Brief Article pubs.acs.org/jmc
Chiral Dihydrobenzofuran Acids Show Potent Retinoid X Receptor− Nuclear Receptor Related 1 Protein Dimer Activation Henrik Sundén,†,⊥ Anja Schaf̈ er,†,§ Marcel Scheepstra,§ Seppe Leysen,§ Marcus Malo,† Jian-Nong Ma,‡ Ethan S. Burstein,‡ Christian Ottmann,§ Luc Brunsveld,§ and Roger Olsson*,†,‡,∥ †
Department of Chemistry and Molecular Biology, Medicinal Chemistry, University of Gothenburg, SE-41296 Gothenburg, Sweden ACADIA Pharmaceuticals Inc., San Diego, California 92130, United States § Department of Biomedical Engineering and Institute of Complex Molecular Systems, Laboratory of Chemical Biology, Technische Universiteit Eindhoven, Den Dolech 2, 5612 AZ Eindhoven, The Netherlands ∥ Chemical Biology & Therapeutics, Department of Experimental Medical Science, Lund University, Sölvegatan 19, BMC DIO, SE-221 84 Lund, Sweden ‡
S Supporting Information *
ABSTRACT: The nuclear receptor Nurr1 can be activated by RXR via heterodimerization (RXR−Nurr1) and is a promising target for treating neurodegenerative diseases. We herein report the enantioselective synthesis and SAR of sterically constricted benzofurans at RXR. The established SAR, using whole cell functional assays, lead to the full agonist 9a at RXR (pEC50 of 8.2) and RXR−Nurr1. The X-ray structure shows enantiomeric discrimination where 9a optimally addresses the ligand binding pocket of RXR.
■
INTRODUCTION Modulation of nuclear receptors, including retinoid X receptors (RXRα, β, and γ), has neuroprotective effects in animal models of neurodegenerative disorders such as Parkinson’s disease (PD), Alzheimer’s disease (AD), and multiple sclerosis (MS).1−5 RXRs are master regulator receptors that heterodimerize with other RXR subtypes and several other nuclear receptors, e.g., retinoid acid receptors (RARα, β, and γ),6 the liver X receptors (LXRα and β),7 the peroxisome proliferatoractivated receptors (PPARα, β, and γ),8 and nuclear receptor related 1 protein (Nurr1) and Nurr77,8 causing specific physiological responses by modulating gene expression. Recently, Cramer et al.3 showed that bexarotene, an RXR agonist approved for cutaneous T-cell lymphoma,9 can improve overall cognitive function in an AD mouse model. In AD, misfolded amyloid β (Aβ) peptide accumulates in the brain. Clearance of Aβ is usually facilitated by apolipoprotein E (apoE). Expression of apoE is induced by activation of RXR. When bexarotene was administered to AD mice, Aβ levels decreased. The activity in the mouse Alzheimer’s models is believed to be mediated through activation of PPARγ−RXR and LXR−RXR heterodimers which induces the expression of apoE and facilitates Aβ clearance, however, contributions by other RXR heterodimers were not evaluated.3 PD is caused by the degeneration of dopamine neurons in the ventral midbrain. Factors that influence neuronal survival © 2016 American Chemical Society
are therefore very important when it comes to developing disease modifying drugs for neurodegenerative disorders.10 In this context, the orphan receptor Nurr1 is highly interesting because of its role in the proper development, function, and survival of dopaminergic neurons.11−13 Thus, drugs that activate Nurr1 have disease modifying potential in PD. Nurr1 deficiencies have been linked to familial PD14−17 and Nurr1 knockout mice show a severe reduction in dopaminergic neurons and perinatal lethality.18,19 Moreover, activation of Nurr1 has been shown to inhibit expression of proinflammatory neurotoxic mediators in microglia and astrocytes and to promote neuronal survival by protecting them from toxic insults.20 In contrast to the majority of nuclear receptors, the ligand binding domain of Nurr1 adopts a “closed conformation”21 and therefore may be difficult to target directly with small molecule drugs. An indirect way to activate Nurr1 would be to target its potential permissive heterodimerization partner RXR. Using bioluminescence resonance energy transfer (BRET2) assays, directly monitoring interactions of Nurr1 with other nuclear receptors, we recently demonstrated that bexarotene activates RXR−Nurr1 heterodimers, rescues dopamine neurons, and restores behavioral function in a rat model of PD.1 Synthetic ligands (Figure 1), Received: November 1, 2015 Published: January 28, 2016 1232
