Post-transcriptional RNA Regulons Affecting Cell Cycle and Proliferation

download Post-transcriptional RNA Regulons Affecting Cell Cycle and Proliferation

of 11

Transcript of Post-transcriptional RNA Regulons Affecting Cell Cycle and Proliferation

  • 8/9/2019 Post-transcriptional RNA Regulons Affecting Cell Cycle and Proliferation

    1/11

    http://localhost/var/www/apps/conversion/tmp/scratch_1/dx.doi.org/10.1016/j.semcdb.2014.05.014mailto:[email protected]://www.elsevier.com/locate/semcdbhttp://www.sciencedirect.com/science/journal/10849521http://localhost/var/www/apps/conversion/tmp/scratch_1/dx.doi.org/10.1016/j.semcdb.2014.05.014http://localhost/var/www/apps/conversion/tmp/scratch_1/dx.doi.org/10.1016/j.semcdb.2014.05.014
  • 8/9/2019 Post-transcriptional RNA Regulons Affecting Cell Cycle and Proliferation

    2/11

    Please citethisarticle inpressas:Blackinton JG,KeeneJD.Post-transcriptional RNAregulonsaffectingcell cycle andproliferation.Semin

    Cell DevBiol (2014), http://dx.doi.org/10.1016/j.semcdb.2014.05.014

    ARTICLE IN PRESSG Model

    YSCDB-1605; No.of Pages11

    2 J.G. Blackinton, J.D. Keene / Seminars in Cell & Developmental Biology xxx (2014) xxxxxx

    are to respond to endogenous and exogenous cell signals effi-

    ciently and economically in order to help coordinate production of

    components of biochemical pathways or macromolecular protein

    complexes; for example, biochemical pathways, subunits of cellu-

    lar pumps, organelles, or other machines that function together

    in cellular compartments. RNA regulons consisting of RNAs and

    RNA-bindingproteinshave beenwidelydocumented to coordinate

    post-transcriptional processes in many species; the general topic

    has been broadly reviewed in recent years [317]. Therefore, this

    article will focus onRNAregulons/PTROs identified todate that are

    involved in coordinating the production of proteins regulating cell

    cycle initiation and progression.

    The concept of post-transcriptional RNAoperons and regulons,

    generally referred to as PTROs or RNA regulons, invokes the clas-

    sical bacterial operons discovered by Jacob,Monod andcoworkers

    and Monod [18] as a didactic tool to make it easily understood.

    However, there are significant differences between the classical

    operons and regulons of bacteria and RNA regulons. These differ-

    ences are not based on bacterial versus eukaryotic mechanisms,

    as RNA regulons have recently been described in archaea and

    bacteria [1924]. Instead, they are due to the fundamental differ-

    ences in the organization of DNA and RNA in any cell. While DNA

    genes in bacterialoperons areconstrainedby covalentarchitecture

    and physical location, RNAs are diverse in quantity and location,affording them vast combinatorial opportunities for RNAprotein

    and RNARNA interactions and transiently dynamic regulation of

    mRNAsubsetsencodingfunctionally relatedproteins.This variabil-

    ity in quantityandlocalization ofRNAs affords RNAregulonsmany

    additional properties. First, multiple copies of each mRNA species

    arecapableof joiningmany differentRNAregulonssimultaneously

    based on different recognition elements in the RNA (RREs or USER

    codes). Second, RNA regulons are agile with reversible interac-

    tions between RNA binding proteins (RBPs) and RNAs. Therefore,

    RNA regulons have highly dynamic RNA targeting and the RNPs

    can be remodeled rapidly because the coordinated RNAs them-

    selvesare combinatorial. Third, RNAregulons cangovern localized

    dynamic regulation of functionally related mRNAs. This is partic-

    ularly important in eukaryotes, as cellular compartmentalizationrequires greater control of localizationamong differentorganelles.

    In sum,multiple copies of eachmRNA can use their multiple RREs

    or USERs to join groups of functionally related mRNAs in time

    and space in order to coordinate multiple functions in combina-

    tion. These overlapping mRNA networks have modular coherence

    because the RNPs that form the RNA regulons can be remodeled

    withdifferent functional statesdependingon signal inputs [2529].

    Physically compartmentalized nuclear and cytoplasmic RNP

    regulonsthat coordinate theproductionof functionallyrelatedpro-

    teins have also been termed quasi-genomes to describe their oft

    remote locations in cells [1,2]. However, the combinatorial use of

    functionally related mRNAs, whether physically clustered or not

    is highly plastic in real-time and that plasticity is not available to

    classical operons and regulons in bacteria. Perhaps a better RNAregulon analogy may be that made to music, where each mRNA

    equates to individual notes from each instrument that must har-

    monizein differentcoordinatedcombinations in time andspace to

    produceamelodically orchestratedsymphony.Moreover,imagina-

    tivecomposers canrearrange thenotes andinstruments toproduce

    different musical scores just as RNA regulons provide an avenue

    of evolutionary opportunity for proteins to gain new functions by

    rewiring existing RNA recognition elements or proteins in real-

    time, orevolvingnew interactingpartners[3032]. Agoodexample

    of RNAprotein rewiring have been observed with PUF3 family

    RBPs, with changes in classes of regulatory targets within fungal

    species [31,32] as well as across many other metazoan species

    [30,33,34]. Presumably as a consequence of these myriad advan-

    tages, RNA regulons have evolved to play functionally important

    rolesin nearlyallspeciesacross the threedomainsof life [3]. Indeed,

    the dynamic plasticity of RNA regulons are just beginning to be

    understood [1,2,7,3538].

    Itwas previously assumed that everymRNA species functioned

    independently in the eukaryotic cell and was regulated indepen-

    dently by distinct RBPs. But, as noted above, we now understand

    thatwith few exceptions every mRNA is a member of one ormore

    RNA regulon(s) and most RBPs regulate large numbers of mRNAs

    [3,3942]. This is because each mRNA has multiple copies and

    each one of those copies has the potential to associate with other

    mRNAsthatencodeproteins thatare functionallyrelated[17]. Thus,

    becausemRNAsarethemselves combinatorial,meaningthat differ-

    ent combinations ofmRNAs canform differentRNAregulons, their

    formation depends on the needs of the cell in time and space [2].

    The first studies to indicate that RBPs can target multiple growth

    regulatory mRNAs used in vitro RNA binding methods and Selex

    procedures withrandomRNAs [4347] andwithmRNAs extracted

    from humanbrain [39].

