Bacteriophages and their potential for treatment of gastrointestinal diseases

Abstract

Although bacteriophages have been overshadowed as therapeutic agents by antibiotics for decades, the emergence of multidrug-resistant bacteria and a better understanding of the role of the gut microbiota in human health and disease have brought them back into focus. In this Perspective, we briefly introduce basic phage biology and summarize recent discoveries about phages in relation to their role in the gut microbiota and gastrointestinal diseases, such as inflammatory bowel disease and chronic liver disease. In addition, we review preclinical studies and clinical trials of phage therapy for enteric disease and explore current challenges and potential future directions.

Introduction

Changes in the human intestinal microbiota have been associated with gastrointestinal and liver diseases including inflammatory bowel disease, colorectal cancer (CRC), alcohol-associated liver disease and nonalcoholic fatty liver disease^1–7. Although most of the changes have been described for bacteria, some studies have revealed changes in the gut virome that are associated with disease and developmental dysfunction^8–14.

The human virome is dominated by bacteriophages (also known as phages), which are viruses that can infect bacteria¹⁵. After their discovery >100 years ago^16–18, phages were widely used as antibacterials. However, the primitive state of microbial biology, decades before Watson and Crick, prevented meaningful scientific development of phage therapeutics, especially in the mid-century context of the discovery and rapid industrialization of small-molecule antibiotics^19,20. In the past few decades, the widespread emergence of multidrug-resistant bacteria has reduced the practical utility of antibiotics^21,22. Moreover, a new understanding of the close relationship between the intestinal microbiota and human health has brought into question the general applicability of broad-spectrum antibiotics^23,24. Finally, modern molecular genetics, structural biology and high-throughput genomics have revealed such a high quantity and diversity of phages that most pathogenic bacteria could be targeted.

In this Perspective, we review the role of phages in maintaining human health and in disease pathogenesis, summarize advances in phage-based therapeutics including the direct use of phages in treating enteric disease, and discuss the manipulation of the gut microbiota by the targeting of specific bacterial species. Finally, we discuss the challenges to clinical application of phages and possible future directions for research.

Phage biology: structure and function

With an estimated population of more than 10³¹ particles, phages are the most abundant and diverse biological entities on Earth^25,26. As natural predators of bacteria^27,28, phages are ubiquitous in bacteria-rich environments, including soil, ocean and the human body^29–36.

Generally, phages consist of a protein capsid (rarely with an internal membrane) that contains genomic nucleic acid, which can be linear double-stranded DNA (dsDNA), linear single- or double-stranded RNA, or circular single-stranded DNA (sscDNA)^37,38. Phages are usually classified according to their structure, based on transmission electron microscopy and genome sequence^39–41. The vast majority of DNA phages in the human gut microbiota belong to the order Caudovirales, which are dsDNA phages with genomic DNAs (gDNAs) of ~15–750 kb⁴². Caudovirales have protein capsids based on icosahedral symmetry and come in three general morphologies defined by a tail structure: siphophages (flexible tail), myophages (contractile tail) and podophages (short tail)⁴³. The tails and associated tail fibres constitute an apparatus that not only defines the target specificity of the virion, but also contributes to the efficient infection⁴⁴. The human gut microbiota also contains substantial numbers of much smaller (~5 kb gDNA) phages of the sscDNA family Microviridae^45,46, which are isometric phages that lack tail structures and are restricted to Gram-negative bacteria such as Enterobacteria⁴⁷.

In general, phages can be categorized as virulent or temperate^48,49. Virulent phages (such as Escherichia coli phage T4) follow only a lytic pathway that begins with specific adsorption to a bacterial surface receptor, which can be a protein, carbohydrate, lipid or other external features such as pili, extracellular polysaccharide or flagella^50–52. This adsorption is followed by injection of the gDNA into the host cytoplasm, a programme of DNA replication and gene expression, assembly of the progeny virions, and, finally, release of the progeny by lysis of the host^53,54 (Supplementary Fig. 1). In contrast, temperate phages (such as E. coli λ phage) initiate infection in the same way but have the option to undergo lysogeny^55,56, in which viral gene expression is shut off by a phage-encoded repressor and a dormant prophage, either integrated into the host chromosome or as a linear or circular self-replicating plasmid, is formed ^57,58 (Supplementary Fig. 1). Importantly, the resultant lysogenic cell is thereby immune to further infection by the same phage because of the presence of the lysogenic repressor⁵⁹.

