*Article* **Individual Pharmacotherapy Management (IPM)—IV: Optimized Usage of Approved Antimicrobials Addressing Under-Recognized Adverse Drug Reactions and Drug-Drug Interactions in Polypharmacy**

**Ursula Wolf 1,\*, Henning Baust <sup>2</sup> , Rüdiger Neef <sup>3</sup> and Thomas Steinke 2,4**


**Abstract:** Antimicrobial therapy is often a life-saving medical intervention for inpatients and outpatients. Almost all medical disciplines are involved in this therapeutic procedure. Knowledge of adverse drug reactions (ADRs) and drug-drug interactions (DDIs) is important to avoid drug-related harm. Within the broad spectrum of antibiotic and antifungal therapy, most typical ADRs are known to physicians. The aim of this study was to evaluate relevant pharmacological aspects with which we are not so familiar and to provide further practical guidance. Individual pharmacotherapy management (IPM) as a synopsis of internal medicine and clinical pharmacology based on the entirety of the digital patient information with reference to drug information, guidelines, and literature research has been continuously performed for over 8 years in interdisciplinary intensive care and trauma and transplant patients. Findings from over 52,000 detailed medication analyses highlight critical ADRs and DDIs, especially in these vulnerable patients with polypharmacy. We present the most relevant ADRs and DDIs in antibiotic and antifungal pharmacology, which are less frequently considered in relation to neurologic, hemostaseologic, hematologic, endocrinologic, and cardiac complexities. Constant awareness and preventive strategies help avoid life-threatening manifestations of these inherent risks and ensure patient and drug safety in antimicrobial therapy.

**Keywords:** antimicrobial; antibiotics; antifungals; adverse drug reaction (ADR); drug-drug interaction (DDI); polypharmacy; multimorbidity; intensive care patients; traumatology; elderly patients; organ failure; multi-organ failure; drug safety; patient safety

## **1. Introduction**

Antimicrobial therapy often means a life-saving medical intervention for in-hospital and outpatients, and almost all medical disciplines are involved in this therapeutic regimen. Knowledge of adverse drug reactions (ADRs) and drug-drug interactions (DDIs) is important to avoid drug-related harm, as severe organ damage or life-threatening conditions can already occur with dosing due to drug-drug-inhibited metabolism. With respect to the expanding elderly patient population with polypharmacy, the almost rapidly increasing drug availabilities, and administration by different specialists in multimorbidity, ARDs and DDIs are turning into a major health problem worldwide. An ADR is "a response to a drug which is noxious and unintended and which occurs at doses normally used in man for prophylaxis, diagnosis, or therapy of disease or for the modification of physiologic

**Citation:** Wolf, U.; Baust, H.; Neef, R.; Steinke, T. Individual Pharmacotherapy Management (IPM)—IV: Optimized Usage of Approved Antimicrobials Addressing Under-Recognized Adverse Drug Reactions and Drug-Drug Interactions in Polypharmacy. *Antibiotics* **2022**, *11*, 1381. https://doi.org/10.3390/ antibiotics11101381

Academic Editor: Dóra Kovács

Received: 24 August 2022 Accepted: 5 October 2022 Published: 9 October 2022

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

**Copyright:** © 2022 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/).

function" (WHO 1972), which has already been defined by the WHO for 50 years [1]. A further, more clarified terminology refers to an ADR as "an appreciably harmful or unpleasant reaction resulting from an intervention related to the use of a medicinal product; adverse effects usually predict hazard from future administration and warrant prevention, or specific treatment, or alteration of the dosage regimen, or withdrawal of the product", and classical type A (augmented) and type B (bizarre) ADRs display variable patterns depending on age, sex, race, diseases, drug category, route of administration, and DDIs, as well as the individual genotypic profile, which is a further determinant for the manifestation of ADRs [2]. The classification into six ADR types (A, B, C, D, E, F) aims to differentiate dose-related (augmented), non-dose-related (bizarre), dose-related and timerelated (chronic), time-related (delayed), withdrawal (end of use), and failure of therapy (failure) ADRs. According to the directive 2001/83/EC of the European Parliament and the 2001 Council on the Community code relating to medicinal products for human use, reported in the Guideline on good pharmacovigilance practices (GVP)-Annex I (Rev 4), an ADR is currently defined as: "A response to a medicinal product which is noxious and unintended. Response in this context means that a causal relationship between a medicinal product and an adverse event is at least a reasonable possibility. An adverse reaction, in contrast to an adverse event, is characterized by the fact that a causal relationship between a medicinal product and an occurrence is suspected. For regulatory reporting purposes, if an event is spontaneously reported, even if the relationship is unknown or unstated by the healthcare professional or consumer as primary source, it meets the definition of an adverse reaction. Therefore all spontaneous reports notified by healthcare professionals or consumers are considered suspected adverse reactions, since they convey the suspicions of the primary sources, unless the primary source specifically state that they believe the event to be unrelated or that a causal relationship can be excluded. Adverse reactions may arise from use of the product within or outside the terms of the marketing authorization or from occupational exposure. Use outside the marketing authorization includes off-label use, overdose, misuse, abuse and medication errors." [3].

As there are different definitions and categorizations for ADRs, which subsume DDIs [4–6], this may further complicate the predicament in the daily routine. Rather, the ADRs being increasingly reported post-marketing with varying drug co-medications may actually represent the result of so-far unknown pharmacokinetic or pharmacodynamic DDIs. In terms of our approach to management, the distinction between an inherently drug-associated ADR and a DDI resulting from combination therapy should always be clear for guiding physicians to avoid critical combinations, in particular. Any drug can cause ADRs to potentially manifest with a variety of clinical effects. Within the broad spectrum of antibiotic and antifungal therapy, most typical ADRs are or should be known to physicians. ADRs account for a high degree of hospitalization, up to 12%, and for a high rate of morbidity and mortality, ranking among the top 10 leading causes of death and illness in developed countries with enormous socioeconomic costs [7–11], irrespective of the presumably high number of unreported cases. This applies, especially, to antibiotics and intensive care patients [12,13]. In order not to confuse ADRs and DDIs with new disease symptoms and, thus, initiate an escalating spiral of therapeutic counter-regulation, the knowledge of specific ADRs is crucial, especially in intensive care medicine and in the polypharmacy of the elderly, who are particularly vulnerable in this context because of their frequent pre-existing severe or multimorbid conditions. Way back in 1997, Rochon and Gurwitz described the risks associated with the prescribing cascade and, in this context, called for the need to optimize drug treatment in the elderly patients most affected; in addition, a further distinction has been made bet- ween prescribing cascades that are sometimes appropriate, but must be regularly re-evaluated and documented, and problematic prescribing cascades [14,15]. Misinterpretation of an ADR as a new symptom or exacerbation of the underlying disease is the beginning of the prescribing cascade, prevalent in patients who already have extensive polypharmacy and impaired organ function and are, therefore, particularly susceptible to ADRs and DDIs. Today, 25 years later, the

problem has become aggravated as the drugs available to treat an almost typical elderly multimorbid patient, e.g., with manifest hypertension, diabetes, heart failure, and COPD or urologic conditions, have become more numerous and diverse, physicians increasingly specialize, and the different disciplines do not even seem to be aware of the ADRs or DDIs of drugs from the other medical specialties. On top of this, changing organ functions of an intensively ill patient and, in addition, the complementary mechanical organ replacement procedures either temporarily or chronically influence the dosing regime and require very close monitoring of laboratory parameters in patients, often in life-threatening conditions.

Since it can be difficult to differentiate ADRs and DDIs from disease progression, for example, in the presence of pre-existing single- or multi-organ hypoxic failure, it is particularly important to focus on ADRs and DDIs from the onset of drug use, although in rare cases these may still occur even after drug discontinuation. The aim of this study was to highlight relevant pharmacological ADR and DDI aspects of antimicrobials administered in the hospital setting, which are not so widely known, and to provide further practical guidance from lessons learned based on long-term interdisciplinary experience with individual pharmacotherapy management (IPM).

#### **2. Results**

Findings from over 52,000 detailed medication analyses continuously performed for over 8 years in interdisciplinary intensive care, transplant, and trauma patients highlight critical ADR and DDI risks, especially in these vulnerable patients with polypharmacy and/or with concomitant organ deterioration. This is where our individual pharmacotherapy management (IPM) (Figure 1) has proven very reliable and valuable in identifying and targeting drug-associated risks at the earliest stage. The ability to have the most comprehensive digital view of the patient as the basis for IPM is the essential prerequisite for adjusting each drug precisely for the individual patient's medical condition as captured via metabolism and/or excretion depending on the specific mode of degradation of each drug, and always with the additional targeted focus on associated DDIs. Reading the Summary of Product Characteristics (SmPC) for each drug of an extensive drug list and, additionally, checking all pharmacokinetic and pharmacodynamic interactions between the drugs means an almost unmanageable procedure in the context of the time pressure of today's physicians' working conditions. We present the essence of often under-recognized but critical aspects of today's typical drug combinations related to antimicrobials to link the almost unmanageable set of risk aspects of ADRs and DDIs in extensive polypharmacy to everyday clinical realities and to enable the sharing of eight years of daily IPM experience for the most critical issues. We explicitly outline the relevant risks of ADRs and DDIs in the clinical pharmacology of antimicrobial agents that are less frequently addressed adequately in terms of neurologic, hemostaseologic, hematologic, endocrinologic, and cardiac complexities.

The IPM strategy, which aims to assess each drug as comprehensively as possible and which accounts for the entire patient's very acute clinical situation simultaneously, is shown in Figure 1. The first data on the preventative impact of this significantly effective IPM with a 90.2% reduction in delirium has been published previously [16]. Focusing on the entire patient condition with all of the information from the fully digitally available records proved to be a crucial and fundamental requirement for performing this extensive, individualized medication analysis, considering the respective, often abruptly changing, acute status in the ICU patients. The IPM process uncovers relevant complexities in almost all different groups of medications, and especially in polypharmacy, that have apparently been insufficiently addressed or not considered at all. Not infrequently, these manifested ADRs or DDIs were the reason for further deterioration of the clinical condition of the patient and his transfer to the ICU before application of IPM. Even therapeutic drug monitoring in antibiotics cannot guarantee drug safety, as it does not eliminate ARDs that are associated with regular dosing already or DDIs related to interacting co-medication. Despite knowledge of the often rapid alterations in renal function or metabolic rate due to

DDIs or hepatic dysfunction, especially in multimorbid elderly patients and in transplant and intensive care patients with polypharmacy, immediate concomitant dose adjustment of the antibiotic regimen to fine-tune the daily dose is required more often than it is practiced. For this purpose, and to differentiate from ADRs and DDIs, close laboratory monitoring of these organ functions is mandatory. abolic rate due to DDIs or hepatic dysfunction, especially in multimorbid elderly patients and in transplant and intensive care patients with polypharmacy, immediate concomitant dose adjustment of the antibiotic regimen to fine-tune the daily dose is required more often than it is practiced. For this purpose, and to differentiate from ADRs and DDIs, close laboratory monitoring of these organ functions is mandatory.

condition of the patient and his transfer to the ICU before application of IPM. Even therapeutic drug monitoring in antibiotics cannot guarantee drug safety, as it does not eliminate ARDs that are associated with regular dosing already or DDIs related to interacting co-medication. Despite knowledge of the often rapid alterations in renal function or met-

*Antibiotics* **2022**, *10*, x FOR PEER REVIEW 4 of 25

**Figure 1.** Comprehensive, reproducible IPM based on the hospital's digital patient record. \* Comprehensive, digitally based IPM, developed, implemented, and practiced by Ursula Wolf, MD, Head of the Pharmacotherapy Management Department, specialist in internal medicine, and expertised in clinical pharmacology, who performed > 52,000 individual medication reviews. **Figure 1.** Comprehensive, reproducible IPM based on the hospital's digital patient record. \* Comprehensive, digitally based IPM, developed, implemented, and practiced by Ursula Wolf, MD, Head of the Pharmacotherapy Management Department, specialist in internal medicine, and expertised in clinical pharmacology, who performed >52,000 individual medication reviews.

Thus, knowledge gaps about ADRs and DDIs within the outpatient medication lists or in combination with in-hospital drugs have been identified repeatedly from more than 52,000 medication reviews performed regularly as part of IPM and therapeutic drug monitoring (TDM). Their presentation in this selection serves as an informative practice support for therapeutic application. Hospitalized patients with reviewed medication included elderly patients >70 years of age from the traumatology department, all patients in the operative ICUs, in the organ and stem cell transplantation departments, and those in further disciplines within the UKH; the patients' entire ambulatory drug therapy was mostly covered. This additional IPM effort intends to identify measures to improve clinicians' recognition of these DDIs and prevent unwarranted ADRs of antimicrobial agents.

Being familiar with ADRs is essential to prevent drug-induced harm. Within the broad spectrum of antibiotic and antifungal therapy, physicians are aware of most typical ADRs. Thus, explicitly for the assessed relevant pharmacological issues in antimicrobial therapy that are less commonly considered, we provide further practical devices.

Our findings, in terms of relevant complexities in antimicrobial pharmacology that are less frequently considered, are presented in Table 1. They refer to ADRs and DDIs often recognized within a polypharmacy regimen or in patients with pre-existing organ deterioration.

#### *2.1. Anidulafungin*

Patients who experience elevated liver enzyme levels during anidulafungin therapy must be monitored for worsening liver function, and continuation of anidulafungin therapy must be subjected to a rigorous benefit–risk assessment. Hypokalemia and hyperglycemia, as well as headache and risk of convulsions, have to be considered [17].

#### *2.2. Carbapenems*

The rapid onset of extensive decline in blood levels of valproic acid when co-administered with carbapenem agents results in a 60–100% decrease in valproic acid levels in about two days [15]. This combination therapy is not manageable and must be avoided, and there are enough antiepileptic alternatives.

Neurologic/psychologic symptoms including hallucinations, drowsiness, dizziness, somnolence, and headache are ADRs to be aware of, and more rare symptoms comprise myoclonus, confusional states, seizures, paresthesias, encephalopathia, focal tremor, impaired sense of taste, and hypotension [18]. Seizures have been reported during treatment with carbapenem, including meropenem [18]. Our impression is that older patients are especially more at risk, often because of other concomitant diseases, cerebrovascular concerns, and medications that additionally increase the risk of seizures. Serum magnesium should be in the upper normal range and, if the risk is correspondingly high, such as after a stroke, concomitant medication with a similar seizure potential should be avoided as far as possible. With carbapenem, attention needs to be paid to thrombocytosis and thrombocytopenia. We observed high-grade thrombocytosis with carbapenem in the therapeutic course that started with onset, and if there was no contraindication, but risk related to thrombocytosis as in coronary patients, we temporarily applied acetylsalicylic acid (ASA) in a low dose. A differential diagnosis of thrombocytosis in antibiotic therapy remains, of course, a reactive thrombopoiesis in the recovery phase of an infection or a concomitant use of cortisone or special affections of the spleen.

#### *2.3. Cotrimoxazole and Rifampicin*

Not infrequently, hyperkalemia and hyponatremia are induced, necessitating simultaneous serum electrolyte monitoring.

Rifampicin decreases the effect of all currently available direct oral anticoagulants (DOACs) by inducing CYP3A4 and p-glycoprotein (P-gp); therefore, concomitant use is contraindicated. The same applies to rifampicin with calcineurin inhibitors and mammalian target of rapamycin inhibitors (mTORIs), where a dose elevation of the immunosuppressants must be assured under close TDM to prevent transplant rejection.







Hepatic dysfunction

*Antibiotics* **2022**, *11*, 1381


69

*Antibiotics* **2022**, *11*, 1381






Typically, an inducer effect lasts a few days after discontinuation of the drug, and also manifests itself only maximally after a few days from the start of therapy in contrast to inhibitory metabolic disorders. This always has to be considered, for example, in the timing of TDM of immunosuppressants for contemporary dose adaptation [19].

#### *2.4. Daptomycin*

Daptomycin can cause anxiety, insomnia, dizziness, and headache. Use may result in false prolongation of prothrombin time (PT) and elevation of the international normalized ratio (INR) with certain recombinant thromboplastin reagents. Other drugs associated with myopathy, such as statins, should be temporarily discontinued; creatine phosphokinase (CPK) levels must be measured at baseline and at regular intervals. Eosinophilic pneumonia is a potential side effect of daptomycin. Daptomycin is one of the still-rare group of drugs whose bioavailability has been studied in relation to obesity. Obesity increases the area under the curve (AUC0-∞) of daptomycin [20].

#### *2.5. Fluoroquinolones*

Simultaneous administration with tube feeds is not recommended due to the minerals, and explicitly, divalent cations. There are interactions with antacids, iron, zinc, magnesium, sucralfate, calcium, didanosine, oral nutritional solutions, and dairy products. Ciprofloxacin should be administered 2 h before or at least 4 h after these products [21].

Systemic and inhaled fluoroquinolones may contribute to an increased risk of aneurysm, aortic dissection, and tendon rupture, but also of valvular regurgitation/insufficiency. A careful risk–benefit assessment is required, and other therapeutic options should be considered in patients at increased risk due to predisposing factors such as existing valvular heart disease, endocarditis, connective tissue diseases, hypertension, rheumatoid arthritis, and infectious arthritis [22].

