**New Insights on Biofilm Antimicrobial Strategies**

Editors

**Lu´ıs Melo Nuno F. Azevedo**

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

*Editors* Lu´ıs Melo Centre of Biological Engineering University of Minho Braga Portugal Nuno F. Azevedo LEPABE – Dep. of Chemical Engineering Faculty of Engineering, University of Porto Porto Portugal

*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 *Antibiotics* (ISSN 2079-6382) (available at: www.mdpi.com/journal/antibiotics/special issues/ biofilm antimicro).

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 978-3-0365-1221-1 (Hbk) ISBN 978-3-0365-1220-4 (PDF)**

© 2021 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: *Antibiotics* **2020**, *9*, 799, doi:10.3390/antibiotics9110799 ............... **153**

v


### **About the Editors**

### **Lu´ıs Melo**

With a degree and MSc in biology and a post-graduation qualification in nethods of DNA analysis, he obtained his PhD in biomedical engineering at the University of Minho, where he is currently a junior researcher. Luis D.R. Melo established his research on exploring phages' interactions with bacterial biofilms and has demonstrated the value of phage proteins on the control and detection of pathogenic bacteria. His experience in flow cytometry, genomics, transcriptomics, and proteomics has led to an increased knowledge on phage/host interactions. He has been involved in different scientifically funded projects, being the PI of two. He is also involved in the organization of international practical courses and conferences within this field of knowledge. The quality and innovative character of the research developed led to the publication of several internationally referenced papers and book chapters and enabled the creation of a patent.

### **Nuno F. Azevedo**

Nuno F. Azevedo (PhD) is an assistant professor at the Faculty of Engineering of the University of Porto (FEUP) and a researcher at the Laboratory for Process Engineering, Environment, Biotechnology, and Energy (LEPABE). His main research interests include exploring the potential of nucleic acid mimics (NAMs) for rapid diagnosis and treatment of infectious agents in multispecies biofilms. He is currently leading the EU-funded project DelNAM, a project that aims to develop a novel therapeutic approach to solve bacterial resistance to antibiotics through the delivery of antibacterial NAMs into bacterial biofilms and cells within the human body. During his research career, he has authored or co-authored more than 100 papers in peer-reviewed international journals, submitted seven patents, and co-edited two books. He was also a co-founder of the biotech company Biomode SA, which was granted >1.5M€ in investment funds.

### *Editorial* **New Insights on Biofilm Antimicrobial Strategies**

**Luís D. R. Melo 1,\* and Nuno F. Azevedo 2,\***


Over the last few decades, the study of microbial biofilms has been gaining interest among the scientific community. These microbial communities comprise cells adhered to surfaces that are surrounded by a self-produced exopolymeric matrix that protects biofilm cells against different external stresses. Biofilms can have a negative impact on different sectors within society, namely in agriculture, food industries, and veterinary and human health. As a consequence of their metabolic state and matrix protection, biofilm cells are very difficult to tackle with antibiotics or chemical disinfectants. Due to this problem, recent advances in the development of antibiotic alternatives or complementary strategies to prevent or control biofilms have been reported. This Special Issue includes different strategies to prevent biofilm formation or control biofilm development and includes full research articles, reviews, a communication, and a perspective.

Regarding the problem per se, Uruén and Chopo-Escuin et al. [1] reviewed the mechanisms by which biofilms are tolerant or resistant to antibiotics, emphasizing the role of the biofilm matrix, physiological heterogeneity of biofilm cells, quorum sensing, horizontal gene transfer, and other mutations on biofilms. In the second part of the review, several alternatives to combat biofilms were discussed. The problem of bacterial resistance was assessed in an original study by Shenkutie et al. [2], where the biofilm-forming ability of 104 *Acinetobacter baumannii* clinical strains was evaluated. Moreover, the authors observed that the minimum biofilm eradication concentrations were significantly higher than the minimum bactericidal concentrations for several antibiotics, a fact that led to an increase in persister cell detection on biofilms. In another study, the influence of *Escherichia coli* diversity in biofilms composed of up to six different strains isolated from urine was evaluated in urinary tract infection conditions. The authors detected that as the number of strains increased, the number of culturable cells also increased but overall the biofilms produced less matrix [3]. The impact of the biofilm matrix on flow cytometry in multi-species biofilms was one of the parameters evaluated by Grainha et al. [4]. Despite the potential of this technique to assess several aspects of biofilms, the authors reported that results are very dependent of the microbial strain used, the morphological state of the cells, and the biofilm matrix.

Another important topic covered in this Special Issue is the prevention of biofilm formation. Alves et al. studied the initial events of *E. coli* adhesion to polydimethylsiloxane and demonstrated that a proper tuning of operational parameters is required to avoid hydrodynamic blocking, which will allow the scientific community to obtain reliable data about cell–surface interactions [5]. In another study performed by the same group, the authors show the effect of pristine and functionalized carbon nanotube incorporation into poly(dimethylsiloxane) materials. Initial *E. coli* adhesion was assessed in conditions simulating urinary tract devices (catheters and stents). The results led the authors to conclude that the incorporation of carbon nanotubes, even at low loading values, might be beneficial for the application of biomedical devices for the urinary tract [6].

**Citation:** Melo, L.D.R.; Azevedo, N.F. New Insights on Biofilm Antimicrobial Strategies. *Antibiotics* **2021**, *10*, 407. https://doi.org/ 10.3390/antibiotics10040407

Received: 26 March 2021 Accepted: 6 April 2021 Published: 9 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/).

Trøstrup et al. and Brum et al. contributed to this Special Issue with two reviews [7,8]. The first is focused on the impact of Pseudomonas aeruginosa biofilms on the local and systemic host response observed in vitro and in vivo. The authors also discussed the implications for clinical wound healing and a possible therapeutic approach using an antimicrobial peptide as immunomodulatory topical treatment [8]. In the second review, a comparison between the use of polyether-ether-ketone (PEEK) and other commonly used materials in implant dentistry (titanium and zirconia) as biofilm-preventing or -controlling agents was comprehensively conducted. The authors concluded that despite pure PEEK being susceptible to biofilm formation, there are numerous strategies that can improve its antibiofilm properties, namely PEEK sulfonation, incorporation of therapeutic/bioactive agents in the PEEK matrix or surface, PEEK coatings, and, finally, the incorporation of reinforcement agents [7].

Different approaches to control biofilms using antibiotic alternative strategies were also submitted to this Special Issue. James D. Boub provided a perspective on the use of phage therapy to combat infectious biofilms [9]. The perspective referred to many aspects of bacteriophage therapy that should be taken into consideration before their broader use. The author suggested the development of standardized protocols that will allow for better and stricter testing of this therapeutic agent in the treatment of biofilm infections. Regarding this topic, Oliveira et al. used a bacteriophage cocktail to control *P. aeruginosa* biofilm formation on endotracheal tubes [10]. Despite some promising results on reducing bacterial colonization, the authors concluded that this strategy could have more potential with the development of new coating strategies. Several natural products are also commonly seen as promising antibiofilm agents. Hoang et al. analyzed the composition and antibiofilm activity of 15 methanolic extracts from *Iris* spp. [11]. *Iris pallida* s.l. leaf extract was the most effective at both preventing biofilm formation and controlling multispecies oral biofilms, with no toxicity observed, suggesting its potential application for oral biofilms. In another study, the antibiofilm activity of cyanobacteria *Arthrospira platensis* extracts (free and nanovectorized) was studied on *Candida albicans* and *Cutibacterium acnes* (single- and dual-species biofilms). Efficacy results varied depending on the microbial species and on the type of biofilm, emphasizing the importance of studying more complex communities such as polymicrobial biofilms [12].

Using a different approach, Mil-Homens et al. reported the application of a synthetic polycationic oligomer (L-OEI-h) as an alternative to treat *Klebsiella pneumoniae* infections [13]. The authors showed that L-OEI-h caused lysis of the cytoplasmic membrane in a panel of different species. This promising compound showed no visible cytotoxicity on the *Galleria mellonella* in vivo model; however, its antibiofilm capacity is yet to be tested.

The last approach published in this Special Issue was the use of electrospun titanium dioxide (TiO2) nanofibers and their activity against *Staphylococcus aureus* and *P. aeruginosa* [14]. Although, on planktonic cultures, TiO2 nanofibers were more active against *P. aeruginosa* than *S. aureus*, biofilms were prevented in both species in the same order of magnitude, suggesting their potential use for coating inanimate objects.

From a different perspective, Verran et al. provided a communication discussing the relevance of public engagement activities in complex phenomena such as biofilms and Antimicrobial resistance (AMR) [15]. The authors describe three different public engagement activities focused on biofilm control, namely hand hygiene, plaque control, and an externally applied antimicrobial coating using quantitative and/or qualitative methods.

This Special Issue collects high-quality original articles and reviews that demonstrate the relevance of biofilm infections as well as the potential of numerous antibiotic alternative strategies to prevent or control them. We hope that these articles will encourage researchers to investigate new antibiofilm strategies and implement them using standardized protocols that will help research to more effectively move towards clinical implementation.

### **References**


### *Review* **Biofilms as Promoters of Bacterial Antibiotic Resistance and Tolerance**

**Cristina Uruén 1,† , Gema Chopo-Escuin 1,†, Jan Tommassen <sup>2</sup> , Raúl C. Mainar-Jaime <sup>1</sup> and Jesús Arenas 1,\***


**Abstract:** Multidrug resistant bacteria are a global threat for human and animal health. However, they are only part of the problem of antibiotic failure. Another bacterial strategy that contributes to their capacity to withstand antimicrobials is the formation of biofilms. Biofilms are associations of microorganisms embedded a self-produced extracellular matrix. They create particular environments that confer bacterial tolerance and resistance to antibiotics by different mechanisms that depend upon factors such as biofilm composition, architecture, the stage of biofilm development, and growth conditions. The biofilm structure hinders the penetration of antibiotics and may prevent the accumulation of bactericidal concentrations throughout the entire biofilm. In addition, gradients of dispersion of nutrients and oxygen within the biofilm generate different metabolic states of individual cells and favor the development of antibiotic tolerance and bacterial persistence. Furthermore, antimicrobial resistance may develop within biofilms through a variety of mechanisms. The expression of efflux pumps may be induced in various parts of the biofilm and the mutation frequency is induced, while the presence of extracellular DNA and the close contact between cells favor horizontal gene transfer. A deep understanding of the mechanisms by which biofilms cause tolerance/resistance to antibiotics helps to develop novel strategies to fight these infections.


**Citation:** Uruén, C.; Chopo-Escuin, G.; Tommassen, J.; Mainar-Jaime, R.C.; Arenas, J. Biofilms as Promoters of Bacterial Antibiotic Resistance and Tolerance. *Antibiotics* **2021**, *10*, 3. https://dx.doi.org/10.3390/antibiotics 10010003

Received: 28 September 2020 Accepted: 19 December 2020 Published: 23 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/).

**Keywords:** biofilms; antibiotic resistance; antibiotic tolerance; multidrug-resistant bacteria; recalcitrance; biofilm control

### **1. Introduction**

During the last few decades, a significant increase in the number of clinical and environmental multidrug-resistant (MDR) bacteria, also called superbugs, has been reported. Particularly problematic in this respect are the major human pathogens, e.g., *Enterococcus faecium*, *Staphylococcus aureus*, *Klebsiella pneumoniae*, *Acinetobacter baumannii*, *Pseudomonas aeruginosa*, and *Enterobacter* spp. Additionally, among veterinary pathogens, particularly those associated with livestock farming and poultry production, the rate of drug resistance has increased. Some of them, such as *Salmonella enterica* and *Campylobacter* spp., are also important zoonotic pathogens.

About 700,000 human deaths are attributed to MDR bacteria each year globally, and this number is expected to exceed 10 million deaths by 2050, at a cumulative global cost of USD 100 trillion [1]. Additionally, the Global Antimicrobial Surveillance System from the World Health Organization (WHO) reported a widespread occurrence of MDR in 2018 among 2,164,568 people with suspected bacterial infections across 66 countries (ranging from high to low income). The proportion of people infected with MDR bacteria varies among countries, with higher levels of resistance to those drugs more widely utilized for treating infections (e.g., ciprofloxacin) [2], and many pathogens have developed mechanisms to survive to practically all of the antibiotic families available on the market, for

example *K. pneumoniae*. Despite the apparent exacerbation of this global issue, the high costs of the development of new antibiotics and the unavoidable emergence of antimicrobial resistances (AMR) have caused antibiotic development to lack economic appeal to the pharmaceutical industry.

MDR infections have been extensively associated with hospital and healthcare settings, for example those caused by *P. aeruginosa*, *A. baumannii*, or *K. pneumoniae*. However, MDR pathogens can also be transmitted within the community, for example *Neisseria gonorrhoeae.* In addition, people can be infected by MDR zoonotic bacteria that were selected through the use of antibiotics in food animals. In fact, antimicrobial resistance (AMR) in zoonotic bacteria recovered either from food animals or their carcasses is widespread. For instance, among *Salmonella enterica* recovered from pig carcasses in the EU in 2017, 53%, 59.5%, and 56.8% were resistant to ampicillin, sulfamethoxazole, and tetracycline, respectively. In the case of *Campylobacter* spp. recovered from poultry meat, high levels of resistance were noted for ciprofloxacin, nalidixic acid, and tetracycline in 54–83% of the isolates [3]. In fact, the relation between the use of some antibiotics in food animals and the subsequent detection of AMR in bacteria isolated from humans has been suggested in several studies [4,5]. Thus, the wide spreading of AMR, the high mortality rates, and the lack of initiatives to discover new antibiotics together make MDR bacteria a critical problem for modern medicine.

In addition to the well-known genetic mechanisms behind the AMR phenomenon and the transfer of antibiotic-resistance genes by horizontal gene transfer (HGT), bacteria are capable of displaying other strategies to withstand an exposure to antimicrobials, one of which is the ability to produce biofilms. Biofilms are highly structured associations of microorganisms embedded in a self-produced extracellular matrix (ECM) and adhered to a biotic or an abiotic surface. The biofilm confers many benefits to the members of the community, including collective recalcitrance. Recalcitrance is a term defined as "*the ability of pathogenic biofilms to survive in presence of high concentrations of antibiotics*" [6]. Indeed, biofilm cells are 10–1000 fold less susceptible to various antimicrobial agents than their planktonic forms [7–9]. Recalcitrance to antibiotics is achieved in biofilms through a variety of mechanisms, some of which can lead to an increase in the number of MDR bacteria, since processes such as HGT or hypermutability are favored within the biofilm environment. In fact, the biofilm is recognized as a reservoir of antibiotic-resistance genes [10].

