- Primary research
- Open Access
Differential effects of HIF-α isoforms on apoptosis in renal carcinoma cell lines
© Doonachar et al.; licensee BioMed Central. 2015
- Received: 17 November 2014
- Accepted: 10 February 2015
- Published: 22 February 2015
Germline mutations in the von Hippel-Lindau (VHL) tumor suppressor gene predispose individuals to clear cell renal carcinomas, hemangioblastomas, and pheochromocytomas. The VHL gene product forms an ubiquitin E3 ligase complex, with regulation of hypoxia-inducible factor alpha (HIF-α) as its best known function. Lack of VHL expression has been shown previously to sensitize renal cells to apoptosis caused by certain cellular stresses. In this report, the role of HIF-α in apoptosis was investigated using two parent VHL-null renal carcinoma cell lines.
786-O and RCC10 renal carcinoma cell lines with manipulated levels of VHL, HIF-1α, or HIF-2α were subjected to cellular stresses and analyzed by western blotting for the abundance of apoptotic markers.
Cell lines expressing mutant VHL proteins that were unable to regulate HIF-α had increased levels of apoptosis when irradiated with ultraviolet (UV) light. The influences of the two major isoforms of HIF-α, HIF-1α and HIF-2α, on apoptosis, were compared by creating cell lines in which levels of each isoform were modulated via short hairpin RNA interference. In UV-irradiated cells, HIF-2α expression was determined to promote apoptosis, whereas HIF-1α was anti-apoptotic. In cells deprived of either glucose or serum, HIF-1α expression was generally anti-apoptotic, while HIF-2α expression was observed to either promote apoptosis or have less of an influence on apoptosis, depending on the cell line used.
HIF-1α and HIF-2α exerted distinct effects in each of the conditions tested, with expression of HIF-1α largely blocking apoptosis and HIF-2α generally promoting apoptosis. These results reinforce that HIF-1α and HIF-2α have distinct biological roles and that their relative expression levels may influence some therapeutic interventions that are dependent on apoptosis.
- Glucose starvation
- Serum starvation
Inactivation of the von Hippel-Lindau (VHL) tumor suppressor gene occurs in VHL disease, an inherited cancer syndrome that predisposes affected individuals to a number of benign and malignant tumors that may affect various organs such as the kidneys, the central nervous system, the retina, and the pancreas . The VHL protein (pVHL) is a part of an ubiquitin E3 ligase complex that targets proteins for proteolysis [2,3]. One important target of pVHL ubiquitination is hypoxia-inducible factor alpha (HIF-α), a subunit of the transcription factor HIF, which up-regulates hypoxia inducible genes [4-6]. Renal cell carcinoma is particularly dependent on loss of functional VHL and the ensuing up-regulation of HIF-α and HIF transcriptional targets .
Although loss of VHL can lead to up-regulation of both HIF-1α and HIF-2α isoforms, there are some functional differences that exist between these two isoforms. While there are many common transcriptional targets, it appears that each HIF-α isoform has some distinct genes that it can transactivate . For example, HIF-1α, but not HIF-2α can direct expression of glycolytic genes  and HIF-2α, but not HIF-1α can up-regulate expression of cyclin D1, transforming growth factor alpha (TGF-α) and vascular endothelial growth factor VEGF [8,10]. Through opposing functional interactions with the cMyc oncoprotein, HIF-1α has been seen to slow down the cell cycle and tumor growth, whereas HIF-2α has been seen to promote cell proliferation . In renal carcinoma, loss of VHL can lead to sustained up-regulation of both HIF-1α and HIF-2α isoforms in some tumors or only HIF-2α in other tumors, but not solely HIF-1α . Accordingly, it is not surprising that HIF-2α, but not HIF-1α, has been seen as the major driver of renal tumorigenesis [7,13].
Previous studies have shown that VHL expression can protect renal cells against some pro-apoptotic cellular stresses. Reintroduction of pVHL into VHL-deficient cells leads to protection from the cytotoxic effects of serum withdrawal, glucose deprivation, impaired protein processing, chemical hypoxia, and UV radiation [14-17]. It is likely that VHL exerts some of these effects through regulation of HIF-α, although it is presently unclear whether regulation of HIF-1α or HIF-2α is more important for this effect. While some differences in the regulation of pro- and anti-apoptotic genes by HIF-1α and HIF-2α have been described [8,9,18], it has not been fully elucidated whether there are distinct influences of HIF-1α versus HIF-2α on apoptosis due to various stimuli. Here, we carry out a set of experiments to investigate whether expression levels of HIF-1α and HIF-2α have differential effects on apoptosis in renal cells subjected to certain stresses.
