Engineering bacteria to modulate host metabolism

Abstract

The microbial community of the gut, collectively termed the gut microbiota, modulates both host metabolism and disease development in a variety of clinical contexts. The microbiota can have detrimental effects and be involved in disease development and progression, but it can also offer benefits to the host. This has led in the last years to the development of different therapeutic strategies targeting the microbiota. In this review, we will focus on one of these strategies that involve the use of engineered bacteria to modulate gut microbiota in the treatment of metabolic disorders. We will discuss the recent developments and challenges in the use of these bacterial strains with an emphasis on their use for the treatment of metabolic diseases.

1. INTRODUCTION

Human health and disease risk are influenced by the microbial communities (bacteria, archaea, bacteriophages, eukaryotic viruses, fungi, and single‐cell eukaryotes), referred to as the microbiota. These microbial ecosystems are found on all human mucosal surfaces, and most are commensal or mutualistic microorganisms that contribute to important functions in the body. These roles vary from training the host immunity, digesting food, modifying drug action and metabolism, eliminating toxins, and producing numerous small molecules that can act as signaling molecules and influence host metabolism. ¹ The gut and its microbiota are in close relationship with the liver and other organs as almost all venous blood from the gut reaches the liver via portal circulation, followed by drainage from the liver into the systemic circulation. Therefore, pathogen‐associated molecular patterns (PAMPs), gut‐produced metabolites, and signaling molecules can reach everywhere in the body explaining partly why the microbiota is important in the body's homeostasis. A substantial body of research from recent years suggests that alterations in the composition and function of the microbiota can modify the host metabolism and drive different diseases, influencing their progression and their response to therapies. These alterations can stem from variations in the beneficial microbial communities or specific members of it, to increase or colonization with microorganisms with deleterious effects on the host or changes in the bacterial metabolite production. ¹ , ² , ³ Therefore, different microbiota‐based strategies were developed to restore or correct these alterations. Their aim is to replace (fecal microbiota transplantation), enhance (probiotics, prebiotics, symbiotics, engineered bacteria), suppress (antibiotics, bacteriophages) the microbiota or members of its community, or harness the microbe‐derived metabolites (also called postbiotics) and their related signaling pathways. ³ The advantages of most of these approaches in different medical conditions have been largely addressed in multiple reviews. ¹ , ² , ³ , ⁴

Additionally, recent advances in synthetic biology have led to the development of genetically engineered bacteria and fungi, which are considered next‐generation microbiota therapeutics. ⁵ Regulatory agencies such as the United States Food and Drug Administration include these therapeutics in the category of live biotherapeutic products (LBP), defined as any biological product that contains live organisms (but is not a vaccine) and that can be used in the prevention and treatment of a disease. ⁶ A recombinant LBP includes microorganisms that have been genetically modified (deletion, addition, or modification of genetic capacities). This definition delimits these products from the classic probiotics, which are live microorganisms that confer a health benefit on the host and that are considered dietary supplements and not treatments. ⁷

In contrast with traditional treatments, recombinant living therapies could be of great interest in chronic illnesses as they could deliver the intervention in a constant manner and for a long period. ⁸ Their field of application is very wide and ranges from bacteria and fungi engineered to deliver therapies that otherwise degrade before reaching their target, reduce systemic drug exposure and systemic side effects, activate the immune system, and record transient signals for noninvasive testing. ⁵ , ⁹

With the increasing understanding of how the microbiota modulates diseases, these engineered microorganisms gained particular interest as they can target specific functions of the microbiota that are impaired in disease and restore them. Microorganisms can be reprogrammed to perform complex tasks by using different genetic approaches and circuits. ⁸ They can also be designed to express functions that address monogenic inborn errors of metabolism (e.g., phenylalanine metabolism in patients with phenylketonuria) ¹⁰ , ¹¹ and other orphan diseases.

