International Journal of Biological Macromolecules 108 (2018) 214–224

Contents lists available at ScienceDirect

International Journal of Biological Macromolecules

j ourna l ho me pa g e: www.elsev ier .com/ locate / i jb iomac

Structure based mimicking of Phthalic acid esters (PAEs) andinhibition of hACMSD, an important enzyme of the tryptophankynurenine metabolism pathway

Neha Singh, Vikram Dalal, Pravindra Kumar ∗

Department of Biotechnology, Indian Institute of Technology Roorkee, 247667, India

a r t i c l e i n f o

Article history:Received 23 October 2017Received in revised form 1 December 2017Accepted 3 December 2017Available online 5 December 2017

Keywords:hACMSDPhthalatesMolecular docking and simulation

a b s t r a c t

Human �-amino-�-carboxymuconate-�-semialdehyde decarboxylase (hACMSD) is a zinc containingamidohydrolase which is a vital enzyme of the kynurenine pathway in tryptophan metabolism. It pre-vents the accumulation of quinolinic acid (QA) and helps in the maintenance of basal Trp-niacin ratio.To assess the structure based inhibitory action of PAEs such as DMP, DEP, DBP, DIBP, DEHP and theirmetabolites, these were docked into the active site cavity of hACMSD. Docking results show that thebinding affinities of PAEs lie in the comparable range (−4.9 kca/mol–7.48 kcal/mol) with Dipicolinic acid(−6.21 kcal/mol), a substrate analogue of hACMSD. PAEs interact with the key residues such as Arg47and Trp191 and lie within the 4 Å vicinity of zinc metal at the active site of hACMSD. Dynamics and sta-bility of the PAEs-hACMSD complexes were determined by performing molecular dynamics simulationsusing GROMACS 5.14. Binding free energy calculations of the PAEs-hACMSD complexes were estimatedby using MMPBSA method. The results emphasize that PAEs can structurally mimic the binding patternof tryptophan metabolites to hACMSD, which further leads to inhibition of its activity and accumulationof the quinolate in the kynurenine pathway of tryptophan metabolism.

© 2017 Elsevier B.V. All rights reserved.

1. Introduction

In the brain, elevated levels of quinolinic acid (QA) are oftenassociated with the pathogenesis of different neurodegenera-tive disorders including Alzheimer’s and Huntington’s disease[1–3]. In kynurenine pathway of tryptophan metabolism, �-amino-�-carboxymuconate-�-semialdehyde (ACMS) is metabolized to�-amino-�-muconate-�-semialdehyde (AMS) via action of impor-tant enzyme ACMS decarboxylase, further AMS converted to acetylCoA as shown in Fig. 1 [4,5]. To maintain the basal Trp-niacin ratio,ACMS is non-enzymatically metabolized to quinolate (QA) whichfurther leads to the NAD formation [6]. Thus, the presence of keyenzyme ACMSD prevents the accumulation of quinolate [7]. Vari-ous studies related to ACMSD show that enzyme is zinc-dependentamidohydrolase maintaining quinolinic acid (QA) and NAD homeo-stasis [8]. Disturbance in the basal levels of QA is associated withmany physiological and pathological conditions related to the cen-tral nervous system (CNS) [9]. Thus, ACMSD act as a checkpoint andregulates the balance between the relative QA levels.

∗ Corresponding author.E-mail address: pravinmcu@gmail.com (P. Kumar).

Several studies have reported that ACMSD is an critical enzymefor tryptophan metabolism [10,11]. Phthalic acid esters (PAEs) arecommonly used for the industrial manufacturing of lubricants, var-ious adhesives, pest repellents, and plastics [12,13]. Several PAEsare long-established environmental endocrine disrupters, perox-isome proliferators and induce reproductive and developmentaltoxicities [14–18]. It has been reported that PAEs can imbalancethe Trp-niacin basal ratio in the tryptophan metabolism pathway.In rats, it has been reported that the conversion ratio of tryptophanto niacin has increased with increase in the dietary concentration ofdi-(2-ethylhexyl) phthalate (DEHP) [19]. This study shows that theincrease in the amount of quinolinic acid with an increase in DEHPconcentration is associated with the inhibition of enzymatic activ-ity of ACMSD. Similarly, di-n-butyl phthalate (DnBP) is reported tobe linked with alteration in trp to niacin ratio in the weaning rats,which were fed with niacin-free and tryptophan limited diet [20].

It has been shown that DEHP degrades to Phthalic acid viaan intermediate mono-(2-ethylhexyl) phthalate (MEHP) [21]. Inanother report, it has been shown that DEHP and its metaboliteMEHP increased QA production in the rats [22]. This study revealsthat the structural similarity of DEHP and MEHP to tryptophanmetabolites is responsible for the noticeable changes in normal

https://doi.org/10.1016/j.ijbiomac.2017.12.0050141-8130/© 2017 Elsevier B.V. All rights reserved.

https://doi.org/10.1016/j.ijbiomac.2017.12.005http://www.sciencedirect.com/science/journal/01418130http://www.elsevier.com/locate/ijbiomachttp://crossmark.crossref.org/dialog/?doi=10.1016/j.ijbiomac.2017.12.005&domain=pdfmailto:pravinmcu@gmail.comhttps://doi.org/10.1016/j.ijbiomac.2017.12.005

N. Singh et al. / International Journal of Biological Macromolecules 108 (2018) 214–224 215

Fig. 1. Kynurenine pathway of tryptophan metabolism. Tryptophan is metabolized to kynurenine and further to 3-Hydroxy anthranilic acid (3-HAA). Then, ACMSD participatesin a reaction (marked with a solid arrow) and directs the conversion of �-amino-�-carboxymuconate-�-semialdehyde (ACMS) to �-amino-�-muconate-�-semialdehyde(AMS), which is further metabolized to Acetyl-CoA. ACMS is also non-enzymatically converted to quinolate (marked with a dotted arrow), which further leads to nicotinamideadenine dinucleotide (NAD) biosynthesis.

tryptophan metabolism and caused the inhibition of ACMSD activ-ity.

Although, numerous studies have shown that phthalatesare involved in disturbing the basal trp to niacin ratio, butthe elucidation of the binding mode and important inter-actions of PAEs with hACMSD have not been reported yet.This study highlights the important interactions of phtha-lates with human �-amino-�-carboxymuconate-�-semialdehydedecarboxylase (hACMSD) which eventually inhibit the hACMSDactivity and leads to the accumulation of quinolate. The crystalstructure of the hACMSD along with substrate analogue Dipicol-inic acid, PDC (PDB ID: 4IH3) is available [8]. PDC binds in the zinccontaining active site of hACMSD and shows the interaction withArg47 and Trp191. In this study, five commonly used PAEs andtheir corresponding monophthalates used for the docking stud-ies with hACMSD are: dimethyl phthalate (DMP), diethyl phthalate(DEP), di-n-butyl phthalate (DnBP), di-isobutyl phthalate (DIBP), di-(2-ethylhexyl) phthalate (DEHP), monomethyl phthalate (MMP),monoethyl phthalate (MEP), mono-n-butyl phthalate (MBP),mono-iso butyl phthalate (MIBP), and mono-(2-ethylhexyl) phtha-late (MEHP), mono-(2-ethyl-5-hydroxyhexyl) phthalate (MEHHP),mono-(2-ethyl-5-oxyhexyl) phthalate (MEOHP). Molecular dock-ing and simulation studies were used to investigate the bindingmode and stability of these PAEs to human ACMSD. The resultsconclude that these phthalates can efficiently bind and can inhibitthe activity of hACMSD. Hence, the binding of PAEs with hACMSDaffect the basal trp to niacin ratio which further accumulates thequinolate in the tryptophan metabolism pathway.

