# SUPRAMOLECULAR METAL-BASED ENTITIES FOR BIOMEDICAL AND BIOLOGICAL APPLICATIONS

EDITED BY : Angela Casini and James D. Crowley PUBLISHED IN : Frontiers in Chemistry

#### Frontiers Copyright Statement

© Copyright 2007-2019 Frontiers Media SA. All rights reserved. All content included on this site, such as text, graphics, logos, button icons, images, video/audio clips, downloads, data compilations and software, is the property of or is licensed to Frontiers Media SA ("Frontiers") or its licensees and/or subcontractors. The copyright in the text of individual articles is the property of their respective authors, subject to a license granted to Frontiers.

The compilation of articles constituting this e-book, wherever published, as well as the compilation of all other content on this site, is the exclusive property of Frontiers. For the conditions for downloading and copying of e-books from Frontiers' website, please see the Terms for Website Use. If purchasing Frontiers e-books from other websites or sources, the conditions of the website concerned apply.

Images and graphics not forming part of user-contributed materials may not be downloaded or copied without permission.

Individual articles may be downloaded and reproduced in accordance with the principles of the CC-BY licence subject to any copyright or other notices. They may not be re-sold as an e-book.

As author or other contributor you grant a CC-BY licence to others to reproduce your articles, including any graphics and third-party materials supplied by you, in accordance with the Conditions for Website Use and subject to any copyright notices which you include in connection with your articles and materials.

All copyright, and all rights therein, are protected by national and international copyright laws.

The above represents a summary only. For the full conditions see the Conditions for Authors and the Conditions for Website Use. ISSN 1664-8714 ISBN 978-2-88945-899-8 DOI 10.3389/978-2-88945-899-8

### About Frontiers

Frontiers is more than just an open-access publisher of scholarly articles: it is a pioneering approach to the world of academia, radically improving the way scholarly research is managed. The grand vision of Frontiers is a world where all people have an equal opportunity to seek, share and generate knowledge. Frontiers provides immediate and permanent online open access to all its publications, but this alone is not enough to realize our grand goals.

### Frontiers Journal Series

The Frontiers Journal Series is a multi-tier and interdisciplinary set of open-access, online journals, promising a paradigm shift from the current review, selection and dissemination processes in academic publishing. All Frontiers journals are driven by researchers for researchers; therefore, they constitute a service to the scholarly community. At the same time, the Frontiers Journal Series operates on a revolutionary invention, the tiered publishing system, initially addressing specific communities of scholars, and gradually climbing up to broader public understanding, thus serving the interests of the lay society, too.

## Dedication to Quality

Each Frontiers article is a landmark of the highest quality, thanks to genuinely collaborative interactions between authors and review editors, who include some of the world's best academicians. Research must be certified by peers before entering a stream of knowledge that may eventually reach the public - and shape society; therefore, Frontiers only applies the most rigorous and unbiased reviews.

Frontiers revolutionizes research publishing by freely delivering the most outstanding research, evaluated with no bias from both the academic and social point of view. By applying the most advanced information technologies, Frontiers is catapulting scholarly publishing into a new generation.

### What are Frontiers Research Topics?

Frontiers Research Topics are very popular trademarks of the Frontiers Journals Series: they are collections of at least ten articles, all centered on a particular subject. With their unique mix of varied contributions from Original Research to Review Articles, Frontiers Research Topics unify the most influential researchers, the latest key findings and historical advances in a hot research area! Find out more on how to host your own Frontiers Research Topic or contribute to one as an author by contacting the Frontiers Editorial Office: researchtopics@frontiersin.org

# SUPRAMOLECULAR METAL-BASED ENTITIES FOR BIOMEDICAL AND BIOLOGICAL APPLICATIONS

Topic Editors: Angela Casini, Cardiff University, United Kingdom James D. Crowley, University of Otago, New Zealand

Research in the discovery of metal supramolecular complexes, mainly formed by the self-assembly of inorganic metal compounds with either inorganic or organic molecules via coordination (or organometallic) bonds, is a rapidly developing and newly rising highlight interdisciplinary field. This Reseach Topic is aimed at providing representative examples of supramolecular metal-based entities for different biological and biomedical applications.

Citation: Casini, A., Crowley, J. D., eds. (2019). Supramolecular Metal-Based Entities for Biomedical and Biological Applications. Lausanne: Frontiers Media. doi: 10.3389/978-2-88945-899-8

# Table of Contents

# EDITORIAL

*05 Editorial: Supramolecular Metal-Based Entities for Biomedical and Biological Applications* Angela Casini and James D. Crowley

# REVIEW PAPERS


# ARTICLES


Han Wang, Zihan Qiu, He Liu, Amarasooriya M. D. S. Jayawardhana, Zhizhou Yue, Hala Daghlas, David J. Bowers, Bansidhar Datta and Yao-Rong Zheng

*52 Self-Assembly of a [1 + 1] Ionic Hexagonal Macrocycle and its Antiproliferative Activity*

Khushwant Singh, Ankit Gangrade, Sourav Bhowmick, Achintya Jana, Biman B. Mandal and Neeladri Das

*63 Eukaryotic Cell Toxicity and HSA Binding of [Ru(Me4 phen)(bb7 )]2+ and the Effect of Encapsulation in Cucurbit[10]uril*

Biyun Sun, Ian F. Musgrave, Anthony I. Day, Kirsten Heimann, F. Richard Keene and J. Grant Collins

*75 Supramolecular Fluorescence Probe Based on Twisted Cucurbit[14]uril for Sensing Fungicide Flusilazole*

Ying Fan, Rui-Han Gao, Ying Huang, Bing Bian, Zhu Tao and Xin Xiao

*83 Directed Self-Assembly of Trimeric DNA-Bindingchiral Miniprotein Helicates*

Jacobo Gómez-González, Diego G. Peña, Ghofrane Barka, Giuseppe Sciortino, Jean-Didier Maréchal, Miguel Vázquez López and M. Eugenio Vázquez

*91* Exo*-Functionalized Metallacages as Host-Guest Systems for the Anticancer Drug Cisplatin*

Ben Woods, Margot N. Wenzel, Thomas Williams, Sophie R. Thomas, Robert L. Jenkins and Angela Casini

# *97 Antimicrobial Activity and Cytotoxicity of Ag(I) and Au(I) Pillarplexes*

Alexander Pöthig, Sara Ahmed, Hanne Cecilie Winther-Larsen, Shengyang Guan, Philipp J. Altmann, Jürgen Kudermann, Adriana Magalhães Santos Andresen, Tor Gjøen and Ove Alexander Høgmoen Åstrand

# *105 Stepwise Synthesis of* Tetra*-imidazolium Macrocycles and Their*  N*-Heterocyclic Carbene Metal Complexes*

Zili Li, Nuchareenat Wiratpruk and Peter J. Barnard

# Editorial: Supramolecular Metal-Based Entities for Biomedical and Biological Applications

Angela Casini <sup>1</sup> \* and James D. Crowley <sup>2</sup>

*<sup>1</sup> School of Chemistry, Cardiff University, Cardiff, United Kingdom, <sup>2</sup> Department of Chemistry, University of Otago, Dunedin, New Zealand*

Keywords: supramolecular complex, MOF (Metal-Organic-Framework), metallacages, metallacycle complexes, therapy, drug delivery systems, cancer, self-assembly

**Editorial on the Research Topic**

### **Supramolecular Metal-Based Entities for Biomedical and Biological Applications**

The biomedical and biological application of supramolecular metal-based structures—including the discrete supramolecular coordination complexes (SCCs) and polymeric metal organic frameworks (MOFs)—is an emergent research area. In fact, in the last decade, a number of studies demonstrated the potential of these supramolecular compounds as novel anticancer drugs, drug delivery systems, and imaging agents, as well as for molecular recognition of relevant biomolecules.

The present article collection includes an overview of the examples of discrete 2D- and 3D-supramolecular metal-based systems, with focus on cancer therapy, in the review of Ahmedova. The review covers the specific case of 2D metalacycles, as well as 3D metallacages, helicates, capsules and barrels, and marks out some key tendencies in relation to their biological activity and potential areas of biomedical application.

Of note, the point of view of one of the pioneers in this research area, Therrien, is provided in a review focused on the role of the second coordination sphere, including H-bonding, hydrophobic, electrostatic, and van der Waals interactions, in the biological activity of arene ruthenium metalla-assemblies. These types of interactions are crucial in the design of supramolecular entities implemented for molecular recognition.

Thus, a number of original papers in this issue report on the use of selected metallacages as cytotoxic agents against human cancer cells. For example, Vasdev et al. reported on [Pd2L4] 4+ metallacages featuring bis-quinoline and bis-isoquinoline-based ligands. The bis-quinolone-based cage was found to be helical while the bis-isoquinoline-based system was lantern shaped. The binding of the anticancer drug cisplatin was also examined. It was shown that only the bisisoquinoline cage interacted with the drug, suggesting that the helical twist interferes with the host-guest interaction. Both new cages were shown to be more cytotoxic against cancer cell lines than related pyridyl based systems.

Interestingly, Wang et al. provided the proof-of-concept of using nanoprecipitation to formulate hydrophobic Pt(II)-organic cages.

Along the same line, the antiproliferative effects of a self-assembled hexagonal Pd(II) macrocycle against a small panel of cancer cell lines were found comparable to cisplatin in the paper by Singh et al.

Furthermore, Sun et al. explored the potential of encapsulating Ru(II) polypyridyl complexes in macrocyclic host compounds, namely cucurbit[10]urils (Q[10]), to reduce the toxicity of Ru(II) complexes. The Q[10] encapsulation enhanced the pharmacokinetics of one of the ruthenium complexes which could potentially enable the complex to be used as a new type of antimicrobial

Edited and reviewed by: *Tony D. James, University of Bath, United Kingdom*

> \*Correspondence: *Angela Casini casinia@cardiff.ac.uk*

#### Specialty section:

*This article was submitted to Supramolecular Chemistry, a section of the journal Frontiers in Chemistry*

Received: *02 April 2019* Accepted: *10 April 2019* Published: *26 April 2019*

#### Citation:

*Casini A and Crowley JD (2019) Editorial: Supramolecular Metal-Based Entities for Biomedical and Biological Applications. Front. Chem. 7:293. doi: 10.3389/fchem.2019.00293*

**5**

agent. The host-guest complex between twisted cucurbit[14]uril and thioflavin T was also explored for sensing of non-fluorescent triazole fungicides by Fan et al.

Peptides are highly versatile building blocks, encoding precise structural and functional information within their amino acid sequence, which can be used to design and synthesize novel supramolecular metal-based helicates structures endowed with molecular recognition properties. In the work by Gómez-González et al. the modification of the T4 Fibritin foldon with metal-chelating bipyridines allowed the obtainment of Fe helicates which display good in vitro DNA binding and selectivity toward three-way DNA junctions.

Among the features of supramolecular-based scaffolds, their host-guest chemistry is certainly interesting to exploit them as drug delivery systems or as biosensors. Thus, Woods et al. reported on a series of exo-functionalized [Pd2L4] 4+ metallacages featuring 2,6-bis(pyridin-3-ylethynyl)pyridine) ligands. The encapsulation of the anticancer drug cisplatin in selected cages has been studied using a combination of NMR spectroscopy methods, and the obtained results show that if the solvent is of sufficient polarity, cisplatin encapsulation can easily occur in the hydrophobic cavity of the cage despite the absence of hydrogen-bond accepting cavity-facing residues.

A novel aspect of this research field concerns the use of supramolecular organometallic complexes, whereby the (organic) linker is forming the organometallic bond to the metal node, for therapy. Thus, the antimicrobial and anticancer properties of organometallic Ag(I) and Au(I) pillarplexes were explored in different bacterial and human cancer cells, respectively (Pöthig et al.). The obtained results showed more relevant bioactivity in the case of the silver compounds.

Within this framework, Barnard et al. developed the synthesis of Au(I) and Pd(II) N-heterocyclic carbene (NHC) complexes derived from tetra-imidazolium linked macrocycles for anion sensing. Interestingly, some of the Ag(I) and the analogous Au(I) complexes adopted intriguing hexanuclear structures with the general formula [M6L3] <sup>6</sup>+. It is envisaged that these novel scaffolds could allow the slow release of metal ions for the treatment of bacterial infections.

Overall, as highlighted in the contributions to this special issue, the myriad of possible metal-based supramolecular structures and their almost limitless modularity and tunability suggests that the biomedical applications of these complex chemical entities will continue to emerge.

# AUTHOR CONTRIBUTIONS

All authors listed have made a substantial, direct and intellectual contribution to the work, and approved it for publication.

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Casini and Crowley. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Biomedical Applications of Metallosupramolecular Assemblies—Structural Aspects of the Anticancer Activity

### Anife Ahmedova\*

*Laboratory of Biocoordination and Bioanalytical Chemistry, Faculty of Chemistry and Pharmacy, Sofia University, Sofia, Bulgaria*

The design and development of metallosupramolecular systems has resulted in construction of a myriad of fascinating structures with highly diverse properties and potential applications. Assessment of the biomedical applications of metallosupramolecular assemblies is an emerging field of research that stems from the recently demonstrated promising results on such systems. After the pioneering works of Therrien and coworkers on organometallic Ru-cages with promising anticancer properties, this topic has evolved to the more recent studies on bioactivity of supramolecular coordination complexes built from different metal ions and various multidentate ligands. Sufficient amount of data on the anticancer activity of metallosupramolecules has already been reported and allows outlining some general tendencies in the structural aspects of the biological activity. The main structural properties of the complexes that can be readily modified to enhance their activity are the size, the shape and charge of the formed complexes. Moreover, the intrinsic properties of the building components could predetermine some of the main characteristics of the overall supramolecular complex, such as its optical properties, chemical reactivity, solubility, etc., and could, thereby, define the areas of its biomedical applications. The unique structural property of most of the metallosupramolecular assemblies, however, is the presence of a discrete cavity that renders a whole range of additional applications resulting from specific host-guest interactions. The encapsulations of small bioactive or fluorescent molecules have been employed for delivery or recognition purposes in many examples. On the other hand, metallosupramolecules have been imbedded into target-specific polymeric nanoparticles that resulted in a successful combination of their therapeutic and diagnostic properties, making them promising for theranostic application in cancer treatment. The aim of this review paper is to mark out some key tendencies in the reported metallosupramolecular structures in relation with their biological activity and potential areas of biomedical application. In this way, a useful set of guidelines can be delineated to help synthetic chemists broaden the application areas of their supramolecular systems by few structural changes.

Keywords: metallacycle, metallacage, anticancer, capsule, encapsulation

#### Edited by:

*James D. Crowley, University of Otago, New Zealand*

#### Reviewed by:

*Yaorong Zheng, Kent State University, United States Dong-Sheng Guo, Nankai University, China*

> \*Correspondence: *Anife Ahmedova ahmedova@chem.uni-sofia.bg*

#### Specialty section:

*This article was submitted to Supramolecular Chemistry, a section of the journal Frontiers in Chemistry*

Received: *30 September 2018* Accepted: *30 November 2018* Published: *14 December 2018*

#### Citation:

*Ahmedova A (2018) Biomedical Applications of Metallosupramolecular Assemblies—Structural Aspects of the Anticancer Activity. Front. Chem. 6:620. doi: 10.3389/fchem.2018.00620*

**7**

# INTRODUCTION

The coordination bond directed self-assembly processes have been broadly exploited in building metallosupramolecular species with diverse composition (Bilbeisi et al., 2014), discrete size and shape (Schmidt et al., 2014; Cook and Stang, 2015), and variety of useful properties. Unarguably, the aesthetics of these highly symmetric metallosupramolecules has been the initial source of inspiration for many chemists. This resulted in not only building a myriad of fascinating structures but also in establishing the main principles in the design and synthesis of metallosupramolecular complexes with predefined shapes by exploiting the bond directionality and the coordination properties of the metal ions, and intelligent design of the linking organic ligands (Datta et al., 2018). Nowadays, the main targets are the specific properties of the metallosupramolecular complexes in view of their practical applications that range from chemical catalysis to multi-purpose and stimuli-responsive advanced materials (McConnell et al., 2015). The demonstrated high potential to purposefully combine the properties of metal ions with that of organic ligands has recently attracted the attention of bioinorganic chemists which added up more facets to the range of practical use of the metallosupramolecules their biological application for therapy and/or diagnostics (Casini et al., 2017; Leung et al., 2017).

The continuous studies related to the clinical use of Ptcomplexes as anticancer drugs have evolved to establishing structure-activity relationships for this class of compounds. It is, therefore, not surprising that the first studies on the potential biological applications of metallosupramolecules had focused on anticancer activity of Pt-linked metallosupramolecular cycles (Mounir et al., 2007) and interaction of Fe- or Ru-linked metallohelicates with DNA fragments (Hotze et al., 2006; Malina et al., 2008). This had also given impetus to explore a broader range of metal complexes including the organometallic halfsandwich structures (Aird et al., 2002). Some of the initial examples for studying in details the anticancer activity profiles of coordination driven self-assemblies are the organometallic Ru(II) metallacycles and cages that have been extensively investigated (Therrien et al., 2008; Smith and Therrien, 2011; Dubey et al., 2013; Vajpayee et al., 2013). Thereby, the topic of metalbased drugs had naturally reached for the advantages that metallosupramolecular entities can provide—the major one is their encapsulation and delivery potential in addition to the intrinsic properties of their building components (Therrien et al., 2008; Schmitt et al., 2012; Therrien, 2013; Ma and Zhao, 2015).

As a multidisciplinary field, the topic of biomedical application of metallosupramolecules requires mutual understanding between synthetic chemists and biologists with the continuing help of specialist in photochemistry, pharmacology, bioimaging, etc. With the ultimate goal to facilitate this process, we intended to describe the recent achievements in the design of metallacycles, metallacages and helicates, and metallosupramolecular capsules and barrels in view of their anticancer properties, recognition of biomolecules with therapeutic implication, as well as improved delivery of bioactive molecules for therapeutic or imaging purposes. The examples of advanced nanomaterials that are based on, or incorporate, metallosupramolecular entities have grown in number and enriched in functionality, and are, therefore, given a major importance in this review paper. While the specific areas of biological applications of the metal-based supramolecules are described, the effect of their characteristic structural features is emphasized. This includes the charge, the shape and the size of the metallosupramolecules, as well as their specific cavity, which reflect the overall stability, reactivity, encapsulation potential, etc.—all being related to the biologically relevant applications. Ultimately, the wide variety of structures with desired composition and properties that can be suitably obtained through supramolecular coordination chemistry, makes this class of compounds a generous area for exploration not only from synthetic point of view but also in view of their biological activity.

# METALLACYCLES

As two-dimensional structures, the properties of metallacycles can be tuned by mere combination of the intrinsic characteristics of the metal ion and the linking organic ligands, thus forming supramolecular entities of variable size, shape, and charge. The stability issues have been overcome by employing inert metal ions as metal hinges or organometallic clips that form more robust metal-carbon bonds. Among the latter are the half-sandwich Ru(II) arene clips with bridging oxalate ligands that have been developed by Süss-Fink and coworkers and employed in the construction of the first organometallic Ru(II) metallamacrocycles (Yan et al., 1997). As the anticancer properties of ruthenium complexes came into focus after their introduction into clinical trials, and the first reports on the in vitro and in vivo cytotoxicity of a series of novel Ru(II) organometallic arene complexes (Aird et al., 2002), the topic expanded further on to the anticancer potency of metallosupramolecular assemblies with the pioneering works of Therrien and coworkers (Therrien et al., 2008; Mattsson et al., 2009). Several groups had greatly contributed in the field of biological applications of metallacycles of different shape and composition. The anticancer properties of series of tetracationic metallacycles, formed by [2+2] assembly between dipyridyl ligands and Ru(II) arene clips with oxalate or dioxydo-1,4 quinonato types of bridges of different size, have been thoroughly studied and later reviewed by Cook, Stang and coworkers (Cook et al., 2013). While this class of organometallic Ru(II)

**Abbreviations:** CDDP, cisplatin; SISO, cervix adenocarcinoma; A-427, lung carcinoma; LCLC-103H, large cell lung cancer; 5637, urinary bladder carcinoma; U2OS, osteosarcoma; VM-1, cisplatin resistant melanoma cell line; MCF7, breast cancer; A549, lung carcinoma; KB, oral cancer; HaCaT, skin keratinocyte; MDA-MB-231, human breast adenocarcinoma cell line; T98G, (brain tumor); SNU-80, (thyroid cancer); HEK-293, human embryonic kidney cells (non-malignant cell lines); HepG2, human hepatocellular carcinoma; DU-145, prostate cancer cells; MCF10A, immortalized non-malignant breast tissue; SKOV-3, Human ovarian cancer; HBL100 and T47D, Human breast-tumor cell lines; HT-29, colorectal adenocarcinoma cells; T-24, bladder cancer cells; SKW-3, T-cell leukemia; HL-60, acute myelocyte leukemia; HL-60/Dox and HL-60/CDDP, Doxorubicin and cisplatin-resistant counterparts of the HL-60 cells.

metallacycles demonstrated promising cytotoxicity profiles and structure-activity relationships could be outlined, the bisruthenium arene clips played further role in building 3D supramolecular cages and boxes by coordinating with tripodal or tetrapodal pyridyl ligands. Thereby, additional functionality has been introduced in terms of guest-encapsulation capability that further empowered the biological potential of the organometallic arene-metallosupramolecules. This topic has been extensively reviewed (Zhang et al., 2017) and therefore shall be left outside the scope of this paper.

Pt(II) and Pd(II)-cornered square-planar complexes have long been utilized in building highly symmetric 2D and 3D structures of desired shape and size. As chemically more inert units, the ethylenediamine-Pt(II) corners have been used to form [2+2] self-assemblies with rhomboidal structures by coordinating with bent-shape bipyridyl ligands. Some recent examples utilized the cationic N-monoalkylated 4,4′ -bipyridinium or 2,7 diazapyrenium based ligands (Terenzi et al., 2012; Domarco et al., 2017) depicted in **Figure 1**. Similarly to earlier reports on DNA-binding capability of tetranuclear Pt(II) squares of linear bipyridyne ligands (Mounir et al., 2007), the formed hexa- and octacationic Pt(II) rectangular metallacycles strongly bind native and G-quadruplex DNA. While the smaller size, unsymmetrical 2,7- diazapyrenium ligand is able to intercalate the DNA, the formed metallacycle (**1**) binds DNA ten times more strongly and bends its structure, thereby blocking the DNA transaction that possibly reflects on the observed anticancer activity against four types of human cancer cell lines (Terenzi et al., 2012). Similarly, the octacationic Pt(II)-cornered metallacycles **2a-c** exhibit moderate anticancer activity as compared to cisplatin


TABLE 1 | Cytotoxicity data of selected metallacycles against a panel of human cancer cells.

(**Table 1**), which has also been related to their strong binding to DNA (Domarco et al., 2017). Interestingly, the size of the rectangular box determines their effect on the DNA binding and the G4 folding process—the smaller Pt-box exhibited lower binding affinity but better selectivity for a particular G4 topology. The middle-sized box **2b** showed highest anticancer activity against the tested cancer cells, including the cisplatin resistant cell line VM-1, which, however, can be associated with the detected moderate toxicity of the free ligand (**L3**) against the U2OS cells (38.2 ± 3.4µM).

More recently, new types of cornered Pt(II) and Pd(II) square-planar complexes (**A1-A3**, **Figure 1**) have been utilized in building metallosupramolecules with potential biomedical use. Notable examples are the highly fluorescent triangles constructed from boron dipyrromethene (BODIPY) based dipodal ligands and Pt(II) or Pd(II) corners, as they combine cellular imaging and cancer treatment potency of the building components. The highly emissive BODIPY-Platinum supramolecular triangles (**3** and **4** in **Figure 1**) proved potent for intracellular visualization by confocal microscopy as well as for photodynamic therapy (PDT) and chemotherapy (Zhou et al., 2018a). Indeed, these metallacycles were used in the form of nanoparticles (NPs) that have been prepared through nanoprecipitation. More favorable cellular internalization of the NPs, formed from the cationic metallacycles **3** and **4**, has been observed in comparison with the free, pyridine functionalized BODIPY ligand. Furthermore, metallacycles **3** and **4** demonstrated strong anticancer activity against HeLa cells, which is intrinsic property of the Pt(II) corners **A1** and **A2** (**Table 2**). By light irradiation, however, the anticancer activity could be further amplified through inducing the photodynamic therapy modality, which is characteristic property of the BODIPY fluorophore, and have resulted in significant enhancement in the cytotoxicity against the cisplatin resistant cell line A2780cis. Thereby, effective theranostic agents have been obtained through combination of the chemotherapeutic potential of the platinum acceptors with the BODIPY donors as imaging probes and photosensitizers.

In another work, BODIPY derived Pd(II) metallosupramolecular triangles have been obtained by using Pd(II) hinges that are cis-capped with 1,3-bisdiphenylphosphino-propane, and the triangle-to-square conversion was studied (**5a-d** in **Figure 1**) (Gupta et al., 2017). The structures were fully characterized by <sup>1</sup>H, <sup>31</sup>P NMR, ESI-MS, and X-ray diffraction (of the triangle). The triangle and square topologies exist in a solvent-dependent equilibrium with the triangles being favored in higher polarity solvents. The stability of the complexes has been evaluated by UV-Vis absorption in presence of chlorides and in phosphate-buffered solution in DMSO or Dulbecco's Modified Eagle Medium, showing no change in the spectra for up to 48 h, except for the partial disruption seen upon addition of chloride ions. These BODIPY derived Pd(II) supramolecular triangles showed selective cytotoxicity against brain cancer (glioblastoma) cells U87 in comparison to that against normal brain cells WI-38. The anticancer activity of the triangles is higher or comparable to that of cisplatin, whereas the estimated selectivity is less pronounced than the selectivity of cisplatin (**Table 2**). These green fluorescent metallacycles have also been tested for fluorescence imaging of the cancer cells; the data demonstrated internalization in the cytoplasm and accumulation of aggregates in the cell membrane but not in the nucleus. Unlike the smaller precursors, the supramolecular triangles show appreciable interaction with the tested biomolecules—dsDNA, ssDNA, and BSA.

Hexagonal metallamacrocycles have been obtained by selfassembly of an organometallic platinum(II) clip **A4** with linear dipodal ligands, 4,4′ -bipyridine or pyrazine, depicted in **Figure 2** (Bhowmick et al., 2018). Both macrocycles (**6a** and **7**) exhibited cytotoxicity higher than cisplatin against the human breast cancer cell line MCF7. In general, the smaller-size macrocycle **6a** showed higher anticancer activity against all cell types (**Table 3**), and induced higher cell membrane damage. The latter has been associated with eventual necrosis of the cancer cells, in addition to the apoptosis, which was related to the swollen morphology of the treated cells seen by fluorescence microscopy imaging. The difference in the observed anticancer activity of the studied macrocycles has been associated with the lower hydrophobicity and lower molecular weight of **6a**; stability in presence of biomolecules has not been tested. Similarly to the macrocycles


*Cell types given in bold are normal non-cancerous cells.* \* *Irradiation by light (400–700 nm, 50 mW/cm*<sup>2</sup> *) for 5 min after 12 h incubation followed by12 h incubation.*

**6a** and **7**, the synthesis and anticancer potency of other flexible [2 + 2] tetracationic metallamacrocycles have been estimated (Jana et al., 2018). The macrocycle **6b** exhibited moderate anticancer activity against the tested cell lines (**Table 3**) as a result of self-assembly of the non-toxic organometallic clip **A5** and the pyrazine ligand. Yet, its cytotoxicity is lower than cisplatin and no selectivity could be attained.

Recently, two supramolecular metallacycles with hexagonal and rhombic shape have been obtained by using pyrazinebased organoplatinum(II) clip **A6** and two isomeric, also pyrazine-based, ditopic pyridyl ligands (Jana et al., 2019). The para- isomer of the ligand **L7** is shown in **Figure 2** and leads to formation of the hexagonal metallacycle **11** upon [3+3] self-assembly with the pyrazine-based organoplatinum(II) acceptor, whereas the meta- isomer of **L7** forms a rhomboidal structure (not shown) upon [2+2] self-assembly with the same organoplatinum(II) acceptor. Both types of metallacycles showed slightly higher anticancer activity than the organoplatinum(II) acceptor **A6** and cisplatin (**Table 3**), which is more appreciable for the larger, hexacationic [3+3] self-assembly **11**.

The rhomboidal Pt(II)-metallacycle **9**, depicted in **Figure 2**, has been evaluated for its anticancer activity both in vivo and in vitro (Grishagin et al., 2014). The formed metallacycle shows 13 times higher emission in the visible region (400– 700 nm) than the parent ligand in aqueous-DMSO solution. This property has been used to monitor the cellular uptake and the stability of the metallacycle by laser-scanning confocal microscopy in A549 and HeLa cells. The data evidenced their stability upon cellular internalization in addition to the high resistance to photobleaching under the conditions of the confocal microscopy experiment. At low concentrations (1 nM−5µM) the rhomboidal metallacycle and the organoplatinum(II) acceptor are not toxic, as seen by MTT test after 48 h of treatment of the cancerous cells. However, in the in vivo experiments in immunocompromised nude mice with tumor xenografts of breast cancer MDA-MB-231, the rhomboid **9** exerted a median tumor volume reduction of 88% (P < 0.001) (**Figure 3**). A maximum dose of 6 mg/kg in PBS:DMSO = 1:1 (v/v) has been applied by i.p. injections of 300 µL per animal, leading to no obvious signs of toxicity in terms of weight loss and visual appearance and behavior of the treated mice. The

tumor growth inhibition (T/C%) value, defined as the ratio of the median tumor volume for the treated vs. the control group at a particular day, has been estimated to 36% at the last day of the experiment of 30 days (**Figure 3** left).

Stang and coworkers prepared metallosupramolecular polymers based on an organometallic Pt-linked rhomboidal metallacycle **8b** (**Figure 2**) that was conjugated with two types of polymers via amidation reaction between alkylamine and N-hydroxysuccinimide-activated carboxylic acid (Zhang et al., 2016). These metallacycle-cored polymers P1 and P2 (**Figure 4**) have been utilized to form nanoparticles at low concentration. All synthetic modifications resulted in enhanced emission due to aggregation-induced emission properties of the tetraphenylethene-based pyridyl donor that is the main building block of the metallacycle. The fluorescence properties of these polymers were further used in cell imaging, which showed a significant enrichment in lung cells after i.v. injection into a mouse bearing an MDA-MB-231 tumor. The observed significant fluorescence of the tumor even 24 h after injection demonstrated that the studied metallosupramolecular polymer (P2) is both chemostable and photostable in vivo. Considering the anticancer activity of previously reported rhomboidal Pt(II) metallacycles of Grishagin et al. (2014), the authors infer that this type of fluorescent metallacycle-cored polymers can have potential applications in lung cancer theranostics.

The aggregation-induced emission could also be exploited in building metallosupramolecular species as has been described for a hexagonal [3+3] self-assembled metallacycle **10** built from


TABLE 3 | Cytotoxicity data of selected metallacycles against a panel of human cancer cells.

*Cell types given in bold are normal non-cancerous cells.*

tetraphenylethylene (TPE)-based organoplatinum(II) acceptors and the ditopic TPE-based dipyridyl ligand **L6** (**Figure 2**), as well as that of octacationic cages that were obtained from a tetrapodal tetrapyridyl ligand similar to **L6** (Yan et al., 2016). Interestingly, both types of supramolecular assemblies showed distinct photophysical properties depending on the degree of flexibility that the ligand attains after self-assembling with the TPE-based platinum(II) acceptors. The hexacationic metallacycle **10** exhibits aggregation-induced emission (AIE) by increasing the amount of hexane in the CH2Cl<sup>2</sup> solutions (up to 90%) due to the formation of nanoaggregates, as seen from the scanning electron microscopy (SEM) images, whereas the cages are intrinsically highly emissive. Taking advantage of the observed aggregation-induced enhanced emission (AIEE) of these systems, the same authors applied advanced approaches to build hybrid metallosupramolecular polymeric and bio-nanoparticles. The hexacationic metallacycle **10** has been assembled with the negatively charged rod-like tobacco mosaic virus (TMV) to form three-dimensional micrometer-sized superstructures via electrostatic interactions (Tian et al., 2016). In this way the fluorescence dramatically increased and provided an easily detectable evidence for the formation of novel metallosupramolecular biohybrid materials. Interestingly, addition of tetrabutylammonium bromide disrupts the metallosupramolecular core which results in lower fluorescence. The same strategy has been used to build similar biohybrid materials with bacteriophage M13 and ferritin, although with different degree of aggregation-induced emission. The demonstrated fluorescence switching upon aggregation with negatively charged protein-based species, forming emissive bionanoparticle/metallosupramolecular hybrid complexes, and their controllable disruption, enrich significantly the possibilities

for design and development of different supramolecular assemblies with dynamic optical properties and high potential to be used for both imaging and delivery of bioactive compounds.

On the other hand, aggregation-induced quenching may significantly compromise the photophysical functionality of some fluorophores such as porphyrins. In a recent paper, an alternative way to produce porphyrin functionalized hexagonal metallacycle has been reported (Chen et al., 2018). To avoid selfaggregation of the porphyrin rings, the metallacycle depicted in **Figure 5** has been produced in the confined cavity of mesoporous carbon FDU-16. Thereby, the molecules of the porphyrin-containing metallacycle have been dispersed within the mesoporous cavities, which resulted in increased stability and photocatalytic activity of the porphyrin-containing metallacycles. These properties have been demonstrated through successful use of the hybrid material in heterogeneous catalysis for photooxidation of sulfides. The six-fold faster generation of singlet (1O2), compared to that of free metallacycles in solution, is an important improvement that has been achieved. Although no biological applications have been discussed yet, the demonstrated strategy to construct welldefined porphyrin-containing metallacycles with improved stability and photoactivity may find implications in the future development of nanohybrids with improved properties for PDT.

Another strategy to build metallosupramolecules with significantly augmented photophysical properties, which allow for their use in photodynamic therapy, has recently been demonstrated for a heterometallic Ru-Pt metallacycle **12** (**Figure 6**) (Zhou et al., 2018b). The authors utilized the excellent photostability and two-photon absorption characteristics of the Ru(II) polypyridyl precursor, which after formation of the metallacycle exhibits shifted luminescence

FIGURE 3 | Effect of rhomboid 9 on tumor growth rate in MDA-MB-231 xenografts. (Left) Representative pictures of tumors, excised from control and mice treated with rhomboid 9. (Right) Localization of the near-infrared contrast agent IR-783 in the tumors of the control and treated mice. The signal was processed with Living Image software with one representative sample for each group presented above. Mice from the rhomboid 9-treated group show lower intensity of the signal originating from the tumor-accumulated contrast agent compared with the control group. Adapted with permission from Grishagin et al. (2014).

to the near-infrared region characterized also with a larger two-photon absorption cross-section and higher singlet oxygen generation efficiency. Thereby, the formed large macrocyclic structure of **12** renders it a very potent twophoton PDT agent. It is proposed that the highly charged (decacationic) structure of **12** facilitates its internalization in cells and localization in mitochondria and nuclei as evidenced by time-dependent inductively coupled plasma MS (ICP-MS) analysis of cellular uptake and the biodistribution of **12** in A549 cells. The metallacycle exhibits very high photocytotoxicity, induced by the generation of intracellular <sup>1</sup>O<sup>2</sup> upon irradiation and damaging the mitochondria and cellular nuclei, whereas the toxicity in the dark is very weak. These observations have also been confirmed in the in vivo studies which demonstrated that **12** is capable to efficiently ablate cancer tissues at low light doses with minimal system toxicity.

The tetracationic heterobimetallic cycles **13** and **14** have been obtained from an N, N′ -bis(4-(pyridin-4-ylethynyl)phenyl)pyridine-2,6-dicarboxamide ligand and cis-blocked palladium(II) or platinum(II) 1,1′ bis(diphenylphosphino)ferrocene triflates (Mishra et al., 2014). Their structure and composition have been characterized by multinuclear NMR spectroscopy, elemental analysis, and high-resolution electrospray mass spectrometry (HR-ESMS) and further modeled by molecular mechanics calculations. The stability of the metallacycles has been tested either by monitoring the cytotoxicity effect after different periods of pre-incubation in cell culture medium (Dulbecco's modified eagle medium—DMEM) or in DMSO (for 0, 10, 25, and 50 h at 37 <sup>0</sup>C) or by fluorescence spectroscopy measurements in physiological buffer solution at pH 7.4 and in presence of redox active compounds (H2O<sup>2</sup> or dithiothreitol). While both metallacycles exhibit cytotoxicity higher than cisplatin against the tested human-cancer cell lines (T98G, KB and SNU-80), they are twice less toxic to the normal human embryonic kidney cells (HEK-293) than cisplatin, which points to a much better cancer cell selectivity (**Table 3**). The stability studies indicated that under physiological conditions and even in presence of redox-active compounds the metallacycles **13** and **14** remain intact for at least 50 h, whereas in presence of DMSO a significant loss of cytotoxicity has been seen for pre-incubation periods of > 25 hours. These observations could suggest that the metallosupramolecular structure of **13** and **14** is crucial for the observed cytotoxicity profiles, since the DMSO induced disassembly is proposed as the main reason for decrease in

the cytotoxicity. The mode of action leading to the observed cytotoxicity has been associated with apoptosis as evidenced by the caspase-3/7 activity assays.

# METALLACAGES AND HELICATES

Metallacages and helicates are 3D structures that have characteristic cavity or cylindrical shape. Thereby, except combining the properties of the building components, metallacages can be excellent hosts for a variety of guest molecules that can be encapsulated in the hollow of the cages. This particular feature of metallacages has long been utilized in the design and development of smart nanostructures for multiple applications, such as molecular flasks (Yoshizawa et al., 2009), containers (Zarra et al., 2015), for safe storage (Yamashina et al., 2014) and recognition (Rodriguez et al., 2017), etc. Metallahelicates, on the other hand, can have specific helicity that is capable of selective interactions with biomolecules with certain conformation, such as DNA fragments. The growing interest in exploring the biomedical applications of these types of 3D metallosupramolecules has progressed to building metallacages that are either capable of encapsulating bioactive molecules or can be structurally decorated with biofunctional fragments. Thereby, the areas of potential applications of metallacages already include drug delivery and bioimaging except the initially explored biological activities that are intrinsically characteristics for their components. Similar expansion in the exploration of the properties of metallahelicates is also in progress.

Some examples of metallacages that had attracted the attention of synthetic and bioinorganic chemists are the [Pd2L4] <sup>4</sup><sup>+</sup> cages that assemble from simple ditopic tripyridyl ligands−2,6-bis(pyridin-3-ylethynyl)pyridine or its bis(pyridin-3-ylethynyl)benzene analog (**Figure 7**). The interest has been initiated with the reported encapsulation of the anticancer drug cisplatin by the supramolecular cage **15** that can be controllably released by external stimuli (Lewis et al., 2012). The cage formation and cisplatin encapsulation proceeds virtually quantitatively and the cage **15** and its host-guest complex with encapsulated two molecules of cisplatin have been characterized by single-crystal X-ray diffraction analyses. While the cage and the host-guest complex are rather stable in organic solvents, the cage readily disassembles in presence of competing amine ligands or chlorides, and thus the encapsulated cisplatin molecules can be released. The authors studied in details the encapsulation phenomenon and highlighted the importance of the hydrogen bonding interactions between the cage and the amine ligands of the cisplatin guest. This conclusion has been suggested by the lack of encapsulation evidences when a Pd2L<sup>4</sup> metallacage built from 1,3-bis(pyridin-3-ylethynyl)benzene ligand (when X=CH in **15**, **Figure 7**) has been used.

Further exploration of the observed stimuli-responsive catch and release of cisplatin by cage **15** has been focused on suitable ligand modifications to include an optically responsive fragment and/or increase the stability of the cage and the host-guest complex under biologically relevant conditions. The Cu(I) catalyzed azide-alkyne cycloaddition "click" reaction (CuAAC) has been utilized to build a library of electrochemically or photochemically responsive cages (**16a-c**, **Figure 7**) (Lewis et al., 2013, 2014). It has been demonstrated that the formed 1,2,3 triazole units in the exo-functionalization does not affect the self-assembling process nor does it interfere with the ability of the formed cages to bind cisplatin as guest molecules. Similarly, single-crystal X-ray structure has been determined for the 1:2 host-guest complex of cage **16a** and cisplatin (Lewis et al., 2013). The cage formation does not alter the one-electron reversible oxidation characteristic for 4-ferrocenyl triazoles. The same strategy has been used to "click" photoresponsive or biological molecules, such as 4-amino-1,8-naphthalimide (in **16b**, **Figure 7**) and fac-[Re(CO)3] complex with the 4-(2-pyridyl)-1,2,3-triazolyl moiety (in **16c**, **Figure 7**) or caffeine, acetate protected-Dglucose, estradiol and dipeptide units (Lewis et al., 2014). The attachment of the organic or inorganic fluorophores (in **16b** and **16c**, respectively) transformed the virtually non-emissive cages into emissive ones, with especially efficient emission seen for cage **16b** showing quantum yield of fluorescence of 0.59.

With the ultimate goal to improve the kinetic stability of the studied Pd2L<sup>4</sup> cages and evaluate their biological activity, Crowley and coworkers further modified the used tripyridyl ligands by adding NH2-susbstituents in the terminal pyridyl rings (Preston et al., 2016). Thereby, cages **17a** and **17b** have been obtained and their kinetic and biological properties have been compared with the non-functionalized cage **18d**. The stability studies showed that 2-amino substitution on the terminal pyridine rings significantly enhances the stability of the formed cage **17a** in presence of chloride ions or histidine and cysteine potent ligands in biological environment. Interestingly, cage **17b** has kinetic properties and cisplatin-encapsulation ability similar to that of **18d**, which points out to the importance of subtle structural modification of the self-assembling ligands on the properties of the formed metallacages. Enhanced stability of **17a** of up to several hours to almost 2 days (in presence of histidine) reflected also to the observed appreciable cytotoxicity against the cisplatin-resistant breast cancer cell line MDA-MB-231 (**Table 4**). In contrast, the **17b** and **18d** did not show appreciable cytotoxicity against the tested lung and breast cancer cell (for 24 h of incubation), which can be in favor to their stimuli-responsive cisplatin encapsulation and the suggested drug-delivery potential.

Further functionalization and cytotoxicity studies on this class of metallacages have been undertaken by Casini, Kühn and coworkers. The antiproliferative effects of Pd(II) cages of hydroxymethyl-functionalized bi- and tripyridyl ligands (**18c** and **18d** in **Figure 7**, respectively) indicated that upon 72 h of incubation the cages exert cytotoxicity against the tested cancer cells either comparable to that of cisplatin (for **18c**) or 4-times lower (for **18d**, **Table 4**) (Schmidt et al., 2016b). The other two examples of Pd(II) cages derived from differently functionalized bipyridyne ligands (**18a** and **18b** in **Figure 7**) have the same order of cytotoxicity effect as **18c**. The observed lower toxicity of **18d** and its distinctive ability to encapsulate cisplatin, as suggested earlier by Crowley and confirmed herein by X-ray data of the host-guest complex, can be regarded as favorable combination of properties for the sake of anticancer drug delivery potential (Schmidt et al., 2016b). To introduce optical functionality to the observed higher cytotoxicity of the Pd(II) cages derived from the bi-pyridyl ligands, the same authors exo-functionalized the Pd2L<sup>4</sup> cages with naphthalene or anthracene fluorophores (Schmidt et al., 2016a). The formed ligands exhibit weak fluorescence and upon self-assembly and formation of the cages **19a** – **19d** (**Figure 7**) the emissive properties of the fluorophore-tagged ligands have been preserved leading to the formation of weakly or non-emissive metallacages, most probably due to the high degree of flexibility. The exofunctionalization with aromatic rings, however, reflected the cytotoxicity of the ligands and the corresponding metallacages **19a, 19b,** and **19d** that increased considerably when compared with the non-functionalized cage **19c** (**Table 4**). Interestingly, by introducing methoxy groups in the terminal pyridine rings the fluorescence of the ligands increases (Kaiser et al., 2016); upon capsule formation, however, virtually non-emissive Pd(II)- and a novel Pt(II) cage have been obtained. Interaction of the Pd(II) cages of metoxy-functionalized bi- and tri-pyridine ligands with cisplatin has been suggested by <sup>1</sup>H NMR measurements. Pd(II)

cage **20** and its analog of a bi-pyridine ligand (structure not shown) have been proposed as suitable drug-delivery platforms as the estimated cytotoxicity is relatively low.

To achieve efficient targeting, functionalization of metallacages with biological fragments has been reported for the Pd2L<sup>4</sup> cages of the di-pyridyl ligands depicted in **Figure 7** (Han et al., 2017). Two different approaches for exo-functionalization with a model linear peptide Ac-NLEFK-Am have been described, which are based on amide bond formation between the carboxylic (or amine) group of the exo-functionalized ligand/cage and that of the side chains of the model peptide. Better results could be obtained with the pre-conjugation approach, where the peptide is coupled to the ligand prior to the following in situ self-assembly and formation of the Pd2L<sup>4</sup> cages. Thereby, metallacages of larger size could be obtained that can more efficiently benefit from the enhance permeability effect (EPR) for passive targeting to cancer cells and tissues. More importantly, this approach could open the prospect to bioconjugation and specific targeting of metallosupramolecules by attaching receptor-specific peptides or antibodies as an way to complement their either drug-delivery potential or intrinsic cytotoxicity (Han et al., 2017).

