Rethinking natural product discovery to unblock the antibiotic pipeline

Antimicrobial resistance is a current and growing problem. Much like climate change, it has been a slow-moving crisis that threatens to reach a tipping point without immediate interventions. Two important papers published in the last 2 years have produced the first global surveys of AMR. These papers present important insight into the complex demographics of AMR and how the burden falls unequally on developing countries [1,2]. They highlight the organisms (mostly ESKAPE pathogens) that pose the greatest threats and to whom. For example, they show, on one hand, deaths in children under 5 from sepsis have dropped by 60% since 1990. However, deaths in older patients from sepsis have doubled over the same period. In addition, while overall infant mortality resulting from infection has reduced, the proportion of those infants with AMR infections has increased. This suggests that children who develop sepsis have infections that are more difficult to treat [1]. This granularity of data will certainly allow more targeted interventions. As with any systemic crisis, the solutions to AMR are multifaceted and will require an interdisciplinary approach from improved sanitation and diagnostics, to changes in medical and agricultural practice [3]. However, even if public health interventions are successful in slowing the spread of resistant infections, the problem of new antibiotic discovery and development remains, as we do not currently produce enough new antimicrobials to outpace resistance [4]. Indeed, returning to the example above, many deaths from sepsis in children result from a lack of adequate antibiotics to treat resistant infections [1].

1. Natural products: exploiting the antibiotic makers

Our current antibacterial agents are mostly (~70%) derived from natural microbial sources (e.g., beta lactams, peptides, and macrolides) with some synthetic classes (e.g., fluoroquinolones). We have thus benefited enormously from the innate ability of microbes to biosynthesise complex molecules to utilize as antibiotics [5]. Actinomycetes (e.g., Streptomyces, Micromonospora) are well-known antibiotic producers responsible for several clinical staples [6]. These slow-growing bacteria and fungi mostly produce specialized metabolites, also known as natural products, to fend off competitors for scarce resources in their own ecological contexts [5]. This year marks the centenary of the first discovery, by Andre Gratia and Sara Dath [7,8], of antimicrobial activity by Streptomyces which have since been found to be prolific producers of biologically active metabolites of various classes including terpenes, polyketides, ribosomal, and non-ribosomal peptides. With the advent of genome sequencing, they have been shown to have huge unexploited biosynthetic potential. Interestingly, the gene clusters that encode natural product biosynthetic pathways also often include resistance genes and indeed in mixed microbial communities such as the soil, the plasticity of bacterial genomes allows antimicrobial resistance to evolve as a defense mechanism to enable bacteria to survive in these competitive environments.

As we know, antimicrobial resistance comes hand in hand with antimicrobial use, so careful antibiotic management is required to slow the spread of resistance and we must also ensure a steady supply of new antimicrobials.

2. Where will new antibiotics come from?

Clinically used natural product antibiotics have historically been isolated via fermentation of a producing bacterial or fungal strain. This is because the exceptional molecular complexity of natural product antibiotics means their chemical synthesis for manufacture is uneconomical and would make them unaffordable. Fermentation to produce natural products pioneered by Chain, Florey, and Heatley began a revolution in industrial microbiology that has enabled the production of myriad antibiotics in a comparatively simple and sustainable fashion. It demonstrated how we can exploit microbes as highly efficient producers of complex molecules.

So, what are our options for new antimicrobial development? There are several approaches available [9], all the subject of ongoing research, for example de novo drug design, i.e., creation of synthetic drugs (including in silico design) based on a known microbial target or combinatorial synthetic libraries for phenotypic testing. In addition, targeting the molecular mechanism of resistance can extend the usefulness of existing antibiotics, e.g., the development and coprescribing of beta lactamase inhibitors to maintain the effectiveness of beta lactam antibiotics.

