- Open Access
Keap1 expression has independent prognostic value in pancreatic adenocarcinomas
© Isohookana et al.; licensee BioMed Central. 2015
- Received: 27 August 2014
- Accepted: 25 March 2015
- Published: 16 April 2015
Oxidative stress and redox-regulating enzymes may potentially accelerate pancreatic carcinogenesis and also affect chemoresistance. Recently major antioxidant response regulator NF-E2-related factor 2 (Nrf2) has been linked to poor prognosis in pancreatic cancer. Nrf2 activity is strictly regulated by oxidative stress sensor Kelch-like ECH-associated protein 1 (Keap1). Oxidative DNA damage can be estimated e.g. by 8-hydroxy-2′-deoxyguanosine (8-OHdG) expression. The aim of this study was to evaluate the expression and possible prognostic role of Keap1 and 8-OHdG in pancreatic cancer.
We assessed immunohistochemically the expression of 8-OHdG and Keap1 in precisely characterized material of 69 pancreatic adenocarcinoma patients.
Nuclear 8-OHdG associated with cytoplasmic Keap1 expression (p = 0.031) and was overexpressed in patients with smaller tumors (p = 0.016) and in tumors without lymph node involvement (p = 0.051). Cytoplasmic 8-OHdG expression associated with higher differentiation (p = 0.023). Cytoplasmic Keap1 immunostaining associated with N0-staging (p = 0.0009) and the absence of distant metastases (p = 0.018). Membranous Keap1 associated with longer relapse-free survival (p = 0.041) and pancreatic cancer-specific survival (median survival 14 vs. 32 months; p = 0.029) and was in multivariate analysis an independent prognostic factor of pancreatic cancer-related death (HR 2.66, 95%CI 1.23-5.75).
Oxidative stress and main redox regulators may participate in pancreatic carcinogenesis and Keap1 appears as a promising prognostic factor in pancreatic cancer. Future studies should also concentrate on potential link between redox regulation and chemoresistance in pancreatic cancer.
The virtual slide(s) for this article can be found here: http://www.diagnosticpathology.diagnomx.eu/vs/4220521801406476
- Antioxidant enzymes
- Oxidative stress
- Pancreatic cancer
Although pancreatic adenocarcinoma is not the one of the most common cancers, it has dismal prognosis, the five year survival being about 5% . At the moment of diagnosis, malignant tissue extends beyond the pancreas in 90% of the cases and metastasis can often be detected . Because of local expansion and early appearance of metastases, tumor is local enough for curative surgical resection only in 10-20% of the cases. Adjuvant chemotherapy with gemcitabine is considered in these cases, with slight improvement in survival .
Reactive oxygen species (ROS) are produced mostly as a consequence of aerobic metabolism in mitochondrial respiratory chain. In addition to their crucial role in signal transduction, ROS production is a cornerstone of radiotherapy and may also participate to gemcitabine resistance via hypoxia-inducible factor 1α- and nuclear factor κB- mediated CXCR4 up-regulation . An expanding amount of literature from in vitro experiments suggests that ROS have a substantial effect on pancreatic carcinogenesis [5-9]. However, compared to most other solid tumors, these data from in vivo studies are rather limited in pancreatic cancer.
By definition, ROS have very limited lifetime, but their interaction with e.g. DNA can be reliably measured with the use of antibodies against specific footprints of oxidative damage. Hydroxyl radical (•OH) is the most reactive ROS with a lifetime of less than 1 nanoseconds and it reacts easily with pyrimidines, purines and chromatin protein resulting in base modifications, genomic instability and alterations in gene expression such as p53 tumor suppressor gene [10,11]. Nucleoside 8-hydroxy-2′-deoxyguanosine (8-OHdG) is a specific end-product of •OH -derived damage in C8 carbonyl group of guanine and it is the most studied marker against oxidative damage in DNA .