Metallosupramolecular systems of composition [Pd2L4] 4+ have been isolated by using smaller size ditopic ligands based on bis-triazoles similar to **L9** in **Figure 8** (McNeill et al., 2015). The formed complexes have the form of quadruplystranded helicates similarly to the crystal structure of **21**, depicted in **Figure 8** and reported earlier (Crowley and Gavey, 2010). These helical structures have improved stability that is much more pronounced in the case of bis-triazole ligand with hexyl pendant arms forming the analogous complex **21-hex** (structure not shown). The increased stability in presence of chlorides, histidine and cysteine correlated with the observed higher cytotoxicity against the tested cancer cell lines in comparison of the metallacage **15** (**Table 4**). Interestingly, the complex **21-hex** attains its half-inhibitory concentration the first 2 h of treatment of the cells in contrast to the known anticancer platinum drugs. This suggested different mode of anticancer action that has been associated with loss of cellular membrane integrity as derived from fluorescence microscope imaging and enzymatic (LDH) activity assay (McNeill et al., 2015). This hypothesis also corroborated with observed lack of selectivity of the cytotoxicity of complex **21-hex**. A large library of bis(bidentate) 2-pyridyl-1,2,3-triazole ligands with various lengths have been explored for their ability to form triple helicates with Fe(II) ions with general composition [Fe2L3] <sup>4</sup><sup>+</sup> (Vellas et al., 2013). While self-assembly and formations of cylinders have been confirmed in solution as well as in the solid state, the poor stability of these complexes hampered their potential biological activity. Molecular docking studies of these complexes suggested they could bind duplex and triplex DNA fragments.

The chiral metallohelical complexes [Fe2(**L10**)3] <sup>4</sup><sup>+</sup> and [Fe2(**L11**)3] <sup>4</sup>+, constructed by Scott, Qu and coworkers, have shown evidences to selectively target and enantioselectively inhibit amyloid-β (Aβ) aggregation (Li et al., 2014). The ability of these metal complexes to inhibit Aβ assembly suggested that they might be useful in blocking Aβ-mediated cellular



toxicity and thereby present a therapeutic strategy for treatment of Alzheimer's disease. The selective targeting of alpha/betadiscordant stretches at the early steps of aggregation has been demonstrated by multiple biophysical and biochemical approaches. More recently, a series of analogs of [Fe2(**L11**)3] 4+, having linking bridges of different flexibility, have been tested for their anticancer activity and evidenced both very good water solubility and stability (Kaner et al., 2016). Interestingly, the socalled flexicates exhibit anticancer activity of sub-micromolar to nanomolar range against the tested human cancer cell lines: MDA-MB-468 (human epithelial breast adenocarcinoma), HCT116 p53+/<sup>+</sup> and HCT116 p53−/<sup>−</sup> that are genetically identical human colorectal cancer cell lines except for the presence or absence of functional p53, which is a common genetic difference associated with increased resistance to chemotherapy. In the cisplatin-sensitive cells MDA-MB-468, the glycolbridged flexicate analogous to [Fe2(**L11**)3] <sup>4</sup><sup>+</sup> have shown a cytotoxicity (IC<sup>50</sup> = 0.2 ± 0.1µM) higher than cisplatin by an order of magnitude. Moreover, some of the tested flexicates were very toxic against the resistant HCT116 p53−/−cells that lack p53 and showing cytotoxicity as low as 40 nM. Another important feature of this class of flexicates is their excellent selectivity, which is estimated to be nearly three orders of magnitude higher than that of cisplatin against healthy human retinal

pigment epithelial (ARPE19) and lung fibroblast (WI38) cells. The helical complex [Fe2(**L10**)3] <sup>4</sup><sup>+</sup> showed limited solubility in water and could not be fully characterized, whereas its mxylenyl analog exhibited cytotoxicity as high as 0.4 ± 0.1µM against the MDA-MB-468 cell lines. In some cases significant enantiomeric difference had been observed pointing to the structural and conformational importance of the helicates. Nevertheless, the investigated mode of action could not be associated with DNA interaction nor apoptosis induced cell death.

The earlier reported metallosupramolecular helicate [Fe2(**L12**)3] <sup>4</sup><sup>+</sup> of Hannon and coworkers are capable of binding DNA Y-shaped junctions and could also unwind DNA and bind preferentially to regular alternating purine–pyrimidine sequences (Malina et al., 2008). More recent studies on the DNA interaction preferences have also revealed that this metallohelicate is capable of recognizing DNA bulges that are DNA defects consisting of one or more unpaired bases in the DNA double helical strands (Malina et al., 2014). These findings complemented the available data on the ability of the [Fe2(**L12**)3] <sup>4</sup><sup>+</sup> helicate to recognize non-canonical structures in DNA that might have relevance with various diseases. Recent report on the ability of the [Fe2(**L12**)3] <sup>4</sup><sup>+</sup> helicate to bind to TAR RNA, which is transactivation responsive region RNA, suggested it biological potential of inhibiting an essential step in the HIV-1 replication cycle (Malina et al., 2016). The observation that the iron(II) supramolecular helicates binds to TAR RNA at nanomolar concentrations and inhibits the interaction between the HIV-1 transactivator protein Tat and TAR, which plays a critical role in HIV-1 transcription, broadens the scope of biological importance of metallosupramolecular helicates.

The anticancer potency of dinuclear ruthenium metallohelicates have been among the first examples of biological activity demonstrated for double-stranded (**22** and **23** in **Figure 8**) (Hotze et al., 2006) and triple-stranded (**24** in **Figure 8**) metallohelicates (Pascu et al., 2007). The two different isomers of the dinuclear [Ru2L2Cl4] complex exhibit slightly different cytotoxicity profiles (**Table 4**) with the symmetric **23** isomer being an order of magnitude more cytotoxic than the cis/trans isomer **22** against the tested cancer cell lines. The analogous triple-stranded helicate **24** (Pascu et al., 2007) demonstrated very high stability and capability of binding DNA. This cylinder is photoresponsive to additions of ct-DNA in terms of fluorescence enhancement from the ruthenium cylinder and a concomitant blue shift (8 nm) in the emission maximum. The observed cytotoxicity, however, is significantly lower than the double-stranded analogs and also lower than that of cisplatin (**Table 4**). More recent examples on the construction and biological activity of double-stranded Ru(II) helicates of **L13** have been reported by Rice and coworkers and suggested the formed helicate **25** as a suitable compound that can target difficult to treat tumors (Allison et al., 2018), such as the human colorectal cancer cell line HCT116. The selectivity of **25** was similar to cisplatin and oxaliplatin toward the HCT116 p53+/<sup>+</sup> cancer cells, but it also showed substantively higher cytotoxicity toward the p53−/<sup>−</sup> cells.

Advanced approaches that utilize metallosupramolecular cages as delivery system for either anticancer drugs and prodrugs or fluorescent agents have gained increasing interest as alternative ways to achieve selectivity and efficient targeting of tumor tissues for both therapeutic and imaging purposes. Among the initial examples are those that use the hexanuclear octahedral metallacage of Fujita (**Pt-cage 1** in **Figure 9**), (Ibukuro et al., 1998), which has been shown to encapsulate four molecules of adamantyl Pt(IV)-complex exhibiting prodrug properties (Zheng et al., 2015). The formed host-guest complex has a diameter of about 3 nm, as characterized by detailed NMR analyses. The cytotoxicity of the host-guest complex is much higher than these of the **Pt-cage 1** or the Pt(IV)-prodrug guest alone, and is comparable to the anticancer activity of cisplatin against a panel of human cancer cells lines. The improved anticancer activity is caused by biological reductions of the Pt(IV)-prodrug, as evidenced by adding ascorbic acid, and release of the cytotoxic cisplatin. Thereby the cisplatin resistance in the tested A280CP70 cell lines could be overcome that was explained with the high cellular uptake of the highly positive charge of the host-guest nanostructure. On the other hand, by changing the type of the Pt(II)-cornering complexes of **Pt-cage 1,** from ethilenediamine to an 1,1′ -bipyridine complex, the formed octahedral cage becomes more cytotoxic than cisplatin against a panel of human cancer cells including the cisplatin-resistant celss (IC<sup>50</sup> = 3.42 ± 0.63µM for A2780CP70 and IC<sup>50</sup> = 7.85 ± 0.21µM for MD-MBA-231 vs. IC<sup>50</sup> values for cisplatin of 6.49 ± 1.40µM for A2780CP70 and 15.4 ± 2.3µM for MDA-MB-231) (Zheng et al., 2016). The detailed study on the mechanism of action of this Pt6L<sup>4</sup> octahedral cage showed that it interacts with DNA in a non-covalent intercalative way leading to DNA damage, and induces apoptosis through upregulation of p53 and p21 proteins.

Another strategy has been tested in building drug-loaded nanoparticles with polymeric shell of anionic block copolymer (methoxy polyethylene glycol-blockpolyglutamic acid, MPEG5k-PGA30) (Yue et al., 2018). The same octahedral metallacage (**Pt-cage 1**) has been used to encapsulate a fluoresceinlinked Pt(IV) prodrug (**Figure 9B**). The Pt(IV) prodrug-loaded supramolecular nanoparticles have spherical shape, and their average size was estimated to 20 nm by cryo-TEM imaging or overall size of 80 nm including the peripheral polymers and according to the dynamic light scattering (DLS) data. The drugrelease profiles speeds up dramatically under acidic conditions (pH = 5.0 in acetate buffer) and reducing environment of 1 mM ascorbic acid, that is estimated by the release half-life being 3 days in PBS and only 23 h in acidic and reducing environment. Intracellular release of the fluorescein-conjugated Pt(IV) prodrug was evidenced by fluorescence microscopy that has been enabled by the appearance of bright green fluorescence of the prodrug released from the otherwise non-fluorescent nanoparticle. Moreover, the cytotoxicity of the nanoparticles against HeLa cells is comparable to that of cisplatin (IC<sup>50</sup> = 5.02 ± 0.67µM for the nanoparticles; 2.95 ± 0.42µM for cisplatin, and 5.98 ± 0.58µM for the Pt(IV) prodrug), whereas the metallacage (**Pt-cage 1**) is essentially non-toxic. This work demonstrates an working strategy to use anionic block copolymer, to include a cytotoxic host-guest complex into negatively charged nanoparticles of appropriate size that are able to slowly release the therapeutic content and show excellent anticancer efficacy accompanied with fluorescence imaging modality.

Alternatively, Stang and coworkers utilized a highly emissive tetraphenylethene metallacage to build supramolecular nanoparticles with 1,2-distearoyl-phosphatidylethanolamine (DSPE)/polyethylene glycol (PEG) shell (**Figure 9C**) showing excellent stability and biotin-receptor targeting ability to be used as cancer-cell selective delivery systems (Yu et al., 2016). Moreover, these supramolecular nanoparticles preserve the coordination triggered aggregation induced emission (AIE) enhancement, which renders them highly emissive in biological media and suitable for imaging purposes. The incorporated metallacage is built from Pt(II) linked triphenyl pillars, decorated with differently modified PEG-arms, and two tetraphenylethene ligands each bearing four pyridines for coordination with the metal. The metal precursor is a cis-Pt(Et3P)2(OTf)<sup>2</sup> complex that exhibits high and unselective cytotoxicity. Upon formation of self-assembled polymeric aggregates with sizes of about 37 nm (as seen by TEM and DLS techniques) selective delivery to cancer cells has been achieved employing the EPR effect additionally to the biotin-mediated internalization in cancer cells. Thereby, the metallacage loaded nanoparticles combine the therapeutic with the imaging modality making them suitable for theranostic applications. Confocal laser scanning microscopy and flow cytometry confirmed the selective

targeting of the biotin functionalized nanoparticles to cancer cells overexpressing the corresponding receptor (HeLa and HepG2 cells) in contrast to normal cells (CHO and HEK-293), which also corroborated with their higher and selective cytotoxicity and high Pt-content in the cancer cells nuclei. The theranostic functions of the biotin-functionalized nanoparticles have been demonstrated also in vivo in HeLa tumor bearing female nude mice confirming their superiority than the clinical formulations of oxaliplatin, carboplatin, or cisplatin used as control. Importantly, while higher accumulation of Pt from the nanoparticles was observed in the tumor in comparison with the control Pt-drugs, administration of nanoparticles resulted in a lower Pt uptake by the liver, spleen, lung, and kidneys, which suggested that the use of PEGylated nanoparticles also provides way to decrease the systemic toxicity of Pt-drugs.

# METALLOCAPSULES AND BARRELS

The design and utilization of large aromatic panels as multidentate donor ligands enabled the construction of 3D metallosupramolecular structures with a characteristic wellconfined cavity, and having capsular or barrel-like shapes. Interesting examples consist of aromatic panels with intrinsic fluorescence properties, such as anthracene and carbazole fragments in the ligands depicted in **Figure 10**. The formation of robust M2L<sup>4</sup> coordination capsules have been described by Yoshizawa and coworkers (Li et al., 2012), and their fluorescence and guest-encapsulation properties have been studied in details (Kishi et al., 2011) for the sake of practical applications for safe storage of reactive compounds (Yamashina et al., 2014) or recognition purposes (Yamashina et al., 2016, 2017).

Recent results on the anticancer properties of Pd(II) and Pt(II) capsules **26** and **27** (**Figure 10**) indicated that they exhibit anticancer activity profiles much superior to cisplatin in terms of higher toxicity against two types of leukemic cells (HL-60 and SKW-3) as well as against cisplatin-resistant cells (HL-60/CDDP) (Ahmedova et al., 2016b). More importantly, the capsules showed more than 5 times better selectivity toward cancerous cells than that of cisplatin, which has been estimated by the IC<sup>50</sup> of normal-to-cancer cell toxicity ratio employing human embryonic kidney cells (HEK-293) as a model for non-malignant cells. The stability tests revealed that both types of coordination capsules remain intact in presence of excess of biologically relevant molecules such as amino acids and DNA bases. However, the addition of sulfur-containing biomolecules such as cysteine, causes disassembly of the capsules that proceeds instantaneously in case of the Pd(II) capsule **26** and takes up to 3 days for the more inert Pt(II) analog **27**. Decomposition of the capsules can be easily monitored by the appearance of the characteristic strong fluorescence of the liberated bis-anthracene ligand **L16**. Further on, the effect of guest encapsulation on the stability and the cytotoxicity of the capsules have been investigated (Ahmedova et al., 2016a). A successful modulation of these properties has been achieved by encapsulation of different types of hydrophobic guest molecules (pyrene and caffeine). The stabilities of the empty hosts and the corresponding host-guest complexes have been tested in presence of glutathione – the sulfur-containing oligopeptide that has relevance to cancer evolution. The observed strong stabilization of the capsules by encapsulation of pyrene guests has resulted in virtually complete loss of cytotoxicity of the otherwise highly toxic empty hosts **26** and **27** against the tested human cancer cell lines (HT-29, T-24, HL-60 and its resistant counterparts HL-60/Dox and HL-60/CDDP). These findings have suggested that the cytotoxicity of **26** and **27** can be readily modulated by encapsulation of large aromatic guests. More interestingly, the observed cytotoxicity profiles correlated with the capsules stability in the presence of glutathione that was estimated by NMR-based kinetic experiments. The observed stability-cytotoxicity correlation allowed proposing that the glutathione-triggered decomposition of the capsules could be the reason for the observed anticancer activity of these robust capsular metallosupramolecules. This hypothesis corroborates with the reported very high selectivity of the capsules to cancerous cells as well as with their cross-resistance cytotoxicity.

Mukherjee and coworkers reported the synthesis and single-crystal X-ray structure of a Pd8L<sup>4</sup> water-soluble molecular barrel **28** that self-assembled from a tetrapyridyl donor (**L17**) and a cis-blocked Pd(II) precursor [cis- (en)Pd(NO3)2] (Bhat et al., 2017). The affinity of the barrel's cavity to hydrophobic guest molecules has been employed in encapsulation of the anticancer drug curcumin, which resulted in significant stabilization of curcumin and its solubilization in water. This reflected on the increased anticancer activity of water solutions of curcumin against HeLa cancer cells demonstrating the successful utility of the metallosupramolecular barrel as a carrier and protector for photosensitive and hydrophobic drugs. Although stability and cytotoxicity of the empty barrels alone have not been investigated, this study clearly demonstrated the increased bioavailability and stability of curcumin in water solutions as seen from the fluorescence microscopy imaging of the cancer cells and the increased cytotoxicity with respect to the curcumin concentration.

A similar tetrafacial barrel with composition Pd8L<sup>4</sup> (**29**, **Figure 10**) has been built from tetraimidazolyl carbazole ligand (**L18**) (Roy et al., 2015). This metallosupramolecule has a tubular morphology that was fully characterized by multinuclear 1D/2D NMR and ESI-MS as well as by single-crystal X-ray diffraction of the coronene enclosing host-guest complex. While the barrel itself is water soluble, its hydrophobic cavity renders it ideal to solubilize hydrophobic guest molecules. Moreover, the inclusion of polyaromatic hydrocarbons with strong optical absorption and emission properties is accompanied with color and/or emission change. This property has been employed in live-cell fluorescence microscopy imaging by the water soluble perylene enclosing barrel, which demonstrated enhanced cell membrane permeability of the inclusion complex. This example gives the hope for further successful utilization of water soluble metalosupramolecules as drug or probe carriers provided their biocompatibility, stability and drug release properties is proved.

The flexible tetra(pyridin-4-yl)benzene-1,4-amine (**L17**, **Figure 10**) has also been used to construct organometallic octanuclear Ru(II) cages from dinuclear p-cymene ruthenium(II) acceptors, used as bridging oxo-pillars that vary from oxalate, benzoquinones, naphthoquinones (Ajibola Adeyemo et al., 2017). The largest box was characterized by X-ray diffraction besides the NMR, ESI-MS and electronic absorption spectra that have been used for characterization of all other boxes. All Ru8 cages showed strong anticancer activity in vitro that is comparable with cytotoxicity of cisplatin against A549 and HeLa cells. The anticancer profiles showed increasing cytotoxicity by enlarging the aromatic unit of the oxo-bridging pillars.

# CONCLUSIONS

From the presented recent examples for the fast-expanding chemotherapeutic applications of metallosupramolecular entities, it can be concluded that this class of compounds provide a rich field for exploration. The diversity in their properties stems from the fact that the metallosupramolecules can intelligently combine metal hinges of different coordination and reactivity properties as well as multidentate ligands that can be readily decorated without affecting the self-assembly processes. This has presented several working strategies to either increase the size or stability of the formed metalosupramolecular structures or to link them with receptor specific vectors. Construction of larger size composites that incorporate metallosupramolecules proved useful to target and deliver both therapeutic and imaging probes. The unique property of the 3D structures of metallacages, capsules, and barrels, is their discrete cavity which opens up further prospects for biological applications. Among the mostly explored ones are the encapsulation and delivery of bioactive molecules. Thereby, one and the same metallosupramolecular host can be used for different biological (therapeutic or diagnostic) purposes depending on the properties of the encapsulated guest molecule. The demonstrated encapsulation of Pt(IV) prodrugs, which can be activated and subsequently released in the intracellular media by selective reduction and formation of the corresponding Pt(II)-drug, is an intriguing advancement to fully explore the potential for biomedical applications of the coordination supramolecules. Moreover, the encapsulation of a fluorescein-bearing Pt(IV)-prodrugs by the **Pt-cage 1**, shown in **Figure 9**, had actually added imaging modality to the selective anticancer treatment. The presented examples are only some of the suggestions on how to augment the cancer treatment potential of such host-guest complexes with additional imaging modality.

The showed examples can expand in the near future toward modifying or multiplying the main therapeutic action by introducing immunomodulatory, anti- or prooxidative molecules, or other bioactive fragments either as guest molecules or as exo-functionalization of the supramolecular hosts. The proof-of-concept for exo-functionalization of metallosupramolecules with large peptides has already been demonstrated and is expected to open additional prospect to construct hybrid systems with larger size (>30 nm) that can benefit from both the targeting properties of the attached biomolecule and the truly efficient EPR effect.

The flexibility to modulate the overall stability of the 3D metallosupramolecular structures, by simply encapsulating appropriate guest molecule is another approach to fine-tune either the delivery properties or their therapeutic potential. In the cases when the increased stability affects the biological action of the supramolecules, guest-encapsulation can be exploited to turn the toxic species into non-toxic or vice versa, and thus, transform them from toxic anticancer agents to safe drug-delivery vehicles. Exploiting also building components with interesting optical or redox properties enriches further the possibilities to adjust the structure and properties of the constructed metallosupramolecules to the desired biomedical application. Inevitably, the presented examples are an expanding source of inspiration for many synthetic and biological chemists

# REFERENCES


to, hopefully, allow them fully reveal the multifaceted potential of the metallosupramolecules.

# AUTHOR CONTRIBUTIONS

The author confirms being the sole contributor of this work and has approved it for publication.

# ACKNOWLEDGMENTS

The National Science Fund of Bulgaria is gratefully acknowledged for the financial support (DFNI-B02/24). The support from H2020-Twinning project (Materials Networking; 692146) is gratefully acknowledged.


**Conflict of Interest Statement:** The author declares that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Ahmedova. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# The Role of the Second Coordination Sphere in the Biological Activity of Arene Ruthenium Metalla-Assemblies

### Bruno Therrien\*

Institute of Chemistry, University of Neuchatel, Neuchatel, Switzerland

For nearly 15 years, the biological and biomedical applications of arene ruthenium metalla-assemblies have flourished. Today, the synthetic strategies to generate arene ruthenium assemblies are well-established, and these compounds offer tremendous possibilities in terms of structural diversities and chemical properties. However, the second coordination sphere is often poorly considered, if not ignored, when designing such arene ruthenium metalla-assemblies. These weak interactions (hydrogen bonding, hydrophobic, ionic, electrostatic, van der Waals, π-π stacking) that take place in the solid state or in solution are generally key interactions for the foreseen applications. Therefore, in this review, we want to emphasize this important property of arene ruthenium metalla-assemblies by showing examples dealing with second coordination sphere interactions and how this can be better integrated in the design of these versatile supramolecular metal-based entities.

Keywords: arene ruthenium, metalla-assemblies, bio-inorganic, piano-stool complexes, weak interactions,

# INTRODUCTION

second coordination sphere

It is now well-known that the second coordination sphere (second shell) plays a major role in metal-based enzymatic transformations (Dudev et al., 2003). These additional interactions that take place in the proximity of the catalytic pocket can either stabilize the metal-substrate complex, stabilize the electronic state of the metal, orient the ligands to enhance the reactivity, act as proton and/or electron mediators, and so on (Ando et al., 1996; Botta, 2000; Steed, 2001; Haviv et al., 2018). More generally, these weak interactions (coordination, ionic, hydrogen bonding, hydrophobic, electrostatic, van der Waals, π-π stacking) are not only extremely important for biological processes, they are also a pillar of supramolecular chemistry. They allow the formation in the solid state of molecular networks (Hosseini, 2003), the preparation of liquid crystalline materials (Kato et al., 2018), the construction of coordination-driven assemblies (Fujita et al., 2005), as well as the generation of many other supramolecular systems (Wu et al., 2008; Zhou et al., 2017).

For many years, we have been involved in the field of coordination-driven self-assembly, using piano-stool complexes (also called half-sandwich complexes) as building blocks, and especially arene ruthenium complexes as biological agents (Therrien and Furrer, 2014; Therrien, 2015). The piano-stool unit provides three coordination sites at 90◦ from each other for a strategic

#### Edited by:

Angela Casini, Cardiff University, United Kingdom

#### Reviewed by:

Gilles Gasser, Université de Sciences Lettres de Paris, France Tatjana N. Parac-Vogt, KU Leuven, Belgium

> \*Correspondence: Bruno Therrien bruno.therrien@unine.ch

#### Specialty section:

This article was submitted to Supramolecular Chemistry, a section of the journal Frontiers in Chemistry

Received: 25 October 2018 Accepted: 20 November 2018 Published: 11 December 2018

#### Citation:

Therrien B (2018) The Role of the Second Coordination Sphere in the Biological Activity of Arene Ruthenium Metalla-Assemblies. Front. Chem. 6:602. doi: 10.3389/fchem.2018.00602

**27**

coordination of ligands on metals (Therrien, 2009), which allows the design of 2D and 3D entities (Cook et al., 2013; Singh et al., 2014; Therrien, 2018). These metalla-assemblies possess different functional groups, situated either at the periphery or at the core of the assembly. They have good stability in solution, showing no dynamic ligand exchange under ordinary conditions (Garci et al., 2014). They come in different sizes, with or without a cavity, and they can be positively charged, thus showing various properties and solubility. Among these properties, host-guest chemistry, biological activity, DNA interactions, recognition, ion binding, and others have been identified. Therefore, in this perspective review, the biological applications of arene ruthenium metalla-assemblies are discussed from the second coordination sphere point of view, to better emphasize the importance of weak interactions in diverse properties. Ultimately, it can provide to those working in the field a different angle to envision the next generation of metallaassemblies in biomedical fields.

# SENSING

The first example of arene ruthenium metalla-assemblies used for sensing was published in 2001 (Piotrowski et al., 2001). The electrochemical property of a trinuclear arene ruthenium metalla-cycle was exploited. Interestingly, upon the binding of alkali chloride salts in the triple-oxo binding site of the metalla-cycle (**Figure 1**), the oxidation potential of the trinuclear assembly was shifted by as much as 450 mV in the presence of LiCl, thus providing a redox-responsive signal upon guest binding.

Later, the possibility of using arene ruthenium metallaassemblies for sensing biologically relevant substrates was explored by Chi and Stang (Vajpayee et al., 2011; Mishra et al., 2012). The cavity of arene ruthenium metalla-rectangles (**Figure 2**) has shown interactions with polyanionic compounds (oxalate, citrate, tartrate). The presence of multiple amido and pyridine groups within the core of the metalla-rectangles was crucial for the recognition process to take place. Moreover, the amido groups gave some structural flexibility to the systems, thus providing binding adaptability to optimize the interactions with anions. In these systems, the size of the cavity and the capacity to form several hydrogen-bonds with anions are crucial elements for the sensing process to take place.

These are simple examples in which the nature of the functional groups incorporated in the building blocks of the arene ruthenium metalla-assemblies has provided valuable second sphere coordination interactions to develop metallaassembled sensors.

# DNA INTERACTIONS

The good water solubility and the presence of positive charges on most arene ruthenium metalla-assemblies are both advantageous properties for interactions with biomolecules. The first study dealing with arene ruthenium metallaassemblies and DNA interactions was published in 2009 (**Figure 3**). The tetracationic bowl-shaped rectangle showed good interaction with calf-thymus DNA (Linares et al., 2009). Binding assays have suggested that the interactions take place in the major groove of the duplex DNA strand. Conformational changes in the DNA strand are probably due to electrostatic interactions between the cationic metalla-rectangle and the negatively charged surface of DNA as well as the good match

between the size of the rectangle and the size of the major groove.

Following this initial study, several other arene ruthenium metalla-assemblies were tested as duplex DNA binders (Linares et al., 2010; Paul et al., 2015; Gupta et al., 2016, 2017). However, in the human genome, DNA sequences can fold into other thermodynamically stable structures such as hairpins and quadruplexes. These kinds of secondary structures are interesting targets for therapeutic applications (Zhao et al., 2010).

et al., 2008).

Consequently, arene ruthenium metalla-assemblies have been used, for example, as quadruplex DNA stabilizers (Barry et al., 2009).

A G-quadruplex is composed of guanine tetrads intercalated by cations, and it possesses a planar aromatic surface (Balasubramanian and Neidle, 2009). Therefore, cationic molecules with π-stacking affinity can potentially interact with G-quadruplexes. This possibility of interaction and stabilization was confirmed from the porphyrin-based arene ruthenium metalla-cubes (**Figure 4**), which showed good stabilization of telomeric and c-myc DNA quadruplexes. In such systems, we can assume that a combination of electrostatic and π-stacking (hydrophobic) interactions generates second coordination sphere interactions.

# PROTEIN INTERACTIONS

Like DNA, proteins can positively interact with cationic compounds, and accordingly with positively charged arene ruthenium metalla-assemblies. Different proteins interacting with such metalla-assemblies have been identified. Interestingly, these studies have shown that metalla-assemblies of different shapes, sizes, structures and numbers of charges are able to interact with proteins, confirming that interaction with biomolecules is a common feature of arene ruthenium metallaassemblies (Dubey et al., 2015; Elumalai et al., 2016, 2017).

Indeed, an arene ruthenium metalla-rectangle with bisamido pyridine containing linkers (**Figure 5**) shows strong interaction with enhanced green fluorescence protein (EGFP) (Mishra et al., 2014). Similarly, an arene ruthenium metallaprism can disrupt the folded structures of albumin, transferrin, cytochrome-c, and other proteins (Paul et al., 2018), showing a great diversity of protein interactions. These examples

suggest that the interactions between metalla-assemblies and proteins are mainly electrostatic. Therefore, to optimize the metalla-assembly protein interactions, and to gain a degree of selectivity for a specific protein, one cannot rely purely on electrostatic, hydrophobic, size-dependent or hydrogenbond interactions. Only the multiplicity of second coordination sphere interactions can generate selective protein binders, which increases the complexity of designing arene ruthenium metallaassemblies.

# DECOMPLEXATION UPON INTERACTIONS

The first biological application of arene ruthenium metallaassemblies was confirmed by the complex-in-a-complex system (**Figure 6**), in which a water-soluble arene ruthenium metallaprism was used to transport the platinum acetylacetonate complex to cells (Therrien et al., 2008). Later, it was demonstrated that, after internalization, the guest complex was released, most likely upon disassembly of the cage compound (Mattsson et al., 2010). Furthermore, the metalla-prism can interact with biomolecules such as arginine, cysteine, glutathione, lysine, histidine (Paul et al., 2012a,b), and biomolecules possessing coordinating functional groups, thus being able to initiate the breakage of the cage compound. This suggests that despite being relatively robust to ligand exchange processes (Garci et al., 2014), second coordination sphere interactions can be used to disassemble arene ruthenium metalla-assemblies.

# OXYGEN INTERACTIONS

In traditional photodynamic therapy (PDT), a photosensitizer interacts with oxygen to produce reactive oxygen species (ROS), and these ROS trigger cell death upon light activation (Patrice et al., 2003). Therefore, to evaluate the potential of using arene ruthenium metalla-assemblies as PDT agents, several metallaassemblies coupled to photosensitizers were prepared and tested (Schmitt et al., 2009). The arene ruthenium units can modulate the solubility of the photosensitizers (Schmitt et al., 2008) as well as modify its photochemical behavior. Moreover, the cavity of arene ruthenium cages can transport and protect the photosensitizer (Schmitt et al., 2012).

Nevertheless, other types of molecules can interact with oxygen to generate ROS. This is the case of BODIPY, an interesting fluorescent dye with a high-quantum yield that can promote the production of ROS. Therefore, BODIPY-based linkers have been recently inserted in arene ruthenium metallarectangles (**Figure 7**), and the cytotoxicity of the compounds on various cancer cell lines was confirmed (Gupta et al., 2017). The fluorescence associated with the BODIPY units was exploited to localize the metalla-rectangle in the cytoplasm of cancer cells. However, the possibility of using such metalla-rectangles as PDT agents remains to be explored.

Anthracene is another molecule that can react with oxygen (Aubry et al., 2003). Anthracene forms in the presence of oxygen and light activation, an endoperoxide intermediate. The endoperoxide formation is reversible, and oxygen can be released in a different environment. Therefore, knowing that arene ruthenium metalla-assemblies can be internalized to cells and can be coupled to photosensitizers, we have recently synthesized an anthracene-based metalla-rectangle (**Figure 8**) (Gaschard et al., 2018). Despite an unsuccessful endoperoxide formation on the metalla-rectangle, the ultimate goal of this project was to transport oxygen and a photosensitizer to cells for an optimization of PDT treatments in hypoxic cancers.

# CONCLUSION

As pointed out in this article, the second coordination sphere plays a major role in most biological and biomedical applications involving arene ruthenium metalla-assemblies. Therefore, the introduction of functional groups that can generate weak interactions with biomolecules on either the arene, the building blocks or the guest molecules is essential for the development of biologically active arene ruthenium metalla-assemblies. In the future, designing metalla-assemblies with second coordination sphere interactions in mind will be challenging, but it could provide the next generation of arene ruthenium derivatives for biological and biomedical applications.

# AUTHOR CONTRIBUTIONS

The author confirms being the sole contributor of this work and has approved it for publication.

# FUNDING

Over the years, we would like to acknowledge the financial support from the Swiss National Science Foundation, the European Union (POLYTHEA and NOAH projects) and the University of Neuchatel.

# ACKNOWLEDGMENTS

The author would like to thank past and present members of his group.

# REFERENCES


**Conflict of Interest Statement:** The author declares that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Therrien. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Anticancer Activity and Cisplatin Binding Ability of Bis-Quinoline and Bis-Isoquinoline Derived [Pd2L4] 4+ Metallosupramolecular Cages

Roan A. S. Vasdev 1,2,3 \*, Lachlan F. Gaudin<sup>1</sup> , Dan Preston<sup>1</sup> , Jackmil P. Jogy <sup>3</sup> , Gregory I. Giles <sup>3</sup> and James D. Crowley 1,2 \*

*<sup>1</sup> Department of Chemistry, University of Otago, Dunedin, New Zealand, <sup>2</sup> MacDiarmid Institute for Advanced Materials and Nanotechnology, Wellington, New Zealand, <sup>3</sup> Department of Pharmacology and Toxicology, University of Otago, Dunedin, New Zealand*

#### Edited by:

*Jonathan Kitchen, Institute of Natural and Mathematical Sciences, Massey University, New Zealand*

### Reviewed by:

*Robert Elmes, Maynooth University, Ireland Nicholas White, Australian National University, Australia*

#### \*Correspondence:

*Roan A. S. Vasdev roan.vasdev@otago.ac.nz James D. Crowley jcrowley@chemistry.otago.ac.nz*

#### Specialty section:

*This article was submitted to Supramolecular Chemistry, a section of the journal Frontiers in Chemistry*

Received: *29 September 2018* Accepted: *31 October 2018* Published: *22 November 2018*

#### Citation:

*Vasdev RAS, Gaudin LF, Preston D, Jogy JP, Giles GI and Crowley JD (2018) Anticancer Activity and Cisplatin Binding Ability of Bis-Quinoline and Bis-Isoquinoline Derived [Pd*2*L*4*]* 4+ *Metallosupramolecular Cages. Front. Chem. 6:563. doi: 10.3389/fchem.2018.00563* New *bis-*quinoline (Lq) and *bis-*isoquinoline-based (Liq) ligands have been synthesized, along with their respective homoleptic [Pd2(L<sup>q</sup> or Liq)4] <sup>4</sup><sup>+</sup> cages (C<sup>q</sup> and Ciq). The ligands and cages were characterized by <sup>1</sup>H, <sup>13</sup>C and diffusion ordered (DOSY) NMR spectroscopies, high resolution electrospray ionization mass spectrometry (HR-ESIMS) and in the case of the *bis*-quinoline cage, X-ray crystallography. The crystal structure of the C<sup>q</sup> architecture showed that the [Pd2(Lq)4] <sup>4</sup><sup>+</sup> cage formed a twisted *meso* isomer where the [Pd(quinoline)4] <sup>2</sup><sup>+</sup> units at either end of the cage architecture adopt the opposite twists (left and right handed). Conversely, Density Functional Theory (DFT) calculations on the Ciq cage architecture indicated that a lantern shaped conformation, similar to what has been observed before for related [Pd2(Ltripy)4] <sup>4</sup><sup>+</sup> systems (where Ltripy = 2,6-*bis*(pyridin-3-ylethynyl)pyridine), was generated. The different cage conformations manifest different properties for the isomeric cages. The Ciq cage is able to bind, weakly in acetonitrile, the anticancer drug cisplatin whereas the C<sup>q</sup> architecture shows no interaction with the guest under the same conditions. The kinetic robustness of the two cages in the presence of Cl<sup>−</sup> nucleophiles was also different. The Ciq cage was completely decomposed into free Liq and [Pd(Cl)4] <sup>2</sup><sup>−</sup> within 1 h. However, the C<sup>q</sup> cage was more long lived and was only fully decomposed after 7 h. The new ligands (Liq and Lq) and the Pd(II) cage architectures (Ciq and Cq) were assessed for their cytotoxic properties against two cancerous cell lines (A549 lung cancer and MDA-MB-231 breast cancer) and one non-cancerous cell line (HDFa skin cells). It was found that L<sup>q</sup> and C<sup>q</sup> were both reasonably cytotoxic (IC50S ≈ 0.5µM) against A549, while Ciq was slightly less active (IC<sup>50</sup> = 7.4µM). Liq was not soluble enough to allow the IC<sup>50</sup> to be determined against either of the two cancerous cell lines. However, none of the molecules showed any selectivity for the cancer cells, as they were all found to have similar cytotoxicities against HDFa skin cells (IC<sup>50</sup> values ranged from 2.6 to 3.0µM).

Keywords: palladium(II), anticancer, self-assembly, metallosupramolecular, quinoline

# INTRODUCTION

Metallosupramolecular architectures (MSAs) (Cook and Stang, 2015) have been attracting increasing attention over the past two decades due to their many potential applications including catalysis (Yoshizawa et al., 2009; Yoshizawa and Fujita, 2010; Martí-Centelles et al., 2018), storage (Mal et al., 2009), and sensing (Wang et al., 2011). Inspired by the success of cisplatin and other metallodrugs (Mjos and Orvig, 2014) there is emerging interest in exploiting MSAs for biomedical purposes (Cook et al., 2013; Therrien, 2015; Casini et al., 2017; Zhou et al., 2017). Several groups have examined MSAs as drug delivery vectors (Therrien et al., 2008; Schmitt et al., 2012; Yi et al., 2012; Zheng et al., 2015; Samanta et al., 2016, 2017; Bhat et al., 2017; Xu et al., 2017; Wang J.-F. et al., 2018; Yue et al., 2018). Additionally, MSAs have been shown to bind DNA (Oleksy et al., 2006; Garci et al., 2017; Zhao et al., 2017) and RNA (Phongtongpasuk et al., 2013; Malina et al., 2015), interact with proteins (Li et al., 2014; Mitchell et al., 2017) and have anticancer (Hotze et al., 2008; Faulkner et al., 2014; Grishagin et al., 2014; Dubey et al., 2015; Zheng et al., 2016; Allison et al., 2018) and antibacterial (Richards et al., 2009; Howson et al., 2012; Wang H. et al., 2018) properties.

Since the pioneering work of McMorran and Steel (McMorran and Steel, 1998) interest in [M2(**L**)4] <sup>n</sup><sup>+</sup> cage-type structures has burgeoned (Schmidt et al., 2014). Some time ago now we reported that a [Pd2(**L**tripy)4] <sup>4</sup><sup>+</sup> cage (where **L**tripy = 2,6-bis(pyridin-3 ylethynyl)pyridine) could host two molecules of the anticancer drug cisplatin within the cavity of the cage (Lewis et al., 2012), and thus had potential as a drug delivery vector. Disappointingly, the binding event, which was governed mainly by hydrogen bonding interactions, was weak (Preston et al., 2015). The host-guest complex formed in acetonitrile (CH3CN) and dimethylformamide (DMF) but unfortunately, in more hydrogen bond competitive solvents such as water and dimethyl sulfoxide (DMSO) no host-guest interaction was observed. Additionally, the parent Pd(II) based cage decomposed rapidly in the presence of nucleophiles (Lewis et al., 2012; McNeill et al., 2015). Thus, in order to exploit these [Pd2(**L**)4] <sup>4</sup><sup>+</sup> cages as cisplatin delivery vehicles these issues need to be addressed. We and others have examined a range of modifications to the parent [Pd2(**L**)4] <sup>4</sup><sup>+</sup> cage system in order to improve the solubility (Lewis and Crowley, 2014; Preston et al., 2015; Han et al., 2017) and other properties (Lewis et al., 2013, 2014; Kaiser et al., 2016; Schmidt et al., 2016a) of the cage. Efforts have also been made to enhance the strength of the host-guest interaction (Kim et al., 2015) and the stability of the cages in the presence of biologically revelant nucleophiles (Preston et al., 2016). However, while some improvements have been made these [Pd2(**L**)4] <sup>4</sup><sup>+</sup> cages still require further modifcations in order to be useful drug delivery vectors.

The [Pd2(**L**)4] <sup>4</sup><sup>+</sup> cages have also been examined for their cytotoxic properties. We showed that the parent [Pd2(**L**tripy)4] 4+ was modestly cytotoxic (IC<sup>50</sup> values range from 40 to 70µM) against a range of cancer cell lines but was less active than related bis-1,2,3-triazole [Pd2(**L**)4] <sup>4</sup><sup>+</sup> helicates (McNeill et al., 2015). We also examined the cytotoxicity of related amino substituted [Pd2(**L**)4] <sup>4</sup><sup>+</sup> cages against the same panel of cancer cells and found that they exhibited similar cytotoxic properties as the parent systems (Preston et al., 2016). Casini, Kühn and coworkers (Kaiser et al., 2016; Schmidt et al., 2016b) have measured the cytotoxicity of a series of related [Pd2(**L**)4] <sup>4</sup><sup>+</sup> cages and observed similar IC<sup>50</sup> values (10–70µM). Additionally, they have measured the cytotoxicity of mixtures of the cages and cisplatin and unsurprisingly have found that those mixtures are more cytotoxic than cage alone (IC<sup>50</sup> = 2–13µM). Yoshizawa, Ahmedova and co-workers have also found that [M2(**L**)4] <sup>4</sup><sup>+</sup> (M = Pd2<sup>+</sup> or Pt2+) cages with similar, but more hydrophobic, dipyridyl anthracenyl ligands (**L**anthracene) display high anticancer activity (IC<sup>50</sup> values range from 0.9 to 37.4µM against HL-60, HL-60/Dox, HT-29, T-24, SKW-3 cancer cell lines) (Ahmedova et al., 2016; Anife et al., 2016).

The majority of the [Pd2(**L**)4] <sup>4</sup><sup>+</sup> cages examined to date feature pyridyl donors, as part of our efforts to improve the biological properties of these systems herein we describe the use of isoquinoline and quinoline-derived ligands for the assembly of two new [Pd2(**L**)4] <sup>4</sup><sup>+</sup> cages. While is well-known that isoquinoline and quinoline ligands can bind with palladium(II) and platinum(II) (Bondy et al., 2004) their use as donor systems in ligands for the generation MSAs has not been extensively explored (Bloch et al., 2016, 2017). These quinoline derived systems feature different electronic and steric properties compared to the parent pyridyl systems thus we also examine the effect these changes have on the host-guest chemistry with cisplatin, the stability of the cages in the presence of nucleophiles and the antiproliferative properties of the cages.

# RESULTS AND DISCUSSION

The synthesis of the new quinoline (**L**q) and isoquinoline (**L**iq) based ligands was facile (**Supplementary Material**). Using sequential Sonogashira carbon-carbon cross coupling reactions from commercially available building blocks the ligands were generated in good yields (**L**<sup>q</sup> = 86% and **L**iq = 78%). <sup>1</sup>H and <sup>13</sup>C NMR spectroscopic data were consistent with the formation of the ligands which was supported by high-resolution electrospray mass spectrometry (HR-ESIMS) (**Figure 1** and **Supplementary Material**). Peaks corresponding to protonated and sodiated ligand were observed at m/z = 382.1320 and 404.1132, respectively, for **L**iq and similar peaks were observed for **L**q (**Supplementary Material**).

With the ligands in hand, the complexation with [Pd(CH3CN)4](BF4)<sup>2</sup> in acetonitrile was examined (**Figure 1**). The cage formation was monitored using <sup>1</sup>H NMR spectroscopy (CD3CN, 298 K) and showed that mixing [Pd(CH3CN)4](BF4)<sup>2</sup> and the **L**iq ligand at room temperature (RT) in a 1:2 ratio led to the rapid (<2 min) and quantitative formation of the expected **C**iq cage (**Figure 1**), similar to what was observed with the parent **L**tripy system (Lewis et al., 2012). Interestingly, the behavior of **L**qwith [Pd(CH3CN)4](BF4)<sup>2</sup> at room temperature was very different. After 5 min at RT the reaction mixture displayed multiple proton resonances, none of which were due to free ligand, consistent with the formation of a mixture of different cage isomers. The reaction was monitored using <sup>1</sup>H NMR spectroscopy for 24 h at RT however little to no changes were

observed after the first hour and the spectrum still displayed multiple proton resonances. A <sup>1</sup>H DOSY experiment (CD3CN, 298 K) on the mixture showed that all the different proton resonances had the same diffusion co-efficient consistent with the postulate that the reaction mixture contains a series of cage isomers (**Supplementary Material**).

The assembly reaction between **L**qwith [Pd(CH3CN)4](BF4)<sup>2</sup> was then carried out at 65◦C, in CD3CN and again monitored using <sup>1</sup>H NMR spectroscopy (**Figure 2**). After 5 min the same complicated series of proton resonance were observed. However, with continued heating this slowly resolved into a single series of resonances (after 7 h), consistent with the formation of a single cage isomer (**Figure 2**). Pleasingly, both cages (**C**iq and **C**q) could be isolated by adding diethyl ether into the acetonitrile reaction mixtures providing the cages as colorless/tan precipitates in 88% (**C**iq) or 92% (**C**q) yield, respectively. <sup>1</sup>H NMR spectroscopy (CD3CN, 298 K) exhibited the expected downfield shifts of the signals pertaining to protons Ha, H<sup>b</sup> and H<sup>f</sup> as well as the anticipated downfield shifts of the rest of the isoquinoline and quinoline protons resonances (**Figure 1**). HR-ESIMS data also supported the formation of the cages, showing ions corresponding to the loss of 2, 3 and 4 tetrafluoroborate (BF<sup>−</sup> 4 ) counterions (m/z = 956.1610 (2+), 608.4424 (3+), and 434.5832 (4+), **Supplementary Material**). <sup>1</sup>H DOSY experiments (CD3CN, 298 K) on the ligands (Diffusion coefficients (D) = 13.1 (**L**q) and 15.0 (**L**iq) x 10−<sup>10</sup> m<sup>2</sup> s −1 ) and cages (D = 4.1 (**C**q) and 4.3 (**C**iq) × 10−<sup>10</sup> m<sup>2</sup> s <sup>−</sup><sup>1</sup> were also consistent with the formation of the larger [Pd2(**L**)4] <sup>4</sup><sup>+</sup> cages (**Supplementary Material**).

