Bacterial carbonic anhydrases: underexploited antibacterial therapeutic targets

Brief history & physiology of targeting human carbonic anhydrases

Carbonic anhydrases (CA) are metalloenzymes that are present in all kingdoms of life. This family of enzymes primarily catalyze the interconversion of CO₂ to a bicarbonate anion and a proton; this reaction is mediated through a hydroxide anion bound to a catalytic Zn²⁺ ion within the active site of the enzyme [1]. This reaction is important for a variety of biological processes that take place within both prokaryotes and eukaryotes. In humans, CAs have been extensively studied and have long been validated as drug targets, beginning with the first clinical use of acetazolamide (AZM) in 1952 as a diuretic [2]. Since that time, 15 isoforms of human CAs have been identified in various tissues, organs and subcellular locations. These isoforms play an important role in many physiological processes, including CO₂ transport in the lungs, regulation of CO₂ and pH homeostasis in other tissues and electrolyte balance, to name a few [3]. Regulation of these processes has made CA inhibitors (CAIs) effective treatment options for glaucoma, epilepsy, congestive heart failure and altitude sickness, in addition to their aforementioned use as a diuretic [4]. More recently, CAs have become targets for cancer therapy, because tumor-associated CA IX is overexpressed in hypoxic conditions to regulate the pH of cancer cells [3].

Recent successes for bacterial CAs as drug targets

Although human CAs have been exploited as therapeutic targets for decades, the hypothesis that CAs are relevant drug targets for the development of anti-infectives has also been known for quite some time [5]. However, bacterial CAs have yet to gain the same traction as their human counterparts as targets for drug development. For example, in the 1980s it was observed that treating patients with AZM yielded around 95% healing of gastric ulcers [6]; at first, this effect was thought to be mediated through pH regulation of the stomach lining. However, it was later found that AZM possessed antibacterial activity against Helicobacter pylori, the bacterial pathogen associated with peptic ulcers, and that the target was likely inhibition of bacterial CAs expressed by H. pylori. Patients treated with AZM reported very low rates of recurrence (6%) compared with those treated with the traditional options of antacids or proton pump inhibitors (associated with 34 and 60% recurrence of peptic ulcers, respectively) [7]. This suggested that the antibacterial activity of AZM against H. pylori may partially contribute to the healing of and prolonged protection against peptic ulcers. Additionally, in the 1960s microbiologists reported that strains of Neisseria species produced CAs; interestingly, AZM as well as a second CAI, ethoxzolamide, inhibited growth of these strains [8]. Even with these two examples of effective CAIs with antibacterial activity, there is a lack of concerted effort to leverage bacterial CAs as viable antimicrobial targets. There are a few groups globally studying CAs as drug targets, with efforts mostly focused on identification, enzymatic characterization and inhibition of bacterial CAs in vitro (recently reviewed by Supuran and Capasso [9]). Several research groups have translated CAIs into antimicrobial assays for drug development [10–13] and recently have demonstrated the utility of AZM in vivo as an effective agent against vancomycin-resistant enterococci [14].

Challenges & opportunities for targeting bacterial CAs

Alhough there have been recent successes in identifying CAIs with antibacterial activity, there are still considerable challenges that may impede bacterial CAs from being fully adopted as viable therapeutic targets. Some of these challenges are: bacteria express many subfamilies of CAs of which relatively little is known compared with the human CA isoforms; bacterial CA diversity is quite high, which may limit the potential to develop broad-spectrum antibiotics; there may be varying degrees of essentiality among bacterial CAs, even within the same species; there is concern over off-target human CA interactions that could preclude CAIs from being viable antimicrobial targets; and CAI drug design may no longer be considered innovative. Targeting bacterial CAs is likely not a one-size-fits-all strategy for broad-spectrum antibiotic development. However, where challenges are present, there may also lie considerable opportunities.

Humans only express the α-family of CAs, while bacteria express the α-class as well as the β-, γ-, ι-classes of CAs [15]. These additional CA classes possess wide mechanistic and structural diversity. For example, most vertebrate α-CAs are monomeric (with the exceptions of CA IX and XII) and contain a deep cavity in the active site wherein three histidine residues are present to co-ordinate a catalytic Zn²⁺ ion, which activates a water molecule for nucleophilic attack of CO₂. β-class CAs are active in dimers or tetramers with two or four identical active sites that contain one histidine and two cysteine residues to co-ordinate the Zn²⁺ ion. This class is also pH dependent because an aspartate residue co-ordinates with the Zn²⁺ ion at pH <8.3, which effectively displaces the water molecule needed for catalysis. At pH >8.3, the aspartate is removed and the water molecule is able to co-ordinate the Zn²⁺ ion for catalytic activity. The γ-CA class is present as a catalytic trimer whose three active sites are located at the interface of each monomeric subunit which provides an extended, deep groove where the metal ion is positioned at one end. This subfamily uses multiple ions, including Zn²⁺, Fe²⁺ and Co²⁺. The ι-CA class was recently discovered in bacterial genomes and is currently being characterized.

