Safety assessment of membrane-associated protoporphyrinogen IX oxidase from Enterobacter cloacae expressed in genetically modified crops

ABSTRACT

Crops tolerant to protoporphyrinogen IX oxidase (PPO)-inhibiting herbicides were developed by expressing an herbicide insensitive PPO, a membrane-associated protein from Enterobacter cloacae. E. cloacae is commonly found in the environment, and PPO is present in thousands of species, including probiotic organisms with a history of safe use. Comparisons with current allergen and protein toxin databases revealed no sequence similarities between PPO and known allergens or toxins. To validate the use of Escherichia coli-produced PPO in safety studies, physicochemical and functional characterization demonstrated that the PPO produced by genetically modified (GM) soybean has comparable immunoreactivity and functional activity to that produced from E. coli, with neither being glycosylated. PPO was fully digested after exposure to pepsin and pancreatin for 2 and 5 minutes, respectively, and its activity is completely lost at temperatures of 55°C or higher. Mice dosed orally with PPO at a level of 5000 mg protein per kg body weight showed no adverse effects, as indicated in body weight gains, food consumption, and clinical observations. This comprehensive safety assessment indicates that PPO protein from GM crops is safe for food and feed consumption. Additionally, we present methods to demonstrate the functional equivalence of a membrane-associated protein from E. coli and plant, along with a novel process for formulating PPO in gram quantities at 76.3 mg/ml.

Introduction

A major challenge in sustainable agriculture is managing weed competition for essential resources such as minerals, water, sunlight, and space.¹ The development of herbicide-tolerant crops has significantly improved weed management.² By 2023, genetically modified (GM) herbicide-tolerant crops were widely adopted in the U.S., accounting for 95% of soybean acreage, 94% of cotton acreage, and 91% of maize acreage, with similar adoption trends observed globally.^3,4 While GM crops tolerant to herbicides like glyphosate and glufosinate have proven invaluable, new herbicide-resistant traits are needed to prevent overreliance on a single herbicide and manage more resistant weed species.

Protoporphyrinogen IX oxidase (PPO), classified under the oxidoreductase superfamily (EC 1.3.3.4), catalyzes the oxidation of protoporphyrinogen IX to protoporphyrin IX.⁵ This reaction is a crucial step in heme production for oxygen transport in animals and in the synthesis of chlorophyll and other essential tetrapyrroles for photosynthesis in plants.^6,7 PPO is widely present in various organisms and categorized into several types, including HemG, HemY, and HemJ, each playing essential roles in different biological pathways.⁸ HemY PPO, primarily found in eukaryotes such as plants and animals, is involved in the biosynthesis of these critical compounds. In plants, there are two isoforms of PPO: one localized in the chloroplasts, which is involved in chlorophyll biosynthesis, and the other in the mitochondria, which plays a role in heme biosynthesis.^9,10 In animals, PPO is localized in the mitochondria, where it functions in the heme biosynthesis pathway.^5,11,12 HemJ is thought to have originated in α-proteobacteria and is also found in other proteobacteria and cyanobacteria, where it is involved in heme biosynthesis.^13,14 On the other hand, HemG is primarily present in γ-proteobacteria including Escherichia coli. It forms membrane-associated oligomers and uses non-covalently bound flavin mononucleotide (FMN) as a cofactor for its catalytic activity.^12,15 Notably, the HemG PPO is an oxygen-independent enzyme that uses menaquinone or menadione, rather than oxygen, as an electron acceptor.¹⁵ Consequently, this class of PPO is also known as menaquinone-dependent protoporphyrinogen IX dehydrogenase.¹⁵ Despite the diversity in protein sequences and structures among various PPO classes, the conservation of PPO enzymatic function across a wide range of organisms underscores its essential role in fundamental life processes, highlighting its evolutionary importance in both prokaryotes and eukaryotes. Remarkably, HemG-type PPO can substitute for HemY function in plants and vice versa,^16,17 demonstrating the conserved role of PPO in protoporphyrinogen IX oxidation across biological systems.

This conservation is particularly significant in the context of herbicide development, as the inhibition of PPO activity by specific inhibitors leads to the accumulation of reactive intermediates, resulting in oxidative stress in plants. Such oxidative stress damages cellular membranes and eventually leads to plant cell death, making PPO inhibitors highly effective as herbicides.^16,18 Given PPO’s essential role in plant metabolism, it has become a major target in herbicide discovery, especially for weed control strategies.¹⁹ A bacterial PPO derived from Enterobacter cloacae, known as H_N90 PPO and belonging to the HemG family of PPO, presents a promising alternative. Unlike the native plant PPO (HemY), which is sensitive to some PPO-inhibiting herbicides, H_N90 PPO (hereafter referred to as PPO) has been engineered for expression in the chloroplasts of maize, cotton, and soybean, conferring tolerance to these herbicides.¹⁶ This genetic modification enables these GM crops to resist herbicidal applications of PPO inhibitors, providing farmers with enhanced options for effective weed management.

However, prior to commercialization, proteins expressed in GM crops must undergo a rigorous safety assessment based on a framework established by the FAO and WHO Codex Alimentarius Commission in 2009.²⁰ This assessment uses a tiered, weight-of-evidence approach to evaluate the safety of the protein for human and animal consumption.^20–26 The first tier includes reviewing the history of safe use (HOSU) of the protein and its host organism, along with bioinformatics analyses to assess whether the new protein resembles existing proteins in food and feed, as well as known toxins or allergens. This tier also involves characterizing the physicochemical and functional properties of the protein, evaluating its susceptibility to digestive enzymes, and determining its stability under heat treatment typical of grain processing and cooking. If a hazard is identified in Tier I, the second tier focuses on assessing potential mammalian toxicity in vivo. Recently, HOSU, bioinformatics analysis, and protein characterization have been recommended as core studies, while assessments of protein susceptibility to digestive enzymes, heat stability and animal toxicity are now considered as supplementary studies.²⁷ Regardless of classification, both approaches emphasize that animal studies should be conducted in a hypothesis-driven manner.