#### *2.6. Linezolid*

Clinical use of linezolid in co-medication with serotonergic agents such as the antidepressant group of selective serotonin reuptake inhibitors (SSRIs) and serotonergic opioids is contraindicated because of the risk of serotonin syndrome. Linezolid is an antibiotic agent with nonspecific MAO-inhibitory activity and a predisposing risk for malignant neuroleptic syndrome. Therefore, it is important to know that excessive amounts of foods and beverages high in tyramine should be avoided, given the potential for significant pressure reactions [Lit]. A risk of myelosuppression (including anemia, leucopenia, pancytopenia, and thrombocytopenia) and lactate acidosis as ADRs needs to be considered with linezolid [23].

#### *2.7. Similarities of Macrolides and Azoles concerning Severe DDIs*

The combination of most macrolides or azoles with the widespread use of statins must be viewed very critically [24]. HMG-CoA reductase inhibitor (statin) ADRs such as myositis and rhabdomyolysis are often dose dependent. The bioavailability of the statins atorvastatin, lovastatin, and simvastatin, when co-administered with CYP3A44 inhibiting antimicrobials, increases, as do their ADRs. In this context, pravastatin, rosuvastatin, and fluvastatin are more inert due to their different metabolic and elimination pathways.

With macrolides or azoles, such as erythromycin, ketoconazole, or voriconazole, rhabdomyolysis and severe hepatic dysfunction have been observed to occur suddenly, especially with propofol plus amiodarone, which is also due to the marked, opposing CY3A4 inhibition of metabolism. This is essential to know because amiodarone itself can induce tachyarrhythmias, which, at the same time, evolve from the QT prolongating potential risk of most macrolides and azoles. These combinations aggravate and prolong the dangerous situation of the metabolism and toxicity of any drug involved via CYP3A4 metabolism and/or P-gp transport. Increasing hepatic dysfunction and even organ failure can rapidly escalate under these circumstances as the drugs become increasingly unmetabolized and

cumulatively manifest their own hepatotoxic risks. These ADRs can accumulate into multi-organ failure, particularly in the presence of pre-existing renal failure. However, hepatic failure can be reversible with timely targeted drug discontinuation. Withdrawal of amiodarone often results in the timely resolution of the sometimes life-threatening problem. With regard to rhabdomyolysis, to prevent renal failure, we would like to point out that besides drug withdrawal or dose adjustment, the earliest administration of bicarbonate is recommended, always with reference to the simultaneously monitoring of adequate respiratory capacity, the partial pressure of carbon dioxide (pCO2), and the acid–base balance.

Fentanyl and buprenorphine must be withdrawn with the onset of macrolides or azoles because of the risk of, among other consequences, respiratory depressant effects due to decreased metabolism [25].

All currently approved DOACs are contraindicated with most macrolides (notice the metabolic properties of three different groups) and azoles because they increase the risk of hemorrhages due to the decreased DOAC degradation by inhibited P-gp and/or CYP3A4 [26]. The same applies to ticagrelor metabolism [27], which is also a substrate of P-gp and CYP3A4, a fact that is often not taken into account postinterventionally in cardiology patients and is further aggravated by concomitant hepatic dysfunction with reduced metabolic capacity, e.g., related to right-sided heart failure.

Both macrolides and azoles interact with several antiviral agents; in this context, and with respect to its current frequent use in the contemporary COVID-19 pandemic, we refer to remdesivir. Although remdesivir is a substrate of the metabolizing enzymes CYP2C8, CYP2D6, and CYP3A4, as well as a substrate of the transporters for organic aniontransporting polypeptide 1B1 (OATP1B1) and P-gp, no clinical interaction studies have been performed with remdesivir [28]. Concomitant administration of potent inhibitors such as azoles and macrolides may result in intolerably increased remdesivir exposure with all its ADRs. Therefore, the combination has to be avoided. Administration of potent P-gp and CYP3A4 inducers such as rifampicin and cotrimoxazole potentially decrease plasma concentrations of remdesivir and, thus, resulting in a reduction or loss of efficacy; therefore, it cannot be administered in combination. The guidance framework is still imprecise here and it is incomprehensible that a substance approved to be used in COVID-19 therapy worldwide is so understudied for interactions, as polypharmacy is often unavoidable in these patients especially.

It is of great importance to consider the same DDIs of most azoles and most macrolides, or rifampicin and cotrimoxazole, respectively, in analogous terms for nirmatrelvir/ritonavir [29]. Nirmatrelvir/ritonavir should not be started immediately after discontinuation of the antibiotics fusidic acid, rifampicin, and cotrimoxazole. This remains critical because it should be administered as early as possible with the onset of COVID-19 symptoms. Precaution is required with regard to inducer and inhibitor effects with other drugs, as these DDIs may lead to clinically significant ADRs, potentially resulting in severe, life-threatening events from greater exposures of nirmatrelvir/ritonavir due to the concomitant-inhibiting medication and vice versa; with inducers comes a loss of the therapeutic effect of nirmatrelvir/ritonavir and the possible development of viral resistance. Nirmatrelvir/ritonavir is a strong inhibitor of p-GP and CYP3A4. To give an impression of these risks, investigations with rifabutin reveal a 4-fold–2.5-fold increase, whereas the metabolite 25-O-desacetyl rifabutin has a 38-fold–16-fold increase. This large increase in the rifabutin area under the curve (AUC) requires the consequent reduction of the rifabutin dose to 150 mg 3 times per week when co-administered with ritonavir as a strong inhibitor. The ritonavir-induced, elevated AUC change of ketoconazole was 3.4-fold [29].

#### *2.8. Posaconazole*

Posaconazole is an azole commonly used in stem cell transplantation. Therefore, it is important to note that its inherent risk of major gastrointestinal side effects [30] with severe diarrhea may mimic or overlap CMV colitis or graft-versus-host disease, or symptomatically mask the success of their therapies and, therefore, must always be considered in differential diagnosis, especially in this accordingly most critical patient population.

#### *2.9. Rifampicin*

As a CYP3A4 and P-gp inducer, rifampicin, conversely, can significantly reduce the availability of CYP3A4 and P-gp substrates and, thus, their efficacy, such as in calcineurin inhibitors and mTORIs; therefore TDM and dose adjustments are required [31]. For these inducing effects, the combination with DOACs is contraindicated. There is also a wide range of DDIs resulting from inducing effects of rifampicin, e.g., with virustatics, antiepileptic drugs, and opioids, that reduce or lose their effects.

Similarly, for *metamizole*, a very common postoperative analgesic in Germany, a potential 2.9-fold hepatocyte CYP3A4 induction has been described [32], whereby a potential reduction, e.g., in the efficacy of various macrolides or azoles, cannot be excluded. We also regularly have to counteract it through dosage elevation in co-administered calcineurin inhibitors and mTORIs. Rifampicin is contraindicated in the presence of jaundice and hepatic dysfunction. Its enzyme induction can enhance the metabolism of endogenous substrates including adrenal hormones, thyroid hormones, and vitamin D. Additionally, the reported incidence of a paradoxical drug reaction ranges between 9.2 and 25%. In these cases, after the initial improvement of the tuberculosis during therapy with rifampicin, the symptoms may worsen again. An excessive immune reaction is suspected as a possible cause, which can be treated symptomatically while continuing antituberculostatic therapy [33].

#### *2.10. Tigecycline*

Tigecycline is known to possibly induce hyperbilirubinemia and prolonged PTT with the elevated risk of bleeding [34]. Especially in the case of concomitant continuous intravenous heparin administration, the latter often has to be adjusted according to the PTT values that need to be monitored. Based on an in vitro study, tigecycline is a P-gp substrate [34]. Co-administration of P-gp inhibitors (e.g., ketoconazole or cyclosporine) affect the pharmacokinetics of tigecycline and increase the bioavailability, including its ADRs. Concomitant P-gp inducers (e.g., rifampicin) may reduce the antibiotic effect [31]. The drug may increase serum concentrations of calcineurin inhibitors, necessitating close TDM. Tigecycline requires close monitoring for the development of superinfection, hyperbilirubinemia, hepatic injury, and pancreatitis [34].

#### *2.11. Vancomycin*

Thrombocytopenia is not infrequently seen with vancomycin. It is worth mentioning a reference to obese patients, whom we increasingly have to consider. Since an adjustment of the usual daily doses of vancomycin may be necessary in obese patients [35], TDM is appropriately helpful in this situation with regular level measurements because of its nephrotoxicity and further risks. Rapid bolus administration (i.e., over several minutes) may be associated with severe hypotension (including shock and rare cardiac arrest), histamine-like responses and maculopapular or erythematous rash ("red man's syndrome" or "red neck syndrome"), and pain and spasm of the chest and spine muscles [35].

#### *2.12. Voriconazole*

The preclinical repeated-dose toxicity studies of the intravenous vehicle sodium betacyclodextrin sulfobutyl ether (SBECD) revealed vacuolization of the urinary tract epithelium and activation of macrophages in the liver and lungs as the main effects [36]. Carcinogenicity studies have not been performed with SBECD, although an SBECD-containing impurity has been shown to be a mutagenic substance with demonstrated carcinogenic potential in rodents; thus, this compound must also be considered to have carcinogenic potential in humans [36].

#### *2.13. Probenecid*

When used concomitantly with probenecid due to its slower excretion, the plasma levels, effect, and ADRs of several antimicrobials may be increased, and dose adjustments may be necessary in penicillin, cephalosporin, quinolone (e.g., ciprofloxacin, norfloxacin), dapsone, sulfonamides, nitrofurantoin, nalidixic acid, and rifampicin [34]. Probenecid is contraindicated in simultaneous treatment with β-lactam-antibiotics and with pre-existing renal impairment [37].

#### *2.14. Proton Pump Inhibitors*

Proton pump inhibitors are known to be associated with an elevated risk of bacterial infections [38]. They should be applied more restrictively, and for stress ulcer prevention, be reduced to the adequate prophylactic dosage of, e.g., daily 20 mg of pantoprazole only.

#### *2.15. Benefits and Preliminary Effects of the Applied IPM Strategy*

We have elaborated the domain aspects extracted and tabulated for antimicrobial ADRs and DDIs and are continuously focusing on their application in the IPM.

The unavailability despite physicians' requests for such compilations attests to the clinical weight of this overview of risks. Studying the innumerable ADRs and DDIs from SmPCs and DDI checks takes time, which is not available in everyday clinical practice. Additionally, since countless ADRs and DDIs have emerged, especially within the broad medication list for the critically ill patients in the ICU and for polypharmacy in elderly traumatology patients who require additional antimicrobial therapy, which are impossible to comprehensively research in everyday life, we provide our insights resulting from a synoptic point of view of internal medicine and clinical pharmacology based on the daily performance of 35 to 52 IPM in an everyday real life condition by this tabulated overview. Applying this focus as one elementary building block among others of the comprehensive IPM enables us to perform a fast medication analysis in 6.5 min, whereas the extensive procedure otherwise is almost unmanageable due to the particularly broad medication lists for each patient within these settings and may take 1 to 2 h.

In addition to this benefit, the strategy employed, including maintained awareness of these knowledgeable, selective ADRs and DDIs and applying this information to each individual patient, also may contribute to the apparently highly effective IPM measures. There are compelling preliminary data suggesting that this preventive IPM strategy outlined here, documenting our IPM attention and strategic aspects, including the broad component of antimicrobials, contributes to risk reduction and optimized medication therapy in patients most vulnerable, such as those with unavoidable polypharmacy who suffer from concomitant organ dysfunction already.

Based on the annually documented reports of the controlling department of the UKH, the first preliminary descriptive results, which have to be statistically analyzed further on for corresponding publications, document that coinciding with the onset of IPM implementation for the operative ICU patients most critically ill, and for whom the ICU was the treatment ward in which thy had the longest length of stay at the UKH, the average length of stay decreased from 29.1 to 23.2 days from 2015 to 2018. In parallel, the number of patients treated even increased, e.g., by 32.7% in the ICU ward, where critically ill patients are often mechanically ventilated and dialyzed. Despite higher numbers of patients treated, medication costs concurrently decreased by 27.9% for the operative ICU wards. These results will be published separately for the entire evaluation of the IPM implementation after further analytical statistics are performed.

Regarding the department of geriatric traumatology, we retrospectively analyzed the association of the overall implemented IPM elements, including the exclusively presented, specific antimicrobial strategies with an ADR and DDI focus on prevention of complicating delirium, in-hospital fall events, and impaired renal function within a subset of 404 patients in a group-matched study. There was an IPM-associated 90.2% relative reduction in

complicating delirium [16] and almost analogous results for the reduction in in-hospital fall events and in renal function impairment (for separate publication).

#### **3. Discussion**

IPM in the context of antimicrobial therapy reveals several clinical important aspects to consider. Most of them should be familiar as more or less typical ADRs and, thus, have not been presented here repeatedly. The selected insights into lesser-known ADRs and essential DDIs result from clinical IPM practice and observations in our patient treatments, and here, represent only excerpts from the respective ADRs and DDIs. A further number of unidentified ADRs must be assumed, since they do not always manifest in their fulminant form or a complete spectrum, but also in a crytogenic or undulating manner. This incompleteness may also apply, in part, to the identified and listed ADRs referred to in the drug product information.

Therefore, there is almost no way to definitively determine a uniform general rate of ADRs for a given drug, but the more vulnerable the patient is due to his pre-existing conditions and illnesses, organ dysfunction, and polypharmacy, the higher the risk of ADR and DDI manifestation. For this reason, post-marketing evaluations, documentations, and studies are very important because the pre-marketing trial phase of the approval studies almost never include these particularly susceptible patients.

Drugs and, especially, antibiotics can rapidly worsen an already developing single- or multi-organ failure, e.g., as a result of pre-existing hypoxia, and require special therapeutic attention with fine-tuning of dose adjustments. The essential knowledge and consideration of drug-specific ADRs, particularly in frequent antimicrobial combination therapy, is a preventive tool in patient and drug safety, both of which are continuously serious concerns as repeatedly claimed by the WHO for over a half century already [1]. A decline in renal function, and hepatic and further organ deterioration up to multi-organ failure always must exclude pharmacological iatrogenic drug-associated causation. This presentation is a sub-aspect of extensive studies of our IPM demonstrating IPM-associated optimization of medication, e.g., effectively preventing delirium by 90.2% in elderly trauma patients [16].

As ADRs are categorized inconsistently and DDIs are subsumed, this makes the important differentiation of these two risky issues less clear, and thus, the preventative, necessary countermeasures might be less appropriate or even missed.

In addition, we must take into account the most frequently prevalent hypoalbuminemia, especially in our demographically increasing elderly patient population, as well as in critically ill intensive care patients. This aspect remains almost neglected and should be focused on more intensely because, besides the predominant, clinically significant pharmacokinetic interactions observed at the metabolic level, another large proportion of competing interactions and ADR manifestations must be considered at the serum proteinbinding level [39]. This further concern has not been addressed within the presented aspects. Nevertheless, we must pay additional attention to it, especially in view of the extremely common manifestation of hypoalbuminemia in our elderly and ICU patients. This can lead to severe and more rapid overexposures of high protein-binding drugs, including their ADRs. The dilemma is further exacerbated by simultaneously competing efforts of different drugs to bind to the remaining serum protein residual capacity.

It is worthwhile to depict that DDIs are of such high clinical relevance, both on the metabolic and protein-binding level, that they are even targeted in drug design and development by pharmaceutical companies. This is found with the combined therapeutic use of nirmatrelvir/ritonavir [29], and also with imipenem plus cilastin [40] and with taxanes; here, the albumin-bound nanoparticle formulation of paclitaxel and nab-paclitaxel exhibits enhanced paclitaxel tissue distribution and enhanced tumor penetration through additional active, selective transport into tumor tissue via target proteins [41,42].

Further IPM efforts are intended for communicating interventions that improve the physicians' recognition of these DDIs and ADRs of antimicrobials. With this in mind, we have outlined the detailed process of effective IPM enabling the early detection of ADRs and DDIs and provide the critical basis for causal, preventive medical countermeasures aimed at successful and safe antimicrobial therapy, even in the most vulnerable risk patients.

Knowledge of documented ADRs and DDIs enable early differentiation from further disease progression, which, if unrecognized, would typically and inevitably lead to additive and accumulative drug therapy to treat these unconscious iatrogenic symptoms, and thus, might be the beginning of a fatal course for the patient. The implementation of the comprehensive IPM has been found to be highly preventive of several complications, as was just published for delirium [16], and thus, IPM guarantees an important step forward in eliminating severe drug and patient safety concerns. In this context, antimicrobials are also known to potentially directly induce iatrogenic delirium, e.g., from anticholinergic properties in ampicillin, clindamycin, and gentamycin or from the dopaminergic and serotonergic agent linezolid and hyponatremia-inducing antibiotics such as cotrimoxazole [43].

In 1975, in their critical review on ADRs, Karch and Lasagna already stated, "The data on adverse drug reactions (ADRs) are incomplete, unrepresentative, uncontrolled, and lacking in operational criteria for identifying ADRs. No quantitative conclusions can be drawn from the reported data as for morbidity, mortality, or the underlying causes of ADRs, and attempts to extrapolate the available data to the general population would be invalid and perhaps misleading" [44]. There have been no sufficiently successful attempts to overcome this serious grievance for nearly 50 years now, although it is becoming an increasingly threatening problem worldwide [45] with an inherent, extreme socioeconomic healthcare burden. Since 1998, the U.S. Food and Drug Administration has been operating the Adverse Event Reporting System, collecting all voluntary reports of adverse drug events submitted directly to the agency or through drug manufacturers. These data show a significant increase in reported deaths and serious injuries related to drug therapy during the observation period from 1998 to 2005. Within a 7-year period, the reported number of serious adverse drug events increased 2.6-fold, from 34,966 to 89,842, and the number of fatal adverse drug events increased 2.7-fold, from 5519 to 15,107 [46]. Davies et al. documented patients with adverse drug events in the hospital that had a longer length of stay and were more likely to be older, female, and taking a higher number of medications. The latter was the only significant predictor of an ADR in the multivariable analysis, with each additional medication increasing the risk of an episode of ADR by 1.14 (95% CI 1.09, 1.20) [47]. A causality and preventability assessment of adverse drug events of antibiotics among inpatients by Saqib et al. revealed 59.3% were preventable and caused by medication errors due to the nonadherence of policies (38.4%) and lack of information about antibiotics (32%) [12].