These prophages can be passively carried by the bacterial host indefinitely; they also often carry genes that affect the bacterial host, including pathogenesis factors and defenses against other phages^59,60. Moreover, either spontaneously at a low frequency or at a high frequency as a consequence of cell stress, the prophage can undergo induction and enter the lytic pathway, resulting in cell death and release of the progeny virions^61–64. Even in undisturbed planktonic culture, all lysogenic strains spontaneously produce a certain concentration of free virions at a level that depends on the stability of repression, which can vary by >6 orders of magnitude⁶⁵.

The host range of phages is primarily determined by the receptors on the host surfaces, the receptor recognition proteins of phages, and their interactions. In addition, there are numerous anti-phage systems that impose blocks at nearly every level of the infection process, including inhibition of DNA penetration into the cell, destruction of the phage DNA, inhibition of phage gene expression, and altruistic suicide of the infected cell⁶⁶. Moreover, phages have mustered countervailing molecular and genetic strategies against these defenses⁶⁷. In sum, there are many factors that define host range. Phages generally have host ranges that are restricted to one bacterial species^68,69; efficient propagation of a single phage on widely different bacterial genera has not been convincingly documented. However, the methods used to isolate phages usually involve enrichment on a particular species, which probably biases searches for phages towards finding ‘specialist’ viruses. Indeed, the famous E. coli P1 phage, which was initially isolated as a prophage , is capable of injecting its DNA into Myxococcus xanthus, a bacterial species belonging to the class Deltaproteobacteria — for comparison, E. coli belongs to the class Gammaproteobacteria⁷⁰.

Phages in the gastrointestinal tract

The human body contains diverse communities of microorganisms, consisting of bacteria, viruses (including phages and eukaryotic viruses), fungi and others^71–75. It is estimated that there are approximately the same number of bacterial cells as human cells in the human body, with most of them in the gut ⁷⁶. Phages are inherently more difficult to quantify in the diverse microenvironments of the gut, but most estimates for the phage-to-cell ratio are in the 0.1–10 range¹⁵. As noted earlier, the predominant phages in the human gut, as in all environments that are rich in bacteria, are dsDNA podophages, myophages and siphophages of the order Caudovirales⁷⁷, followed by the small isometric viruses from the family Microviridae⁷⁸.

Intestinal phageome of healthy individuals

Phages are hardly detectable in faecal samples of newborns⁷⁹, but diverse populations can be detected within a few months^79–81. In the first 2 years of life, the richness of the gut phageome decreases, which correlates with early-life bacterial colonization^80–82. Although a core gut phageome has been proposed^34,83, other researchers have suggested that each individual has a unique intestinal phageome^45,46,84. The intestinal phageome consists of both prophages in bacterial cells and free virions or virus-like particles. Previous studies have described core bacterial members (such as species of the genera Bacteroides and Ruminococcus) in the human gut that are common among different individuals^71,85, thus, intestinal phage sequences that were detected in multiple individuals might be the prophages in those core bacteria rather than free virions. Caudovirales, especially those of a temperate lifestyle, have highly mosaic genomes, meaning that different phages can have clusters of identical gene sequences^86–90, making it a challenge to accurately assign a particular sequence read to a particular phage.

Different sample preparation protocols lead to variance among studies^91–95. Additionally, other factors that can affect acquisition and interpretation of data include the analytical methods (metagenomic sequencing versus microscopy)^34,96–99, which bioinformatics tools and databases are used^100,101, as well as sampling positions and testing materials^102,103. In addition, there is evidence that some phages can exist in ‘carrier’ states, in which they are dormant but not repressed or integrated into the host genome^104,105. Altogether, it is therefore not unexpected to see apparently contradictory results regarding the composition and dynamics of the human intestinal phageome. Further work is needed to develop standardized protocols across the methodological spectrum, from viral DNA and RNA extraction to bioinformatic analyses.

Patients with gastrointestinal diseases

Gut bacterial dysbiosis is commonly seen in patients with gastrointestinal and liver diseases^1–3, and, unsurprisingly, the intestinal phageomes of these patients differ from those of healthy individuals.

Patients with Crohn’s disease (n=27) or ulcerative colitis (n=42) have been reported to have a higher relative abundance of Caudovirales compared with Microviridae, and different compositions of Caudovirales families, compared with healthy individuals (n=61), by metagenomic sequencing of the DNA of virus-like particles from faecal samples^8,9 (Fig. 1). Patients with Crohn’s disease had relatively more temperate phages, and the changes in virome composition were reflected in bacterial alterations (for example, patients with inflammatory bowel disease had reduced abundance of Firmicutes and increased levels of phages targeting Firmicutes)⁹. As the gut microbiota varies with the environment (including diet)^106–110, researchers recruited healthy individuals from the same household for these studies, instead of using matched controls from different households^8,9. Interestingly, metagenomic sequencing of faecal virus-like particles from 55 patients with irritable bowel syndrome and 51 control individuals showed that patients with irritable bowel syndrome had a less diverse faecal virome than controls, but the shift from lytic to temperate phages was not observed, which is different from patients with inflammatory bowel disease¹¹¹.