Crystals of **C**q suitable for X-ray diffraction formed during the cooling of an acetonitrile solution of the cage from 65◦C to room temperature. The structure was solved in the tetragonal space group P4/mnc with the asymmetric unit containing one eighth of the cage and one quarter of a BF<sup>−</sup> 4 counterion (**Figure 3** and **Supplementary Material**). The other BF<sup>−</sup> 4 anions and some acetonitrile molecules could not be modeled sensibly thus the SQUEEZE routine was employed to account for the diffuse electron density (**Supplementary Material**). The data revealed the expected [Pd2(**L**q)4] <sup>4</sup><sup>+</sup> cage structure. The Pd-N bond lengths (Pd1-N2 2.045 Å) were similar to what have been previously observed for the related [Pd2(**L**tripy)4] <sup>4</sup><sup>+</sup> cages where the Pd-N bond lengths range from 2.016 to 2.027 Å (Lewis et al., 2012; Lewis and Crowley, 2014). The **L**q ligands of the cage are twisted giving a V-shaped conformation where the terminal quinoline and central pyridyl heterocyclic units are not co-planar which is quite different to what was observed with the [Pd2(**L**tripy)4] 4+ cages. In X-ray structures of the parent [Pd2(**L**tripy)4] <sup>4</sup><sup>+</sup> cages the **L**tripy ligands were found in a linear conformation with the heterocyclic units coplanar. The twisting also alters the Pd1- Pd1′ distance within **C**<sup>q</sup> related to the [Pd2(**L**tripy)4] <sup>4</sup><sup>+</sup> cages. The Pd1-Pd1′ distances for the parent [Pd2(**L**tripy)4] <sup>4</sup><sup>+</sup> cages range from 11.49 to 12.24 Å, whereas the Pd1-Pd1′ distance was found to be longer (12.506 Å) suggesting that the **C**qcage has a larger cavity despite featuring the same 2,6-diethynylpyridine linker units. The [Pd(**quinoline**)4] <sup>2</sup><sup>+</sup> units at the top and bottom of **C**q are twisted in opposite directions, the top cationic unit has a right handed twist while the bottom cationic unit has a left handed twist giving an overall meso structure (**Figures 3B**,**C** and **Supplementary Material**). Despite extensive efforts we were unable to obtain X-ray diffraction quality single crystals of **C**iq. Thus, to gain further insight into the structure of **C**iq we modeled the cage using Density Functional Theory (DFT) calculations (**Figures 3D,E**). Energy minimization of **C**iq (DFT, BP86 def2-SVP, acetonitrile solvation, **Supplementary Material**) showed that the cage adopted a lantern shape similar to what was previously observed for [Pd2(**L**tripy)4] <sup>4</sup><sup>+</sup> cages (Lewis et al., 2012; Lewis and Crowley, 2014). The calculated Pd – N bond distances (2.049 Å) and the Pd-Pd′ distance (11.758 Å) match well with those observed crystallographically for the related [Pd2(**L**tripy)4] <sup>4</sup><sup>+</sup> cages. The **L**iq ligand adopts a linear conformation with all the heterocyclic units coplanar. The DFT calculations indicated that the **C**iq is structurally very similar to the parent [Pd2(**L**tripy)4] <sup>4</sup><sup>+</sup> cages whereas the **C**<sup>q</sup> is more twisted and provided a cavity of different size and shape to the parent cages and the **C**iq system.

We and others have previously shown that other similar [Pd2(**L**tripy)4] <sup>4</sup><sup>+</sup> cages can encapsulate cisplatin through hydrogen bonding interactions in CH3CN and DMF solvents (Lewis et al., 2012, 2013, 2014; Kaiser et al., 2016; Preston et al., 2016, 2017; Schmidt et al., 2016b). Therefore, we examined the ability of **C**iq and **C**<sup>q</sup> to interact with cisplatin in CH3CN using <sup>1</sup>H NMR spectroscopy. Addition of an excess of cisplatin to a

CD3CN solution of the **C**iq cage resulted in a downfield shift and broadening (1δ = 0.03 ppm) of the internally directed cage proton H<sup>a</sup> (**Figures 4A,B**) indicative of cisplatin binding within the cage cavity, albeit weakly. A similar <sup>1</sup>H NMR experiment was carried out with the **C**q cage (**Figures 4C,D**). However, with the **C**q cage no shifts were observed for any of the cage proton resonances in the presence of an excess of cisplatin suggesting that the more twisted **C**q cage does not interact with the anticancer agent. The behavior was similar to what has been observed with a related twisted [Pd2(**L**2Atripy)4] <sup>4</sup><sup>+</sup> cage (where **L**2Atripy = 2,6-bis[2-(6-amino-3-pyridinyl)ethynyl]-4 pyridinemethanol) (Preston et al., 2016). The [Pd2(**L**2Atripy)4] 4+ cage did not bind cisplatin in DMF solvent and the lack of binding was ascribed to the twisted cage cavity which was not as preorganised as those of the related lantern shaped [Pd2(**L**tripy)4] <sup>4</sup><sup>+</sup> cages. Presumably the different sized cavity and different spatial arrangement of the hydrogen bond donors and acceptors caused by the twisting observed in the crystal structure of **C**<sup>q</sup> impedes the cisplatin-**C**<sup>q</sup> interaction in this case.

The kinetic robustness of the related [Pd2(**L**tripy)4] 4+ architectures in the presence of common biological nucleophiles (chloride (Cl−), histidine and cysteine) has been determined using <sup>1</sup>H NMR competition experiments. When the parent [Pd2(**L**tripy)4] <sup>4</sup><sup>+</sup> architectures were treated with 8 equivalents of tetrabutylammonium chloride the pyridyl substituted cages were rapidly and quantitatively decomposed (in <5 min). To examine the effect of substituting the pyridyl donor units for quinoline heterocycles time-course <sup>1</sup>H NMR competition experiments were carried out in d6-DMSO where 2 mM solutions of each cage (**C**qor **C**iq) were treated with 8 equivalents of tetrabutylammonium chloride at 298 K (**Figure 5** and **Supplementary Material**). Within 30 s of adding Cl<sup>−</sup> to the **C**iq cage, there were multiple species observed in the <sup>1</sup>H NMR spectrum. These were attributed to the **C**iq cage, [Cl⊂**C**iq] 3+, the [Pd2(**L**iq)2Cl4] macrocycle and free ligand based on our own previous results (Preston et al., 2015) and related literature. After 50 min, only uncoordinated ligand was visible in the <sup>1</sup>H NMR spectrum (**Supplementary Material**).

Under the same conditions, **C**q was stable for 1 h before showing signs of decomposition (**Figure 5**). After 3 h, there was no evidence of the **C**<sup>q</sup> cage, and the <sup>1</sup>H NMR spectrum displayed peaks corresponding to free ligand and a second metal-containing species, which based on the observed chemical shifts was most likely the neutral [Pd2(**L**q)2Cl4] macrocycle (**Figure 5H**). This degradation behavior has been seen before with the [Pd2(**L**tripy)4] <sup>4</sup><sup>+</sup> system in DMF (Preston et al., 2015). After 7 h, only free ligand could observed in the <sup>1</sup>H NMR spectrum indicating that all the ligand containing metal complexes had been completely decomposed into [Pd(Cl)4] <sup>2</sup>−(**Figure 5J**).

In comparison to the previously reported [Pd2(**L**tripy)4] 4+ cage (τ1/<sup>2</sup> = 2 min), the isoquinoline cage displayed an identical half-life (τ1/<sup>2</sup> = 2 min), whereas the quinoline system was considerably more robust (τ1/<sup>2</sup> = 2 h). Presumably the observed results reflect the different steric profiles of the two quinoline substituted cages (**C**qor **C**iq). The **C**q cage has the quinoline moieties protecting the external face of the palladium, providing more impediment to nucleophilic attack from that face (**Figure 6**). The **C**iq does not feature the same steric impediment as the benzene units of the isoquinoline heterocycles do not block the top face of the **C**iq cage as much as they do in the quinoline **C**q(**Figure 6**).

FIGURE 3 | Molecular structures of Cq and Ciq. X-ray structure of Cq: (A) ellipsoid side view, (B) tube side view, and (C) tube top view showing paddle-like array of quinoline panels over palladium(II) center. Solvent molecules and counterions have been omitted for clarity. Ellipsoids are shown at 50% probability. DFT optimized (BP86 def2-SVP) model of Ciq; (D) side view and (E) top view showing lantern-shaped structure. Colors: carbon gray, nitrogen blue, palladium magenta, hydrogen white.

To assess biological activity, the cytotoxic effect (as halfmaximal inhibitory concentrations (IC50)) of the ligands and cages were determined against three different cell lines: cisplatin resistant MDA-MB-231 (breast cancer) (Lehmann et al., 2011), A549 (lung cancer) and non-cancerous primary cells: adult human dermal fibroblasts (HDFa) (**Table 1** and **Supplementary Material**). The ligands **L**q and **L**iq exhibited limited solubility, and so data above the concentration of 1µM was unattainable. Below this threshold, **L**iq displayed minimal cytotoxic activity against both cell lines, while **L**q was shown to be cytotoxic against A549 (IC<sup>50</sup> = 0.5µM). Both cages were observed to be cytotoxic against the malignant cell lines, with **C**q showing the same level of toxicity as its ligand against lung cancer cells (IC<sup>50</sup> = 0.5µM). **C**<sup>q</sup> was slightly less cytotoxic against MDA-MB-231 (IC<sup>50</sup> = 1.7µM), whereas **C**iq was less cytotoxic than the quinoline analog, with the IC<sup>50</sup> values ranging from 4.0 to 7.4µM against the cancer cells. Both quinoline cages were found to be considerably more active than the related parent [Pd2(**L**tripy)4] <sup>4</sup><sup>+</sup> cage system (IC<sup>50</sup> = 41.4 and 56.7µM against A549 and MDA-MB-231, respectively) (McNeill et al., 2015). The quinoline cages were also more active than cisplatin against the two cancer lines examined (cisplatin IC<sup>50</sup> values = 41.2 and 9.4µM, against MDA-MB-231 and A549, respectively) (Lo et al., 2015; McNeill et al., 2015). The quinoline cages **C**q and **C**iq were more cytotoxic than all the [Pd2(**L**tripy)4] <sup>4</sup><sup>+</sup> cage systems reported in the literature (IC<sup>50</sup> values for the **L**tripy based systems ranged from 10 to 100µM) (McNeill et al., 2015; Kaiser et al., 2016; Schmidt et al., 2016b). Additionally, **C**q was also more active, albeit against different cancer cell lines (HL-60, HL-60/Dox, HT-29, T-24, SKW-3), than the hydrophobic [Pd2(**L**anthracene)4] 4+ cages of Yoshizawa and Ahmedova (IC<sup>50</sup> values ranged from 0.9 to 37.4µM) (Ahmedova et al., 2016; Anife et al., 2016). TABLE 1 | Half-maximal inhibitory concentrations (IC50) of ligands Lq and Liq, and cages Cq and Ciq architectures at 24 h.


L*tripy,* C*tripy,* L*hextrz,* C*hextrz and cisplatin (Lo et al., 2015) have been added for comparison (McNeill et al., 2015).*

*a IC*<sup>50</sup> *values are given as mean* ± *SE.*

*<sup>b</sup>The* C*tripy and* C*hextrz cages were tested against MCF-10A as a non-cancerous control.*

*<sup>c</sup>Solubility limited the range of concentrations to below 1*µ*M. " – "* = *Not determined.*

**C**q was also more cytotoxic than a hydrophobic bis-hexyl-1,2,3-triazole substituted [Pd2(**L**hextrz)4] <sup>4</sup><sup>+</sup> helicate, **C**hextrz, we developed previously (IC<sup>50</sup> values = 6.9 and 6.0µM against A549 and MDA-MB-231, respectively) (McNeill et al., 2015). We presume that the favorable combination of high hydrophobicity and the kinetic robustness against biological nucleophiles leads to the higher observed activity of **C**q relative to the other [Pd2(**L**)4] <sup>4</sup><sup>+</sup> architectures. Disappointingly, neither of the cages (**C**q and **C**iq) showed any selectivity for the cancer cells, they were all found to have similar cytotoxicity against HDFa skin cells (IC<sup>50</sup> values ranged from 2.6 to 3.0µM).

# CONCLUSION

We have herein reported the synthesis, characterization, cisplatin binding, kinetic robustness and cytotoxicity of two new bis-isoquinoline and bis-quinoline derived [Pd2**(L)**4] <sup>4</sup><sup>+</sup> cage complexes. The crystal structure of **C**q architecture showed that the [Pd2(**L**)4] <sup>4</sup><sup>+</sup> cage formed a twisted meso isomer where the [Pd(**quinoline**)4] <sup>2</sup><sup>+</sup> units at either end of the cage architecture adopt the opposite twists (left and right handed). Conversely, Density Functional Theory (DFT) calculations on the **C**iq cage architecture indicated that a lantern shaped conformation similar to what has been observed before for related [Pd2(**L**tripy)4] 4+ systems was generated. The different cage conformations resulted in different properties for the isomeric cages. The **C**iq cage is able to bind, weakly in acetonitrile, the anticancer drug cisplatin whereas the **C**q architecture shows no interaction with the guest under the same conditions. The kinetic robustness of the two cages in the presence of Cl<sup>−</sup> nucleophiles was also different. The **C**iq cage was completely decomposed into free **L**iq and [Pd(Cl)4] <sup>2</sup>−within 1 h. However, the **C**<sup>q</sup> cage was more long lived and was only fully decomposed after 7 h. The ligands (**L**iq and **L**q) and cages (**C**iq and **C**q) were assessed for their cytotoxic properties against two cancerous cell lines (A549 lung cancer cells and MDA-MB-231 breast cancer cells) and one non-cancerous cell line (HDFa skin cells). It was found that **L**q and **C**q were both reasonably cytotoxic against A549, while **C**iq was slightly less active. The higher observed cytotoxicity of **C**q relative to the other [Pd2(**L**)4] <sup>4</sup><sup>+</sup> architectures was presumed to be due the favorable combination of high hydrophobicity and the kinetic robustness against biological nucleophiles. However, none of the new molecules showed any selectivity for cancer cells, they were all found to have similar cytotoxicity against HDFa skin cells. A range of [Pd2(**L**)4] <sup>4</sup><sup>+</sup> cage systems have now been shown to be cytotoxic. However, in order to advance this class of MSA as anticancer agents more in depth mode of action/mechanistic studies on the origins of the cytotoxic activity are required. Studies to this effect are now underway.

# AUTHOR CONTRIBUTIONS

RV and JC conceived the idea, analyzed the data and wrote the manuscript. RV and LG conducted the synthesis. RV and DP conducted stability studies. RV, DP, and JJ conducted cytotoxicity studies. GG oversaw the cytotoxicity studies and analyzed the data. All authors provided feedback on the manuscript drafts and approved the submission.

# FUNDING

JC thanks the University of Otago, Department of Chemistry and the MacDiarmid Institute for funding. RV was supported by a Claude McCarthy Fellowship. All the authors thank Charlotte Dobson, Pauline Lane and Oceanbridge for their kind donations which supported this work.

# ACKNOWLEDGMENTS

RV and JJ acknowledge the University of Otago for doctoral stipends. Dr. Anna Garden and Dr. David McMorran are thanked for useful discussions. Professor Sally McCormick, Department of Biochemistry, University of Otago is thanked for kindly providing a sample of the HDFa skin cells. The authors acknowledge the contribution of the NeSI high performance computing facilities to the results of this research. New Zealand's national facilities are provided by the New Zealand eScience Infrastructure and funded jointly by NeSI's collaborator institutions and through the Ministry of Business, Innovation and Employment's Research Infrastructure program. URL: https://www.nesi.org.

## REFERENCES


# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fchem. 2018.00563/full#supplementary-material

DNA binding is a potent cytostatic and apoptotic agent without exhibiting genotoxicity. Chem. Biol. 15, 1258–1267. doi: 10.1016/j.chembiol.2008.10.016


antibacterial activity. Nat. Commun. 9:1815. doi: 10.1038/s41467-018-04 247-z


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Vasdev, Gaudin, Preston, Jogy, Giles and Crowley. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Nanoparticles of Metal-Organic Cages Overcoming Drug Resistance in Ovarian Cancer

Han Wang<sup>1</sup> , Zihan Qiu<sup>1</sup> , He Liu<sup>2</sup> , Amarasooriya M. D. S. Jayawardhana<sup>1</sup> , Zhizhou Yue<sup>1</sup> , Hala Daghlas <sup>1</sup> , David J. Bowers <sup>1</sup> , Bansidhar Datta<sup>2</sup> and Yao-Rong Zheng<sup>1</sup> \*

*<sup>1</sup> Department of Chemistry and Biochemistry, Integrated Sciences Building, Kent State University, Kent, OH, United States, <sup>2</sup> Science Research Laboratory, Department of Chemistry and Biochemistry, Kent State University, Kent, OH, United States*

#### Edited by:

*James D. Crowley, University of Otago, New Zealand*

#### Reviewed by:

*Christian Hartinger, The University of Auckland, New Zealand Ross Stewart Forgan, University of Glasgow, United Kingdom*

> \*Correspondence: *Yao-Rong Zheng yzheng7@kent.edu*

#### Specialty section:

*This article was submitted to Supramolecular Chemistry, a section of the journal Frontiers in Chemistry*

Received: *30 September 2018* Accepted: *14 January 2019* Published: *01 February 2019*

#### Citation:

*Wang H, Qiu Z, Liu H, Jayawardhana AMDS, Yue Z, Daghlas H, Bowers DJ, Datta B and Zheng Y-R (2019) Nanoparticles of Metal-Organic Cages Overcoming Drug Resistance in Ovarian Cancer. Front. Chem. 7:39. doi: 10.3389/fchem.2019.00039* A long-standing challenge in the treatment of ovarian cancer is drug resistance to standard platinum-based chemotherapy. Recently, increasing attention has been drawn to the use of self-assembled metal-organic complexes as novel therapeutics for cancer treatment. However, high hydrophobicity that is often associated with these structures lowers their solubility and hinders their clinical translation. In this article, we present a proof-of-concept study of using nanoprecipitation to formulate the hydrophobic metalorganic cages and facilitate their use in treating chemoresistant ovarian cancer. The Pt6L<sup>4</sup> Cage 1 is an octahedral cage formed by self-assembly of six 1,10-phenanthroline-Pt(II) centers and four 2,4,6-tris(4-pyridyl)-1,3,5-triazine ligands (L). Cage 1 is able to trigger DNA damage and exhibits promising *in vitro* potency against a panel of human ovarian cancer cell lines. However, due to the large portion of aromatic components, this cage structure has very limited solubility in cell culture media (<20µM). Notably, upon nanoformulation by using fluorescein (2) and a pegylated anionic polymer (3), the concentration of Cage 1 can reach up to 0.4 mM. Production of the nanoparticles of metal-organic cages (nMOC) is driven by the formation of the 1:1 host-guest complex of 1 and 2 in aqueous solution, which then form nanoprecipitation in presence of poly glutamic acid-*b*-poly ethylene glycol (3). The resulted nMOC are about 100 nm in diameter, and they serve as a delivery platform that slowly releases the therapeutic content. The use of fluorescein facilitates monitoring cell entry of nMOC and drug release using flow cytometry. Finally, comparing to cisplatin, the nMOC exhibit comparable *in vitro* efficacy against a panel of human cancer cell lines, and notably, it shows a much lower resistance factor against chemoresistant ovarian cancer cell lines.

Keywords: metal-organic cages, nanoprecipitation, DNA damage, host-guest chemistry, drug resistance

# INTRODUCTION

Ovarian cancer is the most lethal gynecologic cancer and ranks as the fifth leading cause of death for women in the US. Each year, about 250,000 women are diagnosed with ovarian cancer in the US, and approximately half of these patients die within 5 years after their initial diagnosis (Siegel et al., 2015). The low survival rate of ovarian cancer is mainly attributed to the development of drug resistance to the standard platinum chemotherapy (Holmes, 2015). The FDA-approved platinum drugs, cisplatin and carboplatin, are widely used as the first line treatment for ovarian cancer patients in the US and worldwide (Kelland, 2007). Initially, the patients show good responses to these compounds, but after about 12–24 months, patients commonly develop cancer relapse with drug resistance. Currently, there are very few alternative treatment options after the development of drug resistance to these platinum compounds. Thus, there is an urgent need for new, alternative therapeutics to combat chemoresistant ovarian cancer. In the past, the search of new drug candidates has largely focused on mononuclear metal complexes similar to structure of cisplatin, but there are rarely any clinically relevant breakthroughs (Wilson and Lippard, 2014; Zheng et al., 2014). Macromolecular polynuclear metal complexes emerges as a new type of therapeutic candidates to widen the spectrum of activity of current FDA-approved platinum drugs, however, studies of the therapeutic effects of such compounds are still limited (Farrell, 2015).

Metallosupramolecular coordination complexes represent a novel type of metal-organic complexes constructed by coordination-driven self-assembly (Cook and Stang, 2015). They are ordered, discrete macromolecular structures obtained by controlled aggregation of metal centers and organic ligands, and these ensembles exhibit well-defined sizes and shapes, such as metal-organic polyhedra, polygons, cages, and spheres, etc., (Harris et al., 2013; Xu et al., 2015). Due to the unique structural features, self-assembled metal-organic complexes have led to a large variety of novel research in chemistry, material science, and biology. In recent years, studies about applying such metalorganic complexes in cancer research has attracted increasing attention (Hannon, 2007; Therrien, 2012; Cook et al., 2013; Kaiser et al., 2016; Samanta et al., 2017). Stang, Chi, Therrien, Crowley, and Zheng have respectively published their works on designing new metal-organic polyhedra and cages and testing their anti-cancer efficacy (Lewis et al., 2012; Vajpayee et al., 2013; Grishagin et al., 2014; Zheng et al., 2015; Yue et al., 2018b). However, high hydrophobicity that is often associated with these structures lowers their solubility and hinders their clinical translation (Cook et al., 2013; Cook and Stang, 2015). Therefore, additional formulation of metallosupramolecular coordination complexes is critically important to resolve this issue in order to further advance the field toward clinical translation.

In this article, we present a proof-of-concept study of using nanoprecipitation to formulate the hydrophobic metal-organic cages and facilitate their use in treating chemoresistant ovarian cancer. As shown in **Scheme 1**, we can formulate Cage **1**, a cytotoxic hydrophobic metalorganic cage (**Figure 1A**) with low solubility, using fluorescein (**2**) and a pegylated anionic polymer (**3**). Driven by the host-guest interactions and electrostatic interactions, combination of the three components produces nanoparticles of metal-organic cages (nMOC) covered by PEG, therefore promoting its solubility. The resulted nMOC shows promising anticancer activities against chemoresistant ovarian cancer cells.

# MATERIALS AND METHODS

# General Information

All reagents were purchased from Strem, Aldrich or Alfa and used without further purification. Methoxy polyethylene glycolblock-polyglutamic acid (MPEG5k-PGA50) (**3**) was purchased from Polypeptide Therapeutic Solutions SL (Valencia, Spain). All reactions were carried out under normal atmospheric conditions. Deuterated solvents were purchased from Cambridge Isotope Laboratory (Andover, MA). The <sup>1</sup>H NMR spectra were recorded on a Bruker AMX 400 NMR in the Department of Chemistry and Biochemistry at Kent State University. Chemical shifts in <sup>1</sup>H NMR spectra were internally referenced to solvent signals (DMSO at δ = 2.50 ppm). Elemental analysis was performed on a TruSpec Micro, CHNS analyzer from LECO Corporation. UV-Vis spectra were recorded on a VWR UV-1600PC scanning spectrophotometer. Fluorescence spectra were obtained on a Cary Eclipse fluorescence spectrophotometer. Graphite furnace atomic absorption spectroscopic (GFAAS) measurements were taken on a PerkinElmer PinAAcle 900Z spectrometer. Fluorescence images were acquired using an Olympus IX70 inverted epifluorescence microscope equipped with a digital CCD camera (QImaging). Images were processed and intensities were quantified with ImageJ software (NIH). Dynamic light scattering and zeta-potential analysis were carried out using a Horiba SZ-100 particle analyzer. 3D computational modeling was performed using Schrödinger Macromodel Suite based on MMFF force field. Flow cytometry was carried out on a BD Bioscience Accuri C6 flow cytometer. Cryo-TEM specimens were prepared using thin-film plunge freezing in a FEI Vitrobot at Liquid Crystal Institute at Kent State University. The vitrified specimens were mounted onto a Gatan 626.DH cryo-holder and transferred into a FEI Tecnai F20 TEM. Cryo-TEM observation was performed using low-dose mode. The detailed experimental setup and procedure are reported previously (Gao et al., 2014).

# Synthesis of the Metal-Organic Cage (1)

Synthesis of the Pt6L<sup>4</sup> cage was modified from the previous reports (Zheng et al., 2016; Yue et al., 2018a). Pt(phen)(NO3)<sup>2</sup> (15.0 mg, 30 µmol) and 2,4,6-Tri(4-pyridyl)-1,3,5-triazine (6.0 mg, 19 µmol) were mixed in a microwave reaction vial (2–5 mL), followed by the addition of 4 mL H2O and 4 µL of 65% HNO3. The reaction vial was then sealed and placed in a Biotage Initiator Microwave Synthesizer. The mixture was stirred at 900 rpm and heated using a multi-step sequence (a. 90◦C, 100 W for 30 min; b. 150◦C, 150 W for 60 min). The mixture was centrifuged to remove possible precipitations. Then, the supernatant was collected, and the solvent was removed under reduced pressure. The resulted white solid was washed with 1 mL of water under sonication. After removing water by centrifugation, the final product was dried in vacuum. Yield: 15.7 mg, 78%. <sup>1</sup>H NMR (400 MHz, DMSO-d6) δ 9.75 (d, J = 6.4 Hz, 24H, α-Pyridine), 9.26 (d, J = 8.0 Hz, 12H, phen), 9.16 (d, J = 6.8 Hz, 24H, β-Pyridine), 8.67 (d, J = 4.2 Hz, 12H, phen), 8.52 (d, J = 2.8 Hz, 12H, phen), 8.14 (m, 12H, phen). Anal. Calcd for C144H96N48O36Pt<sup>6</sup> · (H2O)6: C, 39.73; H, 2.50; N, 15.44. Found: C, 39.56; H, 2.25; N, 15.09.

SCHEME 1 | Graphical representation of the formation of nanoparticles of metal-organic cages (nMOC): Cage 1 and fluorescein (2) form a host-guest complex in a 1:1 ratio in aqueous solution, and the use of MPEG5k-PGA50 anionic block copolymer (3) enables the nanoprecipitation of this complex into nMOC, therefore promoting solubility of this cytotoxic hydrophobic cage.

# Preparation of the Nanoparticles of Metal-Organic Cages (nMOC)

To a volume of 1 mL of PBS solution of MPEG5k-PGA<sup>50</sup> (**3**) (1.00 mg/mL) and fluorescein (**2**) (30.0 µL, 13.0 mM) was added the DMSO solution containing the metal-organic cage (**1**) (90.0 µL, 2.00 mM) under shaking at 900 rpm at room temperature (R.T.). The mixture was shaken for 10 min further at R.T. Significant color change from yellow green to red could be observed during the process, which is associated with the formation of the hostguest complex. The resulted red solution was filtered using a 0.2µm PTFE syringe filter. The filtrate was further purified using centrifugal filtration (MWCO = 30 kDa). The solution of the

nanoparticles was concentrated to a volume of 0.8 mL. The Pt content was determined to be 768µM using GFAAS. Each cage molecule contains six Pt centers. Therefore, the concentration of **1** in nMOC is 128µM in PBS. Yield: 56.7%. This sample can be further concentrated up to 0.4 mM in PBS by centrifugal filtration (MWCO = 30 kDa).

# Fluorescence Titration

The 2-mL PBS solution of fluorescein (**2**) (6µM) was titrated by successive addition of Cage **1** (2.5 µL, 2.4 mM in DMSO). The total volume change during the titration is negligible. A spectra range between 500 and 650 nm was monitored during the titration.

# Job Plots From UV-Vis Spectroscopy

A total of 11 samples containing Compound **1** and **2** in PBS were prepared in different molar ratio where each molar fraction varied from 0 to 1.0. The total concentration of **1** and **2** remained as constant as 10µM. Solutions were prepared freshly and observed from 450 to 650 nm. Abs (506 nm) –ε (fluorescein) × [fluorescein] – ε (**1**) × [**1**] was plotted against the molar fraction of fluorescein, where Abs (506 nm) is the absorption of the sample at 506 nm, ε(fluorescein) is the extinction coefficient of fluorescein measured at 506 nm, ε(**1**) is the extinction coefficient of Cage **1** measured at 506 nm, and [**2**] and [**1**] are the concentration of fluorescein and Cage **1** within the sample.

# Dialysis Experiments

A volume of 1 mL PBS solution (pH = 7.4) or acetate buffer (pH = 5.0) containing Cage **1** ([Pt] = 120µM) or nMOC ([Pt] = 240µM) were sealed in micro-dialysis bags (10 kDa MWCO) against 500 mL PBS solution or acetate buffer at R.T. A series of samples were collected within the bags for every few hours and analyzed with GFAAS. All the measurements were done in triplicate.

# Cellular Uptake Evaluated by GFAAS

One million A2780cis cells were seeded on 60 × 10 mm petri dishes and incubated at 37◦C overnight. These cells were treated with cisplatin ([Pt] = 25µM), **1** ([Pt] = 25µM), and nMOC ([Pt] = 25µM) for 4 h at 37◦C. The remaining alive cells were harvested by trypsinization and digested in 200 µL 70% HNO<sup>3</sup> at R.T. overnight. The platinum content in the cells were analyzed by GFAAS. All experiments were performed in triplicate.

# Cellular Uptake Evaluated by Flow Cytometry

A2780cis cells were seeded on 60 × 10 mm petri dishes and incubated at 37◦C overnight. These cells were treated with PBS, fluorescein ([**2**] = 3µM), fluorescein ([**2**] = 3µM) plus polymer, and nMOC ([**2**] = 3µM) for 4 h at 37◦C. The cells were harvested by trypsinization and with the Accuri flow cytometer and the cell populations were analyzed using the FlowJo software.

# Flow Cytometric Analysis

A2780cis cells were seeded on 60 × 10 mm petri dishes and incubated at 37◦C overnight. A2780cis cells were treated with the Pt compounds (nMOC [Pt] = 4µM; cisplatin [Pt] = 10µM) at 37◦C for 72 h. The samples were analyzed with the Accuri flow cytometer and the cell populations were analyzed using the FlowJo software. Annexin V early apoptosis detection kit was used for apoptosis analysis, and the experiments were carried out by following the standard protocol supplied by the manufacturer.

# Immunoblotting Analysis

A volume of 0.6 ml of RIPA buffer was added to the A2780cis cells treated with the different Pt compounds (nMOC [Pt] = 2.5, 5.0, and 7.5µM, 48 h, and Cage **1** [Pt] = 2.5, 5.0, 7.5µM, 48 h). The cells were gently shaken for 15 min at 4 ◦C. Then, the cells were collected with a cell scraper, and transferred to a microcentrifuge tube. A total of 100 µg whole cell lysate with SDS loading dye was boiled for 2– 3 min. A volume of 50 µl of lysate and loading dye mixture per well was loaded for gel electrophoresis (120 Volts for 90 min). Afterward, the proteins were transferred from the gel to UltraCruz <sup>R</sup> Nitrocellulose Pure Transfer Membranes, using an electroblotting apparatus according to the manufacturer's protocols. Non-specific binding was blocked using BlockerTM Blotto (30–60 min at R.T.). The blocked membrane was incubated with primary antibody with 1:1,000 dilution in Blotto for 1 h at R.T. with shaking. The membrane was washed three times for 5 min each with Tris-buffered saline, pH 7.5 with 0.5% Tween 20 (TBST). The membrane was incubated with horse-radish peroxidase (HRP) conjugated secondary antibody with dilution 1:2,000 for 1 h at R.T., with shaking. The membrane was washed three times for 5 min each with TBST. Chemiluminescence Luminol Reagent was used for detecting according to the standard protocol supplied by the manufacturer.

# MTT Assays

Cells were seeded on a 96-well plate (2,000 cells per well) in 200 µL RPMI or DMEM and incubated for 24 h at 37◦C. On the following day, the cells were treated with the platinum compounds of 50 µL, separately at varying concentrations for 72 h at 37◦C. The cells were then treated with 30 µL fresh medium containing 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) (5 mg/mL) and incubated for 4 h at 37◦C. The medium was removed, 200 µL of DMSO was added to the cells, and the absorbance of the purple formazan was recorded at 565 nm using a BioTek Elx800 microplate plate reader. Each experiment was performed in triplicate for each cell line. The IC<sup>50</sup> values were derived from Dose Response curves to obtain the Pt concentration that could reduce cell viability to 50%. All the IC<sup>50</sup> values were obtained by fitting experimental data using Origin.

# LIVE/DEAD Cell Viability Assay

A2780cis cells were cultured on 35 mm sterile glass bottom culture dishes (MATTEK Corporation) for 24 h at 37◦C. The cells were then treated with cisplatin ([Pt] = 10µM), nMOC ([Pt] = 10µM) for 48 h at 37◦C. The LIVE/DEAD assay was carried out by following the standard protocol supplied by the manufacturer.

# RESULTS AND DISCUSSION

# Preparation of the Pt6L<sup>4</sup> Metal-Organic Cage (1)

The M6L<sup>4</sup> type of metal-organic cages was first developed by Fujita (Ibukuro et al., 1998). Recently, Lippard et al. reported that a Pt6L<sup>4</sup> cage bearing a bpy aromatic chelator is able to trigger DNA damage and apoptosis via noncovalent binding (Zheng et al., 2016). In this report, we developed a Pt6L<sup>4</sup> Cage **1** bearing a larger aromatic chelator, 1,10-phenanthroline (phen), toward generating severe DNA damage and high potency. As shown in **Figure 1A**, Cage **1** is an octahedral cage formed by self-assembly of six 1,10 phenanthroline-Pt(II) centers and four 2,4,6-tris(4-pyridyl)- 1,3,5-triazine ligands (L). It is synthesized by following a reported procedure using microwave synthesis, and self-assembly was confirmed by NMR spectroscopy. The proton NMR spectrum of **1** (**Figure 1B**) matched the previous reports and the integrations confirmed a 6:4 ratio between 1,10 phenanthroline-Pt(II) centers and the pyridyl ligands (Zheng et al., 2016).

# Cellular Responses of the Pt6L<sup>4</sup> Metal-Organic Cage (1)

The cytotoxicity profiles of the metal-organic cage (**1**) was determined against a panel of human cancer cell lines using MTT [3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide] assays. As shown in **Figures 2A,B**, five human cancer cell lines were tested, including ovarian cancer cell line A2780, cisplatin-resistant ovarian cancer cell line A2780cis, ovarian cancer cell line SKOV-3, lung cancer cell line A549, and triplenegative breast cancer cell line MDA-MB-231. The cells were incubated with Cage **1** or cisplatin for 72 h, and the IC<sup>50</sup> values (concentrations required to induce 50% viability) were derived from dose–response curves (**Figure 2A**). All of the IC<sup>50</sup> values refer to platinum concentrations. As shown in **Figures 2A,B**, in A2780, A549, and MDA-MB-231, **1** exhibits IC<sup>50</sup> values comparable to those of cisplatin. Notably, **1** exhibits higher cytotoxicity against A2780cis (IC<sup>50</sup> = 4.50 ± 1.38µM) than cisplatin (IC<sup>50</sup> = 11.48 ± 1.09µM). As A2780cis cells are intrinsically resistant to cisplatin, this data shows that **1** has the potential to treat chemoresistant ovarian cells. Furthermore, we used immunoblotting analysis to evaluate the DNA damage triggered by Cage **1**. As shown in **Figure 2C**, A2780cis cells incubated with **1** ([Pt] = 2.5, 5.0, and 7.5µM) for 48 h showed a marked increase in phosphorylated H2AX (γH2AX), a classic biomarker for DNA damage. The results of MTT assays and immunoblotting analysis clearly support that Cage **1** is able to induce DNA damage and exhibit high potency against chemoresistant ovarian cancer cells. However, we found that this cage structure exhibits very limited solubility in cell culture media (<20µM), due to the high hydrophobicity of the large aromatic components. Therefore, additional formulation is essential to promote the solubility of this hydrophobic cytotoxic cage.

# Nanoformulation of the Metal-Organic Cage (1) Into nMOC

We found that combination of Cage **1** with fluorescein (**2**) and MPEG5k-PGA<sup>50</sup> (**3**) can produce nanoparticles of the metalorganic cages (nMOC) in aqueous solution. In the formulation experiment, the DMSO solution containing the metal-organic cage (**1**) (90.0 µL, 2.00 mM) was added to 1 mL PBS solution of MPEG5k-PGA<sup>50</sup> (**3**) (1.00 mg/mL) and fluorescein (**2**) (30.0 µL, 13.0 mM) with shaking at R.T. The nanoparticles were purified via centrifugal filtration, and the yield of the nanoformulation is 56.7%. As determined by GFAAS, upon nanoformulation, the concentration of **1** can reach up to 0.4 mM in aqueous solution and cell culture media. This nanoformulation is driven by the host-guest interactions and electrostatic interactions between the three components. During the formulation, significant color change from yellow green to red was observed, which is associated with the formation of the host-guest complex between Cage **1** and fluorescein (**2**). UV-vis and fluorescence spectroscopy were applied to understand such host-guest interactions. The chromatic change was characterized by UV-vis spectroscopy shown in **Figure 3A**. Increasing the molar ratio of **1** leads to red shift of the absorption spectra of **2**. Using a Job plot extracted from the UV-vis studies (the insert of **Figure 3A**), we determined the binding ratio of the cage and fluorescein to be 1:1. To further measure the dissociation constant (Kd) of the host-guest complex, we performed fluorescence titration of Cage **1** to the fluorescein (**2**) solution. As shown in **Figure 3B**, gradual addition of Cage **1** into the PBS solution of fluorescein (6µM) resulted in the decrease in the fluorescence intensity at 525 nm. As extracted from a Scatchard plot (the insert of **Figure 3B**), the experimental K<sup>d</sup> value was determined as 1.73µM. These results are consistent with our previous report using a similar cage structure, and it is believed that the fluorescein molecule is encapsulated within the cavity of Cage **1** (Yue et al., 2018a). As the final product of the formulation, nMOC encapsulating fluorescein was obtained in the presence of the pegylated anionic polymer (**3**). Driven by the electrostatic interactions, the cationic host-guest complex binds to the anionic polymer, leading their aggregation in aqueous solution. By virtue of the PEG group of the polymer, such aggregations were stabilized on a nanoscale. In the cryo-EM image (**Figure 3C**), the nanoaggregation of the cages was clearly observed, and they are around 20–40 nm. These nanocores are stabilized by their peripheral PEG. As measured by DLS (**Figure 3D**), the average size of the overall nanoparticles is 98.0 ± 8.2 nm. The particle size difference between TEM and DLS is likely due to the peripheral PEG. The PEG layer of the nanoparticles cannot be detected under TEM. In addition, due to peripheral PEG, the zeta potential of the nanoparticles

FIGURE 3 | Nanoformulation of Cage 1 using fluorescein (2) and MPEG5k-PGA50 (3). (A) Quenching of fluorescence emission of fluorescein by titration with 1 (6µM) and the Scatchard plot extracted from fluorescence titration to determine Kd; (B) Changes of the absorption spectra of the mixture of 1 and 2 in a different molar ratio (0–1.0) and the Job plot extracted from titration to determine the binding ratio of 1 and 2.; (C) A cryo-EM image (Scale bar: 20 nm) (D) DLS analysis of size distribution; (E) Zeta-potential analysis; (F) UV-vis spectra of fluorescein and nMOC in PBS ([2] = 20µM).

is close to zero (−1.0 ± 0.4 mV) (**Figure 3E**). The integrity of the cage structure within the nanoparticles was supported by the UV-vis studies. In the UV-vis spectra (**Figure 3F**), the peak corresponding to nMOC is red-shifted compared to that of free fluorescein, and this is due to the encapsulation of fluorescein by Cage **1** within the nanoparticles. The combined evidence from spectroscopic analysis, GFAAS, TEM, and DLS clearly support that the hydrophobic cage can be readily formulated in

analyzed by using flow cytometry.

well-defined nanoparticles and increase its solubility in aqueous solution.

# Cellular Uptake and Drug Release of nMOC

Next, we examined whether the nMOC can enter cancer cells and release Cage **1** and fluorescein (**2**). We determined the cellular uptake of the nanoparticles and compared that with cisplatin and MOCs using GFAAS. The results (**Figure 4A**) show that the cellular uptake of cisplatin was only 50 pmol/million cells in A2780cis cells, while the uptakes of MOCs and Cage **1** were 10 times higher. This results verified that the nanoparticles can readily enter cancer cells. Furthermore, the drug release profiles of these nanoparticles under different pH were analyzed using dialysis in micro-dialysis bags (10 kDa MWCO). In **Figure 4B**, the nanoparticles show a very slow release in PBS (pH = 7.4), and there's only 20% Pt contents released after 100 h. As a control, Cage **1** exhibits fast escape from the dialysis bag under the similar condition, the half-life of which is only less than 2 h. In acidic acetate buffer (pH = 5.0), the nanoparticles rapidly release the Pt contents. As shown in **Figure 4B**, almost half of the Pt contents were released within 72 h in the acetate buffer. Therefore, these nanoparticles are expected to stably hold the therapeutic contents before entering cells and then readily release the cage in the acidic environment of endosomes. Finally, we employed flow cytometry to verify the intracellular drug release. The A2780cis cells were treated with PBS, fluorescein (**2**), **2** + **3**, and nMOC. As a result shown in **Figure 4C**, the sample treated **2** exhibits a relatively low fluorescence intensity, and this is because fluorescein (**2**) itself has very poor cell penetration. As a control, addition of the anionic polymer did not increase the uptake. Notably, the cells treated with the nanoparticles showed marked increase in fluorescent intensity. Considering that the fluorescence signals of **2** is quenched by Cage **1** within the nanoparticles, the intense fluorescence observed in the flow cytometric analysis is resulted from the intracellularly released fluorescein. Therefore, we conclude that nMOC act as a promising system for delivering the hydrophobic cytotoxic metal-organic cages and the encapsulated fluorescein guest.

# Cellular Responses of nMOC

Finally, the in vitro efficacy and cellular responses of the nanoparticles were examined. For A2780, A2780cis, A549, SKOV-3, and MDA-MB-231 cell lines in **Figures 2A,B**, the nanoparticles exhibit IC<sup>50</sup> values comparable to those of Cage **1**. It is notable that nanoparticles maintain a lower IC<sup>50</sup> value against A2780cis cells (IC<sup>50</sup> = 4.65 ± 0.73µM) compared to cisplatin (IC<sup>50</sup> = 11.48 ± 1.09µM). Resistance factors calculated based on IC50(A2780cis)/IC50(A2780) represent an important factor to evaluate how chemoresistant ovarian cancer cells response to the therapeutic candidates. As a result of the MTT assays, the nanoformulation has a much lower resistance factor compared to that of Cage **1** or cisplatin. For nMOC, the value of IC50(A2780cis)/IC50(A2780) is 1.50, while that of Cage **1** is 2.76 and cisplatin is 12.5. Furthermore, Live/Dead cell assay was used to confirm the anticancer activity of nMOC against A2780cis cells (**Figure 2C**). A2780cis cells treated with the nanoparticles contain a higher population of dead cells, but those treated with cisplatin remained viable. In addition, the results of the immunoblotting analysis (**Figure 2D**) show that the nanoparticles act as a DNA damaging agent like Cage **1** and induced phosphorylation of H2AX (γH2AX) in the treated A2780cis cells. Finally, annexin V flow cytometric analysis was used to further verify the apoptosis that is triggered by nanoparticles. Apoptosis represents one major cellular response to DNA damage. In healthy cells, phosphatidylserine (PS) is located on the cytoplasmic surface of the cell membrane, but in cells undergoing apoptosis, PS is translocated to the exterior of the plasma membrane, exposing PS to the external cellular environment where it can be detected by annexin V conjugates. By using annexin V and PI, both early and late stage apoptosis can be identified. As shown in **Figure 2E**, the nanoparticles induced significant populations of cells to under early- and late-stage apoptosis (5.07 and 3.93%, respectively). Variation of IC<sup>50</sup> values (**Figure 2A**) of the platinum compounds across different human cell lines is attributed to different aspects of the mechanism of action. Different cell lines have different capabilities of exporting platinum compounds via efflux, different concentrations of intracellular glutathione that can deactivate platinum compounds, different nuclear DNA damaging repair capability, and different expression levels of anti-apoptotic proteins. Collectively, the cell-based studies suggest that the nanoformulation of Cage **1** maintains its mechanism of action of inducing DNA damage and apoptosis and exhibits high potency against chemoresistant ovarian cancer.

# CONCLUSION

In conclusion, we have presented a novel design of formulating cytotoxic hydrophobic metal-organic cages to facilitate their use in treating chemoresistant ovarian cancer. In this work, we demonstrated that Cage **1**, a cytotoxic hydrophobic metalorganic cage with low solubility, can be formulated into pegylated nanoparticles in presence of fluorescein (**2**) and MPEG5k-PGA<sup>50</sup> (**3**). The nanoformulation is driven by the host-guest interactions between Cage **1** and Compound **2**, confirmed by using UVvis and fluorescence spectroscopy. In presence of the anionic polymer (**3**), such cationic host-guest complexes precipitate in aqueous solution to form the nMOC covered by PEG, characterized by TEM and DLS. In the form of nMOC, the concentration of Cage **1** can reach up to 0.4 mM in aqueous solution or cell culture media. Upon entering cancer cells, the nMOC can slowly releases therapeutic content (**1**) and the fluorophore (**2**). Using MTT assays, we determined that the nMOC exhibits promising in vitro efficacy against a panel of human cancer cell lines, and notably, the nMOC shows a much lower resistance factor against chemoresistant ovarian cancer cell lines comparing to cisplatin. We envision that this study can deliver a threefold impact to the research field that focuses on applying metallosupramolecular coordination complexes in cancer therapy: First, this work demonstrates the feasibility of using pegylated anionic polymers to formulate hydrophobic cationic metal-organic cages toward promoting their solubility and biocompatibility, and ideally, this strategy can be applied to other metal-organic complexes. Moreover, this strategy also introduces the host-guest chemistry that can be further exploited for the delivery of different therapeutic entities, and this may further enhance the therapeutic effects of such systems. Finally, the promising therapeutic effects of the resulted nMOC make them possible therapeutic candidates for future studies toward treating chemoresistant ovarian cancer.

