Submitted:
11 November 2026
Posted:
12 November 2026
You are already at the latest version
Abstract
Bacillus thuringiensis is widely used as a biopesticide but has recently been excluded from further EFSA qualified presumption of safety assessments. Its identification through parasporal crystal production does not necessarily resolve genomic lineage, and enterotoxin genes alone do not establish toxin production or disease. We propose integrating genomic identity, strain-specific toxin activity, exposure and epidemiological evidence to improve food surveillance and safety assessment.
Keywords:
Bacillus cereus group
; biopesticides
; enterotoxins
; foodborne disease
; genome-based taxonomy
; risk assessment
; surveillance
Introduction
Bacillus thuringiensis has long been a cornerstone of microbial pest control. Its spores and parasporal proteins are used in products targeting insect pests, and its pesticidal proteins have also provided important tools for studying insect–microbe interactions. At the same time, B. thuringiensis belongs to the Bacillus cereus sensu lato (s.l.) group, whose members share considerable genomic similarity and include strains associated with foodborne disease. The coexistence of these agricultural and food-safety perspectives makes B. thuringiensis a particularly instructive example of the difficulty of assigning a single safety meaning to a bacterial name (EFSA BIOHAZ Panel, 2016; EFSA BIOHAZ Panel, 2025; EFSA BIOHAZ Panel, 2026).
Recent findings have brought the issue into focus. Schmidt and colleagues detected crystal-forming isolates classified as B. thuringiensis in fresh produce and found that most of a selected, sequenced subset were closely related to commercial biopesticide strains (Schmidt et al., 2027). In parallel, the EFSA Panel on Biological Hazards concluded that B. thuringiensis should not be recommended for qualified presumption of safety (QPS) status and should be excluded from further QPS assessment (EFSA BIOHAZ Panel, 2026). These findings call for careful interpretation, not a choice between treating every B. thuringiensis strain as hazardous or assuming that a long history of agricultural use establishes safety under every exposure scenario.
A further complication concerns what the name B. thuringiensis represents. Parasporal crystal production is an operationally useful phenotype, but genome-based studies have shown that crystal-forming strains can belong to different chromosomal lineages within B. cereus s.l. This commentary considers how phenotype, genomic identity, source, enterotoxin activity and exposure can be integrated to improve the interpretation of food-surveillance findings.
Fresh Produce and the Meaning of a Positive Result
The occurrence of B. thuringiensis on fresh produce and its potential association with commercial biopesticides have been investigated for more than two decades. Early studies identified isolates from fruits and vegetables with typing profiles similar to those of strains used in bioinsecticides (Hendriksen and Hansen, 2006). Subsequent investigations combined phenotypic characterization with whole-genome sequencing and showed close genetic relationships between food-associated isolates and commercial biopesticide strains (Biggel et al., 2022; Frentzel et al., 2020). In spinach, genomic comparisons supported by information on product application linked some isolates to the biocontrol products used during cultivation (Zhao et al., 2022). More recently, surveys in France and Germany have further documented the occurrence of B. thuringiensis or biopesticide-associated isolates in fresh fruits and vegetables (Felten et al., 2025; Frentzel et al., 2026).
Within this broader context, Schmidt et al. (2027) investigated 280 samples representing 30 types of fresh produce collected at a retail distribution facility in Switzerland. Sixty-two samples (22%; 95% confidence interval, 18–27%) yielded at least one crystal-forming isolate and were classified as B. thuringiensis-positive. Of the 219 recovered B. cereus sensu lato isolates, 150 (69%) produced parasporal crystals. In addition, 20 of 22 crystal-forming isolates selected for whole-genome sequencing were closely related to commercial biopesticide strains. However, these isolates were not selected according to a predefined sampling scheme; therefore, this proportion should not be extrapolated to all crystal-forming isolates recovered in the study.
These findings extend previous observations to a broader range of fresh-produce categories and provide a recent dataset for food-safety surveillance. However, they also illustrate the need to distinguish between the detection of a crystal-forming phenotype, the genomic identity of an isolate and the attribution of its presence to a particular product or application.
