Article

pubs.acs.org/jmc

GC-Targeted C8-Linked Pyrrolobenzodiazepine−Biaryl Conjugates with Femtomolar in Vitro Cytotoxicity and in Vivo Antitumor Activity in Mouse Models Khondaker M. Rahman,† Paul J. M. Jackson,† Colin H. James,‡ B. Piku Basu,‡ John A. Hartley,§ Maria de la Fuente,‡ Andreas Schatzlein,‡ Mathew Robson,∥ R. Barbara Pedley,∥ Chris Pepper,⊥ Keith R. Fox,# Philip W. Howard,∇ and David E. Thurston*,†

† Department of Pharmacy, Institute of Pharmaceutical Sciences, King’s College London, 150 Stamford Street, London SE1 9NH, United Kingdom ‡ UCL School of Pharmacy, University College London, 29/39 Brunswick Square, London WC1N 1AX, United Kingdom § Cancer Research UK Drug−DNA Interactions Research Group, UCL Cancer Institute, Paul O'Gorman Building, 72 Huntley Street, London, WC1E 6BT, United Kingdom ∥ UCL Cancer Institute, University College London, Paul O’Gorman Building, 72 Huntley Street, London WC1E 6BT, United Kingdom ⊥ Institute of Cancer & Genetics, Cardiff University School of Medicine, Heath Park, Cardiff CF14 4XN, United Kingdom # Centre for Biological Sciences, University of Southampton, Southampton SO17 1BJ, United Kingdom ∇ Spirogen Ltd., QMB Innovation Centre, 42 New Road, London E1 2AX, United Kingdom

*S Supporting Information

ABSTRACT: DNA binding 4-(1-methyl-1H-pyrrol-3-yl)- benzenamine (MPB) building blocks have been developed that span two DNA base pairs with a strong preference for GC-rich DNA. They have been conjugated to a pyrrolo[2,1- c][1,4]benzodiazepine (PBD) molecule to produce C8-linked PBD−MPB hybrids that can stabilize GC-rich DNA by up to 13-fold compared to AT-rich DNA. Some have subpicomolar IC50 values in human tumor cell lines and in primary chronic lymphocytic leukemia cells, while being up to 6 orders less cytotoxic in the non-tumor cell line WI38, suggesting that key DNA sequences may be relevant targets in these ultrasensitive cancer cell lines. One conjugate, 7h (KMR-28-39), which has femtomolar activity in the breast cancer cell line MDA-MB-231, has significant dose-dependent antitumor activity in MDA-MB- 231 (breast) and MIA PaCa-2 (pancreatic) human tumor xenograft mouse models with insignificant toxicity at therapeutic doses. Preliminary studies suggest that 7h may sterically inhibit interaction of the transcription factor NF-κB with its cognate DNA binding sequence.

■ INTRODUCTION known DNA-targeting agents with a view to designing novel fi DNA is a well-known target for chemotherapeutic intervention, molecules targeted to speci c sequences within genes to allow selective modulation of expression for both research and and the mechanism of action of many currently used anticancer 1,7 drugs involves the inhibition of DNA-related processes such as therapeutic purposes. This has led to substantial progress by 1 a number of groups, the best known example being the hairpin replication and transcription. Many small molecules are known 3,8 to interact in either the major or minor grooves of DNA, or polyamides of Dervan and co-workers. The PBDs are sequence-selective DNA minor-groove insert between base pairs in an intercalative mode. Some bind 9−13 noncovalently, whereas others form monocovalent bonds or binding agents. The naturally occurring PBDs produced inter- or intrastrand cross-links.2 Some molecules such as the by Streptomyces and Micrococcus species are monomeric (e.g., minor-groove noncovalently binding natural products netropsin anthramycin, 3; Figure 1) and form singly alkylated DNA 6,12,19 (1) and distamycin (2) (Figure 1) are selective for AT-rich adducts, whereas the synthetic PBD dimers consist of two sequences of DNA, whereas other families of agents such as the PBD units joined through a C8/C8′-linker and can form minor-groove covalently binding pyrrolo[2,1-c][1,4]- interstrand or intrastrand DNA cross-links in addition to benzodiazepines (PBDs) (e.g., anthramycin, 3) are selective − for GC-rich sequences.3 6 There has been significant interest in Received: December 21, 2012 trying to understand the molecular basis of the selectivity of Published: March 21, 2013

© 2013 American Chemical Society 2911 dx.doi.org/10.1021/jm301882a | J. Med. Chem. 2013, 56, 2911−2935 Journal of Medicinal Chemistry Article

Figure 1. Structures of the AT-targeting minor-groove noncovalently binding agents netropsin (1) and distamycin (2), the GC-targeting covalently binding anthramycin (3), and the C8-linked PBD−polyamide conjugate GWL-78 (4).

− monoadducts.14 17 One PBD dimer, SJG-136, has successfully the sequence selectivity of the two components. These − completed phase I clinical trials18 20 and is presently conjugates were also shown to have excellent cellular/nuclear undergoing phase II evaluation in ovarian and hematological penetration properties, and a degree of correlation was cancers. PBD molecules have a chiral center at their C11a(S)- observed between cytotoxicity and DNA-binding affinity. position, which provides them with an appropriate 3-dimen- Furthermore, a robust sequence-selective blockade of tran- sional shape to fit perfectly within the DNA minor groove.12 scription at sites approximately corresponding to their DNA They also possess an electrophilic N10−C11 moiety (i.e., footprints was observed. interconvertible imine, carbinolamine, or carbinolamine methyl To develop molecules capable of targeting longer GC-rich ether functionalities) that can form a reversible covalent aminal DNA sequences as potential transcription factor inhibitors, we linkage between their C11-position and the nucleophilic C2- decided to explore replacement of the polypyrrole units within 12,21 NH2 group of a guanine base. PBD monomers such as 3 these PBD conjugates with GC-recognizing fragments. The lack typically span three base pairs of DNA with a reported of GC-preference for polypyrrole molecules such as distamycin preference for 5′-Pu-G-Pu-3′ sequences,1,12 although more and netropsin is thought to relate to the wider minor-groove recent data suggest that they have a kinetic preference for 5′- width in GC-rich regions of DNA and the presence of the Py-G-Py-3′ sequences.22 It is known from the literature that the exocyclic 2-amino groups of guanine bases which protrude into DNA binding affinity of PBD molecules correlates well with the minor groove. This prevents these molecules, most of their in vitro cytotoxicity in tumor cell lines,6,15,23 and PBD which are naturally curved, from achieving multiple close van dimers that can cross-link DNA have a greater cytotoxic der Waals contacts with functional groups in the floor and walls potency compared to PBD monomers that can only of the minor groove.3,4 Several studies using empirical force monoalkylate.15 As a result of covalently binding to DNA, fields have concluded that van der Waals contacts are the PBDs can mediate a number of biological effects including the dominant factor in sequence recognition, and Dervan and co- inhibition of endonucleases,24 RNA polymerase,6,25 and workers have devised a sequence rule based on the pairing of transcription factor binding.26,27 For the PBD dimers, the different five-membered heterocycles in a hairpin polyamide most likely basis for their antitumor activity is thought to be DNA motif.8 However, these rules are only valid when preferential DNA repair in healthy cells compared to tumor polyamides are linked through a hairpin motif, and so do not cells, with the repair response depending on the cell type, and apply to linear polyamides. Furthermore, poor cellular uptake the extent and duration of exposure to the agent.28 Tumor cells and nuclear penetration of the relatively high molecular weight are often deficient in one or more relevant DNA repair hairpin polyamides have been major disadvantages for these pathways, thus leading to selective cytotoxicity and antitumor molecules, and this has limited their use as tools in cellular activity in vivo.29 Many PBD molecules also have significant experiments and as potential therapeutic agents. Therefore, in a − antimicrobial activity.30 35 different approach to this problem, we have focused on the In 2006 we reported a series of C8-linked PBD−poly(N- development of a series of biaryl building blocks designed to methylpyrrole) conjugates (e.g., GWL-78, 4; Figure 1) which possess a unique curvature in order to enhance van der Waals demonstrated the synergistic effect of joining a GC-specific contacts within the GC-tracts of the DNA minor groove. Such PBD unit to an AT-recognizing polypyrrole fragment, as building blocks, which span two DNA base pairs, offer a illustrated by a significant increase in DNA binding affinity (up synthetic advantage when designing longer more complex gene- to 50-fold) for 4 compared to its component PBD and targeting molecules, in that fewer building blocks are required tripyrrole fragments.6 DNA footprinting experiments showed to span a greater number of base pairs. that these molecules bind to DNA sequences, with the binding We started by designing a library of fragments to span two site size increasing with molecular length and with the majority DNA base pairs based on simple phenyl-substituted hetero- ′ of sites conforming to the consensus motif 5 -XGXWz (X = any cycles of appropriate length (e.g., Figure 2). With each base but preferably a purine, W = A or T, z =3± 1), reflecting containing one heterocyclic and one carbocyclic component,

2912 dx.doi.org/10.1021/jm301882a | J. Med. Chem. 2013, 56, 2911−2935 Journal of Medicinal Chemistry Article

interaction at one or more of the NF-κB consensus recognition sites, thus inhibiting the NF-κB signaling pathway (Figure 8A). The novel biaryl building blocks developed through this study contribute to the pool of molecular fragments available for the synthesis of DNA-targeting agents, and differentiate themselves from other agents through their CG-recognition properties. The significant in vitro and in vivo activity of 7h, the large differential effect on tumor compared to non-tumor (i.e., Figure 2. (A) Structure of a biaryl building block (MPB) in the amino WI38) cells, and the apparent lack of toxicity in mice suggest ester form. (B) Molecular model showing a biaryl building block that it may have therapeutic potential, especially in NF-κB- (purple) spanning two DNA AT base pairs. dependent tumor types. sufficient diversification of library members was achieved ■ RESULTS AND DISCUSSION through modifications to, and substitutions within, both Synthesis of the Biaryl Polyamides. A linear synthetic components. This allowed the production of libraries of route was devised to prepare the initial building blocks, and building blocks (e.g., Type 1, 5a and 5b, and Type 2, 6a and points of diversity were introduced using parallel chemistry. 6b, Figure 3A) possessing different curvatures resulting in The synthesis was based on the application of HOBT/DIC- greater van der Waals interaction and hydrogen-bonding mediated amide coupling and microwave-assisted Suzuki abilities with the potential to complement functional groups coupling. Two major types of solution-phase libraries were and hydrogen bond donors and acceptors in the walls and floor prepared (Schemes 1 and 2): Type 1 with only one biaryl unit of the minor groove. A fluorescent intercalator displacement in the central position of the polyamide structure, and Type 2 (FID) assay36 was then used to select molecules with a consisting of two joined biaryl units (Figure 3A). preference for GC- rather than AT-rich sequences which For Type 1 building blocks the synthesis started with identified the 4-(1-methyl-1H-pyrrol-3-yl)benzenamine (MPB) commercially available N-methylpyrrole (8), and polyamides of motif (Figure 2A). To demonstrate the potential use of these type 5 were synthesized in a total of seven or eight steps novel building blocks as components of longer more complex (Scheme 1). Initial electrophilic substitution at the C2-position covalently binding DNA-targeting molecules, we then synthe- of the pyrrole ring with trichloroacetyl chloride (90% yield) sized a series of PBD−MPB conjugates (7a−h) in which the followed by nitration with concentrated nitric acid and acetic PBD and MPB fragments were separated by a four-carbon anhydride afforded 10 in good yield (95%). The dimethylamino linker (Figure 4). tail unit was then introduced by coupling to (3- Using an ion pair reversed-phase HPLC assay14,37 in (dimethylamino)propyl)amine to produce intermediate 11,a conjunction with AT-rich and GC-rich oligonucleotides, we reaction that proceeded smoothly without the use of coupling − − showed that MPB units within PBD MPB conjugates prefer to reagents. The nitro group of 11 was then reduced with H2/Pd interact with GC base pairs close to PBD binding sites (Figure Ctoafford the pyrrole amine 12, which was used as a starting 5) in contrast to pyrrole-linked PBDs of type 4 (Figure 1). point for the synthesis of all molecules of Type 1. For example, Similarly, using a FRET-based assay based on oligonucleotides it was coupled to different five-membered bromo heterocyclic containing only one PBD binding site but with modified acids to provide the Suzuki substrates of type 13 in excellent flanking sequences, we showed that PBD−MPB conjugates yield (88−92%). Suzuki coupling between these intermediates stabilize GC-rich oligonucleotides with up to 13-fold greater and a number of amino- or nitroboronic acids proceeded well efficiency compared to AT-rich oligonucleotides, whereas the under microwave conditions to provide intermediates of type opposite was found for pyrrole-linked PBDs which prefer AT 14 in good yields (typically 92−97%). The products of sequences (Table 1). We also confirmed the sequence nitroboronic acids were subsequently reduced to their amine − − selectivity of the PBD MPB conjugates by DNase I foot- forms using H2/Pd C. Amino intermediates of type 14 were printing, with strong footprints observed in GC-rich regions then coupled to either commercially available heterocyclic varying consistently according to the relative position of the carbonyl chlorides or commercially available heterocyclic MPB building block with respect to the PBD moiety (Figure carboxylic acids to give the Type 1 biaryl polyamides (e.g., 5a 6). and 5b) in good yields (87−93%). The 3-(dimethylamino)- These PBD−MPB conjugates were shown to have propyl side chain within all intermediates and final library fi “ subnanomolar IC50 values in MCF7, A431, A2780, A549, members assisted with their puri cation using a catch and MIA PaCa-2, and MDA-MB-231 human tumor cell lines and release” method based on sulfonic acid-containing SCX primary chronic lymphocytic leukemia (CLL) cells, while being resins.38 up to 5-fold less active in the non-tumor cell line WI38 (Table The dimeric Type 2 biaryl polyamides were prepared from 2). In particular, compounds 7g and 7h were active at the different biaryl units as shown in Scheme 2. Four different femtomolar level in the MCF7 and MDA-MB-231 cell lines, biaryl amine building blocks containing a 3-(dimethylamino)- and 7h was found to have significant in vivo antitumor activity propyl tail were synthesized and coupled to a variety of in human tumor xenograft mouse models based on the MDA- commercially available biaryl carboxylic acids. For the N- MB-231 (breast) and MIA PaCa-2 (pancreatic) cell lines, with methylpyrrole-containing biaryl amines, the synthesis started no signs of toxicity at the doses studied (Figure 7). Preliminary from the intermediate 2-(trichloroacetyl)-N-methylpyrrole (9; observations based on transcription factor array experiments, see Scheme 1). Bromination of 9 with N-bromosuccinamide Western blotting studies in 7h-treated primary CLL cells produced the 3-bromopyrrole 16, which was joined to the 3- (Figure 8A,B), staining of tumor cells from biopsies from the (dimethylamino)propyl tail (without the need for a coupling ff − xenograft experiments (Figure 8C), and molecular modeling reagent) to a ord the Suzuki substrate 17 (X = N, N CH3). (Figure 9) suggest that 7h could be working through For furan- and thiophene-containing biaryl amines, the

2913 dx.doi.org/10.1021/jm301882a | J. Med. Chem. 2013, 56, 2911−2935 Journal of Medicinal Chemistry Article

Figure 3. (A) Structures of the different types of noncovalent building blocks all containing the basic biaryl MPB unit (highlighted in blue, Figure 2A) but flanked on either side with pyrrole rings (5a, Type 1) or pyrrole and imidazole rings (5b, Type 1), or with phenylthiazole units attached on one side through either meta (6a, Type 2) or para (6b, Type 2) configurations. Type 1 molecules contain only one MPB unit (highlighted in blue) as the biaryl building block, while Type 2 molecules contain a biaryl unit in addition to the MPB unit. (B) Analysis of the FID assay results showing a comparison between the MPB biaryl-based polyamides 5a,b and 6a,b, distamycin (2), and the AT-selective minor-groove binding agent f−Py−Py− Py for their GC sequence preferences. The vertical bars represent the percentage number of GC-rich versus AT-rich sequences bound within the top 25 sequences. synthesis started from commercially available bromo hetero- from intermediate 17 onward, each library member was purified cyclic acids of type 15 (X = O or S) which were coupled to the by the catch and release method using sulfonic acid-based SCX 3-(dimethylamino)propyl tail using HOBT/DIC to obtain resins.38 intermediates of type 17. The Suzuki coupling reaction Fluorescent Intercalation Displacement (FID) Assay. between 17 and the 4-amino/3-nitroboronic acids proceeded The novel Type 1 and Type 2 biaryl polyamides were screened well under microwave conditions to give (after reduction with against a 512-member combinatorial DNA hairpin library − fi H2/Pd C in the case of the nitro compounds) high yields of containing all possible permutations of ve base pairs using an the biaryl amine building blocks of type 18. Finally, these were optimized version of Boger’s fluorescence-based intercalator coupled to biaryl carboxylic acids of type 19 using HOAT/ displacement (FID) assay36,39 with a 1:2 molar ratio of HATU/DIPEA to obtain Type 2 biaryl polyamides (e.g., 6a oligonucleotide to ligand. The change in fluorescence due to and 6b) in good yields (68−93%). As with the Type 1 library, displacement of ethidium bromide was measured using a

2914 dx.doi.org/10.1021/jm301882a | J. Med. Chem. 2013, 56, 2911−2935 Journal of Medicinal Chemistry Article

Figure 4. Structures of the PBD−MPB C8-conjugates 7a−7h which range from the connection of one (7a) or two (7b) MPB units, to the use of one MPB unit but with one (7c,g) or two (7e) flanking pyrrole rings, one flanking imidazole ring (7d,h), or a combination of pyrrole and imidazole flanking rings (7f).

Figure 5. HPLC chromatograms of the reaction of PBD−MPB 7a with (A) an AT-rich hairpin (Seq-1) showing an insignificant amount of adduct formation after 5 min, and (B) a GC-rich hairpin (Seq-2) showing rapid adduct formation after 5 min; (C) a molecular model of 7a shown in red covalently bound to the AT-rich hairpin sequence, and (D) a molecular model of 7a shown in red covalently bound to the GC-rich hairpin sequence. In both (C) and (D) the hydrogen-bonding interactions are shown in dark blue. More hydrogen-bonding interactions (i.e., three versus two) can be seen between 7a and the GC-rich sequence in (D), thus supporting the HPLC−MS observations. In (C) and (D) the bases to which hydrogen bonds form are shown in different shades of blue.

fluorescence plate reader. The top 25 sequences based on the establish the sequence preference. The results (Figure 3B; binding affinity of the ligands were ranked and analyzed to Table S1, Supporting Information) indicate that library

2915 dx.doi.org/10.1021/jm301882a | J. Med. Chem. 2013, 56, 2911−2935 Journal of Medicinal Chemistry Article ΔT ° − Table 1. m ( C) Values for the PBD MPB Conjugates after 24 h Incubation with AT- and GC-Rich Hairpin Oligonucleotides at a Ratio of 5:1 (Ligand−DNA) at a a Ligand Concentration of 1 μM