Figure 1. Intestinal phageome of healthy individuals and patients with inflammatory bowel disease.

Left: healthy individual; Right: patients with inflammatory bowel disease. Compared with healthy individuals, patients with inflammatory bowel disease have a higher relative abundance of Caudovirales compared with Microviridae, and different compositions of Caudovirales families.

Viromes of colonic mucosa samples from patients with Crohn’s disease contained increased abundance of virus-like particles compared with colonic mucosa samples from healthy individuals¹¹². Rectal mucosa viromes of patients with ulcerative colitis had a higher relative abundance but lower diversity of Caudovirales phages compared with healthy individuals¹⁰, which is consistent with results from analyses of stool samples⁸. Changes in the enteric virome were also observed in patients with CRC using metagenomic sequencing of faecal samples^11,12. In a random forest analysis, researchers identified virome signatures that differentiated patients with CRC from healthy individuals¹¹ and four taxonomic markers associated with patient mortality¹².

Two studies have also reported virome compositions in patients with liver diseases^13,14. One study included 89 patients with alcoholic hepatitis, 36 patients with alcohol use disorder, and 17 controls¹³, while the other study contained 73 patients with nonalcoholic fatty liver disease and 22 individuals as controls¹⁴. Compared with healthy individuals as controls, increased viral diversity was observed in faecal samples from patients with alcoholic hepatitis. Escherichia, Enterobacteria and Enterococcus phages were overrepresented in these patients, and increased abundance of Staphylococcus phages was associated with higher disease severity¹³. Interestingly, patients with more-severe nonalcoholic fatty liver disease had lower intestinal viral diversity, with a significant reduction in the proportion of phages compared with other intestinal viruses¹⁴. In another study, 40 control individuals and 163 patients with cirrhosis were included. The alpha diversity of the faecal virome was similar between groups, while patients with cirrhosis had more phages against Lactobacillales and Enterobacteriaceae¹¹³.

In summary, intestinal phages have been studied predominantly in patients with inflammatory bowel disease; independent cohort studies will be required to extend and validate these findings. No causative links have been built between intestinal phages and diseases, and further studies are therefore needed to determine whether intestinal phageome changes cause disease development or progression or result from disease. Moreover, findings to date have been largely limited to very broad categories of phages, rather than specific phage types or phages of particular hosts. Thus, we are still at an early stage in understanding these ‘dark matters’ of the intestine and their influences on human health and disease.

Phage-based therapy: past and present

Early history

Immediately after proposing the term ‘bacteriophage’ in 1917, Félix d’Herelle started phage treatment in patients with shigellosis¹¹⁴. Patients with advanced disease exhibited dramatic recoveries after being treated with oral doses of a Shigella phage¹¹⁴. Others also reported success using phage therapy against dysentery, including researchers from United States and Australia^115,116. In the late 1920s, d’Herelle and colleagues reported that oral doses of a Vibrio cholerae phage greatly reduced mortality during cholera epidemics in Assam¹¹⁷. Mortality was ~6% in the treated group of patients (n=74), compared with 63% among patients who refused the phage treatment and thereby served as controls (n=124)¹¹⁷. There were multiple contemporaneous reports of the use of phages against other intestinal diseases such as typhoid fever, although the results were not always positive^118–120. Nevertheless, phage-based therapy was widely considered a viable strategy against bacterial infections prior to the discovery of antibiotics.

However, in the 1930s, clinical reviews, especially a major comprehensive study commissioned by the American Medical Association, concluded that phage-based therapies lacked proven efficacy, specifically citing multiple reports in which phage treatment of cholera and other intestinal diseases had failed¹²¹. After that, interest in developing phages as anti-bacterials declined in the West, especially after the industrialization of small-molecule antibiotics during the World War II era²⁰. In retrospect, the use of phages in clinical practice before the era of molecular biology might have been premature. Nevertheless, phage-based therapies are still used today in some eastern European countries/regions¹²²; unfortunately these therapies have not been very well-documented in English language peer-reviewed literature, and we await more solid preclinical studies and better-designed clinical trials.