Beyond the Crystal: Phenotype and Genomic Lineage
Historically, the presence of parasporal crystals has been a key phenotypic feature used to distinguish B. thuringiensis from other members of the B. cereus group. In Bergey’s Manual, the routine recognition of B. thuringiensis was described as relying largely on the microscopic observation of parasporal crystals in sporulating cells (Logan and Vos, 2015). More recently, EFSA also adopted an operational definition of B. thuringiensis for its QPS assessment based on the observation of insecticidal crystals, complemented by molecular identification approaches (EFSA BIOHAZ Panel, 2026). Although this phenotype-based approach remains useful for routine screening and food surveillance, crystal formation does not necessarily correspond to a single genome-defined species.
Genome-based frameworks have emphasized the distinction between chromosomal relatedness and traits such as pesticidal protein production (Carroll et al., 2020). For example, Bacillus toyonensis Bto-UNVM-94 produces parasporal crystals and has insecticidal activity, yet genomic analysis places it within the B. toyonensis lineage (Sauka et al., 2022). Further genome comparisons identified additional crystal-associated strains assigned to B. toyonensis (Sauka et al., 2026). Similarly, the mosquitocidal strain FCC41 has been described as Bacillus wiedmannii biovar thuringiensis on the basis of its genomic placement and phenotype (Lazarte et al., 2018). Pesticidal genes are frequently carried by plasmids or associated with mobile elements, although their location is not universally plasmid-borne. Their presence therefore does not by itself establish chromosomal species identity.
Taxonomic frameworks and databases do not always use the same species boundaries or names. In particular, proposed genomospecies systems and type-strain-based nomenclature may assign different labels to the same lineage. Reports should therefore state the reference database, version, similarity method and nomenclature used, rather than imply that a single naming scheme is universally accepted (Carroll et al., 2020). Biovar designations may communicate a phenotype or gene-based classification, but the criteria used for assignment should be specified.
For the Schmidt survey, a more precise description of the 22% result is that the samples yielded isolates identified as B. thuringiensis using the study’s crystal-based operational definition. The 150 crystal-forming isolates were not all sequenced, and their distribution among genome-defined lineages cannot be determined from the published analysis. One of the two sequenced non-crystal-forming isolates was assigned to B. toyonensis, but this does not demonstrate that the crystal-forming subset included that species. Conversely, the existence of crystal-forming B. toyonensis or B. wiedmannii elsewhere does not establish that either occurred among the survey isolates (Lazarte et al., 2018; Sauka et al., 2026; Sauka et al., 2022; Schmidt et al., 2027). Representative sequencing, coupled with comparisons against type-strain genomes and documented lineage assignments, would address this uncertainty.
From Enterotoxin Genes to Biological Activity
Members of B. cereus s.l. can carry genes encoding the diarrhoeal enterotoxins non-haemolytic enterotoxin (Nhe), haemolysin BL (Hbl) and cytotoxin K (CytK). Detection of relevant genes indicates a potential for production, but does not establish their expression, secretion, assembly into an active toxin complex or biological activity under a particular condition. Gene completeness, allelic variation, regulation and growth conditions may affect the phenotype. The cytK1 and cytK2 variants should be distinguished; neither should be conflated with pesticidal cyt genes encoding Cyt proteins (EFSA BIOHAZ Panel, 2016; EFSA BIOHAZ Panel, 2025).
Functional evidence is therefore essential. Johler et al. (2018) examined isolates from biopesticides, foods and outbreaks and observed variation in enterotoxin-associated cytotoxicity, supporting strain-level assessment. More recently, a study of an isolate obtained from a commercial product and investigated in a hydroponic system reported enterotoxin-associated genes and reduced Caco-2 cell viability after exposure to its cell-free supernatant (Ten et al., 2026). These results establish a functional response under the tested conditions, not the probability of illness after consumption of treated produce.