DOI: 10.1021/acs.jmedchem.5b01702 J. Med. Chem. 2016, 59, 1232−1238
Journal of Medicinal Chemistry
■
Brief Article
RESULTS AND DISCUSSION
Synthesis. Having outlined the basis for the new scaffold design, a set of compounds based on an asymmetric synthesis of quadricyclic dihydrobenzofuran acids was commenced.26 The key intermediate allylic alcohol 4a−e (Scheme 1) was synthesized starting with an amine-catalyzed iodo-Baylis− Hillman reaction of α,β-unsaturated ketones, yielding the αiodo-carbonyls 2a−c in 56−82%.27,28 Iodo-carbonyls 2a−c were easily functionalized through a Pd/C-catalyzed Suzuki reaction to efficiently produce aryl ketones 3a−e in 65−93% yield.29 To introduce the stereogenic center in an enantioselective fashion, 3a and 3b were subjected to an asymmetric oxaborolidine-catalyzed reduction to give cyclic allylic alcohols in high yields and high enantiomeric excess. For example; RCBS catalyzed the reduction of ketone 3a and 3b to yield the (S)-alcohols 4a and 4b in 94% ee and 99% ee, respectively. The enantiomers ent-4a and ent-4b are available by reductions using the S-CBS catalyst.30 The racemic allylic alcohols 4c−e were obtained in excellent yields after reduction of the aryl ketones 3c−e under Luche conditions. Having established a highly enantioselective and robust method, the cyclic allylic alcohols were further reacted using a Mitsunobu reaction protocol with iodophenol 5a to yield the ethers 6a−e (Scheme 2).31 The reactions proceeded in moderate to good yields, however, due to a competing reaction path, the formation of the stereogenic center in 6b did not lead to complete inversion; in reactions toward 6b, and ent-6b we could detect racemization. In spite of the racemization, the arylethers could be obtained in their enantiopure form after a recrystallization in methanol. Subsequently, a palladium catalyzed intramolecular Mizoroki− Heck coupling using Pd(OAc)2, PPh3, and Ag2CO3 converted the ethers 6a−e to the quadricyclic benzofurans 7a−d (and benzofuran 10, vide infra) in generally high yield and very high dr in preference for the cis-fused ring system. The cis-fused ringsystem was determined by NOESY NMR experiments.32,33 In a few cases, the Heck product was isolated together with trace amounts of a double bond isomer. Interestingly, when conducting the Mizoroki−Heck reaction with ring ether 6e, the seven-membered carbocycle 10 is obtained as a 1:1 mixture of diastereomers. The 1:1 trans/cis ratio of 10 is believed to be the result of the higher flexibility of the cycloheptyl ring as compared to cyclohexyl. Furthermore, the intramolecular Heck reaction proceeds with full conversion of the starting material to the product and consequently only filtration as a work up is necessary. In the final step, the double bond was removed by Pd/C under an atmosphere of hydrogen. In the case of the benzyl protected compound 7d, this step yielded the free phenolic compound 14. In all other cases, treatment with aqueous sodium hydroxide produced the carboxylic acids 9a−c in moderate to good yield (24−51%) and diastereomers 12 and 13 in 50% combined yield. The heptacyclic diastereomers 12 (cis-fused) and 13 (transfused) were separated on reversed-phase preparative HPLC, and the stereochemistry was determined by NOESY NMR experiments. Pharmacological Evaluation. The dihydrobenzofuran acids were evaluated in vitro using a luciferase reporter gene assay and a BRET2 assay,1 using bexarotene as a reference (Tables 1 and 2). The racemic forms of 9a−c and 12 showed partial to full agonism with high potencies and efficacies ranging from 47 to 119% at RXR. In addition, the compounds showed selectivity; no activation was seen at estrogen receptors or RAR
Figure 1. Nuclear receptor modulators.
