The Safety of Soy Leghemoglobin Protein Preparation Derived from Pichia pastoris Expressing a Soy Leghemoglobin Gene from Glycine max: In Vitro and In Vivo Studies

Soy leghemoglobin (LegH) protein derived from soy (Glycine max) produced in Pichia pastoris (reclassified as Komagataella phaffii) as LegH Prep is a novel food ingredient that provides meat-like flavor and aroma to plant-derived food products. The safety of LegH Prep has been previously assessed in a battery of in vivo and in vitro testing and found no adverse effects under the conditions tested. In this new work, we present the results of new in vivo and in vitro tests evaluating the safety of LegH Prep. LegH Prep was nonmutagenic in a bacterial reverse mutation assay and nonclastogenic in an in vitro micronucleus assay in human lymphocytes. Systemic toxicity was evaluated in the 90 day dietary study in male and female Sprague–Dawley® rats that included a 28 day recovery period. The study resulted in no animal deaths associated with the administration of LegH Prep at the highest dose (90,000 ppm). There were no significant adverse clinical or physical changes attributed to LegH Prep administration, and no observed adverse effects on either male or female rats over the course of the 28 day recovery phase study. The new 90 day dietary toxicity study established a no observed adverse effect level (NOAEL) of 4798.3 and 5761.5 mg/kg/day, the maximum level tested for male and female rats, respectively. Thus, the results of the studies demonstrate that under the conditions tested, LegH Prep is not toxic for consumption in meat analog products.


Introduction
Animal agriculture is one of the principal contributors to climate change, which through continuous emissions of greenhouse gases places enormous stress on Earth's land, water, and energy resources [1,2].One of the ways to combat the damaging environmental efects of animal agriculture is to harness the power of food biotechnology to develop novel alternative plant-based proteins thereby reducing dependence on animal agriculture's environmental footprint.Te last decade has produced substantial advances in the feld of alternative meat protein engineering.One of these has been the introduction of the soy leghemoglobin (LegH) protein.Te characterization of LegH has been previously described [3,4].LegH protein is bioproduced by the yeast Pichia pastoris (currently reclassifed as Komagataella phaf) via industrial submerged fermentation.LegH Prep has a total protein fraction of at least 65% and is composed of LegH protein, P. pastoris yeast proteins, and food-grade stabilizers [4].New toxicology data support the original safety assessment of this food ingredient derived from a novel source.
Pichia pastoris is a nontoxigenic and nonpathogenic microbial fermentation yeast used in the biomanufacturing of various FDA-notifed GRAS substances [5][6][7][8], including LegH Prep derived from P. pastoris that obtained a "no questions" letter from FDA when a GRAS dossier was notifed (GRN#737) describing use at levels of up to 0.8% soy LegH protein in a low number of diferent types of food products [9], similar to previous notifed GRAS ingredients [8].In 2019, FDA approval was achieved for soy LegH as a color additive [10].LegH Prep derived from P. pastoris has been approved for use in food (meat analog products) in Australia-New Zealand [11], Singapore [12], and Canada [13], as well as meeting regulatory compliance in the United States.LegH Prep from P. pastoris has been used in the meat analog products produced by Impossible Foods since 2016 and has been sold and consumed internationally (over 500 million, ¼ pound (113 g) servings) without any signifcant reports of any safety issues concerning the consumption of soy LegH Prep.
Te safety of LegH Prep has been previously assessed [4] for limited use conditions, using in vivo, in vitro, and in silico testing [14,15].Te potential allergenicity and toxicity risk of LegH Prep was previously evaluated using bioinformatics, proteomics, and a pepsin digestion assay according to CODEX Alimentarius Commission [16] Guideline for the Conduct of Food Safety Assessment of Foods Produced Using Recombinant-DNA Microorganisms (CAC/Gl 46-2003 [15]).Te previous published results demonstrated that the seven residual yeast proteins in LegH Prep (≥1% of the total protein content) displayed no signifcant sequence matches to any known food allergens except the highly conserved wheat glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and similar alignment to homologous proteins from many common yeasts including Saccharomyces sp.[15].Results published by Fraser et al. [4] demonstrated via a pepsin digestion assay that LegH and P. pastoris proteins were swiftly digested, thereby indicating that under the conditions of this assay, LegH Prep is not likely to pose an allergenicity risk.Te authors concluded that there was no risk of cross-reactivity between LegH Prep and GAPDH.However, the published studies did not evaluate long-term administration of the LegH Prep in a preclinical study.
Te safety of LegH Prep produced by P. pastoris (K.phafi) for use in meat analog products at the maximum recommended application rates is supported by previously published toxicological data assessed in Sprague-Dawley ® rats by a 28 day dietary study with an additional 28 day feeding study that evaluated female reproductive health and estrous cycle [4].Te mutagenic potential of LegH Prep was evaluated by a bacterial reverse mutation assay and potential genotoxicity in an in vitro chromosomal aberration assay.Overall, results from these in vitro and in vivo studies confrmed no issues of toxicological concern regarding LegH Prep under the conditions tested [4].
To expand the toxicological body of evidence attesting to the safety of LegH Prep and strengthen the assessment of safety under the intended conditions of long-term ingestion, Impossible Foods conducted new in vivo and in vitro studies to evaluate LegH Prep's potential for general and genetic toxicity.Te in vitro models consist of a bacterial reverse mutation assay and an in vitro micronucleus assay in human peripheral blood lymphocytes (HPBLs).Systemic toxicity was also evaluated via a 90 day dietary feeding study in Sprague-Dawley ® rats that included a 28 day recovery pe- riod.Overall, the results of the studies in this new research underscore the lack of toxicological concern for LegH Prep under the conditions tested.

