Engineered phages are being developed in three roles: as anti-bacterial agents that lyse bacteria antibiotic therapy can no longer reach, as vectors that restore antibiotic sensitivity to resistant strains, and as delivery systems for immune or anti-tumour payloads. The three roles are separated by different engineering requirements, and the field has moved furthest on the first. This overview is drawn from a 2025 peer-reviewed review co-authored by researchers at Shandong University of Technology and Shandong Jiuyi Biotechnology (Microbial Pathogenesis, 198:107052; PMID 39442821), which surveys those directions and what remains open.
Product referenced in this article
Phage Display Technology Platform — application notes, specification table and document downloads for this range.
Why engineered phages rather than wild-type isolates
A wild-type phage that works against one strain rarely works against the next, because host range is determined by the specific receptors on the bacterial surface. A cocktail of wild types improves coverage but does not address the underlying problem, which is that a new strain carrying a different receptor is simply outside the range. Engineering changes the situation by making the recognition step programmable.
The work concentrates on the tail fibre or tail pin protein, which is what makes initial attachment. Three routes have been used: directed evolution against a broader panel, rational design of the receptor-binding domain, and display libraries in which variants with a broadened range are selected directly. Each trades breadth against specificity, and that trade is the central difficulty of the field — a phage that binds too broadly will lyse a bacterium the host depends on.
What the 2025 review maps out: four engineering routes
The review (Cao X., Yu T., Sun Z., Chen M., Xie W., Pang Q., Deng H. Engineered Phages in Anti-Infection and Anti-Tumor Fields: A Review. Microbial Pathogenesis 2025, 198:107052; PMID 39442821) restates why engineering is needed at all: natural phages are highly specific and leave the normal flora untouched, which is exactly why they are attractive, but their narrow host range, the ease with which bacteria tolerate them, and the difficulty of controlling delivery and immunogenicity have kept them at the edge of mainstream use. The argument is that transgene or chemical modification lets each of those limits be addressed directly.
Four routes are surveyed. Host-spectrum modification: replacing or reprogramming the receptor-binding protein, usually on the tail fibre, reshapes or broadens the range so one construct covers more serotypes. Resistance reversal: an engineered phage carries sensitivity back into a resistant strain, after which the antibiotic does the killing. Antigen presentation: displaying a tumour-associated or pathogen antigen on the capsid turns the phage into a vaccine or immunotherapy scaffold. Targeted delivery: the phage is repurposed as a carrier that brings an anti-cancer or anti-infective payload to the infection site or malignant tissue. The review's conclusion is that these modifications take phages from a single killing tool to a programmable biological platform.
Resistance reversal and antibiotic re-sensitisation
The second role is different in kind. Here the phage does not need to kill the bacterium on its own. A phage carrying a small resistance-disrupting payload can select for bacteria that become sensitive to the antibiotic again, at which point the antibiotic does the killing. Because resistance to the phage construct carries a fitness cost while restored antibiotic sensitivity is a large benefit in a treated host, the selection pressure runs in the direction you want.
This is an area where the biology is more advanced than the deployment. The concept is well supported in laboratory study, and the practical obstacles are manufacturing, dosing schedule and the regulatory pathway for a product that is not itself an antibiotic. It also depends on phage biology in ways that matter operationally: a temperate phage integrating into the host genome can be lost when the bacterium divides, so strains carrying the resistance gene in their own genome are preferred for this application.
Antigen presentation and anti-tumour applications
The third role is the most different. Here the phage is not directed at a bacterium but at a mammalian cell, and its value is that it can display a tumour-associated antigen at high copy number without the adjuvant problems of a subunit vaccine. Multiple antigens can be displayed on one particle, which allows a single construct to present a combination, and the same particle can be engineered for a particular adjuvant effect.
The bottleneck is not engineering but delivery and immunogenicity. A multivalent particle is strongly immunogenic, and the useful response to a tumour antigen is not the one that a vaccine is designed to produce. The engineering questions that remain are about controlling where the particle goes and how long it persists, since both determine whether the presentation translates into a response.
What phage display is actually for
Phage display is a different technique and is often confused with the above, so it is worth separating. In display, a gene is fused to a gene encoding a coat protein, and the protein it encodes is presented on the phage surface. A library of billions of variants can then be screened for binding to a target, with the binding phage recovered and amplified. Nothing is lysed and no bacteria are killed; the phage is a display surface that carries a library.
Because the selected variant is linked to the gene that encodes it, this is a route to finding an antibody or a binding protein without having to design one. It is the standard method behind the human single-domain antibody libraries, which are supplied as a screening service: a target is presented, binders are recovered, and the resulting sequences are characterised and expressed in a chosen format.
