A heterotrophically grown, unextracted AURA biomass powder (CCAP 4087/2) was provided to Minnesota Valley Testing Laboratory (MVTL, New Ulm, MN, USA) by Alltech Inc. (Nicholasville, KY, USA). The analytical com­position of the biomass product was determined in compliance with current Good Laboratory Practices (GLP) guidelines (21 CFR part 58): crude protein (AOAC 990.03 [29] ), crude fat (AOAC 954.02 [30] ), Fatty acid composition (AOAC 996.06 [31] ), moisture (AOAC 930.15 [32] ), and ash (AOAC 942.05 [33] ).

A purified monoculture of metabolically active culture was supplied to the Culture Collection of Algae at the University of Texas (Austin, Texas, USA) by Alltech Inc. (Nicholasville, KY, USA) for identification and characterisation within the College of Natural Sciences (University of Texas at Austin, College of Natural Sciences, Austin, TX, USA). Colonies were isolated and multiplied in Medium-H for the following analyses: 18S rRNA identification, Minimum inhibitory concentration and polyketide analysis (Sections 2.3, 2.4 and 2.5). DNA was extracted from the micro-organism using a modification of the JGI protocol for CTAB mediated bacterial genomic DNA isolation (outlined in [34] ) and PCR amplification was performed with universal 18S primer sets. Reading produced a partial 18S sequence which was compared to the NCBI BLAST database to determine sequence similarity.

Compounds, extraction and bacterial strains

Whole cells were selected during mid-exponential phase and concentrated through centrifugation. The growth-medium supernatant was separated and stored for later solid-phase extraction while the remaining cells were prepared for solvent extraction. Solvent extraction involved repeated suspension in cold acetone followed by centrifugation and decantation to remove pigmentation. Remaining cellular mass was then extracted with four methanol washes before being homogenized in 80% 1-propanol/ water solution. Homogenate underwent centrifugation to separate cellular material and the subsequent supernatant joined the methanol extracts. This homogenization process was repeated a second time with the 80% 1-propanol/water solution. The extract mixture was evaporated to remove solvent and resuspended in 2 ml of water. 2 ml of ethyl acetate was used to de-fat this mixture; the aqueous phase was de-fatted a total of four times and again dried. In order to elute the desired analytes, cellular extract was suspended in 10 ml of 25% 1-propanol/water and passed through a chromatography cartridge, washed with 45% 1-propanol/water solution, eluted in 80% 1-propanol/water solution and dried. 30 ml of growth-medium supernatant was mixed with 10 ml of 25% 1-propanol/water solution and the same solid-phase extraction process was followed. The resulting dried cellular extract and supernatants were suspended in 5% methanol/water solution for use in MIC analysis [35]. DH5a E. coli culture was used as the reference culture for the MIC analyses.

Minimum inhibitory concentration determination

The experiment was carried out in accordance with CLSI M02-A12 guidelines using the Disc diffusion method [36]. Whole cells or extracts were tested as follows.

Nine discs were placed onto sterile LB agar plates to accommodate the cell extracts. Eight discs were placed around the plate and one disc was applied to the center of the plate. The central disc received a dose of 5% Methanol solution as a negative control. Four discs received concentrations of the test solutions (Table 1) while the remaining four discs had positive controls applied. For analysis with

whole cells, the four test discs were replaced by spotting four 10 µl drops of dense AURA culture directly onto the agar). After test and control solutions had been applied and the plates allowed to dry, E. coli (DH5a) culture was inoculated onto the agar by means of a misting nebulizer. Duplicate plates were incubated at 37˚C for 12 - 18 hours. Observation of plates established if zones of clearance were present or absent.

The following bacterial stains were selected as per VICH GL36 (R) guidelines [28] and procured from the American Type Culture Collection (Manassas, VA, USA) (Table 2).