The complexity of the CA phylogeny in bacteria may be compounded by varying degrees of essentiality among different bacterial species. To date, there have been limited studies on the essentiality of CAs in different species. For example, the α-CA expressed in Neisseria gonorrhoeae was found to be essential for bacterial survival [16]. In contrast, species of proteobacteria retain a CA gene but do not need it for survival [17]. Clearly, there is significant work to be done to validate bacterial CAs as viable drug targets in different species of bacteria; however, a great reward can be gained by discovering new strategies to combat antibiotic resistance in infections caused by hard-to-treat pathogens.

If CAs are identified as essential for survival in high-priority pathogens, then the diversity of CA classes and architecture may give rise to diversity in the design of novel CAIs. With few exceptions, the majority of CAI drug design has centered on the α-CA class, because this has long been the only clinically relevant subfamily for drug discovery. This includes ligand-bound x-ray crystallographic data of CAIs to drive structure-based drug design. As a case in point, CAIs have never been solved in complex with bacterial CAs other than the α-class. There are a few AZM-bound structures of β-CAs from algae and fungi, but the number still lags far behind the wealth of structural information available for α-CAs. Structural data for γ-CAs is even more limited, with no ligand-bound data to inform designs targeting this subfamily. Furthermore, there are only two structures for the newly discovered ι-CA class.

As CAs from other subfamilies are identified as essential in bacterial pathogens, the incentive to solve ligand-bound structures in these classes for the purpose of structure-based drug design will only increase. Moreover, some bacterial species express multiple types of CAs. If these are shown to all be essential within the same bacteria, a possibility exists for polypharmacology (one drug inhibiting multiple targets) that could reduce the potential for resistance to arise. This possibility will only be realized if more research is done to validate the various bacterial CAs within each pathogen; there is even a likelihood that additional CA genetic families may be discovered.

The final perceived hurdle to bacterial CAs being adopted as viable antibiotic targets is the concern over off-target inhibition of human CAs. Most CAIs possess metal co-ordination functional groups such as sulfonamides or hydroxamic acids which bind to the active site Zn²⁺ ion to inhibit the enzyme. These functionalities would likely still be necessary in order to inhibit bacterial CAs, potentially making it difficult to fully prevent binding to human CAs. Nonetheless, even within the α-CA subfamily, there are exploitable structural differences between bacterial and human α-CAs which are relevant to the design of selective inhibitors for bacterial α-CAs to at least widen the therapeutic window. This has recently been demonstrated as a feasible approach for new inhibitors of Vibrio cholerae CAs. Analogs were designed that displayed low nanomolar potency against the α-CA from V. cholerae with micromolar activity against human CA I and high nanomolar potency versus human CA II [18]. Selectivity against human isoforms was also achieved for CAIs inhibiting the N. gonorrhoeae α-CA [11].

We posit that, as more data emerges implicating bacterial CAs as viable therapeutic targets, the motivation to carry out rational design for selectivity over human CA isoforms will arise and lead to innovative approaches for CAI design that have previously not been necessary. Furthermore, the rationale to reduce human CA binding would likely not be motivated by reducing off-target toxicity. In fact, AZM is often prescribed at doses of 1 g/day (or more) chronically, with few side effects or associated toxicity issues. Safety and tolerability studies with AZM found that 90% of patients tolerated doses >1 g/day, while 44% of patients tolerated up to 4 g/day for a period of 6 months [19]. While newly developed CAIs will likely possess different tolerability profiles, it is encouraging that inhibition of human CAs may not be an impediment for translation of bacterial CAIs for therapeutic development from a toxicity point of view. However, reduction of human CA binding would have value from a drug disposition perspective. Two of the most highly expressed human CAs, I and II, are present in high amounts in red blood cells. This is, at least in part, a reason why CAI doses are high in humans; once the drug is absorbed into the bloodstream, it readily partitions into the red blood cells over plasma at a ratio of about 3:1, which effectively sequesters the molecule from the site of action [20]; therefore reduced binding to one or both of these human CAs will allow more drug to be free to reach the site of action.

Future perspective

In summary, while CA inhibition for antimicrobial drug discovery is not necessarily a new concept, the field is still in its infancy relative to human CA drug discovery. There has been considerable progress recently to validate bacterial CAs as viable therapeutic targets for antibiotic drug discovery, but much more ground needs to be covered. Regardless, with the recent report of bacterial CA inhibitors exhibiting antibacterial efficacy in vancomycin-resistant enterococci animal models, there is evidence that bacterial CAs are worthy targets to pursue for antibiotic drug discovery. Though there are many opportunities for this field to grow, it will require an interdisciplinary effort with microbiologists, biochemists, structural biologists and medicinal chemists working in collaboration to realize the full potential of CAIs with antibacterial activity.