PPO is derived from the bacterium E. cloacae, which is recognized for its adaptability and presence in a variety of environments, including soil, water, plants, the gastrointestinal tracts of humans and animals and other organisms.^28–32 This bacterium has been utilized as a host for cloning and expressing foreign genes,^6,7 and has applications in alternative energy generation,^7,33,34 the biodegradation of low-density polyethylene,³⁵ polymers and pollutants³⁶ and pesticides,^37,38 plant growth promotion⁶ and plant disease control.^39–41

While E. cloacae is ubiquitously present in the environment and has a documented history of safe use in various applications, it is also known for its potential to cause opportunistic infections in immunocompromised patients.^42,43 PPO, including the HemY, HemJ, and HemG types, has been identified in over 1,000 species, with 2,726 distinct variants classified as HemG-type PPO.^8,16 Notably, homologous variants of H_N90 PPO are also present in probiotic bacterial strains such as Lactococcus lactis and Levilactobacillus brevis, based on NCBI gene database searches, as well as in the beneficial strain Bdellovibrio bacteriovorus.⁸ The ubiquitous presence of PPO across various organisms, coupled with human and animal exposure, suggests that it is unlikely to be associated with allergenicity or toxicity from an evolutionary perspective; indeed, there are no reported cases linking PPO to such concerns. Given these well-established safety properties, it could be suggested that an animal toxicity study is not necessary for PPO. However, it is important to note that safety assessment results regarding membrane-associated proteins like PPO remain limited due to their unique characteristics. Consequently, this publication aims to conduct a comprehensive safety assessment of PPO to identify any potential unexpected results related to this membrane-associated protein. The production and storage of active PPO require detergents and other additives,⁴⁴ which complicate protein preparation for safety studies. Nevertheless, we have successfully addressed challenges by developing strategies for PPO equivalence determination, optimizing detergent removal, formulating PPO to a high-concentration, and maintaining PPO stability and activity while advancing methods for this safety assessment. Through this comprehensive evaluation, we present the safety profile for PPO and provide valuable insights for assessing future similar membrane-associated proteins.

Methods & Materials

Bioinformatic Screening of Toxin, Allergen and Protein Databases

The transgenic PPO variants expressed in soybean, maize, and cotton exhibit up to 13 amino acid differences at the N-terminus, which are attributed to variable processing of the chloroplast transit peptide (CTP). Three PPO variant sequences were previously reported.⁴⁴ These PPO protein variant sequences were subjected to bioinformatic screens against databases containing all protein, toxin, and allergen sequences following previously established methods^26,45,46 as outlined in Codex Alimentarius.²⁰ The all-protein database used was the GenBank release 264 ncbi-asn1 all protein dataset.⁴⁷ This database contains 344,558,146 sequences and is herein referred to as PRT_2025. The updated toxin database was derived from the UniProtKB database using the keyword search “(keyword:toxin) AND (reviewed:true)” to isolate curated sequences annotated with the term toxin.⁴⁸ A secondary screen was applied to remove sequences unlikely to be toxins based on keywords such as “anti-toxin,” and sequences out of scope for the analysis (e.g. insecticidal/cry/crystal). The final database contains 7,495 sequences and is herein referred to as TOX_2025. The allergen database utilized was the Health and Environmental Sciences Institute COMPARE (COMprehensive Protein Allergen Resource) database,⁴⁹ which contains 2,836 sequences and is herein referred to as AD_2025. All alignments were generated with FASTA v36.3.8i with the E-value cutoff set to 1.⁵⁰ Returned alignments were deemed significant (i. e. potentially homologous) if displaying E-values of ≤1e-5 (1 ×10⁻⁵), which is recognized as a conservative threshold for protein homology,⁵¹ and has been utilized previously.^45,46

Protein Expression, Purification, Characterization and Equivalence

Both plant- and E. coli-expressed PPOs were produced using the method described previously.⁴⁴ All methods utilized to characterize both plant-produced and E. coli-produced PPO proteins are similar to what have been previously reported.^26,44,45,52 Briefly, the purity-corrected protein concentration of the PPO protein purified from soybean seeds were determined by enzyme-linked immunosorbent assay (ELISA) and total protein concentration of E. coli-produced PPO was determined by amino acid compositional analysis. Apparent molecular weight of plant-produced PPO and both purity and apparent molecular weight of E. coli-produced PPO was determined using densitometric analysis of Coomassie stained SDS – PAGE gels. The identity of both plant-produced and E. coli-produced PPO was confirmed by N-terminal sequence determination and peptide mass fingerprint analysis using nano liquid chromatography tandem mass spectroscopy (LC-MS/MS).⁴⁴ For western blot analysis, each protein was subjected to SDS – PAGE and transferred to a nitrocellulose membrane. The blot was probed with a mouse anti-PPO specific monoclonal antibody.

Glycosylation analysis was conducted following the ECL Glycoprotein Detection method (GE Healthcare) using transferrin as a positive control. The plant-produced PPO protein, E. coli-produced PPO protein, and transferrin (Sigma) were subjected to Tris-glycine 4–20% (w/v) SDS-PAGE followed by electrotransfer to a PVDF (polyvinylidene difluoride) membrane (Invitrogen). Glycosylation analysis on the PVDF membrane was conducted at room temperature according to the manufacturer’s instructions. Carbohydrate moieties, regardless of the type of glycosidic linkage, were detected using a streptavidin Horseradish Peroxidase conjugate, followed by luminol-based detection using an ECL reagent (SuperSignal West Dura Extended Duration Substrate, Thermo Scientific). Blot image was captured with a Bio-Rad ChemiDoc Imager and analyzed using Bio-Rad Image Lab 6.1 Security Edition software.

The method used to determine functional activities of both plant-produced and E. coli-produced PPO proteins were previously reported.⁴⁴ Briefly, all samples were thawed on wet ice just prior to running the assay and all reactions were set-up on wet ice, but the assay was performed at 30°C. Each reaction mixture contains 100 mM Tris, pH 7.4, 1 mM EDTA, 0.1% Tween 80, 2 mM reduced glutathione, 50 μM menadione, 8 μM FMN, and 0.2 μg of the test substance in a total volume of 200 µl and the same mixture without PPO protein as negative control. Additionally, a final concentration of 70 nM herbicide (Flumioxazin, a native plant PPO inhibitor) was added to wells for plant-produced PPO activity assay to inhibit the activity of any possibly co-purified plant native PPO proteins. The specific activity was calculated based on the average nmol of the protoporphyrin (PPN) IX generated per minute by one mg of test substance (nmol/minute/mg) during a 6-minutes reaction time, and reported as nmol/min/mg.

Assessment of PPO Protein Susceptibility to Pepsin and Pancreatin

The susceptibility of PPO protein to degradation by pepsin was assessed following a standardized protocol.⁵³ Briefly, the His-tag PPO protein purified from E. coli was mixed with high purity pepsin (Sigma) to a final protein-to-pepsin ratio of 1 µg total protein:10 U of pepsin. The reaction mixture tube was immediately placed in a 37 ± 2°C water bath. Samples were removed at 0, 0.5, 2, 5, 10, 20, 30, and 60 min and were immediately quenched by the addition of 0.7 M sodium carbonate and SDS-PAGE sample loading buffer. Protein only and pepsin only experimental controls were also prepared and incubated for 60 min in a 37 ± 2°C water bath. All resulting samples were heated at 95–100°C for 5–10 min before being analyzed by SDS-PAGE and western blot analysis.

The susceptibility of PPO to degradation by pancreatin was also assessed. Pancreatin (Thermo Fisher Scientific) was dissolved in 50 mM potassium phosphate buffer (pH 7.5) to a concentration of 10 mg of pancreatin powder/ml as described in the United States Pharmacopoeia.^45,54 The pancreatin solution reaction was formulated so that 55.3 μg of pancreatin powder would be present per μg of PPO. Samples were removed at 0, 5, 15, 30 min and 1, 2, 4, 8, and 24 h (hour), and quenched with SDS-PAGE loading buffer before being analyzed by western blot analysis.