    The first in vivo evidence indicating global coordination by

    endogenous RNA regulons was derived using RNA immunopre-

    cipitation(RIP) followedbymicroarray withmammalianneuronal

    ELAV/Hu proteins known to be involved in cell growth and dif-

    ferentiation [25]. This method was developed to avoid sonication

    or other harsh lysing to maintain physiological interactions, andis therefore entirely different from that used by Mili and Steitz

    when they observed mRNA reassortment [48]. To date, we have

    not observed reassortment when using the original Tenenbaum

    RIP procedure [49]. The Tenenbaum study used both DNA arrays

    and multichannel RNase protection assays to verify that mRNAs

    encodingcell cyclefactorswere among themRNA targetarraysub-

    sets of ELAV/HuB and ELAV/HuR including n-myc, l-myc, b-myc,

    max, and cyclins A2, B1, C, D1 and D2, but not cyclin D3, cyclin

    B2 [25]. (Fig. 1) These findings were consistent with concomitant

    findings by Gorospe and colleagues [50] showing that cyclin A and

    B mRNAs were bound and stabilized by HuR during cell prolifera-

    tion. These findingswere later confirmed in the Gorospe Labusing

    theglobalRIP-chipprocedureofTenenbaumet al.withmicroarrays

    [51].Since then, many global post-transcriptional procedures like

    RIP-chip/seq [25,52], temporal RNA decay and global nuclear run-

    on (reviewed in [13]), polysome profiling, RNA interference, and

    cell synchronization have been used to identify cellular anddevel-

    opmental RNA regulons [315]. RNA regulons have been shown

    to coordinate cellular processes during cellular homeostasis and

    normal replicative events such as cell cycle, as well as during

    developmental programs, biological activations, or environmen-

    tal perturbations. Many RNA regulons have also been discovered

    using genetic approaches while others were observed as dynamic

    patterns of mRNA co-regulation during a regulatory event. The

    trypanosomelifecyclefieldhas employedthis lattermethodexten-

    sively, followed by identifying the key RBPs or global coordinators

    [5,16,28]. Through these techniques, many examples of RNA reg-ulons have emerged over the past decade at nearly every step

    of post-transcriptional RNA processing, including splicing, export,

    translation, RNA stability and localization [6,10].

    An unresolved question, however, iswhether RNAregulons are

    directly or indirectly coupled to one another at each step along

    the path of RNA processing and how mRNA coordination via RNA

    regulons maintains a consistent mRNA target set as the informa-

    tion flows from the chromatin to the ribosome [6]. In other words,

    howismRNAtrafficdirectedthrough theRNAprocessingpathway?

    It would seem efficient for mRNAs encoding functionally related

    proteins to flow together from step to step in a pathway when

    they are destined to function together at the final steps of gene

    expression(Fig.2).Muchof the information regarding thecoupling

    of each of these processes is based upon single mRNAs. However,

    http://localhost/var/www/apps/conversion/tmp/scratch_1/dx.doi.org/10.1016/j.semcdb.2014.05.014http://localhost/var/www/apps/conversion/tmp/scratch_1/dx.doi.org/10.1016/j.semcdb.2014.05.014
  • 8/9/2019 Post-transcriptional RNA Regulons Affecting Cell Cycle and Proliferation

    3/11

    Please citethisarticle inpressas:Blackinton JG,KeeneJD.Post-transcriptional RNAregulonsaffectingcell cycle andproliferation.Semin

    Cell DevBiol (2014), http://dx.doi.org/10.1016/j.semcdb.2014.05.014

    ARTICLE IN PRESSG Model

    YSCDB-1605; No.of Pages11

    J.G. Blackinton, J.D. Keene / Seminars in Cell & Developmental Biology xxx (2014) xxxxxx 3

    Fig. 1. ELAV/Hu RBPs interact with multiplemRNA targets encoding cell cycle reg-

    ulators. Following the RIP procedure, bound mRNA targets were identified using

    a multi component RNAase protection assay. Note that control mRNAs encoding

    ribosomal protein L32 and GAPDH that did not bind to the RBP.

    Reproduced fromTenenbaumetal.,ProcNatlAcadSciU SA97:1408514090 (2000)

    [25].

    evidence for coupling of post-transcriptional interactions globallyis beginning to accumulate, and among the most compelling early

    examplesare fromexperimentswithSaccharomyces cerevisiaeRNA

    regulons.

    2. Global RNA regulons in budding yeast

    Silver and colleagues [53] presented the first solid confir-

    mation of the RNA regulon model in 2003. They demonstrated

    that functionally related mRNAs associated with two yeast RNA

    export factors, Yra1 andMex67 coordinately exportmultiple func-

    tionally related mRNAs to the cytoplasm [53,54]. For example,

    Yra1 associated largely with cell-cycle regulatory factors andpro-

    teinsinvolvedin carbohydratemetabolism,Mex67 associatedwith

    mRNAs encoding membrane proteins, cell-wall components, and

    RNA-regulatory factors (Fig. 3A). This latter set ofMex67 exported

    mRNAs continues to be a consistent finding thatmany RBPs inter-

    actwith mRNAs encoding regulatory RBPs, translation factors, and

    transcription factors. This concept of broad regulation of regula-

    tory factors has been termed regulators of regulators as RNA

    regulons [25,49,55]. While Yra1 associated with approximately

    1000mRNAs, some encoding proteins that repressed S phase and

    induced G1 phase of the cell cycle, and Mex67 associated with

    approximately 1150 mRNAs, some of which encoded membrane

    proteins and translation factors, they both shared349overlapping

    transcriptsencodingcellwallcomponentsandheat-shockproteins.

    This finding of mRNA target overlap was consistent with results

    from Tenenbaum et al. [25] showing that RBPs such as HuR and

    eIF4E each associated with unique as well as overlapping tran-scripts. This would be expected if each mRNA has more than one

    RRE or USER code, thus, demonstrating that networks of mRNAs

    could be modular and overlapping [2]. An exciting aspect of the

    work by Hieronymus and Silver [53] was the connection between

    theYra1mRNA export factorand theAbf1 transcription factor. Not

    only does theAbf1 transcription factor regulate theYra1 gene, but

    the Abf1 mRNA is exported by Yra1 protein. This was confirmed

    genetically suggesting regulatory crosstalk between transcription

    and export.

    A longstanding curiosity has been whether mRNAs exit the

    nucleus randomly as they are passed from the RNA processing

    apparatus through nuclear pores or whether they are organized

    intofunctionally coherentsubpopulations. An earlystudy fromthe

    Cole lab demonstrated mRNA-specific regulation in RNA export.Using yeast mutants of the nucleoporin Rip1p, mRNAs encoding

    certainheatshockproteinswereefficientlyexporteddespiteglobal

    polyA+ mRNA export suppression under stress conditions [56]. A

    Fig.2. Couplingof RNARegulonsfromtranscriptionto translation.A model oftwo distinctRNA processing pathwaysare depictedfor a setofmRNAs.The functionallyrelated

    set of mRNAs in blue are spliced, exported and translated through regulatory events directed by splicing factor A, export factor B and translation factor C in RNA regulons

    that are coupled together through interactionsbetween A, B andC. The functionallyrelated set of mRNAs in redare likewise spliced, exported and translated by D, E and F,

    respectively, through similar coupling. (For interpretation of thereferences to color in this figure legend, thereader is referred to theweb version of thearticle.)