LegH Prep Production and Analysis.
Te soy plant (Glycine max) was the source of the LegH protein sequence which was inserted and expressed into a P. pastoris strain via submerged fed-batch fermentation and obtained using fltration-based methods and food and/or pharmaceuticalgrade reagents.Te P. pastoris production strains were derived from a nontoxigenic and nonpathogenic, safe strain lineage that has a history of safe use in the production of proteins for use in food and other applications [17,18].Te LegH Prep used in the in vivo and in vitro studies (batches LH20-150-160-190FG-301 and PP-PGM2-20-061-302) met internal specifcations.Te strains were derived from the production strain MXY0291 [4,6] and the well-known Pichia strain NRRL Y-11430 [4,19].Both strains were engineered to overexpress the LegH gene as well as eight native enzymes in the Pichia heme biosynthesis pathway (aminolevulinic acid (ALA) synthase, ALA dehydratase, coproporphyrinogen oxidase, ferrochelatase, porphobilinogen deaminase, protoporphyrinogen oxidase, UPG III synthase, and uroporphyrinogen (UPG) III decarboxylase) [4].Postfermentation, the cells were lysed to release the LegH protein.All insoluble materials were removed and the resulting concentrated liquid (LegH Prep) was formulated with food-grade stabilizers and frozen.To ensure that the test article was well incorporated in the animal diet, the LegH Prep was freeze-dried before use.All in vivo testing was performed at Product Safety Labs (PSLs), Dayton, New Jersey, USA.PSL has been accredited and certifed by the Association for Assessment and Accreditation of Laboratory Animal Care (AAALAC).During the 90 day study and 28 day recovery period, the LegH protein concentrations in the neat test article and animal feed samples were analyzed by Impossible Foods using ultra-high-performance liquid chromatography (UHPLC) to assess LegH Prep concentration, homogeneity, and stability.UHPLC was performed using a Waters ACQUITY UHPLC system with an ACQ-UITY BEH SEC 4.6 × 150 mm column (Waters, Milford, MA).LegH protein concentration was quantifed via integration of the 405 nm absorbance at the LegH retention time.
Five bacterial strains (Salmonella typhimurium (ST) TA98 and TA1535 (Moltox, Inc., USA); TA100 and TA1537 (Xenometrix AG, Switzerland), and Escherichia coli (EC) WP2 uvrA (Moltox, Inc., USA)) were used in the plate incorporation and preincubation methods, in the presence and absence of a metabolic activation system (S9 mix, Trinova Biochem GmbH, Gießen, Germany).Sterile water was the negative control and vehicle (except where noted), while sodium azide (NaN 3 ; Sigma), 4-nitro-ophenylenediamine (4-NOPD; Sigma; dissolved in DMSO), methyl methanesulfonate (MMS; MilliporeSigma, USA), and 2-aminoanthracene (2-AA; Alfa Aesar, USA; dissolved in DMSO) were used as positive controls.LegH Prep cytotoxicity and the potential to induce mutations were assessed with tester strains TA98 and TA100 in a preexperiment.Eight concentrations of LegH Prep were tested in triplicate under experimental conditions which were the same as those described below for the main experiment I (plate incorporation test).Te LegH Prep concentrations utilized in the main experiments were chosen based on the results from the pre-experiment (S1, Supplementary Materials).As the pre-experiment results were in accordance with the criteria of validity (Supplementary Info Table S2), these results were reported as a part of the main experiment I.
Te main experiment included an initial test that followed the plate incorporation method using the following materials which were combined and plated using minimal agar: 100 μL of the prepared test solutions, negative (vehicle) control, or prepared positive control substance; 500 μL S9 mix or substitution bufer; 100 μL bacteria suspension (ST or EC); and 2000 μL overlay agar [4].After solidifcation, the plates were inverted and incubated at 37 °C for 48 hours in the dark.Te confrmatory test employed the plate incorporation method allowing for spacing between dose concentrations.Following incubation, revertant colonies were counted using a ProtoCOL counter (Meintrup DWS Labor Laborgeräte GmbH).Te revertant colonies were counted manually if the precipitation of the test article precluded automatic counting.Low spontaneous mutation frequency tester strains TA1535 and TA1537 were counted by hand.Cell toxicity was identifed by clearing or depletion of the background lawn or a reduction in the number of revertants down to a mutation factor of approximately ≤0.5 in relation to the solvent control.For the study to be considered valid, the bacteria (1) must have demonstrated typical responses to ampicillin, (2) the negative control plates (distilled water) with and without the S9 mix must be within the laboratory historical control ranges, (3) the corresponding background growth on both negative control and test plates must be visible, and (4) at least fve diferent concentrations of each tester strain must be analyzable.For each experimental point, the mutation factor (MF) was calculated by dividing the mean value of the revertant counts by the mean values of the solvent control.LegH Prep would be considered mutagenic if a biologically signifcant positive response for at least one of the dose groups is observed in at least one tester strain with or without metabolic activation or if there is a clear and dose-related increase in the number of revertants.
A biologically signifcant positive response was scored if tester strains (TA98, TA100, and E. coli WP2 uvrA) resulted in twice as high the number of reversions and if tester strains TA1535 and TA1537 resulted in at least three times the number of reversions as compared to the control [24].
According to OECD guidelines, the biological relevance of the results serves as the criterion for the interpretation of results, and a statistical evaluation of the results is not regarded as necessary.