The inclusion of ADRs and DDIs as a subset of the International Classification of Diseases (ICD) is an important step to focus on these often serious to life-threatening concerns and weighting them on the level of diseases. This should be addressed more progressively. With 28.4%, Joshua et al. noted a high incidence of serious, even lethal, multiple ADRs in patients admitted to the medical ICU. Similar to our observations, they reported a broad clinical spectrum of ADRs and found infrequently documented ADRs from newer drugs. Antimicrobials (27%) were the commonly involved drugs in these ADRsusceptible patients with pre-existing multi-organ dysfunction and altered pharmacokinetic parameters [13]. In line with our IPM observations, ADRs were significantly associated with comorbidity, polypharmacy, and length of stay [13]. Severe coagulation problems and hemorrhagia, as they may occur from consumptive coagulopathy and sepsis, may also result from the ADRs and DDIs partly listed in our table. The very comprehensive IPM is a helpful tool to differentiate them more easily, particularly through its daily performance and through close patient follow-ups at the most acute, individual level.

The frequent QT prolongation risk of numerous drugs, which is often aggravated by concomitant medication in a cumulative manner, has to be regarded and is not always avoidable. Because most of the risks should be familiar, we left them out of consideration in this presentation, although the clinical consequences may become most critical. As a preventive measure, serum potassium and magnesium should be kept in an upper normal

range and monitored closely, and an acidotic metabolic state should be avoided. Macrolides and azoles, most of which are highly potent inhibitors and pharmacokinetic enhancers, are antimicrobials that are important to a keep in mind, as is fluorchinolone in this context. Due to their transporter- and metabolism-inhibiting capacities, azoles and macrolides even cumulate the tachyarrhythmia risks to a higher degree than merely additively with co-administered QT prolonging drugs, which are respective substrates to the inhibited enzymes. The classification of macrolides into three different groups according to their affinity for CYP3A4, and thus, their propensity to cause pharmacokinetic drug interactions, may help to clinically estimate the severity of potentially resulting DDIs. Group 1 with troleandomycin, erythromycin, and its prodrugs decrease drug metabolism and may engender drug interactions, whereas group 2 with clarithromycin, flurithromycin, midecamycin, midecamycin acetate, josamycin, and roxithromycin should cause less frequent intense interactions [48], although according to our TDM routine data, clarithromycin is associated with a significant elevation in cyclosporine when co-administered, necessitating TDM and the timely dose adjustment of cyclosporine [49]. Additionally, the study of Hill et al. indicates that in elderly patients taking a DOAC, concomitant clarithromycin was associated with a small but statistically significant greater 30-day risk of hospital admission with major hemorrhage compared to azithromycin, a group 3 macrolide [50].

Since CYP450 enzyme generation is downregulated by elevated cytokine levels, such as from IL-1 in chronic inflammation, this would be expected to be renormalized by an IL-1 receptor antagonist such as anakinra. The clinical relevance for CYP450 substrates with a narrow therapeutic range that may be affected currently remains unknown [51]. A further aspect of interindividual varying metabolization and drug effects from mitochondrial genetics addressing personalized medicine and drug toxicity has been reviewed by Penman et al. To date, the implications remain unclear, although clinical studies have reported associations between mitochondrial haplogroup and antiretroviral therapy, chemotherapy, and antibiotic-induced toxicity. Mitochondrial DNA differences may reveal differential functions as a factor in idiosyncrasies, leading to unpredictable adverse effects and druginduced toxicities [52]. The effect of polymorphism of the ABCB1 gene encoding P-gp was assessed by a French group for the substrates rivaroxaban and dabigatran by studying the contribution of ABCB1 genetic polymorphisms, as well as the interaction with clarithromycin, to interindividual variability under dabigatran and rivaroxaban exposure [53]. Whereas the ABCB1 genotype turned out not to be a clinically relevant determinant of both drugs' pharmacokinetics, the co-administration of the P-gp inhibitor clarithromycin with dabigatran or rivaroxaban resulted in a clinically relevant two-fold increase in both drugs' AUC, irrespective of the ABCB1 genotype clearly indicating the risk of DDIs [53]. With regard to almost extremely frequently administered drugs such as atorvastatin, we even find gaps in the knowledge and unawareness of significant DDIs resulting from antimicrobials. E.g., even clarithromycin (a macrolide group 2) induces a 4,5-fold increase in atorvastatin's AUC, and for itraconazole, a 3.3-fold elevated AUC with standard dosages [54]. Given that the dual-interaction mechanism of rifampicin as an inducer of CYP3A4 and an inhibitor of the hepatocellular uptake transporter OATP1B1 is difficult to sum up already, there is even a further varying effect depending on the time of application, with a 5-fold decrease in atorvastatin when it is co-administered with rifampicin simultaneously [54].

ADRs and DDIs, as documented in antimicrobial therapy from our IPM insights, are an increasing challenge in today's healthcare worldwide, especially given the growing complexity of chronic and acute combination therapy interventions, interdisciplinary treatments by different specialists, an aging population with increasing multimorbidity, and additionally, disease interactions from renal impairment or hepatic dysfunction. An alarming deficit of ADR and DDI awareness is burdening the patients, the hospitals, the entire healthcare sector, and the public budget. Urgent intervention strategies for prevention are to be promoted and anchored politically. This may be achieved via the development of increasingly more effective digital medication tools and making institutions of specialized

drug safety managers mandatory in hospitals and health insurance companies to cover the inpatient and outpatient care sectors.

#### **4. Methods**

This study from ongoing, real-life observations was designed to summarize and tabulate the under-recognized, clinically most critical risks of ADRs and DDIs in the context of current medication regimens and in the setting of elderly, multimorbid, or even seriously ill hospitalized patients. Critical, but less recognized, ADR and DDI risks from daily 35–52 medication reviews were selected through comprehensive individual pharmacotherapy management (IPM) in these vulnerable operative intensive care, elderly traumatology, and organ and stem cell transplant patients. The underlying observation period was from 12/2014 to 8/2022. We examined the drug combinations found in current, real-life practice based on medication reviews of >52,000 IPMs. As a consequent part of the study, we also provide examples of our own practice-proven advice on control and countermeasures.

IPM was designed and implemented by a University Hospital Halle (UKH) physician with expertise in internal medicine and clinical pharmacology, and who is responsible for the pharmacotherapy management at the UKH in continuous cooperation with senior physicians of the operative intensive care units (ICUs) and geriatric traumatology. The focus is on patient and drug safety, especially in these most vulnerable groups with polypharmacy, often additionally associated with impaired organ function, as in interdisciplinary intensive care patients, organ or stem cell transplanted patients, and elderly trauma patients. IPM considers the entire digital patient record from the internal clinical information systems for data management, the Integrated Care Manager (ICM) and Orbis software for advice on clinical documentation. Always in synopsis internal medicine/clinical pharmacology, the comprehensive medication review has been performed by the same study physician for each patient daily in the operative ICUs and fortnightly for interdisciplinary patient visitations in geriatric traumatology. The review was conducted primarily based on the Summary of Product Characteristics (SmPC), particularly considering pharmacokinetics, pharmacodynamics, dosage, contraindications, interactions, and all ADRs. In addition, the literature research was carried out to answer extended or unresolved questions. All diagnoses and indications were recorded as the basis for each individual medication review. Further on, current medical guidelines were taken into account.

Recommendations result from the entire consideration of the individual patient's situation, both regarding his chronic and current clinically relevant disease process, and the acute pre- and postoperative situation, considering all available laboratory medical data on organ functions; albumin; lactate and inflammation parameters; electrocardiography and imaging examination results; the continuous course of vital parameters (blood pressure, heart rate, respiratory function, and acid–base balance); body weight; BMI; cognitive disorders; other subjective complaints; pain intensity/profile; the results of assessments and risk scores; demographic data such as age, gender, and living situation; anamnesis/external anamnesis; and the course of clinical examination findings. There is always an additional focus on anticholinergic components and electrolyte disturbances, glucose metabolism, vitamin D, parathormone balance and endocrinological thyroid functions, coagulation parameters, transient organ replacement therapies, and manifest/potential hematologic disorders, in particular, on all forms of anemia. This comprehensive digital overall view of a patient represents the basis for the internistic/clinical pharmacologic intervention with the simultaneous synoptic and adaptive examination of all medications.

By the consequent implementation of this regularly systematic individual pharmacotherapy management, the focus is on prevention and elimination of iatrogenic medicationassociated: 1. severe hypotension from cumulative drug effects, 2. renal injury, 3. single and multiple organ failure, 4. neurologic/psychiatric disorders including delirium, 5. in-hospital fall events, 6. arrhythmias, 7. hemorrhages from hemostaseologic disorders, 8. cerebrocardiovascular events, 9. hematologic/myeloproliferative disorders, 10. all types

of differentiated anemias, 11. venous thrombosis, 12. acid–base and electrolyte imbalance, 13. metabolic/endocrinologic disorders, 14. gastrointestinal effects such as paralytic ileus and ulcer of the upper intestine, 15. multidrug-resistant nosocomial infections, and 16. oropharyngeal dysphagia.

In this context, during the 8 year follow-up of IPM we recognized risks of ADRs and DDIs resulting from antimicrobials, which are of predominant clinical relevance concerning the regarded aspects. As we obviously work with a similar time pressure as most physicians nowadays do, we, exactly for this reason, sum up and compile into a table for overview the most critical risks resulting from our clinical insights of the comprehensive IPM, that always adapts drug information to the individual and acute patient clinical condition.

The findings presented here are a specific sub-focus of lesser referenced ADRs and relevant DDIs associated with the concomitant administration of antimicrobials in the context of our IPM. We explicitly exclude the more familiar nephrotoxic risks, QT prolongation, and myelosuppression as the most typical ADRs of several antimicrobial agents. We covered the relevant pharmacological aspects with which we are not so familiar to provide additionally corresponding further practical guidance through IPM.

#### **5. Strengths and Weaknesses**

The analysis refers to continuous observations on selected wards of the UKH. Nevertheless, the aspects addressed may be representative for the entire patient population. IPM is an individual medication review that is conducted as intensively as possible. Genetic pharmacological aspects are not considered additionally. We did not measure blood values that attribute adverse effects to presumably iatrogenic medical agents due to elevated levels. Only from the time of occurrence during application and reversal of the critical situation by selective drug withdrawal could a possible causal relationship be postulated. However, the causal association remains difficult to definitively prove due to numerous pathophysiological disturbances often occurring in parallel in our severely ill patients. The more than 8 years of daily IPM experience with a critically ill or multimorbid elderly interdisciplinary patient population covers a particularly vulnerable group with polypharmacy and, probably, already reduced organ function. There is almost no possibility to clearly identify the entire frequency of ADR manifestation, but the more vulnerable the patient is as a result of pre-existing illnesses, organ deterioration, and polypharmacy, the higher the ADR risk. The spectrum of ADRs and DDIs addressed is based primarily on the SmPCs, whose completeness and, in rare cases, even the presentation of clearly defined metabolic pathways of active substances and risks from inactive metabolites are not always fulfilled, thus leaving out still unknown grey areas, which may be additionally dangerous for the treated patients.

#### **6. Conclusions and Outlook**

IPM-based awareness of serious but lesser-known ADRs and DDIs of antimicrobial agents can counteract or reverse single- or multi-organ failure, cardiac arrhythmias, life-threatening hemorrhage or thrombosis, hematologic/myeloproliferative disorders, electrolyte imbalances, neurologic/psychiatric disorders, and drug escalations by enabling timely and targeted correction of the underlying causation individually. The conceptionalized IPM strategy applied to optimize drug treatment and prevent ADRs and DDIs, both common risks of polypharmacy, is accompanied by a reduction in length of hospital stay and associated costs. We plan to implement IPM's longstanding experience and lessons learned for wider use on a digitized platform to share a comprehensive compass, in addition to the documented findings on antimicrobials, as an important contribution to improving patient and drug safety collectively when facing the high-risk issue of under-recognized ADRs and DDIs worldwide.

**Author Contributions:** Conceptualization, U.W.; formal analysis, U.W.; investigation, U.W., H.B., R.N. and T.S.; methodology, U.W., H.B., R.N. and T.S.; project administration, U.W., H.B., R.N. and T.S.; validation, U.W.; writing—original draft, U.W.; writing—review and editing, U.W., H.B., R.N. and T.S. Additionally, U.W. conceptionalized and performed the IPM. T.S., H.B. and R.N. put forward the implementation of digital records for patient management. T.S, H.B. and R.N. introduced IPM into interdisciplinary patient care within their departments and have provided continuous and intense collaboration ever since. All authors have read and agreed to the published version of the manuscript.

**Funding:** We acknowledge the financial support from the funding program, Open Access Publishing by the German Research Foundation (DFG).

**Institutional Review Board Statement:** We confirm that all methods were performed in accordance with the relevant guidelines and regulations, such as the ethical standards of the institutional ethics committee and with the 1964 Helsinki Declaration and its subsequent amendments. Ethical review and approval were waived for this study, as no ethical concerns were raised with regard to the anonymized data collection and nonperson presentation. Patient interests worthy of protection are not affected by the completely anonymized data obtained for clinical purposes from routine care.

**Informed Consent Statement:** Requirement for informed consent was waived for the entirely nonperson-related anonymized evaluation and presentation.

**Data Availability Statement:** There was no data collection for the purpose of this article, only observational results and recommendations.

**Acknowledgments:** The authors gratefully acknowledge all team physicians who reliably collaborate with IPM in trauma, transplant, and intensive care medicine, whether during patient rounds or daily telemedicine recommendations. The personal commitment and dedication of every individual physician is what contributes to the highly effective IPM, and the pleasant collegial cooperation on an equal footing promotes patient and drug safety at UKH.

**Conflicts of Interest:** H.B., R.N., and T.S. declare no conflict of interest. U.W. received honoraria for scientific lectures on risks of polypharmacy from Bristol Myers Squibb and Pfizer.

#### **References**


## *Article* **Do Different Sutures with Triclosan Have Different Antimicrobial Activities? A Pharmacodynamic Approach**

**Frederic C. Daoud 1,\* , Fatima M'Zali <sup>2</sup> , Arnaud Zabala <sup>2</sup> , Nicholas Moore <sup>1</sup> and Anne-Marie Rogues 1,3**


**Abstract:** (1) Background: Three antimicrobial absorbable sutures have different triclosan (TS) loads, triclosan release kinetics and hydrolysis times. This in vitro study aims to analyse and compare their antimicrobial pharmacodynamics. (2) Methods: Time-kill assays were performed with eight triclosan-susceptible microorganisms common in surgical site infections (SSIs) and a segment of each TS. Microbial concentrations were measured at T0, T4, T8 and T24 h. Similar non-triclosan sutures (NTS) were used as controls. Microbial concentrations were plotted and analysed with panel analysis. They were predicted over time with a double-exponential model and four parameters fitted to each TS × microorganism combination. (3) Results: The microbial concentration was associated with the triclosan presence, timeslot and microorganism. It was not associated with the suture material. All combinations shared a common pattern with an early steep concentration reduction from baseline to 4–8 h, followed by a concentration up to a 24-h plateau in most cases with a mild concentration increase. (4) Conclusions: Microorganisms seem to be predominantly killed by contact or near-contact killing with the suture rather than the triclosan concentration in the culture medium. No significant in vitro antimicrobial pharmacodynamic difference between the three TS is identified. Triclosan can reduce the suture microbial colonisation and SSI risk.

**Keywords:** suture; antimicrobial; pharmacodynamics; triclosan; surgical site infection; time-kill; contact killing; translational modelling

#### **1. Introduction**

There is a broad array of surgical wound closure methods, including thousands of suture types, staples and surgical adhesives [1]. Sutureless surgery is also being developed in various fields, including maxillofacial and cardiac surgery [2–5]. Minimising the risk of surgical site infection (SSI) is an important consideration when developing surgical wound closure techniques.

Triclosan is a synthetic, hydrophobic bisphenol (5-Chloro-2-(2,4-dichlorophenoxy) phenol) [6]. It is solid below 54 ◦C and displays low solubility in water (10 µg/mL in pure water at 20 ◦C) compared to nonpolar solvents such as olive oil (approximately 600,000 µg/mL) or ethanol (>1 million µg/mL) [7]. Hydrolysis and photodegradation are the two main triclosan degradation pathways. Both are too slow to be measurable over 24 h, providing the assays are protected from intense light. Triclosan has several properties that make it a broad-spectrum antimicrobial, especially its nonpolarity, which brings triclosan molecules together, among other nonpolar substances such as phospholipids influences bacterial cell membranes [8–11]. Triclosan lipid-membranotropism facilitates its concentration inside cell phospholipids and the membranes of gram-positive cocci (mix of peptidoglycan and phospholipids) and gram-negative bacteria (predominantly phospholipids). It also partially explains triclosan's lower ability to penetrate the outer walls

**Citation:** Daoud, F.C.; M'Zali, F.; Zabala, A.; Moore, N.; Rogues, A.-M. Do Different Sutures with Triclosan Have Different Antimicrobial Activities? A Pharmacodynamic Approach. *Antibiotics* **2022**, *11*, 1195. https://doi.org/10.3390/ antibiotics11091195

Academic Editor: Dóra Kovács

Received: 26 July 2022 Accepted: 30 August 2022 Published: 3 September 2022

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

**Copyright:** © 2022 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/).

of *C. albicans*, predominantly consisting of polysaccharides [12]. Triclosan's antimicrobial activity has multiple targets, but the main one is reported to be NADH-dependent enoyl- [acyl carrier protein] reductase (FabI). This inhibits cell membrane fatty acid synthesis, thus disrupting membranes [13,14].

