The search for novel treatment strategies for Streptococcus pneumoniae infections

ABSTRACT This review provides an overview of the most important novel treatment strategies against Streptococcus pneumoniae infections published over the past 10 years. The pneumococcus causes the majority of community-acquired bacterial pneumonia cases, and it is one of the prime pathogens in bacterial meningitis. Over the last 10 years, extensive research has been conducted to prevent severe pneumococcal infections, with a major focus on (i) boosting the host immune system and (ii) discovering novel antibacterials. Boosting the immune system can be done in two ways, either by actively modulating host immunity, mostly through administration of selective antibodies, or by interfering with pneumococcal virulence factors, thereby supporting the host immune system to effectively overcome an infection. While several of such experimental therapies are promising, few have evolved to clinical trials. The discovery of novel antibacterials is hampered by the high research and development costs versus the relatively low revenues for the pharmaceutical industry. Nevertheless, novel enzymatic assays and target-based drug design, allow the identification of targets and the development of novel molecules to effectively treat this life-threatening pathogen.

1. Analogues of already marketed antibiotics are in development to overcome increasing pneumococcal resistance. Several of these are currently undergoing clinical trials. 2. Antibacterial targets, consisting of pneumococcal enzymes essential for viability and survival, are being identified. Target-specific rational drug design, including highthroughput screening (HTS) against a major enzyme followed by elucidation of the structure-activity relationship (SAR) and subsequent lead optimization, shows promise in creating novel antibacterials. Contrarily, discovery of natural antibacterials is being hampered by the difficult growth conditions for microbes producing antibacterials.

INTRODUCTION
Streptococcus pneumoniae, also called the pneumococcus, is a major human pathogen. It is the leading cause of communityacquired bacterial pneumonia (CABP) and can cause otitis media (OM) and meningitis in children, the elderly and immunocompromised patients (Lundbo and Benfield 2017;Peyrani et al. 2019).
In the USA, pneumonia was the eight leading cause of death in 2015 (Jindal et al. 2015). While incidence rates of CABP vary worldwide, overall incidences between 20 and 100 per 10 000 person-years have been observed, with outliers up to 164.3 per 10 000 person-years in patients older than 80 in the USA and up to 294 per 10 000 person-years in Latin-America (Ferreira-Coimbra, Sarda and Rello 2020). For acute OM, 300 million cases are estimated to be caused by S. pneumoniae every year (Monasta et al. 2012;Bergenfelz and Hakansson 2017).
In the USA, current therapy for pneumococcal infections in infants and children typically consists of amoxicillin or, in case of non-IgE-mediated allergy, a cephalosporin. Alternatively, levofloxacin, linezolid, clindamycin or vancomycin can be used (Bradley et al. 2011). Treatment of CABP in adults is generally done with amoxicillin, doxycycline or a macrolide. In case of comorbidities, a fluoroquinolone or combination therapies such as amoxicillin and a cephalosporin or a macrolide and doxycycline are advised by the America Thoracic Society and Infectious Diseases Society of America (Metlay et al. 2019). A similar therapy schedule is proposed for patients with pneumococcal meningitis (van de Beek et al. 2016). In Europe, amoxicillin or a tetracycline is primarily advised in patients with lower respiratory tract infections, while macrolides can only be used in countries with low resistance rates. Alternatively, levofloxacin or moxifloxacin may be considered for general use, while cephalosporins are reserved for hospital use only (Woodhead et al. 2011;Wiersinga et al. 2018). While therapy is in general successful, antibiotic resistance is increasingly observed. According to the United States Centre for Disease Control (CDC), resistance to one or more antibiotics is observed in 30% of pneumococcal infections, and it is estimated that 1.2 million infections per year are caused by resistant pneumococcal strains (Centre for Disease Control (CDC) 2013; Cherazard et al. 2017). Similarly in Europe, 10% of invasive S. pneumoniae isolates reported in 2008 were not susceptible to penicillin, and large regional differences in pneumococcus prevalence have been observed, from less than 5% in Northern Europe to over 40% in some Southern European countries (Woodhead et al. 2011). In 2018, these numbers were confirmed. Resistance occurred in 2.5%-32.3% of all cases in Europe for treatment with macrolides (European Centre for Disease Prevention and Control (ECDC) 2019). Also recently, country-wide resistance rates over 25% have been observed in the USA for macrolides. In Europe, variation is higher, ranging from less than 10% in the northern parts to over 25% in parts of Eastern and Southern Europe (Peyrani et al. 2019). So far, resistance to fluoroquinolones remains low in the USA and European Union (Woodhead et al. 2011;Kim et al. 2016).
Pneumococci asymptomatically colonize the nasopharynx, from where they can migrate to other parts of the airway, thereby generating inflammatory responses and disease. They possess a variety of virulence factors, both bound to the cell wall and excreted, that modulate their virulence. A general overview on the importance of virulence factors is described in detail elsewhere and will not be discussed here (Kadioglu et al. 2008;Brooks and Mias 2018).
The polysaccharide (PS) capsule is considered the most important pneumococcal virulence factor, as it is part of the first recognition by the immune system. This capsule is known to be diverse, giving rise to over 90 different pneumococcal serotypes. Current vaccines consist of PS fragments of a selection of these serotypes to induce an immune response (Kim, Seon and Rhee 2017). The first PS capsule vaccine to be licensed was a 14valent PS vaccine in 1977. This vaccine was quickly expanded to include 23 serotypes, and it is still in use today (Briles et al. 2019). Unfortunately, immunogenicity of this vaccine is rather poor (Westerink, Schroeder and Nahm 2012). To overcome this issue, the heptavalent pneumococcal conjugate vaccine (PCV7) was the first conjugate vaccine to be licensed in the USA in 2000. Since then, it has been followed by a 13-valent conjugate vaccine (Briles et al. 2019). Currently, a 15-valent and a 20valent conjugate vaccine are in development Hurley et al. 2020). The introduction of conjugate vaccines dramatically reduced the rates of pneumococcal meningitis through direct and indirect (herd) protection (van de Beek et al. 2016;Kwambana-Adams et al. 2020). However, replacement of vaccine serotypes by non-vaccine serotypes is occurring, thereby lowering vaccine effectiveness. While in the USA this effect is limited, in the UK non-PCV13 serotypes were responsible for over 40% of invasive pneumococcal diseases in 2017 (Deng et al. 2013;Kwambana-Adams et al. 2020). Still, global vaccination programs are considered essential in the battle against pneumococcal diseases (World Health Organisation (WHO) 2019; Kwambana-Adams et al. 2020).
A recent review by Koulenti et al. describes in depth all antibacterial agents against Gram-positive bacteria currently in clinical trials (phase I to phase III) (Koulenti et al. 2020). Table 1 lists all evaluated compound libraries based on in-use antibiotics since 2010. Such an approach leads to a better understanding of the structure-activity relationship (SAR) of current antibiotics and can therefore lead to the identification of a novel antibiotic-analogue. As such, Table 2 lists a comprehensive overview of the more extensively studied novel antibioticanalogues since 2010. Their mechanism of action is similar to that of an already marketed antibiotic, but in most cases resistance mechanisms differ and/or increased activity towards resistant strains is observed. Several of these analogues are currently undergoing clinical trials.
While evaluating derivatives of known antibiotics can be useful, evaluation of new antibacterial targets may also help to overcome resistance mechanisms. As such, targeting bacterial virulence instead of bacterial physiology should be considered (Rasko and Sperandio 2010). Identifying these novel targets is often done using computational screening, in which bacterial and human genome sequences are screened to identify essential bacterial proteins, without affecting the human host and/or its microbiome (Wadood et al. 2018;Nayak et al. 2019).
Clearly, with antibiotic resistance still increasing worldwide, the search for novel antibacterials remains of utmost importance with several ingoing discovery and development programs. In this review, we will provide a comprehensive overview of the current pneumococcal drug pipeline, hereby excluding known antibiotic analogues and focusing on the discovery of novel drug targets. Overall, novel therapies can be divided into three main categories: (i) modulating the host immune system to lower pneumococcal disease, (ii) interference with pneumococcal virulence and (iii) development of novel antibiotics.