Natural product antibiotic discovery was largely abandoned by industry due to the regular rediscovery of compounds as well as the difficulty isolating low-titer products. However, in recent years, there has been somewhat of a renaissance in the field [10]. Modern natural product discovery benefits from a variety of analytical methods with increased sensitivity, low-cost genome sequencing, and a vast library of microbial genome sequences alongside novel bioinformatic tools that allow “genome mining” for biosynthetic gene clusters [11]. These efforts are aided by a whole suite of new molecular biology tools to enable heterologous expression and manipulation of gene clusters. They have also allowed us to “bioprospect” for microbes from underexplored ecological niches, as well as looking beyond the canonical antibiotic producers to explore new genera, providing a wealth of natural product biosynthetic potential [12,13]. This work is vital in a new expansion of natural product discovery.

However, the derivatization of existing natural product antibiotics may give us the most immediate access to new compounds. Our reliance on microbial fermentation to produce many antibiotics has ironically limited the options to selectively modify the drugs to evade resistance mechanisms. Historically, semisynthesis, where the natural product is chemically modified to change, for example, pharmacological properties, has been employed for this purpose. Indeed, penicillins have been subjected to semisynthesis whereby the key intermediate (+)-6-aminopenicillanic acid is isolated from whole cells then derivatized via selective chemical modification to produce a variety of penicillin like antibiotics [14]. This approach has also been applied to the polyketide natural product, rifamycin, which is derivatized chemically to produce the clinically utilized TB antibiotic, rifampin. Unfortunately, due to the complex nature of many natural products, the ability to selectively chemically modify them is limited. However, a toolkit for derivatization of such molecules is being developed using a variety of modern approaches, including use of enzymes for biocatalytic modification or using pathway engineering.

3. Modern approaches to antibiotic derivatization

With our understanding of biosynthetic machinery, there have been increased efforts to use biosynthetic pathways to generate libraries of derivatives, particularly of polyketides (e.g., erythromycin) and non-ribosomal peptides (e.g., daptomycin and vancomycin). Both classes of natural products are biosynthesized by assembly line like enzymes known, respectively, as polyketide synthases (PKS) and non-ribosomal peptide synthetases (NRPS). The modular nature of these multidomain enzymes has made them extremely attractive as targets for pathway engineering and synthetic biology in efforts to create plug and play systems to generate new antibiotic derivatives [15,16]. There are multiple approaches including precursor directed biosynthesis, mutating substrate selection domains or domain/module swapping [17]. These approaches allow, for example, the incorporation of novel building blocks (e.g. non-natural amino acids) into the polymer (e.g. peptide), thus creating derivatives of the original natural product by altering the structure of the natural product core. These impressive approaches have had some success, but there are still aspects of these complex enzymes that we do not yet understand.

Alternative chemical and biochemical approaches also have significant promise. For example, one of the important structural features of many polyketide and non-ribosomal peptide antibiotics is the fact that most are not linear but cyclic. Cyclized peptides, for example, are more resistant to proteases and create a defined conformation for the molecule to interact with a target. However, macrocyclization is a very tricky chemical reaction, and thus there has been an interest in manipulating biological systems to produce libraries of cyclic peptides, including adapting ribosomal peptide biosynthesis in cells [18]. However, a feature of NRPS directed biosynthesis compared to ribosomal peptide biosynthesis, is their ability to incorporate non-proteinogenic amino acids. Recent efforts have been made to use expanded genetic code techniques to allow ribosomal introduction of non -proteinogenic amino acids [19].

Chemoenzymatic synthesis is the practice of using enzymes in chemical synthesis, usually for chemically challenging transformations. The significant increase in exploration of chemoenzymatic synthesis of complex molecules has been driven by rapid advances in protein biochemistry, from heterologous production to directed evolution, and the availability of genetic data to allow heterologous production and screening of enzymes to catalyse a desired transformation. These technical developments have allowed greater investigation and exploitation of the capacity of enzymes as catalysts for chemical synthesis. For example, the cyclization of large molecules (peptides or polyketides) is a challenging transformation in chemistry but one which nature achieves with comparative ease. Chemoenzymatic synthesis of cyclic peptides, rather than pathway engineering, is particularly attractive due to the ease with which linear peptides can be chemically synthesized using automated synthesizers and the ability to easily incorporate a wide variety of canonical and non-canonical amino acids. Thus, a process that allows the synthesis of peptides and enzymatic cyclization is very appealing. The challenge is in finding cyclases with broad enough substrate tolerance to maximize their utility as biocatalysts [20]. We also recently reported a bioinspired chemical cyclization method where we demonstrate that if we use the biosynthetically encoded linear peptide sequence and activate the termini in the correct fashion, we can generate the corresponding natural product cyclic peptide rapidly and in high conversion and faster than many enzymatic transformations [21].

