**Novel Antibacterial Agents**

Editors

**Fiorella Meneghetti Daniela Barlocco**

MDPI • Basel • Beijing • Wuhan • Barcelona • Belgrade • Manchester • Tokyo • Cluj • Tianjin

*Editors* Fiorella Meneghetti Department of Pharmaceutical Sciences University of Milan Milan Italy Daniela Barlocco Department of Pharmaceutical Sciences University of Milan Milan Italy

*Editorial Office* MDPI St. Alban-Anlage 66 4052 Basel, Switzerland

This is a reprint of articles from the Special Issue published online in the open access journal *Pharmaceuticals* (ISSN 1424-8247) (available at: www.mdpi.com/journal/pharmaceuticals/special issues/Antibacterial Agents).

For citation purposes, cite each article independently as indicated on the article page online and as indicated below:

LastName, A.A.; LastName, B.B.; LastName, C.C. Article Title. *Journal Name* **Year**, *Volume Number*, Page Range.

**ISBN 978-3-0365-2861-8 (Hbk) ISBN 978-3-0365-2860-1 (PDF)**

© 2022 by the authors. Articles in this book are Open Access and distributed under the Creative Commons Attribution (CC BY) license, which allows users to download, copy and build upon published articles, as long as the author and publisher are properly credited, which ensures maximum dissemination and a wider impact of our publications.

The book as a whole is distributed by MDPI under the terms and conditions of the Creative Commons license CC BY-NC-ND.

### **Contents**


Reprinted from: *Pharmaceuticals* **2020**, *13*, 411, doi:10.3390/ph13110411 . . . . . . . . . . . . . . . **137**


**Xiaofei Qin, Celina Vila-Sanjurjo, Ratna Singh, Bodo Philipp and Francisco M. Goycoolea** Screening of Bacterial Quorum Sensing Inhibitors in a *Vibrio fischeri* LuxR-Based Synthetic Fluorescent *E. coli* Biosensor Reprinted from: *Pharmaceuticals* **2020**, *13*, 263, doi:10.3390/ph13090263 . . . . . . . . . . . . . . . **267**


### **About the Editors**

#### **Fiorella Meneghetti**

Fiorella Meneghetti earned her degree with honors in Medicinal Chemistry and Technology at the University of Padova in 1998. In the same year, she obtained the qualification to practice as a pharmacist. In 2002, she received her Ph.D. in Bio-Chemical Sciences from the University of Torino. Then, she joined the University of Milan with a grant entitled "Structural investigation of redox proteins for the nanobiotechologies" (2002-2005). F.M. has been Assistant Professor since January 1st, 2006 (SSD CHIM/08) at the Department of Pharmaceutical Sciences of the University of Milan; she is responsible for the teaching of "Chemical and Toxicological Analysis" and "Drug Analysis". The main area of her research is the crystallization and structural determination of drugs and enzymes. F.M. is co-author of 97 scientific articles published on international journals of relevant impact factor and several presentations (oral/poster) at national and international conferences. Her research activity has been focused on the following fields: 1) structural analysis of enzymes; 2) crystal structures of pharmaceutical compounds; and 3) structural characterization of metal complexes.

#### **Daniela Barlocco**

Daniela Barlocco received her laurea degree in Pharmacy at the University of Pavia (Italy) in 1969 and her laurea degree in Chemistry in 1977 at the same University. In 1978, she joined the University of Milan, Faculty of Pharmacy. In 1992, she moved to the University of Modena for three years as associate professor. In 1995, she returned to the Faculty of Pharmacy of Milan, where she worked as professor of Medicinal Chemistry until her retirement in 2016. She is member of several Editorial Boards, including MRMC and Current Medicinal Chemistry. She spent several months as visiting professor at SmithKline Beecham (Harlow, UK) and at the University College (London). In 2005, she was a member of the Evaluation Panel for the Faculty of Pharmacy in Helsinki. Her research interests include synthetic and theoretical chemistry of pyridazines and medicinal chemistry of several enzyme inhibitors, e.g., AR, TS, ACAT, SSAO, STAT.

### **Preface to "Novel Antibacterial Agents"**

This book was devoted to the latest advances achieved in the antibacterial field, with a focus on the recent efforts made to develop new antimicrobial agents with novel modes of action, and a perspective on future directions of this line of research.

Antimicrobial resistance has become a major threat to global health, and the twenty-two published articles here reported put in evidence that the discovery and development of new antibiotics are extremely challenging. The antimicrobial research covers a wide area, spanning from the design of new compounds, also supported by molecular modeling techniques, their synthesis and characterization, and biological tests.

In this context, the current crisis caused by the COVID-19 pandemic, but also older threats, such as the human immunodeficiency virus or the hepatitis C virus, require greater attention than ever.

The research works described in this book provide an extremely useful example of the results achieved in the field of antibacterial drug development. The search for new chemical entities was approached starting from both natural and synthetic compounds and addressing different targets. In addition, recent findings were presented and discussed highlighting the strategies to fight bacterial resistance. Detailed references to the state-of-the-art can be found in this book.

We strongly encourage the wide group of readers to explore the book that we are presenting, to get inspired to develop new approaches for the diagnosis and treatment of antibacterial diseases, and to circumvent resistance issues.

We are extremely grateful to all the authors for the hard work they have done in producing the chapters. We thank MDPI for the decision to publish this book and Ms. Fendy Fan for the kind assistance and technical support.

> **Fiorella Meneghetti, Daniela Barlocco** *Editors*

### *Editorial* **Special Issue "Novel Antibacterial Agents"**

**Fiorella Meneghetti \* and Daniela Barlocco**

Department of Pharmaceutical Sciences, University of Milan, Via L. Mangiagalli 25, 20133 Milan, Italy; daniela.barlocco@unimi.it

**\*** Correspondence: fiorella.meneghetti@unimi.it

This Special Issue of *Pharmaceuticals* is devoted to significant advances achieved in the field of antibacterial agents. Here, we report recent efforts made to develop new antimicrobials with novel modes of action/resistance, and offer perspectives on the future directions of antibacterial agents.

Antimicrobial resistance has become a major threat to global health and the twentytwo published articles, included here, evidence that the discovery and development of new antibiotics is extremely challenging.

This Special Issue is focused on the search for new chemical entities, starting from both natural and synthetic compounds and addressing different targets. In addition, recent findings are presented and discussed, highlighting strategies of fighting bacterial resistance.

Investigation into antimicrobials covers a wide research area, as emphasized in this Special Issue, spanning from the design, synthesis, and characterization of new compounds, supported by molecular modeling techniques, to the development of biological tests. This is all possible thanks to the contributions of experts.

Natural products, as rich sources of chemical diversity, offer excellent possibilities to identify novel leads in medicinal chemistry. In this direction, S. Garzoli et al. [1], via Headspace-Gas Chromatography/Mass Spectrometry (HS-GC/MS), identified 28 components, mainly belonging to the monoterpenes family, in the essential oils from needles (EOs) of four *Pinaceae*. Both the liquid and vapour phases were evaluated for their antibacterial activity against three Gram-negative (*Escherichia coli*, *Pseudomonas fluorescens*, and *Acinetobacter bohemicus*) and two Gram-positive (*Kocuria marina* and *Bacillus cereus*) bacteria using different assays. Better results were obtained with the vapour phase. In addition, a concentration-dependent antioxidant activity was evidenced for all the EOs. The Authors highlighted the importance of α-pinene as a novel natural antibacterial agent.

Other interesting antioxidant compounds were identified by L.C. Nascimento da Silva et al. [2], in *B. tetraphylla* leaf methanolic extracts (BTME) which, when tested on *Tenebrio molitor* larvae inoculated with heat-inactivated *E. coli*, proved to be able to protect the larvae from the stress. A mixture of aliphatic (terpenes, fatty acids, carbohydrates) and aromatic compounds (phenolic derivatives) were evidenced in BTME using NMR analysis.

Antioxidant properties, in addition to antimicrobial and cytotoxic activity, were reported by A.J.L. Pombeiro et al. [3] in a series of saccharin-tetrazolyl and -thiadiazolyl derivatives. The best antioxidant results were shown for (*N*-(1-methyl-2*H*-tetrazol-5-yl)- *N*-(1,1-dioxo-1,2-benzisothiazol-3-yl) amine, while all the compounds had insignificant toxicity when tested in an *Artemia salina* model.

Several EOs, derived from *Origanum majorana*, *Rosmarinus officinalis*, and *Thymus zygis* medicinal plants were investigated by A. Gaber et al. [4] for their ability to inhibit biofilm formation and eradicate methicillin-resistant *Staphylococcus aureus* (MRSA) isolates. The best activity was found in *T. Zygis*, but all the studied EOs showed interesting properties (the percentage of inhibition ranging from 10.20 to 95.91%, and percentage of eradication ranging from 12.65 to 98.01%), thus, representing potential alternatives to antibiotics.

**Citation:** Meneghetti, F.; Barlocco, D. Special Issue "Novel Antibacterial Agents". *Pharmaceuticals* **2021**, *14*, 382. https://doi.org/10.3390/ph14040382

Received: 14 April 2021 Accepted: 16 April 2021 Published: 19 April 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

The inhibition of biofilm formation was also studied also by B. Citterio et al. [5], who investigated the properties of the marine bisindole alkaloid 2,2-bis(6-bromo-1*H*-indol-3 yl)ethanamine and its fluorinated analogue which were tested both for their potential use as antibiotic adjuvants and antibiofilm agents against *S. aureus* CH 10850 (MRSA) and *S. aureus* ATCC 29213 (MSSA). The fluorinated derivative showed a higher potency in eradicating a preformed biofilm. Both compounds showed a safe profile and were indicated for in vivo application as adjuvants to restore antibiotic treatment against MRSA.

Biofilm formation is an urgent problem in dentistry, due to the high porosity and absorptiveness of the polymers commonly used for the 3D-printed dental prostheses. D.-H. Lee et al. [6] studied the effects of a new chlorhexidine (CHX)-loaded polydimethylsiloxane (PDMS)-based coating material on the surface microstructure, surface wettability and antibacterial activity of 3D-printing dental polymer. The antibacterial was first encapsulated in mesoporous silica nanoparticles (MSN). The significant experimental results supported this approach.

Functionalization of oxidized multi-walled carbon nanotubes (oxCNTs) was studied by Z. Sideratou et al. [7], who used a simple non-covalent modification procedure, using hyperbranched poly(ethyleneimine) derivatives (QPEIs), with various quaternization degrees. The aqueous dispersion of this material was found to be stable over 12 months and to be provided with significant antibacterial and anti-cyanobacterial activity. It was suggested by the Authors for application in the disinfection industry.

W.Y. Bang et al. [8] proposed the use of the cell-free supernatant from *Lactobacillus plantarum* NIBR97 as an alternative to chemical disinfectants. By scanning electron microscopy (SEM), this was shown to cause cellular lysis in bacterial membranes, thus, indicating the involvement of peptides or proteins in its mechanism of action and suggesting the use of proteinase K treatment to support its antibacterial activity. In addition, it showed good antiviral properties, possibly through a different mechanism of action.

A study by A. Baycal et al. [9] on *Moringa oleifera* reported the correlation between the capability of laser-induced breakdown spectroscopy (LIBS) to monitor the elemental compositions of plants and their biological effects. In particular, the bioactive components of the seed (MOS) alcoholic extract were tested against *Escherichia coli* and *Staphylococcus aureus* via an agar well diffusion (AWD) assay and scanning electron microscopy (SEM). An interesting activity against Gram-positive bacteria was evidenced. In addition, the authors reported that MOS extract exhibited significant inhibitory properties on HCT116 cell growth, whereas no effects were noticed in a parallel assay on HEK-293 cells. According to the authors, the antibacterial and anticancer potency of the MOS extract could be attributed to several complexes, e.g., ethyl ester and D-allose and hexadecenoic, oleic and palmitic acids.

J.-H Moon et al. [10] reported their results on 3-pentylcatechol (PC), a synthetic urushiol derivative, able to inhibit the growth of *Helicobacter pylori* in an in vitro assay, in comparison with triple therapy (omeprazole, metronidazole, and clarithromycin). The authors report that PC displayed better effects than triple therapy at all doses. Interestingly, synergism was shown when PC and triple therapy were co-administered.

The composition and biological properties of *Thymus mastichina,* a diffuse semishrub of jungles and the rocky landscapes of the Iberian Peninsula, were reviewed by A.R.T.S. Araujo et al. [11]. Several properties were reported both for the extracts and the essential oils. In particular, activity against methicillin resistant bacteria and antifungal activity should be highlighted, though several other properties (e.g., anticancer, antiviral, insecticidal, repellent, anti-Alzheimer's, and anti-inflammatory) have been investigated. The authors suggest that *Thymus mastichina* should be considered for use in food and cosmetic applications other than in the pharmaceutical field.

Antibiotic resistance was also the target of the paper by M. Sessevmez et al. [12], who indicated bacteriophages as an alternative method, which was first proposed in the early 20th century by d'Herelle, Bruynoghe and Maisin to treat bacterial infections. Different administration methods are possible, including topical application, inhalation, oral or parenteral delivery. *Pseudomonas aeruginosa*, *Mycobacterium tuberculosis* and *Acinetobacter baumannii*, responsible for the main drug resistant infections, are potential targets of phages, which are developed under a strict quality control regime.

Tuberculosis was also the target of F. Meneghetti et al. [13], who described novel furan derivatives as inhibitors of salicylate synthase MbtI, an essential enzyme of mycobacterium, absent in human cells. The best compound, which provided comparable inhibitory properties to the previous leads but endowed a better antitubercular activity, was 5-(3-cyano-5-(trifluoromethyl)phenyl)furan-2-carboxylic acid, which will form the basis of future studies.

Covalent inhibitors of another bacterial enzyme for which there is no human orthologue, namely UDP-*N*-acetylglucosamine enolpyruvyl transferase (MurA), have been developed by G.M. Keserü et al. [14], who indicated bromo-cyclobutenaminones as novel electrophilic probes by screening a small library of cyclobutenone derivatives. The bromine atom has been recognized as an essential requirement, and MS/MS experiments have led to the suggestion that Cys115 is also involved. The stability and bioavailability of these compounds was also assessed.

Another alternative to the use of antibiotics in bacterial infections was proposed by F.M. Goycoolea et al. [15] who studied a library of 23 pure compounds of different chemical structures, assessing their quorum sensing (QS) inhibition activity. The best results were obtained with phenazine carboxylic acid (PCA), 2-heptyl-3-hydroxy-4-quinolone (PQS), 1*H*-2-methyl-4-quinolone (MOQ) and genipin, which exhibited QS inhibition activity without compromising bacterial growth.

A specific disease, characterized by complex aetiological mechanisms, namely atopic dermatitis (AD), was the topic of a review by R. Di Marco et al. [16]. Loss of the skin barrier, linked to dysbiosis and immunological dysfunction, which causes an imbalance in the ratio between the pathogen *Staphylococcus aureus* and/or other microorganisms residing in the skin, is considered as a crucial factor contributing to AD. Though the review suggests several treatments for the disease, including the use of bacteria and/or microbiota transplantation, together with different drug delivery systems, the authors conclude that a standardized process is necessary in order to obtain reliable data.

F. Sparatore et al. [17] addressed the topic of Leishmaniases, the therapy of which is presently based on expensive drugs which are associated with severe side-effects and the treat of resistance. They tested sixteen lucanthone and four amitriptyline analogues in vitro, all of which were characterized by a bulky quinolizidinylalkyl moiety, against *Leishmania tropica* and *L. infantum* promastigotes. All compounds displayed significant activity (IC<sup>50</sup> in the low µM range) and low cytotoxicity. The authors suggest that these analogues could act through trypanothione reductase (TryR) inhibition.

In the search for selective agents that are capable of treating bacterial diarrhoea without affecting the host intestinal microbiota, L. Kokoska et al. [18] investigated ten phytochemicals and their synthetic analogues (berberine, bismuth subsalicylate, ferron, 8-hydroxyquinoline, chloroxine, nitroxoline, salicylic acid, sanguinarine, tannic acid, and zinc pyrithione) in vitro and compared the results with six commercial antibiotics (ceftriaxone, ciprofloxacin, chloramphenicol, metronidazole, tetracycline, and vancomycin) against 21 intestinal pathogenic/probiotic (e.g., *Salmonella* spp. and *bifidobacteria*) bacterial strains and three intestinal cancer/normal (Caco-2, HT29, and FHs 74 Int) cell lines. Several compounds, e.g., chloroxine, ciprofloxacin, nitroxoline, tetracycline, and zinc pyrithione exhibited potent selective growth-inhibitory activity against pathogens, while 8-hydroxyquinoline and sanguinarine provided the best activity towards cancer cells. It should also be noted that none of the compounds were found to be cytotoxic to normal cells. Once more, plants have been suggested as a promising source for novel drugs.

N. D'Amelio et al. [19] investigated the mechanism of action of a synthetic peptide (K11), whose antibacterial properties have been previously reported. Via a liquid and solid-state NMR technique, they studied the interaction of K11 with different biomimetic membranes and reported that this can destabilize them. In addition, via molecular dy-

namic simulations, they suggested that K11 could penetrate the membranes in four steps (anchoring, twisting, helix flip, and internalization) involving several lysine residues.

Additionally, also within the field of synthetic compounds, B. Altava et al. [20] investigated how they could vary the antibacterial properties of imidazole and its imidazolium salts derived from *L*-valine and *L*-phenylalanine, by modulating their lipophilicity. When tested on *E. Coli* and *B. Subtilis* bacterial strains, very encouraging results were obtained. In particular, the minimum bactericidal concentration (MBC) of one compound towards *B. subtilis* was found to be lower than the IC<sup>50</sup> cytotoxicity value for the control cell line, HEK-293. Moreover, the capability of these structures to aggregate in different media was investigated to establish the importance of the monomeric species.

The effects of substituent variation on a different substrate, namely *tris*(1*H*-indol-3 yl)methylium, was investigated by A.S. Trenin et al. [21]. The synthesized compounds were tested on 12 bacterial strains that were either sensitive or resistant to meticillin. The results indicated that antibacterial properties depended on the chain length, with the best activity residing in compounds with C5–C6 chains. The most interesting compounds showed better antibacterial properties than levofloxacin on MRSA and had a very safe profile.

Antibacterial and antifungal activity was investigated in vitro by P. Eleftheriou et al. [22] on a series of 3-amino-5-(indol-3-yl) methylene-4-oxo-2-thioxothiazolidine derivatives. Compounds exhibited significant activity both on Gram-positive and Gram-negative bacteria, demonstrating a potency greater than ampicillin. Similarly, their antifungal activity was superior to that of ketoconazole. Docking studies have suggested that their antibacterial activity could be derived from *E. coli* Mur B inhibition, while CYP51 inhibition would be responsible for the antifungal activity.

The research described in the articles constituting this Special Issue collectively provides extremely useful examples of the results that have been recently achieved in the field of antibacterial drug development.

We hope the readers enjoy this Special Issue and are inspired to develop new approaches for antibacterial disease diagnosis, treatment and to circumvent resistance issues.

**Author Contributions:** F.M. and D.B. contributed equally to this Editorial. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by University of Milan, Linea B.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Acknowledgments:** We are extremely grateful to all authors for their hard work to produce an updated and comprehensive issue on antibacterial agents in a timely fashion. We would also like to thank the Reviewers who carefully evaluated the submitted manuscripts. We thank the Editor-in-Chief of Pharmaceuticals, J.J. Vanden Eynde, for giving us the opportunity to serve as Guest Editors and F. Fan for the kind assistance and technical support.

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

#### **References**


### *Article* **Synthesis, Characterization, and Biological Evaluation of New Derivatives Targeting MbtI as Antitubercular Agents**

**Matteo Mori 1,† , Giovanni Stelitano 2,† , Laurent R. Chiarelli <sup>2</sup> , Giulia Cazzaniga <sup>1</sup> , Arianna Gelain <sup>1</sup> , Daniela Barlocco <sup>1</sup> , Elena Pini <sup>1</sup> , Fiorella Meneghetti 1,\* and Stefania Villa <sup>1</sup>**


**Abstract:** Tuberculosis (TB) causes millions of deaths every year, ranking as one of the most dangerous infectious diseases worldwide. Because several pathogenic strains of *Mycobacterium tuberculosis* (Mtb) have developed resistance against most of the established anti-TB drugs, new therapeutic options are urgently needed. An attractive target for the development of new antitubercular agents is the salicylate synthase MbtI, an essential enzyme for the mycobacterial siderophore biochemical machinery, absent in human cells. A set of analogues of **I** and **II**, two of the most potent MbtI inhibitors identified to date, was synthesized, characterized, and tested to elucidate the structural requirements for achieving an efficient MbtI inhibition and a potent antitubercular activity with this class of compounds. The structure-activity relationships (SAR) here discussed evidenced the importance of the furan as part of the pharmacophore and led to the preparation of six new compounds (**IV**–**IX**), which gave us the opportunity to examine a hitherto unexplored position of the phenyl ring. Among them emerged 5-(3-cyano-5-(trifluoromethyl)phenyl)furan-2-carboxylic acid (**IV**), endowed with comparable inhibitory properties to the previous leads, but a better antitubercular activity, which is a key issue in MbtI inhibitor research. Therefore, compound **IV** offers promising prospects for future studies on the development of novel agents against mycobacterial infections.

**Keywords:** tuberculosis; mycobactins; furan; siderophores; drug design; bioisosterism; drug resistance

#### **1. Introduction**

Tuberculosis (TB), the infectious disease caused by *Mycobacterium tuberculosis* (Mtb), represents a global emergency requiring new therapeutic options, mainly because of the rapid spread of drug-resistant strains, which are causing an alarming rise in clinical cases.

According to the 2020 World Health Organization (WHO) report [1], TB was responsible for around 1.4 million deaths and over 10 million new infections worldwide in 2019; these numbers are expected to rise significantly in 2020 as a consequence of the coronavirus disease 2019 (COVID-19) pandemic. Additionally, it is estimated that Mtb exists in its latent form in approximately one-quarter of the global population [1].

Although the investigation of new pharmaceutical forms for the delivery of current antitubercular drugs may contribute to enhance patient compliance and limit the spread of the disease [2,3], the development of new therapeutic options represents an even more pressing need. While drug-susceptible TB can be cured within 6–8 months with the current standard treatment regimen [4], multi- and extensively drug-resistant (MDR/XDR) infections are treated for at least 20 months with poor outcomes [5], posing a serious threat to human health. The continuous genetic adaptation and rapid propagation of

**Citation:** Mori, M.; Stelitano, G.; Chiarelli, L.R.; Cazzaniga, G.; Gelain, A.; Barlocco, D.; Pini, E.; Meneghetti, F.; Villa, S. Synthesis, Characterization, and Biological Evaluation of New Derivatives Targeting MbtI as Antitubercular Agents. *Pharmaceuticals* **2021**, *14*, 155. https://doi.org/10.3390/ph14020155

Academic Editor: Pascal Sonnet Received: 12 January 2021 Accepted: 9 February 2021 Published: 13 February 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

drug-resistant pathogens have led to an expected drop in the therapeutic efficacy of the current anti-TB drugs, forcing the scientists to face new challenges in the discovery of novel molecular entities to address this issue. Hence, the development of innovative compounds targeting both replicating and dormant Mtb bacilli is critical for the design of more effective and shorter therapies.

To address the urgent need of selective antitubercular drugs with novel mechanisms of action, new drug targets have been recently explored and validated [6–8]. Among them, the mycobactin biosynthetic pathway, which leads to the synthesis of siderophores capable of sequestering host iron, has been identified as a source of promising candidates [9,10]. Indeed, the siderophore biochemical machinery is significantly upregulated under irondeficient conditions, common in infected macrophages, constituting one of the major pathogenic determinants of TB. Moreover, it is absent in humans, thus minimizing the risk of off-target effects.

Among the four enzymes involved in mycobactin biosynthesis and currently under investigation as potential drug targets (i.e., MbtI, MbtA, MbtM, and PPTase), we focused our attention on the Mg2+-dependent bifunctional salicylate synthase MbtI, which catalyzes the first two steps in the production of all mycobacterial siderophores. This enzyme belongs to the family of the chorismate-utilizing enzymes (CUEs) [11] and it catalyzes the reactions shown in Figure 1.

**Figure 1.** Reactions catalyzed by MbtI.

In this context, we developed in recent years a series of furan-based carboxylic acids as MbtI inhibitors [12–15]. Among this class of compounds, **I** and **II** (Figure 2) emerged as the best leads, exhibiting a strong MbtI inhibition, conceivably related to their antitubercular activity, and a negligible cytotoxicity towards eukaryotic cells. When analyzing the structure-activity relationships (SAR) of these compounds, we observed that the activity of the substances was closely related to the presence of an electron withdrawing moiety in a suitable position of the phenyl ring. The removal of the substituent from the phenyl of our furan-based leads (**III**, Figure 2) resulted in a complete loss of activity [14].

**Figure 2.** Chemical structure of the lead compounds **I**, **II**, and **III**.

Encouraged by these studies, and with the aim of enriching our arsenal of MbtI inhibitors with compounds exhibiting enhanced antitubercular activities, we enlarged our set of derivatives to include compounds bearing different heterocyclic scaffolds.

In some literature cases, the furan core was successfully replaced by other heterocycles to improve the cellular activity; indeed, according to Hinsberg's "ring equivalence" theory, the concept of isosterism and bioisosterism can be extended to heterocycles [16]. Here, we investigated whether the furan moiety could be successfully replaced by any of the heterocycles shown in Figure 3, also considering that extensive research efforts have been devoted to the exploration of heterocyclic compounds as antimycobacterial agents [17].

**Figure 3.** Chemical structure of the heterocyclic cores tested in this study: **1** (thiophene), **2** (thiazole), **3** (oxazole), **4** (imidazole), **5** (1,3,4-oxadiazole), **6** (1,2,3-triazole).

We considered the introduction of a thiophene (**1**), because, in several literature examples, the use of this ring has resulted in an improvement of the antimycobacterial activity [18–20]. The furan was then substituted by a thiazole (**2**); aside from being the most common heterocycle in drug design [21], this ring is part of the chemical structure of many compounds endowed with antitubercular activity [22]. To expand our investigations, we synthesized two derivatives bearing an oxazole (**3**), where the sulfur atom of the thiazole ring is substituted by an oxygen, a sulfur isostere [23]. The imidazole was then selected as an attractive isostere of thiazole and oxazole; notably, nitroimidazopyran PA-824 [24] has recently moved to advanced-stage clinical trials, inspiring the development of anti-TB agents featuring this moiety [25]. Finally, we explored the 1,3,4-oxadiazole (**5**), as it was reported to interact with some of the newest anti-TB targets [26], and the 1,2,3-triazole (**6**), whose importance is demonstrated by the antitubercular agent I-A09, which is under preclinical trials [27].

In the first part of this work, we designed, synthesized, and evaluated the biological activity of novel heterocyclic compounds belonging to two homologous series (Table 1), bearing the *m*-CN (series A, compounds **1a**–**6a**) and *p*-NO<sup>2</sup> (series B, **1b**–**6b**) substituent, respectively.


**Table 1.** In vitro activity of compounds **1a,b**–**6a,b**.

\* % residual enzymatic activity at 100 µM; \*\* only for compounds with %RA ≤ 25%.

The modest biological activity of the new derivatives prompted us to reconsider the furan as the best heterocyclic core, suggesting the critical nature of an appropriately substituted furan moiety to maintain a significant enzymatic inhibition and to achieve antitubercular activity.

On this basis, and considering our previous results on disubstituted derivatives, we decided to design and synthesize six new furan-based analogues (**IV**–**IX**, Figure 4). In particular, **IV** was investigated as the isomer of 5-(2-cyano-4-(trifluoromethyl)phenyl)furan-2-carboxylic acid, which exhibited an interesting inhibitory effect (IC<sup>50</sup> of about 18 µM) [13]. The new analogue **IV** bears the CN and CF<sup>3</sup> moieties in the relative *meta* positions to avoid the steric interactions between the two adjacent groups, which had proven to be detrimental for the activity [13]. Compound **V** was synthesized to evaluate the role of the 3-CN moiety, as our past works had shown that this group was superior to other substituents in terms of enzymatic activity [12,13].

**Figure 4.** Chemical structure of compounds **IV**–**IX**.

Finally, compounds **VI**–**IX** were prepared to examine the influence of the substituent at position 5 of the phenyl ring on the biological activity of this class of compounds.

This strategy gave us the opportunity to examine a hitherto unexplored position of the phenyl ring (**IV**–**IX**), leading to the discovery of novel MbtI inhibitors endowed with antimycobacterial activity.

#### **2. Results**

#### *2.1. Chemistry*

The synthetic procedures adopted for the preparation of the compounds are heterogeneous and reflect the diverse approaches needed for the obtainment of the various heterocyclic derivatives. Where possible, the synthetic strategies were designed and optimized to afford the desired compounds, starting from the same commercially available reagents. All the compounds were characterized by means of mono- and bi-dimensional NMR techniques, FT-IR, ESI-MS, and elemental analysis. The procedures for the synthesis of series A and B (compounds **1a,b**–**6a,b**) are depicted in Schemes 1–6; all details regarding procedures and analytical data are reported in the Supplementary Materials.

**Scheme 1.** Synthetic procedure for the preparation of **1a,b**. Reagents and conditions: (**a**) MeOH, conc. H2SO<sup>4</sup> , reflux, overnight; (**b**) Pd(PPh<sup>3</sup> )2Cl<sup>2</sup> , 2 M Na2CO<sup>3</sup> , dry 1,4-dioxane, 90 ◦C, overnight, N<sup>2</sup> atm; (**c**) *1.* LiOH, THF-H2O 2:1, r.t., 2 h for **1a**; 1 M NaOH, EtOH-THF 1:1, reflux, 5 h for **1b**; *2.* 1 M HCl, 0 ◦C.

α

**Scheme 2.** Synthetic procedure for the preparation of **2a,b**. Reagents and conditions: (**a**) NBS, *p*-TsOH, DCM, overnight, r.t, N<sup>2</sup> atm.; (**b**) *1*. hexamine, DCM, 8 h, r.t.; *2*. conc. HCl, EtOH, overnight, r.t.; (**c**) TEA, EtOAc, 3 h, reflux; (**d**) Lawesson's reagent, 1,4-dioxane, 2 h, reflux; (**e**) NaOH, THF-H2O 1:1, 1.5 h, r.t.

**Scheme 3.** Synthetic procedure for the preparation of **3a,b**. Reagents and conditions: (**a**) I<sup>2</sup> , DMSO, 3 h, 130 ◦C; (**b**) NaOH, THF-H2O 1:1, 1.5 h, r.t.

**Scheme 4.** Synthetic procedure for the preparation of **4a,b**. Reagents and conditions: (**a**) SeO<sup>2</sup> , 1,4-dioxane/H2O, reflux, 7 h, N<sup>2</sup> atm; (**b**) NH4OAc, CH3CN, H2O, r.t., 2 h; (**c**) *1.* LiOH, THF-H2O 2:1, r.t., overnight for **4a**; NaOH, THF-H2O 1:1, reflux, 6 h for **4b**; *2.* 3 M HCl, 0 ◦C.

∙ **Scheme 5.** (**A**) Synthetic procedure for the preparation of **5a**. Reagents and conditions: (**a**) dry MeOH, conc. H2SO<sup>4</sup> , reflux, 3 h, N<sup>2</sup> atm; (**b**) NH2NH<sup>2</sup> ·H2O, MeOH, r.t., overnight; (**c**) TEA, DCM, r.t., 2 h; (**d**) TEA, DCM, TsCl, r.t., 2 h; (**e**) *1.* NaOH, THF-H2O 1:1, r.t., 1 h; *2.* Amberlite IR120, 0 ◦C. (**B**) Synthetic procedure for the preparation of **5b**. Reagents and conditions: (**a**) EtOH, conc. H2SO<sup>4</sup> , reflux, overnight; (**b**) NH2NH<sup>2</sup> ·H2O, EtOH, reflux, overnight; (**c**) 86% PPA, 120-130 ◦C, 1.5 h; (**d**) *1.* LiOH, THF-H2O 1:1, r.t., 1 h; *2.* Amberlite IR120, 0 ◦C.