Biofilms are involved in a broad range of infections. Indeed, about 80% of the chronic and recurrent microbial infections in humans are caused by biofilms, some of which result in high mortality and morbidity rates [11,12], particularly those caused by MDR bacteria. Patients with cystic fibrosis or with assisted ventilation are susceptible to chronic infections by biofilm-forming *P. aeruginosa* [13] and *A. baumannii* [14], respectively. These bacteria are also well-known for acquiring MDR, and the resulting biofilms are almost impossible to treat [15]. In addition, biofilms are highly resistant to the host immune defenses and clearance mechanisms. Other nocosomial infections include those caused by biofilms strongly adhered to implants and catheters or medical devices. They are generated by different pathogens, such as *Escherichia coli*, *Proteus mirabilis*, or *K. pneumoniae* [16,17], which can also become MDR. Additionally, *N. gonorrhoeae* forms biofilms on genital mucosa causing chronic infections [18]. The WHO classified this pathogen as high priority because of its extraordinary capacity to persist against all recommended antibiotics to treat the infection [19]. Together, these and many other studies have related the high capacity of several bacteria to survive to antibiotics through biofilms. Understanding the mechanisms that these pathogens utilize to survive antibiotics will help to design adequate surveillance methods and novel strategies to combat these infections. This review focuses on the role of biofilms in the poor susceptibility of bacteria to antibiotics. We will describe the mechanisms responsible for recalcitrance and the currently proposed solutions to treat or prevent biofilm infections.

### **2. Basis of Biofilm-Mediated Antibiotic Survival**

The recalcitrant nature of biofilms to antibiotics depends mostly on (i) the developmental stage of the biofilm, (ii) the ECM composition, and (iii) the biofilm architecture.

### *2.1. Biogenesis of Biofilms*

Biofilm formation is a dynamic process that takes place in a series of sequential steps. It is initiated by the interaction of the bacteria with a surface. Exposure of planktonic cells to stress, which may be provoked by antibiotics, starvation, or other adverse environmental conditions, can initiate biofilm formation by activating gene expression [20]. Additionally, molecules involved in cell-to-cell communication accumulate at high cell density. These molecules, generally referred to as autoinducers, can activate and regulate the process [21] and allow for a coordinated response of the population members, which is known as quorum sensing (QS). The first step of biofilm formation consists of the adhesion of the bacteria to the substratum. This process is often mediated by long, proteinaceous, filamentous fibers that protrude from the bacterial cell surface, such as flagella, fimbriae, or pili. After initial interaction is established, shorter cell surface-exposed structures interact with the substratum, thereby increasing the contact between bacteria and the substratum [12]. Strains of *E. coli* and *Salmonella* produce curli fimbrae that mediate both cell-to-substratum and cell-to-cell interactions [22]. Other proteins such as Bap-family proteins in *S. epidermidis* [23] or CdrA in *P. aeruginosa* [24] are large proteins that interact with ECM components and the bacterial cell surface thereby strengthening the matrix. Autotransporters are proteins secreted through the Type V secretion system in many Gram-negative bacteria and often have demonstrated roles in interbacterial interactions [25]. Then, the bacteria secrete ECM components and proliferate to form a microcolony. ECM serves as a glue element that helps to stabilize interbacterial interactions. Bacteria within the microcolony communicate and organize spatially. Type IV pili act at the junction between cells to form microcolonies and can also contribute to the reorganization of bacteria within the biofilm [26]. Cell-to-cell communication, including QS [21] and also cell-contact-dependent communication systems [27], seem to be relevant for this process. At this stage, the expression of genes for the formation of the ECM increases [12], and biofilms become less vulnerable to antibiotics than earlier biofilm stages [28].

### *2.2. Composition of the ECM*

The ECM consists of a conglomerate of different substances that together provide structural integrity to the biofilm. In general, the ECM can be composed of water, polysaccharides, proteins, lipids, surfactants, glycolipids, extracellular DNA (eDNA), extracellular RNA, membrane vesicles, and ions such as Ca2+. In many bacteria, extracellular polysaccharides and eDNA are prominent components of the ECM [29].

eDNA is constituted of chromosomal DNA that is released into the extracellular milieu through cell lysis, dedicated secretion systems, or membrane vesicles. eDNA is often involved in adhesion, particularly after the first interaction of the cell with the substratum. It mediates acid–base interactions and increases the hydrophobicity of bacterial cells which are favorable for the cell–substratum interaction [30,31]. Indeed, eDNA is used for initiation of biofilm formation in many pathogenic bacteria, including Gram-positive and Gramnegative bacteria and mycobacteria [32–34]. In addition, eDNA facilitates the interaction of the bacteria in the ECM. This is achieved by binding of positively charged segments of cell surface-exposed proteins with the negatively charged eDNA molecules [35]. Various proteins can be implicated in this interaction, such as autotransporters, lipoproteins or twopartner secretion protein A of Gram-negative bacteria, and cell wall-associated proteins in Gram-positive bacteria and fungi [35]. Thus, anchoring the eDNA to the cell surface by DNA-binding proteins is a widespread mechanism for biofilm formation that may also facilitate multispecies biofilms. eDNA can also mediate interactions with other ECM components such as polysaccharides [36]. Together, these interactions are relevant for the structural integrity of biofilms.

The composition of the polysaccharides present in the ECM varies between different bacterial species and even between different isolates of the same species. Most are long linear or branched molecules formed by one (homopolysaccharides) or several different (heteropolysaccharides) residues. They may contain substituents that greatly affect their biological properties. One of the most commonly studied polysaccharides is poly-β-1,6- N-acetyl-D-glucosamine, often named PGA or PNAG. It is synthetized by *E. coli* [37] and *S. aureus* [38], among others. In *E. coli*, PGA is required for initial cell-to-cell and cell-to-substratum attachment [37]. Another polysaccharide present in ECM is cellulose, a linear polymer of β-1,4 linked D-glucose. It is a major component of the ECM of some *E. coli* [39], *Salmonella* [40], and *Pseudomonas* strains [41]. Some *E. coli* strains produce a complex branched polysaccharide called colanic acid [42]. Additionally, *P. aeruginosa* can produce diverse exopolysaccharides. Mucoid *P. aeruginosa* strains produce alginate, a polymer of β-1-4-linked mannuronic acid and α-L-guluronate. Production of alginate confers a mucoid phenotype [43,44], typical of strains isolated from lungs of cystic fibrosis patients with *Pseudomonas* infections that underwent several rounds of antibiotic treatment. Therefore, secretion of alginate is related to pathogenic biofilms [45]. Alginate mediates the establishment of microcolonies at early stages of biofilm formation and provides stability to mature biofilms. Nonmucoid *P. aeruginosa* strains can produce other exopolysaccharides, e.g., Psl or Pel. Pel is a linear, cationic exopolysaccharide formed by 1→4 glycosidic linkages of *N*-acectylglucosamine and *N*-acetylgalactosamine. It has a critical role in maintaining cell-to-cell interactions and pellicle formation [46]. In contrast, Psl is composed of repeating pentasaccharide subunits of D-glucose, D-mannose, and L-rhamnose [47]. Psl mediates attachment to biotic surfaces such as mucin-coated epithelial surfaces and epithelial cells, indicating its relevance for the establishment of *P. aeruginosa* infection [48]. Additionally, *P. aeruginosa* strains can secrete cyclic and linear glucans [49,50] that are formed by β-1,3 linked glucose residues.

The proteinaceous content of the ECM includes proteins that are secreted through active secretion systems or released during cell lysis. The role of many of these proteins in the biofilm matrix is unknown, but some of them have been identified as important contributors to biofilm formation or restructuring in many pathogens. Various are extracellular enzymes. Their substrates can be polysaccharides, proteins, and nucleic acids, present in the ECM. They can function in remodeling of the ECM, detachment of cells from the biofilm, or degradation of polymers for nutrient acquisition.

ECM biogenesis and composition are dynamic and vary between strains of a given species and also depend on environmental conditions, such as nutrient availability and the presence of stressors, and on social crosstalk. Several functions have been attributed to the ECM based on its extraordinary capacity to establish intermolecular interactions between its components, and with surface-exposed structures of the cells, biotic and abiotic substrata, and many environmental molecules [29]. Thereby, the ECM immobilizes cells and keeps them in the biofilm community. By retaining the cells in close proximity, the ECM establishes the optimal conditions for interbacterial communication and exchange of genetic material, which is relevant, amongst others, for the dispersion of antibiotic-resistance genes. The ECM additionally retains water and thereby protects the cells against desiccation. Furthermore, the extracellular enzymes in that hydrated environment generate an external digestive system. In addition, ECM retains several other substances, for instance, nutrients, energy sources, antibiotics, antibiotic-degrading enzymes, and molecules released by cell lysis, thereby constituting a recycling unit [29]. In general, the ECM acts as a protective scaffold.

### *2.3. Biofilm Architecture*

The architecture of biofilms is defined by the organization of the biomass and the spaces in between. The development of this structure depends on the composition of cell-surface structures mediating mutual interactions between cells and interactions of cells with ECM components and with the substratum [35]. The biofilm architecture is responsible for the generation of gradients of dispersion of substances within the biofilm. This will influence the accessibility of these substances to particular niches inside the biofilm, and determines, amongst others, the variation in antibiotic susceptibility of cells within biofilms. Figure 1 illustrates the biofilm architecture of different bacterial species. *P. aeruginosa* strain ATCC 15,692 forms complex biofilms with mushroom-like architectural features consisting of well-defined stalks and caps. *Enterococcus faecalis* ATCC 51,299 biofilms, however, are flat and compact [51], while *Salmonella enterica* strain S12 and *E. coli* strain ESC.1.16 form biofilms constituted of small cell clusters (Figure 1A) [51]. In contrast, biofilms of *Neisseria meningitidis* strain HB-1 are constituted of cell aggregates of different sizes forming defined channel-like structures [33] (Figure 1B).

**Figure 1.** Variable architecture of biofilms. (**A**) Biofilms of five species (*Salmonella enterica*, *Escherichia coli*, *Pseudomonas aeruginosa Staphylococcus aureus*, *Enterococcus faecalis)* were formed under static conditions on abiotic surfaces during 24 h and were stained with Syto9, a green fluorescent nucleic acid marker. Reprinted from [51] with permission from Elsevier. (**B**) Strains of *Neisseria meningitidis* HB-1 and α153 and derivatives, which do or do not produce the autotransporters AutA and AutB (as indicated), formed biofilms under flow conditions during 14 h and were stained with the LIVE/DEAD Backlight bacterial viability stain (where red cells are dead and green cells are live). Reproduced from [52,53].

> Thus, in general, based on their architecture, biofilms can be classified into (i) monolayer biofilms, formed by a compact layer with high surface coverage, or (ii) multilayer biofilms, formed by bacterial clusters of different morphology with a low surface interaction. The biofilm architecture can vary depending on different factors, for instance the expression of surface-exposed proteins. Examples are the meningococcal autotransporters AutA and AutB, whose expression is phase variable and significantly alters the biofilm (Figure 1B) [52,53]. Additionally, the medium composition influences the biofilm architecture. *P. aeruginosa* PAO1 makes monolayer biofilms in the presence of citrate benzoate and casamino acids and multilayer biofilms in presence of glucose [54].

### **3. Mechanisms of Biofilm Recalcitrance**

### *3.1. Types of Antibiotic Recalcitrance*

Biofilm recalcitrance comprises two independent phenomena: antibiotic resistance and antibiotic tolerance. Resistance refers to the capacity of a microorganism to survive and grow at increased antibiotic concentrations for long periods of time and is quantifiable by assessing the minimum inhibitory concentration (MIC) [55]. It involves mechanisms that prevent the binding of an antibiotic to its target, including enzymatic deactivation, active efflux of a drug once it is in the cytoplasm or the cytoplasmic membrane, or reduced influx, among others, and can be generated by HGT or mutations. Together, they preclude antibiotics from altering their target's function and they prevent the production of toxic products that would end up damaging the cell. Resistance can be further classified into intrinsic, acquired, and/or adaptive resistance (expanded in Box 1).

#### **Box 1.** Types of antibiotic resistance.

**Intrinsic resistance.** This is the inherent/natural property of the bacteria to withstand antibiotics. For example, Gram-negative bacteria are, in general, more resistant to antibiotics than Gram-positive ones due to the presence of the outer membrane, which reduces the permeability to many antibiotics [56]. Another example is the wall-less bacterial genus Mycoplasma, which is not affected by antibiotics whose target is the cell wall.

**Acquired resistance.** This arises through genetic modifications of originally sensitive bacteria, either through mutations or by the incorporation of new genes via HGT. Thus, microorganisms that are initially sensitive to an antibiotic become resistant due to spontaneous or induced mutations that alter, for example, the target of the antibiotic or its uptake by the cell or after the acquisition of one or more molecular mechanisms for AMR, such as antibiotic inactivation or, increased antibiotic efflux [57]. These genetic modifications are heritable and will result in a permanent effect if the fitness-cost associated with them is low or null or compensation mechanisms exists [58].

**Adaptive resistance.** This is the capacity of bacteria to vary rapidly gene expression or protein production in response to antibiotics or adverse environmental conditions. The molecular mechanism behind involves epigenetic inheritance, population heterogeneity, gene amplification, and efflux pumps that are regulated by intricate regulatory pathways [59].

**Heteroresistance.** This is the presence, within a given population of bacteria, of one or several subpopulations displaying increased levels of antibiotic resistance compared with the main population [60]. This phenomenon is often related with the presence of unstable genes which would give the bacteria a high likelihood for reversion to susceptibility in the absence of antibiotic selective pressure [61]. This instability makes its detection difficult, increasing thus the risk for treatment failure [62].

> By contrast, antibiotic tolerance is the capacity of bacteria to survive a transient exposure to increased antibiotic concentrations, even those above the MIC. Tolerance is assessed by the minimum bactericidal concentration, that is, the minimum concentration of antibiotic required to kill 99.9% of the cells [63]. Unlike resistance, tolerance is only temporary and after longer exposure periods, the antibiotic will kill the bacteria. It is an adaptive phenomenon that implies a change in cellular behavior, from an active (growing) state to a quiescent (dormant) state [57], and requires large metabolic rearrangements affecting, for example, energy production and nonessential functions. These changes are triggered during poor growth conditions or exposure to stress factors or antibiotics [55,64]. In this case, antibiotics can usually attach to the target molecules, but because their function is no longer essential, the microorganism survives [65]. Tolerance in biofilms is also caused by entrapment of the antibiotics in the ECM, in this case, the antibiotic does not reach its target. In contrast to resistant cells, tolerant cells within the biofilm cannot grow in presence of a bactericidal antibiotic. Persistence is an especial phenomenon of tolerance [64]. Indeed, persistence is a phenomenom that increase the survival of a given population in the presence of bactericidal antibiotics without enhancing the MIC, but in contrast to tolerance, persistence only affects a subset of cells of the population called persisters [64]. Persisters cells are tolerant cells that eventually can be killed at longer exposure times. There are two types of persisters, e.g., type I or triggered persistence, which is induced upon environmental signals, such as starvation, and type II or spontaneous persistence, where a subpopulation of growing bacteria converts into the persister state by a stochastic process [66]. Any how, persistence can be also refereed to as heterotolerance, which is different than heteroresistance (Box 1), as persisters can eventually be killed at longer exposure times. Figure 2 illustrates the mechanisms that govern antibiotic tolerance and antibiotic resistance of biofilms, and they will be further discussed in the next section.

