Unexpected Isomerisation of a Fragment ... - CSIRO Publishing

RESEARCH FRONT CSIRO PUBLISHING

Full Paper

Aust. J. Chem. 2015, 68, 1871–1879 http://dx.doi.org/10.1071/CH15587

Unexpected Isomerisation of a Fragment Analogue During Fragment-Based Screening of HIV Integrase Catalytic Core Domain John H. Ryan,A,E Karen E. Jarvis,A Roger J. Mulder,A Craig L. Francis,A G. Paul Savage,A Olan Dolezal,B Thomas S. Peat,B and John J. DeadmanC,D A

CSIRO Biomedical Manufacturing Program, Bayview Avenue, Clayton, Vic. 3168, Australia. B CSIRO Biomedical Manufacturing Program, 343 Royal Parade, Parkville, Vic. 3052, Australia. C Avexa Pty Ltd, Level 1, 61–63 Camberwell Road, Hawthorn East, Vic. 3123, Australia. D Current address: Chemocopeia Pty Ltd, 157 Arnold Street, Melbourne, Vic. 3054, Australia. E Corresponding author. Email: [email protected]

Fragment-based screening of human immunodeficiency virus type 1 (HIV) integrase revealed several aromatic carboxylic acid fragment hits, some of which bound weakly at the site on the HIV-integrase catalytic core domain that binds the lens epithelium-derived growth factor (LEDGF). Virtual screening of an internal database identified an analogue that bound with higher affinity and in an isomerised form to the LEDGF binding site. The starting lactone was stable in CDCl3; however, an unexpected isomerisation process occurred in [D6]DMSO to give the same isomer found in the LEDGF binding site. This hit led directly to a series of low-micromolar LEDGF inhibitors and, via a scaffold hop, to a series of allosteric binding site inhibitors. Manuscript received: 20 September 2015. Manuscript accepted: 9 October 2015. Published online: 6 November 2015.

Introduction A recent article by Davis and Erlanson described the ‘unknown knowns’ of fragment screening, that is, the pitfalls and artefacts that can befall a fragment-screening program.[1] In fragment screening, the affinities of fragment hits, which are typically low, often push detection methods to their limits and lead to false positives or negatives that can confound the translation of a fragment hit into a lead compound. A range of these problems are associated with validation of the fragment hit or follow-up analogues, in particular identity, purity, and stability of such compounds. Inspired by this article, we report our observations on hit validation from a fragment-based screen against human immunodeficiency virus type 1 (HIV) integrase, which led to identification of compounds that bound to and inhibited the HIV integrase catalytic core domain.[2,3] HIV-1 integrase is a clinically validated therapeutic target for treatment of HIV-1 infection.[4] It is involved in integration of viral DNA into the host DNA. Raltegravir was the first integrase inhibitor to be launched in 2007.[5] Kinetic[6] and structural studies[7] have suggested that Raltegravir binds to a complex of viral DNA, host DNA, and magnesium ions in the catalytic site of the enzyme (the ‘intasome’). Resistance to Raltegravir has emerged owing to mutations in proteins at the catalytic site, driving the need for HIV integrase inhibitors with new modes Journal compilation Ó CSIRO 2015

of action.[8] We initiated a fragment-screening project aimed at developing new inhibitors of HIV integrase and this project resulted in the identification of small-molecule and peptide inhibitors of the binding of lens epithelium-derived growth factor (LEDGF) to the HIV integrase that were active in a whole-cell assay and small-molecule inhibitors of HIV integrase through binding at an allosteric site.[2,3,9] An additional outcome was an improved understanding of the role of the fragmentscreening methodology in identifying hits.[10] In the present paper, we focus on hit validation and expansion that led, directly, to small molecules inhibitors that act by binding to the LEDGF binding site and, indirectly, to inhibitors of an allosteric binding site. Results and Discussion We developed reproducible methods for both immobilising HIV integrase catalytic core domain protein constructs (core3H and core4H) onto a surface plasmon resonance (SPR) sensor chip using biotin–streptavidin methodology, and for crystallising the proteins.[2,3] The initial fragment screen was performed using a library of commercial compounds (Maybridge Ro3), although we have since developed a proprietary fragment library.[11] Screening of the commercial library www.publish.csiro.au/journals/ajc

1872

J. H. Ryan et al.

OH

O

O

O

OH

OH

OH

HN

O

1

O

2

3

4

OH

OH

O O

O

O

O

O

OH

OH

O O

O

N O

N

5

HN HN

6 O

8

OH

O

OH

O

7

N N

N

9

10

O

Fig. 1. Structural formulae of SPR fragment hits that contained an aromatic or heteroaromatic core linked to a carboxylic acid. Fragments found in the fragment binding pocket of integrase core3H are highlighted in the box.