∙

∙

**Scheme 6.** Synthetic procedure for the preparation of **6a,b**. Reagents and conditions: (**a**) *1*. NaN<sup>3</sup> , Cu(OAc)<sup>2</sup> , MeOH, 55 ◦C, 1.5-4 h, N<sup>2</sup> atm; *2.* ethyl propiolate, (+)-sodium <sup>L</sup>-ascorbate, r.t., overnight-24 h; (**b**) *1*. LiOH, THF-H2O 2:1, r.t., 1 h for **6a**; NaOH, THF-H2O 1:1, reflux, 5 h for **6b**; *2.* 3 M HCl, 0 ◦C for **6a**; 1 M HCl, 0 ◦C for **6b**.

∙

The key intermediate (**7**) for the synthesis of **1a**,**b** was obtained through a Fischer– Speier esterification of the commercially available 5-bromo-2-thiophenecarboxylic acid. Then, a palladium-catalyzed Suzuki-Miyaura coupling led to **8a,b**, which were hydrolyzed to the corresponding acids (**1a,b**) under basic conditions (Scheme 1).

The synthesis of the thiazole-based derivatives (**2a,b**) started from the bromination of the appropriate acetophenone, leading to **9a,b**. The hydrochloride salts of the corresponding amines (**10a,b**), obtained through the Delépine reaction, were *N*-acylated with ethyl chlorooxoacetate to afford the corresponding amides (**11a,b**). The formation of the thiazole ring was performed using the Lawesson's reagent, leading to the esters **12a,b**, which were hydrolyzed under basic conditions and isolated as sodium salts (Scheme 2) [28].

Oxazole-based derivatives were obtained through an iodine-promoted formal [3+2] cycloaddition of methyl ketones to α-methylenyl isocyanides: in particular, ethyl isocyanoacetate was reacted with the suitable acetophenone to afford 2,5-disubstituted oxazole esters (**13a,b**). The intermediates were then hydrolyzed under basic conditions and isolated as sodium salts (Scheme 3) [29].

For the synthesis of **4a,b**, the suitably substituted geminal diols **14a,b** were obtained from the commercially available acetophenone in the presence of selenium dioxide. Then, these intermediates were reacted with ethyl 2-oxoacetate and ammonium acetate, affording the imidazole esters **15a,b**, which were finally hydrolyzed to the corresponding acids (**4a,b**) under basic conditions (Scheme 4) [30].

For the synthesis of the *m*-CN-substituted 1,3,4-oxadiazole derivative (**5a**), the commercially available 3-cyanobenzoic acid was converted to the corresponding methyl ester (**16a**) and reacted with hydrazine hydrate to afford the hydrazide **17a**. This intermediate was acylated with ethyl-chlorooxoacetate to **18a**, which was cyclized to **19a** using *p*-toluensulfonyl chloride in the presence of triethylamine. The oxadiazole ester was then hydrolyzed under basic conditions to give **5a** (Scheme 5A) [31].

Concerning the *p*-NO2-substituted 1,3,4-oxadiazole analogue (**5b**), the hydrazide **17b**, obtained as described above, was reacted with ethyl 2-nitroacetate in polyphosphoric acid to afford the ester **19b**, which was hydrolyzed to the corresponding acid (**5b**) under basic conditions (Scheme 5B) [32].

The 1,4-substituted triazole esters **20a,b** were obtained through a one-pot Huisgen cycloaddition, starting from the appropriate phenylboronic acid: the synthesis of the azide was followed by the addition of ethyl propiolate, leading to the desired intermediates. The final acids (**6a,b**) were obtained through a base-catalyzed hydrolysis of the ester function (Scheme 6).

The new furan derivatives **IV**–**IX** were synthesized according to previously published procedures [13]. **V** was obtained by the same approach adopted for **1a**,**b**, employing (3,5-bis(trifluoromethyl)phenyl)boronic acid and methyl 5-bromofuran-2-carboxylate in a traditional Suzuki–Miyaura reaction, followed by a hydrolysis of the ester function [13]. **IV** was obtained by reacting 3-bromo-5-(trifluoromethyl)benzonitrile with (5- (methoxycarbonyl)furan-2-yl)boronic acid in a microwave-assisted Suzuki-Miyaura coupling, followed by a base-catalyzed hydrolysis of the ester function [13]. Similarly, 3-bromo-5-fluorobenzonitrile, 3-bromo-5-methoxybenzonitrile, 3-bromo-5-methylbenzonitrile, and

3-bromo-5-hydroxybenzonitrile were used as starting compounds for **VI**, **VII**, **VIII**, and **IX**, respectively.

#### *2.2. Biological Studies*

To pursue our aim of investigating the role of the heterocyclic core in MbtI inhibition, we decided to compare two sets of data, derived from our previous leads: **I**, characterized by the presence of the *m*-CN group (series A), and **II**, bearing the less druggable *p*-NO<sup>2</sup> group, but capable of potently inhibiting MbtI (series B) [12]. Therefore, keeping the cyano and the nitro group in their original positions, we explored the effects of the variation of the five-membered ring on the activity against the enzyme. The results of the in vitro assays on compounds **1a,b–6a,b**, calculated as previously reported [12], are listed in Table 1.

As for the thiophene analogues, the biological tests showed that **1a** and **1b** are approximately equipotent, with 23% residual enzymatic activity at 100 µM (%RA). The corresponding thiazole derivatives **2a** and **2b** are weaker inhibitors, especially in the presence of the *p*-NO<sup>2</sup> substitution. In some cases, thiazoles have been identified as pan-assay interference compounds (PAINS) [33]. To exclude this possibility, we tested **2a** and **2b**, along with all the other compounds published herein, against the PAINS filters of four online-based services, namely FAF-Drugs4 [34], SmartsFilter (https://chiltepin.health. unm.edu/tomcat/biocomp/smartsfilter accessed on 12 January 2021), SwissADME [35], and Zinc Patterns (http://zinc15.docking.org/patterns/home/ accessed on 12 January 2021). Notably, none of the molecules were identified as potential PAINS. As for the oxazole derivatives, **3a** showed a negligible activity, while **3b** evidenced a modest activity. Although we had envisioned that the structural features of the imidazole ring could be beneficial to form interactions within the MbtI active site, derivatives **4a** and **4b** displayed only a weak activity. Finally, the replacement of the furan with the oxadiazole and triazole cores in **5a**–**b** and **6a**–**b**, respectively, afforded compounds devoid of any significant effect against MbtI.

The minimal inhibitory concentration (MIC99) of the derivatives exhibiting an IC<sup>50</sup> lower than 30 µM was determined against the nonpathogenic *M. bovis* BCG, in ironlimiting conditions (chelated Sauton's medium), using the resazurin reduction assay method (REMA). All of them displayed MIC<sup>99</sup> values greater than 250 µM, which did not represent a significant improvement with respect to the previous leads.

In light of these findings, and considering previous SAR data, we designed the new derivatives **IV**–**IX**. The furan ring was chosen as the central core of these compounds, because it proved to be the best option and an important portion of our pharmacophore model. This additional investigation was undertaken to explore a new position of the phenyl ring, with the final goal of identifying the structural requirements needed to improve the antitubercular potency of these compounds.

The disubstituted derivatives **IV**–**IX** were tested for their effects against the recombinant MbtI, prepared, and assayed as previously reported [12]; their in vitro activities are shown in Table 2.


**Table 2.** In vitro activity of compounds **IV**–**IX**.

\* % residual enzymatic activity at 100 µM; \*\* only for compounds with %RA ≤ 25%.

Compound **IV** was designed with the cyano group in the *meta* position because our past works had shown that this group was superior to other options in terms of enzymatic activity; moreover, it features a trifluoromethyl moiety in 5, which allowed us to explore a hitherto unconsidered substitution pattern on the phenyl ring. **IV** was found to be a potent MbtI inhibitor (≈ 1% RA at 100 µM), with an IC<sup>50</sup> of ≈ 15 µM. Compound **V**, carrying two trifluoromethyl moieties, was also effective against MbtI, though displaying a slightly higher IC<sup>50</sup> with respect to **IV** (≈ 19 µM vs. 15 µM). Then, we tested compounds **VI**–**IX**, maintaining the original cyano group in 3 and featuring different moieties in 5. Firstly, we assayed the fluorine-substituted compound **VI**, which showed an IC<sup>50</sup> value similar to that of **V** (IC<sup>50</sup> ≈17 µM vs. 15 µM). Compound **VII**, bearing a methoxy moiety in 5, displayed a comparable IC<sup>50</sup> with respect to **IV**. Derivatives **VIII** and **IX**, carrying the CH<sup>3</sup> and the OH groups respectively, showed weaker inhibitory properties compared to compound **IV** (IC<sup>50</sup> ≈ 29 µM and 33 µM) (see Table 2).

The four candidates exhibiting promising inhibitory properties (IC<sup>50</sup> < 30 µM) were tested for their antimycobacterial activities against the nonpathogenic *M. bovis* BCG, in iron-limiting conditions (chelated Sauton's medium), using the REMA method. In this assay, compound **IV** showed the best antimycobacterial activity, with a MIC<sup>99</sup> value of 125 µM.

Due to its better bactericidal activity, **IV** emerged as the best inhibitor out of this furan series: its halved MIC<sup>99</sup> compared to **I** (125 µM vs. 250 µM) highlighted the better drugability of this compound with respect to our previous candidates.

Therefore, we submitted compound **IV** to a kinetic analysis, which demonstrated the competitive nature of its inhibition against MbtI, with a K<sup>i</sup> value of 9.2 ± 0.7 µM (Figure 5).

**Figure 5.** Biological characterization of **IV**. (**A**) IC<sup>50</sup> determination of **IV** against MbtI activity. (**B**) Global reciprocal plot of data from MbtI steady-state kinetics analysis towards chorismic acid, in the presence of different concentrations of **IV** (50, 20, 10, 5, 1, and 0 µM). (**C**) MIC<sup>99</sup> determination of **IV** against *M. bovis* BCG growth.

#### **3. Discussion**

In this work, we first applied a bioisosteric replacement strategy introducing in our leads **I** and **II** structural modifications in the five-membered core to alter the compound's electronic distribution and lipophilicity, with the aim of improving the target engagement and the antimycobacterial activity.

Contrary to traditional bioisosteric principles, the biological profiles of the thiophene analogues were not comparable to those of the furan derivatives. Conversely, their inhibitory effect followed the general trend exhibited by the thiazole, oxazole, and imidazole derivatives, suggesting that other factors prevail over the bioisosterism of the two nuclei. Interestingly, a significant decline in the activity was observed in the oxadiazole- and triazole-based compounds. Regarding the substitution of the phenyl ring, the presence

of the *m*-cyano or *p*-nitro groups did not seem to impact significantly on the variations of the activity.

In previous work, we reported the cocrystal structure of MbtI in complex with **I** and described the key interactions of the compound within the active site of the enzyme. Briefly, **I** forms H-bonds through its carboxylic group with Tyr385, Arg405, and an ordered water molecule; the oxygen of the furan interacts with Arg405, while the phenyl ring forms a cation-π interaction with Lys438 and a Van der Waals contact with Thr361. Finally, the cyano group interacts with Lys205, a key amino acid involved in the first step of the catalytic reaction. While the diminished activity of the triazole ring may be justified by the absence of a heteroatom capable of accepting a H bond from Arg405, the formulation of a hypothesis to explain the superiority of the furan with respect to the remaining cores is more arduous. A computational analysis of the binding modes of the tested compounds did not reveal significant disparities with respect to that of the lead molecule (unpublished data), suggesting that other influencing elements must be involved. Similarly, an in-silico comparison of the physicochemical characteristics of the compounds did not lead to meaningful results. Despite the different heterocyclic nuclei impart modifications to the overall properties of the molecules, a correspondence between the alteration of a parameter and the biological activity could not be unequivocally established. Therefore, it is reasonable to assume that the superior activity of the furan derivatives cannot be merely linked to the variation of one single parameter, but rather it is the result of a much more complex intertwinement of unrelated minimal modifications. The inherent multifactorial nature of these processes makes it hard, and potentially misleading, to seek a simplistic, univocal interpretation of these results. Hence, it is our opinion that the biological activity, empirically detected with our assays, is the only meaningful parameter that should be considered while determining the best heterocyclic core for this class of compounds.

In addition, when working with mycobacterial enzymes, in vitro activity may not necessarily correlate with the efficacy against the mycobacteria; therefore, compounds showing a weaker inhibitory effect against the purified target could display a better activity against bacterial growth, for instance due to an improved membrane permeability. On these bases, the MIC of the most potent molecules (IC<sup>50</sup> < 30 µM) was determined against *M. bovis* BCG. Although, in some literature cases, the furan core was successfully replaced by other heterocycles to improve the cellular activity [36], this was not our case. None of the compounds belonging to series A and B exhibited improved antitubercular action, all of them having MIC<sup>99</sup> values greater than 250 µM.

Overall, these biological results confirmed the furan as the best heterocyclic moiety among the options explored in this study and prompted us to reconsider this ring as the best scaffold to gain MbtI inhibition and antimycobacterial activity. In this regard, new modifications to the phenyl ring were attempted to improve the biological profile of our compounds. In our previous work, we discovered that the *m*-CN substitution offered the possibility to achieve better results in terms of enzyme inhibition compared to the other groups [12]. Moreover, in the context of a previously published disubstituted series, we took into account 5-(2-cyano-4-(trifluoromethyl)phenyl)furan-2-carboxylic acid, which proved to possess an interesting inhibitory effect (IC<sup>50</sup> of about 18 µM). Therefore, we decided to explore the activity of its isomer 5-(3-cyano-5-(trifluoromethyl)phenyl)furan-2-carboxylic acid (**IV**, Figure 4), bearing the cyano group in the preferred position 3. Additionally, the relocation of the trifluoromethyl moiety to position 5 to avoid the steric interactions between adjacent groups allowed us to examine a hitherto unexplored substitution site on the phenyl ring. The new isomer revealed an interesting IC<sup>50</sup> value of about 15 µM. Following the same strategy, we synthesized compound **V**, an isomer of 5-(2,4-bis(trifluoromethyl)phenyl)furan-2-carboxylic acid (IC<sup>50</sup> of about 13 µM) [13]. The new analogue exhibited slightly lower inhibitory properties (IC<sup>50</sup> ≈ 19 µM) with respect to the parent compound, while maintaining a promising activity. Subsequently, to further explore the SAR of the 5-substituent, we synthesized and tested compounds **VI**–**IX**, bearing the CN group in 3.

The presence of different substituents at position 5 of compounds **IV**–**IX** did not seem to significantly affect their inhibitory activity against MbtI, with IC<sup>50</sup> ranging from 15 µM to 33 µM. By contrast, the 5-CF<sup>3</sup> group of **IV** was able to significantly ameliorate its antimycobacterial properties with respect to the lead **I**. Indeed, **IV** displayed a MIC<sup>99</sup> value of 125 µM, far better than that of **I** (250 µM).

Further biological studies demonstrated that **IV** was a competitive inhibitor of MbtI, exhibiting a K*i* of about 9 µM, roughly comparable to that of **I**.

Overall, these SAR observations demonstrated the essentiality of the furan core and the advantages of a 3,5-disubstitution of the phenyl ring to achieve a potent in vitro activity against MbtI and a significant antimycobacterial effect.

The improvement of the MIC of new compounds is a common goal in antitubercular drug discovery, as reported in a very recent review, which also supported the importance of the mycobactin biosynthetic pathway for the development of anti-TB agents [37]. The increased antitubercular activity of **IV** with respect to our previous leads opens new avenues for structural modifications towards improved candidates.

#### **4. Material and Methods**

#### *4.1. Chemistry*

All starting materials, chemicals, and solvents were purchased from commercial suppliers (Sigma-Aldrich, St. Louis, MI, USA; FluoroChem, Hadfield, UK; Carlo Erba, Cornaredo, Italy) and used as received. Anhydrous solvents were utilized without further drying. Aluminum-backed Silica Gel 60 plates (0.2 mm; Merck, Darmstadt, Germany) were used for analytical thin-layer chromatography (TLC), to follow the course of the reactions. Microwave-assisted reactions were carried out with a Biotage® Initiator Classic (Biotage, Uppsala, Sweden). Silica gel 60 (40–63 µM; Merck) was used for the purification of intermediates and final compounds, through flash column chromatography. Melting points were determined in open capillary tubes with a Stuart SMP30 Melting Point Apparatus (Cole-Parmer Stuart, Stone, UK). All tested compounds were characterized by means of mono- and bi-dimensional NMR techniques, FT-IR, and ESI-MS. <sup>1</sup>H and <sup>13</sup>C NMR spectra were acquired at ambient temperature with a Varian Oxford 300 MHz instrument (Varian, Palo Alto, CA, USA) or a Bruker Avance 300 MHz instrument (Bruker, Billerica, MA, USA), operating at 300 MHz for <sup>1</sup>H and 75 MHz for <sup>13</sup>C. Chemical shifts are expressed in ppm (δ), while *J*-couplings are given in Hertz. The full decoupling mode was employed for <sup>13</sup>C spectra when the relaxation times of the carbons did not allow for a sufficient resolution using the APT sequence. The 2D-NOESY sequence was employed to unambiguously assign the hydrogen signals, when appropriate. HMBC and HSQC analyses were performed to aid the assignment of <sup>13</sup>C NMR signals, when necessary. ATR-FT-IR spectra were acquired with a Perkin Elmer Spectrum One FT-IR (Perkin Elmer, Waltham, MA, USA), equipped with a Perkin Elmer Universal ATR sampling accessory consisting of a diamond crystal. Analyses were performed in a spectral region between 4000 and 650 cm−<sup>1</sup> and analyzed by transmittance technique with 28 scansions and 4 cm−<sup>1</sup> resolution. MS analyses were carried out with a Thermo Fisher (Waltham, MA, USA) LCQ Fleet system, equipped with an ESI electrospray ionization source and an Ion Trap mass analyzer; ionization: ESI positive or ESI negative; capillary temperature: 250 ◦C; source voltage: 5.50 kV; source current: 4.00 µA; multipole 1 and 2 offset: −5.50 V and −7.50 V, respectively; intermultipole lens voltage: −16.00 V; trap DC offset voltage: −10.00 V. The purity of the tested compounds was assessed by means of elemental analysis using a EuroVector EA 3000 CHNS-O analyzer (EuroVector, Pavia, Italy). All experimental values are within ± 0.40% of the theoretical predictions, indicating a ≥ 95% purity.

All synthetic procedures are reported in the Supplementary Materials (SM).

*5-(3-Cyanophenyl)thiophene-2-carboxylic acid (1a).* The compound was synthesized by a specific procedure, reported in SM. Aspect: white solid. Mp: 207 ◦C. TLC (DCM-MeOH 9:1): R<sup>f</sup> = 0.20. The following analytical data are referred to the sodium salt. <sup>1</sup>H NMR (300 MHz, DMSO-*d6*) δ 8.09 (t, *J* = 1.7 Hz, 1H, H6), 7.91 (ddd, *J* = 7.9, 2.0, 1.2 Hz, 1H,

H8), 7.70 (dt, *J* = 7.7, 1.4 Hz, 1H, H10), 7.57 (t, *J* = 7.8 Hz, 1H, H9), 7.48 (d, *J* = 3.7 Hz, 1H, H3), 7.21 (d, *J* = 3.7 Hz, 1H, H4) ppm. <sup>13</sup>C NMR (75 MHz, DMSO-*d6*) δ 164.80 (COO−), 149.77 (C2), 141.67 (C5), 136.20 (C5'), 131.04 (C8), 130.74 (C9), 130.17 (C10), 128.71 (C6, C3), 125.45 (C4), 119.01 (CN), 112.70 (C7) ppm. FT-IR (ATR) ν = 2235, 1578, 1537, 1450, 1397, 1335, 811, 789, 767, 682, 675 cm−<sup>1</sup> . Anal. calcd. for C12H6NNaO2S: C, 57.37; H, 2.41; N, 5.58; S, 12.76. Found: C, 57.48; H, 2.45; N, 5.61; S, 12.83.

*5-(4-Nitrophenyl)thiophene-2-carboxylic acid (1b).* The compound was obtained according to Procedure A (SM). Starting compound: methyl 5-(4-nitrophenyl)thiophene-2-carboxylate. Yield: 98%. Aspect: yellow solid. Mp: 189 ◦C. TLC (DCM-MeOH 9:1): R<sup>f</sup> = 0.20. The following analytical data are referred to the sodium salt. <sup>1</sup>H NMR (300 MHz, DMSO-*d6*) δ 8.21 (d, *J* = 9.0 Hz, 2H, H7,7'), 7.97 (d, *J* = 9.0 Hz, 2H, H6,6'), 7.57 (d, *J* = 3.7 Hz, 1H, H3), 7.22 (d, *J* = 3.7 Hz, 1H, H4) ppm. <sup>13</sup>C NMR (75 MHz, DMSO-*d6*) δ 164.05 (COO−), 152.14 (C2), 146.32 (C8), 141.46 (C5), 141.24 (C5'), 128.78 (C3), 126.96 (C4), 126.15 (C7,7'), 124.86 (C6,6') ppm. FT-IR (KBr) ν = 3435, 2920, 2550, 1927, 1664, 1622, 1514, 1450, 1365, 995, 833, 704 cm−<sup>1</sup> . ESI-MS (m/z) calcd for C11H6NNaO4S 270.99, found 204.71 [M-CO2Na]−. Anal. calcd. for C11H6NNaO4S: C, 48.71; H, 2.23; N, 5.16; S, 11.82. Found: C, 48.75; H, 2.27; N, 5.14; S, 11.71.

*Sodium 5-(3-cyanophenyl)thiazole-2-carboxylate (2a).* The compound was obtained according to Procedure B (SM). Starting compound: ethyl 5-(3-cyanophenyl)thiazole-2 carboxylate. Yield: 86%. Aspect: white solid. Mp: >300 ◦C (dec.). <sup>1</sup>H NMR (300 MHz, DMSO-*d6*) δ 8.20 (s, 1H, H4), 8.19 (t, *J* = 1.4 Hz, 1H, H6), 7.95 (ddd, *J* = 7.9, 1.9, 1.2, Hz, 1H, H10), 7.77 (dt, *J* = 7.7, 1.2, H8), 7.60 (dt, *J* = 7.9, 0.5 Hz, 1H, H9) ppm. <sup>13</sup>C NMR (75 MHz, DMSO-*d6*) δ 173.60 (C2), 161.61 (COO−), 140.73 (C4), 138.39 (C5), 133.51 (C5'), 131.81 (C8), 131.51 (C10), 130.85 (C9), 129.86 (C6), 118.85 (CN), 112.81 (C7) ppm. FT-IR (ATR) ν = 3354, 2235, 1663, 1641, 1578, 1440, 1407, 1366, 1110, 866, 806, 796 cm−<sup>1</sup> . Anal. calcd. for C11H5N2NaO2S: C, 52.38; H, 2.00; N, 11.11; S, 12.71. Found: C, 52.51; H, 2.02; N, 11.09; S, 12.75.

*Sodium 5-(4-nitrophenyl)thiazole-2-carboxylate (2b).* The compound was obtained according to Procedure A (SM). Starting compound: ethyl 5-(4-nitrophenyl)thiazole-2 carboxylate. Yield: 85%. Aspect: dark green solid. Mp: >300 ◦C (dec.). <sup>1</sup>H NMR (300 MHz, DMSO-*d6*) δ 8.29 (s, 1H, H4), 8.23 (d, *J* = 6.0 Hz, 2H, H7,7'), 7.93 (d, *J* = 6.0 Hz, 2H, H6,6') ppm. <sup>13</sup>C NMR (75 MHz, DMSO-*d6*) δ 174.68 (C2), 161.40 (COO−), 146.99 (C8), 141.97 (C4), 138.75 (C5'), 138.37 (C5), 127.62 (C7,7'), 124.86 (C6,6') ppm. FT-IR (ATR) ν = 3648, 3297, 3100, 2963, 1675, 1645, 1621, 1595, 1514, 1424, 1408, 1365, 1343, 1108, 847 cm−<sup>1</sup> . ESI-MS (m/z) calcd for C10H5N2NaO4S 272.21, found 205.78 [M-CO2Na]−. Anal. calcd. for C10H5N2NaO4S: C, 44.12; H, 1.85; N, 10.29; S, 11.78. Found: C, 44.31; H, 1.87; N, 10.34; S, 11.67.

*Sodium 5-(3-cyanophenyl)oxazole-2-carboxylate (3a).* The compound was obtained according to Procedure B (SM). Starting compound: ethyl 5-(3-cyanophenyl)oxazole-2 carboxylate. Yield: 89%. Aspect: grey solid. Mp: >300 ◦C (dec.). <sup>1</sup>H NMR (300 MHz, DMSO-*d6*) δ 8.17 (t, *J* = 1.6 Hz, H6), 7.99 (dt, *J* = 8.0, 1.6 Hz, 1H, H10), 7.74 (dt, *J* = 8.0, 1.6 Hz, 1H, H8), 7.66 (t, *J* = 8.0 Hz, H9) ppm. <sup>13</sup>C NMR (75 MHz, DMSO-*d6*) δ 161.87 (COO−), 157.88 (C2), 148.06 (C5), 131.96 (C9), 130.85 (C8), 129.69 (C5'), 128.69 (C10), 127.90 (C6), 125.07 (C4), 118.78 (CN), 112.74 (C7) ppm. FT-IR (ATR) ν = 3522, 3383, 2234, 1650, 1616, 1520, 1422, 1389, 1318, 1273, 1216, 965, 825, 817, 795 cm−<sup>1</sup> . ESI-MS (*m*/*z*) calcd. for C11H5N2NaO<sup>3</sup> 236.16, found 169.67 [M-CO2Na]−. Anal. calcd. for C11H5N2NaO3: C, 55.94; H, 2.13; N, 11.86. Found: C, 56.03; H, 2.15; N, 11.93.

*Sodium 5-(4-nitrophenyl)oxazole-2-carboxylate (3b).* The compound was obtained according to Procedure B (SM). Starting compound: ethyl 5-(4-nitrophenyl)oxazole-2 carboxylate. Yield: 80%. Aspect: pale yellow solid. Mp: >300 ◦C (dec.). <sup>1</sup>H NMR (300 MHz, DMSO-*d6*) δ 8.30 (d, *J* = 8.9 Hz, 2H, H7,7'), 7.96 (d, *J* = 6.0 Hz, 2H, H6,6'), 7.87 (s, 1H, H4) ppm. <sup>13</sup>C APT NMR (75 MHz, DMSO-*d6*) δ 162.28 (COO−), 157.95 (C2), 148.37 (C5), 147.06 (C8), 134.32 (C5'), 126.99 (C4), 125.31 (C7,7'), 124.92 (C6,6') ppm. FT-IR (ATR) ν = 3436, 2964, 1645, 1607, 1512, 1388, 1346, 1261, 1108, 854, 818 cm−<sup>1</sup> . ESI-MS (*m*/*z*) calcd. for C10H5N2NaO<sup>5</sup>

256.15, found 189.94 [M-CO2Na]−. Anal. calcd. for C10H5N2NaO5: C, 46.89; H, 1.97; N, 10.94. Found: C, 46.59; H, 1.98; N, 10.92.

*5-(3-Cyanophenyl)-1H-imidazole-2-carboxylic acid (4a).* The compound was synthesized through a specific procedure, reported in SM. Aspect: white solid. Mp: 165 ◦C. TLC (DCM-MeOH 7:3): R<sup>f</sup> = 0.42. <sup>1</sup>H NMR (300 MHz, DMSO-*d6*) δ 12.0-9.0 (broad s exch. D2O, 2H, NH<sup>2</sup> + ), 8.25 (s, 1H, H6), 8.16 (d, *J* = 7.8 Hz, 1H, H8), 7.99 (s, 1H, H4), 7.68 (d, *J* = 7.8 Hz, 1H, H10), 7.58 (t, *J* = 7.8 Hz, 1H, H9) ppm. The compound degrades in solution at room temperature, during the acquisition of the <sup>13</sup>C NMR spectrum. FT-IR (ATR) ν = 3205, 2228, 1666, 1601, 1514, 1473, 1426, 1403, 1334, 1130, 1089, 811, 780, 680 cm−<sup>1</sup> . ESI-MS (m/z) calcd. for C11H7N3O<sup>2</sup> 213.19, found 212.42 [M-H]−. Anal. calcd. for C11H7N3O2: C, 61.97; H, 3.31; N, 19.71. Found: C, 62.03; H, 3.35; N, 19.82.

*5-(4-Nitrophenyl)-1H-imidazole-2-carboxylic acid (4b).* The compound was obtained according to Procedure C (SM). Starting compound: ethyl 5-(4-nitrophenyl)-1*H*-imidazole-2-carboxylate. Yield: 66%. Aspect: red solid. Mp: 137 ◦C. TLC (DCM-MeOH 7:3): R<sup>f</sup> = 0.40. The following analytical data are referred to the sodium salt. <sup>1</sup>H NMR (300 MHz, DMSO-*d6*) δ 8.15-8.09 (m, 4H, H6,6', H7,7'), 7.76 (s, 1H, H4) ppm. <sup>13</sup>C NMR (75 MHz, DMSO-*d6*) δ 162.09 (COO−), 149.48 (C2), 145.46 (C8), 142.17 (C5), 137.58 (C5'), 125.45 (C7,7'), 124.19 (C6,6'), 117.67 (C4) ppm. FT-IR (ATR) ν = 3607, 3156, 1652, 1600, 1494, 1472, 1415, 1342, 1135, 1112, 992, 849, 751 cm−<sup>1</sup> . ESI-MS (m/z) calcd for C10H7N3O<sup>4</sup> 233.18, found 232.32 [M-H]−. Anal. calcd. for C10H6N3NaO4: C, 47.07; H, 2.37; N, 16.47. Found: C, 47.31; H, 2.39; N, 16.36.

*5-(3-Cyanophenyl)-1,3,4-oxadiazole-2-carboxylic acid (5a).* The compound was obtained according to Procedure D (SM). Starting compound: ethyl 5-(3-cyanophenyl)-1,3,4 oxadiazole-2-carboxylate. Yield: quantitative. Aspect: white solid. Mp: 222 ◦C (dec.). TLC (DCM-MeOH 7:3): R<sup>f</sup> = 0.42. The following analytical data are referred to the sodium salt. <sup>1</sup>H NMR (300 MHz, DMSO-*d6*) δ 8.35 (t, *J* = 1.8 Hz, 1H, H6), 8.28 (dt, *J* = 8.0, 1.4 Hz, 1H, H8), 8.06 (dt, *J* = 7.8, 1.4 Hz, 1H, H10), 7.79 (dt, *J* = 7.9, 0.7 Hz, 1H, H9) ppm. The compound degrades in solution at room temperature, during the acquisition of the <sup>13</sup>C NMR spectrum. FT-IR (ATR) ν = 3543, 3384, 2232, 1651, 1614, 1549, 1400, 1343, 1228, 1183, 1086, 812, 807, 679 cm−<sup>1</sup> . Anal. calcd. for C10H4N3NaO3: C, 50.65; H, 1.70; N, 17.72. Found: C, 50.71; H, 1.81; N, 17.87.