Minimum inhibitory concentration determination

The experiments were carried out in accordance with CLSI M07-A9 guidelines using the Broth macro-dilution method at ABC Laboratories (Gainesville, FL, USA) [37]. Each product was tested separately with each microorganism under aerobic, anaerobic or microaerobic growth conditions as appropriate. Inocula were initially prepared on solid media, Trypticase Soy Agar (TSA) under aerobic conditions for E. coli and E. faecalis, TSA with 5% Sheep blood (SBA) under anaerobic organisms for B. fragilis, B. longum, C. aerofaciens, F. canifelinum, P. anaerobius and C. difficile while L. brevis was prepared on de Mann, Rogosa Sharpe (MRS) agar under microaerobic conditions. Colonies were selected and suspended in sterile 0.85% saline in order to obtain optical densities of 2.0 McFarland prior to the experiment.

Aerobic analysis

Serial dilutions (1:1) (n = 13) of algal powder and algal oil suspensions were separately prepared in Mueller Hinton Broth (MHB) to obtain 1% down to 0.0002441406% test concentrations. Positive and negative controls were similarly prepared in MHB. 10 µl volumes were inoculated into each broth suspension to form an initial concentration of ca. 1 × 10⁵ CFU/ml. Dilution tubes were incubated aerobically at 35˚C ± 2˚C for 18 - 20 hours.

Anaerobic analysis

Serial dilutions (1:1) (n = 13) of algal powder and algal oil suspensions were separately prepared in Brucella Broth with Hemin and Vitamin K¹ to obtain 1% down to 0.0002441406% test concentrations. Positive and negative controls were similarly prepared in supplemented Brucella Broth. 10 µl volumes were inoculated

into each broth suspension to form an initial concentration of ca. 1 × 10⁵ CFU/ml. Dilution tubes were incubated anaerobically at 36˚C ± 1˚C for 46 - 48 hours (B. fragilis and B. longum) or at 36˚C ± 1˚C for 70 - 72 hours (C. difficile, C. aerofaciens, F. canifelinum and P. anaerobius).

Microaerobic analysis

Serial dilutions (1:1) (n = 13) of algal powder and algal oil suspensions were separately prepared in MRS Broth to obtain 1% down to 0.0002441406% test concentrations. Positive and negative controls were similarly prepared in MRS Broth. 10 µl volumes were inoculated into each broth suspension to form an initial concentration of ca. 3 × 10⁵ CFU/ml. Dilution tubes were incubated microaerobically at 30˚C ± 2˚C for 46 - 48 hours.

Following incubation, tubes containing MHB, supplemented Brucella Broth and MRS Broth were examined and scored positive or negative for turbidity. The MIC was determined from the lowest concentration of test material that acted to completely inhibit bacterial growth.

Polyketide analysis

The presence of “Golden algae” polyketides was carried out by LC/MS metabolic fingerprinting. Whole cells were selected during mid-exponential phase and concentrated through centrifugation. Cellular extracts and supernatant extracts were prepared as per [35]. Extract fractions were suspended in 5% methanol solution before being diluted to 1 µg/µl for LC/MS analysis. Spectra were collated and peaks in the 900 - 1200 m/z regions were compared to peaks for known polyketides.

Domoic acid and epi-domoic acid analysis

Unextracted Aurantiochytrium limacinum (CCAP 4087/2) biomass (produced by Alltech Inc. KY, USA) was supplied to the Marine Institute (Galway, Ireland). Samples were prepared in 50:50 methanol /water solution and extracted through ultrasonication for 15 minutes. HPLC-DAD and UHPLC-DAD analysis for the presence of domoic acid toxins was carried out according to in-house protocols.

These analyses were carried out by Minnesota Valley Testing Laboratories Inc (Minnesota, USA) in compliance with current Good Laboratory Practices (GLP) guidelines (21 CFR part 58) according to the methods indicated in Table 3.

Proximate analysis confirmed the oleaginous nature of the unextracted Aurantiochytrium biomass with an average 72.85% crude fat in the sample (Table 4). Percentages of crude protein, moisture and ash are all low (11.44%, 3.10% and 3.39% respectively), indicating significant conversion of substrate to desirable fatty acids. Table 5 reveals the fatty acid profile and the lipid composition. The major components are palmitic acid (average 36.35% w/w) and docosahexaenoic acid (average 18.17% w/w). The total polyunsaturated fat content in the AURA biomass is 18.83% w/w of which DHA constitutes the greatest proportion.