For buyers specifying this work, the questions that determine whether a campaign succeeds are practical ones. What is the target and how it is presented, because a monovalent soluble antigen selects differently from a cell or tissue sample. What the selection stringency is across rounds. What the final output format is, because a sequence delivered as a gene is not yet a product. And what the acceptance criteria are for a hit, which should be agreed before the campaign starts rather than interpreted afterwards.
Where the field is genuinely open
Three gaps recur. The first is regulatory: a phage product is classified differently in different jurisdictions, and the classification determines the clinical pathway, which in turn determines what manufacturing evidence is required. The second is manufacturing. Phage production is biological, variable and, at anything beyond laboratory scale, expensive per dose, and the economics of a biological anti-infective depend on producing it cheaply enough for the indication to be worth treating.
The third is delivery and residency. A phage that is cleared quickly must be dosed repeatedly, and in the livestock applications this has so far been the limiting factor rather than the biology. A live replicating phage persists longer, but persistence in the host is exactly what regulators scrutinise, and the balance between the two is unresolved for most applications.
We work on phage display and related platform services rather than on clinical phage therapy, and we are explicit about the boundary. The services we supply are library screening and recombinant antibody discovery, which are research-stage activities. Nothing we supply is a clinical product, and any claim that a display-derived sequence is a treatment would be a claim the work does not support.
Where 91bio's phage work sits: from the lab to the farm
91bio's role in this field is practical rather than clinical. The review's corresponding author is affiliated with both Shandong University of Technology and Shandong Jiuyi Biotechnology, and that collaboration is the link between the two. On the 91bio side the relevant assets are phage strain resources, fermentation process know-how and formulation — the part of the chain that turns a construct designed in a lab into something that can be made, stored and applied at scale.
The near-term industrial use is livestock and aquaculture, where "reduce and replace antibiotics" policies have made phage biocontrol one of the most watched alternatives to in-feed antibiotics. A phage that targets a specific pathogen and spares the rest of the gut or pond microbiota is a good fit, because specificity is the property antibiotics lack. 91bio's work here is product development for those environments — see our phage biocontrol for livestock — not clinical phage therapy.
The same collaboration has produced related, publicly retrievable outputs along a "synthetic biology and new medicine" line: a study on planarian TatD DNase against biofilms in Environmental Research, and a review of CDK inhibitors in ChemistrySelect. These are the research record the engineering work builds on, not product claims.
We are explicit about the boundary. 91bio supplies phage-display and recombinant-antibody discovery as research-stage platform services, and develops phage biocontrol products for livestock and aquaculture. We do not supply clinical phage-therapy products, and any description of a display-derived sequence as a treatment would be a claim the work does not support.
Frequently asked questions
- What is the difference between phage therapy and phage display?
- They use the same biological system for different purposes. In phage therapy, a live phage infects and lyses a target bacterium to treat an infection. In phage display, a gene is fused to a coat-protein gene and the encoded protein is presented on the surface, so a large library can be screened for binding to a target. Display does not kill anything; the phage is a display surface.
- Why do engineered phages need a broader host range?
- Because host range is set by the specific receptors on the bacterial surface, and those receptors differ between strains. A phage selected against one clinical isolate will often not attach to another. Broadening range by engineering the tail fibre or tail pin is what allows coverage across a panel, and the difficulty is doing so without making the phage bind so broadly that it damages a bacterium the host needs.
- Can phages replace antibiotics?
- Not in the sense of a direct substitute, and anyone presenting it that way is overselling. The realistic expectation is complementary use: phages for bacteria that antibiotics no longer reach, phages as a means of restoring antibiotic sensitivity in resistant strains, and antibiotics still doing the killing in the second case. Development also faces real obstacles in manufacturing cost, delivery and regulatory classification.
- What does a phage display service actually deliver?
- Recovered binder sequences from a screened library, together with the characterisation data from the campaign: which clones were selected, at what stringency, and what they bound. The output format matters and should be agreed in advance, because a delivered gene sequence is not yet a product. We supply library screening and recombinant antibody discovery as research-stage services, not clinical products.
- Is this based on a published paper?
- Yes. The directions summarised here are drawn from a 2025 peer-reviewed review, 'Engineered Phages in Anti-Infection and Anti-Tumor Fields: A Review' (Microbial Pathogenesis, 198:107052; PMID 39442821), co-authored by researchers at Shandong University of Technology and Shandong Jiuyi Biotechnology. Related collaboration outputs include a biofilm study in Environmental Research and a CDK-inhibitor review in ChemistrySelect.