Dissecting the ribosomal inhibition mechanism of a new ketolide carrying an alkyl-aryl group at C-13 of its lactone ring Marios G. Krokidis, Ourania N. Kostopoulou, Dimitrios L. Kalpaxis, George P. Dinos
To cite this version:
Marios G. Krokidis, Ourania N. Kostopoulou, Dimitrios L. Kalpaxis, George P. Dinos. Dissect- ing the ribosomal inhibition mechanism of a new ketolide carrying an alkyl-aryl group at C-13 of its lactone ring. International Journal of Antimicrobial Agents, Elsevier, 2010, 35 (3), pp.235. 10.1016/j.ijantimicag.2009.11.002. hal-00556385
HAL Id: hal-00556385 https://hal.archives-ouvertes.fr/hal-00556385 Submitted on 16 Jan 2011
HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non, lished or not. The documents may come from émanant des établissements d’enseignement et de teaching and research institutions in France or recherche français ou étrangers, des laboratoires abroad, or from public or private research centers. publics ou privés. Accepted Manuscript
Title: Dissecting the ribosomal inhibition mechanism of a new ketolide carrying an alkyl-aryl group at C-13 of its lactone ring
Authors: Marios G. Krokidis, Ourania N. Kostopoulou, Dimitrios L. Kalpaxis, George P. Dinos
PII: S0924-8579(09)00511-1 DOI: doi:10.1016/j.ijantimicag.2009.11.002 Reference: ANTAGE 3179
To appear in: International Journal of Antimicrobial Agents
Received date: 20-7-2009 Accepted date: 3-11-2009
Please cite this article as: Krokidis MG, Kostopoulou ON, Kalpaxis DL, Dinos GP, Dissecting the ribosomal inhibition mechanism of a new ketolide carrying an alkyl-aryl group at C-13 of its lactone ring, International Journal of Antimicrobial Agents (2008), doi:10.1016/j.ijantimicag.2009.11.002
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. Edited manuscript
Dissecting the ribosomal inhibition mechanism of a new ketolide carrying an alkyl-aryl group at C-13 of its lactone ring
Marios G. Krokidis, Ourania N. Kostopoulou, Dimitrios L. Kalpaxis, George P.
Dinos *
Laboratory of Biochemistry, School of Medicine, University of Patras, 26504
Patras, Greece
ARTICLE INFO
Article history:
Received 20 July 2009
Accepted 3 November 2009
Keywords:
Antibiotics
Macrolides
Ketolides
Kosan-1325Accepted Manuscript
Coupled transcription/translation system
Slow-binding inhibitors
* Corresponding author. Tel.: +30 261 099 6259; fax: +30 261 096 9167.
Page 1 of 25 E-mail address: [email protected] (G.P. Dinos).
Accepted Manuscript
2
Page 2 of 25 ABSTRACT
Ketolides are effective not only against macrolide-sensitive bacteria but also against some macrolide-resistant strains. Here we present data regarding a new ketolide with an alkyl-aryl side chain at C-13 of its lactone ring. It behaves as a strong inhibitor of protein synthesis in a model coupled transcription/translation system, although it does not affect the accuracy of translation. In addition, detailed kinetic analysis shows that it slowly forms a very tight, slowly reversible complex with prokaryotic ribosomes, a property that could be correlated with its superior activity compared with erythromycin against Escherichia coli both in vivo and in vitro.
Accepted Manuscript
3
Page 3 of 25 1. Introduction
Macrolides represent a large family of protein synthesis inhibitors of great clinical interest. They consist of a 12- to 16-membered lactone ring, to which one or more sugar substituents, some of them amino sugars, are attached [1].
Erythromycin and its second-generation derivatives roxithromycin, clarithromycin and azithromycin are the most widely used macrolide antibiotics.
