INTRODUCTION
Bacteriophages (phages) are among the most abundant biological particles on earth. They are also highly versatile and adaptable to a great number of applications. Phages are viruses that infect bacteria; their self-replication depends on access to a bacterial host. Phages were discovered independently by Frederick Twort in 1915 (
1) and by Félix d'Hérelle in 1917 (
2), and they were used early on as antimicrobial agents. Although the initial results of phage therapy were promising (
3,
4), poorly controlled trials and inconsistent results generated controversy within the scientific community about the efficacy and reproducibility of using phages to treat bacterial infections (
5–7). The discovery of penicillin in 1928 and the subsequent arrival of the antibiotic era further cast a shadow on phage therapy (
5,
6). As a result, phage therapy was discontinued in Western countries, even as its use continued in Eastern Europe and the former Soviet Union (
8–10).
Despite the important success of antibiotics in improving human health, antibiotic resistance has emerged with steadily increasing frequency, rendering many antibiotics ineffective (
11–14). Multidrug-resistant bacteria currently constitute one of the most widespread global public health concerns (
15–17). More than 2 million people are sickened every year in the United States alone as a result of antibiotic-resistant infections, resulting in at least 23,000 deaths per year (
16). The rising tide of antibiotic resistance coupled with the low rate of antibiotic discovery (
17,
18) has revived interest in phages as antibacterial agents (
19–21).
Unlike most antibiotics, phages are typically highly specific for a particular set of bacterial species or strains and are thus expected to have fewer off-target effects on commensal microflora than antibiotics do (
22). The self-replicating nature of phages and the availability of simple, rapid, and low-cost phage production processes are additional advantages for their use as antimicrobials (
22). Phages have been used not only to treat and prevent human bacterial infections (
9,
23–25) but also to control plant diseases (
26–29), detect pathogens (
30–33), and assess food safety (
34–37).
Notwithstanding their antimicrobial potential, some major concerns remain about the use of phages in clinical medicine. The specificity of phages means that they can target bacterial strains precisely; however, because a single phage type is unlikely to target all strains within a given species, cocktails combining various phages are often necessary to be broadly applicable for treating the wide range of bacteria that can cause clinical infections. Obtaining regulatory approval for the therapeutic applications of such cocktails can be challenging because of the significant diversity of phages in terms of structure, life cycle, and genome organization (
22,
38). Like certain antibiotics, phages can cause rapid and massive bacterial lysis and the subsequent release of cell wall components (e.g., lipopolysaccharides [LPS]), which can induce adverse immune responses in the human host (
39,
40). Bacteria frequently live in biofilm communities surrounded by extracellular polymeric substances (EPS), which can act as a barrier to phage penetration (
41). Furthermore, as bacteria evolve, they can develop resistance mechanisms to avoid phage infection (
38,
42,
43). By genetically engineering phages, it may be possible to overcome many of these limitations (
44). The engineering of specific phages and components has been facilitated by the ever-growing abundance of fully sequenced phage genomes in public databases (
45,
46) and by research into elucidating the structures of phage components (
47–51) and the interactions between phages and their host bacteria (
52–54). This review focuses on advances made in phage engineering techniques and applications in the past decade. Specifically, we discuss the use of phages in pathogen control and detection, as well as their broader application in other research areas, including targeted drug delivery and materials engineering.
PHAGE-DERIVED ANTIMICROBIALS
In addition to harnessing phages for their antimicrobial (lytic) properties or as delivery vehicles for antimicrobial agents, phage proteins can be used on their own as direct antimicrobials (
143–145). Once phages have infected bacteria and undergone replication, phage-encoded endolysins degrade the peptidoglycan of the bacterial cell wall from within the cell. Endolysins thus come into play at the terminal stage of the phage replication cycle, causing host cell lysis (
146). Endolysins can be effective when applied to the outside of the bacterial cell, even though they are naturally active from the inside. Experimentally, these enzymes have been expressed, purified, and used mostly against Gram-positive bacteria, which are more susceptible to lysis than Gram-negative bacteria because they lack an outer membrane (
147,
148). Endolysins added externally to Gram-positive bacteria can result in rapid lysis (
147,
149). Furthermore, they have successfully prevented or treated infections caused by Gram-positive bacteria in animal models (
150–155).