via SPR revealed 16 hits.[10] Of these 16 hits, 10 contained aromatic or heteroaromatic ring systems linked to a carboxylic acid group (Fig. 1). All 16 fragments were subjected to soaking experiments with crystals of the integrase core3H construct and six fragments were found in the same binding pocket, which we have termed the ‘fragment binding pocket’.[10] Of these six fragments, fragments 1–3 were (hetero)aromatic carboxylic acids and exhibited nearly identical binding modes (Fig. 2a).[10] A fourth fragment, 4, that contained an ethylene spacer between the ring system and the carboxylic acid group exhibited a slightly different binding orientation for the acid group (Fig. 2b). In addition to the well-defined electron density that was observed for six of the fragments, various fragments showed weak or partial density in the LEDGF binding site, indicating partial occupancy or multiple orientations of the fragment in the binding site.[10] The 16 selected hits had dissociation constants (KDs) in the millimolar range, determined by SPR, indicating the fragments were weakly binding, not unexpectedly with initial fragment hits (Table 1).[10] A range of analogues and derivatives of these fragment hits were sourced from commercial vendors; however, none of these follow-up compounds showed significant improvement in binding. As some of the fragments showed weak or partial density in the LEDGF binding site, we sought to identify compounds that would bind more strongly to this biologically more relevant site. We selected a series of analogues of the carboxylic acidcontaining fragment hits from the CSIRO Compound Library,[11] by using a combination of substructure and similarity searches of the database. One of the follow-up compounds, isobenzofuranone 11,[12] showed promising affinity to both core3H and core4H constructs in the SPR assay (Table 1).[2] Compound 11 was soaked into crystals of the integrase catalytic core domain. Although only weak electron density was found in the fragment binding site, strong electron density was found in the LEDGF binding site.[2] †

The electron density due to the compound bound in the LEDGF binding site did not match any of the diastereomeric or enantiomeric forms of compound 11; instead, it matched the structure of a ring-opened carboxylic acid isomer 12 (Fig. 3). In theory, E- and Z-geometrical isomers of the carbon–carbon double bond of compound 12 are possible, although the density clearly matched the E-isomer. Additionally, rotamers of the E-isomer of 12 are possible owing to the steric hindrance between the two aromatic rings, which means that the two aromatic rings cannot readily adopt a coplanar relationship (Fig. 4). Clearly, a single rotamer of E-12 was found in the LEDGF binding site (Fig. 3). This result raised several questions, including: was the sample of compound 11 in the form as described or was it actually the ring-opened acid form 12, and were the SPR binding data observed for compound 11 actually due to compound 11 or to the ring-opened isomer 12? The experimental conditions for obtaining compound 11 bound in an HIV integrase catalytic core domain crystal involved administering the powdered sample of 11 to the crystal-containing droplet (containing cryo-solution) in a crystallisation plate, then sealing the plate for 24 to 48 h before data collection at the Australian Synchrotron. As the sample of compound 11 was thought to be the sample originally reported in 1980,[12] we first sought to verify the identify and purity of the sample. A sample of the archived compound 11 was analysed using 1 H NMR spectroscopy (CDCl3) and the data obtained were generally consistent with the literature data[12] and with the designated structure 11 (as an ,2 : 1 mixture of diastereomers). In particular, the signals at d 4.32 and 3.96 were assigned to the diastereomeric methine protons (–ArCHCO–) of the indanyl moieties for the minor and major diastereomer, and the geminally coupled doublets at d 3.62 and 3.52 ppm and d 3.61 and 3.49 were assigned to the methylene protons (–ArCH2CO–) of the indanyl ring systems for the major and minor diastereomer respectively.†

The NMR data reported in ref. [12] were obtained at 60 MHz in CCl4 with tetramethylsilane as an internal reference. Ref. [12] states that ‘The crude product, before recrystallisation, contained varying amounts (up to 40 %) of a second isomer, indicated by a doublet at d 4.35.

Isomerisation of Fragments During Screening of HIV Integrase

(a)

1873

(b)

Fig. 2. Overlays of X-ray crystallographic structures with fragments 1–3 (a), and 3 and 4 (b) bound in the fragment binding pocket of the HIV-1 integrase catalytic core domain. The catalytic core domain was solved as a dimer and the fragment binding pocket is formed in close proximity to the interface between the two monomers. One monomer is shown in blue and the other in gold. Table 1. Biophysical and biochemical data from testing fragments, analogues and derivatives[2,10] Compound no. 1 2 3 4 5 6 7 8 9 10 11 12 17 18 19 20 21

KD SPR core3H [mM]

KD SPR core4H [mM]

IC50 AlphaScreen LEDGF inhibition [mM]

IC50 30 -processing strand transfer inhibition [mM]

– 6400 5400 6500 4600 5200 – 13000 .25000 – 1570 337 1375 149 1435 304 595 16 7.6 1.2 – –

– – – – – – – – – – 750 311 763 311 724 347 1180 198 76 11 – –

– – – – – – – – – – – 200 270 220 8.1 – –

– – – – – – – – – – – – – – – 295 5

(a)

(b)

O

O

O

O

CO2H O O O O

11

12

Fig. 3. Crystallographic structure showing overlay of a diasteromeric form of lactone 11 (a), and E-12 (b) into the difference electron-density map (green netting) found in the LEDGF binding site.