Other experimental results illustrate why a single assay cannot settle the issue. Alilat et al. (2026) examined supernatants of five biopesticide strains in an intestinal epithelial model and reported no cytotoxicity after 24 hours under their conditions, although inflammatory and other cellular responses, together with differences in toxin abundance, were observed. Negative results under a particular laboratory condition do not establish universal safety; positive responses likewise require interpretation in relation to dose, test conditions and physiological relevance. Where feasible, toxin-component measurements, assays of biological activity and relevant intestinal models should be combined, with appropriate controls and transparent reporting.
A foodborne diarrheal outcome also depends on events after ingestion, including survival and germination of spores and toxin production in the intestinal environment. Measuring only toxins already present on produce could therefore miss an important route of hazard expression. Conversely, an active culture supernatant prepared in a nutrient-rich medium may not reflect toxin levels produced during actual exposure. Pesticidal crystal proteins and mammalian enterotoxins should be considered separately: their biological targets and roles differ, and evidence about one class should not be substituted for evidence about the other.
Exposure, Outbreaks and Source Attribution
Food safety depends on both hazard and exposure. Relevant variables include viable cell and spore concentrations, the proportion attributable to the strain of interest, application and harvest practices, persistence during distribution, washing and storage, preparation, serving size and consumption patterns. Produce sold for raw consumption is important because it may not undergo a reliable microbial inactivation step. Retail detection establishes a potential route of exposure, while a quantitative assessment requires information closer to the point of consumption.
Epidemiological observations are also relevant. Mozhaitseva et al. (2026) investigated 294 B. cereus s.l. genomes associated with 183 foodborne outbreak investigations in France and reported a B. thuringiensis subsp. kurstaki population, particularly associated with vegetable-based salads. This work strengthens the rationale for considering B. thuringiensis-like strains in outbreak investigations. However, recovery of an isolate from an investigated food or event does not by itself establish that a specific commercial product caused illness. The attribution of individual cases depends on the epidemiological evidence, sampling, strain comparisons and exclusion of alternative explanations.
High-resolution sequencing can connect food, clinical and product isolates more effectively than broad phenotypic categories, but its interpretation requires representative reference genomes, quality control and awareness that closely related strains may share ancestry independently of a documented application. Epidemiological links and information on product use remain important. Retrospective analysis of isolates collected across the food chain, together with contemporaneous exposure records, could improve source attribution and estimates of the contribution of particular lineages to disease.
Interpreting the EFSA QPS Decision
In 2016, EFSA assessed risks associated with B. cereus and other species including B. thuringiensis in food, emphasizing shared hazards and limitations in identification and exposure data (EFSA BIOHAZ Panel, 2016). In 2025, the BIOHAZ Panel evaluated B. thuringiensis for QPS status and concluded that it was not suitable because of safety concerns (EFSA BIOHAZ Panel, 2025). In June 2026, the Panel confirmed that B. thuringiensis remained not recommended and concluded that it should be excluded from further QPS assessment (EFSA BIOHAZ Panel, 2026).
The conclusion concerns the suitability of a taxonomic unit for a general presumption of safety within the QPS process. It does not itself constitute a finding that every B. thuringiensis strain causes disease, a product-specific risk estimate or a blanket prohibition of B. thuringiensis biopesticides. Conversely, the absence of a product-specific prohibition does not resolve the scientific concerns that led to the QPS conclusion. QPS, evaluation of a plant protection active substance and authorization of particular products or uses are related but distinct processes. For example, EU rules for certain B. thuringiensis active substances include specific harvest-related conditions and data requirements (European Commission, 2023).
A key question is how the phenotype-based unit assessed under QPS relates to genome-defined lineages and individual active strains. A naming framework alone cannot settle a safety assessment. Nevertheless, reporting both the operational phenotype and genomic assignment would make it clearer which organisms are represented in evidence considered across regulatory contexts, and would facilitate comparison of findings across studies.