In this review, we summarize data on how engineered microorganisms can be used to influence host metabolic functions and discuss the challenges that this approach in the quest of becoming a widely used treatment. We focus on examples of gain‐of‐function genetic engineering and examples that have been tested in different metabolic diseases in both animal models and clinical trials.

2. BUILDING NEXT‐GENERATION MICROBIOTA THERAPEUTICS

An engineered microorganism must comply with several key requirements to induce a significant metabolic impact. After choosing a function of interest to be manipulated, a key step is the selection of a microbial host organism, or chassis, which would enable environment‐sensing, regulate gene expression, and production of a therapeutic product. ¹² Historically, the first engineered bacteria were designed to target DNA that encodes for a function of interest (mainly enzymes) to induce transcriptional and translational changes. This can be achieved by shearing, splicing, and integrating the target genes into the cell using a wide range of methods (e.g., gene transfer, transfection, transduction, protoplast fusion, conjugative transfer, or lysogenic conversion) (reviewed in ¹³ , ¹⁴ ). Other techniques can be used to directly integrate the gene into the bacterial chromosome (homologous recombination, site‐specific recombination, transposable recombination, and CRISPR–Cas9) (Figure 1). The choice of the method is mainly driven by the size of the DNA that must be inserted and the chassis used. Indeed, for large genes (e.g., larger than 100 kb) bacterial artificial chromosomes, ¹⁵ transformation‐associated recombination or phage recombination systems showed promising results. ¹⁶ Using these techniques, bacteria can be programmed to activate the inserted gene only upon the onset of a certain stimulus as in the case of memory circuits ¹⁷ or several stimuli such as in the case of genetic logic gates. ¹⁸ This is of particular interest as the production of a certain molecule or enzyme can be triggered only when needed. Nevertheless, even with the development of such a wide range of techniques, it is still challenging to create an engineered microorganism as most chassis will inactivate the expression of foreign genes. ¹⁴ Probiotic microorganisms such as Lactobacillus spp. and Bifidobacterium spp., but also well‐characterized members of the microbiota such as Bacillus spp., Streptococcus spp., and Escherichia coli, ¹⁹ have been used as chassis to enable the construction of engineered synthetic living therapies. In addition to these bacteria, probiotic yeasts such as Saccharomyces boulardii have also been engineered in recent years ²⁰ and showed promising results. These chassis were frequently chosen because of their biosafety profile (probiotic bacteria or yeast) and the ease of genetic manipulation and rapid growth in the laboratory (E. coli). The environmental niches a strain can grow in also determine the choice of the chassis as it faces many challenges to survival in the luminal and mucosa‐associated environment, mediated both by the host (e.g., peristalsis, innate and adaptive immunity) and by other native microorganisms (e.g., competition, niche availability). ²¹ Moreover, host baseline characteristics and microbiome composition are the main drivers of the different colonization patterns of probiotics ²² that should be taken into consideration in order to provide long‐term, autonomous, and efficient therapies. For a stable engraftment that will provide a significant and durable metabolic effect, they may need concomitant antibiotic treatment, in order to allow their implantation, or multiple administrations to induce a positive pressure at the gut level. ⁸ Moreover, factors such as abundance, degree of colonization, and interaction with the immune system vary widely between strains and could be critical for the successful implementation of a given application. ⁹ Strategies to overcome these challenges include the utilization of bacterial spores ²³ or the use of native, microbiota‐derived bacteria. Once the strain is implanted, the robustness of engineered functions to time and changing environments must be high and is a major challenge to long‐term cellular therapies. ⁵ Finally, engineered bacteria may be very expensive, particularly if complex genetic editing is involved, and this would limit their translation into clinical practice. ⁸ Once all these technical challenges are resolved, and for the product to be used by humans, it must feature a positive benefit–risk ratio. For this, robust quality, safety, and efficacy data are needed and are the basis of any reglementary evaluation. However, developers of LBP face several additional challenges: although the product has no direct systemic passage, it can modify the host physiology directly and indirectly through its secreted metabolites and local immunomodulatory functions. Furthermore, evaluating toxicity is challenging as it is not always dose dependent. Therefore, the developers of these LBP must use extensive combinations of in vitro, ex vivo, and in vivo models to gather broad data that will prove high safety and low toxicity profiles. ²⁴ Of particular interest in this context are the new ex‐vivo methods that have been developed in the last years such as organoids and organ‐on‐chip approaches. Their use could bring important insights into the evaluation of these products especially since they can be derived directly from human samples. In the following sections, we will focus on how engineered bacteria can be used to rewire the host metabolism and treat metabolic diseases. Technical aspects related to the generation of these bacteria are covered elsewhere. ⁵ , ⁸ , ⁹ , ¹⁹ , ²⁵