# AUTHOR CONTRIBUTIONS

HW and ZY: synthesis and characterization of Cage **1**; HW and HD: spectroscopic studies of the host-guest complex formed by Cage **1** and fluorescein; HW and AJ: preparation and characterization of nMOC; ZQ: cell-based studies; HL and BD: analysis of phosphorylation of H2AX by western blots; Y-RZ: develop the concept and design the experiments; Y-RZ and DB: prepare the manuscript.

# FUNDING

Kent State University, the Startup Fund, Farris Family Innovation Awards, LaunchPad Awards.

# ACKNOWLEDGMENTS

Y-RZ thanks the financial support provided by the startup fund, Farris Family Innovation Awards, and LaunchPad Awards provided by Kent State University. The TEM data were obtained by Dr. Min Gao (or other staff member) at the (cryo) TEM facility at the Liquid Crystal Institute, Kent State University, supported by the Ohio Research Scholars Program Research Cluster on Surfaces in Advanced Materials.

# REFERENCES


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Wang, Qiu, Liu, Jayawardhana, Yue, Daghlas, Bowers, Datta and Zheng. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Self-Assembly of a [1 + 1] Ionic Hexagonal Macrocycle and Its Antiproliferative Activity

Khushwant Singh<sup>1</sup> , Ankit Gangrade<sup>2</sup> , Sourav Bhowmick <sup>1</sup> , Achintya Jana<sup>1</sup> , Biman B. Mandal <sup>2</sup> \* and Neeladri Das <sup>1</sup> \*

*<sup>1</sup> Department of Chemistry, Indian Institute of Technology Patna, Bihta, India, <sup>2</sup> Department of Biosciences and Bioengineering, Indian Institute of Technology Guwahati, Guwahati, India*

#### Edited by:

*James D. Crowley, University of Otago, New Zealand*

#### Reviewed by:

*Sally Elisabeth Plush, University of South Australia, Australia Jyotirmayee Mohanty, Bhabha Atomic Research Centre, India Robert Elmes, Maynooth University, Ireland*

#### \*Correspondence:

*Biman B. Mandal biman.mandal@iitg.ernet.in; mandal.biman@gmail.com Neeladri Das neeladri@iitp.ac.in; neeladri2002@yahoo.co.in*

#### Specialty section:

*This article was submitted to Supramolecular Chemistry, a section of the journal Frontiers in Chemistry*

Received: *07 February 2018* Accepted: *12 March 2018* Published: *03 April 2018*

#### Citation:

*Singh K, Gangrade A, Bhowmick S, Jana A, Mandal BB and Das N (2018) Self-Assembly of a [1* + *1] Ionic Hexagonal Macrocycle and Its Antiproliferative Activity. Front. Chem. 6:87. doi: 10.3389/fchem.2018.00087* A unique irregular hexagon was self-assembled using an organic donor clip (bearing terminal pyridyl units) and a complementary organometallic acceptor clip. The resulting metallamacrocycle was characterized by multinuclear NMR, mass spectrometry, and elemental analyses. Molecular modeling confirmed hexagonal shaped cavity for this metallamacrocycle which is a unique example of a discrete hexagonal framework self-assembled from only two building blocks. Cytotoxicity of the Pt-based acceptor tecton and the self-assembled PtII-based macrocycle was evaluated using three cancer cell lines and results were compared with cisplatin. Results confirmed a positive effect of the metallamacrocycle formation on cell growth inhibition.

Keywords: ionic metallamacrocycle, supramolecular assemblies, cytotoxicity, cancer cell lines, heterocyclic compounds

# INTRODUCTION

Coordination driven self-assembly has been used conveniently in contemporary research for the construction of wide range of discrete supramolecular architectures (Chakrabarty et al., 2011; Cook et al., 2013b; Li et al., 2013; Schmidt et al., 2014; Schoedel and Zaworotko, 2014; Tanaka et al., 2014; Bhowmick et al., 2015a,b; Cook and Stang, 2015; Wang et al., 2016; Jana et al., 2017). In such complex frameworks, the building blocks (aka supramolecular tectons) are held together by using multiple ligand-metal coordination bonds (Lehn, 1995; Chakrabarty et al., 2011; Cook et al., 2013b; Li et al., 2013; Ward and Raithby, 2013; Han et al., 2014; Schmidt et al., 2014; Schoedel and Zaworotko, 2014; Tanaka et al., 2014; Bhowmick et al., 2015a,b; Cook and Stang, 2015; Johnson et al., 2015; Holloway et al., 2016, 2017; Wang et al., 2016; Jana et al., 2017). The dimensions of resultant abiotic supramolecules are guided by chemical information (stereo and geometric) contained in precursor moieties (Oliveri et al., 2008; Newkome and Shreiner, 2010; Inokuma et al., 2011; Cook et al., 2013b; Harris et al., 2013; Yoshizawa and Klosterman, 2014; Cook and Stang, 2015; Wang et al., 2015; Xu et al., 2016; Jana and Das, 2017). Intense research interest in understanding the complexity level of these artificial self-assembly processes is reflected in the quantum of reports published in this research domain (Barbara, 2001; Sauvage, 2001; Balzani et al., 2002; Beuerle, 2016; Fujita et al., 2016). Moreover, metal-mediated self-assembly studies are useful in understanding the fundamental principles of molecular self-organization in nature (Liu et al., 2007; Albertí et al., 2013; Chakraborty et al., 2013, 2015; Cook et al., 2013b; Galstyan et al., 2013; Saha et al., 2014; Sarkar et al., 2014; Chen et al., 2015; Cook and Stang, 2015; Li et al., 2015).

The library of discrete structures designed using the principles of coordination driven self-assembly (Pluth et al., 2009; Yoshizawa et al., 2009; Michito and Makoto, 2010; Thanasekaran et al., 2012; Cook et al., 2013b; Therrien, 2013; Young and Hay, 2013; Mishra and Gupta, 2014; Samanta and Mukherjee, 2014; Xu et al., 2014; Cook and Stang, 2015; Bhowmick et al., 2016) include two dimensional (2-D) structures such as metallamacrocycles (triangles, rectangles, pentagon, hexagon, square etc.) and three dimensional (3-D) frameworks (cages, boxes, barrels, prisms, Archimedean, and Platonic solid) (Stang and Olenyuk, 1997; Fujita et al., 2005; Takezawa and Shionoya, 2012; Cook et al., 2013b; Kishi et al., 2013; Chen et al., 2014; Liu et al., 2014; Lu et al., 2014; Cook and Stang, 2015; Manna et al., 2015; Zhang et al., 2017). Among the 2-D polygons, hexagonal frameworks are most interesting because such a shape is most abundantly observed in nature. Prominent examples of hexagonal motif observed in nature are honeycombs and graphite. Artificial selfassembly of hexagonal nanoscalar entities is challenging, as it requires convergence of several smaller components.

In recent years, an emerging application of these abiological supramolecules is to investigate their nature of interactions with biological systems (Wang and Lippard, 2005; Kelland, 2007; Wheate et al., 2010). These interactions include but are not limited to studies with cancer cells, DNA and proteins. Among purely inorganic complexes, Pt, Ru, and Au based ions are most popular (Mattsson et al., 2009; Ott, 2009; Berners-Price and Filipovska, 2011; Vajpayee et al., 2011, 2013; Lo et al., 2015; Ajibola et al., 2017). For example, Pt based cisplatin has been one of the most successful therapeutic agent in cancer treatment (Rosenberg et al., 1969; Vickers et al., 2004). However cisplatin has limitation (as a chemotherapeutic agent) such as uptake by healthy cells, resistance developed by target cancer cells, harmful side effects (such as nephrotoxicity, neurotoxicity and ototoxicity) and protein inactivation (Jung and Lippard, 2007; Yao et al., 2007; Wang and Guo, 2008; Todd and Lippard, 2009; Kaluderovic and Paschke, 2011; Barry and Sadler, 2013; Farrell, 2015). Considering these limitations of cisplatin's therapeutic use, synthetic organometallic and supramolecular entities are being explored as alternatives with potential application in treatment of cancer. These supramolecules are often constructed from organometallic precursor molecules. In this class of molecules, the metal ruthenium is clearly a leader (Yan et al., 1997; Therrien et al., 2008; Paul et al., 2012; Jo et al., 2017). Cook, Stang, and Chi have reviewed the biological interactions of metallacycles derived using coordination driven self-assembly protocol (Cook et al., 2013a). Majority of these are derived from Ru based precursors. It is obvious from this report that biochemical interactions of supramolecular frameworks bearing Pt(II) centers have not been explored, especially in the context of their potential as anticancer therapeutic agents.

This present work is in continuation of our research efforts to design unique supramolecular coordination complexes (SCCs) wherein we report synthesis of a discrete and nanoscalar hexagonal supramolecular complex using a new donor tecton. Furthermore, we have explored cytotoxicity of the new hexagonal SCC and its organometallic precursor. Additionally the results have been compared with cisplatin under similar conditions.

# EXPERIMENTAL SECTION

# General Details

All chemicals and anhydrous solvents used in this work were purchased from commercial sources and used without further purification. FTIR spectra were recorded in a PerkinElmer Spectrum 400 FT-IR spectrophotometer. <sup>1</sup>H and <sup>31</sup>P{1H} NMR spectra were recorded on Bruker 400 MHz spectrometer. Elemental analyses were carried out using an Elementar Vario Micro Cube elemental analyzer. ESI-MS analysis was performed using a Bruker Impact ESI-Q-TOF system. Theoretical calculations of PM6 semiempirical molecular orbital method were carried out with Gaussian 09. A549 (human lung cancer cell line), KB (human oral cancer cell line) and HaCaT (human skin keratinocyte cell line) were procured from National Centre for Cell Science (NCCS), Pune. MTT [(3-(4, 5-dimethylthiazol-2-yl)-2, 5-diphenyl tetrasodium bromide] was purchased from SigmaAldrich, USA. Ethidium homodimer-1 in 2 mL pbs, propidium iodide, Ribonuclease A were also purchased from SigmaAldrich.

# Synthesis of Compound 5

2,6-bis((3-iodophenyl)ethynyl)pyrazine **3** (0.050g, 0.093 mmol), 4-ethynyl-pyridine **4** (0.019 g, 0.187 mmol), CuI (0.017 g, 0.009 mmol) and bis(-triphenylphosphine)palladium(II) dichloride (6.52 mg, 0.009 mmol) were charged in a 50 ml Schlenk flask in the glove box. Subsequently, 10 ml dry THF and freshly distilled and degassed triethylamine (0.5 ml, 0.372 mmol) were added under nitrogen. The reaction mixture was stirred overnight at room temperature. After overnight stirring, the reaction mixture was filtered through a bed of celite. The filtrate obtained was evaporated to dryness on a rotary evaporator to obtain a crude product which was purified by column chromatography on neutral alumina by eluting with 35% ethyl acetate in hexane to isolate the desired product (**5**) as off white solid.

Yield: 0.037 g, 81%, mp 198–202◦C; <sup>1</sup>H NMR (400 MHz, CDCl3): δ 8.67 (s, 2H, Ar-H), 8.62–8.61 (dd, J = 6 Hz, 2H, Ar-H), 7.81–7.80 (m, 1H, Ar-H), 7.64–7.575 (m, 2H, Ar-H), 7.43–7.37 (m, 3H, Ar-H).13C{1H} NMR (CDCl3, 100 MHz): δ 149.8, 145.7, 139.5, 135.3, 132.9, 132.7, 130.9, 128.9, 125.5, 122.8, 121.9, 92.4, 92.3, 87.6, 86.1. IR (ATR): 3,048, 2,918, 2,849, 2,210, 1,687, 1,585, 1,503, 1,404, 1,279, 1,203, 1,154, 987, 887, 796, 671 cm−<sup>1</sup> . Anal. Calcd. For C34H18N4: C, 84.63; H, 3.76; N, 11.61. Found: C, 84.72; H, 3.84; N, 11.68.HRMS (ESI, m/z): Calculated for C34H18N<sup>4</sup> ([M + H]+): 483.16; Found: 483.16.

# Synthesis of Macrocycle 7

To the solution of **6** (30 mg, 0.020 mmol) in chloroform (4 mL) was added two equivalents of AgNO<sup>3</sup> (7.06 mg, 0.040 mmol) in one portion, and the reaction mixture was stirred overnight in the absence of light at room temperature. The precipitated AgI was filtered off over a bed of Celite, and the filtrate was collected as a yellow colored solution. Subsequently, a methanolic solution of the donor tecton **5** (0.02 mmol, 0.5 mL) was added drop wise to the aforementioned filtrate with continuous stirring. The reaction mixture was stirred for 15 h at room temperature. Solvents were removed by rotary evaporator and the product obtained which was washed several times with n-pentane to obtain a solid that was finally dried in a vacuum. The macrocycle **7** was recrystallized as an off white microcrystalline solid by slow vapor diffusion of diethyl ether in its corresponding concentrated chloroform-methanol solution.

## Macrocycle 7

Yield: 34 mg, 92%; <sup>1</sup>H NMR (400 MHz, CDCl3): δ 8.89–8.87 (dd, J = 8 Hz, 4H, Ar-H), 8.68 (s, 2H, Ar-H), 8.67 (s, 2H, Ar-H), 8.00–7.99 (m, 4H, Ar-H), 7.89–7.87 (dd, J = 8 Hz, 4H, Ar-H), 7.68–7.64 (m, 4H, Ar-H), 7.47–7.45 (m, 4H, Ar-H), 7.33– 7.32 (m, 4H, Ar-H), 1.86–1.83 (m, 8H, -CH2-), 1.26–1.20 (m, 12H, -CH3). <sup>31</sup>P NMR (162 MHz, CDCl3): δ 15.70 (<sup>1</sup> JPPt = 1,162 Hz). FTIR (ATR): 2,972, 2,922, 2,216, 2,116, 1,710, 1,602, 1,511, 1,469, 1,329, 1,145, 1,034, 888, 762, 683 cm−<sup>1</sup> . Anal. Calcd. for C82H88N8O6P4Pt2: C, 54.85; H, 4.94; N, 6.24. Found: C, 54.93; H, 4.98; N, 6.28. ESI-MS m/z found: [**7**−2NO3] <sup>2</sup><sup>+</sup> = 835.76.

# Preparation of Stock Solutions

Organometallic complex **6** and self-assembled macrocycle **7** were solubilized into cell culture grade dimethyl sulfoxide to make 10 mM stock solution. However, cisplatin was solubilized into saline (0.9% sodium chloride) to make 1 mM stock solution. Aliquots of the stock solutions were taken and preserved in −20◦C till further use.

# In Vitro Cytotoxicity

The cytotoxicity of all the compounds (**6, 7** and cisplatin) was assessed using MTT assay against A549, KB, and HaCaT cell lines. First, **6** and **7** were solubilized in DMSO to prepare stock solutions (concentration 10 mM). Subsequently, stock solutions were diluted 1,000 times in cell culture media for the treatment. The compounds were then tested at varying concentration range starts from 2.5 to 30µM. For the assay, all the cells were trypsinized and counted using hemocytometer. Approximately 5 × 10<sup>3</sup> cells were seeded into each well of a 96 well plate and allowed to adhere for 24 h. Post adherence the cells were treated with pre-determined concentration (2.5, 5, 10, 15, 20, 25, and 30µM) of all the compounds and incubated for 48 h. The media was then decanted, and cells were washed with phosphate buffer saline (PBS), pH 7.4. 100 µl of 10% MTT (5mg/ml stock) prepared in FBS free media was then added to each well and incubated at 37◦C for 4 h. The formed formazan crystals were solubilized in DMSO and absorbance was taken at 570 nm using the Multiplate reader (Tecan Infinite Pro, Switzerland). The percentage viability was calculated by normalizing the absorbance value of test sample with anuntreated control, and the seven-point dose-dependent curve was plotted to determine the IC<sup>50</sup> (IC<sup>50</sup> value is the concentration of compound at which 50% of cells are viable) value of each compound.

# Live/Dead Assay

The viability of cells was visualized under a fluorescence microscope using live/dead solution. The cells were seeded in a 96 well plate as mentioned earlier. Post adherence cells were treated for 48 h with **6-7** at 25µM concentrations and cisplatin at 10µM concentration. The media was decanted and after PBS (Phosphate-buffered saline) wash, cells were incubated with 100 µl working live/dead solution (1 µL of 4 mM Calcein AM and 4 µL of 2 mM ethidium homodimer-1 in 2 mL PBS) for 10 min at 37◦C. The live cells fluoresce green (Excitation: 495 nm, Emission: 516 nm) due to Calcein-AM uptake however dead cells fluoresce red (Excitation: 528 nm, Emission: 617 nm) due to Ethidium homodimer-1 uptake.

# RESULTS AND DISCUSSION

# Synthesis and Characterization of the Organic Donor Clip (5)

Commercially available 2,6-dichloropyrazine is the synthetic precursor for both donor and acceptor tectons employed herein for the self-assembly reaction. In the first step, 2,6 diiodopyrazine (**1**) was obtained from 2,6-dichloropyrazine, which was further reacted in two steps to yield 2,6-bis((3 iodophenyl)ethynyl)pyrazine (**3**) via the formation of 2,6 diethynylpyrazine (**2**) (**Scheme 1**) (Bhowmick et al., 2017).

In the final step, **3** was further reacted with 4-ethynyl pyridine (**4**) to yield the desired organic donor clip (**5**) as depicted in **Scheme 2**. 2,6-bis((4-ethynylpyridyl)ethynyl)pyrazine) (**5**) was obtained as a white colored compound (81% isolated yield). **5** is stable in air/moisture and has high solubility in common organic solvents. **5** was fully characterized by FT-IR and NMR spectroscopy, mass spectrometry and elemental analyses.

In the <sup>1</sup>H NMR of **5** (Figure S1), signal appearing at δ = 8.62 ppm and δ = 7.64 ppm corresponds to the α and β pyridyl protons respectively. As expected, signals corresponding to these protons appear as doublets. All characteristic peaks corresponding to pyrazine, phenyl and pyridyl units of **5** were observed in <sup>13</sup>C {1H} NMR spectrum (Figure S2).

# Application of 5 as a Donor Tecton toward Self-Assembly of a Supramolecular Hexagon

Coordination driven self-assembly reactions have been employed for the construction of complex yet discrete metallomacrocycles. Among various shapes reported in the literature, hexagonal macrocycles are especially fascinating because of nature's preference for hexagon motif (honeycomb, graphite, snowflakes, etc.) (Pucci et al., 2000). Consequently, there is interest in mimicking nature in laboratory in the context of design and synthesis of macrocycles with hexagonal cavity. The most commonly reported methodologies involve self-organization of 12 (twelve) or 6 (six) tectons, thereby generating hexagonal assemblies that are referred to as [6 + 6] or [3 + 3] hexagonal metallacycles respectively (Cook and Stang, 2015). These syntheses are challenging as they require self-assembly of a large number of molecular components in a reaction that is entropically unfavorable. More recently, we have reported design and synthesis of hexagonal polygons requiring selfassembly of two donor and two acceptor tectons thereby yielding [2 + 2] hexagons. In this report, we have chosen donor and acceptor in such a manner that construction of a hexagonal macrocycle requires only one donor and one acceptor

tecton. Thus the anticipated framework will be a unique [1 + 1] hexagonal macrocycle requiring the least number of selfassembling molecular tectons for its construction.

In order to perform the above mentioned self-assembly reaction, first the acceptor tecton (**6**) was reacted with two equivalents of AgNO<sup>3</sup> in CHCl<sup>3</sup> to yield the corresponding dinitrate derivative which was further reacted with the donor tecton (**5**) (methanol solution) in 1:1 stiochiometric ratio at room temperature (**Scheme 3**). A solid product was obtained by evaporating the solvents. Subsequent washing (with n-pentane) and recrystallization yielded the desired product (**7**) as an offwhite microcrystalline solid (yield > 90%) that was soluble in organic solvents.

The self-assembled product (**7**) was subjected to various characterization techniques to confirm its purity and composition. In <sup>31</sup>P{1H} NMR spectrum (**Figure 1A**) a sharp singlet (δ = 15.70 ppm), accompanied with a pair of <sup>195</sup>Pt satellite peaks (<sup>1</sup> JPPt = 1,162 Hz), indicated that **7** has a highly symmetrical structure, wherein all phosphorous nuclei are chemically equivalent. The <sup>1</sup>H NMR spectrum of **7** also suggested incorporation of both tectons (**5** and **6**) in it (**Figure 1B** and Figures S3, S4). In <sup>1</sup>H NMR spectrum of **7** (**Figure 1B**), two peaks appearing at 8.67 and 8.68 ppm were assigned to the protons of pyrazine moieties of donor (**5**) and acceptor clip (**6**) respectively. Four sets of signals at 8.00–7.99, 7.68–7.64, 7.49–7.45, 7.33–7.32 ppm are due to protons in phenyl ring. Two sets of signals at 8.89–8.87, 7.89–7.87 ppm are due to protons of the pyridyl ring. The ethyl protons of PEt<sup>3</sup> groups (attached to PtII center) are observed in the range 1.86–1.20 ppm spectrum (Figure S3).

The formation of **7** via coordination of **5** (donor tecton) with **6** (acceptor tecton) was also evident from the observed downfield shift (1δ ∼ 0.25 ppm) of both α and β pyridyl protons present in the terminal pyridine moieties (Figure S5). This has been attributed to the decrease in electron density in pyridine rings of the donor (**5**) due to their coordination (via lone pair on nitrogen) with PtII metal centers of the acceptor tecton (**6**).

Furthermore, the purity of product **7** was confirmed by <sup>1</sup>H DOSY NMR spectroscopy (Figure S6), wherein a single trace spectrum indicated the formation of a single product. This result also confirms absence of other macrocyclic species or oligomers as byproduct and exclusive formation of discrete macrocyclic species. Thus NMR analyses confirm purity of the **7** and indicated formation of single, highly symmetrical species. Sharp peaks in <sup>1</sup>H and <sup>31</sup>P NMR are also clear indications of presence of discrete species and absence of any oligomers. Mass spectrometry was subsequently used to confirm the proposed composition of **7**. The

ESI-TOF-MS spectrum of **7** indicated formation of desired [1 + 1] molecular ensembles (**Figure 2**). The ESI-TOF-MS spectrum of **7** showed a signal corresponding to the consecutive loss of two nitrate counter anions from the expected [1 + 1] macrocycle at m/z = 835.76 [**7**-2NO3] <sup>2</sup>+. The isotopic resolution of this peak was in excellent agreement with the theoretically predicted isotopic distribution pattern of [**7**-2NO3] <sup>2</sup><sup>+</sup> (**Figures 2A,B** and Figure S8). Thus mass spectrometric analysis of **7** confirmed the formation of a discrete [1 + 1] self-assembled macrocycle.

All attempts to grow X-ray quality single crystals of **7** were unsuccessful. In such a scenario, molecular modeling (using PM6 semiempirical MO method) (Stewart, 2007) was employed to obtain useful structural information for **7**. The energy minimized structure of **7** (**Figure 3** and Figure S7) confirms the formation of a macrocycle with hexagonal cavity. The distance between the two exocyclic nitrogen atoms of pyrazine rings was 2.53 nm. The distance between the two platinum centers was found to be 1.30 nm. The lengths of sides of hexagon were found to be 1.62 and 0.68 nm. Thus **7** may be described as an irregular hexagon since all the sides are not of equal length even though the polygonal framework is equiangular. A slight distortion from the square geometry was observed at the two Pt(II) centers.

# Cytotoxicity (in Vitro) Assessment and Estimation of IC<sup>50</sup>

Till date, platinum based molecules have been used most frequently as anticancer therapeutic agents. Cisplatin is undoubtedly the most popular anticancer drug bearing Pt(II). In the first step of its mechanism of action (as anticancer drug), the chlorides of cisplatin are substituted/exchanged by hydroxide ions or water. The resulting diaquo species subsequently bind with nucleophilic sites present in RNA or DNA. Thus cytotoxicity of cisplatin stems from the fact that Pt-Cl bonds are unstable under physiological conditions relative to the new Pt-N bonds that are formed due to adduct formation with nucleobases. Also it is known that Pt-N bonds are essentially irreversible under the same physiological conditions (Farrell, 1999). Moreover, in the context of platinum based complexes as anticancer drugs, it is well known that cisplatin is quite potent while transplatin is inactive. This is a consequence of the difference in the stereospecificity of the labile Pt-Cl bonds in these two isomers. **6** and **7** are quite different from cisplatin in this aspect as these species don't have two labile (halide/hydroxo) groups in cis orientation to facilitate adduct (bifunctional) formation with DNA (as observed with cisplatin). It was therefore our curiosity to study the cytotoxicity of Pt(II) based compounds (**6** and **7**) reported herein that don't posses two cis-oriented labile groups unlike cisplatin.

The cytotoxicity of platinum containing **6** and **7** were examined against three cell lines KB (human oral cancer), A549 (human lung cancer) and HaCaT (human skin keratinocyte) using MTT assay. Cisplatin was used as a positive control (Wang and Lippard, 2005). Toxicity of all the compounds was observed to be dose and cell dependent. The IC<sup>50</sup> value is the inhibitory concentration of a compound at which 50% cells are dead. A seven-point dose-dependent plot (**Figure 4**) was recorded for each compound against each cell line for measurement of IC<sup>50</sup> concentration. **7** exhibited higher toxicity (relatively lesser IC<sup>50</sup> value) relative to **6** against A549 and KB cancer cells. However, for HaCaT cells, both **6** and **7** exhibited comparable IC<sup>50</sup> value (**Table 1**).

From the IC<sup>50</sup> values listed in **Table 1**, it is clear that compounds **6** and **7** have lesser toxicty relatively to cisplatin. This was anticipated considering the absence of two cis-oriented labile groups in **6** and **7**. In other words, the general higher activity of cisplatin over other **6** and **7** is due to the presence of cis-oriented more labile Pt-Cl bonds in cisplatin that tend to hydrolyze

easily under physiological conditions to facilitate binding with DNA/RNA. On the other hand, the PtII-based compounds in this report (**6** and **7**) have Pt-N bonds that tend to be relatively stable under such conditions.

It is however noteworthy that though **6** and **7** don't have labile groups in cis orientation to facilitate adduct (bifunctional) formation with DNA, these species show reasonable cytotoxic effect (**Table 1**). More interestingly, the macrocycle (**7**) exhibited superior cytotoxic effect when compared with cisplatin for A549 (human liver cancer) cells. Additionally, it was observed that upon self-assembly and formation of the macrocycle **7**, the resulting supramolecular framwork exhibited lower IC<sup>50</sup> value relative to that observed for the acceptor tecton (**6**) in case of A549 and KB carcinoma cell. For, HaCaT cell, the IC<sup>50</sup> values are comparable for both **6** and **7**. Considering the literature reported stability of Pt-N bonds under physiological conditions (Farrell, 1999), it may be assumed that macrocycle **7** remains intact under these conditions.

Subsequently, the effect of Pt-based compounds (**6** and **7**) on cell viability and attachment was studied and results compared with cisplatin. Cancer cells (KB, A549, and HaCaT) were treated with compounds (**6**, **7** and cisplatin) and these were stained with fluorescent dyes (Calcein AM and Ethidium homodimer-1). In these fluorescent imaging studies, cancer cells were treated with 25µM **6**-**7** and 10µM cisplatin for 48 h. Post-treatment, cells were stained with a live/dead stainsolution. Fluorescent images are shown in **Figure 5**, wherein live cells (green color) were firmly attached to tissue culture plate (TCP) surface relative to dead cells

compounds treated group compared to the untreated group.

(red color). Variable dead cell population was observed in each case due to either their loose attachment to TCP or difference in their respective IC50. These results suggest that **6** and **7** indeed demonstrate cancer cell growth inhibition to an extent that is comparable cisplatin (a widely used Pt-drug for treatment on various forms of cancer).

# CONCLUSION

In conclusion, we have reported a new donor tecton (**5**) with two pendant pyridine rings that is derived from commercially available 2,6-dichloropyrazine. **5** is a new ditopic donor tecton with 0◦ angular orientation of the two pyridine rings. Selfassembly reaction of **5** with a complementary ditopic acceptor tecton (**6**) resulted in the formation of a single product (**7**). NMR and elemental analyses confirmed its purity while mass spectrometry (ESI-MS) data supported the formation of a [1 + 1] supramolecular species. PM6 molecular modeling suggested


formation of a macrocycle with hexagonal cavity. The product (**7**) is a unique example in literature wherein a nanoscalar, discrete and cationic hexagonal framework has been synthesized using only two tectons—one unit of donor and one unit of acceptor. Previous report of all Pt(II)-based hexagonal macrocycles utilized four, six or twelve tectons resulting in highly charged species. Furthermore the PtII-based species (**6** and **7**) were subjected to cytotoxicity studies using three different cancer cell lines, and the results were compared with cisplatin. Although **6** and **7** are less potent than cisplatin, it was observed that formation of the metallamacrocycle (**7**) improved cytotoxicity in case of treatment with A549 and KB cancer cells. This is in spite of the fact that acceptor tecton (**6)** contains labile Pt-I bonds while macrocycle **7** contains platinum bonds to nitrogen that are essentially irreversible under physiological conditions (Farrell, 1999). Interestingly, the supramolecular ensemble (**7**) showed better A549 cell growth inhibitory effect than cisplatin. The cytotoxic behavior of **6** and **7** may be attributed to their noncovalent interaction with DNA. These results will form the basis of further research on the mechanism of cell-killing action of these Pt-based species which are stereo-chemically quite different from cisplatin. Nevertheless this report is a unique example, wherein a supramolecular hexagonal framework with only two Pt(II) centers has been self-assembled and it potential application in anticancer therapy has studied. Development of similar Pt(II)-based self-assembled structures as therapeutic agents for malignant cells is currently being explored in our laboratory.

# AUTHOR CONTRIBUTIONS

ND conceived the research and supervised the experimental work related to synthesis and characterization of organic/organometallic molecules reported herein. KS synthesized all new compounds reported in this manuscript. AJ optimized the energy-minimized geometry of the metallacycles **7**. SB assisted in the synthesis of some of the literature reported pyrazine precursors. AG carried out the experiments related to evaluation of biological activity of **6, 7** and cisplatin reported herein and these biological studies were supervised by BM. All authors have contributed to interpretation of results, compiled the manuscript and have approved the final manuscript.

# ACKNOWLEDGMENTS

ND thanks the Indian Institute of Technology (IIT) Patna for financial support. KS and AJ thank UGC, New Delhi,

# REFERENCES


India, for a Research Fellowship. SB thanks IIT Patna for an Institute Research Fellowship. The authors also acknowledge SAIF-Panjab University and SAIF-IIT Patna for analytical facilities. BM gratefully acknowledges funding support through the Department of Biotechnology and Department of Science and Technology, Government of India.

# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fchem. 2018.00087/full#supplementary-material

<sup>1</sup>H and <sup>13</sup>C{1H} NMR spectra of **5**, <sup>1</sup>H, <sup>31</sup>P{1H} and <sup>1</sup>H DOSY NMR spectra of the metallamacrocycle (**7**), comparison <sup>1</sup>H NMR spectra of donor ligand **5** and macrocycle **7** and **s**pacefilled model of macrocycle **7** optimized by PM6 semiempirical molecular orbital method.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Singh, Gangrade, Bhowmick, Jana, Mandal and Das. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Eukaryotic Cell Toxicity and HSA Binding of [Ru(Me4phen)(bb7)]2<sup>+</sup> and the Effect of Encapsulation in Cucurbit[10]uril

Biyun Sun<sup>1</sup> , Ian F. Musgrave<sup>2</sup> , Anthony I. Day <sup>1</sup> , Kirsten Heimann3,4, F. Richard Keene5,6 \* and J. Grant Collins <sup>1</sup> \*

*<sup>1</sup> School of Physical, Environmental and Mathematical Sciences, University of New South Wales, Australian Defence Force Academy, Canberra, ACT, Australia, <sup>2</sup> Discipline of Pharmacology, Adelaide Medical School, University of Adelaide, Adelaide, SA, Australia, <sup>3</sup> College of Medicine and Public Health, Flinders University, Adelaide, SA, Australia, <sup>4</sup> College of Science and Engineering, James Cook University, Townsville, QLD, Australia, <sup>5</sup> Department of Chemistry, School of Physical Sciences, University of Adelaide, Adelaide, SA, Australia, <sup>6</sup> Australian Institute of Tropical Health and Medicine/Centre for Molecular Therapeutics, James Cook University, Townsville, QLD, Australia*

#### Edited by:

*James D. Crowley, University of Otago, New Zealand*

#### Reviewed by:

*Julia Lorenzo, Autonomous University of Barcelona, Spain Amanda Eckermann, Hope College, United States*

\*Correspondence:

*F. Richard Keene richard.keene@adelaide.edu.au J. Grant Collins g.collins@adfa.edu.au*

#### Specialty section:

*This article was submitted to Supramolecular Chemistry, a section of the journal Frontiers in Chemistry*

Received: *07 September 2018* Accepted: *14 November 2018* Published: *30 November 2018*

#### Citation:

*Sun B, Musgrave IF, Day AI, Heimann K, Keene FR and Collins JG (2018) Eukaryotic Cell Toxicity and HSA Binding of [Ru(Me*4*phen)(bb*7)*]* 2+ *and the Effect of Encapsulation in Cucurbit[10]uril. Front. Chem. 6:595. doi: 10.3389/fchem.2018.00595* The toxicity (IC50) of a series of mononuclear ruthenium complexes containing bis[4(4′ -methyl-2,2′ -bipyridyl)]-1,*n*-alkane (bbn) as a tetradentate ligand against three eukaryotic cell lines—BHK (baby hamster kidney), Caco-2 (heterogeneous human epithelial colorectal adenocarcinoma) and Hep-G2 (liver carcinoma)—have been determined. The results demonstrate that *cis*-α-[Ru(Me4phen)(bb7)]2<sup>+</sup> (designated as α-Me4phen-bb7, where Me4phen = 3,4,7,8-tetramethyl-1,10-phenanthroline) showed little toxicity toward the three cell lines, and was considerably less toxic than *cis*-α-[Ru(phen)(bb12)]2<sup>+</sup> (α-phen-bb12) and the dinuclear complex [{Ru(phen)2}2{µ-bb12}]4+. Fluorescence spectroscopy was used to study the binding of the ruthenium complexes with human serum albumin (HSA). The binding of α-Me4phen-bb<sup>7</sup> to the macrocyclic host molecule cucurbit[10]uril (Q[10]) was examined by NMR spectroscopy. Large upfield <sup>1</sup>H NMR chemical shift changes observed for the methylene protons in the bb<sup>7</sup> ligand upon addition of Q[10], coupled with the observation of several intermolecular ROEs in ROESY spectra, indicated that α-Me4phen-bb<sup>7</sup> bound Q[10] with the bb<sup>7</sup> methylene carbons within the cavity and the metal center positioned outside one of the portals. Simple molecular modeling confirmed the feasibility of the binding model. An α-Me4phen-bb7-Q[10] binding constant of 9.9 ± 0.2 × 10<sup>6</sup> M−<sup>1</sup> was determined by luminescence spectroscopy. Q[10]-encapsulation decreased the toxicity of α-Me4phen-bb<sup>7</sup> against the three eukaryotic cell lines and increased the binding affinity of the ruthenium complex for HSA. Confocal microscopy experiments indicated that the level of accumulation of α-Me4phen-7 in BHK cells is not significantly affected by Q[10]-encapsulation. Taken together, the combined results suggest that α-Me4phen-7 could be a good candidate as a new antimicrobial agent, and Q[10]-encapsulation could be a method to improve the pharmacokinetics of the ruthenium complex.

Keywords: ruthenium complexes, cytotoxicity, HSA binding, cucurbit[10]uril, supramolecular chemistry

# INTRODUCTION

Due to the increasing resistance of bacteria, particularly Gramnegative species, to the range of drugs currently in clinical use there is significant interest in developing new antimicrobial agents (Boucher et al., 2009). While there is an on-going effort to produce new antimicrobial drugs based upon analogs of known scaffolds, e.g., β-lactam antibiotics (Boucher et al., 2009), there is an increasing focus on the use of metal-based compounds (Richards et al., 2009; Neelakantan et al., 2010; Ng et al., 2013; Pandrala et al., 2013). In particular, and reflecting a greater recognition of their therapeutic potential (Bergamo and Sava, 2007; Moucheron, 2009; Süss-Fink, 2010; Gill and Thomas, 2012; Li et al., 2018), has been the growing development of ruthenium complexes as potential antimicrobial agents (Li et al., 2015a; Southam et al., 2017; Mital and Ziora, 2018). Dwyer and coworkers demonstrated more than 60 years ago the activity of mononuclear polypyridylruthenium(II) complexes against both Gram-positive and Gram-negative bacteria (Dwyer et al., 1952, 1969). However, it has only been over the last 10 years that there has been renewed and widespread interest in examining the antimicrobial properties of a broad range of ruthenium complexes (Kumar et al., 2009, 2016; Bolhuis et al., 2011; Li et al., 2011, 2013a, 2014; Shobha Devi et al., 2013; Gorle et al., 2016).

Previous studies from our group have examined the antimicrobial properties of di-, tri- and tetra-nuclear polypyridylruthenium(II) complexes in which the metal centers are linked by the bis[4(4′ -methyl-2,2′ -bipyridyl)]- 1,n-alkane ligand ("bbn"; see **Figure 1**) (Li et al., 2011, 2016; Gorle et al., 2014). These oligonuclear ruthenium complexes showed excellent activity against Gram-positive bacteria, and maintained the activity against current drug-resistant strains such as methicillin-resistant Staphylococcus aureus (MRSA) and vancomycin-resistant enterococci (VRE) (Gorle et al., 2016). However, the oligonuclear complexes showed variable activity to Gram-negative strains, with a number of species being essentially resistant to the ruthenium complexes (Gorle et al., 2016).

As the toxicity to both bacterial and eukaryotic cells increased with the number of ruthenium centers in the oligonuclear complex (Li et al., 2012, 2015b), we sought to examine mononuclear complexes that contained the bb<sup>n</sup> moiety as a tetradentate ligand—[Ru(phen′ )(bbn)]2<sup>+</sup> complexes (where phen′ = 1,10-phenanthroline and a variety of its derivatives, see **Figure 2**) (Gorle et al., 2015; Sun et al., 2018). Although some of these mononuclear complexes particularly cis-α-[Ru(Me4phen)(bb7)]2<sup>+</sup> (Me4phen = 3,4,7,8 tetramethyl-1,10-phenanthroline), designated as α-Me4phen-7; and cis-α-[Ru(phen)(bb12)]2+, designated as α-phen-12 showed good and uniform activity against both Gram-positive and Gram-negative species, they were slightly less active than the oligonuclear complexes to most bacteria (Sun et al., 2018). However, while the toxicity toward eukaryotic cells has been established for the oligonuclear complexes (Li et al., 2015b), the corresponding data for the mononuclear complexes have yet to be reported.

Another important factor governing the clinical potential of a new compound is its ability to bind serum proteins (Kratochwil et al., 2002; Bohnert and Gan, 2013; Liu et al., 2014). Blood serum protein binding can affect the distribution and clearance of a drug in several ways: only the free (non-protein bound) drug can act as an antimicrobial agent; alternatively, non-protein bound small drugs are cleared more rapidly from the bloodstream. Again, while the binding of the oligonuclear complexes to human serum albumin (HSA)–the most abundant serum protein (0.6 mM in healthy humans)–has been determined for the oligonuclear complexes (Li et al., 2013b), it is yet to be reported for the mononuclear complexes.

In the present study, we have determined the toxicity of the mononuclear [Ru(phen′ )(bbn)]2<sup>+</sup> complexes toward a panel of eukaryotic cells and examined their HSA binding ability. Furthermore, as encapsulation in cucurbit[n]urils {macrocyclic host compounds composed of n glycoluril monomeric units, Q[n]–see **Figure 3** (Lagona et al., 2005; Kim et al., 2007; Isaacs, 2009)} can potentially decrease the toxicity of a drug and modulate the drugs' ability to bind serum proteins, (Wheate et al., 2004; Jeon et al., 2005; Li et al., 2013b) we have also examined the ability of α-Me4phen-7 to form an inclusion complex with Q[10] and determined the subsequent effect on the toxicity of the ruthenium complex toward eukaryotic cells and its ability to bind HSA.

# EXPERIMENTAL

# Materials

The ruthenium complexes used in this study {cis-α- [Ru(phen)(bb7)]2<sup>+</sup> (designated as α-phen-bb7), cis-α- [Ru(Me2phen)(bb7)]2<sup>+</sup> (α-Me2phen-bb7), cis-α-[Ru(Me4 phen)(bb7)]2<sup>+</sup> (α-Me4phen-bb7), cis-α-[Ru(NO2phen)(bb7)]2<sup>+</sup> (α-NO2phen-bb7), cis-α-[Ru(phen)(bb12)]2<sup>+</sup> (α-phen-bb12), [Ru(Me4phen)3] <sup>2</sup><sup>+</sup> and Rubb12} and Q[10] were synthesized as previously described (Liu et al., 2005; Pisani et al., 2010; Gorle et al., 2015; Sun et al., 2018).

# Toxicity

The toxicities of the ruthenium complexes were assayed against three eukaryotic cell lines: BHK (baby hamster kidney); Caco-2 (heterogeneous human epithelial colorectal adenocarcinoma); and Hep-G2 (liver carcinoma). All cell lines were cultured in 75 mL culture flasks in DMEM culture media (Dulbecco's Modified Eagle's Medium, Gibco, Thermo Fisher, UK) supplemented with 10% fetal bovine serum (Gibco, Thermo Fisher, UK), 1% MEM Non-Essential Amino Acid (Sigma, UK) and 1% penicillin-streptomycin (Sigma) at 37◦C in an atmosphere of 5% humidified CO2. Cells used in the study were in the logarithmic growth phase, and unless stated were grown to 70% confluence, and then trypsinized with 0.5% trypsin-EDTA (Gibco, Canada) for detachment and use in the assays.

Half-maximal inhibitory concentrations (IC50) of the ruthenium complexes against BHK, Caco-2 and HepG2 cell lines were performed using the mitochondrial-dependent reduction of 3-(3,4-dimethylthiazol-2-yl)-5-diphenyl tetrazolium bromide (MTT) to formazan as previously described (Sundaraneedi et al., 2017). The cells were cultured in 96-well microtiter plates containing the respective media to a cell density of 3,000 per well at 37◦C in an atmosphere of air containing 5% CO2. Cell viability was assessed after continuous exposure to various concentrations

of drugs (ranging from 0.5 to 400µM) for 24 and 48 h. Cisplatin served as a treatment evaluation standard. Ruthenium complex stocks were made to the required concentration in sterile Milli-Q water, and cisplatin solutions were made to the required concentration in culture medium before the treatment. The amount of MTT reduced to formazan within the cells was quantified by measuring the absorbance at λ = 570 nm using a Fluostar Galaxy Microplate Reader. The average values presented are based on at least three independent experiments. The IC<sup>50</sup> values were determined using GraphPad Prism 7.0 (GraphPad Software, San Diego, USA).

# HSA Binding

All solutions used in the protein binding studies were dissolved in a 0.1 M sodium phosphate buffer (pH = 7.4). Protein solutions (5µM) were titrated with the ruthenium complexes from 2 mM stock solutions, from a metal complex/protein ratio of 0.0 to 8.0. The maximum fluorescence for HSA (Sigma-Aldrich) was observed at λ = 345 nm after excitation at λ = 280 nm. All experiments were carried out in triplicate, with results presented as mean and standard deviation.

# Encapsulation by Q[10] NMR

<sup>1</sup>H NMR titrations were carried out by the addition of solid Q[10] (0.4 mg) into a solution of α-Me4phen-bb<sup>7</sup> (2 mM, 600 µL) to a 1:1 Q[10]:α-Me4phen-bb<sup>7</sup> ratio. ROESY NMR experiments for the free and Q[10]-encapsulated α-Me4phen-bb<sup>7</sup> were conducted with 2048 data points in t2 for 256 t1 values, with a pulse repetition delay set to 1.7 s and mixing times of 350 ms.

#### Molecular Modeling

The encapsulation of α-Me4phen-7 in the Q[10] cavity was examined by molecular modeling using the HyperChem (version 8, HyperCube, Inc.) program. The bb<sup>7</sup> alkyl chain was positioned deep within the Q[10] cavity and the ruthenium metal center positioned at one of the portals in a manner consistent with the NMR results. The Q[10] encapsulated α-Me4phen-7 was then optimized using the Amber99 molecular mechanics forcefield without explicitly treating water molecules (i.e., "in vacuo"). Successive geometry optimizations using slightly different starting conformations and different minimization algorithms produced similar results.

#### Luminescence Titration

Spectra were recorded on a Horiba Scientific FluoroMax-4 fluorescence spectrophotometer at room temperature in a 1 cm quartz cuvette. Samples were excited at λ = 435 nm and emissions were recorded in the λ = 510–800 nm range with slit widths of 5 nm in increments of 1 nm. All luminescence spectroscopy samples were prepared in Milli-Q water. Luminescence titrations were carried out by titration of 20 µL aliquots of Q[10] (12.5µM) stock solution into a α-Me4phen-7 (0.675µM; 2,950 µL) solution to a 1.5:1 ratio. The sample was then sparged with argon to remove any oxygen present and allowed to equilibrate for 3 min before measurements were taken.