such as HX600,22 HX603,22 and XCT0135908,23 the latter found by Perlman and co-workers in a high throughput screening, showed selectivity for RXR−Nurr1 over a broad range of other RXR dimerization partners but with rather low potency. Nevertheless, the latter yielded promising results in a rat neuron PD model;24 the dopaminergic neurons were selectively protected from degeneration, and the effect was only seen in Nurr1-expressing neurons. Hence, RXR−Nurr1 agonists may have promise as future PD treatments but improvements regarding potency and chemotype are urgently required. The dihydrobenzofuran scaffold is a recurring motif in biologically active natural products, e.g., morphine and galanthamine. We recently took advantage of the asymmetry of the dihydrobenzofuran scaffold to design a nonplanar, potent, and highly selective estrogen receptor β (ERβ) agonist (Figure 1).25 Initial overlays with bexarotene indicated that the chiral dihydrobenzofuran scaffold could also be used to synthesize new RXR and RXR−Nurr1 ligands. The two cisfused enantiomers of the dihydrobenzofuran scaffold showed good overlays with two separate low energy conformers of bexarotene (see graphical abstract). Hence the dihydrobenzofuran scaffold presented an opportunity to design a ligand that mimics only the active bexarotene conformer, avoiding potential drawbacks originating from the in-active conformer, e.g., nontarget effects. Of note, RO01 (Figure 1) was designed on the dihydrobenzofuran enantiomer not matching the active bexarotene conformer seen in the crystal structure (Figure 2). Thus, each enantiomer of the dihydrobenzofuran scaffold shows biological activities at different receptors of the nuclear hormone receptor superfamily. In this study, we present a highly potent and selective benzofuran scaffold as a promising candidate for exploring the biological effects of RXR−Nurr1 modulation. 1233
DOI: 10.1021/acs.jmedchem.5b01702 J. Med. Chem. 2016, 59, 1232−1238
Journal of Medicinal Chemistry
Brief Article
Figure 2. (A) Crystal structure of RXR (green) in complex with 9a (shown in orange sticks and surface representation). The cofactor peptide is shown in cyan. (B) Interactions of 9a in the ligand binding pocket of RXR. The gray mesh shows the electron density for 9a, contoured at 0.61 e/Å3 (1.5σ). RXR residues making hydrophobic and hydrogen bonding interactions with 9a are shown as white and green sticks, respectively. For clarity, residues 263−279 belonging to helix 3 are not shown. Of these, I268, A271, A272, and Q275 also make hydrophobic contacts with 9a. A 2D plot of the interactions can be found in the SI. (C) Structural superposition of RXR in complex with bexarotene (purple) and 9a (orange) in the LBD of RXR. The RXR molecules are shown in different shades of green. The arrow marks the structural difference between the ligands. (D) Overlay of 9a (orange) and its enantiomer ent-9a (red) in the LBD of RXR. The arrow indicates how the enantiomer would sterically clash with RXR residue A272.
100-fold drop in potency, with a pEC50 value of 6.0. This trend is accentuated in case of the six-membered ring where the most potent enantiomer, 9b, has a pEC50 value of 7.7 (Table 1, entry 10) and the ent-enantiomer (ent-9b) is inactive (Table 1, entry 11). The fact that one of the enantiomers was inactive is not surprising considering the built-in rigidity of the ligand, causing the enantiomers to have a drastically different molecular topology (vide infra crystallography). Replacing the carboxylic acid with a hydroxyl group led to a completely inactive compound (14), underlining the importance of the carboxylic acid for RXR activity. A BRET2 assay was used to monitor the interactions of the enantiomers of the most active compounds in the dihydrobenzofuran series with RXR−RXR and RXR−Nurr1 heterodimers (Table 2). The pentacyclic dihydrobenzofuran 9a displayed a 3-fold biased interaction with RXR−Nurr1, with a pEC50 of 7.9 and 111% efficacy, as compared to RXR−RXR pEC50 of 7.6 and 85% efficacy. ent-9a was 100-fold less potent in recruiting the RXR−Nurr1 heterodimer and not active at RXR−RXR, in accordance with the trends seen in the reporter gene assay. The six-membered dihydrobenzofuran 9b displayed a similar potency for both RXR−Nurr1 and RXR−RXR dimers but is higher in efficacy for RXR−Nurr1. Again, the antipode ent-9b was almost inactive.