Keap1/Nrf2 pathway is the key sensor of intracellular redox status. Under normal redox state, Keap1 (Kelch-like ECH-associated protein 1) is bound to Nrf2 (nuclear factor erythoid-derived 2-like 2) and Nrf2 activity is repressed by degradation in ubiquitin proteasome pathway [13-16]. When exposed to oxidative stress, Keap1 is inactivated and unable to bind Nrf2, which subsequently translocates to nucleus and binds to antioxidant response element (ARE) in DNA together with small Maf proteins [13,15]. This results in induction of several cytoprotective genes, including antioxidant enzymes, drug transporters and drug-metabolizing enzymes [13,15]. The importance of Keap1 is underlined in Keap1-deficient mice, which die postnatally and have a constitutive Nrf2 activation . In vitro suppression of Keap1 in human prostate and non-small cell lung carcinoma cell lines results in an increased Nrf2 activity and further to sensitization to various chemotherapeutic agents and radiotherapy [18,19].
We recently demonstrated that nuclear Nrf2 expression associates with poor survival in pancreatic adenocarcinomas . Partly as a continuation to that study, our specific aim was here to evaluate the role of both Keap1 and 8-OHdG in an independent material of pancreatic adenocarcinomas.
Tumor size (T)
Nodal metastasis (N)
Distant mestastasis (M)
Type of surgery
Other with curative intention
The paraffin-embedded pancreatic lesions were first sectioned on slides of 3 μm thickness and placed on SuperFrostPlus glass (Menzel–Glāser, Germany). The sections were de-paraffinized in xylene and rehydrated in a graded alcohol series and washed in 10 mM phosphate-buffered saline (PBS). To predigest the sections, they were placed in a microwave oven and boiled in 10 mM citric acid monohydrate (pH 6.0) for 12 minutes (8-OHdG) or in Tris-EDTA (pH 9.0) for 15 minutes (Keap1) and cooled for 30 minutes at room temperature. The sections were flooded in 1% hydrogen peroxide in methanol for 15 min (8-OHdG) or in 3% hydrogen peroxide in methanol for 10 min (Keap1) to consume the endogenous peroxide. The sections were incubated 1 hour at +37°C with mouse monoclonal 8-OHdG antibody (Japan Institute for the Control of Aging, Fukuroi, Japan) using 1:50 dilution. After washing with PBS the slides were incubated with biotinylated secondary antibody (Invitrogen Corporation, California, USA) for 20 minutes at room temperature and then, after washes, incubated for 20 minutes at room temperature in streptavidin-peroxidase (Invitrogen Corporation, California, USA). With Keap1, the slides were incubated with a 1:300 primary antibody dilution Keap1 E20 (Santa Cruz Biotechnology, Inc.) overnight at +4°C. The specificity of used antibody Keap1 E20 has been confirmed previously with western blotting [21,22]. Next morning the sections were incubated with Goat Probe (Biocare Medical, LLC, California, USA) for 15 minutes and after that the sections were incubated with Goat-on-Rodent HRP-Polymer (Biocare Medical, LLC, California, USA) for another 15 minutes. Dako Envision peroxidase detection system (Dako K5007) was used as a chromogen in both immunostainings and the sections were then counterstained with haematoxylin and finally mounted with Immu-Mount (Shandon, Pittsburgh, PA, USA).
Negative controls were produced by using the same procedure except that the primary antibodies (8-OHdG and Keap1) were replaced with PBS.
Immunohistochemical stainings were evaluated by experienced histopathogist (KMH). Both 8-OHdG and Keap1 immunostainings were divided to four class according to their intensity in malignant cells: − = no immunoreactivity observed; + = weak immunostaining; ++ = moderate immunostaining; +++ = strong immunostaining.
IBM SPSS Statistics 126.96.36.199 for Windows was applied for statistical analysis. The reported p-values are from 2-sided chi-square tests, except for survival analysis. Survival was analyzed by using Kaplan-Meier curves with log-rank test and only pancreatic cancer-related death was used as an endpoint. Cox regression analysis was used in multivariate analysis. Relapse-free survival was calculated from the operation date to the confirmed date of relapse. T-class was divided in statistical analyses to either T1-T2 or T3-T4 and grade was divided into either grade I or grade II-III. Probability values below 0.05 were considered significant.
Cytoplasmic Keap1 expression associated with the absence of lymph mode metastases (p = 0.0009) and distant metastases (p = 0.018) and non-significantly with high differentiation (grade I) (p = 0.075). Nuclear 8-OHdG immunostaining was overexpressed in patients with T1-T2 tumors (p = 0.016) and almost significantly in those without lymph node involvement (p = 0.051). Strong cytoplasmic 8-OHdG expression associated with grade I tumors (p = 0.023). Nuclear 8-OHdG and cytoplasmic Keap1 expressions associated with each other (p = 0.031).
