TOXiTAXi is a publicly available database and web application dedicated to gather and analyze the existing and future bioassay results of experiments investigating insecticidal activity and combinatorial effects among Bacillus thuringiensis toxins.
This user-friendly tool provides effortless sorting and analysis of the dataset and it is designed to be used by a bench scientist on an everyday basis. TOXiTAXi is dedicated both to academic communities and implementation units, which develop new bioinsecticides and Integrated Pest Management strategies.

The TOXiTAXi relies on B. thuringiensis toxin nomenclature introduced in 1998 (Crickmore et al., 1998), updated in 2020 (Crickmore et al., 2020) and maintained in pesticidal protein database within Bacterial Pesticidal Protein Resource Center (BPPRC; https://www.bpprc.org/). Currently, TOXiTAXi incorporates data present in manuscripts which contain experiments:

• Investigating interactions among Bacillus thuringiensis Cry, Cyt and Vip toxins
• Performed using B. thuringiensis Cry, Cyt or Vip toxins obtained through expression of cloned genes in hosts not producing other pesticidal toxins (typically E. coli or acrystalliferous B. thuringiensis strains)
• Provide proper quantification of used toxins.

From a selected publication, the available results are incorporated into the database, including activity of toxin compositions as well as separate Cry/Cyt/Vip toxins. Any individual toxin(s)-organism-result association available in a manuscript is considered as a separate experiment. Each experiment recorded in the database includes information regarding:

• Source publication (authorship, publication date, PubMed ID)
• Tested toxic agent/composition (exact toxin name(s) according to current nomenclature, toxin source, expression host, molecular modifications applied, details on toxin preparation)
• Target organism (species name; strain/laboratory line, developmental stage, recognized resistance to toxins)
• Bioassay (toxin administration method, bioassay duration, toxicity measure, observed toxicity with units and confidence intervals)
• Interaction among toxins (expected toxicity of a mixture, interaction type, model used for interaction type estimation, synergism factor)

Experimental records were omitted when toxicity bioassays were performed using:

• Wild type B. thuringiensis strains (usually containing sets of not precisely characterized and/or non-quantified toxins)
• Binary toxins such as Tpp (formerly Bin), Gpp34/Tpp35 (formerly Cry34/Cry35), Vpb1/Vpb2 (formerly Vip1/Vip2) etc.
• Lysinibacillus sphaericus toxins (e.g., Mtx, Mpp)
• B. thuringiensis toxins obtained from transgenic crops – regardless if the above were tested with or without accepted Cry/Cyt/Vip toxins.

Moreover, if for some reason the exact values could not be determined for bioassay parameters (e.g. results presented solely in a chart form) the data was not included. The published data was incorporated into the database without alterations with the following exceptions:

• If synergism factor was not provided by the authors, it was calculated when possible
• Antagonism factors were recalculated to synergism factors
• Toxicity measure units were recalculated to nanograms (e.g., µg/cm² to ng/cm², µg/ml to ng/ml, etc.)

It should be noted that in the current nomenclature, the quaternary rank groups identical or almost identical (a few amino acid substitutions) B. thuringiensis proteins and the differences are considered biologically insignificant (Crickmore et al. 2020; van Frankenhuyzen 2009; Crickmore et al. 1998). Therefore, in TOXiTAXi B. thuringiensis toxins are considered unalike, when they differ in tertiary or higher rank.

TOXiTAXi was designed to be easily expandable and its flexibility enables processing of heterogeneous data regarding toxicity of various biocidal compounds with well-established or just potential use in pest control. In order to test the broad-use potential of the created database, a selection of 16 arbitrarily chosen manuscripts were additionally added to the dataset. These works investigate biocidal activity of agents other than B. thuringiensis Cry/Cyt/Vip proteins, including: chitinolytic enzymes, synthetic and biochemical insecticides, insect midgut proteins, B. thuringiensis Sip proteins, scorpion toxins and fungal spores. For the sake of convenience, in TOXiTAXi, the above-mentioned molecules and pathogens are termed/labeled as “toxins” although some of them do not meet the definition of this term.