Cell membrane lipidic composition is not enough to explain triclosan susceptibility. For example, *Pseudomonas aeruginosa* cell membrane is predominantly lipidic but has an enzymatic membrane efflux pump sufficient to expel triclosan, thus reducing its concentrations and antimicrobial effects [15,16].

Triclosan is added to absorbable sutures to inhibit microbial colonisation and thus reduce the risk of SSI [17–22]. Braided polyglactin-910 sutures are available with a maximum triclosan load of 472 µg/m (V+) [23]. Monofilament polydioxanone (P+) and monofilament poliglecaprone 25 sutures with a maximum of 2360 µg/m (M+) [23–25]. These sutures are also available without triclosan (V, P and M).

A previous study analysed the triclosan release kinetics of V+, P+ and M+ in pure static water and accelerated conditions calibrated to reproduce subcutaneous and intramuscular release in operated large animals [26]. It established the relation between the triclosan release rate and the antistaphylococcal activity in V+. However, the pharmacodynamics of V+, P+ and M+ once implanted in live operated tissues are not documented.

Several surface static agar cultures have measured zones of inhibition (ZOI) of sutures with triclosan (TS) vs. non-triclosan controls (NTS). One assay showed that V+ inhibited the growth of *Staphylococcus aureus* and *Staphylococcus epidermidis* after 24-h exposure, while V caused no inhibition [27]. Others showed similar results with M+ vs. M and P+ vs. P. *Escherichia coli* was inhibited up to approximately 1 cm from P+ [28–30]. Those in vitro experiments also showed the inhibition of methicillin-resistant strains of *S. epidermidis*, *S. aureus* and *Klebsiella pneumoniae*. In vivo experiments in one study showed that subcutaneous TS segments in mice with 7 <sup>×</sup> <sup>10</sup><sup>6</sup> *E. coli* colony-forming units (CFU) inoculum displayed, when removed after 48 h, a 90% microbial reduction, while NTS controls were colonised [28]. The same study showed in guinea pigs with a 4 <sup>×</sup> <sup>10</sup><sup>5</sup> *S. aureus* CFU inoculum a 99.9% microbial reduction.

While these studies confirm triclosan's antibacterial activity, they do not demonstrate the translation of the results to operated human tissues regarding triclosan bioavailability and TS antimicrobial activity. The level, duration and volume of antimicrobial activity around TS are uncertain.

The fact that 88% (22/25) (11,957 patients) of parallel-arm prospective randomised controlled clinical trials (RCT), which are the most comprehensive meta-analyses published to date, are non-significant supports translational uncertainty, although the pooled relative risk (RR) was 0.73 [0.65, 0.82] [31]. The WHO has published a conditional guideline recommending the use of TS to reduce the risk of SSI, stating that the quality of the evidence is moderate [32].

Many factors influence SSI risk, including the suture material, microbial concentration, infected volume, microbial multiplication, susceptibility to triclosan, exposure duration, surgical site characteristics and patient's natural defences [33].

The objective of this study (AD16-174/AST2016-181/IIS15-216/2016-11-09) was to analyse the in vitro pharmacodynamics of the three TS, understand their translational antimicrobial characteristics and compare them (Table S1).

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

#### *2.1. Microbiology: Time-Kill Assays*

Nine microorganisms common in SSIs were selected, representing a range of triclosan minimum inhibitory concentrations (MIC). Those were *E. coli* ATCC 25922 (MIC 0.03 µg/mL), *E. coli* ESBL producer collection clinical strain (MIC 0.03 µg/mL), *S. epidermidis* CIP 8155T (MIC 0.03 µg/mL), *S. aureus* ATCC 29213 (MIC 0.03 µg/mL), Methicillin-resistant *S. aureus* (MRSA) ATCC 33592 (MIC 0.03 µg/mL), MRSA collection clinical strain, *Candida albicans* ATCC 10231 (MIC 4 µg/mL), *C. albicans* collection clinical strain and *P. aeruginosa*

collection clinical strain (MIC 256 µg/mL). V, V+, P, P+, M and M+ challenged all the microorganisms. All sutures had USP 2-0 calibres, i.e., a 0.35 to 0.399 mm diameter, a 35 cm length and, thus, a 0.03 to 0.04 cm<sup>3</sup> volume.

A time-kill assay protocol was specified according to CLSI standards [34]. Sutures and microbial cultures were handled in a safety cabinet. Sutures were unpackaged and immediately cut into various lengths, including 35 cm-long segments. Suture incubation on agar plates for 18 h at 37 ◦C checked sterility.

Agar plate cultures each microorganism's purity. A single colony was inoculated in a 10 mL sterile tryptic soy culture broth (TSB; Difco, BD Diagnostics, Sparks, MD, USA) tube. The suspension cultures were incubated for 4 h at 37 ◦C with continuous shaking until exponential growth was reached with 0.5 McFarland turbidity (approximately 1.5 <sup>×</sup> <sup>10</sup><sup>8</sup> colony-forming units CFU/mL). A densitometric controlled inoculum of the culture was extracted. It was added to 10 mL of the culture medium in a sterile tube setting at an approximately 10<sup>6</sup> CFU/mL baseline microbial concentration. After testing different segment lengths, 35 cm was selected because it enabled the distinguishing of concentration differences between each timeslot. Each culture had one immersed segment incubated for 24 h at 37 ◦C. Microbial concentrations were determined at four timeslots: baseline, 4, 8 and 24 h (T0, T4, T8, T24).

Culture tubes underwent 4 min of 48 Mhz sonication to detach viable microorganisms from the tube walls and suture. A 100 µL sample was drawn and then underwent five serial 1/10 *v*/*v* dilutions. The original sample and dilutions were spread on separate agar plates with Mueller Hinton (MH) medium. The plates were incubated for 24 h at 37 ◦C and used if colonies were countable. Plate colony count multiplied by the dilution factor was the source culture microbial concentration (CFU/mL). Six copies of the 54 sutures × microorganism combinations were performed.

#### *2.2. Data Analysis*

2.2.1. Plots of Time-Kill Assays

For each combination and each timeslot (T0 through T24), the viable microbial concentration was converted to log10(CFU/mL). Repeated time-kill assays were plotted jointly.

#### 2.2.2. Statistical Analysis

A comprehensive panel analysis tested the association of microbial concentration with four factors (1). This model included TS, NTS and *P. aeruginosa*.

$$\mathbf{C}(t) = b\_0 + b\_1 \mathbf{x}\_{\text{trivial}\\ \text{s.u.}} + b\_2 \mathbf{x}\_{\text{material.i}} + b\_3 \mathbf{x}\_{\text{microorganism.i}} + b\_4 \mathbf{x}\_{\text{timestol.i}} \tag{1}$$

A second-panel analysis focused only on TS and triclosan-sensitive microorganisms (2).

$$\mathbf{C}(t) = b\_0 + b\_1 \mathbf{x}\_{\text{material},i} + b\_2 \mathbf{x}\_{\text{micrograms},i} + b\_3 \mathbf{x}\_{\text{timeslot.i}} \tag{2}$$

The *.i* indices refer to the dummy variables defined for each modality of each independent factor. The models were calculated using random effects (RE) panel regression if the Hausman test for random effects was non-significant and in the presence of the heteroskedasticity of microbial concentration residues, confirmed by a significant Breusch– Pagan Lagrange multiplier test (LM). If those conditions were not met, the model would be calculated using the pooled ordinary least squares (OLS) method [35].

Specific findings underwent exploratory post hoc tests.

2.2.3. Pharmacodynamic Fitting of Microbial Concentration

Triclosan bactericidal and antifungal activities and microbial multiplication are two competing dynamics. Therefore, a predictive two-term model was used to predict observations (3).

$$\mathbf{C}(t) = \mathbf{C}\_1 \mathbf{e}^{-tL\_1} + \mathbf{C}\_2 \mathbf{e}^{tL\_2} \tag{3}$$

*C*(*t*) is the microbial concentration in the tube at time t from baseline. The first term describes the microbial decrease over 24 h, with *C*<sup>1</sup> solved using *C*(0)/2, i.e., the half of the mean microbial concentration among the six repetitions at baseline. It is equal to the inoculum divided by the culture medium volume. *Ln*2/*L*<sup>1</sup> is the half-life (*T*11/2) of microbial killing over 24 h. The negative exponent defines an exponential decay.

The second term describes microbial growth, with *C*<sup>2</sup> solved using *C*(0)/2. *Ln*2/*L*<sup>2</sup> is the half-life (*T*11/2) over 24 h. The implicit positive exponent defines the exponential growth.

The four unknown parameters were solved jointly given the observed datasets consisting of (time, microbial concentration) couples for each TS × strain combination and each repetition. The solution targeted minimising the least-squares of observations vs. the predicted values with the following constraints:


$$R\_{adjusted}^2 = 1 - \left(1 - \frac{\sum\_{i=1}^{n} \left(y\_i - \hat{y}\_i\right)^2}{\sum\_{i=1}^{n} \left(y\_i - \overline{y}\right)^2}\right) \frac{n-1}{n - (k+1)}\tag{4}$$

The *R* 2 *adjusted* measures the goodness-of-fit (GoF) of the model and thus the degree of prediction of observations. An *R* 2 *adjusted* equal to 1 indicates the perfect prediction of the observed concentrations, while an *R* 2 *adjusted* equal to 0 shows that the model does not predict any better than random guesses.

Two required initial inputs (the partial concentrations at T0: *C*<sup>1</sup> and *C*2) were defined such that *C*1*e* <sup>0</sup> + *C*2*e* <sup>0</sup> = *C*(0), given *e* <sup>0</sup> = 1.

Data management and statistical analyses were performed with the xt module in Stata 17, StataCorps LLC, College Station, TX, USA.

The fitting of the pharmacodynamic model to the data was performed in Microsoft Excel 2019 version 2205, Microsoft Corporation, Redmond, WA, USA, using the Solver function. Parameter resolutions were checked with Maple 2021.1, Maple Inc., Waterloo, ON, Canada.

#### **3. Results**

#### *3.1. Time-Kill Assays*

The plots show growth with the eight triclosan-sensitive strains with NTS control sutures (Figure 1a–c, Figure 2a–c, and Figure 3a,b) and *P. aeruginosa*, which was used as the triclosan-resistant control strain (Figure 3c).

The plots with TS and triclosan-sensitive strains show an initial rapid microbial concentration reduction between baseline and T4 or T8. The reduction magnitude ranges between about 1.5 log<sup>10</sup> and 3 log10. It is followed by a microbial concentration plateau between T8 and T24. The plateau ranges between a mild decline, a steady concentration and a mild increase.

**Figure 1. (a-c) Figure 1.** Time-kill analyses *S. aureus* and MRSA with the three TS. (**a**–**c**) Time-kill analyses *S. aureus* and MRSA with the three TS.

**Figure 2. (a-c)** Time-kill analyses *C. albicans* and *S. epidermidis* with the three TS. **Figure 2.** (**a**–**c**) Time-kill analyses *C. albicans* and *S. epidermidis* with the three TS.

**Figure 3. (a-c)** Time-kill analyses *E. coli* and *P. aeruginosa* with the three TS. **Figure 3.** (**a**–**c**) Time-kill analyses *E. coli* and *P. aeruginosa* with the three TS.

#### *3.2. Statistical Analysis–Comprehensive Panel Model*

The test eligibility criteria applied to the comprehensive model indicate the applicability of OLS regression (Table 1). The adjustment of the model is moderate, given R<sup>2</sup> = 0.578. The regression coefficients of the model are large and significant when comparing TS to the NTS controls, when comparing timeslots to baseline T0 and when comparing the *P. aeruginosa* triclosan-resistant control to triclosan-sensitive *S. epidermidis*, the reference level in the model. All other coefficients have a small magnitude and are non-significant. This shows that the microbial concentration is significantly associated with the presence or absence of triclosan, timeslots and *P. aeruginosa.* Suture materials and other microbial strains are not associated with microbial concentration in this comprehensive model.

**Table 1.** Comprehensive OLS regression of log<sup>10</sup> microbial concentration depending on triclosan, material, strain and timeslot.


\* *p* < 0.001, \*\* *p* < 0.005, Sig.: significant, df: degrees of freedom, St. Err.: standard error. Hausman test: significance of: fails to meet the asymptotic assumptions. Breusch and Pagan Lagrangian multiplier test (LM) for random effects: *p*-value = 0.2085.

#### *3.3. Statistical Analysis–Focused Panel Model*

The test eligibility criteria applied to the focused model indicated the use of randomeffects panel regression (Table 2). The adjustment of the model is improved with R <sup>2</sup> = 0.687. The regression coefficients of the model are large and significant when comparing the timeslots to the T0 baseline and are significant but mild when comparing the 5 triclosansensitive microbial strains to *S. epidermidis* as a reference level for microorganisms in the model. The coefficients for suture materials have a small magnitude and are non-significant. This model shows that the microbial concentration is significantly associated with timeslots, mildly associated with some microbial strains and not associated with suture materials.


**Table 2.** Focused random effects panel regression of log<sup>10</sup> microbial concentration depending on material, strain and timeslot.

*P. aeruginosa* and NTS not included in the focused model; \* *p* < 0.001, \*\* *p* < 0.005, Sig.: significant, df: degrees of freedom, St. Err.: standard error. Hausman test: *p* = 1. Breusch and Pagan Lagrangian multiplier test (LM) for random effects: *p*-value < 0.0001.

#### *3.4. Statistical Analysis–Post Hoc Paired t-Test*

The mean microbial reduction among triclosan-susceptible microorganisms, from baseline (T0) to trough (T4 or T8), was −2.29 log<sup>10</sup> of CFU/mL [–2.40, –2.19].

The post hoc paired t-test showed a mild but significant mean concentration increase from trough to T24 (+0.36 log<sup>10</sup> of CFU/mL [+0.26; +0.46], *p* < 0.0001).

#### *3.5. Pharmacodynamic Fitting of Microbial Concentration*

Each of the 24 combinations was fitted with a predictive pharmacodynamic function. Examples with *S. epidermidis* are shown in Figure 4, *S. aureus* in Figure 5, *E. coli* in Figure 6, and examples with ESBL-producing *E. coli* in Figure 7. Subfigures (a), (b) and (c) are fittings with V+, P+ and M+, respectively.

The predictive pharmacodynamic functions of the 24 combinations, based on a common algebraic function and fitted parameters, are listed in Table 3. The GoF, estimated by the adjusted *R* 2 *adjusted*, is between 0.61 and 0.89 in 22 fitted functions. There is a poor fit in the two other functions (*R* 2 *adjusted* between 0.38 and 0.49) with *C. albicans ATCC 33529* and the monofilament sutures P+ and M+ (Figures S1–S12).

All functions show an early fast microbial concentration decline from baseline to T4 or T8 h, followed by a plateau until T24. The plateau has a steady concentration in 2 cases (8.33%), a mild concentration decrease in 4 cases (16.7%) and a mild concentration increase in 18 cases (75%).

**Figure 4.** (**a**–**c**) Fitted predictive pharmacodynamic functions *S. epidermidis* with the three TS.

**Figure 4.** (a-c) Fitted predictive pharmacodynamic functions *S. epidermidis* with the three TS*.* 

**Figure 5.** (**a**–**c**) Fitted predictive pharmacodynamic functions *S. aureus* with the three TS.

**Figure 5.** (a-c) Fitted predictive pharmacodynamic functions *S. aureus* with the three TS

**Figure 6.** (**a**–**c**) Fitted predictive pharmacodynamic functions *E. Coli* with the three TS.

**Figure 6.** (a-c) Fitted predictive pharmacodynamic functions *E. Coli* with the three TS

**Figure 7.** (a-c) Fitted predictive pharmacodynamic functions ESBL *E. Coli* with the three TS. **Figure 7.** (**a**–**c**) Fitted predictive pharmacodynamic functions ESBL *E. Coli* with the three TS.


**Table 3.** Fitting pharmacodynamic model (3) microbial concentrations from T0 to T24, excluding NTS and *P. aeruginosa*.

Note: "mild" means a variation of less than 1log10.

#### **4. Discussion**

#### *4.1. Protocol Specifications and Interpretation*

This study performed the first in vitro pharmacodynamics analysis of TS antimicrobial activity using time-kill assays. V, P, M and triclosan-resistant *P. aeruginosa* were the controls. The experimental settings were the same as those used in static water release kinetics [26].

The suspension cultures had enough volume and nutrients to sustain microbial growth beyond the T24 timeslot, as confirmed by the growth in the NTS cultures (Figure 1 to Figure 3). Microbial colony count was proportional to the microbial concentration in the cultures, with a degree of random error between repetitions.

Microbial concentrations exceeding the 10<sup>8</sup> CFU/mL upper boundary could not be accurately estimated, but that had no impact on the analysis. None of them reached the lower detection boundary below 10<sup>2</sup> CFU/mL.

#### *4.2. Key Results*

The plots and statistical analyses showed that the microbial concentration is significantly associated with triclosan, the timeslot and the microorganism. It is not associated with the suture material.

