**4. Bisindole Alkaloid Inhibitors of Methicillin-Resistant** *Staphylococcus aureus* **Pyruvate Kinase from the South African Marine Sponge** *Topsentia* **sp.**

Methicillin-resistant *Staphylococcus aureus* (MRSA), euphemistically also referred to as the "super bug", was initially encountered in public healthcare facilities and remains a significant cause of mortality in these facilities. MRSA is no longer confined to healthcare facilities and has been increasingly reported from the general population and domestic livestock worldwide [25–27]. Annually, MRSA accounts for *ca.* 94,000 infections and 18,000 deaths in the USA and 150,000 infections in the EU [27]. There are no available MRSA mortality data from Southern Africa. However, a 2015 study conducted in three South African academic hospitals reported a MRSA prevalence rate (MRSA infections as a % of all recorded *S. aureus* infections) of 36%, which is comparable to Israel (33.5%) Ireland (38.1%) and the U.K. (35.5%) [28]. The escalating infection and mortality rates associated with the ongoing spread of drug-resistant pathogenic bacteria, e.g., MRSA, are further exacerbated by the dearth of new antibiotics entering the clinic [29].

Paradoxically, the targeting of bacteria-specific proteins in new antibacterial drug development programs is problematic given the concomitant selective pressure that drugs, emerging from this classic drug discovery approach, exert on the pathogens, leading to the proliferation of drug-resistant bacterial strains [30]. Protein target-based antibiotic drug discovery is, however, not redundant. Contemporary genomic and proteomic studies of MRSA [31,32] have increased our understanding of the complex protein-protein interaction networks (interactomes) in this organism. The detailed mapping of interactomes has led to the identification of highly-connected hub proteins, which, given their centrality within the interactomes, are essential for mediating key cellular processes and sustaining MRSA viability [30,31]. Out of necessity, hub proteins are evolutionarily-conserved proteins, given the deleterious effect that mutations of hub proteins would have on the complex interactomes in which they play a key role [30]. Therefore, targeting hub proteins within the MRSA interactomes will minimize the potential for the emergence of drug resistance in MRSA and is a novel strategy for developing much needed new chemotherapeutic interventions against this drug-resistant pathogen [31]. Amongst the suite of hub proteins in a 608-protein interactome network (comprising 23% of the proteome in a hospital-acquired strain of MRSA), Zoraghi *et al.* identified pyruvate kinase (PK) as a suitable target for possible antibiotic drug discovery [30]. Catalyzing the rate-limiting irreversible conversion of phosphoenolpyruvate into pyruvate during glycolysis, pyruvate kinases are, not surprisingly, ubiquitous in both prokaryotes and eukaryotes. Fortuitously, the MRSA PK homotetramer (Figure 6a) has several possible lipophilic binding pockets that are absent in human PK orthologs, allowing potential selective inhibition of this enzyme target [33]. Initially, two parallel strategies were used to generate lead compounds to exploit the inhibition of this key enzyme. The first strategy involved the random screening of >900 marine invertebrate extracts, including those from South African marine invertebrates, for selective MRSA PK inhibition. The second rational drug design strategy coupled knowledge of the detailed structure of the MRSA PK enzyme binding site with contemporary computer-aided drug design techniques to generate new synthetic MRSA PK inhibitors. Both strategies are reviewed in more detail below.

The random screening of 968 marine invertebrate extracts, collected from seven different benthic marine environments around the world, [34], afforded only one extract that was active in the MRSA PK inhibition assay. The methanolic extract of the South African sponge, *Topsentia pachastrelloides* (Figure 1c), showed significant activity in the MRSA PK inhibition assay, and subsequent bioassay-guided fractionation of this extract yielded a cohort of four bisindole alkaloids of which, the two known metabolites *cis*-3,4-dihydrohamacanthin B (**25**, Figure 7) and bromodeoxytopsentin (**26**) proved to be the most active compounds (IC50 16 and 60 nM, respectively). These two compounds also exhibited between 166- and 600-fold selectivity for MRSA PK when compared to similar inhibition data acquired from screening **25** and **26** against four human PK orthologs. X-ray crystallographic analysis of the co-crystallized *cis*-3,4-dihydrohamacanthin B-MRSA PK complex revealed that **25** was neither bound to the recognized activation nor allosteric effector binding sites on this enzyme, but

was instead unexpectedly bound to two identical lipophilic binding sites on the small interface of the MRSA PK homotetramer [34].