Modulating the host immune system
Several recently investigated anti-pneumococcal drug targets focus on enhancing host immune responses after infection. Interfering with these responses is challenging, as this might provoke an unwanted immune cascade leading to an increase in inflammatory damage on the one hand or it might overly inhibit  , Drug targets present at blood-brain barrier. Drug targets in and on macrophages, on epithelial and endothelial cells, in the blood and in the brain are labeled in green. pGSN: plasma gelsolin, GM-CSF: granulocyte/macrophage-colony stimulating factor, iNOS: inducible nitric oxide synthase, NOS3: nitric oxide synthase-3, PsaA: pneumococcal surface antigen A, MIF: macrophage inhibitory factor, IFN-γ : interferon-γ , pIgR: polymeric immunoglobulin receptor, mAb: monoclonal antibody, MASP-2: mannose-binding lectin-associated serine protease, PECAM-1: platelet endothelial cell adhesion molecule.
the immune system leading to an uncontrolled growth and invasion of pathogens. Figure 1 shows a schematic overview of all discussed therapies. Detailed results regarding the in vivo data of these therapies are listed in Table 3.

Induction of nitric oxide synthase (NOS) activity
After macrophage activation by a variety of pro-inflammatory cytokines, constitutively expressed nitric oxide synthase-3 (NOS3) and inducible NOS (iNOS) expression are increased. These synthases show bactericidal activity through the production of nitric oxide (NO) (Hernansanz-Agustín et al. 2013). NO is used as a signaling molecule in low concentrations, while in high concentrations (e.g. during the oxidative burst in neutrophils) it shows direct antimicrobial properties by binding to DNA, proteins and lipids (Schairer et al. 2012). Although an excess production of NO can lead to immunosuppression, NOS inhibitors have previously been reported to reduce tissue injury and mortality in pneumococcal meningitis models. (Fang 2004) It is clear that a strict regulation of the amount of NO is needed to battle an infection. NOS3 expression is mediated by estrogen (Yang et al. 2014b(Yang et al. , 2015. This estrogen-dependency leads to a greater risk of developing pneumonia for males than females (Casimir et al. 2013;Yang et al. 2014b). Statins, known to boost NOS3 activity, have been shown to improve bacterial clearance and survival from secondary pneumococcal pneumonia (Yang et al. 2014b). Furthermore, the use of statins has proven to be beneficial to patients suffering from pneumonia and is associated with a lower risk of hospitalization and mortality (Nielsen et al. 2012; Nishimoto, Rosch and Tuomanen 2020). Administration of sub-cutaneous plasma gelsolin (pGSN), a human blood protein, has also been shown to activate NOS3 and improve outcome of secondary bacterial pneumonia in mice, with reduced acute inflammation and improved bacterial clearance after 24 hours (Yang et al. 2015). The positive effects of pGSN treatment were confirmed in a more clinically relevant murine treatment model using an antibiotic-resistant pneumococcal strain (Yang et al. 2019b). Furthermore, in a 2019 study, patients suffering from community-acquired pneumonia (CAP) admitted to the hospital with low pGSN concentrations were more at risk for developing severe, short-term clinical outcomes such as higher risk of death, septic shock and respiratory failure (Self et al. 2019).
Recently, a phase 1 clinical trial was finalized showing recombinant pGSN was generally safe and well tolerated in patients with mild CAP symptoms and a clinical trial studying the use of recombinant pGSN for the treatment of Covid-19 patients is currently recruiting (Tannous et al. 2020). Apart from NOS3, increased levels of iNOS are also known to contribute to the antibacterial activity of alveolar macrophages against pneumococci. It has also been shown that granulocyte/macrophagecolony stimulating factor (GM-CSF), used in the treatment of leukemia, is able to induce iNOS induction in response to infection, leading to a reduction in bacterial load and inflammation in the lungs of mice (Steinwede et al. 2011).

Macrophage inhibitory factor (MIF) inhibitors
MIF is a component of the innate immune system and constitutively expressed by immune and epithelial cells.

2016)
It is released rapidly after exposure to bacteria and proinflammatory cytokines, promotes expression of numerous other pro-inflammatory molecules and thereby amplifies the response. Apart from its role in inflammation and immunity, it is also important in cell proliferation and oncogenesis (Roger et al. 2007;Bewersdorf et al. 2018). While MIF is essential for pneumococcal clearance after nasopharyngeal colonization (Das et al. 2014), high MIF levels in cerebrospinal fluid (CSF) are associated with poor patient outcome. Moreover, mice treated with MIFneutralizing antibodies show lower bacterial burdens in lungs and a higher survival in a lethal pneumococcal sepsis model (Savva et al. 2016). Similarly, in another study using a murine lung infection model, treatment with the small-molecule receptor antagonist MIF098 improved survival, decreased bacterial burdens by 2 logs and reduced inflammation (Weiser et al. 2015).

Blocking of brain endothelial receptors
Adhesion to and invasion of endothelial and epithelial cells, mediated by several cellular receptors, is an important aspect of invasive pneumococcal disease, such as meningitis. First, the platelet-activating factor (PAF) receptor on activated epithelial and endothelial cells enables the bacteria to enter the basal membrane of the host epithelial cell. Secondly, the pneumococcal choline-binding protein PspC can bind to the epithelial polymeric immunoglobulin receptor (pIgR). Analogous to the PAF receptor pathway, bacteria are transported into the cell. Concordantly, when crossing the blood-brain barrier (BBB), the PAF receptor is used again (Koedel, Scheld and Pfister 2002;Mook-Kanamori et al. 2011). Furthermore, also pIgR is expressed by brain endothelial cells and can be used by pneumococci to adhere. Lastly, platelet endothelial cell adhesion molecule (PECAM-1), one of the major endothelial adhesion molecules, can mediate adhesion of pneumococci to the BBB endothelium (Iovino et al. 2017). Using an in vivo meningitis model, treatment of infected mice with anti-pIgR and anti-PECAM-1 antibodies 1 hour post infection increased survival time and lowered bacterial brain burden, yet all mice eventually still succumbed. A cotreatment strategy with ceftriaxone was however more successful. Ceftriaxone was capable of clearing the blood infection while the anti-pIgR and anti-PECAM-1 antibodies prevented most bacteria from passing the BBB, leading to a decrease in bacterial burdens and an increase in survival. Moreover, neuroinflammation was significantly lower in the combination therapy group compared to untreated mice or mice treated with ceftriaxone alone (Bewersdorf et al. 2018;Iovino, Thorsdottir and Henriques-Normark 2018).