In the last decade, a third-generation macrolide appeared on the market, named telithromycin [2]; another, cethromycin, is currently in the last stages of clinical trials [3]. Both are semisynthetic derivatives of erythromycin, fused with an 11,12- cyclic carbamate group (Fig. 1). In addition, an alkyl-aryl side chain is linked to the lactone ring either through the N-atom of the carbamate group (telithromycin) or through the O-6 of the lactone ring (cethromycin). Moreover, the sugar -L- cladinose at position 3 of the lactone ring is replaced by a keto group and for this reason such compounds are named ketolides. Ketolides may be the future of macrolides since they are effective not only against macrolide-sensitive bacteria but also against some macrolide-resistant strains [4]. They are mainly effective against bacteriaAccepted with inducible macrolide–lincosamide Manuscript–streptogramin B (MLSB) resistance but they have no effect against strains exhibiting constitutive MLSB resistance. Although it was originally thought that ketolides do not induce MLSB resistance, it is now established that they do, although over a small range of concentrations and at a low rate [5].
4
Page 4 of 25 The success of telithromycin and cethromycin in treating infections caused by macrolide-resistant streptococci and staphylococci has triggered an intensive search for newer improved ketolide derivatives. Data regarding a new ketolide,
Kosan-1325 (K-1325) (Fig. 1), were recently published [6] showing that it displays enhanced antimicrobial activity and a distinct mode of action. It binds to the classical macrolide-binding site, just at the entrance of the tunnel, with higher affinity than erythromycin and independently of whether uridine or adenosine is at position 2609 of 23S rRNA (Escherichia coli numbering).
Here we present new data regarding the interaction of K-1325 with the ribosome.
According to these data, K-1325 behaves as a strong inhibitor of a model transcription/translation system without affecting the accuracy of translation.
Additionally, this ketolide behaves as a slow binding and slowly reversible inhibitor, following a one-step mechanism. This is reminiscent of the behaviour of spiramycin and 5-O-mycaminosyltylonolide (OMT), two 16-membered ring macrolides that also follow a one-step mechanism [7,8]. In addition, we demonstrate that, compared with erythromycin, K-1325 forms a tighter complex with E. coli ribosomes, a property that could be correlated with its superior activity againstAccepted bacteria both in vivo and in vitro.Manuscript
2. Materials and methods
L-Phenylalanine, puromycin dihydrochloride (disodium salt), tylosin, erythromycin, GTP, ATP and tRNA from E. coli strain W were purchased from
5
Page 5 of 25 3 Sigma Chemical Co. L-[2,3,4,5,6- H]Phenylalanine was purchased from
Amersham Life Science. Cellulose nitrate filters (type HA, 24 mm diameter, 0.45
m pore size) were from Millipore Corp. Scintillation cocktail Filter Count was purchased from Perkin-Elmer. Ketolide K-1325 was provided by Kosan
Bioscience Inc.
2.1. Biochemical preparations
70S tight-coupled ribosomes were obtained from E. coli K-12 cells as previously described [9]. Heteropolymeric mRNA (MF-mRNA) was prepared with run-off transcription as previously described [10] and was used in a molar ratio seven times that of the ribosomes. Its length is 46 nucleotides with an AUG (Met) codon in the middle, followed by an UUC (Phe) codon. Complex C, i.e. the 70S ribosome•MF-mRNA•Ac[3H]Phe-tRNA complex, was prepared as described previously [8]. Titration with puromycin [8] revealed that 50% of the ribosomes adsorbed on a filter were in the form of complex C, with Ac[3H]Phe-tRNA almost completely bound at the P site.
2.2. CoupledAccepted transcription/translation assay Manuscript Coupled transcription/translation experiments were performed using the E. coli lysate-based system (RTS 100 E. coli HY Kit from Roche) for the expression of the green fluorescent protein (GFP) type cyc3 as described previously [10].
Reaction mixtures were incubated for 5 h at 30 C with shaking (900 rpm) in an
6
Page 6 of 25 RTS ProteoMaster Instrument (Roche) and the results are expressed as mg/mL of GFP [10].
2.3. Puromycin reaction
The reaction between complex C and puromycin was carried out as described previously [8]. Briefly, complex C reacted with puromycin in excess, in the presence or absence of macrolides, and the reaction progress was analysed over a wide range of puromycin and macrolide concentrations. When desired, the reaction was terminated by adding an equal volume of 1 M NaOH. The product,
AcPhe-puromycin, was extracted with ethyl acetate and radioactivity was measured in a liquid scintillation spectrometer. In data processing, the product
(P) was expressed as a percentage of the isolated radioactivity (N0) on the filter
(100 P/N0). Controls without puromycin were included in each experiment and the values obtained were subtracted.