Stability of the PPO Protein at Temperatures Encountered in Cooking and Processing

The His-tag PPO protein purified from E. coli was incubated at 25, 37, 55, 75 or 95 ± 2°C for 15 or 30 min. All samples were returned to wet ice immediately following heat treatment. A control sample was maintained on wet ice throughout the treatment period. Following the heat treatments, all samples were subjected to functional activity analysis as described above.

Acute Oral Toxicity Assessment of PPO Protein

Formulation of Dosing Solutions

To formulate the test dosing solution (TDS), a total of 8.1 L of the His-tag PPO protein samples in a buffer composed of 1x PBS, 56 mM L-arginine (L-Arg), 0.1 mM FMN, 20% glycerol, 60 mM L-histidine, 60 mM imidazole, and 0.18% (3 mM) Thesit®^a (also known as polyethylene glycol dodecyl ether, a type of fatty alcohol ethoxylate that acts as a detergent) at pH 9.1, was removed from a −80°C freezer, thawed at 4°C on ice for ~9 hr, and then slowly diluted to 100 L with Dilution Buffer (5 mM potassium phosphate monobasic, 28 mM L-Arg, 0.1 mM FMN, pH 10.3). The diluted PPO sample was first concentrated to 2 L using a hollow fiber cartridge (100,000 NMWC, 1.15 m² surface area, Cytiva Life Sciences). After concentration, the 2 L sample was buffer exchanged via continuous diafiltration into Dilution Buffer for 50 turnover volumes, concentrated down to 950 ml using the same hollow fiber device, and further concentrated to 250 ml with a smaller hollow fiber cartridge (100,000 NMWC, 1400 cm² surface area, Cytiva Life Sciences). Finally, it was concentrated down to 60 ml with Amicon Ultra-15 Centrifugal filters (100,000 NMWC, EMD Millipore). About 3 ml of Dilution Buffer was used for the final rinse of the filters and pooled with concentrate for a final TDS of 63 ml.

To formulate the control dosing solution (CDS), BSA powder was weighed out and dissolved in 150 ml of Vehicle Dosing Solution (VDS) to a target concentration of at least 75 mg/ml. The pH of the CDS was adjusted to match the pH of the TDS and to a final volume of 200 ml with VDS.

The Vehicle Dosing Solution (VDS) is the permeate collected (2 L) during the final concentration phase through the hollow fiber cartridge in the preparation of TDS, stored at 4°C, and adjusted Thesit® concentration to approximate the level of that in TDS.

Prior to the initiation of the study, the dosing solutions were analyzed to confirm their concentration, homogeneity, Thesit®, pH and conductivity. Thesit® was measured as outlined below. Conductivity was assessed using a METTLER TOLEDO^TM SevenCompact Duo pH/Conductivity meter. Once the dosing solutions met the target specifications, they were placed in 125 ml Nalgene square poly bottles with sealing caps, frozen in a dry ice and ethanol bath, and stored at −80°C before being shipped on dry ice to Charles River Laboratories, Inc. for dosing. Protein stability throughout the dosing period was evaluated by measuring protein concentration with Bio-Rad and assessing the integrity of pre- and post-dosing solutions via SDS-PAGE. PPO activity for both pre- and post-dosing TDS was determined using the previously described method,

Determination of Thesit® Concentration in Dosing Solutions by Nuclear Magnetic Resonance (NMR)

The 3-(Trimethylsilyl)-1-propanesulfonic acid-d6 sodium salt (DSS-d6, Sigma) was used as a standard molecule for the quantitative NMR (qNMR). The standard solution of 100 mM DSS-d6 in D₂O was prepared. Its exact concentration or purity (2.038% w/w) was then measured using the qNMRw method with Milli-Q water as a calibrant.⁵⁵ The qNMR sample was prepared in four steps: (1) Add and weigh 50 µl dosing solution in a 5-mm NMR tube; (2) Add and weigh 10 µl DSS-d6 standard solution to the NMR tube; (3) Add 600 µl methanol-d4 to the NMR tube; and (4) Mix the sample by reversing the NMR tube up-and-down for 20 times or more. The proteins were precipitated and settled at the bottom of the NMR tube.

The PROTON qNMR spectra were acquired using a Bruker 600 MHz NMR spectrometer with a Cryoprobe. The NMR spectral data acquisition parameters include 30° pulse, 5 seconds acquisition time and 55 seconds relaxation delay. The Fourier transformation was done with an exponential window function and 0.3 Hz line broadening. The spectrum was then processed with phasing, baseline correction and peak integration. The Thesit® concentration P (% w/w) was calculated using the following qNMR equation.⁵⁵

p = 3P_rW_rMA/(WM_rA_r)

P_r (2.038, % w/w) is the concentration of DSS-d6 standard solution; W_r and W are weights (mg) of DSS-d6 standard solution and dosing solution, respectively, in the qNMR sample; M (583 Da) and M_r (224.36 Da) are molecular weights of Thesit® and DSS-d6, respectively; A is the area of the NMR peak at 0.90 ppm from the methyl group (3 protons) of Thesit® molecules; and A_r is the area of the NMR peak at 0.00 ppm from methyl groups (9 protons) of DSS-d6.

Acute Toxicity Study

The acute oral toxicity of PPO was assessed in a GLP compliant mice oral acute study following the EPA Health Effects Test Guideline OPPTS 870.1100. The study was conducted at Charles River Laboratories in Ashland, OH. Briefly, CD-1 mice aged approximately 8–9 weeks were allocated to three groups of 20 mice each (10 males and 10 females) using a stratified randomization scheme designed to achieve similar group mean body weights. Mice were fasted for approximately 3 hours prior to dose administration. The test, control, and vehicle dosing solutions were administered in two doses with each dose volume of 33.3 mL/kg body weight (3 to 4 hours apart) by oral gavage on day 1 for a targeted dose of 5000 mg/kg/day. The animals were fasted between doses and feed was returned following the second dose. Mice were dosed with either the TDS, CDS, or VDS. During the 14-day observation period, PMI Nutrition International Certified Rodent Chow No. 5002 and water were available to the rodents ad libitum except during designated procedures. Animals were assessed for mortality twice daily. Detailed clinical observations were conducted within 4 days of receipt, on the day of randomization, a minimum of 2 times post dose on Day 1 and then once daily thereafter. Individual animal body weights were recorded within 4 days of receipt, on the day of randomization, on Day 1 (prior to fasting), on Day 8 and on the day of scheduled necropsy (Day 15). Food consumption was measured quantitatively on Days 1, 8, and 15. Study animals were subject to a complete necropsy under the supervision of a board-certified veterinary pathologist. All statistical tests were conducted at the 5% significance level. All pairwise comparisons were conducted using two-sided tests and were reported at the 1% and 5% levels. Group homogeneity was assessed with Leverne’s test. Either a two-sided t-test or Wilcoxon Rank sum test was used to compare datasets with two groups. Datasets with at least three groups were compared with either a one-way ANOVA F-test or Kruskal-Wallis test.