Phage endolysins have also been engineered. For example, one study reported the construction of four chimeric phage endolysins (Cpl-711, Cpl-771, Cpl-117, and Cpl-177) by shuffling and combining the structural elements (catalytic domain, linker, and cell wall-binding domain) of two pneumococcal phage endolysins, Cpl-1 and Cpl-7S (a synthetic variant of Cpl-7 with improved bactericidal activity) (
156). The bactericidal activity of the new chimeric endolysins against
Streptococcus pneumoniae was evaluated, and Cpl-711 was found to be the most efficient chimera. This chimera was composed of the catalytic domain of Cpl-7S and the linker and cell wall-binding domain of Cpl-1. At a concentration of 0.01 μg/ml, Cpl-711 reduced the number of
S. pneumoniae cells by 2 orders of magnitude after 1 h of treatment, whereas Cpl-1 reduced cell viability by only 15%. At 1 μg/ml, the chimeric enzyme Cpl-711 reduced the number of viable cells in pneumococcal biofilms by 4 orders of magnitude after a 2-h treatment, which was an improvement over the approximately 1.5 orders of magnitude of killing by the parental proteins, Cpl-1 and Cpl-7S (
156). In
in vivo assays, mice infected intraperitoneally with an
S. pneumoniae suspension and treated 1 h after bacterial challenge with Cpl-711 had about 50% greater survival than those treated with Cpl-1 (
156). This study demonstrated the flexibility with which new and improved phage endolysins can be engineered. Because phage endolysins are diverse, engineering them may provide a pipeline of novel antimicrobial agents. In fact, similar work has been performed on phage endolysins isolated from
Listeria spp. (
157),
Streptococcus spp. (
158,
159), and
Staphylococcus spp. (
158–161).
Gram-negative bacteria are difficult to lyse because the outer membrane blocks access of the endolysin to the peptidoglycan. Endolysin-mediated lysis of Gram-negative bacteria has been achieved, however, through the use of permeabilizing agents (
162). There are two classes of outer membrane permeabilizers: (i) polycationic agents, such as polymyxin and its derivatives, which interact with phospholipids in the cell membrane (
163,
164), or lysine polymers, which adsorb to the cell surface and block growth (
164,
165); and (ii) chelators, such as EDTA, which remove ions from the outer membrane, leading to its disintegration (
163,
164), or weak organic acids, which penetrate the cell wall and interfere with bacterial physiology (
163,
164). Nonetheless, it is important to highlight that the
in vivo toxicity of the outer membrane permeabilizers might limit the applicability of this approach. For example, both EDTA and citric acid were found to have cytotoxic effects on macrophages
ex vivo (
166).
Briers et al. combined endolysin EL188 from
P. aeruginosa phage EL with outer membrane permeabilizers and evaluated the antibacterial activity against
P. aeruginosa strains. The permeabilizers tested were polymyxin B, poly-
l-lysine, EDTA, and citric acid (
164).
In vitro antibacterial assays revealed that EDTA was the best permeabilizer: the combination of the endolysin and EDTA reduced the number of
P. aeruginosa PAO1 cells in mid-log phase by more than 4 orders of magnitude in just 30 min (
164). Similarly, another study in which EDTA was used as a permeabilizer reported reductions of up to approximately 3 orders of magnitude of
P. aeruginosa PAO1 cell counts after 30 min of incubation with globular endolysins encoded by phages infecting Gram-negative bacteria (
162). Oliveira et al. reported the lethality of a
Salmonella phage endolysin (Lys68) combined with organic acids against Gram-negative bacteria (
167). The best results were achieved against
Pseudomonas cultures: reductions of
Pseudomonas aeruginosa cells of approximately 2.4, 1.5, and 3.3 orders of magnitude and reductions of
Pseudomonas fluorescens cells of approximately 1.6, 1.4, and 5.4 orders of magnitude were observed 30 min after applying Lys68 in combination with EDTA, citric acid, and malic acid, respectively (
167). Determining the best combination of permeabilizer and lysin for a given target bacterium currently appears to be performed empirically.
To circumvent the problem of outer membrane permeability, Briers et al. engineered endolysins to contain LPS-destabilizing peptides. The resulting endolysins, called Artilysins, penetrate the bacterial outer membrane, which is something that the original endolysins are not capable of doing (
168). With this approach, the LPS ion-based membrane stabilization is disrupted by the physicochemical properties of the synthetic peptides coupled to the endolysins, enhancing their killing effect (
168). Thus, the fusion of a polycationic nonapeptide (PCNP) to the OBPgp279 endolysin (from
P. fluorescens phage OBP) enhanced the bactericidal activity of the native endolysin against
P. aeruginosa PAO1 from 1.10 to 2.61 orders of magnitude of reduction, even in the absence of permeabilizers. Although the PCNP-fused endolysin was found to be effective without permeabilizers, its activity was enhanced by EDTA: viable cell counts were reduced by 5.38 orders of magnitude and 4.27 orders of magnitude for
P. aeruginosa PAO1 and multidrug-resistant
P. aeruginosa Br667, respectively, within 30 min (
168).