2.4. Inactivation of complex C by tylosin and K-1325, both acting in a competitive fashion Complex C adsorbedAccepted on a cellulose nitrate filterManuscript reacted with various concentrations of tylosin and/or K-1325 in 2 mL of buffer A [HEPES-KOH (pH
7.6), 4.5 mM magnesium acetate, 150 mM ammonium acetate, 2 mM spermidine, 0.05 mM spermine and 4 mM -mercaptoethanol]. The reaction was allowed to proceed at 25 C for the desired time interval and was stopped by
7
Page 7 of 25 immersing the filter in 15 mL of cold buffer A. The remaining active complex C, after washing the cellulose nitrate filter to remove traces of tylosin not specifically bound, was titrated with puromycin (2 mM, 2 min at 25 C) as described previously [11].
3. Results
3.1. Inhibitory effect of K-1325 in a coupled transcription/translation system
The most physiological in vitro system for protein synthesis currently available is the coupled transcription/translation system [12]. We utilised this system to express the GFP gene in order to define the half-inhibitory concentration (IC50) of
K-1325 and its effect on translation accuracy. As shown in Fig. 2, the IC50 of K-
1325 is ca. 0.8 M, whilst complete inhibition can be seen at concentrations >5
M. The fluorescent activity of GFP was measured both in sodium dodecyl sulphate (SDS) and native gels, the latter enabling the fully folded, active fraction of GFP (GFPactive) to be quantitatively determined and correlated with the total amount of protein (GFPtotal) from the former. As a consequence, this method allows measurement of the overall misincorporation, since increased misreading reduces the Acceptedratio GFPactive/GFPtotal. Application Manuscript of this approach in this study revealed that the accuracy of translation remains unaffected with increasing concentrations of K-1325, in contrast to that previously observed under the influence of edeine, a well known inducer of misreading [10].
8
Page 8 of 25 3.2. Puromycin reaction in the presence and absence of K-1325
Tylosin and K-1325 were tested as peptidyltransferase (PTase) inhibitors by analysing their effect on the puromycin reaction. This reaction is a model system for testing PTase activity and takes place according to Kinetic Scheme 1:
Ks k3 C + S CS C’ + P Kinetic Scheme 1
where C is the complex of E. coli 70S ribosomes programmed with MF-mRNA and bearing AcPhe-tRNA at the P site, S is puromycin, P is the product AcPhe- puromycin and C’ is a ribosomal complex without bound donor and thus unable to react for a second cycle with puromycin. In the presence of puromycin in excess, the reaction follows pseudo-first-order kinetics. Consequently, the relationship:
Co ln kobs t Co P (Eq. 1)
holds, where Co is the reactive complex C at zero time, kobs is the apparent rate constant of productAccepted formation and t is the reaction Manuscript time. The rate constant kobs is related to the puromycin concentration by Equation 2:
k S k 3 obs K S S (Eq. 2)
9
Page 9 of 25 which allows determination of both k3 and KS constants by double reciprocal plotting [13]. By analysing the effect of tylosin and K-1325 on the puromycin reaction, we observed that K-1325 does not cause any inhibition of product formation, whereas in the presence of tylosin there is strong inhibition and after 2 min the semi-logarithmic plot reaches a plateau (Fig. 3). In the simultaneous presence of K-1325 and tylosin, the inhibition effect is relieved and at high concentrations of K-1325 becomes nullified (data not shown). These results are reminiscent of previous observations [11], according to which tylosin, a well known PTase inhibitor, inactivates complex C versus puromycin. In contrast, K-
1325, like other macrolides with shorter sugar extensions, such as erythromycin, clarithromycin and OMT, does not inhibit PTase but competes with tylosin for common or overlapping binding sites in the ribosome [8,11].
3.3. Competition of K-1325 and tylosin for binding on ribosomal complex C
To study the competition between tylosin and K-1325 for binding to complex C, we analysed the remaining activity of complex C following its exposure to both antibiotics. According to Fig. 4, inactivation of complex C follows first-order kinetics (Fig. 4A) and depends not only on tylosin but also on the K-1325 concentrationAccepted (Fig. 4B). In other words, K-1325, Manuscript although not inhibiting the puromycin reaction, offers protection to complex C from tylosin inactivation. In theory, this type of competition can follow two alternative mechanisms, depicted in Kinetic Scheme 2.