Apoptosis in UV-treated 786-O cells with mutant VHL proteins correlates to HIF-2α levels
Reduction of HIF-2α levels in 786-O provides partial protection from UV-mediated apoptosis
To extend these analyses and determine whether the contribution of HIF-2α toward apoptosis was also observed with other forms of cellular stress, similar analyses were performed following conditions of glucose starvation (Figure 3C) and serum starvation (Figure 3D). Unlike with UV-treatment, reduction of HIF-2α levels had little effect on apoptosis caused by these stresses, whereas VHL was seen to exert a protective effect. Thus, in 786-O cells, VHL’s cell survival effects under glucose and serum starvation conditions are likely to occur independently of HIF-2α regulation.
Reduction of HIF-1α and HIF-2α levels leads to differential effects on apoptosis in RCC10 cells
The RCC10 cell lines were subjected to UV treatment (Figure 4B). Similar to what was observed in 786-O, RCC10 cells with lowered levels of HIF-2α had decreased abundance of cleaved PARP and cleaved caspase-3, comparable to the levels seen in cells with reintroduced VHL. Interestingly, knockdown of HIF-1α caused an increase in cleaved PARP and caspase-3 as compared to control cells. Thus, in UV treated RCC10 cells, HIF-2α expression may promote apoptosis, whereas HIF-1α expression appears to inhibit it.
In VHL-null cells expressing mutant VHL proteins, UV-mediated apoptosis was correlated with type I mutations that had high levels of HIF-α. The distinct contributions of HIF-1α versus HIF-2α toward apoptosis were then investigated using VHL-deficient renal carcinoma cell lines and manipulation of HIF-1α and HIF-2α expression. Reduction of HIF-2α levels led to decreased apoptosis with UV treatment. This result implies that the expression of HIF-2α is pro-apoptotic in UV treated cells. In UV-treated RCC10 cells, reduction of HIF-1α levels had the opposite effect, increasing apoptosis, suggesting that the expression of HIF-1α is normally anti-apoptotic. Therefore, there appears to be distinct effects of the two HIF-α isoforms on apoptosis in UV-treated cells. HIF-2α promotion of apoptosis may be a more important determinant than HIF-1α with UV treatment because RCC10 cells that have up-regulated levels of both HIF-1α and HIF-2α were susceptible to UV-initiated apoptosis (see Figure 2). Alternatively, it is possible that there is a greater abundance of HIF-2α than HIF-1α in RCC10 cells that is responsible for this effect, although this cannot be discerned using available antibodies to these isoforms.
The influences of HIF-1α and HIF-2α expression on apoptosis caused by additional stresses were also determined. For serum and glucose starved cells, knockdown of HIF-2α had little effect in 786-O cells, but decreased apoptosis in RCC10 cells. One important difference between these cell lines is that 786-O cells express only HIF-2α, whereas RCC10 cells express both HIF-α isoforms. Notably, knockdown of HIF-1α in RCC10 cells led to increased apoptosis in glucose-starved cells indicating that HIF-1α is anti-apoptotic for this stress. Moreover, there was an apparent increase of HIF-1α in RCC10 HIF-2α knockdown cells that coincided with reduced apoptosis. Thus, it is possible that the anti-apoptotic effects of HIF-1α are responsible for at least part of the reduction in apoptosis seen in HIF-2α shRNA cells upon glucose or serum withdrawal. This notion would also provide an explanation for the apparent lack of effect of HIF-2α shRNA in glucose and serum starved 786-O cells. However, regardless of the exact mechanism, these findings indicate overall differential influences on apoptosis by the HIF-α isoforms under serum and glucose starvation conditions, similar to UV irradiation.