2. Material and methods

2.1. Protein and ligand preparation

Computational docking and simulation studies were performedin order to assess the interaction of the phthalates with hACMSD.The coordinates of hACMSD were retrieved from the crystal struc-

ture of hACMSD bound to a substrate analogue Dipicolinic acid(PDC) (PDB ID: 4IH3) from the RCSB database [8]. Geometric opti-mization of the protein was done by using the Clean Geometrymodule of Discovery Studio (DS) 4.0 suite by Accelerys (San Diego,CA, USA). Side-chain torsion angles varying more than 30◦ fromthe ideal values were corrected. The conformational quality waschecked using the Minimize and refine protein module of DS suite,in which water molecules were eliminated and the CHARMM27force field was used for the protein receptor [23]. The structurewas minimized for 1000 steps by utilizing the smart-minimizeralgorithm with 0.1 RMS gradient cut-off to remove the steric over-laps.

2.2. Molecular docking of ligands

AutoDock 4.2.6 was used to perform the docking of PAEs withhACMSD [24]. Autodock utilizes a semiempirical free energy forcefield to calculate the binding free energy of a small moleculeto a macromolecule. Receptor molecule was prepared by addingexplicit hydrogen molecules and associated Kollman charges(16.0) by utilizing the AutoDock Tools 1.5.6 and saved in.pdbqtfile format. Five commonly used diphthalates and their cor-responding monophthalates used for the docking studies withhACMSD are: dimethyl phthalate (DMP), diethyl phthalate (DEP),di-n-butyl phthalate (DnBP), di-isobutyl phthalate (DIBP), di-(2-ethylhexyl) phthalate (DEHP), monomethyl phthalate (MMP),monoethyl phthalate (MEP), mono-n-butyl phthalate (MBP),mono-iso butyl phthalate (MIBP), mono-(2-ethylhexyl) phthalate(MEHP), mono-(2-ethyl-5-hydroxyhexyl) phthalate (MEHHP), andmono-(2-ethyl-5-oxyhexyl) phthalate (MEOHP). As a positive con-trol, substrate analogue of hACMSD, dipicolinic acid (PDC) wasdocked and compared with binding affinity scores of PAEs. The3D structures of all the phthalates were drawn using Marvin suite17.13 [http://www. chemaxon.com/marvin/sketch/index.jsp] andminimized using DS suite. The properties of phthalate compoundsand their structures are shown in Table 1. The ligands were pre-

http://www. chemaxon.com/marvin/sketch/index.jsphttp://www. chemaxon.com/marvin/sketch/index.jsphttp://www. chemaxon.com/marvin/sketch/index.jsphttp://www. chemaxon.com/marvin/sketch/index.jsphttp://www. chemaxon.com/marvin/sketch/index.jsphttp://www. chemaxon.com/marvin/sketch/index.jsphttp://www. chemaxon.com/marvin/sketch/index.jsphttp://www. chemaxon.com/marvin/sketch/index.jsp

216 N. Singh et al. / International Journal of Biological Macromolecules 108 (2018) 214–224

Table 1Commonly used Phthalates and their usage. The details of Phthalic acid esters (PAEs) such as DMP, DEP, DBP, DIBP, DEHP and their metabolites along with their differentuses.

S.no. Phthalate category Phthalates Molecular weight (g/mol) Structure Usage

1 Low Molecular weight DMP 194.18 Used as plasticizer, also used in insectrepellents

MMP 180.15

DEP 222.4 Solvent in personal care products, celluloseacetate plasticized films for food packaging

MEP 194.18

DBP 278.34 Enteric coating of medications and foodsupplements and nitrocellulose–coatedregenerated cellulose film (RCF) used inplasticized coatings.MBP 222.24

DIBP 278.34 Used as a substitute for DBP due to thesimilarity in their application properties

MIBP 222.24

2 High molecular weight DEHP 390.56 Production of polyvinyl chloride (PVC)plastics, Used in PVC based, medicalproducts, retail packed food items

MEHP 278.34

MEHHP 294.34

MEOHP 292.33

pared by adding hydrogen atoms and Gasteiger charges and thensaved in.pdbqt format. Ligand flexibility was used to specify thetorsional degrees of freedom in ligand molecule. The atomic poten-tial grid map was generated with a spacing of 0.375 Å by usingAutoGrid 4. The docking receptor grid was created by choosingthe catalytic key residues which show interaction with substrateanalogue, PDC i.e. Arg47 and Trp191 and His174, present in theproximity of zinc metal. The dimensions of the grid box were as fol-lows 70 Å × 70 Å × 70 Å and the center point coordinates were set asX = 0.56, Y = −12.94 and Z = 2.77. For docking purpose, Lamarckiangenetic algorithm and grid supported energy evaluation methodwere adopted. The number of total GA runs was increased from10 to 100. Other docking parameters were used as default. Thepose with the maximum binding affinity score and the correspond-ing interactions was selected and further visually inspected andanalyzed in PyMol 1.3 [25].

2.3. Molecular dynamics simulation

The associated structural and dynamic changes occurring at theatomistic level in hACMSD on the binding of PAEs were analyzedby molecular dynamics simulation. The simulation study was per-formed with Gromacs 5.1.4 suite with GROMOS96 43a1 force fieldon LINUX based workstation [26,27]. PDC and phthalates topol-ogy files were generated using Automated Topology Builder (ATB)[28,29]. The protein complexes were solvated in a cubic box withsimple point charge (SPC) waters and counter ions were addedfor the overall electrostatic neutrality of the system [30]. Energyminimization was performed to minimize the steric clashes byusing Steepest descent algorithm for 50,000 iteration steps andcut-off up to 1000 kJmol−1. Then the system was equilibrated intwo different phases for 50,000 steps. The first phase of equilibra-

tion was done with a constant number of particles, volume, andtemperature (NVT), each step 2 fs. The second phase of equilibra-tion was performed with a constant number of particles, pressure,and temperature (NPT), the ensemble at 300 K. LINCS algorithm wasutilized for covalent bond constraints in the equilibration steps. Forthe calculation of Lennard-Jones and Coulomb interactions, 1.4 nmradius cut-off was used. Long range electrostatics were calculatedby using Particle Mesh Ewald (PME) method with Fourier grid spac-ing of 1.6 Å. The temperature inside the box was regulated by usingV-rescale, a modified Berendsen temperature coupling method.Parrinello-Rahman pressure coupling method was utilized in NPTequilibration.