Results

Bioinformatic Assessing Relationship of PPO to Proteins in the Current Database

The results of the bioinformatic searches conducted with the PPO protein variants against the AD_2025 database resulted in no alignments exceeding the E-value threshold of ≤1e-5, no windows of alignment that displayed >35% identity over 80 amino acids, and no contiguous 8-mer peptide matches (Table 1).

Table 1.

Summary of alignments for the FASTA and sliding window peptide searches of the 2025 databases utilizing the PPO variant protein sequences.

Database	Sequence Namea	Search of the 2025 Sequence Databases
		FASTA search
		8mer	35% ID 80 aa	# Hits	Accession	Description	Identity	aa Overlap	E-value
AD_2025	PPO	No	No	2	P42058.1	Alt a 7, unknown function [Alternaria alternata]	34.4	32	0.33
AD_2025	PPO+3	No	No	2	P42058.1	Alt a 7, unknown function [Alternaria alternata]	34.4	32	0.3
AD_2025	PPO+13	No	No	2	P42058.1	Alt a 7, unknown function [Alternaria alternata]	34.4	32	0.42
TOX_2025	PPO	N/A		1	H1ZZI3	Toxin To14 OS=Tityus obscurus	34.9	43	0.75
TOX_2025	PPO+3	N/A		1	H1ZZI3	Toxin To14 OS=Tityus obscurus	34.9	43	0.85
TOX_2025	PPO+13	N/A		1	H1ZZI3	Toxin To14 OS=Tityus obscurus	34.9	43	0.81
PRT_2025	PPO	N/A		20125	QPW68890.1	Sequence 100 from patent US 10,745,712	100.0	179	2.1e-81
PRT_2025	PPO+3	N/A		20035	WON41786.1	Sequence 110 from patent US 11,629,358	100.0	178	4e-81
PRT_2025	PPO+13	N/A		20041	WQA93882.1	Sequence 11 from patent US 11,758,907	100.0	178	1.7e-81

^aPPO refers to the GM cotton-expressed PPO, PPO+3 to the soybean variant studied in this publication, and PPO+13 to the maize variant.⁴⁴

The search against the TOX_2025 database resulted in no alignments exceeding the E-value threshold of ≤1e-5. For each variant, the single returned alignment was to a small portion of a toxin sequence that displayed just 34.9% identity over 43 amino acids (E-score of 0.75), a level of homology that is not likely to indicate common or synonymous function.

The search of PRT_2025 resulted in numerous alignments to the PPO protein sequence. The top alignment for each variant was the positive self-identification of the PPO sequence as filed in numerous patents (QPW68890.1, WON41786.1, WQA93882.1). Subsequent alignments consisted of other patent entries for the same sequence and other outgroup proteins containing the HemG domain. The positive identification of PPO and its broader family does not indicate a potential for adverse biological effects but rather validates the accurate placement and profiling of the test proteins within this family.

Altogether these searches did not return alignments indicating a capacity for allergenic or toxic effects and accurately identified the PPO protein as a member of the HemG domain containing PPO family. This is unsurprising as this protein family is known to be broadly represented and expressed within both prokaryotic and eukaryotic species without any adverse effects.

Characterization and Equivalency Assessments of Soybean-Produced PPO with E. Coli-Produced PPO

The PPO protein was purified from GM soybean seed and E. coli cell paste as described in the Methods section. Since the PPO expression level in soybean seed was low, <1 ppm, it was necessary to express the protein in a heterologous host system such as E. coli to obtain enough amount of protein to enable safety assessment. Additionally, PPO is a membrane-associated protein with an intrinsic tendency for aggregation, making expression and purification very difficult. Over 700 different conditions related to PPO expression and purification and four different removable tagged constructs were tested, but we failed to purify enough amount of tag-free PPO protein with >90% purity from E. coli for safety assessment.⁴⁴ Therefore, a 6x His-tag was added to the N-terminus of the PPO to facilitate protein expression and purification from E. coli. Previously, both His-tag PPO and non-tag PPO purified from E. coli were shown to be equivalent in immunoreactivity and functional activity.⁴⁴ To demonstrate whether E. coli-produced His-tag PPO is a suitable surrogate for assessing safety of the PPO expressed in GM soybean, characterization and assessment of physiochemical and functional equivalence for both plant-produced and E. coli-produced PPO were conducted.^20,52,56

During the characterization of the plant-produced protein, it was found that CTP was incompletely processed, leaving an additional three amino acids, aligning with the previous report.⁴⁴ The purity and/or apparent molecular weight, identity, and immunoreactivity, and functional activity of GM soybean-produced PPO and E. coli-produced PPO were determined (Table 2 and Figure 1). The apparent molecular weight (MW) of plant-produced and E. coli-produced PPO was 18.8 and 19.8 kDa, respectively. Densitometric analysis of SDS-PAGE gel indicated that the purity of the E. coli-produced PPO was 97%. Since the purity of the plant-produced PPO was too low to be determined, the purity-corrected PPO concentration was determined by ELISA. The sequence identities of both plant-produced and E. coli-produced PPO proteins were confirmed with 85% and 100% coverage of the entire protein sequence, respectively (Table 2). Immunoreactivities of plant-produced and E. coli-produced PPO proteins were assessed using western analysis. The results demonstrated that the two PPO proteins migrate at the expected apparent molecular weights (Figure 1) and show comparable band intensities between plant-produced PPO and E. coli-produced PPO (Figure 1), suggesting that both plant-produced and E. coli-produced PPO have equivalent immunoreactivity. Therefore, the difference in the N-terminal sequence did not impact the protein immunoreactivity.

Table 2.

Summary of PPO characteristics.

Characteristics	Methods	Results
		Soybean-produced PPO	E. coli-produced PPO
Apparent MW (kDa)	SDS-PAGE/Densitometry	18.8	19.8
Purity (%)	SDS-PAGE/Densitometry	Not determined	97
Identity	N-terminal sequence	DASKALVLYST	MHHHHHKALVLYST
	LC-MS/MS coverage	85%	100%
Immunoreactivity	Western blot	confirmed	confirmed
Function activity (nmol/min/mg)	Spectrophotometer	266.7a	258.4b
Glycosylation	GE Glycosylation Detection Module	None detected	None detected

^aValue refers to mean calculated based on one assay with two replicates.

^bValue refers to means calculated from eight assays, and each assay has two replicates.

Figure 1.