While the same general trends were observed using 786-O and RCC10 cells, there were some slight differences in apoptosis (additional to the one mentioned above) that may relate to the HIF-α isoforms in these cells. For example, with 786-O cells, serum starvation caused the smallest induction of apoptosis out of the three stimuli used here, with longer exposures on western blots needed to visualize the cleaved apoptotic markers (data not shown). This effect was not observed with RCC10 cells. 786-O cells were in general more susceptible to apoptosis due to glucose starvation, with incubations past 24 hours resulting in cells that detached from the culture dishes, albeit less so for cells containing VHL (data not shown), whereas incubations of RCC10 cells in glucose-free media were possible for over 48 hours. While the lack of expression of the anti-apoptotic HIF-1α may be invoked for the latter finding, it is also likely that 786-O cells are more malignant than RCC10 (as seen by shorter doubling time for 786-O cells, data not shown), which may also result in a greater need for glucose and increased resistance to serum withdrawal. Of note, these observations also highlight the notion that different apoptotic regulators and/or pathways are likely to be engaged for the different apoptotic stresses. Note that although the apoptosis reported here may occur by different pathways, these pathways all converge on activation of caspases. Caspase-3, used as a marker in this report, is a central player in the apoptotic cascade, functioning as an effector caspase . Cleavage of the other apoptosis marker used here, PARP, is downstream of caspase activation (reviewed in ), although PARP can also be cleaved by caspase-7 . In the majority of assays here, PARP and caspase cleavage coincided, adding strength to their validity.
There are likely many factors that may account for the opposite effects of HIF-1α and HIF-2α on apoptosis observed in this study. Differences in the genes that HIF-1α and HIF-2α transactivate or repress are more likely to be involved. For example, HIF-1α, but not HIF-2α, promotes expression of glycolytic genes, which may cause cells either to be more efficient at glucose utilization or allow them to utilize other molecules as energy sources, resulting in decreased apoptosis in glucose starvation conditions, in agreement with the present findings. Other differential transcriptional targets involved in apoptosis include BNip3, which has been reported to be up-regulated by HIF-1α [8,25], cIAP, which may be up-regulated by HIF-2α , and ARC (Apoptosis Repressor with a Card Domain), which is up-regulated by HIF-1α , to name a few. In addition, some of the differential effects of HIF-1α and HIF-2α on apoptosis may be related to non-transcriptional effects, perhaps through differences in protein interactions. Opposite effects of HIF-1α and HIF-2α on mdm2 binding and p53 transcriptional activity have been noted (reviewed in ). Different roles of HIF-1α and HIF-2α in regulation of mTOR, which is involved in cellular survival decisions, have also been proposed (reviewed in ). Promotion of c-myc activity by HIF-2α and not HIF-1α , which can potentiate apoptosis in some circumstances, may also be a contributing factor and is in agreement with the promotion of apoptosis by HIF-2α observed in this report. Given the complexity of HIF-1α and HIF-2α biology, it is very likely the underlying causes for the differences in apoptosis as a result of HIF-1α or HIF-2α expression observed here are multifactorial. Further studies to explore these mechanisms are indicated, especially since HIF-1α and HIF-2α’s effects on apoptosis are likely to have implications toward clinical interventions that depend on apoptosis as their mode of action.
Cell lines and cell culture
293T and 786-O cells were obtained from the American Type Culture Collection. RCC10 renal carcinoma cells were generously provided by Miguel Esteban (Imperial College, London). All cells were grown in Dulbecco’s Modified Eagle’s Medium (DMEM) containing either 10% Serum Supreme (BioWhitaker) or 10% fetal calf serum (FCS). Media was supplemented with penicillin-streptomycin (100 U/ml and 100 μg/ml, respectively).
Retroviral expression vectors, retroviral infection and cell lines
Wild-type and mutant VHLp30 retroviral expression constructs have been previously described . Two retroviral vectors directing expression of shRNA targeting HIF-1α were created in pSuperRetro as described , using CTGATGACCAGCAACTTGA as one 19-mer HIF-1α target sequence and GCCACTTCGAAGTAGTGCT as another target sequence. Two retroviral shRNA vectors targeting HIF-2α, also pSuperRetro-based , were generously provided by Dr. William Kaelin (Dana Farber Cancer Center).