FIGURE 1.

Common synthetic biology methods used to generate engendered micro‐organisms (CRISPR: clustered regularly interspced short palindromic repeats, Cas9: CRISPR‐associated protein 9, created with BioRender.com).

3. ENGINEERED BACTERIA TO METABOLIZE TOXIC BIOMOLECULES

The gut contains a wide range of metabolites derived from the host, the microbial community, and food. Both the host and the microbiota can use some of these metabolites, and there are numerous overlaps and complementarity between the host and microbial metabolic pathways. ²⁶ Any disruption in these pathways could lead to the accumulation of intermediates, impacting both the host metabolic homeostasis and the composition of the microbiota. Consequently, bacterial cell‐based biosensors could be created to perform activities based on sense and response schemes, detecting a biomarker for disease and inducing a particular metabolic pathway. ⁸ This approach was used in several metabolic conditions (Table 1), in which there is an accumulation of toxic metabolites because the host is not metabolizing them or because the microbiota pathways are defective.

TABLE 1.

Engineered microbes developed to modify host metabolism.

Tageted biomolecule	Chassis	Mechanism	In vivo model	Species	Administration time	Effect	Phase	Study type	Condition	Duration	Effect	Adverse events	NCT
	Preclinical testing						Clinical testing
Ammonia	Lactobacillus plantarum	Hyperconsumption of ammonia	Ornithine transcarbamoylase‐deficient Sparse‐fur mice	Mouse	3 days	↓ Ammonemia
			Carbon tetrachloride rat model	Rat	3 days	↓ Ammonemia
			Thioacetamide‐induced acute liver failure mice	Mouse	3 days	↑ Survival ↓ blood and fecal ammonia, ↓ astrocyte swelling in the brain cortex
	E. coli Nissle (SYNB1020)	Overproduce of arginine	Ornithine transcarbamylase‐deficient spfash mice	Mouse and monkeys	28 days	↓ Systemic ammonemia, ↑ survival	1	First‐in‐human, Oral Single and Multiple Dose‐Escalation, Randomized, Double‐blinded, Placebo‐controlled	Healthy	14 days	↑ Urinary nitrate, plasma 15 N‐nitrate, urinary 15 N‐nitrate		NCT03179878
			Thioacetamide‐induced acute liver failure mice			↓ Ammonemia	1b/2a	Randomized, Double‐blind, Placebo‐controlled Study	Cirrhosis	7 days	Lack of efficacy		NCT03447730
	E. coli Nissle 1917	Overproduce of arginine (S‐ARG)	Bile‐duct ligated (BDL)	Rats	3 and 5 weeks	↓ Ammonemia ↓IP‐10 and MCP‐1 at week 3 no effect on ALT/AST/ALP/GGT/albumin/bilirubin and gene expression of liver function markers (IL‐10/IL‐6/IL‐1β/TGF‐β/α‐SMA/collagen‐1α1/Bcl‐2). ↓Colonic mRNA (TNF‐α/IL‐1β/occludin) markers at both timepoints
		Overproduce of arginine+butyrate production	Bile‐duct ligated (BDL)	Rats	3 and 5 weeks	↓ Ammonemia ↓memory impairment ↓systemic inflammation (IL‐10/MCP‐1/endotoxin) at both timepoints (except 5‐week endotoxin) ↓Colonic mRNA (TNF‐α/IL‐1β/occludin) markers at both timepoints
Phenylalanine	E. coli Nissle (SYNB1618)	Phe degrading enzymes (PAL, LAAD)	Pahenu2/enu2 PKU	Mouse	One dose	↓ Blood Phe independent of dietary protein intake	1/2	Dose‐escalating, randomized, double‐blinded, placebo‐controlled		1–3/day, 7 days	Safe and well tolerated with a maximum tolerated dose of 2 × 1011 colony‐forming units. Bacteria was cleared within 4 days of the last dose. ↑ (Dose‐responsive) in strain‐specific Phe metabolites in plasma (trans‐cinnamic acid) and urine (hippuric acid)	Gastrointestinal and of mild to moderate severity	NCT03516487
