Bioorganic & Medicinal Chemistry 12 (2004) 3451–3469
`
`15-Fluoro prostaglandin FP agonists: a new class of
`topical ocular hypotensives
`
`Peter Klimko,* Mark Hellberg, Marsha McLaughlin, Najam Sharif, Bryon Severns,
`Gary Williams, Karen Haggard and John Liao
`
`Alcon Research, Ltd, Pharmaceutical Products Research, 6201 S. Freeway, MS R2-39, Ft. Worth, TX 76134, USA
`
`Received 10 February 2004; revised 23 April 2004; accepted 28 April 2004
`Available online 25 May 2004
`
`Abstract—A novel series of 15-fluoro prostaglandins with phenoxy termination of the x-chain was synthesized and evaluated for
`binding and functional activation of the prostaglandin FP receptor in vitro and for side effect potential and topical ocular hypo-
`tensive efficacy in vivo. Compounds with the 15a-fluoride relative stereochemistry displayed EC50 values of 6 20 nM, comparable to
`the value for the endogenous ligand PGF2a. Evaluation of selected ester prodrugs of these 15-fluoro prostaglandins in vivo high-
`lighted their generally low propensity to elicit hyperemia or ocular irritation in rabbits and efficacious intraocular pressure-lowering
`property in monkeys. In particular 13,14-dihydro-15-deoxy-15a-fluoro-16-aryloxy-x-tetranor-cis-D4-PGF2a isopropyl ester (24)
`caused relatively little ocular irritation in rabbits while lowering intraocular pressure in conscious ocular hypertensive monkeys by
`39% following a topical ocular dose of 3 lg.
`Ó 2004 Elsevier Ltd. All rights reserved.
`
`1. Introduction
`
`Glaucoma, a heterogeneous family of optic neuro-
`pathies, is one of the leading causes of blindness in the
`developed world. Elevated intraocular pressure (IOP) is
`an important risk factor for loss of visual field due to
`optic nerve damage.1 Endogenous prostaglandins and
`their prodrugs, such as PGF2a isopropyl ester, reduce
`IOP in monkeys and in man, but also cause conjunctival
`hyperemia, foreign-body sensation, and stinging.2 Pro-
`drugs of potent, selective prostaglandin FP receptor
`agonists exhibit similar IOP-lowering potencies and
`efficacies as their endogenous counterparts but elicit
`greatly reduced ocular side effects. Three such synthetic
`prostaglandin analogs-travoprost,3a latanoprost,3b and
`bimatoprost3c––are the active ingredients of topically
`active, once-a-day IOP-lowering medications (Fig. 1).
`Their introduction to clinical practice has revolutionized
`the treatment of glaucoma.
`
`Our continuing interest in discovering novel structures
`with prostaglandin FP receptor agonist activity led us
`
`Keywords: 15-Fluoro prostaglandin; Glaucoma.
`* Corresponding author. Tel.: +1-817-551-6932; fax: +1-817-615-4696;
`e-mail: peter.klimko@alconlabs.com
`
`0968-0896/$ - see front matter Ó 2004 Elsevier Ltd. All rights reserved.
`doi:10.1016/j.bmc.2004.04.034
`
`Figure 1.
`
`to consider the 15-deoxy-15-fluoro (termed 15-fluoro
`hereinafter) structural motif. The replacement of the
`carbon 15-hydroxyl group of PGF2a with a fluorine
`atom should profoundly affect many physicochemical
`properties of the molecule.4 The volume occupied by the
`C–F group (bond distance 1.38 A,5a F van der Waals
`radius 1.47 A6) is smaller than that for the C–O–H
`array (bond lengths: 1.43 A5b for a C–O bond and
`0.96 A5c for an O–H bond; van der Waals radii: 1.52 A6
`for O, 1.20 A6
`for H). Compared to the hydroxyl
`group the carbon-bound fluorine atom has no hydrogen
`bond-donating capacity and has a diminished hydrogen
`
`IPR Page 1/19
`
`Santen/Asahi Glass Exhibit 2021
`Micro Labs v. Santen Pharm. and Asahi Glass
`IPR2017-01434
`
`

