Aflatoxin reduction with bacteria

3.5.1. Potential of lactic acid bacteria to mitigate aflatoxin risks by binding

In several studies, the usage of LAB as a biocontrol method has been promoted by binding, removing, deactivating or degrading the aflatoxins. Thus, the idea has been that LAB could be applied to milk to mitigate the aflatoxin prevalence and activity at the end of the food chain. This could be the last option as all the other measures have failed to mitigate the risk of the aflatoxins.

Different strains of LAB and especially Lactobacillus have been analyzed for AFM1 binding ability during incubation and for the stability of formed bonds during an external disturbance, mainly washing with PBS.

Binding of AFB1 has been analyzed more widely than AFM1, but as AFM1 is only present in milk and AFB1 is absent in milk, studies have focused on AFM1 binding by LAB in milk. In a review from 2015, a number of binding studies and the binding efficiencies have been summarized and analyzed ¹⁴⁵. Overall, there seems to be no one factor affecting the binding stability and predictability throughout the studies. Depending on conditions, one factor enhancing binding in another study, may be decreasing binding in other study conditions.

Increasing AFM1 binding performance by Lactobacillus strains has been associated with longer incubation time

150–153, non-viability of bacteria ^25,154–156 milk/yoghurt as incubation matrix ^150,157, lower pH ¹⁵⁷ and higher pH

154,156, making the whole process highly reliant on all the applied factors (temperature, strain, treatment, time, concentration of AFM1, concentration of LAB etc). But also, viable Lactobacillus bacteria strains have performed better in binding than non-viable strains ^25,151,154, and it has been shown that the binding effect is immediate, incubation being insignificant for the improved binding efficiency 155,156,158,159.

Lb. plantarum in PBS spiked with 0.15 µg/ml of AFM1,after 15 minutes and 24 hours incubation, bound 5 and 8 % as viable cells, and 13 and 14 % as heat killed cells and then releasing 66 and 69 % and 63 and 54 % of initial bound AFM1155 showing considerably lower binding levels than some other studies. Seven different strains (Lb. plantarum, E. avium, P. pentosaceus, Lb. gasseri, Lb. bulgaricus, Lb. rhamnosus, and B. lactis) were tested for their ability to bind AFM1 as viable cells, and as heat-killed cells. All strains showed higher binding capacity than heat killed.

Lb. acidophilus added to traditionally fermented milk as heat-killed cells bound less (78) %) than viable cells (95 %) after the storage period of 28 days. After a day, heat-killed cells had bound 64 %, when viable cells bound 51 % but the viable cells binding was observed to increase significantly during storage, and apparently due to the multiplication of the bacteria cells present in the traditional yoghurt ¹⁵⁴.

Conflicting results between the studies are observed: the concentration of AFM1 did not have an effect on binding ¹⁵⁷, but then also the concentration of the AFM1 seems to affect the binding potential significantly ^153,160. The concentration of bacteria cells applied could also be a significant factor ^154,160 for overall performance. AFM1

concentration was found to have an effect on the binding efficiency in PBS ¹⁶⁰: lower AFM1 concentration had

a higher binding share, 80 %, with viable Lb. plantarum at 0.05 µg/l and 77 % at 0.1 µg/l, also other strains showed similar pattern (Lb. helveticus, Lb. lactic and S. cereviciae).

Higher bacteria concentration induced higher binding shares, maximum was achieved with 10¹⁰ and minimum with 10⁷¹⁶⁰. Lb. plantarum binding dropped from approximately 80 % to 40 % when bacteria concentration decreased from 10¹⁰ cells to 10⁹ cells in two different AFM1 concentrations. Heat killed Lb. acidophilus added to traditionally fermented milk bound better throughout the storage the spiked AFM1 (0.5 ppb) in higher bacteria concentration at 9¹⁰ cfu/ml (61, 94, 99 %), than at 7¹⁰ cfu/ml (51, 91, 95 %) ¹⁵⁴.

Lb. bulgaricus, Lb. rhamnosus and Bifidobacterium lactis, all heat killed, were tested for their ability to bind AFM1

in UHT skimmed milk, at 4 °C and 37 °C ¹⁵⁵. Two strains (Lb. bulgaricus and Lb. rhamnosus) showing higher binding shares (33 % and 25 %) at a higher temperature and only B. lactis binding slightly better at 4 °C (38 %).

Overall, the binding performance of all three strains was significantly less in milk than in PBS.

Three concentrations of AFM1 (0.1, 0.5 and 0.75 µg/l) were added to skimmed milk with lyophilized Lb.

acidophilus LA-5, and with a mixture of Lb. acidophilus LA-5 and yoghurt bacteria (YoFlex: S. thermophiles and Lb. delbrueckii spp. bulgaricus) and the binding was observed after 21 days storage at 4°C ¹⁵³. The binding share increased with increasing AFM1 concentrations, overall, being approximately 90 %.