The present results show that expression of Nrf2’s main regulator, Keap1, is highly significant prognostic factor in pancreatic adenocarcinomas in terms of RFS and PCSS. These data give biological support to our recent study, which demonstrated the association of Nrf2 expression and poor survival .
Continuous activation of proto-oncogene Nrf2 and overexpression of Nrf2-targeted genes such as heme oxygenase-1 (HO-1), glutathione S-transferases and thioredoxin reductase 1 are observed in several solid carcinomas . The alterations in the repressor of Nrf2, Keap1 may activate Nrf2-regulated antioxidant response pathway at least 1) by achievement of somatic Keap1 mutations to disturb Nrf2/Keap1 interaction 2) by epigenetic silencing via methylation of Keap1 resulting in nuclear accumulation of Nrf2 3) by cysteine residual modifications of Keap1 [15,18,23,24]. There are no recognized germ-line mutations of Keap1, but somatic mutations have been reported in various carcinomas . In pancreatic carcinomas only synonymous mutations of Keap1 have been reported  but interestingly in the context of pancreatic cancer biology, Keap1 mutations in other cancers seem to be more considerably prevalent in adenocarcinomas and in patients with smoking history .
According to our knowledge, there is only one previous study evaluating Keap1 expression in pancreatic cancer in humans . The authors reported no associations between Keap1 and clinicopathological parameters, however, material was smaller than ours and no survival analysis was performed. In the present study cytoplasmic Keap1 expression was tightly connected to reduced metastatic tendency, both to lymph nodes and to distant sites. Cytoplasmic Keap1 likely reflects Nrf2-bound complexes, which results in Nrf2’s ubiquitination and inhibition of its function and subsequently ARE and Nrf2 target genes are inactivated. These genes include HO-1, which play a central role in pancreatic cancer metastasis formation by inducing angiogenesis in co-operation with PTEN, which predicts worse survival in several studies [25,27]. The function of membrane-associated Keap1 is not as obvious, but previous immunolocalization study found Keap1 also as a component of cell-cell junctions and also in focal adhesions and the authors suggested that Keap1 may also have actin-binding properties .
Despite of protection against metastatic potential, cytoplasmic Keap1 expression did not have impact to survival curves. Instead, Keap1 expression associated in cell membranes had a significant protective effect on survival in multivariate analysis, with median pancreatic cancer-specific survival more than doubled. The magnitude of Keap1’s protective value may therefore be even more prominent than previously reported Nrf2 expression . There were no long-time survivors in patients with negative membranous Keap1 immunostaining, whereas in those with Keap1 expression there was a plateau in survival curve during five follow-up years. Even more strikingly, in the group having membranous Keap1 expression, 46.3% of patients did not develop a relapse, which can be considered as a substantial proportion among these patients. Rare previous data from other carcinomas shows that in squamous non-small cell lung carcinoma Keap1 expression likewise associated with better survival with HR of 2.09 . In contrast, we and others have previously shown that there is aberrant Keap1 methylation in triple-negative breast cancer (TNBC) and Keap1 overexpression is associated with poor prognosis in TNBC [30,31]. In breast cancer paradoxical connection between Keap1 and worse survival may derive from the sensitive induction of Keap1 in stressed tumors, rather than carcinogenesis promoting features of Keap1 itself as discussed previously .