    http://localhost/var/www/apps/conversion/tmp/scratch_1/dx.doi.org/10.1016/j.semcdb.2014.05.014http://localhost/var/www/apps/conversion/tmp/scratch_1/dx.doi.org/10.1016/j.semcdb.2014.05.014
  • 8/9/2019 Post-transcriptional RNA Regulons Affecting Cell Cycle and Proliferation

    4/11

    Please citethisarticle inpressas:Blackinton JG,KeeneJD.Post-transcriptional RNAregulonsaffectingcell cycle andproliferation.Semin

    Cell DevBiol (2014), http://dx.doi.org/10.1016/j.semcdb.2014.05.014

    ARTICLE IN PRESSG Model

    YSCDB-1605; No.of Pages11

    4 J.G. Blackinton, J.D. Keene / Seminars in Cell & Developmental Biology xxx (2014) xxxxxx

    Fig. 3. Early examplesof RNAregulons.(A) CoordinatedmRNA exportshowingtwo yeast RBPexport factorsMex67 andYra1 as they transportgroups of functionallyrelated

    mRNAs out of the nucleus from derived from the data of Hieronymus and Silver (2003) [53]. Reproduced from Keene,Nat Genet 33: 111112 (2003) [54]. (B) Coordinated

    regulation of theyeastPUF3 RBPtransporting nuclear mRNAs that encodemitochondrial proteins tomitochondria fortranslation anduptakefromGerberet al. (2004) [35].

    Reproduced from Keene,Nat RevGenet 8:533543 (2007) [3].

    unique nuclear pore complex containing Rip1p alongwith several

    nucleoporinsexportsheatshockmRNAs tothecytoplasm.Differen-

    tial mRNA export of this type likely involves distinct RNA-binding

    surfaces, although it is possible that some of the Nup proteins

    themselves either possess RNA-binding properties or are coupledwith specific RBPs, possibly coupled to the splicing apparatus [57].

    While several examples of pre-messenger RNA splicing regulons

    andRNAexport regulonshavebeenpublished, demonstrating cou-

    plingof theseRNAregulonsatthesestepsalongthepathwayof RNA

    processing has remained challenging. As discussed below, a sub-

    sequent study from the Silver lab using Drosophila cells [58], also

    showed export of heat-shock factors whose mRNAs are uniquely

    exported like those of Saavedra et al. [56] but have known roles in

    splicingvia thePCIdomain-containingprotein-2that subsequently

    associateswithpolysomes.However,it isnotknownfromthiswork

    whetherthesubsequenteventsdownstreamof theheat-shock spe-

    cific exportmechanism are coordinated as an RNAregulon(s).

    Brown and colleagues discovered that nuclear-encoded mito-

    chondrial mRNAs bound to the PUF3 RNA-binding protein are

    exportedoutof thenucleus tomitochondria,undergoderepression

    and translation, allowing the new functionally related proteins to

    enterthemitochondrion[35]. Derepression of factors alreadyposi-

    tioned for activation at a remote site is an efficient and accurate

    means to coordinate a group function (Fig. 3B). Several subsequentstudies confirmed and extended these findings using genetic and

    biochemicalmethods[5961].Moreover,the Brown laboratoryfol-

    lowed up this initial discovery by identifying a total of 46 RBPs of

    S. cerevisiae that also bound to unique modular subsets of mRNAs

    encodingfunctionallyrelated proteins[62]. Anumberof theseRBPs

    bound similar messages at different stages in the RNA life cycle,

    hinting at a role for coupling of RBPs.

    In addition to these early confirmations of the RNA regulon

    model, many additional distinct biological processes, pathways

    and macromolecular complexes across the three domains of life

    have been shown to be regulated together at the mRNA level,

    and in nearly every case regulatory RBPs involved in coordinat-

    ing thesefunctionswereidentified[3,10,13,63] (seereview articles

    noted above). Below,we focus on cell cycle functions in yeast and

    http://localhost/var/www/apps/conversion/tmp/scratch_1/dx.doi.org/10.1016/j.semcdb.2014.05.014http://localhost/var/www/apps/conversion/tmp/scratch_1/dx.doi.org/10.1016/j.semcdb.2014.05.014
  • 8/9/2019 Post-transcriptional RNA Regulons Affecting Cell Cycle and Proliferation

    5/11

    Please citethisarticle inpressas:Blackinton JG,KeeneJD.Post-transcriptional RNAregulonsaffectingcell cycle andproliferation.Semin

    Cell DevBiol (2014), http://dx.doi.org/10.1016/j.semcdb.2014.05.014

    ARTICLE IN PRESSG Model

    YSCDB-1605; No.of Pages11

    J.G. Blackinton, J.D. Keene / Seminars in Cell & Developmental Biology xxx (2014) xxxxxx 5

    Fig. 4. Coordination ofmRNAs associated with NUPs andRBPs that are subsequently translated into cell cycle regulators. Several RBPs have been genetically demonstrated

    to influence cell cycle events, includingexport factorNUP98and translationfactorsPUM1,DND1 andGLD-1as describedin thetext. Interestingly, theseRBPs viatheirmRNA

    targets actas repressors of cell growthand have been characterized as tumor suppressors.As describedin thetext, loss of function oftheseRBPscauses increasedcellgrowth

    through defects at cell cycle checkpoints.

    mammalian cells that arepotentially coordinatedas RNAregulons

    at the levels of RNAexport and translation/stability.

    3. Nuclear export RNA regulons during the cell cycle

    Many published examples of splicing events functioning as

    RNA regulons have suggested that discrete subsets of mRNAs

    are coordinately spliced, associated with RBP complexes such as

    the exon junction complex and eventually exported to the cyto-

    plasm[6466]. Silverandcoworkers observedupregulationof RNA

    splice variants encoding prodeath functions as directed by the

    ASF/SF2 RBP [64]. Among thekeyalternativelyspliced targetswere

    the apoptosis regulators Bcl2, Mcl1 and caspase-9. Using a high-

    throughputRNAinterferencescreen, theyfound thatmanyof thesecell cycle regulatorsweregeneticallylinkedto themitoticregulator

    aurora kinase A. Overall their data indicated that these alternative

    splicing regulators coordinate amechanistic balance between pro-

    and anti- apoptosis and cell cycle. Thus, the data suggest that cou-

    pling between the cell cycleandalternative splicing is coordinated

    viaoverlapping RNAregulonsandhas implicationsforoncogenesis

    [64]. Figs. 2 and 4 In addition, there aredefinitiveexamplesof very

    early nuclear RNA imprinting such as the exon-junction complex

    that helps determine the final fate of a given mRNA, aswell as the

    fate of functionally related subsets of mRNAs localized for trans-

    lation in the cytoplasm [15,67,68]. Several studies demonstrated

    the association of export proteins with mRNA subsets in organ-

    isms fromyeast toflies tomammals, someofwhichincludemRNAs

    encoding cell cycle and proliferation factors [9,53,54,58,69].

