|Home | About | Journals | Submit | Contact Us | Français|
A series of indazole-dione derivatives were synthesized by the 1,3-dipolar cycloaddition reaction of appropriate substituted benzoquinones or naphthoquinones and N-carboalkoxyamino diazopropane derivatives. These compounds were evaluated for their effects on human carbonyl reductase. Several of the analogs were found to serve as substrates for carbonyl reductase with a wide range of catalytic efficiencies, while four analogs display inhibitory activities with IC50 values ranging from 3 – 5 μM. Two of the inhibitors were studied in greater detail and were found to be noncompetitive inhibitors against both NADPH and menadione with KI values ranging between 2 and 11 μM. Computational studies suggest that conformation of the compounds may determine whether the indazole-diones bind productively to yield product or nonproductively to inhibit the enzyme.
Anthracyclines such as doxorubicin and daunorubicin (Figure 1A) are anti-neoplastic agents commonly used to treat a variety of cancers.1–5 However, the use of anthracyclines in treating cancer is significantly limited by a potentially lethal cardiotoxicity.3–5 Formation of reactive oxygen species has been proposed as a major pathway for the cause and development of anthracycline-induced cardiotoxicity.6 As a result, extensive studies and tests have been conducted to evaluate the cardio-protective effectiveness of antioxidants against anthracycline-induced cardiotoxicity, but these efforts have resulted in only limited success.7,8 So far, the antioxidant and iron chelator dexrazoxane is the only drug approved by the FDA as an adjuvant therapy for protection against anthracycline-induced cardiotoxicity.6,9 However, this drug offers only partial cardio-protection, and its use is accompanied by dose-dependent leukopenia.7 Therefore, it is necessary to develop new and effective means of stemming anthracycline-induced cardiotoxicity based on mechanisms that are different from those involving the generation of reactive oxygen species. One promising strategy involves inhibiting a pathway by which anthracyclines are converted to alcohol metabolites.
Several reports suggest that the C13 hydroxy metabolites of the anthracyclines (Figure 1) are the principal agents responsible for the observed cardiotoxicity.10–17 Enzymes with NADPH-dependent carbonyl reductase activity are known to reduce the C13 carbonyl of anthracyclines to form the cardiotoxic hydroxyl metabolites. Specifically, several studies have correlated the concentrations of carbonyl reductase (CR; EC 188.8.131.52) with the risk of developing anthracycline-induced cardiotoxicity.18,19 For example, decreased levels of CR from heterozygous mice for a null allele of a CR gene protected against the development of anthracycline cardiotoxity.19 However, overexpression of CR in transgenic mice greatly increased susceptibility to anthracycline cardiotoxicity.18
It was also found that the parent anthracyclines, rather than the alcohol metabolites, are largely responsible for the anti-neoplastic properties,11 while the cardiotoxic alcohol metabolites possess significantly reduced anticancer properties. Thus, anthracycline reduction has been postulated to be a mechanism for anthracycline drug resistance.20 Since anthracycline reduction by CR leads to the formation of alcohol metabolites that are more cardiotoxic and less cytotoxic than the parent anthracyclines, it appears that anthracycline reduction by CR contributes to the cardiotoxicity problem in at least two ways: (1) It reduces the amount of anthracycline in circulation thereby necessitating an increase in the dose needed for effective treatment; (2) An increased anthracycline dosage results in an increase in the amount of cardiotoxic metabolite formed. Consequently, this increase in metabolite concentration increases the risk of cardiotoxicity. Such observations suggest that CR could be a pharmacological target for inhibitors that block its action in order to reduce the risk of cardiotoxicity associated with anthracycline therapy and improve anthracycline anticancer efficacy.