*5-(4-Nitrophenyl)-1,3,4-oxadiazole-2-carboxylic acid (5b).* The compound was obtained according to Procedure D (SM). Starting compound: ethyl 5-(4-nitrophenyl)-1,3,4 oxadiazole-2-carboxylate. Yield: quantitative. Aspect: pale yellow solid. Mp: 220 ◦C (dec.). TLC (DCM-MeOH 7:3): R<sup>f</sup> = 0.44. <sup>1</sup>H NMR (300 MHz, DMSO-*d6*) δ 13.7 (broad s exch. D2O, 1H, COOH), 8.30 (d, *J* = 8.9 Hz, 2H, H7,7'), 8.15 (d, *J* = 8.9 Hz, 2H, H6,6') ppm. <sup>13</sup>C NMR (75 MHz, DMSO-*d6*) δ 166.24 (COOH), 150.51 (C5), 136.98 (C5'), 131.14 (C7,7'), 124.15 (C6,6') ppm. FT-IR (ATR) ν = 2962, 2924, 2853, 1691, 1603, 1520, 1258, 1080, 1013, 789 cm−<sup>1</sup> . Anal. calcd. for C9H5N3O5: C, 45.97; H, 2.14; N, 17.87. Found: C, 46.02; H, 2.17; N, 17.96.

*1-(3-Cyanophenyl)-1H-1,2,3-triazole-4-carboxylic acid (6a).* The compound was synthesized through a specific procedure, reported in SM. Aspect: white solid. TLC (DCM-MeOH 9:1): R<sup>f</sup> = 0.14. <sup>1</sup>H NMR (300 MHz, DMSO-*d6*) δ 8.90 (s, 1H, H1), 8.44 (t, *J* = 2.0 Hz, 1H, H6), 8.30 (ddd, *J* = 1.2, 2.0, 8.1 Hz, 1H, H8), 7.91 (dt, *J* = 1.2, 8.1 Hz, 1H, H10), 7.77 (t, *J* = 8.1 Hz, 1H, H9) ppm. <sup>13</sup>C NMR (75 MHz, DMSO-*d6*) δ 177.11 (C2), 163.73 (COOH), 137.81 (C5'), 132.35 (C8), 131.66 (C9), 124.99 (C10), 124.28 (C6), 123.68 (C1), 118.30 (CN), 113.25 (C7) ppm. FT-IR (ATR) ν = 3389, 3091, 2235, 1589, 1557, 1536, 1403, 1343, 1312, 1021, 794 cm−<sup>1</sup> . Anal. calcd. for C10H6N4O2: C, 56.08; H, 2.82; N, 26.16. Found: C, 56.27; H, 2.91; N, 26.35.

*1-(4-Nitrophenyl)-1H-1,2,3-triazole-4-carboxylic acid (6b).* The compound was obtained according to Procedure A (SM). Starting compound: ethyl 1-(4-nitrophenyl)-1*H*-1,2,3 triazole-4-carboxylate. Yield: 91%. Aspect: white solid. Mp: 175 ◦C. TLC (DCM-MeOH 8:2): R<sup>f</sup> = 0.12. <sup>1</sup>H NMR (300 MHz, DMSO-*d6*) δ 13.4 (broad s exch. D2O, 1 H, COOH), 9.59 (s, 1H, H1), 8.46 (d, *J* = 7.0 Hz, 2H, H7,7'), 8.30 (d, *J* = 7.0 Hz, 2H, H6,6') ppm. <sup>13</sup>C NMR (75 MHz, DMSO-*d6*) δ 161.73 (COOH), 147.60 (C8), 141.69 (C2), 140.96 (C5'), 128.02 (C1), 125.90 (C7,7'), 121.72 (C6,6') ppm. FT-IR (ATR) ν = 3249, 3142, 3102, 3061, 2962, 2913, 2866,

1729, 1704, 1596, 1516, 1342, 1268, 1219, 1151, 1032, 982, 869, 854 cm−<sup>1</sup> . ESI-MS (m/z) calcd for C9H6N4O<sup>4</sup> 234.17, found 161.30 [M-CO2-N2] <sup>−</sup>. Anal. calcd. for C9H6N4O4: C, 46.16; H, 2.58; N, 23.93. Found: C, 46.25; H, 2.60; N, 23.97.

*5-(3-Cyano-5-(trifluoromethyl)phenyl)furan-2-carboxylic acid (IV).* The compound was synthesized according to a previously published procedure [13]. Yield: 70%. Aspect: white solid. Mp: 210 ◦C. TLC (DCM-MeOH 7:3): R<sup>f</sup> = 0.33. <sup>1</sup>H NMR (300 MHz, DMSO-*d6*) δ 13.59-13.10 (broad s. exch. D2O, 1H, COOH), 8.62-8.57 (m, 1H, H7), 8.41-8.33 (m, 2H, H11, H9), 7.52 (d, *J* = 3.7 Hz, 1H, H4), 7.38 (d, *J* = 3.7 Hz, 1H, H3) ppm. <sup>13</sup>C APT NMR (75 MHz, DMSO-*d6*) δ 159.45 (COOH), 152.84 (C5), 146.18 (C2), 132.17 (C7), 132.17-131.78- 131.34-130.90 (q, C10), 131.99 (C6), 129.14 (C9), 128.80-125.19-121.56-117.94 (q, CF3), 125.09 (C11), 120.09 (C3), 117.53 (CN), 114.28 (C8), 111.74 (C4) ppm. FT-IR (ATR): ν = 3130, 3082, 2960, 2917, 2849, 2237, 1688, 1578, 1519, 1455, 1438, 1410, 1344, 1274, 1256, 1163, 1133, 1114, 1085, 1032, 900 cm−<sup>1</sup> . ESI-MS (m/z) calcd for C13H6F3NO<sup>3</sup> 281.03, found 280.14 [M-H]−. Anal. calcd. for C13H6F3NO3: C, 55.53; H, 2.15; N, 4.98. Found: C, 55.47; H, 2.19; N, 5.01.

*5-(3,5-Bis(trifluoromethyl)phenyl)furan-2-carboxylic acid (V).* The compound was synthesized according to a previously published procedure [13]. Yield: 91%. Aspect: white solid. Mp: 168 ◦C. TLC (DCM-MeOH 7:3): R<sup>f</sup> = 0.39. <sup>1</sup>H NMR (300 MHz, DMSO-*d6*) δ 13.65-13.20 (broad s. exch. D2O, 1H, COOH), 8.41-8.36 (m, 2H, H7, H11), 8.12-8.07 (m, 1H, H9), 7.59 (d, *J* = 3.7 Hz, 1H, H3), 7.38 (d, *J* = 3.7 Hz, 1H, H4) ppm. <sup>13</sup>C APT NMR (75 MHz, DMSO-*d6*) δ 159.48 (COOH), 153.19 (C5), 146.03 (C2), 132.34-131.90-131.46-131.03 (q, C8), 132.00 (C6), 128.98-125.36-121.74-118.12 (q, CF3), 125.00 (C11), 122.14 (C9), 120.14 (C3), 111.73 (C4) ppm. FT-IR (ATR): ν = 2960, 2925, 2855, 1689, 1621, 1591, 1526, 1455, 1420, 1366, 1278, 1161, 1124, 1081, 1027, 896 cm−<sup>1</sup> . ESI-MS (m/z) calcd for C13H6F6O<sup>6</sup> 324.18, found 323.16 [M-H]−. Anal. calcd. for C13H6F6O3: C, 48.17; H, 1.87. Found: C, 48.02; H, 1.91.

*5-(3-cyano-5-fluorophenyl)furan-2-carboxylic acid (VI).* The compound was synthesized according to a previously published procedure [13]. Yield: quantitative. Aspect: white solid. Mp: 247 ◦C. TLC (DCM-MeOH 7:3): R<sup>f</sup> = 0.26. <sup>1</sup>H NMR (300 MHz, DMSO-*d6*) δ 13.38 (broad s exch D2O, 1H, COOH), 8.14 (t, *J* = 1.6, 1H, H7), 7.96 (dd, *J* = 7.8, 1.6 Hz, 1H, H11), 7.88 (dd, *J* = 6.0, 1.6 Hz, 1H, H9), 7.43 (d, *J* = 3.7 Hz, 1H, H4), 7.37 (d, *J* = 3.7 Hz, 1H, H3) ppm. <sup>13</sup>C APT NMR (75 MHz, DMSO-*d6*) δ 164.18-160.91 (d, CF), 159.53 (COOH), 153.20-153.16 (d, C5), 145.89 (C2), 133.05-132.93 (d, C6) 124.92-124.87 (d, C7), 120.19 (C3), 119.75-119.41 (d, C11), 117.72-117.68 (d, CN), 116.57-116.25 (d, C9), 114.46-114.32 (d, C8), 111.43 (C4) ppm. FT-IR (ATR): ν = 3113, 2916, 2849, 2663, 2575, 2231, 1675, 1594, 1519, 1435, 1308, 1214, 1173, 1028, 866, 809, 760 cm−<sup>1</sup> . ESI-MS (m/z) calcd for C12H6FNO<sup>3</sup> 231.18, found 230.50 [M-H]−. Anal. calcd. for C12H6F NO3: C, 62.34; H, 2.62. Found: C, 62.53; H, 2.51.

*5-(3-Cyano-5-methoxyphenyl)furan-2-carboxylic acid (VII).* The compound was synthesized according to a previously published procedure [13]. Yield: 80%. Aspect: white solid. Mp: 226 ◦C. TLC (DCM-MeOH 7:3): R<sup>f</sup> = 0.46. <sup>1</sup>H NMR (300 MHz, DMSO-*d6*) δ 13.42-13.19 (broad s. exch. D2O, 1H, COOH), 7.82 (t, *J* = 1.4 Hz, 1H, H7), 7.59 (dd, *J* = 2.5, 1.5 Hz, 1H, H11), 7.45 (dd, *J* = 2.5, 1.3 Hz, 1H, H9), 7.35 (d, *J* = 3.7 Hz, 1H, H3), 7.33 (d, *J* = 3.7 Hz, 1H, H4), 3.87 (s, 3H, CH3) ppm. <sup>13</sup>C NMR (75 MHz, DMSO-*d6*) δ 160.43 (C10) 159.61 (COOH), 154.2 (C5), 145.46 (C2), 132.08 (C6), 120.73 (C3), 120.20 (C7), 118.66 (CN), 117.65 (C9), 115.02 (C11), 113.72 (C8), 110.60 (C4), 56.49 (CH3) ppm. FT-IR (ATR): ν = 3116, 3086, 2926, 2574, 2229, 1693, 1608, 1594, 1572, 1515, 1461, 1427, 1305, 1215, 1167, 1033, 960 cm−<sup>1</sup> . ESI-MS (*m*/*z*) calcd for C13H9O<sup>4</sup> 243.05, found 242.28 [M-H]−. Anal. calcd. for C13H9F3O4: C, 54.56; H, 3.17. Found: C, 54.71; H, 3.23.

*5-(3-cyano-5-methylphenyl)furan-2-carboxylic acid (VIII).* The compound was synthesized according to a previously published procedure [13]. Yield: 90%. Aspect: white solid. Mp: 238 ◦C. TLC (DCM-MeOH 7:3): R<sup>f</sup> = 0.31 <sup>1</sup>H NMR (300 MHz, DMSO-*d6*) δ 13.20 (broad s exch D2O, 1H, COOH), 8.08-8.02 (m, 1H, H7), 7.93-7.88 (m, 1H, H11), 7.67-7.62 (m, 1H, H9), 7.33 (d, *J* = 3.6 Hz, 1H, H3), 7.28 (d, *J* = 3.6 Hz, 1H, H4), 2.39 (s, 3H, CH3) ppm. <sup>13</sup>C NMR (75 MHz, DMSO-*d6*) δ 159.56 (COOH), 154.49 (C2), 145.39 (C5), 140.84 (C10), 132.82 (C9), 130.66 (C6), 129.51 (C11), 125.61 (C7), 120.13 (C3), 118.82 (CN), 112.64 (C8), 110.06 (C4), 20.89 (CH3) ppm. FT-IR (ATR): ν = 3119, 2926, 2849, 2692, 2579, 2228, 1726, 1682, 1584, 1520, 1422, 1299,

1169, 1029, 858, 810, 760 cm−<sup>1</sup> . ESI-MS (m/z) calcd for C13H9NO<sup>3</sup> 227.063, found 226.35 [M-H]−. Anal. calcd. for C13H<sup>9</sup> NO3: C, 68.72; H, 3.99. Found: C, 68.93; H, 4.01.

*5-(3-cyano-5-hydroxyphenyl)furan-2-carboxylic acid (IX).* The compound was synthesized according to a previously published procedure [13]. Yield: 95%. Aspect: white solid. Mp: 280 ◦C (dec.). TLC (DCM-MeOH 7:3 and 3 drops of CH3COOH): R<sup>f</sup> = 0.44. <sup>1</sup>H NMR (300 MHz, DMSO-*d6*) δ 13.22 (broad s exch. D2O, 1H, COOH), 10.54 (broad s exch. D2O, 1H, OH), 7.71 (t, *J* = 1.5, 1H, H7), 7.48 (dd, *J* = 2.4, 1.5 Hz, 1H, H11), 7.32 (d, *J* = 3.6 Hz, 1H, H3), 7.26 (d, *J* = 3.6 Hz, 1H, H4), 7.14 (dd, *J* = 2.4, 1.5, 1H, H9) ppm. <sup>13</sup>C APT NMR (75 MHz, DMSO-*d6*) δ 159.50 (C10), 158.90 (COOH), 154.42 (C5), 145.27 (C2), 132.09 (C6), 120.10 (C3), 119.32 (C7), 118.93 (C9), 118.74 (CN), 115.92 (C11), 113.50 (C8), 110.08 (C4) ppm. FT-IR (ATR): ν = 3400, 3108, 2602, 2228, 1652, 1595, 1516, 1487, 1439, 1310, 1241, 1213, 1174, 1152, 1035, 963, 881, 816, 668 cm−<sup>1</sup> . ESI-MS (m/z) calcd. for C12H7NO<sup>4</sup> 229.19, found 228.29 [M-H]−. Anal. calcd. for C12H<sup>7</sup> NO4: C, 62.89; H, 3.08. Found: C, 62.53; H, 3.05.

#### *4.2. Biological Activities*

#### 4.2.1. MbtI Enzymatic Assays

Recombinant *M. tuberculosis* MbtI was produced and purified as previously reported [14]. Enzyme activity was determined at 37 ◦C, measuring the formation of salicylic acid by a fluorimetric assay, slightly modified from Vasan et al. [38]. Briefly, the reactions were performed in a final volume of 400 µL of 50 mM Hepes pH 7.5, 5 mM MgCl2, containing 1-2 µM MbtI, by the addition of chorismic acid, and monitored using a Perkin-Elmer LS3 fluorimeter (Ex. λ = 305 nm, Em. λ = 420 nm). Inhibition assays were performed in the presence of the compound at 100 µM (stock solution 20 mM in DMSO) and 50 µM chorismic acid. Where possible, compounds were tested both as free acids and sodium salts, providing analogous results. For compounds inhibiting by more than 75% the initial activity, IC<sup>50</sup> values were determined. To this end, the activity was measured at different compound concentrations, and values were calculated according to Equation (1), with Origin 8 software:

$$\mathbf{A}\_{[\mathbf{I}]} = \mathbf{A}\_{[\mathbf{0}]} \times \left( 1 - \frac{[\mathbf{I}]}{[\mathbf{I}] + [\mathbf{C}\_{50}]} \right) \tag{1}$$

where A[I] is the activity at inhibitor concentration [I] and A[0] is the activity in the absence of the inhibitor.

The K<sup>i</sup> was determined at different substrate [S] and compound concentrations, using Equation (2):

$$\mathbf{v}^{\top} = \frac{\mathbf{V\_{max}[\mathbf{S}]}}{[\mathbf{S}] + \mathbf{K\_m} \left(1 + \frac{[\mathbf{I}]}{\mathbf{K\_i}}\right)} \tag{2}$$

#### 4.2.2. MIC Determination

The minimal inhibitory concentration (MIC99) of the most active compounds was determined against *M. bovis* BCG in low-iron chelated Sauton's medium, by the 2-fold microdilution method in U-bottom 96-well microtiter plates, as previously reported [13]. To this purpose, cells were grown in 7H9 medium, sub-cultured in chelated Sauton's medium, and then diluted to an OD<sup>600</sup> of 0.01 in chelated Sauton's containing different concentrations of the test compound. After 15 days of incubation at 37 ◦C, the growth was evaluated by the resazurin reduction assay method (REMA). Thirty microliters of a 0.01% solution of filter-sterilized resazurin sodium salt were added to each well, and the microtiters were re-incubated at the same temperature for 24 h. The MIC was defined as the lowest concentration of the drug that prevented a change in color from blue to pink, which indicates bacterial growth.

#### **5. Conclusions**

In this paper, a SAR study on our series of MbtI inhibitors led to the identification of new candidates, endowed with a potent activity against the enzyme and encouraging bactericidal properties. For the new products, we described the design, synthesis, analytical characterization, and biological activity.

Firstly, two sets of compounds, series A and B, incorporating a variety of heterocyclic motifs were biologically evaluated against MbtI and in whole-cell assays against *M. bovis* BCG. This approach led to the disclosure of **1b, 3b**, and **4b** provided with a moderate activity (IC<sup>50</sup> in the range 18–27 µM), but low bactericidal effects. Overall, the obtained results confirmed that the furan core was a better scaffold to gain MbtI inhibition in comparison with several other heterocycles.

These findings provided the basis for the design of the new furan-based derivatives **IV**–**IX**, which were synthesized, characterized, and tested. The best compound of this series, **IV**, bearing the preferred cyano group at position 3 and a trifluoromethyl moiety in 5, showed a potent MbtI inhibitory effect (%RA ≈ 1%, IC<sup>50</sup> ≈ 15 µM, K*i* ≈ 9 µM), comparable to the previous lead **I**, but exhibiting an enhanced antimycobacterial action (MIC<sup>99</sup> = 125 µM), thus becoming one of the few potent MbtI inhibitors endowed with a promising antitubercular activity.

These observations justify the selection of **IV** as the new lead of our next optimization campaign, which will contribute to strengthen the perspectives of anti-TB drug discovery.

**Supplementary Materials:** The Supplementary Materials are available online at https://www.mdpi. com/1424-8247/14/2/155/s1.

**Author Contributions:** Conceptualization: F.M., M.M., and S.V. Synthesis and characterization of the compounds: E.P., M.M., G.C., A.G., and S.V. Biological experiments: L.R.C. and G.S. Financial resources: S.V., F.M., A.G., and E.P. F.M. supervised the whole study and wrote the paper. F.M., S.V., L.R.C., M.M., and D.B. reviewed and edited the manuscript. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** The data presented in this study are available within the article.

**Acknowledgments:** All authors would like to acknowledge the University of Milan for funding this work (Linea B). Moreover, they gratefully thank Giulia Gwen Ballabio and Matteo Catalano for their helpful contribution.

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

#### **References**


### *Article* **Liquid and Vapor Phase of Four Conifer-Derived Essential Oils: Comparison of Chemical Compositions and Antimicrobial and Antioxidant Properties**

**Stefania Garzoli 1,\* ,† , Valentina Laghezza Masci 2,† , Valentina Caradonna <sup>2</sup> , Antonio Tiezzi <sup>2</sup> , Pierluigi Giacomello <sup>1</sup> and Elisa Ovidi <sup>2</sup>**


**Abstract:** In this study, the chemical composition of the vapor and liquid phase of *Pinus cembra* L., *Pinus mugo* Turra, *Picea abies* L., and *Abies Alba* M. needles essential oils (EOs) was investigated by Headspace-Gas Chromatography/Mass Spectrometry (HS-GC/MS). In the examined EOs, a total of twenty-eight components were identified, most of which belong to the monoterpenes family. α-Pinene (16.6–44.0%), β-pinene (7.5–44.7%), limonene (9.5–32.5%), and γ-terpinene (0.3–19.7%) were the most abundant components of the liquid phase. Such major compounds were also detected in the vapor phase of all EOs, and α-pinene reached higher relative percentages than in the liquid phase. Then, both the liquid and vapor phases were evaluated in terms of antibacterial activity against three Gramnegative bacteria (*Escherichia coli*, *Pseudomonas fluorescens*, and *Acinetobacter bohemicus*) and two Grampositive bacteria (*Kocuria marina* and *Bacillus cereus*) using a microwell dilution assay, disc diffusion assay, and vapor phase test. The lowest Minimum Inhibitory Concentration (MIC) (13.28 mg/mL) and Minimal Bactericidal Concentration (MBC) (26.56 mg/mL) values, which correspond to the highest antibacterial activities, were reported for *P. abies* EO against *A. bohemicus* and for *A. alba* EO against *A. bohemicus* and *B. cereus*. The vapor phase of all the tested EOs was more active than liquid phase, showing the inhibition halos from 41.00 ± 10.15 mm to 80.00 ± 0.00 mm for three bacterial strains (*A. bohemicus*, *K. marina*, and *B. cereus*). Furthermore, antioxidant activities were also investigated by 2,2-diphenyl-1-picrylhydrazyl (DPPH) and 2,2′ -azinobis (3- ethylbenzothiazoline-6 sulfonic acid) diammonium salt (ABTS) assays, and a concentration-dependent antioxidant capacity for all EOs was found. *P. mugo* EO showed the best antioxidant activity than the other Pinaceae EOs. The four Pinaceae EOs could be further investigated for their promising antibacterial and antioxidant properties, and, in particular, α-pinene seems to have interesting possibilities for use as a novel natural antibacterial agent.

**Keywords:** *Pinus cembra* L.; *Pinus mugo* Turra; *Picea abies* L.; *Abies alba* M.; essential oil; chemical investigation; HS-GC/MS; antibacterial activity; antioxidant activity

#### **1. Introduction**

Since ancient times, plants are a source of different kinds of compounds that humans used for their numerous biological activities and as a source for drug development [1]. Nowadays, the studies on antioxidant and antimicrobial activities of natural products are of considerable interest due to the importance of identifying and characterizing new bioactive molecules for applications in different fields as food preservation and packaging, antibiotically resistance phenomenon, and plant diseases.

**Citation:** Garzoli, S.; Masci, V.L.; Caradonna, V.; Tiezzi, A.; Giacomello, P.; Ovidi, E. Liquid and Vapor Phase of Four Conifer-Derived Essential Oils: Comparison of Chemical Compositions and Antimicrobial and Antioxidant Properties. *Pharmaceuticals* **2021**, *14*, 134. https://doi.org/10.3390/ph14020134

Academic Editor: Fiorella Meneghetti; Daniela Barlocco Received: 30 December 2020 Accepted: 4 February 2021 Published: 8 February 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

Among plant secondary metabolites, essential oils (EOs), biosynthesized by glandular trichomes and other secretory structures in plants, are liquids particularly rich in volatile molecules such as monoterpene and sesquiterpene hydrocarbons, oxygenated monoterpenes and sesquiterpenes, esters, aldehydes, ketones, alcohols, phenols, and oxides [2–4]. The chemical composition of EOs can vary from plant to plant and even in the same species and depends on several factors such as post-harvest conservation conditions [5], extraction methods [6–8] and times [9,10], microclimate, and site in which the plant is growing [11,12]. EOs contribute to the plant relations with environment and with other organisms, and humans and animals take advantage of the abundance of such bioactive molecules from the plant kingdom [4]. Numerous papers deal with the biological activities of the EOs such as antioxidant and anti-inflammatory properties, antibacterial and antifungal activities, immunomodulatory effects, and cytotoxic activities against different cancer cell lines [13–19]. Gymnosperms, the Cupressaceae, and the Pinaceae families produce economically important EOs [20]. The Pinaceae family is the largest family of non-flowering seed plants and comprises 11 genera and approximately 230 species of trees, rarely shrubs, which are widely distributed in the Northern Hemisphere [21,22]. The biological activities of Pinaceae EOs reflect the richness in their chemical composition. Antioxidant, antibacterial, antifungal, insect larvicidal, anti-inflammatory, and antiproliferative activities are reported for different genus of the Pinaceae family [16,23–30].

In our searching and studying of natural compounds, in the present paper, we investigated and compared the chemical composition and the antimicrobial and antioxidant properties of the vapor and liquid phase of four Pinaceae EOs from *Pinus cembra* L. and *Pinus mugo* Turra, which belong to the Pinus genus, and *Picea abies* L. and *Abies alba* M., which belong to the Picea and Abies genus, respectively.

#### **2. Results**

#### *2.1. Liquid and Vapor Phases EOs Chemical Composition*

By Gas Chromatography-Mass Spectrometry (GC/MS) and Headspace (HS)-GC/MS analysis, the composition of the vapor and liquid phase of all EOs was described. Twenty components were identified in *P. cembra* and *P. mugo* EOs, and they are listed in Table 1. The most abundant component was α-pinene (44.0% when using GC/MS and 65.6% when using HS/GC-MS) followed by γ-terpinene (19.7% GC/MS; 11.0% HS/GC-MS), limonene (14.8% GC/MS and 8.2%; HS/GC-MS) and β-pinene (12.5% GC/MS; 12.4% HS/GC-MS) in *P. cembra* EO. On the contrary, β-pinene (43.3% GC/MS; 42.3% HS/GC-MS) was the major compound in *P. mugo*. EO followed by α-pinene (16.6% GC/MS; 31.6% HS/GC-MS) and limonene (9.5% GC/MS; 7.8% HS/GC-MS). β-Phellandrene (16.0%) as well as other minor compounds such as p-cymenene (0.1%), copaene (0.1%), and bornyl acetate (3.0%) appeared only in the liquid phase of *P. mugo* EO. On the other hand, α-phellandrene (0.7%) was detected only in *P. mugo* vapor phase EO. In particular, in the vapor phase of both EOs, the components from N◦ 11 to N◦ 20 were missing except for β-caryophyllene (0.1%), which was detected in *P. mugo* EO.

Twenty-one components were identified in *P. abies* and *A. alba* EOs, and they are listed in Table 2. β-Pinene was the principal compound in *P. abies* EO (20.2% when using GC/MS and 34.5% when using HS/GC-MS), while α-pinene (30.8% GC/MS; 51.3% HS/GC-MS) was the principal compound in *P. abies* EO. The second most abundant component was α-pinene (20.2% and 34.5%) in *P. abies* EO and limonene (32.5% and 19.0%) in *A. alba* EO when using GC/MS and HS/GC-MS, respectively. p-Cymene (0.2%; 0.1%), camphor (1.2%; 0.2%), and borneol (2.1%; 0.2%) were detected only in the liquid and vapor phase, respectively of *P. abies* EOs. α-Himachalene (0.3%), citronellol acetate (0.4%), humulene (1.6%), and caryophyllene oxide (0.1) appeared only in the liquid phase of *A. alba* EO. Lastly, in the vapor phase of both EOs, the components from N◦ 12 to N◦ 21 were missing except for borneol (0.2%), which was detected in *P. abies* EO.


**Table 1.** Chemical composition (%) of liquid and vapor phases of *P. cembra* and *P. mugo* EOs.

<sup>1</sup> The components are reported according their elution order on a polar column; <sup>2</sup> Linear retention indices measured on polar column; <sup>3</sup> Linear retention indices from the literature; <sup>4</sup> Percentage values of *P. cembra* EO components (%); <sup>5</sup> Percentage values of *P. cembra* EO components (vapor phase); <sup>6</sup> Percentage mean values of *P. mugo* EO components (%); <sup>7</sup> Percentage mean values of *P. mugo* EO components (vapor phase); -Not detected; tr: traces (mean value < 0.1%).

**Table 2.** Chemical composition (%) of liquid and vapor phases of *P. abies* and *A. alba* essential oils (EOs).


<sup>1</sup> The components are reported according their elution order on polar column; <sup>2</sup> Linear retention indices measured on polar column; <sup>3</sup> Linear retention indices from the literature; <sup>4</sup> Percentage mean values of *P. abies* EO components (%); <sup>5</sup> Percentage mean values of *P. abies* EO components (vapor phase); <sup>6</sup> Percentage mean values of *A. alba* EO components (%); <sup>7</sup> Percentage mean values of *A. alba* EO components (vapor phase); -Not detected; tr: traces (mean value < 0.1%).

> Among the most abundant compounds, particular attention was paid to α-pinene, as it always reached higher percentages in the vapor phase than in the liquid phase of the investigated EOs. The compared values are as follows: (44.0% vs. 65.6%), (16.6% vs. 31.6%),

α

γ

δ

(20.2% vs. 35.5%), and (30.8% vs. 51.3%) liquid and vapor phase in *P. cembra*, *P. mugo*, *P. abies*, and *A. alba* EOs, respectively (Figure 1).

α

α **Figure 1.** Bar graph of α-pinene percentage trend in liquid and vapor phase EOs.

*2.2. Antibacterial Activities of P. Cembra, P. Mughus, P. Abies, and A. Alba EOs*

≤ The antibacterial activities of the Pinaceae EOs were evaluated for three Gram-negative (*Escherichia coli*, *Pseudomonas fluorescens*, and *Acinetobacter bohemicus*) and two Gram-positive bacteria (*Kocuria marina* and *Bacillus cereus* using micro dilution assay to determine Minimum Inhibitory Concentration (MIC) and the Minimal Bactericidal Concentration (MBC), and the MBC/MIC ratio defines an agent as bacteriostatic when the MBC/MIC ratio > 4 and as bactericidal when the MBC/MIC ratio ≤ 4 [31]. Furthermore, the disc diffusion assay by contact with the essential oil determined the diameter of bacterial growth inhibition zone (IZ), and the vapor phase test determined the antibacterial growth inhibition zone (Vapor IZ) by more volatile molecules of the EO in a preservative atmosphere. The antibacterial results of the tested EOs are summarized in Tables 3–6 reporting the MIC, MBC, MBC/MIC ratio, IZ, and vapor IZ following the treatments for each bacterial strain. In Table 3, the treatment with *P. cembra* EO showed MIC and MBC values of 53.12 mg/mL for *E. coli*, *P. fluorescens*, and *K. marina*, while MIC values were 26.56 mg/mL for *A. bohemicus* and *B. cereus*, and MBC values were 26.56 mg/mL and 53.12 mg/mL, respectively. MBC/MIC ratio defined the *P. cembra* EO as bactericidal against all bacterial strains. No effects were observed with the disc diffusion assay and with the vapor phase test for *P. cembra* EO against *E. coli* and *P. fluorescens*. The IZ and vapor IZ values were 17.67 ± 0.58 mm and 67.33 ± 2.52 mm for *A. bohemicus*, 9.33 ± 0.58 mm and 80.00 ± 0.00 mm for *K. marina*, and 11.67 ± 1.15 mm and 80.00 ± 0.00 mm for *B. cereus*, respectively.


**Table 3.** Antibacterial activity of *P. cembra* EO.

<sup>1</sup> Minimal Inhibitory Concentration expressed in mg/mL of EO treatment; <sup>2</sup> Minimal Bactericidal Concentration expressed in mg/mL of EO treatment; <sup>3</sup> Growth inhibition zone by disc diffusion assay expressed in mm; <sup>4</sup> Growth inhibition zone by vapor phase test expressed in mm. Values are expressed as means ± SD. *p* < 0.05 from one-way analysis of variance test (ANOVA).