Modulation of complement activity
The complement system is important in human immunity and comprises several recognition proteins activated in response to pathogens. Activation is triggered by a proteolytic cleavage amplification cascade, which generates fragments capable of binding to microbial surfaces as opsonins and bacterial cell destructions. Opsonins can also promote pathogen phagocytosis and induce inflammatory responses (Andre et al. 2017). In pneumococcal meningitis, complement however has a dual role. While it is needed to initiate complement-mediated bacterial killing, uncontrolled activation can occur and often leads to a worse disease outcome (Bewersdorf et al. 2018). High levels of complement component 5 (C5) have been shown to worsen patient outcome for bacterial meningitis. Furthermore, adjuvant therapy with C5 antibodies showed beneficial effects on survival rates, brain damage and clinical severity in several murine in vivo studies (Woehrl et al. 2011;Kasanmoentalib et al. 2015). Similar results have been found using mannose-binding lectinassociated serine protease (MASP-2), an important activator in the lectin pathway. While MASP-2 has proven to be important in avoiding nasopharyngeal carriage of pneumococci, it is also associated with worsened meningitis outcomes. As with C5, treatment with MASP-2 antibodies after pneumococcal infection increased in vivo murine survival (Kasanmoentalib et al. 2017). Complement factor H (FH), a regulatory protein inhibiting complement component 3 (C3), is known to be important in moderating pneumococcal disease. However, in several mouse models, combination therapy of recombinant FH with ceftriaxone showed no beneficial effects (Van Der Maten et al. 2016;Kasanmoentalib et al. 2019). Lastly and contradictorily to aforementioned therapies, Ali et al. showed increased in vitro opsonization of pneumococci after properdin treatment, i.e. a known positive regulator of complement activation naturally present in humans. Furthermore, animals infected with pneumococci and treated with properdin show a higher chance of survival and lower bacterial blood burden compared to non-treated animals (Ali et al. 2014). Clearly, functions of complement in pneumococcal pathogenesis are diverse. Interfering with its working mechanism might result in major and unforeseen changes in immune responses, and therefore, this interference should be done in a controlled and temporarily manner. Currently, one anti-C5 antibody, eculizumab, is on the market. However, side effects to this drug include increased risk of severe meningococcal meningitis, demonstrating the delicate role of the complement system in bacterial infections. Two other antibodies are currently undergoing clinical trials (Koelman, Brouwer and Van De Beek 2019).

Interferon-gamma (IFN-γ ) inhibitors
Similarly to the aforementioned molecules, IFN-γ also plays a dual role in pneumococcal disease. As a powerful mediator of different innate and adaptive immune pathways during inflammation and infection, it is required for generating an effective response upon bacterial invasion. However, it is also involved in long-term neurological sequelae after pneumococcal meningitis and interfering with IFN-γ responses can lead to a higher survival rate of patients (Too et al. 2014;Yau et al. 2017;Bewersdorf et al. 2018). IFN-γ antibody treatment improved survival after pneumococcal meningitis and overall clinical symptoms were less compared to non-treated animals. Interestingly, bacterial burdens were unaffected by this treatment (Pettini et al. 2015).

Death receptor (DR) agonists
Tumor necrosis factor-related inducing ligand (TRAIL) is a member of the TNF superfamily. There are five human TRAIL receptors known, of which receptors DR4 and DR5 are involved in apoptosis of neutrophils, amongst other effects (Sag et al. 2019). It has been shown that TRAIL deficiency leads to increased inflammation following pneumococcal challenge (Hoffmann et al. 2007). In a murine pneumonia model, treatment with agonistic anti-DR5 antibodies led to a small but significant decrease in lung burden 72 h p.i. and an increase in survival from 30% to 70% 8 days p.i (Steinwede et al. 2012). However, TRAIL is also considered important in protection against viral lung infection, which often precede pneumococcal infections. Therefore, interference with this molecule is considered difficult, making it less favorable as therapeutic target (Braithwaite, Marriott and Lawrie 2018).

Danger-associated molecular pattern (DAMP) inhibitors
Meningitis is triggered after recognition of pathogen-associated molecular patterns (PAMPs). As a result, neutrophils are recruited and release toxins in order to kill the pathogens present. However, these neutrophil-derived toxins can, in turn, cause stress and damage to host cells. These injured cells subsequently release DAMPs, responsible for tissue damage (Wache et al. 2015;Masouris et al. 2017). HMGB1 and MRP14 are DAMPs known to be secreted in large amounts in de CSF of meningitis patients and could thus be a therapeutic target. Recently, the MRP14 inhibitor paquinimod that could be used in autoimmune diseases has been shown to reduce inflammation in a pneumococcal meningitis model, further supporting its potential for meningitis therapy (Wache et al. 2015;Bewersdorf et al. 2018). Similarly, HMGB1 antibodies were also shown to lower inflammation, and improve the outcome of pneumococcal infection (Masouris et al. 2017). As for other immunomodulatory therapies, inhibiting DAMPs can only be done in a very tightly controlled and balanced manner to assure controlled inflammation and subsequent elimination of bacteria (Land 2020).

Pneumococcal surface antigen A (PsaA) and pneumococcal surface protein A (PspA) as targets for therapy
PsaA and PspA are well-known pneumococcal virulence factors and have been studied as potential vaccine candidates. PsaA is an indispensable protein for Mn 2+ transport, protecting against oxidative stress and for adherence to endothelial cells. PspA is important in evasion of the host immune system by interfering with the complement C3 cascade (Tai 2006). Mutants lacking PsaA are significantly less virulent compared to their parent strains, which could be attributed to growth impairment, reduced adherence capacity or hypersensitivity to oxidative stress (Rajam et al. 2008a). PsaA however, seems also essential in the induction of a protective host immune response, since peptide P4, a short amino acid fragment of PsaA, was capable of binding to nasopharyngeal epithelial cells and eliciting an inflammatory response (Rajam et al. 2008b). Furthermore, ex vivo human alveolar macrophages and neutrophils demonstrated improved bacterial killing after P4 exposure and treatment of intranasal infected mice with P4 was shown to increase survival (Rajam et al. 2009(Rajam et al. , 2010Bangert et al. 2013;Morton et al. 2016). No follow-up studies involving P4 have been reported so far. Monoclonal antibodies (mAbs) against PspA have also been studied for therapeutic use. Several mAbs against PspA were produced in mice, and their in vitro opsonization capacity to pneumococci was evaluated. One selected mAb (mAb 140H1) was evaluated in vivo, and administration after infection with pneumococci reduced bacterial lung and blood burden and improved survival rate. Furthermore, combination therapy with the standard antibiotic ceftriaxone showed a synergistic effect (Kristian et al. 2016). Currently, PspA is studied as a vaccine candidate rather than a therapeutic target (Wagner-Muñiz et al. 2018;Akbari et al. 2019).