Numerous compounds have been found to inhibit CR,21,22 with flavonoids such as rutin, quercetin, and quercitrin being among the most potent inhibitors.21 Tanaka et al.23 have recently reported the pyrazolopyrimidine 3-(1-tert-butyl-4-amino-1H-pyrazolo[3,4-d]pyrimidin-3-yl)phenol (3) of the Src family kinase inhibitors as a potent inhibitor of CR-catalyzed NADPH-dependent reduction of menadione to menadiol (IC50 = 788 nM). In order to study the efficacy of CR inhibition on anthracycline efficacy, they prepared an analog which showed poor inhibition against protein kinases but maintained potent inhibition against CR (IC50 = 759 nM). This potent and selective inhibitor of CR, 3-(7-isopropyl-4-(methylamino)-7H-pyrrolo[2,3-d]pyrimidin-5yl)phenol (4), was able to effect a 25 % enhancement of daunorubicin-mediated A549-cell killing, consistent with its ability to inhibit CR-mediated metabolism of daunorubicin.23
While investigating the anticancer properties of novel indazole-dione derivatives, we were prompted to evaluate the biological activities of these compounds on the CR pathway. Our screening revealed this new class of indazole-diones to be inhibitors and substrates of CR, and we herein report the synthesis of several 3-[(N-carboalkoxy)ethylamino]indazole-dione derivatives and their biological activities on human CR.
The indazole-diones in this work were prepared as outlined in Scheme 2. Nitrosation of tetrahydro-2-pyrimidone furnished N-nitrosotetrahydro-2-pyrimidone which reacted with alcoholic potassium alkoxides to generate 3-(N-carboalkoxyamino)-1-diazopropanes. The in situ cycloaddition of the diazopropane intermediates with appropriate benzoquinone and naphthoquinone derivatives in a mixture of THF/Et2O at room temperature furnished the corresponding indazole-dione derivatives.24 Some of the quinones used as dipolarophiles in the cycloaddition reaction were obtained commercially, but the phenylamino-1,4-benzoquinones were synthesized by adding a solution of aniline derivatives (5) in aqueous acetic acid to an aqueous solution of 1,4-benzoquinone (6) in an ice bath (Scheme 1) according to the method of Barakat and coworkers.25 This reaction is known to usually give a mixture of mono- and di-anilino-1,4-benzoquinones, but no purification was done to separate the two as only the mono substituted derivatives react in a 1,3-dipolar cycloaddition reaction to give phenylamino indazole-diones.
In addition, the 1,3-dipolar cycloaddition of the diazopropane intermediates with almost all the quinone dipolarophiles was found to be regioselective and furnished only the 6-substituted indazole-dione. However, the cycloaddition of 3-(N-carbomethoxyamino)-1-diazopropane to 2-chloro-1,4-benzoquinone produced a regioisomeric mixture of 6- and 5-chloroindazol-4,7-dione (19A and 19B, Scheme 3).
We have previously reported24 the synthesis of compounds 20 and 21 while all the other analogs in this study are new.
Quinones are known to be good substrates of carbonyl reductase.21 Most of the indazole-dione derivatives in this study were also found to be substrates of CR with different catalytic efficiencies. Table 1 lists the different kinetic constants of CR for the different indazole-dione substrates. The kcat/Km, which is a measure of an enzyme’s catalytic efficiency, varied by about 450-fold for CR and the indazole-dione substrates in this work. The substrates with the greatest catalytic efficiencies (ranging from 0.087 to 1.3 μM−1s−1) are indazole-diones 19A, 20 and 21, which have either a chlorine or a methoxy substituent at position 6 of the indazole-dione. The substrates with the lowest catalytic efficiencies (ranging from 0.0030 to 0.0051 μM−1s−1) are indazole-diones 14, 15, and 16, which have a phenyl amino substituent at position 6 of the indazole-dione. By way of comparison, menadione, an excellent CR substrate which is often used as a model quinone substrate, has a kcat/Km value of 0.098 μM−1s−1.26 The differences in kcat/Km values between the best and worst substrates most likely results from a combination of electronic and steric factors. On one hand the phenyl amino substituents are far bulkier than the substituents present at position 6 in compounds 19A, 20, and 21 and, therefore, could interfere with binding interactions that promote efficient catalysis. On the other hand, electronic factors seem to underlie the differences in the catalytic efficiencies of compounds 19A and 21. Whereas compound 19A (the best substrate) possesses an electron-withdrawing chlorine group at position 6, compound 21 possesses an electron-donating methoxy group at this position. The slight difference in the substrate efficiencies of compounds 20 and 21 further underscores the importance of steric contributions to the activities of these compounds upon interaction with the binding domain of CR. While both compounds 20 and 21 possess a methoxy substituent at position 6 of the indazole-dione scaffold, compound 20 incorporates a bulkier benzyl group (compared to the smaller methyl group of compound 21) on the carbamate appendage making it a less efficient substrate than 21.