			Heathy	Cynomolgus monkeys	One dose	Inhibited increases in serum Phe after an oral Phe dietary challenge
	E. coli Nissle (SYNB1934)	Phe degrading enzymes with increased PAL activity	Heathy	Cynomolgus monkeys	One dose	Two‐fold increase in in vivo PAL activity compared to SYNB1618	1	Dose‐escalation, Placebo‐ and Active‐Controlled Crossover Study	Healthy	At least 3/day	Completed, results pending		NCT04984525
							2	Open‐label, dual‐arm study of either a SYNB1618 or SYNB1934		15			NCT04534842
Homocystein	E. coli Nissle SYNB1353	Consumes methionine					1	Dose‐escalation, Randomized, Placebo‐Controlled Study	Healthy	7 days	Recruiting		NCT05462132
Oxalic acid	E. coli Nissle (SYNB8802)	Consumes oxalate		Mice, non‐human primate		↓ Urinary oxalate excretion	1	Double‐Blind, Randomized, Placebo‐controlled Study	Gastric bypass surgery or short‐bowel syndrome	12 days	Recruiting		NCT05377112
							1a/b	Multiple dose‐escalation, randomized, double‐blinded, placebo‐controlled study	Healthy volunteers (HV) and subjects diagnosed with enteric hyperoxaluria (EH)	6 days	Recruiting		NCT04629170
Uric acid	SYNB2081	In developpement
Acetaldehyde	Bacteroides subtilis		Rats		90 days	Non toxic
	Bacteroides subtilis	Expression of acetaldehyde dehydrogenase	Alcohol feeding	Mouse	8 days	↓ Blood alcohol, ↓ALT, AST, PAL, liver malondialdehyde levels and ↑ total antioxidant capacity and superoxide dismutase levels in mouse liver.
Bile acids	E. coli	Expression of bile salt hydrolasis	Mixed meal challenge	Mouse	Single oral administration	↑ Total fecal bile acids, ↓ primary fecal ↓ primary conjugated BA, ↓ primary unconjugated bile acids were, increased expression of Fxr, Shp, and Cyp27a1 ↓ postprandial blood glucose ↓ insulin levels
			Ob/Ob	Mouse		No changes in body weight, ↑ insulin sensitivity
Glp‐1	Lactobacillus gasseri ATCC 33323	GLP‐1 secretion	Diabetes (STZ model)	Rat	2/day for 90 days	↑ Insulin, ↑ glucose tolerance, ↑ insulin‐producing cells within the upper intestine
	Lactococcus lactis MG1363	GLP‐1 secretion	Obesity‐Induced by High Fat Diet in Mice	Mice	Every 2 days for another 30 days	↓ The bodyweight ↑ glucose intolerance ↓ serum and liver triglyceride, ↑ liver expressions of peroxisome proliferator‐activated receptors α (PPARα) and its target genes ↑ intestinal microbial diversity.
	Lactobacillus plantarum‐pMG36e‐GLP‐1	GLP‐1 secretion	Spontaneous type 2 diabetes mellitus (T2DM)	Monkeys	7 weeks	↓ Fasting blood glucose ↓weight
Il22	Lactobacillus reuteri	Il22 secretion	Diet‐Induced Obesity	Mouse	8‐week treatment	↓ Liver weight ↓ triglycerides
	Lactobacillus reuteri	Il22 secretion	Alcohol (Lieber–DeCarli)	Mouse	10 days treatment	↓ Liver damage, inflammation, and bacterial translocation to the liver ↑ expression of REG3G in intestine
Butyrate	Bacillus subtilis	Butyrate production	HFD	Mouse	14 weeks	Retard body weight gain induced and visceral fat accumulation of mice, ↑ glucose tolerance and insulin tolerance, ↓ liver damage
	Bacillus subtilis (SCK6)	Butyrate production	HFD		14 weeks	↓ Body weight, body weight gain, and food intake ↓ blood glucose, insulin resistance