The final production step of molecular dynamics simulation wascarried out for 20 ns, each step of 2 fs. Trajectories were saved andresults were analyzed using XMGRACE. Root mean square devia-tion (RMSD) variation in protein backbone was calculated by usingg rms tool which utilizes the least-square fitting method. Overallroot mean square fluctuation (RMSF) in the atomic positions of pro-tein C� backbone was calculated by using the g rmsf tool. A roughmeasure of compactness factor of protein during the course of thesimulation was estimated by using the g gyrate tool of GROMACS.gmx sasa was used for computation of the total solvent accessiblesurface area (SASA). Hydrogen bonds were calculated with 3.5 Ådistance cut-off by using g hbond and the distribution of inter-molecular hydrogen bond lengths throughout the simulation werealso analyzed.

2.4. MMPBSA binding free energy calculation

The binding free energy of the interaction betweenligand–protein complexes were obtained by utilizing the Molecu-lar Mechanic/Poisson-Boltzmann Surface Area (MMPBSA) method

N. Singh et al. / International Journal of Biological Macromolecules 108 (2018) 214–224 217

Table 2Details of molecular docking results. The summary of binding affinities (kcal/mol)and the polar interactions of the PAEs-hACMSD complexes

S.no Ligand Binding affinity(kcal/mol)

Interactions

1. Dipicolinic acid(PDC)

−6.21 Arg47 (NH1)–O3 (2.8 Å)Arg47 (NH2)–O3 (3.4 Å)Arg47 (NH2)–O3 (3.1 Å)Trp191 (NE1)–O4 (3.2 Å)Trp191 (NE1)–N1 (3.3 Å)Asp291 (OD2)–O2 (3.1 Å)

2. DEHP −5.77 Arg47 (NH2)–O4 (2.4 Å)Arg47 (NH2)–O1 (2.7 Å)Arg47 (NE)–O3 (3.4 Å)Trp191 (NE1)–O4 (2.4 Å)Trp191 (NE1)–O4 (2.7 Å)Trp191 (NE1)–O4 (3.5 Å)

3. MEHP −7.48 His8 (NE2)–O3 (2.7 Å)Arg47 (NH2)–O1 (3.5 Å)His174 (NE2)–O3 (2.5 Å)Trp191 (NE1)–O1 (3.8 Å)His224 (NE2)–O3 (2.7 Å)Asp291 (OD1)–O3 (2.6 Å)Asp291 (OD2)–O4 (3.2 Å)

4. MEHHP −6.94 Arg47 (NH1)–O4 (3.0 Å)Arg47 (NH1)–O5 (3.0 Å)Arg47 (NH2)–O4 (3.2 Å)Val78 (N)–O5 (3.6 Å)Trp191 (NE1)–O3 (2.9 Å)Leu296 (N)–O3 (3.5 Å)

5. MEOHP −7.42 His8 (NE2)–O3 (3.4 Å)Arg47 (NH1)–O1 (3.4 Å)Arg47 (NH2)–O1 (2.8 Å)His174 (NE2)–O3 (3.2 Å)Asp177(N)–O3 (2.7 Å)Trp191 (NE1)–O1 (3.6 Å)His224 (NE2)–O3 (2.7 Å)Asp291 (OD1)–O3 (3.5 Å)Asp291 (OD2)–O4 (3.1 Å)

6. DMP −4.96 Arg47 (NH1)–O2 (2.7 Å)Arg47 (NH2)–O2 (3.3 Å)Arg47 (NH2)–O3 (2.8 Å)Trp191 (NE1)–O4 (3.1 Å)Trp191 (NE1)–N1 (2.9 Å)

7. DEP −4.90 Arg47 (NH1)–O4 (2.9 Å)Arg47 (NH1)–O3 (3.1 Å)Arg47 (NH2)–O2 (2.8 Å)Arg47 (NH2)–O3 (3.0 Å)Arg47 (NH2)–O4 (3.5 Å)Trp191 (NE1)–O2 (3.3 Å)

8. DBP −5.65 Arg47 (NH1)–O3 (3.0 Å)Arg47 (NH2)–O3 (2.9 Å)Val78 (N)–O5 (2.5 Å)Trp191 (NE1)–O2 (3.2 Å)

9. DIBP −5.13 Arg47 (NH1)–O4 (2.9 Å)Arg47 (NH1)–O2 (3.2 Å)Arg47 (NH2)–O1 (2.9 Å)Val78 (N)–O3 (3.7 Å)Trp191 (NE1)–O1 (3.2 Å)

10. MMP −5.10 His8 (NE2)–O11 (2.7 Å)Arg47 (NH2)–O1 (3.1 Å)His174 (NE2)–O11 (2.6 Å)Trp191 (NE1)–O1 (3.3 Å)His224 (NE2)–O12 (3.0 Å)Asp291 (OD1)–O11 (2.5 Å)Asp291 (OD2)–O12 (3.4 Å)

11. MEP −5.03 Arg47 (NH1)–O4 (2.8 Å)Arg47 (NH2)–O4 (3.3 Å)Arg47 (NH2)–O1 (3.0 Å)Val78 (N)–O2 (3.4 Å)His174 (NE2)–O3 (3.5 Å)Trp191 (NE1)–O1 (2.8 Å)

Table 2 (Continued)

S.no Ligand Binding affinity(kcal/mol)

Interactions

12. MBP −5.83 Arg47 (NH1)–O3 (3.2 Å)Arg47 (NH2)–O4 (3.2 Å)Arg47 (NH2)–O3 (3.0 Å)Val78 (N)–O4 (3.6 Å)Trp191 (NE1)–O2 (2.8 Å)

13. MIBP −5.25 Arg47 (NH1)–O4 (2.9 Å)Arg47 (NH2)–O3 (3.0 Å)Arg47 (NH2)–O2 (2.6 Å)Trp191 (NE1)–O2 (2.7 Å)

which employs ensembles derived from molecular dynamics(MD) simulation [31]. In the GROMACS module, the g mmpbsaapplication is used for the calculation of different components ofthe binding free energy of PDC-hACMSD and PAEs-hACSMD com-plexes. Here, the binding energy is an average of three energeticterms, i.e. potential energy in the vacuum, polar-solvation energy,and non-polar solvation energy, respectively. In the presentstudy, the snapshots at each 10 ps between 15 and 20 ns werecollected and MMPBSA was performed to predict the bindingenergy.

3. Results and discussion

Due to the structural similarity of the benzene ring of PAEs withnatural substrate analogue, PDC, they are expected to mimic thebinding mode at the active site of hACSMD. Side chains of PAEsenabled them to make analogous interactions with the importantresidues such as Arg47 and Trp191 and they occupy the activesite of hACMSD within the 4 Å vicinity of zinc metal. Therefore,this study illustrates the binding efficiency of commonly used PAEscomparable to that of PDC, a substrate analogue of hACMSD.

