 1+, 10–25% of cells positive; 2+, 25–50% of cells positive; 3+, 50–75% of the cells positive; 4+, > 75% of the cells positive. CD73-close and CD73-distant staining score were means of three high power fields of view. CD73-close included stromal cells positioned close to the tumor cells, and CD73-distant were those most distant from the tumor cells in the same slides.
Tumor cells and fibroblasts grown on 8-well culture slides for 48 h were fixed in 4% paraformaldehyde at 4 °C for 30 min, and permeabilized with 0.1% Triton X-100 in phosphate-buffered saline. After blocking with Image-iT FX Signal Enhancer (Invitrogen) for 1 h at room temperature, the slides were incubated with anti-emmprin (mouse monoclonal, R&D; 1:100) and anti-CD73 (rabbit polyclonal, abcam; 1:100) antibodies at 4 °C overnight. Secondary antibodies and mounting medium were anti-mouse IgG Alexa594 (Invitrogen), anti-rabbit IgG Alexa488 (Invitrogen), and Mounting Medium with DAPI (Abcam). Images of cells were captured using an Biozero BZ-8000 (Keyence, Osaka, Japan).
Quantitative data are presented as mean ± standard deviation (SD) and were analyzed using the Student’s t test. A p value < 0.01 was considered indicative of statistical significance.
The study material comprised 10 epithelioid sarcoma samples from two males and eight female patients (age range: 22–81 y; mean: 61 y) obtained from the soft tissue tumor file of the Department of Pathology, Fukuoka University Hospital, between 1995 and 2015. Use of anonymous and redundant tissue is part of the standard treatment agreement with patients in our hospital when no objection has been expressed.
Identification of molecules which form a complex with emmprin by MS analysis
Both CD73 and CD99 form a complex with emmprin
CD73 and CD99 are expressed in both tumor cells and fibroblasts
CD73 regulates the production of MMP-2 from co-cultures of sarcoma cells with fibroblasts
CD73 is co-expressed and co-localized with emmprin
A PLA signal is detected as a red dot when the recognition sites of two antibodies are in close proximity, making it possible to verify whether two proteins are co-localized. In our case, PLA experiments were performed using antibodies for CD73 and emmprin. With fibroblasts alone, there was no observable signal (Fig. 5e). A PLA signal was detected with tumor cells alone (Fig. 5f) and under co-culture conditions (Fig. 5g). Furthermore, a much stronger signal was obtained under co-culture conditions than with tumor cells alone. When CD73 was suppressed by siRNA, PLA signals were inhibited (Fig. 5h). The PLA signals were quantified by using Image Tool analysis, which confirmed that the signal intensity was clearly decreased by CD73 siRNA vs. control siRNA (Additional file 3: Figure. S3).
These results show that in tumor cells or in co-cultured fibroblasts and tumor cells, CD73 and emmprin are co-expressed and co-localized.
CD73 suppresses MMP-2 production via non-enzymatic activity
Co-expression of emmprin and CD73 in vivo
In this study, we have for the first time reported that CD73 forms a complex with emmprin and regulates the production of MMP-2 from fibroblasts, although there has been a previous report providing evidence that suppression of CD73 leads to down-regulation of MMP-2 . Findings of CD73 complexation with emmprin were confirmed by immunoprecipitation, double immunofluorescence staining, and proximity ligation assay. CD73 siRNA and CD73 neutralizing antibody were used to confirm that CD73 regulates MMP-2 production in co-cultures of sarcoma cells with fibroblasts.
Emmprin acts as a receptor for cyclophilins, S100A9 and platelet glycoprotein VI [10, 27], and also associates tightly with monocarboxylate transporters [7, 10, 28, 29]. Emmprin forms complexes with GLUT1, CD44 and CD98 on the cell membrane [10, 30]. Caveolin-1 [31, 32], MT1-MMP/MMP-14 , AnxA2 [34, 35], and integrins  have also been identified as binding partners in interactions with emmprin, and are thereby suggested to be involved in the regulation of MMP activity. However, it is possible that there are other emmprin binding partners which have not hitherto been reported.
We performed MS analysis to identify the proteins that bound with emmprin when tumor cells were co-cultured with fibroblasts. CD73 and CD99, detected by overlap in MS analyses of three different co-culture experiments, were selected and each protein was further analyzed (Fig. 1). The fact that these proteins were detected as conserved proteins in the analysis of three different co-culture experiments indicated that CD73 and CD99 were the proteins that bound most reproducibly with emmprin under conditions of co-culture of tumor cells and fibroblasts. Indeed, CD73 and CD99 were expressed abundantly both in fibroblasts and in tumor cells (Fig. 3).