`

`3452
`
`P. Klimko et al. / Bioorg. Med. Chem. 12 (2004) 3451–3469
`
`bond-accepting ability, although there is some contro-
`versy as to the magnitude of this effect.7 Due to the
`greater electronegativity of fluorine compared to oxy-
`gen, the fluorinated carbon should be more electroneg-
`ative than its hydroxylated congener. In a related
`fashion, the lower energy of the C–F, as compared to
`the C–O, r orbital can reinforce unusual stereoelec-
`tronic-based conformational properties (e.g., the ano-
`meric effect) due to energetically favorable n=r or p=r
`overlap.8 Finally, fluorine-for-hydroxyl
`substitution
`significantly increases lipophilicity.
`
`Given the structure–activity relationship insight that
`could be gained by this substitution, there are surpris-
`ingly few reports of this motif. This is perhaps due to the
`belief that hydroxyl substitution at carbon 15 is neces-
`sary for potent biological activity.9 The most systematic
`studies in the literature are those of Bezuglov et al., who
`reported the synthesis, ex vivo biological studies, and
`physicochemical properties of a variety of 15-fluoro
`analogs of endogenous prostaglandins,
`including 15-
`fluoro-PGF2a and 15-fluoro-PGE2 (Fig. 2).10
`
`Based on the effect of 15-fluoro-PGF2a on smooth
`muscle tone of several organs in rats, hamsters, and
`guinea pigs, it was theorized that this substitution in-
`creased EP receptor affinity.10c Interestingly, 15-fluoro
`PGE2 itself demonstrated reduced biological activity
`compared to PGE2.
`
`the synthesis and pharmacological
`We now report
`characterization of a series of 15-fluoro-16-aryloxy-x-
`tetranor-PGF2a analogs I (Fig. 3).11;12 Many of these
`compounds potently activated the prostaglandin FP
`receptor in vitro, and several effectively lowered IOP in
`conscious lasered (ocular hypertensive due to trabecu-
`loplasty) monkeys while causing relatively little ocular
`irritation in rabbits. In particular, phenoxy-terminated
`congener 24 elicited minimal ocular hyperemic response
`in rabbits, and a twice-a-day 3 lg dose lowered IOP in
`monkeys by 39%.
`
`Figure 2.
`
`Figure 3.
`
`2. Results and discussion
`
`2.1. Chemistry
`
`The syntheses of the 16-phenoxy terminated 15-fluoro
`compounds are illustrative. Reduction of known enone
`113 with (+)-B-chlorodiisopinocampheylborane14
`[(+)-
`DIP-Cl] afforded predominantly the b-alcohol 2 with
`8:1 stereoselectivity; diastereomerically pure 2 was
`isolated in 35% yield after flash chromatography
`(Scheme 1). In a complementary fashion, reduction with
`the antipode ())-DIP-Cl afforded the diastereomerically
`pure epimeric a-alcohol 3 in 50% yield after chroma-
`tography. Alternatively, carbonyl reduction of 1 with
`CeCl3/NaBH4 provided the alcohols 2 and 3 as a 1:1
`mixture. Fluorination of the allyl alcohol functionality
`with (diethylaminosulfur)trifluoride (DAST) was not
`stereospecific; using either the diastereomerically pure 2
`or the 1:1 mixture 2/3 as the starting material afforded
`the same mixture of four allylic fluorides 4–5. The lack
`of stereospecificity in the fluorination implies ionization
`at carbon 15 to generate the same intermediate
`from either diastereomeric alcohol, followed by attack
`of fluoride with retention or inversion at the ipso- or
`b-positions. One diastereomer of 5 and the two diaste-
`reo-mers 4 coeluted chromatographically, but were
`separable from the other diasteromer of 5. Debenzoy-
`lation of the 4+one diastereomer of 5 mixture provided
`alcohols 6, which were separable from the alcohol de-
`rived from 5. The alcohols 6 were then protected as their
`THP ethers 7. DIBAL-H reduction of the lactone to the
`corresponding lactol, followed by Wittig condensation
`with Ph3P@CH(CH2)4CO2K and esterification of the
`crude product mixture gave 8. Deprotection of 8 and
`HPLC purification afforded the 15a- and 15b-fluoride
`isopropyl esters 9 and 10, which were saponified to their
`corresponding acids 11 and 12 under standard condi-
`tions. In this and all subsequent cases, the carbon 15
`absolute stereochemistry was tentatively assigned based
`on the EC50 values of the acids in the prostaglandin FP
`receptor functional assay (see Table 1), the more potent
`diastereomer being assigned as the 15a-fluoride.15
`
`With respect to the 13,14-dihydro series, hydrogenation
`of b-alcohol 2 followed by DAST fluorination afforded
`13 (Scheme 2). In contrast to the allyl alcohol case,
`DAST fluorination was
`stereospecific:
`the
`same
`sequence applied to a-alcohol 3 afforded the inverted
`product 15. In each case the fluorination product was
`accompanied by about 20% of an HF elimination
`by-product, which could be separated by flash chromato-
`graphy either immediately post-fluorination or at the
`end of the synthesis. Protecting group interchange,
`lactone to lactol reduction, a-chain installation/esterifi-
`cation, and deprotection as above afforded 13,14-dihy-
`dro-15-fluoro prostaglandin esters 17 and 18, which
`were saponified to the acids 19 and 20.
`
`Synthesis of the 13,14-dihydro-cis-D4 congener16 diver-
`ged from 14 (Scheme 3). For convenience, the afore-
`mentioned HF elimination contaminant (see above)
`from the DAST fluorination step was carried forward
`until HPLC isolation of 24 at the end of the synthesis.
`
`IPR Page 2/19
`
`