1874

J. H. Ryan et al.

A close inspection of the final 1H NMR spectrum revealed minor signals due to the starting material 11, indicating that a steady-state equilibrium has been reached. In order to determine if the equilibrium point could be attained more quickly, a freshly prepared solution of 11 in [D6]DMSO was heated using a Biotage Initiator 60TM microwave reactor at 1008C for 1 h. The 1H NMR spectrum of the processed solution was practically identical to that obtained after aging the solution for a day at room temperature. In the 13C NMR spectrum, the expected 18 signals for the major isomer were observed, with the most diagnostic being the signals at d 203.5 and 165.3 ppm due to the ketone and carboxylic acid carbons respectively, and at d 102.4 and 40.7 ppm due to the methylene dioxy carbon and indanyl methylene carbon respectively. This result showed that the conversion of 11 into the acid 12 could be accelerated by heating and demonstrated the stability of the acid 12 under these conditions. In order to conduct further studies, compound 11 was resynthesised via acid-catalysed condensation of indan-2-one 13 with 5,6-(methylenedioxy)phthalaldehydic acid 14, according to the procedure of Prager and Ward (Scheme 2).[12] Analysis of the 1H NMR spectrum of resynthesised lactone 11 revealed it to be a mixture of diastereomers; however, also revealed was the presence of a significant amount of the corresponding 1,3-dialkylated product 15. As originally reported,[12] the lactone 11 could not be readily purified by normal-phase chromatography owing its instability on silica. As the stability of 15 in DMSO was also of interest, the isomerisation studies were performed using the sample of 11 containing impurity 15. The stability of lactone 11 in solution was assessed by 1H NMR spectroscopy at 500 MHz. The monitoring of a CDCl3

Next, we studied the effect of DMSO, the solvent used to dissolve compound 11 and a cosolvent in the SPR and X-ray crystallographic studies, on compound 11. The 1H NMR spectrum of a freshly prepared solution of 11 in [D6]DMSO showed signals consistent with structure 11 (again as an ,2 : 1 mixture of diastereomers). On standing the NMR solution at room temperature for 24 h, the 1H NMR spectrum changed, with the signals due to the starting diasteromers 11 being replaced by that of a new species. There was no detectable further change to the spectrum after 5 days. The NMR signals were consistent with the ring-opened isomer 12, with major and minor signals indicating the presence of an ,5 : 1 ratio of geometrical isomers (Scheme 1).z The 1H NMR spectrum showed the expected number of signals, with diagnostic signals attributed to the major geometrical isomer including a broad singlet at d 13.29 ppm due to the carboxylic acid proton, a singlet at d 7.58 due to the olefinic proton, a singlet at d 6.23 due to the methylenedioxy protons and a singlet at d 3.62 due to the indanyl methylene protons. An additional singlet at d 3.55 is tentatively assigned to the indanyl methylene protons of the minor geometrical isomer.

O

O CO2H

CO2H

O

O

O

O

A

B

Fig. 4. Examples of atropisomeric conformations of E-12 (conformation (A) was found in the LEDGF binding site).

O O

O

O

O

H O

CO2H

[D6]DMSO

O

O

H

O

⫹

CO2H

O O

11

E-12

Z-12

Scheme 1. Conversion of lactone 11 into acid 12.

O

O

O O

O

O ⫹

O

Cat. pTsOH O

HO

O

Benzene, Δ

13

⫹

O

O

O

14

O

O O

O

11 15

O

O O

Scheme 2. Synthesis of lactone 11.

z

Although the E-isomer was found bound in the integrase crystal structure, it is yet to be unequivocally proved that the major geometrical isomer formed in [D6]DMSO is the E-isomer.