A further feature of natural product biosynthesis is so-called “tailoring” whereby additional enzymatic steps are carried out after biosynthesis of the core cyclic peptide or macrolide to introduce further modifications that often modulate biological activity. This is also a principle of medicinal chemistry, known as late-stage functionalisation, whereby a core structure is modified in various ways to alter, e.g., physico-chemical properties, bioavailability, etc. Enzymes are thus highly attractive as reagents for derivatizing complex molecules, as they can often catalyze chemistry in a highly specific manner and carry out reactions that are impossible using standard synthetic chemistry. Biocatalytic late-stage functionalisation could thus enable further semi synthesis of natural products as enzyme catalyzed reactions are milder and more selective than traditional synthesis and thus are likely to be more compatible with complex natural product structures. There are now many examples of chemoenzymatic synthesis in industry [22].

4. The future: opportunities for innovation and collaboration

It is clear that there are an array of exciting approaches to the discovery and development of antibiotics. Indeed, a combined approach can be envisaged whereby starting from an existing antibiotic or novel natural product, a workflow is developed in which chemical synthesis or even whole cells are used to generate a core structure that is modified by a variety of enzymes to create a library of complex molecules. Upon identification of a hit and following optimization, a bacterium could be engineered to scale up production for testing and manufacture. This brings together multiple technologies from high throughput synthesis, biocatalysis, and directed evolution of enzymes to pathway engineering and synthetic biology. We can also be optimistic that machine learning will reduce the experimental workload at several of these stages, speeding up many processes [23,24]. Indeed, outlining what the landscape of antibiotic development and manufacture looks like in the future, considering new technologies and imperatives such as antibiotic waste management and sustainability, may enable some focus on how we overcome some of the barriers to investment and derisk these projects in the future.

Antibiotic production is a biotechnology success story that began over 100 years ago. Successful translation to the clinic was built on decades of fundamental scientific investigation and was enabled by public-private and international partnerships. A century later, it is worth noting that beta lactam antibiotics like penicillin are still ~65% of the antibiotic market [25]. Thus, that initial investment in fundamental science has paid off many times over, in both economic benefit and the myriad medical innovations they enabled (joint replacements, transplants, heart surgery, chemotherapy etc.), improving quality of life and saving millions of lives. The same spirit of innovation now needs to be applied to enable a second golden age of antibiotic discovery and development, to introduce greater diversity into the antibiotic portfolio by accessing the full reach of the chemical space that natural products afford.

Much like climate change, the AMR crisis is not the future, it is already upon us, and thus we cannot wait longer to act. Recently, COVID-19 vaccine development again demonstrated what is possible with concerted international, industry, and government collaboration built on a solid foundation of fundamental science [26,27]. Recent government initiatives for antimicrobial development (e.g., UK subscription model, US Pasteur Act, EU antimicrobial incentive program) are promising, as more investment is needed urgently to secure the future production of these key medicines that underpin global healthcare [28,29]. We can rise to this urgent challenge by capitalizing on innovative fundamental research across the world to ensure we do not return to a pre-antibiotic age.

Funding

Work on natural product synthesis and antimicrobials in the Barry lab is currently funded by an MRC impact Acceleration Award [MR/X502923/1] and a King’s Climate & Sustainability Seed Fund Award.

Disclosure statement

The author is an inventor on patent Ding, Y., Müller, M., Barry. S. M. Bioinspired natural cyclic peptide synthesis. UK Patent Application No 2402551.2 King’s College London.