    For example, Guthrie and colleagues [69] demonstrated that

    hnRNP-type RNA export factors Nab2, Nab4 and Npl3 in S. cere-

    visiae each bind to their own unique specific subset of mRNAs.

    Using the RIP-chip procedure, they found that Npl3 preferentially

    binds to ribosomal protein mRNAs, while Nab4 strongly prefers

    mRNAs encoding factors involved in amino acid metabolism. The

    latter observation was consistent with genetic mutants of Nab4

    that also showed a desensitized growthphenotype when exposed

    to aminoacid stress. Aswith other RIP-chip studies, they identified

    a common set ofmotifs: AAAAG prevalent in Nab2 and UAUAUAA

    or ACACTACA in Nab4 mRNA targets. The authors concluded that

    there are multiple routes of mRNA export for functionally related

    mRNA subsets, and hypothesized that Npl3maybe responsible for

    the bulk mRNA export given that many hnRNP proteins associatewith it, but that these balances may shift during stress conditions.

    In addition, the authors suggest coupling of RNA export with 3

    end processing via interactions between Nab2 and polyA-binding

    protein. It was not clear whether the 3 ends of mRNAs encoding

    the functionally related proteinswould be coordinately cleaved in

    a simultaneous or sequential order by cleavage factor 1 in concert

    with their coordinated export to the cytoplasm via these hnRNP

    proteins. In any case, this landmark paper presents evidence that

    theseimportantproteinscanexportdistinctcombinationsofmRNA

    cargos.

    As noted above, using a genome-wide RNA interference screen,

    Silver andcolleagues identified72 factors,somerequired fordiffer-

    ential nuclear export in Drosophila, and examined specific mRNAs

    encoding heat shock factors that are in turn, coupled to polysomes

    http://localhost/var/www/apps/conversion/tmp/scratch_1/dx.doi.org/10.1016/j.semcdb.2014.05.014http://localhost/var/www/apps/conversion/tmp/scratch_1/dx.doi.org/10.1016/j.semcdb.2014.05.014
  • 8/9/2019 Post-transcriptional RNA Regulons Affecting Cell Cycle and Proliferation

    6/11

    Please citethisarticle inpressas:Blackinton JG,KeeneJD.Post-transcriptional RNAregulonsaffectingcell cycle andproliferation.Semin

    Cell DevBiol (2014), http://dx.doi.org/10.1016/j.semcdb.2014.05.014

    ARTICLE IN PRESSG Model

    YSCDB-1605; No.of Pages11

    6 J.G. Blackinton, J.D. Keene / Seminars in Cell & Developmental Biology xxx (2014) xxxxxx

    and translation via a PCI domain protein that associates with RNA

    [58]. These findings, while not global, have strong implications for

    coupling of RNA regulons from splicing to translation.

    4. Nuclear export RNA regulons inmammalian cell cycle

    4.1. eIF4E RNA regulon of cell cycle and proliferation

    Specific nuclear pore components have been shown in mam-malian cells to be unique for export of mRNA subsets encoding

    proteinsregulating cellulargrowth,DNAdamageandcellcycle reg-

    ulation[7076]. For example, thewell-studiedcapbindingprotein,

    eIF4E, is a major regulator of protein synthesis and has mitogenic

    and oncogenic propertiesdiscovered by Sonenberg and colleagues

    and confirmed in many laboratories [15,7780]. The overexpres-

    sion of eIF4E increased Ras activation and it could be reversed by

    expression ofGTPaseactivating protein, a knownnegative antago-

    nist [77]. More recently,Bordenand colleagues reported that eIF4E

    associates in thenucleuswith RNPs that mediate nuclear exportof

    discrete mRNA subsets as a RNA regulon via a CRM1-dependent

    mechanism [70]. The mRNA targets of this RNA export regulon

    were identified by a RIP-chip/seq type immunoprecipitation and

    included several cyclins,mdm2, c-myc andother cell cycle-relatedfactors. ThesemRNAswere found to have a specific stem-loopRRE

    they termed 4E-SE functioning as a USER. The authors examined

    the 3Dstructural featuresof thestem-loop and confirmed itsstruc-

    ture using nuclease probes, and demonstrated that the 4E-SE was

    required to form stable RNP complexes using cyclin D1 as their

    model. They concluded that eIF4E, acting as a RNA regulon, could

    modulate the coordinated expression of these cell cycle proteins

    efficiently, thereby impacting proliferation and cell survival.

    4.2. Nup96 mRNA subsets encoding cell cycle regulators

    A study in the Blobel and Matunis labs revealed that a poly-

    cistronic cDNA encodes a precursor protein containing both the

    Nup98 and Nup96 nucleoporins [81]. Further analysis indicatedthat this precursor protein is cleaved by a protease to yield these

    two nucleoporins that were later found to have related functional

    roles in nucleocytoplasmic transport. Interestingly, both proteins

    have roles in cell cycle, p53 pathways and carcinogenesis inmany

    species [71,73,75,82,83] . In acute myeloid leukemia, a transloca-

    tion creates a fusion protein between Nup98 and a homeodomain

    DNA binding domain that blocks thenormal functioningof Nup96,

    and thereby, disrupting its putative tumor suppressor function as

    discussed below in Section 4.3 from the data of Singer et al. [75].

    Nup98 can shuttle into andout of thenucleus andappears to inter-

    act directly with the chromatin to affect transcription of cell cycle

    genes inDrosophila [73,83]. In any case, bothNup96 andNup98 are

    suggested to function as part of post-transcriptional RNAregulons

    coordinatingtransportofmRNAs encodingcell cycle, developmentand oncogenesis.

    Nup96hasbeen showntohave several importantfunctions and

    is a componentof theNup107160 complex that is cell cycle regu-

    lated. In some cases, each component of the complex hasdifferent

    stoichiometry in the nuclear pore during stages of the cell cycle

    [71]. Case in point, Fontoura and coworkers demonstrated that at

    mitosis the Nup96 subunit is downregulated preferentially as a

    requirement for T cells to exit G1/S phase of the cell cycle and in

    heterozygous mutant Nup96 mouse T cells that express lowered

    levels of Nup96, cell cycle progression increased. Moreover, levels

    of cyclin Awere also downregulated early inmitosis,whilecyclin B

    was increased intheseT cells.In theNup96+/ heterozygousmouse

    cells, decreasedNup96 increased the rate of cell cycle progression

    that correlated with failure to export specific mRNAs that encode

    cell cycleregulators.ThesemRNA targets ofNup96associatedwith

    cell cycledysfunction includedcyclinD3, CDK6andIB thataccu-

    mulated in the cytoplasm while the nuclear/cytoplasmic ratio of

    GAPDH, ICAM-1 and -tubulin did not change when Nup96 levels

    were lowered [71,82]. Thecytoplasmic predominance of cyclinD3

    andCDK6mRNAs resulted in increasedproductionof their proteins

    andpremature activation of S phase.