To produce retroviral supernatants, the recombinant retroviral vectors were co-transfected with the retroviral packing plasmid, pCL-Ampho , into 293T cells, as previously described [19,21]. To create stable pools of retrovirally-infected cells, cells were incubated overnight in a mixture (1:1) of retroviral supernatants and fresh medium that was supplemented with polybrene (10 μg/ml). The mixture was removed and replaced with fresh media the next day. For the HIF-1α and HIF-2α shRNA retroviruses, a pool of the two retroviruses for each shRNA target was used (with equal amounts of each retroviral supernatant). Three days following infection, the cells were replated and incubated for 10 to 14 days in media containing puromycin (0.5 μg/ml) to allow for proper selection.
786-O cells stably infected with control or HIF-2α shRNA retroviruses or stably transfected with VHLp19 have been described previously [19,21]. RCC10 cells stably infected with wild-type and mutant VHLp30 constructs have been described previously .
Apoptotic cell stresses
Cell stresses were applied to cells grown at confluency. For UV treatment, media was aspirated from culture dishes, which were then placed under the germicidal UV lamp (UVC, 254 nm wavelength) in a tissue culture hood for 20 seconds (approximately 60 J/m2). Media was replaced immediately after UV exposure and the cells were incubated for 24 hours prior to lysis. For glucose and serum starvation, the media was removed from culture dishes and the cells were rinsed with phosphate buffered saline (PBS). The PBS was removed and cells were incubated with either glucose-free DMEM (Gibco/Life Technologies, Grand Island, NY) supplemented with 10% FCS (for glucose starvation) or DMEM lacking FCS (for serum starvation). Cells were grown for 24-96 hours (depending on cell type and treatment), after which they were lysed.
Cells grown on 60-mm culture dishes until confluent were rinsed with PBS and then lysed by incubating with 150 μl of lysis buffer (50 mM HEPES (pH 7.6), 250 mM NaCl, 0.5% Nonidet P-40, 0.5% Triton X-100, 5 mM EDTA, 1 mM phenylmethylsulfonyl fluoride (PMSF), 1 mM Na2VO3 and 2 μg/ml each of aprotinin, bestatin, and leupeptin) at 4°C for 30 minutes. Lysed cells were scraped with a plastic scraper, resuspended by pipetting, collected, and were spun down in a refrigerated microcentrifuge for 15 minutes to remove all insoluble material. The supernatant was collected and a Bradford protein assay (Bio-Rad, Hercules, CA) was performed to determine protein concentrations. Equal amounts of protein for each well of a gel, ranging from 25-50 μg among the different blots performed, were mixed with an equivalent volume of 2× SDS buffer and were separated by SDS-PAGE. The separated proteins were then transferred to a polyvinylidene difluoride (PVDF) membrane overnight at 30 volts for 16 hours and western blotting was then performed as described.
VHL mAb 11E12, which has been previously described , HIF-2α rabbit antibody (Novus Biological, Littleton, CO), GLUT-1 rabbit antibody (Alpha Diagnostic, San Antonio, TX), and caspase-3 and Bcl-2 rabbit antibodies (Cell Signaling, Danvers, MA) were used at a 1:1000 dilution in western blots. HIF-1α mouse antibody (BD Biosciences, Franklin Lakes, NJ) was used at a 1:250 dilution. PARP-1/2 rabbit antibody (Santa Cruz Biotechnology, Santa Cruz, CA) was used at a 1:200 dilution in western blots. Anti-mouse IgG-HRP and anti-rabbit IgG-HRP secondary antibodies (Southern Biotech, Birmingham AL) were used at a 1:2286 dilution.
HIF-1α and HIF-2α exerted contrasting effects in all conditions tested. In UV-irradiated cells, HIF-2α expression promoted apoptosis, whereas HIF-1α was anti-apoptotic. In glucose or serum starved cells, HIF-1α expression was largely anti-apoptotic, while HIF-2α expression generally promoted apoptosis or had less of an influence, depending on the cell line used. These results reinforce that HIF-1α and HIF-2α have distinct biological roles. Moreover, their relative expression levels may influence some therapeutic interventions that are dependent on apoptosis and thus should be taken into account.
Research reported in this publication was supported by the National Cancer Institute of the National Institutes of Health [award number R15CA121992]. The authors would like to thank Anna Michalik and Rita Burmanroy for their help with certain experiments.