Interfering with pneumococcal virulence
Next to modulating host immunity, research focusing on reducing pneumococcal virulence to enable the immune system to overcome the infection shows growing interest. Most proposed targets are based on virulence factors specific to pneumococci and thus potential novel therapies will be pathogen-specific. A schematic overview of the drug targets is shown in Fig. 2. Table 5 lists detailed results regarding the current in vivo data of these therapies.

PS capsule formation inhibitors
As mentioned earlier, the PS capsule is the most important virulence factor. It inhibits macrophage phagocytosis, which forms the first line of defense to pneumococcal invasion (Dockrell et al. 2003;Dockrell, Whyte and Mitchell 2012). Deletion of the capsule increases phagocytosis rates in vitro and decreases virulence in vivo (Preston and Dockrell 2008). However, downregulation of the capsule is needed to initiate nasopharyngeal colonization (Gilley and Orihuela 2014). Also during pneumonia and OM, downregulation of the capsule and subsequent formation of a biofilm is considered part of the pneumococcal immune evasion strategy (Moscoso, Garcia and Lopez 2006;Domenech et al. 2013). In contrast, when the change from a commensal to a pathogenic lifestyle occurs, pneumococci upregulate their PS capsule production (Gilley and Orihuela 2014). The pneumococcal capsule is formed through two pathways, the Wzy-dependent pathway used by most, and the synthase-dependent pathway, only used in serotypes 3 and 37 (Geno et al. 2015). The capsule of serotype 3 consists of glucose (Glc) and glucuronic acid (GlcA) and is the result of 3 genes (cps3D, cps3S and cps3U), of which only the first two are essential genes for capsule production. Serotype 37 only requires one gene, tts, to form its capsule, consisting solely of Glc chains. The Wzy-dependent pathway is more complex. In general, sugar-1-phosphate is transferred to a lipid carrier on the cytoplasmic side of the cell membrane by a glycosyltransfersase. From there, the repeat unit is built and translocated to the extracellular side by flippase Wzx before being polymerized by Wzy. Lastly, the final sugar (usually Glc) is covalently bound to N-acetylglucosamine residue of the cell wall peptidoglycan (Geno et al. 2015;Paton and Trappetti 2019). While these pathways differ, they both use the cps gene locus to build the correct sugar conformation. As the loci slightly differ for each serotype, interfering with them or their products is difficult. Thus far, only CpsB, a tyrosine phosphatase encoded by cpsB, has been suggested to be a potential novel anti-virulence drug target, as cpsB mutants are avirulent in several animal models of infection (Morona et al. 2004;Standish et al. 2012;Monteiro Pedroso et al. 2017). Furthermore, the molecule fascioquinol Ean extract derived from the marine sponge Fasciospongia spp.inhibited CpsB phosphatase activity and increased macrophage attachment in vitro (Standish et al. 2012).
Lastly, regardless of the pneumococcal serotype or capsule pathway, uridine diphosphate glucose (UDP:Glc) is generally considered a key component in the formation of PS capsule. UDP:Glc is part of the Glc and galactose (Gal) metabolism and is made through the interconversion of glucose-1-phosphate (Glc-1-P) by uridine diphosphate glucose pyrophosphorylase (UDPG:PP) (Mollerach, López and García 1998). It has been described that mutants lacking a functional galU gene form a lower amount of PS capsule, are more prone to macrophage phagocytosis in vitro and are less virulent in a Galleria mellonella in vivo model. (Mollerach, López and García 1998;Cools et al. 2018) As the pneumococcal UDPG:PP crystal structure is unknown, modeling of new molecules is challenging. However, recently, two attempts have been made, either using the purified enzyme in an enzymatic assay or using a computational molecular docking model based on other bacterial UDPG:PP's (Zavala et al. 2017;Cools et al. 2020).

Bacterial cell wall modifiers
The cell wall of Gram-positive bacteria, such as the pneumococcus, consists of a thick peptidogylycan (PG) layer, to which teichoic acids and capsular PS are covalently attached. Due to its importance, PG biosynthesis has always been an interesting drug target. Currently, beta-lactam antibiotics and vancomycin are the two most used PG inhibitors (Vollmer, Blanot and De Pedro 2008;Rajagopal and Walker 2017;Vollmer, Massidda and Tomasz 2019).
Lysozyme, a major bacteriolytic component of the immune system, hydrolyses PG chains, which results in lysis of bacteria. Pathogens such as S. pneumoniae, however, modify their glycan strands through deacetylation by peptidoglycan Nacetylglycosamine deacetylase A (PgdA), thereby increasing resistance to lysozyme. Mutant strains lacking PgdA are significantly more susceptible to lysozyme in vitro and show a reduction in virulence in vivo (Vollmer and Tomasz 2002). Since then, several publications have described in vitro inhibitors of PgdA enzymatic activity, yet no follow-up studies on the activity towards the bacterial cell or in vivo use have been performed (Bui et al. 2011;Ariyakumaran et al. 2015;DiFrancesco, Morrison and Nitz 2018).

Pneumolysin (PLY) inhibitors
PLY is formed intracellularly and released in the environment through bacterial lysis. It is known to be cytolytic to all human cell types through the formation of pores. Furthermore, it promotes pro-inflammatory immune responses through activation of the classical and lectin pathways of complement activation (Anderson and Feldman 2017). As PLY is known as an essential virulence factor for bacterial survival in the respiratory tract, inhibiting this enzyme might prove beneficial (Kadioglu et al. 2008;Kim, Seon and Rhee 2017). A variety of natural compounds has been successfully tested for their anti-PLY activity (Table 4). Also statins were proven to have a direct effect on PLY cytotoxicity in vitro and in vivo and could be valuable as adjuvant therapy to existing antibiotics. However, further studies are needed to study the mechanism of action and potential place in pneumococcal therapy (Nishimoto, Rosch and Tuomanen 2020).
Lastly, liposomes consisting of naturally occurring cholesterol and sphingomyelin can be used to sequester PLY (Baumgartner et al. 2016). In vitro, these liposomes protect monocytes from secreted PLY. Furthermore, mice were more prone to survive a pneumococcal lung infection after treatment with liposomes. Besides a small reduction in bacterial burdens in lung and blood, the treatment was also capable of reducing tumor necrosis factor α (TNF-α) levels. This led to less signs of inflammation in the lungs. In an in vivo sepsis model, bacterial blood burden was reduced after treatment, leading to an increase in murine survival. Also in this model, TNF-α levels were decreased (Henry et al. 2015). CAL02, a mixture of liposomes, has undergone clinical trials to verify its safety and activity. While a dosage effect could not be established, the safety and tolerability of CAL02 was promising, with no adverse effects that could be linked to local tolerability events. Furthermore, all patients receiving CAL02 treatment were cured 15 to 22 days after the start of the trial (Laterre et al. 2019).