In Vitro Mammalian Micronucleus Assay in Human
Peripheral Blood Lymphocytes (HPBLs).Te in vitro mammalian micronucleus assay was performed at Eurofns (Munich, Germany) in compliance with the German GLP regulations and under the appropriate OECD [25] and European Commission [26] guidelines.Te study employed human peripheral blood lymphocytes (HPBLs) in both the absence and the presence of chemically induced rat liver S9 metabolic activation system (either prepared at Eurofns Munich or obtained from Trinova Biochem, Giessen, Germany).Blood was collected from a single donor with no known recent exposure to genotoxic chemicals, or radiation to reduce interindividual variability and samples was stored in heparinized tubes at 4 °C for a maximum of 4 hours.Whole blood samples treated with heparin were precultured in the presence of mitogen (phytohemagglutinin, PHA).HPBLs were cultured in complete medium (RPMI 1640 containing 15% heat-inactivated fetal bovine serum (FBS), 2.4 µg/mL of phytohemagglutinin, and 100 units of penicillin/streptomycin solution).A pre-experiment was conducted under identical conditions described for the main experiment I (4 hour incubation) to determine the cytotoxicity of the LegH Prep using the cytokinesis-block proliferation index (CBPI).Te following concentrations were tested with or without S9 mix: 10, 20, 39, 78, 156, 312.5, 625, 1250, 2500, and 5000 µg/mL.Te LegH Prep was suspended and diluted in cell culture medium (RPMI) within 1 hour prior to treatment.After ultrasonication for 30 minutes at room temperature, a stable suspension was obtained.Te pH value was within a physiological range in the test item.All positive control substances used were from Sigma unless specifed.Methyl methanesulfonate (MMS, 50 and 65 µg/ mL-without metabolic activation) and cyclophosphamide (CPA, 15 µg/mL15 µg/mL-with metabolic activation) were used as clastogenic controls and colchicine (0.04 and 0.4 µg/ mL) (without metabolic activation) was used as an aneugenic control.

Experiment I (Metabolic Activation)
. Whole blood samples were precultured (44 to 48 hours) in the presence of PHA prior to LegH Prep dosage.LegH Prep was added to the lymphocytes then incubated for 4 hours in the presence or absence of metabolic (S9) activation.Te treatment medium (complete culture medium without FBS) was removed at the end of the incubation period; the cells were then washed and the cultures were incubated in complete culture medium-+ 6 µg/mL cytochalasin B for 40-42 hours at 37 °C [27].