Table 4 summarizes the antibacterial tests for *P. mugo* EO. MIC and MBC values were 52.16 mg/mL for *E. coli*, *P. fluorescens*, and *K. marina*, while MIC and MBC values were 26.08 mg/mL for *A. bohemicus* and MIC and MBC values were 26.08 mg/mL and

52.16 mg/mL for *B. cereus*, respectively. The MBC/MIC ratio defined as bactericidal the *P. mugo* EO against all bacterial strains. No effects were observed by the disc diffusion assay and by the vapor phase test for *P. cembra* EO on *P. fluorescens*. *P. mugo* EO was not highly active against *E. coli* with an IZ value of 9.67 ± 0.58 mm, while no growth inhibition zone was observed by the vapor phase test. Higher antibacterial activity was observed for the other bacterial strains: IZ and vapor IZ values were 25.33 ± 4.51 mm and 41.00 ± 10.15 mm for *A. bohemicus*, 11.33 ± 1.15 mm and 80.00 ± 0.00 mm for *K. marina*, and 15.67 ± 1.15 mm and 76.67 ± 5.77 mm for *B. cereus*, respectively. The *P. mugo* EO vapor phase was more active than the liquid phase against *A. bohemicus*, *K. marina*, and *B. cereus.*

**Table 4.** Antibacterial activity of *P. mugo* EO.


<sup>1</sup> Minimal Inhibitory Concentration expressed in mg/mL of EO treatment; <sup>2</sup> Minimal Bactericidal Concentration expressed in mg/mL of EO treatment; <sup>3</sup> Growth inhibition zone by disc diffusion assay expressed in mm; <sup>4</sup> Growth inhibition zone by vapor phase test expressed in mm. Values are expressed as means ± SD. *p* < 0.05 from one-way analysis of variance test (ANOVA).

**Table 5.** Antibacterial activity of *P. abies* EO.


<sup>1</sup> Minimal Inhibitory Concentration expressed in mg/mL of EO treatment; <sup>2</sup> Minimal Bactericidal Concentration expressed in mg/mL of EO treatment; <sup>3</sup> Growth inhibition zone by disc diffusion assay expressed in mm; <sup>4</sup> Growth inhibition zone by vapor phase test expressed in mm. Values are expressed as means ± SD. *p* < 0.05 from one-way analysis of variance test (ANOVA).

**Table 6.** Antibacterial activity of *A. alba* EO.


<sup>1</sup> Minimal Inhibitory Concentration expressed in mg/mL of EO treatment; <sup>2</sup> Minimal Bactericidal Concentration expressed in mg/mL of EO treatment; <sup>3</sup> Growth inhibition zone by disc diffusion assay expressed in mm; <sup>4</sup> Growth inhibition zone by vapor phase test expressed in mm. Values are expressed as means ± SD. *p* < 0.05. *p* < 0.05 from one-way analysis of variance test (ANOVA).

*P. abies* EO antibacterial activity is reported in Table 5. MIC and MBC values were 53.12 mg/mL for *E. coli*, *P. fluorescens,* and *K. marina*. For *A. bohemicus*, a lower MIC value was observed (13.28 mg/mL), whereas MBC was 26.56 mg/mL. The antibacterial activity

for *B. cereus* was 26.56 mg/mL and 53.12 mg/mL for the MIC and MBC values, respectively. As obtained by the MBC/MIC ratio, *P. abies* EO was bactericidal against all bacterial strains. No effects were observed by the disc diffusion assay and by the vapor phase test for *P. abies* EO on *P. fluorescens* and *E. coli*. IZ and vapor IZ values were 18.67 ± 1.53 mm and 76.67 ± 5.77 mm for *A. bohemicus*, 9.67 ± 1.15 mm and 80.00 ± 0.00 mm for *K. marina* and 11.67 ± 1.53 mm and 80.00 ± 0.00 mm for *B. cereus*, respectively. The *P. abies* EO vapor phase was more active rather than the liquid phase against *A. bohemicus*, *K. marina*, and *B. cereus*.

The results for *A. alba* antibacterial activity are reported in Table 6. MIC and MBC values were 51.28 mg/mL for *E. coli*, *P. fluorescens*, and *K. marina*. Lower MIC and MBC values were found for *A. bohemicus* and for *B. cereus* (12.82 mg/mL and 25.64 mg/mL, respectively). The MBC/MIC ratio defined as bactericidal the *A. alba* EO against all bacterial strains. No effects were observed in the disc diffusion assay and in the vapor phase test for *A. alba* EO on *E. coli* and *P. fluorescens*. Higher antibacterial activity was detected for the other bacterial strains: IZ and vapor IZ values were 19.67 ± 0.58 mm and 80.00 ± 00 mm for *A. bohemicus*, 7.67 ± 1.15 and 80.00 ± 0.00 for *K. marina*, and 15.00 ± 2.65 and 66.67 ± 11.55 for *B. cereus*, respectively. The vapor phase test revealed that the activities of the *A. alba* EO against *A. bohemicus*, *K. marina,* and *B. cereus* were higher than those of the liquid phase.

#### *2.3. Antioxidant Activity*

To determine the antioxidant activity of the four Pinaceae EOs, 2,2-Diphenyl-1 picrylhydrazyl (DPPH) scavenging activity and 2,2′ -azinobis (3- ethylbenzothiazoline-6-sulfonic acid) diammonium salt (ABTS) radical scavenging assay, based on the reaction of the potential antioxidant with colored radicals, were carried out. The antioxidant activity results are reported in Table 7. In all EOs, a concentration-dependent antioxidant capacity was found. In both tests, *P. mugo* EO showed the highest antioxidant activity than the other Pinaceae EOs. This EO exhibits lower IC<sup>50</sup> values (3.08 µg/mL and 43.08 µg/mL for DPPH and ABTS assays, respectively) and higher TEAC values (7.65 mol/mg and 14.01 mol/mg for DPPH and ABTS assays, respectively). The second effective essential oil was the *A. alba* EO with IC<sup>50</sup> values of 7.84 µg/mL and 44.23 µg/mL and TEAC values of 1.63 mol/mg and 13.26 mol/mg in the DPPH method and ABTS method, respectively. The TEAC values of *P. cembra* and *P. abies* EOs were almost identical: 1.63 mol/mg and 1.68 mol/mg in the DPPH assay, respectively and 13.26 mol/mg for both the EOs in the ABTS assay. Taking into account the ABTS test, the IC<sup>50</sup> amount was 44.90 µg/mL in *P. cembra* EO and 45.00 µg/mL in *A. alba* EO. In the DPPH test, both EOs showed similar IC<sup>50</sup> values, too (13.01 µg/mL for *P. cembra* and 13.05 µg/mL for *P. abies*). *A. alba* EO remains the least effective in antioxidant capacity of the analyzed Pinaceae EOs.



\* µg/mL of essential oil; \*\* µM of Trolox equivalent/mg of essential oil. Values are expressed as means ± SD. *p* < 0.05.

#### **3. Discussion**

The chemical profile of both the vapor and liquid phase and the antibacterial and antioxidant activities of four Pinaceae EOs, obtained from needles by steam distillation, were investigated using different kinds of techniques and assays. In the literature, a few papers reporting the Pinaceae EOs chemical composition are present, and no report describes the volatile composition of the vapor phase of the conifer-derived EOs by HS-GC/MS, as we

applied for our investigation. In our investigations, the chemical constituents resulted primarily monoterpenoids and their contents were higher in the vapor phases of *P. cembra* (99.9%) and *P. mugo* (100.0%) EOs than the vapor phases of the *P. abies* (97.9%) and *A. alba* (95.8%) EOs. The major compounds of the *P. cembra* EO were α-pinene (44.0%), γ-terpinene (19.7%), limonene (14.8%), and β-pinene (12.5%). Similar composition was described by Lis et al. [32], where the needle oil was dominated by α-pinene (48.4%), limonene (7.5%), and β-phellanderene (3.1%); α-pinene was also the major component (69.14%) in needle EO of *P. cembra* growing in Romania [33]. The composition of the EO from twig tips with needles of the *P. cembra* L. growing in Salzburg Alps was represented by α-pinene (43.9–48.3%), β-phellandrene (13.1–17.2%), and β-pinene (6.6–9.3%) [34]. *Pinus cembra* needles EO from Slovakia consisted of α-pinene (53.2%), limonene (11.4%), and β-phellandrene (9.4%) [35].

The main components of *P. mugo* EO were β-pinene (43.3%), α-pinene (16.6%), βphellandrene (16.0%), and limonene (9.5%) with a low percentage of β-caryophyllene (3.6%). A different composition was reported for *P. mugo* EO from needles growing in Poland where 3-carene (23.8 %), myrcene (22.3 %), and α-pinene (10.3 %) resulted as the main components [36]. 3-Carene (31.73%) was also the major compound in EO of *P. mugo* from Kosovo [37], followed by α-pinene (19.95%) and β-phellandrene (13.49%). *P.mugo* needles EOs from Macedonian [38] and Serbia [39] mainly consisted of ∆ 3 -carene (amount up to 35% and 23.9%), α- and β-pinene (up to 20% and 17.9%) and β-phellandrene (amount about 15% and 7.2%), respectively.

In *P. abies* EO, we found β-pinene (44.7%) as the most abundant component followed by α-pinene (20.2%), limonene (14.2%), and camphene (7.2%). A different composition has been described for the EOs from shoots of *P. abies* that grow wild in different locations of Romania, which are characterized by limonene (from 6.27% up to 12.98%), camphene (from 3.89% up to 14.07%), α-pinene (from 2.44% up to 10.42%), and β-myrcene (from 0.44% up to 10.12%) [40].

The chemical composition of *A. alba* EO showed two components such as limonene and α-pinene with a similar percentage (32.5% and 30.8%) followed by camphene (11.2%) and β-pinene (7.5%). The same compounds were listed with an inverted trend in *A. alba* EO from Montenegro where β-pinene (32.8%) was the major component followed by α-pinene (17.3%) and camphene (16.7%) [41]. On the contrary, α-limonene (about 70%) and α-pinene (57%) were the major compounds in *A. alba* EO from seeds and cones respectively [42]. In *A. alba* EO from Poland, limonene was the component with the higher percentage (82.9%) detected in seed EO, whereas α-pinene (50.0%) was the main component in cone EO [43]. According to the literature [44] and on the basis of the reported data, it becomes evident that the chemical composition of the EOs from species belonging to the Pinaceae family can depend by multiple factors such as part of the plant examined, its geographic origin, and also, extraction methods and storage [45].

MIC and MBC values defined by the microwell dilution method were tested against *E. coli*, *P. fluorescens* and *A. bohemicus* Gram-negative bacteria strains and *K. marina* and *B. cereus* Gram-positive bacteria strains. The lowest MIC (13.28 mg/mL) and MBC (26.56 mg/mL) values, which correspond to the highest antibacterial activities, were reported for *P. abies* EO against *A. bohemicus* and for *A. alba* EO against *A. bohemicus* and *B. cereus* with 12.82 mg/mL (MIC) and 25.64 mg/mL (MBC).

The increase of antibiotically resistance phenomenon in human and animal pathologies has determined the intensification of research on new natural antimicrobial substances [19,46,47], and in this view, several studies were carried out to investigate the biological activities of Pinaceae EOs and the roles of their molecules. *P. abies* EO extracted by supercritical carbon dioxide was investigated for antimicrobial properties on *E. coli* using the isothermal calorimetry technique, and it inhibited the growth and interfered with the metabolic activity of the microorganism [48]. Kartnig et al. [49] determined the antibacterial activities of the essential oils of young pine shoots on different bacterial strains also from human patients, and significant activities were revealed against G+ bacteria strains and *Candida* species tested. Apetrei et al. [25] reported that needles and twigs

essential oils of *Pinus cembra* showed high activity against *Sarcina lutea* and *Staphylococcus aureus* and no activity against *B. cereus*, *E. coli* and *Pseudomonas aeruginosa*.

The antibacterial activities of the four Pinaceae EOs were also confirmed by agar diffusion and disk volatilization methods by which the IZ and vapor IZ were measured in mm of inhibition halos. For all the tested EOs, the vapor phases were more active than the liquid phases, showing the inhibition halos from 41.00 ± 10.15 mm to 80.00 ± 0.00 mm for three bacterial strains (*A. bohemicus*, *K. marina*, and *B. cereus*). Concerning *E. coli* and *P. fluorescens*, a very low or null activity was reported. The results showed high activities of the EOs against *A. bohemicus*, *K. marina*, and *B. cereus* and a scarce or null activity against *E. coli* and *P. fluorescens*. The highest activities obtained by vapor phases of all EOs against *A. bohemicus*, *K. marina*, and *B. cereus* could be related with the presence of α-pinene. In the graph bar (Figure 1), the relative percentages of α-pinene were reported. It reached higher percentages in the vapor phase than in the liquid phase of all investigated EOs. In particular, liquid and vapor phase values were as follows: (44.0% vs. 65.6%), (16.6% vs. 31.6%), (20.2% vs. 35.5%), and (30.8% vs. 51.3%), in *P. cembra*, *P. mugo*, *P. abies*, and *A. alba* EOs respectively. These results suggest that α-pinene could play an important role for the detected antibacterial activity. Some papers reported α-pinene from Pinaceae EOs as the main compound showing good biological activity; it was the principal constituent (5.2–37.0%) in five Moroccan Pinus species EOs [50] and in *Pinus peuce* Griseb. EOs (12.89–27.34%) growing on three different locations in R. Macedonia [51].

Different studies confirmed the antibacterial properties of α-pinene [52]. Freitas et al. [53] reported that α-pinene has antibacterial and antibiotic-modulating activities against *S. aureus;* it also increases the activity of norfloxacin against *E. coli* and norfloxacin and gentamicin against *S. aureus*. Furthermore, Hippeli et al. [54] described an antiinflammatory potential of *P. mugo* EOs and its main compound α-pinene, while Cole et al. [55] showed anti-proliferative activity on the MCF-7 cell line. On the other hand, Kurti et al. [37] attributed the antimicrobial activities of some Pinus species EOs from Kosovo to the hexane/diethyl ether fractions, which were mainly composed by oxygenated monoterpenes.

In the present study, the susceptibility of bacteria does not seem to be related with the features of the cell Gram-positive and Gram-negative bacteria wall structure, since the more sensitive bacteria strains, *A. bohemicus*, *K. marina*, and *B. cereus* do not belong to the same group. Generally, Gram-negative are more resistant than Gram-positive bacteria, because the cell wall does not allow the entrance into the cell of hydrophobic molecules present in the essential oils [56,57], although some exceptions have been shown [58,59]. In a comparative study of the essential oils from four Pinus species [30], it was found that the sensitivity of the tested bacterial pathogens cannot be related with the cell wall structure. Different mechanisms of action can explain the EOs antimicrobial activities, and their wide variety of molecular components can act at multiple levels [60].

The DPPH and ABTS assays demonstrated a significant antioxidant activity for all Pinaceae EOs. *P. mugo* EO was the more active with an IC<sup>50</sup> 3.08 ± 0.65 and 43.08 ± 6.95 µg/mL for DPPH and ABTS assays, respectively. The values expressed in Trolox equivalent (TEAC) confirmed identical results. A comparative investigation has been carried out on *P. halepensis* EO chemical composition and antioxidant activities, with respect to the impact of geographic variation and environmental conditions [61].

The variety of compounds that are present in the investigated EOs confers them numerous biological properties, and their antioxidant activities could be related to the presence of monoterpenes. Wang et al. [62] studied the antioxidant activities of seven terpenoids found in wine, and among the tested compounds, α-pinene and limonene had the highest DPPH free radical scavenging and the highest reducing power. Wojtunik-Kulesza [63] reviewed the monoterpenes biological properties and antioxidant activities of α-pinene were also reported.

#### **4. Materials and Methods**

#### *4.1. Materials*

*P. cembra* L., *P. mugo* Turra, *A. alba* M., and *P. abies* L. bio essential oils (IT BIO 013 n◦ BZ-43509-AB) from needles growing in Alto Adige, Italy were obtained by steam distillation for 6 h extraction time and were directly provided by Bergila GmbH Srl (Falzes/Issengo-Bolzano). Methanol, 2,2-Diphenyl-1-picrylhydrazyl (DPPH), 6-hydroxy-2,5,7,8-tetramethylchroman-2-carboxylic acid (Trolox), 2,2′ -azinobis (3-ethylbenzothiazoline-6-sulfonic acid) diammonium salt (ABTS), potassium persulfate (K2S2O8), LB Broth with Agar and Thiazolyl Blue Tetrazolium Bromide (MTT) were from Sigma-Aldrich (Darmstadt, Germany). Gentamicin sulfate was purchased from Biochrom PAN-Bio-Tech GmbH (Aidenbach, Germany).

#### *4.2. Gas Chromatography–Mass Spectrometry (GC–MS) Analysis*

To describe the chemical composition of the EOs, a gas chromatograph with a flame ionization detector (FID) directly coupled to a mass spectrometer (MS) Perkin Elmer Clarus 500 model (Waltham, MA, USA) was used. The GC was equipped with a Restek Stabilwax (fused-silica) polar capillary column. Helium was used as carrier gas at a flow rate of 1 mL/min. The injector was set to a 280 ◦C, and the oven temperature program was as follows: isothermal at 60 ◦C for 5 min, then ramped to 220 ◦C at a rate of 6 ◦C min−<sup>1</sup> , and finally isothermal at 220 ◦C for 20 min. One uL of EO was diluted in 1 mL of methanol, and the injection volume was 1 µL. The Electron Impact-Mass Spectrometer (EI-MS) mass spectra were recorded at 70 eV (EI) and were scanned in the range of 40–500 *m*/*z*. The ion source and the connection parts temperature was 220 ◦C. The injector split ratio was 1:20. The GC-TIC mass spectra were obtained by the TurboMass data analysis software (Perkin Elmer). The identification of components was performed by matching their mass spectra with those stored in the Wiley and NIST 02 mass spectra libraries database. Furthermore, the linear retention indices (LRIs) (relative to C8–C30 aliphatic hydrocarbons, injected in the column at the same operating conditions described above) were calculated and compared with available retention data present in the literature. The relative percentages of all identified components were obtained by peak area normalization from GC-FID chromatograms without the use of an internal standard or correction factors and expressed in percentages. All analyses were repeated twice.

#### *4.3. Headspace GC-MS Analysis*

The volatile chemical profile of essential oils was carried out with a Perkin Elmer Headspace Turbomatrix 40 (Waltham, MA, USA) autosampler connected to GC-MS [64,65]. One mL of the each EO was placed in 20 mL vials sealed with headspace PTFE-coated silicone rubber septa and caps. To optimize the headspace procedure for the determination of volatile organic compounds (VOCs), more operative parameters were optimized. The gas phase of the sealed vials was equilibrated for 20 min at 60 ◦C and was followed immediately by compound desorption into GC injector in splitless mode. Quantification of compounds was performed by GC-FID in the same conditions described in the previous paragraph.

#### *4.4. Antibacterial Activities of the Pinaceae Essential Oils*

The antibacterial activities were investigated by using different methods, the Minimal Inhibitory Concentration (MIC), the Minimal Bactericidal Concentration (MBC), the agar diffusion method, and Vapor Phase Test (VPT).

#### 4.4.1. Bacterial Strains

Five bacterial strains from the culture collections of the Plant Cytology and Biotechnology Laboratory of Tuscia University were tested to evaluate the antibacterial activities of *P. cembra* L., *P. mugo* Turra, *A. alba* M., and *P. abies* L. essential oils: *Escherichia coli* ATCC 25922, *Pseudomonas fluorescens* ATCC 13525, and *Acinetobacter bohemicus* DSM 102855 among Gram-negative and *Kocuria marina* DSM 16420 and *Bacillus cereus* ATCC 10876 among Grampositive. All tested bacterial strains were maintained on LB broth (10 g tryptone, 5 g yeast extract, 10 g NaCl per liter, autoclaved at 121 ◦C for 20 min) with agar. Bacteria cultures were maintained at two different temperatures: 26 ◦C for *P. fluorescens*, *A. bohemicus,* and *B. cereus* and 37 ◦C for *K. marina* and *E. coli*. All inocula were prepared with fresh cultures plated the day before the test.

#### 4.4.2. Minimum Inhibitory Concentration (MIC)

The MIC is defined as the lowest concentration of antimicrobial agent that completely inhibits the growth of the microorganism as detected by the unaided eye and was carried out according to the microwell dilution method. Briefly, 12 dilutions of the four essential oils in LB broth (for *P. cembra* from 53.12 to 0.01 mg/mL; for *P. mugo* from 52.16 to 0.01 mg/mL; for *A. alba* form 51.28 to 0.01 mg/mL and for *P. abies* from 53.12 to 0.01 mg/mL), a control with the same percentage of DMSO (from 6.25% to 0.003%) in Lysogeny broth, a growth control without treatments, a positive control with gentamicin diluted from 100 to 0.05 µg/mL, and a sterility control without bacteria were plated on 96 microwell plates. Then, 50 µL of bacterial inoculum, 10<sup>6</sup> CFU/mL, were added in each well, except for the sterility control, and the plates were incubated for 24 h at the corresponding temperature. The visualization of the inhibition activity was obtained by 20 µL of a solution of 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (200 µg/mL, MTT) added to each well. The assay was carried out in triplicate. The MBC/MIC ratio was reported to interpret the activity of the essential oil, and an antimicrobial agent is considered bacteriostatic when the ratio MBC/MIC > 4 and bactericidal when the ratio MBC/MIC is ≤4 [31].

#### 4.4.3. Minimum Bactericidal Concentration (MBC)

To verify the lowest concentration at which the tested essential oils kill the bacterial cells, which is defined the Minimum Bactericidal Concentration (MBC), 10 µL of the last four dilutions from microwell dilution method in which no bacteria growth was observed were plated on a Petri plate with LB agar and incubated for 24 h. The concentration at which no growth on agar was observed defined MBC values. The assay was carried out in triplicate.

#### 4.4.4. Agar Diffusion Method

To determine the diameter of the halo inhibition of the bacteria growth induced by *P. abies*, *A. alba*, *P. cembra*, and *P. mugo* essential oils, the bacterial strains were suspended in LB broth to obtain a turbidity of 0.5 McFarland (approximately 10<sup>8</sup> Colony-Forming Unit/mL—CFU/mL) and then plated on LB broth with agar in a Petri plate. Sterile disks (6 mm diameter, Oxoid) were placed on the agar and impregnated with 10 µL of samples. Two µL of gentamicin from a stock solution (10 mg/mL) was used as a positive control After 24 h, the inhibitory activities of each essential oil were recorded as mm of halo diameter without growth [58] using a vernier caliper rule. The mean and the respective standard deviation (SD) of the measured halo in three independent experiments were recorded.

#### 4.4.5. Vapor Phase Test (VPT)

The antibacterial activity of the Pinaceae essential oils in the vapor phase was evaluated by the modified disk volatilization method [66,67]. LB agar were poured into an 80 mm plastic Petri dish and a lower amount into its cover. Each bacterial suspension containing 10<sup>8</sup> CFU/mL was plated on the LB agar medium. Then, 10 µL of tested essential oils were added to a 6 mm sterile disk and placed on agar in the covered Petri plate. Liquid LB agar was put in the space between the cover and the base of the Petri dishes to facilitate the sealing and to prevent any vapor leakage. The Petri plates were incubated for 24 h in an inverted position, and afterwards, the inhibition halos were measured. Negative controls were carried out without the essential. All VPTs were carried out in triplicate.

#### *4.5. Antioxidant Activity*

To assess the antioxidant activity of the four Pinaceae essential oils, DPPH radical scavenging activity and ABTS radical scavenging assay, which are based on the reaction of the potential antioxidant with colored radicals, were carried out.

#### 4.5.1. DPPH Scavenging Activity Assay

In DPPH radical scavenging assay, the Pinaceae essential oils antioxidant activities were calculated against the 1,1-diphenyl-2-picrilidrazil radical (DPPH•) using the method described by Sanchez-Moreno et al. [68]. First, 100 µL of fresh solution of a solid crystalline DPPH• (0.2 mM) in methanol were added to 100 µL of 12 geometric dilutions in methanol of each essential oil inside a 96-well plate. Geometric dilutions of the samples in methanol were used as sample blanks. In blank DPPH samples, essential oils were omitted. As a positive control, dilutions were prepared starting from Trolox solution (1 mM) in methanol. The samples were incubated for 30 min in the dark at room temperature, and the absorbances decreases were measured at 517 nm using a Tecan SunriseTM UV-vis spectrophotometer. The assay was repeated three times.

#### 4.5.2. ABTS Radical Scavenging Assay

The radical scavenging activities of the Pinaceae essential oils were also calculated using the ABTS (2,2′ -azino-bis (3-ethylbenzothiazoline-6-sulfonic acid) diammonium salt) assay described by Re et al. [69] with some modifications. The radical cation ABTS+• was produced by reacting ABTS aqueous solution (7 mM) with K2S2O<sup>8</sup> (140 mM) following an incubation for 16 h in the dark at room temperature before use. The ABTS+• solution was diluted with ethanol to reach an absorbance of 0.70 ± 0.02 at 734 nm, and 1980 µL was mixed with 20 µL of the essential oil dilutions in ethanol. The resulting solutions were incubated for 5 min at room temperature. Afterwards, the absorbances were measured at 734 nm using a Jasco V-630 UV-Visible spectrophotometer and using Spectra ManagerTM software. Furthermore, the absorbance of the ABTS+• blank, consisting of 20 µL of ethanol dissolved in 1980 µL of ABTS+• solution, was measured. The assay was repeated three times.

#### 4.5.3. IC<sup>50</sup> and TEAC Calculation

Trolox and samples calibration curves were obtained by plotting the inhibition ratio against sample concentrations. The inhibition ratio was calculated using the following formula:

$$IR\% = \frac{A \, blank - A \, sample}{A \, blank} \times 100.\tag{1}$$

The IC50 parameter was calculated using the sample calibration curve. A lower value of the IC50 parameter correspond to a lower concentration of the EO that can scavenge 50% of DPPH· molecules; therefore, it indicates a higher antioxidant activity.

The Trolox equivalent antioxidant capacity (TEAC) index was obtained from the ratio between the Trolox IC50 (µM) and the sample IC50 (mg/L):

$$TEAC\,\,=\,\frac{IC\_{50\_{trolx}}}{IC\_{50\_{sample}}}.\tag{2}$$

#### *4.6. Statistical Analysis*

The results were expressed as means ± standard deviation (SD). The one-way analysis of variance test (ANOVA) using GraphPad Prism software (GraphPad Prism 5.0, GraphPad Software, Inc., San Diego, CA, USA) was used to evaluate statistical discrepancies between the groups (*p* values < 0.05).

#### **5. Conclusions**

In this study, for the first time, the chemical composition of the liquid and vapor phase of four Pinaceae EOs was investigated by the HS-GC/MS technique. The results of analyses

showed that these EOs are rich in monoterpenoids and highlight that α-pinene, one of the main compounds, is more abundant in the vapor phase of each oil than in the liquid phase. The antimicrobial and antioxidant activities were also reported and compared. The vapor phase of each EO resulted more active against the investigated bacterial strains.

The biological effects of the Pinaceae EOs combined with their bioavailability makes them promising sources for possible application in different fields such as pharmacology, pharmacognosy, and phytochemistry.

**Author Contributions:** Conceptualization, S.G.; V.L.M.; E.O.; investigation, S.G., V.L.M.; V.C.; data curation, S.G.; V.L.M.; E.O.; writing—original draft preparation, S.G.; V.L.M.; E.O.; writing—review and editing, S.G.; A.T.; E.O.; supervision, P.G; A.T.; funding acquisition, P.G.; A.T. All the authors critically edited the manuscript before submission. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Not applicable.

**Acknowledgments:** The authors are thankful to Bergila, GmbH Srl (Falzes/Issengo-Bolzano) Italy, for providing *Pinus cembra* L., *Pinus mugo* Turra, *Picea abies* L. and *Abies Alba* M. essential oils.

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

#### **References**


### *Review* **Bacteriophage Therapy of Bacterial Infections: The Rediscovered Frontier**

**Nejat Düzgüne¸s 1,\* , Melike Sessevmez <sup>2</sup> and Metin Yildirim <sup>3</sup>**


**Abstract:** Antibiotic-resistant infections present a serious health concern worldwide. It is estimated that there are 2.8 million antibiotic-resistant infections and 35,000 deaths in the United States every year. Such microorganisms include *Acinetobacter*, Enterobacterioceae, *Pseudomonas*, *Staphylococcus* and *Mycobacterium*. Alternative treatment methods are, thus, necessary to treat such infections. Bacteriophages are viruses of bacteria. In a lytic infection, the newly formed phage particles lyse the bacterium and continue to infect other bacteria. In the early 20th century, d'Herelle, Bruynoghe and Maisin used bacterium-specific phages to treat bacterial infections. Bacteriophages are being identified, purified and developed as pharmaceutically acceptable macromolecular "drugs," undergoing strict quality control. Phages can be applied topically or delivered by inhalation, orally or parenterally. Some of the major drug-resistant infections that are potential targets of pharmaceutically prepared phages are *Pseudomonas aeruginosa*, *Mycobacterium tuberculosis* and *Acinetobacter baumannii*.

**Citation:** Düzgüne¸s, N.; Sessevmez, M.; Yildirim, M. Bacteriophage Therapy of Bacterial Infections: The Rediscovered Frontier. *Pharmaceuticals* **2021**, *14*, 34. https://doi.org/10.3390/ph14010034

Received: 22 November 2020 Accepted: 29 December 2020 Published: 5 January 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

**Keywords:** lytic infection; antibiotic-resistance; *Mycobacterium tuberculosis*; *Acinetobacter baumannii*; *Pseudomonas aeruginosa*; phage production; magistral phage; pulmonary delivery; oral administration; topical delivery

#### **1. Introduction: Bacteriophage Treatment of a Serious Infection**

A 68-year-old man with diabetes developed necrotizing pancreatitis that was complicated by a pancreatic pseudocyst infected with a multi-drug-resistant strain of *Acinetobacter baumannii* [1]. *A. baumannii* is a Gram-negative nosocomial pathogen involved in bacteremia, meningitis and pulmonary infections with a high mortality rate. It is one of the "ESKAPE" microorganisms that are grouped together because of the common occurrence of multi-drug-resistance in the group. These microorganisms include *Enterococcus faecium*, *Staphylococcus aureus*, *Klebsiella pneumoniae*, *Acinetobacter baumannii*, *Pseudomonas aeruginosa* and *Enterobacter* species. The condition of the patient was deteriorating rapidly despite antibiotic treatment, to which he was obviously not responding. Bacteriophage therapy was initiated as part of an emergency investigational new drug protocol.

Bacteriophages are viruses of bacteria. Phages can cause either lytic or lysogenic infections in bacteria after attaching to a receptor or receptors on the bacterial surface and delivering their genome into the bacteria. In a lytic infection, the phage replicates and the new phage particles lyse the bacterium and continue to infect other bacteria (Figure 1). In a lysogenic infection, a DNA phage inserts its genetic material into the bacterial chromosome, and the genome is passed on to daughter cells as the bacterium divides. The integrated DNA may be activated by changes in environmental conditions to excise itself from the chromosome, producing phage particles that become lytic [2–4].

**Figure 1.** The lytic infection cycle of a bacteriophage. A phage particle attaches to a receptor on the surface of a host bacterium and delivers its genome into the cytoplasm. Phage proteins and replicate genomes are synthesized and selfassemble into new phage particles that eventually lyse the bacterium. The phages then infect other bacteria with the particular receptor (reproduced with permission from Kortright et al., 2019 [4]).

A large number of phage types that could specifically lyse *A. baumannii* were tested against three strains of the bacterium obtained previously from the patient. The phages had been collected by the Biological Defense Research Directorate of the Naval Medical Research Center. Most of the phages, however, were not lytic towards the first clinical bacterial sample obtained from the patient. Six phage types inhibited bacterial growth for 20 h. Four of these phages were pooled and, when tested against the bacterial isolate from the patient, had superior activity compared to each of the phages by themselves. The patient showed clear clinical improvement within 2 days of the administration of the phage cocktail containing a total of 5 × 10<sup>9</sup> particles. Initially, the phage mixture was delivered through the percutaneous catheters draining the pseudocyst cavity, the gallbladder and the intra-abdominal cavity and repeated every 6 to 12 h. Thirty-six hours after the initiation of treatment, the phage cocktail was given intravenously [1]. This treatment reversed the patient's clinical decline, cleared the *A. baumannii* infection and returned the patient to health.