(data not shown). The racemic 9a showed partial agonism displaying a pEC50 value of 7.6 and efficacy of 65% at RXR (Table 1, entry 2). When the alkyl ring is expanded from five to six carbons (rac-9b), the efficacy increases to 96% (Table 1, entry 3). In comparison, the ortho-fluorocarboxylic acid (rac9c) (Table 1, entry 4) has a similar activity as the nonsubstituted rac-9b, indicating that modulating the carboxylic acid by adding an ortho-fluoro had no beneficial effects on activity. In addition, rac-9b and rac-9c are equipotent with a pEC50 value of 7.3. The cis-fused seven-membered benzofuran 12 (Table 1, entry 5) displayed a slightly lower potency and a severely reduced efficacy as compared to the other ring congeners, indicating that the bulk of the ring may be of less importance for potency but more important for efficacy (compare to rac-9a). A reason for this observation might be the increased flexibility of the seven-membered ring. Conversely, the trans-fused diastereomer 13 was inactive (Table 1, entry 6), indicating that the molecular topology that is determined by the stereochemistry at the two stereogenic centers is critical for the activity of the benzofuran scaffold. Thus, the enantiomers of the two most potent compounds were investigated separately (Table 1, entries 8−11). The pentacyclic enantiomer 9a turned out to be a full agonist and the most potent compound in the series, with a pEC50 value of 8.2 and efficacy of 119%. By contrast, ent-9a displays an over 1234
DOI: 10.1021/acs.jmedchem.5b01702 J. Med. Chem. 2016, 59, 1232−1238
Journal of Medicinal Chemistry
Brief Article
Scheme 1. Synthesis of 4a−ea
Scheme 2. Synthesis of 9a−c, 12, 13, and 14a
v a
(a) I2, DABCO, 2a 56%, 2b 82%. (b) 2c: I2, pyridine, 59%. (c) Pd/C, Na2CO3, ArB(OH)2, 3a 68%, 3b 93%, 3c 85%, 3d 91%, 3e 65%. (d) (R)-CBS, BMS, 4a 52%, 94% ee, 4b 77%, 99% ee. (e) CeCl3·7H2O, NaBH4, 4c 95%, 4d 77%, 4e 74%.
Structural Evaluation. To elucidate the structural basis for the highly potent RXR interaction of the dihydrobenzofuran compounds, we set up crystal screens with the most potent compound 9a, RXRα and a TIF2 NR2 cofactor peptide (686KHKILHRLLQDSS-698; for details, see Supporting Information (SI)).34 After five days, crystals were observed and directly flash-cooled in liquid nitrogen for X-ray crystallographic structure determination. 9a fills the classical ligand binding pocket, with helix 12 adopting the typical agonist conformation which facilitates binding of the coactivator peptide (Figure 2A). The clear electron density for enantiomer 9a in the crystal structure reveals how the molecule is accommodated in the ligand binding pocket. Extensive hydrophobic interactions are formed with residues located in RXRα helices 3, 5, 10, and 11, among them Val 342, Val 349, Ile 310, Ile 268, and Leu 436 (Figure 2B and SI, Figure S1). The carboxylic acid functional group of 9a creates a charge-assisted hydrogen bond with Arg 316 and a hydrogen bond with the backbone amide nitrogen of Ala 327. These polar interactions are important for binding as evidenced by the complete inactivity of hydroxyl group congener rac-14. When 9a is superimposed with bexarotene for comparison, the two molecules overlay nicely (Figure 2C). A small but significant difference is that the five-membered carbocycle occupies slightly more space than the ethylene group, filling a cavity formed by amino acids of helix 5. The cavity is most likely more spacious, as the SAR shows that both six- and seven-membered rings also can be accommodated here. Moreover, the X-ray structure provides a structural basis for the observed enantioselectivity of the RXR agonist. Accordingly, benzofuran ent-9a encounters a steric clash with amino acid Ala 272 from helix 2 and fails to establish the hydrogen bond with the backbone nitrogen in Ala 327 and the salt bridge with Arg 316 (Figure 2D), explaining why ent-9a and ent-9b show little or no activity in the biological testing.
a (a) DIAD, PPh3, toluene, 0 °C to rt; 6a 100%, 98% ee; 6b 77%, > 99% ee; 6c 44%; 6d 49%, 6e 43%. (b) Pd(OAc)2, PPh3, Ag2CO3, 7a− d crude yields 94−100%. (c) Pd/C, H2, 14 85%. (d) NaOH (9 M aq) 9a 51%, 9b 24%, 9c 32%, 12 and 13 50% yield (both diastereomers).