ΔT (GC)/ Δ Δ Δ m Tm(AT) Tm(GC) Tm(AT) at 1 ligand Unit 1 Unit 2 at 1 μM at 1 μM μM PBD acid OH 0.4 0.2 0.5 (37) 7a MPB 0.3 3.9 13.0 7b MPB MPB 0.8 3.6 3.6 7c MPB Py 2.1 3.1 1.48 7d MPB Im 0.5 2.8 5.6 7e MPB Py−Py 3.2 4.3 1.34 7f MPB Py− 3.4 5.8 1.70 Im 7g Py MPB 5.0 1.4 0.28 7h Im MPB 6.0 6.2 1.03 4 Py Py 7.4 1.1 0.15 DC-81 0.2 0.0 0 aUnit 1 and Unit 2 are joined by an amide linkage, and the terminal unit contains an ester functionality (except for 37). The final column Δ Δ provides the ratio Tm(GC)/ Tm(AT) for comparative purposes. Py = pyrrole, Im = imidazole, for DC-81 see ref 42. members containing an MPB motif switched binding preference from AT-rich to GC-rich sequences. For example, the percentages of GC bases in the 25 most preferred sequences for compounds Py−MPB−Py (5a), Py−MPB−Im (5b), MPB−phenylthiazole (para−meta)(6a), and MPB− phenylthiazole (para−para)(6b) were 72%, 78.4%, 76%, and 71.2%, respectively, compared to just 12% for distamycin (2) − − − Figure 6. (A) Summary of the key footprints (underscored) of the and 9% for the tripyrrole polyamide f Py Py Py (Supporting − − fi PBD MPB conjugates 7a h. Sequences highlighted in red were Information). Restricting the analysis to the top ve most produced mainly by conjugates with MPB units attached directly next preferred sequences increased the percentage of GC content for to the PBD core (e.g., 7a and 7b), whereas sequences highlighted in some compounds still further to >90% (Figure S1, Supporting blue were more common for conjugates with one pyrrole or imidazole Information). ring between the MPB and PBD units (e.g., 7g and 7h); (B) DNase I Distamycin (2) and f−Py−Py−Py (an AT-selective control footprints showing the interaction of 7h and 4 with HexARev and molecule) showed significant AT-selectivity within their top 25 HexB. Ligand concentrations (μM) are indicated at the top of each gel sequences, and completely failed to bind to hairpins containing lane. Tracks labelled “con” show DNase I cleavage in the absence of “ ” fi four or five GC base pairs. In comparison, MPB biaryl added ligand, while GA tracks are markers speci c for purines. The ′ 32 compounds bound between 5 and 18 four- or five-base-pair GC DNA fragments were labeled at their 3 -ends with P. sequences but failed to bind to a single four- or five-base-pair AT sequence (Table S1, Supporting Information). This upward and out of the minor groove. This change in transition in base-pair preference resulting from changing an conformation is due to the presence of the MPB building N-methylpyrrole unit to an MPB is novel and provides a basis block (which contains an additional phenyl ring) at the center for the design of new generations of minor-groove binding of the 5a and 5b structures, in contrast to distamycin, which ligands with enhanced GC-selectivity. Although the precise contains only pyrrole rings. Free energy values calculated after a molecular basis for this switch in base-pair preference is unclear, 20 ns molecular modeling simulation also support this it is likely that overall molecular shape plays a role as, according observation (Table S3, Supporting Information). Next, these to molecular modeling studies, the biaryl ligands have a unique MPB units were attached to a covalently binding PBD molecule 3-dimensional structure compared to pyrrole-containing to establish their potential use in more-complex GC-targeted compounds such as netropsin and distamycin. It was also ligands. observed that, unlike 2, both 5a and 5b adopt a different Synthesis of the PBD−MPB Conjugates. The biaryl units conformation within the minor groove with the carbonyl described above were coupled to the C8-position of the PBD groups of their amide linkages pointing downward, thus moiety via a four-carbon linker chosen because it had been providing additional hydrogen-bonding interactions with previously optimized during the design of the GWL-78 (4) guanine bases. Distamycin, on the other hand, adopts a series of PBD conjugates.6 The intention was to compare the different conformation with its carbonyl groups pointing sequence selectivity of these new conjugates to the parent

2916 dx.doi.org/10.1021/jm301882a | J. Med. Chem. 2013, 56, 2911−2935 Journal of Medicinal Chemistry Article − Table 2. IC50 Values (nM) for PBD-MPB Conjugates 7a 7h and Control Molecules Determined after 96 h Exposure in a Panel of Human Cancer Cell Lines (A431, A549, A2780, MCF7, MDA-MB-231, and MIA PaCa-2), Primary CLL Cells, and the a Noncancer Cell Line WI38

IC50 (nM) compd A431 A549 A2780 MCF7 MDA-MB-231 MIA PaCa-2 primary CLL WI38 7a 2.31 7.50 1.87 1.91 2.11 1.2 6.2 159.9 7b 2.91 0.54 0.56 0.90 0.46 0.35 4.7 158.7 7c 4.60 0.019 0.96 1.70 2.70 0.34 2.1 425.9 7d 0.19 0.47 0.15 0.37 0.45 0.11 3.0 1240 7e 0.86 2.3 0.24 0.31 0.59 0.31 0.96 41.3 7f 0.064 0.45 ND 0.075 0.015 0.25 0.037 65.6 7g (KMR-28-35) 0.0056 0.056 0.013 0.00002 0.000065 0.0013 0.098 473.8 7h (KMR-28-39) 0.018 0.034 0.021 0.00002 0.00018 0.0021 0.17 129.2 4 0.55 6.1 0.57 0.32 0.12 2.4 1.3 41.3 DC-81 1727.3 4075.4 1055.2 534.8 816.0 NDb NDb 14095.3 PBD acid (37) 1089.9 1495.1 3756.4 702.3 1668.3 1564.0 NDb 4663.1 MPB Unit (23) >20000 >20000 >20000 >20000 >20000 >20000 NDb >20000 aThe MTT colorimetric assay43 was used to measure cytotoxicity in all cell lines except for the primary CLL cells, for which the annexin V assay44 b μ was employed. ND = No data obtained under the conditions of the experiment as the relative cell survival was higher than 50% (IC50 >20 M).

GWL-78 molecule which consists of a PBD and an AT- with tetrakis(triphenylphosphine)palladium(0) and pyrrolidine targeting bis(N-methylpyrrole) fragment joined through a C4- in DCM to afford the final products 7a−f in their N10−C11 linker. As expected from its structure, GWL-78 has a preference imine forms in good yields (92−97%). For target molecules for DNA motifs consisting of a PBD binding site (e.g., Pu-G- with a heterocyclic unit between the PBD linker and MPB units Pu) adjacent to a series of AT base pairs.6 Based on this, it was (e.g., 7g, h), 36 was initially coupled to the appropriate anticipated that the analogous PBD−MPB conjugates should heterocylic ester, followed by hydrolysis to afford intermediates recognize DNA motifs containing a PBD binding site next to 39 and 40. Coupling to the MPB unit afforded intermediates of runs of GC-rich DNA, thus creating novel DNA-targeting types 38g, h, which could be N10/C11-deprotected to afford agents with complementary base-pair recognition properties the target molecules of types 7g, h. The coupling and Alloc/ compared to GWL-78. THP-deprotection reactions were performed in parallel using a The MPB component of the PBD−MPB conjugates was GreenHouse parallel synthesizer, and all compounds were synthesized according to Scheme 3. The chloroacetylated shown to be >95% pure on the basis of LC−MS analysis. bromopyrrole 16 (see Scheme 2) was converted to its methyl Finally, the PBD capping unit 36 was itself deprotected under ester 20 (95%) and then subjected to Suzuki coupling with 4- similar conditions to produce the parent PBD N10−C11 imine ((tert-butoxycarbonyl)amino)phenylboronic acid under micro- 37, which was used as a control in the cell culture experiments. wave conditions to obtain the N-Boc-MPB methyl ester 21 Ion Pair Reversed-Phase HPLC−MS Assay. The (87%), which was then hydrolyzed with NaOH/dioxane to sequence preference of the simplest PBD−MPB conjugate afford the N-Boc-MPB acid 22 (91%). Deprotection of 21 (7a) toward short (i.e., 17 bp) custom-designed oligonucleo- using 4 M HCl in dioxane afforded the core MPB ester 23, tides was evaluated initially using an ion pair reversed-phase which was coupled to 22, and the product deprotected in situ HPLC−MS assay14,37 with the PBD−dipyrrole conjugate to afford the dimeric MPB ester 24. The N-Boc-MPB acid 22 GWL-786 (4) as a control (Figure 5A; Figure S2, Supporting was also coupled to a series of (4-Boc-amino)pyrrole/imidazole Information). This methodology allowed the kinetics of the methyl esters to obtain 25 and 26, and these were Boc- interaction of these covalently binding molecules with DNA to deprotected to afford the MPB−Py/imidazole amines of types be measured. The hairpin-forming oligonucleotides used in the 27 and 28. Boc-protected 25 and 26 were hydrolyzed with study [i.e., 5′-TATA-AGA-TTT-TCT-TATA-3′ (Seq-1) and NaOH/dioxane to obtain the respective acids 29 and 30, and 5′-GCGC-AGA-TTT-TCT-GCGC-3′ (Seq-2)] each contained these were coupled to 4-amino-1-methyl-1H-pyrrole-2-carbox- only one preferred PBD binding triplet (i.e., AGA) suitable for ylate to afford Boc-protected Py/Im−MPB combinations of covalent interaction, and we recently demonstrated that types 31 and 32. These were subsequently Boc-deprotected to hairpin-forming oligonucleotides of this type can form stable obtain 33 and 34, which were combined with the PBD capping adducts with PBDs.14 ff − unit (Scheme 4) to a ord the PBD MPB hybrid compounds Annealed AT-rich Seq-1 alone gave a distinct peak at tR = 7a−h. 26.9 min (Figure S2, Supporting Information). After addition The N10-Alloc-C11-O-THP-protected PBD capping unit 36 of 4 to Seq-1 (4:1 ligand:DNA ratio) followed by immediate (Scheme 4) was synthesized according to a modified version of injection onto the HPLC column, rapid appearance of an 6 a procedure reported by Wells and co-workers but with adduct peak at tR = 28.7 min was observed, along with a improved yield and lower reaction times (see Supporting decrease in unreacted oligonucleotide (tR = 26.9 min). Reaction Information). It was then coupled to the MPB polyamide units was complete after 3 h, with the 26.9 min peak disappearing 23−34 (Scheme 3) using EDCI/DMAP/DCM. For conjugates completely. The identity of the 28.7 min peak was confirmed by with the MPB unit attached directly to the C8-linker of the MALDI-TOF MS as the 1:1 4−Seq-1 covalent hairpin adduct. PBD (e.g., 7a−f), 36 was directly coupled to provide This rapid reaction with the AT-rich Seq-1 was anticipated on intermediates of types 38a−f. The N10-Alloc and C11-O- the basis of previous studies6,14 and also the high pyrrole THP protecting groups were then simultaneously removed content of 4. Next, a similar experiment was carried out with 7a

2917 dx.doi.org/10.1021/jm301882a | J. Med. Chem. 2013, 56, 2911−2935 Journal of Medicinal Chemistry Article

Figure 7. (A) In vivo antitumor activity of 7h in immune-deficient mice bearing a human MDA-MB-231 (breast) xenograft at dose levels of 250 and 300 μg/kg, demonstrating a clear dose response compared to the control (vehicle only). The arrow indicates the day of the last iv dose administered. (B) In vivo antitumor activity of 7h in immune-deficient mice bearing a human MIA PaCa-2 (pancreatic) xenograft at the dose level of 300 μg/kg. Significant tumor growth suppression was found in both cases, with no toxicities observed at the dose administered, or found in major organs after autopsy. using the same sequence. In this case very slow adduct remained constant throughout these experiments, the rate of fi formation (tR = 28.2 min) was observed with insigni cant adduct formation could be directly attributed to the biaryl and reaction at 0 h (i.e., approximately 5 min), and with the dipyrrole components of 7a and 4, respectively, suggesting that reaction still not complete after 24 h (Figure 5A). This result the MPB component prefers GC base pairs near the PBD was consistent with data from the FID assay36 in which binding site, which manifests as a faster rate of covalent polyamides containing biaryl units had a poor tolerance for AT- modification of DNA. Therefore, these results provided rich DNA. However, repeating the experiment with the GC- quantitative evidence that 7a preferred the GC-rich Seq-2 rich Seq-2 gave the opposite result, with 4 failing to provide (67% binding within 5 min) but had poor tolerance for the AT- significant amounts of covalent adduct (27.6 min) after rich Seq-1 (3% binding within 5 min). Conversely, 4 had poor approximately 5 min (Supporting Information), whereas 7a tolerance for the GC-rich Seq-2 (12% binding within 5 min) fi formed a signi cant quantity (>50%) of adduct (tR = 27.6 min) but a strong preference for AT-rich Seq-1 (70% binding within after the same time period, identified as the 1:1 7a−Seq-2 5 min). Molecular modeling studies on the interaction of 7a covalent complex by MALDI-TOF MS (Figure 5B). As the with the two oligonucleotides (Figure 5C,D) supported the structure of the PBD unit and associated linker for both experimental observations and are described in the “Molecular conjugates, and the composition of the oligonucleotides, Modeling Studies” section”.

2918 dx.doi.org/10.1021/jm301882a | J. Med. Chem. 2013, 56, 2911−2935 Journal of Medicinal Chemistry Article

Figure 8. (A) Diagram of the possible role of 7h in the inhibition of NF-κB and the down-regulation of NF-κB-associated genes. (B) Effect of 7h on the expression of phosphor-p65 and phosphor-IKBα in primary CLL cells after 24 h incubation. (C) Results of quantitation of immunohistochemical (IHC) staining of sections taken from MIA PaCa-2 xenografts in mice treated with 7h compared to control, showing down-regulation of BCL2 expression.

FRET-Based DNA Melting Study. The DNA binding of the GC-rich compared to the AT-rich hairpin, with the other affinity of PBD molecules has been evaluated frequently in the PBD−MPB conjugates providing a significant but lower degree past using a calf thymus DNA thermal denaturation assay,6,15,40 of stabilization of the GC-rich oligonucleotides. These results although this methodology has a number of limitations. For correlated well with the HPLC−MS data, which indicated that example, calf thymus DNA contains long genomic DNA the 7a and PBD−dipyrrole (4) molecules had significantly sequences which include a large number of PBD binding sites greater preferences for the GC-rich and AT-rich oligonucleo- that can mask the effect of subtle changes in sequence tides, respectively. selectivity for individual ligands. To more accurately assess the Footprinting Study. The sequence selectivity profiles of DNA binding affinity and sequence preference of the new the PBD−MPB conjugates were studied further by DNase I PBD−MPB conjugates, we developed a FRET-based DNA footprinting using the Hex A and Hex B fragments, which were melting assay based on a duel-labeled hairpin-forming designed to contain all 64 combinations of symmetrical oligonucleotide containing only one PBD binding site (Figure hexanucleotides.41 The reverse fragments (HexARev and S3, Supporting Information). This had the advantage that HexBrev) that contain the same sequences but in the opposite sequences surrounding the binding site could be easily modified orientation were also used in the study. A summary of the to be either AT- or GC-rich, thus enabling observation of the observed footprints for 7a−h, along with representative gels for effect on binding affinity for individual ligands. The study was 4 and 7h with the HexARev and HexB fragments, are shown in carried out to understand the GC sequence preference of the parts A and B, respectively, of Figure 6 (further gels are MPB building block when covalently attached to a PBD core. provided in Figure S4, Supporting Information). The sequences We used a very short hairpin oligonucleotide containing a highlighted in red were mainly protected by conjugates with seven-base-pair stem and a three-base-pair loop, and made MPB units attached directly to the PBD core, whereas changes to the flanking sequence (i.e., AT vs GC) next to the sequences highlighted in blue were more commonly protected PBD binding site (AGA). The same hairpin oligonucleotides by conjugates where pyrrole or imidazole rings had been without the fluorophores were used for the HPLC−MS study introduced between the PBD and MPB units (i.e., PBD−[Py/ described above. The melting temperatures of these short Im]−MPB). oligonucleotides were low, and it was not anticipated that the It is clear that both molecules bind to a large number of sites PBD−MPB compounds would increase their melting temper- on these fragments, and several of the protected regions consist atures significantly. However, the results summarized in Table 1 of overlapping binding sites. Due to the ability of these ligands clearly show that the relative positions of the pyrrole, imidazole, to bind covalently to guanine, it is not possible to define the and MPB units play a significant role in the extent to which an precise binding preference for each one, although some general individual ligand can stabilize a particular DNA sequence. For features can be determined. Useful information about the example, the presence of a pyrrole or imidazole ring sequence preference of the MPB binding motif can be obtained immediately adjacent to the four-carbon linker provides a when the footprinting pattern of 7h (PBD−Im−MPB) is greater DNA-stabilizing effect upon the AT-rich hairpin, while compared with that of the AT-preferring 4 (PBD−Py−Py). On MPB units at this position diminish the stabilizing effect. The the basis of molecular models, conjugates 4 and 7h are opposite effect was observed for the GC-rich hairpin, except in expected to span six and seven base pairs, respectively, with the the case of PBD−Im−MPB (7h), which retained its DNA- PBD core of each spanning three base pairs. When the clearest stabilizing effect on both types of oligonucleotides. This was footprints obtained at the lowest concentrations are compared anticipated, as imidazoles are known to tolerate GC sequences for both compounds, 7h appears to bind to sequences well. Although an MPB unit immediately next to the linker containing an N-G-N-W-N-S-S motif (N = any base; W = A fi Δ signi cantly enhanced the Tm for the GC-rich hairpin, or T; S = G or C), while 4 provides distinct footprints at swapping the position of the MPB unit with a pyrrole, or sequences containing an N-G-N-N-W-W pattern. Crucially, 7h replacement of both MPB units with pyrroles, resulted in a failed to protect these AT-rich sequences at the lowest Δ − decrease in Tm. Interestingly, the simple PBD MPB concentration, while 4 failed to protect the GC-rich N-G-N- conjugate (7a) produced an approximately 13-fold stabilization W-N-S-S sequences protected by 7h. Furthermore, according

2919 dx.doi.org/10.1021/jm301882a | J. Med. Chem. 2013, 56, 2911−2935 Journal of Medicinal Chemistry Article

Figure 9. (A) Molecular models of 7h covalently bound to the second guanine of the 5′-GGGACTGTCC-3′ and 5′-GGGGGTCGCC-3′ duplexes (i.e., consensus NF-κB recognition sequences) showing favorable hydrogen-bonding interactions (green) between the ligand and guanine bases (purple). (B) Energy-minimized molecular model showing inhibition of interaction of NF-κB (red) with its cognate DNA sequence (gray) by 7h (blue). The imidazole ring of 7h is shown interacting with lysine 145 (yellow) of the NF-κB protein, thus sterically hindering approach of the protein to its DNA binding site. to the observed footprinting patterns at the lowest concen- guanines of every possible N-G-N binding triplet within a long tration, the MPB building block lined up with the terminal G/ stretch of DNA, and the noncovalent component of the PBD C-G/C section of the N-G-N-W-N-S-S pattern, while the Py− C8-conjugates (i.e., −Im−MPB in the case of 7h) cannot Py component of 4 aligned with the terminal W-W section of exclusively determine the sequence preference of these the N-G-N-N-W-W motif, consistent with the expected molecules. Therefore, at high concentrations of 7h, protection sequence preferences for these building blocks. of both GC- and AT-rich sequences with a nearby N-G-N A further detailed examination of the footprint patterns for binding triplet was observed, but at low ligand concentrations 7h showed that it protected every occurrence of N-G-N-W-N- enhanced protection of GC-rich sequences occurred, with the S-S motifs that appeared in the Hex-A, Hex-A reverse, Hex-B, opposite effect observed for the AT-preferring GWL-78 (Figure and Hex-B reverse sequences. However, the protection was not 6B). limited to this pattern, and there were a significant number of Cytotoxicity Data. The in vitro cytotoxicities of the PBD− AT-rich sequences protected as well. This broad sequence MPB conjugates 7a−h, GWL-78 (4), the PBD acid 37, the protection may be explained by the fact that the reactive imine MPB ester 23,DC-8142 (49; Supporting Information), moiety of the PBD fragment can potentially covalently bond to anthramycin (3), and the two biaryl polyamides 5a (Type 1)

2920 dx.doi.org/10.1021/jm301882a | J. Med. Chem. 2013, 56, 2911−2935 Journal of Medicinal Chemistry Article

a Scheme 1. Synthesis of Type 1 Biaryl Polyamides

a − ° Reagents and conditions: (i) trichloroacetyl chloride, THF, rt, 4 h; (ii) concd HNO3, acetic anhydride (added slowly), 5 C, 3 h; (iii) (3- (dimethylamino)propyl)amine, dry THF, rt, 6 h; (iv) H2, Pd/C, 4 h; (v) 5-bromo heterocyclic acids, DIC, HOBT, DCM, rt, 16 h; (vi) 4- − aminoboronic acid, (PPh3)4Pd, K2CO3, ethanol/toluene/water (9:3:1), microwave, 12 23 min, (vii) 3-nitroboronic acid, (PPh3)4Pd, K2CO3, − ethanol/toluene/water (9:3:1), microwave, 8 15 min; (viii) H2, Pd/C, EtOAc, 4 h; (ix) heterocyclic carbonyl chloride, dry DMF, rt, 2 h; (x) heterocyclic carboxylic acid, DIC, HOBT, DCM, rt, 16 h.

a Scheme 2. Synthesis of Type 2 Biaryl Polyamides

aReagents and conditions: (i) NBS, dry THF, −10 °C for 2 h, rt for 4 h; (ii) (3-(dimethylamino)propyl)amine, dry THF, rt, 6 h; (iii) (3- (dimethylamino)propyl)amine, DIC, HOBT, DCM, rt, 16 h; (iv) 4-aminoboronic acid, (PPh3)4Pd, K2CO3, ethanol/toluene/water (9:3:1), − microwave, 12 23 min; (v) 3-nitroboronic acid, (PPh3)4Pd, K2CO3, ethanol/tolune/water (9:3:1), microwave, 8-15 min; (vi) H2, Pd/C, EtOAC, 4 h; (vii) DIC, HOBT, DCM, rt, 16 h. and 6a (Type 2) were evaluated in a number of tumor cell lines, and MPB units had negligible activity across the cell lines, primary CLL cells, and the non-tumor cell line WI38 after 96 h whereas the PBD−MPB conjugates 7a−h ranged in activity of exposure using the MTT [3-(4,5-dimethylthiazol-2-yl)-2,5- from submicromolar to femtomolar. These results illustrate the diphenyltetrazolium bromide] colorimetric assay43 for all synergy of joining a PBD unit to an MPB heterocyclic moiety, immortalized cells lines (MCF7, A431, A2780, A549, MIA with the combination providing a significant increase in PaCa-2, MDA-MB-231, WI-38) and the annexin V assay44 for cytotoxicity compared to the component units. For example, − − the primary CLL cells (Table 2). In accord with the thermal PBD Py MPB (7g) had picomolar to femtomolar IC50 values ff denaturation and footprinting data, the nonconjugated PBD in the di erent cancer cell lines (e.g., IC50 = 65 fM in MDA-

2921 dx.doi.org/10.1021/jm301882a | J. Med. Chem. 2013, 56, 2911−2935 Journal of Medicinal Chemistry Article

a Scheme 3. Synthesis of MPB−Heterocycle Conjugates

a fl Reagents and conditions: (i) sodium methoxide, MeOH (anhydrous), H2SO4, 30 min re ux; (ii) 4-((tert-butoxycarbonyl)amino)phenylboronic − acid, (PPh3)4Pd, K2CO3, ethanol/toluene/water (9:3:1), microwave, 8 min; (iii) 2 M NaOH/LiOH in dioxane, 3 6 h; (iv) 4 M HCl in dioxane, 4 h; (v) DIC/HOBT, DCM, 5 h; (vi) 4-amino heterocyclic ester, DIC/HOBT, DCM, 5 h; (vii) methyl 4-amino-1-methyl-1H-pyrrole-2-carboxylate, DIC/HOBT, DCM, 5 h.