`

`P. Klimko et al. / Bioorg. Med. Chem. 12 (2004) 3451–3469
`
`3453
`
`Scheme 1.
`
`Compound
`
`Table 1. FP receptor functional response and binding data
`Ki  SEMa (nM)
`210 ± 64
`1100 ± 110
`52 ± 1
`3900 ± 2200
`690 ± 140
`360 ± 27
`38 ± 12
`230 ± 23
`240 ± 130
`182 ± 15
`160 ± 22
`3800 ± 160
`129 ± 12
`92 ± 14
`31 ± 2
`52 ± 2
`22 ± 5
`
`11
`12
`19
`20
`25
`28
`30
`32
`34
`36
`38
`40
`PGF2a
`Latanoprost acid
`Cloprostenol
`Travoprost acid
`16-Phenoxy-x-tetranor-PGF2a
`a SEM¼ Standard error of the mean.
`
`EC50 ± SEM (nM)
`
`Response (%)
`
`4 ± 0.5
`96 ± 16
`10 ± 2
`154 ± 6
`11 ± 1
`183 ± 41
`20 ± 11
`13 ± 3
`2 ± 0.5
`81 ± 10
`13 ± 5
`131 ± 35
`24.5 ± 0.92
`34.4 ± 5.2
`1 ± 0.04
`2.7 ± 0.28
`1 ± 0.4
`
`104
`97
`79
`65
`81
`91
`64
`85
`84
`76
`83
`44
`92
`75
`100
`100
`100
`
`IPR Page 3/19
`
`

`

`3454
`
`P. Klimko et al. / Bioorg. Med. Chem. 12 (2004) 3451–3469
`
`Scheme 2.
`
`Scheme 3.
`
`the C-11 oxygen afforded lactone 21.
`Silylation of
`Reduction to the lactol using DIBAL-H was followed
`by Wittig condensation to provide enol ether 22. Acidic
`hydrolysis gave homologated lactol 23, which under-
`went olefination in the usual manner and TBAF-medi-
`ated desilylation to provide 24 as a 96:4 mixture of
`4Z:4E olefin geometrical isomers after HPLC purifica-
`tion. Saponification then afforded 25.
`
`As a final synthetic note, we were unable to prepare 15-
`fluoro-PGF2a itself. The immediate post-fluorination
`product 26, an allyl alkyl fluoride,
`largely degraded
`upon standing over 1–2 days, even in solution at 4 °C
`(Fig. 4). An odor of H–F could be detected from the
`degraded sample, and 600 MHz 1H/150 MHz 13C NMR
`analyses of the crude showed a complex mixture whose
`components mostly lacked a C–F bond. Interestingly the
`
`IPR Page 4/19
`
`

`

`P. Klimko et al. / Bioorg. Med. Chem. 12 (2004) 3451–3469
`
`3455
`
`Figure 4.
`
`13,14-dihydro-15-fluoro analog 27, a dialkyl fluoride,
`and products synthetically derived from it were stable at
`room temperature indefinitely. Therefore for compara-
`tive biological evaluation we prepared 13,14-dihydro-15-
`deoxy-15a-fluoro-PGF2a (28) and its isopropyl ester
`(29). We currently have no satisfactory explanation for
`the enhanced stabilities of dialkyl fluorides such as 27
`and allyl (phenoxyalkyl) fluorides like 4 over allyl alkyl
`fluoride 26.17
`
`2.2. Pharmacology
`
`The compounds shown in Figure 5 were evaluated for
`their binding affinity and functional efficacy at the FP
`prostaglandin receptor, for side effect potential in the
`rabbit ocular irritation (ROI) model, for topical ocular
`potency in the cat pupil diameter (CPD) constriction
`model, and for IOP lowering in the lasered ocular
`
`hypertensive monkey model. The carboxylic acid was
`used in all in vitro studies since it is believed to be
`the pharmacologically active form of the compound. The
`corresponding isopropyl ester prodrugs were used in the
`in vivo experiments to facilitate corneal penetration and
`delivery of the carboxylic acid to the aqueous humor.
`
`2.3. In vitro studies
`
`2.3.1. FP receptor functional response and binding. Table
`1 summarizes our evaluation of 15-fluoro prostaglandin
`acids for binding to an FP receptor expressed in
`luteum (Ki),18a and for
`functional
`bovine corpus
`potency (EC50 ¼ effective concentration necessary for a
`compound to attain 50% of its maximal response) and
`efficacy via stimulation of FP receptor-linked phospho-
`inositide turnover in Swiss 3T3 mouse fibroblast cells.18b
`The standards PGF2a, latanoprost acid, cloprostenol,
`
`Figure 5. Structure of prostaglandin analogs evaluated in vitro and in vivo.
`
`IPR Page 5/19
`
`