1875

solution of the lactone showed no perceptible change over a 24-h period. The 1H NMR analyses of lactone 11 in [D6]DMSO indicated equilibration of the phthalide form to give the acid form 12, with equilibrium being attained after ,6 h (Fig. 5). As previously found, the final equilibrium attained had the acid 12 as the major component, weak signals due to the starting lactone 11 were still apparent, and a mixture of geometrical isomers was formed with one predominating. It was also apparent that the bis lactone 15, present in the sample of 11, had isomerised under these conditions to give the corresponding diacid 16 (Scheme 3). The signals for the alkene protons of the diacid 16 were found at d 7.76, well downfield of the alkene proton signal (d 7.57) for the acid 12. These intriguing results indicated the possibility of an unexpected solvent-dependent isomerisation process where the phthalide form is favoured in CDCl3 and the acid form is favoured in [D6]DMSO, a significantly more polar solvent. Although the polarity of the solvent is thought to be a factor in this process, [D6]DMSO is hygroscopic and it is possible that water in the [D6]DMSO may have an influence on the equilibrium position. Additionally, it is yet to be explored as to whether

O

O

O

CO2H

O O

O

O

O

11

12 [D6]DMSO ⫹

⫹

O

O

O

O

O

O

O

O HO2C

O

O

15

16

Scheme 3. 1H NMR study of conversion of lactone 11 into acid 12 in [D6]DMSO.

t⫽6h

t⫽4h

t⫽1h

t⫽0

13

12

11

10

9

8

O O

O

t ⫽ 14 h

14

O

HO2C

7

6

5

4

3

2

1

[ppm]

Fig. 5. Monitoring of the isomerisation of lactone 11 and bis lactone 15 in [D6]DMSO by 1H NMR spectroscopy (500 MHz) (þ, 11; B, 15; , 12 major; o, 12 minor; D, 16).

1876

this phenomenon in [D6]DMSO translates to non-deuterated DMSO, although of course this is quite likely. The pH dependence of the equilibrium between such phthalides and the corresponding ring-opened chalcones is a well-known phenomenon, with the chalcone form being thermodynamically favoured under alkaline conditions (in the deprotonated form), whereas the phthalide form is thermodynamically favoured under acidic conditions (in the neutral form).[13–19] To the best of our knowledge, the effect of solvent polarity on the relative thermodynamic stability of each form has not been studied. The pH values for the SPR and crystallographic studies were 7.4 and 4.6–5.6 respectively, above the expected pKa (¼ log10 Ka; Ka is the acid dissociation constant) for a substituted benzoic acid,[20] which would favour formation of the anionic form of 12. The anionic chalcone 12 is then preferentially sequestered by the protein, in part owing to the interactions of the carboxylate anion moiety with backbone amide residues (Fig. 6).[2] The proposal that the SPR binding activity observed on testing of 11 is primarily due to 12 was supported by analyses of independently synthesised 12.[2] The 1H NMR spectrum of independently synthesised 12 showed the same signals that were observed for 12 formed by isomerisation of 11 in [D6] DMSO. SPR analysis of the independently synthesised 12 gave approximately the same binding constant as that originally obtained from testing of 11 (Table 1).[2] Crystallographic analyses of independently synthesised 12 soaked into the integrase crystals confirmed the binding of 12 in an identical mode to 12 obtained from soaking lactone 11 into integrase crystals (Fig. 6a). The hypothesis that the carboxylic acid moiety in 12 is responsible for the selectivity of the protein for the chalcone form 12 over the closed form 11 is supported by the retention of the interaction between the carboxyl acid group and the backbone amide residues during optimisation of hit 12 into a 33-fold more active LEDGF inhibitor 19 (Fig. 6c and Scheme 4).[2] Reduction of the ketone 12 afforded alcohol 17, which bound in the LEDGF site with similar affinity to 12 (Fig. 6b). Recognition that certain peptides also bound in the same LEDGF binding site led to the design of analogues and serendipitous discovery of acyclic analogues, e.g. 18.[9] Other precedents for screening hits that have become chemically transformed during the assay procedure include during the discovery of the antiviral daclatasvir where the oxidative dimerisation of an iminothiazolidinone hit led to a much more active species.[21] Optimisation of the acyclic hit series resulted in the LEDGF inhibitor 19, with low-micromolar activity in the SPR binding and Alphascreen LEDGF inhibition assays (Table 1) and HIV pseudo-virus replication in a wholecell infectivity assay.[2] The fragment analogue 12 is closely related in structure to an alternative series of inhibitors that bound to an allosteric binding site.[3] During the early stages of the hit-to-lead campaign, indolinone 20 was evaluated as an alternative scaffold to 12 and found to form a complex with integrase and to inhibit the catalytic activity of integrase, with a half maximal inhibitory concentration (IC50) of 295 mM in the 30 -processing strand transfer HIV-1 integrase assay. However, X-ray crystallographic studies showed that with this seemingly small change, indolinone 20 was found to bind in an allosteric binding site. Compound 20 was modified by changing the indolinone group to an isatin group and by addition of an acrylic acid side-chain,