Western blot analysis and immunoreactivity of plant-produced and E. coli-produced PPO proteins. The plant-produced and E. coli-produced PPO proteins were resolved on a pre-cast 15-well Tris-Glycine 4–20% (w/v) polyacrylamide gradient mini-gel by Tris-Glycine-SDS running buffer (Invitrogen) and electro-transferred onto a nitrocellulose membrane. The blot was probed with an anti-PPO specific mAb and developed using an enhanced chemiluminescence system. E. coli-produced and plant-produced PPO proteins were loaded in duplicate at 0.4, 0.8, and 1.6 ng based on purity-corrected protein concentration or protein concentration determined by ELISA, in lanes 2 to 7 and 8 to 13, respectively. MagicMark XP Western Protein Standards (Invitrogen) were loaded in lane 1 for molecular weight reference.

The specific activities of plant-produced and E. coli-produced PPO proteins were assessed, which were 266.7 and 258.4 nmol/min/mg, respectively (Table 2). The activity values are very close to each other and align well with previously published data. These values also fall within the established prediction interval (PI).⁴⁴ PI is a well-established statistical analysis method to set acceptance limits for functional equivalence assessment^57–59 . Since the specific activity of plant-produced PPO falls within the PI, it was deemed to have equivalent functional activity to that of the E. coli-produced PPO.

Glycosylation analysis was conducted to determine if both plant-produced and E. coli-produced PPOs were glycosylated. The results are presented in Figure 2. A clear signal was observed at an apparent MW (~80 kDa) for the transferrin positive control (Figure 2, Lanes 3 and 4). No glycosylation signals around 19.0 kDa were observed for the plant-produced (Figure 2, Lanes 9 and 10) or E. coli-produced (Figure 2, Lanes 6 and 7) PPO. Glycosylated signals were observed in the ~30, ~75, and ~250 kDa regions (Figure 2, Lanes 9 and 10). These correspond to co-purified endogenous, glycosylated soybean proteins, not PPO proteins, as they are not recognized by the PPO monoclonal antibody (Figure 1). All of this weight of evidence demonstrates that the His-tag PPO produced from E. coli is a suitable surrogate for the safety assessment of soy-produced PPO. Therefore, safety studies were conducted using the N-terminal 6xHis PPO purified from E. coli.

Figure 2.

Glycosylation analysis of plant-produced PPO. Transferrin (positive control, Lanes 3 and 4, 75 ng and 150 ng, respectively), E. coli produced PPO (Lanes 6 and 7, 75 ng and 150 ng, respectively) and purified PPO expressed in plants (Lanes 9 and 10, 75 ng and 150 ng, respectively) were subjected to pre-cast Tris-glycine 4–20% (w/v) SDS-PAGE and electro-transferred onto a PVDF membrane. Lane 1: Precision plus protein standards (Bio Rad); Lanes 2, 5 and 8 are empty. Carbohydrate moieties were detected by using ECL reagents and the blot image was captured using a Bio-Rad ChemiDoc Imager.

Susceptibility of the PPO Protein to Pepsin and Pancreatin

The intact PPO protein was incubated with pepsin at 37 ± 2°C and assessed by SDS-PAGE analysis at specifically defined time intervals (Figure 3, Panel A). The intact, full-length PPO protein (∼19 kDa) was degraded to below the limit of detection (LOD) of the Coomassie-stained SDS-PAGE gel after 0.5 min (Figure 3 Panel A, Lane 5). The fragments of 2.5 kDa were visible during the first 2 min of the pepsin digestion (Figure 3, Panel A, Lane 6); after that time, PPO was completely digested. There was no change in the banding pattern for the protein when incubated at 37°C in the absence of pepsin (Figure 3, Panel A, Lanes 3 and 12). Additionally, there was no change in the protein band corresponding to pepsin (∼38 kDa) when incubated at 37°C in the absence of PPO protein (Figure 3, Panel A, Lanes 2 and 13).

Figure 3.

PPO degradation by gastrointestinal proteases. Stability of the PPO protein in the presence of pepsin (Panel A) and pancreatin (Panel B) was assessed by SDS-PAGE and Western blot analysis, respectively. Based on pre-reaction protein concentrations, 1 µg or 10 ng of PPO were loaded onto SDS-PAGE gels for electrophoresis followed by Coomassie staining or western blot analysis, respectively. Lane designations are as follows: Panel A: Lanes 1 and 14, Mark 12 molecular markers (Thermo Fisher Scientific); Lane 2, pepsin only at time zero; Lane 3, PPO only at time zero; Lanes 4–11, time course of PPO incubation with pepsin for 0, 0.5, 2, 5, 10, 20, 30, and 60 min, respectively; Lane 12, PPO only at 60 min; Lane 13, pepsin only at 60 min. Panel B: Lane 1, MagicMark XP Western Protein Standards (Invitrogen); Lane 2, pancreatin only at time zero; Lane 3, PPO only at time zero; Lanes 4–12, the time course of PPO digestion by pancreatin at time points of 0, 5, 15, 30 min, as well as 1, 2, 4, 8 and 24 h (hour), respectively; Lane 13, PPO only at 24 h; Lane 14, pancreatin only at 24 h.

The intact PPO protein was also incubated with pancreatin and assessed by western blot analysis at specifically defined time intervals (Figure 3, Panel B). Only western blot analysis was carried out due to the potential ambiguous detection from pancreatin which is a crude extract with multiple enzymes and components. The full-length protein was degraded to below the LOD of the western blot within the first 5 min (Figure 3, Panel B, Lane 5). A small amount of PPO protein appeared as a higher molecular weight ladder in the 24-hour no pancreatin control (Figure 3, Panel B, Lane 13), likely due to aggregation of this membrane-associated protein during prolonged incubation. However, the majority of PPO protein was still observed at approximately 19 kDa, with an intensity similar to that seen at zero min in the absence of pancreatin (Figure 3, Panel B, Lane 3). This indicates that the degradation of PPO protein was due to the proteolytic activity of pancreatin, rather than degradation during incubation in the buffer throughout the experiment.

Heat Treatment of PPO at Temperatures Common to Cooking and Processing

The thermal stability of PPO protein was evaluated by determining the percent residual enzyme activity following heat treatment at 25°C, 37°C, 55°C, 75°C or 95°C for 15 or 30 min compared with the activity of untreated PPO protein. The enzymatic activity of PPO was stable at an incubation temperature of 25°C but was reduced when incubated at 37°C. Enzymatic activities of PPO were not detected following incubation at temperatures equal or above 55°C (Figure 4) indicating that they were all equally heat labile.

Figure 4.

Heat stability of PPO. Purified PPO protein was heated over a temperature range of 25–95°C for 15 or 30 min and then subjected to functional activity analysis. Mean specific activity was determined from duplicate assay replicates. The PPO activity of non-heat-treated samples was assigned 100% active. Relative activity = [specific activity of heat-treated sample/specific activity of non-heat-treated sample] × 100. Error bars represent the standard deviation of the duplicates.