Four indazole-dione derivatives (11, 12, 22, and 23) were found to be inhibitors of CR with IC50 values of 3–5 μM against menadione and 11–25 μM against daunorubicin (Table 2). Further studies with indazole-diones 11 and 12 revealed that they are noncompetitive inhibitors against both NADPH and menadione (Table 3). Similar inhibition patterns were observed for the similarly structured pyrazolopyrimidine inhibitors.23,27 Such a finding suggests that these inhibitors bind to several forms of the enzyme, as indicated in Figure 2.
The chemical structures of these inhibitors are only slightly different from those of the indazole-dione substrates discussed above. For example, among the benzene-fused indazole-diones, the substitution of hydrogen at position 8 of substrate 17 with a hydroxyl group gives the indazole-dione 23, which is an inhibitor of CR. The small structural changes on the benzene-fused indazole-diones that give rise to the inhibitory activities may involve interactions, perhaps through hydrogen bonding with amino acids in the active site, that disrupt the proper positioning of the substrate carbonyls for reduction by NADPH. In fact, X-ray structure analysis of CR with the pyrazolpyrimidine inhibitor 3 embedded in the binding site showed the hydroxyl group on 3 to make key H-bond interactions with some residues of the active site of CR.23 A similar interaction with indazole-dione 23 may also be responsible for the inhibitory activity.
For indazole-diones with a phenylamino substituent, replacement of hydrogen attached to nitrogen of the phenylamino group by a methyl or ethyl group converted the indazole-dione derivative from a substrate to an inhibitor. In an attempt to explain why such a substitution converts a phenylamino-substituted indazole-dione from a substrate to an inhibitor, computational methods were employed. A comparison of the energy minimized structures for indazole-diones 11, 12, and 15 shows that the orientation of the phenyl ring with respect to the indazole-dione ring is markedly different when the amino hydrogen is replaced with either a methyl or an ethyl substituent (Figure 3). For example, the phenyl ring of the substrate indazole-dione 15 is almost coplanar with the indazole-dione ring. However, substituting the hydrogen on the 6-amino group with an ethyl or methyl substituent (11 and 12, respectively) presumably induces unfavorable eclipsing interactions that cause the phenyl ring to twist perpendicular to the indazole-dione ring. This apparent difference in orientation could result in different binding modes for the substrate and inhibitor phenylamino indazole-diones such that the catalytic residues and NADPH are positioned properly for catalysis when the phenyl ring is more in plane with the indazole-dione ring. Further work is underway to gain additional insight into the structural requirements for a substrate versus an inhibitor of CR for this class of compounds.
We have synthesized a small library of novel indazole-diones and assessed their biological activities on human carbonyl reductase (CR). Most of the indazole-diones were found to be substrates, while four analogs were found to be inhibitors of human CR with IC50 values of 3–5 micromolar. These inhibitors offer new leads and the potential to improve anthracycline therapy by reducing the cardiotoxicity associated with anthracyclines, while also increasing the anticancer efficacy of anthracyclines. More importantly, this study serves as a lead in a pursuit of the understanding of the structural requirements for substrate versus inhibitory effects on CR.
All reactions were carried out using laboratory grade materials and solvents. Purification by column chromatography was done using 70–230 mesh silica gel. Melting points were determined in open capillary tubes on a Mel-Temp melting point apparatus and are uncorrected. The IR spectra were performed in KBr on a Nicolet Magna 560 or a Nicolet Avatar 370 FTIR spectrometer. The 1H-NMR spectra were obtained on a Bruker Avance 400 MHz spectrometer in deuterated dimethyl sulfoxide (DMSO-d6). Chemical shifts are in δ units (ppm) with (CH3)2SO (2.50 ppm) as the internal standard. Peaks due to methylene protons α to the urethane group of the indazole-diones in this study are buried or overlap significantly with the water peak at 3.25 – 3.50 ppm. This was confirmed by obtaining the 1H-NMR spectrum of one indazole-dione (compound 11) in deuterated chloroform (CDCl3). The 1H-NMR spectrum of 11 in CDCl3 displayed a 2H multiplet centered at 3.53 ppm for the methylene protons α to the urethane group, whereas, the peak for the same protons was buried in the water peak at 3.33 ppm in the DMSO-d6 spectrum. For most of the compounds, mass spectral measurements were done using a Finnigan Navigator mass spectrometer; while for three compounds, high resolution mass spectra (EI or FAB) were recorded on a VG Analytical 70-SE mass spectrometer equipped with a 11-250J data system. Methanol was used as the solvent, and all the mass spectra showed signals corresponding to the molecular ion plus sodium (Na). N-Nitrosotetrahydro-2(H)-pyrimidone was prepared as previously reported.24
Benzoquinone (6, 2.0884 g, 19.32 mmol) was dissolved in hot distilled water (200 mL) and the solution was filtered. The filtrate was cooled down to −8 °C, and a solution of N-ethylaniline (5a, 1.2 mL, 9.506 mmol) was added to it (The solution of N-ethylaniline was prepared by adding N-ethylaniline (1.2 mL) into water (75 mL) and then adding glacial acetic acid until all the N-ethylaniline dissolved). The resulting mixture was stirred at 0 °C for 15 min, and filtered under suction to yield a dark purple solid (0.7320 g) which was used without purification for the next experiment.