- Latif F, Tory K, Gnarra J, Yao M, Duh FM, Orcutt ML, et al. Identification of the von Hippel-Lindau disease tumor suppressor gene. Science. 1993;260(5112):1317–20.View ArticlePubMedGoogle Scholar
- Iwai K, Yamanaka K, Kamura T, Minato N, Conaway RC, Conaway JW, et al. Identification of the von Hippel-Lindau tumor-suppressor protein as part of an active E3 ubiquitin ligase complex. Proc Natl Acad Sci U S A. 1999;96(22):12436–41.View ArticlePubMed CentralPubMedGoogle Scholar
- Lisztwan J, Imbert G, Wirbelauer C, Gstaiger M, Krek W. The von Hippel-Lindau tumor suppressor protein is a component of an E3 ubiquitin-protein ligase activity. Genes Dev. 1999;13:1822–33.View ArticlePubMed CentralPubMedGoogle Scholar
- Maxwell PH, Wiesener MS, Chang GW, Clifford SC, Vaux EC, Cockman ME, et al. The tumour suppressor protein VHL targets hypoxia-inducible factors for oxygen-dependent proteolysis. Nature. 1999;399(6733):271–5.View ArticlePubMedGoogle Scholar
- Cockman ME, Masson N, Mole DR, Jaakkola P, Chang GW, Clifford SC, et al. Hypoxia inducible factor-alpha binding and ubiquitylation by the von Hippel-Lindau tumor suppressor protein. J Biol Chem. 2000;275(33):25733–41.View ArticlePubMedGoogle Scholar
- Ohh M, Park CW, Ivan M, Hoffman MA, Kim TY, Huang LE, et al. Ubiquitination of hypoxia-inducible factor requires direct binding to the beta-domain of the von Hippel-Lindau protein. Nat Cell Biol. 2000;2(7):423–7.View ArticlePubMedGoogle Scholar
- Kondo K, Kim WY, Lechpammer M, Kaelin Jr WG. Inhibition of HIF2alpha is sufficient to suppress pVHL-defective tumor growth. PLoS Biol. 2003;1(3):E83.View ArticlePubMed CentralPubMedGoogle Scholar
- Raval RR, Lau KW, Tran MG, Sowter HM, Mandriota SJ, Li JL, et al. Contrasting properties of hypoxia-inducible factor 1 (HIF-1) and HIF-2 in von Hippel-Lindau-associated renal cell carcinoma. Mol Cell Biol. 2005;25(13):5675–86.View ArticlePubMed CentralPubMedGoogle Scholar
- Hu CJ, Wang LY, Chodosh LA, Keith B, Simon MC. Differential roles of hypoxia-inducible factor 1alpha (HIF-1alpha) and HIF-2alpha in hypoxic gene regulation. Mol Cell Biol. 2003;23(24):9361–74.View ArticlePubMed CentralPubMedGoogle Scholar
- Covello KL, Simon MC, Keith B. Targeted replacement of hypoxia-inducible factor-1alpha by a hypoxia-inducible factor-2alpha knock-in allele promotes tumor growth. Cancer Res. 2005;65(6):2277–86.View ArticlePubMedGoogle Scholar
- Gordan JD, Bertout JA, Hu CJ, Diehl JA, Simon MC. HIF-2alpha promotes hypoxic cell proliferation by enhancing c-myc transcriptional activity. Cancer Cell. 2007;11(4):335–47.View ArticlePubMed CentralPubMedGoogle Scholar
- Gordan JD, Lal P, Dondeti VR, Letrero R, Parekh KN, Oquendo CE, et al. HIF-alpha effects on c-Myc distinguish two subtypes of sporadic VHL-deficient clear cell renal carcinoma. Cancer Cell. 2008;14(6):435–46.View ArticlePubMed CentralPubMedGoogle Scholar
- Maranchie JK, Vasselli JR, Riss J, Bonifacino JS, Linehan WM, Klausner RD. The contribution of VHL substrate binding and HIF1-alpha to the phenotype of VHL loss in renal cell carcinoma. Cancer Cell. 2002;1(3):247–55.View ArticlePubMedGoogle Scholar
- Pause A, Lee S, Lonergan KM, Klausner RD. The von Hippel-Lindau tumor suppressor gene is required for cell cycle exit upon serum withdrawal. Proc Natl Acad Sci U S A. 1998;95(3):993–8.View ArticlePubMed CentralPubMedGoogle Scholar