MB-231), and a significant selectivity between the tumor cell demonstrate that conjugation of a single MPB fragment to a μ fi lines and the non-tumor cell line WI38 (i.e., IC50 = 0.47 M), PBD unit at a preferred distance and in a con guration that providing an up to 9 orders of magnitude difference. allows binding within the DNA minor groove confers potent From a structure−activity relationship (SAR) standpoint, a cytotoxicity in a number of tumor cell lines. The lower single MPB unit was found to enhance cytotoxicity, with PBD− cytotoxicity of these compounds in the non-tumor cell line − − − Py MPB (7g) and PBD Im MPB (7h) having lower IC50 WI38 (by at least 3 orders of magnitude) suggests that values compared to conjugates containing two (PBD−MPB− molecules of this type may have therapeutic potential. MPB, 7b) or no (PBD−Py−Py, 4) MPB units. However, the In Vivo Human Tumor Xenograft Studies. After relative position of the MPB unit in the conjugate was also observation of significant and selective in vitro cytotoxicity in found to be important for optimal cytotoxicity, as PBD−MPB− a number of cell lines, both 7g and 7h were progressed to an Py (7c) and PBD−MPB−Im (7d) had significantly reduced initial small-scale MTD study in Swiss Webster mice, using an activity compared to 7g and 7h, respectively. These results intraperitoneal (ip) dosing regimen. The compounds were

2922 dx.doi.org/10.1021/jm301882a | J. Med. Chem. 2013, 56, 2911−2935 Journal of Medicinal Chemistry Article

a Scheme 4. Synthesis of the PBD−MPB Conjugates 7a−7h

a Reagents and conditions: (i) 2 M NaOH in dioxane, rt, 5 h; (ii) EDCI, DMAP, DMF, rt, 4 h; (iii) (PPh3)4Pd, pyrrolidine, PPh3, rt, 30 min. generally well-tolerated without any signs of toxicity up to a In vivo studies were performed in both breast (MDA-MB- dose level of 400 μg/kg/day. A small but significant weight loss 231) and pancreatic (MIA PaCa-2) cancer xenografts grown in was observed for 7g at this dose level, so the experiment was MF1 nude mice. The results of the MDA-MB-231 breast terminated after 10 days (Figure S11, Supporting Information). xenograft study are shown in Figure 7A. The drug-treated μ groups (six mice per group) were given an iv dose of either 250 A repeat MTD experiment using only 7h at 350 g/kg/day μ μ fi confirmed a complete lack of toxicity at this dose level (Figure g/kg/day or 300 g/kg/day for ve consecutive days followed by two drug-free days, and this cycle was repeated for four S12, Supporting Information). As 7h appeared to have a fi fi weeks. 7h produced signi cant dose-dependent antitumor marginally better toxicity pro le in the MTD study, it was activity compared to control mice (vehicle alone) at both decided to carry out in vivo xenograft studies with this dose levels with no signs of toxicity. Interestingly, at the 300 molecule. Intravenous (iv) administration was used as the μg/kg dose level, the tumor did not regrow for up to three preferred route of drug delivery for the xenograft studies on the weeks after administration of the last iv dose. For the MIA basis of literature reports that this is the optimal delivery PaCa-2 xenograft study, an iv dose of 300 μg/kg/day was again method for the PBD dimer SJG-136.45 administered for five consecutive days followed by two drug-

2923 dx.doi.org/10.1021/jm301882a | J. Med. Chem. 2013, 56, 2911−2935 Journal of Medicinal Chemistry Article

Table 3. Differences in Free Energy (kcal/mol) upon Binding of Distamycin (2), Py−MPB−Py (5a), and Py−MPB−Im (5b) in a Forward and Reverse Orientations within the Minor Groove of the DNA Duplexes Shown in the First Column

DNA sequence 2 forward 2 reverse 5a forward 5a reverse 5b forward 5b reverse 5′-GCTATTAGGC-3′ −48.23 −44.56 −42.14 −42.54 −44.39 −42.77 3′-CGATAATCCG-5′

5′-GCCCGGCGGC-3′ −42.76 −43.14 −46.68 −44.30 −47.47 −44.74 3′-CGGGCCGCCG-5′ aForward = amine tail toward left (5′-end) of duplex as depicted in table above. Reverse = amine tail toward right of duplex (3′-end). free days, and the cycle repeated for three weeks. At this dose drug-treated control animals using standard immunohistochem- level, 7h also produced significant antitumor activity compared ical techniques. 7h was shown to down-regulate BCL2 to the control mice (vehicle only) with no signs of toxicity expression in treated MIA PaCa-2 tumor tissue compared to (Figure 7B). the control in a statistically significant manner on the basis of Preliminary Studies on NF-κB Inhibition. There is quantification of staining (Figure 8C). Although these growing evidence that PBD monomers (which cannot cross- preliminary results suggest NF-κB inhibition as a potential link DNA) exert their pharmacological effect through tran- mechanism of action for 7h, and a possible reason for its − scription factor inhibition.26,27,46 51 The significant cytotoxicity unusually high potency, further studies are under way to of 7h in a panel of tumor cell lines, and the in vivo antitumor confirm this. activity observed in the MDA-MB-231 and MIA PaCa-2 mouse Molecular Modeling Studies. Molecular models were xenografts, prompted us to carry out preliminary investigations constructed to gain insight into the GC sequence preference of on the possible effect of 7h on different transcription factors. MPB building blocks, and the potential NF-κB inhibitory Therefore, 7h was evaluated initially in a commercial activity of 7h. The study initially focused on the noncovalently Transcription Factor Activation Profiling Array assay (Signosis) interacting biaryl polyamide MPB building blocks, and using the HeLa cell line. In this assay the activities of 48 molecular dynamics simulations were performed with Py− transcription factors could be monitored simultaneously using a MPB−Py (5a), Py−MPB−Im (5b), and distamycin (2)in collection of biotin-labeled DNA probes based on the conjunction with the hairpin DNA constructs. In addition, consensus sequences of individual transcription factor DNA nonhairpin DNA duplexes were constructed containing 5′- binding sites. The top five transcription factors whose activities GCCCGGCGGC-3′ and 5′-GCTATTAGGC-3′ sequences, as were at least 30% down-regulated by 7h at a concentration of these were the most preferred for the GC-targeting 5a and 5b 10 nM for 4 h are shown in Figure S12 (Supporting and the AT-targeting 2, respectively. Dynamics simulations Information). After 4 h of treatment at 10 nM, 7h reduced were performed over 20 ns in implicit solvent, and the free the activity of NF-κB by almost 50%. It is noteworthy that four energy of binding was calculated using the AMBER MM_PBSA of the other transcription factors affected (i.e., NFAT, EGR, approach. In this method, internal energies and nonbonded SMAD, and OCT-4; Figure S12) are GC-rich and have interactions (long-range cutoff) derived from molecular consensus sequences similar to that of NF-κB. This finding was mechanics were combined with the generalized Born (GB) consistent with the previous report of a DC-81/indole continuum solvent method. Structures for the free energy conjugate inducing apoptosis through inhibition of the NF-κB calculations consisted of approximately 200 models derived pathway.48 from the molecular dynamics simulation taken at equal On the basis of the hypothesis that 7h may down-regulate intervals. The procedure was repeated for each ligand when the expression of NF-κB-dependent genes (e.g., IκB, BCL2, oriented in both 5′ to 3′ and 3′ to 5′ directions in the minor κ BCLXL) by binding to the cognate DNA sequence of NF- B groove. Prior to each simulation, the free ligand was positioned and thereby blocking interaction of the transcription factor just outside the minor groove and energy minimized with atom protein and inhibiting transcription of a number of genes restraints applied to the DNA structure which were gradually (Figure 8A), it was decided to explore this possibility in CLL reduced in stages to zero. Over the course of the dynamics − cells52 54 in which NF-κB signaling is known to be active and simulation, the ligands became gradually embedded in the closely correlates with the initiation and progression of minor groove, with 5a showing superior points of contact malignancy. Using levels of phosphorylated IκB and p65 as within the GC-rich sequences. During the simulation, both 5a surrogates for NF-κB activity in comparison to the actin protein and 5b traveled up and down the minor groove between C2 as a control, Western blotting indicated that, after 24 h of and G8, forming, breaking, and re-forming up to three incubation, as anticipated 7h caused a significant suppression of hydrogen bonds of 3.2 Å distance between downward-facing κ α phosphorylated I B at concentrations down to 0.1 nM, with carbonyls and the exocyclic NH2 groups of guanines within the only a marginal effect on phosphorylated p65 (Figure 8B). GC-rich sequence. These hydrogen bonds appeared to play a Finally, NF-κBis“antiapoptotic” and capable of enhancing crucial role in the GC-specificity of these molecules. This cell survival when tumor cells are under stress to progress to contrasted with the AT-containing sequence 5′-GCTAT- apoptosis. The mechanisms by which NF-κB inhibits apoptosis TAGGC-3′, where 5a and 5b traveled up and down the generally involve BCL2 family members that act as inhibitors of minor groove in a similar manner but failed to form hydrogen programmed cell death.55,56 Therefore, immunohistochemical bonds. Conversely, 2 displayed superior interactions with the staining was performed on sections taken from biopsy material AT-rich sequences (Figure S6, Supporting Information). The from the xenografted MIA PaCa-2 tumors of 7h-treated mice to changes in free energy (Table 3) show that 2 has a higher assess this possibility. The tumor sections were stained for binding affinity (in both orientations) compared to both 5a and BCL2 expression and compared to tumor sections from non- 5b for the AT-rich sequence. Conversely, for the GC-rich

2924 dx.doi.org/10.1021/jm301882a | J. Med. Chem. 2013, 56, 2911−2935 Journal of Medicinal Chemistry Article sequence, both 5a and 5b are slightly favored in both component and DNA bases in the N-G/C-G/C region. The orientations compared to 2. This result is consistent with modeling studies confirmed that, in the case of the sequence observations from the FID, FRET, and footprinting studies, 5′AAGAAGGCAA-3′, which contains an idealized 7h binding thus supporting a GC sequence preference for MPB-containing sequence (underlined), hydrogen bonds are formed between biaryl polyamides. the NH linker of the Im−MPB unit and the carbonyl of T16, Molecular dynamics simulations were performed over 2 ns and between the ring nitrogen of the imidazole and G6. As a for both 7a and 4 in their noncovalently and covalently bound result, 7h spans seven base pairs, an arrangement further states for both of the hairpin sequences used in the HPLC−MS supported by simulations of the noncovalently bound molecule study to understand the faster reaction and GC sequence in the same sequence in which 7h remained over the idealized preference of 7a. For the noncovalently bound ligands, the free PBD binding sequence for the full 20 ns of the simulation. energy of binding as measured according to the differences in Interestingly, hydrogen bonds were formed from the imidazole free energy of the complex compared to the DNA and ligand ring and NH of the Im−MPB−Py group intermittently, with an alone [ΔG = G(complex) − G(DNA) − G(ligand)] was extra hydrogen bond occurring between the carbonyl of the consistent with the experimental observations. The calculated terminal ester and the exocyclic amine of G13, suggesting that free energies for 7a noncovalently bound to the GC- and AT- 7h may form different hydrogen bonds when locating in its rich hairpins were −41.56 and −40.49 kcal/mol, respectively, idealized sequence compared to those which hold it in place indicating the relative preference of 7a for the GC-rich when it is covalently bound (Figure S9, Supporting sequence. On the other hand, for the AT-preferring 4, the Information). free energy of the 4/AT-rich hairpin DNA complex was lower Finally, molecular modeling studies were carried out to (−50.68 kcal/mol) compared to that of the 4/GC-rich hairpin rationalize the possible binding of 7h to NF-κB-recognition complex (−46.83 kcal/mol), again consistent with the sequences (i.e., GGGRNNYYCC, where R corresponds to a experimental observations. Interestingly, during the simulations, purine, Y represents a pyrimidine, and N can be any base). A the ligands moved backward and forward within the minor series of molecular dynamics simulations in implicit solvent groove, possibly reflecting, in part, flexing of the DNA during over 20 ns were conducted for 7h both covalently bound and the simulation, with the ligands reaccommodating accordingly. noncovalently bound to two NF-κB consensus sequences (5′- For models of covalently bound ligands, H-bonds were seen GGGACAGCCC-3′ and 5′-GGGGGTCGCC-3′). During to form and break transiently throughout the simulations. The simulations with 5′-GGGACAGCCC-3′, 7h located to its differences in hydrogen bond formation and intensity may ideal binding sequence at approximately 1 ns and remained account, in part, for the experimental observations. For restrained for the entire simulation, held by favorable hydrogen- example, after covalent attachment and subsequent molecular bonding interactions between the carbonyl of its terminal ester rearrangement of bonds within the PBD B-ring, a potential H- and G7 and between the N of its imidazole ring and the ′ bond common to all structures was observed to form between exocyclic NH2 of G15. Simulations with 5 -GGGGGTCGCC- the 10-NH of the B-ring and the oxygen (O2) of T13. In the 3′ showed similar interactions, with the phenyl group locating case of 7a, torsions within the linker region allowed the NH of directly over A15 for the duration of the simulation and the linker to form a H-bond with the oxygen of the A5 ribose hydrogen bonds forming between the ring nitrogen of the sugar in both sequences (Figure S7, Supporting Information), imidazole and the exocyclic NH2 of G5, between the carbonyl with the CO group H-bonding to the NH2 functionality of G14 of the terminal ester of 7h and the exocyclic NH2 of G13, and for a limited duration. The O-methyl of the free end of 7a also between N10 of the PBD and the reacting guanine (Figure 9A). formed a transient H-bond to the NH2 of G1 and G3 (Figure A further study involving 7h and the crystal structure of the S7). These bonds may contribute to the greater affinity of 7a p50−p65 heterodimer of the NF-κB protein docked in its for the GC-rich compared to the AT-rich sequence. For 4 in consensus sequence (PDB ID 1VKX) revealed that the the AT-rich sequence, a transient H-bond was observed to form imidazole ring of 7h may interact with lysine-145 of the NF- between the NH of the linker and the A4 N atom, whereas in κB protein (p50, highlighted in yellow), thus sterically the GC-rich sequence a transient H-bond formed between the hindering approach of the protein to its DNA binding site O of the free end of 4 and the NH2 of G16 (Supporting (Figure 9B). This observation provides a future opportunity to Information). In summary, the differences in the electrostatic explore the SAR of analogues of 7h modified at the imidazole environment of the floor of the minor groove in relation to the position. ligands, the van der Waals forces arising from the closeness of fit of the ligands, and the hydrogen-bonding interactions all ■ CONCLUSIONS appear to be factors contributing to the different binding This is the first report of a heteroaryl building block (i.e., the affinities. Although after initial minimization prior to the MPB unit) that can drive minor-groove noncovalently binding simulation 7a appeared not to follow the minor groove shape as ligands toward GC-rich sequences which, under normal well as 4, it was later observed to adjust its conformation in the circumstances, are difficult to target. The MPB biaryl motif linker region to become accommodated in the groove as either alone or conjugated to a covalently binding DNA- illustrated by the space-filling models shown in Figure S10 interactive PBD molecule has a strong preference for GC-rich (Supporting Information). sequences as demonstrated by the results of FID, HPLC−MS, Molecular modeling studies were also carried out to FRET-based, and DNA footprinting methodologies. Molecular rationalize the preference of 7h for the sequence N-G-N-W- modeling experiments support these observations, suggesting N-G/C-G/C as observed in the footprinting studies. In this that the high affinity of MPB-based molecules for GC-rich case, the N-G-N triplet is the PBD binding site and the sequences may be due to a combination of overall shape and methylene chain between the PBD and MPB is positioned over the generation of additional hydrogen bonds for some the strongly van der Waal interacting “W” region, with a combinations of ligands and sequences. Two of the PBD− number of hydrogen bonds formed between the MPB−Py MPB conjugates, 7g and 7h,havesignificant in vitro

2925 dx.doi.org/10.1021/jm301882a | J. Med. Chem. 2013, 56, 2911−2935 Journal of Medicinal Chemistry Article

