Pathogenic polyglutamine expansion length correlates with polarity of the flanking sequences
© Kim; licensee BioMed Central Ltd. 2014
Received: 13 June 2014
Accepted: 23 October 2014
Published: 6 November 2014
Polyglutamine (polyQ) repeat expansion within coding sequence of a soluble protein is responsible for eight autosomal-dominant genetic neurodegenerative disorders. These disorders affect cerebellum, striatum, basal ganglia and other brain regions. The pathogenic polyQ-expansion threshold in these proteins varies from 32Q to 54Q. Understanding the reasons for variability in pathogenic polyQ threshold may provide insights into pathogenic mechanisms responsible for development of these disorders.
Here we established a quantitative correlation between the polarity of the flanking sequences and pathogenic polyQ-expansion threshold in this protein family. We introduced an “edge polarity index” (EPI) to quantify polarity effects of the flanking regions and established a strong correlation between EPI index and critical polyQ expansion length in this protein family. Based on this analysis we subdivided polyQ-expanded proteins into 2 groups – with strong and weak dependence of polyQ threshold on EPI index. The main difference between members of the first and the second group is a polarity profile of these proteins outside of polyQ and flanking regions. PolyQ proteins are known substrates for proteasome and most likely mechanistic explanation for the observed correlation is that proteasome may have an impaired ability to process continuous non-polar regions of proteins.
The proposed hypothesis provides a quantitative explanation for variability in pathogenic threshold among polyQ-expansion disorders, which we established to correlate with polarity of flanking regions. To explain these results we propose that proteasome is not efficient in processing continuous non-polar regions of proteins, resulting in release of undigested and partially digested fragments. If supported experimentally, our hypothesis may have wide implications for further understanding the pathogensis of polyglutamine expansion disorders.
Spinal cord, Brain stem
Cerebellum, Basal ganglia
Cerebellum, Brain stem
Cerebellum, Brain stem
Ventral pons, Substantia nigra
Cerebellum, Brain stem, Spinal cord
Calculation of protein polarity profiles and polarity edge indexes
The Zimmerman Polarity index  for each protein was calculated by using ProtScale software package  with the following options: window size: 9, Relative weight of the window edges compared to the window center: 100%, Weight variation model: linear. The polarity edge index was derived by calculating change in polarity on the edges of polyQ sequence induced by the flanking regions. The amino terminal polarity index ( N PI) and carboxy-terminal polarity index ( C PI) are shown in Table 1. The total edge polarity index ( E PI) was calculated as a sum of N PI and C PI indexes.
Primary protein-sequence comparison of polyQ-expanded protein family
Sequence polarity and proteasomal processing of protein
Polarity of polyQ-flanking regions and pathogenic polyQ threshold
The expanded polyQ stretch corresponds to an extended low polarity sequence. As discussed above, we propose that such a sequence is a poor substrate for proteasomal processing. This argument may explain the accumulation of undigested ubiquitinated fragments of polyQ-expanded proteins in cells [23, 24, 31–36]. We further reasoned that the pathogenic threshold of polyQ expansion may be related to potential influence of flanking regions on proteasomal degradation of naked polyQ sequence. Specifically, if polyQ sequence is surrounded by highly polar flanking regions, then these regions can increase the effective polarity on the edges of polyQ stretch, promoting proteasomal processing. On the other hand, if polyQ region is flanked by low polarity regions, then the polarity of polyQ sequence edges remains low, impairing proteasomal processing. This hypothesis predicts that a polyQ sequence embedded within polar flanking sequences must have longer expansion to reach pathogenic threshold than the polyQ sequence embedded within low polarity flanking sequences.
Pathogenic polyQ threshold and influence of protein context
Interestingly, each member of the first group has been reported to be associated with proteasome in biochemical experiments. In some cases association with proteasome occurred via ubiquitinated form of the protein - such as AR [37, 38], Atxn3 [39, 40], or Htt [41, 42]. In some cases association with proteasome did not require ubiquitination, such as for Atxn7 [43, 44] and for TBP [45–47]. In contrast, members of the second group have not been reported to associate with proteasome in biochemical experiments. The only known interaction is proteasomal association of Atxn1 that is mediated by HSP/CHIP [48, 49] and requires partial unfolding of Atxn1 to be initiated. Although correlative, this argument further suggests that the members of the first group are better substrates for proteasomal degradation than the members of the second group.