One such metabolite is ammonia, a by‐product of nitrogen metabolism, produced both by the host through the action of glutaminase located within enterocytes of the small intestine and colon, and by the microbiota through the action of urease‐producing bacteria. Ammonia derived from the gut is absorbed into the hepatic portal circulation and is metabolized by the liver under normal physiological conditions. However, in the context of chronic liver disease, the capacity to metabolize ammonia is impaired and the accumulation of ammonia leads to encephalopathy, a broad spectrum assortment of neuropsychiatric disturbances associated with both acute and chronic liver failure. ²⁷

As a majority of systemic ammonia is generated in the gut by bacterial protein degradation and microbial urease activity, engineered bacteria on a Lactobacillus plantarum chassis ²⁸ and probiotic E. coli Nissle, SYNB1020, ²⁹ were developed. These strains have an increased arginine biosynthesis that accelerates the conversion of ammonia from the gut to nontoxic L‐arginine. One strain of E. coli Nissle‐producing arginine was further modified to additionally synthesize butyrate. ³⁰ All these strains showed promising results in preclinical models of hyperammonemia, in lowering ammonia and improving the survival of mice and rats. Moreover, rats receiving the strain producing both arginine and butyrate featured an additional memory preservation. ³⁰ E. coli Nissle SYNB1020 was further tested in humans and in a phase I study, it was well tolerated for up to 14 days, and reached a steady state by the second day of dosing. Excreted bacterial cells were alive and metabolically active but were no longer detectable in feces 2 weeks after the last dose. ²⁹ However, the randomized, double‐blind, placebo‐controlled phase Ib/IIa study (NCT03447730) in patients with cirrhosis and elevated blood ammonia yielded negative results, with no difference in blood ammonia or changes in other exploratory endpoints noted relative to placebo despite evidence of activity of the strain as measured by plasma and urinary nitrate increases. Development of this strain was halted in view of these results (Synlogic website https://www.synlogictx.com).

In a similar approach, a modified E. coli Nissle is being tested as a potential biotherapeutic for phenylketonuria (PKU), a rare genetic caused by biallelic mutations in the PAH gene that results in an inability to convert phenylalanine (Phe) to tyrosine and that leads to severe neurological complications due to elevated blood Phe levels. ²⁶ Using the probiotic chassis, the genes encoding phenylalanine ammonia‐lyase and l‐amino acid deaminase were inserted into the bacterial genome, thereby allowing for bacterial consumption of Phe within the gastrointestinal tract (SYNB1618). This approach showed positive results in animal models. ¹¹ In a phase 1/2a randomized, placebo‐controlled, double‐blind, multicenter study in PKU patients, the strain was well tolerated, cleared within 4 days of the last dose, and showed a dose‐responsive increase in strain‐specific Phe metabolites in plasma and urine. ¹⁰ This strain was further engineered (SYNB1934) to generate higher bacterial PAL activity. ³¹ The strains were then compared in a Phase 2 open‐label 28‐day study (Synpheny‐1 Study) in patients with PKU and both strains demonstrated clinically meaningful reductions in fasting plasma Phe (www.synlogictx.com). Based on these positive results, the company developing these strains is planning to start a pivotal phase 3 study in the first half of 2023 (www.synlogictx.com).

