Heat‐killed Mycolicibacterium aurum Aogashima: An environmental nonpathogenic actinobacteria under development as a safe novel food ingredient

Abstract Over the last few decades, a wealth of evidence has formed the basis for “the Old Friends hypothesis” suggesting that, in contrast to the past, increasingly people are living in environments with limited and less diverse microbial exposure, with potential consequences for their health. Hence, including safe live or heat‐killed microbes in the diet may be beneficial in promoting and maintaining human health. In order to assess the safety of microbes beyond the current use of standardized cultures and probiotic supplements, new approaches are being developed. Here, we present evidence for the safety of heat‐killed Mycolicibacterium aurum Aogashima as a novel food, utilizing the decision tree approach developed by Pariza and colleagues (2015). We provide evidence that the genome of M. aurum Aogashima is free of (1) genetic elements associated with pathogenicity or toxigenicity, (2) transferable antibiotic resistance gene DNA, and (3) genes coding for antibiotics used in human or veterinary medicine. Moreover, a 90‐day oral toxicity study in rats showed that (4) the no observed adverse effect level (NOAEL) was the highest concentration tested, namely 2000 μg/kg BW/day. We conclude that oral consumption of heat‐killed M. aurum Aogashima is safe and warrants further evaluation as a novel food ingredient.

. According to the revised definition of the Food and Agriculture Organization (FAO)/World Health Organization (WHO), as well as in the public perception, probiotics are nonpathogenic live microorganisms that, when administered in adequate amounts, confer a health benefit to the host, such as improvement in metabolism and intestinal flora and modulation of immune functions (Aponte et al., 2020;FAO & WHO, 2002;Hill et al., 2014;Wilkins & Sequoia, 2017). The probiotic market is growing rapidly, buoyed by both foods and supplements intended to enhance wellness in healthy individuals, and by preparations for the dietary management of diseases (Grumet et al., 2020).
NTB mycobacteria, which include the mycolicibacteria, are not a permanent constituent of the microbiome, but because they have been regularly encountered in the diet and the environment, there is evidence for their evolutionary adaptedness (Rook, 2010). Interestingly, akin to the bacteria which make up the gut and skin microbiota, researchers have now identified communities of several of these operational mycobacterial taxonomic units in the oral cavity of healthy individuals (Macovei et al., 2015). This is presumably a reflection of the significant exposure to environmental NTB mycobacteria by this route. The extent to which NTB mycobacteria such as mycolicibacteria hold promise, like probiotics, for influencing human well-being is the subject of ongoing research. Nevertheless, "the Old Friends hypothesis" makes a case for their benefit to human health as revealed by the drastic reduction of exposure to saprophytic environmental NTB mycobacteria in modern living conditions (Flandroy et al., 2018;Lowry et al., 2016;Rook et al., 2004Rook et al., , 2013. Until recently, the assumption has been that probiotics should be viable to exert positive effects. Instead, there is now increasing evidence to show that nonviable probiotics maintain their healthpromoting benefits and a new term "postbiotic" has been coined to indicate preparations of inanimate microorganisms and/or their components that confer a health benefit to the host (Aguilar-Toalá et al., 2018;Barros et al., 2020;Seminen et al., 2021;Taverniti & Guglielmetti, 2011). From a commercial standpoint, the use of nonviable bacteria has several advantages, including easing the challenges associated with product storage to maintain viability, reduction of safety concerns arising from horizontal virulence gene transfer from pathogenic bacteria, and the ability to deliver exact numbers of microorganisms per dose. In light of these issues, nonviable bacteria are now under consideration as novel food ingredients. In this report, we present evidence for the safety of heat-killed Mycolicibacterium aurum Aogashima as a novel food ingredient. This is an environmental saprophytic organism which may not have the documented history of safe use that food-associated probiotics have, but nonetheless, is likely to have been evolutionarily present in the diet, through exposure to untreated and even treated water supplies. The safety of this novel food was determined using the decision tree approach developed by Pariza and colleagues which relies on assessment of lack of allergenicity risk, confirmation that resistance to various antimicrobials is intrinsic and nontransmissible and that no harmful effects are detected in standard toxicology testing (Pariza et al., 2015). Our data support the conclusion that heat-killed M. aurum Aogashima is safe as a novel food ingredient. Good Laboratory Practice (GLP) principles. The organism is grown in a bioreactor of either five or twenty-five liters. Once an appropriate biomass is reached, the bacteria are recovered by centrifugation and resuspension in water, before heat inactivation at 121ºC for ≥20min.