3.1. Molecular docking of ligands

Molecular docking is an extensively used computationalapproach to validate the binding of the suitable orientation ofsmall molecule with the receptor protein. In order to character-ize the molecular interactions, molecular docking of co-crystallizedsubstrate analogue, Dipicolinic acid (PDC) along with PAEs was per-formed within the binding pocket of hACMSD using AutoDock 4.2.6.All the generated binding poses were ranked and clustered accord-ing to their root mean- standard deviation (RMSD) value, bindingaffinity score, and the vicinity of the zinc metal. The AutoDockresults show that the diphthalates and their monophthalates havea binding affinity in the range of −4.9 to −7.48 kcal/mol which iscomparable to a substrate analogue, PDC (−6.21 kcal/mol) as shownin Table 2.

All phthalates have shown interaction with the key amino acidresidues such as Arg47 and Trp191, similar to substrate analogue,PDC as shown in Figs. 2–4, , . Moreover, the monophthalates ofDEHP such as MEHP, MEHHP, and MEOHP have shown maximumbinding affinities with hACMSD. PAEs are present within the rangeof 4 Å from the zinc metal and occupied the active site cavity com-prising of residues such as His8, Val76, His174, Asp291, Phe294,and Leu296, in the same manner as that of PDC. MEHP and MEOHPalso have shown interaction with other active site residues suchas His8, His224, His174, Asp177, and Asp291. The crystal struc-ture of hACMSD in complex with 1,3-dihydroxyacetone phosphate(DHAP) (PDB ID: 2WM1), a glycolytic intermediate which is a potentinhibitor of the enzyme also shows interaction with key residuessuch as Arg47, Asp291, and Trp191(Garavaglia et al., 2009). Theseresults suggest that all the studied PAEs can efficiently bind in the

218 N. Singh et al. / International Journal of Biological Macromolecules 108 (2018) 214–224

Fig. 2. The ribbon diagram of hACMSD binding active site cavity with PDC, DEHP and its metabolites. A) DEHP (green color), B) MEHP (magenta color), C) MEHHP (pink color),D) MEOHP (pale yellow color) and E) PDC (grey color). All the PAEs occupy the active site cavity within the 4 Å vicinity of zinc metal. Residues of the hACMSD are shown instick form, zinc metal is shown as a sphere in metallic blue color and the red dotted line represents intermolecular hydrogen bond interactions. (For interpretation of thereferences to colour in this figure legend, the reader is referred to the web version of this article.)

active site of hACMSD. Moreover, DEHP along with its monophtha-lates seems to be a potent inhibitor for hACMSD. Hence, inhibitionof hACMSD by PAEs can disbalance the basal ratio of Trp-niacin inthe kynurenine pathway.

3.2. Molecular simulation results

Molecular dynamics simulation studies provide suitable meansto understand the changes occurring at the atomistic level in theprotein-ligand system and emphasizes on the stability of the com-

plex. Therefore, a simulation study was performed in order tounderstand the dynamics involved during binding of phthalates tohACMSD. In the present study, different parameters such as RMSD,RMSF, radius of gyration (Rg), solvent accessible surface area (SASA)and the hydrogen bond formation and length distribution duringthe course of the simulation have been studied.

3.2.1. Root-mean-square deviation (RMSD)It represents the dynamic stability of the protein and predicts

the conformation changes occurring in the protein backbone dur-

N. Singh et al. / International Journal of Biological Macromolecules 108 (2018) 214–224 219

Fig. 3. The cartoon representation of hACMSD binding with low molecular weight diphthalates. A) DMP (purple color), B) DEP (white color), C) DBP (orange color) and D)DIBP (yellow color). hACMSD residues are shown in stick form, zinc metal is shown as a sphere in metallic blue color and the red dotted line represents intermolecularhydrogen bond interactions of low molecular phthalates with hACMSD. (For interpretation of the references to colour in this figure legend, the reader is referred to the webversion of this article.)

Fig. 4. The binding interaction of low molecular weight Monophthalates with hACMSD. A) MMP (pale green color), B) MEP (forest green color), C) MBP (light red color) andD) MIBP (light blue color). The interacting residues of hACMSD residues are represented in stick form, zinc metal is shown as a sphere in metallic blue color and the red dottedline shows the intermolecular hydrogen bonds. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

220 N. Singh et al. / International Journal of Biological Macromolecules 108 (2018) 214–224

Fig. 5. RMSD graph of hACMSD with PDC and PAEs. RMSD profile of the C� backbone of hACMSD during the 20 ns of simulation at 300 K: A) PDC, DEHP and its metabolites,B) PDC and low molecular weight diphthalates such as DMP, DEP, DBP and DIBP, and C) PDC and low molecular weight monophthalates such as MMP, MEP, MBP and MIBP.

Table 3Average values of RMSD and RMSF of hACMSD −PAEs complexes. Average RMSDand RMSF values corresponding to variation in hACMSD backbone on the binding ofPAEs and PDC.

S.No Compound Average RMSD (Å) Average RMSF (Å)

1. PDC 0.252 0.1122. DEHP 0.261 0.1263. MEHP 0.307 0.1154. MEHHP 0.257 0.1255. MEOHP 0.221 0.1256. DMP 0.278 0.1277. DEP 0.275 0.1178. DBP 0.255 0.1189. DIBP 0.295 0.111

10. MMP 0.231 0.10211. MEP 0.262 0.13612. MBP 0.223 0.11013. MIBP 0.279 0.132

ing the simulation. Here, in the present study, RMSD values ofPDC-hACMSD and PAEs-hACMSD complexes were analyzed. RMSDplots show that most of the system acquires equilibrium within10–12 ns during the course of the simulation and were stable up to20 ns as shown in Fig. 5. The average RMSD values of the PAEs-hACMSD complexes were compared with the reference ligandPDC-hACSMD complex as shown in Table 3. The high fluctuationof the RMSD values for all the complexes lies within in the rangeof 2 Å to 3 Å. Complexes of lower molecular weight monophtha-lates with hACMSD showed less average RMSD as compared totheir corresponding diphthalates. Complexes of hACMSD with DEPand MEP shows initial fluctuations and attained equilibrium after7 ns of simulation. RMSD results suggest that there is no majorchange in the backbone RMSD patterns of PAEs-hACMSD com-plexes as compared to PDC-hACMSD complex. Moreover, resultsreveal that lower molecular weight monophthalates have a lessaverage root mean square deviation and are more stabilized ascompared to their corresponding diphthalates. The RMSD resultsanalysis implies that the binding of phthalates at the catalytic site

of hACMSD is stable and does not vary the protein backbone sta-bility.