J. H. Ryan et al.

resulting in 21, which was a 45-fold improved inhibitor in the strand transfer assay (Scheme 5). Conclusion A fragment-screening campaign revealed fragments that bound in the fragment binding pocket of the HIV integrase catalytic core domain (Fig. 7a). The screening of analogues and derivatives from the CSIRO Compound Library by SPR resulted in identification of a follow-up hit that was proved by X-ray crystallography to bind in an isomeric form in an alternative binding site, the LEDGF binding site (Fig. 7b). The follow-up hit initiated development of low-micromolar inhibitors of LEDGF binding to the HIV integrase protein and, via scaffold hop, a series of inhibitors of another binding site allosteric to the integrase catalytic site (Fig. 7). A broader study of the isomerisation phenomenon is under way and the results will be reported in due course. Experimental The SPR and crystallographic methods have been reported previously.[2,3,10] General Analytical Methods 1 H and 13C NMR spectra were recorded on a Bruker Biospin Av500 spectrometer at 500 and 125 MHz respectively, a Bruker Av400 spectrometer at 400 and 100 MHz, or a Bruker Av200 spectrometer at 200 and 50 MHz, at 258C unless otherwise note. For 1H NMR spectra, the peak due to residual solvent (d 7.24 for CDCl3, 2.50 for [D6]DMSO) was used as the internal reference. The reference for proton-decoupled 13C NMR spectra in [D6] DMSO was the central peak (d 39.51) of the deuterated [D6] DMSO septet. Low-resolution electron impact mass spectra (LRMS-EI) were run on a ThermoQuest MAT95XL mass spectrometer or a Thermo Scientific DFS mass spectrometer, using ionisation energy of 70 eV. Positive and negative chemical ionisation (CI) mass spectra were collected on a Thermo Scientific DFS mass spectrometer using an ionisation energy of 150 eV and methane as the reagent gas. Accurate mass measurements (high-resolution mass spectra (HRMS)-EI) were obtained with a mass resolution of 5000–10000 using PFK (perfluorokerosene) as the reference compound. Materials Indan-2-one 13 was supplied by Aldrich. 6-Hydroxy-[1,3] dioxolo[4,5-e]isobenzofuran-8(6H)-one 14 was synthesised according to the method of Borchardt et al.[22] Experimental Methods 6-(2-Oxo-2,3-dihydro-1H-inden-1-yl)-[1,3]dioxolo[4,5e]isobenzofuran-8(6H)-one 11 (Original Sample Archived in CSIRO Compound Library) The sample with designated code CSIRO-045513 was provided to CSIRO by Prager and Ward.[10] The 1H NMR data for 11 (CSIRO-045513) obtained in CDCl3 at 400 MHz was consistent with that originally reported (obtained in CCl4 at 60 MHz with tetramethylsilane as an internal reference) and with that obtained from freshly prepared 11 (see below) and indicated that this sample of 11 was an ,2 : 1 mixture of diastereoisomers. The 1 H NMR data for a freshly prepared solution of 11 (CSIRO045513) collected in [D6]DMSO at 400 MHz also indicated that this sample of 11 was an ,2 : 1 mixture of diastereoisomers and was consistent with the data obtained from freshly prepared 11 (see below).


(a)

1877

(b)

(c)

Fig. 6. Crystallographic representations of integrase–ligand interactions showing overlays of crystal structures from soaking of 11 (magenta) and independently synthesised 12 (blue) (a), 12 (blue) and alcohol 17 (light blue) (b), and 12 (blue) and optimised analogue 19 (yellow) (c), clearly showing in each case the interaction between the carboxylate residue of the ligands and the backbone amides of residues 170 and 171 in the LEDGF binding pocket.[2]

R2 OH

O

O

CO2H

CO2H

R3

NH CO2H

O

O O

N R1

O

12

17

O O

18 R1 ⫽ Me, R2 ⫽ p-MeO-Ph, R3 ⫽ H 19 R1 ⫽ allyl, R2 ⫽ R3 ⫽ p-MeO-Ph Scheme 4. Optimisation of fragment analogue 12 into low-micromolar HIV inhibitor 19.[2]

O O

O

CO2H

CO2H N O

N O

O

O

20

21

HO2C

Scheme 5. Optimisation of fragment analogue 20 into low-micromolar allosteric integrase inhibitor 21.[3]

(a)

(b)

Fig. 7. Crystallographic representations of HIV integrase dimer–ligand interactions showing the integrase monomers (brown and light blue), fragment 1 bound in the fragment binding pocket (dark blue) (a), compound 12 bound in the LEDGF binding site (magenta) (b). The allosteric binding site is shaded grey.