**Figure 6.** (**a**) X-ray structure of MRSA PK (PDB Accession Number 3T07) with the large and small interfaces and the *cis*-3,4-dihydrohamacanthin B binding sites indicated; (**b**) X-ray co-crystal generated diagram of **25** (green) in the *cis*-3,4-dihydrohamacanthin B binding site of MRSA PK with **26** (grey) overlaid in its highest scoring docked conformation with its bromines displayed as CPK models; (**c**) highest scoring docked conformation of the synthetic compound **33**, with its chlorines displayed as CPK models, in the MRSA PK *cis*-3,4-dihydrohamacanthin B binding site.

The small interface in MRSA PK is postulated to be crucial for establishing the rigidity of MRSA PK necessary for catalytic activity, and the binding of either **25** or **26** to this region of the protein is therefore thought to induce flexibility and, subsequently, to reduce enzyme activity [33]. The symmetrical *cis*-3,4-dihydrohamacanthin B binding sites are characterized by two lipophilic pockets with an appropriate spatial arrangement to readily accommodate the bromine substituents of **25** and **26** (Figure 6b). In addition, the histidine residues (HIS365) on the neighboring parallel MRSA PK α-helices rearrange to anchor the indole rings through π-interactions [34,35]. Interestingly, sequence alignment between MRSA and human PK isoforms indicated that access to the analogous binding sites in human PK orthologs is hindered by a group of amino acids that effectively shield these sites from potential ligands [33]. This structural difference around the entrance to the binding sites is consequently thought to account for the greater selectivity of **25** and **26**, and related synthetic inhibitors *vide infra*, for MRSA PK over human PK orthologs [33,35].

**Figure 7.** Chemical structures of compounds **25**–**33**.

From the preliminary screening of an *in silico* library, coupled with functional enzyme assays, Zoraghi *et al.* identified the benzimidazole compound IS-130 (**27**, Figure 7) as a potent MRSA PK inhibitor (IC50 91 nM) with good specificity (>1000-fold) for MRSA PK over human isoforms, but with poor *in vivo* antibacterial activity at the cellular level against a methicillin-susceptible strain of *S. aureus* (MSSA) (minimum inhibitory concentration (MIC) >187 μg/mL) [30]. Nevertheless, the structural motif of **27** provided the starting point for a medicinal chemistry program aimed at improving selectivity, potency and antibacterial activity of potential MRSA PK inhibitors. Compound AM-168 (**28**) exhibited only slightly reduced potency (IC50 126 nM) and substantially increased antibacterial activity (MIC 9.7 μg/mL), which was attributed by Zoraghi *et al.* to increased cell membrane penetration due to the increased lipophilicity imparted by the C11 ethyl substituent to this compound [30]. Accordingly, additional alkyl substitution, e.g., NSK5-15 (**29**), further enhanced antibacterial activity against several strains of MSSA and MRSA (MIC 1.4–5.8 μg/mL) with a further decrease in MRSA PK inhibitory potency (185 nM). Kumar *et al.* [33] extended Zoraghi *et al.*'s preliminary study and prepared a series of >70 compounds in which systematic structural changes were made to the heteroaromatic ring, the phenolic moiety and the central linker unit of the hit compound **27**. This series of compounds was screened against MRSA PK and methicillin-susceptible *S. aureus*, with Kumar *et al.* reporting varying levels of potency (IC50 15–380 nM) and antibacterial activity (MIC 1–>194 μg/mL), respectively. Interestingly, co-crystallization of **27** and **28** with MRSA PK followed by X-ray analysis revealed that these compounds were also bound to the *cis*-3,4-dihydrohamacanthin B binding sites of the MRSA PK enzyme (Figure 6a) [33]. Unfortunately, Kumar *et al.*'s synthetic program did not shed any light on the structure activity relationships that might conclusively link potency (IC50) with antibacterial activity (MIC). Ultimately, *N*-methylindole (**30**) provided the best combination of *in vitro* MRSA PK inhibition (IC50 79 nM) and antibacterial activity (MIC 1 μg/mL), possibly warranting further exploitation of **30** and analogous compounds as potential antibiotics effective against MRSA [33].