REFERENCES:
• Crickmore N, Zeigler DR, Feitelson J, et al (1998) Revision of the nomenclature for the Bacillus thuringiensis pesticidal crystal proteins. Microbiol Mol Biol Rev 62:807–813. doi: 1092-2172/98
• Crickmore, N, Berry, C, Panneerselvam, S, et al (2020) A structure-based nomenclature for Bacillus thuringiensis and other bacteria-derived pesticidal proteins. J Invertebr Pathol (in press), doi: 10.1016/j.jip.2020.107438
• Frankenhuyzen K van (2009) Insecticidal activity of Bacillus thuringiensis crystal proteins. J Invertebr Pathol 101:1–16. doi: 10.1016/j.jip.2009.02.009

The TOXiTAXi database interface is designed to be used by a bench scientist on an everyday basis. Following a simplicity rule, the interface of TOXiTAXi has been built on only one result window and simple two types of querying systems, despite the vast amount and variability of data contained in the database.

The data selection process is provided in the form of a hierarchical expandable menu to supply the functionality for selecting any combination of toxins or target species. In this way, users can select toxins either by marking individual proteins (e.g., Cry1Ab3 and/or Vip3Aa20) or groups of proteins (e.g., all Cry and/or Vip proteins) from the tree-like menu reflecting the proteins hierarchical nomenclature or by typing protein names in the input text box. Analogously, users can select individual target species of interest (e.g., Helicoverpa armigera) or different combinations of taxonomic groups (e.g., all Lepidoptera and Coleoptera species). The real potential of the TOXiTAXi service is based on its ability to combine different datasets into one display - there is no limit as to the number and type of selected toxins and target species that can be combined in a single query.

The default result window provides basic information on experiments found for a given toxin(s) or target species specified during data selection process. This information includes toxin quantity, observed toxicity including biocidal activity measure (e.g., LC₅₀, LC₉₀, and mortality rate) and its unit (e.g., ng/cm²), combinatorial effect among toxins in composition (i.e., additivity, synergism, and antagonism), and reference publication. User can generate customizable and integrated results by adding additional information concerning experiments such as:

• Developmental stage of target species (e.g., larval instar)
• Recognized resistance in target species (e.g., resistance to Cry1Ac protein)
• Bioassay duration (e.g., 7 days)
• Toxin administration method (e.g., surface contamination)
• Expected toxicity (i.e., theoretical toxicity value of toxin combination assuming lack of synergism and antagonism)
• Confidence intervals of the observed toxicity
• Synergism factor (SF; ratio of expected to observed toxicity)
• Model used for interaction type estimation

The order of the result category columns can be manually adjusted by the user, and eight filters enable further selection of desired information. Such customized tables can be further downloaded from the web page in common tabular formats (i.e. Excel, CSV, and PDF files). Also, by clicking on reference publication, the user is forwarded to related publication in PubMed repository or other online source.

For even more detailed information on each separate experiment, the user can view record window by clicking on toxin(s) name in the result table. The window is divided into three distinct sections. First section is dedicated to the individual components of a given toxin or combination of toxins (e.g., Cry1Ac1 + Cry9Aa) and apart from data available in result table it also provides details such as toxin modification, toxin preparation and source (i.e., expression host). The second section concerns target species and gives additional notification i.e., on particular strain of tested organism. The third section shows detailed information on the experiment results such as toxicity value with confidence intervals or expected toxicity of toxin combination.

Results of toxin combinatorial effects are presented in the TOXiTAXi database result window as they appear in the original works. Therefore, the synergism/addition/antagonism mark is based on model assumptions adopted by authors of each work. It should be noted however, that the theoretical basis and model assumptions used for estimation of toxin interactions (not limited to insecticidal toxins) are subject of academic debates since the early XXth century (Roel et al., 2017; March and de March, 1987; Ashford, 1981; Bliss, 1939; Loewe and Muischnek, 1926). One of the discussed aspects is the SF value threshold, below which synergism should not be assumed. For example, recent works (Walters et al., 2018; Rodea-Palomares et al., 2015; Cedergreen, 2014) suggest that SF value should be at least 2 to “consider a result as being more than additive”. The authors state also that more than 10-fold increase in activity due to synergism is observed in bioassays very seldom. To acknowledge this important issue and to improve information display for the user, a categorization has been added to the website that indicates the magnitude of toxin interactions (both synergistic and antagonistic). For this purpose, a SF scale is used, where the “strength” of synergistic and antagonistic interactions is classified into three categories: "weak/doubtful", "moderate", and "strong".