N-Nitrosotetrahydro-2-pyrimidone (9, 0.7040 g, 5.4523 mmol) was added in small portions to a cooled mixture of KOH (1.0044 g) in methanol (10 mL) and diethyl ether (50 mL). The resultant mixture was stirred at room temperature for 25 min and washed with saturated NaCl solution (2 × 25 mL), 50 % aqueous NaCl (2 × 25 mL), and saturated NaCl (2 × 25 mL). The ether layer was added to a solution of 2-(N-ethylphenylamino)-1,4-benzoquinone (7a, 0.6606 g, 2.9068 mmol) in THF (25 mL), and the resultant mixture was stirred at room temperature for 3 days. The mixture was concentrated in vacuo and diethyl ether (100 mL) was added to it. The resulting mixture was placed in the freezer overnight and filtered under suction to furnish a brick red solid of 11 (282.7 mg, 26.40 %). Mp 95–96 °C. IR (cm−1) 3336, 3121, 1697, 1615, 1548. 1H-NMR (DMSO-d6) δ 1.10 (t, 3H, J = 6.5 Hz), 2.99 (t, 2H, J = 6.8 Hz), 3.51 (s, 3H), 3.78 (m, 2H), 5.84 (s, 1H), 7.13 (d, 2H, J = 7.5 Hz), 7.18–7.24 (m, 2H), 7.36 (t, 2H, J = 7.8 Hz), 14.00 (s, 1H). ESI MS m/z 391.0 ([M+Na]+ calcd 391.1). 1H-NMR (CDCl3) δ 1.23 (t, 3H, J = 7.0 Hz), 3.17 (dist. t, 2H), 3.53 (m, 2H), 3.63 (s, 3H), 3.84 (q, 2H, J = 7.0 Hz), 5.52 (br s, 1H), 5.82 (s, 1H), 7.10 (d, 2H, J = 8.0 Hz), 7.25 (t, 1H, J = 7.4 Hz), 7.37 (t, 2H, J = 7.73 Hz), 12.43 (br s, 1H).
Mp 194–196 °C. IR (cm−1) 3354, 3125, 1696, 1603, 1554. 1H-NMR (DMSO d6) δ 3.02 (t, 2H, J = 6.8 Hz), 3.52 (s, 3H), 5.93 (s, 1H), 7.13 (d, 2H, J = 7.6 Hz), 7.18 (t, 1H, J = 7.3 Hz), 7.27 (br t, 1H, J = 5.4 Hz), 7.34 (t, 2H, J = 7.8 Hz), 14.00 (s, 1H). ESI MS m/z 377.2 ([M+Na]+ calcd 377.1).
Mp 212–214 °C. IR (cm−1) 3335, 3250, 3118, 1696, 1624, 1591. 1H-NMR (DMSO- d6) δ 2.31 (s, 3H), 3.00 (t, 2H, J = 7.0 Hz), 3.51 (s, 3H), 5.74 (s, 1H), 7.22 (s, 5H), 8.87 (s, 1H), 14.10 (s, 1H). FAB HRMS calcd for C18H19N4O4 m/z 355.1406 (M+ + 1), found 355.1440.