Current potential

Treating bacterial infection

Over the past 2 decades, some clinical trials and case studies reported the use of phages to treat gastrointestinal diseases (Table 1). The safety and efficacy of oral administration of E. coli T4-like phages have been tested in healthy individuals and patients with bacterial diarrhoea in several small-scale studies including both adults and children^123–126. No severe adverse effects were reported, but no efficacy was observed either. Similar results were obtained in a large clinical trial using phages to treat bacterial diarrhoea in Bangladeshi children (n=120)¹²⁷. In these studies, faecal phages against the target bacterial hosts (E. coli) were increased in treated children, but the titres did not show substantial intestinal phage replication; E. coli was low in absolute abundance, so applying higher titres of phages might have achieved better results.

Table 1.

Clinical trials and case reports of phage-based therapy against bacterial infections in gastrointestinal diseases

Infectious agent targeting	Phage and dose	Population and treatment method	Outcome and interpretation	Reference
E. coli	Phage T4 (Dose A 105 PFU/ml, dose B 103 PFU/ml)	15 healthy individuals Oral administration	Phage T4 is safe, E. coli abundance not changed	123
E. coli	9 T4-like phages (Dose A 3x109 PFU/ml, dose B 3x107 PFU/ml)	15 healthy individuals Oral administration	Phage cocktail is safe, gut microbiota profile not affected	124
E. coli	Commercial phage cocktail ColiProteus 20ml for adults, 10ml for children, and 10-fold dilution	5 healthy adults, 10 healthy children Oral administration	Phage cocktail is overall safe, with occasional reported adverse effect not relevant to dosage	125
E. coli	11 T4-like phages (3.6x108 PFU) or ColiProteus (1.4x109 PFU)	120 male children with diarrhoea Oral administration	Safe but lack of efficacy	127
E. coli	T4-like phage cocktail (108 or 106 PFU for older children, 107 or 105 PFU for younger children) or ColiProteus (5x108 or 109 PFU)	20 older children, 20 younger children Oral administration	Both cocktails are safe	126
Acinetobacter baumannii	9 phages in 3 cocktails (5x109 PFU intravenous)	68-year-old male patient with necrotizing pancreatitis complicated by pancreatic pseudocyst Intracavitary and Intravenous	Patient completely recovered	128
Klebsiella pneumoniae	1 phage (107 PFU orally, 106 PFU intra-rectally)	57-year-old female patient with Crohn’s disease, with multi-site infection (gastrointestinal tract, urinary tract, etc) Oral and intra-rectal	The original host (Klebsiella pneumoniae) was no longer detected	201

English language publications only. E. coli, Escherichia coli.

In 2016, a 68-year-old male patient with diabetes, infected with multidrug-resistant Acinetobacter baumannii, developed necrotizing pancreatitis complicated by a pancreatic pseudocyst¹²⁸. Despite multiple antibiotic courses and a percutaneous drainage of a pancreatic pseudocyst, the patient deteriorated over a four-month period. On the basis of an emergency Investigational New Drug (IND) permission from the US Food and Drug Administration, phage therapy was initiated (intracavitary and intravenous) and the patient returned to health after approximately five months¹²⁸. Although this is only a case report, the obvious downward clinical course before phage treatment and the clear turning point after phage administration generated wide publicity and brought renewed hope that phage-based therapies might be used to treat bacterial infections (especially multidrug-resistant bacteria). Multiple case reports in other emergency IND situations have accumulated in the past 4 years against the multidrug-resistant bacteria Pseudomonas aeruginosa and Mycobacterium abscessus^129,130. However, standardized clinical trials will be required to further determine the efficacy of phage therapies for different infectious diseases.

Phages have also been evaluated for disease prophylaxis in preclinical models. Oral administration of a three-phage cocktail to infant mice 24-hours prior to V. cholerae challenge had significant reductions in bacterial colonization of the intestine¹³¹. In addition, using an infant rabbit model, administration of phage before bacterial challenge protected them from cholera-like diarrhoea¹³¹. As cholera epidemics are seasonal and self-limiting¹³², phage prophylaxis might be used to control disease spread and protect high-risk individuals during outbreaks. More studies should be performed to explore the potential protective effect shortly after bacterial challenge, thereby aiming to reduce bacterial colonization and prevent disease.