CD73 is also known as ecto-5′-nucleotidase (ecto5′-NT). It is mainly present on cell membranes and functions as an enzyme which converts AMP to adenosine; it also has non-enzymatic activity [24, 25]. CD73 has been shown to be expressed in tumor cells and in the stroma of colorectal cancer [37, 38], prostate cancer [39, 40], gastric cancer , breast cancer , ovarian cancer  and squamous cell carcinoma . Whether high expression of CD73 in cancer predicts poor prognosis or favorable prognosis is controversial, with published reports supporting both outcomes [25, 37, 38, 39, 40, 42, 44]. According to our literature review, the expression of CD73 in sarcoma and stromal cells of sarcoma has not been previously reported. In this study, APCP, an inhibitor of CD73’s enzymatic activity did not inhibit MMP-2 production from co-culture cells from one (data not shown) to 7 days in a concentration of 20–80 μM (Fig. 6a). These results signify that inhibition of CD73 enzymatic activity does not effect MMP-2 production from the co-culture cells. These results support the possibility of the importance of emmprin-CD73 interaction that is needed for MMP-2 production, in that CD73 inhibition by siRNA and neutralizing antibody induced the suppression of MMP-2 production from co-cultured cells (Fig. 6b). The possibility cannot be denied that CD73 neutralizing antibody may disturb the interaction of CD73 and emmprin.
In this study, we analyzed the role of CD73 in cancer using an epithelioid sarcoma cell line and in vivo specimens taken from cases of epithelioid sarcoma, which is a highly aggressive non-epithelial neoplasm. Epithelioid sarcoma is a very rare tumor; however, the tumor cells exhibit an epithelioid morphology. It is therefore relatively easy to distinguish tumor cells from stromal cells in epithelioid sarcoma specimens histologically (Fig. 7). In these in vivo cases, emmprin was expressed in tumor cells and CD73 expression was observed in tumor cells as well as in fibroblasts (Fig. 7, Additional file 6: Table S1). CD73 was more highly expressed in the stromal fibroblasts close to the tumor cells, although the difference was not statistically significant (Additional file 6: Table S1, Additional file 4: Figure. S4). As we used biopsy samples, stromal CD73 expression of CD73-close and -distant were examined in the same samples: this was because, during the biopsies, no tissue suitable for use as a control were collected. It is necessary to examines CD73 expression in fibroblasts more distant from the tumor cells in the other tumor specimens. CD73 is expressed in both tumor cells and fibroblasts and emmprin is mainly expressed in tumor cells as shown in in vitro (Figs. 3 and 5) and in vivo (Fig. 7). Co-expression of emmprin and CD73 were observed mainly in tumor cells by fluorescent immunostaining (arrows in Figs. 5 and 7). These results suggest that CD73 and emmprin form a complex at the adhesion site of the tumor cells and adjacent fibroblasts. These findings also suggest the possibility of tumor-stromal interaction associated with CD73 in stromal cells.
CD99 is a transmembrane glycoprotein of 25 kDa and 32 kDa, coded by the gene MIC2 in the short arms of the X and Y chromosomes. CD99 is a marker of Ewing sarcoma and primitive neuroectodermal tumors. Its expression has been confirmed in malignant tumors of hematopoietic neoplasm, synovial sarcoma, solitary fibrous tumors, testicular, ovarian sex cord tumors, and breast cancer [45, 46]. Ours is the first study to demonstrate formation of a complex between CD99 and emmprin. While the CD99-emmprin complex was detected, inhibition of CD99 expression did not suppress MMP-2 production by fibroblasts. The precise function of the CD99 and emmprin complex is, as yet, unclear. Both CD99 and CD73 were expressed in tumor cells as well as in fibroblasts, although, more abundant expression was observed in fibroblasts than in tumor cells (Fig. 3). Previously, it has been reported that the expression of CD99 in lung cancer tumor stroma is associated with poor prognosis , which, together with our observation of CD99 with emmprin, suggests the possibility that CD99 also contributes to an interaction between tumor and stroma.
Expression of emmprin has been confirmed in many malignant tumors, and it has been reported to be associated with the progression of cancer . In some sarcomas, a correlation between emmprin expression in sarcoma cells and prognosis has also been reported [18, 19, 20]. Therapeutic options are limited, especially with advanced sarcoma compared to epithelial neoplasms. Our study demonstrated a new insight into the mechanism of sarcoma invasion, that may possibly become a target of therapy.