Acute Toxicity Testing of PPO Protein

One of the most challenges in formulating TDS for an acute toxicity study is to remove or reduce concentration of the detergent Thesit® from 0.18% in the starting material (purity-corrected concentration of PPO protein is 1.5 mg/ml) while maintaining PPO activity. Although Thesit® in the starting material is crucial for maintaining PPO activity during long-term storage, its safety profile raises concerns when considered for use in animal dosing studies. According to the safety sheet from Millipore Sigma, the oral LD50 of Thesit® in rats is 1000 mg/kg, but there is no established safety range available for dosing in mice. Therefore, we ran a preliminary dosing study, where we administered Thesit® via oral gavage to mice for 7 consecutive days (repeat dosing) at doses of 10, 100, and 200 mg/kg/day (Supplemental material S1). All doses were well tolerated, with no effects on mortality or significant adverse findings. However, lower body weight and weight gains or weight loss were observed at the 200 mg/kg/day dose. Based on these results, the No Observable Adverse Effect Level (NOAEL) for the study was 100 mg/kg/day. Since the concentration of PPO must remain above 75 mg/ml for the acute study, the NOAEL indicates that the amount of Thesit® must be less than 0.02 mg per mg of PPO in the TDS. Given that the starting material contains 1.2 mg of Thesit® per mg of PPO, this necessitates a reduction of more than 60-fold in the concentration of Thesit® from the starting materials.

Following extensive testing and optimization to reduce Thesit®, the starting material was first diluted more than 12-fold to lower the concentration of Thesit® in the sample to <0.015%. Second, since active PPO exists as a multimer in solution according to our observations by size exclusion chromatography, a 100 kDa cutoff hollow fiber cartridge was used to remove the detergent from the sample while retaining PPO. The majority of Thesit® was removed during the first stage of concentration from 100 L of the diluted sample to 2 L concentrate. In order to formulate at least 75 mg PPO/ml to meet the targeted dose level, the 2 liters of sample were finally concentrated to about 63 ml through a three-step procedure, resulting in the removal of more than 99% of Thesit® from the starting material according to NMR analysis. Following this, the protein concentration, conductivity, and pH were analyzed and presented in Table 3. The concentration of Thesit® was found to be 0.071%, that resulted in a 47.3 mg/kg dose in the acute study, well below the NOAEL determined in the preliminary study. These results demonstrate that the TDS met the requirements for dosing mice at 5000 mg/kg. After the TDS was formulated, the pH of the CDS was adjusted to match that of TDS. Meanwhile, the concentration of Thesit® in VDS was also adjusted to match its concentration in the TDS (Table 3). Those adjustments for the CDS and VDS ensure alignment for all components except proteins.

Table 3.

Summary of the dosing solution characteristics.

Samples	Purity-corrected protein concentration (mg/ml)a	Conductivity (µS/cm)	pH	Thesit® concentration (%, w/w)b
TDS CDS VDS	76.3 87.3 NDc	1439 2979 1389	9.95 9.92 9.50	0.071 0.063 0.066

^aEach value refers to purity-corrected protein concentration. Total protein concentration was determined by Bio-Rad Protein Assay and represents the mean of six replicates.

^bEach value refers to the mean of three replicates. Representative NMR spectra for TDS, CDS and VDS are in the Supplemental material S2.

^cND refers to “Not Detectable.”

Although there was not any hazard identified in the initial core studies, the toxicity of PPO was evaluated in a GLP compliant mice oral acute study following the EPA Guideline OPPTS 870.1100 to comply with regulatory requirements. The study showed that there were no test substance-related effects of PPO protein when administered by oral gavage at a dose of 5000 mg protein/kg body weight in male and female CD-1 mice. All animals survived to the end of the study to their scheduled euthanasia. There were no test substance-related deaths or clinical observations. Additionally, no test substance-related effects on body weights, body weight gain, food consumption, or gross pathology findings were observed. Table 4 shows body weight, body weight gains, food consumption and mortality data. A statistically significant increase in food consumption was noted (p ≤ .01) in TDS (PPO treated) females when compared to VDS (buffer control) females from days 1 to 15. This difference was not considered test substance-related or adverse, as it did not correlate with changes in body weights. Furthermore, there were no statistically significant food consumption changes in TDS males when compared to VDS males. Therefore, NOAEL for PPO protein was determined to be greater than or equal to 5000 mg protein/kg body weight, the highest dose tested.

Table 4.

Summary of the acute toxicity study results for PPO.

Treatment	Mean Body Weights in g (SD) day 1		Mean Body Weights in g (SD) day 15		Mean BW Gains (SD) (Day 1–15)		Mean Food Consumption in g/animal/day (SD) Days 1–15		Mortality
	Male	Female	Male	Female	Male	Female	Male	Female
VDS (Buffer)	36.75 (2.41)	27.32 (2)	36.72 (2.79)	27.5 (1.83)	−0.03 (0.73)	0.18 (0.66)	4.7 (0.7)	4.2 (0.4)	No Mortality
CDS (5000 mg/kg BSA)	35.71 (2.18)	25.84 (2.32)	36.4 (2.16)	27.19 (2.81)	0.69 (0.65)	1.35 (1.05)	4.8 (0.6)	5.1 (1.1)	No Mortality
TDS (5000 mg/kg PPO)	36.17 (2.23)	25.71 (1.53)	36.36 (2.08)	26.45 (2.31)	0.19 (1.34)	0.74 (1.5)	4.5 (0.8)	5.9 (1.5)*	No Mortality

VDS is vehicle (buffer) control, CDS is the protein control (5000 mg BSA/kg of body weight) and TDS is the treatment (5000 mg PPO/kg of body weight).

*A statistically significant increase in food consumption was noted (p≤.01) in TDS females when compared to VDS females from days 1 to 15. This difference was not considered test substance-related or adverse, as it did not correlate with changes in body weights and there were no statistically significant food consumption changes in TDS males when compared to VDS males. The NOAEL for PPO Protein was determined to be greater than or equal to 5000 mg protein/kg body weight, the highest dose tested.

Discussion

PPO Has a History of Safe Use

Transgenic maize, cotton, and soybean conferring tolerance to PPO-inhibiting herbicides have been developed with expression of HemG PPO.¹⁶ This study focuses on the comprehensive safety assessment of PPO using a weight-of-evidence approach. Given the importance of HOSU in assessing the safety of newly expressed proteins, three key components can be considered: safety of the donor organism, familiarity, and historical consumption.

The source organism, E. cloacae, from which the H_N90 PPO gene is derived, is ubiquitously present in various environments such as soil, water, plants, the gastrointestinal tracts of humans and animals, and other organisms.^28–32 It has a well-documented history of safe use in applications such as cloning,^6,7 biodegradation,^35–38 and bioenergy generation.^7,33,34 However, it is important to note that E. cloacae can cause rare opportunistic secondary infections in humans.^42,43

PPO is essential for the biosynthesis of critical compounds like heme and chlorophyll across a wide range of organisms, including bacteria, plants, and animals. PPO proteins are present in various foods, including grains, vegetables, and fruits that are commonly consumed raw, and no health issues have been reported related to their presence. Currently, over 1,000 species have been reported to contain various PPOs, among which 2,726 variants of HemG PPO have been identified,^8,16 with no documented cases of pathogenicity or allergenicity attributed to the protein.