Manipulating the gut microbiota

Strategies to manipulate the gut microbiota include faecal microbiota transplantation (FMT)¹³³, use of prebiotics and probiotics¹³⁴, and adjustments to diet and nutrient intake¹⁰⁸. In the past decade, phages have also been used for precision editing of the gut microbiota (Table 2). In 2017, a United States patent was granted for PreforPro (Deerland Probiotics and Enzymes, Kennesaw, GA), a mix of phages targeting E. coli¹³⁵. Two placebo-controlled trials have been conducted to determine the safety and efficacy of PreforPro in improving intestinal health by altering gut bacterial composition. One trial evaluated the effect of the phage cocktail alone¹³⁶, while the other trial tested the additive effect of PreforPro on probiotics Bifidobacterium animalis subsp. lactis BL04¹³⁷. Both trials included healthy individuals who reported having mild-to-moderate gastrointestinal distress but no diagnosed gastrointestinal disorders. Over the 28-day study in both trials, encapsulated PreforPro was found to be safe and tolerated, but the evidence of efficacy was not clear-cut ^136,137. The connection between E. coli and abdominal symptoms has not been well established; thus, more studies are needed to better evaluate the potential efficacy.

Table 2.

Studies of phage-based strategies for gut microbiota modulation in gastrointestinal diseases

Infectious agent targeting	Status	Phage and dose	Population and treatment method	Outcome and interpretation	Reference
E. coli	Clinical trial, complete	PreforPro (4 phages) 1 capsule daily for 28 days	32 healthy individuals with mild-to-moderate gastrointestinal distress Oral administration	Phage cocktail is safe and tolerable, but no difference from placebo	136
E. coli	Clinical trial, complete	PreforPro (4 phages) together with probiotics Bifidobacterium animalis subspecies lactis strain BL04 1 capsule daily for 28 days	68 healthy individuals with mild-to-moderate gastrointestinal distress Oral administration	Phage supplement is tolerated, but no compelling evidence of efficacy. Only marginally significant effects on self-diagnosed gastrointestinal inflammation are reported as evidence of a benefit	137
E. coli	Clinical trial, active	EcoActive (phage cocktail) Twice daily for 15 days	30 patients with Crohn’s disease in remission Oral administration		202
Enterococcus faecalis	Preclinical	Phage cocktail (3-4 phages) 1010 PFU, one day before sacrifice	Germ-free mice colonized with stool samples from patients with alcoholic hepatitis Oral administration	Phage cocktail is beneficial for alcohol-related liver disease	141

English language publications only. E. coli, Escherichia coli.

Adherent-invasive E. coli has been implicated in the pathogenesis of inflammatory bowel disease over the past 2 decades^138,139. Phages against such E. coli strains have been proposed as a treatment option. Conventional mice colonized with adherent-invasive E. coli were given drinking water with 2% dextran sodium sulfate (DSS) to induce mild symptoms of colitis. After one week, a three-phage cocktail was orally gavaged to the mice, followed by 2% DSS drinking water for another two weeks¹⁴⁰. Mice receiving phage treatment were found to be protected from DSS-induced colitis¹⁴⁰, and E. coli colonization was reduced¹⁴⁰. Active phage replication was detected in ileal biopsy samples spiked with E. coli from patients with Crohn’s disease, providing additional evidence for the killing potential of phages in such environments¹⁴⁰. A phase I/IIa randomized, double-blind, placebo-controlled clinical trial is underway to assess the safety and efficacy of oral administration of phages that target intestinal adherent-invasive E. coli in patients with Crohn’s disease in remission (NCT03808103).

Researchers have shown that faecal levels of Enterococcus faecalis, a commensal member of the human gut microbiota with low abundance, is significantly increased in patients with alcoholic hepatitis compared with patients with alcohol use disorder and non-alcoholic individuals. The presence of E. faecalis strains that produce cytolysin (a bacterial exotoxin) correlates with worse outcomes and with mortality in patients with alcoholic hepatitis¹⁴¹. Oral administration of cytolysin-positive E. faecalis exacerbated ethanol-induced liver disease in conventional mice. To extend these findings to humans, gnotobiotic mice were colonized with faecal samples from cytolysin-positive and cytolysin-negative patients with alcoholic hepatitis. Phages specifically targeting cytolysin-positive E. faecalis were administered by oral gavage and they reduced ethanol-induced liver disease, whereas phages against cytolysin-negative E. faecalis did not have any beneficial effect¹⁴¹. Besides the implications for potential treatment of ethanol-induced liver disease, this study can be regarded as one of the first documented examples of precision editing of the gut microbiota, by extirpation of a subpopulation of E. faecalis (Fig. 2). Larger studies are needed to validate these results and a clinical trial is necessary to test the therapeutic effects in patients with alcoholic hepatitis.

Figure 2. Manipulation of the gut microbiota by phages.