Experiment II (No Metabolic Activation).
Whole blood cultures were precultured in the presence of PHA for 44 to 48 hours prior to exposure to LegH Prep and were added in a complete culture medium.An hour later, 6 µg/mL cytochalasin B was added and the cells were incubated for a further 43 hours at 37 °C.Te culture medium was removed at the end of the treatment period, and the cells were prepared for microscopic analysis.Duplicate cultures were analyzed at each dose level except for the pre-experiment.Table 1 outlines the study design.

Culture Preparation.
At the end of incubation, the complete culture medium was removed, and the cells treated with a cold hypotonic solution (0.075 M potassium chloride) at room temperature then centrifuged.Te pellet was resuspended with a fxation solution and centrifuged.Te collected cells were fxed with a methanol (3 parts) + glacial acetic acid (1 part) solution, resuspended, and loaded onto clean glass slides and dried and stained with acridine orange solution.

Analysis of Micronuclei.
For each dose group, at least 2000 binucleated cells (if possible) per concentration (1000 binucleated cells per slide) were analyzed for micronuclei [28].Mononucleated and multinucleated cells and cells with more than six micronuclei were not considered [29].

Cytokinesis-Block Proliferation Index (CBPI).
To properly evaluate cytotoxicity, a cytokinesis-block proliferation index (CBPI) was determined from 500 cells according to the following formula: where c 1 is mononucleate cells, c 2 is binucleate cells, c x is multinucleate cells, and n is the total number of cells.
Te CBPI was used to calculate the % cytostasis, which indicates the inhibition of cell growth of treated cultures in comparison to control cultures: where CBPI T is the cytokinesis-block proliferation index of treated cultures and CBPI C is the cytokinesis-block proliferation index of control cultures.

Statistical Analysis.
Signifcance was decided at a probability value of p < 0.05.Te nonparametric χ 2 test was performed to analyze the results in both experiments.

14 Day Dietary Toxicity/Palatability Study in Rats.
Te 14 day toxicity/palatability study followed OECD Guidelines 407 [30] and was compliant with US FDA guidelines [31].® rats with a 28 day recovery period was conducted at PSL according to GLP and OECD guidelines [32] and the U.S. FDA guidelines [31].Te protocol for this in vivo study was preapproved (P703) by the IACUC of the laboratory (Dayton, NJ), and the laboratory has been accredited by the AAALAC organization.Te animals were quarantined/acclimated at PSL (as described in Section 2.4) for fve days before the study starts.All animals were individually housed and provided a form of object (Nylabone ® or IChew) enrichment.Animal body weight and clinical observation data were recorded at least twice before the study starts.Typical of protocols for studies completed for regulatory evaluations, including the European Chemicals Agency (ECHA) [33], four groups of adult CRL Sprague-Dawley CD ® IGS rats (10/sex/group) were maintained on diets prepared to target daily intakes of 1875, 3750, and 5625 mg/kg/day LegH Prep for Groups 2-4, respectively (Supplementary Table S4), with fve additional animals from Groups 1 and 4 remained on the study for an additional 28 day recovery period (Supplementary Figure S1).Rats in the control group were provided diet that was consistent with diets with the other groups but did not contain LegH Prep.Diet information is provided as a supplement to this manuscript (Supplementary Table S5).Te neat LegH Prep was monitored for stability throughout the study and was found to be stable.Homogeneity and dietary stability analyses showed that LegH Prep was homogeneously distributed and stable in the dietary matrix during a 4 day preparation interval (data not shown).Te dietary concentrations to provide 3750 and 5625 mg/kg bw/day were considered to have met the target concentrations.All animals received an ophthalmological evaluation by focal illumination and slit lamp biomicroscopy, prior to study initiation, and for all animals on Day 87.All animals were observed once a day for any sign of toxicity, survivability, and behavior and weekly for detailed clinical observations.Body weights were