HydroPhage Sustains Bacteriophage Delivery for Seven Days and Reduces Drug-Resistant Pseudomonas Wound Infections in Mice
A new Nature Communications study addresses a problem that remains poorly defined in bacteriophage therapy: how much phage should be administered, how often it should be given, and whether systemic or local delivery is preferable when treating an established wound infection. Researchers led by Yung-Hao Lin, Tejas Dharmaraj and Qingquan Chen, with Ovijit Chaudhuri and Paul L. Bollyky jointly supervising the work, developed a murine Pseudomonas aeruginosa wound model specifically designed to compare these variables and then used the results to engineer a sustained-release phage hydrogel called HydroPhage. The study was accepted on September 9, 2026.
The work involved researchers from Stanford University and Stanford Medicine together with collaborators from University Hospital Münster, Radboud University, New Jersey Institute of Technology and other institutions. Its experimental focus was not simply whether a bacteriophage could kill P. aeruginosa, but how pharmacological variables such as route, local retention, dose and exposure time determine efficacy once an infection has already been established. This is relevant because much preclinical phage work uses prophylactic treatment or phage administration close to the time of bacterial inoculation, conditions that do not reproduce treatment of an already colonized wound.
The investigators adapted a delayed-inoculation wound model. Mice were depilated three days before treatment, wounded two days before treatment and inoculated with bioluminescent P. aeruginosa PAO1-Xen41 24 hours before phage administration. This delay allowed a provisional wound matrix to develop before bacterial challenge and produced stable local infections without implantation of foreign material or induced immunosuppression. The animals showed local redness and irritation but minimal systemic illness, 100% survival and no detectable bacteria in the spleen, indicating that the model represented localized wound infection rather than bacteremia or sepsis.
Four obligatory lytic P. aeruginosa phages — PAML-31-1, LPS-5, Luz24 and OMKO1 — were initially evaluated, and PAML-31-1 was selected because it produced the strongest suppression of PAO1-Xen41. PAML-31-1 is a Pbunavirus that uses the pseudomonal O-specific antigen/lipopolysaccharide region as a receptor. The authors then used colony-forming units from harvested wounds together with daily bioluminescence imaging to quantify bacterial burden, eradication and subsequent regrowth.
The first experiment compared three topical dosing strategies. A low-dose group received 1 × 10^9 PFU once, a high-dose group received 4 × 10^9 PFU once, and a repeated-dose group received 1 × 10^9 PFU per day for four days, giving the same cumulative 4 × 10^9 PFU exposure as the single high-dose treatment. All phages were administered directly to the infected wound underneath a semi-occlusive Tegaderm dressing.
All three topical strategies significantly reduced bacterial burden relative to PBS. Mean reductions were 1.61 log10 CFU/mg with the single low dose, 2.33 log10 with the single high dose and 2.37 log10 with daily treatment. Complete culture-negative eradication at the endpoint occurred in 15.8% of low-dose mice, 18.2% of high-dose mice and 28.6% of repeatedly treated mice, compared with 4.2% of PBS controls. The eradication percentages were reported descriptively rather than subjected to significance testing, so the higher proportion with repeated dosing should be interpreted as a signal rather than proof that repeated dosing is statistically superior.
Bioluminescence revealed another important feature: phage treatment initially suppressed bacterial growth, but bacterial populations frequently rebounded during subsequent days. The high-dose group showed significantly lower bioluminescent burden than the low-dose group on day 3, while endpoint differences between the topical regimens became smaller. The data therefore suggested that increasing local exposure improves early bacterial control but does not necessarily prevent later regrowth.
The researchers then compared topical treatment with systemic delivery. A single intravenous dose contained the same total 4 × 10^9 PFU used in the high-dose topical regimen. Intravenous treatment produced a modest 0.90-log reduction in wound CFU compared with PBS, reaching P = 0.0223, but it produced almost no difference in eradication: 5% of intravenously treated mice had culture-negative wounds compared with 4% of controls. Bioluminescence-based killing rates were similarly low, at approximately 6% versus 0%.