`

`3456
`
`P. Klimko et al. / Bioorg. Med. Chem. 12 (2004) 3451–3469
`
`Table 2. Cat, rabbit, monkey, and lipophilicity data
`
`Ester (corresponding acid)
`
`ROI15
`
`a (lg)
`
`CPD,b ED5 (lg)
`
`Monkey IOP, % change (dose in lg)
`
`RRIc
`
`––d
`9 (11)
`0.3
`0.05
`10 (12)
`1
`0.5
`––
`17 (19)
`10
`0.2
`––
`)39% (3)
`24 (25)
`10
`0.04
`)33% (10)
`31 (30)
`100
`1
`33 (32)
`100
`10
`––
`)18% (1)
`35 (34)
`10
`0.1
`37 (36)
`>100
`1
`––
`39 (38)
`10
`1.2
`––
`)38% (1)
`PGF2a isopropyl ester
`<0.1
`0.03
`)29% (0.3)
`Travoprost
`3
`0.015
`)27% (3)
`Latanoprost
`1.8
`0.13
`)39% (1)
`Cloprostenol isopropyl ester
`0.3
`0.013
`a ROI15 ¼ Dose estimated to produce conjunctival hyperemia in 15% of the tested rabbits over 4 h.
`b CPD¼ Cat pupil diameter constriction.
`c RRI¼ Relative retention index.
`d Not tested.
`
`1.79
`––
`2.41
`2.46
`4.13
`4.96
`2.96
`––
`3.51
`1.00
`1.16
`1.05
`––
`
`travoprost acid, and 16-phenoxy-x-tetranor-PGF2a are
`included for comparison.
`
`tended to have increased receptor binding affinity but
`were less potent in the functional assay.
`
`2.4. In vivo studies
`
`Conjunctival hyperemia was studied in New Zealand
`Albino rabbits.3a ROI15 denotes the dose estimated to
`produce 15% incidence of hyperemia over the 4 h course
`of the study. As a preliminary assay of topical ocular
`potency, the ability of a test compound to constrict the
`cat pupil over time was measured and is expressed as an
`ED5 value,3a indicating the dose estimated to produce a
`5 unit area (mm h) in a graph of the difference in pupil
`diameter in the dosed eye versus time. Acute IOP-low-
`ering efficacy was measured in conscious ocular hyper-
`tensive cynomolgus monkeys.3a These parameters for
`selected compounds are shown in Table 2.
`
`The high hydrophobicity of prostaglandins in general
`and these 15-fluoro analogs in particular caused exper-
`imental determination of octanol–water partition co-
`efficients to be imprecise and not very useful
`for
`comparison purposes. As an alternative, for selected
`compounds an HPLC-based relative retention index
`(RRI) was used as a surrogate measure of distribution.
`Retention times were measured for the compounds on a
`reverse-phase HPLC column using acetonitrile/pH 3
`ammonium phosphate buffer elution. The RRI is de-
`fined as the retention time of the test item divided by
`that for PGF2a isopropyl ester. In general, the higher
`a compound’s RRI value, the more lipophilic it is.
`
`The corresponding data for the standards cloprostenol
`isopropyl ester,
`latanoprost,
`travoprost, and PGF2a
`isopropyl ester are included in Table 2 for comparison.
`
`The 15-fluoro prostaglandin analogs had approximately
`10-fold lower affinity for, while being 2–5-fold less po-
`tent for functional activation of the FP receptor as
`compared to the corresponding 15-hydroxy analogs (11,
`34, and 38 vs 16-phenoxy-x-tetranor-PGF2a, cloproste-
`nol, and travoprost acid). The 13,14-dihydro analogs
`
`Topical ocular efficacy of a prostaglandin analog de-
`pends on its inherent activity, metabolic stability, and
`bioavailability. To be effective, an ester prodrug must be
`absorbed and hydrolyzed by ocular tissue and the
`resulting carboxylic acid delivered to the trabecular
`meshwork or ciliary muscle, which are the presumed
`target tissues. In the 15-fluoro series there is neither a
`strong correlation between FP receptor binding affinity
`and functional efficacy nor between either in vitro
`binding affinity or functional efficacy and in vivo effi-
`cacy. However within a series of 15a-fluoride ester
`prodrugs whose acids had similar Ki values (e.g., com-
`pounds 11, 32, 34, and 38, with Ki ¼ 160–240 nM), less
`lipophilic compounds were more potent in the cat pupil
`diameter assay. Similarly the two 15a-fluoride acids 11
`and 34 and the 15-hydroxy compound travoprost acid
`all had similar functional potency (EC50 ¼ 1–4 nM), yet
`for their corresponding isopropyl esters lower lipophi-
`licity afforded higher cat pupil diameter potency.
`
`The cis-D4 analog 25 is noteworthy because although it
`had modest affinity for the FP receptor (Ki ¼ 690 nM), it
`was a fairly potent agonist in the functional assay
`(EC50 ¼ 11 nM). The corresponding isopropyl ester 24
`was surprisingly potent in the cat pupil diameter assay,
`while demonstrating good separation between its cat
`ED5 (0.04 lg) and ROH15 (10 lg) values. The compound
`was also very effective in lowering IOP in ocular
`hypertensive monkeys. The cis-D4 modification is known
`to inhibit a-chain metabolism of PGF2a in monkeys.19
`Thus the in vivo efficacy of 24 may be due in part to its
`enhanced metabolic stability, both in the a-chain due to
`the D5 to D4 olefin shift and in the x-chain due to the
`lack of the 15-hydroxy group.
`
`3. Conclusion
`
`Substitution of a fluorine for the 15-hydroxy group of
`select prostaglandin FP agonists decreased in vitro
`
`IPR Page 6/19
`
`