Toward an Integrated Assessment and Surveillance Framework
A practical framework would connect five layers. First, surveillance should document how isolates were recovered, counted and identified, including limits of detection and the distinction between sample-level and isolate-level prevalence. Crystal screening can provide an accessible initial indication, while its specificity and sensitivity should be evaluated against representative genomic and phenotypic references. A lack of visible crystals under one set of conditions should not automatically exclude pesticidal potential.
Second, representative isolates should be characterized genomically using transparent reference sets, type-strain comparisons and reproducible methods. Species or lineage assignment, pesticidal gene content, mobile elements and similarity to commercial strains should be reported as distinct results. Selection should capture commodity, season and phenotypic diversity; the uncertainty introduced by non-random selection should be stated.
Third, the presence and integrity of enterotoxin-associated genes should guide functional testing, rather than act as a stand-alone decision criterion. Measurements of toxin components and activity should be interpreted with strain-specific growth, sporulation and environmental conditions. Fourth, exposure assessments should quantify viable organisms along the route from application and harvest to consumption, including background environmental sources and uncertainties in strain-specific counts. Fifth, outbreak investigations should link food and clinical data with high-resolution strain comparison and information on application histories whenever available.
This framework would not assign a universal safe threshold or presume that individual test results are decisive. Its purpose is to make explicit what has been measured, where evidence remains uncertain and which next observation would most improve a particular assessment. It could also distinguish a taxonomic question from a source-tracing question: a food isolate may be close to a commercial strain yet require separate evidence to establish the route and timing of contamination.
Conclusions
Fresh-produce surveys, functional experiments and outbreak investigations collectively justify continued attention to B. thuringiensis and related B. cereus s.l. strains in the food chain. At the same time, crystal formation, genomic lineage, enterotoxin genes, toxin activity and human disease represent different layers of evidence. The distinction matters both for interpreting the 2026 EFSA QPS conclusion and for designing meaningful surveillance.
Going beyond the crystal means retaining useful operational methods while connecting them to genomic identity, strain-specific function, exposure and epidemiology. Such an approach can support proportionate food-safety decisions and more informative assessment of microbial pest-control products.
Author Contributions
Conceptualization, L.P., methodology, C.P. and L.P.; formal analysis, C.P. and L.P.; investigation, C.P. and L.P.; resources, L.P.; data curation, L.P.; writing–original draft preparation, L.P.; writing, review and editing, C.P. and L.P.; visualization, L.P.; supervision, L.P.; project administration, L.P.; funding acquisition, L.P. All authors have read and agreed to the final version of the manuscript.
Conflicts of interest
The authors declare there are no conflicts of interest.
Funding information
This work was supported by MCIN/AEI and the European Union through the Ramón y Cajal programme (RYC2023-043507-I). L.P. was supported by the Ramón y Cajal research contract (RYC2023-043507-I). C.P. was funded by the additional allocation of the Ramón y Cajal grant RYC2023-043507-I.
Data availability
Not applicable. .
Acknowledgments
Leopoldo Palma gratefully acknowledges the Spanish Ministry of Science, Innovation, and Universities, the Spanish State Research Agency, and the European Union for funding his Ramón y Cajal contract (grant ref. RYC2023-043507-I).
Declaration of generative AI use
The authors used ChatGPT (OpenAI) to assist with the wording of the manuscript and to improve its English language and readability. The scientific ideas, interpretation and conclusions were developed by the authors, who reviewed and edited the text and take full responsibility for the final manuscript.
References
- Alilat, S., et al., 2026. Multi-omics characterisation of the virulence and inflammatory potential of Bacillus thuringiensis in an in vitro intestinal epithelial model. Science of The Total Environment. 1029, 181742.
- Biggel, M., et al., 2022. Whole Genome Sequencing Reveals Biopesticidal Origin of Bacillus thuringiensis in Foods. Frontiers in Microbiology. Volume 12–2021.
- Carroll, L. M., et al., 2020. Proposal of a Taxonomic Nomenclature for the Bacillus cereus Group Which Reconciles Genomic Definitions of Bacterial Species with Clinical and Industrial Phenotypes. mBio. 11, 1–15.