Repeated systemic treatment was also tested using four daily intraperitoneal injections because repeated intravenous access was technically impractical in mice. This regimen delivered the same cumulative 4 × 10^9 PFU but did not significantly reduce CFU relative to control and produced only a 7% eradication rate. Phages could nevertheless still be detected in wound tissue, demonstrating that simply detecting viable phage at the endpoint was not sufficient to predict antibacterial efficacy.
The authors explicitly caution against extrapolating this result to systemic phage therapy in humans. Their wound model did not produce bacteremia, and practical limitations in mice prevented evaluation of continuous or repeatedly administered intravenous infusions similar to regimens that might be used clinically. The study therefore demonstrates superiority of local exposure over the particular systemic regimens tested in this model, not a general conclusion that intravenous phage therapy is ineffective for human wound infections.
These dosing experiments led the team to focus on maintaining high local phage concentrations without requiring a patient to receive daily topical applications. The researchers designed HydroPhage as a soft injectable hydrogel composed of thiolated hyaluronic acid and polyethylene glycol-based crosslinkers. Stable covalent thioether bonds are combined with reversible hemithioacetal crosslinks generated using PEG-aldehyde and PEG-benzaldehyde molecules. The reactions occur under physiological conditions without harsh solvents or catalysts, and phages can be mixed directly with the polymer precursors before gelation. The resulting material can be delivered through a 28-gauge needle and conforms to irregular wound surfaces.
The optimized HydroPhage formulation contained 5 mg/mL thiolated hyaluronic acid, 1 mg/mL hyperbranched PEG multi-acrylate and 75 mg/mL four-arm PEG crosslinkers composed of 90% aldehyde and 10% benzaldehyde species. For most experiments, PAML-31-1 was incorporated at 1 × 10^11 PFU/mL.
Phage loading concentration substantially affected recovery from the material. When HydroPhage contained 10^11 PFU/mL, approximately 62% of encapsulated PAML-31-1 was recovered during the release experiment, compared with 28% at 10^10 PFU/mL and 21% at 10^9 PFU/mL. Comparable release behaviour was observed with two other P. aeruginosa phages, LPS-5 and Luz24, suggesting that the release principle was not unique to PAML-31-1.
The major difference from a conventional alginate gel was the release profile. The ionically crosslinked alginate comparator released nearly all encapsulated phage during the first 24 hours, whereas HydroPhage maintained phage release at approximately 10^9 PFU per millilitre of hydrogel or higher for as long as seven days. Release closely correlated with loss of polymer mass, indicating that hydrolytic erosion of the network rather than a simple initial diffusion burst was the principal release mechanism.
The estimated hydrogel mesh size ranged from approximately 20 to 200 nm, which is relevant because PAML-31-1 measures about 174 nm in length with a head diameter around 61 nm. The authors propose that relatively low stable-crosslink density prevents the network from permanently trapping phage particles while the dynamic component provides viscosity, adhesion and mechanical adaptability. Around 30–35% of incorporated phages had been released by 18 hours, followed by continued release as the matrix progressively eroded.
HydroPhage retained antibacterial activity after release. Phages recovered on day 7 still suppressed planktonic P. aeruginosa, and phages were able to migrate from the hydrogel across semi-solid agar and inhibit bacterial growth beyond the material itself. The authors also tested mature PAO1-GFP biofilms, where free PAML-31-1 reduced biofilm thickness by 43% after 21 hours and HydroPhage reduced it by 52%. Biofilm density fell by 57% with free phage and by 65% with HydroPhage, whereas density increased by 55% with control hydrogel and 72% in untreated biofilms.