The binding constant (K<sup>b</sup> ) for α-Me4phen-7 binding to Q[10] was calculated using Equation 1 (Pisani et al., 2010):

$$K\_b = \frac{[\text{MC}]\_b}{[\text{MC}]\_f \times [\text{Q10}]\_f} \tag{1}$$

where [MC]<sup>f</sup> and [Q10]<sup>f</sup> are the concentrations of free α-Me4phen-7 and Q[10], respectively, at each titration point. [MC]<sup>b</sup> was determined using Equation 2:

$$[\text{MC}]\_b = \frac{[\text{MC}]\_t}{1 - \left(\frac{\delta\_{\text{obs}} - \delta\_b}{\delta\_f(\text{x}) + \delta\_b(\text{x})}\right)} \tag{2}$$

where [MC]<sup>t</sup> is the concentration of metal complex in the sample, δobs is the luminescence intensity maxima observed at each titration point, δ<sup>b</sup> is the luminescence intensity maximum of the bound metal complex, δ<sup>f</sup> is the luminescence intensity maximum of the free metal complex and x is the ratio of bound/free metal complex at each titration point. The binding constant at each point was calculated and averaged to give an approximate binding constant.

#### Confocal Microscopy

The trypsinized BHK cells were seeded on sterile, poly L-lysine coated, coverslips in a 24-well plate. The ruthenium complex was applied to the cells in growth media to make the desired concentration (25µM) and incubated at 37◦C with 5% CO<sup>2</sup>

for 1 h. Following the incubation, 100 nM Mitotracker <sup>R</sup> Green FM (Invitrogen) was added for mitochondrial staining. Staining was carried out in DMEM medium under standard cultivation conditions as per the manufacturer's instructions. Following staining, the coverslips were gently rinsed with phosphate buffer solution (PBS, pH = 7.1) prior to confocal laser scanning microscopy.

The cellular accumulation of the ruthenium complex was determined using a laser scanning confocal microscope (FV3000, Olympus). Samples were viewed under 60× silicone immersion using the following excitation (λex) and emission (λem) wavelengths. Ruthenium complex (λex = 450 nm, λem = 630 nm) and Mitotracker Green FM (λex = 490 nm, λem = 516 nm) were excited using a blue argon laser (λex = 488 nm), and emissions were collected at λ = 600–650 nm for the ruthenium complex and λ = 500–550 nm for Mitotracker Green. Image data acquisition and processing was performed using Olympus FV31S-SW software. The luminescence intensity of cis-α-[Ru(Me4phen)(bb7)]2<sup>+</sup> inside the BHK cells was quantified using the surface rendering function of Imaris 9.2.1 software, and the data analyzed for statistical difference using the Wilcoxon-test and t-test methods.

# RESULTS

# Toxicity Against Eukaryotic Cells

The toxicities of the [Ru(phen′ )(bbn)]2<sup>+</sup> complexes against three eukaryotic cell lines {baby hamster kidney (BHK), human epithelial colorectal adenocarcinoma (Caco-2) and human hepatocellular carcinoma (Hep-G2)} were determined and compared to the values obtained for the dinuclear complex Rubb12, [Ru(Me4phen)3] <sup>2</sup><sup>+</sup> {designated (Me4phen)3, the most active of the mononuclear complexes originally identified by Dwyer and co-workers} and the control anticancer agent cisplatin. The BHK and Hep-G2 cell lines were chosen so that the IC<sup>50</sup> results could be compared to those from our previous studies with multinuclear ruthenium complexes (Li et al., 2015b). It was also of interest to examine the toxicity of the ruthenium complexes against a third eukaryotic cell line that was distinct from kidney or liver cells—consequently, the Caco-2 cell line was also used in this study. The results are summarized in **Table 1**. The IC<sup>50</sup> values determined for Rubb<sup>12</sup> against the BHK and Hep-G2 cell lines in this study were similar to those obtained in an earlier study (Li et al., 2015b). Of note, α-Me4phen-7 showed no toxicity (>400µM) against BHK and Caco-2 cells, and very low activity (≈ 150µM) against Hep-G2 cells for a 24 h incubation and very low (BHK and Hep-G2) or no toxicity (Caco-2) for a 48-h incubation. By contrast, α-phen-12 was (2–5)-fold more toxic than α-Me4phen-7, while Rubb<sup>12</sup> was generally much more toxic to the eukaryotic cells than either of the mononuclear complexes.

# HSA Binding

The binding of the [Ru(phen′ )(bbn)]2<sup>+</sup> complexes to human serum albumin (HSA) was examined by fluorescence spectroscopy. HSA has one fluorescent tryptophan residue that can be used as a fluorophore in drug-binding experiments. TABLE 1 | IC50 values (µM) of the ruthenium complexes against the BHK, Hep-G2 and Caco-2 cell lines for 24- and 48-h incubations.


The relative changes in HSA fluorescence upon binding of selected ruthenium complexes are shown in **Figure 4**.

All ruthenium complexes showed significant binding affinity with HSA. Consistent with a previous HSA binding study of the Rubb<sup>n</sup> complexes (Li et al., 2013b), the decrease in the tryptophan fluorescence was found to be due to static quenching, a positional change of the tryptophan residue induced by the binding of the ruthenium complex to the protein.

The static quenching via binding of the ruthenium complexes to HSA can be represented as:

$$\text{nRu} + \text{HSA} \rightarrow \text{nRu} \cdots \text{HSA}$$

where nRu...HSA denotes the quenched ruthenium complex/HSA species. A double-logarithm equation of log10[(F0-F)/F] vs. log10[Ru] can then be used to express the equilibrium between the free and bound ruthenium(II) complex to determine the apparent binding constant (Kapp) and the binding number (n) (Krause-Heuer et al., 2012).

$$\log\left[\frac{F\_0 - F}{F}\right] = \log\_{10} K\_{\text{app}} + \ln \log\_{10} \text{[Ru]}$$

It was not possible to determine a single apparent HSA binding constant, as a linear relationship was not observed over the entire concentration range of added ruthenium complex in the plot of log10[(F<sup>0</sup> – F)/F] vs. log10[Ru]. However, two distinct linear sections could be found in the analysis of the double logarithm, corresponding to ruthenium complex to HSA ratios of (a) 0 to 2.4 and (b) 2.4 to 8. As previously observed (Li et al., 2013b), the two binding ratio regions represent an initial 1:1 binding at low ruthenium complex concentration (ratios 0–2.4), and then upon addition of further ruthenium complex, a region (2.4–8) where multiple ruthenium complexes (≥ 2:1) are bound to HSA. The higher affinity 1:1 HSA binding constant Kapp and the binding numbers n for the ruthenium complexes determined at low ratios of added ruthenium complex are summarized in **Table 2**.

For the [Ru(phen′ )(bb7)]2<sup>+</sup> complexes, the HSA binding affinity increased with the number of methyl substituents on

the 1,10-phenanthroline ligand; however, the α-NO2phen-bb<sup>7</sup> complex showed the highest binding affinity. Interestingly, the α-Me4phen-7 complex bound HSA with significantly higher affinity than the [Ru(Me4phen)3] <sup>2</sup><sup>+</sup> and α-phen-bb<sup>12</sup> complexes, and with approximately the same affinity as previously determined for Rubb<sup>12</sup> (Kapp = 4.7 × 10<sup>5</sup> M−<sup>1</sup> ) (Li et al., 2013b).

## Q[10] Encapsulation of α-Me4phen-7 NMR Studies

As α-Me4phen-7 exhibited the best differential (≈ 20-fold) between high antimicrobial activity (previously demonstrated in Sun et al., 2018) and low toxicity, it was selected for the Q[10] binding studies. **Figure 5** shows the <sup>1</sup>H NMR spectrum of α-Me4phen-7 and the spectra of the metal complex with added Q[10] at various Q[10] to α-Me4phen-7 ratios (R). The <sup>1</sup>H NMR resonances of α-Me4phen-7 and the Q[10]-bound ruthenium complex were assigned from a combination of DQFCOSY and ROESY experiments. In particular, the resonances from the bb<sup>7</sup> ligand were assigned through the observation of an ROE from the spin-coupled bpy H5 and H5' protons (coupled to the H6



and H6' protons) to the substituted-bpy ligand methyl (Me) and chain methylene protons respectively. The bpy H3 and H3' were similarly assigned from the observed ROEs to the bpy Me and α-CH<sup>2</sup> protons, respectively. The phenanthroline H2/9 resonances were assigned through the observation of an ROE to the bb<sup>7</sup> H6 protons.

One set of exchange-broadened resonances was observed for the protons from both α-Me4phen-7 and Q[10], indicating intermediate-exchange kinetics on the NMR timescale at 25◦C for R < 1. Of note, the resonances from the CH<sup>2</sup> groups in the alkyl chain of the bb<sup>7</sup> ligand shifted significantly upfield (0.3– 0.5 ppm). This indicates the alkyl chain is positioned deep within the Q[10] cavity, as many previous studies have established that resonances from guest protons located inside a Q[n] cavity shift significantly upfield, with protons positioned toward the middle of the Q[n] cavity exhibiting the largest (up to 1 ppm) upfield shifts (Mock and Shih, 1986; Jeon et al., 1996). This suggests that α-Me4phen-7 binds Q[10] with the bb<sup>7</sup> ligand deep within the cavity and the Me4phen ligand projecting out of the portal. Consistent with this proposal are the significant upfield shifts (≈ 0.5 ppm) for the bb<sup>7</sup> protons on the pyridyl ring containing the alkyl chain (H3', H5' and H6') coupled with the observed downfield shifts of the Me4phen H2/9 and H5/6 resonances.

Interestingly, at a Q[10] to α-Me4phen-7 ratio of 1 (as determined from the integration of the respective resonances), two sets of resonances for the bb<sup>7</sup> aromatic protons and the Q[10] CH<sup>2</sup> resonances were observed (see **Figures 5**, **6**), indicating slow-exchange binding kinetics. The inequivalence of the Q[10] methylene resonances that project toward the portals (5.83 and 5.64 ppm) and toward the cavity center (4.23 and 4.14 ppm) is due to the non-symmetric encapsulation of α-Me4phen-7 in Q[10]. As intermediate-exchange kinetics were observed at R < 1 but slow-exchange at R = 1, it is concluded that for all values of R < 1 there must be some 1:2 (Q[10] to α-Me4phen-7) binding with the rate of exchange between the 1:1 and 1:2 binding modes being in the fast-exchange regime. In the R = 1 spectrum, the resonances from the bpy CH<sup>3</sup> groups were also inequivalent, with one peak shifted downfield and the other upfield compared to equivalent bpy CH<sup>3</sup> resonances in the free ruthenium complex. This suggests that one bpy CH<sup>3</sup> is positioned inside the Q[10] cavity with the second bpy CH<sup>3</sup> located outside of the portal. Possibly due to the broadness of the resonances from α-Me4phen-7, very few intermolecular ROEs were observed in ROESY spectra of the Q[10]-encapsulated ruthenium complex.

However, ROEs was observed from the Me4phen H2/9 and 3/8 methyl resonances to the Q[10] methylene protons that project toward the portal.

### Molecular Modeling

In order to confirm the feasibility of the mode of binding suggested by the NMR data, the encapsulation of α-Me4phen-7 within the Q[10] cavity was examined by molecular simulations using the HyperChem modeling software. The methylene chain of α-Me4phen-7 was folded within the Q[10] cavity and the ruthenium metal center positioned at the portal in a manner consistent with the observed intermolecular ROEs observed in ROESY spectra. The system was then successively optimized using the Amber99 molecular mechanics forcefield. As shown in **Figure 7**, stable low energy conformations could be obtained with the methylene chain folded in the cavity and the ruthenium metal

Q[10] using *HyperChem* software. The color of the atoms of α-Me4phen-7 are: H, white; C, cyan; N, blue; and Ru, red.

center positioned outside the portal consistent with the NMR data.

### Luminescence Spectroscopy

In order to obtain an equilibrium binding constant, the association of α-Me4phen-7 with Q[10] was studied by luminescence spectroscopy. The luminescence intensity of α-Me4phen-7 decreased upon the addition of each aliquot of Q[10] up to R = 1.5 (see **Figure 8**). Binding curves were generated by plotting luminescence intensity (counts per second) against the molar ratio of added Q[10], from which a 1:1 binding constant of 9.9 ± 0.2 × 10<sup>6</sup> M−<sup>1</sup> was determined.

### Biological Properties of α-Me4phen-7 Encapsulated in Q[10]

Upon encapsulation in Q[10], α-Me4phen-7 exhibited no toxicity to any of the cell lines for both 24- or 48-h incubations (see

**Table 1**). Based upon the 48-h incubation toxicity to BHK and Hep-G2 cells, it can be concluded that Q[10]-encapsulation decreases the toxicity of α-Me4phen-7 by at least (2–3)-fold. Of note, and contrary to what was observed for Rubb12, Q[10] encapsulation increases the affinity of α-Me4phen-7 for HSA (see **Figure 9**). A Kapp of 2 ± 0.2 × 10<sup>6</sup> M−<sup>1</sup> was determined for the Q[10]-encapsulated ruthenium complex.

Reducing the cellular accumulation of α-Me4phen-7 is the simplest mechanism by which Q[10]-encapsulation could reduce toxicity. Consequently, confocal microscopy was used to obtain both a qualitative and quantitative measure of the effect of Q[10] encapsulation on the cellular accumulation of α-Me4phen-7 in BHK cells after a short incubation time. **Figure 10** shows confocal microscopy images of BHK cells after a 1 h incubation with free or Q[10]-encapsulated α-Me4phen-7. Mitotracker Green (which selectively localizes in the mitochondria) was used to visualize the cells, with the nucleus of the cell being the dark circular region outlined by the green luminescence. As it appeared that the cellular accumulation was not significantly different between the free or Q[10]-encapsulated α-Me4phen-7 samples, the luminescence intensity from α-Me4phen-7 per cell was quantified over all the images taken. The results are shown in **Figure 11**. As seen in the "box and whiskers" diagram, and confirmed by statistical analysis (Wilcoxon-test and a t-test; p = 0.13 and 0.11, respectively), there was no difference in the level of the cellular accumulation of α-Me4phen-7 between the experiments where the ruthenium complex was added to the BHK cells in either the free or Q[10]-encapsulated forms.

# DISCUSSION

We have previously shown that some mononuclear ruthenium(II) complexes containing bb<sup>n</sup> as a tetradentate ligand (particularly α-phen-12 and α-Me4phen-7) exhibit good antimicrobial activity (MICs = 1–10µM) against both Gram-positive and Gram-negative bacteria (Gorle et al., 2015; Sun et al., 2018). In this study we have examined their associated toxicity to eukaryotic cells. Importantly, although α-Me4phen-7 was as active against Gram-negative bacteria as α-phen-12 (Sun et al., 2018), the results from the present study showed that

FIGURE 10 | Top, left to right: α-Me4phen-7 accumulation in BHK cells at 25µM after a 1-h incubation, stained by Mitotracker Green and α-Me4phen-7 (red). Bottom, left to right: Q[10]-encapsulated α-Me4phen-7 accumulation in BHK cells at 25µM after a 1-h incubation, stained by Mitotracker Green and α-Me4phen-7 (red). Scale bar = 10µm.

α-Me4phen-7 was (2–5)-fold less toxic to eukaryotic cells than α-phen-12. It was also previously shown that α-phen-12 (log P = −0.9) accumulated in bacterial cells more rapidly and to a greater degree than did α-Me4phen-7 (log P = −1.33), consistent with their respective lipophilicities (Sun et al., 2018). Consequently, α-Me4phen-7 appears to be more toxic to Gramnegative bacteria than α-phen-12 on a molar basis. This could be due to the greater DNA binding ability of α-Me4phen-7, compared to α-phen-12 (Sun et al., 2018). Of relevance in the present study, the preliminary confocal microscopy experiments indicated that α-Me4phen-7 did not localize in the nucleus of BHK cells. As a consequence, α-Me4phen-7 might not be inherently more toxic to eukaryotic cells than α-phen-12. Hence, based upon lipophilicity, it would be expected that α-phen-12 would also accumulate to greater degree in eukaryotic cells than α-Me4phen-7, and thereby be more toxic. It was also noted that α-Me4phen-7 was significantly less toxic than the lead oligonuclear complex, Rubb12. In absolute terms, α-Me4phen-7 showed very little toxicity to the eukaryotic cells and was found to be at least 20-fold more active against Gram-negative bacteria (Sun et al., 2018) than eukaryotic cells over a 24 h time period.

In terms of HSA binding, α-Me4phen-7 bound the serum protein approximately 10-fold more tightly than α-phen-12 and α-Me2phen-7, and with similar affinity as Rubb12. Previous studies have demonstrated that there are a number of factors governing the binding affinity of drugs with HSA (Kratochwil et al., 2002; Bohnert and Gan, 2013; Liu et al., 2014). Lipophilicity is generally considered to be important, particularly within a series of structurally-related compounds. In this study it was found the degree of methylation of the phenanthroline ligand, and hence lipophilicity, did correlate with stronger binding. However, the α-NO2phen-7 complex (log P = −1.58) bound HSA with higher affinity than α-Me4phen-7 (log P = −1.33). This observation suggests that the dispersion of the 2+ charge of the ruthenium complex is also important, which is consistent with the approximate charge of HSA at physiological pH being −17 (Fogh-Andersen et al., 1993).

It is often assumed that low serum protein binding is a "good" drug property, as it increases the concentration of the free drug; however, in a recent survey of newly approved drugs, Liu and co-workers (Liu et al., 2014) found that 45 and 24% of the drugs exhibited high (>95% bound) or very high (>99%) serum protein binding, respectively. Consequently, Liu et al and others proposed that serum protein binding is neither an intrinsically "good" or "bad" property. However, a recent pharmacokinetic study of Rubb<sup>12</sup> in mice demonstrated that the amount of the ruthenium complex in the serum decreased very rapidly after administration by intravenous injection (Li et al., 2017). This result suggests that potential new ruthenium complexes would need to bind serum proteins at least as strongly, if not with greater affinity, than Rubb12. Based upon this criterion, it is noted that α-Me4phen-7 binds HSA with at least similar affinity to Rubb12.

Given the antimicrobial potential of α-Me4phen-7, it was of interest to examine the ability of the ruthenium complex to bind Q[10], and determine its subsequent effect on cytotoxicity and HSA binding. The results of this study showed that α-Me4phen-7 does form a high-affinity water-soluble association complex with Q[10], with the bb<sup>7</sup> alkyl chain positioned deep within the cavity and the ruthenium metal center positioned at one of the portals. This binding mode, with the non-polar segment of a metal complex in the hydrophobic cavity and the cationic metal center positioned at the carbonyl-rimmed hydrophilic portals, is consistent with previous studies (Pisani et al., 2010; Alrawashdeh et al., 2016). The magnitude of the α-Me4phen-7-Q[10] binding constant (≈10<sup>7</sup> M−<sup>1</sup> ) indicates that a large proportion of the ruthenium complex would be Q[10]-bound at the micro-molar concentrations that would be found in the blood after administration.

The effect of Q[n] encapsulation on the cytotoxicity of a variety of compounds toward eukaryotic cell lines has been previously examined. For example, Kim et al. reported a (5–10) fold decrease in activity for oxaliplatin in Q[7] (Jeon et al., 2005), while Wheate et al showed that a dinuclear platinum complex maintained its activity toward L1210 cells, but was 2-fold less active against the cisplatin resistant cell line L1210/DDP (Wheate et al., 2004). Furthermore, there have been several recent reports where encapsulation in Q[n] was proposed to have increased the uptake of a cytotoxic agent (Konda et al., 2017; Shinde et al., 2018). In the present study, Q[10]-encapsulation decreased the toxicity of α-Me4phen-7 toward BHK and Hep-G2 cells, while no conclusions could be drawn for the Caco-2 cells, given the lack of activity of the free ruthenium complex for both 24 and 48-h incubations. The most obvious explanation for the decreased toxicity of α-Me4phen-7 encapsulated in Q[10] is that the concentration of the free, unbound, ruthenium complex is significantly lower in the assay medium and the Q[10]-bound ruthenium complex can not cross the cell membrane. This would consequently result in a slower and lower accumulation of the ruthenium complex inside the cells. However, the results from the confocal microscopy experiments indicated that the level of accumulation of α-Me4phen-7 inside the cells is not significantly affected by encapsulation over a 1-h incubation.

In terms of HSA binding, Q[10]-encapsulation increased the affinity of α-Me4phen-7 for the serum protein. Interestingly, in an earlier study it was determined that Q[10]-binding of Rubb<sup>12</sup> decreased the affinity of the ruthenium complex for HSA binding (Li et al., 2013b). As noted above, given the short residence time found for Rubb<sup>12</sup> after intravenous injection in mice, the (3–4)-fold higher HSA binding affinity is likely to be beneficial.

## CONCLUSIONS

In conclusion, the results of this study have demonstrated that the most active antimicrobial mononuclear ruthenium complex against Gram-negative bacteria, α-Me4phen-7, is also the least toxic to eukaryotic cells. Furthermore, although it exhibits slightly lower (≈ 2-fold) antimicrobial activities to Gram-negative bacteria than the lead oligonuclear complex

## REFERENCES


Rubb<sup>12</sup> (Sun et al., 2018), α-Me4phen-7 is also significantly less toxic to eukaryotic cells compared to the dinuclear complex. Fluorescence assays demonstrated that α-Me4phen-7 bound HSA with similar affinity to that previously reported for Rubb12. Taken together, the results of this study suggest α-Me4phen-7 is a suitable candidate for further studies of its antimicrobial properties. In addition, it has been shown that α-Me4phen-7 can form a high-affinity, water-soluble, inclusion complex with Q[10]. Q[10]-encapsulation decreases the toxicity of α-Me4phen-7 toward eukaryotic cells and increases its HSA binding affinity. As a consequence, administration of α-Me4phen-7 encapsulated in Q[10] may increase the potential of the ruthenium complex as an antimicrobial agent. However, the effect of Q[10] encapsulation on the antimicrobial properties of α-Me4phen-7 needs to be examined.

# AUTHOR CONTRIBUTIONS

BS synthesized the ruthenium complexes and carried out the experiments. AD provided the Q[10], while IM, KH, FK, and JC were involved in planning the research, analysing the data, and writing the paper.

# FUNDING

On invitation to contribute, it was indicated there would be no charges for this paper in the Research Topic Supramolecular Metal-Based Entities for Biomedical and Biological Applications, in the Frontiers in Chemistry—Supramolecular Chemistry Section (hosted by Angela Casini and James D. Crowley). This is submitted on the basis that there will be NO CHARGE.

# ACKNOWLEDGMENTS

BS thanks UNSW Canberra for a Ph.D. scholarship and a Travel Grant to undertake the biology experiments at the University of Adelaide. The authors thank Prof H.H. Harris (University of Adelaide) for the gift of some cisplatin, and Jane Sibbons (University of Adelaide) for help in quantifying the confocal microscope images.


membrane permeability in gram-negative bacteria–in particular Pseudomonas aeruginosa. Chem. Eur. J. 21, 10472–10481. doi: 10.1002/chem.2015 00385


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Sun, Musgrave, Day, Heimann, Keene and Collins. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Supramolecular Fluorescence Probe Based on Twisted Cucurbit[14]uril for Sensing Fungicide Flusilazole

Ying Fan1,2, Rui-Han Gao<sup>2</sup> , Ying Huang<sup>2</sup> , Bing Bian<sup>3</sup> , Zhu Tao<sup>2</sup> \* and Xin Xiao<sup>2</sup> \*

*<sup>1</sup> State Key Laboratory Breeding Base of Green Pesticide and Agricultural Bioengineering, Key Laboratory of Green Pesticide and Agricultural Bioengineering, Ministry of Education, Guizhou University, Guiyang, China, <sup>2</sup> Key Laboratory of Macrocyclic and Supramolecular Chemistry of Guizhou Province, Guizhou University, Guiyang, China, <sup>3</sup> College of Chemical and Environmental Engineering, Shandong University of Science and Technology, Qingdao, China*

The host-guest complex of the common dye, thioflavin T (ThT), and twisted cucurbit[14]uril (*t*Q[14]) was selected as a fluorescent probe to determine non-fluorescent triazole fungicides, including flusilazole, azaconazole, triadimefon, tebuconazole, tricyclazole, flutriafol, penconazole, and triadimenol isomer A, in an aqueous solution. The experimental results reveal that the ThT@*t*Q[14] probe selectively responded to flusilazole with significant fluorescence quenching and a detection limit of 1.27 × 10−<sup>8</sup> mol/L. In addition, the response mechanism involves not only a cooperation interaction—ThT occupies a side-cavity of the *t*Q[14] host and the triazole fungicide occupies another side-cavity of the *t*Q[14] host—but also a competition interaction in which both ThT and the triazole fungicide occupy the side-cavities of the *t*Q[14] host.

### Edited by:

*Angela Casini, Cardiff University, United Kingdom*

#### Reviewed by:

*Dong-Sheng Guo, Nankai University, China Simin Liu, Wuhan University of Science and Technology, China*

#### \*Correspondence:

*Zhu Tao gzutao@263.net Xin Xiao gyhxxiaoxin@163.com*

#### Specialty section:

*This article was submitted to Supramolecular Chemistry, a section of the journal Frontiers in Chemistry*

Received: *30 August 2018* Accepted: *01 March 2019* Published: *21 March 2019*

#### Citation:

*Fan Y, Gao R-H, Huang Y, Bian B, Tao Z and Xiao X (2019) Supramolecular Fluorescence Probe Based on Twisted Cucurbit[14]uril for Sensing Fungicide Flusilazole. Front. Chem. 7:154. doi: 10.3389/fchem.2019.00154* Keywords: twisted cucurbit[14]uril, thioflavin T, fluorescent probe, triazole fungicides, flusilazole

# INTRODUCTION

Triazole systemic fungicides, particularly flusilazole (1-{[bis(4-fluorophenyl) (methyl)silyl] methyl}-1H-1,2,4-triazole), are widely used in fruit, vegetable, and grain crops during cultivation and storage with high efficiency and good sterilization ability (**Figure 1**) (Zhang Y. H. et al., 2016). However, they still have somewhat toxic or other undesirable side effects on non-target organisms. Extensive or inappropriate use can cause soil and water pollution, and thus threaten human health (Ma et al., 2016). Therefore, it is necessary to develop sensitive and selective methods for the analysis of triazole fungicides, which are usually present in trace amounts.

The most common analytical method used for the trace determination of triazoles, particularly flusilazole residues in water, is chromatography e.g., liquid chromatography-tandem mass spectrometry (LC-MS/MS) (García-Valcárcel and Tadeo, 2011; Fu et al., 2017), high-performance liquid chromatography (HPLC) with ultraviolet light, diode-array detection (DAD), photodiodearray (PDA) detection (Bordagaray et al., 2013, 2014; Qi et al., 2014; Ma et al., 2016; Zhang Y. H. et al., 2016), gas chromatography (GC) with nitrogen-phosphorous detection (NPD) or electron capture detection (ECD) (Lozowicka et al., 2015; Im et al., 2016), and GC-mass spectrometry (GC-MS) (Tseng et al., 2014; Chu et al., 2015), and with tandem MS (GC-MS/MS) (Xu et al., 2013). Nevertheless, these chromatographic techniques necessitate experienced workers, costly devices, and lengthy specimen preparation. In comparison with the above-mentioned methods, spectrofluorimetry is the most favorable analytical strategy employed to analyze different biological specimens due to its innate simplicity, high sensitivity, and accessibility in the majority of quality-control and clinical laboratories (Yao et al., 2013). However, determination of the existence of many common triazoles cannot be performed directly using typical fluorimetric techniques since the aqueous solutions with triazoles do not have native fluorescence.

It is noteworthy that the complexation of a fluorescent dye with a macrocycle host can induce significant change in its fluorescence (Zhou et al., 2008; Zhang et al., 2012; Kogawa et al., 2014). Such assays depend on an indicator-displacement technique in which the analyte competitively supplants a fluorescent guest that results in an alteration in its fluorescence intensity (You et al., 2015; Sayed and Pal, 2016). Various macrocyclic compounds, including cyclodextrins, calix[n]arenes, and pillar[n]arenas, have long been employed as the hosts in several macrocyclic-dye supramolecular systems (Choudhury et al., 2010; Lau and Heyne, 2010; Liu et al., 2015b; Sun et al., 2015). In particular, cucurbit[n]urils (Q[n]s) are made up of n glycoluril units connected via 2n methylene bridges (Day et al., 2000; Kim et al., 2000; Cong et al., 2016). These hosts have highly polar carbonyl-fringed portals with hydrophobic cavities that can create remarkably stable complexes with different guest molecules (Kim, 2002; Dsouza et al., 2011; Gao et al., 2017; Liu J. et al., 2017; Murray et al., 2017; Yang et al., 2018). The origination of inclusion complexes frequently improves or interrupts the photo-physical and photo-chemical characteristics of the guest molecules. For example, using different Q[n]s to encapsulate some dyes can change the fluorescent characteristics of the guest molecules (Praetorius et al., 2008; Baumes et al., 2009, 2011; Choudhury et al., 2009). The dye@Q[n] complexes can be composed of supramolecular fluorescent probes with high sensitivity and selectivity to recognize and detect analytes, such as metal ions, amines, pesticides, drugs, and DNA (Wheate, 2008; Mohanty et al., 2009; Nau et al., 2009; Zhou et al., 2009; Day and Collins, 2012; Xing et al., 2013; Elbashir and Aboul-Enein, 2015; Liu W. Q. et al., 2017; Xi et al., 2017; Tang et al., 2018; Wang et al., 2018).

Thioflavin T, 3,6-dimethyl-2-(4-dimethylaminophenyl) benzthiazolium cation (ThT) is a benzthiazolium dye (**Figure 1**). An aqueous solution of ThT reveals a weak native fluorescence. Nevertheless, as revealed by previous evaluations, the fluorescence of ThT in an aqueous solution can be vastly improved in the presence of Q[7], Q[8], and twisted cucurbit[14]uril (tQ[14]). It was found that ThT can form ThT@Q[n] inclusion complexes with different stoichiometric ratios, which can be applied to identify both alkali-metal and alkaline-earth-metal ions (Choudhury et al., 2009, 2010; Mohanty et al., 2009; Wang et al., 2018). In the present study, a procedure was proposed to determine the residues of triazole fungicides, such as azaconazole, triadimefon, tebuconazole, tricyclazole, flutriafol, penconazole, and triadimenol isomer A in aqueous solution (**Figure 1**).

# RESULTS AND DISCUSSION

# Fluorescence Quenching of ThT@tQ[14] With Flusilazole

Previous studies have proven that the fluorescence of ThT in an aqueous solution over a wide range of interaction ratios can be enhanced by its interaction with tQ[14] via different interaction models due to the multiple cavity features of the tQ[14] molecule and two active moieties (the benzothiazole and dimethylaminobenzene moieties) in ThT (Wang et al., 2018). In the current work, we focused on the inclusion complexes of ThT@tQ[14] with 1:1 and 1:5 interaction ratios in the presence of the eight selected triazoles, the corresponding fluorescence spectra of which can be seen in **Figure 2** and **Figures S1–S3**. A significant fluorescence quenching of the ThT@tQ[14] inclusion complex probe (1:1) was only observed upon increasing the amount of flusilazole; ∼60% of the fluorescence intensity was lost when Cflusilazole/CThT@tQ[14] was ∼4 (**Figure 2A)**. **Figure 2B** shows the plot of the fluorescence intensity at λem = 488 nm upon increasing the flusilazole concentration; Kobserved was calculated by the nonlinear-leastsquares method to be (8.2 ± 0.4) ×10<sup>5</sup> L/mol obtained (**Figure S4**), which is likely due to the influence of the interaction of tQ[14] with flusilazole on the interaction of tQ[14] with ThT. Moreover, the ThT@tQ[14] fluorescent probe presented high selectivity for flusilazole because the probe showed little or almost no response to the other seven triazole fungicides studied (**Figure 3** and **Figure S1**). The ThT@tQ[14] inclusion complex probe at a molar ratio of 1:5 was also selected to investigate its response to triazole pesticides with the addition of flusilazole also leading to a fluorescence quenching of the ThT@tQ[14] inclusion complex probe; only ∼22% of the fluorescence intensity was lost when Cflusilazole/CThT@tQ[14] was ∼4 (**Figure S2**). However, the ThT@tQ[14] fluorescent probe also showed little or almost no response to the other triazole fungicides studied (**Figure S3**).

# Analytical Characteristics of ThT@tQ[14] Probe

The standard calibration curve (**Figure S5**) of the complexes of flusilazole with ThT@tQ[14] (1:1) was obtained using the plot of fluorescence intensity F compared to flusilazole concentration. The linear range was (0.0–1.0) × 10−<sup>6</sup> mol/L for flusilazole with a correlation coefficient of 0.9888. For flusilazole, the limit of detection (LOD) was 1.27 × 10−<sup>8</sup> mol/L and the linear regression equation F = −216.24C+576.26. Although the ThT@tQ[14] (1:5) inclusion complex was less sensitive to flusilazole, it still exhibited a very low LOD (2.80 × 10−<sup>8</sup> mol/L) and the linear regression equation was F = −98.25C+863.26 (**Figure S6**).

# Effects of Interfering Substances

In the present work, the impacts of different common interfering substances on the determination of flusilazole using the ThT@tQ[14] probe were also investigated. The tolerance limit was established as the concentration of an added interfering substance that causes a relative error of < ±5% in the flusilazole determination. The samples consisted of a fixed amount of flusilazole (1.0µM) and ThT@tQ[14] (1.0µM) with an increasing amount of different interfering substances; the corresponding results are shown in **Table 1** and **Figure S7**. It is clear that the determination was free from interference in the presence of the common metal ions and anions in aqueous solutions, except Ca2+. Structural analysis revealed that tQ[14] had a similar portal size to those of Q[6] and

FIGURE 1 | Structures of *t*Q[14], ThT, and eight fungicide triazoles, namely flusilazole, azaconazole, triadimefon, tebuconazole, tricyclazole, flutriafol, penconazole, and triadimenol isomer A.

Q[7] (Cheng et al., 2013; Liu et al., 2015a; Li et al., 2016; Zhang J. et al., 2016; Zhang et al., 2017), and a higher portal carbonyl intensity (Cheng et al., 2013; Liu et al., 2015a; Zhang J. et al., 2016; Zhang et al., 2017), which could exhibit a higher affinity for metal cations, particularly the Ca2<sup>+</sup> cation (Cheng et al., 2013; Qiu et al., 2017; Wang et al., 2018). The titration fluorescence spectra recorded for the tQ[14]-ThT-flusilazole (1:1:1) system upon increasing the amount of Ca2<sup>+</sup> showed that the fluorescence of the ternary interaction system could be further decreased due to the influence of the Ca2<sup>+</sup> cation (**Figure S8**). The titration <sup>1</sup>H nuclear magnetic resonance (NMR) spectra for the tQ[14]-ThTflusilazole (1:1:1) system in the presence of Ca2<sup>+</sup> also showed that the guest molecules, ThT and flusilazole, were gradually pushed out from the cavity area upon increasing the amount of Ca2<sup>+</sup> (**Figure S9**).

# Preliminary Exploration of the Response Mechanism of ThT@tQ[14] Fluorescent Probe Toward Flusilazole Titration <sup>1</sup>H NMR Spectra

We have found that the ThT@tQ[14] fluorescent probe was selectively sensitive toward flusilazole and could be used to detect flusilazole via a significant fluorescence quenching process. Why is the ThT@tQ[14] fluorescent probe selectively sensitive to flusilazole? How does the flusilazole molecule influence the interaction between the ThT molecule with the tQ[14] host molecule? What could the interaction mode be? To understand the selectivity and response mechanism of the ThT@tQ[14] fluorescent probe toward flusilazole, titration <sup>1</sup>H NMR spectra were recorded upon the gradual addition of one of the selected triazole fungicides into the solution of the ThT@tQ[14] (1:1) fluorescent probe. The detailed interaction of ThT and tQ[14]

TABLE 1 | Impact of interfering substances (tolerance error ± 5%).


has been discussed in a previous work (**Figures 4A,B**) (Wang et al., 2018). **Figure 4F** shows the <sup>1</sup>H NMR spectrum for the flusilazole in the presence of tQ[14]. When compared with the <sup>1</sup>H NMR spectrum for the unbound flusilazole molecule (**Figure 4G**), the proton resonances corresponding to H5 and H6 on the triazole moiety of the bound flusilazole molecule underwent slightly down-field shifts (0.01 and 0.01 ppm, respectively), whereas the other proton resonances experienced slightly up-field shifts, suggesting that a weak interaction existed between tQ[14] and the flusilazole guest molecule or π · ·· π stacking of the aryl groups of the flusilazole guest molecules. **Figures 4C–E** show the titration <sup>1</sup>H NMR spectra for the tQ[14]- ThT complex upon increasing the amount of flusilazole. The addition of flusilazole appeared to pull the bound ThT out of the interaction area of the tQ[14] host because the proton resonances were closer to those of the unbound ThT molecule. However, the proton resonances of flusilazole, in particular, the H5 and H6 experienced down-field shift, suggesting that ThT was still interacting with the host and flusilazole was located in the de-shielding region of tQ[14] (probably at portal area). Similar phenomena could be observed for the other tQ[14]-ThT-triazole fungicide systems, in which the addition of the triazole fungicide could lead to a change in the interaction between tQ[14] and ThT. Moreover, both ThT and the triazole fungicide exhibited an interaction with the tQ[14] host (**Figures S10–S16**). At first, it seemed that it was difficult to obtain the required information using the titration <sup>1</sup>H NMR method. A closer inspection revealed that the addition of different triazole fungicides could lead to different chemical shifts for the proton resonances corresponding to the ThT molecule, and the effects of the different triazole fungicides on the bound ThT were compared with the chemical shift corresponding to proton Hg. The data in **Table 2** reveal that flusilazole resulted in the larger change in the chemical-shift values (1δ = 0.05) when the tQ[14]-ThT-triazole fungicide ratio was 1:1:2.

2D diffusion-ordered NMR spectroscopy (DOSY) experiments were performed to afford further evidence for the formation of the tQ[14]-ThT-flusilazole (1:1:1) ternary interaction species. **Figure S17** depicts the DOSY spectra of tQ[14], ThT, flusilazole, and tQ[14]-ThT-flusilazole (1:1:1) ternary interaction species in D2O at 298 K and the corresponding diffusion coefficients (D) are 3.49 × 10−<sup>10</sup> , 5.46×10−10, 3.88 × 10−10, and 2.92 × 10−10, respectively. According to the values of four species, the tQ[14]-ThTflusilazole (1:1:1) ternary interaction species is the smallest, suggesting that the ternary species could be the largest species, moreover, all the proton signals of the host and the guest display the same diffusion coefficient (D = 2.92 × 10−<sup>10</sup> · m<sup>2</sup> · s −1 ), indicating that they are part of the same species.

## Isothermal Titration Calorimetry

Isothermal titration calorimetry (ITC) experiments were conducted to determine the association constants and thermodynamic parameters of the host-guest interaction between tQ[14] and ThT with the triazole fungicides in an aqueous solution to further explore the fluorescence-quenching mechanism of ThT@tQ[14] with flusilazole (**Figures S18–S26**). From the data acquired (**Table 3**), all the association constants obtained for the triazole fungicides@tQ[14] complexes

Frontiers in Chemistry | www.frontiersin.org

Guest δ [Q14-ThT (1:1)] δ [Q14-ThT-G (1:1:2)] 1δ

Flusilazole 2.469 2.419 0.050

TABLE 2 | Chemical shifts corresponding to proton Hg of ThT.


The fluorescence enhancement mechanism of ThT could be due to the restriction of the freely rotating dimethylamine group on the benzene moiety, making the lone pair electrons on nitrogen atoms conjugate to the ThT aromatic system. The shell-like cavity structure of tQ[14] can provide such controlled environment, the inclusion of dimethylamino phenyl moiety of ThT inhibited dimethylamino free rotation, moreover, the shell-like cavity structure of tQ[14] could prevent ThT from threading through the side cavity of tQ[14] (Liu et al., 2015a; TABLE 3 | Thermodynamic parameters of *t*Q[14] and ThT with eight triazole pesticides in aqueous solution at 298.15 K.

*t*Q[14]-ThT (1.28 ± 0.03) × 10<sup>5</sup> −48.4 −19.2 *t*Q[14]-flusilazole (8.39 ± 0.05) × 10<sup>5</sup> −38.4 −4.6 *t*Q[14]-azaconazole (5.45 ± 0.08) × 10<sup>5</sup> −36.8 −4.0 *t*Q[14]-triadimefon (1.46 ± 0.04) × 10<sup>5</sup> −64.3 −34.9 *t*Q[14]-tebuconazole (3.00 ± 0.09) × 10<sup>5</sup> −42.7 −11.4 *t*Q[14]-tricyclazole (5.01 ± 0.12) × 10<sup>5</sup> −36.8 −4.2 *t*Q[14]-flutriafol (1.97 ± 0.05) × 10<sup>5</sup> −64.6 −34.3 *t*Q[14]-penconazole (5.23 ± 0.08) × 10<sup>5</sup> −38.8 −6.2 *t*Q[14]-triadimenol isomer A (4.31 ± 0.10) × 10<sup>5</sup> −39.0 −6.9



Complex Ka (m <sup>−</sup><sup>1</sup>


) 1H(kJ/mol) T1S(kJ/mol)

toward the free ThT, in particular, the interaction of the azole moiety and quaternary ammonium moiety could weaken the electron-withdrawing capacity of quaternary ammonium moiety, resulting in partial quenching of the fluorescence of the tQ[14]- ThT-flusilazole interaction system (**Figure 5**). According to the suggested interaction mode, whether the fluorescence quenching of ThT@tQ[14] is caused mainly depends on the interaction of the azole moiety of pesticides and quaternary ammonium moiety of ThT and the ability of pesticides to pull ThT out of the side cavity of tQ[14]. While the NMR and ITC measurement results showed that the flusilazole has the biggest impact, although the differences from other triazole fungicides are subtle.

# CONCLUSIONS

In order to further expand the application of Q[n]-based host-guest chemistry, especially the host-guest complexes with fluorescent properties, in detection and recognition, a known host-guest complex of ThT@tQ[14] (1:1) was used as a fluorescent probe to determine non-fluorescence triazole fungicides, including flusilazole, azaconazole, triadimefon, tebuconazole, tricyclazole, flutriafol, penconazole, and triadimenol isomer A. This new and simple fluorometry method proved to be highly selective and sensitive to one of triazole fungicides—flusilazole, and the determination was free from interference by the common metal ions and anions in aqueous solutions, except Ca2+. The investigation of the response mechanism revealed that a side-cavity of the tQ[14] host includes the dimethylamino phenyl moiety of ThT, resulting in the ThT fluorescence enhancement; the addition of flusilazole results in the interaction of the azole moiety of flusilazole and quaternary ammonium moiety of ThT, which could weaken the electron-withdrawing capacity of quaternary ammonium moiety, resulting in partial quenching of the fluorescence of the tQ[14]-ThT-flusilazole interaction system. This unusual phenomenon results from the novel structural feature of tQ[14], namely that tQ[14] possesses a central-cavity and two of the same side-cavities.

# EXPERIMENT

# Materials

tQ[14] was set up and purified in our laboratory according to a procedure detailed in the literature (Cheng et al., 2013). Analytical grade flusilazole (99.5%), azaconazole (99.5%), triadimefon (99.0%), tebuconazole (99.0%), tricyclazole (99.0%), flutriafol (98.6%), penconazole (99.5%), and triadimenol isomer A (95.4%) were purchased from Dr. Ehrenstorfer GmbH (Augsburg, Germany) and used as-received without any further purification. Double-distilled water was used for each of the experiments.

# <sup>1</sup>H NMR

Each of the <sup>1</sup>H NMR spectra, including those for titration experiments, were documented at 25◦C on a JEOL JNM-ECZ400s spectrometer using SiMe<sup>4</sup> as an internal reference. D2O was utilized as a field-frequency lock and the chemical shifts documented in parts per million (ppm).

# ITC

Microcalorimetric experiments were conducted with a Nano ITC (TA, USA) isothermal titration calorimeter. Then, 25 consecutive 10-µL aliquots of a 1 mM tQ[14] solution were introduced into the microcalorimetric reaction cell, which contained 1.3 mL of a 0.1-mM guest molecule solution at 25◦C. The heat of reaction was

corrected for the heat of dilution of the guest molecule solution, which was determined in a separate experiment. Each of the solutions were de-gassed before the titration via ultrasonication. Computer simulations (curve fitting) were conducted with Nano ITC analysis software.

# Fluorescence Titration

The fluorescence spectra of the host-guest complexes were documented at 25◦C using a Varian Cary Eclipse spectrofluorometer (Varian, Inc., Palo Alto, CA, USA). Stock solutions of tQ[14] (1 × 10−<sup>3</sup> mol L−<sup>1</sup> ), ThT (1 × 10−<sup>3</sup> mol L −1 ), and flusilazole (1 × 10−<sup>4</sup> mol L−<sup>1</sup> ) were set up in water. Working solutions were set up by diluting the stock solutions to the necessary concentrations.

A tQ[14]-ThT (1:1) complex solution was set up at a fixed concentration of 1 × 10−<sup>6</sup> mol·L −1 in H2O, which was subsequently combined with flusilazole at guest/host ratios of 0, 0.25:1, 0.5:1, 0.75:1, 1:1, . . . , and 4:1. Fluorescence spectrophotometric titrations were established as detailed prior (λex = 448 nm and λem = 488 nm). For each experiment, three replicate measurements were recorded.

# LOD Measurement

The calculation technique used for the LOD was based on the standard derivation of 10 measurements without the guest molecule (σ) and the slope of the linear calibration curve (K) based on the formula LOD = 3σ/K.