13 δ cytotoxicity in certain tumor cell lines (i.e., 65 fM in the case of 2.26 (6H, s), 1.69 (2H, m). C NMR (CDCl3, 100 MHz): 161.7, 7g in MDA-MB-231), and 7h has promising antitumor activity 129.6, 124.2, 115.3, 102.6, 59.2, 45.54, 39.46, 36.23, 25.84. EIMS (m/ + in immune-compromised mice bearing MDA-MB-231 or MIA z): 225.15 (MH ). PaCa-2 xenografts. 4-Bromo-N-(5-((3-(dimethylamino)propyl)carbamoyl)-1-methyl- 1H-pyrrol-3-yl)-1-methyl-1H-pyrrole-2-carboxamide (13a). HOBt The results of preliminary experiments based on Western κ (1.776 g, 13.14 mmol, 2 equiv) and DIC (1.77 mL, 11.50 mol, 1.75 blotting in NF- B-dependent CLL cells, a commercially equiv) were added to a solution of 4-bromo-1-methyl-1H-pyrrole-2- available transcription factor activation kit, and histological carboxylic acid (1.3 g, 6.43 mmol) in dry DCM, and the was solution studies from the xenograft experiments support the hypothesis maintained at room temperature for 1 h. 12 (1.44 g, 6.8 mmol, 1 that 7h may interact at the NF-κB DNA consensus recognition equiv) was then added and the reaction mixture stirred at room site. This proposed mechanism of action is also supported by temperature for 16 h, after which reaction was complete. The reaction molecular modeling studies which suggest that part of the mixture was then passed through an SCX-2 cartridge (sorbent mass = 10 × product mass) and was then washed through with DCM (3 × 50 structure of 7h (the imidazole ring) protrudes out of the minor × × groove, sterically interacting with NF-κB. Further studies are mL), DMF (3 50 mL), and MeOH (2 50 mL). The product was under way to confirm this mechanism of action. released from the cartridge by washing through with 2 M NH3 in MeOH (30 mL) and was concentrated in vacuo to obtain 13a as a light brown solid (2.24 g, 94.1%). 1H NMR (DMSO, 400 MHz): δ ■ EXPERIMENTAL DETAILS 10.03 (1H, s), 8.22 (1H, t, J=5.6 Hz), 7.29 (1H, dd, J=3.6, 1.6 Hz), Synthesis. Type 1 Biaryl Polyamides. 2-(Trichloroacetyl)-1- 7.25(1H, dd, J=3.6, 1.6 Hz), 7.11 (1H, 1H, dd, J=3.6, 1.6 Hz), methylpyrrole (9). 6.91(1H, d, J=1.6 Hz), 3.97 (3H, s), 3.91 (3H, s), 3.31 (2H, q, J=5.6 A solution of N-methylpyrrole (113.06 g, 1.39 13 mol, 1.0 equiv) in dry ether (350 mL) was added dropwise over a Hz), 2.38 (2H, t, J=6.0 Hz), 2.27 (6H, s), 1.73 (2H, m). C NMR δ period of 70 min to a stirred solution of trichloroacetyl chloride (254 (DMSO, 100 MHz): 161.1, 156.7, 142.8, 127.3, 126.2, 123.1, 122.2, 117.9, 113.8, 111.4, 103.8, 56.9, 45.1, 36.9, 36.3, 35.7 27.1. EIMS (m/ g, 1.39 mol, 1.0 equiv) in dry ether (350 mL) in a 2 L three-necked + flask. HCl gas produced in the reaction was removed by flushing with z): 411.42 (M + H ). nitrogen. The reaction mixture was allowed to stir for 1.5 h until TLC N-(3-(Dimethylamino)propyl)-1-methyl-4-(1-methyl-4-(4-nitro- and LC−MS indicated that reaction was complete. The mixture was phenyl)-1H-pyrrole-2-carboxamido)-1H-pyrrole-2-carboxamide (14a). A catalytic amount of tetrakis(triphenylphosphine)palladium then quenched with K2CO3 solution (1 M, 250 mL) and extracted with ethyl acetate (3 × 200 mL), and the combined organic layers [Pd(PPh3)4] (0.35 g, 0.30 mmol, 0.1 equiv) was added to a mixture of were concentrated in vacuo. The crystalline residue was washed with 13a (0.12 g, 0.29 mmol, 1 equiv), 4-nitrophenylboronic acid (0.067 g, n-hexane and dried under vacuum to afford 9 (281.18 g, 79.5%) as a 1.2 equiv), and K2CO3 (0.125 g, 3 equiv) in EtOH/toluene/water ν −1 (9:3:1) in a 5 mL microwave vial containing a magnetic stirrer, white solid. IR ( max,cm ): 3299, 3121, 3008, 2954, 1789, 1674, 1 57 flushing with nitrogen during each addition. The reaction mixture was 1521, 1406, 1206, 1100, 980, 881, 757. H NMR (CDCl3, 400 MHz): δ 7.42 (1H, dd, J = 4.4, 1.6 Hz), 6.97 (1H, t, J = 1.6 Hz), 6.22 sealed under an inert N2 atmosphere and heated with microwave 13 radiation in an EMRYS Optimizer microwave station (Personal (1H, dd, J = 4.4, 2.4 Hz), 3.97 (3H, s). C NMR (CDCl3, 400 MHz): ° − δ 133.6, 124.0, 122.4, 108.9, 38.5. Chemistry) at 100 C for 8 min, after which LC MS analysis 4-Nitro-2-(trichloroacetyl)-1-methylpyrrole (10). Fuming nitric indicated completion of reaction. The cooled reaction mixture was then passed through an Isolute SCX-2 cartridge which was washed acid (37.5 mL) was added dropwise over 1 h to a mechanically × × × stirred solution of 9 (100 g, 0.47 mol) in acetic anhydride (475 mL) with DCM (3 20 mL), DMF (3 20 mL), and MeOH (2 20 which was maintained at −5 °C using an acetone/dry ice bath mL). The product was then released from the cartridge by washing (caution: risk of sudden exotherm if temperature exceeds 10 °C). After with 2 M NH3 in MeOH (10 mL), followed by concentration in vacuo ff 1 addition was complete, the temperature was raised gradually to 10 °C to a ord 14a as a light brown solid (0.12 g, 91%). H NMR (DMSO, δ over a period of 3 h with continuous stirring. Next, the mixture was 400 MHz): 10.03 (1H, s), 8.13 (1H, t, J = 5.6 Hz, NH), 8.24 (2H, d, cooled to −30 °C using an acetone/dry ice bath and then diluted with J = 8.8 Hz), 7.78 (2H, d, J = 8.8 Hz), 7.75 (1H, d, J = 2.0 Hz), 7.44 2-propanol (500 mL). The resulting precipitate was collected using a (1H, d, J = 1.6 Hz), 7.20 (1H, d, J=1.6 Hz), 6.90 (1H, d, J = 2.0 Hz), Buchner flask and dried in vacuo to afford 10 as a pale yellow solid (96 3.93 (3H, s), 3.82 (3H, s), 3.20 (2H, m), 2.32 (2H, t, J = 6.8 Hz), 2.20 ν −1 (6H, s), 1.63 (2H, m). EIMS (m/z): 453.40 (MH+). g, 80%). IR ( max,cm ): 3055, 2166, 1620, 1466, 1310, 1180, 1083, 1 57 δ N-(3-(Dimethylamino)propyl)-1-methyl-4-(1-methyl-4-(4-(1- 847, 763. H NMR (CDCl3, 400 MHz): 7.95 (1H, d, J = 1.6 Hz), 7.71 (1H, d, J = 1.6 Hz), 4.06 (3H, s). 13C NMR (CDCl 400 MHz): methyl-1H-pyrrole-2-carboxamido)phenyl)-1H-pyrrole-2-carboxa- 3, mido)-1H-pyrrole-2-carboxamide (Py−MPB−Py, 5a). δ 173.7, 135.3, 130.25, 121.47, 117.46, 94.83, 39.80. EIMS (m/z): 1-Methyl-1H- 272.93 (MH+). pyrrole-2-carbonyl chloride (50 mg, 0.345 mmol, 1.2 equiv) was added to a stirred solution of 14 (100 mg, 0.23 mmol, 1.0 equiv) in dry DMF N-(3-(Dimethylamino)propyl)-1-methyl-4-nitro-1H-pyrrole-2-car- − boxamide (11). (3-(Dimethylamino)propyl)amine (6.3 mL, 0.038 (5 mL) using a micropipet and the mixture stirred for 3 h until LC mol) was added to a mechanically stirred solution of 10 (10.36 g, MS indicated completion of reaction. The reaction mixture was then × 0.038 mol, 1.0 equiv) in THF. The mixture was allowed to stir for 1 h passed through an Isolute SCX-2 cartridge followed by DCM (3 20 × × fi and was then diluted with n-hexane to produce a cream-colored mL), DMF (3 20 mL), and MeOH (2 20 mL). The nal product crystalline precipitate that was collected by vacuum filtration and dried was released from the cartridge using 2 M NH3 in MeOH (10 mL) ff ν −1 and the eluent concentrated in vacuo to obtain crude 5a as a light in vacuo to a ord 11 as a white solid (9.6 g, 99%). IR ( max,cm ): 1 δ brown solid (90 mg, 72%). Purification by reversed-phase preparative 2956, 2245, 1627, 851, 774. H NMR (CDCl3, 400 MHz): 7.54 (1H, d, J=1.6 Hz), 6.92 (1H, d, J=2.0 Hz), 4.01 (3H, s), 3.50 (2H, m), HPLC using a C18 column and acetonitrile/water/0.1% formic acid 2.52 (2H, m), 1.73 (2H, m). EIMS (m/z): 255.15 (MH+). (65%:45% gradient) as the mobile phase afforded 5a, which was fl ff ν freeze-dried to obtain a u y white solid (64 mg, 52.5%). IR ( max, 4-Amino-N-(3-(dimethylamino)propyl)-1-methyl-1H-pyrrole-2- − carboxamide (12). 11 (3 g, 11.79 mmol) was dissolved in a minimum cm 1): 3254, 2047, 1629, 1539, 1461, 1432, 1324, 1251, 1063, 873, volume of ethyl acetate, and a catalytic amount of 10% Pd/C (0.3 g) 784, 764, 733. 1H NMR (DMSO, 400 MHz): δ 9.87 (1H, s), 9.72 was added as a slurry in ethyl acetate (150 mL). The reaction mixture (1H, s), 8.09 (1H, t, J=5.6 Hz), 7.71 (2H, d, J=8.8 Hz), 7.49 (2H, d, was then shaken for 3 h at 45 psi in a Parr hydrogenator, after which J=8.8 Hz), 7.40 (1H, d, J=2.0 Hz), 7.27 (1H, d, J=2.0), 7.19 (1H, reduction was complete. The suspension was filtered using a Buchner d, J=2.0), 7.03 (1H, dd, J=4.0, 1.6 Hz), 7.00 (1H, t, J = 2.0 Hz), 6.84 flask (caution: possible combustion) and the filtrate concentrated to (1H, d, J=2.0 Hz), 6.10 (1H, m), 3.91 (3H, s), 3.89 (3H, s), 3.82 ff 1 a ord 12 as a colorless oil (2.4 g, 90.7%). H NMR (CDCl3, 400 (3H, s), 3.21 (2H, q, J=6.8), 2.37 (2H, t, J = 7.2 Hz), 2.24 (6H, s), MHz): δ 7.55 (1H, s), 6.24 (1H, d, J = 2.0 Hz), 6.00 (1H, d, J=2.0 1.66 (2H, m). 13C NMR (100 MHz, DMSO): δ 163.8, 161.2, 159.7, Hz), 3.82 (3H, s), 3.42 (2H, q, J=6.0 Hz), 2.42 (2H, t, J=6.0 Hz), 158.3, 149.8, 137.0, 136.1, 135.1, 134.1, 129.6, 128.7, 124.7, 124.3,

2926 dx.doi.org/10.1021/jm301882a | J. Med. Chem. 2013, 56, 2911−2935 Journal of Medicinal Chemistry Article

ν −1 120.4, 117.7, 113.5, 109.4, 106.7, 103.9, 56.7, 44.7, 39.7, 36.8, 36.3, 98.4%). IR ( max,cm ): 3315, 1627, 1508, 1435, 1397, 1281, 1176, + 1 δ 35.9, 26.7. HRMS (EI, m/z): calcd for C29H35N7O3 (MH ), 530.2874; 1118, 720, 694. H NMR (DMSO, 400 MHz): 8.07 (1H, t, J=5.6 found, 530.2855. Hz), 7.17 (2H, dd J=8.8, 1.6 Hz), 7.11 (1H, d, J=1.6 Hz), 6.97 (1H, N-(4-(5-((5-((3-(Dimethylamino)propyl)carbamoyl)-1-methyl-1H- d, J=2.0 Hz), 6.55 (2H, d, J=8.0, 2.0), 3.83 (3H, s), 3.22 (2H, q, J= pyrrol-3-yl)carbamoyl)-1-methyl-1H-pyrrol-3-yl)phenyl)-1-methyl- 6.8), 2.56 (2H, m), 2.37 (6H, s), 1.72 (2H, m). EIMS (m/z): 301.35 1H-imidazole-2-carboxamide (Py−MPB−Im, 5b). HATU (136.8 mg, (MH+). 0.36 mmol, 2 equiv) and DIPEA (70 μL, 0.54 mmol, 3.0 equiv) were 4-(4-(3-(2-Methylthiazol-4-yl)benzamido)phenyl)-N-(3- added to a magnetically stirred solution of 1-methyl-1H-imidazole-2- (dimethylamino)propyl)-1-methyl-1H-pyrrole-2-carboxamide (6a). carboxylic acid (34 mg, 0.27 mmol, 1.5 equiv) at 0 °C. After 30 min A solution of 19 (50 mg, 0.22 mmol, 1.2 equiv) in anhydrous DMF 76.25 mg of 14 (0.18 mmol, 1 equiv) was added, and the reaction was added to a three-necked flask under a nitrogen atmosphere and mixture was stirred for 4 h at rt, at which point LC−MS indicated was cooled to −5 °C in an ice/acetone bath. HOAt (60 mg, 0.44 completion of reaction. The reaction mixture was passed through an mmol, 2 equiv), HATU (108 mg, 0.286 mmol, 1.3 equiv), and DIPEA Isolute SCX-2 cartridge and was then washed through with DCM (3 × (85.14, 0.66 mmol, 3 equiv) were added with continuous stirring. After 20 mL), DMF (3 × 20 mL), and MeOH (2 × 20 mL). The product 15 min of further stirring, 19 (58 mg, 0.19 mmol, 1.0 equiv) in was released from the cartridge using a mixture of 2 M NH3 in MeOH anhydrous DMF was added to the reaction vial using a cannula. The (10 mL), which was concentrated in vacuo to afford crude 5b as a light reaction mixture was allowed to stir for 3 h until TLC and LC−MS brown solid. This was purified with reversed-phase preparative HPLC indicated that reaction was complete. The reaction mixture was then using a C18 column and acetonitrile/water/0.1% formic acid passed through an Isolute SCX-2 cartridge and was washed through (68%:32% gradient) as the mobile phase to afford pure 5b, which with DCM (3 × 20 mL), DMF (3 × 20 mL), and MeOH (2 × 20 fl ff was freeze-dried to obtain a u y white solid (54 mg, 43.9%). IR mL). The product was released from the cartridge using 2 M NH3 in ν −1 (FTIR, max,cm ): 3281, 2358, 2186, 1589, 1520, 1463, 1430, 1360, MeOH (10 mL) and was concentrated in vacuo to obtain a light 1278, 1104, 933, 816, 659. 1H NMR (DMSO, 400 MHz): δ 10.31 brown solid. This was purified using a silica gel column with a mixture (1H, s), 9.75 (1H, s), 8.03 (1H, t, J=6), 7.71 (2H, d, J=8.8), 7.53 of DCM, MeOH, and 2 M NH3 (9:0.5:0.5) as the eluent. The eluent (1H, s), 7.52 (1H, s, 7.48 (2H, d, J=8.8), 7.42 (1H, s), 7.04 (2H, m), was evaporated in vacuo to afford 6a as a light cream solid (75 mg, ν −1 6.1 (1H, t, J=4), 3.94 (3H, s), 3.92 (3H, s), 3.89 (3H, s), 3.27 (2H, t, 79%). IR ( max,cm ): 3291, 2759, 1651, 1629, 1590, 1524, 1460, J=6.4), 2.26 (2H, t, J=7.2), 2.15 (6H, s), 1.64 (2H, m). 13C NMR 1325, 1287, 1000, 925, 831, 806, 793, 720. 1H NMR (DMSO, 400 (100 MHz, DMSO): δ 161.2, 159.7, 158.5, 156.2, 149.8, 137.0, 136.1, MHz): δ 10.33 (1H, s), 8.49 (1H, d, J=1.54), 8.24 (1H, s), 8.13 (1H, 135.1, 134.1, 129.6, 128.7, 127.6, 125.3, 124.4, 122.1, 120.4, 113.5, dd, J=8.0, 1.2 Hz), 8.12 (1H, t, J=5.6 Hz), 8.06 (1H, s), 7.90 (1H, d, 110.7, 106.7, 56.8, 45.1, 40.2, 36.9, 36.5, 36.3, 34.8, 27.1. HRMS (EI, J=8.0 Hz), 7.78 (2H, d, J=8.4 Hz), 7.61 (1H, t, J=7.6 Hz), 7.49 + m/z): calcd for C28H34N8O3 (MH ), 531.2825; found, 531.2819. (2H, d, J=8.8 Hz), 7.35 (1H, d, J=1.6 Hz), 7.13 (1H, d, J=1.6 Hz), Type 2 Biaryl Polyamides. 1-(4-Bromo-1-methyl-1H-pyrrol-2-yl)- 7.06 (1H, d, J=1.6 Hz), 6.80 (1H, d, J = 2.0 Hz), 3.82 (3H, s), 3.21 2,2,2-trichloroethanone (16). N-Bromosuccinamide (2.36 g, 13.24 (2H, q, J=6.8), 2.76 (3H, s), 2.37 (2H, t, J=8.0), 2.23 (6H, s), 1.65 mmol, 1.0 equiv) was added to a stirred solution of 9 (3 g, 13.24 (2H, m). 13C NMR (100 MHz, DMSO): δ 165.8, 163.8, 153.1, 145.6, mmol, 1.0 equiv) in anhydrous THF (35 mL) at −10 °C. The reaction 143.2, 130.4, 128.8, 126.9, 126.7, 126.4, 124.3, 124.2, 120.8, 113.3, mixture was maintained at −10 °C for 2 h and then allowed to warm 108.9, 103.2, 92.8, 56.5, 56.4, 44.6, 37.3, 26.7, 26.4. HRMS (EI, m/z): + + to room temperature over approximately 4 h. Excess THF was calcd for C28H28N5O2S (MH ), 502.2271; found, 502.2255 (MH ). evaporated in vacuo, and the resulting solid was redissolved in a PBD−MPB Conjugates. Methyl 4-Bromo-1-methyl-1H-pyrrole-2- mixture of EtOAc/n-hexane (1:9). This solution was filtered through a carboxylate (20). To a stirred solution of 16 (5.5 g, 18.01 mmol, 1 plug of silica and the filtrate evaporated in vacuo to give a solid which equiv) in dry MeOH (40 mL) was added a solution of sodium was recrystallized from n-hexane to afford pure 16 (3.55 g, 88%). IR methoxide (5 mL) through a syringe. The sodium methoxide solution ν −1 ( max,cm ): 3148, 2956, 1669,1458, 1215, 1189, 1062, 923, 842, 823, was prepared from NaH (60%) in mineral oil (84 mg, 3.5 mmol, 0.2 1 δ 748, 678. H NMR (CDCl3, 400 MHz): 7.46 (1H, d, J = 2.0 Hz), equiv), which was previously washed with hexane. The solution was 13 δ fl 6.95 (1H, d, J = 1.6 Hz) 3.95 (3H, s). C NMR (CDCl3, 100 MHz): heated at re ux for 30 min, after which TLC indicated that all starting + 172.4, 132.8, 124.6, 132.2, 96.1, 38.7. EIMS (m/z): 306.86 (MH ) material had been consumed. A few drops of concentrated H2SO4 4-Bromo-N-(3-(dimethylamino)propyl)-1-methyl-1H-pyrrole-2- were then added to neutralize the base and bring the pH to 2.0. The carboxamide (17). (3-(Dimethylamino)propyl)amine (86 mg, 0.84 excess MeOH was evaporated in vacuo and the resulting oil mmol, 1 equiv) was added to a stirred solution of 16 (257 mg, 0.84 redissolved in EtOAc (50 mL) and then washed with water (40 mmol, 1 equiv) in THF (20 mL). The reaction mixture was stirred for mL). The aqueous layer was extracted with EtOAc (3 × 40 mL), and fi fi 3 h and then puri ed using Isolute SCX-2 cartridges using the method the organic phases were combined, dried (MgSO4), ltered, and described for compound 14a. The resulting product was dried in vacuo concentrated in vacuo to afford 20 as a pale white solid (3.80 g, 97%). ff 1 ν −1 to a ord 17 as a pale yellow solid (230 mg, 94%). H NMR (CDCl3, IR ( max,cm ): 3138, 2948, 1692, 1472, 1334, 1245, 1115, 1082, 921, δ 1 δ 400 MHz): 7.61 (1H, s), 6.46 (1H, d, J=2.0 Hz), 6.00 (1H, d, J= 823, 753. H NMR (400 MHz, CDCl3): 6.89 (d, 1H, J = 2.0 Hz), 2.0 Hz), 3.84 (3H, s), 3.36 (2H, m), 2.45 (2H, t, J=7.2 Hz), 1.72 6.76 (d, 1H, J = 2.0 Hz), 3.89 (s, 3H), 3.81 (s, 3H). 13C NMR (100 13 δ δ (2H, m). C NMR (CDCl3, 100 MHz): 173.2, 131.6, 114.3, 103.2, MHz, CDCl3): 160.8, 128.7, 122.9, 119.2, 95.1, 51.2, 36.9. EIMS (m/ 57.8, 45.1, 39.5, 37.1, 25.7. EIMS (m/z): 289.06 (MH+). z): 219.26 (MH+). 4-(4-Aminophenyl)-N-(3-(dimethylamino)propyl)-1-methyl-1H- Methyl 4-(4-((tert-Butoxycarbonyl)amino)phenyl)-1-methyl-1H- pyrrole-2-carboxamide (18). A catalytic amount of Pd(PPh3)4 (79 pyrrole-2-carboxylate (21). A catalytic amount of Pd(PPh3)4 (0.47g, mg, 0.069 mmol, 0.1 equiv) was added to a mixture of 17 (200 mg, 0.413 mmol, 0.06 equiv) was added to a mixture 20 (1.5 g, 6.88 mmol, 0.69 mmol, 1 equiv), 4-aminophenylboronic acid (112 mg, 0.82 mmol, 1 equiv), 4-((tert-butoxycarbonyl)amino)phenylboronic acid (1.57 g, 1.2 equiv), and K2CO3 (285 mg, 2.06 mmol, 3 equiv) in EtOH/ 6.88 mmol, 1.20 equiv), and K2CO3 (2.856 g, 3 equiv) in EtOH/ toluene/water (9:3:1, 5 mL) in a 10 mL microwave vial containing a toluene/water (9:3:1, 13.5 mL) in a 10−20 mL microwave vial magnetic stirrer, flushing with nitrogen during each addition. The vial containing a magnetic stirrer. The reaction vessel was flushed with was sealed under N2 and then heated with microwave radiation in an nitrogen during each addition. The vial was sealed in a N2 atmosphere EMRYS Optimizer microwave station (Personal Chemistry) at 100 °C and then heated with microwave radiation in an EMRYS Optimizer for 9 min, after which LC−MS indicated that reaction was complete. microwave station (Personal Chemistry) at 100 °C for 12 min, after The cooled reaction mixture was passed through an Isolute SCX-2 which LC−MS and TLC indicated that reaction was complete. The cartridge and was then washed through with DCM (3 × 50 mL), DMF cooled reaction mixture was diluted with water (50 mL) and extracted (3 × 50 mL), and MeOH (2 × 50 mL). The product was released with EtOAc (3 × 40 mL), and the organic phases were combined, from the cartridge using 2 M NH3 in MeOH (10 mL) and was dried (MgSO4), and concentrated in vacuo to give a colorless oil. This concentrated in vacuo to afford 18 as a light brown solid (205 mg, was purified by flash chromatography on silica gel using n-hexane/