`

`P. Klimko et al. / Bioorg. Med. Chem. 12 (2004) 3451–3469
`
`3457
`
`receptor affinity markedly and functional potency less
`dramatically. As compared to their 15-hydroxy cong-
`eners, the 15-fluoro prostaglandin ester prodrugs were
`less potent but still effective in the cat, and tended to be
`less hyperemic in the rabbit. Notably compound 24
`(13,14-dihydro-15-deoxy-15a-fluoro-16-aryloxy-x-tetra-
`nor-cis-D4-PGF2a isopropyl ester) potently constricted
`the cat pupil, induced minimal hyperemic response in
`the rabbit, and lowered monkey IOP by 39%. This
`profile is consistent with other prostaglandin analogs
`shown to be effective ocular hypotensive agents in
`humans.
`
`4. Experimental section
`
`4.1. Chemistry general methods
`
`Abbreviations used include: calcd, calculated; DAST
`(diethylamino)sulfur trifluoride; DBU, 1,8-diazabicylo-
`[5.4.0]undec-7-ene; DHP, 3,4-dihydro-2H-pyran; DMAP,
`4-(dimethylamino)pyridine; DIBAL-H,
`diisobutyl-
`aluminum hydride; (+)- and ())-DIP chloride, (+)- and
`())-B-chlorodiisopinocampheylborane; MF, molecular
`formula; MW, molecular weight; PTFE, poly(terfluoro-
`ethylene); p-TsOH, p-toluenesulfonic acid monohydrate;
`TBAF, tetra-n-butylammonium fluoride. Unless other-
`wise noted, all 1H NMR and 13C NMR spectra were
`acquired in CDCl3 solvent. All 200 MHz 1H NMR and
`50 MHz 13C NMR spectra were acquired on a Varian
`Gemini 200 spectrometer. All 600 MHz 1H NMR and
`150 MHz 13C NMR spectra were acquired on a Bruker
`DRX600 spectrometer. For reactions without added
`water, solvents used were anhydrous grade from Aldrich
`Chemical Company and were used without further
`purification. PGF2a and its isopropyl ester and 16-
`phenoxy-x-tetranor-PGF2a were purchased from Cay-
`man Chemical Company, Ann Arbor, Michigan, USA,
`and were used as received. Latanoprost and its acid20
`was synthesized in-house according to published pro-
`cedures, as were travoprost and its acid and cloprostenol
`and its isopropyl ester.21 Unless otherwise stated, all
`reactions were run under a positive pressure of nitrogen,
`and all temperatures quoted refer to external tempera-
`tures. Concentration refers to removal of solvent in
`vacuo on a rotary evaporator. Reactions were moni-
`tored by TLC on E. Merck Silica Gel 60 F254 plates, with
`visualization by UV light or staining with either ethan-
`olic phosphomolybdic acid or 2% aqueous KMnO4.
`Column chromatographic purifications were performed
`under positive air flow using 230–400 mesh silica gel
`from E.M. Science. Chromatography solvents used were
`HPLC grade from E.M. Science. Electrospray low
`resolution mass spectra (ES-LRMS) were acquired on a
`Finnegan TSQ 46 triple quadrupole mass spectrometer
`operating in the positive electrospray mode. Matrix-
`assisted laser desorption ionization low resolution mass
`spectra (MALDI-LRMS) were acquired on a Voyager
`RP laser desorption time-of-flight mass spectrometer.
`High resolution mass spectra (HRMS) were acquired by
`Analytical Instrument Group, Raleigh, NC; the spectra
`were acquired using the Fast Atom Bombardment
`Mode.
`
`[3aR,4R(1E,3S),5R,6aS]-5-Benzoyloxy-4-(3-hydr-
`4.1.1.
`oxy-4-phenoxy-1-buten-1-yl)-hexahydro-2H-cyclopenta-
`[b]furan-2-one (2). To a mixture of enone 113 (2.97 g,
`7.31 mmol) in THF (25 mL) at 0 °C was added via syr-
`inge over 20 min a solution of (+)-DIP chloride (5.0 g,
`15.6 mmol) in THF (20 mL). After 90 min the reaction
`mixture was warmed to room temperature and stirred
`overnight. Methanol (5 mL) was added, the mixture was
`stirred for 15 min, and 1 M HCl (75 mL) was added. The
`solution was extracted with ethyl acetate (3 · 75 mL),
`dried (MgSO4), filtered, and concentrated, and the res-
`idue was dissolved in 100 mL of 1:1 CH3CN–hexane.
`The bottom CH3CN layer was isolated and extracted
`with hexane (1 · 50 mL), concentrated, and chromato-
`graphed on a 15 cm tall · 45 mm diameter silica gel col-
`umn eluting with 1:1 ethyl acetate–hexane to afford
`three components; the slowest eluting was a diastereo-
`merically pure ( P99:1) sample of alcohol 2 (1.03 g,
`35%), the middle eluting was a 9:1 mixture of 2 and its
`diastereomeric alcohol 3 (820 mg, 27%), and the fastest
`eluting was a 1:1 mixture of 2:3 (579 mg, 19%). Spec-
`tral data for 2: 13C NMR (50 MHz): d 176.36 (C), 166.03
`(C), 158.29 (C), 133.37 (CH), 131.14 (CH), 131.00 (CH),
`129.65 (CH), 129.57 (CH), 128.54 (CH), 121.39 (CH),
`114.60 (CH), 83.28 (CH), 79.03 (CH), 71.56 (CH2),
`70.41 (CH), 54.30 (CH), 42.67 (CH), 37.57 (CH2), 34.94
`(CH2). MALDI-LRMS, m=z calcd for C24H24O6Na
`[(M+Na)þ], 431; found 431.
`
`[3aR,4R(1E,3RS),5R,6aS]-5-Benzoyloxy-4-(3-flu-
`4.1.2.
`oro-4-phenoxybutenyl)-hexahydro-2H-cyclopenta[b]furan-
`2-one (4) and [3aR,4R(2E,1RS),5R,6aS]-5-Benzoyloxy-4-
`(1-fluoro-4-phenoxy-2-buten-1-yl)-hexahydro-2H-cyclo-
`penta[b]furan-2-one (5).
`
`4.1.2.1. Demonstration run. To a solution of 2
`(182 mg, 0.45 mmol) in CH2Cl2 (1.6 mL) at 0 °C was
`added dropwise via syringe (diethylamino)sulfur triflu-
`oride (DAST; 110 mg, 0.69 mmol). After 1 h saturated
`NaHCO3 (3 mL) and ethyl acetate (4 mL) were added,
`and the mixture was extracted with ethyl acetate
`(3 · 4 mL). The combined organic layers were dried
`(MgSO4), filtered, and concentrated. The residue was
`flash chromatographed on a 7 cm tall · 40 mm diameter
`silica gel column eluting with 2:1 hexane–ethyl acetate to
`afford two components. The first to elute, 13.5 mg, was
`tentatively assigned as a single diastereomer (the olefin
`geometry was assumed to be E) of 5, the allyl fluoride
`resulting from formal SN20 fluorination of the starting
`material, on the basis of 1H NMR spectral data for this
`compound (yield¼ 7%). The 1H NMR spectrum for this
`compound (200 MHz) showed the proton on the fluo-
`rine-containing carbon resonating as a doublet of mul-
`tiplets between d ¼ 5:1 and 5.4 ppm with an estimated J
`value of about 50 Hz, and being mostly obscured by the
`two multiplets due to two oxygenated methine protons
`d ¼ 5:4–5:5 ppm and
`d ¼
`resonating
`between
`5:1–5:2 ppm. Furthermore,
`the CH2OPh protons
`appeared as a broad singlet between d ¼ 4:5–4:6 ppm,
`which was distinct from that for the formal SN2 fluori-
`nation product 4 (vide infra). MALDI-LRMS, calcd for
`
`IPR Page 7/19
`
`