Other rare metabolic diseases such as homocystinuria (HCU) or hyperoxaluria are also considered candidates for the use of engineered bacteria. Homocystinuria is a rare genetic disorder characterized by high levels of homocysteine and risks including thromboembolism, lens dislocation, skeletal abnormalities, developmental delay, and intellectual disability. A candidate biotherapy that consumes methionine in the gastrointestinal tract (SYNB1353) is in phase 1 clinical development (NCT05462132). Enteric hyperoxaluria is a metabolic disorder characterized by an accumulation of oxalate that leads to chronic kidney stones. E. coli Nissle, engineered to metabolize oxalate has shown promising results in mice and nonhuman primates. ³² The drug is now in phase 1 clinical testing (NCT04629170 and NCT05377112). Other clinical targets for modified E. coli Nissle capable of lowering toxic metabolites include gout, but the development of this strain (SYNB2081) is still in the early phases.

Engineered bacteria could also be used to clear toxic metabolites resulting from alcohol exposure. Alcohol is metabolized in hepatocytes to acetaldehyde and, further on, to acetate. Acetaldehyde is extremely toxic and carcinogenic. ³³ Using homologous recombination on a food‐grade probiotic chassis, Bacteroides subtilis, both alcohol dehydrogenase from Saccharomyces cerevisiae (scADH) and aldehyde dehydrogenase from Issatchenkia terricola (istALDH) were integrated into the bacterium. In vivo experiments in a mouse model of short‐term alcohol exposure showed that this strain alleviates some features of alcohol liver disease. ³⁴ A similar strain expressing aldehyde dehydrogenase (patent US10849938B2) featured a safe profile in rats ³⁵ and its clinical development is ongoing (https://zbiotics.com/).

Most of these approaches used probiotic strains for their biosafety profile. However, these strains need multiple administrations as they do not colonize the gut. To overcome this issue, native gut bacteria (hence termed ‘precision probiotics’) could be used as chassis. In a recent study, a native strain of E. coli was modified to increase its bile salt hydrogenase activity. ¹² Bacterial BSH is a key enzyme in the homeostasis of bile acids and alterations in the microbiota and the subsequent alterations in its ability to metabolize bile acids are associated with several metabolic diseases including obesity, NAFLD, and alcoholic liver disease. ³⁶ In this proof‐of‐concept study, native E. coli expressing a high BSH activity durably colonized mice modified their bile acid pool and induced improved insulin sensitivity and glucose tolerance upon consumption of a high‐fat diet. Moreover, using a genetic model of type 2 diabetes, the Ob/Ob mice, mice colonized with this strain by a single gavage, 3 months earlier, had significant improvements in insulin sensitivity compared with the control cohort. ¹² This seminal study opens new avenues in using engineered microorganisms as personalized microbiota editing strategies (Figure 2). As the functional capacity of the microbiota is altered in different diseases, these changes could be further characterized in an individualized manner and harnessed to be used as therapeutic targets. In addition to bile acids and BSH, other bacterial‐related pathways such as indole derivatives and SCFA production could be modulated as synthetic biology techniques exist for both of these pathways and showed positive results in animal models of metabolic diseases. ³⁷ , ³⁸

FIGURE 2.

Example of the potential use of engineered bacteria in treating metabolic syndrome. Patients suffering from obesity and metabolic syndrome present changes in the intestinal microbiota functions such as decreased bile salt hydrolase activity. The use of host bacteria engineered to express BSH could improve metabolic syndrome features such as glycemic response and dyslipidemia (BSH, bile salt hydrolase, created with BioRender.com).

These examples show that transgene delivery using probiotic or native chassis can metabolize host metabolites and modify the serum metabolome and host physiology and disease risk.