2927 dx.doi.org/10.1021/jm301882a | J. Med. Chem. 2013, 56, 2911−2935 Journal of Medicinal Chemistry Article

EtOAc (9:1) as the eluent to afford 21 as a light yellow oil (2.2 g, added to a stirred solution of 22 (0.45 g, 1.42 mmol, 1.2 equiv) in ν −1 97%). IR ( max,cm ): 3353, 2975, 1696, 1521, 1441, 1366, 1264, DMF (8 mL), and stirring was continued for 30 min at rt, after which 1 δ 1235, 1209, 1058, 822, 799, 657. H NMR (400 MHz, CDCl3): 7.40 methyl 4-amino-1-methyl-1H-pyrrole-2-carboxylate (0.18 g, 1.18 (d, 2H, J = 8.8 Hz), 7.33 (d, 2 H, J = 8.8 Hz), 7.16 (d, 1H, J = 2.0 Hz,), mmol, 1.0 equiv) was added. The reaction mixture was stirred for a 7.02 (d, 1H, J = 2.0), 6.45 (br s, 1H), 3.95 (s, 3H), 3.83 (s, 3H), 1.52 further 6 h at rt, when TLC indicated that reaction was complete. The 13 δ (s, 9H). C NMR (100 MHz, CDCl3): 161.7, 152.8, 136.5, 129.5, reaction was quenched by pouring the mixture into ice/water (250 125.9, 125.6, 123.7, 123.0, 119.0, 114.6, 80.5, 51.1, 36.9, 28.4. EIMS mL), and the resulting mixture was extracted with ethyl acetate (3 × (m/z): 330.46 (MH+). 150 mL). The combined organic extracts were sequentially washed 4-(4-((tert-Butoxycarbonyl)amino)phenyl)-1-methyl-1H-pyrrole- with citric acid (100 mL), saturated aqueous NaHCO3 (100 mL), 2-carboxylic Acid (22). An aqueous solution of NaOH (0.5 M, 2.0 water (100 mL), and brine (100 mL) and then dried over MgSO4. The equiv) was added to a solution of 21 (1.0 g, 3.027 mmol) in 1,4- ethyl acetate was evaporated in vacuo and the resulting crude product dioxane (40 mL). The mixture was allowed to stir at room 25 (0.58 g, 90.6%) used directly for the next Boc-deprotection step temperature for 6 h, when TLC indicated that reaction was complete. without further purification. 25 (0.29 g, 0.63 mmol) was dissolved in Excess 1,4-dioxane was evaporated in vacuo and the residue diluted MeOH/4 M HCl in dioxane (50:50, 15 mL) and the mixture stirred with water (60 mL). The resulting solution was acidified with HCl (0.5 for 6 h at rt, when TLC indicated that reaction was complete. The M) to pH 2.0 and the product extracted with ethyl acetate (100 mL × solvent was evaporated in vacuo to afford crude 27 as a brown solid. 2). The combined organic phases were washed with brine, dried This was purified by flash chromatography on silica gel using n- ff (MgSO4), and concentrated in vacuo to a ord 22 as a light yellow oil. hexane/EtOAc (9:1) as the eluent. The combined relevant fractions This was purified using flash chromatography with silica gel and ethyl were evaporated in vacuo to afford 27 as a light brown solid (0.21 g, acetate/n-hexane (2:8) as the eluent. Concentration of the combined 95%). 1H NMR (DMSO, 400 MHz): δ 9.75 (1H, s), 7.94 (1H, s), relevant eluted fractions afforded 22 as a colorless oil (0.92 g, 96.8%). 7.71 (2H, d, J=8.8 Hz, 7.48 (1H, s), 7.48 (1H, s), 7.44 (2H, d, J=8.8 ν −1 IR ( max,cm ): 3371, 2979, 1698, 1671, 1522, 1445, 1367, 1285, Hz), 7.42 (1H, s), 7.19 (1H, d, J=2.0), 7.13 (1H, s), 3.96 (3H, s), 1161, 1112, 1047, 823, 803, 762, 714, 631. 1H NMR (400 MHz, 3.91 (3H, s), 3.87 (3H, s). EIMS (m/z): 353.64 (MH+). δ CDCl3): 8.33 (1H, s), 7.55 (d, 2H, J = 8.8 Hz), 7.50 (d, 2 H, J = 8.8 Methyl 4-(4-(4-Aminophenyl)-1-methyl-1H-pyrrole-2-carboxami- Hz), 7.36 (d, 1H, J = 2.0 Hz,), 7.22 (d, 1H, J = 2.0), 3.97 (s, 3H), 1.50 do)-1-methyl-1H-imidazole-2-carboxylate (28). The Boc-protected 13 δ (s, 9H). C NMR (100 MHz, CDCl3): 162.3, 153.7, 138.6, 123.0, 22 (0.3 g, 0.94 mmol, 1.2 equiv) was dissolved in DMF (5 mL), and 127.1, 126.0, 124.4, 124.0, 119.5, 115.1, 79.9, 36.9, 28.6. EIMS (m/z): EDCI (302 mg, 1.58 mmol, 2.0 equiv) and DMAP (241 mg, 1.97 315.16 (MH+). mmol, 2.5 equiv) were added. The mixture was allowed to stir for 30 Methyl 4-(4-Aminophenyl)-1-methyl-1H-pyrrole-2-carboxylate min at rt, after which methyl 4-amino-1-methyl-1H-imidazole-2- (23). The N-Boc-protected methyl ester 21 (1 g, 3.03 mmol) was carboxylate (0.12 g, 0.79 mmol, 1.0 equiv) was added. The reaction dissolved in MeOH/4 M HCl in dioxane (50:50, 15 mL) and the mixture was allowed to stir for a further 6 h, at which point TLC reaction mixture stirred for 3 h, when TLC indicated that reaction was indicated that reaction was complete. The reaction was then quenched complete. The solvent was evaporated in vacuo to produce crude 23 as by pouring the mixture into ice/water (150 mL), and the resulting a brown solid. This was purified by flash chromatography with silica mixture was extracted with ethyl acetate (3 × 60 mL). The combined gel using n-hexane/EtOAc (9:1) as the eluent. Evaporation of the organic phase was washed sequentially with saturated aqueous ff combined relevant fractions in vacuo a orded 23 as a brown solid NaHCO3 (50 mL), water (50 mL), and brine (50 mL) and then ν −1 (0.65 g, 94.2%). IR ( max,cm ): 3366, 2987,1688, 1629, 1566, 1422, dried (MgSO4), and the ethyl acetate was evaporated in vacuo to 1372, 1262, 1181, 1103, 1067, 951, 821, 784, 756. 1H NMR (400 afford the crude product 26 (0.48 g), which was used directly for the δ MHz, CDCl3): 7.28 (2H, d, J = 8.4 Hz), 7.11 (1H, d, J = 2.0 Hz), next Boc-deprotection step. The crude intermediate was dissolved in 6.96 (1H, d, J = 2.0 Hz), 6.68 (d, 2 H, J = 8.0 Hz), 3.94 (s, 3H), 3.83 MeOH (5 mL), and 4 M HCl in dioxane (5 mL) was added slowly to 13 δ (s, 3H). C NMR (100 MHz, CDCl3): 161.7, 144.7, 126.2, 125.4, the stirred solution. The reaction mixture was then stirred for 2 h, 125.2, 115.5, 114.4, 51.0, 36.8. EIMS (m/z): 231.1 (MH+). when TLC indicated that reaction was complete. The solvent was Methyl 4-(4-(4-(4-Aminophenyl)-1-methyl-1H-pyrrole-2- evaporated in vacuo to obtain the crude product as a brown solid. This carboxamido)phenyl)-1-methyl-1H-pyrrole-2-carboxylate (24). was purified by flash chromatography (n-hexane/EtOAc, 9:1) to afford EDCI (200 mg, 1.04 mmol, 2.0 equiv) and DMAP (159 mg, 1.3 pure 28 as a light brown solid (0.35 g, 81% over two steps). 1H NMR mmol, 2.5 equiv) were added to a solution of the N-Boc-protected (DMSO, 400 MHz): δ 9.75 (1H, s), 8.03 (1H, s), 7.71 (2H, d, J=8.8 amino acid 22 (0.2 g, 0.63 mmol, 1.2 equiv) in DMF (5 mL), and the Hz, 7.53 (1H, s), 7.52 (1H, s), 7.48 (2H, d, J=8.8 Hz), 7.42 (1H, s), mixture was stirred for 30 min at rt. The amino ester 23 (120 mg, 0.52 7.19 (1H, d, J=2.0), 3.94 (3H, s), 3.91 (3H, s), 3.89 (3H, s). EIMS mmol, 1.0 equiv) was added and the reaction mixture allowed to stir (m/z): 354.42 (MH+). for a further 3 h, after which TLC indicated that reaction was Methyl 4-(4-(4-(4-Aminophenyl)-1-methyl-1H-pyrrole-2-carboxa- complete. The reaction was quenched by being pouring into ice/water mido)-1-methyl-1H-pyrrole-2-carboxamido)-1-methyl-1H-pyrrole- (approximately 150 mL), and the resulting mixture was extracted with 2-carboxylate (33). Lithium hydroxide (68 mg, 1.65 mmol, 3 equiv) ethyl acetate (3 × 50 mL). The combined organic phase was washed was added to 25 (0.25 g, 0.55 mmol) dissolved in aqueous dioxane (8 sequentially with saturated aqueous NaHCO3 (50 mL), water (50 mL of dioxane, 4 mL of water) at room temperature. The reaction mL), and brine (50 mL) and then dried (MgSO4). The ethyl acetate mixture was stirred for 3 h, when TLC showed that reaction was was evaporated in vacuo to afford the product, which was used directly complete. The dioxane was evaporated in vacuo and the residue in the Boc-deprotection step without purification. The crude product diluted with water (100 mL). The resulting solution was acidified with was dissolved in MeOH/4 M HCl in dioxane (50:50, 5 mL) and the 1 M citric acid (to pH 3) and then extracted with ethyl acetate (2 × 50 solution stirred for 2 h, after which TLC indicated that reaction was mL). The combined organic phase was washed with brine (50 mL), complete. The solvent was evaporated in vacuo to afford crude 24 as a dried (MgSO4), and evaporated in vacuo to obtain the intermediate 29 brown solid. This was purified by flash chromatography using silica gel as a white solid (0.23 g, 91.6%), which was used directly for the next and n-hexane/EtOAc (9:1) as the eluent. Evaporation of the combined step without purification. EDCI (200 mg, 1.04 mmol, 2.0 equiv) and relevant fractions in vacuo afforded pure 24 as a brown solid (0.18 g, DMAP (158 mg, 1.3 mmol, 2.5 equiv) were added to a stirred solution 1 δ 80.7%). H NMR (400 MHz, CDCl3): 7.72 (1H, s), 7.69 (1H, s), of 29 (0.23 g, 0.52 mmol) in DMF, followed by stirring at rt for 20 7.57 (2H, d, J = 8.0 Hz), 7.46 (4H, d, J = 8.0 Hz), 7.41 (2H, d, J = 8.0 min. Methyl 4-amino-1-methyl-1H-pyrrole-2-carboxylate (80.1 mg, Hz), 7.20 (1H, d, J = 2.0 Hz), 7.06 (1H, d, J = 2.0 Hz, 7.02 (1H, d, J = 0.52 mmol, 1.0 equiv) was then added and the reaction mixture 1.6 Hz), 6.92 (1H, s). EIMS (m/z): 429.26 (MH+). allowed to stir for a further 3 h, when TLC indicated that reaction was Methyl 4-(4-(4-Aminophenyl)-1-methyl-1H-pyrrole-2-carboxami- complete. The reaction was quenched by pouring the mixture into ice/ do)-1-methyl-1H-pyrrole-2-carboxylate (27). EDCI (542 mg, 2.82 water (150 mL), and the resulting mixture was extracted with ethyl mmol, 2 equiv) and DMAP (433 mg, 3.55 mmol, 2.5 equiv) were acetate (3 × 50 mL). The combined extracts were washed sequentially

2928 dx.doi.org/10.1021/jm301882a | J. Med. Chem. 2013, 56, 2911−2935 Journal of Medicinal Chemistry Article with saturated aqueous NaHCO3 (50 mL) and brine (50 mL) and to stir for 30 min. Methyl 4-amino-1-methyl-1H-pyrrole-2-carboxylate then dried (MgSO4). The ethyl acetate was evaporated in vacuo to (0.5 g, 3.243 mmol, 1.0 equiv) was then added and the mixture afford crude 31, which was used directly for the final Boc-deprotection allowed to stir for a further 6 h, when TLC indicated that reaction was step. The intermediate 31 was dissolved in MeOH (5 mL), and 4 M complete. The reaction was quenched by pouring the mixture into ice/ HCl in dioxane (5 mL) was added slowly to the stirred solution. The water (500 mL), and the resulting mixture was extracted with ethyl reaction mixture was stirred for 3 h, when TLC indicated that reaction acetate (3 × 150 mL). The combined organic phase was washed was complete. The solvent was evaporated in vacuo to give the crude sequentially with citric acid (200 mL), saturated aqueous NaHCO fi fi fl 3 nal product as a brown solid. This was puri ed by ash (250 mL), water (250 mL), and brine (250 mL) and dried (MgSO4), chromatography (n-hexane/EtOAc, 8:2) to afford 31 as a light and the ethyl acetate was evaporated in vacuo to afford the crude brown solid (0.20 g, 83% over three steps). 1H NMR (DMSO, 400 product (1.88 g), which was used directly in the next hydrolysis step to MHz): δ 9.72 (1H, s), 8.09 (1H, t, J=5.6 Hz), 7.71 (2H, d, J=8.8 give 39. Lithium hydroxide (0.24 g, 5.71 mmol, 3 equiv) was added to Hz), 7.49 (2H, d, J=8.8 Hz), 7.40 (1H, d, J=2.0 Hz), 7.27 (1H, d, J a solution of the intermediate (1.88 g, 2.87 mmol) in aqueous dioxane = 2.0), 7.19 (1H, d, J=2.0), 7.03 (1H, dd, J=4.0, 1.6 Hz), 7.00 (1H, t, (75 mL of dioxane, 11.5 mL of water) at rt, followed by stirring for 3 h, J = 2.0 Hz), 6.84 (1H, d, J=2.0 Hz), 6.10 (1H, m), 3.89 (3H, s). when TLC indicated that reaction was complete. The dioxane was EIMS (m/z): 475.35 (MH+). evaporated in vacuo and the residue diluted with water (200 mL). The 4-((10-((Allyloxy)carbonyl)-7-methoxy-5-oxo-11-((tetrahydro-2H- resulting solution was acidified with 1 M citric acid (to pH 3) followed pyran-2-yl)oxy)-2,3,5,10,11,11a-hexahydro-1H-pyrrolo[2,1-c][1,4]- by extraction with ethyl acetate (2 × 100 mL). The combined organic benzodiazepin-8-yl)oxy)butanoic Acid (36). NaOH (0.5 M) was phase was washed with brine (100 mL), dried (MgSO4), and added to a solution of 35 (9.4 g, 17.67 mmol, 1.0 equiv) in dioxane concentrated in vacuo to afford 44 as a white solid (1.68 g, 74% over (250 mL) at rt. The reaction mixture was stirred for 4 h at rt, when two steps). 1H NMR: δ 9.09 (1H, s), 7.39 (1H, d, J = 2.0 Hz), 7.14 TLC indicated that reaction was complete. The dioxane was then (1H, s), 7.12 (1H, s), 6.96 (1H, s), 6.76 (1H, d, J = 2.0 Hz), 5.86−5.75 evaporated in vacuo and the residue diluted with water (250 mL). The (2H, m), 5.13−4.84 (3H, m), 4.61−4.21 (2H,m), 4.06−3.88 (3H, m), resulting solution was acidified with 1 M citric acid (to pH 3) and then − − × 3.87 (3H, s), 3.87 (3H, s), 3.86 (3H, s), 3.53 3.44 (3H, m), 2.55 extracted with ethyl acetate (2 100 mL). The combined organic 2.45 (2H, m), 2.13−1.88 (6H, m), 1.70−1.39 (6H). EIMS (m/z): phase was washed with brine (100 mL), dried (MgSO4), and 641.57 (MH+). concentrated in vacuo to afford 36 as a white solid (8.7 g, 94%). 1H 6 δ 4-(4-(((11S,11aS)-10-((Allyloxy)carbonyl)-7-methoxy-5-oxo-11- NMR (400 MHz, CDCl3): 7.2 (2H, s), 6.90 (1H, s), 6.58 (1H, s), − ((tetrahydro-2H-pyran-2-yl)oxy)-2,3,5,10,11,11a-hexahydro-1H- 5.85 (2H, d, J = 9.2 Hz), 5.73 (2H, d, J = 9.2 Hz), 5.03 5.13 (m, 6H), benzo[e]pyrrolo[1,2-a][1,4]diazepin-8-yl)oxy)butanamido)-1-meth- 4.68−4.35 (m, 4H), 4.09−4.01 (m, 4H), 3.91−3.82 (m, 8H), 3.69− yl-1H-imidazole-2-carboxylic Acid (40). EDCI (1.13 g, 6.48 mmol, 3.46 (m, 8H), 2.60−2.55 (m, 4H), 2.18−2.00 (m, 10H), 1.76−1.55 2.0 equiv) and DMAP (0.99 g, 8.1 mmol, 2.5 equiv) were added to a − 13 δ (m, 4H), 1.53 1.43 (m, 8H). C NMR (100 MHz, CDCl3): 177.6, stirred solution of the Alloc-THP-protected PBD acid 36 (1.68 g, 3.25 167.6, 149.8, 132.1, 131.9, 126.7, 117.3, 114.9, 110.8, 100.7, 96.0, 91.7, mmol, 1.1 equiv) in DMF (50 mL), and the mixture was allowed to 88.5, 67.9, 66.6, 63.6, 60.1, 56.1, 46.5, 31.1, 30.3, 28.8, 25.2, 24.1, 23.2, stir for 30 min at rt. Ethyl 4-amino-1-methyl-1H-imidazole-2- + 20.0. EIMS (m/z): 519.26 (MH ). carboxylate (0.5 g, 2.95 mmol, 1.0 equiv) was then added and the (11aS)-Allyl-7-methoxy-8-(4-((4-(5-(methoxycarbonyl)-1-methyl- mixture allowed to stir for a further 6 h, when TLC indicated that 1H-pyrrol-3-yl)phenyl)amino)-4-oxobutoxy)-5-oxo-11-((tetrahydro- reaction was complete. The reaction was quenched by pouring the 2H-pyran-2-yl)oxy)-2,3,11,11a-hexahydro-1H-pyrrolo[2,1-c][1,4]- mixture into ice/water (500 mL), and the resulting mixture was benzodiazepine-10(5H)-carboxylate (38). EDCI (2.49 g, 13.02 × mmol, 2.0 equiv) and DMAP (1.989 g, 16.28 mmol, 2.5 equiv) extracted with ethyl acetate (3 150 mL). The combined organic were added to a stirred solution of the Alloc-THP-protected PBD acid phase was washed sequentially with citric acid (200 mL), saturated 36 (3.72 g, 7.16 mmol, 1.2 equiv) in DMF at rt. The mixture was aqueous NaHCO3 (250 mL), water (250 mL), and brine (250 mL) allowed to stir for 30 min, after which the MPB ester 23 (1.5 g, 6.514 and then dried (MgSO4). The ethyl acetate was evaporated in vacuo to mmol, 1.0 equiv) was added, and the mixture was allowed to stir for a give the crude product (1.97 g), which was used directly for the next further 2 h, when TLC indicated that reaction was complete. The hydrolysis step. Lithium hydroxide (0.338 g, 5.71 mmol, 3 equiv) was reaction was quenched by pouring the mixture into ice/water (500 added to the crude product (1.97 g) in aqueous dioxane (75 mL of mL), and the resulting mixture was extracted with ethyl acetate (3 × dioxane, 15.1 mL of water) at rt and the mixture stirred for 3 h, when 150 mL). The combined organic phase was washed with citric acid TLC indicated that reaction was complete. The dioxane was (200 mL), saturated aqueous NaHCO (250 mL), water (250 mL), evaporated in vacuo, the residue dissolved in water (200 mL), and 3 the resulting solution acidified with 1 M citric acid (to pH 3) and and brine (250 mL) and dried (MgSO4), and the ethyl acetate was × evaporated in vacuo to afford the crude final product as a white solid. extracted with ethyl acetate (2 100 mL). The combined organic This was purified by flash chromatography using silica gel (MeOH/ phase was washed with brine (100 mL), dried (MgSO4), and ff evaporated in vacuo to afford 40 as a white solid (1.38 g, 66.67%). 1H CHCl3, 2:8) to a ord 38 as a white foamy solid (4.05 g, 85.5%). IR − δ (ν ,cm 1): 2949, 2362, 1704, 1600, 1514, 1436, 1372, 1269, 1203, NMR: 9.09 (1H, s), 7.39 (1H, d, J = 2.0 Hz), 7.14 (1H, s), 7.12 (1H, max − − − 1107, 1021, 964, 765. 1H NMR (400 MHz, CDCl ): δ 7.82 (1H, s), s), 6.96 (1H, s), 5.86 5.75 (2H, m), 5.13 4.84 (3H, m), 4.61 4.21 3 − 7.48 (2H, m), 7.41 (1H, d, J = 2.0 Hz), 7.40 (1H, d, J = 2.4 Hz), 7.23 (2H, m), 4.06 3.88 (3H, m), 3.87 (3H, s), 3.87 (3H, s), 3.86 (3H, s), − − − − (2H, d, J = 8.4 Hz), 7.17 (1H, d, J = 2.0 Hz), 7.04 (1H, d, J = 2.0 Hz), 3.53 3.44 (3H, m), 2.55 2.45 (2H, m), 2.13 1.88 (6H, m), 1.70 + 5.93−5.65 (2H, m), 5.09−5.4.97 (m, 4H), 4.68−4.32 (m, 4H), 4.15− 1.39 (6H). EIMS (m/z): 642.78 (MH ). 4.10 (m, 4H), 3.94−3.82 (m, 12H), 3.68 (m, 2H), 3.59−3.49 (m, 6H), (S)-Methyl 4-(4-(4-((7-Methoxy-5-oxo-2,3,5,11a-tetrahydro-1H- 2.60−2.57 (m, 3H), 2.15−2.00 (m, 8H), 1.88−1.80 (m, 2H), 1.79- pyrrolo[2,1-c][1,4]benzodiazepin-8-yl)oxy)butanamido)phenyl)-1- − 13 δ methyl-1H-pyrrole-2-carboxylate (7a). Palladium tetrakis- 1.70 (6H), 1.60 1.44 (m, 12H). C NMR (100 MHz, CDCl3): μ 177.1, 170.5, 167.3, 161,6, 149.1, 136.3, 132.1, 131.9, 130.4, 128.9, (triphenylphosphine) (5.60 mg, 4.8 M, 0.05 equiv) was added to a 127.1, 125.9, 125.4, 123.5, 123.1, 120.3, 117.3, 114.6, 110.8, 91.5, 88.6, mixture of the Alloc-THP-protected PBD conjugate 38 (70 mg, 0.097 68.2, 66.5, 64.3, 63.6, 60.3, 56.0, 51.1, 46.4, 36.8, 31.1, 30.9, 29.1, 25.1, mmol), pyrrolidine (8.36 mg, 0.117 mmol, 1.2 equiv), and 24.6, 23.2, 21.0, 20.1. EIMS (m/z): 731.67 (MH+). triphenylphosphine (8.62 mg, 0.25 equiv) in DCM (5 mL), and the 4-(4-(((11S,11aS)-10-((Allyloxy)carbonyl)-7-methoxy-5-oxo-11- reaction mixture was stirred at rt for 2 h, when TLC indicated that ((tetrahydro-2H-pyran-2-yl)oxy)-2,3,5,10,11,11a-hexahydro-1H- reaction was complete. The DCM was evaporated in vacuo and the benzo[e]pyrrolo[1,2-a][1,4]diazepin-8-yl)oxy)butanamido)-1-meth- resulting residue stored in vacuo for 4 h to remove excess pyrrolidine. yl-1H-pyrrole-2-carboxylic Acid (39). EDCI (1.24 g, 6.48 mmol, 2.0 This crude product (60 mg) was purified by column chromatography equiv) and DMAP (0.99 g, 8.1 mmol, 2.5 equiv) were added to a (n-hexane/EtOAc, 65:35) to afford 7a as a yellow solid (40 mg, 77%). α 22.7 ν −1 stirred solution of the Alloc-THP-protected PBD acid 36 (1.85 g, 3.57 [ ]D : +165 (c = 0.046, CHCl3). IR ( max,cm ): 3297, 2944, 2358, mmol, 1.2 equiv) in DMF (50 mL) at rt, and the mixture was allowed 1701, 1598, 1567, 1508, 1442, 1374, 1264, 1212, 1181, 1106, 1072,