`

`3458
`
`P. Klimko et al. / Bioorg. Med. Chem. 12 (2004) 3451–3469
`
`C24H23O5FNa [(M+ Na)þ], 433; found, 433. The second
`component to elute, 80.0 mg, was assigned as a mixture
`of the two diastereomers of the formal SN2 product 4
`and the other diastereomer of 5 (yield¼ 43%). The
`proton NMR spectrum (200 MHz) for this three-com-
`pound mixture was more complicated than that for the
`single compound faster-eluting component. However, a
`key region of the spectrum was between d ¼ 4–5 ppm. A
`broad singlet between d ¼ 4:5–4:6 ppm was likely due to
`the resonance of the CH2OPh protons for 5 (vide supra).
`two multiplets, one about d ¼
`The appearance of
`4:0 ppm and the other about d ¼ 4:1 ppm, are consistent
`with resonances due to the individual CH2OPh for 4.
`The integration values indicated a ratio of 3:1 ratio of
`4:one diastereomer of 5.
`
`Particularly diagnostic for the slower-eluting component
`was the 13C and DEPT NMR spectra. The first key re-
`gion of these spectra was that between d ¼ 85–95 ppm.
`The appearance of three doublets due to a CH carbon
`bonded to fluorine (d ¼ 92:16 ppm, CH, d, J ¼ 174 Hz;
`d ¼ 90:42 ppm, CH, d, J ¼ 172 Hz; d ¼ 90:20 ppm, CH,
`d, J ¼ 174 Hz) indicate that there were three-fluorinated
`products. The second key region of the spectra was be-
`tween d ¼ 65–70 ppm, with the relevant resonances
`being at d ¼ 69:44 ppm (CH2, d, J ¼ 24 Hz) and at
`d ¼ 66:83 ppm (CH2). The third key region was between
`d ¼ 50–60 ppm, with resonances at d ¼ 56:98 ppm (CH,
`d, J ¼ 21 Hz), d ¼ 54:14 ppm (CH), and d ¼ 54:00 ppm
`(CH). HRMS analysis of the slower eluting component
`exhibited an m=z ratio for Mþ of 410.15283 (calculated
`for C24H23O5F, 410.152839). These data are most con-
`sistent with the assignments in Table 3 below.
`
`4.1.2.2. Synthesis run. To a solution of 2 (2.90 g,
`7.11 mmol) in CH2Cl2 (45 mL) at )78 °C was added
`DAST dropwise via syringe (1.71 g, 10.6 mmol). After 5 h
`the reaction mixture was quenched by the addition of
`methanol (3 mL), the solution was warmed to room
`temperature, and saturated NaHCO3 was added (30 mL).
`The mixture was extracted with CH2Cl2 (2 · 75 mL),
`dried (MgSO4), filtered, and concentrated. The residue
`was flash chromatographed on a 20 cm tall · 53 mm
`silica gel column, 2:1 fi 3:2 hexane–ethyl
`diameter
`acetate gradient elution, to afford the slower-eluting
`component described in the demonstration run above,
`which consisted of a mixture of 4 and one diastereomer of
`allylically transposed fluoride 5 (1.48 g, 51%).
`
`4.1.3. [3aR,4R(1E,3RS),5R,6aS]-4-(3-Fluoro-4-phenoxy-
`butenyl)-5-hydroxy-hexahydro-2H-cyclopenta[b]furan-2-
`one (6) and [3aR,4R(2E,1RS),5R,6aS]-4-(1-Fluoro-4-phen-
`oxy-2-buten-1-yl)-5-hydroxy-hexahydro-2H-cyclopenta[b]-
`furan-2-one. To a solution of the above mixture of 4 and
`
`5 (830 mg, 2.03 mmol) in methanol (20 mL) was added
`(320 mg, 2.31 mmol). After 1 h the reaction
`K2CO3
`mixture was quenched by the addition of saturated citric
`acid (25 mL) and the solution was extracted with ethyl
`acetate (4 · 30 mL). The combined organic layers were