    While it appears that Nup96 suppresses export of these mRNA

    specific subsets, these investigatorssuggestedthatNup96 couldbe

    a potentialRBP. However, further studies areneededto definehow

    these cell cycle-specific mRNAs are selected from the up-stream

    processing apparatusandmoved through the pipeline. These find-

    ingsledWozniakandGoldfarb[82] to suggest that thesefindingsfit

    thecriteriafor aRNAregulonand speculated thatNup96itself could

    be the transacting RBP that controls the RNA regulon. (Fig. 2) How

    discrete subsets of functionally related mRNAs are able to locate

    andassociatewiththenuclearpore subunits isnot knownbutcould

    involve a coupling of export to thenascent splicing apparatus that

    are themselves function as an early organizational step in the RNA

    regulon pathway of RNA coordination [6,64].

    4.3. Nup98 mRNA subsets encoding cell cycle regulators

    Nup98is another astounding example ofa potentialRNA exportregulon that is involved in cell cycle and p53 regulation in mam-

    maliancells. Privesand coworkers reported thatNup98 is required

    for the expression of the p21 (CDKN1A) oncogenic protein on the

    post-transcriptional levelandtherebyaffectsseveralproteinsin the

    p53 pathway [75]. In an RNA interference screen, Nup98 appeared

    tobe involved in targetingof p53regulated genes.Previous studies

    demonstrateda linkbetweenNup98andleukemiaviaanunrelated

    mechanism[83]. However, it is exceedingly interesting tonotethat

    Nup98andNup96 (discussed above in Section 4.2.) areproducts of

    a cleavage event of a precursor protein [8183]. The p21 regula-

    tion reported in the Prives Lab showed that Nup98 forms an RNP

    complexwithp21mRNAvia its 3 UTRandpreventsdegradationof

    the p21message and can be rescued byblocking the RNA degrada-

    tion exosome system. Using a RIPprocedure, the 3 UTR ofp21wasshown to directly bind to Nup98 in HepG2 cells, and further anal-

    ysis suggested a C-rich RRE thatwas also present in other mRNAs.

    For example, their data based on computational analysis indicated

    that mRNA encoding the 14-3-3 protein, another p53 pathway

    factor, is similarly bound to Nup98 and protected from degrada-

    tion.Moreover, theauthors confirmedusingRIP-seq and qPCR that

    the14-3-3proteinhasa similar C-richRRE forbinding. Theywent

    on to demonstrate that Nup98 is required for cells to undergo p53

    mediated senescence and could partially protect cells from DNA

    damage-mediated apoptosis. Nup98 as well as 14-3-3 levels of

    expression were both decreased in concert followingDNA damage

    as demonstrated usingmurinehepatocellular carcinomas and in a

    cohort of patient samples [75]. One important conclusion is that if

    Nup98 is a tumor suppressor, it would be advantageous for cancercells to inactivateNup98 during their evolution. It will be interest-

    ing to see how the functions of Nup98 are compromised among

    diverse collections of carcinoma or leukemia cells.

    In summary, the findings thatNup98 can function as a RBP that

    binds to a subset of p53 target genes and regulate the stability of

    their mRNAs is both exciting and unique (Fig. 4). It is a poten-

    tially separate oncogenic pathway whereby Nup98 can regulate

    p53 genes as a tumor suppressor. The authors suggest that this

    may representa p53post-transcriptional RNAregulon thatcouples

    with p53 transcriptional regulation to coordinate otherwise nor-

    mal functions whichare potentially dysregulated at several stages

    along the gene expression pathway. It will be interesting to use

    genomewideRIP-chip/seqand PAR-CLIPmethodstodeterminethe

    globalset ofmRNAsthat interactwithNup98undervariousgrowth

    http://localhost/var/www/apps/conversion/tmp/scratch_1/dx.doi.org/10.1016/j.semcdb.2014.05.014http://localhost/var/www/apps/conversion/tmp/scratch_1/dx.doi.org/10.1016/j.semcdb.2014.05.014
  • 8/9/2019 Post-transcriptional RNA Regulons Affecting Cell Cycle and Proliferation

    7/11

    Please citethisarticle inpressas:Blackinton JG,KeeneJD.Post-transcriptional RNAregulonsaffectingcell cycle andproliferation.Semin

    Cell DevBiol (2014), http://dx.doi.org/10.1016/j.semcdb.2014.05.014

    ARTICLE IN PRESSG Model

    YSCDB-1605; No.of Pages11

    J.G. Blackinton, J.D. Keene / Seminars in Cell & Developmental Biology xxx (2014) xxxxxx 7

    conditions.OthermorediverseexamplesofRNAregulonsandRBPs

    affectingcell cycle and DNAdamageresponsesarediscussedin the

    next section.

    5. Coordinated RNA targeting in cell cycle, p53 and DNA

    damage

    5.1. RNA regulons responding to the DNA damage response

    Among the most prevalent mutational events in cancer is that

    of thecheckpoint kinase, ataxia telangiectasiamutated (ATM)[84].

    TheATMresponseactivateswhenDNA undergoesdouble stranded

    breaksby irradiation, oxidative stress or other forms of physical or

    chemical assault causingthe kinaseto phosphorylatea seriesof cell

    cyclecheckpoint regulators includingcheckpoint kinase2 or Chk2

    and otherproteins in thep53pathwayinvolved intheDNAdamage

    response and repair. This response helps tominimize the effects of

    DNA damage by delaying the cell cycle or inducing apoptosis to

    eliminate damaged cells. Several RBPs including HuR, AUF1, TIAR,

    RPL26,nucleolin andothers aredocumented to respond to induced

    stressandDNAdamage[72,74,76,8587], in some cases, becoming

    relocalized to stress granules. The role of stress granules resulting

    fromoxidative stress isbelievedtotemporarilystoreRBPs andtheirRNA cargo in order to protect the mRNAs until the DNA is repaired

    [88]. Forexample, after exposingmammalian cells toUVClight the

    levels of p53 mRNA did not change, but translation of its mRNA

    to protein was increased. The increased production of p53 under

    these conditions was shown to involve the HuR RBP acting on the

    3 UTR of the p53 mRNA. These results were confirmed using RNA

    interference against HuR to demonstrate the expected inhibition

    of p53 production. It is logical to expect thatwhen the synthesis of

    p53protein is altered, thedownstream transcription targets of p53

    would be affected accordingly.

    Gartenhaus and coworkers took advantage of these observa-

    tions to examinewhetherHuRparticipates in the ATMresponseon

    a global level via RNA operons and regulons [29]. The Chk2 kinase

    has been shown to phosphorylate HuR at residues 88, 100, and118, and thereby alter its specificity of binding to the SIRT1 mRNA

    followingoxidative stress induction. A previousmodel was gener-

    ated in the Gorospe lab proposing thatATM and Chk2 cooperate to

    regulatethebindingof specificmRNA subsets byHuR[89]. UsingB-

    lymphocytecelllinesanda RIP-chipribonomicsapproach, followed

    byconfirmationof individual target effects,Gartenhaus andcollab-

    orators concluded thatmodular networksofHuRregulatedtargets

    encoding functionally related proteinsof the p53and other growth

    pathways are coordinated globally in responses to DNA damage

    andstress[29]. Forexample, this study identified numerousmRNA

    targets of HuR that encoded proteins involved in cell proliferation,

    cell cycle, cell arrest, MAP kinase regulation and stress responses.