Antimicrobial peptides (AMPs)
AMPs are a large and diverse group of molecules, produced by both pro-and eukaryotes. In humans, they are important in innate immunity, while in bacteria AMPs are produced as a way to compete with and kill other bacteria (Mahlapuu et al. 2016). In a 2015 study, analogues of indolicidin and ranalexin, two natural peptides with known antibacterial activity against Grampositive bacteria, showed in vitro bactericidal activity towards antibiotic susceptible and resistant pneumococcal strains, without cytotoxicity for eukaryotic cells (Jindal et al. 2015). In silico molecular docking pointed towards interactions with autolysin and/or PLY, two known pneumococcal virulence factors as a mechanism of action, while a later study revealed that cell membrane integrity is highly impacted by these AMPs in vitro (Jindal et al. 2015(Jindal et al. , 2017. This study also confirmed the in vivo effect of these compounds in a murine bacteremia model, leading to an increase in survival, decrease in bacterial burden and decrease in tissue damage in lungs and spleen (Jindal et al. 2017).

Quorum sensing (QS) inhibitors
Biofilm formation is known to play an important role in OM and cochlear implant infections (Cevizci et al. 2015). QS systems are known to control the maturation stage of these biofilms, as they allow communication between bacteria in a cell-density dependent manner (Brackman and Coenye 2014). Targeting this system, rather than targeting the bacteria itself, has been gaining scientific attention and S. pneumoniae uses several QS systems that have the potential for intervention. First, the ComABCDE pathway is regulated by competence-stimulating peptide (CSP). This system allows for induction of competence and controlling genetic transformation, when a biofilm has matured and pneumococcal density is high enough. Bacteriocins, which inhibit the growth of competing bacteria, are produced as a response to the BlpABCSRF pathway, which operates similarly to ComABCDE. Also, autoinducer-2 (AI-2) is known as a common QS system of Gram-positive bacteria and is also present in pneumococci. The enzyme LuxS activates AI-2, which, in turn, facilitates initial attachment of bacteria to a surface (Galante et al. 2014). Apart from LuxS, DNA adenine methyltransferase (DAM) plays an important role in the biosynthesis of AI-2, as it is part of the activated methyl cycle. It catalyzes a methyl transfer from S-adenosyl-L-methionine (SAM) to adenine, a feature unique in bacteria (Yadav et al. 2015). Lastly, in 2015, Dimarchi et al. discovered a novel QS system, TprA/PhrA, that controls the expression of bacteriocins called lantibiotics (Hoover et al. 2015). This system has been shown to be crucial for pneumococcal virulence in pneumonia, meningitis and OM models (Motib et al. 2017). Several biofilm disruption strategies have been studied. Cevizci et al. tested the possibility of analogues of N-acyl homoserine lactone (AHL), a signaling molecule known to play a role in QS systems, to prevent pneumococcal biofilm formation in an in vivo cochlear implant model. While these results are preliminary, prolonged treatment with the AHL analogue benzoxazolone after cochlear implant surgery prevented biofilm formation on these implants (Cevizci et al. 2015). However, it should be noted that AHL is a molecule solely attributed to Gram-negative bacteria, leaving the exact mechanism of these inhibitors against pneumococci unknown. More recently, a linear molecularly imprinted polymer (LMIP) targeting the TprA receptor and its signaling peptide PhrA was evaluated. Addition of LMIP-PhrA to in vitro cultures reduced growth rate and neuraminidase activity, which is important in pneumococcal colonization and invasiveness. In a pneumonia mouse model, dissemination from lungs to blood was prevented by LMIP-PhrA, resulting in longer survival of animals (Motib et al. 2017(Motib et al. , 2019. Lastly, Yadav et al. studied the effect of interfering with DAM or SAM. In 2012, the effect of 5-azacytidine (5-aza) on in vitro planktonic and biofilm growth and the effect on gene expression was assessed. 5-aza is a hypomethylating drug used in leukemia treatment. Only a minor effect on planktonic growth was observed, while biofilm formation was adequately and dose-dependently inhibited in vitro. Using scanning electron microscopy (SEM), 5-aza-treated biofilms were visualized and observed to be thinner, more scattered in clumps and disorganized compared to non-treated biofilms. Also genes involved in AI-2 synthesis were downregulated after 5-aza treatment (Yadav, Chae and Song 2012). In a very similar study, sinefungin, a natural nucleoside and analogue of SAM known for its inhibitory effects on transmethylation reactions and overall antifungal, antiviral and antiprotozoal activities, was evaluated. Also in this study, in vitro biofilm formation was inhibited, and AI-2 synthesis genes were downregulated. Furthermore, in vivo biofilm formation in an OM rat model showed a decrease in CFU/bulla after sinefungin treatment (Yadav et al. 2014). Inhibition of DAM by the small molecule pyrimidinedione proved equally successful in inhibiting in vitro biofilm formation and downregulating virulence-related genes, such as ply and lytB, as well as the competence-related gene comC (Yadav et al. 2015).

Development of novel antibiotics
Lastly, a variety of novel antibiotics is being investigated, which in most cases work against a wider spectrum of pathogens. However, as for in-use antibiotics, mutations leading to resistance  (Yadav et al. 2014) are of concern. An overview of the antibiotics tested against pneumococci is shown in Fig. 3. Detailed information regarding the described in vivo experiments is listed in Table 6.

Bacterial topoisomerase II inhibitors
Bacterial topoisomerases (DNA gyrase and topoisomerase IV) are targets of the well-known quinolones. However, a new class of topoisomerase II inhibitors, called novel bacterial topoisomerase II inhibitors (NBTIs), has been put forward as a new way of treating quinolone-resistant infections. These inhibitors also bind to DNA gyrase and topoisomerase IV, yet do so on a slightly different binding site, thereby evading existing resistance mechanisms. Gepotidacin represent this new drug class.
It shows in vitro activity against a variety of Gram-positive bacteria, including quinolone-resistant pneumococci (Bax et al. 2010;Biedenbach et al. 2016;. Gepotidacin is currently recruiting for phase III trials, evaluating efficacy, safety and applicability of the drug (Kolarič, Anderluh and Minovski 2020;Koulenti et al. 2020). Other classes of bacterial topoisomerase II inhibitors are the pyrrolamides, tetrahydropyrans, tricyclics, pyrimidines and quinolines. For some of these lead compounds, only in vitro efficacy has been tested, while others also show promising effects in vivo. Regardless, none of these classes have progressed into clinical trials for pneumococcal infections (Zhang et al. 2011b;Eakin et al. 2012;Mitton-Fry et al. 2013;Odagiri et al. 2013Odagiri et al. , 2018Uria-Nickelsen et al. 2013;Surivet et al. 2015;Lepak et al. 2016;Miles et al. 2016).