Mp 217–220 °C. IR (cm−1) 3334, 3113, 1701, 1582. 1H-NMR (DMSO-d6) δ 3.02 (t, 2H, J = 6.8 Hz), 3.51 (s, 3H), 3.85 (s, 3H), 5.54 (s, 1H), 7.03 (dt, 1H, J = 7.6, 1.3 Hz), 7.15 (dd, 1H, J = 8.3, 1.2 Hz), 7.20–7.28 (m, 2H), 7.33 (dd, 1H, J = 7.8, 1.5 Hz), 8.34 (s, 1H), 14.10 (s, 1H). EI HRMS calcd for C18H18N4O5 m/z 370.1277, found 370.1271
Mp °C 202–203. IR (cm−1) 3401, 3327, 3109, 1697, 1578. 1H-NMR (DMSO-d6) δ 3.02 (t, 2H, J = 6.2 Hz), 3.51 (s, 3H), 5.83 (s, 1H), 7.17–7.21 (m, 1H), 7.24–7.27 (m, 1H), 7.35 (d, 2H, J = 7.4 Hz), 7.39–7.44 (m, 2H), 8.95 (s, 1H), 14.10 (s, 1H). ESI MS m/z 363.1 ([M+Na]+ calcd 363.1).
Mp 191 °C (dec.). IR (cm−1) 3324, 3222, 1696, 1613, 1571. 1H-NMR (DMSO- d6) δ 3.00 (t, 2H, J = 6.9 Hz), 3.51 (s, 3H), 3.77 (s, 3H), 5.63 (s, 1H), 6.99 (d, 2H, J = 9.0 Hz), 7.22–7.31 (m, 3H), 8.86 (s, 1H), 14.10 (s, 1H). ESI MS m/z 393.2 ([M+Na]+ calcd 393.1).
N-Nitrosotetrahydro-2-pyrimidone (9, 0.6974 g, 5.4012 mmol) was added in small portions to a cooled mixture of KOH (0.9962 g) in methanol (10 mL) and diethyl ether (50 mL). The resultant mixture was stirred at room temperature for 25 min and washed with saturated NaCl (2 × 25 mL), 50 % aqueous NaCl (2 × 25 mL), and saturated NaCl (2 × 25 mL). The ether layer was added to a solution of 1,4-naphthoquinone (0.4852 g, 3.0680 mmol) in THF (20 mL). The resultant mixture was stirred at room temperature for 1.5 h and filtered under suction to furnish a pale yellowish solid of 17 (444.8 mg, 48.44 %). Mp 249-252 °C. IR (cm−1) 3289, 3122, 1680, 1596. 1H-NMR (DMSO-d6) δ 3.13 (t, 2H, J = 6.5 Hz), 3.51 (s, 3H), 7.25 (br t, 1H), 7.84–7.91 (m, 2H), 8.12–8.16 (m, 2H), 14.40 (s, 1H). ESI MS m/z 322.1 ([M+Na]+ calcd 322.1).
Mp 204–206 °C. IR (cm−1) 3305, 3183, 3117, 1685, 1642, 1560. 1H-NMR (DMSO-d6) δ 2.02 (d, 3H, J = 1.5 Hz), 3.00 (t, 2H, J = 6.9 Hz), 3.50 (s, 3H), 6.66 (d, 1H, J = 1.6 Hz), 7.25 (t, 1H, J = 5.7 Hz), 14.00 (s, 1H). ESI MS m/z 286.1 ([M+Na]+ calcd 286.1).
Mp 192–193 °C. IR (cm−1) 3322, 3179, 3109, 1700, 1642, 1572. 1H-NMR (DMSO-d6) δ 3.03 (t, 2H, J = 6.8 Hz), 3.51 (s, 3H), 7.18–7.26 (m, 2H), 14.35 (s, 1H). EI HRMS calcd for C11H10N3O4Cl m/z 283.0360, found 283.0320.
See ref 24.
See ref 24.
Mp 216–219 °C. IR (cm−1) 3357, 3117, 1676, 1534. 1H-NMR (DMSO-d6) δ 3.10 (br s, 2H), 3.51 (s, 3H), 3.86 (s, 3H), 3.87 (s, 3H), 7.25 (br s, 1H), 14.30 (s, 1H). ESI MS m/z 450.2 ([M+Na]+ calcd 450.0).