4. ENGINEERED BACTERIA TO PRODUCE BIOACTIVE BIOMOLECULES FOR THE TREATMENT OF METABOLIC DISEASES

Bacteria can also be used to produce beneficial metabolites that can improve host metabolism and disease. One of the most active fields of research in this aspect is diabetes mellitus, a metabolic disorder in which glucose metabolism is disrupted due to the lack of insulin‐producing pancreatic β cells (type 1) or reduced insulin responsiveness (type 2). The pathogenesis of diabetes is complex and includes the intervention of multiple hormones and signaling molecules at different body sites (pancreas, liver, gut, muscle). Glucagon‐like peptide‐1 (GLP‐1) is one such example, as it is produced in the intestine upon food consumption, which stimulates the production of insulin in the pancreas. ³⁹ GLP‐1 agonists have shown good results in the treatment of obesity, ⁴⁰ diabetes, ⁴¹ and NASH, ⁴² but their application is limited due to the peptide's short half‐life and the need for regular administration. To overcome these shortcomings, engineered bacteria were developed to produce GLP‐1 directly in the gut. An engineered Lactobacillus gasseri that expresses an inactive full‐length version of GLP‐1 showed that this approach is efficient in a rat model of diabetes. This strain reprogrammed intestinal cells into glucose‐responsive insulin‐secreting cells leading to increase glucose tolerance. ⁴³ Similar approaches using different chassis, such as Lactococcus lactis ⁴⁴ and Lactobacillus plantarum, ⁴⁵ had comparable results in other models of diabetes and host species (mice, monkeys). These strains were able to improve also features of the metabolic syndrome such as steatosis and lipid metabolism.

Bacterial metabolites have an important role in modulating host metabolism. Among these metabolites, short‐chain fatty acids (SCFAs), provide energy to colonic epithelial cells, regulate energy harvest, immune responses, and multiple other host functions including energy homeostasis, lipid metabolism, and insulin sensitivity. ⁴⁶ The production of SCFA is impaired in several metabolic diseases and SCFA supplementation or the administration of their precursors is effective in animal models of several metabolic diseases and preliminary clinical trials (reviewed in ⁴⁶ ). However, they have low bioavailability, and direct supplementation has a poor sustained effect. ⁴⁷ , ⁴⁸ To overcome these limitations, an engineered Bacillus subtilis producing butyrate‐retarded body weight gain induced by a high‐fat diet and visceral fat accumulation in mice, improved glucose tolerance and insulin tolerance, thereby reducing liver damage. ³⁸ , ⁴⁹ In a similar approach that utilized E. coli Nissle engineered to produce butyrate, this strain exerted antagonistic effects on the endocannabinoid system by downregulating and upregulating the endocannabinoid‐synthesizing and degrading enzyme expression, respectively, while alleviating cardiometabolic disease. ⁵⁰

The microbiota also controls metabolic diseases indirectly through the modulation of cytokine production and inflammatory responses both at the intestinal level and at the systemic level. Several metabolic diseases feature an imbalance between anti‐inflammatory and proinflammatory cytokines. One such example is interleukin‐22 (IL‐22), a member of the IL‐10 family of cytokines, expressed predominantly by subsets of innate lymphoid cells (ILCs), RORγt+, and activated T cells. In the gut, IL‐22 regulates the production of antimicrobial peptides, a key element in the maintenance of gut barrier homeostasis. IL‐22 and antimicrobial peptides are decreased in several metabolic diseases including alcoholic liver disease ⁵¹ and other metabolic disorders. ⁵² In ALD for instance, type 3 ILCs produce lower levels of IL‐22 because of lower intestinal levels of indole‐3‐acetic acid (IAA), a microbiota‐derived ligand of the aryl hydrocarbon receptor (AHR), which regulates the expression of IL‐22. However, the administration of Lactobacillus reuteri engineered to produce IL‐22 restored the phenotype, upregulated expression of REG3G in the intestine, and protected mice from alcohol‐induced liver injury. ⁵³ Using the same type of chassis, L. reuteri, the strain capable to produce IL‐22 also prevented weight gain and improved features of NAFLD in a mouse model of diet‐induced obesity. ⁵⁴ Despite these promising results, there is no ongoing study in humans, to our knowledge, assessing these engineered bacteria and their possible bioactive actions.