recorded twice during acclimation, including prior to test initiation on Day 0, and weekly thereafter until Day 91 (main test) and on Day 119 (recovery).Food consumption measurements were taken to coincide with body weight measurements.Food efciency and dietary intake were calculated.Clinical pathology and a thyroid hormone assessment were performed on both the main test and the recovery phase animals.Urine samples (utilizing metabolism cages) and blood samples (via sublingual bleeding under isofurane anesthesia) were collected on Days 92/120 for males and Days 93/120 for females for the main test and recovery animals, respectively.Blood was collected (∼500 μL) in a precalibrated tube containing K 2 EDTA for hematological tests.Te whole blood samples were centrifuged (refrigerated) and approximately 1 mL of serum was collected into a preservative-free tube for serum chemistry tests utilizing the COBAS C311 automated analyzer.Hematological analysis completed on an ADVIA 120 Hematology System included RBC, HCT, MCV, MCH, Absolute ARET, WBC, diferential leukocyte count, MCHC, hemoglobin (HGB), MCV, RDW, and PLT.Coagulation analysis on a Siemens Sysmex CA620 system included prothrombin time (PT) and activated partial thromboplastin time (APTT).Clinical chemistry determined on a COBAS C311 analyzer included AST, SDH, BILI total, CREA, TRIG, total serum protein, globulin (GLOB), inorganic phosphorus, K, ALT, ALKP, BUN, total CHOL, GLUC, ALB, CALC, NA, and CL.Urinalysis included quality, color (COL), clarity, urine volume (UVOL), microscopic urine sediment examination, pH, GLUC, specifc gravity, protein (UMTP), ketone, bilirubin, blood, and urobilinogen.Study animals were euthanized under isofurane anesthesia, and blood was collected for evaluation of coagulation parameters.Vaginal smears were collected from all female rats on the day of terminal sacrifce (day 93 for the main test animals and day 120 for the recovery phase animals) to determine the stage of estrus.Gross necropsies were performed on all animals, and histological evaluation of selected organs and tissues was performed on Groups 1 and 4. All clinical pathology sample analyses were performed at PSL (Dayton, NJ).Study animals underwent a gross necropsy, which entailed a detailed assessment of the animal's physical appearance, body orifces, the musculoskeletal system, and all organs associated with the cranial, thoracic, abdominal, and pelvic cavities.Tissues and organs were collected and preserved in 10% neutralbufered formalin except for the eyes, testes, and epididymis, which were preserved in Davidson's fxative before utilizing a gradient transfer process, with fnal storage in absolute ethanol prior to shipment of tissues to the pathology lab for histology processing.A subset of tissues/organs was weighed wet immediately after dissection to avoid desiccation including adrenal glands, kidneys, spleen, brain, liver, thymus, testes, epididymis, ovaries with oviducts, uterus, and heart.All preserved animal tissues were sent to StageBio (Mount Journal of Toxicology Jackson, VA) for further processing and analysis by a boardcertifed veterinary pathologist.Tissues from all the main study animals in the control and high dose group, the female reproductive organs (from all dose levels), as well as all gross lesions from all animals were processed, embedded in parafn, sectioned, and stained with hematoxylin and eosin (H&E).For the 14 day study, a two-way analysis of variance (ANOVA) was used to compare all in-life endpoints in both treatment and control groups that were classifed as having multiple measurements of continuous data over time (e.g., body weight parameters, food consumption, and food efcacy), thereby testing the efects of both time and treatment, with methods accounting for repeated measures in one independent variable (time) [34].Groups where variance is found to be signifcantly diferent were compared using a nonparametric method such as the Kruskal-Wallis nonparametric analysis of variance.If a nonparametric ANOVA was signifcant, a comparison of treated groups to control was performed (e.g., Dunn's test).If warranted by sufcient group sizes, the incidence of clinical observations may be evaluated through sequential application of a trend test [35].