The British bacteriologist Ernest Hankin (1896) noted the anti-*Vibrio cholerae* activity of water from two rivers in India and suggested that a substance that could pass through porcelain filters caused this observation and possibly limited cholera epidemics.

The Russian bacteriologist Gamaleya reported a similar phenomenon with *Bacillus subtilis* [5]. The English bacteriologist Frederick Twort hypothesized in 1915 that the antibacterial effect could be mediated by a virus [6]. The French-Canadian microbiologist Felix d'Herelle, at the Pasteur Institute in Paris, identified non-bacterial microorganisms from the stools of patients suffering from severe hemorrhagic dysentery. These microorganisms formed plaques in cultures of *Shigella* isolated from the patients [7]. In 1919, d'Herelle used a phage preparation to treat a boy with dysentery who recovered within a day and then three additional patients who started to recover within a day [8,9]. The first report of phage therapy was published in 1921 by Richard Bruynoghe and Joseph Maisin [10]. They injected bacteriophages into and around surgically opened staphylococcal skin lesions, which regressed within 1–2 days.

#### **2. Phages as Pharmaceuticals**

#### *2.1. Phage Isolation and Enrichment*

The key processes in phage therapy protocols are phage selection and isolation. The wrong choices can have fatal consequences [11]. Generally, two methods are used when

choosing the appropriate phage for therapy: (1) A phage cocktail, such as Pyophage and Intestiphage. These preparations have a broader spectrum of activity than a single phage component and do not allow resistance to develop within a short time. (2) A pathogenspecific phage. Bacteria are isolated from the infection and tested for susceptibility to particular phages isolated previously [11].

Samples for phage isolation are taken from environments where the bacterial host can often be found, including soil, plant residues, fecal matter, wastewater and sewage (Figure 2). Phages against *Shigella dysenteriae* 2308 were isolated from the New York City sewage by Dubos et al. [12]. B\_VpS\_BA3 and vB\_VpS\_CA8 phages against *Vibrio parahaemolyticus* were isolated from sewage collected in China [13]. The vB\_KpnP\_IME337 phage against carbapenem-resistant *Klebsiella pneumoniae* was isolated from hospital sewage in China [14]. Li et al. [15] isolated 54 novel phages against the same organism from medical and domestic sewage wastewater. The newly isolated phage P545 had a relatively wide host range and strong antibacterial activity.

Although there are differences in phage isolation, the basic principle of the methods is the same as that developed by d'Herelle, and they are generally characterized as enrichment procedures [16]. First, the presence of phages is detected in the collected sample. Selection of the bacterial host is vital for the isolation of a phage in a recently acquired sample from the environment. Solid samples are mixed with sterile broth or buffer and then subjected to centrifugation and filtration [17]. Bacteria of interest are incubated overnight with the environmental sample. The bacteria that have survived the attack of the lytic phages are removed from the mixture by centrifugation or filtration, or both. The presence of phages in the filtrate is then assessed by plaque assay or by qPCR. The isolated phages have to be analyzed for their virulence, i.e., their ability to lyse target bacteria and the range of bacterial types they are able to infect. In an alternate method, samples from the environment are plated directly onto a lawn of particular bacteria and the presence of plaques resulting from bacterial lysis is detected. The latter method has been used to discover phages that lyse *Escherichia coli* and various bacteria from dental plaque and the oral cavity [17,18].

In the procedure described in detail by Luong et al. [19], the target bacterial strain is isolated and incubated with the phages. Then, after several agar plaque isolations, a single plaque is cultivated overnight. The isolated phage genome is sequenced to screen and identify lysogenic and deleterious genes. Phages are grown at liter scale, and the lysate is purified to eliminate any bacteria and cellular debris by pressure-driven filtration through filters of 0.8-, 0.45- and 0.22-µm pore size, followed by cross-flow ultra-filtration to eliminate debris smaller than 100 kD. This process eliminates endotoxin, exotoxins, peptidoglycan, nucleic acids and flagella. The phage particles are purified by CsCl density gradient centrifugation and dialysis to eliminate the CsCl. Any residual endotoxin molecules are removed by lipopolysaccharide-affinity chromatography. The last step ensures that the phage preparations do not cause inflammation or endotoxic shock when administered to patients [19].

Phages are expected to be found where the host bacteria reside. For example, phages that infect intestinal bacteria can be isolated from fecal material, and phages against epidermal bacteria such as *Staphylococcus aureus* are most likely isolated from skin samples or wound exudates. Identifying a phage against a particular bacterium is not straightforward, however. Whereas phages that lyse antibiotic-resistant *Klebsiella pneumoniae* and *Pseudomonas aeruginosa* were readily isolated from sewage samples, phages against antibiotic-resistant *Acinetobacter baumannii* were not found as frequently [20]. Furthermore, phages against methicillin-resistant *Staphylococcus aureus* were identified only rarely.

**Figure 2.** Stages of phage preparation. The environment (e.g., wastewater, farms and soil) is a source for all types of phages. The presence of phages in the tested sample is determined by different methods such as the double layer agar method, spot assays, the colorimetric method, the enrichment method or electron microscopy. The plaques that indicate lytic activity are picked up and transferred for the determination of phage type, specificity, etc. A phage lysate is prepared. At this stage, multiple procedures are performed to check for sterility (microbial contamination), toxicity (bacterial endotoxin or lipopolysaccharide (LPS) quantification), bacterial DNA contamination and phage titer. The purified phage preparation is stored.

The choice of a host for phage isolation may also depend on the ease of culturing the bacteria, as in the use of *Mycobacterium smegmatis* to isolate phages that will infect other *Mycobacterium* species. *M. smegmatis* grows much faster than *M. tuberculosis* and thus can produce a lawn on the appropriate agar surface for testing phage activity [17,20,21]. The isolated phages would then be tested further on the specific target *Mycobacterium* species.

Swanstrom et al. [22] investigated the variables contributing to the generation of high-titer phage stocks, using the agar layer method and coli phage T4r. The numbers of virus particles and bacteria per plate, the incubation period, the amount of soft agar in the agar layer as well as the broth volume used for virus extraction from the agar were found to be significant factors. When these factors were optimized, stock concentrations in the range of 1011–10<sup>12</sup> infectious particles/mL could be obtained [22].

Echeverría-Vega et al. [23] used a straightforward protocol for the isolation of bacteriophages from coastal organisms. They also validated the protocol for the isolation of lytic bacteriophages for the fish pathogen bacterium *Vibrio ordalii*. This method has particular utility for the recovery of bacteriophages for use as natural antimicrobial agents in aquacultures. In the enrichment method, samples are added to the host produced in a suitable medium, incubated and then centrifuged. The suspension containing the phage is filtered and applied at different concentrations onto an agar medium with target bacteria. The formed plaques are then counted. Thanks to the enrichment method, phages at low concentration can reach the desired level in culture. Enrichment is an advantageous method in cases where the amount of phages is low [24]. Numerous lytic phages were isolated against *Caulobacter* and *Asticcacaulis* bacteria using the enrichment method [25]. Methods such as spot testing, plaque testing, culture lysis and the calorimetric method are used in the detection of newly isolated bacteriophages (Figure 2) [26–31]. In the spot assay [26], bacteria are grown in Luria-Bertani broth, and after they are in the early log phase, they are mixed with soft agar and poured onto a Petri plate with previously poured agar. A phage filtrate is then placed on the soft agar and the plates are incubated overnight at 37 ◦C, after which bacterial lysis zones are counted [28]. In the double layer agar method, a bacterial culture in the log phase is mixed with a purified phage preparation and incubated briefly to allow for phage adsorption. This mixture is combined with soft agar and poured onto a previously solidified agar layer to form a homogeneous layer. After incubation at 37 ◦C for 24 h, plaque formation is observed, indicating phage activity. The plaques are resuspended in Mg (SM) buffer [28].

#### *2.2. Phage Production*

Bacteriophages need a host cell to reproduce. Understanding the interactions between host bacteria and bacteriophages is a crucial step in estimating the risks in production, including possible mutations in either microorganism [32]. The production process may also be affected by the nutrient composition, oxygenation, temperature and pH [33].

The substrate and temperature chosen for phage infection and bacterial growth are important factors. Fermentation is an important stage for host bacteria to multiply and produce bacteriophages. The sterilization step is performed to destroy undesired microorganisms. The bioreactor can be sterilized with heat, medium or a combination of these. During the fermentation process, the injected air is filtered through an in-line membrane. The air released after fermentation is filtered after it condenses [34].

Phages are grown basically in shaker flasks or stirred tank bioreactors. The latter are used to carry out industrial-scale production of bacteriophages, which has been divided into three different systems: batch, semi-continuous and continuous [35]. Each system has brought about its distinct benefits and drawbacks, discussed earlier by Merabishvili et al. [36]. Mancuso et al. [37] developed a production process that makes it possible to obtain high titers of *E. coli* T3 phages at high concentrations (10<sup>11</sup> PFU mL−<sup>1</sup> ) using two continuous stirred tank bioreactors. The first bioreactor is just for propagation of the host bacteria at a steady-state growth rate by using controllable dilution rates and growth-

limiting substrate (glucose). The second bioreactor is used for bacteriophage production and is fed from the host bacteria of the first bioreactor. Besides achieving high phage productivity of bacteriophages via the production process, the mutation risk of the host bacteria potentially caused by bacteriophages is suppressed.

#### *2.3. Phage Purification and Quality Control*

For the pharmaceutical application of phages, it is necessary to first carry out the purification process. The bacteriophage of interest is separated from host bacteria cells and debris by centrifugation, microfiltration or by using these methods together. The potential presence of any toxins in the preparations would be detrimental to the final pharmaceutical product. A Chamberland filter of 0.1–1 µm was used for bacteriophage preparations to be used in human trials [38]. It was recently clarified with a 0.2-µm filter pore size. Purification procedures of phages should follow the Critical Quality Attributes (CQA) specification [34]. The process of removing endotoxins from phages is complex because lipopolysaccharide forms micelles that have approximately the same size as phages. Therefore, extra purification methods such as ion exchange, affinity chromatography and solvent extraction are needed for lysates of phage-infected Gram-negative bacteria [33].

Endotoxin. In bacteriophage products, endotoxin measurement is critical. Gel clot, turbidimetric and chromatic methods are used for endotoxin determination in bacteriophage products. The *Limulus* amoebocyte lysate (LAL) assay is the most commonly used method [39].

Transmission electron microscopy (TEM). The specific morphology of phages in a final product can be viewed by transmission electron microscopy. Merabishvili et al. [40] used TEM for confirmation of the presence of the expected virion morphologic particles as well as their specific interaction with the target bacteria.

Titer. The process of determining phage concentration by dilution and plating with susceptible cells is called titering or the plaque assay. A bacteriophage capable of productively infecting a cell is named a plaque-forming unit (PFU/mL) [41].

pH. In a therapeutic formulation, the pH value is very important. According to the European Pharmacopoeia, the pH should be in the range 6.0–8.0 [42].

Nucleic Acid Contaminants. Because phages break down bacterial DNA, the presence and concentration of nucleic acid residues in final products should be determined. qPCR can be used for this purpose [33].

#### *2.4. Phage Stability and Storage Conditions*

Once solutions of phages are prepared, the biological properties of the phages have to be preserved during storage. Freeze-drying, spray-drying or encapsulation methods can be used to increase phage stability, as well as adding stabilizing additives to their solutions [43–46]. The quality, safety and storage conditions of phages to be prepared for use in treatment should be validated [47]. González-Menéndez et al. [48] investigated different preservation techniques for the storage of *Staphylococcus* phages (phiIPLA88, phiIPLA35, phiIPLA-RODI and phiIPLA-C1C). They evaluated the stability of phages at different temperatures (−20, −80 and −196 ◦C) and time periods (1, 6, 12 and 24 months). They also investigated various stabilization enhancing agents, including disaccharides, glycerol, sorbitol and skim milk. They showed that at −80 and −196 ◦C, all phages showed good viability after 24 months, regardless of the stabilizer [48].

#### *2.5. Therapeutic Phages*

Hyman et al. [17] proposed the following characteristics of phages to be used for therapeutic purposes: (a) The phage should be virulent and be able to cause complete cytotoxicity to the target bacterium. (b) It should be exclusively lytic and should not become temperate (i.e., lysogenic). (c) The phage should have the potential to transduce the host bacteria. (d) It should have the desired host range. (e) It should be screened for toxin genes that can affect the patient. *Myoviridae, Siphoviridae and Podoviridae* families are used com-

monly for phage therapy [49,50]. There are approximately 800 phages against pathogens such as *Escherichia*, *Morganella*, *Klebsiella*, *Enterococcus*, *Pseudomonas*, *Staphylococcus* and *Salmonella* [31].

#### *2.6. Magistral Phage*

A "magistral preparation" or a "compounded prescription drug product", in Europe and the U.S., respectively, is defined as "any medicinal product prepared in a pharmacy in accordance with a medical prescription for an individual patient" [33,34,48,51,52]. Such preparations for a particular patient are mixed by a pharmacist from their individual ingredients based on a prescription from a physician. The magistral formula is a practical way for a medical doctor to personalize patient treatments to specific needs and to make medications available that do not exist commercially. Some medicines, including natural hormone combinations, are made as magistral preparations. It is expected that magistral preparations will become more readily available as novel medicines are developed to treat rare conditions.

A magistral phage preparation is prepared from a phage bank, which is a repository of well-characterized microorganisms. A phage as an active pharmaceutical ingredient (API) is produced using a suitable bacterial host. An approved laboratory then carries out External Quality Assessments to test the API's properties and quality. Active phage APIs are evaluated for activity against the target. Finally, phage APIs are mixed with a suitable carrier system. There are currently no guidelines on the preparation, formulation and use of magistral phages [52].

#### *2.7. Topical Administration of Phages*

Several studies have shown that local and topical phage applications are successful. In the treatment of infections caused by Staphylococci, *Klebsiella*, *Pseudomonas*, *Proteus* and *Escherichia* such as conjunctivitis, otitis, gingivitis, furunculosis, decubitis ulcer, open wound infection, burns, osthitis (caused by fractures) and chronic suppurative fistulae, phage cocktails have been applied locally [53–56]. A commercial product called PhagoBioDerm, which targets *P. aeruginosa*, *S. aureus* and *Streptococcus* spp. and contains phages as well as cipro-floxacin, can be applied directly over infected wounds. Goode et al. [57] eliminated *Salmonella* contamination on chicken skin by using a lytic bacteriophage. Vieira et al. [58] performed phage therapy against multidrug-resistant *P. aeruginosa* that had caused skin infections. Thanks to phage therapy, the amount of *P. aeruginosa* 709 present in human skin decreased by four orders of magnitude.

#### *2.8. Pulmonary Phage Delivery*

The first studies of inhaled phage therapy were carried out in the early 1960s. Such treatments at a more advanced level were performed in Russia, Poland and Georgia. Although there have been many successful trials, some treatments have not had a positive outcome because of a lack of phage variety, quality control and technical knowledge [59]. Phages may be encapsulated in polymers, nanoparticles and liposomes for stability during storage, including storage as a freeze-dried preparation [60]. Liposome encapsulation was found to facilitate phage entry into macrophages. Treatment of experimental *K. pneumoniae*-induced lobar pneumonia was more effective with liposome-entrapped phages administered intraperitoneally as late as 3 days post-infection, whereas free phages provided a therapeutic effect only if they were administered at 1 day after infection [61]. Systemic side effects were reduced by the use of liposomal phage.

Liquid formulations using intranasal instillation and nebulization in phage studies against respiratory infections on animal models are quite popular. Liquid phage formulations are stable, easily aerosolizable and easy to formulate compared to other carrier systems [59]. Carrigy et al. [62] tested pre-exposure prophylactic aerosol delivery of the anti-tuberculosis bacteriophage D29 as an option for protection against *Mycobacterium tuberculosis* infection and proposed that mycobacteriophage aerosols at sufficient doses may be

protective against *M. tuberculosis* infection. The same group studied the titer reduction and phage delivery rate of three inhalation devices (Vibrating Mesh Nebulizer, Jet Nebulizer and Soft Mist Inhaler) with the mycobacteriophage D29 and showed that this method of administration is suitable for phage delivery to lung tissue [63]. Golshahi et al. [64] determined that the inhaled formulation of bacteriophages gives successful results in the treatment of cystic fibrosis pulmonary infections.

#### *2.9. Parenteral Phage Application*

Phages are rapidly eliminated by the immune system when administered intravenously. Lin et al. [65] investigated the intravenous administration of the anti-pseudomonal phage øPEV20 in *P. aeruginosa*-infected rats and demonstrated dose-dependent pharmacodynamics. Intravenous administration of øPEV20 at a dose of >10<sup>4</sup> PFU/mouse resulted in rapid bacterial killing and >8-log<sup>10</sup> CFU/mL reduction in bacterial load compared with the initial inoculum and untreated controls at 2.5 h. However, treatment at a dose of <10<sup>4</sup> øPEV20 PFU/mouse was ineffective against pan-drug-resistant *P. aeruginosa*. McVay et al. [66] injected a *P. aeruginosa* phage cocktail with three different administration methods (subcutaneous (s.c.), intramuscular (i.m.) and intraperitoneal (i.p.)) to *P. aeruginosa*-infected mice. Without treatment, the survival rate was 6%, and i.p. administration of phage resulted in the highest rate of survival (87%). According to the results of pharmacokinetic studies on phages, compared to other administration routes, phages reached the target at higher concentrations and faster when given via the i.p. route.

#### *2.10. Oral Phage Therapy*

Oral formulations of bacteriophages are generally used to target acute gastrointestinal infections. However, there are a number of factors for the treatment to be successful, including stability and effective phage dose at the site of infection. A significant decrease in phage titers occurs before the phages reach the site of infection. Phage viability and activity decrease as a result of gastric acidity and digestive enzymes such as pepsin and pancreatin. Thus, it is necessary to prepare new dosage forms. Vinner et al. [67] encapsulated enteric bacteriophage K1F against *E. coli* in a pH-responsive solid formulation and examined the viability of these bacteriophages at different pH values. They found that the microencapsulation process preserved phages for an extended period in the gastric acid environment. The encapsulated phages were active in killing *E. coli* co-incubated with human epithelial cells, which are normally stressed in the presence of the bacteria alone. There were no stability problems for the encapsulated phages that were refrigerated for 4 weeks. Stanford et al. [68] used polymer-encapsulated wV8, rV5, wV7 and wV11 phages, which are targeted to *E. coli*. Then, the phages were exposed to pH 3.2 for 20 min. The unencapsulated phages lost their activity while the encapsulated phages recovered 13.6% of their activity. Vinner et al. [69] prepared the encapsulated bacteriophage Felix O1, which is specific to *Salmonella*, by spray-drying, employing a commercially available pH-sensitive copolymer of methyl methacrylate and methacrylic acid. The inclusion of trehalose in the formulation protected the phages from the effects of spray-drying, maintaining the original phage titer. In a different approach, Colom et al. [70] encapsulated the phages UAB\_Phi20, UAB\_Phi78 and UAB\_Ph87 individually in a complex mixture of lipids that produced a net positive charge on the ensuing liposomes. The encapsulation efficiencies were relatively high, in the range of 47–49%, which is most likely the result of phage binding to the net cationic lipid mixture. The encapsulated phages were more effective than plain phages in *S. enterica* ser. Typhimurium-infected chickens at only 8 days following treatment, with a 3.9 log<sup>10</sup> reduction [70].

#### **3. Mycobacteriophage Therapy of** *Mycobacterium tuberculosis*

There are more than 170 *Mycobacterium* species that have great variety in terms of their pathogenicity in humans [71]. In addition to *M. tuberculosis*, *M. ulcerans* and *M. leprae* cause Buruli ulcer and leprosy, respectively [72]. *M. tuberculosis* is a well-known example

of an intracellular bacterium that localizes inside phagosomes of macrophages of the host and causes tuberculosis (TB), which primarily affects the lungs [73]. Multi-drug-resistant (MDR) TB cases have emerged in the late 1980s and early 1990s. These strains are resistant to the first-line drugs against TB, rifampicin and isoniazid. In 2018, the World Health Organization (WHO) reported that 484,000 new TB patients failed to respond to rifampicin. Seventy-eight percent of these patients were infected with MDR-TB [74].

Alternative treatment approaches for MDR-TB have become crucial to managing the disease. One of these approaches is mycobacteriophage therapy. More than 70 years have passed since mycobacteriophages were isolated for the first time [75]. So far, 11,282 mycobacteriophages have been isolated [76].

Bacteriophages can enter macrophages by four main routes [77] (Figure 3): (1) Endocytotic uptake of the bacteriophage alone; (2) entry into the macrophage via pathogenic bacteria together with the bound phage; (3) uptake of the bacteriophage and non-pathogenic bacteria; (4) internalization of the bacteriophage that has been encapsulated by poly-mers or liposomes. Relatively non-pathogenic vectors, such as *M. smegmatis,* can be used to deliver phages to the same intracellular compartments where *M. tuberculosis* is found [78]. The lytic mycobacteriophage TM4 was delivered in this manner to *M. tuberculosis*-infected RAW264.7 macrophages and reduced the bacterial counts. By contrast, the phage alone was ineffective. The administration of the *M. smegmatis*-TM4 complex to *M. avium*-infected mice significantly decreased the bacterial counts in the spleen, whereas TM4 or *M. smegmatis* alone had no effect [79]. The authors suggested that phage resistance (which was observed in their study) could be overcome by the use of phage cocktails.

**Figure 3.** Possible cellular entry pathways of bacteriophages. The pathways 1–4 are described in the text. Dark red hexagons, bacteriophage; brown-filled ovals, bacteria (pathogenic bacteria, P; non-pathogenic bacteria, N); curly lines within bacteria, bacteriophage nucleic acids; orange ovals, endosomal vesicles; blue-gray circles and M, microcapsules; dotted lines, degrading bacterial membrane (reproduced with permission from Nieth et al., 2015 [77]).

The mycobacteriophage D29 was used to treat *M. tuberculosis* H37Rv inside RAW 264.7 macrophages [80]. The phage, administered twice over a 24-h period, caused an eight-fold reduction in the CFUs, indicating that it was able to access the intracellular compartment occupied by the bacteria. The phage was encapsulated in (or associated with) liposomes comprising phosphatidylcholine, cholesterol and Tween-80 and which were sized by extrusion through membranes of 400-nm diameter. This formulation applied to infected macrophages resulted in a two-fold improvement of the antimycobacterial effect over that of the free phage. In an in vitro model of tuberculous granuloma developed from peripheral blood mononuclear cells of patients with TB, liposomal phage was about nine-fold more effective than free D29 [81].

Aerosolized bacteriophage D29 was used to investigate the possibility of protecting mice against *M. tuberculosis* infection [62]. This treatment significantly decreased the *M. tuberculosis* counts in the lungs 1 day and 3 weeks after challenge. The authors suggested that aerosolized mycobacteriophages may be useful in conferring additional protection to

healthcare workers who may be at risk of exposure to tuberculosis. D29 was also employed in a murine footpad model in treating Buruli ulcer, which is caused by *Mycobacterium ulcerans* [82]. In infected patients, the bacterium causes necrosis of the skin, subcutaneous tissue as well as bone. If the disease reaches advanced stages, surgical resection of the skin may be necessary. The subcutaneous injection of D29 resulted in a decrease in pathology and mycobacterial counts. It also caused increased production of cytokines, including IFN-γ, in the footpads and draining lymph nodes. Endolysins are bacteriophage-encoded peptidoglycan-disrupting enzymes synthesized at the last stage of the phage life cycle in the infected bacteria [83]. One endolysin, lysine B, was found to lyse *M. ulcerans* infecting the footpad of experimental mice [84].

Developing mycobacteriophages into efficient therapeutic pharmaceuticals has focused on improving their uptake into macrophages and co-localization with the intracellular mycobacteria. In this process, however, it is essential to maintain the stability of the formulation and the vitality of the mycobacteriophages. In the next step, well-established in vitro and in vivo studies of effective and stable mycobacteriophage formulations are expected to translate into clinical studies with successful outcomes.

#### **4. Bacteriophage Therapy of** *Pseudomonas aeruginosa*

*P. aeruginosa* are Gram-negative aerobic bacteria classified as Gammaproteobacteria that can cause severe necrotizing bronchopneumonia, burn wound infections, urinary tract infections, otitis externa, eye infections and bacteremia [85].

In a murine model of sepsis caused by *P. aeruginosa* via the gut, the lytic phage KPP10 administered orally increased the survival rate from 0% in the controls to 67% [86]. The number of viable bacteria in the liver, spleen and blood were reduced in the phagetreated group, as were the levels of inflammatory cytokines in the liver and blood. Imipenemresistant *P. aeruginosa* delivered i.p. resulted in bacteremia and killed 100% of experimental mice within 24 h [87]; the i.p. administration of the phage ØA392 within 1 h of infection was able to rescue all the animals. The phages were found in blood within 2 h. However, delivery of the phage at 3 h post-infection resulted in only 50% survival. In a murine burn-wound model, fatal infection by *P. aeruginosa* could be reduced to 87% survival when a three-phage cocktail was given i.p. [67]. The phages rapidly distributed to the blood, liver and spleen. In a similar study, i.p. delivery of multi-drug-resistant *P. aeruginosa* caused fatal bacteremia in mice within 2 d [88]. A phage strain that had lytic activity against numerous multi-drug-resistant *P. aeruginosa* given 45 min after bacterial infection resulted in 100% survival. Fifty percent of the animals could be saved even when the therapy was applied at a point where the animals were sick. The therapeutic effect of the phage was also shown not to be the result of a non-specific immune response.

When mice with acute lung infection with intranasally administered bioluminescent *P. aeruginosa*, which resulted in the death of all the animals within 2 days, were treated with bacteriophage PAK-P1-to-bacterium ratios of 1:1 and 10:1 via the same route, they survived until the end of the 12-day experiment [89]. Bacteriophage treatment also prevented lung infection when administered 24 h before inoculation of bacteria. Two phages were isolated from wastewater, the myovirus ϕNH-4 and the podovirus ϕMR299-2, and used to treat *P. aeruginosa* infection in murine lungs [90]. The pathogen was reduced by three to four orders of magnitude in 6 h. A mixture of the two phages could kill biofilms of mucoid and nonmucoid strains of *P. aeruginosa* on CFBE41o-cystic fibrosis bronchial epithelial cells, and the phages were shown to multiply over 24 h.

Phage GNCP treatment of multi-drug-resistant *P. aeruginosa* infection in diabetic and non-diabetic mice, which caused fatal bacteremia within 2 d, at a 10:1 ratio of phage:bacteria resulted in protection of 90% of diabetic animals and 100% of non-diabetic animals [91]. Bacteriophages were also effective in reducing inflammation in a murine acute infection model of *P. aeruginosa* [92]. The titer of phage PEV31 delivered intratracheally to mice without bacterial infection decreased with a t1/2 of about 8 h. In mice infected with *P. aeruginosa*, the phage titer increased by about two orders of magnitude in 16 h, and

bacterial growth was suppressed, whereas it increased exponentially in the untreated animals [93].

#### **5. Clinical Cases Treated with Bacteriophages**

Lung transplant recipients with life-threatening multi-drug-resistant *P. aeruginosa* or *Burkholderia dolosa* infections were treated with lytic bacteriophages targeting the bacterial strains, together with antibiotics [94]. Two patients with *P. aeruginosa* infection responded to the treatment and could leave the hospital. The patient with recurrent *B. dolosa* infection did not respond to bacteriophage therapy. The safety and feasibility of phage treatment of patients with various infections at a single center in the U.S. was established, although two of the 10 patients described did not respond to therapy [95].

*P. aeruginosa* can infect prosthetic vascular grafts that often do not respond to antibiotic therapy [96]. The bacteriophage OMKO1, together with ceftazidime, was used to treat infection of an aortic graft, which was resolved and did not recur.

Bacteriophage therapy was applied to chronic non-healing wounds that were infected with *E. coli, S. aureus* and *P. aeruginosa* and that did not respond to antibiotic therapy [97]. The application of a cocktail of bacteriophages over the wounds every other day resulted in the resolution of the infection after 3–5 doses. The wounds healed completely in seven out of 20 patients and formed healthy granulation tissue and margins in the other patients.

In a trial that included 48 patients with non-healing wounds, a single phage against a particular bacterial infection or multiple phages targeting multiple bacteria were applied every other day for 5 to 7 days [98]. The cure rate was 81%, with diabetic patients having a lower rate (74%).

A 65-year-old woman with a post-operative left-eye corneal abscess and interstitial keratitis was treated for many years with various antibiotics but remained positive for vancomycin-intermediate sensitivity *S. aureus* in the nasal cavity, skin and eye [99]. She then underwent topical and intravenous phage therapy with the bacteriophage SATA-8505. This phage strain is active against the methicillin-resistant *S. aureus* strain USA300 and has been patented. Ocular and nasal cultures from the patient 3 and 6 months after therapy showed no infection.

#### **6. Conclusions**

The importance of phage therapy for bacterial infections has been recognized by both academic institutions and the pharmaceutical industry. The Center for Phage Applications and Therapeutics at the University of California San Diego, the Center for Phage Technology at Texas A&M University at College Station and the Pittsburgh Bacteriophage Institute at the University of Pittsburgh are examples of academic institutions. Companies focusing on phage therapy include the Eliava Institute and affiliated companies in Tblisi, InnoPhage in Porto, Adaptive Phage Therapeutics in Gaithersburg, Intralytics in Columbia, Maryland, and Armata Pharmaceuticals in Marina Del Ray. Thus, it appears that phage therapy will be widely available, next to newly developed antibiotics, to teat multi-drug-resistant infections.

Although small-scale studies have demonstrated the potential of phage therapy for bacterial infections, especially in cases of severe antibiotic resistance, the widespread applicability of this therapy has not been shown in clinical trials [100]. In a clinical trial involving patients with urinary tract infections, phages administered directly into the bladder were no more effective than placebo or antibiotics [101]. Burn-wound infections with *P. aeruginosa* were treated with either the phage PP1131 or 1% sulfadiazine silver emulsion cream, the standard of care, in a multi-center clinical trial. Phage treatment at the relatively low dose of 10<sup>6</sup> plaque-forming units per mL was not as effective as the standard of care [102].

To supplement phage therapy, it may be possible to utilize antibiotics and phages simultaneously in some circumstances. In an in vitro study of *P. aeruginosa* and *S. aureus* biofilms, either alone or in combination, the phage EPA1 that infects *P. aeruginosa* and different antibiotics, the simultaneous application of the two agents drastically increased the cytotoxicity against the bacteria [103]. The addition of gentamicin or ciprofloxacin after a 6-h treatment with the phage appeared to eradicate the bacterial biofilms, with higher gentamicin concentrations being necessary for treating combined biofilms.

Clinical trials of bacteriophage therapy of bacterial infections are still at an early stage. Optimal conditions of phage use, including their concentration, the time and sequence of administration and their combination with the appropriate antibiotics, are likely to establish the effectiveness and reliability of this medicine. Even until such standards are established, their ability to save the patient described in the Introduction is a most welcome addition to the practice of medicine.