Thymidylate kinase (TMK) inhibitors
TMK is an essential enzyme in the DNA synthesis pathway. It transfers phosphate from adenosine triphosphate (ATP) to thymidine monophosphate (dTMP), which leads to the formation of thymidine diphosphate (dTDP), an essential component of the thymidine triphosphate (dTTP) pathway (Keating et al. 2012). As such, TMK is essential for bacterial growth and has been put forward as an interesting drug target (Petit and Koretke 2002). Structure-based drug design led to the development of several compounds, of which TK-666 has proven to be active in vitro against Gram-positive bacteria, while not being harmful to eukaryotic cells. Furthermore, it was shown to be equally active against antibiotic resistant and susceptible pneumococci (Keating et al. 2012;Martínez-Botella et al. 2012). Currently, TMK inhibitors are mainly evaluated for their use against Mycobacterium tuberculosis infections (Jian et al. 2019(Jian et al. , 2020Venugopala et al. 2020).

PyrG inhibitors
PyrG, a bacterial cytidine triphosphate (CTP) synthase, produces CTP from uridine triphosphate (UTP) and glutamine and is essential in the pyrimidine de novo biosynthetic pathway (Endrizzi et al. 2004). Furthermore, it is required for growth of bacteria such as Haemophilus influenzae. PyrG inhibitors were first reported in the literature over 40 years ago as antitumor drugs, however, thus far no PyrG inhibitors have reached the market. After an enzyme-based HTS of PyrG inhibitors, enamine proved the most interesting. However, while the IC 50 value in the enzymatic assay was low (0.091 μM), the MIC was more than 128 μg/mL for pneumococci. In contrast, MICs ranging from 16 to 64 μg/mL were observed for other pathogens such as H. influenzae, E. coli and S. aureus. While this demonstrates the necessity of whole-cell based assays in drug discovery, it is also a first step towards the development of pneumococcal PyrG inhibitors (Yoshida et al. 2012). Similarly to TMK inhibitors, PyrG inhibitors are now primarily studied as a way to combat mycobacterial infections (Chiarelli et al. 2018).

NAD + -dependent DNA ligase (LigA) inhibitors
LigA is an essential enzyme in viability for both Gram-positive and Gram-negative bacteria. It is completely unrelated to eukaryotic DNA ligases, making it an interesting drug target (Pascal 2008;Mills et al. 2011). Selective inhibitors of LigA, substituted adenosine analogues, were reported to show a broad spectrum of bacterial inhibition, ranging from E. coli, Mycoplasma pneumoniae to S. pneumoniae. No binding affinity to eukaryotic DNA ligases or cytotoxicity against human red blood cells or alveolar epithelial cells A549 was observed. Furthermore, an in vivo murine lung infection model showed a dose-dependent reduction of pneumococci after treatment with one of these adenosine analogues (Mills et al. 2011). Optimization of compound classes pyridopyrimidines and naphthyridines has also been studied in vitro, leading to LigA inhibiting molecules for a wide range of bacteria, including pneumococci (Murphy-Benenato et al. 2015). More recently, the mechanism of action of the long known broad-spectrum antibiotic cordycepin was attributed to its binding to LigA (Zhou et al. 2016). However, no follow-up research on any of the aforementioned compounds has been performed.

Peptide deformylase (PDF) inhibitors
PDF is a metalloenzyme used in bacterial peptide elongation. It cleaves a formyl group from the terminal N-methionine of a  (Pi et al. 2020) newly synthesized polypeptide following ribosomal translation and elongation. Removal of this formyl group is vital for bacterial viability. Importantly, human PDF is structurally different from its prokaryotic counterpart, therefore bacterial PDF is considered an interesting novel drug target (Sangshetti, Khan and Shinde 2014). In 2011, GSK1322322 was introduced as a potential novel antibiotic against multidrug resistant S. aureus and CABP (Ross et al. 2011). It showed good in vitro activity against pneumococci and was able to reduce pneumococcal lung burdens in a murine model (Sutcliffe 2011). This molecule was undergoing clinical trials, however trials were ended due to the identification of potentially harmful metabolites.(United States National Institute of Health (NIH) 2019) IDP-73 152, another PDF inhibitor, went through a phase I clinical study in 2013, showing no adverse effects after oral administration in healthy volunteers. The results were published in 2019, but no follow-up studies have been announced (Shin et al. 2019).

Inhibition of cell wall synthesis
Beta-lactams, the most widely used antibiotics, interfere with cell wall synthesis by targeting the transpeptidase activity of penicillin-binding proteins (PBPs). They block transpeptidation by covalently binding to these enzymes, thereby killing the bacteria. However, resistance is increasing for this antibiotic class. (Macheboeuf et al. 2006) Therefore, attempts have been made to discover novel inhibitors. One of these strategies led to the screening of noncovalent inhibitors, as they are postulated to be less susceptible to resistance-inducing mutations of PBPs and production of beta-lactamases . In 2009, the first virtual screening of noncovalent pneumococcal inhibitors was performed. However, no biological assays were performed to confirm these results (Miguet et al. 2009). Another article from 2011 reported on 2 molecules showing promising in vitro activity against antibiotic-resistant pneumococci . Since then, no results on noncovalent PBP inhibitors have been published, suggesting this approach was unsuccessful. Using a completely different mechanism, Lewis et al. published the finding of the first novel natural bacterium-derived antibiotic in decades (Durand, Raoult and Dubourg 2019). Teixobactin inhibits peptidoglycan biosynthesis by binding to lipid II and lipid III, precursors of peptidoglycan and teichoic acid, respectively. In vitro, it is active against a variety of Grampositive bacteria, but not against Gram-negatives. Teixobactin showed a dose-dependent reduction in pneumococcal lung burden in a murine model (Ling et al. 2015). Since its discovery, numerous analogues have been synthesized, however still no clinical trials involving teixobactin or one of its analogues have been setup (McCarthy 2019; Koulenti et al. 2020).
Lastly, undecaprenyl pyrophosphate synthetase (UPPS) synthesizes undecaprenyl pyrophosphate (UPP) from isopentenyl pyrophosphate (IPP). UPP is essential in the bacterial cell wall synthesis, as it is needed for cross membrane transport of carbohydrates, and as such, UPPS has been shown to be essential for pneumococcal growth (Apfel et al. 1999). However, only few attempts have been made to develop UPPS inhibitors against pneumococci. In 2010, tetramic acids were evaluated as pneumococcal UPPS inhibitors. While these compounds showed interesting in vitro activity, no whole-cell based assays were performed (Lee et al. 2010). Later, Danley et al. reported in vitro UPPs inhibition by UK-106 051, a carboxamide analogue, leading to inhibition of pneumococcal growth, yet no in vivo confirmation has been attempted (Danley et al. 2015). Since then, few papers on the use of UPPS inhibitors against a variety of bacteria were published, all reporting early stage preclinical data (Concha et al. 2016;Inokoshi et al. 2016;Wang et al. 2016;Cherian et al. 2017;Jukic et al. 2019). While these data show potential for this target, currently there are no UPPS inhibitors in development.