Lood et al. built a genomic library based on prophages induced from the Gram-negative organism
Acinetobacter baumannii to screen for genes encoding antibacterial endolysins (
169). They identified and isolated several endolysins active against
A. baumannii.
In vitro studies showed that phage lysin PlyF307, the one with the greatest activity, reduced exponentially growing cultures of
A. baumannii clinical isolates by >5 orders of magnitude within 2 h.
In vitro treatment of
A. baumannii biofilms with PlyF307 for 2 h reduced the number of cells by 1.6 orders of magnitude. Furthermore, the lysin also rescued mice from lethal
A. baumannii bacteremia: while 90% of the buffer-treated mice died within 2 days, 50% of PlyF307-treated mice survived the lethal dose of
A. baumannii. This was the first study to use native endolysins without additional factors for the treatment of Gram-negative infections in mice (
169).
PHAGE ENGINEERING FOR BACTERIAL DETECTION AND DIAGNOSTICS
Most of the methods used to detect and identify bacterial pathogens in food, hospital, and industrial environments are time-consuming, in part because they require enrichment steps for increased sensitivity and/or specificity (
170,
171). Traditional plating techniques not only are laborious but also often fail to detect pathogens present in samples at low levels (
170–172). Other methods, such as antibody-based ones, do not usually perform well for complex samples without enrichment to amplify the bacterial targets (
173). Techniques such as PCR or hybridization-based assays can be very sensitive but are not able to discriminate between live and dead cells without bacterial enrichment, and they also require the careful design of primers to avoid off-target hits and the misidentification of species (
171,
173). Recent advances in genetic engineering and synthetic biology, particularly the development of phage-based tools for pathogen detection, have made it possible to overcome such limitations.
Loessner et al. described a rapid, easy, and sensitive method for using engineered phage to detect
Listeria monocytogenes in contaminated food (
59). This method consisted of inserting, by homologous recombination, a
Vibrio harveyi luxA and
luxB gene fusion (
luxAB) downstream of the major capsid protein gene of
Listeria phage A511 (
59). Upon infection of the targeted bacteria, this engineered phage generated light. Detectable luminescence was generated rapidly, within 2 h of application, even on food contaminated with as few as 5 × 10
2 L. monocytogenes cells/ml (
30). When an enrichment step was included, levels of <1 CFU/g could be detected by use of the engineered phage (
59). Sarkis et al. described a similar method for detecting live mycobacteria. They cloned a luciferase gene into the tRNA region of the genome of the L5 mycobacteriophage and used the recombinant mycobacteriophage to identify
Mycobacterium smegmatis cells. Aliquots of cultures with hundreds of
M. smegmatis cells produced a positive signal (above the background) in just a few hours. Even samples with as few as 12.2 and 2.7 CFU/100 μl produced positive signals, though only after 2 and 3 days, respectively (
62).
Phages expressing green fluorescent protein (GFP) have been proposed as a fast and accurate method for detecting
E. coli (
64,
65,
174). The
gfp gene, originally carried on a plasmid, was inserted by homologous recombination into the genomes of phages T4 (wild type), T4e
− (a gene
e amber mutant phage) (
64), and PP01 (
65) such that it was displayed on the small outer capsid (SOC) of these phages, resulting in phages T4wt/GFP, T4e
−/GFP, and PP01-SOC/GFP (GFP introduced into the C terminus of SOC) or PP01-GFP/SOC (GFP introduced into the N terminus of SOC), respectively. The
gfp gene was also inserted into phages IP008 and IP052, within the
e gene, which encodes a phage lysozyme, resulting in phages IP008e-/GFP and IP052e-/GFP, which exhibited suppressed lytic activity (
174). Incubating T4wt/GFP with
E. coli K-12 led to increased fluorescence intensity during the initial stages of infection but then resulted in cell lysis, which made it difficult to identify phage-infected cells by fluorescence microscopy. On the other hand,
E. coli incubated with the engineered phage T4e
−/GFP exhibited fluorescence, the intensity of which increased with the infection time (
64). The GFP-labeled PP01 phage was able to specifically detect
E. coli O157:H7, and fluorescence could be observed by microscopy after as little as 10 min of incubation (
65). Engineering phages to express multiple
gfp genes can enhance the detectable signal. For example,
E. coli B
e cells infected with phage IP008e-/GFP or IP052e-/GFP exhibited low fluorescence intensity. When
gfp was additionally fused to the
soc gene in phages IP008e-/GFP and IP052e-/GFP, resulting in IP008e-/2xGFP and IP052e-/2xGFP, the fluorescence intensity was stronger and increased with incubation time (
174). The detection limit of this approach has not yet been evaluated.
