Phytic Acid Degrading Lactic Acid Bacteria in Tef-Injera Fermentation

International Journal of Food Microbiology 190 (2014) 54–60

Contents lists available at ScienceDirect

International Journal of Food Microbiology

journal homepage: www.elsevier.com/locate/ijfoodmicro

Phytic acid degrading lactic acid bacteria in tef-injera fermentation

Maren M. Fischer a,⁎, Ines M. Egli a,1, Isabelle Aeberli a, Richard F. Hurrell a, Leo Meile b a Laboratory of Human Nutrition, Institute of Food, Nutrition and Health, ETH Zurich, Schmelzbergstrasse 7, 8092 Zurich, Switzerland b Laboratory of Food Biotechnology, Institute of Food, Nutrition and Health, ETH Zurich, Schmelzbergstrasse 7, 8092 Zurich, Switzerland article info abstract

Article history: Ethiopian injera, a soft pancake, baked from fermented batter, is preferentially prepared from tef (Eragrostis tef) Received 14 April 2014 ﬂour. The phytic acid (PA) content of tef is high and is only partly degraded during the fermentation step. PA Received in revised form 20 July 2014 chelates with iron and zinc in the human digestive tract and strongly inhibits their absorption. With the aim to Accepted 12 August 2014 formulate a starter culture that would substantially degrade PA during injera preparation, we assessed the poten- Available online 20 August 2014 tial of microorganisms isolated from Ethiopian household-tef fermentations to degrade PA. Lactic acid bacteria

Keywords: (LAB) were found to be among the dominating microorganisms. Seventy-six isolates from thirteen different tef Phytic acid fermentations were analyzed for phytase activity and thirteen different isolates of seven different species were Iron detected to be positive in a phytase screening assay. In 20-mL model tef fermentations, out of these thirteen iso- Zinc lates, the use of Lactobacillus (L.) buchneri strain MF58 and Pediococcus pentosaceus strain MF35 resulted in lowest Lactic acid bacteria PA contents in the fermented tef of 41% and 42%, respectively of its initial content. In comparison 59% of PA Tef remained when spontaneously fermented. Full scale tef fermentation (0.6 L) and injera production using Injera L. buchneri MF58 as culture additive decreased PA in cooked injera from 1.05 to 0.34 ± 0.02 g/100 g, representing a degradation of 68% compared to 42% in injera from non-inoculated traditional fermentation. The visual appearance of the pancakes was similar. The ﬁnal molar ratios of PA to iron of 4 and to zinc of 12 achieved with L. buchneri MF58 were decreased by ca. 50% compared to the traditional fermentation. In conclusion, selected LAB strains in tef fermentations can degrade PA, with L. buchneri MF58 displaying the highest PA degrading potential. The 68% PA degradation achieved by the application of L. buchneri MF58 would be expected to improve human zinc absorption from tef-injera, but further PA degradation is probably necessary if iron absorption has to be increased. © 2014 Elsevier B.V. All rights reserved.

1. Introduction fermenting batter, ensuring the desired textural quality of the baked injera pancake. Due to the use of ersho, the pH drops rapidly (Yigzaw Injera, the staple food consumed widely in Ethiopia (Umeta et al., et al., 2004) and reaches values below pH 4 within hours (Baye et al., 2005), is a pancake prepared from tef (Eragrostis tef), an ancient cereal, 2013; Stewart and Getachew, 1962). As would be expected with acidifi- indigenous to Ethiopia (Zegeye, 1997). Traditional preparation of cation, microbial analysis of tef-injera batter has reported lactic acid Ethiopian tef-injera has not changed over decades (Stewart and bacteria (LAB) to be the major fermentative microbes accompanied by Getachew, 1962; Yetneberk et al., 2004) and involves a fermentation yeast, Enterobacteriaceae and Bacillus spp., but the latter was detected step of 2–3 days based on continuous backslopping, a process in to a lesser extent (Nigatu and Gashe, 1998). These findings are compa- which the new batter is inoculated with a leftover from the previous fer- rable to other sourdough preparations (Minervini et al., 2014). mentation, called “ersho” in Amharic. Ersho, tef flour and water are Tef is rich in phytic acid (PA), myo-inositol hexakisphosphate, which thoroughly kneaded and allowed to rest. Liquid that gets separated on is the main phosphate storage form in most cereals, legumes and nuts. top is discarded and replaced with fresh water. Before the batter is final- With its 6 phosphate groups, PA has a large number of negatively ly poured onto a clay plate and baked, a so-called “absit” is prepared by charged residues over a broad pH range and a strong chelating potential boiling a portion of the batter with water and adding it back to the for divalent cations such as Ca2+,Mg2+,Fe2+ and Zn2+ in the human digestive tract (Schlemmer et al., 2009). The PA-mineral chelates are in- soluble and prevent the absorption of nutritionally important minerals ⁎ Corresponding author at: ETH Zurich, Institute of Food, Nutrition and Health, LFV D and trace elements from plant-based foods (Hurrell, 2004). This is a 16.2, Schmelzbergstrasse 7, 8092 Zurich, Switzerland. Tel.: +41 44 632 43 69. major concern in developing countries, where cereals and legumes are E-mail addresses: maren.fi[email protected] (M.M. Fischer), [email protected] the major source of minerals for large population groups, and where (I.M. Egli), [email protected] (I. Aeberli), [email protected] fi (R.F. Hurrell), [email protected] (L. Meile). plant based diets are associated with nutritional iron and zinc de cien- 1 Present address: ETH Board, Haeldeliweg 15, 8092 Zurich, Switzerland. cies (Ramakrishnan, 2002).