2929 dx.doi.org/10.1021/jm301882a | J. Med. Chem. 2013, 56, 2911−2935 Journal of Medicinal Chemistry Article

1 δ 824, 730. H NMR (500 MHz, CDCl3): 7.68 (1H, s), 7.65 (1H, d, J J = 8.0 Hz), 7.45 (1H, d, J = 1.6 Hz), 7.31 (2H, d, J = 8.0 Hz), 7.20 = 4.5 Hz), 7.52 (1H, s), 7.46 (2H, dd, J = 8.4, 2.0 Hz), 7.40 (2H, dd, J (1H, s), 6.96 (1H, s), 6.89 (1H, br s), 6.81 (1H, s), 6.78 (1H, d, J = 1.6 = 8.4, 2.0 Hz), 7.16 (1H, d, J = 2.0 Hz), 7.03 (1H, d, J = 1.6 Hz), 6.82 Hz), 6.71 (1H, br s), 4.11−4.16 (2H, m), 3.97 (3H, s), 3.92 (3H, s), (1H, s), 4.12−4.20 (2H, m), 3.94 (3H, s), 3.88 (3H, s), 3.88 (3H, s), 3.88 (3H, s), 3.84 (3H, s), 3.79 (3H, s), 3.68−3.71 (1H, m), 3.55− 3.68−3.71 (1H, m), 3.50−3.60 (2H, m), 2.58−2.62 (2H, m), 2.26− 3.60 (2H, m), 2.56−2.61 (2H, m), 2.22−2.28 (4H, m), 1.99−2.04 − 13 δ + 2.31 (4H, m), 1.50 1.54 (2H, m). C NMR (125 MHz, CDCl3): (2H, m). HRMS (EI, m/z): calcd for C41H43N9O8 (MH ), 790.3313; 164.5, 162.4, 161.6, 150.5, 147.8, 140.7, 125.9, 125.5 (2C), 123.6, found, 790.3314. 123.1, 120.3, 114.6, 111.8, 111.0, 94.4 (2C), 68.0, 63.7, 56.1, 53.7, 51.0, (S)-Methyl 4-(4-(4-(4-(4-((7-Methoxy-5-oxo-2,3,5,11a-tetrahydro- 46.6, 36.8, 31.9, 29.6, 25.2, 24.8, 24.1, 20.2. HRMS (EI, m/z): calcd for 1H-pyrrolo[2,1-c][1,4]benzodiazepin-8-yl)oxy)butanamido)phenyl)- C H N O (MH+), 545.2400; found 545.2422. 1-methyl-1H-pyrrole-2-carboxamido)-1-methyl-1H-imidazole-2- 30 32 4 6 α 22.7 Compounds 7b−h were synthesized in an analogous manner, but carboxamido)-1-methyl-1H-pyrrole-2-carboxylate (7f). [ ]D : +142 ν −1 using different reaction intermediates as outlined in Scheme 4. See (c = 0.043, CHCl3). IR ( max,cm ): 3408, 2358, 2168, 2148, 2019, 1978,1938, 1718, 1534, 1260, 1118, 757. 1H NMR (500 MHz, Table S2 (Supporting Information) for reaction yields. δ (S)-Methyl 4-(4-(4-(4-(4-((7-Methoxy-5-oxo-2,3,5,11a-tetrahydro- CDCl3): 8.72 (1H, s, NH), 8.12 (1H, s, NH), 7.71 (1H, s), 7.65 1H-pyrrolo[2,1-c][1,4]benzodiazepin-8-yl)oxy)butanamido)phenyl)- (1H, d, J = 4.4 Hz), 7.53 (1H, s), 7.48 (2H, d, J = 8.0 Hz), 7.47 (1H, 1-methyl-1H-pyrrole-2-carboxamido)phenyl)-1-methyl-1H-pyrrole- s), 7.42 (2H, d, J = 1.6 Hz), 7.40 (2H, d, J = 8.0 Hz), 7.03 (1H, d, J = α 22.7 ν −1 − 2-carboxylate (7b). [ ]D : +134 (c = 0.038, CHCl3). IR ( max,cm ): 1.6 Hz), 6.95 (1H, s), 6.82 (1H, s), 6.81 (1H, d, J = 1.6 Hz,), 4.12 3850, 3732, 3619, 2443, 2354, 2228, 2169, 2091, 1971,1859, 1729, 4.21 (2H, m), 4.07 (3H, s), 4.00 (3H, s), 3.91 (3H, s), 3.89 (3H, s), 1679, 1521,1265, 734, 629. 1H NMR (500 MHz, CDCl ): δ 7.72 (1H, 3.81 (3H, s), 3.69−3.72 (1H, m), 3.55−3.61 (2H, m), 2.58−2.63 (2H, 3 − − s), 7.69 (1H, s), 7.66 (1H, d, J = 4.0 Hz), 7.57 (2H, d, J = 8.0 Hz), m), 2.26 2.32 (4H, m, CH2), 2.02 2.07 (2H, m). HRMS (EI, m/z): + 7.53 (1H, s), 7.46 (4H, d, J = 8.0 Hz), 7.41 (2H, d, J = 8.0 Hz), 7.20 calcd for C41H43N9O8 (MH ): 790.3313; found, 790.3314. (1H, d, J = 2.0 Hz), 7.06 (1H, d, J = 2.0 Hz), 7.02 (1H, d, J = 1.6 Hz), (S)-Methyl 4-(4-(4-(4-((7-Methoxy-5-oxo-2,3,5,11a-tetrahydro- 6.92 (1H, s), 6.84 (1H, s), 4.12−4.20 (2H, m), 4.00 (3H, s), 3.96 (3H, 1H-pyrrolo[2,1-c][1,4]benzodiazepin-8-yl)oxy)butanamido)-1- methyl-1H-pyrrole-2-carboxamido)phenyl)-1-methyl-1H-pyrrole-2- s), 3.88 (3H, s), 3.84 (3H, s), 3.70−3.73 (1H, m), 3.55−3.61 (2H, m), − carboxylate (7g). α 22.7 ν 1 2.58−2.62 (2H, m), 2.29−2.31 (2H, m), 1.93−2.06 (4H, m). 13C [ ]D : +197 (c = 0.052, CHCl3). IR ( max,cm ): NMR (125 MHz, CDCl ): δ 164.5, 162.4, 161.7, 150.7, 147.3, 139.2, 3330, 2360, 2214, 2180, 2041, 2020, 1999, 1967, 1698, 1517, 1438, 3 1265, 1180, 1119, 756, 722, 696, 667, 630. 1H NMR (500 MHz, 126.0, 125.6, 125.4 (2C), 125.2 (2C), 123.0, 120.4 (2C), 114.6 (2C), δ 111.4, 94.6 (2C), 68.3, 63.7, 56.1, 51.6 (2C), 41.0, 36.9, 31.9, 29.6, CDCl3): 7.77 (1H, s), 7.68 (1H, s), 7.67 (2H, d, J = 8.0 Hz), 7.64 (1H, d, J = 5.0 Hz), 7.55 (2H, d, J = 8.0 Hz), 7.47 (1H, d, J = 2.0 Hz), 25.2, 24.2, 24.1, 20.2. HRMS (EI, m/z): calcd for C H N O (MH+), 42 42 6 7 7.43 (1H, s), 7.18 (1H, d, J = 2.0 Hz), 7.09 (1H, d, J = 2.0 Hz), 7.05, 743.3193; found, 743.3193. (1H, d, J = 2.0 Hz), 6.83 (1H, s), 4.09−4.16 (2H, m), 3.95 (3H, s), (S)-Methyl 4-(4-(4-(4-((7-Methoxy-5-oxo-2,3,5,11a-tetrahydro- − − 1H-pyrrolo[2,1-c][1,4]benzodiazepin-8-yl)oxy)butanamido)phenyl)- 3.90 (6H, s), 3.84 (3H, s), 3.67 3.71 (1H, m), 3.54 3.57 (2H, m), − − − 13 1-methyl-1H-pyrrole-2-carboxamido)-1-methyl-1H-pyrrole-2-car- 2.53 2.56 (2H, m), 2.23 2.30 (4H, m), 2.00 2.05 (2H, m). C α 22.7 ν −1 δ boxylate (7c). [ ]D : +128 (c = 0.037, CHCl3). IR ( max,cm ): NMR (125 MHz, CDCl3): 169.8, 164.5, 162.6, 161.6, 159.5, 150.7, 3321, 2237, 2107, 2041, 1967, 1860, 1685,1517,1435, 1254, 1180, 147.9, 140.8, 133.1, 132.2, 132.1, 131.9, 131,7, 128.5, 128.4, 125.9, 1 δ 1118, 749, 722, 696, 667. H NMR (500 MHz, CDCl3): 7.98 (1H, 125.5, 123.7, 123.1, 121.5, 120.7, 120.4, 119.7, 114.7, 112.0, 111.4, s), 7.88 (1H, s), 7.68 (1H, s), 7.65 (1H, d, J = 4.0 Hz,), 7.64 (2H, d, J 103.9, 68.1, 56.2, 53.7, 51.0, 46.7, 36.8, 33.2, 29.6, 25.1, 24.1. HRMS + = 8.0 Hz), 7.54 (1H, d, J = 1.6 Hz), 7.52 (1H, d, J = 1.6 Hz), 7.45 (1H, (EI, m/z): calcd for C36H38N6O7 (MH ), 667.2880; found, 667.2882. d, J = 2.0 Hz), 7.33 (2H, d, J = 8.0 Hz), 6.97 (1H, s), 6.89 (1H, s), (S)-Methyl 4-(4-(4-(4-((7-Methoxy-5-oxo-2,3,5,11a-tetrahydro- 4.08−4.18 (2H, m), 3.97 (3H, s), 3.89 (3H, s), 3.84 (3H, s), 3.79 (3H, 1H-pyrrolo[2,1-c][1,4]benzodiazepin-8-yl)oxy)butanamido)-1- methyl-1H-imidazole-2-carboxamido)phenyl)-1-methyl-1H-pyrrole- s), 3.66−3.70 (1H, m), 3.55−3.60 (2H, m), 2.56−2.61 (2H, m), 2.23− − 2-carboxylate (7h). [α]22.7: +188 (c = 0.052, CHCl ). IR (ν ,cm 1): 2.32 (4H, m), 2.00−2.05 (2H, m). 13C NMR (125 MHz, CDCl ): δ D 3 max 3 3301, 2169, 2136, 2018, 1978, 1937, 1680, 1564, 1518, 1439, 1265, 162.5, 161.6, 159.1, 150.4, 147.7, 138.4, 132.8, 132.1 131.9 (2C), 1181,1108, 750, 722. 1H NMR (500 MHz, CDCl ): δ 8.90 (1H, s), 128.6, 128.4 (2C), 125.4(2C), 124.8, 123.0, 121.0, 120.4 (2C), 116.2, 3 7.98 (1H, s, NH), 7.67 (1H, s), 7.63 (1H, d, J = 4.4 Hz), 7.59 (2H, d, J 114.6 (2C), 109.9, 94.2, 67.4, 63.6, 57.1, 53.7, 51.1, 46.7, 36.9, 36.7, + = 8.4 Hz), 7.46 (2H, d, J = 8.4 Hz), 7.42 (1H, s), 7.19 (1H, d, J = 2.0 34.0, 29.6, 24.2. HRMS (EI, m/z): calcd for C36H38N6O7 (MH ), Hz), 7.06 (1H, d, J = 1.6 Hz), 6.83 (1H, s), 4.10−4.22 (2H, m), 4.07 667.2880; found, 667.2881. (3H, s), 3.96 (6H, s), 3.84 (3H, s), 3.67−3.70 (1H, m), 3.54−3.58 (S)-Ethyl 4-(4-(4-(4-((7-Methoxy-5-oxo-2,3,5,11a-tetrahydro-1H- (2H, m), 2.57−2.67 (2H, m), 2.26−2.31 (4H, m), 1.98−2.05 (2H, m). pyrrolo[2,1-c][1,4]benzodiazepin-8-yl)oxy)butanamido)phenyl)-1- 13 δ methyl-1H-pyrrole-2-carboxamido)-1-methyl-1H-imidazole-2-car- C NMR (125 MHz, CDCl3): 169.5, 164.5, 162.5, 161.6, 156.4, α 22.7 ν −1 boxylate (7d). [ ]D : +122 (c = 0.028, CHCl3). IR ( max,cm ): 150.4, 147.8, 135.6, 132.1, 131.9 (2C), 128.5, 128.4, 126.0, 125.6, 3324, 2355, 2157, 2109, 2032, 1913, 1600,1533, 1465, 1262, 1179, 123.5, 123.1, 121.5, 119.7, 114.6 (2C), 111.6, 111.0, 67.7, 56.1, 53.7, 1109, 751. 1H NMR (500 MHz, CDCl ): δ 8.47 (1H, s), 7.72 (1H, s), 51.1, 46.6, 36.9, 35.8, 33.9, 29.6, 24.7, 24.1. HRMS (EI, m/z): calcd for 3 + 7.66 (1H, d, J = 4.0 Hz), 7.55 (1H, s), 7.52 (1H, d, J = 2.0), 7.49 (2H, C35H37N7O7 (MH ), 668.2833; found, 668.2838. fi d, J = 8.0 Hz), 7.37 (2H, d, J = 8.0 Hz), 7.16 (1H, d, J = 1.6 Hz), 7.03 FID Assay. This assay was a modi cation of a procedure described 36 (1H, s), 6.91 (1H, s), 4.39−4.43 (2H, m) 4.13−4.22 (2H, m), 4.01 by Boger and co-workers. The hairpin deoxyoligonucleotides were μ (3H, s), 3.99 (3H, s), 3.83 (3H, s), 3.68−3.72 (1H, m), 3.55−3.60 purchased from Genbase Inc. as 1600 M solutions in water. Prior to μ fi (2H, m), 2.58−2.63 (2H, m), 2.24−2.32 (4H, m), 2.00−2.07 (2H, m), use they were diluted to a concentration of 10 M using puri ed water ° 1.41−1.45 (3H, m). 13C NMR (125 MHz, CDCl ): δ 163.1, 162.5, and stored at 0 C. Each well of a Costar Black 96-well plate was 3 μ μ 158.7, 150.4, 147.7, 140.7, 137.2, 131.5, 125.6 (2C), 125.4 (2C), 123.8, loaded with 5 L of a unique hairpin from the 10 M stock solutions fi μ μ ff 123.0, 120.4 (2C), 114.5, 111.6, 110.8, 109.9, 100.0, 67.4, 61.5, 56.1, ( nal concentration 1 M) and 44 L of Tris bu er containing EtBr fi μ 53.7, 51.1, 46.7, 37.0, 36.0, 34.0, 29.6, 24.8, 24.2, 14.4. HRMS (EI, m/ ( nal concentration 4 M). The plate was centrifuged for 45 s at 1500 + rpm and incubated for 15 min at 25 °C in the dark with mixing to z): calcd for C36H39N7O7 (MH ), 682.2989; found, 682.2986. (S)-Methyl 4-(4-(4-(4-(4-((7-Methoxy-5-oxo-2,3,5,11a-tetrahydro- ensure equilibration. Finally, 1 μL of drug solution (0.1 mM in 1H-pyrrolo[2,1-c][1,4]benzodiazepin-8-yl)oxy)butanamido)phenyl)- DMSO) was added to each well (final concentration 2 μM). After 1-methyl-1H-pyrrole-2-carboxamido)-1-methyl-1H-pyrrole-2-car- being centrifuged for 45 s, the plate was incubated for 45 min at 25 °C α 22.7 boxamido)-1-methyl-1H-pyrrole-2-carboxylate (7e). [ ]D : +149 (c in the dark. Each well was read (average of 10 reads) on an EnVision ν −1 fl λ λ = 0.054, CHCl3). IR ( max,cm ): 3310, 2947, 2358, 2168, 2153, 2132, uorescence plate reader ( ex = 544 nm, em = 595 nm), both before 2070, 2011, 1989, 1651, 1538, 1434, 1402, 1257, 1107, 753. 1H NMR the drug was added and after the final incubation period with the drug. δ (400 MHz, CDCl3): 8.02 (1H, s), 7.88 (1H, d, J = 5.2 Hz), 7.68 Compound assessments were conducted in triplicate, with some wells (1H, s), 7.67 (1H, d, J = 1.6 Hz), 7.64 (1H, d, J = 1.6 Hz), 7.53(2H, d, used as a control for no drug (e.g., EtBr and hairpin, 100%