Statistical Analysis
For the 90 day study, the following parameters were calculated and analyzed by the Bartlett test for homogeneity of variances and normality [36]: body weights, food consumption, UVOL, hematology, blood chemistry, absolute and relative organ weights, averages, and standard deviations.One-way analysis of variance (ANOVA) was used to compare treated and control groups and the Bartlett test indicated homogenous variances.When ANOVA was signifcant, a comparison of the treated groups to control by the Dunn test for multiple comparisons was performed [37,38].Where variances were considered signifcantly diferent by the Bartlett test, groups were compared using a nonparametric method (Kruskal-Wallis nonparametric ANOVA) [39].When nonparametric ANOVA was significant, a comparison of treated groups to control was performed using the Dunnett test [40].Clinical pathology was preliminarily tested via the Levene test [41] for homogeneity and via the Shapiro-Wilk test [42] for normalcy followed by ANOVA and the Dunnett test [37,38].

Bacterial Reverse Mutation Assay.
Te pre-experiment analysis found no limiting toxicity nor limiting precipitation of the test item was observed in either tester strain used at the maximum recommended concentration of 5000 µg/plate (with and without metabolic activation; Supplementary Table S1).Terefore, concentrations of 31.6 to 5000 µg/plate were selected for the main experiments.Data results for Experiments I and II are shown in Tables 2-5.
In Experiment I, LegH Prep precipitation was observed in tester strains TA98 and TA100 at ≥1000 µg/plate (with and without metabolic activation), and in tester strains TA1535, TA1537, and E. coli WP2 uvrA at ≥316 µg/plate (with and without metabolic activation) (Table 2).In Experiment II, precipitation was observed in all tester strains at ≥316 µg/plate (with and without metabolic activation) (Table 4).Te observed precipitation did not interfere with the scoring; thus, it did not impact the results.In both experiments the mutation factors were within typical ranges (Tables 3 and 5).Four plates in Experiment I (TA100, 2500 µg/plate; TA1535, 31.6 µg/plate; TA1537, 5000 µg/plate; E. coli WP2 uvrA, 5000 µg/plate; without metabolic activation) exhibited microbial contamination but did not afect the quality, integrity, or evaluation of the results as the microbial contamination could be clearly distinguished from the tester strain revertants.In Experiment I, cytotoxic efects of LegH Prep were observed in tester strain TA1535 at 5000 µg/plate (without metabolic activation) (Table 2).In Experiment II, cytotoxic efects of the test item were noted in tester strain TA1537 at ≥2500 µg/plate (without metabolic activation) (Table 4).No further cytotoxic efects of the test item were noted in Experiments I or II.No biologically relevant increases in revertant colony numbers of any of the fve tester strains were observed following treatment with LegH Prep at any concentration level neither in the presence nor in the absence of metabolic activation in Experiments I and II.

In Vitro Mammalian Micronucleus Assay in Human
Lymphocytes.Te potential of LegH Prep to induce micronuclei in human peripheral blood lymphocytes (HPBLs) in the absence and presence of metabolic activation with S9 was evaluated.Te concentrations used in the main experiments (I and II) were based on the pre-experiment (Supplementary Table S6), with precipitation of LegH Prep observed in the pre-experiment at ≥312.5 µg/mL with and without metabolic activation.LegH Prep was analyzed at 650 and 250 µg/mL with and without metabolic activation, respectively, in Experiment I.In Experiment II, 750 µg/mL was selected as the highest concentration (with and without metabolic (S9) activation) for microscopic analysis of micronuclei.Te concentrations evaluated for micronuclei frequencies are provided in Table 6.
LegH Prep precipitation at the end of treatment was observed at ≥650 µg/mL without metabolic activation and at ≥250 µg/mL with metabolic activation in Experiment I and at ≥750 µg/mL in Experiment II (Tables 7 and 8).No increase 6 Journal of Toxicology       S4).Te feed formulation was held constant throughout the study.LegH Prep remained stable and homogenous throughout the study (data not shown).Mean dietary intakes were calculated to be 4646.9,8843.5, and 13035.9mg/kg/day for males and 4175.9, 8686.0, and 12401.5 mg/kg/day for females, respectively.

Journal of Toxicology
No mortalities and no changes in mean body weight (Table 9), LegH Prep intake (Table 10), mean daily body weight gain, food consumption, and food efciency (Supplementary Tables S7-S9) that were ascribed to the administration of LegH Prep occurred during the 14 day study.In-life clinical signs were comparable between the control and LegH Prep dose groups.

Pathology.
Dietary exposure to LegH Prep for 14 days in both male and female rats did not induce any biologically adverse changes in hematology and clinical chemistry parameters.Signifcant increases in mean phosphorus levels in Group 4 males and potassium levels in Group 3 and Group 4 males from control Group 1 were observed.All the changes in hematology parameters were considered unrelated to LegH Prep administration, including those that attained statistical signifcance because they occurred sporadically and were considered unrelated due to biological variance among rats as the magnitude of variation was minimal.Clinical chemistry parameters for male and female rats in Groups 2-4 were generally comparable to control Group 1 throughout the study except for statistically signifcant decreases (p < 0.05-0.01) in mean phosphorus levels for Group 4 males and in potassium levels for Groups 3 and 4; see Tables 11 and 12 for pathology results showing summary tables describing the mean, hematology, and clinical chemistry results.Te signifcant potassium and phosphorus results were within historical control ranges and the control levels for these parameters were on the very low end of the historical control range for the laboratory and rat strain, which resulted in a statistically signifcant response but was not a biologically signifcant efect.Te nonsignifcant increase in the female Group 4 AST value was not considered toxicologically signifcant, as there was a high degree of variability in the results indicating potential issues with the samples and not a consistent, toxicology-related efect.