The plots with all TSs and triclosan-susceptible microorganisms consisted of an initially rapid microbial reduction from T0 to T4 or T8, with a mean reduction of −2.29 log<sup>10</sup> of CFU/mL followed by a plateau without a microbial concentration change, mild decrease or mild increase. The mean change between the T8 and T24 was a mild but significant increase.

The underlying mechanisms of the predictive pharmacodynamic models were assumed to be microbial kill and multiplication. The fitting was good in all combinations except for two. The models differed little between TS types for a given microorganism. Most differences were between microorganisms. The functions reproduced the initial microbial concentration rapid reduction and subsequent plateau and showed a mild increase in 75% of combinations.

#### *4.3. Interpretation of The Time-Kill Assays*

These assays show the ability of V+, P+ and M+ to reduce a high microbial concentration over a 4-to-8-h period, with very little difference given each microorganism despite the 5-fold difference in the triclosan load and suture structure. The plateau and frequent mild increase are unexpected.

The key question is whether the predominant killing was by the triclosan concentration in the culture medium or by the contact killing at the suture surface or near it, where the triclosan concentration is high. Indeed, triclosan is a hydrophobic solid whose dissolution follows the Noyes & Whitney principle, which explains the triclosan gradient between the solid surface and the bulk of the solvent in a static tube. This gradient disappears when the tube is shaken. Therefore, its dissolution rate is slow around the suture, whose volume is 0.3 to 0.4% of the culture volume. Therefore, unless the tube is shaken, the triclosan diffusion layer forms a gradient with a maximum close to 10 µg/mL at the suture surface, decreasing with the distance from the source [7,26,36–39].

The culture conditions in static TSB (water + 3% organic nutrients and minerals) were closer to pure static water release kinetic determinations than to the ethanol/water 3.3% w/w solution with 24 rounds-per-minute constant rotation [26]. The triclosan release in 10 mL of pure static water is with V+ 2.3, 3.3 and 6.8 µg at 4, 8 and 24 h, respectively, with P+, 5.3, 6.7 and 6.1 µg and with M+, 7.6, 9.6 and 7.4 µg [26]. After sonication, the triclosan concentrations were homogeneous in the bulk of the cultures, i.e., with V+ 0.23, 0.33 and 0.68 µg/mL at 4, 8 and 24 h, respectively, with P+, 0.53, 0.67 and 0.61 µg/mL and with M+, 0.76, 0.96 and 0.74 µg/mL.

The triclosan minimal bactericidal concentrations (MBC) of *S. aureus* (0.03 to 2 µg/mL), *E. coli* (0.03 to 16 µg/mL) and *C. albicans* (0.12 to 16 µg/mL) are presented [40,41]. Therefore, the triclosan concentrations were within the MBC ranges, as the first sonication. This should have caused microbial killing through T24 in all TS ×microbial combinations. The plateau

and a mild increase in 75% of combinations after T8 suggest that the triclosan concentration in the medium was too low to prevent microbial multiplication.

The available data do not provide proof of the mechanism. However, a potential explanation is that the dispersed triclosan after the T4 and T8 sonications were captured in the lipids of the killed bacteria. Therefore, the plateau between T8 and T24 is likely to be an experimental artefact when the tubes are still, and the two terms of the pharmacodynamic functions offset each other.

The obtained in vitro data do not show why the mild microbial concentration increased during the T8–T24 plateau in 75% of the assays. One potential explanation is the gradual decrease in the release rate while exponential microbial growth continued.

#### *4.4. Comparison with Other Preclinical Studies*

These in vitro pharmacodynamic models and underlying release kinetics are compatible with the absence of *E. coli* and *S. aureus* surface growth on agar plates with TS segments explanted from rodents after 48 h [26,28].

One study attempted to measure the duration and level of TS antimicrobial activity with a TS segment transferred consecutively from one static surface agar culture plate to another for up to 30 days [42]. The conclusions were that TSs display antimicrobial activity from about one week to one month depending on the TS × microorganism.

The methods of these in vitro experiments must be considered when interpreting the results. (1) The two-dimensional diffusion around the TS segments on the agar surface cultures represents, at most, the amount that would be contained in the three-dimension diffusion layer of the suspension culture. (2) The water/air surface diffusion meets less resistance than it does in full immersion. Therefore, ZOI overestimates, by several-fold, the antimicrobial volume of TS. (3) The TS triclosan release in static cultures is 15 to 60 times slower than that observed in large animal subcutaneous or intramuscular explants [26]. Therefore, the in vitro antimicrobial activity duration is also overestimated.

#### *4.5. Translational Interpretation to Live Operated Human Tissues*

The translational application of this pharmacodynamic study to operated human tissues is limited.

(1) The TS release rate in the time-kill assays is 15 to 60 times slower than it is in operated tissues. (2) Surgical sites are much larger than 10 mL tubes, so the suture volume is much less than 0.3 to 0.4% of the surgical site. (3) The permanent in vivo motion maintains a thin diffusion layer around the sutures, so the contact or near-contact volume with drifting microorganisms is probably negligible compared to the surgical site volume at risk. (4) When natural defences are functional, scattered bacteria are rapidly killed, and few encounter the TS. When natural defences are weak, scattered microbial multiplication requires antibiotics and/or reintervention

Therefore, preventing microbial colonisation while the triclosan release rate is efficacious, i.e., a few hours after implantation, is the only result of this pharmacodynamic study that can translate to a surgical site. However, that can relieve natural defences by reducing the risk of microbial colonisation of the suture. The three TS types share similar in vitro antimicrobial activity. There is no indication they would have significantly different in vitro antimicrobial activities in operated tissues. Measurements of microbial dose or concentration reduction cannot translate to operated tissues because they do not consider the complexity of surgical sites and natural defences.

#### **5. Conclusions**

This in vitro study shows that triclosan sutures kill susceptible microorganisms that come in direct contact or near contact with their surface. The in vitro antimicrobial profiles of the braided polyglactin-910, monofilament polydioxanone and monofilament poliglecaprone 25 sutures present no significant difference, and no difference in the operated tissues is predicted.

The in vitro pharmacodynamics suggest a significant reduction in the microbial dose close to the sutures as of implantation. Triclosan can minimise the suture colonisation risk early on, relieve natural defences and reduce the risk of surgical site infection.

**Supplementary Materials:** The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/antibiotics11091195/s1. Figure S1: MRSA ATCC 33592 with V+; Figure S2: MRSA ATCC 33592 with P+; Figure S3: MRSA ATCC 33592 with M+; Figure S4: MRSA clinical with V+; Figure S5: MRSA clinical with P+; Figure S6: MRSA clinical with M+; Figure S7: C. albicans ATCC 10231 with V+; Figure S8: *C. albicans* ATCC 10231 with P+; Figure S9: *C. albicans* ATCC 10231 with M+; Figure S10: *C. albicans* clinical with V+; Figure S11: *C. albicans* clinical with P+; Figure S12: *C. albicans* clinical with M+; Table S1: Pharmacodynamics raw data.

**Author Contributions:** Conceptualisation, F.C.D., A.-M.R., N.M. and F.M.; methodology, F.M., A.Z. and F.C.D.; microbiological data assays, F.M. and A.Z.; quality assessment, F.M.; validation, A.-M.R. and F.C.D.; data analysis, F.C.D.; interpretation, F.C.D., F.M. and A.-M.R.; writing—original draft preparation, F.C.D.; writing—review and editing, F.M., A.Z., A.-M.R. and N.M.; supervision, N.M. and A.-M.R. All authors have read and agreed to the published version of the manuscript.

**Funding:** The project specification, organisation and analysis were funded by Université de Bordeaux, INSERM U1219 Bordeaux Population Health. The microbiology assays were performed at Université de Bordeaux, Aquitaine Microbiologie, UMR 5234 CNRS, and they were supported by an unrestricted grant from Ethicon, a division of Johnson and Johnson Medical Limited (investigator-initiated project ID.: IIS15-216/2016-11-09).

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

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** All source data is in the Supplementary Materials File (raw data table).

**Acknowledgments:** Ethicon, a division of Johnson and Johnson Medical Limited, shared their experience with in vitro assays with those products without interfering with the protocol design, conduct and analysis.

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

#### **References**


## *Article* **Analysis of a Library of** *Escherichia coli* **Transporter Knockout Strains to Identify Transport Pathways of Antibiotics**

**Lachlan Jake Munro <sup>1</sup> and Douglas B. Kell 1,2,\***


**Abstract:** Antibiotic resistance is a major global healthcare issue. Antibiotic compounds cross the bacterial cell membrane via membrane transporters, and a major mechanism of antibiotic resistance is through modification of the membrane transporters to increase the efflux or reduce the influx of antibiotics. Targeting these transporters is a potential avenue to combat antibiotic resistance. In this study, we used an automated screening pipeline to evaluate the growth of a library of 447 *Escherichia coli* transporter knockout strains exposed to sub-inhibitory concentrations of 18 diverse antimicrobials. We found numerous knockout strains that showed more resistant or sensitive phenotypes to specific antimicrobials, suggestive of transport pathways. We highlight several specific drug-transporter interactions that we identified and provide the full dataset, which will be a useful resource in further research on antimicrobial transport pathways. Overall, we determined that transporters are involved in modulating the efficacy of almost all the antimicrobial compounds tested and can, thus, play a major role in the development of antimicrobial resistance.

**Keywords:** transporters; antibiotics; *Escherichia coli*

**Citation:** Munro, L.J.; Kell, D.B. Analysis of a Library of *Escherichia coli* Transporter Knockout Strains to Identify Transport Pathways of Antibiotics. *Antibiotics* **2022**, *11*, 1129. https://doi.org/10.3390/ antibiotics11081129

Academic Editor: Dóra Kovács

Received: 12 July 2022 Accepted: 15 August 2022 Published: 19 August 2022

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

**Copyright:** © 2022 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/).

#### **1. Introduction**

Antibiotic resistance is a major global healthcare burden, with over 1 million deaths attributable to antibiotic resistance in 2019, and with the World Health Organization listing antimicrobial resistance as one of the top 10 global public health threats facing humanity [1,2]. Despite a pressing need for novel antibiotics that can overcome resistance, few novel drugs have been developed and approved in the last 20 years, with a general downward trend in new approvals since 1983 [3], albeit the last decade has seen some increase in new approvals [4]. While systems have been proposed to overcome the economic obstacles to incentivize antibiotic development [5], other solutions are needed.

In order to exert their effects, most antibiotics must first cross the bacterial cell membrane. There is now strong evidence to suggest that in order to enter cells, nearly all compounds must pass through membrane transporters and that "passive bilayer diffusion is negligible" [6–8]. The importance of transporters is further highlighted by the wellestablished fact that a major mechanism of antibiotic resistance is through adaptations in transport systems. These adaptations reduce the intracellular concentration of antibiotics through either the increased efflux or the decreased influx of antibiotics through membrane transporters [9–11]. Antibiotic adjuncts that inhibit the efflux of the active drug have been proposed as a means to counteract resistance [12,13].

While modelling suggests that mutations that cause antibiotic resistance will predominantly be in exporters rather than importers [14], there are also examples where adaptations that reduce the import of antibiotics confer resistance. This has been observed in the case of fosfomycin [15], aminoglycosides [16] and chloramphenicol [17].

Of the ~4401 genes in the *Escherichia coli* K-12 chromosome, an estimated 598 encode established or predicted membrane transporters [18]. Despite extensive research, around a

quarter of these transporters are orphans (also known as y-genes), in that they have no assigned substrate. Recent studies have illustrated that accumulation of cationic fluorophores involves multiple influx and efflux transporters, with a library of transporter knockouts spanning a 30–70-fold difference in fluorescence level when exposed to the fluorophores SYBR Green or diS-C3(5) [19]. We hypothesized that accumulation and excretion of antimicrobial agents may similarly involve numerous membrane transporters. Previously, we have used a high-throughput screening method to identify a melatonin exporter [20]. In this, growth of a library of knockout transporters was measured in the presence of melatonin, and inhibited growth was used as a proxy to identify when an exporter had been knocked out (conversely, a resistant strain potentially has an importer knocked out).

In the present study, which forms part of an ongoing project to deorphanize all orphan *E. coli* transporters, we investigated the previously developed library of 447 *E. coli* transporter knockouts [19,20] for growth, in the presence of 18 structurally diverse antimicrobial agents. We included novel and recently approved antibiotics, such as cefiderocol and flumequine, as well as drugs not traditionally used against Gram-negative bacteria. We also developed an automated workflow for the liquid-handing steps and data processing. The full dataset encompasses approximately 8000 growth curves (in duplicate), and the entire raw dataset and code used for processing are provided in the Supplementary Materials. We identified several transporter knockout strains that showed resistance or sensitivity to many of the antimicrobials tested, including previously unannotated transporters or y-genes. Finally, we illustrate the potential utility of large-scale analysis of the dataset for predicting substrates of orphan transporters.

#### **2. Results**

#### *2.1. Antimicrobial Selection*

We selected a range of compounds that were available and readily soluble in our LB media. We endeavored to include compounds from several major antibiotic classes as well as compounds with activity that have previously not been tested extensively against Gramnegative bacteria (ornidazole and paraquat). Cefiderocol was very recently approved, so was included due to novelty.

#### *2.2. Determination of Antibiotic Concentrations for High-Throughput Screening*

The workflow for the screening of each compound (Supplementary Figure S1) initially required identifying a sub-inhibitory concentration of the antibiotic in the wild-type strain BW1556 (WT, parent strain for the Keio collection). Minimum inhibitory concentrations (MICs) were determined by measurement of *E. coli* growth in a two-fold serial dilution of the relevant antibiotic in LB. We defined MIC values as those with OD levels at 48 h less than 10% of the antibiotic-free media condition. For screening of the transporter library, we selected concentrations that showed some inhibition of growth in the WT strain without causing full inhibition. Generally, this was a concentration half that of the MIC; however, when effects on growth (reduction in maximum OD, increased lag time, decreased growth rate, or some combination thereof) could be clearly seen at concentrations significantly lower than that, then these concentrations were used. The 18 antimicrobial compounds included in this study, as well as the MIC and screening concentrations, are listed in Table 1, and chemical structures of the antibiotics used are shown in Supplementary Figure S2.

#### *2.3. Baseline Characteristics of Transporter Library*

Generally, the growth of nearly all transporter knockout strains in LB without supplementation was similar. The empirical area under the curve (AUC) (a general measure of growth) had a mean of 30.0 and an interquartile range (IQR) of 1.90, while the maximum growth rate had a mean of 1.19 and an IQR of 0.14 h−<sup>1</sup> . The WT growth rate was slightly higher than the mean growth rate, possibly due to the burden of expression of the constitutively active kanamycin cassette present in all knockout strains [21,22]. Histograms illustrating the distribution of the AUC and growth rate are shown in Figure 1A,B. We

also observed good correlation between replicates with regards to growth rate and AUC (Figure 1C,D). It should be noted that the automated inoculation step occasionally did not dispense a droplet into the well. In cases where this was clearly the case (i.e., where we observed growth in one but not another replicate), the replicate missing growth was filtered from the results.

**Table 1.** Antibiotics and antimicrobials used in this study, with minimum inhibitory concentration (MIC) in the wild-type (WT) strain, and the concentrations used for high-throughput screening experiments are shown.


\* For these compounds, growth was detected at the highest concentration tested, despite clear inhibition at lower concentrations. *Antibiotics* **2022**, *11*, x FOR PEER REVIEW 4 of 18

**Figure 1.** Histograms illustrating distributions of (**A**) growth rate and (**B**) area under the curve (AUC) for transporter library growth in LB in the absence of antibiotics. Scatter plots showing correlation between replicates for (**C**) growth rates and (**D**) AUC for transporter library growth in LB alone. **Figure 1.** Histograms illustrating distributions of (**A**) growth rate and (**B**) area under the curve (AUC) for transporter library growth in LB in the absence of antibiotics. Scatter plots showing correlation between replicates for (**C**) growth rates and (**D**) AUC for transporter library growth in LB alone.

of antibiotics, the rank-order curves show a much wider spread of values. Figure 3 shows a box and whisker plot illustrating the distribution of AUC values for all compounds

tested.

Compared to growth in LB, we saw a much larger degree of variation in growth in

Compared to growth in LB, we saw a much larger degree of variation in growth in the presence of the antimicrobial compounds, consistent with multiple transporter knockouts influencing the level of intracellular antibiotic accumulation. Figure 2A shows rank-order curves for the mean AUC for growth of the transporter library in LB, while Figure 2B–D shows the same information for growth in the indicated antibiotics. In the presence of antibiotics, the rank-order curves show a much wider spread of values. Figure 3 shows a box and whisker plot illustrating the distribution of AUC values for all compounds tested. *Antibiotics* **2022**, *11*, x FOR PEER REVIEW 5 of 18 *Antibiotics* **2022**, *11*, x FOR PEER REVIEW 5 of 18

**Figure 2.** Transporter knockout library ordered by the mean area under the curve (AUC). Data shown for growth in LB (**A**) and subinhibitory concentrations of antibiotics/antimicrobials indicated (**B**–**D**). Y-genes are shown in blue, and annotated transporters are shown in red, while the arrow indicates the position of the WT strain. **Figure 2.** Transporter knockout library ordered by the mean area under the curve (AUC). Data shown for growth in LB (**A**) and subinhibitory concentrations of antibiotics/antimicrobials indicated (**B**–**D**). Y-genes are shown in blue, and annotated transporters are shown in red, while the arrow indicates the position of the WT strain. **Figure 2.** Transporter knockout library ordered by the mean area under the curve (AUC). Data shown for growth in LB (**A**) and subinhibitory concentrations of antibiotics/antimicrobials indicated (**B**–**D**). Y-genes are shown in blue, and annotated transporters are shown in red, while the arrow indicates the position of the WT strain.

of the transporter library in LB and antimicrobial compounds. **Figure 3.** Box and whisker plot showing distribution of mean area under the curve (AUC) for growth of the transporter library in LB and antimicrobial compounds. **Figure 3.** Box and whisker plot showing distribution of mean area under the curve (AUC) for growth of the transporter library in LB and antimicrobial compounds.