AFM1 binding has been reported being a relatively immediate phenomenon with no difference in incubation time 154,155,158 but some strains experienced increased levels of binding after more prolonged incubation ¹⁵⁹. A mixture of heat-killed strains (Lb. delbrueckii spp. bulgaricus, Lb. rhamnosus and B. lactis), bound approximately 11 % of added AFM1 in UHT skimmed milk, showing no difference between the 30- and 60-minutes incubation

158. During fermentation storage for 28 days, Lb. acidophilus binding increased from day one share of 51 % up to 95 %, which might have been due to increasing bacteria population in the product ¹⁵⁴.

Some studies have observed 100 % binding efficiency with different strains; Lactobacillus helveticus, mixture with S. cerevisiae and traditional yoghurt fermenting bacteria with Lb. acidophilus have been recorded to bind added AFM1 at the level of 99 - 100 % 154,158,160.

3.5.2. Stability of the formed bond complex and bioaccessibility

For biocontrol purpose, not only the binding efficiency is relevant, but also the stability of the formed bond.

A weak bond releasing the AFM1 would not be useful despite the initial binding efficiency. If the binding phenomenon is only temporary, the suitability as a biocontrol method will be impaired due to its unstable nature and risks that may arise in different uncontrollable conditions. Several studies have found different levels of release of AFM1 from formed AFM1 and LAB bonds under different conditions.

The binding of AFM1 by LAB has been proven to be an unstable structure as washing disturbs the formed bond of LAB and AFM1 by releasing the AFM1155,156,160. In a process described as washing, the stability of viable

Lb. plantarum and AFM1 binding was tested in three sequential treatments releasing, 50 %, 9 % and 1 % from initial concentration ¹⁶⁰. Bacteria mixtures released even up to 68 % of initially bound AFM1160, seven different viable and heat-killed strains released 40 – 87 % ¹⁵⁵, questioning the potential of the LAB as a biocontrol to the aflatoxin problem as a binder. Some very low shares were reported to be released in washing: varying amounts by viable strains ¹⁶¹, Lb. acidophilus, Lb. reuteri, Lb. rhamnosus, Lb. johnsonii and B. bifidum released 1 – 5 % of initially bound AFM1159.

The binding potential has been suggested to be based on physical adhesion to bacteria cell wall components such as polysaccharides and peptidoglycans, which are denatured during heat treatment, and at least in theory, increases the binding effect with increased hydrophobic nature ¹⁵⁵. It has also been suggested that binding strength is based on covalent binding between bacteria components and AFM1, or reduced bioavailability caused by the degradation of AFM1 by LAB metabolites ¹⁶². The weak binding structure results to weak stability where repeated PBS washing releases the bound AFM1, suggesting a non-covalent bond on the hydrophobic sites of bacteria surface ^155,163.

Bioaccessibility of aflatoxins in the presence of suggested binding components have been assessed in a few studies. The relative bioaccessibility of AFM1 was reduced after three hours from the initial 22 % to 45 % in a digestive stimulation process for milk ¹⁵⁹. In another study, bioaccessibility of AFM1 was assessed with six strains in a digestion model that resulted in a reduction of bioaccessibility from the initial binding of 80 – 83 % down to 15 – 31 % ¹⁶⁴.

3.5.3. Aflatoxin reduction through biodegradation

The possibility for aflatoxin control in foods is not only in mechanical binding, but biodegradation of aflatoxin by components produced by bacteria and other external factors and components have been considered, one such option is through aflatoxin detoxification interactions. Aflatoxin accumulation in nature has been concluded not to occur, thus a conclusion towards biodegradation by biological components present in nature can be drawn.

There is speculation if the degradation of aflatoxins could be a possible pathway in the phenomenon of aflatoxin “reduction”. This can be achieved by two suggested possibilities; physical binding or cell absorption followed by actual, physical biodegradation ¹⁶⁵. In the AFB1 reduction study with salt-tolerant Candida versatilis CGMCC 3790, authors were able to speculate about these possibilities with identified metabolites from AFB1

degradation process ¹⁶⁵.

Non-pathogenic Escherichia coli has been found to degrade AFB1, based on the detected metabolites of bio-transformed AFB1166. Extracellular components secreted by the cells were concluded to be more efficient in biodegrading, than mere intercellular extracts. Also, degradation was mediated by heat-resistant proteins. The formed metabolites of AFB1 were suggested as being less toxic based on in vivo experiments. The study compared with the culture supernatant of Bacillus pumilus E-1-1-1, cells and extracts showed that the AFM1

detoxification mechanisms were more of degradation than binding or absorption ¹⁶⁷. Other findings highlight

the cell-free supernatant of Bacillus velezensis DY3108 potential and in cytotoxicity assay biodegraded products displayed significantly lower levels of cytotoxic effects than AFB1168. Similar results were obtained with the crude enzyme solution of Bacillus licheniformis (BL010) with detected biotransformation products resulting in conclusions of biodegradation having occurred ¹⁶⁹.