3.2.2. Root-mean-square fluctuation (RMSF)RMSF determines the flexibility of the polypeptide chain after

fitting it to a reference frame. It is the fluctuation of C� atomcoordinates from their average position during the simulation.Generally, in proteins loosely organized loops are characterizedby high RMSF values while secondary structural elements showless flexibility. In the present context, residue mobility was cal-culated for each of the PAEs-hACMSD complexes and was plottedagainst the residue number based on the trajectory of MD simu-lation. Results show a higher RMSF peak in the residues range of180–193, a loop region, which facilitates the entry of ligands intothe active site of hACMSD as shown in Fig. 6. The partial helicalregion from the residue 232–250, present towards the C terminal ofhACMSD also shows the higher RMSF fluctuation. These results sug-gest that active site residues were not considerably perturbed uponbinding of the ligands. Results illustrate that the RMSF fluctuationprofiles of PAEs-hACMSD complexes were almost similar to PDC-hACMSD complex. RMSFs results imply that the atomic mobilityof PAEs-hACMSD complexes is in consent with PDC-hACMSD com-plex. Thus, the PAEs form stable complexes with hACMSD and caninhibit this important enzyme of tryptophan metabolism pathway.

3.2.3. Radius of gyration (Rg)Radius of gyration (Rg) factor is associated with the compactness

of protein during the molecular simulation. It is simply a measure ofthe distance between the center of mass of the protein atoms andits terminal in a given time step. In general, a stably folded pro-tein tends to maintain a relatively less variation in Rg value whichdetermines its dynamic stability. In the present study, variationoccurring in Rg value is plotted against time as shown in Fig. 7. Theradius of gyration results shows that compactness of phthalate-hACMSD complexes is comparable to the PDC-hACMSD complex.Rg results reveal that secondary structures are compactly packed

N. Singh et al. / International Journal of Biological Macromolecules 108 (2018) 214–224 221

Fig. 6. RMSF molecular dynamics simulation results of hACMSD for 20 ns. A) PDC, DEHP and its metabolites with hACMSD, B) PDC and low molecular weight diphthalatessuch as DMP, DEP, DBP and DIBP with hACMSD, and C) PDC and low molecular weight monophthalates such as MMP, MEP, MBP and MIBP.

Fig. 7. Radius of gyration (Rg) plots of PDC-hACMSD and PAEs-hACSMD. The radius of gyration results associated with the compactness of the hACMSD protein for thesimulation of 20 ns: A) PDC, DEHP and its metabolites, B) PDC and low molecular weight diphthalates such as DMP, DEP, DBP and DIBP, and C) PDC, low molecular weightmonophthalates such as MMP, MEP, MBP and MIBP.

in the case of diphthalates and their metabolites to form stablecomplexes with hACMSD.

3.2.4. Solvent accessible surface area (SASA)Solvation free energy of each atom in a protein is contributed

by its polar and non polar residue interactions. Solvent accessiblesurface area (SASA) is the surface area monitored by the probe ofthe solvent molecule when it traces the Van der Waals surface ofthe receptor molecule. Mostly structural alterations are monitoredin the residues forming the loop region in the vicinity of the active

site cavity. In general, hydrophobic residues mostly contribute tothe rise of SASA value. This is also apparent by the raised value of thesolvent accessible surface area (SASA) in that region. SASA resultsof PAEs-hACMSD complexes are similar to PDC-hACMSD complexas shown in Fig. 8.

3.2.5. Hydrogen bond analysisHydrogen bond number and distribution in the PAEs-hACMSD

complexes were studied to determine the stability of the systemduring the 20 ns simulation period. The g hbond utility of GRO-

222 N. Singh et al. / International Journal of Biological Macromolecules 108 (2018) 214–224

Fig. 8. Solvent accessible surface area profile of hACMSD with PDC and PAEs. SASA results of hACMSD-PDC and hACMSD-PAEs complexes during the simulation of 20 ns: A)PDC, DEHP and its metabolites, B) PDC and low molecular weight diphthalates such as DMP, DEP, DBP and DIBP, and C) PDC and low molecular weight monophthalates suchas MMP, MEP, MBP and MIBP.

Fig. 9. Hydrogen bond number results of PDC-hACMSD and PAEs-hACMSD complexes during the 20 ns of simulation. A) PDC, DEHP and its metabolites, B) PDC and lowmolecular weight diphthalates such as DMP, DEP, DBP and DIBP, and C) PDC and low molecular weight monophthalates such as MMP, MEP, MBP and MIBP.

MACS was employed to compute the hydrogen bond numbers anddistribution profiles of the complexes. Hydrogen bond numbersresults show that the most of the PAEs-hACMSD complexes havemaintained a minimum of two hydrogen bonds throughout thecourse of simulation as shown in Fig. 9. Moreover, monophtha-late metabolites of DEHP such as MEHP, MEHHP, and MEOHP forma maximum number of hydrogen bonds with the hACMSD dur-ing the course of 20 ns simulation. The average number of h bondsduring the MD phase shows their continuous contribution in pro-viding stability to the complex. The results of the distribution ofhydrogen bond lengths also indicate that the PAEs-hACMSD com-

plexes have form high to low affinity hydrogen bonds, which isin consent with PDC-hACMSD complex as shown in Fig. 10. Thehydrogen bond results help in understanding the functionality andability of these harmful phthalate compounds to efficiently hinderthe activity hACMSD in the kynurenine pathway.

3.3. MMPBSA binding free energy calculation

The free energy calculation analysis is useful in assessing thebinding potential of ligands as it provides a quantitative estimationof the binding free energy. MMPBSA, a utility within GROMACS was

N. Singh et al. / International Journal of Biological Macromolecules 108 (2018) 214–224 223

Fig. 10. Bar representation of the hydrogen bond numbers distribution of hACMSD with PDC and PAEs. Hydrogen bond distribution profile of PDC-hACMSD and PAEs-hACMSDcomplexes for the course of 20 ns simulation at 300 K, A) PDC, DEHP and its metabolites, B) PDC and low molecular weight diphthalates such as DMP, DEP, DBP and DIBP, andC) PDC and low molecular weight monophthalates such as MMP, MEP, MBP and MIBP.

Table 4Binding energies of hACMSD-PAEs complexes by MMPBSA. Van der Waals energy, electostatic energy, polar solvation energy and SASA energy components contributing tothe total binding free energy of hACMSD-PAEs and hACMSD-PDC complexes.

