Can probiotic yeast carry metals out of the gut?

Start with the strain and the material

Saccharomyces boulardii belongs within S. cerevisiae diversity. Khatri and colleagues’ comparative genomes placed boulardii isolates near wine-associated strains, while identifying differences among genomes. Shared species identity therefore supports mechanistic comparison, not interchangeability. A laboratory yeast, a probiotic product and brewery waste can differ in growth history, wall architecture and processing. An experiment should name its isolate before a result becomes a claim about “probiotic yeast.” ^[6]

One particularly useful identity check connects two primary papers. Stathatou and colleagues described trace-lead removal by S. cerevisiae, but their Methods identify ATCC MYA-796. Batista and colleagues had sequenced that same accession as S. cerevisiae var. boulardii, originating from Ultra Levure. This is a documented boulardii-lineage result. It is still a result for an experimentally prepared material, not evidence that a retail capsule has the same adsorption capacity inside a person. ^{[1] [2]}

Biosorption means association with biological material; it need not require metabolism. Cell-wall adsorption occurs outside the plasma membrane. Membrane transport moves a substance across a selective boundary, and intracellular sequestration stores it elsewhere. These processes may coexist, but a fall in dissolved metal after centrifugation does not by itself identify which occurred. Earlier free-versus-immobilized yeast experiments also implicated the wall in Pb(II) capture, supporting this as a material property worth investigating. ^{[7] [8]}

Explore unexpected metal biology across drugs, plant compounds and microbes →

Lead: a strong material result with a specific boundary

In the MYA-796 study, washed, freeze-dried biomass contacted lead-nitrate solutions at pH 5 and 25°C. At 100 μg/L initial Pb, equilibrium was reached within five minutes; uptake approached 12 mg/g biomass at 1,000 μg/L initial Pb. Spectroscopy implicated wall functional groups, with mannoproteins and β-glucans proposed as key contributors. “Inactive” describes the non-growing assay material; lack of growth in that experiment does not establish that every cell had been killed. ^[1]

The wall hypothesis raises a useful evidence question: across published preparations, does adsorption track wall chemistry more closely than cellular activity? The 2009 study of free and immobilized biomass offers a comparison, although its performance cannot be transferred to another strain or matrix. For environmental treatment, separation of the loaded sorbent defines removal from the treated water. An oral application adds a different requirement: evidence that the associated metal remains unavailable for absorption until the material leaves the intestine. ^[8]

Follow lead through chemical form, exposure and biological compartments →

Three elements, three different evidence problems

Arsenic: protection can occur without strong binding

Clemente and colleagues tested As(III) and As(V) separately. Yeast removal from solution was below 10%, yet some microorganism conditions reduced arsenic transport across intestinal cells. That mismatch is informative: a barrier-related effect is a competing explanation, rather than proof of efficient adsorption. The study’s highlighted rice result belonged to Lactobacillus plantarum and should not be reassigned to Saccharomyces. The accessible report does not justify identifying its yeast panel as a particular boulardii strain. ^[4]

A 2026 rat study by Thomas and colleagues supplies a different observation: oral co-treatment with a boulardii preparation improved several antioxidant and hematologic measures during sodium-arsenate exposure. The journal abstract reports 30 male Wistar rats across five groups over 28 days. It does not report arsenic mass balance or a strain accession. Its docking calculations concern proposed interactions with antioxidant enzymes, not measured yeast–arsenic binding. Protection from injury and reduced retained arsenic are therefore distinct hypotheses. ^[9]

Living yeast also has its own priorities. In laboratory S. cerevisiae, Ghosh, Shen and Rosen used gene disruption and radiotracer experiments to separate Acr3-dependent arsenite exclusion from Ycf1-dependent vacuolar sequestration involving glutathione. A cell can survive by exporting arsenic, potentially returning it to its surroundings. These laboratory pathways offer competing explanations for yeast survival; their operation during human gastrointestinal transit has not been established here. Metal tolerance is not automatically metal capture. ^[7]