Several strategies are available to overcome the inhibition of PA stocks, either bought at a local market or grown on their own land. on iron absorption, including the addition of ascorbic acid or EDTA For later enumeration and isolation of microorganisms, approximately to meals (Hurrell, 2002), but these strategies do not improve zinc ab- 5 g fermenting batter (before the absit preparation step) was collected sorption. One promising measure that has been reported to increase in duplicate from each family in a sterile falcon tube prefilled with both iron and zinc absorption from foods is enzymatic degradation of 1 mL glycerol. The samples were stored on ice during the study-day, PA (Egli et al., 2004; Troesch et al., 2009). Phytases hydrolyze inositol before the final storage at −20 °C. phosphates and have been reported in microorganisms, plants and ani- mals, although their activity and optimum conditions for reaction differ 2.4. Enumeration of microbes widely (Konietzny and Greiner, 2002). The selection of an appropriate enzyme adapted to the processing conditions of a specificfoodisthere- At ETH Zurich, thawed Ethiopian tef-injera batter samples from fore critical. Phytases are present in cereals (Egli et al., 2002) but have 6 households were analyzed in duplicate for total microbial content not been reported in tef, although PA has been reported to be partially and a further 7 tef-injera batter samples for LAB only. To evaluate the degraded during tef-injera fermentation (Abebe et al., 2007; Umeta model fermentations performed at the laboratory in Zurich, samples et al., 2005; Urga and Narasimha, 1998). Although PA degradation has of about 1 g fresh batter were taken for immediate analysis. been reported to occur to varying extents, the loss is only partial unlike All batter samples were diluted tenfold (w/w) in peptone-solution the almost complete reduction reported when wheat is added to injera and 0.1 mL of subsequent serial ten-fold dilutions were spread on selec- preparation (Baye et al., 2013). tive agar media in duplicates for enumeration of bacteria and fungi. In addition to cereal phytases, microbial phytases can degrade PA Presumptive lactobacilli were selected on MRS agar medium (1.5% during fermentation. Effective PA degradation was achieved by yeasts, agar), incubated anaerobically (AnaeroGen packs, Oxoid, UK) at 37 °C. isolated from Tanzanian togwa, during togwa fermentation (Hellström YM agar medium with chloramphenicol (100 mg/L) against bacterial et al., 2012). But for the application of microbial phytase producers growth was used to cultivate yeasts at 25 °C. General bacterial counts as starter culture in food production the safety aspect should not be were performed on aerobic plate count (PC) agar medium at 30 °C as neglected. This is why LAB, which are involved in many food processes well as presumptive Bacillus spp. spore counts after a heat treatment and also have been noted in injera fermentations (Gashe, 1985) are step at 85 °C for 15 min (Kastner, 2008). Presumptive enterococci of high interest as starter cultures. Selected strains of LAB species orig- were enumerated on KFS agar medium incubated at 43 °C for 48 h. inating from European sourdough fermentations for bread production VRBD agar medium was used to detect Enterobacteriaceae after a 48 h as well as from an African pearl-millet fermentation for gruel making incubation at 37 °C. After incubation, cell counts were calculated in have been reported to be capable of degrading PA (De Angelis et al., colony forming units (cfu) per g of batter. 2003; Lopez et al., 2000; Songré-Ouattara et al., 2008). The present study was designed to identify and isolate microorgan- 2.5. Isolation, cultivation and characterization of LAB isms from tef-injera fermentation that have phytase activity and are eligible as starter culture for tef-injera fermentation. LAB, dominating Dominant, presumptive LAB were picked (2–3 isolates per colony in Ethiopian tef-injera fermentations, were isolated and screened for morphology) from MRS agar media, containing between 30 and phytase activity. The LAB isolates showing PA degrading potential 300 cfu obtained from diluted injera batter, and were purified by re- were characterized and further tested in model tef fermentations as peated streaking on the same medium. Strains were assessed for well as in full scale tef-injera preparation for their ability to degrade Gram-classification (3% KOH), catalase activity (3% H2O2), and by PA within the food-matrix. phase contrast light microscopy. Liquid cultures were grown in MRS media and stored at −80 °C in 25% (v/v) glycerol as a cryoprotectant. 