10
Page 10 of 25 KTyl k4 KTyl k4 C + I CI C*I C + I CI C*I k5 k5 + +
A A
k6 k7 KA
k6 C* A CA C*A k7
One-step mechanism Two-step mechanism
In Kinetic Scheme 2 there are two alternative kinetic schemes representing the competition between K-1325 (A) and tylosin (I) for binding to common or overlapping binding sites in ribosomal complex C. K-1325 may theoretically bind to ribosomal complex C using a one- or a two-step mechanism.
Kinetic equations regarding the two mechanisms have been derived previously
[8,11] and are presented in Table 1. Fixing our data to these equations gives us the opportunity to distinguish which of the two mechanisms is really followed by
K-1325. As shown in Fig. 4A, inactivation of complex C follows first-order kinetics. From the slope of each plot an apparent rate constant for the inactivation Acceptedof complex C, termed F, can be calculated.Manuscript The linearity of F versus the concentration of K-1325 supports the left-side model of Scheme 2, according to which K-1325 interacts with complex C via a one-step mechanism. Otherwise, it might follow equation a’, which corresponds to a non-linear plot (Table 1). To further support this model, 1/[C*I] was plotted versus either K-1325
11
Page 11 of 25 concentration (Fig. 5A) or the inverse of tylosin concentration (1/[I]) (Fig. 5B).
[C*I] represents the concentration of complex C*I at infinite time, when equilibrium between C, tylosin and K-1325 is established. Both plots are linear and are fixed well to Equation b of Table 1, a fact further confirming the one-step model.
The slope of the plot of F versus the concentration of K-1325 is equal to k6KTyl/(KTyl+[I]) (see Equation a in Table 1). By re-plotting the slope against [I], a
4 –1 –1 value of k6 equal to 0.36 10 M s was estimated. The dissociation rate constant k7 was measured as described previously [6] and was confirmed to be
–1 equal to 0.002 min . k6 and k7 constants, estimated in this manner, provide the overall dissociation constant KA (= k7/k6) with a value of 9.2 nM, similar to that measured previously by equilibrium binding studies [6].
4. Discussion
The increasing incidence of antibiotic resistance and the toxicity associated with some of the available compounds is a stimulus for further research aiming to discover new antibiotics. Moreover, the ribosome is a particularly versatile target for drug developmentAccepted [14]. Under these circumstances, Manuscript and taking into account the side effects of telithromycin in clinical applications [15], further exploitation of the ribosome as a target for new ketolides appears warranted. Such a new compound is K-1325. Its advantages are known from previous work [6], but detailed analysis of its interaction with the ribosome was so far missing.
12
Page 12 of 25 According to the data in this study, K-1325 behaves as a slow binding and slowly reversible inhibitor following a one-step mechanism [16]. Its slow association constant (0.36 104 M–1s–1) confirms that K-1325 acts in a time-dependent manner, resembling other macrolides tested in our laboratory such as erythromycin and tylosin [11]. However, in contrast to that observed for erythromycin and tylosin, a rapidly formed encounter complex CA (Kinetic
Scheme 2) is not clear for K-1325. Nevertheless, such an initial step cannot be easily excluded because sometimes it cannot be detected under the experimental conditions and antibiotic concentrations used [17].
Comparing K-1325 with erythromycin [11], we conclude that although the association rate constants do not differ significantly, the dissociation rate constant of K-1325 is significantly lower, suggesting that additional binding forces keep K-1325 tightly bound to the ribosome. It is tempting to speculate that this additional binding energy is coming from the interaction of the alkyl-aryl side chain of K-1325 with the ribosome. In support of this notion, chemical protection experiments performed in a previous study revealed that K-1325 has idiosyncratic interactions with the E. coli ribosome [6]. Nevertheless, fixation of a drug to the ribosomeAccepted depends not only on theManuscript molecular features of the drug but also on the stereochemistry of the drug-binding pocket on the ribosome, which may differ among bacterial species. Crystallographic analysis of drug–ribosome complexes from different microorganisms, studies regarding the resistance of
13
Page 13 of 25 different strains against an antibiotic, and kinetic studies investigating the drug– ribosome interaction have indicated that accommodation of a drug with the ribosome may be species-specific [18–22]. Therefore, to identify the most effective antibiotic against a certain bacterial species, both parameters must be taken into account with caution and a large suite of antibiotics must be available. K-1325 shows promise and could be useful in the future to treat bacterial infections.