Why arsenic oxidation state changes the biological question →

Mercury: the food can defeat the binder

Jadán-Piedra and colleagues compared inorganic Hg(II) and methylmercury in simulated digestion. Their supplementary food comparison included wine yeasts VRB and T73 and the preparation labeled Ultralevura. Reduced bioaccessibility in mushrooms did not carry over to swordfish. In a separate BY4741 experiment, cysteine and albumin reduced mercury retention. The authors proposed competition from mercury–cysteine or peptide complexes as an explanation for the seafood result. That mechanism remains an interpretation; the matrix-specific failure itself is observed. ^[5]

This is why “binds mercury” is too broad to predict performance. A useful cross-study question is whether differences in reported retention follow mercury species and food composition after accounting for the amount initially present. Lower dissolved mercury alone would be ambiguous if the remaining fraction contained a more readily transported complex. The relevant evidence must describe both quantity and chemical form. The existing contrast between mushrooms and seafood makes the meal’s chemistry central to interpretation.

Compare inorganic mercury and methylmercury →

The exit problem: adsorption can reverse

Mirmahdi and colleagues tested ATCC 9763 after 24 hours of contact with a Pb/As/Hg/Cd mixture, then applied sequential simulated gastric and intestinal conditions. Prepared fluids used pH 2 pepsin and pH 8 pancreatin/bile, with two-hour stages at 37°C. Untreated, autoclaved, ethanol-treated and alkali-treated cells did not behave identically. Release occurred under some conditions, while retention of some metals increased elsewhere. These are conditional adsorption/desorption results, not a universal verdict that all yeast complexes remain bound or all fail. ^[3]

Human transit is independently demonstrable. Klein and colleagues recovered administered S. boulardii from healthy volunteers’ feces; ampicillin changed recovery and steady-state fecal levels. But recovering yeast does not establish recovery of its original metal cargo. A complete balance must distinguish unabsorbed dietary metal, metal newly secreted by the host, and metal released from yeast along the way. Stool concentration alone is inadequate because changing stool mass can change concentration without changing total elimination. ^[10]

A targeted search for this review did not identify a strain-defined S. boulardii human intervention demonstrating increased Pb, As or Hg elimination or a reduced body burden. That search result is not evidence that the mechanism is impossible. It identifies the missing link between material chemistry and a human outcome. Prevention of new absorption is also different from extracting metal already retained in tissues; the latter requires a demonstrated route back into the intestinal compartment.

Compare luminal binding with biliary secretion and fecal elimination →

A living carrier introduces additional biology

The same preparation could alter microbial interactions, intestinal permeability or transit without removing appreciable metal. Conversely, survival mechanisms such as export could undermine retention. Published observations involving living yeast, inactive biomass and cell-free material therefore answer different questions. If protection appears where durable adsorption is absent, a barrier-related or secreted effect becomes a competing explanation. A synthesis should retain those alternatives rather than describe every favorable response as “binding.” ^{[7] [9]}

Viability also changes the safety question. Rannikko and colleagues found an association between probiotic use and Saccharomyces fungemia in Finnish hospital records. The study does not estimate absolute risk for healthy users, and strain matching was incomplete. Its clinical setting also underscores why evidence about a nonliving sorbent cannot establish the safety of a living product. A water-treatment adsorbent and an orally administered living yeast are different interventions, each requiring evidence appropriate to its destination. ^[11]

Evidence that would distinguish capture from clearance

The most informative evidence would be agreement across scales: a defined metal species remains associated with yeast-derived material, intestinal delivery is lower, and total elimination and internal-exposure measurements tell a consistent story. A mismatch would still clarify the mechanism—for example, strong water-treatment performance with no demonstrated oral benefit, or protection from injury without reduced metal retention. These distinctions give future evidence a clear interpretive purpose while keeping established observations separate from the proposed explanation.

Read how this guide separates observations, hypotheses and discriminating predictions →