1878

m/z (HRMS-EI) 308.0683 (Mþ); C18H12O5 requires 308.0679. m/z (EI) 308 (5 %, Mþ), 178 (9), 177 (100). 5-((2-Oxo-2,3-dihydro-1H-inden-1-ylidene)methyl) benzo[d][1,3]dioxole-4-carboxylic Acid 12 (Prepared from 11)(as a Mixture of Major and Minor E/Z Isomers) A solution of 11 (original sample CSIRO-045513) in [D6] DMSO was stored at room temperature and 1H NMR spectra were collected at periodic intervals. There was no perceptible change in the spectrum obtained after 24 h compared with that obtained after 5 days. The predominant signals were due to 12 as an ,5 : 1 mixture of geometrical isomers. The final spectrum was identical to that obtained from heating the sample solution at 1008C (see data below). A solution of 11 (original sample CSIRO-045513) in [D6] DMSO was heated at 1008C for 2 h using a Biotage Initiator 60TM microwave reactor. dH (400 MHz, [D6]DMSO) 13.29 (broad s, 1H, –CO2H), 7.58 (s, 1H, –C¼CH–Ar), 7.44–7.27 (m, 3H, ArH), 7.19–7.07 (m, 3H, ArH), 6.23 (s, 2H, –OCH2O–), 3.62 (s, 2H, ArCH2CO–) (major geometrical isomer); 7.67 (s, 1H, –C¼CH–Ar), 6.18 (s, 2H, –OCH2O–), 3.55 (s, 2H, ArCH2CO–) (partial data for minor geometrical isomer). dC (125MHz, [D6]DMSO) 203.5 (C¼O), 165.3 (CO2H), 148.9, 148.3, 138.9, 137.1, 133.7, 132.5, 129.1, 128.3, 126.9, 125.6, 122.7, 122.5, 113.8, 110.7, 102.4 (–OCH2O–), 40.7 (ArCH2CO–) (major geometrical isomer). 5-((2-Oxo-2,3-dihydro-1H-inden-1-ylidene)methyl) benzo[d][1,3]dioxole-4-carboxylic Acid 12[2](as a Mixture of E/Z Isomers) To a solution of indan-2-one 13 (551 mg, 4.17 mmol) in EtOH (10 mL) was added 6-hydroxy-[1,3]dioxolo[4,5-e] isobenzofuran-8(6H)-one 14 (270 mg, 1.39 mmol) in one portion, followed by piperidine (3 mL). The reaction mixture was stirred at room temperature for 2 h and then HCl (conc., 1.0 mL) was added. The resultant mixture was stirred for another 10 min, then the solvent was removed under vacuum. The residue was diluted with H2O (20 mL) and the mixture was extracted with EtOAc (2 30 mL). The combined extract was dried (Na2SO4) and concentrated under vacuum to give compound 12 (350 mg, 81 % yield) as a yellow solid. m/z (HRMS-EI) 308.0673 (Mþ); C18H12O5 requires 308.0679. dH (400 MHz, [D6]DMSO) 7.57 (s, 1H, –C¼CH–Ar), 7.42– 7.38 (m, 1H), 7.37–7.34 (m, 2H), 7.31–7.27 (m, 3H), 6.21 (s, 2H, –OCH2O–), 3.60 (s, 2H, ArCH2CO–) (major geometrical isomer); 7.66 (s, 1H, –C¼CH–Ar), 6.16 (s, 2H, –OCH2O–), 3.53 (s, 2H, ArCH2CO–) (partial data for minor geometric isomer). 6-(2-Oxo-2,3-dihydro-1H-inden-1-yl)-[1,3]dioxolo [4,5-e]isobenzofuran-8(6H)-one 11[10] Phthalide derivative 11 was synthesised according the procedure of Prager and Ward.[10] A mixture of indan-2-one 13 (0.680 g, 5.15 mmol), 6-hydroxy-[1,3]dioxolo[4,5-e] isobenzofuran-8(6H)-one 14 (1.00 g, 5.15 mmol), and a crystal of p-toluenesulfonic acid in benzene (50 mL) was heated at reflux for 5 h, using a Dean–Stark apparatus to remove water. The mixture was concentrated to dryness and the solid residue