2930 dx.doi.org/10.1021/jm301882a | J. Med. Chem. 2013, 56, 2911−2935 Journal of Medicinal Chemistry Article

fluorescence, first fluorescence reading) and some containing only the citrate in 1 mL of water, mixed 1:1 for the matrix), 2:1, 1:1, or 1:5 drug and EtBr in water to act as a control for no DNA (e.g., EtBr, (sample:matrix), to determine the most effective ratio. A 1 μL volume water, and drug, 0% fluorescence). Fluorescent readings were recorded of the sample/matrix mixture was spotted onto the MALDI target after addition of the drug to the DNA sequences intercalated by EtBr plate and allowed to dry. Analyses were carried out on a Voyager DE- (second fluorescence reading). Fluorescence readings were reported as Pro with a nitrogen laser in positive linear mode using delayed the percentage decrease in fluorescence relative to the control, but extraction (500 ns) and an accelerating voltage of 25 000 V. were converted to relative binding percentages using the formula 100 Acquisition was between 4000 and 15000 Da with 100 shots/ − (first fluorescence reading/second reading × 100). spectrum. Ion Pair Reversed-Phase (RP) HPLC−MS Assay. General Molecular Modeling. Molecular models were constructed using Informaiton. The single-stranded (SS) oligonucleotides were obtained AMBER (version 11) software.58 Each DNA sequence was in a lyophilized form from Eurogentec, United Kingdom. Each constructed using nab, and antechamber was used to convert the oligonucleotide was dissolved in 1 M ammonium acetate (Sigma- structures to mol2 files with the application of Gasteiger charges. Aldrich, United Kingdom) to form a stock solution of 1 mM. To Missing parameters were generated for 2, 4, 5a, 5b, 7a, and 7h using ensure hairpin formation, the oligonucleotide solutions were heated to parmchk. The gaff and DNA optimized parm99bsc0 force fields were ° 90 C for 10 min in a heating/cooling block (Grant Bio, United loaded for DNA, and xleap was used to manually dock each ligand Kingdom) and the solutions then allowed to cool slowly to room individually and in the case of 4 and 7h to create the covalent bond − ° temperature followed by storage at 20 C overnight to ensure between the exocyclic NH of the reacting guanine and the C10−N11 completion of the annealing process. Working solutions of hairpin 2 μ imine of the PBD. Each adduct was minimized in a stepwise manner oligonucleotides of 50 M were prepared by diluting the stored by initially restraining the DNA with a high positional force constant solutions with 100 mM ammonium acetate. to enable the ligand to find its energy minimum, followed by a GWL-78 (4). The PBD C8-conjugate 4 was provided by Spirogen reduction of the DNA restraints in a gradual manner until zero. Once Ltd (Batch no. SG2274.005) and was dissolved in methanol to form a the full system was minimized, production simulations were run for a stock solution of 10 mM which was stored at −20 °C for no longer μ period of 20 ns using pmemd. The generalized Born/surface area (GB/ than four months. Working solutions of the drug of 200 M were SA) implicit solvent model was used with monovalent electrostatic ion prepared by diluting the 10 mM stock solution with 100 mM − ° screening simulated with SALTCON set to 0.2M. The dynamics ammonium acetate. These were stored at 20 C for not more than integration time step was 0.002 ps while constraining all bonds to one week and thawed to room temperature for use when required. hydrogen atoms using the SHAKE algorithm. A temperature of 300 K Ligand−DNA Complex Preparation. Ligand−DNA complexes was maintained using the Langevin thermostat, and a long-range were prepared by incubating the ligands with hairpin oligonucleotides nonbonded cutoff of 100 Å was used. in a 4:1 molar ratio at room temperature. Samples were withdrawn at Initial studies involved the interaction of 2 (distamycin), 4 (GWL- various time intervals and subjected to ion pair RPLC and mass 78), 5a (Py−MPB−Py) and 5b (Py−MPB−Im) with hairpin DNA to spectrometry analysis as described below. ascertain GC-specificity of the noncovalent MPB moiety. Ligands were Ion Pair Reversed-Phase Liquid Chromatography. Chromatog- initially positioned outside of the DNA minor groove, and raphy was performed on a Thermo Electron HPLC system equipped minimization was performed on the system with DNA restrained for with a 4.6 × 50 mm Xterra MS C18 column packed with 2.5 μM particles (Waters Ltd., United Kingdom), a UV 1000 detector, an the initial stages to allow the ligands to locate in the minor groove at AS3000 autosampler, an SCM1000 vacuum degasser, and Chromquest their preferred binding sequences. Restraints were then removed in a software (version 4.1). A gradient system of 100 mM triethylammo- gradient manner, and the DNA and ligand minimized. Simulations ff were also undertaken with noncovalent and covalent adducts involving nium bicarbonate (TEAB) as bu er A and 40% acetonitrile in water ′ ′ ′ (HPLC grade, Fisher Scientific, United Kingdom) as buffer B was nonhairpin DNA containing 5 -GCTATTAGGC-3 ,5- ff ff GGGGGCGCGG-3′, and 5′-GCCCGGCGGC-3′ sequences with used. For bu er A, a 1 M preformulated bu er of TEAB (Sigma- − Aldrich, United Kingdom) was diluted to 100 mM with HPLC grade the same ligands. Furthermore, 20 ns simulations of ligand DNA fi adducts were performed both noncovalently and covalently bound in water (Fisher Scienti c, United Kingdom). The gradient was ramped ff from 90% A at 0 min to 50% A at 20 min, 65% A at 30 min, and finally an e ort to identify the recognition and binding mechanism of 7h to 10% A at 45 min. UV absorbance was monitored at 254 nm, and the sequence N-G-N-W-N-G/C-G/C (where W is adenine or fractions containing separated components were collected manually, thymine), the preferred binding sequence observed in the footprinting combined when appropriate, lyophilized, and analyzed using MALDI- studies. TOF mass spectrometry as described below. The free energy of binding was calculated using the AMBER Mass Spectrometry Analysis (ESI-MS). ESI-MS spectra were MMPBSA approach. The procedure was repeated for each ligand acquired on a Micromass Q-TOF Global tandem mass spectrometer when oriented in the reverse direction in the minor groove. For the (Waters, United Kingdom) fitted with a NanoSpray ion source. unbound ligands, the free energy of binding was measured according ff Negative mode was used for data acquisition, and the instrument was to the di erences in free energy of the complex compared to the DNA Δ − − calibrated with ions produced from a standard solution of taurocholic and ligand alone [ G = G(complex) G(DNA) G(ligand)]. acid (10 pmol/μL) in acetonitrile. The HPLC fractions collected were Further molecular modeling studies were undertaken to rationalize lyophilized (Speedvac, Thermo Electron, United Kingdom) and mixed the mechanism of action of 7h. A series of molecular dynamics with a 1:1 (v/v) mixture of 40% acetonitrile/water and 20 mM simulation studies in implicit solvent over 20 ns were conducted with triethylamine (TEA)/water (Fisher Scientific, United Kingdom), 7h both covalently bound and noncovalently bound to two NF-κB which was also used as the electrospray solvent. A 3−5 μL volume consensus sequences (5′-GGGACAGCCC-3′ and 5′- of sample was loaded into a metal-coated borosilicate electrospray GGGGGTCGCC-3′) using the procedures described above. Molec- 59 needle with an internal diameter of 0.7 mm and a spray orifice of 1−10 ular models were visualized using the UCSF Chimera program. μm (NanoES spray capillaries, Proxeon Biosystems, United Kingdom) For the preliminary docking study on the inhibition of interaction of which was positioned at ∼10 mm from the sample cone, giving a flow the NF-κB protein with its cognate DNA sequence by 7h, the rate of ∼20 nL/min. Nitrogen was used as the API gas, and the structures of the NF-κB DNA consensus sequence (5′- capillary, cone, and RF lens 1 voltages were set at 1.8−2.0 kV, ∼35 V, TGGGGACTTTCC-3′) and the NF-κB protein were separately and 50 V, respectively, to ensure minimum fragmentation of the isolated from the crystal structure of the complex of the NF-κB p50− ligand−DNA adducts. The collision and MCP voltages were set as 5 p65 heterodimer with DNA (PBD ID 1VKX). 7h was docked, and 2200 V, respectively. Spectra were acquired over the m/z range of covalently bound into the NF-κB recognition sequence, and energy 1000−3000. minimized in a stepwise manner using the AMBER software. Once Mass Spectrometry Analysis (MALDI-TOF). Samples were diluted minimized, the DNA−7h adduct was aligned with a copy of the with matrix (37 mg of THAP in 1 mL of ACN, 45 mg of ammonium original consensus DNA sequence from the crystal structure, thereby

2931 dx.doi.org/10.1021/jm301882a | J. Med. Chem. 2013, 56, 2911−2935 Journal of Medicinal Chemistry Article occupying identical positional coordinates, and the 7h adduct and control compounds. The cells were grown in normal cell culture ° fi protein were visualized together using Chimera. conditions at 37 C under a 5% CO2 humidi ed atmosphere, either in FRET DNA Thermal Denaturation Assay. Solutions (400 nM) Dulbecco’s modified Eagle’s medium or in modified Eagle’s medium of fluorescence-tagged Seq-3 (5′-Fam-TATA-AGA-TTT-TCT-TATA- (depending on the cell line) supplemented with 10% fetal bovine Tamra-3′) and Seq-4 (5′-Fam-GCGC-AGA-TTT-TCT-GCGC- serum (Biosera, United Kingdom), 1% L-glutamine, 1% nonessential Tamra-3′) (Eurogentec, United Kingdom) in FRET buffer (50 mM amino acids, and 0.05% hydrocortisone (Gibco, Invitrogen, United potassium cacodylate, pH 7.4) were prepared by diluting a 20 μM States). The cells were then seeded into 96-well plates in a total stock solution in water. These solutions were heated at 85 °C for 5 volume of 160 μL and allowed to reach a 30−40% degree of min before being cooled to room temperature over 5 h to promote confluence before the experiment was started. The test compounds annealing. Ligand solutions were prepared initially in concentrations were dissolved in sterilized ultrapure water at a maximum double that required for the final solutions, and dilutions from the concentration of 100 μM, and serial decimal dilutions were prepared initial 10 mM DMSO stock solutions were carried out using FRET and added to the cells in a volume of 40 μL. After 96 h of continuous buffer. A 50 μL volume of annealed DNA and 50 μL of ligand solution exposure to each compound, the cytotoxicity was determined using the were placed into each well of a 96-well plate (MJ Research Inc., United MTT (Lancaster Synthesis Ltd., United Kingdom) colorimetric States) and then processed in a DNA Engine Opticon (MJ Research). assay.43 Absorbance was quantified by spectrophotometry at λ = 570 ° Fluorescence readings were taken at intervals of 0.5 C over the range nm (ELx808, Bio-Tek Instruments, Inc., United States). IC50 values of 30−100 °C, with a constant temperature maintained for 30 s prior were calculated by a dose−response analysis using the Origin 6.0 to each reading. The incident radiation was 450−495 nm with software. detection at 515−545 nm. The raw data were imported into the Origin Annexin V Assay. Freshly isolated peripheral blood CLL cells (1 × program (version 7.0, OriginLab Corp., United States) and the graphs 106 mL−1) were cultured in RPMI medium (Invitrogen, Paisley, U.K.) smoothed using a 10-point running average prior to normalization. supplemented with 100 U/mL penicillin, 100 mg/mL streptomycin, Determination of melting temperatures was based on obtaining values and 10% fetal calf serum. The cells were incubated at 37 °Cina fi fi at the maxima of the rst derivative of the smoothed melting curves humidi ed 5% CO2 atmosphere in the presence of each compound. using a script. The difference between the melting temperature of each All compounds were dissolved in DMSO and were evaluated in serial Δ sample and that of the blank (i.e., Tm) was used for comparative dilutions against the primary CLL cells. In addition, control cultures purposes. were carried out to which no drug was added. The cytotoxic effects of DNase I Footprinting. DNase I footprinting was performed as the compounds were quantified using an annexin V/propidium iodide previously described60 using the HexA and HexB DNA fragments. flow cytometry assay (Bender Medsystems, Vienna, Austria). All assays These synthetic cloned fragments were designed so that between them were performed in duplicate, and LD50 values were calculated from they contained all 64 symmetrical hexanucleotide sequences. To sigmoidal dose−response curves using the Prism 6.0 software facilitate examination of the binding sites that are located at either end (Graphpad Software Inc., San Diego, CA). The sigmoidal dose− of these sequences, each fragment was cloned in both orientations response curves were derived by plotting log[compound concen- (HexA with HexArev and HexB with HexBrev) as previously tration] against the percentage apoptosis induced by that concen- described.41 These footprinting templates were considered to be tration. A wide range of concentrations were used to establish the especially useful for testing the symmetrical ligands, rather than natural biologically active range for each individual compound. restriction fragments that contain a random mixture of potential Human Tumor Xenograft Studies on 7h. The human breast binding sites. Radiolabeled DNA fragments were obtained by digesting (MDA-MB-231, 5 × 106 cells) and pancreatic (MIA PaCa-2, 5 × 106 the parent plasmids with HindIII and SacI (HexA) or EcoR1 and Pst1 cells) cancer cell lines were employed to establish subcutaneous (HexB), and then labeled at the 3′-end of the EcoR1 or HindIII sites xenografts in the flanks of female MF1 nude mice (2−3 months old, with [α-32P]dATP using reverse transcriptase. The radiolabeled weighing 20−25 g). Subsequent passage was by the subcutaneous fragments of interest were separated from the remainder of the implantation of small tumor pieces (approximately 1 mm3) into the plasmid DNA on 6% (w/v) polyacrylamide gels. The DNA was eluted flank. All experiments were in compliance with the “Guidelines for the and dissolved in 10 mM Tris−HCl, pH 7.5, containing 0.1 mM EDTA welfare and use of animals in cancer research”.61 When the tumors to give ∼10 cps/μL, as determined on a hand-held Geiger counter reached approximately 0.06 cm3 (three weeks postimplantation), the (<10 nM). A 1.5 μL volume of radiolabeled DNA was then mixed with mice were divided into three groups of six. The test compound 7h was 1.5 μL of ligand solution (dissolved in 10 mM Tris−HCl, pH 7.5, dissolved in DMSO and diluted in saline to the required containing 10 mM NaCl). After equilibration of the ligand−DNA concentrations. It was delivered as an intravenous injection into the complex, the mixture was digested by adding 2 μL of DNase I (about tail vein: Group 1 received 250 μg/kg, Group 2 received 300 μg/kg, × 0.01 U/mL) diluted in NaCl (20 mM), MgCl2 (2 mM) and MnCl2 (2 and Group 3 was the untreated control. The drug was administered 5 mM). The DNase I (Sigma-Aldrich, United Kingdom) was stored at daily for three weeks and then 2× daily in week 4. Tumors were −20 °C as a stock solution of 7200 U/mL in 0.15 M NaCl containing measured using the π-based ellipsoid volume formula62 (length × μ × × π − 1 mM MgCl2. After 1 min the reaction was stopped by adding 4 Lof width height /6) every 3 4 days, and the mice were also DNase I stop solution (10 mM EDTA, 1 mM NaOH, 0.1% weighed and examined at the same time to determine any signs of bromophenol blue, 80% formamide). Before being loaded onto a toxicity from the drug (e.g., animals exhibiting rapid weight loss of gel, the DNA was denatured by incubation at 100 °C for 3 min >20% preoperative weight). All mice were culled when the tumors followed by rapid cooling on ice. The gel consisted of denaturing reached 1.5 cm3. polyacrylamide prepared from 16 mL of Sequagel (National Transcription Factor Profiling Assay. Analysis of the activity of Diagnostics, United Kingdom), 5 mL of 10× TBE buffer containing 48 transcription factors (TFs) was performed according to the 8 M urea, and 27 mL of diluent (50% urea). The gels (40 cm long, 0.3 manufacturer’s instructions using the TF Activation Profiling Plate mm thick) were run at 1500 V for approximately 2 h until the dye Array I (Signosis). Nuclear protein extracts were prepared from 1 × reached the bottom of the gel. The gel plates were then separated, and 107 HeLa cells treated with 7h for 4 h, and also from untreated control the gel fixed by immersion in 10% (v/v) acetic acid before being cells, according to the manufacturer’s instructions using the Nuclear transferred to Whatman 3MM paper and dried in vacuo at 80 °C. The Extraction Kit (Signosis, Inc., Sunnyvale, CA). A 10 μg sample of dried gel was then exposed to a phosphor imager (Kodak) screen nuclear protein extracts was assayed per sample. overnight before scanning. Western Blotting Experiment. CLL cells were washed with PBS Cytotoxicity Studies. MTT Assay. A panel of several types of and lysed by resuspension in lysis buffer consisting of HEPES (50 immortalized human cancer cell lines, including epidermoid (A431), mM), sodium fluoride (5 mM), iodoacetamide (5 mM), sodium lung (A549), ovarian (A2780), breast (MCF7 and MDA-MB-231), chloride (75 mM), NP40 (1%), PMSF (1 mM), sodium orthovanadate and pancreatic (MIA PaCa-2), as well as the lung fibroblast cell line (1 mM), protease inhibitors (Sigma) (1%), phosphatase inhibitor WI-38, were used to determine the cytotoxicity of 7a−h and associated cocktail 2 (Sigma) (1%), and phosphatase inhibitor cocktail 3 (Sigma)