**Figure 2.** Mechanisms of biofilms recalcitrance. Biofilm recalcitrance comprises a combination of antibiotic (ATB) resistance mechanisms (right panel) and ATB tolerance mechanisms (left panel). Resistance mechanisms confer the ability to survive and grow at increased ATB concentrations for long periods and involve horizontal genetic transfer (HGT), hypermutation, and quorum sensing (QS), leading to transport of antibiotics via efflux pumps, reduced permeability of the outer membrane, or production of enzymes that inactivate ATB. The type of AMR (Box 1) is indicated in green. In contrast, tolerance mechanisms lead microorganisms to survive at increased ATB concentrations temporally, and involve activation of stress responses (SOS response, stringent response SR)) and hypoxia, leading to activation of a quiescent state, anaerobic metabolism, decrease of membrane potential, and moderate increase in efflux pump expression. Arrowhead lines indicate the interrelation between mechanisms. CM: cytoplasmic membrane.

Overall, biofilm recalcitrance does not depend on one unique mechanism but is a combination of both antibiotic tolerance and antibiotic resistance mechanisms. Such combination varies depending upon aspects such as the bacterial species or strain, the antimicrobial agent, the developmental stage of the biofilm, and the biofilm growth conditions [63].

### *3.2. The Protective ECM Barrier*

The components of the ECM can impact the efficacy of antibiotics on biofilm-forming cells. ECM influences the biofilm architecture, which in turn generates gradients of dispersion that affect the access of antibiotics to the biofilm members. The flow of substances through biofilm varies. In the channels, the flow is convective while it occurs by diffusion within cell clusters [67]. Thus, antibiotics can penetrate rapidly through the channels of the biofilm but may be retained locally in cell aggregates. Due to the differences in biofilm

architecture (Figure 1), the time for an antibiotic to reach the interior of the biofilm varies between strains [68].

The diffusion of antibiotics through the biofilm can also be limited by their interaction with particular ECM components, which affects antibiotic effectivity. This is illustrated in the literature by several examples including *P. aeruginosa.* Alginate is a polyanionic exopolysaccharide that protects *Pseudomonas* biofilms from aminoglycosides [44]. Additionally, the highly anionic cyclic glucans present in these biofilms interact with the aminoglycoside kanamycin [69]. Presumably, the high negative charge of alginate and the cyclic glucans helps to establish ionic interactions with these positively charged antibiotics. However, in strains that do not secrete alginate, the polysaccharides Pel and Psl are involved in the establishment of biofilms. Pel provides protection against the aminoglycosides tobramycin and gentamicin, but not against ciprofloxacin [70]. Unlike alginate, Pel is a cationic exopolysaccharide, and, therefore, resistance to aminoglycosides cannot be explained by direct charge interaction. However, Pel binds eDNA [46], which is negatively charged and can interact with aminoglycosides. Additionally, Pel could bind negatively charged portions of other antibiotics such as ampicillin, though this hypothesis has not been explored yet. The exopolysaccharide Psl also binds eDNA [71] and plays a role in resistance to colistin, polymyxin B, tobramycin, and ciprofloxacin at early stages of biofilm formation [72]. This protective effect was also observed in the non-Psl producing species *E. coli* and *S. aureus* if they were present in a mixed biofilm together with *P. aeruginosa* [72].

Studies in *Pseudomonas* evidence that eDNA enhances resistance of biofilms to aminoglycosides but not to fluoroquinolones or β-lactams [73,74]. eDNA also enhances resistance of biofilms of *Staphylococcus epidermidis* to the glycopeptide antibiotic vancomycin [75]. Likely, the negatively charged eDNA binds positively charged aminoglycosides and glycopeptides. The latter study [75] also demonstrated that the binding constant of vancomycin and eDNA is up to 100-fold higher than that of vancomycin and its target, the D-Ala-D-Ala peptide in peptidoglycan precursors. Thus, within the biofilm environment, eDNA may compete with D-Ala-D-Ala peptide, and, because of the higher affinity of vancomycin for eDNA, it may be retained in the ECM. Besides direct interaction of eDNA with antibiotics, accumulation of eDNA creates a cation-limited environment by chelating cations such as Mg2+. In *P. aeruginosa* and *S. enterica* serovar Typhimurium (*S.* Typhimurium), reduction of the Mg2+ concentration triggers the two-component regulatory systems PhoPQ and PmrAB, which are linked to AMR [74,76]. Activation of these systems generates modifications in the lipid A moiety of lipopolysaccharides (LPS) through (i) the expression of the outer membrane (OM) protein PagP, which adds a palmitoyl residue to the lipid A, (ii) the substitution of the phosphate groups with 4-amino-4-deoxy-L-arabinose and/or phophoethanolamine, and (iii) the production of LpxO, which adds a hydroxyl group onto the second carbon atom of one of the fatty acyl chains. The first stage of aminoglycoside uptake involves the binding of the polycationic antibiotics to the negatively charged components of the bacterial membrane, such as LPS of Gram-negative organisms. This is followed by displacement of Mg2+ ions [77,78], which leads to disruption of the OM and initiation of aminoglycoside uptake [79,80]. The modifications in the lipid A generated by the activation of PhoPQ and PmrAB alter considerably the lipid A charge and the OM permeability, which could explain the involvement of eDNA in aminoglycoside resistance. On the other hand, the activity of antimicrobials can also promote the release of eDNA to the ECM. For instance, the amount of eDNA in biofilms of S. epidermidis doubled by treatment with vancomycin [75]. The released eDNA can then bind the positively charged antibiotics and prevent them from reaching the cells and exerting their activity. Considering that eDNA is a constituent of the ECM of many bacteria, it is tempting to speculate that this phenomenon could also occur in other bacteria. Additionally, antibioticmodifying enzymes can be released and located into the biofilm matrix. In mixed-species biofilms, the presence of a single species that secretes such enzymes would be beneficial for antibiotic-sensitive species in the same biofilm. Examples include *Moraxella catarrhalis*

which secretes β-lactamases and, thereby, protects *S. pneumoniae* [81] and *H. influenzae* [82] from amoxicillin and ampicillin treatments, respectively, in mixed biofilms.

To summarize, components of the ECM alter the biomass organization affecting diffusion of certain antibiotics within the biofilm and, thus, altering the level of exposition of cells located in specific biofilm niches particularly those within dense biomass. Additionally, certain ECM components can interact with antibiotics, preventing them from reaching their targets within cells. Furthermore, the ECM can retain antibiotic-modifying enzymes, which is particularly relevant in mixed biofilms, where susceptible bacteria can be protected by such enzymes released from resistant cohabitants.

### *3.3. Physiological Heterogeneity*

The architecture and organization of the biofilm also generates gradients of dispersion of nutrients, oxygen, pH, signaling molecules, and waste products. Oxygen and nutrient depletion occur in particular niches, such as inside the microcolonies and in the lower cell layers, and these conditions can induce a variety of physiological states involving different metabolism (aerobic, microaerobic, and fermentative) and growth rates (fast and slow growth, dormant cells, and persister cells) [83,84]. Nongrowing and slowly growing bacteria are less vulnerable to antibiotics as a consequence of the inactivity of antibiotic targets, a phenomenon called "drug indifference". In contrast, persister cells are phenotypic variants that constitute a part of the population with tolerance to antibiotics (see Section 3.1) that can resume growth after antibiotic removal. Persisters are present in both biofilm and planktonic cultures; however, biofilms typically harbor more persisters than planktonic cultures [66]. Ultimately, the biofilm is constituted of cells with different physiological states and chances to survive an external drug insult. Indeed, this repertoire of physiological cell states is relevant for tolerance to multiple antibiotics. For example, slowly growing cells are tolerant to antibiotics such as tobramycin and ciprofloxacin [85], which target protein synthesis machinery and DNA gyrase, respectively, and thus, exert their activity on fast-growing cells. In contrast, slowly growing cells are susceptible to antibiotics such as colistin [86] that act on the membrane.

Bacteria respond to starvation and stress conditions through specific adaptative mechanisms, such as activation of the stringent response (SR) and the SOS response. The SR is induced by amino-acid, carbon, and iron starvation [87]. Under amino-acid deprivation, the ribosomes are stalled by the presence of uncharged tRNA in the A site (Figure 3A). (p)ppGpp synthetases, e.g., RelA and SpoT in β- and γ-proteobacteria, sense stalled ribosomes and synthetize (p)ppGpp, also known as *alarmone*, which initiates transcriptional reprogramming and regulates various metabolic pathways, such as phosphate, amino-acid and lipid metabolism, among others [87] (Figure 3A). (p)ppGpp also modulates the repressor CodY, a master regulator of many genes triggered during environmental stress. Ultimately, the SR shuts down almost all metabolic processes, and, thus, cells become tolerant to antibiotics that target such processes. For example, the SR in *E. coli* has been linked to tolerance to inhibitors of cell-wall biosynthesis, such as penicillins [88], cephalosporins and carbapenems [89], and of cell division, such as norfloxacin [90] and ofloxacin [91]. The SR has been related to reduced susceptibility to ofloxacin [92,93], meropenem, colistin [92], and gentamicin [92,94,95] in *P. aeruginosa* biofilms, and to tolerance to ampicillin and vancomycin in *S. aureus* [96].

**Figure 3.** The stringent response, toxin-antitoxin, and SOS response pathways. (**A**) The stringent response (SR) is triggered by several stress conditions (amino-acid, carbon and iron deprivation or membrane damage) that activate the production (p)ppGpp by the synthetases RelA and SpoT and homologues. (p)ppGpp reprograms cell metabolism through the interaction with proteins involved in translation, transcription, replication, amino-acid metabolism, and nucleotide metabolism. (**B**) The SOS response is triggered by damaged DNA. Single-stranded (ssDNA) is detected by RecA. In the presence of (d)ATP, RecA is activated causing self-cleavage of LexA. LexA is a dimer that represses the transcription of SOS genes, which harbor an SOS box in their promoter. Cleavage of LexA leads to activation of the SOS genes inducing a repertoire of activities as indicated. The *lexA* gene also contains an SOS box; therefore, LexA production is self-regulated. Once the DNA is repaired, LexA represses the SOS response. (**C**) Several toxin-antitoxin (TA) modules are dispersed on the chromosome and are further classified according to the nature of the antitoxin and its mechanism of action. They are constituted by two genes, one encoding a toxin, with specific activities against target molecules (DNA, RNA, membrane, cell wall synthesis, ATP), and the other an antitoxin that binds to the toxin and inhibits its toxic activity. Under normal conditions, toxin and antitoxin are equally produced and thus the toxin does not exert its function. However, under stress conditions, the antitoxin can be degraded, and the toxin is free for toxic reactions. SR and SOS responses increase the production of ClpXP and Lon proteases, which activates the toxin by degradation of the antitoxin TA modules production. The toxic activity has several repercussions on cell biology.

The SOS response also contributes to antibiotic tolerance. It is generated by stress conditions such as DNA damage. Single-stranded DNA, generated by disruption of the DNA, activates RecA, which, in the presence of (d)ATP, stimulates self-cleavage of the repressor LexA leading to de-repression of SOS genes [97] (Figure 3B). Under regular physiological conditions, LexA is bound to a specific DNA sequence (SOS box) located upstream of several genes participating in DNA repair, mutagenesis and cell growth, and represses SOS gene expression (Figure 3B). The SOS response plays an important role in tolerance to antibiotics that cause DNA damage such as fluoroquinolones [98]. Studies in *E. coli* have demonstrated the function of the SOS response in enhancing biofilm tolerance to fluoroquinolones [91]. Together with the SR, the SOS response can also activate the expression of toxin-antitoxin (TA) modules [91,99], although TA modules can also be activated by stress-induced proteases like ClpXP and Lon in response to antibiotics. TA modules are genetic elements composed of two genes. One gene encodes a stable toxin protein that inhibits bacterial growth by interfering with essential cellular processes such as DNA replication, translation, cell wall synthesis, and membrane homeostasis, among others [100] (Figure 3C). The other gene encodes a cognate antitoxin that typically prevents or impairs toxin function. Antitoxins can be labile molecules that are degraded under stress conditions, a circumstance that allows the toxins to exert their harmful effects in cellular functions. Thus, by inactivating antibiotic targets, TA confers antibiotic tolerance [101] and increases persister formation [102] (Figure 3C). Several studies have shown upregulation of TA modules in persister cells. A well-studied example is MqsA and MqsR, a classical TA module where the MqsR functions as the toxin and MqsA as the antitoxin [103,104]. MqsR production is stimulated during biofilm formation and enhances cell motility in *E. coli* [103], and MqsA has been linked to the regulation of the general stress responses, such as oxidative stress [104]. MqsA represses the stress regulator RpoS, which decreases the concentration of the messenger 3,5-cyclic diguanylic acid and, consequently, biofilm formation is inhibited. However, upon stress conditions, such as oxidative stress, MqsA is unstable and rapidly degraded by Lon and ClpXP proteases, causing the accumulation of RpoS [104]. Then, SR is activated, and bacteria initiate biofilm formation. However, this has recently been disputed. Frainkin et al. (2019) reported that MqsA is not a global regulator and does not regulate *rpoS* expression. Moreover, authors showed that *mpsRA* production is not regulated by stress and that mutation of mpsRA has no clear effect on biofilm formation [105].

TAs are often involved in the stabilization of plasmids [106] and genomic islands that carry integrative and conjugative elements, which can mediate resistance to multiple antibiotics [107]. Considering that these genomic elements are commonly involved in promoting HGT [108], the role of TAs in antibiotic resistance can be notable. The SOS response and the SR participate in the dissemination of antibiotic resistance through integrons. Integrons are genetic elements involved in the capture antibiotic resistance genes. As they are located on mobile genetic elements, such as transposons, they contribute to the dissemination of these genes among Gram-negative bacteria [109]. An integron is composed of a gene encoding an integrase, a specific recombination site, and a promoter that controls the expression of promoter-less genes embedded within gene cassettes [110], some of which can be located in genetic mobile elements and contain antibiotic resistancegene cassettes. Within the biofilm environment, transposases are activated under SOS response and SR regulates integrase expression [111], thereby promoting the dissemination of the antibiotic resistance genes located in mobile elements within members of the biofilm.

Depletion of oxygen in the interior of biofilms [112,113] and the presence of oxygen gradients have been demonstrated in different biofilm models [114,115]. Hypoxia affects metabolism substantially, because it shifts aerobic respiration to alternative metabolic routes such as denitrification and fermentation. Consistent with the hypoxic conditions in biofilms is the detected production of nitrous oxide, an intermediate of denitrification, in sputum of cystic fibrosis patients with chronic *P. aeruginosa* infection [116]. Earlier work using 70 different Gram-negative and 23 Gram-positive bacteria growing in aerobic and anaerobic conditions showed that microorganisms were more tolerant to aminoglycosides and highly tolerant to tobramycin under anaerobic conditions [117]. Accordingly, *Pseudomonas* biofilms formed under anaerobic conditions were more tolerant to antibiotics such as tobramycin, ciprofloxacin, carbenicillin, ceftazidime, chloramphenicol, and tetracycline than those formed in aerobiosis [112]. Indeed, hypoxia reduces membrane potential conferring tolerance to antibiotics such as aminoglycosides that require an intact membrane potential to be transported into the cytoplasm [118].