- EFSA BIOHAZ Panel, 2016. Risks for public health related to the presence of Bacillus cereus and other Bacillus spp. including Bacillus thuringiensis in foodstuffs. EFSA Journal. 14, e04524.
- EFSA BIOHAZ Panel, 2025. Update of the list of qualified presumption of safety (QPS) recommended microbiological agents intentionally added to food or feed as notified to EFSA 21: Suitability of taxonomic units notified to EFSA until September 2024. EFSA Journal. 23.
- EFSA BIOHAZ Panel, 2026. Update of the list of qualified presumption of safety (QPS) recommended microbiological agents intentionally added to food or feed as notified to EFSA 24: Suitability of taxonomic units notified to EFSA until March 2026. EFSA Journal. 24, e10155.
- European Commission, Commission Implementing Regulation (EU) 2023/1002 of 23 May 2023 renewing the approval of the active substance Bacillus thuringiensis subsp. aizawai strain ABTS-1857 in accordance with Regulation (EC) No 1107/2009 of the European Parliament and of the Council, and amending Commission Implementing Regulation (EU) No 540/2011. Official Journal of the European Union, Vol. L 136. European Union, 2023, pp. 28–34.
- Felten, A., et al., 2025. Occurrence and identification of Bacillus thuringiensis in fresh vegetables and fruits in France. International Journal of Food Microbiology. 111477.
- Frentzel, H., et al., 2020. Indications of biopesticidal Bacillus thuringiensis strains in bell pepper and tomato. International Journal of Food Microbiology. 321, 108542.
- Frentzel, H., et al., 2026. Levels of biopesticide Bacillus thuringiensis strains on tomatoes, leaf lettuce and head lettuce from retailers and producers in Germany. Applied Food Research. 6, 101666.
- Hendriksen, N. B., Hansen, B. M., 2006. Detection of Bacillus thuringiensis kurstaki HD1 on cabbage for human consumption. FEMS Microbiology Letters. 257, 106–111.
- Johler, S., et al., 2018. Enterotoxin Production of Bacillus thuringiensis Isolates From Biopesticides, Foods, and Outbreaks. Front Microbiol. 9, 1915.
- Lazarte, J. N., et al., 2018. Bacillus wiedmannii biovar thuringiensis: A Specialized Mosquitocidal Pathogen with Plasmids from Diverse Origins. Genome Biol Evol. 10, 2823–2833.
- Logan, N. A., Vos, P. D., Bacillus. Bergey’s Manual of Systematics of Archaea and Bacteria, 2015, pp. 1–163.
- Mozhaitseva, K., et al., 2026. Population structure of Bacillus cereus sensu lato associated with foodborne outbreaks in France between 2004 and 2023. Food microbiology. 133, 104882.
- Sauka, D. H., et al., 2026. Bacillus toyonensis biovar Thuringiensis: an overlooked entomopathogen? Journal of Bacteriology. 208, e00039–26.
- Sauka, D. H., et al., 2022. Bacillus toyonensis biovar Thuringiensis: A novel entomopathogen with insecticidal activity against lepidopteran and coleopteran pests. Biol. Control. 167:104838.
- Schmidt, C., et al., 2027. Biopesticide-associated Bacillus thuringiensis on retail fresh produce: occurrence, typing and implications for food-safety surveillance. Food Microbiology. 143, 105358.
- Ten, M. M. Z., et al., 2026. Insights into the food safety implications of a commercial Bacillus thuringiensis product in hydroponic systems. Applied and Environmental Microbiology. 92, e00885–26.
- Zhao, X., et al., 2022. Identification and characterization of Bacillus thuringiensis and other Bacillus cereus group isolates from spinach by whole genome sequencing. Front Microbiol. 13, 1030921.
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content. |
© 2026 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
Copyright: This open access article is published under a Creative Commons CC BY 4.0 license, which permit the free download, distribution, and reuse, provided that the author and preprint are cited in any reuse.