#### *2.4. Resistant and Sensitive Transporter Results*

We observed that many transporter knockouts had effects on the sensitivity to the compounds tested. Due to the large volume of data generated, detailed descriptions of all relevant results are beyond the scope of this paper. As we intend to perform targeted follow-up validation experiments, given that this is an initial exploratory study, we are hesitant to define specific criteria for sensitivity or resistance. Indeed, one advantage of our study over previous high-throughput growth assays is the generation of full growth curves rather than the reduction in growth of a single metric, as we elaborate on in the discussion.

Selected specific results are discussed below. Generally, when searching for novel results, we looked at strains ranked highest or lowest in normalized growth rate or AUC and then manually inspected the growth curves. We selected those highlighted in this paper based on interactions with y-genes or where we felt there were plausible and interesting mechanisms for discussion.

Our study replicated some previously established drug–transporter interactions. First, we observed that knockout of acrB, a promiscuous drug-efflux protein well-known to play a role in export of many xenobiotics [23], caused increased sensitivity to 10 out of the 18 antimicrobial compounds tested. Interestingly we saw the increased inhibition caused by knockout of acrB take different forms: growth rate decreases including increased lag time, decreased max OD and AUC and complete inhibition (Figure 4). Fosphomycin has been shown to gain access to the cell via the hexose-6-phosphate:phosphate antiporter uhpT and the glutamate/aspartate: H+ symporter gltP [15,24,25]. In our phosphomycin experiment, knockouts of each of these transporters were among the most resistant of the transporter knockout strains tested, consistent with these knockouts causing reduction in influx (Supplementary Materials). MacAB-Tolc has been reported to efflux macrolide antibiotics, and expression of this complex in a strain hypersensitive to antibiotics conferred resistance to macrolides including azithromycin [26]. In our study, the macB knockout strain (∆*macB*) did not result in azithromycin sensitivity, and we also could not see evidence of increased sensitivity to azithromycin in a previously published dataset [27]. *Antibiotics* **2022**, *11*, x FOR PEER REVIEW 7 of 18

**Figure 4.** Growth curves for WT (black) and *∆acrB* (red) in LB (**A**), azithromycin, (**B**) D-cycloserine (**C**) and chloramphenicol (**D**). Shaded areas represent the standard deviation for the WT growth curve. **Figure 4.** Growth curves for WT (black) and ∆*acrB* (red) in LB (**A**), azithromycin, (**B**) D-cycloserine (**C**) and chloramphenicol (**D**). Shaded areas represent the standard deviation for the WT growth curve.

sensitive strains (by normalized AUC) are shown in Table 2.

Ornidazole is a nitroimidazole generally used to treat protozoan infections, and, in this study, we found it to have an MIC in *E. coli* BW1556 of 800 mg/L (Table 1). The transporter nimT (previously yeaN) has been shown to function as an exporter of 2-nitroimid-

in ornidazole sensitivity in ∆*nimT* compared to WT (Figure 5A). We did, however, observe a clear increased sensitivity to ornidazole in ∆*argO* (Figure 5B)*,* a strain with a probable arginine exporter knocked out [29]. Some strains also showed an ornidazole-resistant phenotype. These included ∆*narU*, a nitrite/nitrate exchanger knockout (Figure 5C). Possibly, this transporter is interacting in some way with the nitro-group on ornidazole, and the knockout of this reduces influx into the cell. We also saw a resistant phenotype in ∆*ygaH*  (Figure 5D)*,* an uncharacterized orphan transporter. The other most resistant and most

Ornidazole is a nitroimidazole generally used to treat protozoan infections, and, in this study, we found it to have an MIC in *E. coli* BW1556 of 800 mg/L (Table 1). The transporter nimT (previously yeaN) has been shown to function as an exporter of 2-nitroimidazole, a compound structurally similar to ornidazole [28]. We saw only a modest increase in ornidazole sensitivity in ∆*nimT* compared to WT (Figure 5A). We did, however, observe a clear increased sensitivity to ornidazole in ∆*argO* (Figure 5B), a strain with a probable arginine exporter knocked out [29]. Some strains also showed an ornidazole-resistant phenotype. These included ∆*narU*, a nitrite/nitrate exchanger knockout (Figure 5C). Possibly, this transporter is interacting in some way with the nitro-group on ornidazole, and the knockout of this reduces influx into the cell. We also saw a resistant phenotype in ∆*ygaH* (Figure 5D), an uncharacterized orphan transporter. The other most resistant and most sensitive strains (by normalized AUC) are shown in Table 2. *Antibiotics* **2022**, *11*, x FOR PEER REVIEW 8 of 18

**Figure 5.** Growth curves for WT (black) and knockout strains (red) in 400 mg/L ornidazole. ∆*nimT* (**A**), ∆*argO* (**B**), ∆*narU* (**C**) and ∆*ygaH* (**D**) are shown. Shaded areas represent the standard deviation for the WT growth curve. **Figure 5.** Growth curves for WT (black) and knockout strains (red) in 400 mg/L ornidazole. ∆*nimT* (**A**), ∆*argO* (**B**), ∆*narU* (**C**) and ∆*ygaH* (**D**) are shown. Shaded areas represent the standard deviation for the WT growth curve.

**Table 2.** Growth parameters for the 15 most sensitive and most resistant strains in Ornidazole by normalized AUC. Normalized values show the value in ornidazole/value in LB for the given strain. **Strain Max OD Rate AUC Normalized Max OD Normalized Rate Normalized AUC**  argO 0.74 0.35 6.36 0.29 0.45 0.23 ydjE 0.58 0.71 6.55 0.71 0.37 0.26 tolQ 0.60 1.05 7.98 1.16 0.38 0.29 Azithromycin is a macrolide antibiotic, widely used in the treatment of clinical infections. We saw several transporter knockout strains that had dramatically decreased growth in the presence of 7.5 mg/L azithromycin compared to WT. These included the choline transporter knockout ∆*betT* and a tyrosine symporter knockout ∆*tyrP* (Figure 6A,B). Near-complete inhibition was also seen in the multidrug exporter knockout strain ∆*mdtN*. We also observed resistant phenotypes, specifically in the orphan knockout strains ∆*yhdW* and ∆*ydfJ* (Figure 6C,D). Growth parameters for the most resistant and sensitive strains are shown in Table 3.

glvB 0.75 0.57 9.78 0.54 0.44 0.34 sapB 0.64 0.73 9.05 0.74 0.41 0.35 yebQ 0.83 0.59 8.96 0.89 0.53 0.36 ddpF 0.76 0.80 11.43 0.64 0.44 0.37 uup 0.72 0.85 10.07 0.81 0.45 0.37 cysW 0.64 1.62 10.22 1.40 0.41 0.38 ydcS 0.65 0.54 9.25 0.70 0.39 0.38

guaB 0.69 1.08 11.22 1.18 0.42 0.40 gntU 0.87 1.04 12.13 0.83 0.52 0.40 ynfA 0.77 0.93 9.97 1.31 0.52 0.41 ybhF 1.15 0.95 19.30 0.96 0.66 0.65 arnE 1.14 1.03 17.97 1.14 0.70 0.66 sapC 1.11 0.74 17.12 0.77 0.67 0.66 cycA 1.17 1.07 19.65 0.83 0.69 0.66 narU 1.14 0.86 19.88 0.67 0.67 0.66


**Table 2.** Growth parameters for the 15 most sensitive and most resistant strains in Ornidazole by normalized AUC. Normalized values show the value in ornidazole/value in LB for the given strain.

**Figure 6.** Growth curves for WT (black) and knockout strains (red) in 400 mg/L azithromycin. ∆*betT* (**A**), ∆*tyrP*(**B**), ∆*yhdW* (**C**) and ∆*ydfJ* (**D**) are shown. Shaded areas represent the standard deviation for the WT growth curve. **Figure 6.** Growth curves for WT (black) and knockout strains (red) in 400 mg/L azithromycin. ∆*betT* (**A**), ∆*tyrP*(**B**), ∆*yhdW* (**C**) and ∆*ydfJ* (**D**) are shown. Shaded areas represent the standard deviation for the WT growth curve.


**Table 3.** Growth parameters for the 15 most sensitive and most resistant strains in ornidazole. Normalized values represent the value in ornidazole/value in LB for the given strain and values for the maximum OD, rate and area under the curve are shown.

We also included compounds not traditionally used as antibiotics, which have antimicrobial activity. Paraquat (also known as methyl viologen) is a widely used herbicide, which exerts toxic effects, after conversion to a superoxide radical, once inside the cell [30]. We found that concentrations up to 257 mg/L did not cause full inhibition of *E. coli* growth; however, there was evidence of growth inhibition at concentrations of 32 mg/L (Figure 7A). We screened the growth of the transporter knockout library at 67 mg/L and observed several strains that had increased sensitivity including the y-gene knockout ∆*ydcZ* and in two metal cation exporter knockouts ∆*cusA* (Figure 7B) and ∆*fieF*, suggestive of the ability of these transporters to export paraquat, which notably is also a cation. We also observed strains with resistance; interestingly, these included ∆*aroP*, an aromatic amino acid permease knockout. Given the aromatic structure of paraquat, it is very plausible that one of the means of its entry into the cell is via this aromatic amino acid permease (Figure 7C). The most sensitive and resistant strains in paraquat are shown in Table 4.

maximum OD, rate and area under the curve are shown.

**Figure 7.** Growth features of paraquat. (**A**) Concentration inhibition data for growth of WT *E. coli* in the indicated concentration of paraquat. Shaded areas represent the standard deviation, n = 3. Growth curves for WT (black) and knockout strains (red) in 32 mg/L paraquat. ∆*cusB* (**B**) and ∆*aroP* (**C**) are shown. **Table 4.** Growth parameters for the 15 most sensitive and most resistant strains in paraquat. Normalized values represent the value in paraquat/value in LB for the given strain and values for the **Figure 7.** Growth features of paraquat. (**A**) Concentration inhibition data for growth of WT *E. coli* in the indicated concentration of paraquat. Shaded areas represent the standard deviation, n = 3. Growth curves for WT (black) and knockout strains (red) in 32 mg/L paraquat. ∆*cusB* (**B**) and ∆*aroP* (**C**) are shown.

**Strain MaxOD Rate AUC Normalized MaxOD Normalized Rate Normalized AUC**  ydcZ 0.11 1.55 0.39 1.35 0.06 0.01 potH 0.74 0.57 1.79 0.50 0.44 0.06 **Table 4.** Growth parameters for the 15 most sensitive and most resistant strains in paraquat. Normalized values represent the value in paraquat/value in LB for the given strain and values for the maximum OD, rate and area under the curve are shown.


#### *2.5. Large-Scale Data Analysis*

The generation of a dataset such as this one also allows the large-scale investigation of relationships between transporters. The heat map of correlations between AUC values for strains across growth conditions is shown in Figure 8A. We also extracted the *p* values for the correlation and found a peak near zero (Figure 8B), indicating likely true positives in the significant results. We found 35 correlations that met the conditions for significance at *p* < 0.05 after full Bonferroni correction [31] (with a corrected *<sup>p</sup>* value of 2.5 <sup>×</sup> <sup>10</sup>−<sup>7</sup> ). These are shown in Table 2. One possible application of this type of large-scale data analysis is in predicting substrates of unannotated transporters by their relationship to transporters with annotations (a "guilt by association" methodology) (Table 5). For instance, the growth parameters of ∆*sapB*, a putrescine exporter knockout, have a high correlation with the knockout of orphan transporter knockout ∆*ydjE* (Figure 8C), suggesting transport of compounds similar to putrescine or at least some overlap in substrate selectivity and function. *Antibiotics* **2022**, *11*, x FOR PEER REVIEW 13 of 18

**Figure 8.** (**A**). Heatmap showing correlations between area under the curve values for transporters in all antibiotics. Zoom inset highlights a cluster of highly correlated transporter knockout strains. (**B**) Histogram of *p*-values for all correlations showing the peak around 0, indicating likely true positive results. (**C**) Scatter plot indicating high degree of correlation between ∆*sapB* and ∆*ydjE*. **Figure 8.** (**A**). Heatmap showing correlations between area under the curve values for transporters in all antibiotics. Zoom inset highlights a cluster of highly correlated transporter knockout strains. (**B**) Histogram of *p*-values for all correlations showing the peak around 0, indicating likely true positive results. (**C**) Scatter plot indicating high degree of correlation between ∆*sapB* and ∆*ydjE*.

**Transporter 1 Transporter 2 Correlation** *p* **Value**  yihP yhdX 0.91 2.46 × 10−<sup>7</sup>

uup gsiC 0.91 2.36 × 10−<sup>7</sup> yicL yhdX 0.91 2.26 × 10−<sup>7</sup> btuC araJ 0.91 2.22 × 10−<sup>7</sup> mdtG copA 0.91 2.04 × 10−<sup>7</sup> rarD adeQ 0.91 2.02 × 10−<sup>7</sup>

**Table 5.** Transporters that are highly correlated.


**Table 5.** Transporters that are highly correlated.

#### **3. Discussion**

Identifying substrates and describing the structure activity relationship of bacterial transporters is a challenging task. Given the importance of transport systems in the function of most antibiotics, understanding the influx and efflux pathways of antimicrobial agents is highly relevant to mitigating problem of drug resistant bacteria. In this study, we have sought to gain insights into the potential influx and efflux pathways of a set of compounds with antimicrobial activity in *E. coli*. We have used an automated method to generate a dataset of growth curves for 447 transporter knockout strains against subinhibitory concentrations of 18 structurally diverse compounds with antibacterial activity. Our data showed results consistent with some previously reported pathways for antibiotic influx and efflux. We also report on numerous novel specific observations of transporter knockouts that reduce or enhance growth in the presence of antimicrobials as well as discuss potentially novel observations from analysis of the full dataset. All data are provided as a resource in the Supplementary Materials.

There were numerous novel transporter–compound interactions identified in this study, which suggest the need for follow-up experiments. We have previously developed a workflow for confirmatory work on the results that involves: (1) PCR to confirm the strain is correctly labelled, and (2) MIC determination to validate the observed result [20]. Work to automate these processes is ongoing and will be detailed in future publications. Once a specific transporter–compound interaction has been confirmed, the task of finding other, potentially higher affinity substrates for the transporter is feasible, as often the identification of an initial substrate is the most difficult step. Tools that can identify compounds "closest" to the antimicrobial in chemical space [32] will allow detailed investigation of the structure activity relationship of transporters.

A limitation to the approach described in this study for transporter pathway identification is that growth is used as a proxy for transport with no direct measurement. While the simplest mechanism by which a transporter knockout changes growth in the presence of an antimicrobial is through the direct alteration of transport, there are other possible mechanisms. Gene knockouts are well-known to cause pleiotropic effects, with knockouts of a single gene often causing altered expression in many other genes that may affect transporter sensitivity [33]. Further validation of predicted effects can be achieved by use of the 'ASKA' overexpression collection [34], where, if the opposite effect is observed (e.g., resistance in an overexpression strain and sensitivity in a knockout strain), this could be seen as providing further evidence that the compound is indeed a substrate of the given transporter [19,35,36]. Given the large degree of redundancy in transporters (i.e., a given substrates may be transported by several transporters), more robust follow-up results could be achieved through investigation of double knockouts, as has recently been demonstrated in yeast [37].

There are studies that have investigated the growth of the full Keio collection of *E. coli* knockouts under different conditions (including numerous antibiotics) on solid media [27,38,39]. While the use of solid media and the application of advanced robotics allows a much higher throughput (with all of the above studies run in a 1536-well format), this does come with limitations. This high-density solid-media growth introduces a "neighbor effect", where crowding of the bacterial colonies causes growth inhibition that needs to be corrected for [38]. Additionally, these studies report one metric or at most two metrics for growth, which potentially obscures the relevant information only available from the full growth curve. It should also be noted that while the authors of the above papers all commendably shared their data, in two out of the above three publications, the provided links for accessing data are no longer active [27,38], fitting with an observed trend of data accessibility decreasing in the years following publication [40].

There is a large degree of homology found between transporters of different pathogenic bacterial families. Given this, as well as the recent advances in prediction of protein structure from sequence [41,42], we envision that the dataset generated in this paper will be of use for understanding and predicting the interaction between antibiotics and membrane transporters in other clinically relevant bacterial species.

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

Antibiotics were sourced from Sigma, apart from rifampicin, ceftriaxone and azithromycin, which were purchased from Tokyo Chemical Industries (Zwijndrecht, Belgium).

Plates were prepared by first inoculating deep-well plates with 1 mL Lucia Broth (LB) from glycerol stocks of the transporter library. The inoculated LB was grown overnight under agitation at 37 ◦C. Following overnight growth, the cultures were mixed with 1 mL 50% glycerol. CR1496c polystyrene plates (Enzyscreen, NL, Heemstede, The Netherlands) were prepared for growth assays, by dispensing a 3 µL droplet of the culture and glycerol mixture into the bottom of the well. These loaded plates were then stored at −20 ◦C for up to 3 weeks or −80 ◦C for longer storage.