Interaction type	Weak/Doubtful	Moderate	Strong
Synergism	SF ∈ [1, 2)*	SF ∈ [2, 10)	SF ∈ [10, ∞)**
Antagonism	SF ∈ [0.5, 1)	SF ∈ [0.1, 0.5)	SF ∈ [0, 0.1)

* SF = 2 is suggested as a threshold, below which toxin interactions should not be considered as synergistic.

** SF > 10 is considered high in magnitude and rather seldom reported.

Combinatorial effect magnitude labels are presented on the TOXiTAXi website in the “Interaction” column of the result window three. However, the proposed scale is fixed and thus can only provide a rough estimate of interaction magnitude. For this reason, a manifestation of SF values in the form of percentiles has also been implemented. Percentiles are dynamically generated on the website based on the distribution of all SF values (separately for each synergism and antagonism), currently deposited in the database. Percentiles as a measure of interaction magnitude inform the user about a relative standing of a given SF in comparison to all synergistic/antagonistic interactions present in the database. For example, the synergism factor reported for Cry1Ac + Cry1F (SF = 26.3) scores above the 90th percentile, which places the Cry1Ac + Cry1F interaction among the top 10% of synergistic interactions stored in the database. The percentiles are displayed in the "Synergism factor" column in the result window.

REFERENCES:
• Ashford JR (1981) General Models for the Joint Action of Mixtures of Drugs. Biometrics 37:457. doi: 10.2307/2530559
• Bliss CI (1939) The toxicity of poisons applied jointly. Ann Appl Biol 26:585–615. doi: 10.1111/j.1744-7348.1939.tb06990.x
• Cedergreen N (2014) Quantifying synergy: A systematic review of mixture toxicity studies within environmental toxicology. PLoS One 9:. doi: 10.1371/journal.pone.0096580
• de March BGE (1987) Simple similar action and independent joint action - two similar models for the joint effects of toxicants applied as mixtures. Aquat Toxicol 9:291–304. doi: 10.1016/0166-445X(87)90029-4
• Loewe S, Muischnek H (1926) Über Kombinationswirkungen - Mitteilung: Hilfsmittel der Fragestellung. Arch für Exp Pathol und Pharmakologie 114:313–326. doi: 10.1007/BF01952257
• Rodea-Palomares I, González-Pleiter M, Martín-Betancor K, et al (2015) Additivity and Interactions in Ecotoxicity of Pollutant Mixtures: Some Patterns, Conclusions, and Open Questions. Toxics 3:342–369. doi: 10.3390/toxics3040342
• Walters FS, Graser G, Burns A, Raybould A (2018) When the Whole is Not Greater than the Sum of the Parts: A Critical Review of Laboratory Bioassay Effects Testing for Insecticidal Protein Interactions. Environ Entomol 1–14. doi: 10.1093/ee/nvx207