90 Day Dietary Study in Rats with a 28 Day Recovery
Period.Te neat LegH Prep was monitored for stability and deemed stable over the course of the study to within an acceptable margin of variability.Homogeneity and dietary stability analyses showed that LegH Prep was evenly distributed and was stable in the dietary matrix during the 4 day preparation interval.Homogeneity analysis of Day 0 dietary preparations reported a relative standard deviation of 2.82, 1.12, and 0.19% for dietary concentrations of 30,000, 60,000, and 90,000 ppm dose group formulations.Stability testing found that the test substance was at 90.4, 95.7, and 99.5% at Day 4 of nominal concentrations of 30,000, 60,000, and 90,000 ppm of the LegH Prep for Groups 2-4, respectively.Te dietary concentrations of 60,000 and 90,000 ppm, the intermediate and highest levels tested, were considered to have met target concentrations.Week 13 concentration verifcation for 30,000 ppm was below target.
No mortalities occurred over the course of the study.Tere were no clinical observations attributed to the dosing of LegH Prep.All clinical observations noted were considered incidental and of no toxicological relevance, as there were no trends in observations that increased with the dietary level.Also, there were no changes in body weight (Figure 1, Tables 13 and 14), body weight gain (data not shown), food consumption, or food efciency of male and female rats over the course of this main study phase or recovery phase attributed to the dietary intake of LegH Prep (Supplementary Tables S10-S13).Te daily intake of LegH Prep was calculated by body weight and food consumption measurements collected over the course of the study.Mean weekly body weights and mean daily body weight gains for male and female rats in Groups 2-4 (30,000-90,000) were comparable to control Group 1 (0 ppm) throughout the 90 day study and recovery period.Te mean overall (Days 0-91) daily intake for the main test rats fed 30,000, 60,000, and 90,0000 ppm of the LegH Prep was calculated to be 1637.3,3202.3, and 4820.4 mg/kg/day of LegH Prep for males and 2024.8,4127.9, and 5930.8 mg/kg/day LegH Prep for females, respectively (Table 15).Tese values were generally in good agreement with targeted LegH Prep exposure concentrations in mg/kg/day for males and females.12 Journal of Toxicology      16-23).All statistically signifcant changes were within historical control ranges, without histopathological correlate, and were not considered adverse.A slight increase in LDL cholesterol in the intermediate dose (60,000 ppm male group) was considered toxicologically insignifcant as there was no dose progression (Table 19).Dietary exposure to LegH Prep at levels of up to 90,000 ppm for at least 90 days resulted in no test articlerelated macroscopic observations, organ weight changes, or microscopic fndings.Signifcant changes in absolute (Tables 16 and 17) and relative thymus weight, as well as epididymis-to-body weight, and signifcant increases in kidney-to-brain weights were observed in a nondosedependent manner and were therefore not considered related to LegH Prep consumption (Tables S14 and S15).No signifcant alterations were found in the estrous cycle distribution between the control and high dose groups during the main phase or recovery phase of the 90 day study (Supplementary Table S16).Journal of Toxicology