F I G U R E 1
The resulting M. aurum Aogashima biomass is then further diluted with water and stored prior to use.

| Safety evaluation process
The safety of M. aurum Aogashima was assessed based on the decision tree approach developed by Pariza and colleagues (2015). A flow chart describing the steps is depicted in Figure 1.  (OECD, 1998). At the end of the treatment period, all animals were subjected to a gross necropsy, internal organs were weighed, and organ tissues from the control and high dose animals were examined microscopically.

| Genome sequencing and analysis
DNA was extracted from a culture of M. aurum Aogashima as described in Amaro et al., (2008). Genome sequencing was performed using an Illumina MiSeq instrument, as previously described (Sangal et al., 2015). The genomes were assembled into contigs using SPAdes 3.9.0 with a kmer length of 127 and subsequently annotated using the Rapid Annotation of microbial genomes Subsystems Technology (RAST) server (Aziz et al., 2008;Bankevich et al., 2012).

| Antimicrobial resistance gene assessment
The presence of genes coding for antibiotic resistance (AMR) was assessed in the genome of M. aurum Aogashima. The whole genomic sequence was compared against ResFinder databases and the Comprehensive Antibiotic Resistance Database (CARD) (Alcock et al., 2020;Zankari et al., 2012). Briefly, in silico genome analysis for the AMR genes was carried out by ResFinder 3.2 webserver which encompasses 15-drug classes in its database: aminoglycoside, beta-lactam, colistin, a fluoroquinolone, fosfomycin, fusidic acid, glycopeptide, macrolide-lincosamide-streptogramin B, nitroimidazole, oxazolidinone, phenicol, rifampicin, sulphonamide, tetracycline, and trimethoprim (Zankari et al., 2012). The percent identity and perfect alignment were set at 70% and 60%, respectively. The minimum length or the number of nucleotides that must overlap a resistant gene to count as a hit was set at the default of 60%.
The genome sequence of strain M. aurum Aogashima was interrogated for the presence of AMR genes based on CARD and using Resistance Gene Identifier (RGI) software for resistome analysis and prediction (Alcock et al., 2020). Each predicted AMR gene was manually mapped and annotated using the SEED and the RAST server (Aziz et al., 2012). Protein domains of AMR genes were confirmed after comparison with those available in the Conserved Domains Database (CDD) of NCBI (Marchler-Bauer et al., 2015). Any hits were reported and analyzed.

| Pathogenic gene clusters and virulence factors assessment
The draft genome sequence of M. aurum Aogashima was screened for pathogenic island and virulence factors using the Virulence Factor database (VFDB) (Liu et al., 2019). Experimentally validated virulence factors of major medically important bacterial pathogens belonging to 24 genera were considered. In addition, predicted coding sequences were identified using the GLIMMER3 system (system for finding genes in microbial DNA) prior to using the VFanalyzer

tool (Virulence Factor analyzer tool). Lastly, blastp and Conserved
Domain tools of NCBI were used to identify the virulence factors associated amino acid sequences of M. aurum and determine their functional similarity with those of Mycobacterium tuberculosis H37Rv. The established threshold of 60% for functional protein similarity was adopted (Addou et al., 2009).

| Statistical Analysis
In vivo data were analyzed using Graph Pad Prism to give group mean values and standard error. Where appropriate and within each sex, one-way ANOVA followed by Sidak's multiple comparisons test was used to determine statistical differences upon comparison of groups receiving different doses of heat-killed M. aurum Aogashima versus the control group. food intake revealed no significant differences between groups.