S.No Compound Van der Waalsenergy (kJ/mol)

Electostaticenergy (kJ/mol)

Polar solvationenergy (kJ/mol)

SASA energy(kJ/mol)

Binding energy(kJ/mol)

1 PDC −139.405 +/− 0.329 −9.747 +/− 0.331 67.745 +/− 0.423 −11.127 +/− 0.027 −92.546 +/− 0.5672 DEHP −213.316 +/− 0.602 −12.924 +/− 0.346 157.375+/− 1.362 −20.392 +/− 0.056 −89.257 +/− 1.5153 MEHP −182.956 +/− 0.609 −35.861 +/− 0.383 104.625+/− 0.742 −15.679 +/− 0.045 −129.888 +/− 0.6864 MEHHP −201.045 +/− 0.928 −10.651 +/− 0.404 78.052 +/− 0.629 −15.373 +/− 0.059 −149.068+/− 0.8065 MEOHP −228.297 +/− 0.522 −22.370 +/− 0.309 118.271+/− 0.579 −17.246 +/− 0.043 −149.646+/− 0.8476 DMP −158.565 +/− 0.343 −25.771 +/− 0.241 122.446+/− 0.451 −12.060 +/− 0.029 −73.961 +/− 0.5077 DEP −142.667 +/− 0.468 −42.500 +/− 0.454 135.593+/− 1.449 −12.310 +/− 0.040 −61.884 +/− 1.3478 DBP −167.730 +/− 0.465 −27.721 +/− 0.459 124.417+/− 0.716 −14.253 +/− 0.042 −85.287 +/− 0.8109 DIBP −183.035 +/− 0.477 −9.031 +/− 0.340 125.265+/− 0.891 −17.092 +/− 0.041 −83.893 +/− 0.91010 MMP −149.449 +/− 0.410 −65.586 +/− 0.430 139.931+/− 0.756 −10.851 +/− 0.026 −85.955 +/− 0.99011 MEP −145.130 +/− 0.379 −23.799 +/− 0.557 103.311+/− 1.398 −12.793 +/− 0.039 −78.411 +/− 1.38512 MBP −182.239 +/− 1.320 −20.841 +/− 0.394 108.814+/− 0.860 −13.533 +/− 0.098 −107.794+/− 1.40213 MIBP −196.513 +/− 0.367 −9.160 +/− 0.191 122.351+/− 0.323 −13.708 +/− 0.029 −97.022 +/− 0.489

utilized to calculate the binding free energy of PAEs-hACMSD com-plexes. The trajectories of the last 5 ns of PAEs-hACMSD complexeswere generated and the MMPBSA was utilized for the prediction ofthe binding energy of the complexes. MMPBSA results show thatthe binding free energies of the PAEs are comparable to the PDCwhile the monophthalates metabolites of DEHP have higher bind-ing free energy than other PAEs as shown in Table 4. Binding freeenergy results of MMPBSA indicate that PAEs-hACMSD complexesare stable. MMPBSA results confirm that these phthalates can effi-ciently bind in the active site of hACMSD and inhibit enzymaticactivity. Highest binding free energies of DEHP and its metabo-lites reveal that they are can bind with high affinity to hACSMDas compared to PDC.

4. Conclusion

In kynurenine pathway, human �-amino-�-carboxymuconate-�-semialdehyde decarboxylase (hACMSD) controls the level ofquinolinic acid. The inhibition of hACMSD activity may lead tothe elevation of quinolate levels which is often associated with

several neurological disorders. PAEs are the ubiquitous environ-mental pollutants which have endocrine disruption and teratogenicproperties. The in-silico techniques such as molecular docking andsimulation studies were done to analyze the interactions of com-monly used PAEs with hACMSD. This computational study showsthat the PAEs-hACMSD complexes are stable and have bindingaffinities similar to natural substrate analogue complex i.e. PDC-hACMSD. Hence, PAEs and their metabolites can efficiently bindto hACMSD and inhibit its activity in the kynurenine pathway oftryptophan metabolism. Our study emphasizes on the inhibitoryeffects of phthalates on hACMSD activity and simultaneously pro-vides a regulatory link between disturbances in trp to niacin ratioon the administration of PAEs in the diet. Further studies shouldfocus on the in-vitro assessment of phthalate toxicity and correla-tion needs to establish between the elevations of quinolinic acidlevel in urine along with corresponding phthalate exposure.

Disclosure statement

No conflict of interest is reported by the authors

224 N. Singh et al. / International Journal of Biological Macromolecules 108 (2018) 214–224

Acknowledgement

PK would like to thank DBT, Govt. Of India, Ministry of Sci-ence and Technology for providing the financial support throughNational Bioscience Award and grant no. BT/HRD/NBA/37/01/2015(VII). NS thanks, UGC and VD thanks DBT for financial support.

References

[1] H. Baran, K. Jellinger, L. Deecke, Kynurenine metabolism in Alzheimer’sdisease, J. Neural Transm. 106 (1999) 165–181.

[2] S.-I. Fukuoka, K. Ishiguro, K. Yanagihara, A. Tanabe, Y. Egashira, H. Sanada, K.Shibata, Identification and expression of a cDNA encoding human�-amino-�-carboxymuconate-�-semialdehyde decarboxylase (ACMSD) a keyenzyme for the tryptophan-niacine pathway and quinolinate hypothesis, J.Biol. Chem. 277 (2002) 35162–35167.

[3] R. Schwarcz, P. Guidetti, K.V. Sathyasaikumar, P.J. Muchowski, Of mice, ratsand men: revisiting the quinolinic acid hypothesis of Huntington’s disease,Prog. Neurobiol. 90 (2010) 230–245.

[4] A.H. Mehler, K. Yano, E.L. May, Nicotinic acid biosynthesis: control by anenzyme that competes with a spontaneous reaction, Science 145 (1964)817–819.

[5] F. Moroni, Tryptophan metabolism and brain function: focus on kynurenineand other indole metabolites, Eur. J. Pharmacol. 375 (1999) 87–100.

[6] L. Pucci, S. Perozzi, F. Cimadamore, G. Orsomando, N. Raffaelli, Tissueexpression and biochemical characterization of human 2-amino3-carboxymuconate 6-semialdehyde decarboxylase, a key enzyme intryptophan catabolism, FEBS J. 274 (2007) 827–840.

[7] T.W. Stone, G.M. Mackay, C.M. Forrest, C.J. Clark, L.G. Darlington, Tryptophanmetabolites and brain disorders, Clin. Chem. Lab. Med. 41 (2003) 852–859.

[8] L. Huo, F. Liu, H. Iwaki, T. Li, Y. Hasegawa, A. Liu, Human�-amino-�-carboxymuconate-�-semialdehyde decarboxylase (ACMSD): Astructural and mechanistic unveiling, Proteins 83 (2015) 178–187.

[9] R. Beninger, A. Colton, J. Ingles, K. Jhamandas, R. Boegman, Picolinic acidblocks the neurotoxic but not the neuroexcitant properties of quinolinic acidin the rat brain: evidence from turning behaviour and tyrosine hydroxylaseimmunohistochemistry, Neuroscience 61 (1994) 603–612.

[10] S. Garavaglia, S. Perozzi, L. Galeazzi, N. Raffaelli, M. Rizzi, The crystal structureof human �-amino-�-carboxymuconate-�-semialdehyde decarboxylase incomplex with 1, 3-dihydroxyacetonephosphate suggests a regulatory linkbetween NAD synthesis and glycolysis, The FEBS Journal 276 (2009)6615–6623.