Acknowledgment
The authors thank Kosan Bioscience Inc. for providing the new ketolide K-1325.
Funding
This work was partially supported by the Research Committee of Patras
University, Greece (program ‘K. Karatheodoris’).
Competing interests
None declared.
Ethical approval Not required.Accepted Manuscript
14
Page 14 of 25 References
[1] Omura S. Macrolide antibiotics: chemistry, biology and practice. Orlando, FL:
Academic Press; 2002.
[2] Ackermann G, Rodloff AC. Drugs of the 21st century: telithromycin (HMR
3647)—the first ketolide. J Antimicrob Chemother 2003;51:497–511.
[3] Hammerschlag MR, Sharma R. Use of cethromycin, a new ketolide, for
treatment of community-acquired respiratory infections. Expert Opin Investig
Drugs 2008;17:387–400.
[4] Katz L, Ashley GW. Translation and protein synthesis: macrolides. Chem Rev
2005;105:499–528.
[5] Bailey M, Chettiath T, Mankin AS. Induction of erm(C) expression by
noninducing antibiotics. Antimicrob Agents Chemother 2008;52:866–74.
[6] Kouvela EC, Kalpaxis DL, Wilson DN, Dinos GP. Distinct mode of interaction
of a novel ketolides antibiotic that displays enhanced antimicrobial activity.
Antimicrob Agents Chemother 2009;53:1411–9.
[7] Dinos G, Synetos D, Coutsogeorgopoulos C. Interaction between the
antibiotic spiramycin and a ribosomal complex active in peptide bond
formation. Biochemistry 1993;32:10638–47.
[8] KarahaliosAccepted P, Kalpaxis D, Fu H, Katz L, WilsonManuscript D, Dinos G. On the
mechanism of action of 9-O-arylalkyloxime derivatives of 6-O-
mycaminosyltylonolide, a new class of 16-membered macrolide antibiotics.
Mol Pharmacol 2006;70:1271–80.
15
Page 15 of 25 [9] Blaha G, Stelzl U, Spahn CMT, Agrawal RK, Frank J, Nierhaus KH.
Preparation of functional ribosomal complexes and the effect of buffer
conditions on tRNA positions observed by cryoelectron microscopy. Methods
Enzymol 2000;317:292–309.
[10] Dinos G, Wilson D, Teraoka Y, Szafrarski W, Fucini P, Kalpaxis D, et al.
Dissecting the ribosomal inhibition mechanisms of edeine and pactamycin:
the universally conserved residues G693 and C795 regulate P-site tRNA
binding. Mol Cell 2004;13:113–24.
[11] Dinos G, Kalpaxis D. Kinetic studies on the interaction between a
ribosomal complex active in peptide bond formation and the macrolide
antibiotics tylosin and erythromycin. Biochemistry 2001;39:11621–8.
[12] Iskakova MB, Szaflarski W, Dreyfus M, Remme J, Nierhaus KH.
Troubleshooting coupled in vitro transcription–translation system derived from
Escherichia coli cells: synthesis of high-yield fully active proteins. Nucleic
Acids Res 2006;34:e135.
[13] Synetos D, Coutsogeorgopoulos C. Studies on the catalytic rate constant
of ribosomal peptidyltransferase. Biochim Biophys Acta 1987;923:275–85.
[14] Hermann T. Drugs targeting the ribosome. Curr Opin Struct Biol 2005;15:355Accepted–66. Manuscript [15] Shlaes DM, Moellering RC. Telithromycin and the FDA: implications for
the future. Lancet Infect Dis 2008;8:83–5.
[16] Morrison JF, Walsh C. The behavior and significance of slow-binding
enzyme inhibitors. Adv Enzymol Relat Areas Mol Biol 1988;61:201–301.