Veale *et al.* [35] used the chemical structure of the naturally-occurring hit compound **26** identified in the South African sponge extract as a starting point for an extensive ligand-receptor docking study of various analogs of **26** with the *cis*-3,4-dihydrohamacanthin B binding site. Postulating that a dihalogenated analog of **26** would better exploit the opportunities offered by the symmetrical *cis*-3,4-dihydrohamacanthin B binding site, in particular the two terminal lipophilic binding pockets, Veale *et al.* prepared the 6 , 6"dihalogenated (F, Cl, Br and I, Figure 7) analogues of **26** and the debrominated compound, deoxytopsentin (**31**), for a comparative MRSA PK inhibition study. The target halogenated synthetic compounds were readily accessed in reasonable overall yield (10%–32% over five steps) via the dehydrative cyclocondensation of the respective *N*-Boc-protected 6-halo-indolyl-3-glyoxals with ammonium acetate in ethanol, followed by thermolytic cleavage of

the *N*-Boc groups. As expected, the MRSA inhibition activity of the non-halogenated compound **31** (IC50 240 nM) was less active than the naturally-occurring monobrominated compound **26** (IC50 60 nM), while both the dibrominated and dichlorinated (Figure 6c) analogs (**32** and **33**) were an order of magnitude more potent (IC50 2 and 1.5 nM, respectively) than **26** coupled to improved selectivity for MRSA PK over the four human orthologs assayed [35].

Veale *et al.* further evaluated the importance of the imidazole ring to MRSA PK inhibition in dihalogenated bisindole alkaloids by preparing a similar series of dihalogenated bisindoles in which the imidazole ring was replaced by a thiazole moiety, e.g., **34** [36] (Figure 8). Coupling of α-oxo-1*H*-indole-3-thioacetamide (**35**) with the α-bromoketone (**36**) in a regiospecific Hantzsch thiazole ring formation reaction afforded the targeted halogenated bisindole thiazoles. The μM activity of the synthetic bisindole thiazoles (e.g., IC50 5 μM for **35**) indicated that bioisosteric replacement of the imidazole ring with a thiazole had a negative impact on MRSA PK potency [36]. The antibacterial activity of both the synthetic bisindole imidazoles and bisindole thiazoles was not recorded.

**Figure 8.** Chemical structures of compounds **34**–**36**.

Furthermore, acknowledging the significance of a bisindole motif to increased inhibition of MRSA PK, Sperry and co-workers have recently described the selective MRSA PK inhibition of a cohort of eleven, variously-substituted, synthetic 1,2-bis(3-indolyly)ethane compounds, e.g., **37** [37] (Figure 9). Compound **37**, accessed via palladium catalyzed heteroannulation of the aldehyde (**38**) and 1-iodo-2-amino-4-nitrobenzene (**39**) [38], was the most potent of the series (IC50 0.9 μM) and exhibited a 20–106-fold selectivity for MRSA PK over four human PK isoforms. Replacing the C6 nitro substituent with chloro, nitrile, methoxy and methyl functionalities had a deleterious effect on MRSA PK inhibition, with the nitrile and methoxy analogs inactive and the chloro and methyl analogs two orders of magnitude less active (IC50 272 and 294 μM, respectively) [37].