# REFERENCES


The standard deviation of 10 measurements without the guest molecule could be determined based on the following relationship: σ = s 1 n−1 Pn i=1 (x<sup>i</sup> − x) 2 ,where n is the number of measurements (n = 11).

# AUTHOR CONTRIBUTIONS

YF, R-HG, YH, and BB undertook the acquisition and analysis of data for the work. ZT and XX drafted the work and revised it critically for important intellectual content.

# ACKNOWLEDGMENTS

We acknowledge the support of the National Natural Science Foundation of China (Grant No. 21562015), the Creative Research Groups of Guizhou Provincial Education Department (2017-028), and the Innovation Program for High-level Talents of Guizhou Province (No. 2016-5657).

# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fchem. 2019.00154/full#supplementary-material

the determination of fungicides in wine. Food Chem. 185, 377–382. doi: 10.1016/j.foodchem.2015.04.015


crystal structures of cucurbit[n]uril (n = 5, 7, and 8). J. Am. Chem. Soc. 122, 540–541. doi: 10.1021/ja993376p


fungicides and its application in living cell imaging. Spectrochim. Acta Part A Mol. Biomol. Spectrosc. 191, 372–376. doi: 10.1016/j.saa.2017.10.042


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Fan, Gao, Huang, Bian, Tao and Xiao. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Directed Self-Assembly of Trimeric DNA-Bindingchiral Miniprotein Helicates

Jacobo Gómez-González <sup>1</sup> , Diego G. Peña<sup>2</sup> , Ghofrane Barka<sup>1</sup> , Giuseppe Sciortino3,4 , Jean-Didier Maréchal <sup>3</sup> \*, Miguel Vázquez López <sup>1</sup> \* and M. Eugenio Vázquez <sup>2</sup> \*

<sup>1</sup> Centro Singular de Investigación en Química Biolóxica e Materiais Moleculares (CiQUS), Departamento de Química Inorgánica, Universidade de Santiago de Compostela, Santiago de Compostela, Spain, <sup>2</sup> Centro Singular de Investigación en Química Biolóxica e Materiais Moleculares (CiQUS), Departamento de Química Orgánica, Universidade de Santiago de Compostela, Santiago de Compostela, Spain, <sup>3</sup> Departament de Química, Universitat Autònoma de Barcelona, Cerdanyola, Spain, <sup>4</sup> Dipartimento di Chimica e Farmacia, Università di Sassari, Sassari, Italy

We propose that peptides are highly versatile platforms for the precise design of supramolecular metal architectures, and particularly, for the controlled assembly of helicates. In this context, we show that the bacteriophage T4 Fibritin foldon (T4Ff) can been engineered on its N-terminus with metal-chelating 2,2′ -bipyridine units that stereoselectively assemble in the presence of Fe(II) into parallel, three-stranded peptide helicates with preferred helical orientation. Modeling studies support the proposed self-assembly and the stability of the final helicate. Furthermore, we show that these designed mini-metalloproteins selectively recognize three-way DNA junctions over double-stranded DNA.

#### Edited by:

Angela Casini, Cardiff University, United Kingdom

#### Reviewed by:

Olga Iranzo, Center National de la Recherche Scientifique Marseille, France Guzman Gil-Ramirez, University of Lincoln, United Kingdom

#### \*Correspondence:

Jean-Didier Maréchal jeandidier.marechal@uab.cat Miguel Vázquez López miguel.vazquez.lopez@usc.es M. Eugenio Vázquez eugenio.vazquez@usc.es

#### Specialty section:

This article was submitted to Supramolecular Chemistry, a section of the journal Frontiers in Chemistry

Received: 20 August 2018 Accepted: 09 October 2018 Published: 30 October 2018

#### Citation:

Gómez-González J, Peña DG, Barka G, Sciortino G, Maréchal J-D, Vázquez López M and Vázquez ME (2018) Directed Self-Assembly of Trimeric DNA-Bindingchiral Miniprotein Helicates. Front. Chem. 6:520. doi: 10.3389/fchem.2018.00520 Keywords: metallopeptide, self-assembly water, DNA recognition, enantioselectivity, peptide motifs, coordination chemistry

# INTRODUCTION

Peptides are excellent supramolecular building blocks that encode precise structural and functional information within their amino acid sequence. Accordingly, researchers have explored diverse peptide motifs, such as coiled-coils, β-hairpins, or peptide amphiphiles, as the basis of biofunctional devices and materials (Matsuura et al., 2005, 2010; Gazit, 2007; Ulijn and Smith, 2008; Apostolovic et al., 2010; Robson Marsden and Kros, 2010; Boyle and Woolfson, 2011; Lai et al., 2012; Pazos et al., 2016). Curiously, despite the enormous potential for controlling stereochemistry, nuclearity and stoichiometry, the controlled supramolecular assembly of inorganic complexes with peptide motifs has been somewhat overlooked, and only a handful of systems based on modified coiled-coil motifs have been reported (Lieberman and Sasaki, 1991; Ghadiri et al., 1992; Li et al., 2000; Peacock et al., 2012; Ball, 2013; Berwick et al., 2014; Luo et al., 2016). On the other hand, helicates are discrete metal complexes in which one or more organic ligands are coiled around—and coordinating—two or more metal ions (Piguet et al., 1997; Albrecht, 2001, 2005) as a result of ligand coiling, helicates are inherently chiral species that can appear as two enantiomers according to the orientation in which the ligands twist around the helical axis defined by the metal centers. Besides their intrinsic interest in basic supramolecular chemistry, helicates have shown promising DNA-binding properties that have been associated with antimicrobial and antitumoral effects (Howson et al., 2012; Kaner et al., 2015). However, more than 20 years after the pioneering studies by Prof. Jean-Marie Lehn (Lehn et al., 1987; Ulijn and Smith, 2008), helicates are still not viable alternatives to traditional

**83**

DNA-binding agents. The slow development in the applied chemistry of metal helicates ultimately derives from the shortcomings associated with the classic synthetic approaches with organic ligands that complicate the structural control of the final helicates (i.e., oligomerization state, relative orientation of asymmetric ligands, supramolecular helicity) and hampers their efficient structural and functional optimization. Indeed, despite some noteworthy examples (Haino et al., 2009; Cardo et al., 2011; Howson et al., 2012; Chen et al., 2017; Mitchell et al., 2017; Guan et al., 2018), no general approach for the efficient and versatile stereoselective synthesis of helicates is yet available, making of these systems a challenging test case to demonstrate the potential of peptides for the controlled assembly of metallostructures.

Our strategy relied in the selection of a synthetically-accessible and structurally well-defined trimeric peptide domain as scaffold for the programmed assembly of the helicate. As an alternative (and orthogonal) platform to the ubiquitous leucine zippers, we focused our attention on the C-terminal domain of the bacteriophage T4 Fibritin foldon (T4Ff), a trimeric β-propellerlike structure formed by the self-assembly of a short 27-amino acid peptide (Tao et al., 1997; Papanikolopoulou et al., 2004; Habazettl et al., 2009). The intrinsic stability and structural resilience of the T4Ff scaffold has been exploited for the stabilization of trimeric structures of a number of peptides and engineered proteins (Stetefeld et al., 2003; Du et al., 2011; Berthelmann et al., 2014; Kobayashi et al., 2015), and given those precedents we envisioned that the T4Ff could also be used as a robust platform for the programmed assembly of chiral dinuclear helicates, thus offering an alternative for the integration of coordination and peptide chemistry beyond other widely explored peptide scaffolds.

# MATERIALS AND METHODS

# General

All reagents were acquired from the regular chemical suppliers. All solvents were dry and synthesis grade, unless specifically noted (NH4)Fe2(SO4)<sup>2</sup> • 6 H2O salt from Sigma-Aldrich was used as Fe(II) ion source. Reactions were followed by analytical UHPLC-MS with an Agilent 1200 series LC/MS using a SB C18 (1.8µm, 2.1 × 50 mm) analytical column from Phenomenex. Standard conditions for analytical UHPLC consisted on a linear gradient from 5 to 95% of solvent B for 12 min at a flow rate of 0.35 mL/min (A: water with 0.1% TFA, B: acetonitrile with 0.1% TFA). Compounds were detected by UV absorption at 222, 270, and 330 nm. Electrospray Ionization Mass Spectrometry (ESI/MS) was performed with an Agilent 6120 Quadrupole LC/MS model in positive scan mode using direct injection of the purified peptide solution into the MS detector.

# Computational Methods

The model for the <sup>33</sup>-[**(**β**AlaBpy)2-T4Ff**]3Fe+<sup>4</sup> 2 helicate was built with UCSF chimera1.12 (Pettersen et al., 2004), starting from the NMR resolved structure of the trimeric Foldon of the T4 phagehead fibritin (PDB code: 1RFO) mutating the carboxyl C-Termini to amide groups (see Results and Discussion section). Based on previous work, the model of <sup>33</sup>-[**(**β**AlaBpy)2-T4Ff**]3Fe+<sup>4</sup> 2 helicate were built connecting the N-termini of the T4Ff peptides. Molecular Dynamics (MD) simulations were set up with the xleap, solvating the model with a box of pre-equilibrated TIP3P water molecules and the total charge was balanced with Cl<sup>−</sup> ions (ions94.lib library). The AMBER14SB force field was used for standard residues (Hornak, Abel, Okur, Strockbine, Roitberg and Simmerling., 2006), while the GAFF force field was adopted for the remaining atoms. Fe-bonding force constants and equilibrium parameters were obtained through the Seminario method, using Gaussian09 to compute the geometry and harmonic frequencies at DFT level (Frisch et al., 2010), with the B3LYP functional (Yanai et al., 2004), combined with scalar-relativistic Stuttgart–Dresden SDD pseudopotential and its associated double-ζ basis plus a set of f polarization functions for the metal ion (Ehlers et al., 1993). The 6-31G(d,p) basis set was used for H, C, O, and N. Point charges were derived using the RESP (Restrained ElectroStatic Potential) model (Bayly et al., 1993). The force field building operations were carried out using the MCPB.py (Li and Merz, 2016). The solvent and the whole system were sequentially submitted to 3,000 energy minimization steps to relax possible steric clashes. Then, thermalization of water molecules and side chains was achieved by increasing the temperature from 100 K up to 300 K. MD simulations under periodic boundary conditions were carried out during 100 ns with OpenMM engine through OMMProtocol (Eastman et al., 2017; Pedregal et al., 2018). Analysis of the trajectories was carried out by means of cpptraj implemented in ambertools16 (Case et al., 2016).

# Solid-Phase Peptide Synthesis (SPPS)

All peptide synthesis reagents, as well as the Fmoc amino acid derivatives were purchased from GL Biochem (Shanghai) Ltd., Fmoc-β-Ala-OH was from Sigma Aldrich. C-terminal amide natural T4Ff peptides were synthesized following standard Fmocpeptide synthesis protocols on a 0.1 mmol scale using a 0.5 mmol/g loading H-Rink amide ChemMatrix resin (35–100 mesh) from Sigma Aldrich with a Liberty Lite automatic microwave assisted peptide synthesizer from CEM Corporation. The amino acids were coupled in 5-fold excess using oxyme as an activating agent. Couplings were conducted for 4 min at 90◦C. Deprotection of the temporal Fmoc protecting group was performed by treating the resin with 20% piperidine in DMF for 1 min at 75◦C. Once the synthesis is finished, the peptide was acetylated with a solution of 0.8 ml AcOH, 2 ml of DIEA/DMF (0.2 M) and 3.2 ml of DMF. The last non-natural Fmoc-β-Ala-Bpy-OH residues were coupled by hand in 4-fold excess using HATU as activating agent. Each amino acid was activated for 1 min in DIEA/DMF 0.2 M before being added onto the resin. These manual couplings were conducted for 60 min. Deprotection of the temporal Fmoc protecting group was performed by treating the resin with 20% piperidine in DMF for 20 min. Cleavage and deprotection of the peptide were simultaneously performed using standard conditions by incubating the resin for 2.5 h with an acidic mixture containing 50 µL CH2Cl2, 25 µL of H2O, 25 µL of TIS (triisopropylsilane), and 900 TFA µL. The resin was filtered, and the TFA filtrate was concentrated under a nitrogen stream to an approximate volume of 1 mL, and then added onto ice-cold diethyl ether (20 mL). After 10–30 min, the precipitate was centrifuged and washed again with 5 mL of ice-cold ether. The solid residue was dried under argon and redissolved in acetonitrile/water 1:1 (2–5 mL) and purified by semi-preparative RP-HPLC.

Peptides were purified by preparative RP-HPLC with an Waters 1500 series Liquid Chromatograph using a Sunfire Prep C18 OBD (5µm, 19 × 150 mm) reverse-phase column from Waters. Standard conditions for analytical and preparative RP- HPLC consisted on an isocratic regime during the first 2 min, followed by a linear gradient from 15 to 75% of solvent B for 30 min (A: water 0.1% TFA, B: acetonitrile 0.1% TFA). Compounds were detected by UV absorption (222 nm) and by ESI/MS. The fractions containing the products were freeze-dried and their identity was confirmed by ESI/MS and MALDI-TOF. Matrix-assisted laser desorption/ionization mass spectrometry (MALDI/MS) was performed with a Bruker Autoflex MALDI/TOF model in positive scan mode by direct irradiation of the matrix-absorbed peptide.

# Spectroscopic Measurements

UV measurements were made in a Jasco V-630 spectrophotometer coupled to a Jasco ETC-717 temperature controller, using a standard Hellma semi-micro cuvette (108.002- QS) with a light path of 10 mm. Measurements were made at 20◦C. Luminescence experiments were made with a Varian Cary Eclipse Fluorescence Spectophotometer coupled to a Cary Single Cell peltier accessory (Agilent Technologies) temperature controller. All measurements were made with a Hellma semi-micro cuvette (108F-QS) at 20◦C. Circular dichroism measurements were made with a Jasco J-715 coupled to a Neslab RTE-111 termostated water bath, using a Hellma 100-QS cuvette (2 mm light pass).

# Electrophoretic Mobility Shift Assays

EMSA were performed with a BioRad Mini Protean gel system, powered by an electrophoresis power supplies PowerPac Basic model, maximum power 150 V, frequency 50–60 Hz at 140 V (constant V). Binding reactions were performed over 30 min in 1.8 mM Tris-HCl (pH 7.5), 90 mM KCl, 1.8 mM MgCl2, 0.2 mM TCEP, 9% glycerol, 0.11 mg/mL BSA, and 2.2% NP-40. For the experiments we used 200 nM of the DNAs (twDNA and dsDNA), and a total incubation volume of 20 µL. After incubation for 30 min at room temperature, products were resolved by PAGE using a 10% non-denaturing polyacrylamide gel and 1 × TBE buffer (0.445 M Tris, 0.445 M Boric acid) for 35 min at 25◦C, and analyzed by staining with SyBrGold (Molecular Probes: 5 µL in 50 mL of 0.5 × TBE) for 10 min and visualized by fluorescence (BioRad GelDoc XR+ molecular imager).

# RESULTS AND DISCUSSION

As metal-chelating unit we chose 2,2′ -bipyridine, a ligand that has been extensively used in coordination chemistry and yields stable complexes with a variety of metal ions (Kaes et al., 2000). Furthermore, we have previously described an Fmocprotected 2,2′ -bipyridine dipeptide derivative that can be readily implemented into standard Fmoc solid-phase peptide synthesis (SPPS) protocols, and have showed that the structure of this chelating unit, in which the 2,2′ -bipyridine ligand is integrated in the peptide backbone, effectively couples the conformational preferences of the peptide chain with the geometry of the resulting metal complexes (Rama et al., 2012; Gamba et al., 2013, 2014, 2016; Salvadó et al., 2016).

The chelating 2,2′ -bipyridine residue was obtained following an optimized synthetic route (Rama et al., 2012), based on the work carried out by the Newkome and Imperiali groups (Newkome et al., 1997; Torrado et al., 1998). The key step in the synthesis being the desymmetrization of a diethyl [1,1′ -biphenyl]-4,4′ -dicarboxylate intermediate with hydrazine monohydrate under conditions that allow the selective precipitation of the monocarbohydrazide, which is oxidized into the corresponding azyl azide, and then transformed into a carbamate through a Curtius rearrangement (Rama et al., 2012). Simultaneous hydrolysis of the carbamate and the ester group gives the desired bipyridine amino acid, which is derivatized in the form of a dipeptide to obtain the Fmoc-βAlaBpy-OH building block for increased solubility, stability, and solubility that allow its use following standard solid-phase peptide synthesis protocols (Ishida et al., 2006).

Inspection of the structure of T4Ff (PDB IDs 4NCU or 1RFO; Güthe et al., 2004; Berthelmann et al., 2014) showed that the N-terminal Gly residues are relatively close to each other and could accommodate the chelating 2,2′ -bipyridine units without noticeable distortion of the T4Ff scaffold upon metal coordination. Moreover, we envisioned that the natural twist of the N-terminal polyproline helices in the folded T4Ff trimer should induce a 33-configuration (M helicity) on its derived helicate (Tao et al., 1997), which would be the preferred chirality for the efficient recognition of three-way DNA junctions (Oleksy et al., 2006; Gamba et al., 2016). Therefore, we synthesized the desired **(**β**AlaBpy)2-T4Ff** helicate precursor ligand following standard Fmoc SPPS methods as outlined in **Figure 1** (Coin et al., 2007). The final peptide ligand was purified by HPLC and its identity confirmed by ESI-MS.

Having at hand the desired peptides we proceeded with the study of their metal binding properties. Surprisingly, while 2,2′ -bipyridine is weakly emissive, and is even considered non-fluorescent (Dhanya and Bhattacharyya, 1992; Yagi et al., 1994), we found that the asymmetric 5′ -amido-[2,2′ -bipyridine]- 5-carboxamide unit within the βAlaBpy residue was highly emissive, displaying intense band at c.a. 420 nm with a quantum yield of 0.37 (Dong et al., 2017). Additionally, the emission was quenched by coordination to Fe(II) ions, which could be exploited to monitor the formation of the β-annulus helicate. Thus, we recorded the emission spectra of a 3µM solution (9µM monomer) of [**(**β**AlaBpy)2-T4Ff**]<sup>3</sup> in phosphate buffer (1 mM, pH 6.5) in the presence of increasing concentrations of (NH4)2Fe(SO4)<sup>2</sup> • 6 H2O (Mohr salt) as source of Fe(II) ions (λexc = 305 nm), and observed a concentration-dependent quenching of the emission intensity of the bipyridine ligands. The emission intensity profile of the titration nm could be fitted to a 1:2 binding mode with dissociation constants for the first, and second iron coordination of KD<sup>1</sup> = 5.5 ± 3.3µM and a

KD<sup>2</sup> = 6.6 ± 0.7µM, respectively (**Figure 2**, left; Kuzmic, 1996, 2009). UV/Vis titrations were also qualitatively consistent with the fluorescence data, showing a weak MLCT at about 535 nm in the presence of Fe(II) ions (See **Supplementary Material**). The formation of the expected [[**(**β**AlaBpy)2-T4Ff**]3Fe2] <sup>4</sup><sup>+</sup> complex was also confirmed by mass spectrometry of the final solution of the titrations, which showed a peak at the expected mass of the molecular ion (m/z = 11084.6).

In order to study the chirality induction around the metal centers we measured the circular dichroism spectra of the trimeric [**(**β**AlaBpy)2-T4Ff**]<sup>3</sup> ligand, and its Fe(II) complex [**(**β**AlaBpy)2-T4Ff**]3Fe+<sup>4</sup> 2 . As expected from the original structural analysis, the observed positive Cotton effect at c.a. 330 nm is consistent with the formation of a 33-helicate. Furthermore, the small change in the CD spectra upon addition of Fe(II) also suggests that the bipyridine ligands are strongly preorganized, even in absence of the metal, and that only a small rearrangement of the chromophores takes place upon coordination (**Figure 2**, right). This is consistent with earlier computational studies with related bis-bipyridyl peptide ligands, which showed that the bipyridine residues have a large tendency to stack on top of each other (Rama et al., 2012). This stacking interaction will presumably rigidify the bisbipyridyl trimer and facilitate the helical induction by the foldon domain.

In order to gain some insight into the structure and stability of the peptide helicate we performed Molecular Dynamics (MD) simulations in explicit solvent and periodic boundary conditions (see Methods section for details). The structure of the 33- [**(**β**AlaBpy)2-T4Ff**]3Fe+<sup>4</sup> 2 unit appears highly stable along all the MD trajectory retaining its helicity conformation and the Fe(II) octahedral coordination geometry. Moreover, the T4Ff scaffold appears stable during the simulation showing no appreciable deformations as a result of the introduction of the artifical (βAlaBpy)<sup>2</sup> unit. The root-mean square deviation (RMSD) of the whole system was computed along the MD using the minimized initial structures as a reference, the trajectories attain relative stable RMSD after the first ∼20 ns, that reach up to 1.99 ± 0.62 Å in average (See **Supplementary Material**). A cluster analysis was performed on the full length MD experiments showing a predominant conformations occupying about ∼40% of the total conformation repartition. Overall, the results highlight that the computed model is very stable along the 100 ns of the MD and results consistent with the experimental data. Interestingly, the

[(βAlaBpy)2 -T4Ff]3 with increasing concentrations of Fe(II). Inset shows emission at 420 nm upon excitation at 305 nm with increasing concentrations of Fe(II), and the best fit to a 1:2 binding mode (Hellman and Fried, 2007; Peberdy et al., 2007). Experiments were made in triplicate. Right. Circular Dichroism of a 6µM solution (18µM monomer) of [(βAlaBpy)2 -T4Ff]3 (dashed line) and in the presence of 90µM Fe(II) (solid line). All experiments were made in 1 mM phosphate buffer, pH 6.5, 10 mM NaCl at 20◦C.

MD analysis revealed a hinge region with increased flexibility connecting the more rigid helicate and foldon domains, which suggests the replacement of the N-terminal Gly reside for a more conformationally restricted residue in future designs.

Having made a preliminary characterization of the T4Ff helicate, we studied its DNA binding properties by titrating a 2µM solution of [**(**β**AlaBpy)2-T4Ff**]<sup>3</sup> (6µM mononer) in the presence of saturating concentrations of Fe(II) according to the previous fluorescence titrations (20µM) with increasing concentrations of a three-way DNA junction (**tw-DNA**), and measuring the fluorescence anisotropy of the bipyridine fluorophores at 420 nm after each addition of DNA. The titration profile could be fitted to a 1:1 binding mode, with a

dissociation constant of 2.17 ± 0.45µM of the [**(**β**AlaBpy)2- T4Ff**]3Fe<sup>2</sup> complex to **tw-DNA**. Titrations under the same conditions with a model double stranded DNA (**ds-DNA**) led to a small, monotonic increase in the anisotropy, which is in tune with the the formation of weak complexes or non-specific binding (**Figure 3**). The low affinity to dsDNA is consistent with previous studies with other helicates (**Figure 4**; Tuma et al., 1999; Oleksy et al., 2006; Gamba et al., 2016). Control titrations adding with [**(**β**AlaBpy)2-T4Ff**]<sup>3</sup> foldon in absence of metal did not show any response to added DNA (See **Supplementary Material**), thus confirming that the formation of the helicate structure is required for DNA recognition, and the foldon only have a structural role in the formation of the helicate.

In addition to the spectroscopic studies, we also studied the DNA binding properties of the [**(**β**AlaBpy)2-T4Ff**]3Fe<sup>2</sup> helicate by electrophoretic mobility assays (EMSA) in polyacrylamide gel under non-denaturing conditions (Liebler and Diederichsen, 2004), visualizing the DNA in the gel using SybrGold staining (Vázquez et al., 2007). In agreement with the fluorescence titration studies discussed previously, incubation of the target **tw-DNA** with the [**(**β**AlaBpy)2-T4Ff**]3Fe<sup>2</sup> helicate resulted in the concentration-dependent appearance of a new retarded band, which is consistent with the formation of the expected **tw-DNA**/[**(**β**AlaBpy)2-T4Ff**]3Fe<sup>2</sup> complex (**Figure 5**, lanes 1– 6). Additionally, the overall intensity of the lanes of the gel is progressively reduced in the presence of increasing concentrations of the [**(**β**AlaBpy)2-T4Ff**]3Fe<sup>2</sup> complex, which suggests the formation of higher-order aggregates with the threeway junction DNA in the gel conditions (Chanvorachote et al., 2009; Thordarson, 2010). On the other hand, incubation of a model double-stranded DNA with the peptide helicate did not show any new slow-migrating bands (**Figure 5**, lanes 7-10), which is in agreement with the expected low affinity for this form of DNA, and demonstrates that the small increase observed

in the fluorescence anisotropy titration of dsDNA (**Figure 4**) arises from weak interactions that are not seen at the lower concentrations used in the EMSA experiment.

# CONCLUSIONS

In summary, we have shown the potential of small protein domains for the precise structural organization of coordination complexes. Modification of the T4 Fibritin foldon with metalchelating bipyridines results allows the assembly of unique three-strand helicates in which the parallel orientation of the three helicate ligands is directed by the self-assembled T4Ff domain, and the chirality of the dinuclear helicate (M helicity or 33-configuration in the metal complexes) is selected by the relative orientation of the natural polyproline helices at the Nterminus of the T4Ff trimer. The final supramolecular peptide

# REFERENCES


helicate [**(**β**AlaBpy)2-T4Ff**]3Fe<sup>2</sup> displays good in vitro DNA binding and selectivity toward three-way DNA junctions. We are currently exploring alternative peptide sequences to improve the solubility of the peptide/DNA complexes, and modifications with positively charged residues that might increase the overall affinity.

# AUTHOR CONTRIBUTIONS

JG-G and DGP performed the experimental work (synthesis of the bipyridine building block, peptide synthesis, metal and DNA binding studies), GB did preliminary studies with the (βAlaBpy)2-T4Ff peptide. GS and J-DM did the computational work and contributed to the preparation of the final manuscript. MVL and MEV conceived the project, supervised the experimental work. MEV wrote the manuscript with the collaboration of MVL, and prepared the graphic material.

# ACKNOWLEDGMENTS

Financial support from the Spanish grants CTQ2015-70698-R, CTQ2017-87889-P, the Xunta de Galicia (Centro singular de investigación de Galicia accreditation 2016–2019, ED431G/09) and the European Union (European Regional Development Fund - ERDF), is gratefully acknowledged. JG-G, thanks the Spanish MINECO for his FPI fellowship, GB thanks the ERC for her EU METALIC-II 2013-2442/001-001-EMA2 mobility scheme fellowship, and GS. thanks the Universitat Autònoma de Barcelona for its support to his PhD. J-DM and GS are thankful for the support given by the Generalitat de Catalunya 2017SGR1323. Support of COST Action CM1306 is kindly acknowledged. MEV, also wish to acknowledge the generous support by the Fundación Asociación Española Contra el Cáncer AECC (IDEAS197VAZQ grant).

# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fchem. 2018.00520/full#supplementary-material


improved protein backbone parameters. Proteins Struct. Funct. Bioinf. 65, 712–725. doi: 10.1002/prot.21123


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Gómez-González, Peña, Barka, Sciortino, Maréchal, Vázquez López and Vázquez. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Exo-Functionalized Metallacages as Host-Guest Systems for the Anticancer Drug Cisplatin

Ben Woods, Margot N. Wenzel, Thomas Williams, Sophie R. Thomas, Robert L. Jenkins and Angela Casini\*

School of Chemistry, Cardiff University, Cardiff, United Kingdom

Within the framework of designing new self-assembled metallosupramolecular architectures for drug delivery, seven [Pd2L4] <sup>4</sup><sup>+</sup> metallacages (L = 2,6-bis(pyridine-3-yl ethynyl)pyridine) featuring different groups in exo-position, selected to enhance the cage solubility in aqueous environment, were synthesized. Thus, carboxylic acids, sugars, and PEG groups were tethered to the bispyridyl ligands directly or via disulfide bond formation, as well as via click chemistry. The ligands and respective cages were characterized by different methods, including NMR spectroscopy and high-resolution electrospray mass spectrometry (HR-ESI-MS). While the two ligands featuring carboxylic acid-functionalized groups showed improved solubility in water, the other ligands were soluble only in organic solvents. Unfortunately, all the respective self-assembled cages were also insoluble in water. Afterwards, the encapsulation properties of the anticancer drug cisplatin in selected [Pd2L4]X<sup>4</sup> cages (X <sup>=</sup> NO<sup>−</sup> 3 , BF<sup>−</sup> 4 ) were studied by <sup>1</sup>H, <sup>1</sup>H DOSY, and <sup>195</sup>Pt NMR spectroscopy. The effect of the counter ions as well as of the polarity of the solvent in the drug encapsulation process were also investigated, and provided useful information on the host-guest properties of these experimental drug delivery systems. Our results provide further experimental support for previous studies that suggest the desolvation of guests from surrounding solvent molecules and the resulting solvent rearrangement may actually be the primary driving force for determining guest binding affinities in metallacages, in the absence of specific functional group interactions.

Keywords: metallacages, supramolecular coordination complexes, water solubility, host-guest chemistry, cisplatin

# INTRODUCTION

Supramolecular chemistry has its roots in biology, and in recent years the unique and often advantageous properties of supramolecular materials have led to their extensive exploration in the fields of biomolecular recognition, drug delivery, disease diagnosis, and imaging. In this context, supramolecular coordination complexes (SCCs) hold great promise (Cook et al., 2013; Casini et al., 2017). In fact, the number of reports on the bioactivity of three-dimensional (3D) SCCs with different shapes has substantially increased (Kaner et al., 2016; Preston et al., 2016; Casini et al., 2017) and includes helicates (Schmitt et al., 2012), metallacages (Vajpayee et al., 2011), cubes (Ahmedova et al., 2016), prisms(Ahmad et al., 2015), and capsules (Therrien, 2013). These systems are of general formula MnLm, where M is usually Fe(II), Pd(II), Pt(II), or half-sandwich organometallic clips based on Ru(II), Os(II), or Ir(III) and Rh(III), and L is the ligand of the coordination complex.

#### Edited by:

Leyong Wang, Nanjing University, China

#### Reviewed by: Xiao-Yu Hu,

Nanjing University of Aeronautics and Astronautics, China Laia Vilà Nadal, University of Glasgow, United Kingdom Robert Elmes, Maynooth University, Ireland

> \*Correspondence: Angela Casini casinia@cardiff.ac.uk

#### Specialty section:

This article was submitted to Supramolecular Chemistry, a section of the journal Frontiers in Chemistry

Received: 05 November 2018 Accepted: 24 January 2019 Published: 18 February 2019

#### Citation:

Woods B, Wenzel MN, Williams T, Thomas SR, Jenkins RL and Casini A (2019) Exo-Functionalized Metallacages as Host-Guest Systems for the Anticancer Drug Cisplatin. Front. Chem. 7:68. doi: 10.3389/fchem.2019.00068

In 2012, Lewis et al. reported on the encapsulation properties of the anticancer drug cisplatin within cationic [Pd2L4] 4+ cages (L = 2,6-bis(pyridine-3-ylethynyl)pyridine as the bidentate ligand) studied by NMR and X-ray diffraction analysis (Lewis et al., 2012). Following these promising results, as part of our efforts to develop functional metallosupramolecular drug delivery vectors, we have recently expanded the family of exofunctionalized [Pd2L4] <sup>4</sup><sup>+</sup> cages coupled to fluorescent groups for imaging by fluorescence microscopy (Kaiser et al., 2016; Schmidt et al., 2016a,b), as well as to peptides for targeted drug delivery of cisplatin to cancer cells (Han et al., 2017, 2018). It is worth mentioning that while cisplatin occupies a crucial role in the treatment of various malignant tumors, its efficacy, and applicability are heavily restricted by severe systemic toxicities and drug resistance. Thus, our recent studies have exploited the host-guest properties of targeted metallacages to enhance the activity of cisplatin in cancer cells, protecting it from metabolism (Han et al., 2017). Of note, the reduced toxicity of cisplatin encapsulated in integrin targeted metallacages has also been demonstrated by us in an ex vivo model (Han et al., 2017).

As the initially developed cage architectures displayed scarce water solubility, we were interested in improving their hydrophilicity via derivatization of the ligand scaffold. Thus, here, a small library of bis(pyridyl) ligands—of general scaffold 3,5-bis(3-ethynylpyridine)phenyl (**Scheme 1**) - exofunctionalized with chemical groups envisioned to improve the aqueous solubility, were synthesized and self-assembled into the corresponding [Pd2L4] <sup>4</sup><sup>+</sup> cages. The products were characterized by NMR, HR-ESI-MS, and IR. The solubility of the resulting metallacages was then tested in water, as well as in a range of common organic solvents. Furthermore, in order to elucidate the host-guest chemistry of the metallacages and the factors that favor cisplatin's retention in the supramolecular complexes, essential aspects to exploit the cages as drug delivery systems, the drug encapsulation was studied by <sup>1</sup>H, <sup>1</sup>H DOSY, and <sup>195</sup>Pt NMR spectroscopy under different conditions.

# RESULTS AND DISCUSSION

Adapting previously reported procedures (Schmidt et al., 2016b), the synthesis of the exo-functionalized ligand scaffold with carboxylic acid groups, **L1,** and **L2** was first achieved by Sonogashira cross-coupling of a benzyl-protected 1,3 dibromophenyl precursor with 3-ethynylpyridine, followed by basic deprotection of the benzyl ester to afford the free acid, forming the highly conjugated bispyridyl ligand scaffold in good yields (**Scheme S1**). A polyethylene glycol exo-functionalized ligand (**L3**) was also synthesized. In this case, a ligand bearing a central phenol aromatic ring was synthesized via Sonogashira cross-coupling, which was then was coupled to 1-chloro-2-(2-methoxyethoxy) ethane under basic conditions to achieve ligand **L3** (**Scheme S2**). In addition, ligands **L4** and **L5,** featuring water-soluble thio-β-D-glucose moieties, were synthesized for the first time. This was achieved by first synthesizing an alkyl bromide exo-functionalized ligand via classic Appel reaction using triphenylphosphine, carbon tetrabromide, and 3,5-bis(3-ethynylpyridine)-benzylalcohol; the latter having been synthesized via Sonogashira cross-coupling (**Scheme S3**). The 1-thio-β-D-glucose tetraacetate group was then conjugated under basic conditions following a SN2 type reaction to afford the ligand **L4**. Ligand **L5** was achieved by deacetylation of **L4** using Amberlite-IRA 401 (OH−) ion exchange beads, which were stirred together at room temperature overnight, following previously reported procedures (Lewis et al., 2014). Finally, the ligands **L6** and **L7** were synthesized using standard CuAAC "Click" reaction conditions (**Scheme S4**) following similar procedures reported for analogous tris-pyridyl ligand scaffolds (Lewis et al., 2013). Noteworthy, the latter were reported to be soluble in water, and therefore, offered a logical starting point for functionalization of the herein selected 3,5-bis(3-ethynylpyridine) phenyl ligands. All the ligands were characterized by NMR (1H, <sup>13</sup>C) and HR-ESI-MS. A summary of the synthesized ligands and cages can be found in **Scheme S5**.

The solubility of the ligands in water was assessed by overnight sonication in D2O (10 mg/mL, 20◦C), before filtration (pore size 450µm). The filtrate was analyzed by <sup>1</sup>H NMR to determine if detectable amounts of ligand were present. Unfortunately, the exo-functionalized bispyridyl ligand scaffolds were found to be scarcely soluble, and were all undetectable in D2O solution, except for the carboxylic acid functionalized ligands (**L1** = 10 mg/mL, **L2** > 100 mg/mL). The latter could be analyzed by <sup>1</sup>H NMR in D2O (data not shown).

Afterwards, [Pd2L4] <sup>4</sup><sup>+</sup> cage formation was achieved via self-assembly by reacting two equivalents of Pd(II) nitrate pentahydrate with four equivalents of the functionalized ligand (**L1-L7**) in DMSO at room temperature, to form the corresponding homoleptic metallacages (**C1-C7**) (**Scheme 1**). The products were analyzed by NMR (1H, <sup>13</sup>C, <sup>11</sup>B, and <sup>19</sup>F), HR-ESI-MS (**Figures S1–S3**, **S5–S9**) and IR. Following previously reported examples (Schmidt et al., 2016c), quantitative cage formation was monitored by <sup>1</sup>H NMR, whereby the peaks corresponding to the protons alpha to the coordinating nitrogen of the terminal pyridine rings of the ligand undergo a large downfield shift, attributable to the withdrawal of electron density from the aromatic system following coordination to palladium (added in stoichiometric amounts). A representative <sup>1</sup>H NMR spectra for cage **C2** formation is shown in **Figure S4**.

In order to assess whether the synthesized metallacages were soluble in water, the products were suspended in D2O (8 mg per mL) and sonicated at 37◦C for 24 h. The fine suspension was then filtered, and the filtrate was analyzed by <sup>1</sup>H NMR. Despite the tetra-cationic nature of the metallacages, and the incorporation of solubilizing entities to the ligand scaffold, none of the metallacages (**C1-C7**) were soluble in D2O. In fact, the filtrate solution was colorless and no trace of a proton signal could be found in the <sup>1</sup>H NMR spectra.

As a second attempt to water solubilization, and taking inspiration from previous work by Altmann and Pöthig (2016) on the synthesis of water soluble Ag(I) and Au(I) pillarplexes,

we synthesized new cages featuring acetate as the counterion by the use of palladium(II) acetate as the precursor in the self-assembly. However, this attempt produced an insoluble residue that could not be dissolved in either organic solvents or water.

Concerning cisplatin encapsulation, in the case of previously reported [Pd2L4]BF<sup>4</sup> cages (L = (2,6-bis(pyridin-3 ylethynyl)pyridine)) it was hypothesized that their central cavities are lined with four hydrogen bond accepting pyridine units which enable the encapsulation of two cisplatin molecules within the metallosupramolecular architecture through hydrogen bonding interactions between the cage and the amine ligands of the cisplatin guest (Lewis et al., 2012). This idea has been supported by NMR spectroscopy studies (Lewis et al., 2012), where chemical shifts of hydrogens in the trispyridyl ligands were observed upon cisplatin encapsulation. Control experiments confirm the importance of the hydrogen bonding interaction between the cage and the amine ligands of the cisplatin guest. It was also observed that upon addition of small amounts of D2O to a solution of the cisplatin-cage host-guest adduct in acetonitrile, the Ha, and H<sup>b</sup> proton signals of the cage sharpen and shift upfield again, indicating evacuation of cisplatin from the cavity (Lewis et al., 2012).

In this context, information on the encapsulation process of cisplatin within the more hydrophobic cavity of the bispydridyl ligand forming our [Pd2L4] <sup>4</sup><sup>+</sup> cages, as opposed to the less hydrophobic cavity of trispyridyl-ligand Pd2L<sup>4</sup> metallacages (Lewis et al., 2012, 2014), needs to be achieved still. Thus, we investigated if the encapsulation of the anticancer drug cisplatin could occur in the metallacages using <sup>1</sup>H NMR spectroscopy. In order to avoid DMSO as the solvent, since it is known to exchange cisplatin's chlorido ligands, the experiment was carried out in DMF-d7. Cage **C1** [Pd2L4](NO3)<sup>4</sup> was selected to study the encapsulation process due to its solubility in this solvent. **Figure 1** shows the variation of the chemical shifts of selected cage protons upon cisplatin encapsulation. Specifically, we could clearly observe the chemical shifts of peaks H<sup>a</sup> and Hb: the peak of Ha, a cavity-facing proton, showed a small upfield shift of δ −1.0 Hz. The exo-facing proton, Hb, showed a downfield shift upon cisplatin encapsulation, in accordance with previously reported studies (Lewis et al., 2012). Unfortunately, the cavity facing peak He, likely to interact with the guest Pt(II) complex, was masked by the peak of H<sup>f</sup> , and therefore, could not be analyzed.

It is worth mentioning that the observed smaller chemical shifts of the cavity facing proton Ha, with respect to those reported by Lewis et al. (2012) can be due to the lack of the H-bonding with the cisplatin guest (via N—H3N). Moreover, it has been previously reported that NO<sup>−</sup> 3 counterions are encapsulated within the cage architecture of trispyridyl Pd2L<sup>4</sup> metallacages featuring a nitrogen in endo-position, while BF<sup>−</sup> 4 ions are not (Lewis and Crowley, 2014). Thus, we hypothesized

that the observed small chemical shift of H<sup>a</sup> following cisplatin encapsulation was due to the presence of NO<sup>−</sup> 3 ions inside the cage, which already induced chemical shifts of H<sup>a</sup> and H<sup>b</sup> and thus, acted to "mask" the effect of the cisplatin encapsulation process. In fact, upon evacuation of the anionic NO<sup>−</sup> 3 counterion, the "de-shielding" effect of the cavity facing protons from NO<sup>−</sup> 3 would be reduced as the neutral guest, cisplatin, occupies the metallacage cavity.

Afterwards, we also investigated if cisplatin would preferentially be encapsulated within our bispyridyl type Pd2L<sup>4</sup> metallacages featuring a different counter ion. Therefore, an analogous metallacage was synthesized with BF<sup>−</sup> 4 as the counterion, **C1.BF<sup>4</sup>** and studied for its cisplatin encapsulation properties by <sup>1</sup>H NMR spectroscopy. **Supporting Figure S10** shows the stacked <sup>1</sup>H NMR spectrum of **C1.BF4**, alone or encapsulating cisplatin [**C1.BF4**⊂**(cisplatin)**] in DMF-d7. Indeed, both H<sup>a</sup> and H<sup>b</sup> show marked downfield shifts upon cisplatin encapsulation of 0.044 and 0.033 ppm, respectively. In addition, H<sup>e</sup> also shows a downfield shift of 0.03 ppm. Furthermore, the peaks of both H<sup>a</sup> and H<sup>b</sup> of **C1** (before addition of cisplatin) are already shifted further downfield (**Figure 1**, ca. 0.095 and 0.001 ppm, respectively) when compared to H<sup>a</sup> and H<sup>b</sup> of **C1.BF<sup>4</sup>** (**Figure S10**) consistent with encapsulation of NO<sup>−</sup> 3 .

To continue our investigation, the effect of the solvent polarity was hypothesized to be the driving force of the cisplatin encapsulation process. Thus, the latter was also tested using **C1.BF<sup>4</sup>** in MeCN-d3, with reduced polarity with respect to DMF-d7. However, upon addition of 2 eq. cisplatin to the MeCN-d<sup>3</sup> solution of **C1.BF<sup>4</sup>** (1 equation) a precipitate was formed, even after several hours of sonication. The suspension was filtered, and the resulting solution was analyzed by <sup>1</sup>H NMR spectroscopy. However, the resulting spectra showed no indication of cisplatin encapsulation (data not shown). Instead, the large upfield shift of the peaks corresponding to the αprotons of the coordinating pyridyl is indicative of the ligands no longer coordinated to Pd(II) ions. Moreover, the NMR spectrum suggests that the structure of the ligand has also been altered.

Therefore, in order to study the effects of the solvent on cisplatin encapsulation, we synthesized a derivative of cage **C1.BF4**, but where the carboxylic function was protected by a benzyl group, namely cage **C1Bn.BF4**. Thus, cisplatin encapsulation was followed by <sup>1</sup>H NMR in MeCN-d<sup>3</sup> and the results are presented in **Figure S11**. As it can be observed, no chemical shifts could be seen for any of the cage's proton peaks even after 24 h incubation of cisplatin at room temperature. We could conclude that in acetonitrile, cisplatin encapsulation is not favored. In order to check that metallodrug encapsulation was not prevented by the presence of the benzyl group on the cage, we also monitored the process once more in DMF-d7. The results are depicted in **Figure S12** and indicate that, in these conditions, cisplatin encapsulation occurs, as shown by the chemical shifts of H<sup>a</sup> (downfield, δ = +0.009 ppm), H<sup>b</sup> (downfield, δ = +0.021 ppm), and a small shift for the cavityfacing proton H<sup>e</sup> (upfield, δ = −0.002 ppm). The encapsulation of cisplatin in cage **C1Bn.BF<sup>4</sup>** was further studied in DMF-d<sup>7</sup>

using <sup>195</sup>Pt NMR to see if a complementary shift could be observed for the guest cisplatin molecules. Indeed, an upfield shift of −5 ppm was observed upon addition of either 1 eq. or 2 eq. of cisplatin in the presence of metallacage **C1Bn.BF<sup>4</sup>** (**Figure S13**), supporting cisplatin encapsulation. The obtained results are in agreement with previously reported ones from our group on similar cage systems exo-functionalized with different type of ligands (Schmidt et al., 2016c), indicating that surface functionalization does not prevent the guest from entering the cavity.

The encapsulation of cisplatin in **C1Bn.BF<sup>4</sup>** was also studied by <sup>1</sup>H DOSY NMR in DMF-d7. Cisplatin alone shows a broad signal at 4.18 ppm, HNH<sup>3</sup> with a diffusion coefficient = −8.9 × 10−<sup>10</sup> m<sup>2</sup> /s (**Figure S14**, red traces). Clear quenching of this signal was observed upon addition of 1 eq. of metallacage **C1Bn.BF<sup>4</sup>** (**Figure S14**, blue traces). The cisplatin peak reappears slightly upon addition of a second equivalent of cisplatin to **C1Bn.BF4**⊂**(cisplatin)** (**Figure S14**, black traces), which may suggest that the metallacage cavity has been saturated to form **C1Bn.BF4**⊂**(cisplatin)<sup>2</sup>** and residual cisplatin remains free in solution, or that upon addition of a second equivalent of cisplatin, a dynamic equilibrium occurs.

# CONCLUSIONS

In conclusion, we report here on a series of exo-functionalized [Pd2L4] <sup>4</sup><sup>+</sup> metallacages featuring bispydridyl ligands. Despite the addition of hydrophilic groups to the ligands' scaffold, or the use of different counter ions, the increase in water solubility of the resulting cages was not achieved. Thus, different strategies are necessary to address this problem, including the conjugation of the ligands to peptides. The latter approach has already proved to be successful (Han et al., 2017).