**Funding:** This research received no external funding.

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

#### **References**


### *Article* **Antimicrobial Peptide K11 Selectively Recognizes Bacterial Biomimetic Membranes and Acts by Twisting Their Bilayers**

**Francisco Ramos-Martín 1,\* , Claudia Herrera-León 1 , Viviane Antonietti <sup>2</sup> , Pascal Sonnet <sup>2</sup> , Catherine Sarazin <sup>1</sup> and Nicola D'Amelio 1,\***


**Abstract:** K11 is a synthetic peptide originating from the introduction of a lysine residue in position 11 within the sequence of a rationally designed antibacterial scaffold. Despite its remarkable antibacterial properties towards many ESKAPE bacteria and its optimal therapeutic index (320), a detailed description of its mechanism of action is missing. As most antimicrobial peptides act by destabilizing the membranes of the target organisms, we investigated the interaction of K11 with biomimetic membranes of various phospholipid compositions by liquid and solid-state NMR. Our data show that K11 can selectively destabilize bacterial biomimetic membranes and torque the surface of their bilayers. The same is observed for membranes containing other negatively charged phospholipids which might suggest additional biological activities. Molecular dynamic simulations reveal that K11 can penetrate the membrane in four steps: after binding to phosphate groups by means of the lysine residue at the N-terminus (anchoring), three couples of lysine residues act subsequently to exert a torque in the membrane (twisting) which allows the insertion of aromatic side chains at both termini (insertion) eventually leading to the flip of the amphipathic helix inside the bilayer core (helix flip and internalization).

**Keywords:** antimicrobial peptide; biomembranes; ESKAPE; antibiotic resistance; NMR; molecular dynamics; biophysics; sequence alignment

### **1. Introduction**

The persistent use of antibiotics, self-medication and exposure to nosocomial infections has provoked the emergence of multidrug resistant (MDR) bacteria worldwide [1–4]. The term "ESKAPE" was adopted to refer to some of the most relevant pathogens associated with the highest risk of mortality by the World Health Organization (WHO) [5], namely *Enterococcus faecium*, *Staphylococcus aureus*, *Klebsiella pneumoniae*, *Acinetobacter baumannii*, *Pseudomonas aeruginosa* and *Enterobacter* spp.

In the quest for new molecules able to overcome this major health issue, antimicrobial peptides (AMPs) are promising alternatives to classical antibiotics, due to their low tendency to resistance [6]. AMPs are natural peptides found in all life kingdoms which can be considered components of the innate immunity against bacteria but also fungi, parasites, virus and cancer. Their reduced tendency to resistance is intrinsically due to their mechanism of action causing the selective disruption of bacterial membranes by acting on the lipidic organization of membranes whose lipid composition cannot be changed by a simple point mutation. While exceptions exist [7], their efficacy is proven by the fact that they have been evolutionarily optimized over millions of years, their fast killing rate discourages the rise of drug-resistant mutants [8] and horizontal transfer of resistance genes against AMPs is infrequent [9]. As opposed to standard antibiotics, many AMPs are able to rapidly

**Citation:** Ramos-Martín, F.; Herrera-León, C.; Antonietti, V.; Sonnet, P.; Sarazin, C.; D'Amelio, N. Antimicrobial Peptide K11 Selectively Recognizes Bacterial Biomimetic Membranes and Acts by Twisting Their Bilayers. *Pharmaceuticals* **2021**, *14*, 1. https:// dx.doi.org/10.3390/ph14010001

Received: 18 November 2020 Accepted: 19 December 2020 Published: 22 December 2020

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/ licenses/by/4.0/).

permeate bacteria and cause irreversible damage to their cell membranes, leading to the death of microorganisms [10,11]. In some cases, their action is also intracellular [12,13].

Several AMPs have been rationally optimized and in this work we focus on K11, a synthetic AMP which was reported to exert antimicrobial action against many of the mentioned ESKAPE bacteria such as *Acinetobacter baumannii*, methicillin-resistant *Staphylococcus aureus*, *Pseudomonas aeruginosa*, *Staphylococcus epidermidis*, and *Klebsiella pneumoniae* [14,15]. K11 has also been successfully used in-vivo as a topic hydrogel solution against *A. baumannii*-infected wounds [15]. Its mechanism of action deserves special attention considering that many of its bacterial targets [14,15] cause complex infections because of their ability to form biofilms [16–18] or change their membrane composition. For example, *A. baumannii* is not only able to form biofilms on biotic and abiotic surfaces but it can also develop resistance to colistin by incorporating phospholipids such as phosphatidylethanolamine (PE), cardiolipin (CL) and monolysocardiolipin to remodel its lipid composition [18,19].

From the point of view of the sequence, K11 (KWKSFIKKLTKKFLHSAKKF-NH2) is an example of synthetic peptide inspired by natural AMPs (cecropin A1, melittin and magainin) [14,15]. More specifically, K11 is one member of a group of peptides synthesized from the CP-P designed antibacterial scaffold (KWKSFIKKLTSKFLHLAKKF). This template was created [14] from the N-terminus of CP26 peptide (inspired by cecropin A1 and melittin) and C-terminus from P18 peptide (inspired by cecropin A1 and magainin) [20]. While CP26 has been reported to target bacterial lipopolysaccharides (LPS) [21], P18 also displays anticancer activity [20]. Most importantly, both CP26 and P18 display antimicrobial activity and negligible toxicity. The introduction of a lysine in position 11 in the CP-P template (hence the name) led to the K11, a peptide with improved values of the therapeutic index (320) [14]. It is believed that the introduction of lysine 11, besides changing the net positive charge, would also alter its amphipathic structure. However, more structural studies are needed to elucidate its mode of action [14].

The interesting properties of K11 prompted us to investigate its interaction with biomimetic membranes by liquid and solid-state NMR spectroscopy (ssNMR) and Molecular Dynamic simulations (MD). Nowadays many different lipidic systems have been optimized for such kinds of studies, going from dodecylphosphocholine (DPC) micelles to bicelles and liposomes with variable phospholipid and sterol compositions reproducing those of the target organisms. The membrane of K11 bacterial targets contains PE, phosphatidylglycerol (PG) and CL in various amounts, as most bacteria. In particular, *Pseudomonas aeruginosa* [22], *Escherichia coli* [23], *Salmonella paratyphi* [24], *Acinetobacter baumannii*, and *Klebsiella pneumoniae* are rich in PE, as expected for the outer membrane of many gramnegative bacteria [25]. Some of its gram-positive targets such as *Bacillus subtilis* and *Bacillus pumilus*, contain PE and PG (although the distribution of phospholipids is unclear) [26–28], while PG or CL clearly prevail in others, such as *Staphylococcus epidermidis* [29,30], *Staphylococcus aureus* [25] and *Micrococcus luteus* [31]. Independently of the relative composition of PG and PE, a special network of H-bond or water-bridged interactions can be established between the two phospholipids [25,32,33], whose ratio can be modulated by bacteria in response to external agents or conditions [22,28,34]. For example, *S. aureus* and *S. epidermidis* can increase their amount of CL under high salt conditions [29,30].

In this work, we show that K11 is able to penetrate biomimetic membranes reproducing the phospholipid composition found in bacteria. Most intriguing, we show that the peptide might act by twisting the membrane using couples of lysine residues. According with this mechanism, the introduction of lysine 11 (whose introduction in the related CP-P peptide significantly improved the therapeutic index) would act in couple with lysine 12 and synergically with all other lysines to torque the membrane, thus facilitating the insertion of aromatic residues at both termini (phenylalanine or tryptophan) and eventually the full peptide in the innermost part of bacterial bilayers. Additionally, for the first time we have observed an interaction with phosphatidylserine (PS), a phospholipid often involved in a wide range of biological processes including viral infection and carcinogenesis [35–37].

#### **2. Results and Discussion**

The work was organized as follows. First, property-alignment [38] was used to highlight important motifs along the sequence and to explore further possible activities of K11. Second, we studied the structure of the peptide in solution and in the presence of simple biomimetic models (micelles and isotropic bicelles). Third, ssNMR was used to characterize the effect of K11 on the lipid assembly in vesicles containing various compositions of PE, PG and CL, due to their importance in bacterial membranes. In the attempt to understand the selectivity and low toxicity of K11, membranes containing phosphatidylcholine (PC) were used to mimic the outer leaflet of eukaryotic cells. PS was also considered to explain further predicted activities. Finally, the studied systems including an even larger variety of phospholipids were studied by molecular dynamic simulations for a deeper understanding of the experimental results and of the mechanism of action.

#### *2.1. Property-Sequence Alignment of K11 Highlight Antibacterial Motifs and Predicts Further Activities*

In order to highlight structure-function relations, we performed property-alignment in the ADAPTABLE web server [38]. It is important to point out that property alignment clusters sequences with specific activities (antibacterial in this case). The K11 peptide is one of a series of synthetic peptides obtained by designed mutations of the CP-P template [14]. All of them are present in the ADAPTABLE database but they are not meaningful when evaluating the importance of conserved residues among evolutionary-distant sequences. Excluding these entries (peptides 2–22), the sequence-related family (SR family) (Figure S1A) shows that the KWK motif at the N terminus and a large portion of the Cterminus seem to be recurrent in peptides with antibacterial activity. Interestingly, eight out of nine meaningful sequences exhibit activity towards a large variety of cancers [39] (Figure S1B). Such predicted activity could be explained by the fact that one of K11 precursors, P18 has also anticancer properties [20]. One related peptide exhibits antifungal activity against *Candida albicans* and *Trichosporon beigelii* [40]. It should be noted that PS plays a relevant role in both cancer development [35,41–43] and *Candida albicans* virulence [44–48] for which PG [49] and PI [47] are also important.

#### *2.2. K11 Peptide Is Unstructured in Aqueous Solution*

The <sup>1</sup>H and <sup>13</sup>C NMR assignment of K11 is reported in Table S1 and Figure 1A,B. The deviations from random coil values [50–52] indicate that the peptide is mainly unstructured in solution (Figure 1C,D), as also confirmed by circular dichroism (CD) (Figure 1E). The formation of a stable helix (theoretical helical wheel in Figure 1F), which would approach many positive charges arising from eight lysine residues, is probably disfavored in the absence of charge-compensating molecular partners.

α α α β **Figure 1.** (**A**,**B**) The <sup>1</sup>H and <sup>13</sup>C NMR assignment of K11 on the amide/Hα (**A**) and side chain (**B**) regions of <sup>1</sup>H, <sup>1</sup>H-TOCSY and <sup>1</sup>H,13C-HSQC spectrum, respectively. (**C**,**D**) Chemical shift deviations from random coil values of amide HN and Hα protons (**C**) and Cα and Cβ carbons (**D**); (**E**) CD spectrum of K11 in solution; (**F**) helical wheel showing the disposition of side chains in alpha helical conformation.

#### 2.2.1. K11 Peptide Assumes Alpha Helical Conformation in a Lipidic Environment

The titration of K11 with a concentrated solution of DPC induces drastic changes in the <sup>1</sup>H NMR spectrum as shown in Figure 2A. New peaks appear in the spectrum while the signals originating from the unbound peptide gradually disappear as the concentration of DPC increases. The slow exchange regime is clearly exemplified by the isolated signal of W2 Hδ1, disappearing at its original shift of 10.2 ppm and reappearing at 10.8 ppm (see Figure 2A). A closer analysis of <sup>1</sup>H,13C-HSQC spectrum (Figure S2A–D), reveals that all aromatic residues (Figure S2A,B) are deeply affected by the presence of the micelles with the exception of H15, whose signals only slightly shift and partially lose intensity (see Figure S2A). A significant shift is also observed for aliphatic signals of F5, I6, L9, T10, F13, L14, A17, F20 (Figure S2C,D) which would be located on the same molecular face in case an alpha helix is formed. A clear proximity of the aromatic (Figure S2E) and aliphatic (Figure S2F) side chains to the DPC acyl chains is demonstrated by the NOE cross peaks in the NOESY spectrum. NOEs with the acyl chain of DPC (whose assignment was based on the literature [53]) but not with its choline headgroup testify a rather deep insertion of the peptide into the micelle and the absence of a specific interaction with the headgroup. While most <sup>13</sup>C backbone signals are lost in the <sup>1</sup>H,13C-HSQC spectrum, we were able to assign all Hα protons. Their values were used to predict the secondary structure [50–52] of the peptide bound to DPC micelles. The negative deviations from theoretical random coil values unequivocally indicate that the peptide assumes an alpha helical conformation (see Figures 2B and 1C for comparison). Accordingly, all the weak HNi/HNi−1/i+1 NOEs observed in the free peptide gain in intensity (data not shown). All-atom MD simulations are in perfect agreement with NMR data (Figure 2C), showing how the peptide conserves its alpha helical conformation along the full trajectory of 500 ns.

The radial distribution function [54] of each oxygen and nitrogen atom of the membrane from all O/N atoms of the peptide can be used to highlight key interatomic interactions. By focusing on its maximum value in the distance range of H-bond and salt bridges, we obtain a quantification of the frequency at which these interactions occur all along the last 250 ns of the simulations. Results are shown in Figure 2D, revealing that lysine side chains are able to recognize the phosphate groups of the phospholipids. Interestingly, only K1, 3, 8 and 12 interact frequently while K7, 11, 18 and 19 seem slightly less involved (Figure S3). While K7 and 11 might not be at the optimal distance, the absence of interaction of K19 might be due to the high curvature of the micelle, not allowing all lysines of the helix to interact at the same time.

#### 2.2.2. In the Presence of Biomimetic Bicelles K11 Peptide Possibly Assumes a Conformation Similar to That Found with Micelles

In order to better understand the influence of the curvature, we studied the interaction of K11 peptide with isotropic bicelles, better representing a more extended surface in solution [55]. Isotropic bicelles can be formed by a mixture of DMPC (1,2-dimyristoyl-snglycero-3-phosphocholine) and DHPC (1,2-dihexanoyl-sn-glycero-3-phosphocholine) [56]. The short acyl chain of DHPC is able to stabilize the bilayer formed by DMPC, whose myristoyl hydrophobic chains would be otherwise exposed to the solvent [57]. Fast tumbling isotropic bicelles can be obtained at a DMPC/DHPC ratio 1:2 (*q* = 0.5). As in the case of micelles, the <sup>1</sup>H NMR spectrum of K11 peptide drastically changes in the presence of bicelles (70 mM) but NMR signals become too broad for a new assignment. However, the NMR spectrum resembles the one observed in the presence of micelles (see Figure S4) suggesting that the same helical conformation is formed.

α **Figure 2.** (**A**) <sup>1</sup>H NMR spectra of K11 1 mM in the presence of DPC at concentrations 1, 2, 4, 10, 20, 30, 60 mM. The NMR assignment in the presence of micelles is also shown. (**B**) Chemical shift deviations from random coil values of Hα protons whose negative deviations indicate an alpha helical conformation; (**C**) MD snapshot of K11 interacting with DPC micelles; (**D**) Polar contacts recurrence (H-bonds and salt bridges) along MD simulation.

In DMPC/DHPC bicelles part of DMPC lipids can be substituted by phospholipids with different headgroups to mimic different biological membranes [57]. In this way, bicelles containing PE, PG, and PS were formed and tested in their interaction with K11. Figure S4 shows that, although the spectra are qualitatively similar, the linewidth is significantly larger in the case of DMPG (1,2-dimyristoyl-sn-glycero-3-phospho-(1′ -racglycerol)) and DMPS (1,2-dimyristoyl-sn-glycero-3-phospho-L-serine), probably indicating that the peptide is less mobile inside these bilayers because of a stronger interaction. Interestingly, both DMPG and DMPS introduce negative charges in the bicelles which might stabilize the structure of the positively charged K11 peptide.

2.2.3. K11 Selectivity Perturbs the Core of Liposomes with Bacterial Phospholipid Compositions

In order to ascertain the effect of K11 on the lipid assembly of the membrane, we studied the interaction of K11 with multilamellar vesicles (MLVs) by ssNMR.

MLVs are more suitable to mimic lipid bilayers of biological membranes because of their hydration state and the lower curvature than bicelles. They also allow to vary the phospholipid composition more freely [55] while bicelles always require a large part of DMPC and DHPC. Moreover, MLVs can be prepared by using commercially available phospholipids bearing deuterated palmitic chains allowing to sense the organization of the hydrophobic core of the lipid bilayer by <sup>2</sup>H NMR. Thus, the order parameter for each C-2H bond of the chain can be measured by means of <sup>2</sup>H quadrupolar splitting [58–60]. Figure 3 shows <sup>2</sup>H spectra of MLVs with various phospholipid compositions and the effect of the presence of K11. Each spectrum results from a superposition of the quadrupolar doublet arising from different C-2H bonds. Since the mobility of unsaturated chains in bilayers increases as we move away from the headgroup, the quadrupolar splitting also decreases, with the consequence that methyl groups appear at the center of the envelope while the carbon in position 2 appears at the extremities of the spectrum, with all the remaining C-2H in-between. Even though we could not measure the order parameters for each C-H moiety because of the low resolution, the overall behavior is very clear when observing the width of the superimposed signals. The presence of K11 does not perturb the <sup>2</sup>H spectra of POPC (1-palmitoyl-2-oleoyl-glycero-3-phosphocholine) membranes and POPE (30% POPC) (1-palmitoyl-2-oleoyl-sn-glycero-3-phosphoethanolamine), suggesting that the peptide is not able to penetrate deeply into these bilayers. This observation stresses the importance of the curvature in biomimetic models. Despite the presence of the same headgroup, K11 deeply penetrates DPC micelles but not POPC MLVs, where lipids are more closely compacted because of the locally almost planar surface as opposed to the high curvature of micelles [55].

Quite interestingly, K11 deeply affects the <sup>2</sup>H spectrum of POPG (1-palmitoyl-2 oleoyl-sn-glycero-3-phospho-(1′ -rac-glycerol)) and POPS (1-palmitoyl-2-oleoyl-sn-glycero-3-phospho-L-serine) MLVs. These headgroups are commonly found in bacteria and cancer cells, respectively, but also fungi, as hypothesized in the activity prediction based on property-alignment by ADAPTABLE. Our data reproduce qualitatively what was found with bicelles, where the increased linewidth observed with PG and PS headgroups suggested a stronger binding (see Section 2.2.2). For both POPG and POPS MLVs, the apparent reduction of the quadrupolar splitting and the loss of resolution reflects a drastic increase in the phospholipid acyl chain mobility, most probably due to the internalization of the peptide in the bilayers. Encouraged by these results, we prepared MLVs using a mixture of PE and PG headgroups, typically found in bacteria [33]. As shown in Figure 3, K11 is able to perturb the fluidity of such bilayers even more and the effect becomes really important in the presence of CL, also found in bacteria [61]. Figure 3 shows how K11 affects also membranes mainly constituted by CL (CL 50% / POPC 50%). It should be noted that in the cases of CL and POPE, the addition of POPC was necessary for the formation of a MLV, due to their intrinsic shape and different *T*<sup>m</sup> [32,62–68].

**Figure 3.** Static <sup>2</sup>H NMR spectra of various multilamellar vesicles (MLVs) in the absence (blue) and in the presence (red) of K11 peptide.

#### *2.3. MD Simulations Provide a Molecular Picture of the Interaction*

In order to get insight into the details of the interaction between K11 peptide and biomembranes, we performed MD simulations using a variety of phospholipid combinations involved in bacteria, cancer and fungi. Figure 4 shows the most significant snapshot for each run, and in particular the frames where the peptide comes close to the membranes.

In the case of POPC, K11 stays away from the bilayer during most of the simulation and even when it approaches the membrane, there is no evidence of a specific interaction. For POPE (containing 30% of POPC to reproduce the experimental conditions), the situation is only slightly different. In this case, the peptide can lay on the surface sporadically and very few interactions are established.

The situation is radically different for POPG and in general for all negatively charged phospholipids: POPS, POPI (1-palmitoyl-2-oleoyl-sn-glycero-3-phosphoinositol), and CL. A clearer picture comes from the analysis of polar interactions shown in Figures S5 and S6. As in Section 2.2.1, we calculated the recurrence of H-bond and salt bridges (Figures S5 and S6). In the case of POPG we observe an important interaction of the N-terminal lysine with the oxygen atoms of POPG phosphate groups by means of both the backbone and side chain amines. Such an interaction is consistently observed when the peptide significantly interacts, suggesting that K11 peptide approaches the membranes with the first lysine residue. Furthermore, also the side chains of lysine residues in position 3, 7 and 8 establish similar contacts. The selectivity for negatively charged phospholipids like POPG is probably due to the electrostatic attraction leading the positively charged K11 peptide (+8 at physiological pH) towards the negative charges introduced by POPG headgroup but also to the availability of multiple oxygen atoms provided by its glycerol moiety (inositol and carboxylate in the case of POPI and POPS, respectively), which are available for hydrogen bonding or the formation of extra salt bridges in the case of POPS. Not surprisingly, a strong interaction is also observed with CL due to the structural similarity to POPG, the exposition of phosphate groups at the membrane surface and the doubly negative charge. When PG or CL are in mixtures, they are involved in the large majority of the peptide-membrane interactions (see Figure S6).

**Figure 4.** MD snapshots representative of K11 peptide interacting with several membranes of variable phospholipid compositions. Color code: (**A**) POPC black (body) and light gray (choline group); (**B**) POPE dark green (body), turquoise (headgroup) and light green (amine of the headgroup); (**C**) POPG dark violet (body), violet (headgroup) and light violet (hydroxyls of the headgroup); (**D**) POPS brown (body), gold (headgroup), light yellow (amine of the headgroup) and orange (carboxyl of the headgroup); (**E**) POPI blue (body), light blue (headgroup) and cyan (hydroxyls of the headgroup); (**F**) CL dark red (body) and light red (headgroup). Panels G-H show lipid mixtures typically found in bacteria: (**G**) POPE/POPG and (**H**) POPE/POPG/CL. Panel (**I**) represents an example of calculation with eight peptides while panel (**J**) refer to a calculation of one peptide interacting with a pure POPE membrane, differing from that in panel B for the absence of POPC. Snapshots in panels (**K**), (**L**) refer to examples of simulations where K11 is purposely placed inside the membrane at the start of the calculation. In all panels the phosphorus atom of phospholipids is shown as a yellow sphere; for clarity, only functional moieties of headgroups are represented as spheres either in the upper leaflet, or in both leaflets (panels K, L). K11 peptide is shown as a "tube" colored from blue (N-terminus) to red (C-terminus) except in panel I where each of the eight peptide has a different color. Side chains are shown as sticks with the following color code: positively charged (blue), negatively charged (red), non-polar (light gray), polar (yellow).

The Coulomb attraction is clearly a key factor to understand such selectivity but also other factors play a role: the inter-lipid spacing (modulated by the curvature [69,70]), the steric hindrance of the headgroups and the amount of inter-lipid interactions [34,35]. The first factor explains why K11 can penetrate DPC micelles (whose PC headgroup is not found in bacteria) but doesn't seem to affect POPC liposomes significantly. In this case, the high curvature makes the inter-lipid spacing wider, facilitating the access to the micelle core. The second factor explains the preference for CL over PE and PG when the three are present (see Figure S6); in the case of CL phosphate moieties are directly accessible as its headgroup doesn't present steric hindrance. The third factor is also important, because the insertion of the peptide implies breaking a number of favorable interactions (like the ones present in PE/PG membranes [25,32,33]). Besides these effects, the presence of negative charges on the membranes does incentivize the binding not only in the case of PS but also with PG and PI (Figures S5 and S6). In similar conditions of curvature, we can expect that the accessibility to phosphate moieties is modulated by the steric hindrance of headgroups but also the network of inter-lipid interactions. Last but not least, the mobility of the peptide in the complex and the degree of lipid-order destabilization might contribute to the overall process as entropic contribution [71].

#### 2.3.1. K11 Exerts a Twisting Effect of Its Target Membranes

Quite interestingly, the membrane planarity is heavily perturbed and almost twisted when K11 peptide interacts. A closer analysis of the helix wheel reveals that the peptide could act as a screw twisting the membrane by means of couples of lysine residues (Figure 5). In all MD simulations in the presence of membranes containing PG, PS, PI or CL we observe the same behavior (see Figure 5A): the peptide approaches the membrane with the N-terminal lysine (step 1, anchoring), grabs onto the available oxygen atoms (most frequently phosphate oxygen O13 and O14 but also headgroup oxygens) and deforms the membrane (ste p2, twisting). The deformation allows the insertion of terminal aromatic groups (F20 but in some cases also W2) inside the bilayer (step 3, insertion). F20 and W2 are the only amino acids whose hydrophobic side chains are readily available and this is due to the fact that the amphipathic helix formed by K11 approaches the membrane with its hydrophilic face. In particular, in the simulation containing POPE/POPG, we observe a further important step. The insertion of the aromatic ring of F20 eventually determines the flip of the full hydrophobic face of the helix into the bilayer thus reaching the hydrophobic core (step 4, helix flip and internalization). The peptide remains inserted even prolonging the simulation up to 2 µs.

A closer look to the relative disposition of lysine residues in an alpha helix conformation can explain the twisting effect on the membrane (Figure 5B). We believe that K11 peptide actually works as a screw. By landing on the surface with the first lysine, the peptide anchors to the available oxygen atoms. These may arise from the phospholipid phosphate groups or oxygens in the headgroups. Such an anchoring is quite effective because lysine 1 bears two amine moieties that can bind in a bidentate fashion. Figures S5 and S6 shows how such an interaction is present in almost all simulations involving charged phospholipids with high occurrence. We can imagine dissecting the helix of the peptide with three planes almost orthogonal to its long axis, each containing two lysine residues (K7 and 8, K11 and 12, K18 and 19). The K1 anchoring step is followed by the establishment of interactions involving lysines 7 and 8 with available nearby membrane oxygen atoms. These bindings have the synergic effect of rotating the membrane in their plane (see Figure 5B). Such rotation is subsequently reproduced in the plane of lysines 11 and 12 and in that of lysines 18 and 19. As the couples of lysines 7–8, 11–12 and 18–19 are located with different phases in the helix wheel, these subsequent rotations have the effect of a twist. In particular, while lysines 7–8 and 11–12 would determine a clockwise rotation, lysines 18–19 would act in the opposite sense because of their intermediate phase in the wheel.

**Figure 5.** (**A**) Proposed mechanism of action of K11 peptide in four steps. The peptide first anchors to the membrane by the lysine residue in position 1 (anchoring) which can bind membrane oxygen atoms in a bidentate fashion. The peptide then twists the membrane (as described in panel B) thus allowing the insertion of terminal aromatic side chains. Finally, the peptide flips inside the bilayer. For color codes refer to the caption of Figure 4. (**B**) Mechanism by which K11 might exert a torque on the target membrane. Yellow circles represent oxygen atoms on the surface of the membrane and available for H-bonding or salt bridges with lysine side chains. The helix formed by K11 is represented as a cylinder. Lysine residues and membrane planes are represented in blue color whose intensity degrades with the distance from the observer. The torque is achieved in three subsequent steps, each rotating the membrane in the plane described by each couple of lysine residue. A geometrical representation of the effect on the membrane for each step is exemplified under each step. Image generated with the help of CalcPlot3D software [72].

Significant perturbation of the membrane can be more easily visualized by monitoring the area per lipid along the MD trajectories. Although the perturbation can be detected in simulations with one peptide, the effect is amplified with the introduction of several peptides (see Figure S7). With the exception of POPC and POPE, we observe a decrease in area per lipid of the upper leaflet and an increase in that of the lower leaflet, indicating that the peptide exerts a pressure causing the membrane to invaginate (negative curvature).

#### 2.3.2. K11 First Rigidifies the Membrane and Subsequently Makes It More Fluid

The last step of the mechanism proposed in Figure 5 (step 4, helix flip and internalization) is essential because it explains the reduction of the lipid chain order experimentally demonstrated for the phospholipid acyl chains by the perturbation of <sup>2</sup>H NMR spectra (Figure 3). It should be stressed that peptide anchoring (step 1 in Figure 5A) and membrane twisting (step 2 in Figure 5A) actually increase the order parameters of acyl side chains while the internalization (step 4 in Figure 5A) of the peptide in the hydrophobic core reduces it (Figure 6), as experimentally observed (Figure 3). The description of the proposed mechanism of action in four steps may require extending our 500 ns simulations further. The complete helix flip (step 4 in Figure 5A) is only observed in one of the three repetitions of the simulation with POPE/POPG membranes. We can hypothesize that in POPE/POPG mixtures (which better represent the bacterial membrane with respect to pure POPG) the activation energy for the penetration of the peptide is lower, thus allowing its detection in our 500 ns-long simulations. This is a reasonable hypothesis when considering that the PE headgroup has a smaller steric hindrance than that of PG and could facilitate the entrance of the peptide, as can also be rarely observed in simulations with pure POPE (Figure 4J). For this reason, we have extended the POPE/POPG calculation up to 2 µs.

μ **Figure 6.** Order parameter of C-H moieties in palmitoyl side chains in membranes containing POPE (70%) and POPG (30%) as calculated from MD simulations in the absence (2 repetitions in black labeled as 1 and 2) and in the presence of K11 peptide. The order parameter varies in different ways along the 2 µs trajectory. The membrane is rigidified upon interaction (brown curve) but becomes more fluid as the peptide penetrates (red) and becomes internalized (orange curve).

It should be noted that the formation of the complex takes place in slow exchange in the NMR time scale (Figure 2A), meaning that kex << |∆ω|. kex is the exchange rate constant (kex = kon[L] + koff where kon and koff are the on and off-rate constants for the formation of the complex between the peptide P and the membrane M according to

the equation P + M → PL) and |∆ω| is the chemical shift difference between the free and the bound form of the peptide [73]. Our deviations are on the order of 0.2 ppm (Figure 2B), meaning that at 500 MHz |∆ω| is ~600 s−<sup>1</sup> (|∆ω| = 2π|∆ν|, where |∆ν| is the chemical shift difference in Hz). Our slow exchange conditions therefore limit the value of kex and koff to a maximum of 600 s−<sup>1</sup> and the lifetime of the complex to a minimum of 1.7 milliseconds or much more, including the case of irreversible binding (the lifetime is the inverse of the off-rate constant). The detection of such long processes [74,75] would require more advanced sampling algorithms including dual-resolution MD [76], coarse-grain simulations, steered MD [77], umbrella-sampling [78,79], metadynamics [80,81], or replica exchange, among others [75,82,83]. This is beyond the scope of this work that aims at characterizing the steps at the very beginning of the interaction, in order to unravel the mode of action. The choice of all-atom MD allows us to directly compare the calculation with NMR data providing specific information on hydrogen and carbon atoms.

Peptide concentration plays an important role in the mechanism of action of AMPs because antimicrobials can act synergically to destabilize the target membrane using different strategies (carpet, pore formation by toroidal or barrel-stave models [76,84–86]). In order to confirm the hypothesis of an initial rigidification we calculated the order parameter for all membranes (Figure S8) increasing the number of peptides to simulate a higher concentration (one snapshot example of such calculation is shown in Figure 4I). Figure S8 shows a rigidification clearly visible for PG and PS membranes, as expected. The increase of order observed upon peptide binding is not so uncommon, and depends on factors like lipid composition of the membrane, temperature, and charge [58,60,87–89]. Sometimes, peptides that attach to the surface of the bilayer can increase acyl chain packing [90,91], especially when a strong electrostatic attraction is established [91]. The rigidification effect upon binding is also consistent with the observed hydrophobic thickness (Figure S9), that greatly increases for POPG, POPG containing membranes and CL as compared to POPC. Furthermore, the observed decrease in the electron density (Figure S9) can be a consequence of more water molecules being located near the polar head groups due to more loosely packaging caused by the presence of the peptides [88,92,93].

The effect of fluidification following the internalization was confirmed (Figure S10) by placing the peptide inside the bilayers at the beginning of the simulations (see example snapshots in Figure 4K,L).