Antibiotics can induce oxidative stress by increasing cellular hydroxyl radical levels and enhancing production of lethal reactive oxygen species (ROS) [119,120]. This could be caused by an increased oxidation rate of tricarboxylic-acid cycle-derived NADH that perturbs iron homeostasis. Ferrous iron from iron-sulfur clusters is oxidized to ferric iron in the Fenton reaction which yields extremely reactive and harmful hydroxyl radicals that oxidize vital macromolecules such as DNA, proteins, and lipids [120,121]. To counteract, bacteria stimulate production of catalases and superoxide dismutases. Indeed, *Pseudomonas* mutants lacking catalase generate biofilms more susceptible to ciprofloxacin than wild-type biofilms [122]. The SR increases catalase and superoxide dismutase levels and represses the production of 4-hydroxy-2-alkylquinolines, which are intercellular signaling molecules with prooxidant properties [92]. Additionally, persisters downregulate genes encoding proteins involved in the generation of ROS, including a ferredoxin reductase, which is involved in recycling Fe3+ to Fe2+ and thus drives the Fenton reaction, and upregulates genes involved in ROS detoxification. Thus, persister cells are, to some extent, protected against the detrimental effects of ROS produced upon antibiotic treatment. Considering that the SR and persister formation are more highly activated in biofilm cells than in planktonic cultures, biofilm cells can better deal with ROS induced by antibiotics.

### *3.4. Traffic of Substances across the Cell Envelope*

Several proteins in the bacterial membranes function in the recognition and transport of substances, including antibiotics, into or out of the cell. This activity is facilitated by efflux pumps and porins (in Gram-negative bacteria) that mediate an active and passive transport, respectively. Efflux pumps can be divided into six families, e.g., the multidrugand toxin-extrusion (MATE), small multidrug-resistance (SMDR), major facilitator (MF), ATP-binding cassette (ABC), resistance-nodulation-division (RND) and proteobacterial antimicrobial compound-efflux (PACE) families. These families display large differences concerning transporter structure, function, and substrate specificity and energy source [123]. All families use protein motif force for as driving force except for the ABC transporters, which use ATP hydrolysis, and some members of MATE family that use sodium gradient instead.

The MATE family comprises proteins of 400–700 amino-acid residues organized in 12 α-helices [124]. These pumps participate in the extrusion of diverse antibiotics, such as ciprofloxacin, chloramphenicol, streptomycin, kanamycin, norfloxacin, and ampicillin [125]. Representative examples of this family in MDR bacteria are YdhE in *E. coli*, which transports kanamycin and acriflavin, amongst others [126], PmpM in *P. aeruginosa* [127], and AbeM in *A. baumannii* [128]. The SMR family is constituted of small proteins composed 100– 120 amino-acid residues organized as homodimers with four transmembrane helices in each subunit. EmrE [129] of *E. coli* and Smr/QacC in *S. aureus* [130], both transporters of acriflavine, belong to this family. The MF family is constituted of membrane proteins with 400–600 amino-acid residues organized in 12–14 transmembrane α-helices. This family of transporters facilitates the passage of ions and carbohydrates across membranes, as well as antimicrobial agents such as tetracyclines and fluoroquinolones [131,132]. NorA, LmrS, and MdeA of *S. aureus* are well-known members of this family [133]. The ABC transporters are active transporters, often constituted of a transmembrane channel, formed by one or two proteins, and a dimeric cytoplasmic ATPase. The MacB transporter of *E. coli*, which operates in concert with the outer-membrane protein TolC and transports azithromycin, clarithromycin, and erythromycin, belongs to the ABC family [134]. Members of the PACE

family transport acriflavine, proflavine, benzalkonium, acriflavine, and chlorhexidine [135]. Transporters of the RND family form a protein complex constituted of about 1000 aminoacid residues organized in a 12-helical structure in the membrane, but, in contrast to MF transporters, RND proteins possess large periplasmic domains. In Gram-negative bacteria, members of MF (e.g., EmrB), RND (e.g., MdtK), and ABC (e.g., MacB) families can be organized in a tripartite protein complex formed of an inner membrane protein, a membrane fusion protein, and an OM protein [123]. Collectively, this complex spans the entire cell envelope and allows for efficient excretion of antibiotics into the external medium. The AcrAB-TolC complex of *E. coli* [136], MexAB-OprM of *P. aeruginosa* [137], and AdeABC of *A. baumannii* [138], all RND family members that participate in tripartite transporters, have been implicated in bacterial biofilm resistance and biofilm formation.

Several lines of evidence relate efflux-pump production to biofilm formation and, directly or indirectly, to AMR/tolerance. First, some efflux-pump-encoding genes are upregulated in biofilms as compared to planktonic cells. This was detected in different transporter families. Examples are the *mdpF* gene of *E. coli*, which encodes a component of the MdtEF efflux pump (RND family) that participates in the tolerance to nitrosylmediated toxicity and that was upregulated in anaerobic conditions [139], a condition found in biofilms. Another example is the multidrug efflux genes *acrA* and *acrB* of *S.* Typhimurium [140]. Second, the exposition of cells to efflux-pump inhibitors reduces biofilm formation in several pathogens such as *E. coli* and *K. pneumoniae* [141], *P. aeruginosa* [142], and *S.* Typhimurium [143]. Third, mutants lacking known efflux systems exhibit a marked reduction in biofilm formation, for instance pump mutants in *Salmonella* showed reduced production of curli [143], which are implicated in biofilm formation. Fourth, several studies link efflux-pump production and acquired resistance of biofilms to antibiotics. Within *Pseudomonas* biofilms, MexAB-OprM and MexCD-OprJ are essential for resistance to azithromycin [144], and MexAB-OprM also mediates resistance to colistin [145]. Additionally, the ABC transporter encoded by the PA1874–1877 operon conferred protection to tobramycin in biofilms [146]. Biofilms of *E. coli* formed by mutants in genes that participate in the AcrAB-TolC system exhibited sensitivity to tobramycin, tetracycline, and the antiseptic benzalkonium, while mutants in the EmrAB system exhibited sensitivity to tobramycin [147]. Overall, these studies evidence that efflux pumps can directly contribute to expel antibiotics during biofilms that contribute to AMR but also ECM components that ultimately contribute to biofilm tolerance.

The OM of Gram-negative bacteria forms a barrier for both hydrophobic and hydrophilic solutes. Porins control the access of antibiotics from the environment to the periplasm. Porins are trimers of 16-stranded β-barrels located in the OM that provide selective access of small hydrophilic molecules to periplasm by diffusion through a water-filled channel present in each of the subunits [148]. Not surprisingly, MDR clinical isolates of *Enterobacteriaceae* often exhibit loss of porin production [149,150]. Several genetic mechanisms reduce or prevent porin synthesis, including downregulation of expression, premature stop codons or insertion elements. Besides, missense mutations can alter the permeability properties. Efflux pumps and porin production act in synergy and have been associated to biofilm production, particularly in *Enterobacteriaceae* [141,151]. In *K. pneumoniae*, the gene coding for the porin OmpK36 was downregulated and the *acrB* gene, coding for a component of a major multidrug-efflux pump, was upregulated in biofilms as compared to planktonic cells [152].

The production of efflux pumps and porins can be up- and downregulated, respectively, to reduce the intracellular accumulation of antibiotics, an adaptive phenomenon based on the transport regulation. Curiously, upregulation of the MDR transporter MdfA of *E. coli* not only makes cells resistant to the aminoglycosides neomycin and kanamycin but also increases their susceptibility to spectinomycin by an, as yet, unexplained mechanism [153,154]. In general, efflux pump production could be regulated by local and global transcriptional regulators, modulators, various substances, including antibiotics, small RNAs and two-component regulatory systems whose activation depends on environmental

stimuli [155]. MarA, BrlR, SoxS, Rob, and AcrR are very well-known regulators of pump production in pathogenic bacteria [155]. Pump production can be regulated at multiple levels. AcrAB-TolC of *E. coli* is the best-studied example of regulation under a complex regulatory network. AcrAB synthesis is negatively regulated by the local repressor AcrR that represses *acrAB* expression [156]. In addition, the repressor AcrS regulates *acrAB* negatively [157], while the histone-like nucleoid structuring protein H-NS has a role in the network repressing the expression *acrS* [158]. Therefore, by negatively regulating *acrS*, H-NS is a positive regulator of *acrAB*. Furthermore, the two-component regulatory systems EvgAS and/or PhoQP activate *tolC* expression, while EvgAS also activates *acrAB* expression [159]. Finally, the global regulators SdiA [160], MarA, SoxS, and Rob activate expression of *acrAB*, and the latter three regulators also activate *tolC* and *micF* expression [155]. *micF* transcripts inhibit the translation of the mRNA of porin OmpF. As OmpF plays an important role in the influx of antibiotics, the bacterium thus controls the efflux and influx of antibiotics by activating AcrAB-TolC and abolishing OmpF production, respectively. Interestingly, while MarA upregulates pump production, it downregulates biofilm formation through activation of the *ycgZ-ymgABC* operon which eventually reduces curli formation [161]. Possibly, MarA helps to activate a mechanism for cells to escape from biofilms as defence to the antibiotic insult. However, since pump expression is regulated by multiple mechanisms, it is difficult to speculate about the precise biological role of MarA within the complex regulatory network. In *P. aeruginosa*, the production of the efflux pumps MexAB-OprM and EmrAB is differentially regulated by the regulators MdrR1 and MdrR2 [162]. These regulators activate EmrAB but repress MexAB-OprM. Their expression varies between cells located in different layers of a biofilm, which leads to different susceptibility to antibiotics dependent on the position within the biofilm biomass. As these efflux pumps have different substrate specificities, it has been proposed that MdrR1 and MdrR2 could act as master modulators controlling the activities of various pumps in different microenvironments within the stratified biofilm structure [162]. All together, these studies illustrate that the regulation of pump production can be controlled by a complex repertoire of regulators, some of which act on a variety of genes that participate in biofilm production, and/or by varying conditions, which are generated within the biofilm environment.

### *3.5. Interbacterial Communication*

QS is a population-density-dependent regulatory mechanism by which bacteria communicate via signaling molecules, called autoinducers. Bacteria produce autoinducers, which accumulate in the environment with the increase in the cell density. These autoinducers are recognized by cell-surface receptors or in the cytoplasm. After receptor recognition, gene transcription is activated, involving genes coding for surface proteins, transcription factors, virulence factors, and proteins involved in biofilm development [163,164]. Peptides are used as autoinducers in Gram-positive bacteria in contrast to the acylated homoserine lactones used in Gram-negative bacteria. Autoinducer 2 is used in Gram-negative and Gram-positive bacteria for intra- and interspecies communication.

QS seems to contribute to biofilm recalcitrance. Biofilms formed by QS mutants or wild-type bacteria treated with QS inhibitors are more susceptible to antibiotics. For example, *P. aeruginosa* biofilms formed by a mutant strain lacking *lasR* and *rhlR*, which is deficient in QS, were significantly more susceptible to tobramycin than wild-type biofilms [165]. Additionally, mixed *Pseudomonas* biofilms formed by QS mutants and wild-type bacteria exhibited a decreased resistance to tobramycin as compared to those formed only with wild-type bacteria [166]. In *S. aureus*, a QS-deficient *agrD* mutant exhibited a biofilmspecific decrease in resistance to rifampin compared to wild type [167]. Additionally, *fsrA* and *gelE* mutants of *E. faecalis*, which are deficient in QS and a QS-controlled protease, respectively, were impaired in biofilm formation in the presence of gentamicin, daptomycin, or linezolid, but not in the absence of these antibiotics [168]. The involvement of QS in biofilm recalcitrance may have multiple origins. QS is involved in biofilm formation and structuration; therefore, QS-defective mutants produce less structured biofilms. Considering that the architecture of the biofilm is relevant for its recalcitrant properties, the resulting biofilms would be more susceptible to antibiotics. Alternatively, QS may have other contributions. For instance, QS in *P. aeruginosa* regulates the production of 2-n-heptyl-4-hydroxyquinoline-N-oxide (HQNO), which inhibits the respiratory chain by binding to the cytochrome *bc*<sup>1</sup> complex [169]. This results in the accumulation of ROS and the reduction in membrane potential and eventually in autolysis. Autolysis releases DNA, which, as previously discussed, promotes biofilm formation and confers resistance against positively charged antibiotics. Additionally, by reducing the electrochemical gradient, the sensitivity to aminoglycosides [170], tetracycline, and macrolides [171] is reduced. QS can also contribute to drug resistance within mixed species biofilms. *Stenotrophomonas maltophilia* and *P. aeruginosa* can form mixed biofilms when they coinfect cystic fibrosis patients. *P. aeruginosa* recognizes signal factors produced by *S. maltophilia* and induces the PmrAB two-component system that regulates resistance to cationic antimicrobial peptides [172]. Thus, QS signals and the resulting downstream consequences can elicit an ample range of physiological changes that alter the antimicrobial susceptibility of cells within a biofilm.

Some intercellular communication requires direct cell-to-cell contact. Well-described examples are the two-partner secretion system (TPS). In the TPS system, a large surfaceexposed protein, generically called TpsA, is secreted by an OM protein, called TpsB [173]. Some bacteria produce several TPS systems. TpsAs can have several functions in biofilm formation. They can function in adhesion to biotic surfaces and in interbacterial interactions [173,174]. In *N. meningitidis*, TpsA contributes to the maturation of biofilms, and its synthesis is upregulated during biofilm formation [175]. In many microorganisms, TpsA functions in inhibiting the growth of related bacteria [176,177]. In the proposed model, TpsA interacts with a conserved receptor on a target cell, after which a small C-terminal part is proteolytically released and transported into the target cell where it displays toxic activities [174]. Kin target cells produce an immunity protein that inhibits the toxic activity by specifically binding to the toxin. In this case, the imported toxin moiety functions as a signaling molecule and stimulates community associated behaviors, such as biofilm formation, as was demonstrated in *Burkholderia* [178]. Killed target bacteria release intracellular components, including DNA, which contributes to the biofilm formation. Additionally, this activity mediates resistance to cefotaxime in *E. coli* [179]. TpsA induces persister formation upon direct contact with cells lacking sufficient levels of immunity protein [179]. Very likely, more recently discovered secretion systems that deliver toxins to the target cells [180,181] also contribute to biofilm formation and recalcitrance, a research area that needs to be addressed.

### *3.6. HGT in Biofilms*

HGT can involve the exchange of AMR genes between bacteria and is carried out through five different mechanisms. Three of them are generally known, namely conjugation (a direct transfer of genes between cells), transformation (acquisition of DNA from the environment), and transduction (gene transfer between cells via bacteriophages). The other mechanisms involve the release of membrane vesicles (MVs), which act as DNA reservoirs, or elongated membranous structures called nanotubes, which are employed for direct cell-to-cell contact. HGT can occur at a higher rate in biofilms than in planktonic cells [182]. Indeed, biofilms play an important role in the dissemination of AMR genes, and they are considered as reservoirs of resistance genes [10]. HGT would be favored within a biofilm for three main reasons: (i) the polymicrobial nature of biofilms that make them reservoirs of genetic diversity, (ii) the structure of biofilm, which restricts bacterial motility, increases cell density and promotes interbacterial interactions, and (iii) the presence of eDNA, which is released by cell lysis or by active secretion systems and that is retained in the ECM and establishes contacts among biofilm members. Probably even more important than its role as a glue for bacterial interactions, the eDNA can be taken up by transformation, one of the main HGT mechanisms. In addition, other factors involved in HGT would be the biomass

surface, as it has been shown that high surface/volume ratios (in well-structured biofilms) increase the efficiency of plasmid transfer [183,184].