Effects of Histidine insertion
A unique clinical case provides an indirect support to our hypothesis. A Japanese SCA1 patient was discovered to have an insertion of 2 His residues within polyQ stretch, resulting in sequence Q45HQHQ10. An expected age of disease onset for a typical SCA1-58Q patient is 22 years of age. In contrast, the SCA1-Q45HQHQ10 patient displayed first symptoms of disease at the age of 50 . In addition, the brain stem atrophy of this patient was much milder than expected for a typical SCA1 patient with similar repeat length . What is an explanation for dramatic protective effects of His insertion? Biophysical studies [51–53] and our own crystallographic experiments  suggested that insertion of His has minimal effect on secondary structure of polyQ region. Thus, it is not likely that insertion of 2 His residues disrupted the “toxic conformation” of the 58Q stretch. However, insertion of 2 His residues is expected to introduce a polarity peak within polyQ sequence. Indeed, the polarity profile of Atxn1-55Q2Н (Q45HQHQ10) contains a significant polar peak (Figure 6D). We propose that such polar insertion enhances proteasomal processing of His-containing protein. As a result, “effective” low polarity polyQ region is shortened to approximately 36Q (Figure 6D), which is consistent with the very mild clinical phenotype of this particular patient .
In this paper we established a quantitative correlation between the polarity of the flanking regions and the pathogenic polyQ expansion threshold for the soluble polyQ-containing proteins. The quantitative analysis enabled us to divide soluble polyQ-expanded proteins into 2 groups – with strong and weak dependence of polyQ threshold on the polarity of the flanking regions. The main difference between members of the first group (Htt, Atxn7, AR, TBP, Atxn3) and the second group (Atxn2, Atxn1, ATN1) is related to polarity profile of remainder of these proteins. All members of the first group composed of regular low-high polarity sequences, whereas members of the second group are composed primarily of low polarity sequence regions. PolyQ proteins are known substrates for proteasomal degradation. We analyzed experiments performed with the model proteasomal substrates and concluded that proteasome has impaired ability to process continuous non-polar regions of proteins. We propose that polarity of flanking regions may have an important modulatory effects on ability of proteasome to process continuous polyQ stretches, resulting in accumulation of polyQ-expanded proteins and partially digested protein fragments in cells. Such proteins can then exert “toxic gain of function” effects by interfering with essential neuronal signaling pathways. These ideas need to be tested experimentally. However, indirect support to our hypothesis is provided by partial protective effects of His insertion within polyQ stretch of SCA1 patient, which has a significant effect on polarity profile of polyQ stretch. We propose that such polar insertion can facilitate proteasomal degradation of polyQ-expanded Ataxin 1, which may explain less severe pathology in these patients.
I thank to Leah Taylor for administrative assistance, members of Bezprozvanny and Kim laboratories for helpful discussions and Drs Ilya Bezprozvanny (I.B.) and George De Martino for comments on the paper. MWK is a Young Investigator of the National Ataxia Foundation and supported by the Hereditary Disease Foundation. This work was also supported by the contract with the Russian Ministry of Science 11.G34.31.0056 (I.B), by the Russian Scientific Fund grant 14-25-00024 (IB), and by the NIH grants R01NS074376 and R01NS056224 (I.B.).
