Gnotobiotic zebrafish: control the microbiome

Raising zebrafish without detectable microorganisms, or with a deliberately defined microbial community, can turn a microbiome association into a testable causal question. That control comes with a cost: diet, stocking density, colonisation timing and verification methods become part of the model. Three studies published in 2025 and 2026 now extend these experiments beyond the earliest days of life. They expand the research toolkit, but also show why “germ-free” or “gnotobiotic” must be treated as a measured experimental state rather than a permanent attribute of the fish.

Germ-free and gnotobiotic describe different conditions

A germ-free, or axenic, animal is maintained without microorganisms detectable by the monitoring methods used. A gnotobiotic animal has a known microbial status: it may be germ-free, mono-associated with one strain or colonised by a defined consortium. These conditions answer different questions and require different controls.

An experiment may ask what changes in the absence of microbes, what one bacterial strain contributes, or how a simplified community behaves. Unrecognised contamination compromises the first comparison. Uneven deliberate colonisation can add animal-to-animal variability to the second and third, even when every vessel receives the same inoculum.

The microbiome contributes to digestion, nutrient acquisition, immune development and xenobiotic metabolism. Removing it is therefore not a neutral technical step. Conventional controls, the effect of the sterilised diet itself and developmental stage need to remain distinct in the design. A phenotype in germ-free fish may be the phenomenon of interest, but it is not automatically representative of conventionally raised animals.

Irradiated feed extends culture to 55 days post-fertilisation

In 2026, Okyere, Di Fulvio and Gaulke tested gamma-irradiated dry chow for long-term germ-free husbandry. Their protocol maintained zebrafish without detectable microbes to 55 days post-fertilisation, well beyond the early larval window that has constrained many gnotobiotic studies. In conventionally raised fish, the irradiated diet did not affect growth or survival and had only modest effects on microbial community composition and diversity.

The germ-free animals did not become physiological equivalents of conventional fish. At 55 days they were smaller and lighter, and their gene-expression profiles differed in pathways involving immune responses, xenobiotic metabolism, organ development, liver function and lipid metabolism. Those changes make the model informative, but they also prevent the absence of a microbiome from being treated as an inconsequential husbandry setting.

For facilities, a stable sterilised feed can reduce reliance on live prey that are difficult to control microbiologically. It does not remove the need to document irradiation dose, batch storage, nutritional integrity, actual intake and animal performance. Compatibility in this study is not automatic validation for every formulation, zebrafish line, density or scientific endpoint.

Live feed may deliver or reshape the intended community

Larval zebrafish respond strongly to moving prey. Márquez Rosales and colleagues therefore used UV irradiation to reduce the bacterial burden of rotifers while retaining their movement and nutritional role. The treatment did not preserve sterility in initially germ-free fish. It did, however, allow pre-existing bacterial communities to persist through experiments lasting eight days of feeding, up to 13 days post-fertilisation. Persistence varied among bacterial species.

This is an important operational distinction: substantially reducing the microbial load of live feed is not the same as making it sterile. If the endpoint requires germ-free animals, microbiological checks of both prey and fish remain essential. If the aim is to maintain a defined consortium, the facility instead needs evidence about which strains persist, in what abundance and with how much variation among individuals.

Byatt and colleagues took another approach in 2026, using bacteria-loaded Tetrahymena thermophila as a live vector. Under their conditions, live feed increased the initial bacterial load in the larval gut by approximately one order of magnitude. Larval density and exposure timing also affected colonisation, which rose from 10^4 to 10^6 bacterial cells over four days. The authors demonstrated colonisation with several bacterial strains, including Aeromonas veronii, Pseudomonas aeruginosa and Vibrio cholerae.

These numbers belong to a specific experimental system. They should not become universal target loads, feeding schedules or release criteria. A different bacterial strain, host line, vessel geometry or analytical method may produce a different colonisation pattern.

Define verification before the experiment starts

A robust protocol begins with a measurable definition of the intended microbial state. For a germ-free group, the plan should identify the tested matrices — water, feed, fish and vessel —, culture and molecular methods, detection limitations, controls and the action triggered by an uncertain result. For a deliberately colonised group, it should also record strain identity, inoculum viability and dose, timing, and any required measurement of abundance or location.

Traceability should connect every result to the feed batch, sterilisation treatment, vessel, density, medium changes and handling events. Negative controls should not be opened or fed after colonised groups with the same equipment. Consumables, pipettes and waste need directional workflows that protect the intended microbial status, even when the organisms used are not being presented as pathogens.

Predefined exclusion rules are equally important. Removing a contaminated vessel, an uncolonised fish or an animal with atypical growth only after seeing the outcome may change the analysed population and bias the conclusion. Teams should decide in advance when to repeat, exclude, analyse separately or retain a deviation as biologically meaningful evidence.

What the new studies support — and what they do not

The three studies solve different technical problems: maintaining fish longer without introducing detectable microbes, preserving a known community while providing live prey, and improving bacterial delivery to the larval gut. Together they make later developmental stages and longer host–microbe interactions more accessible. They do not establish one universal husbandry system for every gnotobiotic programme.

They also do not erase the biological distance between a host deprived of microbes and an animal living with a complex conventional community. When interpreting a phenotype, researchers need to ask whether it reflects the introduced strain, the absence of other microbes, diet, developmental delay, stocking conditions or an interaction among these factors. Results in gnotobiotic larvae or juveniles can identify mechanisms within a controlled system; they do not automatically describe conventional zebrafish or another fish species.

Vetofish can support aquatic research facilities with risk assessment, microbiological monitoring plans, husbandry procedures, sampling workflows and predefined decision criteria. The practical objective is to make microbial status demonstrable, deviations traceable and conclusions proportionate to the model that was actually achieved.