Table 1. Biological Evaluation: Luciferase Reporter Gene Assay for RXR Bindinga RXR entry
compd
1 2 3 4 5 6 7 8 9 10 11
bexarotene rac-9a rac-9b rac-9c rac-12 rac-13 rac-14 9a ent-9a 9b ent-9b
pEC50
efficacy (%)
± 0.1 ± 0.1 ± 0.1 ± 0.1 ± 0.1 NA NA 8.2 ± 0.1 6.0 ± 0.1 7.7 ± 0.1 NA
101 ± 5 65 ± 7 96 ± 6 88 ± 22 47 ± 7
7.5 7.6 7.3 7.3 6.9
119 ± 16 59 ± 4 94 ± 20
a NA = not active. pEC50 is the negative logarithm of EC50 in molar. Agonist efficacies were compared to that of bexarotene (100%). Values represent the mean ± SD of three or more independent experiments (n ≥ 3).
■
CONCLUSION In conclusion, we have presented an efficient seven-step asymmetric synthesis for a new class of conformationally rigid RXR agonists. A focused library of highly potent compounds has been synthesized and pharmacologically evaluated using 1235
DOI: 10.1021/acs.jmedchem.5b01702 J. Med. Chem. 2016, 59, 1232−1238
Journal of Medicinal Chemistry
Brief Article
62.45, 41.11, 35.89, 35.21, 35.12, 34.63, 34.06, 32.27, 32.13, 32.02, 32.00, 29.70, 24.83; m/z (ESI) 391 ([M + H]+, 100%), 279 (9), 186 (5); [α]20 D = +16.1 (c 0.3, methanol). HPLC: purity, 97.1%; retention time, 15.85 min. HRMS (ESI) calcd for C26H30O3 [M + H]+, 391.2267; found, 391.2243 (ent-9a); [α]20 D = −20.8 (c 0.8, methanol). HPLC: purity:, 97.4%; retention time, 15.85 min 4-[(11S,16R)-4,4,7,7-Tetramethyl-17-oxatetracyclo [8.7.0.03,8.011,16]heptadeca-1(10),2,8-trien-11-yl]benzoic Acid (9b). 1H NMR (400 MHz, CDCl3) δ 8.04 (d, J = 8.6 Hz, 2H), 7.50 (d, J = 8.6 Hz, 2H), 6.80 (s, 1H), 6.75 (s, 1H), 4.81 (t, J = 3.9 Hz, 1H), 2.38−2.26 (m, 1H), 2.07−1.98 (m, 1H), 1.92−1.83 (m, 1H), 1.75−1.49 (m, 9H), 1.27 (s, 6H), 1.19 (s, 3H), 1.13 (s, 3H). 13C NMR (101 MHz, CDCl3) δ 169.70, 156.58, 151.46, 145.12, 137.58, 134.78, 130.06, 127.91, 126.99, 121.10, 107.30, 89.15, 52.12, 35.23, 35.11, 34.66, 34.11, 33.97, 32.23, 32.05, 32.01, 31.99, 26.49, 21.72, 19.78; [α]20 D = +16.1 (c 0.9, CHCl3); retention time HPLC (min), 16.53. Purity: 95.0%. HRMS (ESI) calcd for C27H33O3 [M + H]+, 405.2423; found, 405.2402 (ent-9b); [α]20 D = −13.6 (c 0.5, CHCl3); retention time HPLC (min), 16.53. Purity: 95.0%, 3 mg, 11% yield. 2 - F l u o r o- 4- {4 , 4 , 7 , 7 - t e t ra m e t h y l - 17 - o x at e t r a c yc l o [8.7.0.03,8.011,16]]heptadeca-1(10),2,8-trien-11-yl}benzoic Acid (rac-9c). 