Choline binding protein (CBP) interference
CBPs are a large group of proteins located on the pneumococcal cell surface, involved in pneumococcal autolysis (LytA, LytC, CbpD, CbpF), cell separation after division (LytB), interference with complement activation (PspA), adherence to host cells (PspC, PcpA, CbpG) and modulating the amount of choline (Pce). There are two main strategies to use CBP as a novel antipneumococcal approach, (i) through direct inhibition of CBPs and (ii) through the administration of endolysins (Maestro and Sanz 2016).
Maestro et al. published a detailed review in 2016 on the use of direct CPB inhibitors (Maestro and Sanz 2016). All inhibitors were based upon choline, such as esters of bicyclic amines (EBAs) which can be used as monomers or nanoparticle dendrimers, consisting of several ligands. As CBPs contain several binding sites, the binding affinity increases dramatically (Mammen, Choi and Whitesides 1998). De Gracia Retamosa et al. showed EBAs were capable of lysing in vitro planktonic cultures in a dose-dependent manner and increased survival in a zebrafish model (De Gracia Retamosa et al. 2015). In a 2019 study, EBAs were confirmed to in vitro lyse planktonic cultures using a LytAdependent mechanism. Furthermore, also in vitro formation of biofilms was blocked (Roig-Molina et al. 2019). Lastly, phagocytosis of bacteria by microglial cells increased in in vitro dendrimertreated pneumococcal cultures (Ribes et al. 2013).
Phage therapy has repeatedly been suggested in the battle against multidrug resistant bacteria. Bacteriophages have been observed to kill antibiotic resistant bacteria. In addition, some of their products, such as endolysins, show promising results (Cisek et al. 2017). Interestingly, phage resistance mostly comprises a loss of bacterial virulence (Caflisch, Suh and Patel 2019). Endolysins are produced by phage-infected bacteria to enable the release of novel phages, thereby destroying the bacterial cell wall. (Stoffels et al. 2017) In pneumococci-and other Gram-positive bacteria-these lysins belong to the CBP family. Therefore, they recognize choline residues in the teichoic acids of pneumococci (Vázquez and García 2019). Administration of purified endolysins to in vitro pneumococcal cultures led to a rapid decrease in cell density (Diez-Martinez et al. 2015). Furthermore, mice and rats showed a greater likelihood of survival after endolysin therapy following a pneumococcal challenge (Grandgirard et al. 2008;Diez-Martinez et al. 2015). Synergy between different phages has been observed both in vitro on planktonic and biofilm cultures and in a zebrafish model, leading to an increase in zebrafish survival (Vázquez and García 2019). Several clinical trials involving phage therapy have already been done, or are ongoing or planned (Vázquez, García and García 2018;Caflisch, Suh and Patel 2019). While none of these trials have focused on pneumococci, this does indicate phages or lysins are promising antibacterial drug candidates.

PsaA inhibitors
Apart from enhancing the immune system through passive immunization (see earlier), PsaA has also been proposed as a direct drug target. Inhibitors have been studied for their antipneumococcal properties. In 2015, Bajaj et al. virtually screened a library of small molecules for their binding properties to PsaA (Couñago et al. 2014). Hits were further optimized and tested in an in vitro model. Two molecules showed promising inhibitory properties against pneumococci, however the authors stated that more optimization of these compounds was needed (Bajaj et al. 2015). This experimental setup was later repeated by another research group, confirming PsaA as an alternative drug target. However, this research group also failed to identify a sufficiently inhibiting molecule, as its most active compounds possessed a flexible tail, which is considered a poor development prospect (Obaidullah et al. 2018).

Flavin adenine dinucleotide synthetase (FADS) inhibitors
FADSs synthesize flavin mononucleotide (FMN) and flavin adenine dinucleotide (FAD). As cofactors of flavoproteins, FMN and FAD are present in all living organisms and insufficiency of either leads to cell death. In bacteria, FADS synthesizes FMN and FAD from riboflavin (RF) in two steps: production of FMN from RF (RFK module) and subsequent production of FAD from FMN (FMNAT module). While there is some similarity between proand eukaryotes in the RFK module, this is not the case for the FMNAT module, making it an interesting novel drug target (Serrano et al. 2013;Sebastián et al. 2017). Using an enzyme-based HTS approach based on the crystal structure of Corynebacterium ammoniagenes, several lead compounds were identified. However, biological activity against pneumococci remained low, suggesting they do not reach inhibitory concentrations at the intracellular level (Sebastián et al. 2018). Lastly, a recent study using virtual screening identified four FADS inhibitors inhibiting pneumococcal growth in vitro (Lans et al. 2020).

Coenzyme A (CoA) pathway inhibitors
CoA is an essential cofactor in all living organisms in the metabolism of fatty acids. It is synthesized from pantothenate (vitamin B5), cysteine and ATP. First, pantothenate is phosphorylated to 4'-phosphopantothenate by pantothenate kinase (CoaA). This molecule is then condensated with cysteine and subsequently decarboxylated to obtain 4'phosphopantetheine. Using the enzyme phosphopantetheine adenylyltransferase (PPAT) and ATP, 4'-phosphopantetheine is converted to dephospho-CoA, which is subsequently phosphorylated to yield CoA (Leonardi et al. 2005).
PPAT is an essential enzyme within the CoA pathway. It has been put forward as a novel drug target numerous times, because of its structurally attractive site, high degree of conservation among bacterial species, distinct differences between bacterial and human PPAT, known kinetics and available purified enzyme (Miller et al. 2010). In 2013, de Jonge et al. identified several lead compounds capable of inhibiting PPAT through structure-based HTS. Moreover, these compounds inhibited in vitro growth of macrolide-resistant pneumococci, as well as other antibiotic-resistant Gram-positive bacteria. In a murine lung infection model, treatment with PPAT inhibitors led to a stasis of bacterial burden (De Jonge et al. 2013).
Furthermore, pantothenamides, derivates of pantothenate (vitamin B5), are known to possess antibiotic activity in vitro. They are substrates of CoaA, leading to the formation of inactive CoA analogues (Zhang, White and Rock 2006). In mammals, vanins-also called panthetheinaseshydrolyze pantetheine into vitamin B5, as a way of vitamin recycling (Jansen et al. 2013a). In 2013, studies showed that antibiotic pantothenamides were also hydrolyzed and thus inactivated by vanins, leading to the combination of pantothenamides and vanin inhibitors as a novel antibacterial strategy against Gram-positive bacteria. Vanin inhibitors were able to protect pantothenamides from degradation by host panthetheinases, making pantothenamides more stable for use (Jansen et al. 2013b). Furthermore, pantothenamides resistant towards vanin activity have been reported recently. However, activity of these modified pantothenamides towards pneumococci has not been assessed yet, as they are now mainly studied for their antiplasmodial activity (Jansen et al. 2019;Spry et al. 2020).
In turn, CoA is used in a series of reactions to produce fatty acids. In these subsequent reactions, enoyl-acyl carrier protein reductase (ENR) catalyzes the last and rate-limiting step in each round of chain elongation. Different isoforms of ENR exist, however, pneumococci only possess one of them, called FabK (Heath and Rock 1995). In silico docking revealed several potential inhibitors for FabK, however, to date, none of these have been tested further (Zhang et al. 2011a). Some ENR inhibitors have reached clinical trials, however, none of them are specific to FabK (Rana et al. 2020).