J. H. Ryan et al.

was triturated with ether and filtered to give 11 (1.22 g, 77 %) as a yellow solid. Analysis of the NMR and MS data indicated the product contained ,20 % of dialkylated indanone 15. dH (500 MHz, CDCl3) 7.33–6.86 (m, 5H), 6.31–5.99 (m, 4H), 4.32 (d, J 3.2, 1H, CH, minor diastereomer), 3.96 (broad s, 1H, CH, major diastereomer), 3.62 (d, J 22.9, 1H, CH2, major diastereomer), 3.61 (d, J 23.1, 1H, CH2, minor diastereomer) 3.52 (d, J 22.9, 1H, CH2, major diastereomer), 3.49 (d, J 23.1, 1H, CH2, minor diastereomer). Multiple minor signals between d 4.32 and 3.90 were assigned to the diastereomeric methine protons of the dialkylated side product 15. Ratio of 11 : 15 , 4 : 1. dH (500 MHz, [D6]DMSO) 7.50–6.99 (m, 5H), 6.34–6.09 (m, 4H), 4.50 (broad s, 1H, CH minor diastereomer) 4.45 (broad s, 1H, CH, major diastereomer), 3.68 (d, J 23.0, 1H, CH2, major diastereomer), 3.64 (d, J 22.8, 1H, CH2, minor diastereomer), 3.48 (d, J 23.0, 1H, CH2, major diastereomer), 3.45 (d, J 22.8, 1H, CH2, minor diastereomer). The signals of the diastereomeric methine protons of the dialkylated side-product 15 were embedded within the signals at d 4.50 and 4.45. m/z (EI) 308 (5 %, Mþ), 264 (2), 178 (13), 177 (100). m/z (CIþ) 309 (10 %, [M þ H]þ), 292 (18), 291 (100), 263 (15), 177 (58). m/z (CI) 308 (34 %, M), 307 (66, [M H]), 132 (100). For 15: m/z (CIþ) 485 (22 %, [M þ H]þ), 468 (30), 467 (100). m/z (CI) 485 (30 %, [M þ H]), 484 (100, M), 483 (36, [M þ H]). Isomerisation of Lactone 11 to Give Acid 12 A freshly prepared solution of compound 11 (containing ,20 % of dialkylated product 15) in [D6]DMSO was monitored by 1H NMR spectroscopy at 500 MHz until the isomerisation had reached equilibrium with the product consisting of 12 (containing ,20 % of diene 16). dH (500 MHz, [D6]DMSO) For 12: 13.30 (broad s, 1H, –CO2H), 7.57 (s, 1H, –C¼CH–Ar), 7.42–7.25 (m, 3H, ArH), 7.17–7.05 (m, 3H, ArH), 6.22 (s, 2H, –OCH2O–), 3.61 (s, 2H, ArCH2CO–) (major geometrical isomer); 7.66 (s, 1H, –C¼CH– Ar), 6.17 (s, 2H, –OCH2O–), 3.53 (s, 2H, ArCH2CO–) (partial data for minor geometrical isomer). For 16: 7.76 (s, 2H, –C¼CH–Ar), 6.23 (s, 4H, –OCH2O–). m/z (EI) 308 (46, Mþ), 263 (36), 262 (32), 179 (60), 178 (48), 176 (40), 164 (40), 149 (40), 120 (30). m/z (CIþ) 308 (28, Mþ), 307 (90, [M H]þ), 291 (100), 279 (34), 263 (32), 193 (14), 179 (15), 178 (26), 177 (90). m/z (CI) 323 (100, [M þ CH3]), 308 (35, M), 307 (60, [M H]). For 16: m/z (EI) 484 (5 %, Mþ). m/z (CIþ) 483 (10 %, [M H]þ). m/z (CI) 498 (44 %, [M þ CH3]), 485 (32, [M þ H]), 484 (92, M), 482 (32), 466 (18), 440 (14). Acknowledgements We thank Jamie Freemont, Carl Braybrook, and Jo Cosgriff for provision of MS and NMR data, as well as the beamline scientists at the Australian Synchrotron and the C3 facility for help with data collection and crystallisation of HIV integrase respectively. We also thank OpenEye Scientific Software for a licence to Afitt. Compound 12 was synthesised by the team of Dr Xian Bu at SYNthesis med-chem, Level 4, Buiding 21A, 528 Ruiqing Road, Heqing, Pudong, Shanghai 201201.