Conjugative plasmids and integrative and conjugative elements (ICEs) are transferred mostly via conjugation. Conjugation is carried out by a conjugation system based on sex pili that mediate direct contact between two cells, the donor and the recipient. After pilus retraction, intimate contact between donor and recipient allows for DNA transfer. This is probably the most common mechanism for the transfer of AMR genes within the biofilm environment. A study in *S. aureus* that showed a 16,000-fold higher transfer rate of the conjugative plasmid pGO1, which includes trimethoprim- and gentamicin-resistance genes, in biofilms than in planktonic cells, serves as a good example [185]. Likewise, in vitro biofilm experiments have demonstrated inter-family transfer by conjugation of a *bla*NDM-1 gene encoding a carbapenemase from *Enterobacteriaceae* into *P. aeruginosa* and *A. baumannii* [186]. This mechanism occurs more intensely in biofilms than in free-living bacteria because of the proximity between cells in this structure. Apart from conjugation, where cell–cell contact is established by pili, nanotubes can transport nonconjugative plasmids between closely related strains of *Bacillus subtilis* and of *E. coli* [187]. These structures have also been detected in MDR-related bacteria, such as *Acinetobacter baylyi* [188]. Future research will elucidate their involvement in HGT among biofilm members.

Chromosomal DNA and nonconjugative plasmids are exchanged through transformation. Additionally, this mechanism is favored within biofilms because of the presence of large amounts of eDNA in the ECM. An experiment that compared the transformation rate in planktonic and biofilm cells of *N. gonorrhoeae* demonstrated that the transfer efficiency of two resistance genes, *ermC* and *aadA*, was higher at early stages of biofilm formation but decreased with biofilm age [189]. However, the transformation efficiency was shown not to depend on biofilm architecture. Spreading of transformants was observed in loose biofilms under selection pressure but was hardly observed from dense biofilms. Interestingly, even conjugative transposons of the Tn916 family, coding for tetracycline resistance, were shown to be transferred through this mechanism in in vitro grown biofilms of a multispecies consortium of oral bacteria [190].

Alternatively, genomic DNA, which may include AMR genes, can be transferred by transduction, although the diffusion of some phages through biofilms could be hampered by the biofilm matrix (discussed in Section 4). It can serve as an example, as a study showed that a temperate, Shiga-toxin (Stx)-encoding bacteriophage with a chloramphenicolresistance gene (*cat*) inserted as a marker into the *stx* gene, could transfer this gene to *E. coli* within a biofilm [191]. Finally, although MV are released within biofilms and were demonstrated to transfer AMR genes, such as the β-lactamase-encoding *bla*OXA-24 gene in *A. baumannii* [192], their relevance in HGT in a biofilm environment remains to be studied.

### *3.7. Mutation and Biofilms*

Mutations in the bacterial genome can also give rise to AMR [193]. They may occur spontaneously, that is, in the absence of strong selective pressure. Spontaneous mutations are found to occur at a common rate of 10<sup>−</sup>10–10<sup>−</sup><sup>9</sup> per nucleotide per generation for many bacteria; therefore, mutants are usually already present as a minority within a population [194]. The mutation rate can increase significantly by exposure to agents that elicit oxidative stress. Oxidative stress is associated with the build-up of ROS, which may cause direct DNA damage and mutations. In certain circumstances, such as the exposure to sub lethal doses of bactericidal antibiotics, the accumulation of ROS is low, and it may promote resistance by the induction of the synthesis of multidrug efflux pumps and by mutagenesis [195]. This is a common situation when, for instance, subtherapeutic doses of antibiotics are used as growth promoters in animal production. However, more importantly, this may also occur within biofilms where antibiotic diffusion depends on the biofilm architecture. Additional mutations can also appear as a consequence of the SOS response, which induces the synthesis of error-prone DNA polymerases. Thus, stress responses can induce high mutation frequencies through different pathways. The mutation

frequency may further increase after mutations are generated in the DNA failure-prevention or repair systems. The most frequent cause is related to defects in the methyl-directed mismatch repair system, e.g., in genes such as *mutS*, *mutL*, and *uvrD*. This can lead to a 100- to 1000-fold increase in the mutation rate [196,197]. The occurrence of microorganisms with this phenotype, called hypermutators, can represent an evolutionary advantage under selective pressure by increasing the possibility of acquiring favorable mutations, including mutations leading to AMR [198].

The hypermutator phenotype in biofilms has been detected in chronic infections in patients with cystic fibrosis, where 53% of the *Pseudomonas*isolates were hypermutable [199]. The frequency of mutants resistant to rifampicin and ciprofloxacin was higher in *Pseudomonas* biofilms than in free-living bacteria [200]. This state of hypermutability has also been reported in other bacteria isolated from cystic fibrosis patients, such as *S. aureus* and *H. influenzae* [201,202] but not in clinical isolates of the *Enterobacteriaceae* family from acute urinary tract infections governed by biofilms [203]. Thus, the hypermutation may be favored in some biofilm environments but is not a general mechanism. Altogether, bacteria in biofilms could be in a highly mutable state due to growth restrictions and nonlethal selective pressure, possibly further increased by antibiotic treatment, high- and hypermutability is also disadvantageous due to the accumulation of deleterious mutations. A solution could be transient hypermutability [204], where always a part of a population is transiently in a hypermutable state and prone to selection for favorable mutations, whereas the rest is rather stable. To the best of our knowledge, a transient hypermutator stage has not been experimentally demonstrated in biofilms. However, main conditions are accomplished: slow growth and nonlethal selective pressure. This raises the question whether transient hypermutability is a natural biological phenomenon that contributes to biofilm resistance.

In general, mutations that affect the bacterial susceptibility to antibiotics were described (i) to alter an antibiotic target, (ii) to increase the production of efflux pump, (iii) to lead to changes in the cell membranes, or (iv) to increase the production or alter the substrate specificity of enzymes that inactivate antibiotics. For example, mutations affecting the antibiotic target of aminoglycosides were in the *rspL* gene [205], which code for 16S rRNA and the S12 ribosomal protein, respectively. Mutations in the *mexZ* gene in clinical isolates of *Pseudomonas* resulted in overproduction of the efflux system MexXY-OprM [206]. Colistin resistance has been associated with mutations in the genes coding for the PmrAB two-component regulatory system that regulates the addition of aminoarabinose to lipid A [207]. Additionally, mutations resulting in increased production of β-lactamases, e.g., by mutations in the promoter of the chromosomal *ampC* gene [208] or by increase of plasmid copy number [150] have been described, among others. In addition, the genes have evolved over the years, showing a large number of β-lactamase variants with point mutations in the gene resulting in changes in the amino-acid sequence [209]. This has led to the development of extended-spectrum β-lactamases (ESBLs) that degrade also first, second, and third generation cephalosporins and/or became resistant to β-lactamase inhibitors [210].

### **4. Control of Biofilm Infections**

### *4.1. Lessons from Recalcitrant Mechanisms*

As biofilms contribute to bacterial pathogenicity and recalcitrance, novel strategies and agents are required to deal with this issue. We have now clear evidence that the antibiotics used for the treatment of biofilm infections should be carefully selected, and such selection should consider the mechanisms of resistance and tolerance of biofilms. The use of cocktails of antibiotics would probably be more successful than a single antibiotic, but the antibiotic combination should also be thoroughly considered. Antibiotics should cover the heterogenic nature of biofilms. While one of the antibiotics in the combination should be active against persisters (e.g., colistin), others should target growing cells (e.g., ciprofloxacin, tobramycin, or β-lactams). In addition, the selection of antibiotics will benefit from the characterization of ECM composition, particularly the sorption and charge of the matrix, as these properties are relevant contributors to AMR.

Many alternatives to antibiotics have been proposed to inhibit and/or eradicate biofilms. Their nature and their mechanisms of action are ample. In general, they possess one or several activities as (i) biofilm inhibitors, (ii) biofilm dispersers, and (iii) antimicrobials. An overview of these substances is listed in Table 1 and briefly discussed here. QS inhibitors can act as biofilm inhibitors or biofilm dispersers. Several plant-derived compounds exhibit this property, including halogenated furanones, which are molecules similar to N-acyl-homoserine lactones that prevent these QS signaling molecules to interact with their receptor, e.g., a LuxR family member. Thus, they function as antagonist of LuxR and repress expression of QS-induced genes [211,212]. Flavonoids, such as quercetin, can also interfere with QS. Quercetin represses the production of exopolysaccharides in *S. aureus*, required for initiation of biofilm formation [213,214]. However, QS involves a large variety of molecules in different organisms and their role in biofilm formation is species specific, thus, the activity of QS inhibitors is limited. Alternatively, enzymes that degrade QS signaling molecules such as lactonases that degrade lactone rings or phosphorylases have shown great promise [215,216], but again substrate specificity may limit their use. Other substances could contribute to inhibiting biofilm formation by interfering with the SR. For example, the 12-residue peptide 1018 interacts with (p)ppGpp and inhibits the accumulation of the alarmone and, thereby, persister formation, although this mechanism was later disputed [217]. The peptide prevents biofilm formation of different Gram-negative and Gram-positive bacteria [218] and revealed significant synergistic activity to eradicate biofilms in combination with antibiotics [219]. Eugenol is a secondary metabolite from clove (*Syzigium aromaticum*) with antibacterial activity. It inhibits biofilm formation and downregulates *relA* [220] leading to inhibition of the alarmone activation.

**Table 1.** Proposed alternatives to antibiotics with antimicrobial or antibiofilm activities. The substance, the mechanism of action (including anti-biofilm activity), and the target bacterial species are indicated for each agent.




**Table 1.** *Cont.*


OM: outer membrane; MRSA: methicillin-resistant *S. aureus*; QS: quorum sensing; SR: stringent response; STAMP: selectively targeted antimicrobial peptide; IM: inner membrane; EDTA: ethylenediaminetetraacetic acid; AHL: *N*-acyl-homoserine actones; PNAG: poly-(β-1,6)- *N*-acetylglucosamine; DNase: deoxyribonuclease; ECM: extracellular matrix.

> ECM-disrupting enzymes are potentially also suitable inhibitors and dispersers of biofilms. Addition of exogenous enzymes such as Dispersin B or DNase I, which hydrolyze PNAG and eDNA, respectively, in combination with antibiotics eradicate biofilms of different species [290–292]. Other enzymes, such as alginate lyase [274], lysozyme [293], and lysostaphin [278], showed promising results in this respect. Additionally, proteases that cleave proteins of the ECM or proteins located at the bacterial cell surface with a function in biofilm formation disrupted streptococcal [279] and *S. aureus* biofilms [280]. Additionally, addition of exopolysaccharides of the ECM from biofilms of some bacteria can be used to inhibit biofilm formation of other microorganisms. For instance, Psl and Pel, which are produced in *Pseudomonas* biofilms, eradicate biofilms of *S. epidermidis* [289], and the polysaccharide A101 from *Vibrio* sp. QY101 disperses *Pseudomonas* biofilms [294]. Probably, these charged polysaccharides outcompete structures essential for biofilm integrity. Another molecule that destabilizes the ECM is ethyl pyruvate, which, in combination with the Ca2+-chelator EDTA, inhibits biofilms of many microorganisms [288].

> Molecules that inhibit the adhesion properties of bacteria prevent the initiation of biofilm formation. Mannosides are small molecules that inhibit FimH [287], a mannosebinding component of the Type I pili that facilitates adhesion of uropathogenic *E. coli*. Mannosides can be used in combination with antibacterial agents to prevent biofilms on catheters [295]. Similarly, pilicides, which are small ring-fused 2-pyridones, inhibit Type I piliation [286]. Small peptides, such as FN075 and BibC6, block the assembly of curli and pili by disrupting protein–protein interactions during assembly and thereby inhibit the formation of *E. coli* biofilms [285]. Overall, biofilm inhibitors and dispersers utilize different mechanisms that ultimately disrupt intermolecular interactions required for the biogenesis and establishment of biofilms or they degrade these components. As these activities do not affect bacterial viability, they must be provided in combination with antimicrobials for bacterial eradication.

### *4.2. Antimicrobial Substances*

New antimicrobial substances, some of which exhibit good penetration in biofilms, have been proposed as alternatives to antibiotics. Among them, antimicrobial peptides, bacteriophages, and essential oils stand out as most promising and several examples are listed in Table 1. Antimicrobial peptides (AMPs) are small peptides of about 12–50 amino-acid residues, containing a considerable number of hydrophobic residues (≈50%) and positively charged residues [296]. They are produced by the innate immune system of animals, insects, plants, and humans to prevent bacterial, fungal, and viral infections [297,298]. They disrupt bacterial membranes through either one of three different mechanisms, (i) detergent-like membrane packing disruption, (ii) formation of pores in the barrel-stave model and (iii) toroidalpore model [299]. In addition, they can inhibit DNA, RNA, and protein synthesis. Hence, AMPs have a broad activity spectrum against microbes and, consequently, the probability of AMR development is relatively low compared to conventional antibiotics. Some AMPs from different sources have shown a good combination of antimicrobial and antibiofilm activities against superbugs (see examples in Table 1 and expanded in the AMP database: http://aps.unmc.edu/AP). An example is melittin, which is a major component of honeybee venom [300]. This cationic linear peptide of 26 amino-acid residues inserts into bacterial membranes forming short-lived pores and it also inhibits biofilm formation of several bacteria, including *P. aeruginosa* [221,223] and *K. pneumoniae* [222],

among others (Table 1). However, natural AMPs exhibit drawbacks for their application in vivo, including low efficiency, low biostability due to enzymatic degradation, toxicity at the required concentrations, and inefficient delivery to the infection niche. In an effort to improve their utility, several strategies are being conducted, comprising the design of synthetic AMPs, combination with antibiotics, or conjugation to carriers (Table 1). As an example, cyclic derivatives of peptide1018 have been created to enhance the proteolytic stability and reduce aggregation of the peptide [301]. When 1018 was coadministrated with antibiotics, a high synergistic ability to prevent and eradicate biofilms of many bacteria was observed [219]. Some AMPs were encapsulated in vehicles such as polymers, nanoparticles, micelles, carbon nanotubes, and others [reviewed in 302]. The AMPs-carrying vehicles can diffuse through tissue layers and expose simultaneously a large number of peptides improving their effectivity while lowering toxicity and reducing degradation. One of the most commonly used delivery systems is gold nanoparticles (AuNPs) [302]. They have been proposed as conjugate to AMPs because they are of small size, high solubility, stability, and biocompatibility. Esculentin-1a conjugated to AuNPs [AuNPs@Esc(1–21)] exhibited about 15-fold higher activity than the peptide alone and, in contrast to the peptide, it was not toxic. In addition, it showed high resistance to proteases [238].