(a) Phage therapy in Crohn’s disease. Adherent-invasive E. coli stimulates antigen-presenting cell (APC) driving Th17 responses, phages targeting adherent-invasive E. coli were found to be beneficial of DSS-induced colitis. (b) Phage therapy in alcoholic liver disease. Cytolysin positive E. faecalis translocates from the gut to the liver, directly damaging hepatocytes. Phages against cytolysin positive E. faecalis could protect mice from alcoholic liver disease. (c) Phage therapy in colorectal cancer. Streptococcus gallolyticus subsp. gallolyticus (Sgg) upregulates β-catenin, stimulating cancer cell proliferation. Phage therapy might be a promising treatment option.

After first being reported by McCoy and Mason in 1951¹⁴², several studies have revealed that Streptococcus gallolyticus subsp. gallolyticus (Sgg), a cause of septicaemia and infective endocarditis, is associated with CRC (for a detailed review, see Abdulamir et al.¹⁴³ and Boleij et al.¹⁴⁴). Fusobacterium nucleatum is more abundant in faecal samples from patients with CRC than in control individuals^145–147; it is also over-represented in tumours versus matched control tissue specimens from patients with CRC¹⁴⁸. Preclinical studies have shown that a number of bacterial species of the gut microbiota, in particular F. nucleatum, Bacteroides fragilis, E. coli and E. faecalis, are also associated with CRC development and progression via different mechanisms^149–155. In vitro and in vivo studies using mouse models have suggested a CRC tumour-promoting role for Sgg by upregulating β-catenin, a central signalling molecule in colon tumorigenesis¹⁵⁶. In addition, inhibiting intestinal E. coli overgrowth by oral administration of sodium tungstate reduced gut inflammation and the incidence of colitis-associated colonic tumours in two mouse models (azoxymethane/dDSS colitis model and azoxymethane-treated Il10-deficient mouse model)¹⁵⁷. Although more studies are required to confirm the causative link between intestinal bacteria and CRC, phage-mediated precision editing of the gut microbiota might be worth exploring as a promising treatment option (Fig. 2).

It is important to note, however, that it will be necessary to have some understanding of what determines the phage specificity in such studies in order to achieve ‘precision’. E. coli phage T4 uses different receptors for different E. coli strains¹⁵⁸. A similar E. coli phage, Ox2, can change its receptor from one outer membrane protein (the usual receptor of phage Ox2 is OmpA) to another (OmpC and/or OmpX) or to different carbohydrate residues in the lipopolysaccharide as a result of single mutations in the tail fibre¹⁵⁸. Moreover, some phage genomes encode arrays of tail fibre genes that can be switched in and out by high-frequency recombination processes^159,160. Amazingly, phages against Bordetella spp. that encode an error-prone reverse transcriptase that causes extreme hypermutation of the receptor-binding domain of the tail fibre have been isolated¹⁶¹. It is therefore important to identify the receptors of phages to precisely target the host bacteria.

Phage therapy as a precision medicine approach

Phages can not only precisely edit the gut microbiota, they can also deliver drugs to a specific location. As the natural predators of bacteria, phages propagate in environments where their hosts reside. With the development of more powerful tools to engineer phages, drugs can be attached to the phage surface to be released when phages reach their destinations¹⁶². Thus, site-specific administration of high doses might be possible, enabling reduced concentrations of drugs in the circulation and decreased toxic effects on non-target tissues^163,164.

Several preclinical studies have tested this approach. In an in vitro study, thousands of molecules of the antibiotic chloramphenicol were attached to a phage surface via ester linkage, enabling it to be slowly released by serum esterases¹⁶⁵. The phages were able to target Staphylococcus aureus, providing local high concentrations of chloramphenicol that were sufficient to inhibit the growth of previously resistant S. aureus cells¹⁶⁵. A similar idea has also been applied in vivo in a mouse model¹⁶⁶. . F. nucleatum was largely found in CRC tumours and promoted CRC resistance to chemotherapy in mice¹⁶⁷. F. nucleatum targets Toll-like receptor 4 and specific microRNAs to activate the autophagy pathway, thus altering the CRC chemotherapeutic response¹⁶⁷. In a study using a CRC mouse model, Zheng et al. coated F. nucleatum phages with irinotecan, a first-line treatment for CRC¹⁶⁶. These phages target F. nucleatum, which reside in CRC tumour tissues. Phages therefore accumulated in CRC tumours and precisely delivered the drug to its destination, with minimal adverse effects to non-tumour tissues¹⁶⁶. Oral administration of these irinotecan-coated phages also decreased the abundance of F. nucleatum, thus re-sensitizing tumour cells to chemotherapy¹⁶⁶. A similar approach has been used in another study in which the researchers assembled anti-bacterial silver nanoparticles on the phage surface. Those phages specifically targeted F. nucleatum and accumulated in CRC tumour cells. Compared with the mice receiving only chemotherapy, mice receiving a combination of both phage therapy and chemotherapy showed less tumour growth and a longer survival time ¹⁶⁸. Thus, the idea of phage-mediated drug delivery has promise for wide application in clinical practice.