- Gusella JF, MacDonald ME: Molecular genetics: unmasking polyglutamine triggers in neurodegenerative disease. Nat Rev Neurosci. 2000, 1: 109-115.View ArticlePubMedGoogle Scholar
- Havel LS, Li S, Li XJ: Nuclear accumulation of polyglutamine disease proteins and neuropathology. Mol Brain. 2009, 2: 21-10.1186/1756-6606-2-21.PubMed CentralView ArticlePubMedGoogle Scholar
- Banfi S, Chung MY, Kwiatkowski TJ, Ranum LP, McCall AE, Chinault AC, Orr HT, Zoghbi HY: Mapping and cloning of the critical region for the spinocerebellar ataxia type 1 gene (SCA1) in a yeast artificial chromosome contig spanning 1.2 Mb. Genomics. 1993, 18: 627-635. 10.1016/S0888-7543(05)80365-9.View ArticlePubMedGoogle Scholar
- Pulst SM, Nechiporuk A, Nechiporuk T, Gispert S, Chen XN, Lopes-Cendes I, Pearlman S, Starkman S, Orozco-Diaz G, Lunkes A, DeJong P, Rouleau GA, Auburger G, Korenberg JR, Figueroa C, Sahba S: Moderate expansion of a normally biallelic trinucleotide repeat in spinocerebellar ataxia type 2. Nat Genet. 1996, 14: 269-276. 10.1038/ng1196-269.View ArticlePubMedGoogle Scholar
- Kawaguchi Y, Okamoto T, Taniwaki M, Aizawa M, Inoue M, Katayama S, Kawakami H, Nakamura S, Nishimura M, Akiguchi I, Kimura J, Narumiya S, Kakizuka A: CAG expansions in a novel gene for Machado-Joseph disease at chromosome 14q32.1. Nat Genet. 1994, 8: 221-228. 10.1038/ng1194-221.View ArticlePubMedGoogle Scholar
- Toru S, Murakoshi T, Ishikawa K, Saegusa H, Fujigasaki H, Uchihara T, Nagayama S, Osanai M, Mizusawa H, Tanabe T: Spinocerebellar ataxia type 6 mutation alters P-type calcium channel function. J Biol Chem. 2000, 275: 10893-10898. 10.1074/jbc.275.15.10893.View ArticlePubMedGoogle Scholar
- David G, Abbas N, Stevanin G, Durr A, Yvert G, Cancel G, Weber C, Imbert G, Saudou F, Antoniou E, Drabkin H, Gemmill R, Giunti P, Benomar A, Wood N, Ruberg M, Agid Y, Mandel JL, Brice A: Cloning of the SCA7 gene reveals a highly unstable CAG repeat expansion. Nat Genet. 1997, 17: 65-70. 10.1038/ng0997-65.View ArticlePubMedGoogle Scholar
- Nakamura K, Jeong SY, Uchihara T, Anno M, Nagashima K, Nagashima T, Ikeda S, Tsuji S, Kanazawa I: SCA17, a novel autosomal dominant cerebellar ataxia caused by an expanded polyglutamine in TATA-binding protein. Hum Mol Genet. 2001, 10: 1441-1448. 10.1093/hmg/10.14.1441.View ArticlePubMedGoogle Scholar
- Tobin AJ, Signer ER: Huntington's disease: the challenge for cell biologists. Trends Cell Biol. 2000, 10: 531-536. 10.1016/S0962-8924(00)01853-5.View ArticlePubMedGoogle Scholar
- Ross CA: Polyglutamine pathogenesis: emergence of unifying mechanisms for Huntington's disease and related disorders. Neuron. 2002, 35: 819-822. 10.1016/S0896-6273(02)00872-3.View ArticlePubMedGoogle Scholar
- Rubinsztein DC: Lessons from animal models of Huntington's disease. Trends Genet. 2002, 18: 202-209. 10.1016/S0168-9525(01)02625-7.View ArticlePubMedGoogle Scholar
- Li SH, Li XJ: Huntingtin-protein interactions and the pathogenesis of Huntington's disease. Trends Genet. 2004, 20: 146-154. 10.1016/j.tig.2004.01.008.View ArticlePubMedGoogle Scholar
- Bezprozvanny I: Calcium signaling and neurodegenerative diseases. Trends Mol Med. 2009, 15: 89-100. 10.1016/j.molmed.2009.01.001.PubMed CentralView ArticlePubMedGoogle Scholar
- Cha JH: Transcriptional signatures in Huntington's disease. Prog Neurobiol. 2007, 83: 228-248. 10.1016/j.pneurobio.2007.03.004.PubMed CentralView ArticlePubMedGoogle Scholar