    Among them was p21, FOXO3, MEK1, MEK2 and DUSP10 whose

    mRNA binding was directly assessed for downstream ATM/Chk2signaling using HuRmutants at Chk2 phosphorylation sites S88A,

    S100A and T118A, as well as testing the effects of Chk2 inhibitors

    in the presence or absence of irradiation. Overall, this network

    of mRNA targets represented an ATM/Chk2 cascade affecting the

    phosphorylation state of HuR thatsubsequently targeted function-

    allyrelatedmRNAsubpopulations encodingfactors involvedin cell

    proliferation and cell cycle regulation.

    These dynamic datasetswere similar to earlier studies inwhich

    the Gartenhaus group compared HuR and AUF1 RBPs and demon-

    strated changes in overlapping functionally related mRNA subsets

    using MCT-1 oncogene-mediated transformation of immortalized

    breast epithelial MCF10A cells [90]. This work demonstrated the

    combinatorial binding of multiple RBPs to some of the same

    mRNAs as suggested in other studies [25,27,35,76], and others),

    but also showed that overlapping mRNA targets of HuR and AUF1

    involved in cell cycle and cell signaling pathways were differen-

    tially expressed during tumorigenesis [90]. In sum, these global

    mRNA targeting studies with ELAV/Hu proteins have consistently

    identified many of the samemRNA targets encoding cell prolifer-

    ation and cell cycle proteins in support of the RNA regulon model

    and many of its dynamic properties (Fig. 2).

    5.2. RNA regulons targeting p53 and DNA damage at translation

    A striking study from the Tofilon laboratory comparing tran-

    scriptional with translational regulation of genes expressed in

    glioblastoma cell lines following gamma irradiation revealed that

    a response at the level of the ribosome determines the changes

    and obviously the differences in protein production [91,92]. The

    initial study compared three independently derived cell glioma

    lines before and after 7 Grays of irradiation using mRNA analy-

    sis by microarray [91]. By quantifying mRNAs that increased or

    decreased in each of these cell lines after irradiation, the authors

    definitively demonstrated that about 500600 RNAs shifted on

    polysome gradients from the inactive to the translationally active

    regions,whileonlyabout100mRNAs changedin thetranscriptome

    (totalmRNA)data.Moreover,comparingthe threecelllinesshowed

    zero overlap among the transcriptome mRNAs that changed, butapproximately 300 RNAs that changed consistently among the

    translatome datasets. These dramatic results are compelling and

    unexpected, and were confirmed and extended in a subsequent

    paper that compared other types of cancer by these criteria

    [92].

    L et al. [91] used western blotting to confirm that the protein

    products of the mRNAs found to increase or decrease their posi-

    tions on polysome gradients corresponded to the observed up or

    down shifts of the mRNAs. As one might expect, the identifica-

    tion of these mRNAs and their encoded proteins showed that they

    were functionally related. For example, the key proteins included

    primarily cell cycle regulators, cell death, repair and cell prolifer-

    ation factors [91]. As with other studies of this type, many DNA

    andRNA regulatory proteinswere included as well reminiscent ofthe regulators of regulators findings. The authors went on to sug-

    gest that these results are consistent with thepost-transcriptional

    RNAregulonmodel,anddiscussed translation factors such aseIF4E

    and4E-binding proteins aswell as RBPs like fragile Xmental retar-

    dation and CUGBP that could be involved in globally coordinating

    these growth regulatorymRNAs.

    5.3. Global translational dynamics during the cell cycle

    While many RNA regulons have been described in the litera-

    ture to date, those coordinating cell cycle and cell proliferation,

    and those implicated in oncogenesis are compelling.Moreover, the

    global role of translational RNA regulons are particularly exciting

    because the coordinated production of functionally related pro-teins is themost important endgame forbiological systems [15]. A

    recent study by Ruggero and colleagues provided strong evidence

    for translational events determining the expression of cell cycle

    progression, lipid metabolism and the DNA damage/repair path-

    ways at each stage of the cell cycle [93]. Several previous reports

    characterizing changes in mRNA populations throughout the cell

    cycle have identified responsive gene sets.

    Toexaminetranslational regulationduringthecell cycle,Stumpf

    and coworkers [93] used ribosomal profiling/occupancy and high-

    throughput sequencing of RNA from synchronized HeLa cells to

    identify mRNAs unique to G1, S phase, and mitosis. The authors

    found that 12% of the transcripts in the cell change ribosome occu-

    pancy (higher or lower)at some stageof thecell cycle. As expected,

    duringmitosis there were fewer changes inmRNA levels in either

    http://localhost/var/www/apps/conversion/tmp/scratch_1/dx.doi.org/10.1016/j.semcdb.2014.05.014http://localhost/var/www/apps/conversion/tmp/scratch_1/dx.doi.org/10.1016/j.semcdb.2014.05.014
  • 8/9/2019 Post-transcriptional RNA Regulons Affecting Cell Cycle and Proliferation

    8/11

    Please citethisarticle inpressas:Blackinton JG,KeeneJD.Post-transcriptional RNAregulonsaffectingcell cycle andproliferation.Semin

    Cell DevBiol (2014), http://dx.doi.org/10.1016/j.semcdb.2014.05.014

    ARTICLE IN PRESSG Model

    YSCDB-1605; No.of Pages11

    8 J.G. Blackinton, J.D. Keene / Seminars in Cell & Developmental Biology xxx (2014) xxxxxx

    G1orSphasestogetheror separately.They identified coregroups

    of mRNAs that change in all phases of the cell cycle. The patterns

    of clustering transcripts with ribosome occupancy were distinctly

    differentthan themRNAs expressedby thetranscriptome, suggest-

    ing to theauthors that the two processes areuncoupledduringthe

    cell cycle.While there were features of the 5 UTR such as GC con-

    tent of certain transcripts noted these were not deemed to be the

    chiefdeterminants of cell cycle translational changes.On theother

    hand, specific mRNAs encoding known regulators of cell cycling

    were identified. These included increases in cyclin CCNE2 mRNA

    translation in G1 and S phases, and decreases in occupancy of the

    E2F1transcriptionfactormRNA, possibly resulting frommicroRNA

    functions.

    The authors noted distinct changes in ribosome occupancy of

    mRNAs encoding mitotic checkpoint factors and strong increase

    in translation of the RICTOR subunit of the mTOR complex 2 from

    G1 to S phase that was shown to be dependent upon its 5 UTR.