Phage therapy involves the use of lytic bacteriophages to kill bacteria [303]. It has some advantages compared with other antimicrobials, for example, their natural origin, lack of toxicity for humans or nontarget microbes, and their effectiveness against antibioticresistant bacteria. Moreover, they are self-replicating in the presence of host cells and disappear without host. As a disadvantage, phages are strain specific; hence, a successful treatment requires a full understanding of bacteriophage–host interactions, involving identification of the specific phage. Although the chance seems to be low, bacteria can acquire phage resistance at high frequency. A simple point mutation in the phage receptor on the bacterial cell surface already suffices for the bacteria to escape phage attack. In addition, bacteria have several broad strategies to escape from phage, e.g., CRISPR-Cas, restriction enzymes, O-antigen, etc. To overcome these limitations, a combination of phages (phage cocktails) is often recommended instead of a single phage. This therapy is already supported by authorities in certain countries where commercial products against bacteria, such as *Listeria monocytogenes*, *Salmonella enterica* and *E. coli*, as surface disinfectants or processing aids are available. Yet, while phage receptors are fully available in planktonic cells, their accessibility in biofilms is compromised. ECM structures can establish electrostatic interactions with phage particles preventing them from reaching the cell surface. However, some phages may carry polysaccharide-degrading enzymes and thus gain access to receptors on the bacterial cell wall. Additionally, the ECM contains released phage receptors from cell lysis that ultimately compete with cell-surface receptors. Enzymes contained in ECM, such as proteases, inactivate phages. On the other hand, the architecture of biofilms may limit phage diffusion. Biofilms with dense cell clusters established by tight cell–cell binding can limit the access of the phage to the entire community. Dormant cells within the biofilm are less susceptible to phages, as phage replication requires active bacterial metabolism [304,305]. Furthermore, biofilms can generate a state of hypermutability that stimulates the occurrence of phage resistance. Indeed, the emergence of phage-resistant populations among bacteria after phage therapy has been reported [251,306]. Overall, although lytic phages have bactericidal activities, only some hold some promise in the treatment biofilm infections (Table 1). For example, phage EFDG1 showed success in eliminating biofilms in vitro and preventing infection by *E. faecalis* and *E. faecium* [240], and bacteriophage vB\_EfaH\_EF1TV was recently shown to kill clinical *E. faecalis* strains and disrupt their biofilms [241]. Yet, to overcome phage-therapy limitations, different strategies are currently followed including combination with antimicrobials, phage cocktails, and genetically manipulated phages (Table 1). Examples of the latter include phages producing biofilm degrading enzymes such as dispersin B [307] or inhibiting enzymes or enzymes that that contribute to antibiotic penetration such as OmpF porin to enhance antibiotic penetration [308].

Essential oils extracted from plants comprise complex mixtures of volatile substances, including terpenes, terpenoids, and phenols, among others. Some of these compounds possess antimicrobial activity as they constitute part of the immune defense mechanism of plants against infectious agents. Several studies reported the antibiofilm activity of some essential oils against bacteria (Table 1). Many of them damage the bacterial membranes leading to the release of cytoplasm, although their mechanism of action is not uniquely caused by this route. Other essential oils also regulate the expression of genes involved in biofilm formation and biofilm dispersal. For example, essential oils from thyme, cinnamon, and clove exhibited a high antibiofilm activity against many bacteria, including ESKAPE bacteria [257]. Cinnamon oil was earlier proven to inhibit the production of rhamnolipids, proteases, and alginate as well as swarming motility in *P. aeruginosa* [256], which is consistent with inhibition of QS. Another essential oil, tea tree oil (TTO), has shown antibacterial and antibiofilm activity. TTO eradicates *S. aureus* biofilms by affecting the expression of 304 genes participating in many metabolic routes [260] and regulators such as SarA. The global regulator SarA positively controls expression of genes involved in biofilm formation. Additionally, the expression of *cidA* that encodes a murein-hydrolase regulator was downregulated whereas the expression of the *lgrA* and *B* operons, which inhibit autolysis, was upregulated. Together, this reduces the release of eDNA, which is a key component of the *S. aureus* ECM. Other studies have investigated the synergistic action of essential oils or their active principles with other antimicrobial molecules such as synthetic antimicrobial polymers (Table 1). Thus, these and other studies demonstrate that essential oils have a repertoire of killing activities and that they are promising as treatments against biofilms. Yet, their extraction is one of the most effort-requiring and time-consuming processes, which increases the costs of their application.

### *4.3. Alternative Methods*

Physical methods hold promise for eradication and inhibition of biofilms. This is particularly important on surfaces such as chronic wounds [309,310], infected prosthetics, implants, and medical devices. Good examples of these methods include nanoparticles, sonication, irradiation (ultraviolet, visible, or infrared light), or biomaterials. Indeed, blue light, for example, was effective against biofilms formed by *A. baumannii*, *P. aeruginosa*, and *N. gonorrhoeae* although less so against biofilms of *E. coli* and *E. faecalis* [311]. In vivo studies in mouse burns showed that blue-light exposure could drastically reduce bacterial load and effectively protect mice from lethal infection with *P. aeruginosa* [312]. Blue light presumably exerts its effect on bacterial cells by exciting porphyrins which then generate ROS, as suggested by the resistance to blue light exhibited by a *P. aeruginosa* mutant defective in porphyrin biosynthetis [313]. Additionally, ultraviolet C light has been shown to efficiently eradicate *Pseudomonas* biofilms on precontaminated catheter-like tubes [314]. In general, irradiation has only an effect on superficial epidermal layers or the surface of materials because of poor accessibility of deeper tissues. On the other hand, nanoparticles, prepared from diverse materials, including both organic and inorganic materials, have a broad spectrum of antibacterial and antibiofilm activities, e.g., by disrupting bacterial membranes, interacting with proteins or DNA, or promoting the production of ROS [315]. Alternatively, different materials with topographic patterning have been developed to prevent biofilm formation. As the nature, hydrophobicity, and topology of the materials are relevant for substrate–bacteria interactions, these characteristics are conveniently modified in biomedical polymeric surfaces to generate catheters, implants, or devices with reduced bacteria-binding capacity [316,317]. Their effectivity is enhanced in combination with other antibacterial strategies, e.g., the use of nitric oxide-releasing materials, which, together, showed high synergic activity in the inhibition of bacterial growth and biofilm formation of *S. epidermidis* [318]. To summarize, compared with conventional antimicrobial agents, physical methods exhibit a broad-spectrum effectiveness under ambient conditions, are easy to operate at low cost, and low maintenance. Important disadvantages are long exposition times, and their application is mostly restricted to surfaces. Another interesting strategy for

controlling biofilms is the use of probiotics, e.g., live microorganisms with demonstrated health benefits that inhibit pathogenic biofilms. Probiotics have been considered for human therapeutic applications, and even bacterial strains have been genetically modified to kill pathogenic strains and inhibit biofilm production. Probably the best example is the *E. coli* strain Nissle 1917 that has been extensively used for treatment of intestinal disorders [319]. This strain inhibits biofilm formation of pathogenic and nonpathogenic *E. coli*, *S. aureus*, and *S. epidermidis* [320]. In an attempt to improve its therapeutic potential, the strain was genetically modified to synthesize an antibiofilm enzyme, dispersin B, in response to the detection of autoinducers secreted by *P. aeruginosa*. The recombinant strain was active against *P. aeruginosa* gut infection in animal models [321].

### **5. Concluding Remarks**

The capacity of microorganisms to evolve and adapt to environmental cues has led to a health crisis as they became resistant to most, or almost all, commercial antibiotics. Biofilm formation is an ancient form of bacterial adaptation that contributes substantially to the problem because of their recalcitrance to treatment. Indeed, biofilms are the origin of significant morbidity and mortality. As discussed here, biofilm recalcitrance integrates many mechanisms, including metabolic heterogeneity, stress responses, efflux pump regulation, entrapment and inactivation of antibiotics in the ECM, interbacterial communication, increased mutability, and exchange of genetic material. Many of these factors have been discovered particularly in strains of *P. aeruginosa*. However, the specificity and multifaceted nature of the described mechanisms indicate the necessity of studying them also in other bacteria. Even more challenging, but necessary, will be to study biofilms in natural infections, where heterogeneous bacterial populations are common, and many environmental factors, including host defenses or diffusion of antibiotics in tissues, are present.

The understanding of the mechanisms that mediate recalcitrance will definitely guide therapeutic strategies to successfully deal with biofilm infections. These should be accompanied with methodologies for rapid diagnosis of biofilm infections and characterization of the biofilm biology and composition in vivo. Additionally, the availability of a panel of substances to inhibit and disperse biofilms will contribute to the selection of adequate therapeutic strategies to deal with particular biofilm infections.

**Author Contributions:** Conceptualization, J.A.; Methodology, C.U., G.C.-E., J.T., R.C.M.-J., J.A. Writing—Original draft preparation, C.U., G.C.-E., R.C.M.-J., J.A.; Writing—Review and editing, R.C.M.-J., J.T., J.A.; Visualization, C.U., G.C.-E.; Supervision, J.A. All authors have read and agreed to the published version of the manuscript.

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

**Acknowledgments:** We wish to thank Bridier Arnaud (Fougères Laboratory, ANSES, France) and Romain Briandet (Université Paris-Saclay, France) for kindly providing a high-resolution figure about the architecture of MDR biofilms.

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

### **References**


### *Article* **Biofilm-Induced Antibiotic Resistance in Clinical** *Acinetobacter baumannii* **Isolates**

**Abebe Mekuria Shenkutie <sup>1</sup> , Mian Zhi Yao 1, Gilman Kit-hang Siu 1, Barry Kin Chung Wong <sup>2</sup> and Polly Hang-mei Leung 1,\***


Received: 22 October 2020; Accepted: 14 November 2020; Published: 17 November 2020 -

**Abstract:** In order to understand the role of biofilm in the emergence of antibiotic resistance, a total of 104 clinical *Acinetobacter baumannii* strains were investigated for their biofilm-forming capacities and genes associated with biofilm formation. Selected biofilm-formers were tested for antibiotic susceptibilities when grown in biofilm phase. Reversibility of antibiotic susceptibility in planktonic cells regrown from biofilm were investigated. We found 59.6% of the strains were biofilm-formers, among which, 66.1% were non-multidrug resistant (MDR) strains. Presence of virulence genes *bap*, *csuE*, and *abaI* was significantly associated with biofilm-forming capacities. When strains were grown in biofilm state, the minimum biofilm eradication concentrations were 44, 407, and 364 times higher than the minimum bactericidal concentrations (MBC) for colistin, ciprofloxacin, and imipenem, respectively. Persisters were detected after treating the biofilm at 32–256 times the MBC of planktonic cells. Reversibility test for antibiotic susceptibility showed that biofilm formation induced reversible antibiotic tolerance in the non-MDR strains but a higher level of irreversible resistance in the extensively drug-resistant (XDR) strain. In summary, we showed that the non-MDR strains were strong biofilm-formers. Presence of persisters in biofilm contributed to the reduced antibiotic susceptibilities. Biofilm-grown *Acinetobacter baumannii* has induced antibiotic tolerance in non-MDR strains and increased resistance levels in XDR strains. To address the regulatory mechanisms of biofilm-specific resistance, thorough investigations at genome and transcription levels are warranted.

**Keywords:** *Acinetobacter baumannii*; biofilm; antibiotic resistance; antibiotic tolerance; persister

### **1. Introduction**

*A. baumannii* is a significant opportunistic pathogen responsible for a high proportion of healthcare-associated infections [1]. Due to the critical impact of the multidrug-resistant *A. baumannii* on public health, the World Health Organization has categorized this organism as a priority I pathogen among the antibiotic-resistant microorganisms [2–6]. Recent reports have shown that the environmental reservoir is the major source of multidrug-resistant *A. baumannii* outbreaks in hospital environments [7]. The ability of *A. baumannii* to form biofilm facilitates its survival and persistence in the hospital environments [8,9], this in turn contributes to the extensive spread of this pathogen across the globe [10].

Biofilm is a multilayer of highly coordinated microorganisms attached to surfaces in the presence of moisture. Bacterial cells within the biofilm are highly coordinated and undergo phenotype switch to produce a community that is resistant to the adverse external environment. Such phenotype switch also promotes the emergence of antibiotic resistance through the expression of the antibiotic-resistance genes, genetic mutation, or transfer of genes associated with antibiotic-resistance [11].

Virulence genes associated with biofilm formation in *A. baumannii* include *bap*, *csu* locus, *adeFGH*, *ompA*, and *abaI* [10]. *bap* encodes a large bacterial surface protein consisting of 8621 amino acids [11]. The predicted structure of Bap was similar to bacterial adhesins of the immunoglobulin-like fold superfamily and may function as an intercellular adhesin that supports the development of the mature biofilm structure [12]. *csuE* belongs to the *csu* operon, which encodes the chaperone-usher pili assembly system as production of pili is required in the early steps of biofilm formation on abiotic surfaces [13]. *ompA* encodes a porin protein that is involved in adhesion to the epithelial cell, antibiotic resistance, and biofilm formation [8]. *adeFGH* encodes a resistance–nodulation–cell division antibiotics efflux system, which is involved in the synthesis and transportation of autoinducer molecules during biofilm formation [14]. *abaI* encodes an autoinducer synthase, which is an enzyme involved in quorum sensing. Mutation of *abaI* fails to produce acyl-homoserine lactone signals and impairs biofilm maturation [5]. Certain biofilm-associated genes influence the expression of antibiotic resistance, suggesting the link between biofilm formation and antibiotic resistance.

In the past decade, most of the published studies on antimicrobial resistance of *A. baumannii* focused on the planktonic states of the pathogen [5,15,16]. However, biofilm of *A. baumannii* is responsible for various types of catheter-related infections, antimicrobial resistance performed using planktonic cells may not be representative for biofilms. In order to understand the virulence potential and the influence of biofilm growth on antibiotic susceptibility, we characterized the biofilm-producing capacities of the *A. baumannii* strains isolated from clinical specimens and to examine the responses of biofilm cells to antimicrobial agents. It has been documented that antibiotic-resistant strains of *A. baumannii* were strong biofilm formers [10], such properties are beneficial to the survival and dissemination of the resistant strains. However, antibiotic-sensitive strains are vulnerable to antibiotic challenges, in order to protect themselves from antibiotics present in the surrounding environments, the sensitive strains may have the abilities to form biofilm. Therefore, the study investigated the relationship between antibiotic susceptibility and biofilm-forming ability. The results obtained from the phenotypic characterization of the biofilm-producing capacity and biofilm susceptibility tests may offer valuable insights into the development of preventive strategies for biofilm-associated infections in hospital environments.

### **2. Materials and Methods**

### *2.1. Clinical Isolates*

A total of 104 archived and nonduplicate *Acinetobacter species* isolates were collected from different hospitals in Hong Kong. All strains of *Acinetobacter* species were collected from sputum, blood, urine, soft tissue, and hospital environments. The collected isolates were kept in Luria-Bertani (Oxoid Ltd., Basingstoke, UK) broth containing 20% glycerol at −80 ◦C until use. The reference strains *A. baumannii* ATCC 19606 and *E. coli* ATCC 25922 were used as control strains in the biofilm assay or antibiotic susceptibility tests.