- Truant R, Atwal RS, Desmond C, Munsie L, Tran T: Huntington's disease: revisiting the aggregation hypothesis in polyglutamine neurodegenerative diseases. FEBS J. 2008, 275: 4252-4262. 10.1111/j.1742-4658.2008.06561.x.View ArticlePubMedGoogle Scholar
- Takahashi T, Katada S, Onodera O: Polyglutamine diseases: where does toxicity come from? what is toxicity? where are we going?. J Mol Cell Biol. 2010, 2: 180-191. 10.1093/jmcb/mjq005.View ArticlePubMedGoogle Scholar
- Pennuto M, Palazzolo I, Poletti A: Post-translational modifications of expanded polyglutamine proteins: impact on neurotoxicity. Hum Mol Genet. 2009, 18: R40-R47. 10.1093/hmg/ddn412.View ArticlePubMedGoogle Scholar
- Zimmerman JM, Eliezer N, Simha R: The characterization of amino acid sequences in proteins by statistical methods. J Theor Biol. 1968, 21: 170-201. 10.1016/0022-5193(68)90069-6.View ArticlePubMedGoogle Scholar
- Gasteiger EHC, Gattiker A, Duvaud S, Wilkins MR, Appel RD, Bairoch A: Protein identification and analysis tools on the ExPASy server. The Proteomics Protocols Handbook. Edited by: John M, Walker JM. 2005, New York City, USA: Humana Press, 571-607.View ArticleGoogle Scholar
- Kim MW, Chelliah Y, Kim SW, Otwinowski Z, Bezprozvanny I: Secondary structure of Huntingtin amino-terminal region. Structure. 2009, 17: 1205-1212. 10.1016/j.str.2009.08.002.PubMed CentralView ArticlePubMedGoogle Scholar
- Kim M: Beta conformation of polyglutamine track revealed by a crystal structure of Huntingtin N-terminal region with insertion of three histidine residues. Prion. 2013, 7: 221-228. 10.4161/pri.23807.PubMed CentralView ArticlePubMedGoogle Scholar
- Raspe M, Gillis J, Krol H, Krom S, Bosch K, van Veen H, Reits E: Mimicking proteasomal release of polyglutamine peptides initiates aggregation and toxicity. J Cell Sci. 2009, 122: 3262-3271. 10.1242/jcs.045567.View ArticlePubMedGoogle Scholar
- Holmberg CI, Staniszewski KE, Mensah KN, Matouschek A, Morimoto RI: Inefficient degradation of truncated polyglutamine proteins by the proteasome. EMBO J. 2004, 23: 4307-4318. 10.1038/sj.emboj.7600426.PubMed CentralView ArticlePubMedGoogle Scholar
- Venkatraman P, Wetzel R, Tanaka M, Nukina N, Goldberg AL: Eukaryotic proteasomes cannot digest polyglutamine sequences and release them during degradation of polyglutamine-containing proteins. Mol Cell. 2004, 14: 95-104. 10.1016/S1097-2765(04)00151-0.View ArticlePubMedGoogle Scholar
- Jana NR, Zemskov EA, Wang G, Nukina N: Altered proteasomal function due to the expression of polyglutamine-expanded truncated N-terminal huntingtin induces apoptosis by caspase activation through mitochondrial cytochrome c release. Hum Mol Genet. 2001, 10: 1049-1059. 10.1093/hmg/10.10.1049.View ArticlePubMedGoogle Scholar
- Ghoda L, van Daalen Wetters T, Macrae M, Ascherman D, Coffino P: Prevention of rapid intracellular degradation of ODC by a carboxyl-terminal truncation. Science. 1989, 243: 1493-1495. 10.1126/science.2928784.View ArticlePubMedGoogle Scholar
- Blake NW, Moghaddam A, Rao P, Kaur A, Glickman R, Cho YG, Marchini A, Haigh T, Johnson RP, Rickinson AB, Wang F: Inhibition of antigen presentation by the glycine/alanine repeat domain is not conserved in simian homologues of Epstein-Barr virus nuclear antigen 1. J Virol. 1999, 73: 7381-7389.PubMed CentralPubMedGoogle Scholar
- Lin L, Ghosh S: A glycine-rich region in NF-kappaB p105 functions as a processing signal for the generation of the p50 subunit. Mol Cell Biol. 1996, 16: 2248-2254.PubMed CentralView ArticlePubMedGoogle Scholar