| Subchronic oral toxicity study
All groups gained a similar amount of weight ( Figure 2) and ate a similar amount of food (data not shown) when compared to control groups. Animals showed no evidence for treatment-related neurotoxicity based on functional observation battery assessments.
Indeed, there were no effects on functional arena observations or on grip strength or motor activity and sensorimotor responses to visual, acoustic, and proprioceptive stimuli (data not shown). At the end of the treatment period, all animals were subjected to a gross necropsy where organs were weighed and examined macroscopically. We detected no effect on organ weights in the male groups regardless of treatment. In the female group, we observed only a significant decrease in the liver weight and only in the group receiving 20 μg/kg/day (Table 1)  Similarly, levels in control female rats (163.7 ± 4.5) were comparable to levels in rats receiving 20 μg/Kg/day (179.8 ± 6.7) but significantly lower than those in rat receiving 200 and 2000 μg/ Kg/day (191.7 ± 6 and 190.3 ± 5.2, respectively). As glucose levels were already higher than normal (106-184 in males and 89-163 mg/dl in females) in the control groups and rats were not fasted overnight prior to blood sampling, we consider these may be normal biological variations due to food consumption and circadian rhythms.
Hematological parameters were also assessed. There were no statistically significant changes in total white blood cell and immune cell population specific counts in the female groups, aside from a significant decrease in monocytes in the group receiving 200 μg/Kg/day (Table 3). However, when cell population percentages were calculated, no significant differences were detected in any cell populations, including monocytes, regardless of dose received (Table 4). In males, we observed a significant decrease in white blood cell, neutrophils, and lymphocytes absolute counts only in the group receiving 200 μg/Kg/day (Table 3). However, we found no evidence for changes in the percentages of these populations or in any other immune cell populations (Table 4). Moreover, values remained within normal range (white blood cells 1.98-11.06 x 10 3 /μl; neutrophils .33-1.98 10 3 /μl). Due to the lack of a dose relationship, given no differences were detected in the highest dose

| Allergenicity
The allergenicity potential of M. aurum Aogashima was assessed by an innovative 3D-modeling-based analysis, using the AllerCatPro database. Only fifteen potentially allergenic protein sequences were detected with linear sequence window identity above the thresholds of 35% (Table 5). Most of the detected amino acids sequences in the genome of M. aurum Aogashima corresponded to allergenic proteins previously found in fungi (56%), while 26%, 15%, and 3% of the predicted proteins belong to foods, arthropods, and mammals (just one sequence), respectively. None of these proteins showed a 3D epitope identity, and therefore, we concluded that there was no evidence for allergenicity following consumption of heat-killed M.

| Antimicrobial resistance gene assessment
We found no hits between the genome of M. aurum Aogashima and the AMR genes included in ResFinder databases. Instead, 3 hits above 70% identity: rbpA (RbpA bacterial RNA polymerase-binding protein), mtrA (resistance-nodulation-cell division antibiotic efflux pump), and murA transferase (Mycobacterium tuberculosis intrinsic murA conferring resistance to fosfomycin) were reported using CARD webserver. These genes confer resistance to rifampicin, penam, and fosfomycin, respectively. However, as reported in Table 6, all hits were below the general reference value for gene homology (97%).
Furthermore, these genes have been reported to be widely present in the Mycobacteriaceae and, therefore, not surprisingly also in the M. aurum type strain DSM 43999 T (Table 6).

| Pathogenic gene clusters and virulence factors assessment
The whole genome sequence of M. aurum Aogashima was screened to identify genetic element sequences that encode for virulence factors or protein toxins. We found no evidence for pathogenic islands.
Screening of the genome of M. aurum Aogashima for all known virulence factors associated genes showed that most of the predicted genes were found in nonpathogenic or commensal bacteria and are involved in host interaction, survival, and maintenance of basic functions (Table 7). As shown in Table 7, several proteins with experimentally verified virulence factors were present in the genome of M.
aurum Aogashima but their amino acid sequence similarity with that of the pathogenic M. tuberculosis is below the 60% cutoff value for functional homology (Table 7). and mucosal surfaces as well as regular microbial contact in the air, surfaces, and in food and beverages (Rook, 2010). Until very recently, diet provided the most exposure through raw, minimally processed or fermented foods and beverages and through untreated water. A link between consumption of live microbes-such as those found in fermented food-and health has been reported in both intervention and associative studies as well as randomized controlled trials (Marco et al., 2017;Sanlier et al., 2019). As health evidence is mounting, there have been calls to include recommendations for the con- Interestingly, the organism under study in this report, heat-killed M.