Association between nuclear 8-OHdG and cytoplasmic Keap1 expressions probably derives from decreased nuclear DNA damage after Keap1 downregulation and resulting activation of Nrf2’s target genes in ARE. Former studies using Comet assay or HPLC-EC have suggested higher levels of oxidative DNA damage in pancreatic tumors than in benign pancreatic tissue , but studies correlating 8-OHdG and clinicopathological parameters or survival are lacking. We found no association between 8-OHdG expression and either RFS or PCSS, but the presence of nuclear 8-OHdG was connected to smaller primary tumor size and the absence of lymph node metastases. Cytoplasmic 8-OHdG also correlated with better differentiation. In nuclei 8-OHdG is considered as damage marker after •OH attacks to guanosine and based on our recent data from transmission electron microscope studies on Hodgkin lymphomas, cytoplasmic 8-OHdG expression is likely to reflect mitochondrial DNA adducts and damage in mRNA, tRNA, or rRNA . 8-OHdG is one of the most applied markers of oxidative stress and its immunohistochemical expression has been connected to poor survival in melanoma, colorectal cancer, diffuse large B-cell lymphoma and in ovarian carcinoma [34-39]. Nevertheless, excessive oxidative stress may lead to growth restriction or ultimately to apoptosis . This has been suggested as a hypothesis why 8-OHdG expression is considerably attenuated in invasive breast carcinomas compared with non-invasive breast lesions and why high 8-OHdG levels have been constantly associated with better prognosis in breast carcinomas [30,38,39]. Furthermore, 8-OHdG expression is dependent from DNA repair enzymes, especially from the function of human 8-oxoguanine glycosylase (hOGG1). Association between more aggressive pancreatic carcinoma characteristics and lower 8-OHdG expression may therefore result from the induction of DNA repair enzymes under oxidative stress, although this hypothesis remains to be confirmed.
In conclusion, this study adds data to the growing knowledge that highlights the importance of Keap1/Nrf2 pathway in human carcinomas. Although we had at least sufficient follow-up and carefully collected clinicopathogical data, sample size should be more considerable in future studies validating the current results. We also recommend the possible association between Keap1/Nrf2-axis and chemoresistance to be assessed in prospective studies.
The study was supported by the Finnish Cancer Society and the Finnish Anti-tuberculosis Association.
- Yadav D, Lowenfels A. The epidemiology of pancreatitis and pancreatic cancer. Gastroenterology. 2013;144:1252–61.View ArticlePubMed CentralPubMedGoogle Scholar
- Boyle P, Levin B. Pancreas Cancer. In: World Cancer Report 2008. International Agency for Research on Cancer. 2008. http://www.iarc.fr/en/publications/pdfs-online/wcr/2008/wcr_2008.pdf. Accessed 7 Apr 2015.
- Oettle H, Post S, Neuhaus P, Gellert K, Langrehr J, Ridwelski K, et al. Adjuvant chemotherapy with gemcitabine vs observation in patients undergoing curative-intent resection of pancreatic cancer: a randomized controlled trial. JAMA. 2007;297:267–77.View ArticlePubMedGoogle Scholar
- Arora S, Bhardwaj A, Singh S, Srivastava SK, McClellan S, Nirodi CS, et al. An undesired effect of chemotherapy: gemcitabine promotes pancreatic cancer cell invasiveness through reactive oxygen species-dependent, nuclear factor κB- and hypoxia-inducible factor 1α-mediated up-regulation of CXCR4. Biol Chem. 2013;288:21197–207.View ArticleGoogle Scholar
- Acharya A, Das I, Chandhok D, Saha T. Redox regulation in cancer: a double-edged sword with therapeutic potential. Oxid Med Cell Longev. 2010;3:23–4.View ArticlePubMed CentralPubMedGoogle Scholar