Interference with iron transport systems
Iron is an essential nutrient for bacterial growth and survival. As the concentration of free iron in the host is low, bacteria developed highly specific iron-acquisition systems on their membrane surfaces. In S. pneumoniae, PiaABC, PiuABC and PitABC are the three known iron-transport systems, respectively responsible for the acquisition of heme, ferrichrome and ferric irons (Brown et al. 2002;Cheng et al. 2013). PiaA, PiuA and PitA located on the cell surface bind these free iron-molecules. A Ru(II) complex has been tested for its inhibitory activity towards PiuA, needed for ferrichrome transport. This complex was capable of inhibiting pneumococcal growth without affecting an alveolar epithelial cell line in vitro (Yang et al. 2014a). A similar study conducted with a Rh(II) complex showed the same effects, consolidating the use of metal complexes as potential novel anti-pneumococcal drugs (Yang et al. 2019a). Since metal complexes only recently drew attention as potential antibacterial agents, development is still in a very early stage (Frei 2020).

Repurposing of existing drugs
Repurposing of existing drugs has important benefits. As these drugs are already in use, their safety, tolerability and toxicity has been extensively studied, reducing the costs for development (Cragg, Grothaus and Newman 2014).
Robenidine is an anticoccidial agent used worldwide in poultry and rabbits. Recently, it has been evaluated for its activity on bacteria such as pneumococci and S. aureus. Robenidine, together with two analogues, showed in vitro bactericidal activity against pneumococci by disrupting the cell membrane potential, leading to a thicker cell membrane and a wider periplasmic space (Ogunniyi et al. 2017). Recently, treatment of septic mice with one of these analogues, NCL195 showed a minor reduction in bacterial burden 18 h p.i. Survival was also prolonged, as all mice died 16 h p.i. in the control group. In the treatment group, 60% was still alive at this point, but eventually all mice succumbed at 46 h p.i. As such, NCL195 is not suitable for further development, but its scaffold could be used in further research (Pi et al. 2020).
Choline kinase (ChoK) is a mediator of cell growth and division of eukaryotic cells. As such, it is a drug target for tumor cells. ChoK inhibitor RSM-932A is undergoing clinical trials. However, multiple bacteria including pneumococci also express ChoK. Human ChoK inhibitors MN58b and RSM-932A have been shown to inhibit pneumococcal growth in vitro. However, as these inhibitors also inhibit human ChoK and potentially other bacterial ChoK, including those of the human microbiome, other inhibitors need to be identified (Zimmerman, Lacal and Ibrahim 2019).
Antioxidants N-acetyl-L-cysteine and cysteamine are mucolytics and have been proposed suitable in the treatment of Huntington's and Parkinson's diseases, as well as cystic fibrosis and malaria. In vitro, these two compounds have antibacterial activity against pneumococci in mixed-species biofilms with H. influenzae, killing 98% of all bacteria (Domenech and García 2017). However, it should be noted pneumococci used in this study were non-encapsulated to promote biofilm formation, which might affect results.
Lastly, auranofin, a compound used in the treatment of rheumatoid arthritis, has been tested in vitro against multiple bacteria including multidrug resistant pneumococci. Auranofin and derivative MH05 were subsequently tested in a murine sepsis model. They were shown to significantly reduce mortality and bacterial burden of infections with several pneumococcal strains (Aguinagalde et al. 2015). Currently, auranofin is in phase 2 clinical trials as therapy against M. tuberculosis (Butler and Paterson 2020).

CONCLUSION
Mostly in academia, major efforts are underway to identify potential new drug targets and to improve infection outcomes. Currently, there are three main strategies: (i) boosting host immunity by interfering with its immune responses, (ii) lowering pneumococcal virulence and (iii) developing novel antibacterials with a new mechanism of action (MOA). Each of these strategies has its own benefits and limitations. The first strategy often leads to adjuvant therapies, which can be used in concurrence with current antibiotics. While some of these therapies show highly promising results, no clinical trials have been set up to date. Importantly, interfering with host immune responses is a very delicate process. The biggest limitation to developing therapies interfering with host immunity is the risk of creating immune imbalances which can pose serious health threats (Bewersdorf et al. 2018). Still, this strategy is gaining attention, as publications increase and clinical trials are initiated. Secondly, lowering pneumococcal virulence can be done in multiple ways. The general idea is to inhibit important virulence factors to enable our own immune system to overcome the infection, without the need for bactericidal compounds. However, most of the proposed drug targets are pathogen specific. Notwithstanding their beneficial effects on disease outcome, use of these inhibitors requires certainty regarding the causative agent, which cannot always be guaranteed. Lastly, the development of novel natural antibacterials is hampered by the difficult culturing methods of soil bacteria, the primary source for natural antibiotics. Followed by the discovery of teixobactin, this issue was partially resolved by the development of the iChip technology (Ng and Chan 2016). Furthermore, while rational drug discovery (e.g. based on enzymatic screenings) shows promising results, the pharmaceutical industry is not keen on investing in novel antibiotics as investment costs are high, while the profits could be small (Arias and Murray 2015;Jackson, Czaplewski and Piddock 2018). Even when a potential hit is successfully identified, the emergence of resistance is always on the lure, lowering return on investments (Wright 2015). Therefore, the Infectious Diseases Society of America (IDSA) presented the 10 x '20 initiative ten years ago, in 2010, to develop and approve 10 novel, efficacious and safe systemically administered antibiotics by 2020 (Gilbert et al. 2010). While the initial goal is met, in 2019 IDSA released a publication stating that even the development of 20 novel antibiotics might still not be sufficient to tackle future drug resistance problems (Talbot et al. 2019). Therapies targeting the bacteria without killing them, i.e. anti-virulence therapies, might provide an answer to this issue. However, most research involving these targets is still in an early stage, with only liposomes as PLY inhibitors currently going through clinical trials.

Conflict of interest.
None declared.