References [1] B. J. Davis, D. A. Erlanson, Bioorg. Med. Chem. Lett. 2013, 23, 2844. doi:10.1016/J.BMCL.2013.03.028 [2] T. S. Peat, D. I. Rhodes, N. Vandegraaff, G. Le, E. D. Jones, J. Smith, J. A. V. Coates, N. Thienthong, J. Newman, O. Dolezal, J. H. Ryan, G. P. Savage, C. L. Francis, J. J. Deadman, PLoS One 2012, 7, e40147. doi:10.1371/JOURNAL.PONE.0040147 [3] D. I. Rhodes, T. S. Peat, N. Vandegraaff, D. Jeevarajah, G. Le, E. D. Jones, J. A. Smith, J. A. V. Coates, L. J. Winfield, N. Thienthong, J. Newman, D. Lucent, J. H. Ryan, G. P. Savage, C. L. Francis, J. J. Deadman, Antivir. Chem. Chemother. 2011, 21, 155. doi:10.3851/ IMP1716 [4] A. M. Walji, R. I. Sanchez, S.-D. Clas, R. Nofsinger, M. de Lera Ruiz, J. Li, A. Bennet, C. John, D. J. Bennett, J. M. Sanders, C. N. Di Marco, S. Hope Kim, J. Balsells, S. S. Ceglia, Q. Dang, K. Manser, B. Nissley, J. S. Wai, M. Hafey, J. Wang, G. Chessen, A. Templeton, J. Higgins, R. Smith, Y. Wu, J. Grobler, P. J. Coleman, ChemMedChem 2015, 10, 245. doi:10.1002/CMDC.201402393 [5] V. Summa, A. Petrocchi, F. Bonelli, B. Crescenzi, M. Donghi, M. Ferrara, F. Fiore, C. Gardelli, O. G. Paz, D. J. Hazuda, P. Jones, O. Kinzel, R. Laufer, E. Monteagudo, E. Muraglia, E. Nizi, F. Orvieto, P. Pace, G. Pescatore, R. Scarpelli, K. Stillmock, M. V. Witmer, M. Rowley, J. Med. Chem. 2008, 51, 5843. doi:10.1021/ JM800245Z [6] E. P. Garvey, B. Schwartz, M. J. Gartland, S. Lang, W. Halsey, G. Sathe, H. L. Carter, III, K. L. Weaver, Biochemistry 2009, 48, 1644. doi:10.1021/BI802141Y [7] D. P. Maskell, L. Renault, E. Serrao, P. Lesbats, R. Matadeen, S. Hare, D. Lindemann, A. N. Engelman, A. Costa, P. Cherepanov, Nature 2015, 523, 366. doi:10.1038/NATURE14495 [8] K. B. Ingale, M. S. Bhatia, Antivir. Chem. Chemother. 2011, 22, 95. doi:10.3851/IMP1740 [9] D. I. Rhodes, T. S. Peat, N. Vandegraaff, D. Jeevarajah, J. Newman, J. Martyn, J. A. V. Coates, N. J. Ede, P. Rea, J. J. Deadman, ChemBioChem 2011, 12, 2311. doi:10.1002/CBIC.201100350

1879

[10] J. Wielens, S. J. Headey, D. I. Rhodes, R. J. Mulder, O. Dolezal, J. J. Deadman, J. Newman, D. K. Chalmers, M. W. Parker, T. S. Peat, M. J. Scanlon, J. Biomol. Screen. 2013, 18, 147. doi:10.1177/ 1087057112465979 [11] C. L. Francis, P. W. Kenny, O. Dolezal, S. Saubern, M. Kruger, G. P. Savage, T. S. Peat, J. H. Ryan, Aust. J. Chem. 2013, 66, 1473. doi:10.1071/CH13325 [12] G. I. Hutchison, P. A. Marshall, R. H. Prager, J. M. Tippett, A. D. Ward, Aust. J. Chem. 1980, 33, 2699. doi:10.1071/CH9802699 [13] M.-J. Luche-Ronteix, S. Borg, M. Dvolaitzky, R. Lett, A. Marquet, Bull. Soc. Chim. Fr. 1970, 2564. [14] E. W. Bousquet, M. D. Moran, J. Harmon, A. L. Johnson, J. C. Summers, J. Org. Chem. 1975, 40, 2208. doi:10.1021/JO00903A016 [15] H. K. Neudeck, Monatsh. Chem. 1989, 120, 597. doi:10.1007/ BF00810848 [16] H. K. Neudeck, Monatsh. Chem. 1996, 127, 417. doi:10.1007/ BF00810885 [17] R. D. Larsen, E. G. Corley, A. O. King, J. D. Carroll, P. Davis, T. R. Verhoeven, P. J. Reider, M. Labelle, J. Y. Gauthier, Y. B. Xiang, R. J. Zamboni, J. Org. Chem. 1996, 61, 3398. doi:10.1021/JO952103J [18] K. Adachi, T. Iwabuchhi, H. Sano, S. Harayama, J. Bacteriol. 1999, 181, 757. [19] P. Padaratz, M. Fracasso, F. de Campos-Buzzi, R. Correˆa, R. Niero, F. D. Monache, V. Cechinel-Filho, Basic Clin. Pharmacol. Toxicol. 2009, 105, 257. doi:10.1111/J.1742-7843.2009.00441.X [20] C. A. Hollingsworth, P. G. Seybold, C. M. Hadad, Int. J. Quantum Chem. 2002, 90, 1396. doi:10.1002/QUA.10362 [21] M. Belema, N. A. Meanwell, J. Med. Chem. 2014, 57, 5057. doi:10.1021/JM500335H [22] A. K. Sinhababu, R. T. Borchardt, J. Org. Chem. 1983, 48, 2356. doi:10.1021/JO00162A011