[11] H. Matsuda, M. Sato, M. Yakushiji, M. Koshiguchi, S. Hirai, Y. Egashira,Regulation of rat hepatic �-amino-�-carboxymuconate-�-semialdehydedecarboxylase, a key enzyme in the tryptophan-NAD pathway, by dietarycholesterol and sterol regulatory element-binding protein-2, Eur. Journal ofNutri. 53 (2014) 469–477.

[12] Y. Guo, K. Kannan, Challenges encountered in the analysis of phthalate estersin foodstuffs and other biological matrices, Anal. Bioanal. Chem. 404 (2012)2539–2554.

[13] M. Wittassek, H.M. Koch, J. Angerer, T. Brüning, Assessing exposure tophthalates–the human biomonitoring approach, Mol. Nutri. Food Res. 55(2011) 7–31.

[14] Å. Bergman, J.J. Heindel, S. Jobling, K. Kidd, T.R. Zoeller, State of the Science ofEndocrine Disrupting Chemicals 2012: Summary for Decision-makers, 2013(W.H. Organization).

[15] B. Desvergne, J.N. Feige, C. Casals-Casas, PPAR-mediated activity of phthalates:a link to the obesity epidemic? Mol. Cell. Endocrinol. 304 (2009) 43–48.

[16] E. Diamanti-Kandarakis, J.-P. Bourguignon, L.C. Giudice, R. Hauser, G.S. Prins,A.M. Soto, R.T. Zoeller, A.C. Gore, Endocrine-disrupting chemicals: anendocrine society scientific statement, Endocr. Rev. 30 (2009) 293–342.

[17] J.L. Lyche, A.C. Gutleb, Å. Bergman, G.S. Eriksen, A.J. Murk, E. Ropstad, M.Saunders, J.U. Skaare, Reproductive and developmental toxicity of phthalates,J. Toxicol. Environ. Health 12 (Pt. B) (2009) 225–249.

[18] M. Sarath Josh, S. Pradeep, K. Vijayalekshmi Amma, S. Balachandran, U. AbdulJaleel, M. Doble, F. Spener, S. Benjamin, Phthalates efficiently bind to humanperoxisome proliferator activated receptor and retinoid X receptor �, �, �subtypes: an in silico approach, J. Appl. Toxicol. 34 (2014) 754–765.

[19] T. Fukuwatari, M. Ohta, E. Sugimoto, R. Sasaki, K. Shibata, Effects of dietary di(2-ethylhexyl) phthalate a putative endocrine disrupter, on enzyme activitiesinvolved in the metabolism of tryptophan to niacin in rats, Biochim. Biophys.Acta (BBA)-General Subjects 1672 (2004) 67–75.

[20] K. Shibata, T. Fukuwatari, A. Enomoto, E. Sugimoto, Increased conversion ratioof tryptophan to niacin by dietary di-n-butylphthalate, J. Nutr. Sci. Vitaminol.(Tokyo) 47 (2001) 263–266.

[21] N. Singh, V. Dalal, J.K. Mahto, P. Kumar, Biodegradation of phthalic acid esters(PAEs) and in silico structural characterization of mono-2-ethylhexylphthalate (MEHP) hydrolase on the basis of close structural homolog, J.Hazard. Mater. 338 (2017) 11–22.

[22] T. Fukuwatari, S. Ohsaki, S. -i. Fukuoka, R. Sasaki, K. Shibata, Phthalate estersenhance quinolinate production by inhibiting�-amino-�-carboxymuconate-(-semialdehyde decarboxylase (ACMSD), a keyenzyme of the tryptophan pathway, Toxicol. Sci. 81 (2004) 302–308.

[23] A.D. MacKerell Jr., D. Bashford, M. Bellott, R.L. Dunbrack Jr., J.D. Evanseck, M.J.Field, S. Fischer, J. Gao, H. Guo, S. Ha, All-atom empirical potential formolecular modeling and dynamics studies of proteins, J. Phys. Chem. B 102(1998) 3586–3616.

[24] G.M. Morris, R. Huey, W. Lindstrom, M.F. Sanner, R.K. Belew, D.S. Goodsell, A.J.Olson, AutoDock4 and AutoDockTools4. Automated docking with selectivereceptor flexibility, J. Comput. Chem. 30 (2009) 2785–2791.

[25] W.L. DeLano, The PyMOL Molecular Graphics System, 2002 http://pymol. org.[26] D. Van Der Spoel, E. Lindahl, B. Hess, G. Groenhof, A.E. Mark, H.J. Berendsen,

GROMACS: fast, flexible, and free, J. Comput. Chem. 26 (2005) 1701–1718.[27] W.F. van Gunsteren, S.R. Billeter, A.A. Eising, P.H. Hünenberger, P. Krüger, A.E.

Mark, W.R. Scott, I.G. Tironi, Biomolecular Simulation: the {GROMOS96}Manual and User Guide, 1996.

[28] K.B. Koziara, M. Stroet, A.K. Malde, A.E. Mark, Testing and validation of theAutomated Topology Builder (ATB) version 2.0: prediction of hydration freeenthalpies, J. Comput. Aided Mol. Des. 28 (2014) 221–233.

[29] A.K. Malde, L. Zuo, M. Breeze, M. Stroet, D. Poger, P.C. Nair, C. Oostenbrink, A.E.Mark, An automated force field topology builder (ATB) and repository:version 1.0, J. Chem. Theory Comput. 7 (2011) 4026–4037.

[30] H.-C. Lee, W.-C. Hsu, A.-L. Liu, C.-J. Hsu, Y.-C. Sun, Using thermodynamicintegration MD simulation to compute relative protein?ligand binding freeenergy of a GSK3� kinase inhibitor and its analogs, J. Mol. Graphics Modell. 51(2014) 37–49.

[31] R. Kumari, R. Kumar, A. Lynn, O.S.D.D. Consortium, G mmpbsa—A GROMACStool for high-throughput MM-PBSA calculations, J. Chem. Inf. Model 54 (2014)1951–1962.