- Vaquero EC, Edderkaoui M, Pandol SJ, Gukovsky I, Gukovskaya AS. Reactive oxygen species produced by NAD(P)H oxidase inhibit apoptosis in pancreatic cancer cells. J Biol Chem. 2004;279:34643–34654.8.View ArticlePubMedGoogle Scholar
- Masamune A, Shimosegawa T. Signal transduction in pancreatic stellate cells. J Gastroenterol. 2009;44:249–60.View ArticlePubMedGoogle Scholar
- Shimojo Y, Akimoto M, Hisanaga T, Tanaka T, Tajima Y, Honma Y, et al. Attenuation of reactive oxygen species by antioxidants suppresses hypoxia-induced epithelial-mesenchymal transition and metastasis of pancreatic cancer cells. Clin Exp Metastasis. 2013;30:143–54.View ArticlePubMedGoogle Scholar
- Du J, Nelson ES, Simons AL, Olney KE, Moser C, Schrock HE, et al. Regulation of pancreatic cancer growth by superoxide. Mol Carcinog. 2013;52:555–67.View ArticlePubMed CentralPubMedGoogle Scholar
- Valko M, Izakovic M, Mazur M, Rhodes CJ, Telser J. Role of oxygen radicals in DNA damage and cancer incidence. Mol Cell Biochem. 2004;266:37–56.View ArticlePubMedGoogle Scholar
- Karihtala P, Soini Y. Reactive oxygen species and antioxidant mechanisms in human tissues and their relation to malignancies. APMIS. 2007;115:81–103.View ArticlePubMedGoogle Scholar
- Klaunig JE, Kamendulis LM. The role of oxidative stress in carcinogenesis. Annu Rev Pharmacol Toxicol. 2004;44:239–67.View ArticlePubMedGoogle Scholar
- Itoh K, Mimura J, Yamamoto M. Discovery of the negative regulator of Nrf2, Keap1: a historical overview. Antioxid Redox Signal. 2010;13:1665–78.View ArticlePubMedGoogle Scholar
- Kansanen E, Kuosmanen S, Leinonen H, Levonen AL. The Keap1-Nrf2 pathway: mechanisms of activation and dysregulation in cancer. Redox Biol. 2013;1:45–9.View ArticlePubMed CentralPubMedGoogle Scholar
- Villeneuve NF, Lau A, Zhang DD. Regulation of the Nrf2-Keap1 antioxidant response by the ubiquitin proteasome system: an insight into cullin-ring ubiquitin ligases. Antioxid Redox Signal. 2010;13:1699–712.View ArticlePubMed CentralPubMedGoogle Scholar
- Wakabayashi N, Itoh K, Wakabayashi J, Motohashi H, Noda S, Takahashi S, et al. Keap1-null mutation leads to postnatal lethality due to constitutive Nrf2 activation. Nat Genet. 2003;35:238–45.View ArticlePubMedGoogle Scholar
- Zhang P, Singh A, Yegnasubramanian S, Esopi D, Kombairaju P, Bodas M, et al. Loss of Kelch-like ECH-associated protein 1 function in prostate cancer cells causes chemoresistance and radioresistance and promotes tumor growth. Mol Cancer Ther. 2010;9:336–46.View ArticlePubMed CentralPubMedGoogle Scholar
- Zhan L, Zhang H, Zhang Q, Woods CG, Chen Y, Xue P, et al. Regulatory role of KEAP1 and NRF2 in PPARγ expression and chemoresistance in human non-small-cell lung carcinoma cells. Free Radic Biol Med. 2012;53:758–68.View ArticlePubMed CentralPubMedGoogle Scholar
- Soini Y, Eskelinen M, Juvonen P, Kärjä V, Haapasaari K-M, Saarela A, et al. Nuclear Nrf2 expression is related to a poor survival in pancreatic adenocarcinoma. Pathol Res Pract. 2014;210:35–9.View ArticlePubMedGoogle Scholar
- Li CQ, Kim M, Godoy L, Thiantanawat A, Trudel L, Wogan G. Nitric oxide activation of Keap1/nrf2 signaling in human colon carcinoma cells. Med Sci. 2009;106:14547–51.Google Scholar
- Ma J, Cai H, Wu T, Sobhian B, Huo Y, Alcivar A, et al. PALB2 interacts with KEAP1 To promote NRF2 nuclear accumulation and function. Mol Cell Biol. 2012;32:1506–17.View ArticlePubMed CentralPubMedGoogle Scholar
- Hanada N, Takahata T, Zhou Q, Ye X, Sun R, Itoh J, et al. Methylation of the KEAP1 gene promoter region in human colorectal cancer. BMC Cancer. 2012;12:66.View ArticlePubMed CentralPubMedGoogle Scholar
- Adam J, Hatipoglu E, O’Flaherty L, Ternette N, Sahqal N, Lockstone H, et al. Renal cyst formation in Fh1-deficient mice is independent of the Hif/Phd pathway: roles for fumarate in KEAP1 succination and Nrf2 signaling. Cancer Cell. 2011;20:524–37.View ArticlePubMed CentralPubMedGoogle Scholar