`dried (MgSO4), filtered, and concentrated, and the res-
`idue was chromatographed on a 23 cm tall · 53 mm
`diameter silica gel column, 1:1 fi 3:2 ethyl acetate–hex-
`ane gradient elution.
`
`The first compound to elute was an 82 mg fraction
`consisting of one diastereomer of the SN20 fluorination-
`derived product [3aR,4R(2E,1RS),5R,6aS]-4-(1-fluoro-
`4-phenoxy-2-buten-1-yl)-5-hydroxy-hexahydro-2H-cyclo-
`penta[b]furan-2-one (yield¼ 13%). The proton NMR
`spectrum for this compound (200 MHz) showed the
`proton on the fluorinated carbon resonating at
`d ¼ 4:92 ppm as a doublet of triplets, J ¼ 5 Hz and
`48 Hz, due to coupling to the protons on the adjacent
`carbons (the 5 Hz triplet) and to the directly attached
`fluorine (the 48 Hz doublet). Also, the CH2OPh protons
`resonated as a broad singlet about d ¼ 4:58 ppm.
`
`13C NMR (50 MHz): d 176.98 (C), 158.08 (C), 129.79 (d,
`J ¼ 12 Hz, CH), 129.49 (CH), 128.25 (d, J ¼ 19 Hz,
`CH), 121.16 (CH), 114.69 (CH), 93.09 (d, J ¼ 171 Hz,
`CH), 84.12 (CH), 73.57 (d, J ¼ 4 Hz, CH), 66.87 (CH2),
`58.26 (d, J ¼ 20 Hz, CH), 40.89 (CH2), 39.62 (d,
`J ¼ 4 Hz, CH), 35.81 (CH2). The appearance of the
`most downfield CH2 signal at d ¼ 66:87 ppm, due to the
`CH2OPh carbon, as a singlet indicates that a C–F bond
`was not formed on the adjacent carbon; otherwise a two
`bond C–F coupling constant would be present. A two-
`bond C–F doublet with a 20 Hz coupling constant is
`found instead for a CH signal at d ¼ 58:26 ppm. This is
`most consistent with an SN20 fluorination-derived
`product structural assignment for this compound, as is
`the presence of two CH signals as doublets in the ali-
`phatic region of the spectrum with 4 Hz coupling con-
`stants due to three bond C–F couplings: one at
`d ¼ 73:57 ppm and one at d ¼ 39:62 ppm.
`
`The second component to elute was a 593 mg fraction,
`which was assigned as the SN2 fluorination-derived
`product 6 (nominal yield ¼ 95%; there was likely a
`weighing error as the nominal combined yield exceeds
`100%). An inspection of the proton NMR spectrum
`(200 MHz) for this component showed several key dif-
`ferences to that SN20 fluorination-derived product from
`above. First, the appearance of the olefin protons was
`much sharper. Second, the proton on the fluorinated
`carbon resonated as a doublet of quartets at d ¼ 5:24 ppm
`with J ¼ 4 Hz and 50 Hz that is, about 0.3 ppm downfield
`from the fluorinated CH for the SN20 fluorination-derived
`product. Third, the CH2OPh protons resonated as a
`doublet of multiplets about d ¼ 4:1 ppm.
`
`Table 3. Selected carbon 13 NMR spectroscopy peaks for compounds 4 and 5
`
`Compound
`
`4, major fluoride diastereomer
`4, minor fluoride diastereomer
`5, one fluoride diastereomer
`
`Fluorinated carbon resonance
`d ¼ 90:42 ppm, d, J ¼ 172 Hz
`d ¼ 90:20 ppm, d, J ¼ 174 Hz
`d ¼ 92:16 ppm, d, J ¼ 174 Hz
`
`CH2OPh resonance
`d ¼ 69:44 ppm, d, J ¼ 24 Hz
`d ¼ 69:44 ppm, d, J ¼ 24 Hz
`d ¼ 66:83 ppm, singlet
`
`R2CHO2CPh resonance
`d ¼ 54:14 ppm, singlet
`d ¼ 54:00 ppm, singlet
`d ¼ 56:98 ppm, d, J ¼ 21 Hz
`
`IPR Page 8/19
`
`