http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0005http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0005http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0005http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0005http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0005http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0005http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0005http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0005http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0005http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0005http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0005http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0005http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0005http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0005http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0005http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0005http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0005http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0005http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0005http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0010http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0010http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0010http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0010http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0010http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0010http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0010http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0010http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0010http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0010http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0010http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0010http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0010http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0010http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0010http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0010http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0010http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0010http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0010http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0010http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0010http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0010http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0010http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0010http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0010http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0010http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0010http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0010http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0010http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0010http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0010http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0010http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0010http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0010http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0010http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0010http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0010http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0010http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0010http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0010http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0010http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0010http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0010http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0010http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0010http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0010http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0010http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0010http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0015http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0015http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0015http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0015http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0015http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0015http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0015http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0015http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0015http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0015http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0015http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0015http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0015http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0015http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0015http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0015http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0015http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0015http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0015http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0015http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0015http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0015http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0015http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0015http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0015http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0015http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0015http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0015http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0020http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0020http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0020http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0020http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0020http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0020http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0020http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0020http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0020http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0020http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0020http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0020http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0020http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0020http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0020http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0020http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0020http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0020http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0020http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0020http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0020http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0020http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0020http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0020http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0020http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0025http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0025http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0025http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0025http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0025http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0025http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0025http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0025http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0025http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0025http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0025http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0025http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0025http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0025http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0025http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0025http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0025http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0025http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0025http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0025http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0025http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0025http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0030http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0030http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0030http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0030http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0030http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0030http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0030http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0030http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0030http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0030http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0030http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0030http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0030http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0030http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0030http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0030http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0030http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0030http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0030http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0030http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0030http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0030http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0030http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0030http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0030http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0030http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0030http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0030http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0030http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0030http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0030http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0030http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0030http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0030http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0035http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0035http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0035http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0035http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0035http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0035http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0035http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0035http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0035http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0035http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0035http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0035http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0035http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0035http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0035http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0035http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0035http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0035http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0035http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0035http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0035http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0035http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0035http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0035http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0040http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0040http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0040http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0040http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0040http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0040http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0040http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0040http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0040http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0040http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0040http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0040http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0040http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0040http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0040http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0040http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0040http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0040http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0040http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0040http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0040http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0040http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0040http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0040http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0040http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0040http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0040http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0040http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0040http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0040http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0040http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0040http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0045http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0045http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0045http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0045http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0045http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0045http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0045http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0045http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0045http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0045http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0045http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0045http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0045http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0045http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0045http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0045http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0045http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0045http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0045http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0045http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0045http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0045http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0045http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0045http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0045http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0045http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0045http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0045http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0045http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0045http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0045http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0045http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0045http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0045http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0045http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0045http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0045http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0045http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0045http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0045http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0045http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0050http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0050http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0050http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0050http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0050http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0050http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0050http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0050http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0050http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0050http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0050http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0050http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0050http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0050http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0050http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0050http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0050http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0050http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0050http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0050http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0050http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0050http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0050http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0050http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0050http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0050http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0050http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0050http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0050http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0050http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0050http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0050http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0050http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0050http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0050http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0050http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0050http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0050http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0050http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0050http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0050http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0050http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0050http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0050http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0055http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0055http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0055http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0055http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0055http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0055http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0055http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0055http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0055http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0055http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0055http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0055http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0055http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0055http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0055http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0055http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0055http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0055http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0055http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0055http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0055http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0055http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0055http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0055http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0055http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0055http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0055http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0055http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0055http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0055http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0055http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0055http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0055http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0055http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0055http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0055http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0055http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0055http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0055http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0055http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0055http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0055http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0055http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0055http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0055http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0055http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0055http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0060http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0060http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0060http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0060http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0060http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0060http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0060http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0060http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0060http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0060http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0060http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0060http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0060http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0060http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0060http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0060http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0060http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0060http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0060http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0060http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0060http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0060http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0060http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0060http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0060http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0060http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0065http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0065http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0065http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0065http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0065http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0065http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0065http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0065http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0065http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0065http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0065http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0065http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0065http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0065http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0065http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0065http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0065http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0065http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0065http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0065http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0065http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0065http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0065http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0065http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0065http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0065http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0070http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0070http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0070http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0070http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0070http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0070http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0070http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0070http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0070http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0070http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0070http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0070http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0070http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0070http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0070http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0070http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0070http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0070http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0070http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0070http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0070http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0070http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0070http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0070http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0070http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0075http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0075http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0075http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0075http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0075http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0075http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0075http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0075http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0075http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0075http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0075http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0075http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0075http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0075http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0075http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0075http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0075http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0075http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0075http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0075http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0075http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0075http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0075http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0075http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0080http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0080http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0080http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0080http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0080http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0080http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0080http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0080http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0080http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0080http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0080http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0080http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0080http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0080http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0080http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0080http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0080http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0080http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0080http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0080http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0080http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0080http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0080http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0080http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0080http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0080http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0080http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0080http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0080http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0080http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0085http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0085http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0085http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0085http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0085http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0085http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0085http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0085http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0085http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0085http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0085http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0085http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0085http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0085http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0085http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0085http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0085http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0085http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0085http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0085http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0085http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0085http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0085http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0085http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0085http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0085http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0085http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0085http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0085http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0085http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0085http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0085http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0085http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0090http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0090http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0090http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0090http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0090http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0090http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0090http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0090http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0090http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0090http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0090http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0090http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0090http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0090http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0090http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0090http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0090http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0090http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0090http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0090http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0090http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0090http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0090http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0090http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0090http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0090http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0090http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0090http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0090http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0090http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0090http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0090http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0090http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0090http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0090http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0090http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0090http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0090http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0090http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0090http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0090http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0090http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0090http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0090http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0090http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0090http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0090http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0090http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0090http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0090http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0095http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0095http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0095http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0095http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0095http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0095http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0095http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0095http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0095http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0095http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0095http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0095http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0095http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0095http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0095http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0095http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0095http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0095http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0095http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0095http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0095http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0095http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0095http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0095http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0095http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0095http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0095http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0095http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0095http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0095http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0095http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0095http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0095http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0095http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0095http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0095http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0095http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0095http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0095http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0095http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0095http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0095http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0095http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0100http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0100http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0100http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0100http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0100http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0100http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0100http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0100http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0100http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0100http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0100http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0100http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0100http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0100http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0100http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0100http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0100http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0100http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0100http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0100http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0100http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0100http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0100http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0100http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0100http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0100http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0100http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0100http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0105http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0105http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0105http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0105http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0105http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0105http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0105http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0105http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0105http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0105http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0105http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0105http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0105http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0105http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0105http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0105http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0105http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0105http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0105http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0105http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0105http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0105http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0105http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0105http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0105http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0105http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0105http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0105http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0105http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0105http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0105http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0105http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0105http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0105http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0105http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0105http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0105http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0105http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0105http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0110http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0110http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0110http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0110http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0110http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0110http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0110http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0110http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0110http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0110http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0110http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0110http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0110http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0110http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0110http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0110http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0110http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0110http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0110http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0110http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0110http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0110http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0110http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0110http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0110http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0110http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0110http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0110http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0110http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0110http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0110http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0110http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0110http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0110http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0110http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0110http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0110http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0110http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0115http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0115http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0115http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0115http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0115http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0115http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0115http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0115http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0115http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0115http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0115http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0115http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0115http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0115http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0115http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0115http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0115http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0115http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0115http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0115http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0115http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0115http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0115http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0115http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0115http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0115http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0115http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0115http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0115http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0115http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0115http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0115http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0115http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0115http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0115http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0115http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0115http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0115http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0115http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0115http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0115http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0115http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0120http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0120http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0120http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0120http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0120http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0120http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0120http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0120http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0120http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0120http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0120http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0120http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0120http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0120http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0120http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0120http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0120http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0120http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0120http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0120http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0120http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0120http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0120http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0120http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0120http://refhub.elsevier.com/S0141-8130(17)34149-1/sbref0120http://refhub.elsevier.com/S