    They noted a total of 353 differentially expressedmRNAs between

    the three phases of G1, S and M. There was clear evidence that

    these transcripts encoded functionally related cell cycle regulatory

    proteins. Thus, the authors constructed modular network models

    that coordinate the translation of nuclear transport, metabolism,

    cohesins, condensins andDNA repair components. In sum, this is a

    very interesting study that demonstrates dynamic changes in thetranslatomeduring thecell cycle consistent with our own concept

    of RNAregulons [15].

    5.4. Developmental RNA regulons of p53 and cell cycle pathways

    There have been various examples of RNA regulons found in

    cell lines and several that appear to coordinate various processes

    in tissues and organs[3,10,13]. Some of the clearest examples have

    come from genetic studies in mice and C. elegans. With respect to

    cell cycle regulation in developing systems, we will consider a C.

    elegansGLD1 RBP that coordinates cell cyclemRNAs in developing

    embryos, a mouse Pumilio 1 RBP that coordinates components of

    the p53/MAP kinase/cell cycle pathways during spermatogenesis,

    and theDND1 (deadendhomlog1)RBPthat affects cell cyclearrestand differentiation of testicular teratomas inmouse.

    The C. elegans GLD1 RBP is a member of the STAR family of

    RBPs that are regulators of germ cell development and can func-

    tionas tumor suppressors. GLD1mutantsaffect sex determination,

    mitosis-meiosis decisions and germ cell identity and have phe-

    notypes involving cell division, cell cycle and DNA replication,

    and appear to function as translational repressors. Rafal Ciosk and

    coworkers used RIP-chip to quantify GLD1RBP-mRNA targetbind-

    ing strengths in whole worms and by confirming with cell lines

    to identifymRNAtargets andestablishapproximate bindingaffini-

    ties [94]. Using cellstransfectedwithepitopetaggedGLD1, extracts

    of transgenic adult worms were used for the RIP procedure and

    mRNA targets quantified by microarray. These experiments took

    advantage of the earlier methods that allow one to use cell type-specific RBPs to RIP mRNAs away from endogenous RBPs in the

    same tissues [25,49]. The data were highly reproducible (r=0.96)

    and nontagged controls were used, and the previously demon-

    strated heptanucleotide repeat UACUAAC/U with minor sequence

    variants.

    Approximately,950mRNAtargetswereidentifiedthatencoded

    functionally related proteins that participate in cytokinesis, cell

    division, reproduction, cell cycle and DNA replication in a man-

    ner consistent with the RNA regulon model. Among the targets, it

    seems that GLD1may relay its tumor suppression by translation-

    ally repressing cyclin E mRNA, and possibly B-type cyclins as well

    [95]. Overall, these functional groups were deemed to be consis-

    tentwithmeiotic cell cycleprogression, sexdetermination andcell

    fate specification phenotypes observed in earlier studies [94,96].

    (Fig.4) Notonlywere theauthors able tomonitor dynamic changes

    in individual targets in the embryos, theywere also able toquantify

    apparent binding affinities of the targets in the extracts based on

    bindingmotifsusingfluorescence polarization, and tomatch them

    with the in vitro binding and translational repression of the tar-

    gets byGLD1. Interestingly, a GLD1 PAR-CLIP study in C. elegansby

    Rajewsky and coworkers [97] found an approximate 70% identity

    with theGLD1RIP-chipmRNA targets ofWright et al. [94] and they

    represented thesamefunctionallyrelated proteinsconsistentwith

    theRNAregulon coordination model.

    TheCapel laboratory investigated the functions of DND1 RBP in

    formation of mouse testicular teratomas, and its role in cell cycle

    arrest as well as thedifferentiationof primordial germ cells (PGC).

    The DND1 RBP is necessary for preventing cell death of primor-

    dial germ cells in gonads, as demonstrated using PGC defects in

    mouse embryos with DND1 Termutations [98,99]. These investi-

    gatorsshow that DND1regulatesmitotic arrest inmale PGCcellsat

    the level of translation acting principally on cell cycle genes. They

    hypothesize that the balance of positive and negative cell cycle

    regulators, consistent with the observation that cell cycle arrest

    by DND1, is able to prevent tumor formation. Using the RIP-chip

    procedure with DND1, they identified functionally related mRNA

    subsets that regulate the cell cycle and tumor-suppressor signal-

    ing pathways in mouse germ cells at stage E14.5. In particular,mRNAs encoding p21, p27, Lats2, p53 and Pten, representing cell

    cycle inhibitors were highlyenriched in the RIP-chip study. On the

    other hand, Akt2, Akt3, cyclins E2, cyclinE1 and Akt1 representing

    cellcycleactivatorswerenot enriched(Fig.4).Theauthorsconclude

    that these results support the notion thatDND1regulatescell cycle

    events that areconsistentwith itsestablishedphenotypesas a RNA

    regulon [99].

    Another related example of a cell cycle regulatory RNA regu-

    lon was discovered in the Lin laboratory while investigating loss

    of spermatocytes in the mouse germline [74]. The Pumilio 1 mam-

    malian gene is a member of the PUF RBP family that is known to

    suppress expression ofmRNAs towhichthey bind [30,33,35]. Dur-

    ingspermatogenesis in thetestes anoverwhelminglylargenumber

    of sperm cells is generated by mitosis and must be eliminatedthrough apoptosis to keep the organ healthy. Genetic deletion of

    Pumilio1 inmicecauses increasedactivationofp53-mediatedapo-

    ptosis, resulting in reduction in sperm numbers and infertility. On

    theother hand,deletionofp53in the Pumilio 1 nullmice decreased

    apoptosisandrescuedthistesticularhypotrophyphenotype. In this

    study, RIP-chip studies of Pumilio 1 targeting in testicular lysates

    revealedthatit interactswithmRNAsthatencodep53pathway,cell

    cycle and MAP kinase-signaling components, and represses their

    translation[74]. Usinga variety of cell imaging andmoleculartech-

    niques, the authorsconfirmedthesetargets invitroand invivo, and

    report that a distinct subset of these mRNAs encode key members

    of the p53 and cell cycle pathways, thus, functioning as a Pumilio 1

    RNA operon/regulon (Fig. 4).

    6. Conclusions and common themes

    A theme of this article is the challenging problem of under-

    standinghow each step of the RNA processingpathway is coupled

    physically to the next step, and whether RNA regulons at each

    step are maintained and/or transferred down the pathway. The

    traditional assumption of many biologists in the past has been

    that transcription is the sole coordinator of cellular events and

    that protein production is generally a more passive process. But

    while transcription is the initial process that starts gene expres-

    sion,accumulating evidencesupportsa viewthat transcriptionand

    translation are intertwined and co-dependent [1,6,13,15,100,101] .