The same system was then evaluated in infected animals. A single topical HydroPhage application produced a mean 2.33-log10 reduction in bacterial burden relative to PBS, with n = 20 and P < 0.0001. This was essentially identical to the 2.33-log reduction obtained using the single high-dose free-phage treatment. Culture-based eradication occurred in 25% of HydroPhage-treated animals compared with 18% receiving high-dose free phage and approximately 4% of controls. Bioluminescence-based killing rates were 40% with HydroPhage, 25% with free high-dose phage and 0% with PBS.
HydroPhage therefore did not produce a larger average endpoint CFU reduction than an equivalent amount of free topical phage in this model. Its potential advantage was sustained exposure from a single administration. The authors note that the Tegaderm dressing used experimentally retained free liquid phage at the wound site and may consequently have reduced the apparent difference between a retention hydrogel and freely applied phage solution.
The study also tested an extensively drug-resistant clinical P. aeruginosa isolate, CPA012, obtained from the Stanford Health Care Clinical Microbiology Laboratory. CPA012 was resistant to amikacin, aztreonam, cefepime, ceftazidime, ciprofloxacin, imipenem, levofloxacin, meropenem and tobramycin, with intermediate resistance to tazobactam. PAML-31-1 retained lytic activity against this isolate and disrupted its biofilm in vitro.
Mice infected with CPA012 were assigned to PBS, systemic tobramycin at 150 mg/kg/day, combined topical plus intravenous free phage, or a single topical HydroPhage treatment. HydroPhage significantly reduced endpoint bacterial burden relative to PBS, with P = 0.0110. Culture-negative eradication occurred in 3 of 13 HydroPhage-treated animals, or 23.1%, compared with 1 of 11 animals receiving tobramycin, 0 of 7 receiving topical plus intravenous free phage and 0 of 11 PBS controls. These eradication proportions were descriptive and were not statistically compared, and the authors stress that the treatment regimens were methodologically different, so the experiment does not demonstrate that HydroPhage is clinically superior to tobramycin.
The inflammatory analysis provided a secondary biological readout. P. aeruginosa infection increased several inflammatory cell populations in the wound, including CD45-positive cells, NK cells and neutrophils. Among HydroPhage-treated animals in which bacteria were successfully eradicated, lower proportions of CD45-positive cells and neutrophils were observed compared with infected HydroPhage-treated wounds, while the overall pattern was consistent with reduced inflammatory cellular infiltration after bacterial clearance. Macrophage and T-cell populations remained low even after eradication, indicating that bacterial elimination did not immediately normalize the complete immune-cell composition of the tissue.
The hydrogel itself was also examined using RAW264.7 murine macrophage-like cells. Neither the control HA-PEG gel nor PAML-31-1 caused detectable cytotoxicity or induced the M1/M2 polarization measured in the study. These experiments are limited to an in vitro macrophage model and do not establish human biocompatibility, but they provided no obvious toxicity signal for the formulation under the conditions tested.
One of the most important results was that optimized delivery did not eliminate the evolutionary problem of phage resistance. Isolates collected from every phage-treated group — low dose, high dose, repeated topical dosing, HydroPhage and systemic treatment — displayed substantially increased resistance when re-exposed to PAML-31-1. Resistance therefore arose even in regimens that produced little antibacterial benefit, and its magnitude was not correlated with the extent of bacterial reduction.
Whole-genome sequencing of completely resistant isolates recovered from HydroPhage-treated wounds identified nonsynonymous mutations in genes involved in LPS biosynthesis. Because the O-specific antigen/LPS region is the receptor used by PAML-31-1, these mutations provide a plausible genetic explanation for resistance through alteration of phage receptor structures. The sequencing result was performed on selected resistant isolates rather than every bacterial population in the study, so it establishes a likely mechanism in those isolates rather than a universal resistance pathway.