2932 dx.doi.org/10.1021/jm301882a | J. Med. Chem. 2013, 56, 2911−2935 Journal of Medicinal Chemistry Article for 30 min at 4 °C followed by centrifugation at 16000g. Clarified (3) Dervan, P. B. Molecular recognition of DNA by small molecules. lysates were subjected to electrophoresis using NuPage precast 4−12% Bioorg. Med. Chem. 2001, 9, 2215−2235. Bis-Tris gels (Invitrogen, Paisley, U.K.) followed by transfer to PVDF (4) Neidle, S. DNA minor-groove recognition by small molecules. membranes (GE Healthcare UK Ltd., Little Chalfont, U.K.). Nat. Prod. Rep. 2001, 18, 291−309. Immunoblotting was performed with antibodies against phospho- (5) Smellie, M.; Bose, D. S.; Thompson, A. S.; Jenkins, T. C.; IκB, phospho-p65 (New England Biolabs, Hitchin, U.K.), and actin Hartley, J. A.; Thurston, D. E. Sequence-selective recognition of (Abcam, Cambridge, U.K.). duplex DNA through covalent interstrand cross-linking: kinetic and Staining Studies for BCL2 Activity in Xenograft Biopsy molecular modeling studies with pyrrolobenzodiazepine dimers. μ Material. Immunohistochemistry was performed on 10 m thick Biochemistry 2003, 42, 8232−8239. fi cryostat sections of MIA PaCa-2 tumors after they were xed in (6) Wells, G.; Martin, C. R. H.; Howard, P. W.; Sands, Z. A.; acetone for 10 min at room temperature. Following blocking of Laughton, C. A.; Tiberghien, A.; Woo, C. K.; Masterson, L. A.; endogenous peroxidase with 0.3% H2O2/methanol for 15 min at room fi Stephenson, M. J.; Hartley, J. A.; Jenkins, T. C.; Shnyder, S. D.; temperature, sections were rst incubated in 3% normal goat serum/ Loadman, P. M.; Waring, M. J.; Thurston, D. E. Design, synthesis, and phosphase-buffered saline (PBS) for 30 min, followed by 1 h of μ biophysical and biological evaluation of a series of pyrrolobenzodia- incubation with 2 g/mL BCL2 antibody at room temperature. After zepine−poly(N-methylpyrrole) conjugates. J. Med. Chem. 2006, 49, washing with PBS, the sections were then incubated with a horseradish 5442−5461. peroxidase conjugated goat antihuman antibody (DAKO UK Ltd., Ely, (7) Collins, I.; Workman, P. New approaches to molecular cancer U.K.) for 1 h at room temperature, rinsed in PBS, and then incubated − − therapeutics. Nat. Chem. Biol. 2006, 2, 689 700. with the avidin biotin complex (Vector Laboratories Ltd., Orton (8) Dervan, P. B.; Burli, R. W. Sequence-specific DNA recognition by Southgate, U.K.) following the manufacturer’s instructions. The polyamides. Curr. Opin. Chem. Biol. 1999, 3, 688−693. sections were developed with 3,3-diaminobenzidine (Sigma-Aldrich (9) Cipolla, L.; Araujo, A. C.; Airoldi, C.; Bini, D. Pyrrolo[2,1- Co. Ltd., Gillingham, U.K.) and counterstained with hematoxylin. c][1,4]benzodiazepine as a scaffold for the design and synthesis of Finally, the sections were mounted in dibutyl phthalate/xylene and anti-tumour drugs. Anti-Cancer Agents Med. Chem. 2009, 9,1−31. visualized under a light microscope. (10) Thurston, D. E.; Bose, D. S. Synthesis of DNA-interactive − ■ ASSOCIATED CONTENT pyrrolo[2,1-c][1,4]benzodiazepines. Chem. Rev. 1994, 94, 433 465. (11) Antonow, D.; Thurston, D. E. Synthesis of DNA-interactive *S Supporting Information pyrrolo[2,1-c][1,4]benzodiazepines (PBDs). Chem. Rev. 2011, 111, Synthesis of intermediates, FRET, HPLC, and FID data, 2815−2864. footprinting gels, MTD data, and molecular modeling data. (12) Thurston, D. E. Advances in the study of pyrrolo[2,1- This material is available free of charge via the Internet at c][1,4]benzodiazepine (PBD) antitumour antibiotics. In Molecular http://pubs.acs.org Aspects of Anticancer Drug-DNA Interactions, Neidle, S., Waring, M. J., Eds.; The Macmillan Press Ltd.: London, 1993; Vol. 1,pp54−88. ■ AUTHOR INFORMATION (13) Kamal, A.; Rao, M. V.; Laxman, N.; Ramesh, G.; Reddy, G. S. K. Recent developments in the design, synthesis and structure-activity Corresponding Author relationship studies of pyrrolo[2,1-c][1,4]benzodiazepines as DNA- *Phone: +44 (0)207 848 4279 (D.E.T.). E-mail: david. interactive antitumour antibiotics. Curr. Med. Chem.: Anti-Cancer [email protected] (D.E.T.). Agents 2002, 2, 215−254. Notes (14)Rahman,K.M.;Thompson,A.S.;James,C.H.; The authors declare no competing financial interest. Narayanaswamy, M.; Thurston, D. E. The pyrrolobenzodiazepine dimer SJG-136 forms sequence-dependent intrastrand DNA cross- ■ ACKNOWLEDGMENTS links and monoalkylated adducts in addition to interstrand cross-links. J. Am. Chem. Soc. 2009, 131, 13756−13766. Commonwealth Commission UK (Grant BDCA 05/01 to (15) Gregson, S. J.; Howard, P. W.; Hartley, J. A.; Brooks, N. A.; K.M.R.) and Cancer Research UK (CRUK; Grant C180/A1060 Adams, L. J.; Jenkins, T. C.; Kelland, L. R.; Thurston, D. E. Design, to D.E.T.) are acknowledged for financial support for part of synthesis, and evaluation of a novel pyrrolobenzodiazepine DNA- this work. interactive agent with highly efficient cross-linking ability and potent cytotoxicity. J. Med. Chem. 2001, 44, 737−748. ■ ABBREVIATIONS USED (16) Rahman, K. M.; James, C. H.; Bui, T. T. T.; Drake, A. F.; Thurston, D. E. Observation of a single-stranded DNA/pyrrolobenzo- Alloc, (allyloxy)carbonyl; CLL, chronic lymphocytic leukemia; − ′ diazepine adduct. J. Am. Chem. Soc. 2011, 133, 19376 19385. DIC, N,N -diisopropylcarbodiimide; DIPEA, N,N-diisopropy- (17) Rahman, K. M.; James, C. H.; Thurston, D. E. Effect of base lethylamine; DMAP, 4-(dimethylamino)pyridine; EDCI, 1- sequence on the DNA cross-linking properties of pyrrolobenzodiaze- ethyl-3-(3-(dimethylamino)propyl)carbodiimide); FID, fluores- pine (PBD) dimers. Nucleic Acids Res. 2011, 39, 5800−5812. cent intercalator displacement; FRET, fluorescence resonance (18) Janjigian, Y. Y.; Lee, W.; Kris, M. G.; Miller, V. A.; Krug, L. M.; energy transfer; HOBT, hydroxybenzotriazole; MPB, 4-(1- Azzoli, C. G.; Senturk, E.; Calcutt, M. W.; Rizvi, N. A. A phase I trial of methyl-1H-pyrrol-3-yl)benzenamine; MTD, maximum toler- SJG-136 (NSC#694501) in advanced solid tumors. Cancer Chemother. ated dose; MTT, 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenylte- Pharmacol. 2010, 65, 833−838. trazolium bromide; NF-κB, nuclear factor κ B; PBD, pyrrolo- (19) Narendran, A.; Jayanthan, A.; Singh, A.; Incoronato, A.; Desai, [2,1-c][1,4]benzodiazepine; Pu-G-Pu, purine-guanine-purine; S.; Whitlock, J. SJG-136 (NSC 694501), a unique DNA minor groove interstrand cross-linking agent shows significant activity and drug Py-G-Py, pyrimidine-guanine-pyrimidine; THP, tetrahydropyr- combinability against aggressive type of pediatric leukemia. Pediatr. an-2-yl Blood Cancer 2010, 54, 814−814. (20) Hochhauser, D.; Meyer, T.; Spanswick, V. J.; Wu, J.; Clingen, P. ■ REFERENCES H.; Loadman, P. M.; Cobb, M.; Gumbrell, L.; Begent, R. H. J.; Hartley, (1) Thurston, D. E. Chemistry and Pharmacology of Anticancer Drugs; J. A.; Jodrell, D. I. Phase I study of sequence-selective minor groove CRC Press (Taylor & Francis): Boca Raton, FL, 2006; Vol. 1, pp 283. DNA binding agent SJG-136 in patients with advanced solid tumors. (2) Dickinson, L. A.; Burnett, R.; Melander, C.; Edelson, B. S.; Arora, Clin. Cancer Res. 2009, 15, 2140−2147. P. S.; Dervan, P. B.; Gottesfeld, J. M. Arresting cancer proliferation by (21) Rahman, K. M.; James, C. H.; Thurston, D. E. Observation of small-molecule gene regulation. Chem. Biol. 2004, 11, 1583−1594. the reversibility of a covalent pyrrolobenzodiazepine (PBD) DNA

2933 dx.doi.org/10.1021/jm301882a | J. Med. Chem. 2013, 56, 2911−2935 Journal of Medicinal Chemistry Article adduct by HPLC/MS and CD spectroscopy. Org. Biomol. Chem. 2011, efficiency in oligonucleotides of varying sequences. Anal. Biochem. 9, 1632−1641. 2008, 374, 173−181. (22) Rahman, K. M.; Vassoler, H.; James, C. H.; Thurston, D. E. (38) Organ MG, D. C.; Mayhew, D.; Parks, D. J.; Arvanitis, E. A. The DNA sequence preference and adduct orientation of pyrrolo[2,1- use of a supported base and strong cation exchange (SCX) c][1,4]benzodiazepine antitumor agents. ACS Med. Chem. Lett. 2010, chromatography to prepare a variety of structurally-diverse molecular 1, 427−432. libraries prepared by solution-phase methods. Comb. Chem. High (23) Kamal, A.; Rajender; Reddy, D. R.; Reddy, M. K.; Balakishan, Throughput Screening 2002, 5, 211−218. G.;Shaik,T.B.;Chourasia,M.;Sastry,G.N.Remarkable (39) Tse, W. C.; Boger, D. L. A fluorescent intercalator displacement enhancement in the DNA-binding ability of C2-fluoro substituted assay for establishing DNA binding selectivity and affinity. Acc. Chem. pyrrolo[2,1-c][1,4]benzodiazepines and their anticancer potential. Res. 2004, 37,61−69. Bioorg. Med. Chem. 2009, 17, 1557−1572. (40) Jenkins, T. C.; Hurley, L. H.; Neidle, S.; Thurston, D. E. (24) Puvvada, M. S.; Hartley, J. A.; Jenkins, T. C.; Thurston, D. E. A Structure of a covalent DNA minor-groove adduct with a quantitative assay to measure the relative DNA-binding affinity of pyrrolobenzodiazepine dimer: evidence for sequence-specific inter- pyrrolo[2,1-c][1,4]benzodizepine (PBD) antitumour antibiotics based strand cross-linking. J. Med. Chem. 1994, 37, 4529−4537. on the inhibition of restriction endonuclease BamHI. Nucleic Acids Res. (41) Hampshire, A. J.; Fox, K. R. Preferred binding sites for the 1993, 21, 3671−3675. bifunctional intercalator TANDEM determined using DNA fragments (25) Puvvada, M. S.; Forrow, S. A.; Hartley, J. A.; Stephenson, P.; that contain every symmetrical hexanucleotide sequence. Anal. Gibson, I.; Jenkins, T. C.; Thurston, D. E. Inhibition of bacteriophage Biochem. 2008, 374, 298−303. T7 RNA polymerase in vitro transcription by DNA-binding pyrrolo- (42) Thurston, D. E.; Murty, V. S.; Langley, D. R.; Jones, G. B. O- − [2,1-c][1,4]benzodiazepines. Biochemistry 1997, 36, 2478 2484. Debenzylation of a pyrrolo[2,1-c][1,4]benzodiazepine in the presence ’ (26) Kotecha, M.; Kluza, J.; Wells, G.; O Hare, C. C.; Forni, C.; of a carbinolamine functionality: synthesis of DC-81. Synthesis 1990, Mantovani, R.; Howard, P. W.; Morris, P.; Thurston, D. E.; Hartley, J. 81−84. A.; Hochhauser, D. Inhibition of DNA binding of the NF-Y (43) Skehan P, S. R.; Scudiero, D.; Monks, A.; McMahon, J.; Vistica, transcription factor by the pyrrolobenzodiazepine-polyamide conjugate D.; Warren, J. T.; Bokesch, H.; Kenney, S.; Boyd, M. R. New − GWL-78. Mol. Cancer Ther. 2008, 7, 1319 1328. colorimetric cytotoxicity assay for anticancer-drug screening. J. Natl. (27) Hsieh, M.-C.; Hu, W.-P.; Yu, H.-S.; Wu, W.-C.; Chang, L.-S.; Cancer Inst. 1990, 82, 1107−1112. Kao, Y.-H.; Wang, J.-J. A DC-81-indole conjugate agent suppresses (44) van Engeland, M.; Nieland, L. J. W.; Ramaekers, F. C. S.; melanoma A375 cell migration partially via interrupting VEGF Schutte, B.; Reutelingsperger, C. P. M. Annexin V-affinity assay: a production and stromal cell-derived factor-1α-mediated signaling. − review on an apoptosis detection system based on phosphatidylserine Toxicol. Appl. Pharmacol. 2011, 255, 150 159. exposure. Cytometry 1998, 31,1−9. (28) Andreassen, P. R.; Ren, K. Q. Fanconi anemia proteins, DNA (45) Hartley, J. A.; Spanswick, V. J.; Brooks, N.; Clingen, P. H.; interstrand crosslink repair pathways, and cancer therapy. Curr. Cancer − McHugh, P. J.; Hochhauser, D.; Pedley, R. B.; Kelland, L. R.; Alley, M. Drug Targets 2009, 9, 101 117. C.; Schultz, R.; Hollingshead, M. G.; Schweikart, K. M.; Tomaszewski, (29) Hartley, J.; Hamaguchi, A.; Suggitt, M.; Gregson, S.; Thurston, J. E.; Sausville, E. A.; Gregson, S. J.; Howard, P. W.; Thurston, D. E. D.; Howard, P. DNA interstrand cross-linking and in vivo antitumor SJG-136 (NSC 694501), a novel rationally designed DNA minor activity of the extended pyrrolo[2,1-c][1,4]benzodiazepine dimer groove interstrand cross-linking agent with potent and broad spectrum SG2057. Invest. New Drugs 2012, 30, 950−958. antitumor activity. Cancer Res. 2004, 64, 6693−6699. (30) Osada, H.; Uramoto, M.; Uzawa, J.; Kajikawa, K.; Isono, K. New (46) Pepper, C. J.; Hambly, R. M.; Fegan, C. D.; Delavault, P.; pyrrolobenzodiazepine antibiotics, RK-1441 A and B. II. Isolation and Thurston, D. E. The novel sequence-specific DNA cross-linking agent structure. Agric. Biol. Chem. 1990, 54, 2883−2887. (31)Hadjivassileva,T.;Thurston,D.E.;Taylor,P.W. SJG-136 (NSC 694501) has potent and selective in vitro cytotoxicity in human B-cell chronic lymphocytic leukemia cells with evidence of a Pyrrolobenzodiazepine dimers: novel sequence-selective, DNA-inter- − active, cross-linking agents with activity against Gram-positive bacteria. p53-independent mechanism of cell kill. Cancer Res. 2004, 64, 6750 J. Antimicrob. Chemother. 2005, 56, 513−518. 6755. (32) Fotso, S. Naturally occuring pyrrolo[1,4]benzodiazepines in (47) Pepper, C.; Lowe, H.; Fegan, C.; Thurieau, C.; Thurston, D. E.; bacteria. Mini-Rev. Org. Chem. 2010, 7,68−74. Hartley, J. A.; Delavault, P. Fludarabine-mediated suppression of the (33) Fotso, S.; Zabriskie, T. M.; Proteau, P. J.; Flatt, P. M.; Santosa, excision repair enzyme ERCC1 contributes to the cytotoxic synergy D.A.;Sulastri;Mahmud,T.LimazepinesA−F, pyrrolo[1,4]- with the DNA minor groove crosslinking agent SJG-136 (NSC benzodiazepine antibiotics from an Indonesian Micrococcus sp. J. Nat. 694501) in chronic lymphocytic leukaemia cells. Br. J. Cancer 2007, 97, − Prod. 2009, 72, 690−695. 253 259. (34) Rahman, K. M.; Rosado, H.; Moreira, J. B.; Feuerbaum, E.-A.; (48) Hu, W.-P.; Tsai, F.-Y.; Yu, H.-S.; Sung, P.-J.; Chang, L.-S.; − Fox, K. R.; Stecher, E.; Howard, P. W.; Gregson, S. J.; James, C. H.; de Wang, J.-J. Induction of apoptosis by DC-81 indole conjugate agent κ la Fuente, M.; Waldron, D. E.; Thurston, D. E.; Taylor, P. W. through NF- B and JNK/AP-1 pathway. Chem. Res. Toxicol. 2008, 21, − Antistaphylococcal activity of DNA-interactive pyrrolobenzodiazepine 1330 1336. (PBD) dimers and PBD-biaryl conjugates. J. Antimicrob. Chemother. (49) Brucoli, F.; Hawkins, R. M.; James, C. H.; Wells, G.; Jenkins, T. 2012, 67, 1683−1696. C.; Ellis, T.; Hartley, J. A.; Howard, P. W.; Thurston, D. E. Novel C8- (35) Rosado, H.; Rahman, K. M.; Feuerbaum, E.-A.; Hinds, J.; linked pyrrolobenzodiazepine (PBD)-heterocycle conjugates that Thurston, D. E.; Taylor, P. W. The minor groove-binding agent ELB- recognize DNA sequences containing an inverted CCAAT box. Bioorg. 21 forms multiple interstrand and intrastrand covalent cross-links with Med. Chem. Lett. 2011, 21, 3780−3783. duplex DNA and displays potent bactericidal activity against (50) Borgatti, M.; Rutigliano, C.; Bianchi, N.; Mischiati, C.; Baraldi, methicillin-resistant Staphylococcus aureus. J. Antimicrob. Chemother. P. G.; Romagnoli, R.; Gambari, R. Inhibition of NF-kB/DNA 2011, 66, 985−996. interactions and HIV-1 LTR directed transcription by hybrid (36) Boger, D. L.; Fink, B. E.; Brunette, S. R.; Tse, W. C.; Hedrick, molecules containing pyrrolo [2,1-c] [1,4] benzodiazepine (PRD) M. P. A simple, high-resolution method for establishing DNA binding and oligopyrrole carriers. Drug Dev. Res. 2003, 60, 173−185. affinity and sequence selectivity. J. Am. Chem. Soc. 2001, 123, 5878− (51) Baraldi, P. G.; Cacciari, B.; Guiotto, A.; Romagnoli, R.; Spalluto, 5891. G.; Leoni, A.; Bianchi, N.; Feriotto, G.; Rutigliano, C.; Mischiati, C.; (37) Narayanaswamy, M.; Griffiths, W. J.; Howard, P. W.; Thurston, Gambari, R. [2,1-c][1,4]Benzodiazepine (PBD)-distamycin hybrid D. E. An assay combining high-performance liquid chromatography inhibits DNA binding to transcription factor Sp1. Nucleosides, and mass spectrometry to measure DNA interstrand cross-linking Nucleotides Nucleic Acids 2000, 19, 1219−1229.

2934 dx.doi.org/10.1021/jm301882a | J. Med. Chem. 2013, 56, 2911−2935 Journal of Medicinal Chemistry Article

(52) Harhaj, E. W.; Sun, S. C. IKKγ serves as a docking subunit of the IκB kinase (IKK) and mediates interaction of IKK with the human T- cell leukemia virus Tax protein. J. Biol. Chem. 1999, 274, 22911− 22914. (53) Mori, N.; Fujii, M.; Ikeda, S.; Yamada, Y.; Tomonaga, M.; Ballard, D. W.; Yamamoto, N. Constitutive activation of NF-κBin primary adult T-cell leukemia cells. Blood 1999, 93, 2360−2368. (54) Pepper, C.; Hewamana, S.; Brennan, P.; Fegan, C. NF-κBasa prognostic marker and therapeutic target in chronic lymphocytic leukemia. Future Oncol. 2009, 5, 1027−1037. (55) Karin, M.; Lin, A. NF-κB at the crossroads of life and death. Nat. Immunol. 2002, 3, 221−227. (56) Wang, C.-Y.; Cusack, J. C.; Liu, R.; Baldwin, A. S. Control of inducible chemoresistance: enhanced anti-tumor therapy through increased apoptosis by inhibition of NF-κB. Nat. Med. 1999, 5, 412− 417. (57) Baird, E. E.; Dervan, P. B. Solid phase synthesis of polyamides containing imidazole and pyrrole amino acids. J. Am. Chem. Soc. 1996, 118, 6141−6146. (58) Case, D. A.; Darden, T. A.; Cheatham, T. E., III; Simmerling, C. L.; Wang, J.; Duke, R. E.; Luo, R.; Walker, R. C.; Zhang, W.; Merz, K. M.; Roberts, B.; Wang, B.; Hayik, S.; Roitberg, A.; Seabra, G.; Kolossvary,́ I.; Wong, K. F.; Paesani, F.; Vanicek, J.; Liu, J.; Wu, X.; Brozell, S. R.; Steinbrecher, T.; Gohlke, H.; Cai, Q.; Ye, X.; Wang, J.; Hsieh, M.-J.; Cui, G.; Roe, D. R.; Mathews, D. H.; Seetin, M. G.; Sagui, C.; Babin, V.; Luchko, T.; Gusarov, S.; Kovalenko, A.; Kollman, P. A. AMBER 11; University of California: San Francisco, 2010. (59) Pettersen, E. F.; Goddard, T. D.; Huang, C. C.; Couch, G. S.; Greenblatt, D. M.; Meng, E. C.; Ferrin, T. E. UCSF chimeraa visualization system for exploratory research and analysis. J. Comput. Chem. 2004, 25, 1605−1612. (60) Hampshire, A. J.; Rusling, D. A.; Broughton-Head, V. J.; Fox, K. R. Footprinting: a method for determining the sequence selectivity, affinity and kinetics of DNA-binding ligands. Methods 2007, 42, 128− 140. (61) Workman, P.; Aboagye, E. O.; Balkwill, F.; Balmain, A.; Bruder, G.; Chaplin, D. J.; Double, J. A.; Everitt, J.; Farningham, D. A. H.; Glennie, M. J.; Kelland, L. R.; Robinson, V.; Stratford, I. J.; Tozer, G. M.; Watson, S.; Wedge, S. R.; Eccles, S. A. Guidelines for the welfare and use of animals in cancer research. Br. J. Cancer 2010, 102, 1555− 1577. (62) Tomayko, M. M.; Reynolds, C. P. Determination of subcutaneous tumor size in athymic (nude) mice. Cancer Chemother. Pharmacol. 1989, 24, 148−154.

2935 dx.doi.org/10.1021/jm301882a | J. Med. Chem. 2013, 56, 2911−2935