### *2.2. Confirmation of Bacterial Identities*

Identities of the *A. baumannii* isolates were confirmed using a multiplex PCR assay. Three pairs of primers targeting *recA*, *gyrB*, and ITS were used [17]. PCR was performed in a final reaction volume of 40 μL containing 5X Phusion HF Buffer, 10 mM dNTP, and Phusion DNA Polymerase (all from New England BioLabs, MA, USA), and 0.5 μM of each primer using a Veriti thermocycler (Applied Biosystems, Foster City, CA, USA) under the following conditions: initial denaturation at 98 ◦C for 5 min; 35 cycles of denaturation (95 ◦C, 30 s), annealing (54 ◦C, 30 s) and extension (72 ◦C, 1 min); a final extension step at 72 ◦C for 10 min. PCR products were separated by 1.5% agarose gel electrophoresis the gel was stained with 0.5 μg/mL RedSafe nucleic acid staining solution

(iNtRON biotechnology, Gyeonggi, Korea). The stained gel was visualized using a Gel Doc System XR (Bio-Rad Laboratories, Richmond, CA, USA).

### *2.3. Multilocus Sequence Typing (MLST)*

The Oxford scheme of MLST targeting seven chromosomal housekeeping genes was performed according to the method described in the MLST database (http://pubmlst.org/abaumannii). The seven housekeeping genes (*gltA*, *gyrB*, *ghdB*, *recA*, *cpn60*, *gpi*, and *rpoD*) of 104 *A. baumannii* isolates were amplified and sequenced according to the method described by Bartual et al. [18]. The allele number of each gene was obtained by comparing its sequence with the reference sequences in the database. The sequence type of a given isolate was identified by matching the seven locus numbers obtained. If a sequence did not match with any reference sequences in the database, it was designated as a new allele. In addition, if the seven loci did not match with any existing allele combinations in the database, the isolates were regarded as new sequence types.

### *2.4. Biofilm Formation*

Biofilm-forming capacities of the isolates were evaluated with a Calgary Biofilm Device (CBD) with slight modifications of the procedures described by Ceri et al. [19] and the manufacturer (Innovotech, Edmonton, AB, Canada). Briefly, 3–5 colonies were picked from an overnight LB agar plate and inoculated into 10 mL Maximum Recovery Diluent to a turbidity equivalent to 0.5 McFarland standard. The inoculum was mixed 1:1 with LB broth. For each *A. baumannii* strain, 150 μL of bacterial suspension was inoculated into the well of a 96-well microtiter plate. The plate was then covered with a plastic lid with 96 pegs (Innovotech). The microtiter plate was incubated at 37 ◦C for 48 h on a platform shaker set at 110 rpm. After incubation, planktonic cells were aspirated from the wells and discarded; the wells were washed three times with Maximum Recovery Diluent. Both the plates and lids with pegs were inverted and allowed to dry for 2 h at room temperature. Biofilm mass quantification was performed in octuplicate, and each assay was repeated on three separate days.

### *2.5. Quantification of Biofilm Mass*

Biofilm masses formed on the 96-well microtiter plate was quantified using the crystal violet staining method. After drying of the microtiter plate and the lid with pegs, 200 μL of 0.1% aqueous crystal violet solution was added into each well of the microtiter plate, which was covered with the lid with pegs. The set up was incubated at room temperature for 15 min. After staining, the wells and the pegs were washed three times with Maximum Recovery Diluent to remove the excess stain and air-dried. After drying, 200 μL of 33% (*v*/*v*) acetic acid was added to each well of the microtiter plate to extract the crystal violet bound to the biofilm. Absorbance was measured at 570 nm using a microplate spectrophotometer (BIO-RAD). The average optical density (OD) for each *A. baumannii* isolate was calculated, and the biofilm-forming capacity was interpreted according to the guideline described by Stepanovi´c et al. [20]. The cut-off OD (ODc) was defined as three standard deviations above the mean OD of the uninoculated control. Classification criteria are shown in Table 1.

**Table 1.** Classification of biofilm-forming capacity based on optical density (OD) value.


### *2.6. Detection of Biofilm-Associated Genes*

PCR detection of biofilm-associated genes (*bap*, *csuE*, *ompA*, *adeFGH*, *abaI*) was performed on a Veriti thermocycler (Applied Biosystems) using sets of primers shown in Table S1. PCR was performed in a 20-μL reaction consisting of 1 μL extracted genomic DNA, 10 μL Luna Universal qPCR Master Mix (New England Biolabs, USA), 1 μL (10 μM) of each primer. The reaction condition was initial denaturation at 95 ◦C for 7 min, followed by 35 cycles of denaturation at 95 ◦C for 30 s, annealing at 60 ◦C for 1 min and extension at 72 ◦C for 30 s, and a final extension of 72 ◦C for 10 min. PCR products were separated by 1.5% agarose gel electrophoresis the gel was stained with 0.5 μg/mL RedSafe nucleic acid staining solution (iNtRON Biotechnology, Korea). The stained gel was visualized using a Gel Doc System XR (Bio-Rad Laboratories).

### *2.7. Determination of Minimal Inhibitory Concentration (MIC) and Minimum Bactericidal Concentration (MBC)*

Antimicrobial susceptibility testing was determined by broth microdilution techniques according to the procedures described in the CLSI guidelines [21]. Interpretation of susceptible, intermediate, and resistant was based on CLSI guidelines. Ten antimicrobials covering seven categories of antimicrobials used for the treatment of *A. baumannii* infections were included in the determination of MIC and MBC. These included penicillin-β-lactamase inhibitor (ampicillin-sulbactam), cephalosporin (cefotaxime, ceftazidime), a carbapenem (imipenem, meropenem), aminoglycoside (gentamycin), a fluoroquinolone (ciprofloxacin, levofloxacin), tetracycline (tetracycline), lipopeptide (colistin) were included in the study. *E. coli* ATCC 25922 and *A. baumannii* ATCC 19606 (*A. baumannii* ST1861) were used as control strains. The strains were regarded as multidrug-resistant (MDR) if they were resistant to at least three classes of antimicrobial agents, including penicillin and cephalosporin (including inhibitor combinations), fluoroquinolones, and aminoglycosides. MDR strains that were resistant to carbapenem were regarded as extensively drug-resistant (XDR) [22]. XDR strains were resistant to colistin, and all other classes of antibiotics were regarded as pan drug-resistant (PDR).

After MIC determination, 10 μL aliquots from the wells that showed no visible bacterial growth were inoculated onto LB agar plates and incubated at 37 ◦C for 24 h. The MBC was the lowest concentration of antibiotic that killed 99.9% of the initial bacterial population.

### *2.8. Determination of Minimal Biofilm Inhibitory Concentration (MBIC) and Minimum Biofilm Eradication Concentration (MBEC) Assay*

The antibiotic susceptibility test for biofilm was performed according to the procedures described by Moskowitz et al. [23]. After the biofilm was formed on the CBD pegs, the pegs were rinsed three times in a 96-well plate containing 150 μL of Maximum Recovery Diluent. The lid with pegs was then transferred to the new a standard 96-well plate with wells containing 150 μL Mueller–Hinton broth with antibiotics diluted two-fold serially (ranged from 2 to 1024 μg/mL). The 96-well plate was incubated overnight at 37 ◦C. Following incubation, turbidity in each well was examined visually. MBIC was defined as the minimal antibiotic concentration at which no bacterial growth was observed, which meant the minimal antibiotic concentration that inhibited the release of planktonic bacteria from the biofilm.

Following the MBIC examination, MBEC was determined. The lid with pegs was removed and rinsed three times in a 96-well plate containing 150 μL of Maximum Recovery Diluent to remove the planktonic cells. The lid was then placed in a second 96-well plate containing 150 μL Mueller–Hinton broth. The plate was shaken at a speed of 180 rpm for 10 min to detach the biofilm cells from the pegs into the Mueller–Hinton broth. The viability of the biofilm was determined by plate count after 24 h of incubation at 37 ◦C. MBEC was defined as the minimal antibiotic concentration required to eradicate the biofilm.

### *2.9. Confocal Laser Scanning Microscopy Imaging of Biofilm*

Confocal laser scanning microscopy (CLSM) was performed to estimate the proportion of viable cells in the biofilm formed on the pegs. After the formation of biofilm, the pegs were separated from the lid using sterilized pliers and washed three times with Maximum Recovery Diluent to remove the planktonic cells [19]. Biofilm cells on the pegs were dual stained with a mixture of 3.35 μM SYTO-9 and 20 μM propidium iodide according to the instructions of the Film Tracer Live/Dead Biofilm Viability Kit (Cat no: L10316 Invitrogen). CLSM images were acquired using a Leica TCS SPE Confocal Microscope (Leica) with a 63x objective lens. The live and dead cells embedded in subpopulations of biofilm cells were estimated using a software BioFilmAnalyzer [24].

### **3. Reversibility of Antibiotic Resistance of the Biofilm Cells**

A study was performed to evaluate the antibiotic susceptibility profile after the biofilm cells were grown in the planktonic state. Two hyper-biofilm forming *A. baumannii* (ST1894 and ST373) and one weak biofilm former (ST195) were selected for this part of the study. ST1894 and ST373 were non-MDR strains; ST195 was an XDR strain. The biofilm cells of these three *A. baumannii* strains were subcultured on LB agar. MBICs of three antibiotics (colistin, imipenem, and ciprofloxacin) for the biofilm cells and MICs for the resumed planktonic cells were determined according to the procedures described above.

### *3.1. Enumeration of Persister Cells from Planktonic and Biofilm Populations*

Persister cells in the planktonic and biofilm populations were enumerated according to the procedures described by Marques [25] with slight modification. A hyper-biofilm-producing and non-MDR *A. baumannii* ST1894 strain were selected for the isolation of persister cells. The strain was streaked onto a LB agar plate and incubated at 37 ◦C for 24 h. One to two colonies were inoculated into 5 mL LB broth, which was incubated at 37 ◦C with agitation (180 rpm) for 12 h. After incubation, the broth culture was diluted to 1% in 20 mL of fresh LB broth and incubated at 37 ◦C with agitation (180 rpm) for 24 h.

To enumerate persister cells in the planktonic population, 10 mL of overnight LB broth culture of ST1894 was collected by centrifugation at 8000 rpm for 10 min at 4 ◦C and resuspended in 10 mL of saline at 4 ◦C. The washing step was repeated one more time. Cell density was adjusted to an absorbance of 0.8 at 600 nm. A duplicated set of bacterial cell suspension was prepared in the same way. Ninty-eight microliters of 2048 μg/mL ciprofloxacin (20× MIC) or 98 μL saline containing 0.1% acetic acid was added to the 10 mL bacterial cell suspension and incubated at 37 ◦C with agitation at 180 rpm for 24 h. Viable bacterial cells were enumerated using the plate count method. One hundred microliters of the experiment and control samples were inoculated onto the LB agar plate containing 1% MgCl2.7H2O to neutralize the ciprofloxacin. Plate count was performed in triplicate at time points 0, 1, 3, 5, 7, 9, 12, 15, and 24 h after incubation. The number of persister cells present in the planktonic population was determined according to the formula shown below:

$$\text{The number of persons in planktonic cells/mL} = \frac{\text{Number of colonies} \times \text{dilution factor}}{\text{Volume plotted}}$$

To enumerate persister cells in the biofilm population, an overnight LB broth culture of ST1894 was used to grow biofilm using a CBD according to the procedures described elsewhere in this article. After incubation for 48 h, the pegs were removed from the CBD using pliers and each peg was placed in a centrifuge tube with 1 mL Maximum Recovery Diluent. The biofilm cells formed on the pegs were detached by centrifuging at 18,000 rpm for 10 min and were resuspended in 10 mL saline at 4 ◦C. Cell density was adjusted to an absorbance of 0.8 at 600 nm. Ninty-eight microliters of 2048 μg/mL ciprofloxacin (20 × MIC) or 98 μL saline containing 0.1% acetic acid was added to the 10 mL bacterial cell suspension and incubated at 37 ◦C with agitation at 180 rpm for 24 h. The number of viable biofilm cells for each treated and nontreated sample was counted according to the steps as described for planktonic cells. The number of persister cells isolated from the biofilm population was determined according to the formula shown below:

Number of persisters in biofilm/mL <sup>=</sup> Number of colonies×dilution factor×total sample volume Volume plated×surface area of peg

### *3.2. Statistical Analysis*

The association between biofilm-forming abilities and biofilm-associated genes was analyzed using chi-squared tests. SPSS version 24 (IBM Corporation, Armonk, NY, USA) was used for statistical analysis. *p*-value ≤0.05 was considered as statistically significant for the data analyzed in this study.

### **4. Results**

### *4.1. Antibiotic Susceptibility Profiles of the A. baumannii Strains*

The MICs for 10 antibiotics (listed in Table 2) covering seven major antibiotic classes were determined for the 104 *A. baumannii* isolates. The antibiotic susceptibility profiles of the isolates are summarized in Table 2. The results showed that 79.8% (82/104) of the isolates were susceptible to colistin, but only 4.8% (5/104) of the isolates were susceptible to ampicillin-sulbactam. There were 29.8% (31/104) to 52.9% (55/104) of isolates susceptible to the remaining eight antibiotics. Among the 104 isolates, 46.2% (48/104) of strains were non-MDR strains, 30.8% (32/104) were XDR strains, and 23.1% (24/104) were PDR strains (Table 3 and Figure S1).

**Table 2.** Antimicrobials susceptibility profiles the *Acinetobacter baumannii* isolates (n=104). S, I, and R represent susceptible, intermediate, and resistant, respectively.


**Table 3.** The relationship among biofilm-producing ability of *A. baumannii* with antibiotic resistance.


### *4.2. Biofilm-Forming Capacities of the A. baumannii Isolates*

Among the 104 *A. baumannii* isolates studied, 59.6% were biofilm formers, of which 25% were strong biofilm producers, 14.4% were moderate biofilm producers, and 20.2% were weak biofilm. The biofilm control strain *A. baumannii* ATCC19606 was a strong biofilm former. Table 3 summarizes the distribution of biofilm-forming abilities of isolates with various antibiotic susceptibility and biofilm-associated gene profiles. As shown in Table 3, the result reveals that 66.1% of the biofilm-forming isolates were non-MDR strains. In addition, 83.3% of the non-biofilm formers were resistant to multiple antibiotic classes. Overall, these results indicated that the more antibiotic susceptible *A. baumannii* isolates were able to form biofilm than the resistant isolates (*p* = 3 × 10<sup>−</sup>6).

The biofilm-forming capacities of each sequence type are also summarized in Figure S1. The results showed that the isolates with the same sequence type had different biofilm-producing capacities, as in the case of ST2028, ST1417, ST195, ST1860, and ST2037.