The growth assays were initiated by adding 297 µL LB, containing the appropriate dose of antibiotics to the pre-inoculated plates, and sealing with CR1396b Sandwich covers (Enzyscreen, NL, Heemstede). Growth was assessed using the Growth Profiler 960 (Enzyscreen, NL, Heemstede), which uses camera-based measurements to estimate growth rates simultaneously in up to 10 96-well plates. As our library encompassed 5 plates, this

allowed us to run the full library in duplicate. The Growth Profiler 960 was set to 37 ◦C with 225 rpm shaking (recommended settings for *E. coli*), with pictures taken every 20 min.

Inoculation and media loading were performed using an Opentrons OT2 robot fitted with a 20 µL multichannel pipette and a 300 µL multichannel pipette. Scripts used in operation can be found at github.com/ljm176/TransporterScreening (11 August 2022). Growth plates were sterilized between uses by washing and UV in accordance with the instructions of the manufacturer.

*G*-values were obtained from the plate images using the manufacturer's software. *G*-values were converted to *OD*600 values using the formula:

$$OD\_{600} = a \ast \left(G\_{\text{value}} - G\_{Blank}\right)^b$$

With the predetermined values *a* = 0.0158 and *b* = 0.9854, which were found by measurement of a standard curve in accordance with the instructions of the manufacturer.

Data analysis and generation of figures was performed in R (version 4.1.2). Growth rates were determined by using the R package Growthcurver [43]. Growthcurver fits growth curves to the equation:

$$N\_t = \frac{K}{1 + \left(\frac{K - N0}{N0}\right)e^{-rt}}$$

where *N<sup>t</sup>* is total cell population, *K* is the carrying capacity, initial cell population is *N*0 and *r* is the intrinsic growth rate. For strains that showed growth in only a single replicate, the replicate without growth was filtered from analysis, as this was determined to be the result of missed inoculation during automated loading.

Figures were generated using ggplot2 for R. The R script used in data analysis and figure generation is available at github.com/ljm176/TransporterScreening (11 August 2022).

**Supplementary Materials:** The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/antibiotics11081129/s1. Figure S1: Workflow for high throughput screening of transporter knockout library against antibiotics. Figure S2: Structures of antibiotics used in this study; Table S1: Growth Parameters.

**Author Contributions:** Conceptualization L.J.M. and D.B.K.; study design and experimental work L.J.M.; writing, review and editing L.J.M. and D.B.K.; funding acquisition, D.B.K. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the University of Liverpool and Novo Nordisk Foundation (grant NNF20CC0035580).

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

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Datasets are available in the Supplementary Materials for this paper. The code used in analysis is available at github.com/ljm176/TransporterScreening.

**Acknowledgments:** We thank the University of Liverpool and the Novo Nordisk Foundation (grant NNF20CC0035580) for the financial support.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript; or in the decision to publish the results.

#### **References**


## *Article* **Effects of Levofloxacin, Aztreonam, and Colistin on Enzyme Synthesis by** *P. aeruginosa* **Isolated from Cystic Fibrosis Patients**

**Arianna Pani \* , Valeria Lucini, Silvana Dugnani, Alice Schianchi and Francesco Scaglione**

Oncology and Hemato-Oncology, University of Milan, 20122 Milan, Italy

**\*** Correspondence: arianna.pani@unimi.it

**Abstract:** (1) Background: Cystic fibrosis (CF) is characterized by chronic pulmonary inflammation and persistent bacterial infections. *P. aeruginosa* is among the main opportunistic pathogens causing infections in CF. *P. aeruginosa* is able to form a biofilm, decreasing antibiotic permeability. LOX, a lipoxygenase enzyme, is a virulence factor produced by *P. aeruginosa* and promotes its persistence in lung tissues. The aim of this study is to evaluate if antibiotics currently used for aerosol therapy in CF are able to interfere with the production of lipoxygenase from open isolates of *P. Aeruginosa* from patients with CF. (2) Methods: Clinical isolates of *P. aeruginosa* from patients with CF were grown in Luria broth (LB). Minimum inhibitory concentration (MIC) was performed and interpreted for all isolated strains according to the European Committee on Antimicrobial Susceptibility Testing (EUCAST) guidelines. We selected four antibiotics with different mechanisms of action: aztreonam, colistin, amikacin, and levofloxacin. We used human pulmonary epithelial NCI-H929 cells to evaluate LOX activity and its metabolites according to antibiotic action at increasing concentrations. (3) Results: there is a correlation between LOX secretion by clinical isolates of *P. aeruginosa* and biofilm production. Levofloxacin exhibits highly significant inhibitory activity compared to the control. Amikacin also exhibits significant inhibitory activity against LOX production. Aztreonam and colistin do not show inhibitory activity. These results are also confirmed for LOX metabolites. (4) Conclusions: among the evaluated antibiotics, levofloxacin and amikacin have an activity on LOX secretion.

**Keywords:** cystic fibrosis; *P. aeruginosa*; antibiotics

#### **1. Introduction**

Bacterial infection and linked chronic pulmonary inflammation are a pathological trait associated with the genetic disease cystic fibrosis (CF) [1]. These bacterial infections are characterized by robust inflammatory responses, with an elevation in the levels of proinflammatory cytokines and neutrophil accumulation in the CF airway. Unfortunately, responses triggered by the inflammation are not effective in clearing pathogenic microbes in the CF lung [2], creating instead a hyperinflammatory status which can lead to the damage of host tissues and respiratory failure, leading to transplantation or death [3].

Most adult patients with CF have a chronic infection of the airways caused by the opportunistic bacterial pathogen *P. aeruginosa*, which is frequently associated with morbidity and mortality. *P. aeruginosa* grows in the context of hyperinflammation in CF lungs and is able to form mechanically robust biofilms which are resistant to clinically achievable levels of antibiotics [4].

Antibiotic classes currently approved in many countries for use by inhalation include the aminoglycosides (tobramycin and amikacin), monobactams (aztreonam), polymyxins (colistimethate) and fluoroquinolones (levofloxacin). Although all of the antibiotics used are beneficial, none of them are capable of completely eradicating *P. aeruginosa* from bronchial secretions in CF. However, relatively recent studies have reported that levofloxacin, administered by aerosol in CF, produces positive effects that go beyond its simple antibacterial effect [5–7].

**Citation:** Pani, A.; Lucini, V.; Dugnani, S.; Schianchi, A.; Scaglione, F. Effects of Levofloxacin, Aztreonam, and Colistin on Enzyme Synthesis by *P. aeruginosa* Isolated from Cystic Fibrosis Patients. *Antibiotics* **2022**, *11*, 1114. https://doi.org/10.3390/ antibiotics11081114

Academic Editors: Dóra Kovács and Jeffrey Lipman

Received: 8 June 2022 Accepted: 9 August 2022 Published: 17 August 2022

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

**Copyright:** © 2022 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/).

*P. aeruginosa* also persists in the airways by interfering with host defense via secreted bacterial virulence factors and small molecules. Recently, it has been demonstrated that several *P. aeruginosa* clinical isolates express *LoxA*, a gene which encodes for the lipoxygenase enzyme (LOX) that oxidizes polyunsaturated fatty acids [8]. In the lungs, LOX is able to process a wide range of host polyunsaturated fatty acids, with a consequent production of bioactive lipid mediators (including lipoxin A4). LOX is also able to inhibit major chemokine expression, such as macrophage inflammatory proteins (MIPs) and keratinocytes-derived chemokines (KCs), and to recruit leukocytes. Importantly, LOX is able to promote *P. aeruginosa* survival in lung tissues, suggesting a LOX-dependent interference between the host lipid pathways and *P. aeruginosa* lung pathogenesis.

In this study, we wanted to verify whether the antibiotics currently used via aerosols in CF could interfere with the production of LOX from isolates of *P. aeruginosa* from patients with CF.

#### **2. Results**

#### *2.1. Antibiotic Intrinsic Activity (MICs)*

In order to evaluate the activity of antibiotics approved for aerosol use in cystic fibrosis on the production of LOX by *P. aeruginosa*, we chose four antibiotics representing different mechanisms of action: levofloxacin, amikacin, aztreonam, and colistin. First, we evaluated the antibiotic intrinsic activity against 12 clinically isolated strains of *P. aeruginosa*.

Table 1 shows the minimum inhibitory concentrations (MICs) of the levofloxacin, amikacin, aztreonam, and colistin studied against the clinically collected strains of *P. aeruginosa*. Strains 1, 2, 3, 9, 10, 11, and 12 were resistant to levofloxacin (MIC > 2). Only strain 12 was resistant to amikacin (MIC > 16) and strains 9 and 10 to aztreonam (MIC > 16). Strains 1, 2, and 11 were resistant to colistin (MIC > 4). Low MICs (≤0.25 mg/L) were detected for levofloxacin, amikacin, aztreonam, and colistin in strains 4 and 8.


**Table 1.** MICs of levofloxacin, amikacin, aztreonam and colistin against *P. aeruginosa strains isolated from patients with cystic fibrosis*.

#### *2.2. Biofilm Formation*

Secondly, we evaluated the ability of the 12 clinically isolated strains to produce biofilm. Figure 1 shows the biofilm formation by the 12 selected strains. As expected, there are different abilities to form biofilms. Strains 2, 4, 6, and 10 demonstrated the highest ability to produce biofilms at 12, 24, and 48 h. Strain 2 showed a CV absorbance of 0.620 (0.042) at 12 h, 1.040 (0.085) at 24 h, and 1.600 (0.141) at 48 h. Strain 4 showed a CV absorbance of 0.440 (0.127) at 12 h, 0.980 (0.170) at 24 h, and 1.600 (0.141) at 48 h. Strain 6 showed a CV absorbance of 0.400 (0.014) at 12 h, 0.995 (0.148) at 24 h, and 1.700 (0.141) at 48 h. Strain 10 showed a CV absorbance of 0.445 (0.134) at 12 h, 1.150 (0.636) at 24 h, and 2.020 (0.113) at 48 h.

Figure 1 shows the biofilm formation by the 12 selected strains. As expected, there are different abilities to form biofilms. Strains 2, 4, 6, and 10 demonstrated the highest ability to produce biofilms at 12, 24, and 48 h. Strain 2 showed a CV absorbance of 0.620 (0.042) at 12 h, 1.040 (0.085) at 24 h, and 1.600 (0.141) at 48 h. Strain 4 showed a CV absorbance of 0.440 (0.127) at 12 h, 0.980 (0.170) at 24 h, and 1.600 (0.141) at 48 h. Strain 6 showed a CV absorbance of 0.400 (0.014) at 12 h, 0.995 (0.148) at 24 h, and 1.700 (0.141) at 48 h. Strain 10 showed a CV absorbance of 0.445 (0.134) at 12 h, 1.150 (0.636) at 24 h, and 2.020

**Figure 1. Ability to form** biofilm for the 12 clinically isolated *P. aeruginosa* strains. Biofilm was evaluated at 12, 24, and 48 h. CV absorbance measures biofilm formation. Each colored line (numbers **Figure 1. Ability to form** biofilm for the 12 clinically isolated *P. aeruginosa* strains. Biofilm was evaluated at 12, 24, and 48 h. CV absorbance measures biofilm formation. Each colored line (numbers from 1 to 12, see Table 1) refers to a different *P. aeruginosa* strain.

#### from 1 to 12, see Table 1) refers to a different *P. aeruginosa* strain. *2.3. LOX Activity*

(0.113) at 48 h.

*2.3. LOX Activity*  Since the expression of the loxA gene differs from one *P. aeruginosa* strain to another, especially in biofilm growth conditions [9], we measured the LOX activity of the 12 clini-Since the expression of the loxA gene differs from one *P. aeruginosa* strain to another, especially in biofilm growth conditions [9], we measured the LOX activity of the 12 clinically isolated strains. The aim of our evaluations of biofilm and LOX activity was to select the best strain on which to continue our evaluation on the impact of antibiotics on LOX production.

cally isolated strains. The aim of our evaluations of biofilm and LOX activity was to select the best strain on which to continue our evaluation on the impact of antibiotics on LOX production. Figure 2 shows the LOX activity of the 12 selected strains. From the comparison of Figures 1 and 2 it emerges that the strains with the greatest ability to make biofilms are those that have the greatest LOX activity. By performing a one-way ANOVA with Bonferroni's correction, the LOX activity differs statistically significantly between the 12 strains (*p* < 0.001). In fact, strains 2, 4, 6, and 10 demonstrated the highest LOX activity at 12, 24, and 48 h. Mean absorbance of strain 2 was 1.040 (0.085) at 12 h, 1.200 (0.141) at 24 Figure 2 shows the LOX activity of the 12 selected strains. From the comparison of Figures 1 and 2 it emerges that the strains with the greatest ability to make biofilms are those that have the greatest LOX activity. By performing a one-way ANOVA with Bonferroni's correction, the LOX activity differs statistically significantly between the 12 strains (*p* < 0.001). In fact, strains 2, 4, 6, and 10 demonstrated the highest LOX activity at 12, 24, and 48 h. Mean absorbance of strain 2 was 1.040 (0.085) at 12 h, 1.200 (0.141) at 24 h, and 1.450 (0.071) at 48 h. Mean absorbance of strain 4 was 1.350 (0.212) at 12 h, 1.500 (0.141) at 24 h, and 1.550 (0.212) at 48 h. Mean absorbance of strain 6 was 1.350 (0.212) at 12 h, 1.450 (0.212) at 24 h, and 1.500 (0.141) at 48 h. Mean absorbance of strain 10 was 1.550 (0.071) at 12 h, 1.400 (0.283) at 24 h, and 2.050 (0.212) at 48 h.

h, and 1.450 (0.071) at 48 h. Mean absorbance of strain 4 was 1.350 (0.212) at 12 h, 1.500 (0.141) at 24 h, and 1.550 (0.212) at 48 h. Mean absorbance of strain 6 was 1.350 (0.212) at Considering these results, the number 10 strain of *P. aeruginosa* was chosen to perform the subsequent experiments.

12 h, 1.450 (0.212) at 24 h, and 1.500 (0.141) at 48 h. Mean absorbance of strain 10 was 1.550

Considering these results, the number 10 strain of *P. aeruginosa* was chosen to per-

form the subsequent experiments.

(0.071) at 12 h, 1.400 (0.283) at 24 h, and 2.050 (0.212) at 48 h.

*Antibiotics* **2022**, *11*, x FOR PEER REVIEW 4 of 10

**Figure 2.** Lipoxygenase (LOX) activity of the 12 selected *P. aeruginosa* strains. LOX activity was evaluated at 12, 24, and 48 h. Each colored line (numbers from 1 to 12, see Table 1) refers to a different **Figure 2.** Lipoxygenase (LOX) activity of the 12 selected *P. aeruginosa* strains. LOX activity was evaluated at 12, 24, and 48 h. Each colored line (numbers from 1 to 12, see Table 1) refers to a different *P. aeruginosa* strain. MIC—0.700 (0.141) vs. 1.650 (0.071)), while aztreonam and colistin do not show inhibitory activity (mean absorbance at four times the MIC, respectively 1.250 (0.071) and 1.200 (0.141)). It should be noted that at eight times the MIC all antibiotics show inhibitory ac-

#### *P. aeruginosa* strain. *2.4. Effects of Antibiotics on LOX Activity* tivity. This is most likely related to bactericidal activity on the strain rather than to the

*2.4. Effects of Antibiotics on LOX Activity*  We then evaluated the effect of the levofloxacin, aztreonam, amikacin, and colistin on the LOX activity of strain 10 of *P. aeruginosa*. Results are shown in Figure 3. activity inhibiting LOX synthesis.

**Figure 3.** Effects of levofloxacin, aztreonam, amikacin, and colistin on the LOX activity of a clinically isolated strain of *P. aeruginosa* with high LOX activity. Antibiotics were tested at increasing **Figure 3.** Effects of levofloxacin, aztreonam, amikacin, and colistin on the LOX activity of a clinically isolated strain of *P. aeruginosa* with high LOX activity. Antibiotics were tested at increasing concentrations (0.5- to 8-fold the MICs). (LEVO = levofloxacin, AZT = aztreonam, AMIKA = amikacin, COL = colistin, Control = untreated strain).

**Figure 3.** Effects of levofloxacin, aztreonam, amikacin, and colistin on the LOX activity of a clinically isolated strain of *P. aeruginosa* with high LOX activity. Antibiotics were tested at increasing

Levofloxacin exhibits highly significant inhibitory activity compared to the control (mean absorbance at four times the MIC—0.250 (0.071) vs. 1.650 (0.071)). To a lesser extent, amikacin also exhibits significant inhibitory activity (mean absorbance at four times the MIC—0.700 (0.141) vs. 1.650 (0.071)), while aztreonam and colistin do not show inhibitory activity (mean absorbance at four times the MIC, respectively 1.250 (0.071) and 1.200 (0.141)). It should be noted that at eight times the MIC all antibiotics show inhibitory activity. This is most likely related to bactericidal activity on the strain rather than to the activity inhibiting LOX synthesis.

We compared the effects of the four different antibiotics on LOX activity. Table 2 reports the results of the analysis. Differences between levofloxacin and aztreonam, colistin and aztreonam, and colistin were statistically significant (Table 2).