Mp 205–207° C. IR (cm−1) 3342, 3251, 1662, 1646, 1628, 1600. 1H-NMR (DMSO) δ 3.14 (br s, 2H), 3.51 (s, 3H), 7.14–7.44 (m, 2H), 7.58–7.87 (m, 2H), 12.60 (d, 1H), 14.50 (s, 1H). ESI MS m/z 338.1 ([M+Na]+ calcd 338.1).
Recombinant human carbonyl reductase was expressed in E. coli, BL21 DE3 cells harboring pET-5a-HCBR.26 The cells were grown in 6 liters of LB-broth according to published methods.26 Recombinant human liver carbonyl reductase was purified to homogeneity following a previously described method.26 Throughout the preparation, human carbonyl reductase was distinguished from the bacterial enzyme by sensitivity to rutin inhibition.28 Purity was confirmed via SDS-PAGE and the resulting protein was found to have kcat and Km values for menadione and 4-benzoylpyridine consistent with previously reported values for human carbonyl reductase.21,29 Enzyme concentration was determined spectrophotometrically (ε280 = 0.699 mg−1ml) and protein concentrations were determined by the method of Bradford using bovine serum albumin as the standard.30
Kinetic measurements were made for the indazole-dione substrates with different substrate concentrations and 50 μM NADPH in 100mM potassium phosphate. It was necessary to use up to 1 % DMSO or 2 % methanol for several of the carbonyl compounds to enhance their solubility. In control experiments using the standard assay described below, the enzyme was unaffected by methanol concentrations up to 2 % and DMSO concentrations up to 1 %. Carbonyl substrate concentrations were determined from mass or spectrophotometrically (menadione, ε340 = 2500 M−1cm−1; daunorubicin, ε486 = 12,100 M−1cm−1). Assays were measured by following NADPH oxidation to NADP+ spectrophotometrically (ε340 = 6220 M−1cm−1) as a function of time using a Varian Cary 100 Bio.31 Initial slopes of the resulting progress curves were estimated from fits of the curves to either a line or a parabola, and these slopes were used to calculate the initial velocities. Steady-state kinetic data were fit to the Michaelis-Menten equation using HYPERW.32
Inhibitors were evaluated by determining the IC50 values in the following standard assay. Enzyme was added to different concentrations of inhibitor, 250 μM menadione, and 50 μM NADPH in 100 mM potassium phosphate, pH 7.0 at 25°C. IC50 values were extrapolated from the plot of % activity remaining vs. inhibitor concentration. For more complete inhibition studies, concentrations of indazole-diones 11 and 12 were varied against varied menadione or NADPH concentrations. For experiments with varied menadione concentrations the NADPH concentrations were fixed at either 10 μM, 50 μM, or 300 μM. For experiments with varied NADPH concentrations, the menadione concentrations were held constant at either 50 μM or 250 μM. The resulting data were fit to competitive, uncompetitive, and noncompetitive inhibition equations using COMP, UNCOMP, and NONCOMP, respectively.32
Indazole-diones 11, 12, 15, 17, 22, and 23 were subjected to a conformation search using Hartree-Fock 3-21G* with Spartan 04.33 The two lowest energy conformers for each compound were then geometry optimized in Spartan 04 using Hartree-Fock 6-31G*. The remaining structures were subjected to a conformer search excluding the carbamate side-chain using Hartree-Fock 3-21G* with Spartan 04. For each indazole-dione, the lowest energy conformers within a calculated 2 kcal/mol range of each other were carried forward by appending the carbamate side-chain from the initial studies onto the quinone containing core. Another geometry optimization was done with the side-chain intact using Hartree-Fock 3-21G*. These structures were then imported into Gaussian 0334 and subjected to a geometry optimization and frequency calculation using Hartree-Fock 6-31G*. For each indazole-dione, the lowest energy structures were analyzed for conformational and electronic differences and similarities.
The work described in this paper was supported in part by NIH NCRR P20-RR16454 (HAC and AS), NIH NCI 1 R15 CA102119-01 (HAC and CL), Research Corporation Cottrell College Science Award CC5404 (HAC), Howard University New Faculty Grant (OB) and grant 2 G12 RR003048 from the RCMI Program, Division of Research Infrastructure, National Center for Research Resources, NIH.
Publisher's Disclaimer: 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 citable 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.