Furthermore, the encapsulation of the anticancer drug cisplatin in selected cages has been studied by NMR spectroscopy, and the obtained results show that if the solvent is of sufficient polarity, metallodrug encapsulation can easily occur in the hydrophobic cavity of the cage despite the absence of hydrogen-bond accepting central pyridine units (Lewis and Crowley, 2014). Conversely, polar solvent molecules capable of forming hydrogen-bond networks (including DMF and water) are likely to prefer not to be encapsulated by hydrophobic cage cavities, as in our case, and will not compete with the cisplatin molecules. This hypothesis is corroborated by previous studies on M4L<sup>6</sup> metallacages for which guest encapsulation in polar protic solvents, such as water, appears to be driven by initial desolvation of the guest with concomitant rearrangement of the hydrogen bond networks in solution, more than by host-guest interactions alone (Leung et al., 2008). In general, energetically favorable solvent rearrangement during guest desolvation also explains the ability of the supramolecular host to encapsulate neutral guests such as cisplatin, in spite of the lack of possible electrostatic attraction or H-bond interactions with the charged host, as in our case. Further studies are ongoing in our lab to validate this hypothesis.

# AUTHOR CONTRIBUTIONS

AC and BW contributed conception and design of the study. BW, MW, and ST contributed to the design, synthesis, and characterization of the ligands, and of the palladium cages. TW contributed to the HR-ESI-MS analysis of the cages. BW and RJ performed the encapsulation studies by NMR spectroscopy. AC and BW wrote the first draft of the manuscript, while MW wrote sections of the manuscript. All authors contributed to manuscript revision, read, and approved the submitted version.

# REFERENCES


# ACKNOWLEDGMENTS

Authors thank Cardiff University for funding.

## SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fchem. 2019.00068/full#supplementary-material


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Woods, Wenzel, Williams, Thomas, Jenkins and Casini. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Antimicrobial Activity and Cytotoxicity of Ag(I) and Au(I) Pillarplexes

Alexander Pöthig<sup>1</sup> \*, Sara Ahmed<sup>2</sup> , Hanne Cecilie Winther-Larsen<sup>2</sup> , Shengyang Guan<sup>1</sup> , Philipp J. Altmann<sup>1</sup> , Jürgen Kudermann<sup>1</sup> , Adriana Magalhães Santos Andresen<sup>2</sup> , Tor Gjøen<sup>2</sup> and Ove Alexander Høgmoen Åstrand<sup>3</sup> \*

<sup>1</sup> Department of Chemistry and Catalysis Research Center, Technical University of Munich, Garching, Germany, <sup>2</sup> Department of Pharmaceutical Biosciences, School of Pharmacy, University of Oslo, Oslo, Norway, <sup>3</sup> Department of Pharmaceutical Chemistry, School of Pharmacy, University of Oslo, Oslo, Norway

The biological activity of four pillarplex compounds featuring different metals and anions was investigated. The toxicity of the compounds against four bacterial strains [Bacillus subtilis (ATCC6633), Staphylococcus aureus (ATCC6538), Escherichia coli (UVI isolate), Pseudomonas aeruginosa], one fungus (Candida albicans), and a human cell line (HepG2) was determined. Additionally, a UV-Vis titration study of the pillarplexes was carried out to check for stability depending on pH- and chloride concentration changes and evaluate the applicability in physiological media. All compounds are bioactive: the silver compounds showed higher activity against bacteria and fungi, and the corresponding gold pillarplexes were less toxic against human cells.

Keywords: NHC, silver, gold, antimicrobial, cytotoxicity, organometallic, supramolecular, stability

# INTRODUCTION

Since the early 2000s, coinage metal complexes featuring N-heterocyclic carbenes (NHC)—a ligand class with a facile tunability toward sterics, electronics, and solubility—have been employed as bioactive compounds (Herrmann, 2002; Mercs and Albrecht, 2010; Hopkinson et al., 2014). As first examples, silver (I) NHC complexes have been used as antimicrobial compounds, pioneered by Youngs et al. (Kascatan-Nebioglu et al., 2004; Melaiye et al., 2004), and a respective applicability of such compounds has been shown for a variety of complexes ever since (**Figure 1**) (Kascatan-Nebioglu et al., 2007; Hindi et al., 2009; Oehninger et al., 2013; Liang et al., 2018). Hereby, a slow release of silver ions originating from the decomposition of the NHC complexes is expected to be the cause of their activity, which can be rationalized by the comparably labile metal-carbene bond (with respect to other late transition metal-NHC bonds) (Kascatan-Nebioglu et al., 2007). The more stable gold (I) NHC complexes were also employed in studies investigating their antibiotic potential (Lazreg and Cazin, 2014). One possible target are (seleno)-cysteine moieties in proteins, e.g., thioredoxin reductase, accompanied by the inhibition of the enzyme, which is similar to the mode of action proposed for the approved metallodrug Auranofin (Baker et al., 2005; Schuh et al., 2012). This is expected in particular for gold(I) mono-carbene complexes, which can dissociate one (labile non-NHC) ligand to coordinate the sulfur or selenium atom (Rubbiani et al., 2011, 2013; Cheng et al., 2014; Meyer et al., 2014; Arambula et al., 2016; Bertrand et al., 2017; Karaca et al., 2017a; Schmidt et al., 2017; Zhang et al., 2018). In case of the di-NHC complexes, which are more stable toward dissociation, a different mode of action can be observed. Casini and coworkers were able to show stacking of Au(I) di-caffeine NHC complexes in G4 quadruplex DNA

#### *Edited by:*

Angela Casini, Cardiff University, United Kingdom

#### *Reviewed by:*

Zoe Pikramenou, University of Birmingham, United Kingdom Stephane Bellemin-Laponnaz, Centre National de la Recherche Scientifique (CNRS), France

#### *\*Correspondence:*

Alexander Pöthig alexander.poethig@tum.de Ove Alexander Høgmoen Åstrand o.a.h.astrand@farmasi.uio.no

#### *Specialty section:*

This article was submitted to Supramolecular Chemistry, a section of the journal Frontiers in Chemistry

*Received:* 30 September 2018 *Accepted:* 08 November 2018 *Published:* 27 November 2018

#### *Citation:*

Pöthig A, Ahmed S, Winther-Larsen HC, Guan S, Altmann PJ, Kudermann J, Santos Andresen AM, Gjøen T and Høgmoen Åstrand OA (2018) Antimicrobial Activity and Cytotoxicity of Ag(I) and Au(I) Pillarplexes. Front. Chem. 6:584. doi: 10.3389/fchem.2018.00584

**97**

FIGURE 1 | (Left) Overview of related known biomedical relevant compound classes. (Right) NHC based organometallocavitands including precursor investigated in this study.

structures, inhibiting telomerase activity (Bertrand et al., 2014; Bazzicalupi et al., 2016; Karaca et al., 2017b). Hereby, the overall structure of the intact complex (being planar, cationic, and possessing a conjugated system for stacking) determines the ability to interact in a non-covalent binding, forming supramolecular aggregates. A related supramolecular recognition of biomolecules causing bioactivity was discovered by Michael Hannon and coworkers, who were using cylindrical metal helicates—a class of supramolecular coordination complexes (SCCs, **Figure 1**)—to interact with different DNA structures (Meistermann et al., 2002; Oleksi et al., 2006; Hannon, 2007; Ducani et al., 2010; Phongtongpasuk et al., 2013; Malina et al., 2016). They showed, that the overall charge of the compounds (4+) as well as the aromatic parts of the ligands were crucial for supramolecular recognition of the negatively charged DNA. In general, such supramolecular coordination compounds are discussed as a promising class for future applications as metallodrugs or drug delivery systems (Casini et al., 2017).

We recently introduced the pillarplexes (**Figure 1**), a new family of SCCs which are structurally similar to Hannon's cylindrical helicates yet additionally exhibit a pore which allows for encapsulation of guest molecules inside the complex (Altmann and Pöthig, 2016). These compounds are octanuclear coinage metal complexes with two coordinating macrocyclic NHC ligands. Due to their in-built functionality (e.g., luminescence, easy tunable solubility) caused by the metal-complex character, the pillarplexes are even more versatile than their highly successful organic relatives—the pillararenes (Ogoshi et al., 2008, 2016). The latter have also been applied in biomedical applications very recently, for reducing cytotoxicity, and improving the anticancer bioactivity of oxaliplatin (Hao et al., 2018). In case of the pillararenes and metallocage systems (Casini et al., 2017), the cavitand itself shows no bioactivity and therefore can be used to modulate the selectivity and activity of an actual metallodrug.

Our pillarplexes combine the possibilities to behave like NHC complexes, i.e., as a metallodrug themselves, with the possible applications of cavitands. Therefore, to explore the future potential of our pillarplexes in the biomedical context, we conducted a toxicity study. We tested the antimicrobial activity of the four metal complexes (**3–6**), the metal salts as well as the ligand precursor salts (**1**, **2**) toward four different bacterial strains [Bacillus subtilis (ATCC6633), Staphylococcus aureus (ATCC6538), Escherichia coli (UVI isolate), Pseudomonas aeruginosa] as well as one fungus (Candida albicans). We also evaluated the toxicity of the complexes toward a human cell line (HepG2) in order to clarify if related future research directions might be promising to follow. Finally, we conducted a stability study of the pillarplexes toward changes in pH and chloride ion concentration, which has implications on the use of the compounds under physiological conditions.

# EXPERIMENTAL SECTION

# General Details

Compounds **1**–**6** were prepared according to the reported procedures (Altmann et al., 2015; Altmann and Pöthig, 2016). Chemicals were purchased from commercial suppliers and used without further purification if not stated otherwise. Liquid NMR spectra were recorded on a Bruker Avance DPX 400 and a Bruker DRX 400 at room temperature if not stated otherwise. Chemical shifts are given in parts per million (ppm) and the spectra were referenced by using the residual solvent shift as internal standards. Emission spectra were recorded on a Agilent Cary 60 UV-Vis. Nutrient agar plates were prepared according to the instructions provided by Oxoid where 28 g of nutrient

agar (CM0003) was needed to make 1 L of nutrient agar broth. 11.2 g of the agar was added to three 400 ml glass bottles. Four hundred milliliters of distilled water was added into each bottle containing the nutrient agar and was dissolved by stirring. After sterilization, the nutrient agar bottles were cooled to 50◦C and then placed into a 50◦C water bath for the temperature to remain constant. The nutrient agar was then poured halfway into 9 cm sterile petri dishes in HEPA filtered laminar flow cabinets to minimize the risk of contamination. The nutrient agar plates were then left to solidify and were refrigerated at 4 ◦C. Mueller-Hinton agar plates were prepared from Mueller-Hinton agar medium (Sigma-Aldrich) and agar (Oxoid LP0011). Twenty-two grams of the Mueller-Hinton medium was added into 1 L of distilled water in a volumetric flask and dissolved with a magnetic flea at speed 6–7 and temperature 300◦C for ∼10 min (IKA Labortechnik). Fifteen grams of agar was added to the mixture and the stirring continued at speed 5–6 and at temperatures between 200 and 250◦C until the mixture began to boil. After sterilization, the Mueller-Hinton broth was cooled to 50◦C and stirred slightly with a magnetic flea for ∼1 min (IKA Labortechnik). Sixty milliliters of the Mueller-Hinton broth was poured into each 13 cm petri dish using the media dispensing machine (IBS Integra Biosciences Technomat) using aseptic techniques. The Mueller-Hinton plates were allowed to cool and then stored in a refrigerator at 4◦C. Plates were sterilized and stored at 4◦C in the refrigerator before use. One liter of a 0.9% solution of sodium chloride was prepared and sterilized at 121◦C for 20 min at 1 atm. The bacteria were streaked onto a nutrient agar plate using a sterile loop and incubated at 37◦C overnight. The fungus was streaked using a sterile loop onto a TSA plate and incubated at 25◦C for 48 h. Fresh streaks were prepared for each disc diffusion assay.

# Disc Diffusion Assays for Antimicrobial Activity

Antimicrobial activity was measured using the disc diffusion assay essentially as described in guidelines from Clinical and Laboratory Standards Institute CLSI (2012). The bacteria were maintained on Nutrient agar (Oxoid), while the fungus was maintained on Sabouraud dextrose agar (Oxoid). An inoculum of the test microorganisms were made by resuspending freshly overnight grown colonies into 2 mL of a sterile salt solution (0.9% NaCl). The test organism was diluted to McFarland standard density no. 2 and mixing thoroughly (McFarland, 1907). For the Gram-negative bacteria and fungus, 60 µL of the inoculum was added to 25 mL of sterile salt solution, while 120 µL was added for the Gram-positive bacteria.

To prepare the plates for the disc diffusion assay Mueller-Hinton agar 2 (Sigma-Aldrich) were covered with 5 mL of the freshly made inoculate. The surplus inoculate was removed and the plates were then left in a laminar flow hood until the surface of the plates were completely dry.

Six millimeter filter discs were impregnated with 10 µL volume of the ligand precursor compounds **1** [L(PF6)4] and **2** [L(OTf)4], silver pillarplexes **3** [Ag8L2(PF6)4] and **4** [Ag8L2(OAc)4], and gold pillarplexes **5** [Au8L2(PF6)4] and **6** [Au8L2(OAc)4]. The concentration of the compounds used were 10 mM. Further filter discs were also impregnated with 10 µl of: dimethyl sulfoxide (DMSO) acting as a negative control; 10 mM of silver nitrate, 10 mM of gold chloride acting as model compounds for free metal ions; and antibiotic discs including pre-impregnated 30 mg/ml gentamycin sulfate discs (BD BBL Sensi-Disk) (E. coli, S. aureus, P. aeruginosa), pre-impregnated 30 mg/ml tetracycline discs (BD Sensi-Disc) (B. subtilis), and 10 mM of Miconazole nitrate discs (Sigma-Aldrich) (C. albicans), acting as positive controls. The filter discs were placed evenly on 13 cm Mueller-Hinton agar plates separated to avoid overlapping inhibitions zones. The plates were incubated overnight at 32◦C for the bacteria or 25◦C for the fungus. The inhibition zones were measured with a caliper. All experiment was performed at least three times.

# *In vitro* Toxicity in HepG2 Liver Cells

Human hepatocarcinoma cell line HepG2 (HB-8065, ATCC, Manassas, VA, USA) was cultured in MEM-Glutamax (5.5 mM glucose) supplemented with 10% fetal bovine serum (Gibco, Life Technologies AG, Basel, Switzerland), 100µg/mL streptomycin, and 100 units/mL penicillin (both from Gibco, Life Technologies AG, Basle, Switzerland). Cells were incubated at 37◦C under a 5% CO<sup>2</sup> atmosphere. For viability assays, cells were seeded in white 96-well Nunc plates at a density of 20,000 cells/well and left overnight to adhere before experiments were conducted.

The compounds were dissolved in DMSO at concentrations ranging from 10−<sup>3</sup> to 10−<sup>6</sup> M and were added to white 96-well plates (maximum DMSO concentration in wells was lower than 1%) containing 20,000 HepG2 cells/well. Plates were incubated for 24 h at 37◦C in a 5% CO<sup>2</sup> atmosphere. After 24 h, AlamarBlue cell viability reagent (Thermo Fisher, Carlsbad, CA, USA) was added as a 10% solution, and plates were placed back in the incubator for 4 h. AlamarBlue is a redox indicator yielding a fluorescence signal proportional to the number of viable cells in each well (O'Brien et al., 2003). The fluorescence signal was measured in a microplate reader (Clariostar, BMG Labtech, Ortenberg, Germany) at 550 nm/603 nm (excitation/ emission). Data from four replicates were used to calculate the halfmaximal inhibitory concentration (IC50) using Sigmoidal, 4PL, where X is log(concentration) analysis, and a four-parameter logistic regression from GraphPad Prism 7 (GraphPad Software Inc., USA). The experiment was repeated twice with similar results.

# UV-Vis Experiments

### Stability Tests of Pillarplexes Against Chloride

The titrations of the silver pillarplex **4** and gold pillarplex **6** against chloride ions were carried out by stepwise addition of an increasing volume of a 3.072 M sodium chloride solution to 2 mL of a 1.38 · 10−<sup>5</sup> M aqueous pillarplex solution followed by thorough mixing in a quartz cuvette. The UV-Vis absorption spectra were recorded immediately after the addition. The measured

absorbance was corrected for the increase of the sample volume.

### Stability Tests of Pillarplexes Against pH

The stability of silver pillarplex **4** and gold pillarplexes **6** in different concentrations of trifluoromethanesulfonic acid (HOTf) was monitored by UV-Vis spectroscopy. One milliliter of a 2.76 · 10−<sup>5</sup> M aqueous pillarplex solution were injected into an equal volume of 1 mL HOTf solution with pH-values 2, 4, 5, and 6 in the quartz cuvette. The absorption spectra were recorded after 1 min, 1, 7, 24, 48, and 72 h (see **Supplementary Material**).

# RESULTS AND DISCUSSION

## Antimicrobial Activity Studies

The results of the antimicrobial studies are summarized in **Table 1**.

Both silver compounds (entries 3 and 4) show antimicrobial activity against all bacterial strains as well as the fungus. Hereby, the activity is independent of the anion present, as the results are identical within the margin of errors. In comparison to the positive controls (entries 10–12) the overall activity is moderate, however, (by means of statistic uncertainty) it is identical to that of AgNO<sup>3</sup> (entry 7), which has been used as an antibiotic since ancient times (Danscher and Locht, 2010). Hence we suspect the release of silver ions via decomposition of the pillarplexes, which is in agreement with the general behavior of silver(I) NHC complexes, as stated above.

The gold pillarplexes show lower to no activity (entries 5 and 6). Compound **6**, the completely water soluble acetate, shows no activity against any of the microbes, whereas, the more lipophilic compound **5** shows a selective moderate activity against Gram-negative E. coli and Gram-positive S. aureus. In contrast, AuCl<sup>3</sup> shows activity against all bacterial strains (interestingly not against the fungus), which of course might be additionally influenced by the redox activity of the gold(III) ion. However, we suspect the gold pillarplexes being more stable in the physiological environment, therefore not releasing uncoordinated metal ions, which would explain the lower activity. Similarly, if the gold complexes would decompose, a similar toxicity as in case of the free ligand precursors would be expected. In general, such imidazolium salts are known to be potentially toxic, depending on different factors, e.g., lipophilicity or anions (Gravel and Schmitzer, 2017). In our case, the two macrocyclic polyimidazolium ligand precursors (entries 1 and 2) show only moderate and very selective toxicity only against the Gram-positive bacteria S. aureus and B. subtilis. Gram-positive bacteria lacks the outer membrane surrounding the cell wall. This outer membrane excludes, by various mechanisms, certain drugs from penetrating the bacterial cell (Hancock, 1997) and could be the reason for antimicrobial selectivity of compound **2**. For the latter, no activity at all was observed in case of the gold pillarplexes, why we rule out a possible decomposition.

# Cell Toxicity Studies

The results for the toxicity study of the compounds against human HepG2 liver cells are summarized in **Table 2**. The IC50 values were determined for all compounds, however, the silver pillarplexes (3 and 4) as well as AgNO<sup>3</sup> and AuCl<sup>3</sup> all showed precipitation to some degree. This can influence both uptake of the compounds by the cells and the absorbance read, resulting in ambiguous measurement results, which we have pointed out by an asterisk in **Table 2**.

In general, all tested compounds exhibit biological activity. Both ligand precursors (**1** and **2**) exhibited low toxicity levels which corresponds to the determined IC50−values. In contrast, high cell toxicity was observed in concentration higher than 100µM for all pillarplexes (see figures in the **Supplementary Material**). According to the determined IC50 values, the gold congeners are more active within the pairs of pillarplexes with the same anions (**3** vs. **5** and **4** vs. **6**).

TABLE 1 | Results from the disc diffusion assay for compounds 1–8 (10 mM) as well as reference substrates (30 mg/ml for tetracycline and gentamycine sulfate, and 10 mM for miconazole nitrate).


\*negative control, \*\*positive control.

However, they also show a higher base RFU compared to the silver compounds, indicating the gold compounds to be less toxic. With regard to the effect of the anions, the more water-soluble compounds (**2**, **4**, **6**: triflates or acetates) show higher activity than the less water-soluble hexafluorophosphate salts (**1**, **3**, **5**). In general, the same trend as in the antimicrobial assay are observed with the HepG2 cells. The silver pillarplexes appear to be more toxic and more active than their gold counterparts. Precipitation was observed in case of the silver pillarplexes as well as for AgNO<sup>3</sup> and AuCl3, whereas the gold pillarplexes did not exhibit any stability or solubility issues.

# Stability Tests

To evaluate possible reasons for the observations made during the bacterial and cell tests, we conducted an UV-Vis titration study. In detail, we checked the influence of a varying chloride and proton concentration on the stability or solubility of the pillarplex compounds. Therefore, we first evaluated the absorption properties of the two water-soluble pillarplex acetates **4** and **6** in aqueous solution, as well as the ligand precursor (**Figure 2A**). All compounds absorb in the UV range: the silver complex **4** shows an absorption maximum at 226 nm whereas the gold complex **6** absorbs at 245 nm. The ligand precursor absorbs at 209 nm. The molar extinction coefficients for the pillarplex compounds at the wavelengths of the individual maximal absorption are 9.33 · 10<sup>4</sup> ± 6.41 · 10<sup>2</sup> M−<sup>1</sup> cm−<sup>1</sup> (**4**) and 1.21 · 10<sup>5</sup> ± 4.64 · 10<sup>3</sup> M−<sup>1</sup> cm−<sup>1</sup> (**6**).

The titration results of the pillarplexes against an increasing amount of chloride ions present in aqueous solution show a very different behavior of the silver compared to the gold compound (**Figure 2B**). The absorption signal of silver complex **4** immediately drops up to addition of 0.5 mmol NaCl (which is about 17,000-fold excess of chloride). After that, no significant change can be observed in the absorption spectra upon addition of more equivalents of chloride. Apparently, this is close to the physiological concentration of chloride (0.9%) which might be a possible explanation why precipitation was observed in the biological tests for the silver containing pillarplexes. The gold compound **6** also shows a decay if the absorption signal upon chloride addition. However, the drop is less pronounced and at 0.9% chloride concentration, there is still a significant absorption (85% of the initial value). At higher chloride contents we observed a higher variation of the measured values, which we cannot explain up to now. However, even after addition of 1 mmol NaCl (about 35,000-fold excess) the characteristic absorption band at 245 nm can be observed for compound **6** (see **Supplementary Material**), strongly indicating that the gold complex is significantly less effected by chloride addition and still present in solution under physiological conditions.

**Figure 3** shows the pH-dependent decay of pillarplex compounds **4** and **6** over time. From our previous work on pillarplex rotaxanes we already knew, that in case of silver, the metal ions can be released quickly in the presence of an excess of the strong trifluoromethanesulfonic acid (Altmann and Pöthig, 2017). This was reproduced also in case of the empty pillarplex **4** (**Figure 3A**) for which an immediate drop of absorption signal at 226 nm was observed at pH 2, indicating very


\*Precipitation occurred, therefore IC<sup>50</sup> results are ambiguous and no SE could be estimated for these compounds (see *Supplementary Material*).

statistical uncertainties.

fast decomposition to the protonated imidazolium precursor. The resulting UV-Vis spectrum is also in agreement to that measured for the ligand precursor (**Figure 2A**). At higher pHvalues, the decomposition of **4** is significantly slower and almost identical for pH 4–6. A similar behavior was observed for the gold complex **6** although the decay at pH 2 is significantly slower than that of its silver analog (**Figure 3B**). Interestingly, at the higher pH-values the relative drop of the absorption signal is more pronounced compared to the silver complex. However, in case of **6** the resulting absorption spectrum after the assumed decomposition is not resembling that of the ligand precursor, and rather corresponds to the spectrum of **6** just with lower absorption intensity. Therefore, we additionally conducted a NMR experiment to check for protonation of the NHC ligands at pH 2. As a result, no protonated species was detected strongly indicating that the gold pillarplexes are stable even at low pH (see **Supplementary Information Figure S15**).

# CONCLUSION

In general, the silver pillarplexes behave like similar silver complexes and show antimicrobial and antifungal activity as well as moderate toxicity toward human HepG2 cells. The corresponding gold complexes were inactive against most bacterial strains and fungi, as well as had lower HepG2 toxicity. The observed effects originate most likely from the increased stability of the gold pillarplexes compared to the silver pillarplexes, as evident by the UV-Vis titration and the <sup>1</sup>H NMR experiment. The fact that the gold complexes seem comparably non-toxic and stable opens up the possibility of them being used as drug carriers for selective drug delivery or modified release of drugs that could fit inside the cavity in the pillarplexes.

# AUTHOR CONTRIBUTIONS

AP: project conception and supervision, manuscript composition, and writing. PA: synthesis and characterization of pillarplexes. SG: synthesis and characterization of pillarplexes, UV-Vis studies. JK: UV-Vis studies. OH: biological testing supervision, data analysis, manuscript writing. SA biological testing, data analysis. HWL: biological testing, supervision, data analysis. AS: biological testing, data analysis. TG: biological testing supervision, data analysis.

# FUNDING

This work was funded by FCI (AP), CSC (SG), TUM-GS (SG), and RCN (OH).

# ACKNOWLEDGMENTS

AP thanks the Fonds der chemischen Industrie (FCI) for funding of the project (Sachkostenzuschuss) as well as the Leonhard-Lorenz-Stiftung for financial support. SG thanks the CSC for a personal scholarship and the TUM Graduate School for financial support. OH thanks the Research Council of Norway (RCN) for a mobility grant (grant number: 240215).

# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fchem. 2018.00584/full#supplementary-material

Additional information on the IC<sup>50</sup> determinations as well as the UV-Vis experiments is available as supplementary material.

# REFERENCES


Zhang, C., Maddelein, M.-L., Wai-Yin Sun, R., Gornitzka, H., Cuvillier, O., and Hemmert, C. (2018). Pharmacomodulation on Gold-NHC complexes for anticancer applications - is lipophilicity the key point? Eur. J. Med. Chem. 157, 320–332. doi: 10.1016/j.ejmech.2018.07.070

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Pöthig, Ahmed, Winther-Larsen, Guan, Altmann, Kudermann, Santos Andresen, Gjøen and Høgmoen Åstrand. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Stepwise Synthesis of Tetra-imidazolium Macrocycles and Their N-Heterocyclic Carbene Metal Complexes

### Zili Li, Nuchareenat Wiratpruk and Peter J. Barnard\*

Department of Chemistry and Physics, La Trobe Institute for Molecular Science, La Trobe University, Melbourne, VIC, Australia

A modular stepwise synthetic method has been developed for the preperation of tetra-imidazolium macrocycles. Initially a series of three bis(imidazolylmethyl)benzene precursors were alkylated with 1,2-dibromoethane to produce the corresponding bis-bromoethylimidazolium bromide salts. In the second step the bis-bromoethylimidazolium bromide salts were reacted with selected bis(imidazolylmethyl)benzene molecules to produce a series of two symmetrical and three asymmetrical tetra-imidazolium macrocycles. These tetra-imidazolium salts act receptors for anions and <sup>1</sup>H-NMR titration studies were used to determine the association constants between two of the macrocycles and the halide anions chloride, bromide and iodide. The tetra-imidazolium salts are precursors for N-heterocyclic carbene (NHC) ligands and the corresponding silver(I), gold(I), and palladium(II) NHC complexes have been prepared. Varied structures were obtained, which depend on the chosen macrocyclic ligand and metal ion and in the case of the coinage metals Ag(I) and Au(I), mono, di, and hexanuclear complexes were formed.

Keywords: N-heterocycle carbene, macrocycle, anion receptor, N-heterocarbene-gold(I) complexes, tetraimidazolium

# INTRODUCTION

Imidazolium linked macrocycles have attracted significant recent attention because of their capacity to act as anion receptors and to function as pro-ligands for the synthesis of N-heterocyclic carbene metal complexes. Due to the great importance of negatively charged anionic species in biology, the preparation of receptor molecules designed to recognize and sense anions is an area of great research interest (Beer and Gale, 2001; Gale, 2003; Martínez-Máñez and Sancenón, 2003). Imidazolium groups are now well-recognized for their favorable features for the generation of anion receptors, which result both from electrostatic and hydrogen bonding interactions (Alcalde et al., 1999, 2007; Chellappan et al., 2005; Wong et al., 2005; Yoon et al., 2006; Xu et al., 2010). For example, a series of tetra-imidazolium linked macrocyclic compounds e.g., **I** (**Figure 1**) were shown to bind to biologically relevant anions such as chloride and hydrogen sulfate (Wong et al., 2005; Serpell et al., 2011). Additionally, compounds of this type have been used as selective luminescent sensors for nucleic acids and nucleotide derivatives such as DNA, RNA, ATP, and GTP and to sense and image RNA in the living cells, as a result of strong C-H<sup>+</sup> · · ·A <sup>−</sup> hydrogen bonding interations (Neelakandan and Ramaiah, 2008; Ahmed et al., 2011, 2012; Shirinfar et al., 2013).

#### Edited by:

Angela Casini, Cardiff University, United Kingdom

#### Reviewed by:

Alexander Pöthig, Technische Universität München, Germany Greta Bergamaschi, Istituto di Chimica del Riconoscimento Molecolare (ICRM), Italy

> \*Correspondence: Peter J. Barnard p.barnard@latrobe.edu.au

#### Specialty section:

This article was submitted to Supramolecular Chemistry, a section of the journal Frontiers in Chemistry

Received: 17 December 2018 Accepted: 02 April 2019 Published: 24 April 2019

#### Citation:

Li Z, Wiratpruk N and Barnard PJ (2019) Stepwise Synthesis of Tetra-imidazolium Macrocycles and Their N-Heterocyclic Carbene Metal Complexes. Front. Chem. 7:270. doi: 10.3389/fchem.2019.00270

A range of imidazolium linked macrocyclic compounds have been previously reported and both direct macrocyclization (Bass et al., 2010; Altmann et al., 2015; Toure et al., 2016) and stepwise macrocyclization (Mesquida et al., 2013) processes have been previously described for the synthesis of compounds of this type. In the direct macrocyclization approach, equal quantities of a bis-imidazole precursor, and a dihaloalkane are combined under high dilution conditions to produce the desired macrocyclic compound. For example, the imidazolium linked macrocycles **II** and **III** were prepared using the direct macrocyclization approach (Hahn et al., 2008; Schulte to Brinke et al., 2013). This approach can be limited due to unintended reactivity between the precursor compounds, which can produce complex reaction mixtures and poor yields of the desired product. Alternatively, stepwise macrocyclization can be used for the synthesis of polyimidazolium linked macrocyclic compounds (Schulte to Brinke and Hahn, 2015) and the synthesis of the tetra-imidazolium macrocycle **I** was achieved via sequential alkylation of bis-imidazolium precursors (Wong et al., 2005).

Polyimidazolium linked macrocycles have also been utilized as precursors for N-heterocyclic carbene metal complexes and previously the Au(III) complex **IV** (**Figure 2**) was prepared the reaction of KAuCl<sup>4</sup> with a macrocyclic tetra-imidazolium salt **II** (Mageed et al., 2017). Additionally, dinuclear Au(II) and mixed valence Au(I)/Au(III) complexes of have been prepared by the oxidation of dinuclear Au(I) complexes of macrocyclic NHC ligands (Mageed et al., 2018). In the past decade, a number of Ag(I) complexes of macrocyclic NHC ligands that display wide range of structures have been reported (Mckie et al., 2007; Hahn et al., 2008; Schulte to Brinke et al., 2013; Altmann et al., 2016; Fei et al., 2017; Lu et al., 2017, 2018). For example, a sandwich like tetranuclear Ag(I) complex featuring two tetra-NHC ligands was prepared **V** (**Figure 2**) and this complex was utilized as a precursor for the preparation of Au(I), Ni(II), Pd(II), and Pt(II) complexes (Altmann et al., 2016). In addition, macrocyclic tetra-NHC ligands have been shown to act as tetradentate ligand for square-planer metals such as Pd(II), Ni(II), Cu(I), and Pt(II) (Fei et al., 2017) and the interesting square-planar Pt(II) complex **VI** (**Figure 2**) was prepared by metal-template reaction from the tetrakis(trimethylphosphane)platinum(II) triflate and 2-azidophenylisocyanide (Hahn et al., 2005). The biological properties of metal complexes of NHC-based cyclophane and macrocyclic ligands have also been of significant recent interest and Ag(I) and Au(I) complexes have been shown to possess potent antimicrobial and anticancer activities, respectively (Aweda et al., 2013; Shah et al., 2013; Nomiya et al., 2018; Pöthig et al., 2018). Youngs et al. have been particularly active in this field (Hindi et al., 2009; Johnson et al., 2017) and have described the potent antimicrobrial properties of a Ag(I)-NHC complexes of cyclophane ligands (Melaiye et al., 2005). Additionally, a series of Au(I) complexes of related cyclophane-based NHC ligand systems were shown to be selectively toxic to cancer cells as a result of an antimitochondrial mechanism (Barnard et al., 2004, 2006). Meyer and more recently Kühn have also extended the application of macrocyclic tetra-NHC ligands to the synthesis of iron complexes, that provide fascinating models of reactive intermediates that are generated in the catalytic cycles of a range of heme and non-heme iron enzymes (Meyer et al., 2013; Anneser et al., 2015).

As imidazolium linked macrocycles offer the potential for the generations of novel sensors for biologically significant anions and as precursors for NHC metal complexes of metals that are well-known for their biological properties, we became interested in developing new strategies for the synthesis of compounds of this type. In the present paper, we report a novel modular stepwise synthetic approach for the synthesis of tetraimidazolium macrocycles. This approach involves the initial synthesis of bis-bromoethylimidazolium bromide precursors, which can then be utilized for the formation of either symmetrical or asymmetrical tetra-imidazolium macrocycles. These tetra-imidazolium salts bind anions in solution and association constants between two of the macrocycles and the halide anions chloride, bromide, and iodide were determined for two macrocycles. A range of silver(I), gold(I), and palladium(II) NHC complexes have been prepared from these pro-ligands.

# RESULTS AND DISCUSSION

# Synthesis of Tetra-imidazolium Macrocycles

The ortho-phenylene linked tetra-imidazolium macrocycle **9**·**Br<sup>4</sup>** was prepared via a stepwise macrocyclization procedure. Initially, 1,2-bis(imidazolylmethyl)benzene **1**(Baker et al., 2001) was alkylated with an excess of 1,2-dibromoethane (30 equivalents) to obtain the bis-bromoethylimidazolium bromide salt **5**·**Br<sup>2</sup>** in a moderate yield (**Scheme 1**). The <sup>1</sup>H-NMR spectrum of **5**·**Br<sup>2</sup>** shows the imidazolium C2-H proton resonates as a singlet signal at 9.50 ppm, while the ethylene group protons resonate as two upfield shifted triplet signals at 4.00 and 4.68 ppm. The macrocycle **9**·**Br4**, was prepared by heating an equimolar mixture of **5**·**Br<sup>2</sup>** and **1** in a solvent system consisting of a 1:4 mixture of DMF and acetonitrile under high dilution conditions (**Scheme 1**) and was isolated in a moderate yield of 29.7%. Compound **9**·**Br<sup>4</sup>** gave a simple <sup>1</sup>H-NMR spectrum consistent with its high symmetry (point group D2h) with downfield shifted signals for the imidazolium protons which resonated at 9.45 (C2- H) and 7.98 and 7.88 ppm (C4-H/C5-H). The protons of the ethylene linker groups resonated as a singlet signal at 4.81 ppm. A similar method was adopted for the synthesis of the metaphenylene linked macrocycle **10**·**Br<sup>4</sup>** from a mixture of **2** and **6**·**Br<sup>2</sup>** (**Scheme 1**).

With the successful synthesis of the symmetrical macrocycles **9**·**Br<sup>4</sup>** and **10**·**Br4**, we were interested in further exploring the versatility of our stepwise synthetic methodology with the synthesis of asymmetric tetra-imidazolium salts. In initial studies, reaction of either **3** with **5**·**Br<sup>2</sup>** or **1** with **7**·**Br<sup>2</sup>** (**Scheme 1**) under high dilution condition did not successfully produce the desired asymmetrical tetra-imidazolium salts. Additionally, attempts to use tetra-n-butylammonium bromide (Bu4N·Br) as a templation reagent in the reaction of **3** with **5**·**Br<sup>4</sup>** also did not give the desired product. By contrast, the reaction of compound **1** with

**7**·**Br<sup>2</sup>** under high dilution condition in presence of Bu4N·Br gave the desired product **11**·**Br<sup>4</sup>** in a low yield. The <sup>1</sup>H-NMR spectrum of **11**·**Br<sup>4</sup>** showed the imidazolium C2-H proton as singlet signal at 9.76 ppm and the methyl group on the imidazolium C2 carbon resonated as a singlet signal at 2.64 ppm. In addition, consistent with the lower symmetry structure (point group C2h) the benzylic proton resonates as two singlet signals at 5.47 and 5.52 ppm. The asymmetrical pro-ligands **12**·**Br<sup>4</sup>** and **13**·**Br<sup>4</sup>** were prepared in a similar manner from the reaction of either **2** and **8**·**Br<sup>2</sup>** or **2** and **5**·**Br2,** respectively, in presence of Bu4N·Br. Again, the <sup>1</sup>H-NMR spectrum for **13**·**Br<sup>4</sup>** is consistent with the lower symmetry structure, with the C2-H protons resonating as two singlet signals at 8.99 and 9.13 ppm.

# Synthesis of Ag(I), Au(I), and Pd(II) Complexes

The Ag(I) complex **14**·**(PF6)<sup>6</sup>** was prepared by the reaction of pro-ligand **9**·**Br<sup>4</sup>** with Ag2O in DMF with the exclusion of light (**Scheme 2**) and product was isolated as a white crystalline solid in a yield of 21%. The <sup>1</sup>H-NMR spectrum of **14**·**(PF6)<sup>6</sup>** showed no imidazolium C2-H proton signal, indicating that the C2 carbon is deprotonated and coordinated to the metal center as a carbene. The ortho-substituted phenyl-linker group protons resonate as two set of doublets (5.68 and 6.94 ppm) and triplet (7.17 and 7.52 ppm) signals, consistent with a more complex magnetic environment for the Ag(I) complex when compared to the pro-ligand. The <sup>13</sup>C-NMR spectrum for **14**·**(PF6)<sup>6</sup>** revealed a downfield shifted signal at 182 ppm, for which <sup>107</sup>Ag-13C (d, 1 J = 182.32 Hz) and <sup>109</sup>Ag-13C (d, <sup>1</sup> J = 209.99 Hz) couplings were observed, which is also consistent with coordination of the C2 carbon to Ag(I). The high-resolution mass spectrum for **14**·**(PF6)<sup>6</sup>** produced a series of peaks consistent with a hexanuclear structure with the general formula [Ag6L3] <sup>6</sup><sup>+</sup> (where L is the macrocyclic tetra-carbene ligand). For example, a peak was observed at m/z = 372.0334, which corresponds to the formula [C96H96N24Ag6] <sup>6</sup><sup>+</sup> (calculated = 372.0419).

Using a similar approach, the Ag(I) complex **17**·**(PF6)<sup>3</sup>** was prepared from the pro-ligand **10**·**Br4**, however for this compound the <sup>1</sup>H-NMR spectrum showed a signal at 8.37 ppm, consistent with the C2-H proton being present in the complex. However, the <sup>13</sup>C-NMR spectrum of **17**·**(PF6)<sup>3</sup>** showed a downfield shifted signal at 180.4 ppm that displays <sup>107</sup>Ag-<sup>13</sup>C (d, <sup>1</sup> J = 183.58 Hz) and <sup>109</sup>Ag-13C (d, <sup>1</sup> J = 211.24 Hz) couplings, indicating coordination of the imidazole C2 carbon to Ag(I). These NMR results are consistent with a mononuclear complex, where two of the imidazole units are coordinated to the metal, while the other two remain uncoordinated imidazolium units (**Scheme 3**). To further investigate this result, a different synthetic method was undertaken where the proligand **10**·**(PF6)<sup>4</sup>** was reacted with AgNO<sup>3</sup> in present of NH4OH, however the same mononuclear Ag(I) complex was obtained (**Scheme 3**).

The hexanuclear Au(I) complex **15**·**Br<sup>6</sup>** was prepared by the reaction of **9**·**Br<sup>4</sup>** with (THT)AuCl in present of the mild base sodium acetate (**Scheme 2**) and the complex was obtained as an off-white solid in 54.2% yield. The same approach was used in an attempt to prepare the Au(I) complex of proligand **10**·**Br4**, however no complex could be isolated from the reaction mixture. In the next set of reactions, the pro-ligand **11**·**Br<sup>4</sup>** (with both normal and C2-blocked imidazolium groups) was reacted with (THT)AuCl in presence of sodium acetate. It was anticipated that this ligand might produce a complex displaying both "normal" and "abnormal" NHC coordination modes, however the dinuclear Au(I) complex **18**·**(PF6)<sup>2</sup>** was obtained, which displayed only the "normal" NHC coordination mode (**Scheme 4**). Using the same method, a dinuclear Au(I) complex derived from **19**·**(PF6)<sup>4</sup>** was also prepared.

Due to their potential to act as tetradentate ligands with metals that adopt square-planar coordination geometries, the Pd(II) complex of pro-ligand **9**·**Br<sup>4</sup>** was prepared. A range of approached have been previously employed for the preparation of Pd(II)-NHC complexes, including in situ deprotonation and metallation (Baker et al., 2001; Fei et al., 2017) and transmetallation via an intermediate Ag(I) complex (Schulte to Brinke et al., 2013; Andrew et al., 2016). In this work the former in situ deprotonation and metallation approach was initially investigated by reacting **9**·**(PF6)<sup>4</sup>** with Pd(OAc)<sup>2</sup> in DMSO, however the desired Pd(II) complex could not be isolated. As described previously, a hexanuclear Ag(I) complex could be prepared from the pro-ligand **9**·**(PF6)<sup>4</sup>** and in a second attempt to synthesize the Pd(II) complex, pro-ligand **9**·**Br<sup>4</sup>** was first reacted with Ag2O to form the Ag(I) complex in situ followed by addition of K2PdCl<sup>4</sup> (**Scheme 5**). <sup>1</sup>H-NMR analysis of Pd(II) complex **16**·**(PF6)<sup>2</sup>** showed a relatively simple spectrum with the C4/5 protons of the NHC groups resonating as two sets of doublets at 7.48 and 7.83 ppm. The benzylic protons resonate as two sets of doublet signals at 5.20 and 6.44 ppm (AX pattern), which is consistent with a rigid molecular structure in solution. Furthermore, the <sup>13</sup>C-NMR spectrum showed a downfield shifted signal at 167.41 ppm, which corresponds to the NHC carbene carbon coordinated to the Pd(II) metal center.

The Ag(I) transmetallation approach did not successfully produced a Pd(II) complex of pro-ligand **11**·**Br4**. In this attempt, Ag2CO<sup>3</sup> was used instead of Ag2O to avoid the undesirable oxidative cleavage of the C2-blocking methyl group which has been observed previously (Chianese et al., 2004). In a second synthetic attempt, the pro-ligand **11**·**Br<sup>4</sup>** was reacted with K2PdCl<sup>4</sup> in presence of NaOAc. The <sup>1</sup>H-NMR of the crude product showed no peak for the C2-H proton for the "unblocked" imidazole units indicating that these groups are coordinated to the metal center. The C4/5 protons of the C2 "blocked" imidazolium group at 7.34 and 7.66 ppm suggesting

SCHEME 3 | Synthesis of Ag(I) complex 17·(PF<sup>6</sup> )3 from the pro-ligand 10·Br<sup>4</sup> . (a) Ag2O (4.0 eq.), DMF, 70◦C, 3 d. (b) AgNO<sup>3</sup> (4.0 eq.), NH4OH (eq.), CH3CN, RT, overnight.

that C2 blocked imidazolium unit did not react with the metal. Furthermore, the benzylic proton signals at 5.24 and 5.26 ppm and the ethylene linker signal at 4.61–4.70 ppm also indicated an asymmetrical structure of the title complex. The crystal structure in **Figure 5** also suggested the Pd(II) complex **20**·**(PF6)<sup>2</sup>** was successfully synthesized. Unfortunately, it was unable to separate this complex as pure product.

# Structural Studies

Compounds **5**·**Br2**, **9**·**Br4**, **10**·**Br3PF6**, **11**·**Br4**, **15**·**Br2(PF6)4,** and **20**·**(PF6)<sup>2</sup>** were characterized by X-ray crystallography. A representation of the precursor compound **5**·**Br<sup>2</sup>** is shown in **Figure S1** while representations of the tetra-imidazolium macrocycles **9**·**Br4**, **10**·**Br3PF6,** and **11**·**Br<sup>4</sup>** are shown in **Figure 3**. In all cases these imidazolium salts display hydrogen bonding interactions between various hydrogens on the cationic imidazolium units and the bromide counter anions. For example, the shortest C2· · · Br distances for **5**·**Br2**, **9**·**Br4**, **10**·**Br3PF6,** and **11**·**Br4,** respectively, are: 3.4496(3); 3.42219(5) Å; 3.4822(4) Å; and 3.6552 (5) Å, respectively, which fall within a typical range of C-H· · · Br distance caused by hydrogen bonding (Yuan et al., 2002).

Two representations of the X-ray crystal structure for the Au(I) complex **15**·**Br2(PF6)<sup>4</sup>** are shown in **Figure 4**. This remarkable structure reveals that the Au(I) complex adopts a cyclic hexanuclear supramolecular assembly with the overall formula [Au6L3] <sup>6</sup>+. In the cation, each of the Au(I) centers adopt linear two-coordinate geometries, and they can be divided into two distinct groups of three atoms each. For the first group, each Au(I) atoms (Au1, Au4, and Au5) is bound to two NHC units on opposite sides of each of the three ligand molecules. While each of the Au(I) atoms in the second group (Au2, Au3, and Au6) are coordinated to NHC donors from adjacent ligand molecules, resulting in the formation of a metallo-macrocycle. In each case, the ligand molecules are bowl shaped to accommodate the two Au(I) coordination modes. Interestingly, the metallo-macrocycle is arrayed around an encapsulated central bromide counterion. This bromide ion displays interactions with Au(I) atoms Au1, Au4, and Au5 with the distances being 3.19488(5), 3.24731(5), and 3.23064(4) Å, respectively. Previously, a hexanuclear Ag(I) complex was reported for a tetra-carbene ligand linked by aliphatic butyl chains (Fei et al., 2017).