Gut microbial metabolites have important roles in human health and also contribute to diseases^169,170. Short-chain fatty acids (such as butyrate) produced by gut bacteria inhibit tumour growth and stimulate antitumour immune responses in mouse models of CRC^171,172. Zheng et al. showed that phages can be covalently linked with dextran nanoparticles that promote proliferation of Clostridium butyricum, which increased faecal levels of short-chain fatty acids in mice and inhibited tumour growth¹⁶⁶. These phages targeted pathogenic F. nucleatum without affecting C. butyricum. Multi-functional phage particles can be administered orally, and might be used not only to deliver drugs to specific locations, but also to increase treatment efficacy by modulating the intestinal microbiota (that is, reducing the amount of pathogenic bacteria and promoting the growth of beneficial bacteria). This novel and convenient route of administration (that is, oral administration of chemically-coated phages) deserves more attention. Studies are needed in different models of diseases and to assess long-term effects.

Challenges and future directions

A century after their discovery, phages are the subjects of renewed interest for treatment of bacterial infections, especially for gastrointestinal diseases, in which systemic introduction of phages into the bloodstream would not be required. These bacterial predators have broad applications, but there are many challenges to overcome.

Most studies have reported phage-based therapies to be safe^124,136,173, as phages only propagate in bacteria. However, using a mouse model, one study showed that filamentous Pseudomonas phages can directly interact with human leukocytes, with phage RNA being produced and stimulating interferon production¹⁷⁴. This observation indicates that filamentous phages might interact with the human immune system and have direct effects on human health. Several preclinical studies have also assessed the immune response induced by phages. Some reported that orally administrated phages could stimulate inflammatory cytokine production and induce inflammation, mostly in mouse models of intestinal inflammation and dysbiosis^175–177. On the other hand, phage administration in vitro either had no effect on the inflammatory response or exerted an anti-inflammatory response on mammalian cells, as measured by the levels of inflammatory cytokines^{176,178–180}. Given the long-term presence of bacteria and phages in the mammalian gut, it will not be surprising to find out that phages are capable of many interactions with the human immune system and other diverse cell types.

The narrow host range is another apparent limitation of phage-based treatment. Therefore, the superb specificity of phages, which enables precise targeting of bacteria, is also a potential problem, because the narrow host range could limit wide therapeutic utility. One option is to create a phage cocktail, comprised of multiple phages that each target a different receptor. However, this increases the complexity and safety risk of the treatment, because under current guidelines, the safety of each individual phage, as well as each different combination of phages, would need to be tested. Many factors therefore must be considered in developing a therapeutic phage cocktail, including the host range, the receptors, and the infectious efficiency (for detailed guidelines, see Merabishvili et al.¹⁸¹). Another possible strategy involves ‘phage training’, or phage adaptation. This process selects for evolved phages with broader host ranges or that can overcome bacterial resistance through experimental procedures performed in the laboratory. Phages that can target multiple hosts can be obtained via multiple rounds of selection using either different bacterial isolations or resistant mutants (for detailed protocols, see Betts et al.¹⁸² and Friman et al.¹⁸³). This approach has the additional attraction of being ‘natural’ by avoiding recombinant DNA methods and the complications of genetically modified organism (GMO) classification. On the other hand, rapid development of synthetic biology has made engineering phages attractive, albeit subject to GMO regulation. By identifying phage proteins that are responsible for host recognition, genetic modifications could be made to broaden host ranges¹⁸⁴ or reduce the potential for the emergence of phage resistance¹⁸⁵. To achieve this goal, cutting-edge genomic editing tools and more knowledge about host determination will be indispensable.

The research field of phage therapy is still at an early stage, with many scientific questions remaining to be answered. In addition, to obtain a better result, many factors should be considered and carefully evaluated before phage administration^186,187. One of the most important is to screen all the individuals for the presence of the target bacterial host in their gut. The bacterial host should also be tested in vitro to confirm its sensitivity to the selected phages. Another critical point is to determine the dose of phages being applied. Multiple studies have reported the safety of using a relatively high dose of phages (for example, 10⁹ PFU, both orally and intravenously)^124,127,188. Applying phages in high concentrations might be necessary, especially for oral administration as gastric acid could decrease the amount of surviving phages¹⁸⁹. Concurrent administration of an acid-neutralizing reagent and phage encapsulation methods might be helpful in this regard^126,190,191. Additionally, the pharmacokinetic and pharmacodynamic properties of phages also need to be evaluated¹⁹². Multiple studies found phages being cleared in the circulation system in a few hours, in both mice and humans^188,193, which might pose a problem in maintaining sufficient number of phages for therapeutic purposes. However, some researchers have reported that orally administrated phages could still be detected in the gut after several days^124,126. Modifications of the phage capsid proteins might improve its half-life, as such changes might help phages better evade phagocytosis^194–196.