This study had the following limitations. We demonstrated the complex formation and co-expression of CD73 and emmprin; further, we showed that CD73 was associated with the regulation of MMP-2 production in co-cultures of sarcoma cells with fibroblasts. However, we did not demonstrate directly whether the complex formation of CD73 and emmprin is essential for the regulation of MMP-2 production in co-cultures of sarcoma cells with fibroblasts. We performed this study with the sarcoma cell line FU-EPS-1. However, the finding that emmprin-CD73 interaction regulates MMP-2 production in co-cultures of tumor cells with fibroblasts has also been confirmed using squamous cell carcinoma cell lines obtained from head and neck tumor (data not shown). The function of CD99, which also forms a complex with emmprin, likewise remains unsolved. Clarification of these unclear points will be addressed in further detailed studies.
In conclusion, using a combination of in vitro co-culture assays, mass spectrometry, and biological/molecular analysis, we demonstrated that CD73 forms a complex with emmprin to regulate the production of MMP-2 from fibroblasts. In vivo studies performed in tumor tissue sections confirmed colocalization of CD73 and emmprin. Inhibition of CD73 by either a neutralizing antibody or an siRNA, suppressed MMP-2 production from fibroblasts. Our study provides a novel insight into the role of CD73 in emmprin-mediated cancer invasion and metastasis.
The authors thank Ms. M. Onitsuka and Ms. H. Fukagawa, Department of Pathology, Fukuoka University School of Medicine and Hospital for their excellent technical assistance.
MA and KN substantial contributions to the conception or design of the work. MA, KN and MH interpreted the data and edited the manuscript. MA and MM contributed to data acquisition. NK, MO, HKH and MS contributed to data analysis. KN and KK contributed to data interpretation for all analyses. MA, KN, MM and BPT have drafted the work or substantively revised it. KN and BPT supervised the whole process and gave constructive advice. All authors have read and approved the manuscript.
No specific funding is received for this study. This work was supported in part by grants from the Research Center for Advanced Molecular Medicine, Fukuoka University. The grants had no role in the design of the study and collection, analysis, and interpretation of data and in writing the manuscript.
Ethics approval and consent to participate
Anonymous use of redundant tissues is part of the standard treatment agreement with patients at the study hospital when no objection is expressed. The Fukuoka University Hospital Institutional Ethics Committee approved the study protocol (approval number 12–7-13), which allowed a waiver of informed consent for the study.
Consent for publication
The authors declare that they have no competing interest related to this work.
- 17.Nishio J, Iwasaki H, Nabeshima K, Ishiguro M, Naumann S, Isayama T, Naito M, Kaneko Y, Kikuchi M, Bridge JA. Establishment of a new human epithelioid sarcoma cell line, FU-EPS-1: molecular cytogenetic characterization by use of spectral karyotyping and comparative genomic hybridization. Int J Oncol. 2005;27(2):361–9.PubMedGoogle Scholar
- 20.Sameshima T, Nabeshima K, Toole BP, Yokogami K, Okada Y, Goya T, Koono M, Wakisaka S. Glioma cell extracellular matrix metalloproteinase inducer (EMMPRIN) (CD147) stimulates production of membrane-type matrix metalloproteinases and activated gelatinase a in co-cultures with brain-derived fibroblasts. Cancer Lett. 2000;157(2):177–84.CrossRefGoogle Scholar
- 23.Overall CM, Sodek J. Concanavalin a produces a matrix-degradative phenotype in human fibroblasts. Induction and endogenous activation of collagenase, 72-kDa gelatinase, and Pump-1 is accompanied by the suppression of the tissue inhibitor of matrix metalloproteinases. J Biol Chem. 1990;265(34):21141–51.PubMedGoogle Scholar
- 24.Morandini AC, Savio LE, Coutinho-Silva R. The role of P2X7 receptor in infectious inflammatory diseases and the influence of ectonucleotidases. Biom J. 2014;37(4):169–77.Google Scholar
- 46.Byun HJ, Hong IK, Kim E, Jin YJ, Jeoung DI, Hahn JH, Kim YM, Park SH, Lee H. A splice variant of CD99 increases motility and MMP-9 expression of human breast cancer cells through the AKT-, ERK-, and JNK-dependent AP-1 activation signaling pathways. J Biol Chem. 2006;281(46):34833–47.CrossRefGoogle Scholar
Open AccessThis 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.