A representation of the X-ray crystal structure for the Pd(II) **20**·**(PF6)<sup>2</sup>** is shown in **Figure 5**. The molecular structure shows that the ligand is coordinated to the metal center through the "normal" NHC groups, with the C2 "blocked" groups present as cationic imidazolium units. The Pd(II) center is fourcoordinate, with the two-remaining sited being occupied by bromide ions.

# Anion Binding Studies

Due to the tetracationic charge of the imidazolium linked macrocycles prepared in this work, the propensity of **9**·**(PF6)<sup>4</sup>** and **10**·**(PF6)<sup>4</sup>** to bind to the halide anions F<sup>−</sup> Cl−, Br−, and I <sup>−</sup> (as their tetra-n-butylammonium halide salts) was evaluated using <sup>1</sup>H-NMR titration experiments. In initial studies, it was

found that the addition of Bu4N·F to **10**·**(PF6)<sup>4</sup>** in d6-DMSO caused an immediate color change to pale yellow. The <sup>1</sup>H-NMR analysis showed that the color change occurred concurrently with a significant broadening and downfield shift of the imidazolium C2-H signal. In addition, the appearance of new unidentified <sup>1</sup>H-NMR signals were observed which were consistent with decomposition of the macrocyclic receptor. As such this anion was not studied further.

included for clarity. Atomic displacement ellipsoids are shown at the 50% probability level.

Addition of increasing equivalents of Bu4N·Cl (0.25– 14.0 eq.) to a solution of **9**·**(PF6)<sup>4</sup>** in d6-DMSO caused a significant downfield shift of the resonance corresponding to the imidazolium C2-H signal from 8.96 to 9.64 ppm. In addition, the benzylic proton signal was also shifted downfield from 5.34 to 5.76 ppm. Unfortunately, this study was hampered by the gradual precipitation of the imidazolium salt at higher Cl<sup>−</sup> concentrations. In a similar manner, addition of increasing

equivalents of Bu4N·Br and Bu4N·I to **9**·**(PF6)<sup>4</sup>** caused downfield shifts in the imidazolium C2-H signal, although to a lesser extent than that seen for Bu4N·Cl, however for these anions no precipitation of the macrocycle was seen. **Figure 6** shows the <sup>1</sup>H-NMR titration between **9**·**(PF6)<sup>4</sup>** and Bu4N·Br, while **Figure 7** shows the change in the imidazolium C2-H chemical shift for **9**·**(PF6)<sup>4</sup>** in response to increasing equivalents of the added halide anion (solid black lines). Similar results were seen for the macrocyclic receptor **10**·**(PF6)<sup>4</sup>** and the plot of experimental titration data is shown in **Figure S2**. Job plot analysis was then used to evaluate the stoichiometry of the interactions between the macrocyclic receptors and the anions Cl−, Br−, and I−. The results of these studies **Figures S3**, **S4** showed that maxima occurred at close to χ = 0.5 indicating 1:1 stoichiometry for both receptor **9** <sup>4</sup><sup>+</sup> and **10**4<sup>+</sup> with these anions.

Binding (association) constants K<sup>a</sup> (M−<sup>1</sup> ) for the tetraimidazolium macrocycles **9**·**(PF6)<sup>4</sup>** and **10**·**(PF6)<sup>4</sup>** were determined by analysis of the <sup>1</sup>H-NMR titration data using the computer program HypNMR 2018 (Frassineti et al., 1995, 2003). The fitted binding isotherms are shown in **Figure 6** and **Figure S2** (red dashed lines) and the calculated binding constants are given in **Table 1**. The values obtained for the compounds studied in this work are similar in magnitude to those reported previously for related compounds. For example, Beer et al. reported association constants of 420(23), 241(3), and 120(1) M−<sup>1</sup> for a tetra-imidazolium macrocycle in d6-DMSO solution with the anions Cl−, Br<sup>−</sup> and I−, respectively (Serpell et al., 2011).

# CONCLUSION

In conclusion, a novel stepwise synthetic strategy is reported that allows for the synthesis of both symmetrical and asymmetrical tetra-imidazolium linked macrocyclic compounds. The synthetic strategy is modular as initially a range of bis-bromoethylimidazolium bromide precursors were synthesized, which when combined with chosen bis(imidazolylmethyl)benzene molecules produced a range of tetra-imidazolium linked macrocycles. These tetra-imidazolium salts are of significant interest as they bind anions in solution and offer the potential for the development of sensors for biologically relevant anions. Using <sup>1</sup>H-NMR titration studies, the association constants between **9**·**(PF6)<sup>4</sup>** and **10**·**(PF6)<sup>4</sup>** and the halide anions Cl−, Br−, and I<sup>−</sup> were determined and these ranged between 501 and 20 M−<sup>1</sup> . Imidazolium salts are precursors for N-heterocyclic carbene (NHC) ligands and given that Ag(I) and Au(I) complexes of NHC ligands have been shown to display antimicrobial (Aweda et al., 2013; Shah et al., 2013) and anticancer(Barnard et al., 2004) properties, respectively, the Ag(I), Au(I), and Pd(II) NHC complexes derived from these tetra-imidazolium linked macrocycles were prepared. The Ag(I) complexes **14**6<sup>+</sup> and the analogous Au(I) complex **15**6<sup>+</sup> adopted intriguing hexanuclear structures with the general formula [M6L3] <sup>6</sup>+. Currently we are exploring the development of silver complexes of NHC ligands as potential antimicrobial agents and supramolecular assemblies such as complex **14**6<sup>+</sup> are a particular focus for the slow release of silver ions and the results of these studies will be reported in due course.

# EXPERIMENTAL

# General Details

All solvents and chemicals were purchased from Sigma-Aldrich, Chem Supply, Alfa Aesar, and were used as received unless otherwise stated. Where necessary, solvents were further purified using an Innovative Technology Pure Solv solvent purification system. All experiments were performed under an atmosphere of N<sup>2</sup> unless otherwise stated. <sup>1</sup>H-NMR and <sup>13</sup>C-NMR spectra were recorded on Bruker Advance 500 (500.023 MHz for <sup>1</sup>H and 125.74 MHz for <sup>13</sup>C) and spectra were referenced to solvent resonances. Where required, COZY, HSQC, HMBC, and NOESY 2-dimensional experiments were used to assist assignments. Mass spectra were obtained using an Agilent 6530 Q-TOF LC/MS mass spectrometer fitted with an Agilent electrospray ion (ESI) source.

# X-ray Crystallography

Single crystals suitable for X-ray diffraction studies were grown as follows: **5**·**Br<sup>2</sup>** and **9**·**Br<sup>4</sup>** diffusion of diethyl ether into an methanol solution of the title compound; **10**·**Br3PF<sup>6</sup>** diffusion of diethyl ether into an acetonitrile solution of the title compound containing two drops of a solution of tetra-n-butylammonium bromide in acetonitrile; **11**·**Br<sup>4</sup>** slow evaporation of a methanol solution of the title compound; **15**·**Br2(PF6)<sup>4</sup>** diffusion of ethyl acetate into an acetonitrile solution of the title compound; **20**·**(PF6)**<sup>2</sup> was grown by slow diffusion of diethyl ether into an acetonitrile solution of the titled compound. Crystallographic data for all structures determined are given in **Table S1**. For all samples, crystals were removed from the crystallization vial and immediately coated with paratone oil on a glass slide. A suitable crystal was mounted in Paratone oil on a glass fiber

and cooled rapidly to 173 K in a stream of cold N<sup>2</sup> using an Oxford low temperature device. Diffraction data were measured using an Rigaku Oxford Diffraction SuperNova X-ray Diffraction System mounted with Mo-Kα λ = 0.71073 Å and Cu-Kα λ = 1.54184. Data were reduced and corrected for absorption using the CrysAlis Pro program. The SHELXL2013-2 program was used to solve the structures with Direct Methods, with refinement by the Full-Matrix Least-Squares refinement techniques on F2. The non-hydrogen atoms were refined anisotropically and


<sup>a</sup>Precipitate forms after four equivalents of Bu4N·Cl added; <sup>b</sup>precipitate forms rapidly and measurement of association constant not possible.

hydrogen atoms were placed geometrically and refined using the riding model. Coordinates and anisotropic thermal parameters of all non-hydrogen atoms were refined. All calculations were carried out using the program Olex2. Further XRD details are provided in the **Supporting Information**. CCDC 1885400-1885405 contains the supplementary crystallographic data for this paper. These data can be obtained free of charge from The Cambridge Crystallographic Data Centre via www.ccdc.cam.ac.uk/data\_request/cif.

# <sup>1</sup>H-NMR Titration Studies

A solution of the title compound (10 mg) in d6-DMSO (600 µL) and a 1.5 M solution of Bu4N·X (X = Cl, Br and I) in d6-DMSO were prepared, respectively. To the solution of title compound, increasing equivalents (0.25–14.0 eq.) of 1.5 M Bu4N·X solution was added and the resultant solution was thoroughly mixed. The <sup>1</sup>H-NMR spectrum was recorded ∼2 min after each addition at 302 K.

# Jobs Plot Analysis

A solution of the title compound (10 mg/mL) in d6-DMSO and a 0.050 M solution of Bu4N·X (X=Cl, Br and I) in d6-DMSO were prepared, respectively. A varied fraction of title compound solution and Bu4N·X solution was added and diluted with d6-DMSO to 600 µL to maintain the total concentration of substance at 10 mM. The resultant solution was thoroughly mixed and <sup>1</sup>H-NMR spectrum was recorded at 302 K.

# Synthesis

**4**. Sodium hydride (1.14 g, 47.35 mmol) was added to a solution of 2-methylimidazole (3.11 g, 37.88 mmol) in DMF (50 mL) cooled to 0◦C and the resultant mixture was stirred at RT for 1 h and α,α ′ -dibromo-m-xylene (5.00 g, 18.94 mmol) was then added. Stirring was continued at RT for 12 h and the mixture was then diluted with water (100 mL). The mixture was then extracted with CH2Cl<sup>2</sup> (5 × 10 mL) and the combined organic extracts were washed with water (5 × 50 mL) and then brine (20 mL). The organic layer was dried with MgSO<sup>4</sup> and the solvent was evaporated in vacuo yielding a yellow oil. Yield: 2.18 g, 43.2%. <sup>1</sup>H-NMR (500.02 MHz, d6-DMSO): δ = 7.33 (t, <sup>3</sup> JH−<sup>H</sup> = 7.5 Hz, 1H, ArH), 7.08 (d, <sup>3</sup> JH−<sup>H</sup> = 1.0 Hz, 2H, Himi), 7.02 (d, <sup>3</sup> JH−<sup>H</sup> = 8.0 Hz, 2H, ArH), 6.96 (s, 1H, ArH), 6.76 (d, <sup>3</sup> JH−<sup>H</sup> = 1.5 Hz, 2H, Himi), 5.12 (s, 4H, CH2), 2.19 (s, 6H, CH3). <sup>13</sup>C-NMR (125.74 MHz, d6-DMSO): δ = 144.36 (Cq), 138.55 (Cq), 129.64 (CAr), 126.96 (Cimi), 126.59 (CAr), 126.07 (CAr), 120.70 (Cimi), 48.94 (CH2), 13.18 (CH3). HRESI-MS<sup>+</sup> (CH3CN): C14H15N + <sup>4</sup> m/z = 267.1543, calcd = 267.1604.

**5**·**Br2**. To a solution of 1,2-dibromoethane (21.69 mL, 251.7 mmol) in CH3CN (50 mL) stirred at 110◦C, was added dropwise a solution of **1** (2.0 g, 8.39 mmol) in CH3CN (100 mL) over a period of 3 h. The resultant mixture was stirred at the same temperature for 12 h and then filtered whilst still hot. The filtrate was then evaporated in vacuo and the resulting solid was recrystallized from hot ethanol (20 mL) yielding a white crystalline solid. Yield: 2.19 g, 41.2%. <sup>1</sup>H-NMR (500.02 MHz, d6-DMSO): δ = 9.50 (s, 2H, Himi), 7.98 (t, <sup>3</sup> JH−<sup>H</sup> = 1.8 Hz, 2H, Himi), 7.88 (t, <sup>3</sup> JH−<sup>H</sup> = 1.8 Hz, 2H, Himi), 7.49 (dd, <sup>3</sup> JH−<sup>H</sup> = 5.8 Hz, <sup>4</sup> JH−<sup>H</sup> = 3.5 Hz, 2H, ArH), 7.30 (dd, <sup>3</sup> JH−<sup>H</sup> = 5.5 Hz, <sup>4</sup> JH−<sup>H</sup> = 3.5 Hz, 2H, ArH), 5.75 (s, 4H, CH2), 4.69 (t, <sup>3</sup> JH−<sup>H</sup> = 6.0 Hz, 4H, CH2), 4.01 (t, 3 JH−<sup>H</sup> = 6.0 Hz, 4H, CH2). <sup>13</sup>C-NMR (125.74 MHz, d6- DMSO): δ = 137.56 (Cimi), 133.38 (Cq), 130.10 (CAr), 129.83 (CAr), 123.42 (Cimi), 123.38 (Cimi), 50.76 (CH2), 49.61 (CH2), 32.90 (CH2). HRESI-MS<sup>+</sup> (CH3OH): C18H22N4Br2<sup>+</sup> <sup>2</sup> m/z = 227.0094, calcd <sup>=</sup> 227.0090, C18H22N4Br<sup>+</sup> <sup>3</sup> m/z = 532.9367, calcd = 532.9369.

**6**·**Br2**. This compound was prepared using the same method as described for **5**·**Br<sup>2</sup>** from 1,2-dibromoethane (10.80 mL, 125.90 mmol) and **2** (1.00 g, 4.20 mmol). The crude product was purified by a trituration with diethyl ether (3 × 10 mL) yielding a light brown oil. Yield: 0.62 g, 24.0%. <sup>1</sup>H-NMR (500.02 MHz, d6-DMSO): δ = 9.63 (s, 2H, Himi), 7.95 – 7.97 (m, 4H, Himi), 7.65 (s, 1H, ArH), 7.44 – 7.50 (m, 3H, ArH), 5.56 (s, 4H, CH2), 4.69 (t, <sup>3</sup> JH−<sup>H</sup> = 5.9 Hz, 4H, CH2), 4.01 (t, 3 JH−<sup>H</sup> = 5.9 Hz, 4H, CH2). <sup>13</sup>C-NMR (125.74 MHz, d6-DMSO): δ = 137.31 (Cimi), 136.02 (Cq), 130.19 (CAr), 129.08 (CAr), 128.90 (CAr), 123.36 (Cimi), 123.16 (Cimi), 52.05 (CH2), 50.76 (CH2), 32.90 (CH2). HRESI-MS<sup>+</sup> (CH3OH): C18H22N4Br2<sup>+</sup> 2 m/z = 227.0102, calcd = 227.0090.

**7**·**Br2**. This compound was using the same method as described for **5**·**Br<sup>2</sup>** from 1,2-dibromoethane (4.85 mL, 56.31 mmol) and **3** (0.50 g, 1.88 mmol). The crude product was recrystallized from ethanol yielding a white solid. Yield: 0.62 g, 51.6%. <sup>1</sup>H-NMR (500.02 MHz, d6-DMSO): δ =7.90 (d, <sup>3</sup> JH−<sup>H</sup> = 2.1 Hz, 2H, Himi), 7.68 (d, <sup>3</sup> JH−<sup>H</sup> = 2.1 Hz, 2H, Himi), 7.40 (dd, <sup>3</sup> JH−<sup>H</sup> = 5.5 Hz, <sup>4</sup> JH−<sup>H</sup> = 3.4 Hz, 2H, ArH), 6.82 (dd, 3 JH−<sup>H</sup> = 5.7 Hz, <sup>4</sup> JH−<sup>H</sup> = 3.5 Hz, 2H, ArH), 5.67 (s, 4H, CH2), 4.68 (t, <sup>3</sup> JH−<sup>H</sup> = 5.9 Hz, 4H, CH2), 4.00 (t, <sup>3</sup> JH−<sup>H</sup> = 5.9 Hz, 4H, CH2), 2.69 (s, 6H, CH3). <sup>13</sup>C-NMR (125.74 MHz, d6- DMSO): δ = 146.15 (Cq), 132.62 (Cq), 129.28 (CAr), 126.95 (CAr), 122.65 (Cimi), 122.43 (Cimi), 49.29 (CH2), 48.85 (CH2), 31.96 (CH2), 10.70 (CH3). HRESI-MS<sup>+</sup> (CH3OH): C20H26N4Br2<sup>+</sup> 2 m/z = 241.0306, calcd = 241.0246.

**8**·**Br2**. This compound was prepared using the same method as described for **5**·**Br<sup>2</sup>** from 1,2-dibromoethane (8.15 mL, 94.61 mmol) and **4** (0.84 g, 3.15 mmol). The crude product was recrystallized from ethanol yielding a white solid. Yield: 0.76 g, 37.5%. <sup>1</sup>H-NMR (500.02 MHz, d6-DMSO): δ = 7.81 (s, 2H, Himi), 7.79 (s, 2H, Himi), 7.46 (t, <sup>3</sup> JH−<sup>H</sup> = 7.8 Hz, 1H, ArH), 7.33 (s, 1H, ArH), 7.27 (d, <sup>3</sup> JH−<sup>H</sup> = 7.7 Hz, 2H, ArH), 5.47 (s, 4H, CH2), 4.62 (t, <sup>3</sup> JH−<sup>H</sup> = 6.0 Hz, 4H, CH2), 3.93 (t, <sup>3</sup> JH−<sup>H</sup>

= 5.9 Hz, 4H, CH2), 2.67 (s, 6H, CH3). <sup>13</sup>C-NMR (125.74 MHz, d6-DMSO): δ = 145.47 (Cq), 135.76 (Cq), 130.32 (CAr), 128.15 (CAr), 127.59 (CAr), 122.40 (Cimi), 122.35 (Cimi), 50.88 (CH2), 49.17 (CH2), 31.78 (CH2), 10.37 (CH3). HRESI-MS<sup>+</sup> (CH3OH): C20H26N4Br2<sup>+</sup> <sup>2</sup> m/z = 241.0294, calcd = 241.0246, HRESI-MS<sup>+</sup> (CH3OH): C20H26N4Br<sup>+</sup> <sup>3</sup> m/z = 560.9702, calcd = 560.9682.

**9**·**Br4**. Solutions of **1** (0.31 g, 1.30 mmol) in CH3CN (100 mL) and **5**·**Br<sup>2</sup>** (0.80 g, 1.30 mmol) in DMF (100 mL) were added simultaneously dropwise to 150 mL of CH3CN heated at 110◦C over a period of 3 h. The mixture was then stirred at the same temperature for a further 5 d during which time a precipitate formed. The precipitate was collected and washed with CH3CN (3 × 5 mL) and recrystallized from a mixture of methanol and isopropanol yielding a white crystalline solid. Yield: 0.33 g, 29.7%. <sup>1</sup>H-NMR (500.02 MHz, d6-DMSO): δ = 9.45 (s, 4H, Himi), 7.90 (t, <sup>3</sup> JH−<sup>H</sup> = 1.5 Hz, 4H, Himi), 7.77 (t, <sup>3</sup> JH−<sup>H</sup> = 1.5 Hz, 4H, Himi), 7.52 (dd, <sup>3</sup> JH−<sup>H</sup> = 5.8 Hz, <sup>4</sup> JH−<sup>H</sup> = 3.0 Hz, 4H, ArH), 7.40 (dd, <sup>3</sup> JH−<sup>H</sup> = 5.5 Hz, <sup>4</sup> JH−<sup>H</sup> = 3.5 Hz, 4H, ArH), 5.68 (s, 8H, CH2), 4.81 (s, 8H, CH2). <sup>13</sup>C-NMR (125.74 MHz, d6-DMSO): δ = 136.75 (Cimi), 133.13 (Cq), 131.04 (CAr), 130.38 (CAr), 123.66 (Cimi), 123.51(Cimi), 49.95 (CH2). HRESI-MS<sup>+</sup> (CH3CN): C32H36N 4+ <sup>8</sup> m/z = 133.0655, calcd = 133.0760, C32H36N8P3F + 18 m/z = 967.2020, calcd = 967.1967.

**10**·**Br4**. This compound was prepared using the same method as described for **9**·**Br4**. from **2** (0.24 g, 1.01 mmol) and **6**·**Br<sup>2</sup>** (0.62 g, 1.01 mmol). Yield: 0.16 g, 18.0%. <sup>1</sup>H-NMR (500.02 MHz, d6-DMSO): δ = 9.53 (s, 4H, Himi) 7.83 (t, 3 JH−<sup>H</sup> = 1.7 Hz, 4H, Himi), 7.80 (t, <sup>3</sup> JH−<sup>H</sup> = 1.7 Hz, 4H, Himi), 7.51 (s, 2H, ArH), 7.36 (t, <sup>3</sup> JH−<sup>H</sup> = 7.7 Hz, 4H, ArH), 7.26–7.28 (m, 4H, ArH), 5.45 (s, 8H, CH2), 4.84 (s, 8H, CH2). <sup>13</sup>C-NMR (125.74 MHz, d6-DMSO): δ = 137.50 (Cimi), 135.80 (Cq), 130.70 (CAr), 128.92 (CAr), 128.87 (CAr), 128.39 (Cimi), 52.21 (CH2), 48.83 (CH2). HRESI-MS<sup>+</sup> (CH3CN): C32H36N 4+ <sup>8</sup> m/z = 133.0686, C32H36N 4+ 8 calcd = 133.0760, C32H36N8P2F 2+ <sup>12</sup> m/z = 411.1099, C32H36N8P2F 2+ <sup>12</sup> calcd = 411.1163, C32H36N8P3F + <sup>18</sup> m/z = 967.1990, C32H36N8P3F + 18 calcd = 967.1967.

**11**·**Br4**. Solutions of **1** (0.19 g, 0.80 mmol) in CH3CN (50 mL) and **7**·**Br<sup>2</sup>** (0.50 g, 0.80 mmol) in DMF (50 mL) were added simultaneously dropwise to a solution of Bu4N·Br (1.55 g, 4.80 mmol) in CH3CN (150 mL) heated at 110◦C over a period of 3 h. The mixture was stirred at the same temperature for a further 5 d during which time a precipitate formed. The precipitate was collected and washed with CH3CN (3 × 5 mL) and then recrystallized from a mixture of methanol and diethyl ether yielding a white crystalline solid. Yield: 0.10g, 7.5%. <sup>1</sup>H-NMR (500.02 MHz, d6-DMSO): δ = 9.76 (s, 2H, Himi), 8.00 (s, 2H, Himi), 7.65 (s, 2H, Himi), 7.49–7.53 (m, 10H, ArH, Himi), 7.23 (dd, <sup>3</sup> JH−<sup>H</sup> = 5.4 Hz, <sup>4</sup> JH−<sup>H</sup> = 3.6 Hz, 4H, ArH), 5.52 (s, 4H, CH2), 5.47 (s, 4H, CH2), 4.81 – 4.83 (m, 4H, CH2), 4.70 – 4.72 (m, 4H, CH2), 2.64 (s, 6H, CH3). <sup>13</sup>C-NMR (125.74 MHz, d6-DMSO): δ = 145.98 (Cq), 137.44 (Cimi), 132.92 (Cq), 132.34 (Cq), 131.15 (CAr), 130.47 (CAr), 129.97 (CAr), 129.56 (CAr), 123.98 (Cimi), 123.42 (Cimi), 122.83 (Cimi), 122.67 (Cimi), 49.48 (CH2), 49.38 (CH2), 49.19 (CH2), 47.90 (CH2), 10.02 (CH3). HRESI-MS<sup>+</sup> (CH3CN): C32H36N8P2F 2+ <sup>12</sup> m/z = 425.1335, calcd = 425.1324, C34H40N8P3F + <sup>18</sup> m/z = 995.2323, calcd 995.2296.

**12**·**Br4**. This compound was prepared using the same method described for **11**·**Br<sup>4</sup>** from **2** (0.11g, 0.47 mmol) and **8**·**Br<sup>2</sup>** (0.30 g, 0.47 mmol). The crude product was recrystallized from methanol yielding a white crystalline solid. Yield: 0.068 g, 16.5%. <sup>1</sup>H-NMR (500.023 MHz, d6-DMSO): δ = 9.53 (s, 2H, Himi), 7.80-7.84 (m, 4H, Himi), 7.62 (d, <sup>3</sup> JH−<sup>H</sup> = 2.2 Hz, 2H, Himi), 7.48 (d, <sup>3</sup> JH−<sup>H</sup> = 2.2 Hz, 2H, Himi), 7.46 (s, 1H, ArH), 7.40 (t, <sup>3</sup> JH−<sup>H</sup> = 7.7 Hz, 1H, ArH), 7.34 (s, 1H, ArH), 7.24 – 7.30 (m, 3H, ArH), 7.12 (dd, <sup>3</sup> JH−<sup>H</sup> = 7.7 Hz, <sup>4</sup> JH−<sup>H</sup> = 1.3 Hz, 2H, ArH), 5.42 (s, 4H, CH2), 5.38 (s, 4H, CH2), 4.74 (s, 8H, CH2), 2.63 (s, 6H, CH3). <sup>13</sup>C-NMR (125.74 MHz, d6-DMSO): δ = 145.85 (Cq), 137.61 (Cimi), 135.54 (Cq), 135.39 (Cq), 128.47 (CAr), 128.04 (CAr), 123.50 (Cimi), 123.41 (Cimi), 122.30 (Cimi), 122.23 (Cimi), 52.19 (CH2), 50.92 (CH2), 48.33 (CH2), 47.73 (CH2), 10.49 (CH3). HRESI-MS<sup>+</sup> (CH3CN): C32H36N8Br2<sup>+</sup> 2 m/z = 360.0861, calcd = 360.0865.

**13**·**(PF6)<sup>4</sup>** This compound was prepared using the same method described for **11**·**Br<sup>4</sup>** from **2** and **5**·**Br<sup>2</sup>** (0.52 g, 0.85 mmol) and Bu4N·Br (1.37 g, 4.25 mmol). The crude product was then dried in vacuo and re-dissolved in water (5 mL) and then filtered through a plug of celite. To this solution, a solution of KPF<sup>6</sup> saturated in aqueous (3 mL) was added to obtain a white precipitate. The precipitate was washed with isopropanol (5 mL) and recrystallized by vapor diffusion of CH3CN/diethyl ether to obtain a white crystalline. Yield: 0.15 g, 16.0%. <sup>1</sup>H-NMR (500.023 MHz, d6-DMSO): δ = 9.13 (s, 2H, Himi), 8.99 (s, 2H, Himi), 7.84 (t, <sup>3</sup> JH−<sup>H</sup> = 1.8 Hz, 2H, Himi), 7.72 (t, <sup>3</sup> JH−<sup>H</sup> = 1.8 Hz, 2H, Himi), 7.71 (t, <sup>3</sup> JH−<sup>H</sup> = 1.8 Hz, 2H, Himi), 7.65 (t, <sup>3</sup> JH−<sup>H</sup> = 1.8 Hz, 2H, Himi), 7.39 – 7.43 (m, 3H, ArH), 7.11 (s, 1H, ArH), 7.02 (dd, <sup>3</sup> JH−<sup>H</sup> = 7.7 Hz, <sup>4</sup> JH−<sup>H</sup> = 1.1 Hz, 2H, ArH), 6.92 – 6.96 (m, 2H, ArH), 5.37 (s, 4H, CH2), 5.36 (s, 4H, CH2), 4.72–4.80 (m, 8H, CH2). <sup>13</sup>C-NMR (125.74 MHz, d6-DMSO): δ = 137.40 (Cimi), 137.28 (Cimi), 135.56 (Cq), 132.55 (Cq), 130.16 (CAr), 130.04 (CAr), 129.30 (CAr), 127.75 (CAr), 127.35(CAr), 123.99 (Cimi), 123.74 (Cimi), 123.72 (Cimi), 123.57 (Cimi), 52.14 (CH2), 49.50 (CH2), 49.20 (CH2). HRESI-MS<sup>+</sup> (CH3CN): C32H36N 4+ 8 m/z = 133.0750, calcd = 133.0760, C32H36N8P2F 2+ <sup>12</sup> m/z = 411.1158, calcd = 411.1168, C32H36N8P3F + <sup>18</sup> m/z = 967.1985, calcd = 967.1983.

**14**·**(PF6)6**. A slurry of **9**·**Br<sup>4</sup>** (0.10 g, 0.12 mmol) and Ag2O (0.11 g, 0.47 mmol) in DMF (10 mL) was stirred at 50◦C for 3 d with the exclusion of light. Diethyl ether (∼50 mL) was then added to the mixture and a gray precipitate formed which was collected and dissolved in hot water (5 mL). The solution was clarified by filtration through syringe filter and a saturated solution of KPF<sup>6</sup> (2 mL) was added to obtain an off-white precipitate. The solid collected and washed with hot isopropanol (2 mL) and then recrystallized from a mixture of CH3CN and diethyl ether to obtain a white crystalline solid. Yield: 0.026 g, 21.4%. <sup>1</sup>H-NMR (500.02 MHz, d6-DMSO): δ = 7.70 (s, 6H, Himi), 7.68 (s, 6H, Himi), 7.52 (d, <sup>3</sup> JH−<sup>H</sup> = 7.5 Hz, 6H, ArH), 7.17 (tapp, J = 7.5 Hz, 6H, ArH,), 7.09 (s, 6H, Himi), 6.94 (tapp, J = 7.5 Hz, 6H, ArH), 6.76 (s, 6H, Himi), 5.75 (d, <sup>2</sup> JH−<sup>H</sup> = 15.0 Hz, 12H, CH2), 5.68 (d, <sup>3</sup> JH−<sup>H</sup> = 7.5 Hz, 6H, ArH), 5.60 (d, <sup>2</sup> JH−<sup>H</sup> = 17.0 Hz, 6H, CH2), 4.98 (d, <sup>2</sup> JH−<sup>H</sup> = 14.5 Hz, 6H, CH2), 4.65 (t, <sup>3</sup> JH−<sup>H</sup> = 12.5 Hz, 6H, CH2), 4.52 (t, <sup>3</sup> JH−<sup>H</sup> = 12.5 Hz, 6H, CH2), 4.21 (d, <sup>3</sup> JH−<sup>H</sup> = 12.0 Hz, 6H, CH2), 2.48 (s, 6H, CH2). <sup>13</sup>C-NMR (125.74 MHz, d6-DMSO): δ = 186.31 (d, <sup>1</sup> J = 182.32 Hz, <sup>107</sup>Ag-Ccarbene), 186.31 (d, <sup>1</sup> J = 209.99 Hz, <sup>109</sup>Ag-Ccarbene), 178.75 (d, 1 J = 182.32 Hz, <sup>107</sup>Ag-Ccarbene), 178.75 (d, <sup>1</sup> J = 209.99 Hz, <sup>109</sup>Ag-Ccarbene), 135.80 (Cq), 131.60 (Cq), 130.88 (CAr), 129.41 (CAr), 128.14 (CAr), 124.63 (CAr), 123.90 (Cimi), 123.78(Cimi), 121.75 (Cimi), 55.37 (CH2), 52.58 (CH2), 51.66 (CH2), 51.24 (CH2), 50.50 (CH2). HRESI-MS<sup>+</sup> (CH3CN): C96H96N24Ag6<sup>+</sup> 6 m/z = 372.0334, calcd = 372.0419, C96H96N24Ag6P3F 3+ <sup>18</sup> m/z = 889.0490, calcd = 889.0485.

**15**·**Br6**. A slurry of **9**·**Br<sup>4</sup>** (0.10 g, 0.12 mmol) and (THT)AuCl (0.083 g, 0.26 mmol) in DMF (10 mL) was stirred at 110◦C for 0.5 h. To this mixture NaOAc (0.050 g, 0.59 mmol) was added and the solution was stirred at the same temperature for 1 h. The resultant mixture was cooled to RT and then diethyl ether (50 mL) was added to obtain a white precipitate. The precipitate was collected and recrystallized from hot methanol (∼5 mL). Yield: 0.048 g, 54.2%. <sup>1</sup>H-NMR (500.02 MHz, d6-DMSO): δ = 7.83 (s, 6H, Himi), 7.72 (s, 6H, Himi), 7.57 (d, <sup>3</sup> JH−<sup>H</sup> = 7.5 Hz, 6H, ArH), 7.21 (tapp, J = 7.5 Hz, 6H, ArH) 7.11 (s, 6H, Himi), 6.93 (tapp, J = 7.5 Hz, 6H, ArH) 6.84 (s, 6H, Himi), 6.07 (d, <sup>2</sup> JH−<sup>H</sup> = 16.5 Hz, 6H, CH2), 5.74 – 5.78 (m, 12H, ArH, CH2), 5.66 (d, <sup>3</sup> JH−<sup>H</sup> = 16.5 Hz, 6H, CH2), 5.06 (d, <sup>3</sup> JH−<sup>H</sup> = 14.0 Hz, 12H, CH2), 4.49 (t, <sup>3</sup> JH−<sup>H</sup> = 12.8 Hz, 6H, CH2), 4.37 (d, <sup>3</sup> JH−<sup>H</sup> = 12.5 Hz, 6H, CH2), 2.33 (d, <sup>3</sup> JH−<sup>H</sup> = 12.5 Hz, 6H, CH2). <sup>13</sup>C-NMR (125.74 MHz, d6-DMSO): δ = 186.31 (Ccarbene), 181.48 (Ccarbene), 135.84 (Cq), 131.37 (Cq), 131.02 (CAr), 129.62 (CAr), 127.99 (CAr), 124.59 (CAr), 123.95 (Cimi), 123.64 (Cimi), 121.86 (Cimi), 51.89 (CH2), 51.27 (CH2), 50.22 (CH2). HRESI-MS<sup>+</sup> (CH3CN): C96H96N24Au6<sup>+</sup> <sup>6</sup> m/z = 461.1022, calcd = 461.1035.

**16**·**(PF6)2**. A slurry of **9**·**Br<sup>4</sup>** (0.050 g, 0.059 mmol) and Ag2O (0.030 g, 0.13 mmol) in DMF (5 mL) was stirred at 50◦C for 12 h with the exclusion of light. To this mixture K2PdCl<sup>4</sup> (0.019 g, 0.059 mmol) was added and stirring was continued for a further stirred for 12 h at 80◦C. The reaction mixture was clarified by centrifugation and diethyl ether (30 mL) was added to the supernatant yielding a gray precipitate. The precipitate was collected and re-dissolved in hot water (5 mL) and the solution filtered through a syringe filter. To the filtrate, a saturated solution of KPF<sup>6</sup> (2 mL) was added to obtain an off-white precipitate. The precipitate was dried in vacuo and then recrystallized from a mixture of CH3CN and diethyl ether to obtain a white crystalline solid. Yield: 0.0052 g, 9.6%. <sup>1</sup>H-NMR (500.02 MHz, d6-DMSO): δ = 7.87 – 7.92 (m, 4H, ArH), 7.83 (d, <sup>3</sup> JH−<sup>H</sup> = 2.0 Hz, 4H, Himi), 7.48 (d, <sup>3</sup> JH−<sup>H</sup> = 2.0 Hz, 4H, Himi), 7.42 – 7.47 (m, 4H, ArH), 6.44 (d, <sup>3</sup> JH−<sup>H</sup> = 15.0 Hz, 4H, CH2), 5.20 (d, <sup>3</sup> JH−<sup>H</sup> = 14.7 Hz, 4H, CH2), 4.98 – 5.06 (m, 4H, CH2), 4.54 – 4.61 (m, 4H, CH2). <sup>13</sup>C-NMR (125.74 MHz, d6-DMSO): δ =167.41 (Ccarbene), 135.56 (Cq), 131.73 (CAr), 129.83 (CAr), 124.96 (Cimi), 122.68 (Cimi), 50.79 (CH2), 47.39 (CH2). HRESI-MS<sup>+</sup> (CH3CN): C32H32N8Pd2<sup>+</sup> m/z = 317.0889, calcd = 317.0887.

**17**·**(PF6)3**. This compound was prepared using the same method as described for **14**·**(PF6)<sup>6</sup>** from **10**·**Br<sup>4</sup>** (0.050 g, 0.059 mmol) and Ag2O (0.057 g, 0.24 mmol). Yield: 0.0050 g, 6.2%. <sup>1</sup>H-NMR (500.02 MHz, CD3CN): δ = 8.37 (t, J = 1.5 Hz, 2H, Himi), 7.45 – 7.53 (m, 12H, ArH, Himi), 7.21 (t, <sup>3</sup> JH−<sup>H</sup> = 1.8 Hz, 2H, Himi), 6.98 (s, 2H, ArH), 6.95 (t, <sup>3</sup> JH−<sup>H</sup> = 1.9 Hz, 2H, Himi) 5.31 (s, 4H, CH2), 5.08 (s, 4H, CH2), 4.60 (d, <sup>3</sup> JH−<sup>H</sup> = 4.3 Hz, 4H, CH2), 4.55 (d, <sup>3</sup> JH−<sup>H</sup> = 4.7 Hz, 4H, CH2). <sup>13</sup>C-NMR (125.74 MHz, CD3CN): δ = 180.39 (d, 1 J = 183.58 Hz, <sup>107</sup>Ag-Ccarbene), 180.39 (d, <sup>1</sup> J = 211.24 Hz, <sup>109</sup>Ag-Ccarbene), 138.62 (Cq), 136.79 (Cq), 136.18 (Cimi), 130.86 (CAr), 129.82 (Cimi), 129.48 (Cimi), 126.74 (CAr), 125.47 (CAr), 125.42 (CAr), 124.47 (Cimi), 124.12 (Cimi), 122.39 (CAr), 122.35 (CAr), 55.33 (CH2), 53.45 (CH2), 52.73 (CH2), 50.96 (CH2). HRESI-MS<sup>+</sup> (CH3CN): C32H34N8Ag3<sup>+</sup> m/z = 213.0643, calcd = 213.0646, C32H34N8AgPF2<sup>+</sup> <sup>6</sup> m/z = 391.2845, calcd = 391.0794, C32H34N8AgP2F + <sup>12</sup> m/z = 927.1228, calcd = 927.1235.

**18**·**(PF6)2**. This compound was prepared using the same method as described for **15**·**Br<sup>6</sup>** from **11**·**Br<sup>4</sup>** (0.050 g, 0.057 mmol), (THT)AuCl (0.040 g, 0.13 mmol), and NaOAc (0.019 g, 0.23 mmol). For exchange of the bromide anion to hexafluorophosphate the crude product was dissolved in water (3 mL) and the solution filtered through celite. To this solution, a saturated solution of KPF<sup>6</sup> (2 mL) was added to obtain a white precipitate which was then recrystallized from a mixture of CH3CN and diethyl ether yielding a white crystalline solid. Yield: 0.020 g, 29.9%. <sup>1</sup>H-NMR (500.02 MHz, CD3CN): δ = 7.55 – 7.57 (m, 1H, ArH), 7.46 – 7.48 (m, 1H, ArH), 7.44 (d, <sup>3</sup> JH−<sup>H</sup> = 2.0 Hz, 1H, Himi), 7.41−7.42 (m, 1H, ArH), 7.38 (s, 1H, Himi), 7.23 (d, <sup>3</sup> JH−<sup>H</sup> = 2.1 Hz, 1H, Himi), 7.21-7.22 (m, 1H, ArH), 7.15 (d, <sup>3</sup> JH−<sup>H</sup> = 1.9 Hz, 1H, Himi), 5.53 (s, 2H, CH2), 5.49 (s, 2H, CH2), 4.66 – 4.67 (m, 2H, CH2), 4.57 – 4.59 (m, 2H, CH2), 2.54 (s, 3H, CH3). <sup>13</sup>C-NMR (125.74 MHz, CD3CN): δ = 145.20 (Ccarbene), 133.61 (Cq), 130.51 (CAr), 130.43 (CAr), 130.22 (CAr), 129.75 (CAr), 122.74 (Cimi), 122.50 (Cimi), 122.24 (Cimi), 122.05 (Cimi), 51.84 (CH2), 50.05 (CH2), 49.69 (CH2), 48.48 (CH2), 10.36 (CH3). HRESI-MS<sup>+</sup> (CH3CN): C34H38N8Au2Br2<sup>+</sup> <sup>2</sup> m/z = 556.0369, calcd = 556.0443.

**19**·**(PF6)2**. This compound was prepared using the same method as that described for **15**·**Br<sup>6</sup>** from **11**·**Br<sup>4</sup>** (0.030 g, 0.034 mmol), (THT)AuCl (0.024 g, 0.075 mmol), and NaOAc (0.11 g, 0.14 mmol). For exchange of the bromide anion to hexafluorophosphate the crude product was dissolved in water (3 mL) and the solution filtered through celite. To this solution, a saturated solution of KPF<sup>6</sup> (2 mL) was added to obtain a white precipitate which was then recrystallized from a mixture of CH3CN and diethyl ether yielding a white crystalline solid. Yield: 0.0070 g, 1.5%. <sup>1</sup>H-NMR (500.02 MHz, d6-DMSO): δ = 7.71 (d, <sup>3</sup> JH−<sup>H</sup> = 1.9 Hz, 2H, Himi), 7.62 (d, <sup>3</sup> JH−<sup>H</sup> = 1.9 Hz, 2H, Himi), 7.56 (d, <sup>3</sup> JH−<sup>H</sup> = 1.9 Hz, 2H, Himi), 7.39−7.43 (m, 3H, ArH, Himi), 7.28 – 7.38 (m, 3H, ArH), 7.14 – 7.19 (m, 1 H, ArH), 7.07 – 7.14 (m, 1H, ArH), 6.56 – 6.69 (m, 2H, ArH), 5.31 (s, 4H, CH2), 5.06 (s, 4H, CH2), 4.67 (s, 4H, CH2), 4.66 (s, 4H, CH2), 2.40 (s, 6H, CH3). <sup>13</sup>C-NMR (125.74 MHz, d6-DMSO): δ = 172.52 (Ccarbene), 144.61 (Cq), 136.79 (Cq), 134.11 (Cq), 130.12 (CAr), 127.96 (CAr), 125.29 (CAr), 123.45 (Cimi), 122.28 (Cimi), 121.98 (Cimi), 121.91 (Cimi), 52.82 (CH2), 50.44 (CH2),

49.84 (CH2), 48.11 (CH2), 9.19 (CH3). HRESI-MS<sup>+</sup> (CH3CN): <sup>C</sup>34H38N8Au2Br2<sup>+</sup> <sup>2</sup> m/z = 556.0422, calcd = 556.0443.

**20**·**(PF6)2**. To a solution of **11**·**Br<sup>4</sup>** (0.050 g, 0.057 mmol) and K2PdCl<sup>4</sup> (0.019 g, 0.057 mmol) in DMSO (5 mL) at 85◦C, NaOAc (0.019 g, 0.23 mmol) was added and the reaction mixture was stirred same temperature for 12 h. The resultant mixture was then diluted with acetone (5 mL) followed by adding diethyl ether (20 mL) to form a precipitate. The crude precipitate was collected and then re-dissolved in water and filtered through a plug of Celite followed by addition of a saturated solution of KPF<sup>6</sup> (3 mL). The resultant precipitate was collected and washed with isopropanol (5 mL) and diethyl ether (2 × 5 mL) and then recrystallized from a mixture of CH3CN and diethyl ether yielding the product as a pale yellow solid. NMR analysis showed this material to be impure and the <sup>1</sup>H-NMR spectrum is reported for the impure material. <sup>1</sup>H-NMR (500.02 MHz, CD3CN): δ = 7.66 (s, 2H, Himi), 7.55 – 7.57 (m, 4H, ArH), 7.35−7.40 (m, 2H, ArH), 7.34 (s, 2H, Himi), 7.22 – 7.24 (m, 2H, Himi), 7.03 (s, 2H, Himi), 7.02 (s, 2H, Himi), 5.26 (s, 4H, CH2), 5.24 (s, 4H, CH2), 4.61 – 4.70 (m, 8H, CH2), 2.57 (s, 6H, CH3). A crystal suitable for single crystal X-ray diffraction analysis of **20**·**(PF6)<sup>2</sup>** was grown from diffusion of diethyl ether into an acetonitrile solution of **20**·**(PF6)2**.

# REFERENCES


# AUTHOR CONTRIBUTIONS

PB conceived of the presented idea. ZL synthesized compounds and recorded NMR spectra. NW recorded and analyzed high resolution mass spectra. PB and ZL collected X-ray diffraction data. All authors discussed the results and contributed to the final manuscript.

# FUNDING

This research was financially supported by The Australia Research Council, DP150102741 and ZL acknowledges receipt of a La Trobe University postgraduate award.

# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fchem. 2019.00270/full#supplementary-material

Supporting Information | Synthetic details for compounds 1-3, Crystallographic data for compounds 5·Br<sup>2</sup> , 9·Br<sup>4</sup> , 10·Br3PF<sup>6</sup> , 11·Br<sup>4</sup> , 15·Br<sup>2</sup> (PF6 )4 , and 20·(PF<sup>6</sup> )2 and further X-ray crystallographic details and anion binding Job plots for compounds 9·(PF<sup>6</sup> )<sup>4</sup> and 10·(PF<sup>6</sup> )4 and 1H and 13C NMR spectra for tetra-imidazolium salts and metal complexes.


copper(I)–N-heterocyclic carbene (NHC) complexes derived from basketshaped dinuclear Ag(I)–NHC complex. Inorg. Chem. 57, 11322–11332. doi: 10.1021/acs.inorgchem.8b00011


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Li, Wiratpruk and Barnard. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.