- Gorospe M, Egan JM, Zbar B, Lerman M, Geil L, Kuzmin I, et al. Protective function of von Hippel-Lindau protein against impaired protein processing in renal carcinoma cells. Mol Cell Biol. 1999;19(2):1289–300.PubMed CentralPubMedGoogle Scholar
- Schoenfeld AR, Parris T, Eisenberger A, Davidowitz EJ, De Leon M, Talasazan F, et al. The von Hippel-Lindau tumor suppressor gene protects cells from UV-mediated apoptosis. Oncogene. 2000;19(51):5851–7.View ArticlePubMedGoogle Scholar
- Devarajan P, De Leon M, Talasazan F, Schoenfeld AR, Davidowitz EJ, Burk RD. The von Hippel-Lindau gene product inhibits renal cell apoptosis via Bcl-2-dependent pathways. J Biol Chem. 2001;276(44):40599–605.View ArticlePubMedGoogle Scholar
- Razorenova OV, Castellini L, Colavitti R, Edgington LE, Nicolau M, Huang X, et al. The apoptosis repressor with a CARD domain (ARC) gene is a direct hypoxia-inducible factor 1 target gene and promotes survival and proliferation of VHL-deficient renal cancer cells. Mol Cell Biol. 2014;34(4):739–51.View ArticlePubMed CentralPubMedGoogle Scholar
- Bangiyeva V, Rosenbloom A, Alexander A, Isanova B, Popko T, Schoenfeld A. Differences in regulation of tight junctions and cell morphology between VHL mutations from disease subtypes. BMC Cancer. 2009;9(1):229.View ArticlePubMed CentralPubMedGoogle Scholar
- Krieg M, Haas R, Brauch H, Acker T, Flamme I, Plate KH. Up-regulation of hypoxia-inducible factors HIF-1alpha and HIF-2alpha under normoxic conditions in renal carcinoma cells by von Hippel-Lindau tumor suppressor gene loss of function. Oncogene. 2000;19(48):5435–43.View ArticlePubMedGoogle Scholar
- Hughes MD, Kapllani E, Alexander AE, Burk RD, Schoenfeld AR. HIF-2alpha downregulation in the absence of functional VHL is not sufficient for renal cell differentiation. Cancer Cell Int. 2007;7:13.View ArticlePubMed CentralPubMedGoogle Scholar
- Slee EA, Adrain C, Martin SJ. Serial killers: ordering caspase activation events in apoptosis. Cell Death Differ. 1999;6(11):1067–74.View ArticlePubMedGoogle Scholar
- Elmore S. Apoptosis: a review of programmed cell death. Toxicol Pathol. 2007;35(4):495–516.View ArticlePubMed CentralPubMedGoogle Scholar
- Boucher D, Blais V, Denault JB. Caspase-7 uses an exosite to promote poly (ADP ribose) polymerase 1 proteolysis. Proc Natl Acad Sci U S A. 2012;109(15):5669–74.View ArticlePubMed CentralPubMedGoogle Scholar
- Sowter HM, Ratcliffe PJ, Watson P, Greenberg AH, Harris AL. HIF-1-dependent regulation of hypoxic induction of the cell death factors BNIP3 and NIX in human tumors. Cancer Res. 2001;61(18):6669–73.PubMedGoogle Scholar
- Keith B, Johnson RS, Simon MC. HIF1alpha and HIF2alpha: sibling rivalry in hypoxic tumour growth and progression. Nat Rev Cancer. 2012;12(1):9–22.Google Scholar
- Brummelkamp TR, Bernards R, Agami R. Stable suppression of tumorigenicity by virus-mediated RNA interference. Cancer Cell. 2002;2(3):243–7.View ArticlePubMedGoogle Scholar
- Naviaux RK, Costanzi E, Haas M, Verma IM. The pCL vector system: rapid production of helper-free, high-titer, recombinant retroviruses. J Virol. 1996;70(8):5701–5.PubMed CentralPubMedGoogle Scholar
- Schoenfeld A, Davidowitz EJ, Burk RD. A second major native von Hippel-Lindau gene product, initiated from an internal translation start site, functions as a tumor suppressor. Proc Natl Acad Sci U S A. 1998;95(15):8817–22.View ArticlePubMed CentralPubMedGoogle Scholar
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. 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.