**Table 2.** One-way ANOVA effects of different antibiotics on LOX activity.

#### *2.5. Production of 15-LOX-Dependent Metabolites in Lung Epithelial Cells Infected by P. aeruginosa*

To evaluate LOX effects on the host response, we measured LOX metabolites in human lung epithelial NCI-H292 cells challenged by *P. aeruginosa* strain 10. We measured 15-Hydroxyeicosatetraenoic acid (15-HETE), 17-hydroxydocosahexaenoic acid (17-HDoHE) and lipoxin A4 production (LXA4). Figure 4 shows the results relating to the effect of the various antibiotics on the production of 15-HETE. Levofloxacin exhibits significant inhibitory activity compared to the control (mean difference −112.1, 95% CI 27.99–196.2, *p* < 0.005), while amikacin aztreonam, and colistin show no significant inhibitory activity. In this case it should also be noted that at eight times the MIC all antibiotics show inhibitory activity. This is most likely related to bactericidal activity on the strain rather than to the activity inhibiting LOX synthesis.

Figure 5 shows the results relating to the effect of the various antibiotics on the production of 17-HDoHE. Only levofloxacin exhibits significant inhibitory activity compared to the control (mean difference −112 95% CI 55.4–514.8, *p* < 0.005), while amikacin, aztreonam, and colistin show no significant inhibitory activity. In this case it should also be noted that at eight times the MIC all antibiotics show inhibitory activity.

Figure 6 shows the results relating to the effect of various antibiotics on LXA44 production. Only Levofloxacin exhibits significant inhibitory activity compared to the control (mean difference −794.9, 95%CI 225–1364, *p* < 0.005). While amikacin aztreonam and colistin show no significant inhibitory activity. In this case it should also be noted that at eight times the MIC all antibiotics show inhibitory activity.

concentrations (0.5- to 8-fold the MICs). (LEVO = levofloxacin, AZT = aztreonam, AMIKA = amika-

istin and aztreonam, and colistin were statistically significant (Table 2).

**Table 2.** One-way ANOVA effects of different antibiotics on LOX activity.

We compared the effects of the four different antibiotics on LOX activity. Table 2 reports the results of the analysis. Differences between levofloxacin and aztreonam, col-

**parison Test Mean Difference t** *p* **< 0.05 95% CI** 

Levofloxacin vs. aztreonam −0.7200 4.237 Yes −1.273 to −0.1673 Levofloxacin vs. amikacin −0.4800 2.824 No −1.033 to 0.07266 Levofloxacin vs. colistin −0.6680 3.931 Yes −1.221 to −0.1153 Levofloxacin vs. control −1.180 6.943 Yes −1.733 to −0.6273 Aztreonam vs. amikacin 0.2400 1.412 No −0.3127 to 0.7927 Aztreonam vs. colistin 0.05200 0.3060 No −0.5007 to 0.6047 Aztreonam vs. control −0.4600 2.707 No −1.013 to 0.09266 Amikacin vs. colistin −0.1880 1.106 No −0.7407 to 0.3647 Amikacin vs. control −0.7000 4.119 Yes −1.253 to −0.1473 Colistin vs. control −0.5120 3.013 No −1.065 to 0.04066

*2.5. Production of 15-LOX-Dependent Metabolites in Lung Epithelial Cells Infected by P. aeru-*

To evaluate LOX effects on the host response, we measured LOX metabolites in human lung epithelial NCI-H292 cells challenged by *P. aeruginosa* strain 10. We measured 15-Hydroxyeicosatetraenoic acid (15-HETE), 17-hydroxydocosahexaenoic acid (17- HDoHE) and lipoxin A4 production (LXA4). Figure 4 shows the results relating to the effect of the various antibiotics on the production of 15-HETE. Levofloxacin exhibits significant inhibitory activity compared to the control (mean difference −112.1, 95% CI 27.99– 196.2, *p* < 0.005), while amikacin aztreonam, and colistin show no significant inhibitory activity. In this case it should also be noted that at eight times the MIC all antibiotics show inhibitory activity. This is most likely related to bactericidal activity on the strain rather

cin, COL = colistin, Control = untreated strain).

than to the activity inhibiting LOX synthesis.

**Bonferroni's Multiple Com-**

*ginosa* 

**Figure 4.** Concentration of 15-HETE (pg/mg of protein) in extracts of human lung epithelial NCI-H292 cells infected or noninfected with *P. aeruginosa*, 24 h post-infection. (LEVO = levofloxacin, AZT = aztreonam, COL = colistin, Control inf = control in infected cells, Control non inf = control in noninfected cells). duction of 17-HDoHE. Only levofloxacin exhibits significant inhibitory activity compared to the control (mean difference −112 95% CI 55.4–514.8, *p* < 0.005), while amikacin, aztreonam, and colistin show no significant inhibitory activity. In this case it should also be noted that at eight times the MIC all antibiotics show inhibitory activity.

**Figure 5.** Concentration of 17-HDoHE (pg/mg of protein) in extracts of human lung epithelial NCI-H292 cells non-infected and treated with antibiotics, 24 h. (LEVO = levofloxacin, AZT = aztreonam, AMIKA = amikacin, COL = colistin, Control inf = control in infected cells, Control non-inf = control in noninfected cells) **Figure 5.** Concentration of 17-HDoHE (pg/mg of protein) in extracts of human lung epithelial NCI-H292 cells non-infected and treated with antibiotics, 24 h. (LEVO = levofloxacin, AZT = aztreonam, AMIKA = amikacin, COL = colistin, Control inf = control in infected cells, Control noninf = control in noninfected cells).

Figure 6 shows the results relating to the effect of various antibiotics on LXA44 pro-

istin show no significant inhibitory activity. In this case it should also be noted that at

eight times the MIC all antibiotics show inhibitory activity.

**Figure 6.** Concentration of LOXa4 (pg/mg of protein) in extracts of human lung epithelial NCI-H292 cells infected, noninfected, and treated with antibiotics after 24 h. (LEVO = levofloxacin, AZT = aztreonam, AMIKA = amikacin, COL = colistin, Control inf = control in infected cells, Control non inf **Figure 6.** Concentration of LOXa4 (pg/mg of protein) in extracts of human lung epithelial NCI-H292 cells infected, noninfected, and treated with antibiotics after 24 h. (LEVO = levofloxacin, AZT = aztreonam, AMIKA = amikacin, COL = colistin, Control inf = control in infected cells, Control non inf = control in noninfected cells).

#### = control in noninfected cells). **3. Discussion**

**3. Discussion**  *Pseudomonas aeruginosa* is a highly versatile bacterium. One basis for its versatility is the arsenal of enzymes that helps this pathogen to adapt to its environment. Our study shows that lipoxygenase (LOX) is secreted by clinical isolates of *P. aeruginosa*, producing biofilm, and this enzyme may contribute to the lung pathogenesis triggered by this opportunistic pathogen. In a series of elegant experiments, Morello et al. [8] have shown that LOX activity decreases release of chemokines such as KC (CXCL-1) and macrophage inflammatory proteins (MIP-1α/CCL-3, MIP-1β/CCL-4, and MIP-2/CXCL-2), producing a *Pseudomonas aeruginosa* is a highly versatile bacterium. One basis for its versatility is the arsenal of enzymes that helps this pathogen to adapt to its environment. Our study shows that lipoxygenase (LOX) is secreted by clinical isolates of *P. aeruginosa*, producing biofilm, and this enzyme may contribute to the lung pathogenesis triggered by this opportunistic pathogen. In a series of elegant experiments, Morello et al. [8] have shown that LOX activity decreases release of chemokines such as KC (CXCL-1) and macrophage inflammatory proteins (MIP-1α/CCL-3, MIP-1β/CCL-4, and MIP-2/CXCL-2), producing a lower recruitment of immune cells in the airspaces. More importantly, they also observed that LOX activity promotes the spread of bacteria in lung tissues.

lower recruitment of immune cells in the airspaces. More importantly, they also observed that LOX activity promotes the spread of bacteria in lung tissues. The role of 15-LOX has been implicated in various inflammation-related diseases. Increasing evidence highlights the controversial nature of 12/15-LOX in inflammation, as its metabolites have been shown to have both pro- and anti-inflammatory properties [10].

The role of 15-LOX has been implicated in various inflammation-related diseases. Increasing evidence highlights the controversial nature of 12/15-LOX in inflammation, as its metabolites have been shown to have both pro- and anti-inflammatory properties [10]. The pro-inflammatory role of 15-LOX and its metabolite 15(S)-HETE was demonstrated by various studies [11,12]. It is important to note that expression of pro-inflammatory cytokines IL-6, IL-12, CXCL9, and CXCL10, which is LPS-induced, is reduced by inhibition of 12/15-LOX in macrophages [13]. Furthermore, 15- LOX is able to regulate the expression of pro-inflammatory eoxins in epithelial airway cells and eosinophils. Eoxins can cause endothelial cell dysfunction and enhance vascular permeability [14]. Addition-The pro-inflammatory role of 15-LOX and its metabolite 15(S)-HETE was demonstrated by various studies [11,12]. It is important to note that expression of pro-inflammatory cytokines IL-6, IL-12, CXCL9, and CXCL10, which is LPS-induced, is reduced by inhibition of 12/15-LOX in macrophages [13]. Furthermore, 15-LOX is able to regulate the expression of pro-inflammatory eoxins in epithelial airway cells and eosinophils. Eoxins can cause endothelial cell dysfunction and enhance vascular permeability [14]. Additionally, the increased expression of some of these inflammatory molecules is dependent on nuclear factor κB activation [15,16]. Moreover, recently it has been reported that 15-LOX may interact with secretory phospholipase A2 (sPLA2), contributing to sterile inflammation in chronic conditions with differences from classical inflammation on the cytokine level [17].

ally, the increased expression of some of these inflammatory molecules is dependent on nuclear factor κB activation [15,16]. Moreover, recently it has been reported that 15-LOX may interact with secretory phospholipase A2 (sPLA2), contributing to sterile inflammation in chronic conditions with differences from classical inflammation on the cytokine level [17]. On the other hand, demonstrations of the anti-inflammatory properties of 12/15-LOX and its metabolites have been reported. The exact mechanisms by which 12/15-LOX explicates its anti-inflammatory effects are not fully understood, but they may be due to its pro-resolving mediators such as lipoxins, resolvins, and protectins, which are able to induce a potent and direct anti-inflammatory response in various cell types. In summary, it is becoming clear that 12/15-LOX and its metabolites have both pro- and anti-inflammatory

On the other hand, demonstrations of the anti-inflammatory properties of 12/15-LOX and its metabolites have been reported. The exact mechanisms by which 12/15-LOX explicates its anti-inflammatory effects are not fully understood, but they may be due to its

duce a potent and direct anti-inflammatory response in various cell types. In summary, it is becoming clear that 12/15-LOX and its metabolites have both pro- and antieffects, and some of these differential effects of the same metabolites could be due to their different concentrations [10].

In this study, we analyzed the activity of three antibiotics currently used in cystic fibrosis as antipseudomonal inhalation therapy. Among the antibiotics used via aerosol in CF, only levofloxacin exhibited significant inhibitory activity of LOX activity compared to the control, while aztreonam and colistin show no significant inhibitory activity. Levofloxacin acting on the synthesis of enzymes by *P. aeruginosa* determines an indirect anti-inflammatory effect, which is very useful in CF. These data may explain the positive effects exceeding those attributable to the antibacterial activity alone, which have been obtained with levofloxacin during clinical trials in CF [6,18].

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

#### *4.1. Bacterial Strains*

Nonduplicate clinical isolates of *P. aeruginosa* from a permanent collection for the storage of bacterial strains of unrecognizable patients with CF were grown in Luria broth (LB) (Sigma-Aldrich, Milan, Italy). Minimum inhibitory concentrations (MICs) were determined using the reference broth microdilution methodology according to the Clinical and Laboratory Standards Institute (CLSI) guidelines [19] for levofloxacin, colistin, aztreonam, and amikacin for all isolated strains. EUCAST v.12 clinical breakpoints were used for interpretation of susceptibility data, where available. After 18 h of incubation at 37 ◦C, the MICs of the aforementioned antibiotics were defined by no visible growth on plates. Since it has been shown that LOX is mainly produced in Pseudomonas strains that produce biofilms, we have preliminarily evaluated the ability of selected strains to produce biofilm [20].

#### *4.2. Biofilm Quantification*

One microliter of a late-log-phase culture was added to 99 µL LB in a 96-well microtiter plate, incubated for 10 h at 37 ◦C. Biofilm production was quantified by measuring the absorbance of crystal violet (Sigma-Aldrich-, Italy). Biofilms were fixed by heat at 60 ◦C for about 1 h for the crystal violet assay. Subsequently, wells with 150 µL of crystal violet solution (2.3% prepared in 20% ethanol) were incubated for 15 min at room temperature. Excesses of crystal violet were removed by washing and then 200 µL of 33% glacial acetic acid was used to dissolve dye fixed to the biofilm and it was incubated for 1 h at room temperature. CV absorbance was measured at 570 nm using a microplate spectrophotometer (BioTek™ Epoch 2 Microplate Spectrophotometer; BioTek Instruments, Winooski, VT, USA). Biofilm formation was evaluated at 12, 24, and 48 h.

#### *4.3. Antibiotics*

To evaluate the effect of antibiotics, we decided to select three antibiotics with different mechanisms of action. Aztreonam inhibits bacterial cell wall formation, colistin interferes with membrane phospholipids, and levofloxacin inhibits protein synthesis. Aztreonam, colistin, and levofloxacin were purchased from the market (colistin sulfate, aztreonam and levofloxacin for microbiological assay, European Pharmacopoeia (EP) Reference Standard, Sigma Aldrich S.r.l., St. Louis, MO, USA). Antibiotic activity was evaluated against LOX activity and lipid production in a medium containing antibiotics at increasing concentrations (0.5- to 8-fold the MICs in broth).

#### *4.4. Lipoxygenase Assay*

*P. aeruginosa* strains were grown as single colonies on LB agar overnight. Then, a single colony was inoculated in 10 mL of LB broth and grown to stationary phase under static conditions at 28 ◦C. Subsequently the culture broth was centrifuged at 8000× *g* for 15 min and the supernatant was collected in a syringe, sterilized, and stored at −80 ◦C. 10 mL of sample (concentrated supernatants) was mixed with 100 mL of solution A (0.5 mM purified lipoxygenase assay, 10 mM (dimethyl-amino)-benzoic acid (DMAB) (Sigma-Aldrich-Italy) prepared in 100 mM phosphate buffer (pH 6) and incubated for 20 min before the addition

of 100 mL of supernatant from each well was transferred into a new plate, and their absorbance was measured at 598 nm.

#### *4.5. Culture and Infection of Pulmonary Epithelial Cells*

Human pulmonary epithelial NCI-H929 cells were obtained from the American Type Culture Collection (ATCC, Manassas, VA, USA). Cells were grown in RPMI 1640 Glutamax (Gibco, Life Technologies, Rodano, Italy) medium supplemented with 10% heat-inactivated Fetal Bovine Serum (FBS) (Sigma-Aldrich-Italy) in a humidified incubator with 5% CO<sup>2</sup> at 37 C. For infection experiments, cells were cultivated in 24-well plates until confluence (5.105 cells per well). Exponential growth phase bacteria (LB, 37C, 180 rpm, OD 600 nm) were washed twice in ice-cold PBS before addition to freshly dispensed cell culture medium to obtain MOI = 0.1. After 20 h, supernatants were then centrifuged at 8000× *g* for 10 min, collected in a syringe, and sterilized with a 0.22 mm polymer filter before immediate snap-freezing and storage in liquid nitrogen until lipid mediator extraction.

#### *4.6. Lipid Extraction and Liquid Chromatography/Tandem Mass Spectrometry (LC-MS/MS)*

We then evaluated the 15-LOX-dependent metabolites, 15-hydroxy-octadecadienoic acid (15-HETE) and 17-HDoHE, and lipoxin A4 (LXA4) in cell lysates and supernatants. Solid phase extraction was performed with HRX-50 mg 96-well plates. Simultaneous separation of the lipids of interest was performed as well as LC-MS/MS analysis on an ultra-high-performance liquid chromatography system (Q Exactive™ Plus Hybrid Quadrupole-Orbitrap™ Mass Spectrometer Thermo Scientific™, Waltham, MA, USA).

#### *4.7. Statistical Methods*

Continuous variables are expressed as mean and standard deviations (sd). To compare the effects on LOX, 15-HETE, 17-HDoHE, and LOXa4 activity according to the different antibiotics at increasing concentrations, we used a two-way analysis of variance (ANOVA) followed by post hoc comparison using Bonferroni *t*-test. *p*-values of <0.05 were considered significant.

Statistical analyses were performed using GraphPad InStat 8 (GraphPad Software Inc., La Jolla, CA, USA).

#### **5. Conclusions**

Among clinically isolated strains of *P. aeruginosa* from patients with CF, we selected the strain with highest activity of biofilm production and LOX activity.

Testing levofloxacin, aztreonam, colistin, and amikacin at increasing concentrations, only levofloxacin showed a significant activity on the secretion of LOX. Levofloxacin also showed a significant activity on 15-HETE, 17-HDoHE, and LXA4 production. In conclusion, levofloxacin is the only antibiotic, among those studied according to the mechanism of action and the approval for aerosol use, which can impact the secretion of LOX in a strain with a high ability of biofilm production.

**Author Contributions:** Conceptualization, A.P. and F.S.; methodology, F.S., V.L. and S.D.; formal analysis, F.S.; investigation, V.L.; data curation, S.D.; writing—original draft preparation, A.P., F.S. and A.S.; writing—review and editing, A.P. and A.S.; visualization, A.S.; supervision, F.S.; project administration, F.S.; funding acquisition, F.S. All authors have read and agreed to the published version of the manuscript.

**Funding:** An unrestricted grant has been provided by Chiesi Farmaceutici S.p.A to partially fund research activities. Grant number is not applicable.

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

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Data available on request from the authors upon reasonable request.

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

## **References**