Edgar et al. proposed a biodetection system that combines
in vivo biotinylation of an engineered phage followed by conjugation of the phage to streptavidin-coated quantum dots (QDs), semiconductor nanocrystals that give a fluorescence signal (
175). The T7 coliphage was engineered to display a small biotinylation peptide on its major capsid protein. After propagation of the recombinant phage in the bacterial host, the biotinylated progeny phage could be detected by the fluorescence of the streptavidin-coated QDs. If the host was not present, biotinylated phage were not produced, and the functionalized QDs did not bind and were washed away. This method is fast, sensitive, and specific: as few as 10 and 20
E. coli cells/ml were detected by fluorescence microscopy within 1 h for experimental and environmental samples, respectively (
175).
Piuri et al. genetically engineered the mycobacteriophage TM4 to carry a fluorescence-encoding reporter gene, namely,
gfp or
ZsYellow (
176). The engineered mycobacteriophages detected
Mycobacterium tuberculosis by delivering the reporter genes into the cells; the fluorescence could then be monitored by microscopy or flow cytometry (
176). With this rapid and sensitive method, fewer than 100 cells present in the 5-μl aliquots used for microscopy could be detected, and bacterial antibiotic susceptibility could be determined in less than 24 h, as fluorescence was suppressed only in rifampin- or streptomycin-susceptible cells when the corresponding antibiotics were added (
176,
177).
Diagnostic phage technologies are being translated into actual use outside research labs (
173). For example, the first enrichment-free pathogen diagnostic system for
Listeria was recently released for commercial applications (
178). In addition, for phages to be used as personalized antimicrobials in the era of precision medicine, rapid and accurate diagnostics are needed to identify pathogens and determine which phages are most suitable for therapy. We envision that phage-engineering technologies will play an important role in a wide range of settings where rapid microbial detection is desirable.
PHAGE ENGINEERING FOR MATERIALS SCIENCE
In addition to their applications in human health, veterinary health, and food safety, phages have been adapted for use in materials science. By combinations of phage display and genetic engineering techniques, phages have been used to build novel nanostructured materials with various applications, such as energy generation and storage (
201–204), biosensing (
205–207), and tissue regeneration (
208–210). The well-defined shape of M13 and the possibility of displaying functional peptides on its surface have made it the phage of choice most often used for the assembly of new materials (
201,
209,
211). Genetically engineered M13 phages have been adapted to assemble and arrange quantum dots (
212,
213), build liquid crystals and films (
214–216), and fabricate nanorings (
217) and micro- and nanofibers (
218).
In 2006, Nam et al. reported the use of M13 to synthesize and assemble nanowires of cobalt oxide for the fabrication of battery electrodes (
202). M13 was first engineered to display gold-binding peptides with affinity for cobalt ions on its major coat protein (
202). The engineered M13 phages were then used to form nanowires of gold-cobalt oxide, improving the storage capacity of lithium ion batteries (
202). Later, the same group, using M13-based cobalt oxide nanowires, built and characterized microbattery electrodes with full electrochemical functionalities (charge storage capacity and performance rate) (
219). Cobalt manganese oxide nanowires made by M13 phage-mediated synthesis have also been used to build high-capacity lithium-oxygen battery electrodes (
220).
M13 phages have also served as templates for the integration of single-walled carbon nanotubes (SWNTs) into photovoltaic devices for highly efficient electron collection (
221). This method stabilized the SWNTs while maintaining their electronic properties and increasing the power conversion efficiency in dye-sensitized solar cells (
221).
Phage-based materials can also serve medical aims. M13-functionalized SWNTs have been used as effective probes for noninvasive fluorescence imaging of prostate tumors in mice (
207), as well as to target SPARC and to visualize deep, disseminated tumors in mouse models of human ovarian cancers (
222). By attachment of an antibacterial antibody to the p3 minor coat protein of the M13-SWNT complex, probes were made that could be used to image bacterial infections
in vivo (
223).