- Gañán-Gómez I, Wei Y, Yang H, Boyano-Adánez MC, García-Manero G. Oncogenic functions of the transcription factor Nrf2. Free Radic Biol Med. 2013;65:750–64.View ArticlePubMedGoogle Scholar
- Lister A, Nedjadi T, Kitteringham NR, Campbell F, Costello E, Lloyd B, et al. Nrf2 is overexpressed in pancreatic cancer: implications for cell proliferation and therapy. Mol Cancer. 2011;13:10–37.Google Scholar
- Sunamura M, Duda DG, Ghattas MH, Lozonschi L, Motoi F, Yamauchi J, et al. Heme oxygenase-1 accelerates tumor angiogenesis of human pancreatic cancer. Angiogenesis. 2003;6:15–24.View ArticlePubMedGoogle Scholar
- Velichkova M, Guttman J, Warren C, Eng L, Kline K, Vogl A, et al. A homologue of Drosophila kelch associates with myosin-VIIa in specialized adhesion junctions. Cell Motil Cytoskeleton. 2002;51:147–64.View ArticlePubMedGoogle Scholar
- Solis LM, Behrens C, Dong W, Suraokar M, Ozburn NC, Moran CA, et al. Nrf2 and Keap1 abnormalities in non-small cell lung carcinoma and association with clinicopathologic features. Clin Cancer Res. 2010;16:3743–53.View ArticlePubMed CentralPubMedGoogle Scholar
- Barbano R, Muscarella LA, Pasculli B, Valori VM, Fontana A, Coco M, et al. Aberrant Keap1 methylation in breast cancer and association with clinicopathological features. Epigenetics. 2013;8:105–12.View ArticlePubMed CentralPubMedGoogle Scholar
- Karihtala P, Kauppila S, Soini Y, Jukkola-Vuorinen A. Oxidative stress and counteracting mechanisms in hormone receptor positive, triple-negative and basal-like breast carcinomas. BMC Cancer. 2011;11:262.View ArticlePubMed CentralPubMedGoogle Scholar
- Li D, Firozi PF, Zhang W, Shen J, DiGiovanni J, Lau S, et al. DNA adducts, genetic polymorphisms, and K-ras mutation in human pancreatic cancer. Mutat Res. 2002;513:37–48.View ArticlePubMedGoogle Scholar
- Bur H, Haapasaari K-M, Turpeenniemi-Hujanen T, Kuittinen O, Auvinen P, Marin K, et al. Oxidative stress markers and mitochondrial antioxidant enzyme expression are increased in aggressive Hodgkin lymphomas. Histopathology. 2014; In press.Google Scholar
- Sheridan J, Wang LM, Tosetto M, Sheahan K, Hyland J, Fennelly D, et al. Nuclear oxidative damage correlates with poor survival in colorectal cance. Brit J Cancer. 2009;100:381–8.View ArticlePubMed CentralPubMedGoogle Scholar
- Karihtala P, Soini Y, Vaskivuo L, Bloigu R, Puistola U. DNA adduct 8-hydroxydeoxyguanosine, a novel putative marker of prognostic significance in ovarian carcinoma. Intl J Gynecol Cancer. 2009;19:1047–51.View ArticleGoogle Scholar
- Murtas D, Piras F, Minerba L, Ugalde J, Floris C, Maxia C, et al. Nuclear 8-hydroxy-2’deoxyguanosine as survival biomarker in patients with cutaneous melanoma. Oncol Rep. 2010;23:329–35.PubMedGoogle Scholar
- Soini Y, Haapasaari K-M, Vaarala MH, Turpeenniemi-Hujanen T, Kärjä V, Karihtala P. 8-hydroxydeguanosine and nitrotyrosine are prognostic factors in urinary bladder carcinoma. Int J Clin Exp Pathol. 2011;4:267–75.PubMed CentralPubMedGoogle Scholar
- Pasanen AK, Kuitunen H, Haapasaari K-M, Karihtala P, Kyllönen H, Soini Y, et al. Expression and prognostic evaluation of oxidative stress markers in an immunohistochemical study of B-cell derived lymphomas. Leuk Lymphoma. 2012;53:624–31.View ArticlePubMedGoogle Scholar
- Sova H, Jukkola-Vuorinen A, Puistola U, Kauppila S, Karihtala P. 8-Hydroxydeoxyguanosine: a new potential independent prognostic factor in breast cancer. Brit J Cancer. 2010;102:1018–23.View ArticlePubMed CentralPubMedGoogle Scholar
- Karihtala P, Kauppila S, Puistola U, Jukkola-Vuorinen A. Divergent behavior of oxidative stress markers 8-hydroxydeoxyguanosine (8-OHdG) and 4-hydroxy-2-nonenal (HNE) in breast carcinogenesis. Histopathology. 2011;58:854–62.View ArticlePubMedGoogle 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.