#### 2.3.3. K11 Approaches Phospholipids Head Groups from Opposite Leaflets Possibly Leading to Membrane Disassembly after Entering the Bilayer

The simulations of the fully internalized peptide can be thought of as a "prolongation" for those in which the peptide is able to access the membrane core. These simulations allow us to bypass the longer time scales needed to observe the full process. Two snapshots are shown in Figure 4K,L and they testify to a quite interesting phenomenon. The length of K11 helix is slightly shorter than the membrane thickness with the result that both the N-terminus and the C-terminus of K11 tend to recall polar head groups in the membrane core by binding with their oxygen atoms. Polar head groups on opposite leaflets almost come in close proximity. This is possible because polar head groups are initially grabbed by peripheral lysines residues and subsequently "walk" by detaching and attaching to the ones in the center of the helix. This is particularly evident in bacterial biomimetic membranes (Figure 4K,L). Once inside the bilayer, this mechanism would allow K11 to disassemble the membrane.

#### 2.3.4. PS Targeting Opens the Way to Possible New Biological Activities

The data presented in this work indicates that K11 destabilizes PS containing membranes (Figures 3 and 4 but also Figures S5, S7 and S8), and this could be an indication of a possible anticancer activity, as already shown for some members [39] of its SR family (see Section 2.1 and Figure S1). It has to be noted that K11 was created as a combination of CP26 peptide (inspired by cecropin A1 and melittin) and a C-terminus from P18 peptide (inspired by cecropin A1 and magainin), which displays anticancer activity [20]. Similarly

to what happens in apoptotic cells, cancerous cells tend to expose PS, a phospholipid normally found in the inner leaflet of the membrane [35]. A specific interaction with PS is probably the reason why a considerable number of antimicrobial peptides produced in eukaryotes (or inspired by them like K11) display anticancer activity while displaying low hemolytic activity and toxicity to healthy ones. Their eukaryotic origin explains their selectivity [94–96] As PS-targeting has proved to be effective as anti-cancer [36] or antiviral [97] therapies, the selective recognition of PS by K11 should not be undervalued.

One member of the K11 SR family also displays activity against fungi like *Candida albicans*. As shown in this work, K11 targets PS and PI, both being relevant for *Candida* virulence [44–48], together with PG [49].

#### **3. Materials and Methods**

#### *3.1. Synthesis of K11 Peptide*

Fmoc(9-fluorophenylmethoxy)-amino acids, Fmoc-Tyr(tBu)-AC TentaGel® resin (0.22 mmol/g, particle size: 90 µm) and Fmoc-TentaGel®-S RAM resin (0.24 mmol/g, particle size: 90 µm) were purchased from Iris Biotech (Germany). The other chemical compounds were purchased from VWR Chemicals, Iris Biotech or Acros and used without further purification. The peptides were synthesized on an CEM Liberty 1 Microwave Peptide Synthesizer, using standard automated continuous-flow microwave solid-phase peptide synthesis methods. Five-fold molar excess of the above amino acids was used in a typical coupling reaction. Fmoc-deprotection was accomplished by treatment with 20% (*v*/*v*) piperidine in *N*-methyl-2-pyrrolidone (NMP) at 75 ◦C. The coupling reaction was achieved by treatment with 2-(1*H*-benzotriazol-1-yl)-1,1,3,3-tetramethyluronium hexafluorophosphate (HBTU) and *N*,*N*-diisopropylethylamine (DIEA) in NMP using a standard microwave protocol (75 ◦C). The peptide was cleaved and side-chain deprotected by treatment of the peptide resin with a mixture of 1.85 mL of trifluoroacetic acid (TFA), 50 µL of triisopropylsilane, 50 µL H2O and 50 mg of DL-dithiothreitol, in respective percent proportions, 92.5/2.5/2.5/2.5, during 4 h at room temperature. The solid support was removed by filtration, the filtrate concentrated under reduced pressure, and the peptide precipitated from diethyl ether. The precipitate was washed several times with diethyl ether and dried under reduced pressure. The peptides were purified on an RP-HPLC C18 column (Phenomenex® C18, Jupiter 4µ Proteo, 90 Å, 250 × 21.20 mm) using a mixture of aqueous 0.1% (*v*/*v*) TFA (A) and 0.1% (*v*/*v*) TFA in acetonitrile (B) as the mobile phase (flow rate of 3 mL/min) and employing UV detection at 210 and 254 nm. The purity of all peptides was found to be >95%.

K11 peptide was obtained as a white powder, with a total yield of 21.7%, after purification by reverse-phase HPLC (96% analytical purity) (see Figure S11). The concentration of the sample was determined by dissolving a precise amount of the powder in a precise volume of the buffer. The concentrated solution was subsequently divided in aliquots and lyophilized. Once redissolved in buffer, the concentration was confirmed by the absorbance at 280 nm, using a molar extinction coefficient of 5500 cm−<sup>1</sup> , M-1 (only one tryptophan is present) estimated by the ProtParam tool [98] of Expasy server (https://web.expasy.org/protparam/).

#### *3.2. Sequence Alignment by ADAPTABLE Web Server*

The family of peptides sequence-related to K11 (KWKSFIKKLTKKFLHSAKKF) was created by the family generator page of ADAPTABLE webserver (http://gec.u-picardie. fr/adaptable/) using "Create the family of a specific peptide" option with the following parameters: "antibacterial = y"; "activity (µM) = 1"; "Substitution matrix = Blosum45"; "Minimum % of similarity = 51". As ADAPTABLE continuously updates with new entries sequence-related families might change slightly with the time [38].

#### *3.3. Sample Preparation, NMR Experiments and Analysis*

Backbone and sequential resonance assignments were achieved by <sup>1</sup>H,13C-HSQC, <sup>1</sup>H,1H-TOCSY (mixing of 60 ms), and <sup>1</sup>H,1H-NOESY (mixing of 200 ms) recorded on a 500 MHz Bruker spectrometer equipped with a 5 mm Broadband Inverse (BBI) probe. Deuterated sodium 3-(trimethylsilyl)propionate-d4 (TSP-d4) at a concentration of 100 µM was used as internal reference for chemical shift. Reference random coil values in our experimental conditions (*T* = 278 K, pH 6.6 and ionic strength 0.02 M) were calculated by POTENCY web server (https://st-protein02.chem.au.dk/potenci/) [99].

CD spectra were obtained in the far-UV (260–185 nm) on a J-815 Jasco spectropolarimeter (Tokyo, Japan). The CD measurement was performed at 5 ◦C, using a 1 mm path cell, with 5 accumulations for a 216.0 mg/mL sample in 10 mM sodium phosphate buffer, pH 6.6. All CD spectra measured were baseline corrected by subtracting the buffer spectrum.

A 1 mM sample of K11 (90% 10 mM phosphate buffer/10% D2O, pH 6.6) was titrated with a 1 M stock solution of DPC to a final DPC concentration of 60 mM. Titration was followed by 1D <sup>1</sup>H-NMR at 278 K. For the assignment of the interacting form of the peptide 2D <sup>1</sup>H,1H-NOESY and <sup>1</sup>H,13C-HSQC were recorded at total DPC concentrations of 60 mM.

Bicelles were prepared as follows. A mixture of 33.3% DMPC and 66.7% DHPC in chloroform was used to obtain isotropic bicelles at a molar (q) ratio of 0.5. The solvent was evaporated under a nitrogen flow and the samples were then lyophilized and resuspended in a 10 mM phosphate buffer (pH 6.6) to reach a final concentration of 1 M (stock solution). DMPG, DMPS and DMPE containing bicelles were prepared as described above, except part of DMPC was replaced by DMPG (25%), DMPS (25%) or DMPE (10%) reproducing previous experiments [57]. A 1 mM sample of K11 (90% 10 mM phosphate buffer/10% D2O, pH 6.6) was titrated with bicelles up to a final lipid concentration of 70 mM and monitored at 278 K by a 1D <sup>1</sup>H-NMR spectrum recorded after each addition.

MLVs containing deuterated palmitoyl chains were prepared according to the conventional protocol [100–103] using the following proportions: 50%:50% POPC/POPC:d31, 50%:50% POPG/POPG:d31, 50%:50% POPS/POPS:d31; 70%:30% POPE:d31/POPG, 50%:50% CL/POPC:d31, 70%:30% POPE:d31/POPC and 67%:27%:6% POPE:d31/POPG/ CL. Lipids were solubilized in chloroform and solutions were mixed in order to obtain the right proportions in a total lipid amount of 60 mM. The resulting solution was evaporated under nitrogen gas flow. The sample was hydrated with ultrapure water, well-vortexed to promote a total hydration and lyophilized overnight to remove the traces of solvents. The resulting powder containing lipids was hydrated by 80 µL of ultra-pure water (for non-charged lipids) or 10 mM phosphate buffer pH 6.6 100 mM NaCl (for charged lipids), vortexed and homogenized using four free-thaw cycles involving one step of freezing (−80 ◦C, 15 min) followed by thawing (40 ◦C, 15 min) and shaking. Finally, the MLV samples were placed in a 7-mm ssNMR rotor to perform the experiments. 2.4 mM of peptide were added for interaction studies.

ssNMR experiments were recorded at 310 K on a Bruker Avance Biospin 300 WB (7.05 T) equipped with a CP-MAS 7-mm probe (Bruker Biospin, Karlsruhe, Germany). Static <sup>2</sup>H NMR was carried out applying a phase cycled quadrupolar echo pulse sequence (90◦x-τ-90◦y-τ-acq) [104]. The parameters used are listed below: spectral width of 150 kHz, π/2 pulse of 5.25 µs, an interpulse delay of 40 µs, a recycled delay of 1.5 s, and a number of acquisitions ranging from 8 k to 14 k depending on samples. For all spectra, an exponential line broadening of 100 Hz was applied before Fourier-transform from the top of the echo signal.

#### *3.4. Molecular Dynamics Simulations*

Systems for simulations were prepared using CHARMM-GUI [105–107]. A total of 128 lipid molecules were placed in each lipid bilayer (i.e., 64 lipids in each leaflet) and peptide molecules were placed over the upper leaflet at non-interacting distance (>10 Å). Lysine residues were protonated while histidine residue was protonated only on nitrogen

in position δ. Initial peptide structure was obtained via I-TASSER [108] prediction tool, that produced a similar construct as the one produced by PEPFOLD [109,110] software. This structure was almost completely helical. Amidation of the C-terminus was achieved via the CHARMM terminal group patching functionality which is fully integrated in the CHARMM-GUI workflow. In case of calculations with eight peptides, they were placed next to each other but not in contact. A water layer of 50-Å thickness was added above and below the lipid bilayer which resulted in about 15,000 water molecules (30,000 in the case of CL) with small variations depending on the nature of the membrane. Systems were neutralized with Na<sup>+</sup> or Cl<sup>−</sup> counterions.

MD simulations were performed using GROMACS software [111] and CHARMM36 force field [112] under semi-isotropic (for bilayers) and isotropic (for micelles) NPT conditions [113,114]. The TIP3P model [54] was used to describe water molecules. Each system was energy-minimized with a steepest-descent algorithm for 5000 steps. Systems were equilibrated with the Berendsen barostat [115] and Parrinello-Rahman barostat [116,117] was used to maintain pressure (1 bar) semi-isotropically with a time constant of 5 ps and a compressibility of 4.5 × 10−<sup>5</sup> bar−<sup>1</sup> . The Nose-Hoover thermostat [118,119] was chosen to maintain the systems at 310 K with a time constant of 1 ps. All bonds were constrained using the LINear Constraint Solver (LINCS) algorithm, which allowed an integration step of 2 fs. PBC (periodic boundary conditions) were employed for all simulations, and the particle mesh Ewald (PME) method [120] was used for long-range electrostatic interactions. After the standard CHARMM-GUI minimization and equilibration steps [113], the production run was performed for 500 ns (except when mentioned explicitly) and the whole process (minimization, equilibration and production run) was repeated once in the absence of peptide and twice in its presence. Convergence was assessed using RMSD and polar contacts analysis (see Figure S12).

All MD trajectories were analyzed using GROMACS tools [121,122] and Fatslim [123]. MOLMOL [124] and VMD [125] were used for visualization. Graphs and images were produced with GNUplot [126] and PyMol [127].

#### **4. Conclusions**

In this work, we have shown how the K11 peptide, largely unstructured in solution, assumes alpha helical conformation in the presence of biomimetic membranes. The interaction has very different consequences on the stability of the membrane depending on its nature. While PC and PE/PC bilayers are largely unaffected, PG, PS, PI and CL strongly interact with lysine residues. When examining bacterial-like mixtures containing both PG and PE, the large majority of the peptide-membrane interactions takes place with PG and the structurally related CL, if present. However, the same mechanism might well be active in the presence of PS, often exposed on the outer leaflet of cancer cells, which would suggest a potential anticancer activity of K11, as already described for its related peptides. The analysis of polar contacts reveals that lysine side chains tend to interact with oxygen atoms of the phosphate moiety (or the carboxylate of the serine in PS) rather than the OH of the glycerol or inositol head group, indicating that the recognition is based on the formation of salt bridges rather than H-bonds. This explains the lower affinity for PC and PE where the negative charge is neutralized by the choline and ethanolamine moieties, respectively. Once the salt bridges are formed, the peptide might penetrate as a screw, anchoring to the target with its N-terminus and twisting the membrane by further subsequent salt bridges involving pairs of lysine residues. The torque allows then the insertion of terminal hydrophobic side chains and eventually the internalization of the full peptide. Once inside, K11 can approach phospholipid head groups on opposite leaflets causing an effective disruption of the membrane potentially leading to the bacterial death.

**Supplementary Materials:** The following are available online at https://www.mdpi.com/1424-8 247/14/1/1/s1, Figure S1: Sequence alignment of K11 peptide used as a bait in the ADAPTABLE web server; Figure S2: <sup>1</sup>H,13C-HSQC spectral regions and assignment of K11 in solution and in the presence of DPC micelles, Figure S3: Minimum distance of each lysine side chain amine (atom

name NZ) from membrane phosphorus atoms along the simulation trajectory of K11 interacting with DPC micelles, Figure S4: <sup>1</sup>H NMR normalized spectra of K11 in the presence of DMPC/DHPC, DMPC/DHPC/DMPE, DMPC/DHPC/DMPG, and DMPC/DHPC/DMPS bicelles, Figures S5 and S6: Occurrence of polar atom contacts between K11 peptide and various membrane bilayers, Figure S7: Area per lipid in bilayers containing various phospholipids compositions as calculated from MD simulations in the presence of eight K11 peptides, Figure S8: Order parameter of C-H moieties in palmitoyl side chains in membranes containing various phospholipids compositions as calculated from multiple repetitions of MD simulations in the absence (2 repetitions in black labeled as 1 and 2) and in the presence (3 repetitions in red labeled from 1 to 3) of eight K11 peptides. The panel in the right bottom corner is an example of MD snapshot with POPE/POPG bilayer (color code in the caption of Figure 4). TOCL2 refers to CL, Figure S9: Electron density profiles for POPC, POPG and POPE/POPG/CL in presence of eight K11 peptides, Figure S10. Order parameter of C-H moieties in palmitoyl side chains in membranes containing various phospholipids compositions as calculated from multiple repetitions of MD simulations in the absence (2 repetitions in black labeled as 1 and 2) and in the presence (3 repetitions in red labeled from 1 to 3) of K11 peptide initially placed inside the bilayer. The panel in the right bottom corner is an example of MD snapshot with POPE/POPG/CL bilayer (color code in the caption of Figure 4). TOCL2 refers to C, Figure S11: Analytical purity of K11 peptide; Table S1: <sup>1</sup>H and <sup>13</sup>C NMR assignment of K11 peptide, Figure S12: Convergence analysis of the simulation of K11 peptide in the presence of POPE/POPG membrane. (A) Peptide RMSD (Cα carbon); (B) Polar contact block analysis in time intervals.

**Author Contributions:** Conceptualization, F.R.-M. and N.D.; Data curation, N.D.; Formal analysis, F.R.-M. and C.H.-L.; Funding acquisition, C.S. and N.D.; Investigation, F.R.-M., C.H.-L. and V.A.; Methodology, F.R.-M. and N.D.; Project administration, N.D.; Resources, P.S. and C.S.; Software, F.R.-M. and N.D.; Supervision, N.D.; Validation, N.D.; Visualization, F.R.-M. and N.D.; Writing original draft, F.R.-M. and N.D.; Writing—review & editing, F.R.-M., C.H.-L., P.S., C.S. and N.D. All authors have read and agreed to the published version of the manuscript.

**Funding:** Francisco Ramos-Martín's PhD scholarship was co-funded by Conseil régional des Hautsde-France and by European Fund for Economic and Regional Development (FEDER); Claudia Herrera-León's PhD scholarship was funded by the National Council for Science and Technology (CONACYT). This work was partly supported through the ANR Natural-Arsenal project. Publication fees were partly funded by the University of Picardie Jules Verne.

**Acknowledgments:** We would like to thank Zakaria Bouchouireb, Professor Manuel Dauchez, University of Reims Champagne-Ardenne (URCA), UMR CNRS 7369 MEDyC for useful discussion and Dominique Cailleu for his competence in the setting up of NMR experiments. We also thank the Matrics platform at the University "Picardie Jules Verne" and the "Méso-centre de Calcul Scientifique Intensif" at the University of Lille for providing computing resources.

**Conflicts of Interest:** The authors declare that they have no competing interest.

#### **References**


### *Article* **Imidazole and Imidazolium Antibacterial Drugs Derived from Amino Acids**

**Adriana Valls <sup>1</sup> , Jose J. Andreu <sup>1</sup> , Eva Falomir <sup>1</sup> , Santiago V. Luis <sup>1</sup> , Elena Atrián-Blasco 2,3,\* , Scott G. Mitchell 2,3,\* and Belén Altava 1,\***


Received: 18 November 2020; Accepted: 17 December 2020; Published: 21 December 2020 -

**Abstract:** The antibacterial activity of imidazole and imidazolium salts is highly dependent upon their lipophilicity, which can be tuned through the introduction of different hydrophobic substituents on the nitrogen atoms of the imidazole or imidazolium ring of the molecule. Taking this into consideration, we have synthesized and characterized a series of imidazole and imidazolium salts derived from *L*-valine and *L*-phenylalanine containing different hydrophobic groups and tested their antibacterial activity against two model bacterial strains, Gram-negative *E. coli* and Gram-positive *B. subtilis*. Importantly, the results demonstrate that the minimum bactericidal concentration (MBC) of these derivatives can be tuned to fall close to the cytotoxicity values in eukaryotic cell lines. The MBC value of one of these compounds toward *B. subtilis* was found to be lower than the IC<sup>50</sup> cytotoxicity value for the control cell line, HEK-293. Furthermore, the aggregation behavior of these compounds has been studied in pure water, in cell culture media, and in mixtures thereof, in order to determine if the compounds formed self-assembled aggregates at their bioactive concentrations with the aim of determining whether the monomeric species were in fact responsible for the observed antibacterial activity. Overall, these results indicate that imidazole and imidazolium compounds derived from *L*-valine and *L*-phenylalanine—with different alkyl lengths in the amide substitution—can serve as potent antibacterial agents with low cytotoxicity to human cell lines.

**Keywords:** imidazole and imidazolium salts; amino acid; antibacterial agents; aggregation; lipophilicity

#### **1. Introduction**

Aromatic heterocycles, particularly the imidazole ring, have been used in the last decades as structural skeletons to obtain different types of bioactive compounds with antibacterial, antifungal, anticancer, antiviral, antidiabetic, and other properties [1–4]. The search for new potent drug molecules derived from imidazole continues to be an intense area of investigation in medicinal chemistry [5–7]. Moreover, pharmaceutical research, manufacture, and regulation are enhancing the development of solid active ingredients, delivered as powders or tablets; however, many solid drugs which perform well in in vitro evaluation remain too insoluble for the body to absorb [8,9]. Most of the bioactive agents sold for pharmaceutical or food industries are salts [10,11], and in this context, ionic liquids (ILs) represent a promising class of drug candidates whose physicochemical and pharmaceutical properties can be easily tuned [12–17]. In this regard, the imidazolium skeleton can be transformed into ionic liquids with promisingly potent pharmacological properties [18–21]. Consequently, monoimidazolium [22–24] and bisimidazolium [25–28] salts have been explored as a new generation of antibacterial agents. In this

context, amino acid-based monoimidazolium salts with good bacterial toxicity have been reported in the literature [29].

Our research group has an ongoing interest in the biomimetic and bioactive capacity of imidazole or imidazolium amino acid derivatives [30–33]. Herein, imidazole, monotopic, and ditopic imidazolium salts derived from *L*-valine and *L*-phenylalanine with different alkyl lengths in the amide substitution were synthesized and characterized comprehensively. The antibacterial activity of these amino acid-based imidazolium salts against Gram-negative *Escherichia coli* DH5-α (herein *E. coli*) and Gram-positive *Bacillus subtilis* 1904-E (herein *B. subtilis*) were evaluated and their cytotoxicity was also studied using a human embryonic kidney cell line (HEK-293). Finally, due to the amphiphilic character of these compounds and the strong tendency towards self-aggregation of ionic liquid-related surfactants based on imidazolium salts [34–37], we investigated the spontaneous aggregation behavior of these compounds in water and in bacterial cell culture medium. Through optical and scanning electron microscopy, as well as UV-vis and fluorescence spectroscopy, we have extracted structure-property relationships between the degree of aggregation/self-assembly of the *L*-valine and *L*-phenylalanine derivatives and their corresponding antibacterial activity and cytotoxicity [38]. α

#### **2. Results**

#### *2.1. Synthesis*

The imidazole-amino acid derivatives **1a**, **2a**, and **3a** were obtained from the corresponding α-amino amide as previously described [31]. Monotopic and ditopic -imidazolium salts **1b–3b** and **1c–3c** were obtained in high yield by treatment of the corresponding imidazole with benzyl bromide or 1,3-bromomethylbenzene, respectively, as described in our previous publications (Scheme 1) [30,32]. α

**Scheme 1.** Synthesis of the amino acid-based imidazolium salts in this report.

#### *2.2. Antibacterial and Cytotoxicity Studies*

μ The in vitro antibacterial activities of the synthesized compounds were examined against *E. coli* and *B. subtilis*. Bacteria were incubated in culture media with varying concentrations of the examined compound and the antibacterial properties were determined by observation of the optical density at 560 nm (bacteriostatic activity) and by the Resazurin cell viability assay (bactericidal activity). The corresponding minimal bactericidal concentration (MBC) and minimum inhibitory concentration (MIC) values that were obtained are summarized in Table 1 (please refer to Table S1 for MIC and MBC values in µM).

The outer membrane of Gram-negative bacteria such as *E. coli* includes porins, which allow the passage of small hydrophilic molecules across the membrane, and lipopolysaccharide molecules that extend into extracellular space. Thus, the observed trend in the activity results could be explained by the relative lipophilicity of the compounds combined with their capacity to disrupt the cell membrane [39,40]. The relative lipophilicity of the compounds was determined theoretically using VCCLab and Molinspiration softwares (LogP values Table 1) and experimentally (retention time values from HPLC, Table 1, Figures S3–S8). The HPLC method used was first validated using different lipophilic commercial compounds with LogP values from 1 to 4.5 (Figure S3a). The structure-activity relationships of the compounds will be further discussed in Section 3.

**Table 1.** Minimum inhibitory concentration (MIC, µg/mL), minimal bactericidal concentration (MBC, µg/mL), the half maximal inhibitory concentration IC<sup>50</sup> (µg/mL) values and the partition coefficient log P values and HPLC retention times for the different compounds.


<sup>a</sup> Average of values calculated using VCCLab and Molinspiration software. <sup>b</sup> Retention time in HPLC C18 reverse phase, CH3CN/H2O 70/30 (0.1% HCO2H). <sup>c</sup> Value for the protonated forms. <sup>d</sup> Calculated for the imidazolium cations. <sup>e</sup> MIC/MBC values were obtained from a minimum of three separate experiments. Please refer to Supporting Information for MIC/MBC/IC <sup>50</sup> values in µM. <sup>f</sup> Cytotoxicity against MRC-5 cells.

The MBC values obtained were plotted against the logP values of the corresponding compounds to study the possible correlation between activity and lipophilicity (Figure 1).

**Figure 1.** Plot of Log P vs. MBC. \* For the sake of clarity, MBC values > 2000 µg/mL are given the value of 3000 µg/mL.

1

Electron microscopy is a powerful tool to further assess the effect of the imidazole derivatives and imidazolium salts on bacterial cell growth, inhibition, and death. Two compounds which showed from moderate to good antibacterial activity—the bisimidazolium salt **1c** and the monoimidazolium salt **3b**, respectively—were chosen for electron microscopy characterization. For these studies, *E. coli* and *B. subtilis* were inoculated for 20 h with each compound at their corresponding MIC (Figure 2) and <sup>1</sup> <sup>2</sup> MIC (Figures S1 and S2) concentrations and fixed with glutaraldehyde.

μ

**Figure 2.** Scanning electron microscopy (SEM) images of *E. coli* and *B. subtilis* without treatment (-) and after incubation with compounds **1c** and **3b** at their corresponding MIC (60,000×). See Supporting Information for additional SEM images.

#### *2.3. Aggregation Studies*

The self-assembly of the compounds in aqueous medium and in the bacterial cell culture medium (LB broth) was investigated by optical and scanning electron microscopy, as well as UV-vis and fluorescence spectroscopy.

#### 2.3.1. Fluorescence Spectroscopy

To investigate the microenvironment of the critical aggregation concentration (CAC) for self-assembly in water and in the bacterial cell culture medium by fluorescence, the intensity ratio of two of the peaks (*I*1/*I*3) of the pyrene fluorescence spectrum was used [43–45]. Plots of the pyrene *I*1/*I*<sup>3</sup> ratio as a function of the total surfactant concentration show a typical sigmoidal decrease in the region where self-assembly takes place. At low concentrations, this ratio is larger as it corresponds to a polar environment for pyrene. When the surfactant concentration increases this ratio decreases rapidly, as the self-assembly favors the location of pyrene in a more hydrophobic environment, until reaching a roughly constant value because of the full incorporation of the probe into the hydrophobic region of the aggregates. Different approaches have been used to estimate CAC values from *I*1/*I*<sup>3</sup> ratios [46]. The most common approach is the use of the break points, either directly or by extrapolating the values from the intersection of the two straight lines defined at the constant and variable regions of the *I*1/*I*<sup>3</sup> sigmoidal curve [43,46–48]. As CAC represents the threshold of concentration at which self-aggregation starts, the corresponding value can be estimated from the break point at lower concentration (see Figures S9–S14) [49–52].

Fluorescence studies were carried out using MilliQ® water and 1/1 MilliQ® water/bacterial cell culture medium, because with the pure cell culture medium, a strong broad fluorescence emission band was observed precluding an accurate analysis. Furthermore, compound **3c** could not be studied due to solubility problems. The corresponding CACs obtained in water and in the 1/1 mixture of water/bacterial cell culture medium by fluorescence are shown in Table 2.

Furthermore, the MBC values of the compounds against *B. subtilis* were compared to their CAC (Figure 3). This comparison can shed light on the active form of the molecules exerting the antibacterial action, i.e., monomeric or aggregated structures.

**Table 2.** Estimated critical aggregation concentration (CAC) values obtained in aqueous and bacterial cell culture medium using fluorescence spectroscopy at 25 ◦C.


<sup>a</sup> CAC values from the break point. <sup>b</sup> In water. <sup>c</sup> In the 1/1 bacterial cell culture medium/water. <sup>d</sup> Low solubility.

**Figure 3.** Correlation of MBC for *B. subtilis* and CAC for the different series of compounds in the bacterial cell culture medium: (**a**) valine derivatives with long alkyl chain; (**b**) valine derivatives with short chain; and (**c**) phenylalanine derivatives with long alkyl chain.

#### 2.3.2. Optical Microscopy and Scanning Electron Microscopy (SEM)

The morphology of the aggregates in water, in 1/1 water/bacterial cell culture medium, and in the cell culture medium at concentrations above the CAC, were studied by optical microscopy (Figures S15–S17) and SEM (Figure 4 and Figure S18).

**Figure 4.** SEM images for **3b** (**a**) 0.5 mM in water; (**b**) 0.7 mM in 1/1 water/bacterial cell culture medium; and (**c**) 0.7 mM in the bacterial cell culture medium.

#### 2.3.3. UV-Vis Spectroscopy

The aggregation and stability of the aggregates in the different solvents, water, 1/1 water/bacterial cell culture, and bacterial cell culture medium for **1a–c** were studied by UV-vis at 25 ◦C measuring the absorbance at 600 nm, for 1 mM colloidal solutions (Figure 5, Figures S19 and S20) [53,54].

**Figure 5.** (**a**) Change in absorbance at 600 nm with respect to time for **1b** at 1 mM in different media. (**b**) Change in absorbance at 600 nm with respect to time for **1a–1c** at 1 mM in the bacterial cell culture medium.

#### **3. Discussion**

The Gram-positive cell wall is composed of a thick, multilayered peptidoglycan sheath outside of the cytoplasmic membrane, while the Gram-negative cell wall is composed of an outer membrane

μ

linked by lipoproteins to thin, mainly single-layered peptidoglycan. The peptidoglycan is located within the periplasmic space that is created between the outer and inner membranes [55]. It therefore follows that all the tested compounds were more active against *B. subtilis* (Gram-positive) than *E. coli* (Gram-negative) (i.e., see Table 1, entries 1 and 7). Compound **3b** (entry 8) proved to be the most active antibacterial agent possessing an MBC as low as 4 and 128 µg/mL against *B. subtilis* and *E. coli* respectively; while those compounds with shorter alkyl chains presented the lowest activity. Some of the compounds (Table 1, entries 2, 3, 6, 7, and 8) presented MIC values lower than MBC values. An antimicrobial compound is considered to be bactericidal whenever its MBC to MIC ratio is less than or equal to four. Compounds with MBC/MIC >4 are considered to be bacteriostatic [56]. In all the cases in which we could obtain exact MBC and MIC values (Table 1, entries 2, 3, 8 and 9), their ratio was equal to or minor than four. Therefore, compounds **1b**, **1c**, **3b**, and **3c** can be considered to possess true bactericidal activity.

The structural element which clearly relates to a better activity is the use of longer alkyl chains (compounds **1** and **3**) in contrast to shorter alkyl chains (compounds **2**), probably due to an increased lipophilicity. The toxicity towards miscellaneous bacterial strains of alkyl imidazolium salts has been reported to increase with the length of the alkyl chain [57]. The monoimidazolium salts with these longer alkyl chains (**1a-b** and **3a-b**) were more active against *B. subtilis* and *E. coli* than the bisimidazolium counterparts, with the monotopic salt **3b** showing the lowest MIC and MBC values (entries 2 and 8, Table 1), this indicates that the introduction of two hydrophobic alkyl chains contributes greatly to decrease the activity, opposite to the trends observed in the literature [27].

Regarding amino acid nature, the phenylalanine monotopic salt with long alkyl chain (**3b**) presented lower MIC and MBC values against *E. coli* than the analogous valine compound (**1b**) (entries 2 and 8, Table 1), however for ditopic salts, the behavior is the opposite, **3c** presented higher MIC and MBC values than **1c** (entries 3 and 9, Table 1).

Although a similar biological activity for the Gram-positive and Gram-negative organisms is preferred, it is not always the case with different strains of microorganisms. Some authors have described that Gram-positive organisms preferred a more lipophilic molecule than the Gram-negative ones [58]. This has been attributed to the difference in the cell outer membrane between bacterial types and strains: while Gram-positive bacteria have a very simple cell wall, the outer membrane of Gram-negative bacteria contains lipopolysaccharides which are cross-bridged by divalent cations, adding strength to the membrane and impermeability to lipophilic molecules. This agreed with the results obtained in Table 1 where the more lipophilic compounds showed less activity against *E. coli*.