S. aureus endocarditis in mice was visualized by this method, and deeply buried infections were detected with high contrast and high specificity (
223). In order to image tumors in mice
in vivo, M13 was modified to display a SPARC-binding peptide on the p3 minor coat protein and a triglutamate motif on the p8 major coat protein for the templated assembly of magnetic iron oxide nanoparticles (
224). This strategy improved the magnetic resonance contrast of prostate cancer in mice compared with that of traditional nanoparticles used clinically, as each SPARC-targeting phage particle delivered a large number of detectable nanoparticles into the target cells (
224).
Mao et al. synthesized phage-based fibers and coatings with antibacterial properties by engineering phage M13 to express negatively charged glutamic acid peptides on its p8 major coat protein. Silverized phage fibers were then created by the electrostatic binding of silver ions, which have antibacterial properties (
225).
In vitro studies showed that these phage-based fibers exhibited bactericidal activity against
Staphylococcus epidermidis and
E. coli strains, which was visualized by fluorescence-based live-dead staining and zones of inhibition. Silverized phage fibers may thus be useful as anti-infective materials (
225).
Genetically engineered M13 phages have also been used to construct novel tissue-regenerating materials. Merzlyak et al. engineered the M13 phage to display cell signaling motifs (laminin peptides RGD and IKVAV) at the N terminus of the p8 major coat protein (
209). These phage building blocks self-assembled into structurally aligned liquid crystalline-like matrices that could maintain the viability, proliferation, and differentiation of hippocampal neural progenitor cells, as well as control their directional growth (
209). The same research group reported the use of engineered M13 phages to fabricate directionally organized two- and three-dimensional phage-based scaffolds, which showed good cytocompatibility and supported the directional growth and encapsulation of fibroblast cells (
226). Similarly, Wang et al. assembled M13-based matrices that provided a biomimetic microenvironment with controlled biochemical and biophysical cues for the directed differentiation of induced pluripotent stem cells (
227).
CONCLUSIONS
The arrival of the synthetic biology era married with the prodigious diversity of phages has led to powerful applications for therapeutics, diagnostics, and materials science. The introduction of new genetic engineering technologies has led to a more precise and accelerated modification of phage genomes for basic science as well as engineering. Phages have already been used to create new anti-infective agents, diagnostics, drug delivery systems, and vaccines, as well as new materials for nanoscale devices, imaging, and tissue scaffolds.
Despite the advances described above, phage research is still in its infancy. The tremendous diversity of phage types and structures in nature (
228,
229) has not yet been fully tapped. In fact, most naturally occurring phages have not yet been propagated in the lab. Of those that are known, many have not yet been characterized or are not yet amenable to genetic manipulation. Thus, phage engineering has so far involved only a small percentage of existing phage types. For example, most materials science applications have been based on the M13 phage, even though phages with other morphologies and sizes might extend the practical applications of phages in this field.
Next-generation sequencing technologies (
230,
231) have the potential to deposit phage genomes or phage-derived sequences into bioinformatic databases in large quantity without the need to first isolate these phages in the lab. These sequences can then be mined either to recreate natural phages via direct digital-sequence-to-DNA synthesis or to engineer novel phages that combine parts derived from various phages. New technologies are further needed to accelerate the design-build-test cycle for creating specialized phages and to make it possible to translate proof-of-concept academic work into real-world use more efficiently. As described above, highly reliable and rapid strategies that can be generalized to a wide range of phages are still lacking. Many strategies for engineering phages require the ability to genetically modify their bacterial hosts or to efficiently deliver exogenous DNA into these hosts, which is still a challenge for many bacterial species. Thus, new tools for genetic manipulation or DNA transformation are needed. Ideally, it would be possible to introduce multiple genetic alterations into phage genomes with high efficiency and at precise locations.
Finally, the vast majority of the work described in this review has resulted in genetically modified phages that may have significant benefits for diagnosing and treating bacterial infections, for treating nonbacterial diseases, or for constructing new materials. Despite the potential benefits, the acceptance of genetically modified phages for real-world applications may vary across different regions of the world. Strategies for inactivating phages so that they cannot propagate outside the lab, for example, by deleting essential protein genes from the phage genome and supplying these in trans in production hosts, may help to address such concerns. In the case of human use, the choice of compelling areas of tremendous medical need (e.g., for use against Gram-negative pathogens that are highly resistant to antibiotics and other antimicrobials) and explicit demonstrations of safety will both be important. Furthermore, techniques to contain the use of genetically modified phages for diagnostic and materials science applications and to inactivate the phages after use may also help to mitigate these issues. In summary, phage engineering is an area of research that is attracting intense interest and has great potential utility, but it has yet to be fully exploited.