**Figure 9.** Chemical structures of compounds **37**–**39**.

Similarly, Kumar *et al.* [39] changed direction from their earlier medicinal chemistry program based on the putative MRSA PK inhibitor, **27**, to focus on potential bisindole inhibitors of MRSA PK in line with the chemical structures of the potent naturally-occurring bisindole MRSA PK inhibitors, **25** and **26**. Central to their strategy was varying the linker units between the indole rings in order to uncover the relationship between activity and indole orientation relative to the linker unit in addition to providing further insight into size constraints within the MRSA PK binding site. Their initial cohort of directly-linked 2,2 -biindoles (**40**–**42**, Figure 10) was synthesized through a Suzuki–Miyaura coupling of boronic acid precursor **43** and iodinated indole **44**. This cohort of compounds were generally found to potently inhibit MRSA PK at concentrations as low as 1 nM. Initial inhibitory data obtained by Kumar *et al.* supported previous observations made between deoxytopsentin analogues **26** and

**32** that 6-6 dibrominated bisindoles, such as 40 (IC50 7 nM), display superior MRSA PK inhibition than a corresponding monobrominated analogue, e.g., **41** (IC50 21 nM). However, the opposite trend was observed with regard to MIC values (16 and 2 μg/mL, respectively) against MSSA strains [39]. Interestingly, the 6,5 dibrominated analogue (**42**) also displayed potent MRSA PK inhibition (IC50 2.2 nM) coupled to a significantly improved MIC against *S. aureus* (0.3 μg/mL). The structure activity relationship of the linker group between indoles was further explored through insertion of acetylene, ethylene and ethyl moieties between the two substituted indole rings (**45**–**48**, Figure 10) using standard synthetic protocols. Similar MPSA PK inhibitory activity was observed between this group and the 2,2 -biindoles. However, MIC activity against *S. aureus* was generally lost, with the single exception of the 6,5 dibrominated analogue **46**. An additional set of aryl linked bisindole analogues (**49**–**52**) were prepared with a view toward exploiting possible interactions with the aromatic histidine residues present in the binding site. The dibrominated compound (**49**) was found to be comparatively less active than compound **40**, while activity was restored with the mono-brominated compound (**50**), leading the authors to suggest that Compound **50** defines the maximum permissible length within the MRSA PK binding site. Interestingly, analogues of **50** featuring substitutions on the aryl ring (**51**, **52**) were found to be the most active in the series against MRSA PK (IC50 *ca.* 2 nM) while the nitro-containing compound **52** showed encouraging activity against *S. aureus* (MIC 2.0 μg/mL). While no conclusive rationale for the differences between MRSA PK inhibitory activity and MIC values was postulated, Kumar *et al.* determined, through co-administration of their synthetic compounds with the calcium channel blocker verapamil (**53**), that several bisindoles were actively removed from the cells via cellular efflux mechanisms, possibly accounting for the contrasting antibacterial activities observed in their study [39].

**Figure 10.** Chemical structures of compounds **40**–**53**.

### **5. Conclusions**

Interest from Pettit and others in the anti-cancer potential of **1** has remained undiminished for nearly three decades and is likely to continue for the foreseeable future. Realization of the true potential and possible further drug development of the cephalostatins has been hampered by access to commercially-viable synthesis of sufficient quantities of either **1** or similarly-bioactive congeners. Although accessible by laboratory synthesis, potential future drug development interest in **16** will only resume if conflicting cancer cell cytotoxicity data reported for naturally-occurring and synthetic **16** can be explained. The negative impact of drug-resistant pathogens, e.g., MRSA on human health is

steadily increasing, and the need for new antibiotics against these pathogens is continually emphasized. Although the *cis*-3,4-dihydrohamacanthin B binding site of MRSA PK has been identified as a potential selective anti-biotic drug target, resolving the conundrum between potent MRSA PK inhibition and poor *in vivo* MRSA antibacterial activity will define the future of this approach to MRSA antibiotic drug discovery.

**Acknowledgments:** The authors acknowledge with gratitude the ongoing support for South African marine bio-discovery by the South African National Research Foundation, Department of Environmental Affairs (Oceans and Coasts), Medical Research Council, Rhodes University and the University of the Western Cape.

**Conflicts of Interest:** The authors declare no conflict of interest.