The partial 18S sequence assembled from the genetic identification can be seen in Figure 1. When compared to the NCBI BLAST database, the test strain was determined to be most comparable to the following six protists; Aurantiochytrium Sp. ST-2012 clone BJ61 (Genbank: JQ982491.1), Aurantiochytrium Sp. TF81 (Genbank: KM023695.1), Aurantiochytrium Sp. TF23 (Genbank: KM023689.1), Aurantiochytrium Sp. LY-2012 isolate PKU#Sed1 (Genbank: JX847370.1), Aurantiochytrium Sp. LY-2012 isolate PKU#MN7 (Genbank: JX847363.1), Aurantiochytrium Sp.

9.2669 ISU—internal method code for Iowa State University; Chromagar/Difco BBL—proprietary enumeration method on chromogenic agar.

Results are reported in terms of weight of ingredient.

isolate SL1101 (Genbank: JN986842.1). Furthermore, taking into account the fatty acid profile, the protist reveals itself to be similar to Aurantiochytrium sp. as determined by [51].

As revealed by the results in Table 6, uninhibited growth of E. coli (DH5a) was observed at all test application points. Concurrently, inhibition was observed in the presence of the positive control organisms while no inhibition was observed with the negative controls. These results indicate that at low concentrations, Aurantiochytrium cells and cellular extracts thereof pose no observable antimicrobial threat to the reference organism through DHA production or the presence of secondary metabolites.

The experimental results manifested in Table 7 and Table 8 display the turbidity scores for each microorganism against the respective concentration of test material. Turbidity was evident in test suspensions between 1% and 0.0625% concentrations (owing to the presence of the test materials) prior to incubation and was noted to have increased following incubation. All 13 dilution tubes for each microorganism and test material combination were positively scored following incubation, indicating that none of the nine intestinal microorganisms were inhibited by DHA-rich algae powder nor the extracted oil at concentrations up to at least 0.03125% (owing to the initial turbidity at concentrations above 0.03125%).

LC/MS spectra for both the cell fraction extract and the supernatant fraction extracts (Figure 2 and Figure 3) were compared to the Golden Algae spectra in

(+) indicates growth around the treatment point; (−) indicates inhibition of growth around the treatment point.

+ indicates turbidity (growth) in the culture medium; − indicates an absence of turbidity (no growth) in the culture medium.

+ indicates turbidity (growth) in the culture medium; − indicates an absence of turbidity (no growth) in the culture medium.

[35]. Polyketide prymnesinspectral peaks, based on spiked controls, would have been evident at 985.89²⁺ m/z and 1141.44²⁺ m/z for Prym1 or 919.88²⁺ m/z and 1141.44²⁺ m/z for Prym2. While phenotypically similar, Aurantiochytrium limacinum is phylogenetically distant from golden algae and proved to have a dissimilar metabolic fingerprint when compared to the spectra of polyketide producing Prymnesium parvum. None of the spectral peaks aligned with known polyketides, indicating that this organism is unlikely to pose a threat in terms of production of such algal toxins. Moreover, as evidenced by the additional analyses for the presence of Domoic and epi-domoic acids in unextracted AURAbiomass samples (Table 9); Aurantiochytrium limacinum has not been found to produce any of the frequently observed marine algal toxins.

The results of the further assessment for heavy metals, pesticides and other undesirables from industrial fermentation can be seen in Table 10 and Table 11. These analyses are significant in that they demonstrate that the Aurantiochytrium

cfu = colony forming units; ND = not detected; meq = milliequivalent; ppb = parts per billion; ppm = parts per million; *Assuming ND = 0.

DDD = Dichlorodiphenyldichloroethane; DDE = Dichlorodiphenyldichloroethylene; DDT = dichlorodiphenyltrichloroethane.

production process itself is safe and fit for purpose. Industrial scale heterotrophic fermentation did not show any indications of elevated heavy metal accumulation, hygiene indicator organisms and microbial pathogens were all absent and mycotoxin contaminant levels were below the limits of detection. Similarly, pesticide residue levels were below the limits of detection for these analytical methods. Copper and selenium levels were shown to be low, which is pertinent as AURA is intended to be used as a DHA rich feed material for food producing animals, many of which can be sensitive to the presence of high concentrations of these minerals.