5. CHALLENGES AND PERSPECTIVES

The microbiota's impact on human metabolism and metabolic diseases is providing new diagnostic and therapeutic opportunities for disease treatment. Through the examples discussed above, engineered bacteria have shown a capacity to regulate host metabolism by providing alternative metabolic functions (Figure 3). While these approaches showed positive results in animal models of disease, their efficacity in the context of human diseases is still to be determined. Some of these approaches proved to be disappointing once tested in a human context. To improve this issue, new model systems are being developed for both in vivo and in vitro testing, in better characterizing modified strains before clinical testing. ⁵⁵ Several issues still need to be addressed, such as the relevant metabolic paths to be targeted, and treatment dose, frequency, and duration. Most of the engineered bacteria that use a probiotic chassis need regular and repeated administrations and are lost after treatment discontinuation. Human diseases and the subsequent microbial alterations that are noted in 'multi‐factorial' diseases are complex, and only partially deciphered to date. Targeting just one metabolic function could be insufficient in modulating some of these diseases. Nevertheless, as the engineering tools further develop, they may allow for the development of more complex systems that can sense and respond to environmental cues in a dynamic manner, display targeted and prolonged colonization, or combine concerted alteration of several metabolic functions. ⁵⁵ To achieve long‐term colonization, such live therapeutics will need to take into consideration interpersonal differences and micro‐niche changes induced by disease. Personalized engineered bacteria using native‐derived chassis could help overcome this limitation and increase their efficacy. ¹² On the other hand, while stable colonization is desirable for prolonged host interaction, continued and long‐term colonization may not enable discontinuation of treatment with changing clinical contexts or upon development of adverse effects.

FIGURE 3.

Building an engineering microbiota therapeutics and its potential applications. An engineered bacteria must comply with several key requirements to induce a significant metabolic impact, including stable engraftment in the gut, precise function, and a favorable biosafety profile. Engineered microbes can be used to modify host metabolism by either providing alternative metabolic pathways to reduce toxic metabolites or producing active biomolecules that would improve disease states (Glp‐1: glucagon‐like peptide 1, IL22: interleukin 22, created with BioRender.com).

To achieve such colonization reversibility, a biological containment system would enable the killing of the bacteria in a controlled suicide process. ⁹ Among the containment systems, kill switches can induce the death of the bacterium in a temperature‐dependent manner, ⁵⁶ while synthetic auxotrophy uses a strategy by which the bacterial growth depends on a non‐native amino acid. ⁵⁷ These mechanisms can also be used to hinder the spread of genetically engineered bacteria to the surroundings ⁸ or of mobile genetic elements and genetic elements (plasmids, prophages, transposons, insertion sequences) that can be transmitted vertically and horizontally between bacteria. While both animal models and clinical trials are focusing on the efficacy of engineered bacteria, the long‐term effects of therapeutic microbes interacting with the commensal microbes of the host are yet to be completely comprehended and merit further studies.

Overall, engineered microbes constitute promising strategies for microbiota‐based therapies and may be used to restore intestinal and/or systemic homeostasis in metabolic diseases. While many factors remained to be refined in the engineering of these bacteria to transform them into clinically effective and safe therapeutics, they show promise as potentially effective ‘live therapies’ that could prevent or treat disease, while improving patients' quality of life.

CONFLICT OF INTEREST STATEMENT

E.E. is a scientific founder of Daytwo and BiomX, and a paid consultant to Hello Inside, Igen, Purposebio, and Aposense. The other author declares that there are no conflicts of interest.

Ciocan D, Elinav E. Engineering bacteria to modulate host metabolism. Acta Physiol. 2023;238:e14001. doi: 10.1111/apha.14001