`

`P. Klimko et al. / Bioorg. Med. Chem. 12 (2004) 3451–3469
`
`3459
`
`The carbon and DEPT NMR spectra for 6 had the
`following resonances (50 MHz): d 176.91 (C), 158.27
`(C), 134.14 (d, J ¼ 10 Hz, CH), 129.61 (CH), 127.55 (d,
`J ¼ 18 Hz, CH), 121.50 (CH), 114.69 (CH), 90.56 (d,
`J ¼ 172 Hz, CH), 82.68 (CH), 76.74 (due to major dia-
`stereomer, d, J ¼ 2 Hz, CH), 76.65 (due to minor dia-
`stereomer, d, J ¼ 2 Hz, CH), 69.57 (d, J ¼ 24 Hz, CH2),
`56.24 (CH), 42.46 (CH), 40.11 (CH2), 34.49 (CH2). Note
`the appearance of the CH2OPh carbon as a doublet at
`d ¼ 69:57 ppm with J ¼ 24 Hz due to 2-carbon C–F
`coupling, as opposed to its appearance as a singlet in the
`13C NMR spectrum for the regioisomeric fluoride (vide
`supra).
`HRMS data for 6: m=z calcd for C17H20O4F [(M+H)þ],
`307.134634; found, 307.134634.
`
`4.1.4. [3aR,4R(1E,3RS),5R,6aS]-4-(3-Fluoro-4-phenoxy-
`butenyl)-5-(tetrahydropyran-2-yloxy)-hexahydro-2H-cy-
`clopenta[b]furan-2-one (7). To a solution of 6 (590 mg,
`1.93 mmol) and DHP (202 mg, 2.41 mmol) in CH2Cl2
`(11 mL) at 0 °C was added p-TsOH (73 mg, 0.38 mmol).
`After 30 min the reaction mixture was quenched by the
`addition of NEt3 (110 mg, 1.1 mmol) and concentrated,
`and the residue was chromatographed on a 14 cm
`tall · 41 mm diameter silica gel column eluting with 3:2
`ethyl acetate–hexane to afford 7 (437 mg, 58%). HRMS,
`m=z calcd for C22H28O5F [(M+H)þ], 391.192125; found,
`391.19213.
`
`4.1.5. (5Z,13E)-(9S,11R,15RS)-15-Fluoro-9-hydroxy-16-
`phenoxy-11-(tetrahydropyran-2-yloxy)-17,18,19,20-tetra-
`nor-5, 13-prostadienoic acid isopropyl ester (8). To a
`solution of 7 (426 mg, 1.09 mmol) in THF (9 mL) at
`)78 °C was added a 1.5 M solution of DIBAL-H in
`toluene (1.2 mL, 1.8 mmol). After 3 h methanol was
`added (1 mL) and the solution was warmed to room
`temperature. Saturated NH4Cl (3 mL), ether (15 mL),
`and saturated sodium potassium tartrate (15 mL) were
`added and the suspension was stirred for 15 min to break
`the emulsion. The layers were separated, the aqueous
`phase was extracted with ether (2 · 25 mL), and the
`combined organic layers were dried (MgSO4), filtered,
`and concentrated to afford the lactol
`[3aR,4R(1E,
`3RS), 5R,6aS]-4-(3-fluoro-4-phenoxybutenyl)-5-hydroxy-
`hexahydro-2H-cyclopenta[b]furan-2-ol
`(406 mg, 93%).
`MALDI-LRMS, m=z calcd for C22H29O5FNa [(M+
`Na)þ], 415; found, 415.
`To a suspension of Ph3Pþ(CH2)4CO2HBr (970 mg,
`2.19 mmol) in THF (10 mL) at 0 °C was added a 1 M
`solution of potassium t-butoxide in THF (4.6 mL,
`4.6 mmol). After 15 min a solution of the above lactol
`(426 mg, 1.09 mmol) in THF (10 mL) was added. After
`2.5 h the reaction mixture was quenched by the addition
`of saturated KH2PO4 and was warmed to room tem-
`perature. The solution was extracted with ethyl acetate
`(4 · 30 mL), dried (Na2SO4), filtered, and concentrated
`to afford a crude oil.
`
`This oil was dissolved in acetone (15 mL) and cooled to
`0 °C. DBU (760 mg, 5.01 mmol) was added, and after
`40 min isopropyl iodide was added (850 mg, 5.0 mmol).
`The reaction mixture was warmed to room temperature
`and stirred overnight. Saturated KH2PO4 (20 mL) was
`added to quench the reaction mixture, saturated brine
`(20 mL) was added, and the solution was extracted with
`ethyl acetate (3 · 50 mL). The combined organic layers
`were dried over MgSO4, filtered, and concentrated. The
`residue was chromatographed on a 10 cm tall · 41 mm
`diameter silica gel column eluting with 40% ethyl acetate
`in hexane to afford 8 (236 mg, 45%). HRMS, m=z calcd
`[(M+H)þ],
`for C30H44O6F
`519.312738;
`found,
`519.31274.
`
`4.1.6. (5Z,13E)-(9S,11R,15R)-9,11-Dihydroxy-15-fluoro-
`16-phenoxy-17,18,19,20-tetranor-5-prostadienoic
`acid
`isopropyl ester (9) and (5Z,13E)-(9S,11R,15S)-9,11-dihy-
`droxy-15-fluoro-16-phenoxy-17,18,19,20-tetranor-5-pro-
`stadienoic acid isopropyl ester (10). To a solution of THP
`ether 8,
`isopropanol (10 mL), and water (1 mL) was
`added 12 M HCl (1.5 mL). After 45 min the reaction
`mixture was quenched