Although there are still many concerns and challenges (Box 1), phage therapy has great potential for use in clinical practice (Fig. 3). Well-designed, placebo-controlled clinical trials showing safety and efficacy could help the field and attract more scientists and physicians. However, the development of this field not only depends on the ‘researchers’ (scientists, physicians and even patients) but also on the regulatory context. Phages are considered as medicinal products in the US and the European Union¹⁹⁷, which are under very strict constraints related to their production and marketing authorization, such as compliance to Good Manufacturing Practice (GMP)¹⁹⁷. As a customized treatment, therapeutic phages need to be selected and produced ad hoc, making it impractical to be an immutable pre-defined medicinal product¹⁹⁸. Manufacturing GMP-certified medicinal products is overall costly and time-consuming, making it even harder to initiate a phage-based clinical trial¹⁹⁹. Recently, the Belgian government classified therapeutic phages as magistral preparations, providing more flexibilities to phage treatments²⁰⁰. In European law, magistral preparation (compounded prescription drug product in the US) is defined as “any medicinal product prepared in a pharmacy in accordance with a medical prescription for an individual patient” (Article 3 of Directive 2001/83 and Article 6 quater, § 3 of the Law of 25 March 1964). Although this might be less likely to be approved by the US Food and Drug Administration (FDA) due to stricter rules and more concerns, phage therapeutics would require some specific rules and regulations that are different from other standardized medicines. Phages might be considered for GRAS (generally regarded as safe) materials by the US FDA, given their ubiquity in the human body and environment and their fundamental inability to attack human tissues. This designation might set the stage for conducting a few set-piece clinical trials of phage cocktails; assuming positive results in terms of safety and efficacy, regulatory approval might then be advanced for other combinations of phages prepared and formulated in the same way, much like the way new flu vaccines are approved each season. In any case, it is clear that more government-funded, phage-based clinical trials are required to better explore the therapeutic potential of phages in a broad range of gastrointestinal diseases.

Box 1. Important questions for further studies of phage-based therapies.

Although phages were discovered a century ago, phage therapy is still a relatively new research area, with many challenges and problems as well as open questions and opportunities.

• Is phage-based therapy safe for clinical practice? What will the regulations and rules be?

• Will phage-based therapy replace antibiotic treatment? If not, when to choose which option? Or both simultaneously?

• How to decide the best administration route and the dose for each phage therapy?

• Which is better: a single phage or phage cocktail? How to decide which one to use in clinical practice?

• Biofilm is a big challenge for antibiotic treatment; can phages be found that will propagate efficiently in biofilms?

• Could phages be used against intracellular bacterial infections?

• More phages can always be found. Is it possible to quickly and easily ascertain whether new isolates are going to be helpful, adding power to the therapy? Would bioinformatic tools help?

• What are the long-term effects of phage-based treatments on the intestinal microbiota and humans?

Figure 3. Potential applications of phages.

Apart from targeting their bacterial hosts, phages might also interact with the human body and could thereby have multiple effects on human health and diseases.

Conclusion

Bacteriophages have been used to combat bacterial infections for over a century. Discoveries of the association between the gut microbiota and human diseases have prompted renewed attention to this research area. Phages are powerful weapons not only against pathogenic bacterial infections, but they can also precisely edit the intestinal microbiota and harbour promising therapeutic effects for many different gastrointestinal diseases. Multiple therapeutic possibilities have been proposed, but more basic and preclinical studies, as well as properly designed randomized, double-blind, placebo-controlled trials are required to help the field move forward. Still at the early stage, the field has significant problems and challenges to be solved, such as the beneficial or harmful effects of potential phage–human interactions, the evolving nature of phages as vital biological entities, and the long-term effects of a phage-modulated gut microbiota on human health. Overall, phage-based therapies could become promising and powerful approaches to treat many gastrointestinal and possibly extra-intestinal diseases, and are deserving of greater attention and further exploration.

In this Perspective, Duan, Young and Schnabl explore the effects of bacteriophages on the gut microbiota and the potential applications of phage therapy for treatment of gastrointestinal diseases. Limitations and challenges of phage therapy for gastrointestinal diseases are also discussed.