Searching allergen, toxin and all protein databases resulted in no results indicating a potential for adverse effects. This is further validated by direct comparisons to a select group of established probiotics where it is shown the H_N90 PPO shares 63.5% sequence identity with the homolog in L. brevis (GenBank: MEQ6219595.1), 21% sequence identity with the homolog in L. lactis (GenBank: AAK04807.1) and 31% sequence identity with the homolog in B. bacteriovorus (GenBank: CAE80677.1). Notably, L. lactis^60–62 and L. brevis^63,64 are well-established probiotic organisms known for their health benefits, particularly in promoting gut health and enhancing the immune system. These bacteria are naturally present in various fermented foods, such as yogurt, cheese, and kefir, which are commonly consumed by humans. Regular intake of these foods helps maintain a balanced gut microbiota. Besides dietary sources, these beneficial bacteria are available in probiotic supplements, formulated to deliver concentrated amounts of these organisms. In animal husbandry, probiotics containing these strains are often added to feed to support digestive health and overall well-being.^65,66 Additionally, they are found in the environment, including in soil and the gastrointestinal tracts of animals, providing further exposure through natural contact. Together, L. lactis and L. brevis significantly contribute to the health of humans and animals. B. bacteriovorus, a beneficial bacterium, employs predation strategies that have promising applications in water treatment, food preservation, enhancement of industrial processes, and potential combination therapies with bacteriophages and/or antibiotics to combat multi-drug resistant organisms.^67,68 Both humans and animals have been exposed to homologous H_N90 PPO proteins without health concerns. Therefore, H_N90 PPO has a history of safety of use.

The E. Coli -Produced PPO is Equivalent to the Plant-Produced PPO

Due to the membrane association properties of PPO and very low expression level (<1 ppm) in GM crops, it is technically unfeasible to produce large quantities of PPO protein from GM crops. Consequently, the significant amount of H_N90 PPO protein required for safety studies was produced from E. coli.⁴⁴ Therefore, it is essential to demonstrate the equivalence of E. coli -produced PPO protein to plant-produced PPO protein prior to using the E. coli -produced PPO protein for safety assessments.

The equivalence study was conducted using a weight of evidence approach in accordance with Codex and other published guidelines,^20,52,56 which involved comparisons of sequence identity, molecular weight, immunoreactivity, glycosylation, and functional activity. The transgenic plant-produced PPO contains three additional N-terminal amino acids (DAS) derived from an incomplete processing CTP, as previously reported, while the E. coli-produced PPO includes a 6xHis-tag at the N-terminal to facilitate large-scale protein production.⁴⁴

Some eukaryotic proteins undergo post-translational modifications, including glycosylation, through the enzymatic addition of carbohydrate moieties.⁶⁹ The PPO assessed for safety in this publication is derived from E. cloacae, a prokaryotic organism. As prokaryotes generally lack the cellular machinery for glycosylation, PPO is not expected to be glycosylated in its native context. In GM crops, the transgenic PPO is targeted to the chloroplast, an organelle that lacks the glycosylation pathways present in the endoplasmic reticulum and Golgi apparatus of eukaryotic cells. Therefore, glycosylation is not anticipated to occur in PPO produced from transgenic plants, nor is it required for its biological activity or processing.

Nevertheless, in silico analysis of the PPO amino acid sequence revealed several putative glycosylation sites, including a few potential O-linked sites and two N-linked consensus sequences. Given that glycosylation can influence a protein’s physicochemical properties such as stability, solubility, and immunogenicity, glycosylation analysis is not only an important aspect of transgenic protein characterization and safety assessment, but also a critical criterion for establishing the physicochemical and functional equivalence of transgenic proteins derived from different sources.^{26,52,56,70,71} To meet regulatory requirements, glycosylation analysis was conducted using a detection kit. Our results confirmed that PPO produced in both bacterial and plant systems is free of glycosylation.

Mass spectrometry analysis further demonstrated that both plant-produced and microbe-produced H_N90 PPO exhibit no modifications other than incomplete N-terminal processing in the plant-produced PPO. Western blot analysis confirmed equivalent immunoreactivity for both forms of PPO. Due to the enzymatic activity of PPO being very labile during the prolonged and multi-step purification process, the transgenic plant-produced PPO was not purified to homogeneity in order to preserve its activity. However, the determination of protein concentration using Bio-Rad, amino acid analysis (AAA) or other total protein measurement methods can be affected by contaminants. Therefore, the plant-produced PPO concentration was measured using ELISA, a method that is less susceptible to interference from impurities. Given that endogenous plant PPO activity is sensitive to the flumioxazin inhibitor,¹⁶ activity assays for transgenic plant-produced PPO were conducted in the presence of flumioxazin to eliminate any potential activity from endogenous plant PPO. As expected, the activity assays clearly demonstrated that both PPOs from different sources exhibit equivalent functional activity. These results confirmed that the E. coli-produced PPO protein is a suitable surrogate for safety assessment of the transgenic plant produced PPO.

N-Terminal Extensions Pose No Safety Concerns

Due to incomplete chloroplast processing of the CTP, transgenic soybean- and maize-produced PPOs exhibit distinct N-terminal extensions.⁴⁴ The soybean-produced PPO contains three additional N-terminal amino acids, while the maize-produced PPO has 13 additional N-terminal amino acids. Both of these CTP are derived from naturally occurring genes in planta. The three additional amino acids in soybean PPO are present in the chaperonin ClpA/B of Corchorus capsularis and the chaperone protein ClpB3 from Lactuca sativa. Similarly, the 13 N-terminal amino acids in maize PPO originate from chloroplast albino and pale green 6 in Arabidopsis thaliana and also found in casein lytic proteinase B3 in Arabidopsis thaliana. Therefore, these naturally occurring peptides have a history of safe use and are not expected to pose any safety concerns for humans or animals. Bioinformatics analyses further indicated that PPO variants with these N-terminal extensions show no matches to proteins in toxin and allergen databases. Structural predictions also confirmed that these extensions do not affect the structure of the PPO and functional activity assays demonstrated that these N-terminal extensions do not alter PPO functional activity.⁴⁴ Additionally, His-tagged proteins have been shown to be suitable surrogates for non-tagged protein safety assessments, with no associated safety concerns.⁴⁶ Therefore, the His-tagged PPO serves as a suitable surrogate for these PPO variants produced in GM crops, and the safety results can be applied to these homologous PPO variants.