Discussion/Conclusion
Impossible Foods developed an innovative approach to bring change to the alternative protein market via the discovery of soy leghemoglobin's unique organoleptic properties which mimic the taste and aroma of animal meat.By replacing animal protein with sustainable plantbased options, consumers are empowered to make changes in real-time that signifcantly help reduce greenhouse gas emissions by choosing plant-based products over animal meat.LegH Prep is manufactured via a genetically modifed P. pastoris (K.phafi) production strain that overexpresses LegH protein under submerged fermentation.During this process, the cells are lysed and the LegH is collected using a fltration-based recovery process.Te LegH Prep contains LegH protein, host proteins, and food-grade stabilizers [4].Health Canada [13] previously reviewed the petition to add LegH Prep to foods at a maximum soy leghemoglobin protein level of 0.8% into a variety of meat analog products and concluded that the ingredient was safe for human consumption at the intended levels of intake.Te present work builds on that conclusion of safety, as no mutagenic, genotoxic, or general toxicological adverse efects due to LegH Prep administration were found in the current set of studies.Te current in vivo studies increase the length of LegH Prep administration from 28 days [4] to 90 days, with a 28 day recovery period, reconfrming the safety of longterm ingestion of LegH Prep as demonstrated by classical in vivo toxicity studies conducted according to OECD protocols.
A battery of in vivo and in vitro testing has already been performed on LegH Prep to determine its safety in foods [4].Tese studies have shown, under their respective testing conditions, that LegH Prep is safe to consume at the intended intake levels and does not pose any signifcant risk of dietary allergy or toxicity to consumers.Since 2016, LegH Prep has been incorporated into over 500 million servings of ¼ pound (113 g) meat analog products without any reported adverse efects.A new set of genotoxicity studies (bacterial reverse mutation assay and an in vitro Mammalian micronucleus assay in human lymphocytes) were performed to evaluate the potential of LegH Prep to induce mutations.In conclusion, based on the data collected in the mutagenicity and under the experimental conditions reported, LegH Prep derived from P. pastoris (K.phafi) did not cause gene mutations by base pair changes or frameshifts in the genome of the fve bacteria tester strains used and up to a maximum dose of 5000 μg LegH/plate.Terefore, LegH Prep is nonmutagenic in this bacterial reverse mutation assay.Similarly, LegH Prep was found to be nonclastogenic/nonaneugenic in the in vitro mammalian micronucleus assay using human lymphocytes, which evaluated LegH Prep's potential to induce micronuclei in human lymphocytes.Precipitation of the test item in the cultures at the end of treatment was observed at 650 μg/mL and higher without metabolic activation and at 250 μg/mL and higher with metabolic activation in Experiment I and at 750 μg/mL and higher in Experiment II.LegH Prep did not induce structural and/or numerical chromosomal damage in human lymphocytes, in agreement with the results of Fraser et al. [4].Overall, under the conditions of this assay, the results show that LegH Prep is nonmutagenic and nonclastogenic.
Further adding to the body of work demonstrating the safety of LegH Prep as a food ingredient, a 90 day dietary study was performed in rats to evaluate the potential subchronic toxicity of LegH Prep with the addition of a 28 day recovery phase designed to follow up on any potential adverse efects observed during the 90 day study.No adverse efects were observed due to the dietary intake of LegH Prep at the maximum dose tested.Te study resulted in no mortalities and no clinical observations: body weight, ophthalmological, clinical pathology, or histopathological changes due to LegH Prep administration.LegH Prep is not intended for consumption on its own but as a component of plant-based meat products.Tis 90 day dietary toxicity study in rats established a NOAEL of 4798.3 and 5761.5 mg/kg/day, the maximum level consumed by male and female rats, respectively.Consequently, the results of all the studies presented in this article demonstrate that the dietary consumption of LegH Prep which contains soy LegH and P. pastoris proteins from the production strain is not toxic under the conditions tested.
Adjusting our diets to replace animal meat with plantbased options signifcantly reduces the environmental impact inficted by the animal agriculture industry.Impossible Foods' mission is to create safe food technologies that enable us to choose delicious and sustainable plantbased alternatives to animal meat, while simultaneously decreasing the environmental carbon footprint of animal agriculture.

Table 1 :
Study design-in vitro mammalian micronucleus assay in HPBLs using LegH Prep.
-14 and 90 Day Dietary Feeding Study in Rats.Statistical analysis on all the data collected during the in-life phase of both the 14 and 90 day studies was performed by PSL.Te probability value of p < 0.05 was set for signifcance.Te mean and standard deviation were calculated for all quantitative data.Male and female rats were evaluated separately.Statistical analysis was performed on all quantitative data for in-life and organ weight parameters using Provantis ™ version 10, tables and statistics, Instem LSS, Stafordshire, UK.For the 14 day study, the following programs were used for analysis INSTAT or Prism Biostatistics, GraphPad Software, San Diego, CA; Statview, version 5, SAS Institute Inc., Cary, NC; and SigmaStat, version 2, SYSTAT Software, San Jose, CA.
B-naphthofavone-induced Sprague-Dawley rat liver microsomal fraction; −S9 � without S9 microsomal fraction.Data are shown as mean ± SD revertants/plate for three replicates for each concentration in each experiment.
B-naphthofavone-induced Sprague-Dawley rat liver microsomal fraction; −S9 � without S9 microsomal fraction.Data are shown as mean ± SD revertants/plate for three replicates for each concentration in each experiment.

Table 5 :
Summary of bacterial reverse mutation assay results of LegH Prep (Experiment II; up to 5,000 µg/plate)-mutation factor.

Table 6 :
LegH Prep concentrations used with and without metabolic (S9) activation in the in vitro mammalian micronucleus assay.

Table 7 :
Summary result table for Experiments I and II, micronucleus induction in human lymphocytes, 4 h treatment, 44 h fxation interval; without metabolic activation (−S9).

Table 9 :
Summary of mean body weights (g)-14 day study *

Table 10 :
Summary table of mean daily dietary intake of LegH prep-14-day study.

Table 14 :
Summary of mean body weights in 90 day study (g) recovery phase a, * .

Table 15 :
Summary of mean daily dietary intake of LegH Prep (mg/kg/day)-90 day dietary study a, * .

Table 16 :
Summary of mean terminal body weights and organ weights (g)-90 day dietary study.

Table 17 :
Summary of mean terminal body weights and organ weights (g)-90 day dietary study-recovery phase.

Table 19 :
Summary of clinical chemistry parameters-90 day dietary study # .

Table 20 :
Urinalysis-90 day dietary study1 1 Urine obtained on Day 92 relative to start date.* n: inappropriate for statistics.