- Hoyt MA, Zich J, Takeuchi J, Zhang M, Govaerts C, Coffino P: Glycine-alanine repeats impair proper substrate unfolding by the proteasome. EMBO J. 2006, 25: 1720-1729. 10.1038/sj.emboj.7601058.PubMed CentralView ArticlePubMedGoogle Scholar
- Tian L, Holmgren RA, Matouschek A: A conserved processing mechanism regulates the activity of transcription factors Cubitus interruptus and NF-kappaB. Nat Struct Mol Biol. 2005, 12: 1045-1053. 10.1038/nsmb1018.View ArticlePubMedGoogle Scholar
- DiFiglia M, Sapp E, Chase KO, Davies SW, Bates GP, Vonsattel JP, Aronin N: Aggregation of huntingtin in neuronal intranuclear inclusions and dystrophic neurites in brain. Science. 1997, 277: 1990-1993. 10.1126/science.277.5334.1990.View ArticlePubMedGoogle Scholar
- Haacke A, Broadley SA, Boteva R, Tzvetkov N, Hartl FU, Breuer P: Proteolytic cleavage of polyglutamine-expanded ataxin-3 is critical for aggregation and sequestration of non-expanded ataxin-3. Hum Mol Genet. 2006, 15: 555-568. 10.1093/hmg/ddi472.View ArticlePubMedGoogle Scholar
- Landles C, Sathasivam K, Weiss A, Woodman B, Moffitt H, Finkbeiner S, Sun B, Gafni J, Ellerby LM, Trottier Y, Richards WG, Osmand A, Paganetti P, Bates GP: Proteolysis of mutant huntingtin produces an exon 1 fragment that accumulates as an aggregated protein in neuronal nuclei in Huntington disease. J Biol Chem. 2010, 285: 8808-8823. 10.1074/jbc.M109.075028.PubMed CentralView ArticlePubMedGoogle Scholar
- Young JE, Gouw L, Propp S, Sopher BL, Taylor J, Lin A, Hermel E, Logvinova A, Chen SF, Chen S, Bredesen DE, Truant R, Ptacek LJ, La Spada AR, Ellerby LM: Proteolytic cleavage of ataxin-7 by caspase-7 modulates cellular toxicity and transcriptional dysregulation. J Biol Chem. 2007, 282: 30150-30160. 10.1074/jbc.M705265200.View ArticlePubMedGoogle Scholar
- Suzuki Y, Nakayama K, Hashimoto N, Yazawa I: Proteolytic processing regulates pathological accumulation in dentatorubral-pallidoluysian atrophy. FEBS J. 2010, 277: 4873-4887. 10.1111/j.1742-4658.2010.07893.x.View ArticlePubMedGoogle Scholar
- Verhoef LG, Lindsten K, Masucci MG, Dantuma NP: Aggregate formation inhibits proteasomal degradation of polyglutamine proteins. Hum Mol Genet. 2002, 11: 2689-2700. 10.1093/hmg/11.22.2689.View ArticlePubMedGoogle Scholar
- Beitel LK, Elhaji YA, Lumbroso R, Wing SS, Panet-Raymond V, Gottlieb B, Pinsky L, Trifiro MA: Cloning and characterization of an androgen receptor N-terminal-interacting protein with ubiquitin-protein ligase activity. J Mol Endocrinol. 2002, 29: 41-60. 10.1677/jme.0.0290041.View ArticlePubMedGoogle Scholar
- Poukka H, Karvonen U, Janne OA, Palvimo JJ: Covalent modification of the androgen receptor by small ubiquitin-like modifier 1 (SUMO-1). Proc Natl Acad Sci U S A. 2000, 97: 14145-14150. 10.1073/pnas.97.26.14145.PubMed CentralView ArticlePubMedGoogle Scholar
- Burnett B, Li F, Pittman RN: The polyglutamine neurodegenerative protein ataxin-3 binds polyubiquitylated proteins and has ubiquitin protease activity. Hum Mol Genet. 2003, 12: 3195-3205. 10.1093/hmg/ddg344.View ArticlePubMedGoogle Scholar
- Kuhlbrodt K, Janiesch PC, Kevei E, Segref A, Barikbin R, Hoppe T: The Machado-Joseph disease deubiquitylase ATX-3 couples longevity and proteostasis. Nat Cell Biol. 2011, 13: 273-281. 10.1038/ncb2200.View ArticlePubMedGoogle Scholar