### *4.3. Distribution of Biofilm-Associated Virulence Genes in the Clinical A. baumannii Strains*

The percentages of *A. baumannii* strains carrying *bap*, *csuE*, *adeFGH*, *ompA*, and *abaI* were 9.6%, 85.6%, 95.2%, 89.4%, and 82.7%, respectively. The relationship between the presence of biofilm-associated genes and biofilm-forming capacities are presented in Table 4. The presence of *bap*, *csuE*, and *abaI* was significantly associated with biofilm-forming capacities of the isolates (*p*-values ranging from 0.005 to 0.033). The results showed that 100% of *bap*-positive isolates were biofilm formers. On the other hand, 55.3% of biofilm formers were *bap*-negative. Based on the results, the presence of *Bap* indicates the strains are biofilm-formers but the reverse is not true. Over 80% of the *csuE*-negative and *abaI*-negative isolates were biofilm formers, but more than 50% biofilm formers carried these two genes. Similar patterns were observed in *adeFGH* and *ompA* but were not statistically significant (*p*-values = 0.07 and 0.193, respectively).


**Table 4.** Relationship between the presence of biofilm-associated genes and biofilm-forming capacities among the *A. baumannii* isolates.

### *4.4. Comparison of MIC and MBIC*

The MICs and MBICs for colistin, ciprofloxacin, and imipenem were tested against nine different sequence types. These strains were biofilm-forming and susceptible to the three antibiotics being investigated. Antibiotic resistance profiles for biofilm compared to counter-plankton cells are summarized in Table 5. The MBICs increased drastically when compared to those of the planktonic cells. The fold-increase ranged from 2 to 32 fold for colistin, 4 to 64 fold for ciprofloxacin, and 4 to 2048 fold for imipenem. When compared between MICs and MBICs, the average fold-increase was 21, 31, and 386 for colistin, ciprofloxacin, and imipenem, respectively. The susceptibility profiles of most of the strains changed from sensitive to resistant when grown in biofilm state. However, the biofilm formed by ST1990 and ST1417 remained sensitive to ciprofloxacin and imipenem, and biofilm formed by ST1855 remained sensitive to colistin.

**Table 5.** Antibiotic susceptibility profiles of selected sequence types of *A. baumannii* planktonic and biofilm cells.


\* This strain is *Acinetobacter baumannii* ATCC 19606, the control strain for biofilm assay.

### *4.5. Comparison of MBC and MBEC*

Following the determination of MICs and MBICs, we evaluated the MBCs and MBECs of nine *A. baumannii* isolates. The results showed that the MBECs of the nine isolates have increased. Up to 64-fold increase in the concentration of colistin and up to a 1024-fold increase in the concentration of ciprofloxacin and imipenem was required to eradicate *A. baumannii* biofilm compared to the planktonic cells (Table 6). Although the three strains, ST1990, ST1417, and ST1855, had MBICs falling within the sensitive ranges, the MBECs were much higher than the MBICs (Table 5). When compared between MBCs and MBECs, the average fold-increase was 44, 407, and 364 for colistin, ciprofloxacin, and imipenem, respectively. To visualize the viability of biofilm cells after antibiotic treatment, biofilm preparations of a hyper-biofilm-producing strain ST1894 were treated with antibiotics, stained with a LIVE/DEAD BacLight bacterial viability kit and examined using CLSM. The results showed that viable cells in the biofilm can still be detected even after treating with antibiotics at 32–256 times the MBC (Figure 1). The findings of this study suggest that biofilm-producing strains *A. baumannii* cannot be eradicated with the same concentration of antimicrobials required to eradicate planktonic cells.

**Table 6.** Comparison of MBC and MBEC of *A. baumannii* of different MLSTs.


\* This strain is *Acinetobacter baumannii* ATCC 19606, the control strain for biofilm assay.

### *4.6. Reversibility of Antibiotic Susceptibility in Planktonic Cells Regrown from Biofilm*

Reversibility of antibiotic resistance was analyzed for two non-MDR strains *A. baumannii* ST1894, ST373, and one XDR strain *A. baumannii* ST195. Biofilm cells of the three strains were regrown into the planktonic phase and then treated with colistin, imipenem, and ciprofloxacin. The MIC, MBIC, and MIC of the reverted planktonic cells are summarized in Table 7. The results showed that biofilm cells of *A. baumannii* ST1894 reverted to sensitive phenotype when the strain was regrown into planktonic cells. For *A. baumannii* ST373, reversion to sensitive phenotype occurred in colistin and imipenem but not in ciprofloxacin. This suggests that biofilm cells of that strain might have developed mutation associated with ciprofloxacin resistance. Besides, other mechanisms might be involved, such as mutations in the efflux pumps or regulators of efflux pumps could also lead to ciprofloxacin resistance. *A. baumannii* ST195 was resistant to ciprofloxacin and imipenem but sensitive to colistin. The MICs of the reverted planktonic cells were the same as the MBICs for the three antibiotics. The results showed that the antibiotic susceptibility of the strain did not revert to the original pattern. The level of resistance to ciprofloxacin and imipenem has increased, and the strain did not revert to its sensitive phenotype for

colistin. Together the results showed that biofilm formation in *A. baumannii* promotes either antibiotic tolerance that is reversible or the emergence of antibiotics resistance that is irreversible.


**Table 7.** Reversibility of antibiotic resistance of biofilm cells.

### *4.7. Isolation of Persister Cells in Planktonic and Biofilm Cells*

When *A. baumannii* ST1894 was treated with ciprofloxacin at a concentration equaled 20 times of its MIC, all planktonic cells were eradicated after 16 h of exposure to the antibiotic (Figure 2). However, 100 ± 30 CFU/peg biofilm cells survived after 24-h exposure to ciprofloxacin, which supports our hypothesis that persister cells in the biofilm was one of the reasons for the reduced susceptibility to antibiotics. Besides, these persister cells were responsible for the regrowth of biofilm cells after the cells were treated with antibiotic. Further examination of biofilm using CLSM demonstrated the presence of 1.9% persister cells when the biofilm was treated with ciprofloxacin at concentration of 1024 times of its MIC (Figure 3). Hence, the presence of persister cells in biofilm could be the reason for multidrug tolerance to different classes of antibiotics.

**Figure 2.** Detection of persisters from biofilm and planktonic cells of *A. baumannii* ST 1894. The number of viable biofilm and planktonic cells at different time points after treatment with 2048 μg/mL ciprofloxacin. Red solid and dotted lines represent untreated biofilm and planktonic cells, respectively. Blue solid and dotted lines represent biofilm and planktonic cells treated with ciprofloxacin, respectively.

**Figure 3.** CLSM image of *A. baumannii* ST1894 biofilm treated with ciprofloxacin at 1024 × MIC. Viable bacterial cells in the biofilm was detected by a Live/Dead Biofilm Viability Kit. Persister cells appear green in color, dead cells appear red in color.

### **5. Discussion**

The ability of *A. baumannii* to produce biofilm enhances its survival in adverse environments and increases the risk of healthcare-associated infections. The present study evaluated the biofilm-formation abilities of the clinical *A. baumannii* strains and the role of biofilm production for reduced susceptibility to antibiotics. The result showed that 59.6% of the clinical *A. baumannii* isolates were able to form a biofilm. This result is similar to the findings reported by Anghel et al. that about 63% of *A. baumannii* clinical strains were biofilm producers [26]. Our results showed that the strains with the same sequence types might have different biofilm-producing capacities, as in the case of ST2028, ST1417, ST195, ST1860, and ST2037.

Previous studies have documented that biofilm-forming strains were more resistant to antibiotics than the non-biofilm-forming strains [10,27]. However, in this study, we found that a significantly higher proportion of biofilm-forming isolates were non-MDR strains. Our results indicated an inverse relation between antibiotic resistance and biofilm-forming ability of the isolates. This finding is also in line with other studies that biofilm-forming *A. baumannii* was more susceptible to antibiotics [28,29]. A possible explanation for this might be that biofilm formation ensures the survival of the susceptible strains when exposed to antibiotics. However, the variations in methodologies involved in assessing biofilm formation may lead to different observations among different research groups. A standardized biofilm assay is needed for more objective comparison between different studies.

In this study, we observed that the majority of XDR and PDR *A. baumannii* isolates were weak or non-biofilm producers. The results are in accordance with the recent findings that biofilm-forming isolates exhibited lower rates of carbapenem resistance than the non-biofilm-forming isolates [30,31]. This result may be explained by the fact that energy required for expressing the carbapenemase and β-lactamases might reduce the biofilm formation in the isolates harboring those genes.

We also evaluated the relationship between biofilm-associated virulence genes and biofilm-forming abilities of the *A. baumannii* strains. *bap* was detected in a small proportion of strains tested, and all *bap*-positive strains were biofilm-formers, indicating the role of *Bap* in biofilm formation. It has been documented that *Bap* protein is necessary for mature biofilm formation on various abiotic surfaces; disruption of *Bap* resulted in a reduction of biofilm mass [32]. *csuE, adeFGH*, *ompA*, and *abaI* were detected in 82.7–95.2% of the *A. baumannii* strains; almost 50% of these strains were non-biofilm-forming

strains. It is, therefore, necessary to compare the expression levels of the biofilm-associated genes in the biofilm producers and non-biofilm producers.

To examine the emergence of antibiotic resistance during biofilm formation, MIC, MBIC, MBC, and MBEC of colistin, ciprofloxacin, and imipenem against nine selected *A. baumannii* strains were studied. These nine strains were non-MDR and strong biofilm formers. The results showed that biofilm cells of the nine strains developed a high-level of antibiotic resistance and required as high as 2048 times the MIC to inhibit the release of planktonic bacterial cells from biofilm or as high as 1024 times the MBC to eradicate the biofilm cells. Among the three antibiotics tested, imipenem had a high average fold-increase in both MBIC (386-fold) and MBEC (364-fold). As imipenem is uncharged and can penetrate the negatively charged biofilm matrix, thus, the high level of imipenem resistance was unlikely due to poor penetration. Resistance could be a result of the expression of β-lactamase, as a previous study reported that the activity of the β-lactamase promoter was elevated in biofilm cells of *Pseudomonas aeruginosa* [33]. Other possibilities for the emergence of imipenem resistance in the biofilm include the expression of the drug efflux system *adeFGH*, which is involved in the synthesis and transportation of autoinducer molecules during biofilm formation [14].

Ciprofloxacin had much higher fold-increase in MBEC than that in MBIC (407-fold vs. 31-fold), as ciprofloxacin is actively against metabolically active bacterial cells [34], it will be more difficult to eradicate persister cells developed in the biofilm, which are metabolically dormant. Colistin had the lowest average fold-increase in MBIC (21-fold) and MBEC (44-fold) compared to imipenem and ciprofloxacin. Studies also showed that colistin is effective against the metabolically inactive bacterial population in the biofilm [34]. Another study also showed that the bactericidal activity of colistin increased under anaerobic conditions, as the bactericidal action of the antibiotic was independent of the hydroxyl radical formed during aerobic respiration [34]. Overall, *A. baumannii* developed antibiotic resistance during biofilm formation, and the level of biofilm-specific resistance varied according to the induced responses of the biofilm population and the action mechanisms of the antibiotics.

CLSM images of a non-MDR *A. baumannii* ST1984 strain showed that 0.5%–3.7% of biofilm cells were viable after treatment with colistin, ciprofloxacin, and imipenem at 256–4096 times the MICs of these antibiotics. It is possible that the percentages of viable cells were higher because the presence of extracellular DNA in the biofilm might be stained by the propidium iodide stain and lead to an overestimation of dead cell proportions. The presence of persister cells is worrisome since colistin is the last resort for treatment of MDR *A. baumannii*, the emergence of biofilm-specific resistance makes the pathogen untreatable with the conventional antibiotics.

Although Qi et al. have reported the enhancement of antibiotic resistance by biofilm formation [28], our study investigated the reversibility of antibiotic resistance developed in the biofilm. We re-grew the biofilm cells to planktonic state and assessed the changes in MICs in three *A. baumannii* strains. Our novel findings demonstrated that reversion from resistant to susceptible occurred in two non-MDR strains but not in the XDR strain. Such a transient increase in antibiotic resistance in the non-MDR strains was due to the tolerance of antibiotics by the biofilm cells. There are multiple mechanisms involved in the development of antibiotic tolerance in biofilm. The reduced metabolic activity of biofilm and induction of stress responses due to the limitation of nutrients in the biofilm environment triggers antibiotic tolerance [35]. The presence of persister cells in the biofilm population activates the toxin/antitoxin systems, which leads to inhibition of protein translation, initiation of dormancy, and tolerance to antibiotics [36]. In this study, we demonstrated the presence of persister cells in *A. baumannii* strain ST1894, one of the two non-MDR strains, using CLSM imaging and persister cell isolation after treating with a high concentration of ciprofloxacin. The presence of persister cells made the pathogen extremely multidrug tolerant and could eventually lead to the emergence of antibiotic resistance. For the XDR strain (*A. baumannii* ST195), the planktonic cells regrown from biofilm cells exhibited a higher level of antibiotic resistance. This could be caused by mutations of drug targets upon exposure to high antibiotic concentrations. There are two noteworthy issues regarding this XDR strain. First, irreversible colistin resistance was developed during biofilm formation,

making the physicians barehanded to treat the infected patients. Second, although the strain was a weak biofilm former, high-level of irreversible antibiotic resistance was enhanced during biofilm formation. Detailed investigation at the genetic level is needed to understand the underlying regulatory mechanisms involved in the emergence of biofilm-mediated antibiotic resistance.

### **6. Conclusions**

We showed a negative relationship between biofilm-forming capacity and antibiotic susceptibility, in which the strong biofilm-formers were non-MDR strains. We also demonstrated the presence of persisters in the biofilm cells, which accounted for reduced susceptibilities of the *A. baumannii* strains. Another important finding was that growth in biofilm induced reversible antibiotic tolerance in the non-MDR strains but a higher level of irreversible resistance in the XDR strain, including the conversion from colistin sensitive to resistant. To address the regulatory mechanisms of biofilm-specific resistance, research should be undertaken to investigate the alterations at the genome and transcription levels of *A. baumannii* grown in biofilm status. A thorough understanding of the various contributing factors facilitates the design of new therapeutics that target biofilm-specific resistance.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2079-6382/9/11/817/s1, Figure S1: Biofilm-producing abilities and antibiotic susceptibility profiles of *A. baumannii* strains with different MLSTs, Table S1: Primers used for amplification of biofilm-associated genes.

**Author Contributions:** Conceptualization, A.M.S. and P.H.-m.L.; methodology, A.M.S. and M.Z.Y.; validation, A.M.S.; formal analysis, A.M.S.; investigation, A.M.S. and M.Z.Y.; resources, G.K.-h.S. and B.K.C.W.; writing original draft preparation, A.M.S.; writing—review and editing, P.H.-m.L.; visualization, A.M.S. and P.H.-m.L.; supervision, P.H.-m.L.; project administration, P.H.-m.L.; funding acquisition, P.H.-m.L. All authors have read and agreed to the published version of the manuscript.

**Funding:** A.M. Shenkutie is supported by the Hong Kong Fellowship Scheme PH16-01625.

**Acknowledgments:** The CLSM imaging was performed at the Hong Kong Polytechnic University Research Facility in Life Sciences with the support of Clara H.L. Hung.

**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/).

*Article*