    Both processes appear to be coordinated independently, but

    http://localhost/var/www/apps/conversion/tmp/scratch_1/dx.doi.org/10.1016/j.semcdb.2014.05.014http://localhost/var/www/apps/conversion/tmp/scratch_1/dx.doi.org/10.1016/j.semcdb.2014.05.014
  • 8/9/2019 Post-transcriptional RNA Regulons Affecting Cell Cycle and Proliferation

    9/11

    http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0175http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0175http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0175http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0175http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0175http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0175http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0175http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0175http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0175http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0175http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0175http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0175http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0175http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0175http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0175http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0175http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0175http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0175http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0175http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0175http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0175http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0175http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0175http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0175http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0170http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0170http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0170http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0170http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0170http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0170http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0170http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0170http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0170http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0170http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0170http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0170http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0170http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0170http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0170http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0170http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0170http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0170http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0170http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0170http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0170http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0170http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0170http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0170http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0170http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0170http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0170http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0170http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0170http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0170http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0170http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0170http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0170http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0170http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0170http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0170http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0170http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0170http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0165http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0165http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0165http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0165http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0165http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0165http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0165http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0165http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0165http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0165http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0165http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0165http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0165http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0165http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0165http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0165http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0165http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0165http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0165http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0165http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0165http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0165http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0165http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0165http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0165http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0165http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0165http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0165http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0165http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0165http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0165http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0165http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0160http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0160http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0160http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0160http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0160http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0160http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0160http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0160http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0160http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0160http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0160http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0160http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0160http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0160http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0160http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0160http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0160http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0160http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0160http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0160http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0160http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0160http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0160http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0160http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0160http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0155http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0155http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0155http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0155http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0155http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0155http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0155http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0155http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0155http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0155http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0155http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0155http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0155http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0155http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0155http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0155http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0155http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0155http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0155http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0155http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0155http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0155http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0150http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0150http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0150http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0150http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0150http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0150http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0150http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0150http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0150http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0150http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0150http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0150http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0150http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0150http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0150http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0150http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0150http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0150http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0150http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0150http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0150http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0150http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0150http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0150http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0150http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0150http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0150http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0150http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0150http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0150http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0145http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0145http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0145http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0145http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0145http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0145http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0145http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0145http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0145http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0145http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0145http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0145http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0145http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0145http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0145http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0145http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0145http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0145http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0145http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0145http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0145http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0145http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0145http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0145http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0145http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0145http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0145http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0145http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0145http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0140http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0140http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0140http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0140http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0140http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0140http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0140http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0140http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0140http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0140http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0140http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0140http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0140http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0140http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0140http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0140http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0140http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0140http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0140http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0140http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0140http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0140http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0140http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0140http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0140http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0140http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0140http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0140http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0140http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0135http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0135http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0135http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0135http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0135http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0135http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0135http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0135http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0135http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0135http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0135http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0135http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0135http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0135http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0135http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0135http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0135http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0135http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0135http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0135http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0135http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0135http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0130http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0130http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0130http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0130http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0130http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0130http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0130http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0130http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0130http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0130http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0130http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0130http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0130http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0130http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0130http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0130http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0130http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0130http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0130http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0130http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0130http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0130http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0130http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0130http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0130http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0130http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0130http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0130http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0130http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0130http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0130http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0130http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0130http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0125http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0125http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0125http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0125http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0125http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0125http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0125http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0125http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0125http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0125http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0125http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0125http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0125http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0125http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0125http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0125http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0125http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0125http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0125http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0125http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0125http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0125http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0125http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0125http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0125http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0125http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0125http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0125http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0125http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0120http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0120http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0120http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0120http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0120http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0120http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0120http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0120http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0120http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0120http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0120http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0120http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0120http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0120http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0120http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0120http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0120http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0120http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0115http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0115http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0115http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0115http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0115http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0115http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0115http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0115http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0115http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0115http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0115http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0115http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0115http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0115http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0115http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0115http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0115http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0115http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0115http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0115http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0115http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0115http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0115http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0115http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0115http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0110http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0110http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0110http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0110http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0110http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0110http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0110http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0110http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0110http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0110http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0110http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0110http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0110http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0110http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0110http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0110http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0110http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0110http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0110http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0110http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0110http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0110http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0110http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0110http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0110http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0110http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0110http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0105http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0105http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0105http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0105http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0105http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0105http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0105http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0105http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0105http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0105http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0105http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0105http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0105http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0105http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0105http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0105http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0105http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0105http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0105http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0105http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0105http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0105http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0105http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0105http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0105http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0105http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0105http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0105http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0105http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0105http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0105http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0105http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0105http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0105http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0105http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0100http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0100http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0100http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0100http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0100http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0100http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0100http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0100http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0100http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0100http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0100http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0100http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0100http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0100http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0100http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0100http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0100http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0100http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0100http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0100http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0100http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0100http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0100http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0100http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0100http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0100http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0100http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0100http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0100http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0100http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0100http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0100http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0100http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0095http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0095http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0095http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0095http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0095http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0095http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0095http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0095http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0095http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0095http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0095http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0095http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0095http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0095http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0095http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0095http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0095http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0095http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0095http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0095http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0095http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0095http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0095http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0095http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0095http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0095http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0095http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0095http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0095http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0095http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0090http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0090http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0090http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0090http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0090http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0090http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0090http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0090http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0090http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0090http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0090http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0090http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0090http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0090http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0090http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0090http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0090http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0090http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0090http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0090http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0090http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0090http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0090http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0090http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0090http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0090http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0090http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0090http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0090http://localhost/var/www/apps/conversion/tmp/scratch_1/dx.doi.org/10.1093/nsr/nwu004http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0080http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0080http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0080http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0080http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0080http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0080http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0080http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0080http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0080http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0080http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0080http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0080http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0080http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0080http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0080http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0080http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0080http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0080http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0080http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0080http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0080http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0080http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0080http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0080http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0080http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0080http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0080http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0080http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0080http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0080http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0080http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0075http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0075http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0075http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0075http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0075http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0075http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0075http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0075http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0075http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0075http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0075http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0075http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0075http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0075http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0075http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0075http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0075http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0075http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0075http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0075http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0070http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0070http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0070http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0070http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0070http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0070http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0070http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0070http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0070http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0070http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0070http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0070http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0070http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0070http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0070http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0070http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0070http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0070http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0070http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0070http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0070http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0070http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0065http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0065http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0065http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0065http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0065http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0065http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0065http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0065http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0065http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0065http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0065http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0065http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0065http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0065http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0065http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0065http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0065http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0065http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0065http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0065http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0065http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0065http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0065http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0065http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0065http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0065http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0065http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0060http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0060http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0060http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0060http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0060http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0060http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0060http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0060http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0060http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0060http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0060http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0060http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0060http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0060http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0060http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0060http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0055http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0055http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0055http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0055http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0055http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0055http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0055http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0055http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0055http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0055http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0055http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0055http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0055http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0055http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0055http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0055http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0055http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0055http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0055http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0055http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0055http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0055http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0055http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0055http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0055http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0055http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0055http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0050http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0050http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0050http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0050http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0050http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0050http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0050http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0050http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0050http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0050http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0050http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0050http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0050http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0050http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0050http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0050http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0050http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0050http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0050http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0050http://refhub.elsevier.com/S1084-9521(14)00160-8/sbref0045http://refhub.elsevier.com/S1084-9521(14)00160-8/sbre