A good correlation was obtained between the theoretical logP, calculated using the average values from VCCLab software and Molinspiration, and the retention time observed from HPLC (Figure S1). The activity observed against *B. subtilis* increases for compounds with logP > 3 with MBC ≤ 64 mg/mL. However, MBC values greater than 2000 µg/mL were obtained for both lipophilic and lipophobic compounds, with the exception of the lipophilic compounds **1b**, **1c**, and **3b** which present lower MBC values (Figure 1).

From the SEM images in Figure 2, both bacteria strains incubated with compounds **1c** and **3b** show clear signs of damage: from morphological changes to disruption of the cell membrane and leakage of cytoplasmatic material, ending with the disintegration of the bacteria into small fragments. Most images, especially of *B. subtilis*, show an "implosion" of bacteria, with a marked depression in the middle of the cell (also refer to Figures S1 and S2). In some of the images, aggregates of the compounds can be seen surrounding the bacteria, many of which are attached to the cell membrane.

Human embryonic kidney cells, HEK-293, were chosen as a cell model to evaluate the cytotoxicity of the compounds. The HEK-293 cells were incubated in the cell culture medium with varying concentrations of the examined compound and the impact of treatment was measured using the MTT cell viability assay [46]. The results indicated that compounds **1a–1c**, **3a**, and **3c** were considerably toxic with IC<sup>50</sup> values lower than 36 µM (Table 1 and Table S1). Surprisingly, compound **3b** derived

from phenylalanine was less toxic to the HEK-293 cell line at concentrations 15 times higher than the MIC and MBC for the *B. subtilis* strain (Table 1, entry 8) [59].

Comparing results with the commercial antibiotic alamethicin, the corresponding MIC value of **3b** against *B. subtilis* is lower than alamethicin, (Table 1, entry 10) [41], whereas the toxicity of alamethicin against MRC-5 human cells is similar to the cytotoxicity of the imidazolium salt **3b** against HEK-293 line [42].

To gain a more detailed understanding on the mechanism of cytotoxicity of these imidazole and imidazolium salts on bacteria, we studied the possible structure–activity relationship between their antibacterial activities and their aggregation behavior in bacterial cell medium [23,38].

From pyrene fluorescence studies in pure water, the plot of the *I*1/*I*<sup>3</sup> ratio for the corresponding emission spectra vs. concentrations showed one single break point (Figure S9, Figure S11, and Figure S13) reaching in all cases values of *I*1/*I*<sup>3</sup> ≈ 1.3 or lower after the break point. However, in the bacterial cell culture medium, the plot of the *I*1/*I*<sup>3</sup> ratios for the corresponding emission spectra vs. concentration presented two single break point and, in some cases, even three points. The two break points observed for all the compounds suggest the presence of two different processes. The first one takes place at *I*1/*I*<sup>3</sup> values observed for pyrene in the absence of compounds (*I*1/*I*<sup>3</sup> = 1.16 in water/1/1 bacterial cell culture medium) and reveals that pyrene is fully exposed to the polar solvent mixture in this first aggregation step. At the second break point, at higher concentrations, this ratio reaches lower values suggesting the formation in this region of aggregates in which the probe molecule is less solvent exposed [60]. For example, for compound **1b**, the first break point at 6 µM leads to aggregates with an appreciable solvent exposed probe (*I*1/*I*<sup>3</sup> ≈ 1.1) and the second process starts at *ca.* 48 µM affording aggregates, providing a low polarity microenvironment to pyrene, reaching *I*1/*I*<sup>3</sup> values ≈ 0.8 for 1 mM concentration.

The CAC values obtained for compounds with long alkyl chain were in the µM range while for compounds **2a-c** with short alkyl chains, the CAC values obtained were in the mM range (i.e., entries 1 and 4, Table 2), where the lowest CAC values for the ditopic salts were found in water. In general, for the imidazole and monotopic salts, changing the medium from water to water/bacterial cell culture medium led to a decrease in the CAC values (i.e., Table 2, entries 1 and 2), however for ditopic salts, the CAC values did change significantly (Table 2, entries 3 and 6).

Comparing the first CAC values obtained in 1/1 water/bacterial cell culture medium and the MBC for *B. subtilis* for the different series of compounds in Figure 3, it can be observed as compounds **1a–1c** presented the CAC below the MBC, implying that these compounds exist in an aggregated form at the MBC concentration, meaning fewer imidazolium monomers will be present at these concentrations, less than is needed to produce a significant biologic effect, thus increased overall concentrations are needed to obtain the desired bactericidal effects if the monomeric form is the responsibility of the corresponding bioactivity. However, different behavior was observed for the series **3a–3b** derived from phenylalanine. Figure 3c shows how the CAC line intersects the MBC line, indicating that compound **3b** is not aggregated at the corresponding MBC value and exerts a high bactericidal effect, as observed in Table 2 (entry 8). Finally, for the **2a–2c** series, the CAC is below the MBC for **2a** but above for **2b** and **2c**, illustrating that the monomeric form is responsible for the corresponding antibacterial activity (Table 2, entry 6).

In addition to the results above, the compounds containing dodecyl chains can easily align with lipids and hence, accumulate within the bacterial cell membrane. In this regard, compounds with longer alkyl tails have CACs in the µM range in the bacterial cell culture medium and thus easily self-assemble, leading to an easy accumulation within the cell membrane. Therefore, resulting in a lower effective concentration at the site of action within the cellular cytoplasm lower. This accumulation could lead to a biocidal mechanism based on [38], as is observed in Figure 2. Furthermore, it appears that the shorter chain length results in reduced membrane interaction and an energetically unfavorable micelle formation, meaning low self-assembling capability, which leads to a lower overall bacterial cytotoxicity as seen from the corresponding MIC and MBC values (Table 1).

Consequently, by comparing the MBC for *B. subtilis* and CAC of the imidazole and imidazolium series, this study has provided a better understanding of the relationship between the biological activities of these compounds correlated with their aggregation capabilities. The results demonstrate that the compounds with longer alkyl chains provide excellent antimicrobial activity although most of them are aggregated at the antimicrobial response concentration, with only compound **3b** existing in its monomeric form at its corresponding MBC value.

Optical microscopy confirmed the formation of spherical aggregates between 0.5–20 µm diameter in size in the three different media (Figures S15–S17). Regarding the medium, in general, in the culture medium, the dispersity of the aggregates decreased for compounds with longer alkyl chains (see Figures S15 and S17). Furthermore, in the culture medium, compounds **2a–c** and **3b** were able to form worm-like aggregates at the studied concentrations (Figures S16 and S17).

SEM images for **1c** and **3b** in water, in 1/1 water/bacterial cell culture medium and in the bacterial cell culture medium revealed the formation of different aggregates morphologies. Compound **1c** produced spherical aggregates with <3 µm diameter size in all three media, with the aggregates in water being more distorted (Figure S18). Compound **3b** was able to form spherical aggregates in water and water/bacterial cell culture medium (Figure 4a,b), while in the pure culture medium, different morphologies were observed, such as spherical aggregates <1 µm in diameter coexisting with fibrillary aggregates (Figure 4c). When viewed at higher magnification, it is observed that the larger spherical aggregates consisted of several smaller aggregates or dendritic fibrillary aggregates for **3b** and **1c,** respectively (Figure 4 and Figure S18).

Regarding the stability of the aggregates formed, studies by UV-vis spectroscopy for **1a–c** are gathered in Figure 5. Figure 5a shows the change in absorbance (−∆A600) of compound **1b** (1 mM) at 600 nm with respect to time in the different media. The initial rate for the change in absorbance associated to the destabilization of the aggregates is defined by the slope of the linear region of the initial −∆A<sup>600</sup> versus time plot. The slope and the total change in the absorbance was the smallest when using the water/bacterial cell culture medium, being the rate at the initial region for pure water and bacterial cell culture medium in the same ranges. However, it must be highlighted that the absorbance decreases until reaching a zero value after 500 min for pure water. A different behavior was obtained for **1a** and **1c** (see Figures S19 and S20). For compound **1a**, the absorbance at 600 nm decreased with time in the three media with similar rates, while for **1c**, the rate followed the order water > bacterial cell culture medium > 1/1 water/bacterial cell culture medium, reaching almost zero values in the tree media after 24 h.

Overall, the results obtained show that in the pure culture medium, the stability of the aggregates follows the order **1a** > **1b** > **1c** (Figure 5b), indicating that the introduction of the two headgroups and hydrophobic alkyl chains in **1c** contributes to a minor stabilization of the aggregates in this medium.

#### **4. Materials and Methods**

#### *4.1. Materials*

#### 4.1.1. Reagents and Culture Media

Resazurin sodium salt and dimethyl sulfoxide (DMSO) were bought from Sigma-Aldrich. Luria-Bertani (LB) liquid broth (Miller's formulation) and nutrient broth (NB) were freshly prepared and sterilized by autoclave. Broth powders were bought from Scharlab. Tryptone soy agar plates were purchased from Thermo Scientific. Glutaraldehyde was purchased in solution at 25% in H2O and Grade II from Sigma Aldrich and used as provided. Phosphate buffer was prepared from the solid salts NaH2PO<sup>4</sup> and Na2HPO4, both purchased from Aldrich at qualities 99% and 99.5% respectively, by dissolving them in MilliQ® water and adjusting pH with NaOH and HCl solutions.

Cell culture media for cytotoxicity studies were purchased from Gibco (Grand Island, NY, USA). Fetal bovine serum (FBS) was obtained from HyClone (UT, USA). Supplements and other chemicals not listed in this section were obtained from Sigma Chemical Co. (St. Louis, MO, USA). Plastics for

cell culture were supplied by Thermo Scientific BioLite (Madrid, Spain). All tested compounds were dissolved in DMSO at a concentration of 10 mM and stored at −20 ◦C until use. HEK-293 cell lines were maintained in Dulbecco's modified Eagle's medium (DMEM) containing glucose (1 g/L), glutamine (2 mM), penicillin (50 µg/mL), streptomycin (50 µg/mL), and amphotericin B (1.25 µg/mL) supplemented with 10% FBS.

Reagents and solvents, including NMR solvents, were purchased from commercial suppliers and were used without further purification except for pyrene, used for fluorescence studies, that was crystallized twice from methanol. Deionized water was obtained from a MilliQ® equipment (Burlington, MA, USA). Imidazoles **1a** and **2a** and imidazolium salts **1b** and **1c** were prepared as previously described [30,32].

#### 4.1.2. Synthesis and Characterization

Imidazole **3a** and compounds **2b–c** and **3b–c** were prepared following the synthetic protocols.

*General procedure for compound 3a:* To a mixture of glyoxal (40% aq., 1.1 equiv, 2.6 mL) and formaldehyde (37% aq., 5.0 equiv., 7.7 mL), the (*S*)-2- amino-N-dodecyl-3- phenylpropanamide compound (1.0 equiv., 6.9 g, 20.8 mmol) and ammonium acetate (1.1 equiv, 1.8 g, 23.4 mmol) were dissolved previously in methanol and added. The reaction mixture was stirred at room temperature for 48 h. The solvent was evaporated under reduced pressure and the resulting crude residue was treated with saturated Na2CO<sup>3</sup> solution, extracted with CH2Cl<sup>2</sup> (3×), dried with anhydrous MgSO4, filtered, and concentrated.

*General procedure for compounds 2b–3b:* To a mixture of compound **2a–3a** (1.1 equiv) and bromomethylbencene (1.0 equiv) were dissolved in acetonitrile (5 mL). The reaction was carried out under microwave irradiation using 120 W, 1.72 × 10<sup>6</sup> Pa, 150 ◦C, and 1 h. After solvent evaporation, the remaining solid was washed with diethyl ether (×3) to afford the desired compound.

*General procedure for compounds 2c–3c:* To a mixture of compound **2a–3a** (2.2 equiv) and 1, 3-(bis-bromomethyl)benzene (1.0 equiv) were dissolved in acetonitrile (5 mL). The reaction was carried out under microwave irradiation using 120 W, 1.72 × 10<sup>6</sup> Pa, 150 ◦C, and 1 h. After solvent evaporation, the remaining solid was washed with diethyl ether (×3) to afford the desired compound.

*Compound 3a:* yellow liquid (7 g, 88%), [α] 25 <sup>D</sup> = −7.11 (c = 0.01, MeOH); m.p= 56.1 ◦C. <sup>1</sup>H NMR (400 MHz, CDCl<sup>3</sup> and CD3OD) δ 7.26–7.07 (m, 4H), 7.00 (m, 1H), 6.97–6.89 (m, 3H), 6.48 (t, J = 5.7 Hz, NH), 4.67 (dd, J = 9.1, 6.0 Hz, 1H), 3.46 (dd, J = 14.0, 6.0 Hz, 1H), 3.19–3.00 (m, 3H), 1.30 (m, 2H), 1.26–1.04 (m, 18H), 0.85–0.77 (m, 3H). <sup>13</sup>C NMR (101 MHz, CDCl3) δ 168.3, 134.0, 136.3, 129.6, 128.8, 128.7, 127.2, 118.0, 62.9, 39.8, 39.3, 31.9, 29.6, 29.6, 29.5, 29.3, 29.2, 26.8, 22.7, 14.1. MS (ESI) (m/z) calcd. for C24H37N3O [M+H]<sup>+</sup> = 384.3; found 383.4 (100%), 767.7 (35%, [M+M+H]+). IR (ATR) = 3309, 2953, 2919, 2850, 1656, 1549, 1493, 1469, 1454 cm−<sup>1</sup> . Calculated for C24H37N3O·4H2O: C 63.27, H 9.96, N 9.22; found C 62.97, H 9.74, N 9.58.

*Compound 2b:* yellow oil (150 mg, 93%); [α] 25 <sup>D</sup> = 41.07 (c = 0.01, MeOH); m.p. 33 ◦C. <sup>1</sup>H NMR (400 MHz, CDCl3) δ 9.93 (s, 1H), 8.62 (t, J = 5.7 Hz, 1H), 7.75 (t, J = 1.8 Hz, 1H), 7.38–7.21 (m, 5H), 7.11 (s, 1H), 5.66 (d, J = 10.6 Hz, 1H), 5.47–5.27 (dd, J = 10.6, 2.2 Hz, 2H), 3.34–3.17 (m, 1H), 3.12–2.96 (m, 1H), 2.45–2.37 (m, 1H), 1.54–1.40 (m, 2H), 1.32–1.21 (m, 2H), 1.03 (d, J = 6.5 Hz, 3H), 0.81 (t, J = 7.3 Hz, 3H), 0.76 (d, J = 6.6 Hz, 3H). <sup>13</sup>C NMR (101 MHz, CDCl3) δ 167.0, 136.1, 131.8, 130.1, 129.8, 128.6, 121.8, 120.9, 67.7, 53.9, 39.6, 31.2, 31.0, 20.2, 18.8, 18.3, 13.7. MS (ESI) (m/z) calcd. for C19H28N3O [M]<sup>+</sup> = 314.2; found 314.5 (100%); IR (ATR)= 3220, 3063, 2962, 2933, 2873, 1672, 1550, 1497, 1456, 1327, 1225, 1152 cm−<sup>1</sup> . Calculated for C19H28N3OBr: C 57.87, H 7.16, N 10.66; found C 57.80, H 6.98, N 11.01.

*Compound 3b:* yellow viscous solid (129 mg, 90%); [α] 25 <sup>D</sup> = −31.07 (c = 0.01, MeOH); m.p = 16 ◦C. <sup>1</sup>H NMR (300 MHz, CDCl3) δ 9.53 (s, 1H), 8.59 (t, J = 5.5 Hz, NH), 7.74 (s, 1H), 7.41–7.14 (m, 7H), 7.04–6.95 (m, 2H), 6.92 (s, 1H), 6.55 (m, 1H), 5.14 (q, J = 14.7 Hz, 2H), 3.51–2.95 (m, 4H), 1.83 (s, 3H), 1.44 (m, 2H), 1.17 (s, 16H), 0.92–0.70 (m, 3H). <sup>13</sup>C NMR (101 MHz, CDCl3) δ 166.6, 136.1, 134.3, 131.9, 129.8, 129.7, 129.1, 129.0, 128.2, 127.5, 121.8, 120.9, 62.3, 53.6, 40.0, 39.0, 31.9, 29.7, 29.6, 29.5, 29.4, 29.2, 28.9, 27.0, 22.7, 14.1. MS (ESI) (m/z) calcd. for C31H44N3O [M]<sup>+</sup> = 474.4; found 474.7 (100%); IR (ATR) = 3297, 3061, 2966, 2922, 2851, 1654, 1557, 1495, 1453 cm−<sup>1</sup> . Calculated for C31H44N3OBr·H2O: C 65.02, H 8.10, N 7.34; found C 65.46, H 8.24, N 7.68.

*Compound 2c:* yellow oil (104 mg, 90%); [α] 25 <sup>D</sup> = 6.93 (c = 0.01, MeOH); m.p.= 85 ◦C. <sup>1</sup>H NMR (500 MHz, CDCl3) δ 10.01 (s, 2H), 8.40 (t, J = 5.6 Hz, 2H), 8.12 (s, 2H), 7.68 (d, J = 1.3 Hz, 2H), 7.47–7.36 (m, 2H), 7.34–7.23 (m, 2H), 5.73 (d, J = 14.4 Hz, 2H), 5.55–5.38 (dd, J = 10.7 Hz, J = 3.5 Hz, 4H), 3.37–3.25 (m, 2H), 3.15–3.00 (m, 2H), 2.53–2.38 (m, 2H), 1.64–1.43 (m, 4H), 1.39–1.20 (m, 4H), 1.08 (d, J = 6.6 Hz, 6H), 0.87 (t, J = 7.3 Hz, 6H), 0.82 (d, J = 6.7 Hz, 6H). 13C NMR (126 MHz, CDCl3) δ 206.8, 166.7, 136.5, 134.1, 130.6, 130.5, 129.9, 122.4, 120.8, 77.3, 77.0, 76.8, 68.1, 53.2, 39.5, 31.1, 30.9, 30.9, 20.1, 18.8, 18.4, 13.6. MS (ESI) (m/z) calcd. for C32H50N6O<sup>2</sup> [M]2<sup>+</sup> = 275.2; found 275.3 (100%); IR (ATR)= 3412, 3228, 3125, 3066, 2962, 2933, 2873, 1671, 1550, 1465, 1360, 1298, 1226, 1152 cm−<sup>1</sup> . Calculated for C32H50N6O2Br2·2H2O: C 51.48, H 7.29, N 11.26; found C 50.84, H 7.32, N 11.46.

*Compound 3c:* yellow viscous solid (126 mg, 82%); [α] 25 <sup>D</sup> = 17.33 (c = 0.01, MeOH); m.p. = 60 ◦C. <sup>1</sup>H NMR (300 MHz, CDCl3) δ 9.47 (s, 2H), 8.21 (t, J = 5.7 Hz, NH), 7.81–7.51 (m, 6H), 7.29–7.06 (m, 16H), 6.13 (t, J = 8.0 Hz, 2H), 5.31 (s, 4H), 3.37 (dd, J = 13.6, 7.2 Hz, 2H), 3.25–3.10 (m, 4H), 3.00–2.84 (m, 2H), 1.41–1.29 (m, 4H), 1.23–1.05 (m, 32H), 0.80 (m, 6H). <sup>13</sup>C NMR (101 MHz, CDCl3) δ 166.3, 136.3, 134.3, 133.6, 130.6, 130.3, 129.7, 129.2, 128.9, 127.5, 122.3, 121.0, 62.6, 53.0, 39.9, 38.9, 31.9, 29.7, 29.7, 29.6, 29.5, 29.4, 29.2, 28.9, 26.9, 22.7, 14.1. MS (ESI) (m/z) calcd. for C56H82N6O<sup>2</sup> [M]2<sup>+</sup> = 435.3; found 435.7 (100%); IR (ATR)= 3294, 3063, 2923, 2852, 1656, 1554, 1495, 1454, 1362 cm−<sup>1</sup> . Calculated for C56H82N6O2Br2: C 65.23, H 8.02, N 8.15; found C 65.76, H 8.57, N 8.34.

#### 4.1.3. Microorganisms and Growth Conditions

Two bacterial strains were used in the antibacterial assays: *Escherichia coli* DH5α as a Gram-negative model and *Bacillus subtilis* 1904-E as a Gram-positive model. Both bacterial strains were donated to our laboratory and can be provided on request by contacting the corresponding authors. Both bacterial strains were incubated at 37 ◦C and the pre-inoculum incubation time was of 24 h. Liquid Luria-Bertani (LB) medium was used for *E. coli* DH5α and nutrient broth (NB) for *B. subtilis*.

#### *4.2. Methods*

#### 4.2.1. Bacterial Proliferation Assay in Presence of Imidazole Derivatives

The bacteria cell bank suspensions were thawed and inoculated in the appropriate liquid broth for 24 h at 37 ◦C with mild agitation. A dilution from these culture solutions was used for the following tests, corresponding to an inoculum of 1 × 10<sup>7</sup> CFU/mL. Stock solutions of all the tested compounds were prepared in DMSO at a concentration of 100 mg/mL, aliquoted, and stored at −20 ◦C.

*(A) Bacterial growth inhibition assay:* Conditions here described are for testing 6 different concentrations of the compounds, with triplicates of each condition. Therefore, 4 compounds were tested per plate. An adapted version of the microdilution method was used. Firstly, the imidazole derivatives were dissolved in the corresponding broth at 2× the highest tested concentration. Then, 100 µL of the 2× solutions were added to the first (A) and second (B) row wells of a 96-well plate. In addition, 100 µL of liquid medium had been previously added to rows B to F. Subsequent dilutions at 1:2 are prepared in rows B to F, by withdrawal of 100 µL from the previous row (more concentrated) to the next row (half diluted), mixing well. Then, 100 µL were discarded from the last row (F). By now, there are 100 µL in each well, and 100 µL of bacterial suspension at 10<sup>7</sup> CFU/mL were added to each well. Then, the 96-well plates were incubated for 24 h at 37 ◦C under mild agitation. Bacterial growth was controlled both by visual observation of the turbidity in each well and by measuring the optical density (OD) at 560 nm at time 0 h and 24 h. Results are recorded as the lowest concentration of antimicrobial agent that inhibits visible growth of the bacteria, and were compared with the OD variation of a control culture containing *E. coli* or *B. subtilis* (+ control) and of solution of the tested compounds without bacteria (- control).

*(B) Bacterial cell viability assay:* Cell viability was analyzed using a Resazurin (7-Hydroxy-3Hphenoxazin-3-one 10-oxide) assay in a 96-well plate. Once the bacterial cultures of growth inhibition assay had been grown for a total of 24 h, 25 µL of a 0.1 mg/mL resazurin (prepared in LB or NB medium) were added to each well and incubated in the dark at 37 ◦C for 1 h under stirring. Resazurin has a blue color at the testing pH and turns pink when reduced by the viable bacteria to resorufin. Therefore, pink wells indicate metabolizing bacteria, while blue wells are indicative of bacteria that have lost their ability to convert resazurin to resorufin. Different controls were made in order to corroborate the MBC value obtained by the resazurin assay. The change of color was confirmed at 1, 4, and 24 h after its addition. The viability of bacteria was verified (either confirmed or rejected) by the colony plate-counting method, by seeding 10 µL from the cell culture onto tryptone soy agar plates and observing the presence or absence of bacterial growth after 24 h at 37 ◦C.

#### 4.2.2. Log P Calculation and Retention Time Determination

LogP values for the different compounds were calculated using VCCLab (ALOGPS 2.1) and Molinspiration (miLogP2.2) softwares. We used LogP as the average of these values. The protonated forms for the imidazole derivatives were considered. Reverse phase HPLC (equipment: Agilent technologies 1100 series, column: Xterra MS C18 4.6 × 150 mm (5 µmol/L)) was also used for measuring the relative lipophilicity of these compounds, since the retention time of each molecule on the reverse phase column is related to its lipophilicity. All the products were dissolved in MeOH at 2 mmol/L concentration and eluted using 70/30 acetonitrile/water and 0.1% of formic acid for 15 min and flow rate 0.2 mL/min at 25 ◦C. λ used was 254, 280, and 220 nm taking the corresponding chromatogram with higher mAU (see Supporting Information).

#### 4.2.3. H NMR Studies

NMR experiments were carried out on a Varian INOVA 500 spectrometer (500 MHz for <sup>1</sup>H and 125 MHz for <sup>13</sup>C), on a Bruker Avance III HD 400 spectrometer (400 MHz for <sup>1</sup>H and 100 MHz for <sup>13</sup>C) or on a Bruker Avance III HD 300 spectrometer (300 MHz for <sup>1</sup>H and 75 MHz for <sup>13</sup>C) at 25 ◦C. Chemical shifts are reported in ppm using TMS as the reference.

#### 4.2.4. Fluorescence Spectroscopy Measurements

Pyrene was used as a fluorescence probe to determine the CAC of the compounds in water and 1/1 water/bacterial cell culture medium at 25 ± 1 ◦C. Fluorescence measurements were performed with a Spex Fluorolog 3-11 instrument equipped with a 450 W xenon lamp (right angle mode). Firstly, a stock pyrene solution of 1.98 × 10−<sup>4</sup> mol/L was prepared in ultrapure methanol. Then, solutions of the imidazole and imidazolium salt compounds (ranging from 6 to 3 × 10−<sup>3</sup> mmol/L) were prepared in different vials and 5 µL of pyrene solution was transferred into the vials, reaching a final pyrene concentration of 9.89 × 10−<sup>7</sup> mol/L in each vial. Fluorescence spectra of pyrene were recorded from 200 to 650 nm after excitation at 337 nm, and the spectra were not corrected for the Xe lamp spectral response. The slit width was set at 5 nm for both excitation and emission. The peak intensities at 373 and 385 nm were determined as *I*<sup>1</sup> and *I*3, respectively. The ratios of the peak intensities at 373 and 385 nm (*I*1/*I*3) for the emission spectra were recorded as a function of the logarithm of concentration. The CAC values were taken from the break point. Samples were excited with a 337 nm NanoLED.

#### 4.2.5. Optical Images

Images were recorded with OLYMPUS COVER-018 microscopy, BX51TF model, at 25 ◦C. Experiments were carried out in water, 1/1 water/bacterial cell culture medium, and in bacterial cell culture medium.

#### 4.2.6. Scanning Electron Microscopy (SEM)

SEM images of the compounds were obtained using a JEOL 7001F microscope with a digital camera; while SEM images of the incubated bacteria were obtained using an Inspect F50 microscope, at 10 kV and spot size of 3.0, with a digital camera. Bacteria solutions at ca. 0.5 × 10<sup>7</sup> CFU/mL were incubated overnight without and with compounds **1c** and **3b** at their <sup>1</sup> <sup>2</sup> MIC and MIC. After this, bacteria were washed with sterile PBS and fixed by incubation for 2 h in a 2.5% glutaraldehyde solution in phosphate buffer 10 mmol/L at pH 7.2. The fixed bacteria were subsequently washed once with phosphate buffer saline solution and four times with MilliQ water to remove any residual salts and glutaraldehyde. Finally, bacteria were resuspended in MilliQ water and 10 µL of these solutions were placed on silicon wafers and allowed to dry by evaporation overnight. Samples were coated with platinum using the sputtering technique in which microscopic particles of platinum are rejected from the surface after the material is itself bombarded by energetic particles of a plasma or gas. Experiments were carried out in water, 1/1 water/bacterial cell culture medium, and in bacterial cell culture medium.

#### 4.2.7. UV-Vis Spectroscopy

UV-Vis absorption spectra of the colloidal solutions were recorded on a Hewlett-Packard 8453 spectrophotometer at 25 ◦C. Experiments were carried out in water, 1/1 water/bacterial cell culture medium, and in bacterial cell culture medium.

#### 4.2.8. Cell Proliferation Assay for Cytotoxicity Studies

In 96-well plates, 3 × 10<sup>3</sup> HEK-293 cells per well were seeded and incubated with serial dilutions of the tested compounds (from 200 to 0.2 µM) to a total volume of 100 µL of their growth media. The 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT; Sigma Chemical Co.) dye reduction assay in 96-well microplates was used, as previously described [18]. After 2 days of incubation (37 ◦C, 5% CO<sup>2</sup> in a humid atmosphere), 10 µL of MTT (5 mg/mL in phosphate-buffered saline, PBS) was added to each well, and the plate was incubated for a further 3 h (37 ◦C). The supernatant was discarded and replaced by 100 µL of DMSO to dissolve formazan crystals. The absorbance was then read at 540 nm by MultiskanTM FC microplate reader. For all concentrations of compound, cell viability was expressed as the percentage of the ratio between the mean absorbance of treated cells and the mean absorbance of untreated cells. Three independent experiments were performed, and the IC<sup>50</sup> values (i.e., concentration half inhibiting cell proliferation) were graphically determined using GraphPad Prism 4 software.

Statistical analysis: GraphPad Prism v4.0 software (GraphPad Software Inc., La Jolla, CA, USA) was used for statistical analysis. For all experiments, the obtained results of the triplicates were represented as means with standard deviation (SD).

#### **5. Conclusions**

A series of novel imidazole and imidazolium salts derived from *L*-valine and *L*-phenylalanine containing different hydrophobic groups have been synthesized and their antibacterial activity studied against *E. coli* and *B. subtilis*. The results demonstrate that an optimum lipophilicity of the alkyl chain and the amino acid side chain is needed to achieve antibacterial activity. The compounds presented better antibacterial activity against *B. subtilis* than *E. coli*, where compound **1a–1b** and **3a–3b** were the most active against *B. subtilis*, showing MBC values corresponding to 16 µg/mL or lower. Monotopic compound **3b** was 15 times less active against human embryonic kidney cells HEK-293 than toward *B. subtilis,* thus demonstrating its potential as an effective antibacterial agent with good biocompatibility. Aqueous aggregation studies revealed CAC values for compounds **1a–1c** and **3a–3c** in the µM range in water alone, however these CAC values decreased for imidazole and monotopic species when water was replaced with bacterial cell culture medium. Optical microscopy and SEM

images confirmed the formation of these spherical aggregates. It is important to note that most of the bioactive compounds were aggregated to some extent at their MIC/MBC concentrations, however the monotopic compound **3b** was not aggregated at its corresponding MBC, suggesting that the monomeric species was responsible for the observed antibacterial activity.

#### **Supplementary Materials:** The following are available online at http://www.mdpi.com/1424-8247/13/12/482/s1.

**Author Contributions:** Conceptualization, B.A. and S.G.M.; methodology, B.A. and S.G.M.; software, A.V. and J.J.A.; validation, B.A., S.G.M. and E.A.-B.; formal analysis, B.A.; investigation, A.V., J.J.A., E.A.-B., and E.F.; resources, B.A.; data curation, S.G.M.; writing—original draft preparation, B.A.; writing—review and editing, B.A., S.G.M. and E.A.-B.; supervision, B.A., S.G.M. and S.V.L.; funding acquisition, S.V.L., E.A.-B., and S.G.M. All authors have read and agreed to the published version of the manuscript.

**Funding:** E.A.B. and S.G.M acknowledge funding from the Ministerio de Ciencia e Innovación (Spain) (PID2019-109333RB-I00) and the European Union's Horizon 2020 research and innovation program (Marie Skłodowska-Curie grant agreement No 845427). S.V.L and B.A. acknowledge funding from Ministerio de Ciencia e Innovación, RTI2018-098233-B-C22 and Pla de Promoció de la Investigació de la Universitat Jaume I, UJI-B2019-40. A.V. was funded by Ministerio de Ciencia e Innovación within the predoctoral fellowship program, grant FPU15/01191.

**Acknowledgments:** The electron microscopy characterization was conducted at the Laboratorio de Microscopias Avanzadas (LMA) at Universidad de Zaragoza. Authors acknowledge the LMA for offering access to their instruments and expertise. Technical support from the SECIC of the UJI is acknowledged.

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

#### **References**


**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