• Toxin – name or short description of the toxic agent used in bioassay. In case of toxin combinations, separate constituents are separated by plus sign. In case of chimeric/fusion proteins, the names of fused molecules are separated by hyphen. In case of B. thuringiensis toxins nomenclature introduced by Crickmore et al. (1998) is used. For the sake of convenience, various agents (i.e. enzymes, insect midgut proteins) and entomopathogens are termed/labeled as “toxins”, although they do not meet the definition of this term.
• Target species – binominal species name according to current NCBI taxonomy
• Target developmental stage – developmental stage (larval instar, imago, etc.) of organism tested in bioassay
• Recognised resistance in target species – list of toxins or other factors to which the target organism resistance was recognised
• Bioassay duration (days) – bioassay data scoring time (days after treatment)
• Toxin quantity – amount of toxin(s) administered to tested organism in a bioassay, along with dose/concentration unit. In case of experiments calculating dose-dependent measure (LC₅₀, LC₉₀, EC₅₀, etc.) of toxin combination the used toxin ratio is given (i.e., 1:1 ratio, 3:1 ratio, etc.). In case of experiments calculating dose-dependent measure of a single toxin this field is empty.
• Toxin administration method – method of delivering the toxin to the tested organism
  Currently TOXiTAXi contains experiments using seven various toxin administration methods including:
  • Contact
  • Diet incorporation
  • Droplet feeding
  • Environment contamination
  • Force feeding
  • Leaf dip
  • Surface contamination
  To obtain details on method used in each experiment the reader is referred to relevant source manuscripts.
• Toxicity measure – measure used in a bioassay to assess toxin potency. Currently in TAXiTOXi the below mentioned measures are used:
  • EC₅₀ (50% effective concentration) - concentration of toxin(s) causing certain specified effect (i.e., abnormal development) in 50% of exposed population
  • ED₅₀ (50% effective dose) - dose of toxin(s) causing certain specified effect (i.e., abnormal development) in 50% of exposed population
  • IC₅₀ (50% inhibitory concentration) - concentration of toxin(s) causing growth inhibition in 50% of exposed population
  • ID₅₀ (50% inhibitory dose) - dose of toxin(s) causing growth inhibition in 50% of exposed population
  • LC₅₀ (50% lethal concentration) – concentration of toxin(s) lethal for 50% of exposed population
  • LC₉₀ (90% lethal concentration) – concentration of toxin(s) lethal for 90% of exposed population
  • LC₉₅ (95% lethal concentration) – concentration of toxin(s) lethal for 95% of exposed population
  • LC₉₉ (99% lethal concentration) – concentration of toxin(s) lethal for 99% of exposed population
  • LD₅₀ (50% lethal dose) – dose of toxin(s) lethal for 50% of exposed population
  • Mortality (mortality rate) – percentage of dead insects in relation to the entire exposed population
  • Weight reduction – percentage weight loss of exposed population in relation to control
  • Weight reduction 50% - the concentration of the toxin(s) causing 50% reduction in larval weight in comparison with control larvae
  • TI (toxicity improvement) - this measure has been introduced to express the influence of a toxic agent on the biocidal activity of heterogeneous toxin combinations i.e., the influence of chitinase enzyme on the insecticidal activity of B. thuringiensis parasporal crystals containing different, undetermined Cry/Cyt proteins. Expanding the mentioned example, TI can be calculated using equation: TI_chitinase = LC_{50 (chitinase + parasporal crystals)}/LC_{50(parasporal crystals)}. The higher is the TI value the higher is the potency of a toxic agent. TI measure can be successfully used to overcome difficulties in expressing the results of some bioassays in a simple manner, especially for database purposes. It should be noted however, that TI value is only relative and inaccurate value.
• Observed toxicity – calculated value of toxin(s) potency along with appropriate unit
• Confidence intervals – 95% fiducial limits calculated for observed toxicity. In some cases, standard error (SE) is provided in this field.
• Expected toxicity – expected, theoretical value of toxin combination potency calculated upon observed toxicities of constituents tested separately, assuming additive effect between the constituents
• Interaction – estimated combinatorial effect among two or more toxins. Possible options include: synergism, antagonism and additivity. When toxin composition was bioassayed but no interaction studies were performed “N/A” appears in the cell. In case of bioassays performed with individual toxins this cell remains empty.
• Synergism factor (SF) – indicator of synergism (or antagonism) magnitude. This value is calculated by dividing Expected toxicity by Observed toxicity. In experiments indicating synergism, higher SF value means higher magnitude of synergism. In experiments indicating antagonism, lower SF value means higher magnitude of antagonism.
• Interaction estimation model – model used to estimate expected, theoretical value of toxin combination activity, assuming additivity among constituents (lack of synergism/antagonism). Typically, three models are used:
  • Loewe Additivity (simple similar action) – referenced in a number of works (Ashford 1981; March and de March 1987; Roell et al. 2017) and as an example used by Tabashnik (1992)
  • Bliss Independence (similar joint action) – referenced in a number of works (Ashford 1981; March and de March 1987; Roell et al. 2017) and as an example used by Fernández-Luna et al. (2010)
  • Empirical – intuitive approach, which generally relies on comparison of effects exerted by composition and individual constituents, as an example used by Kaikaew et al. (2016)
• Publication – for each separate experiment, the first author and publication date of the related manuscript is given (in case of manuscripts authored by two researchers both names are provided). This field is linked with manuscript online source (preferentially PubMed repository)
• Target species strain – information regarding certain target organism strain/laboratory line
• Toxin source – the species was given from which the toxin originates, was isolated, or from which toxin-coding gene was cloned. Alternatively, “chemical synthesis” is stated when toxin has no source in nature. In case of synthetic pesticides and botanicals the information on commercial pesticide containing the active ingredient was given when possible.
• Toxin form – in case of B. thuringiensis toxins this column shows whether protoxin was used in experiment or toxin after activation step (typically using trypsin or insect gut juice)
• Toxin modification – details regarding substantial modifications made to a toxin comparing to its native state
• Expression host – in case of toxins produced by heterologous expression, a host species name is provided
• Toxin preparation – method of toxin preparation and form (toxin purity) administered to target species:
  • Culture supernatant – cell-free filtrate obtained after incubation
  • Cells – whole cells without cell lysis step
  • Cell lysate – crude cell lysate without any purification step
  • Cell lysate (partial purification) – host cells containing expressed toxin were subjected to cell lysis followed by partial purification step
  • Inclusion bodies – inclusion bodies isolated from cell lysate
  • Endospores and parasporal crystals – B. thuringiensis spore-crystal suspension
  • Parasporal crystals – protein crystals isolated from B. thuringiensis spore-crystal suspension
  • Purified proteins – toxin obtained with protocols including purification step (usually with a chromatographic method)