- Kalchman MA, Graham RK, Xia G, Koide HB, Hodgson JG, Graham KC, Goldberg YP, Gietz RD, Pickart CM, Hayden MR: Huntingtin is ubiquitinated and interacts with a specific ubiquitin-conjugating enzyme. J Biol Chem. 1996, 271: 19385-19394. 10.1074/jbc.271.32.19385.View ArticlePubMedGoogle Scholar
- Steffan JS, Agrawal N, Pallos J, Rockabrand E, Trotman LC, Slepko N, Illes K, Lukacsovich T, Zhu YZ, Cattaneo E, Pandolfi PP, Thompson LM, Marsh JL: SUMO modification of Huntingtin and Huntington's disease pathology. Science. 2004, 304: 100-104. 10.1126/science.1092194.View ArticlePubMedGoogle Scholar
- Ansorge O, Giunti P, Michalik A, Van Broeckhoven C, Harding B, Wood N, Scaravilli F: Ataxin-7 aggregation and ubiquitination in infantile SCA7 with 180 CAG repeats. Ann Neurol. 2004, 56: 448-452. 10.1002/ana.20230.View ArticlePubMedGoogle Scholar
- Matilla A, Gorbea C, Einum DD, Townsend J, Michalik A, van Broeckhoven C, Jensen CC, Murphy KJ, Ptacek LJ, Fu YH: Association of ataxin-7 with the proteasome subunit S4 of the 19S regulatory complex. Hum Mol Genet. 2001, 10: 2821-2831. 10.1093/hmg/10.24.2821.View ArticlePubMedGoogle Scholar
- Ferrell K, Wilkinson CR, Dubiel W, Gordon C: Regulatory subunit interactions of the 26S proteasome, a complex problem. Trends Biochem Sci. 2000, 25: 83-88. 10.1016/S0968-0004(99)01529-7.View ArticlePubMedGoogle Scholar
- Chew BS, Siew WL, Xiao B, Lehming N: Transcriptional activation requires protection of the TATA-binding protein Tbp1 by the ubiquitin-specific protease Ubp3. Biochem J. 2010, 431: 391-399.View ArticlePubMedGoogle Scholar
- Trachtulec Z, Hamvas RM, Forejt J, Lehrach HR, Vincek V, Klein J: Linkage of TATA-binding protein and proteasome subunit C5 genes in mice and humans reveals synteny conserved between mammals and invertebrates. Genomics. 1997, 44: 1-7. 10.1006/geno.1997.4839.View ArticlePubMedGoogle Scholar
- Al-Ramahi I, Lam YC, Chen HK, de Gouyon B, Zhang M, Perez AM, Branco J, de Haro M, Patterson C, Zoghbi HY, Botas J: CHIP protects from the neurotoxicity of expanded and wild-type ataxin-1 and promotes their ubiquitination and degradation. J Biol Chem. 2006, 281: 26714-26724. 10.1074/jbc.M601603200.View ArticlePubMedGoogle Scholar
- Choi JY, Ryu JH, Kim HS, Park SG, Bae KH, Kang S, Myung PK, Cho S, Park BC, do Lee H: Co-chaperone CHIP promotes aggregation of ataxin-1. Mol Cell Neurosci. 2007, 34: 69-79. 10.1016/j.mcn.2006.10.002.View ArticlePubMedGoogle Scholar
- Matsuyama Z, Izumi Y, Kameyama M, Kawakami H, Nakamura S: The effect of CAT trinucleotide interruptions on the age at onset of spinocerebellar ataxia type 1 (SCA1). J Med Genet. 1999, 36: 546-548.PubMed CentralPubMedGoogle Scholar
- Jayaraman M, Kodali R, Wetzel R: The impact of ataxin-1-like histidine insertions on polyglutamine aggregation. Protein Eng Des Sel. 2009, 22: 469-478. 10.1093/protein/gzp023.PubMed CentralView ArticlePubMedGoogle Scholar
- Sharma D, Sharma S, Pasha S, Brahmachari SK: Peptide models for inherited neurodegenerative disorders: conformation and aggregation properties of long polyglutamine peptides with and without interruptions. FEBS Lett. 1999, 456: 181-185. 10.1016/S0014-5793(99)00933-3.View ArticlePubMedGoogle Scholar
- Sen S, Dash D, Pasha S, Brahmachari SK: Role of histidine interruption in mitigating the pathological effects of long polyglutamine stretches in SCA1: a molecular approach. Protein Sci. 2003, 12: 953-962. 10.1110/ps.0224403.PubMed CentralView ArticlePubMedGoogle Scholar
This article is published under license to BioMed Central Ltd. 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.