16
Page 16 of 25 [17] Erion MD, Walsh CT. 1-Aminocyclopropanephosphonate: time-dependent
inactivation of 1-aminocyclopropanecarboxylate deaminase and Bacillus
stearothermophilus alanine racemase by slow dissociation behavior.
Biochemistry 1987;26:3417–25.
[18] Vester B, Douthwaite S. Macrolide resistance conferred by base
substitutions in 23S rRNA. Antimicrob Agents Chemother 2001;45:1–12.
[19] Wilson DN, Harms JM, Nierhaus KH, Schluenzen F, Fucini P. Species-
specific antibiotic–ribosome interactions: implications for drug development.
Biol Chem 2005;386:1239–52.
[20] Schlunzen F, Harms J, Franceschi F, Hansen H, Bartels H, Zarivach R, et
al. Structural basis for the antibiotic activity of ketolides and azalides.
Structure 2003;11:329–38.
[21] Tu D, Blaha G, Moore PB, Steitz TA. Structures of MLSBK antibiotics
bound to mutated large ribosomal subunits provide a structural explanation
for resistance. Cell 2005;121:257–70.
[22] Petropoulos AD, Kouvela EC, Starosta AL, Wilson DN, Dinos GP,
Kalpalxis DL. Time-resolved binding of azithromycin to Escherichia coli
ribosomes. J Mol Biol 2009;385:1179–92. Accepted Manuscript
17
Page 17 of 25 Fig. 1. Chemical structures of erythromycin, telithromycin, cethromycin and
Kosan-1325 (K-1325).
Fig. 2. Expression of green fluorescent protein (GFP) in a transcription/translation system in the presence of increasing concentrations of
Kosan-1325 (K-1325). The total amount of GFP was determined by separating the protein by sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS-
PAGE) and by measuring the intensity of the corresponding band after staining with Coomassie Blue (A), whereas the active GFP fraction was calculated by measuring the fluorescence of the GFP band separated on a native gel (B). (C)
Total GFP protein (●), GFP active fraction (○) and ratio of active GFP to total
GFP (▼) as a function of K-1325 concentration.
Fig. 3. First-order time plots for AcPhe-puromycin formation in the absence or presence of antibiotics. Complex C free of unbound donor reacted with 0.4 mM puromycin alone (■) or in a mixture with either 3 M Kosan-1325 (K-1325) (●) or
3 M tylosin (▼).
Fig. 4. (A) ProgressAccepted curves of complex C inactivation Manuscript by tylosin in the presence of Kosan-1325 (K-1325). Complex C absorbed on a cellulose nitrate filter was exposed to a solution containing either 2 Μ tylosin and 6 M K-1325 ( ) or 3 M tylosin and 10 M K-1325 (○). After the desired period of exposure (t), the reaction was stopped by diluting the reaction mixture in cold buffer. The
18
Page 18 of 25 remaining active complex C (Pt) was then titrated with puromycin. P0 represents the total active complex C before its exposure to antibiotics and P the remaining active complex after infinite time of exposure to tylosin and K-1325. The slope of each plot is equal to the apparent rate constant of inactivation (F). (B) Variation of F as a function of K-1325 concentration. The tylosin concentration was either 2
M ( ) or 6 M (○).
Fig. 5. Variation of 1/[C*I] as a function of the concentration of Kosan-1325 (K-
1325) or the inverse of tylosin concentration (1/[I]).
Accepted Manuscript
19
Page 19 of 25 Edited Table 1
Table 1
Kinetic equations for one-step and two-steps mechanisms (Kinetic Scheme 2) of
drug binding to ribosomes
Function One-step mechanism Two-step mechanism
F k4 I k6 KTyl A I A (a) k4 k6 K I K K Tyl Tyl A (a’) I A 1 KTyl K A
1 1 k K A 1 k K A 6 Tyl (b) 6 Tyl (b’) [C * I ] [C]o k4[C]o I [C]o k4[C]o K A I
Accepted Manuscript
Page 20 of 25 Edited Figure 1
Accepted Manuscript
Page 21 of 25 Edited Figure 2
Accepted Manuscript
Page 22 of 25 Edited Figure 3
Accepted Manuscript
Page 23 of 25 Edited Figure 4
Accepted Manuscript
Page 24 of 25 Edited Figure 5
Accepted Manuscript
Page 25 of 25