REFERENCES:
• Ashford JR (1981) General Models for the Joint Action of Mixtures of Drugs. Biometrics 37:457. https://doi.org/10.2307/2530559
• Crickmore N, Zeigler DR, Feitelson J, et al (1998) Revision of the nomenclature for the Bacillus thuringiensis pesticidal crystal proteins. Microbiol Mol Biol Rev 62:807–813. doi: 1092-2172/98
• Fernández-Luna MT, Tabashnik BE, Lanz-Mendoza H, et al (2010) Single concentration tests show synergism among Bacillus thuringiensis subsp. israelensis toxins against the malaria vector mosquito Anopheles albimanus. J Invertebr Pathol 104:231–233. doi: 10.1016/j.jip.2010.03.007
• Frankenhuyzen K van (2009) Insecticidal activity of Bacillus thuringiensis crystal proteins. J Invertebr Pathol 101:1–16. doi: 10.1016/j.jip.2009.02.009
• Kaikaew A, Promptmas C, Angsuthanasombat C (2016) Importance of Thr328 and Thr369 for functional maintenance of two receptor-binding β-hairpins of the Bacillus thuringiensis Cry4Ba toxin: Implications for synergistic interactions with Cyt2Aa2. Biochem Biophys Res Commun 469:698–703. doi: 10.1016/j.bbrc.2015.11.115
• March BGE de, de March BGE (1987) Simple similar action and independent joint action - two similar models for the joint effects of toxicants applied as mixtures. Aquat Toxicol 9:291–304. https://doi.org/10.1016/0166-445x(87)90029-4
• Roell KR, Reif DM, Motsinger-Reif AA (2017) An Introduction to Terminology and Methodology of Chemical Synergy—Perspectives from Across Disciplines. Front in Pharmacol 8.: https://doi.org/10.3389/fphar.2017.00158
• Tabashnik BE (1992) Evaluation of synergism among Bacillus thuringiensis toxins. Appl Environ Microbiol 58:3343–3346

To ensure database development and its usability we encourage the users to indicate additional publications that should be included into the dataset.
To maintain the reliability and accuracy of the database, we encourage the users to report any noticed errors concerning database content and application functionality.

TOXiTAXi has been created on a research platform established by:

Department of Microbiology
Institute of Experimental Biology
Adam Mickiewicz University
Uniwersytetu Poznańskiego 6
61-614 Poznań
Poland

and

Department of Computational Biology
Institute of Molecular Biology and Biotechnology
Adam Mickiewicz University
Uniwersytetu Poznańskiego 6
61-614 Poznań
Poland

Authors:
Jakub Baranek
Bartłomiej Pogodziński
Norbert Szipluk
Andrzej Zielezinski

If you make use of the data presented here, please cite the following article:

In matters regarding TOXiTAXi dataset, development, and error reports please contact Jakub Baranek: jakbar@amu.edu.pl
In matters regarding informatic part of the project please contact Andrzej Zielezinski: andrzejz@amu.edu.pl