Bioluminescence showed bacterial rebound within approximately 72 hours after monophage treatment in many animals. HydroPhage was the only localized strategy that significantly reduced persistent regrowth compared with intravenous treatment, with P = 0.0064, but even HydroPhage did not prevent resistance or produce eradication in the majority of animals. The authors therefore argue that delivery pharmacokinetics and bacterial evolution must be considered together: repeated or sustained exposure can maintain high concentrations during the initial therapeutic window, but resistant populations may subsequently become dominant.
This distinction between administered dose and effective exposure is central to the paper. Endpoint phage titres in wound tissue were similar after local and systemic administration despite substantially different antibacterial outcomes. The authors therefore conclude that a single terminal PFU measurement cannot reconstruct earlier phage concentrations, retention or productive replication inside the wound. Daily dosing and HydroPhage were the two strategies that maintained renewed phage exposure over time and were also among the most effective at controlling early bacterial growth.
The study does not establish HydroPhage as a treatment ready for human use. The in vivo work relied primarily on a single phage and a mouse wound model, while human wounds can contain multiple bacterial species, heterogeneous biofilms, vascular disease, diabetes, immune dysfunction and much more complex tissue architecture. The researchers did not compare HydroPhage directly with an antibiotic-loaded wound hydrogel, nor did they test a combined phage-and-antibiotic hydrogel. They identify both comparisons as priorities for future work.
The model itself also creates an important experimental caveat. Tegaderm was required to prevent environmental contamination and retain liquid treatments, but by physically trapping free phages in the wound it may have artificially improved retention in the free-phage groups and therefore reduced the measurable advantage of HydroPhage. Systemic treatment was additionally constrained by the inability to perform clinically realistic repeated intravenous infusions in mice, meaning the local-versus-systemic comparison cannot be translated directly to human treatment protocols.
Another limitation is the use of monophage therapy. PAML-31-1 resistance emerged rapidly through mutations consistent with receptor modification, and the authors suggest that phage cocktails could potentially reduce this problem. They also emphasize, however, that robust preclinical and clinical evidence proving general superiority of cocktails over single-phage treatment remains limited, and future experiments are planned to compare monophage and cocktail delivery directly, including cocktail-loaded hydrogels.
The translational concept proposed by the researchers is therefore relatively specific. HydroPhage is intended as an advanced wound dressing able to maintain a high local phage concentration for approximately one week rather than requiring repeated applications of liquid phage. The design was conceived with existing wound-care workflows in mind, including approximately weekly debridement visits, but the study did not test the material in humans or demonstrate that weekly treatment would be sufficient clinically.
The main contribution of the work is consequently pharmacological rather than simply microbiological. In this model, local exposure was more effective than the systemic regimens tested, high or repeated local dosing improved early bacterial suppression, HydroPhage maintained viable phage release for seven days, and a single hydrogel application reduced both laboratory-strain and extensively drug-resistant P. aeruginosa wound burdens. At the same time, the rapid appearance of phage-resistant bacteria prevented reliable eradication in most animals and identifies resistance management as a separate problem that improved delivery alone cannot solve.
The study therefore supports a development strategy in which phage therapy for wounds is treated as a drug-delivery problem as well as a phage-selection problem. Sustained local concentrations, the temporal pattern of exposure, bacterial resistance, biofilm penetration and combination with conventional antimicrobials may all determine whether a phage active in vitro produces durable bacterial control in vivo. HydroPhage provides a preclinical platform for testing these variables, but additional phages, cocktails, antibiotic combinations, polymicrobial models and ultimately human wound studies will be required before its clinical value can be determined.
Sources :
Yung-Hao Lin, Tejas Dharmaraj, Qingquan Chen et al. Dosing and delivery of bacteriophage therapy in a murine wound infection model. Nature Communications, 2026.
https://doi.org/10.1038/s41467-026-77992-1
Primary article manuscript and supplementary methodological information used for the dosing, HydroPhage formulation, resistance, biofilm and animal-model results described above.
.png){kind=logo}
Comments
Post a Comment