| D ISCUSS I ON
aurum Aogashima, may fall within the definition of postbiotic, should a health benefit for this preparation be shown in separately presented studies. The purpose of the work described here, however, is solely to present and assess the evidence for the safe use of heatkilled M. aurum Aogashima as a novel food ingredient.
This environmental saprophytic organism is likely to have been long present in the diet as a harmless water contaminant (Falkinham et al., 2001;Le Dantec et al., 2002a, 2002bVaerewijck et al., 2005).

Safety of heat-killed M. aurum Aogashima as a novel food ingredient
was assessed according to the decision tree approach developed by Pariza and colleagues (2015). The interest in expanding the number of microbes being considered as novel food, beyond the current standardized cultures and probiotics supplements, has driven a new approach to assess safety. This new framework is also pertinent to those cultures that are perceived to lack an established history of safe use for their intended application. We provide evidence that

Genetic elements associated with pathogenicity or toxigenicity
were investigated by extensive in silico analysis and showed no evidence of pathogen-specific virulence factors in M. aurum Aogashima.
Indeed, the virulence factors associated genes identified were common to both pathogenic and nonpathogenic and commensal bacteria and associated with highly conserved functions such as amino acid and purine metabolism and the catabolism of cholesterol and were not located on pathogenic island (Niu et al., 2013). Hence, these highly conserved coding sequences are not considered appropriate markers of pathogenicity of M. aurum Aogashima. In this context, the presence of secreted protein associated genes (e.g., fbpA) is expected because they play a fundamental role in cell envelope maintenance (Belisle et al., 1997). The same can be said with respect to the genes ptpA and ptpB which are widely distributed among pathogenic and nonpathogenic mycobacterial species and also found in the genomes of other prokaryotes, including Lactobacillus spp. (Altermann et al., 2005;Boekhorst et al., 2006). It should also be Antimicrobial resistance gene assessment was made by screening the genome using both ResFinder and CARD webservers to ensure coverage of all AMR determinants (i.e., acquired resistance genes, resistant mutations of housekeeping genes, efflux overexpression, etc.), drug targets, antibiotic molecules and drug classes, and the molecular mechanisms of resistance (Alcock et al., 2020;Zankari et al., 2012). We found no evidence for any resistance genes associated with the most common antimicrobial compounds of concern in food (namely, Ampicillin, Chloramphenicol, Kanamycin, Streptomycin; Erythromycin, Gentamycin, Tetracyclin, Vanomycin, and Lincomycin). We did, however, detect similarities with rbpA, mtrA, and murA. While these genes are known to confer resistance to rifampicin, penam, and fosfomycin, respectively, their identity values were close to, but still below, the cutoff of 97% homology.
Moreover, these genes are commonly present in mycobacteria as they are likely involved in essential cell functions (Maitra et al., 2019;Newell et al., 2006). Finally, there is no evidence for transferability. Hence, the absence of significant resistance genes in M. aurum TA B L E 6 AMR genes detected in the genome sequence of M. aurum Aogashima and its relative M. aurum type strain DSM 43999 T with an identity value ≥70%  (given no differences were detected in the highest dose groups which received doses 10 times of those where differences were observed), we consider these differences part of normal biological variation rather than any effect of consumption of heat-killed M. aurum Aogashima.
Based on the findings of the work and analysis described here, our conclusion is that the use of heat-killed M. aurum Aogashima in food products is safe and that it is suitable for being evaluated as a novel food ingredient.

ACK N OWLED G M ENT
We thank Professor Hans-Peter Klenk and his colleagues for assistance with the work conducted at the University of Newcastle (UK).
We also wish to thank Roberto Suarez of Pen & Tec Consulting for critically reviewing the manuscript.

CO N FLI C T O F I NTE R E S T
TD is a senior executive and holds stock in Aurum Switzerland AG.
IN has no conflict of interest to declare.

D at a Av a i l a b i l it y St ate m e nt
The data that support the findings of this study are available from the corresponding author upon reasonable request.