1H NMR (400 MHz, CDCl3) δ 7.97 (t, J = 8.1 Hz, 1H), 7.26 (dd, J = 8.3, 1.8 Hz, 1H), 7.19 (dd, J = 12.9, 1.7 Hz, 1H), 6.80 (s, 1H), 6.76 (s, 1H), 4.77 (t, J = 4.1 Hz, 1H), 2.25 (dd, J = 14.1, 5.6 Hz, 1H), 1.99 (dd, J = 11.5, 6.0 Hz, 1H), 1.95−1.82 (m, 1H), 1.79−1.49 (m, 9H), 1.27 (s, 6H), 1.20 (s, 3H), 1.15 (s, 3H). 13C NMR (101 MHz, CDCl3) δ 167.90 (d, J = 6.5 Hz), 162.43 (d, J = 260.8 Hz), 156.60 (s), 154.75 (d, J = 7.9 Hz), 145.52 (s), 137.82 (s), 133.84 (s), 132.47 (s), 123.48 (d, J = 3.2 Hz), 121.03 (s), 116.46 (d, J = 23.5 Hz), 115.56−114.54 (m), 107.52 (s), 88.89 (s), 52.24 (s), 35.23 (s), 35.12 (s), 34.71 (s), 34.14 (s), 33.71 (s), 32.26 (s), 32.08 (s), 32.01 (s), 32.00 (s), 26.59 (s), 21.53 (s), 19.67 (s). 19F NMR (376 MHz, CDCl3) δ −75.80 (s, ethyl trifluoroacetate), −107.75 (dd, J = 12.8, 7.9 Hz). Retention time HPLC (min): 16.18. Purity: 92.4%. HRMS (ESI) calcd for C27H31FO3 [M + H]+, 423.2329; found, 423.2307 4-{13,13,16,16-Tetramethyl-9-oxatetracyclo[8.8.0.02,8.012,17]octadeca-1(18),10,12(17)-trien-2-yl}benzoic Acid (12 and 13). 12: 1H NMR (400 MHz, CDCl3) δ 8.01 (d, J = 8.7 Hz, 2H), 7.40 (d, J = 8.7 Hz, 2H), 6.88 (s, 1H), 6.77 (s, 1H), 4.89 (dd, J = 7.3, 1.8 Hz, 1H), 2.41 (dd, J = 14.5, 10.3 Hz, 1H), 2.21 (m, 1H), 2.11 (dd, J = 14.42, 8.30 Hz, 1H), 1.91 (m, 1H), 1.64 (m, 8H), 1.42 (m, 1H), 1.29 (s, 3H), 1.27 (s, 3H), 1.25 (m, 1H), 1.22 (s, 3H), 1.20 (s, 3H). 13C NMR (101 MHz; CDCl3) δ 170.46, 157.08, 155.61, 145.63, 137.08, 130.36, 126.82, 126.47, 122.69, 106.22, 93.69, 57.88, 38.19, 35.24, 35.17, 34.67, 34.05, 32.31, 32.16, 32.08, 32.01, 31.81, 31.04, 29.71, 25.27, 23.43; m/z (ESI) 419 ([M + H]+, 100%), 282 (40). Retention time HPLC (min): 17.99. Purity: 93.5%. HRMS (ESI) calcd for C28H34O3 [M + H]+, 419.2580; found, 419.2600. 13: 1H NMR (400 MHz, CDCl3) δ 7.97 (d, J = 8.6 Hz, 2H), 7.56 (d, J = 8.6 Hz, 2H), 6.80 (s, 1H), 6.78 (s, 1H), 4.88 (dd, J = 11.5, 6.6 Hz, 1H), 3.02 (m, 1H), 2.24 (m, 2H), 1.89 (m, 3H), 1.59 (m, 6H), 1.47 (m, 1H), 1.25 (s, 3H), 1.21 (s, 3H), 1.18 (s, 3H), 1.04 (s, 3H), 0.90 (m, 1H). 13C NMR (101 MHz; CDCl3) δ 171.41, 155.88, 149.44, 145.02, 137.78, 136.84, 130.23, 126.85, 126.40, 120.54, 107.76, 92.46, 54.40, 35.15, 35.01, 34.61, 34.07, 32.38, 32.03, 31.89, 31.77, 29.70, 27.09, 26.31, 25.02, 23.15; m/z (ESI) 419 ([M + H]+, 100%), 254 (48). Retention time HPLC (min): 16.75. Purity: 98.9%. HRMS (ESI) calcd for C28H34O3 [M + H]+, 419.2580; found, 419.2597.
Table 2. Biological Evaluation Using BRET2 Assays for RXR−Nurr1 and RXR−RXR Dimersa RXR−Nurr1 entry
compd
pEC50
1 2 3 4 5
bexarotene 9a ent-9a 9b ent-9b
8.0 ± 0.1 7.9 ± 0.1 6.3 ± 0.1 7.9 ± 0.3