The Weight of Evidence from Core and Supplementary Studies Supports the Safety of PPO

Results regarding the safety of PPO have been generated using a weight of evidence approach. PPO has a documented history of safe use and is not structurally or functionally related to known toxins or allergens. This eliminates the need for further hypothesis-driven evaluations. Nevertheless, to meet international regulatory requirements including those established by Codex Alimentarius and other global authorities,^{20,21,24,70,72} additional studies were conducted to assess the fate of the PPO protein under conditions of heat and enzymatic digestions. Protein stability is a key determinant in the safety assessment of novel proteins. Understanding how PPO responds to heat and digestive enzymes provides critical insight into its potential to remain structurally intact and functionally active under typical exposure conditions. Enzymatic activity following thermal treatment can serve as a proxy for structural integrity, indicating whether the protein retains its native conformation and biological function.^22,23 Conversely, rapid degradation by mammalian digestive enzymes suggests that the protein is unlikely to persist in the gastrointestinal tract, thereby reducing the likelihood of exposure to the intact protein and mitigating potential safety concerns associated with its consumption. Experimental results show that the E. coli-produced PPO is functionally and physiochemically equivalent to plant-produced PPO, is readily digested by mammalian digestive enzymes, and is heat-labile, completely losing its activity at temperatures equal or above 55°C. These findings support the conclusion that PPO is unlikely to remain structurally intact or functionally active following typical food and feed processing and consumption. Furthermore, the presence of a 6xHis-tag on the N-terminal end of the protein had no impact on the susceptibility of PPO to pepsin or pancreatin and heat stability. Therefore, the weight of evidence from these studies supports that PPO is not a hazard to humans and animals. However, the animal study was conducted to further assess the potential toxicity of PPO, ensuring consistency between results of the animal study and the core studies for a membrane-associated protein. NOAEL is defined as the dose that causes no adverse effects in test animals, establishing a safe level of exposure for humans. A protein toxicity test with a dose of 5000 mg PPO/kg body weight showed no adverse impacts on the treated animals, establishing 5000 mg PPO/kg as the NOAEL. This study confirmed that safety study results for a membrane-associated protein are consistent between the animal study and the core study assessments.

This NOAEL, along with estimates of dietary intake of PPO, was used to calculate margins of exposure (MOEs) for PPO in soybean products derived from MON 94115 (i.e., assuming 100% of soybeans consumed are MON 94115 and no PPO is lost during storage, processing, or cooking). Based on soybean consumption of 2.0 g/kg body weight/day,⁷³ the MOE is calculated to be 2.6 x10⁶, which is more than 6 orders of magnitude higher than generous estimates for human exposure to PPO from consuming MON 94115. Moreover, given the low expression of PPO in MON 94115 (average 0.97 mg/kg dw), a 60 kg human would need to consume approximately 3 x10⁵ kg of MON 94115 soybeans in a single day to achieve an intake level similar to the NOAEL from the acute toxicity study (5000 mg/kg). These results indicate that there are no meaningful risks to human health from dietary exposure to PPO protein derived from MON 94115. Similarly, there are no meaningful risks to human and animal health from dietary exposure to PPO protein derived from MON 80616 maize, as the PPO expression level is also low in MON 80616 maize grain (< 1 ppm). Humans generally do not consume whole cottonseed due to the natural toxicity of gossypol in the seed; even when animals eat processed cottonseed, the exposure to cotton PPO poses no meaningful risk to either humans or animals.

The Formulation of Gram Quantities of Active PPO is Achievable

PPO is a membrane-associated protein with hydrophobic characteristics that requires a detergent for purification. Thesit®, a nonionic detergent, has been successfully employed for the purification of PPO from E. coli.⁴⁴ The purified PPO is stored in a buffer composed of 1x PBS, 56 mM L-arginine, 0.1 mM FMN, 20% glycerol, 60 mM L-histidine, 60 mM imidazole, and 0.18% Thesit® at pH 9.1 for long-term storage, which helps maintain its enzymatic activity. However, the presence of detergents and high salt concentrations presents challenges for animal toxicity studies. L-arginine is recognized as a Generally Recognized as Safe (GRAS) substance by the US Food and Drug Administration (CFR) and has been shown to enhance protein solubility, stabilize proteins, suppress aggregation, and reduce viscosity in high-concentration protein formulations.^74,75 We have also demonstrated that the vast majority of Thesit®, due to its relatively small micellar particle size and nonionic nature, can be effectively removed by dialysis during the PPO formulation process. In alignment with these findings, we developed a simplified buffer suitable for animal studies consisting of 5 mM sodium phosphate, 28 mM arginine, 0.1 mM FMN, and pH 10.3,⁴⁴ which also maintains PPO activity for a short-term storage.

To effectively remove detergents and other components from the PPO storage buffer and facilitate the concentration of the protein to high levels, the stored protein sample was initially diluted more than 12-fold, reducing Thesit® to below 0.015%. After extensive testing, a 100 kDa cutoff dialysis cartridge was identified as effective for removing Thesit® while retaining the PPO protein. By combining dilution, concentration and dialysis against an excess volume of optimized buffer, the Thesit® concentration was reduced to 0.071% while achieving a PPO concentration of 76.3 mg/ml. This represents more than 99% reduction in Thesit® from the starting material. Further reduction below 0.071% was not achievable under the tested conditions. According to the Thesit® NOAEL (100 mg/kg/day) that we established, the amount of Thesit® in the TDS administered to animals (47.3 mg/kg/day) is well below this threshold, making it suitable for use in animal studies. To accurately quantify residual Thesit®, we developed an NMR-based quantification method to measure Thesit® concentration. Using this approach, we successfully concentrated 4.8 grams of PPO protein from 8.1 L of the starting material into 63 ml of solution at over 75 mg/ml, with maintaining enzymatic activity. This method also provides valuable insights for the formulation of future membrane-associated proteins.

Conclusions

The safety of PPO, a member of a ubiquitous protein family, has been assessed using a weight-of-evidence approach. These assessments include the core study evaluations, which consist of HOSU assessment, bioinformatic analysis and PPO characterization. The supplementary studies comprise digestibility assessment, heat stability assays, and a mouse dosing study with a NOAEL of 5000 mg/kg. Notably, a 60 kg human would need to consume approximately 3 x10⁵ kg of MON 94115 soybeans in a single day to achieve an intake level similar to the NOAEL from the acute toxicity study. The results from supplementary studies are consistent with those from core studies, demonstrating that PPO poses no hazard. This study aligns with many previously published findings,^24–26 concluding that if no hazard is identified in core studies, then none is expected in supplementary studies. This comprehensive assessment provides valuable insights for future membrane safety evaluations.

Additionally, PPO expressed in different crops exhibits varying N-terminal extensions due to incomplete CTP processing, which are naturally occurring. Bioinformatic analyses further confirm that these extensions, which are up to 13 amino acids long, as seen in the case of transgenic maize PPO, do not share any similarity with known toxins or antigens. Therefore, based on safety results obtained using the His-tag H_N90 PPO, there are no safety concerns associated with PPO variants that have extensions of up to 13 amino acids.

Funding Statement

The author(s) reported there is no funding associated with the work featured in this article.

Disclosure Statement

The authors are employees of Bayer Crop Science.

Supplementary Material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/21645698.2025.2572188.