2. Materials and methods For DNA extraction from LAB, a single colony was picked and suspended in 0.1 mL buffer and treated further as Goldenberger et al. 2.1. Chemicals and flours described in procedure B (Goldenberger et al., 1995). For later amplification, the final DNA preparations were kept at −20 °C. Unless otherwise specified, chemicals of p.a. grade were purchased To differentiate the particular LAB strains used, rep-PCR fingerprints from Sigma-Aldrich Chemie GmbH, Switzerland and Merck KGaA, were produced by DNA amplification with the (GTG)5 primer (Gevers Germany. Media and components for cultivation of microbes were or- et al., 2001). PCR products were visualized by UV light on ethidium bro- dered from Becton Dickinson AG, Switzerland, if not stated differently. mide stained agarose (1.8%) after gel electrophoresis in 1× TAE-buffer Tef for fermentations performed at the laboratory in Zurich was bought (pH 8.0), using 1-kb and 100-bp DNA ladders (Generuler, Fermentas at a local market in Debre Zeyit (Bishoftu), Ethiopia and milled to GmbH, Switzerland) as reference. wholegrain flour in a mill at the same location. To follow the fate of applied LAB starter strains in fermentations, 2–4 colonies from the two highest injera batter dilutions were randomly 2.2. Phytase application picked from MRS agar medium and analyzed by rep-PCR fingerprints as described above. Aspergillus niger phytase euphoVida™ 20000G was kindly provided The isolates were identified to the species level by partial 16S rRNA by DSM Nutritional Products, Kaiseraugst, Switzerland. To ensure com- gene sequencing. These regions were amplified with the help of the plete PA degradation during 48 h of tef fermentation (see Section 2.7), primer pairs bak4 (5′-AGGAGGTGATCCARCCGCA-3′)(Greisen et al., double the amount of phytase units (380 FTU/g PA) theoretically 1994)/bak11w (5′-AGT TTG ATC MTG GCT CAG-3′)(Goldenberger needed to degrade all the PA was applied to tef flour. The enzyme is et al., 1997)or7f(5′-AGA GTT TGA TYM TGG CTC AG-3′)/1510r (5′- inactivated by baking. ACG GYT ACC TTG TTA CGA CTT-3′)(Nielsen et al., 2007)basedon a protocol described by Dasen et al. (1998) and Nielsen et al. 2.3. Sampling of Ethiopian tef-injera batter (2007), respectively. The product was purified (GFX PCR DNA & Gel Band Purification Kit, GE Healthcare, UK) and prepared for standard The traditional injera preparation at household level was studied in Sanger sequencing reactions performed at GATC-Biotech (Konstanz, the Debre Zeyit (Bishoftu) area, which is situated on the elevated plain Germany). The resulting sequences were edited in FinchTV (version of central Ethiopia. Families from the urban and surrounding rural areas 1.4, Geospizza Inc., USA), assembled with the help of BioEdit (version who were willing to demonstrate their way of tef-injera preparation 7.0.9, Ibis Biosciences, USA) and compared to sequences reported in were selected. The tef flour used by the families was from their own GenBank (NCBI, USA), using the blastn algorithm. 56 M.M. Fischer et al. / International Journal of Food Microbiology 190 (2014) 54–60

2.6. Screening for phytase activity pancakes in a frying pan (Miami 28 cm, Migros, Switzerland) covered with a thin layer of peanut oil (MClassic, Migros, Switzerland). Maxi- Based on a previously described two-step phytase assay (Bae et al., mum heat was used until the batter was no longer liquid when heat 1999), a screening of PA degradation ability was performed on a modi- was reduced to a third to complete the cooking. fied MRS (mMRS) agar medium, containing 1.5% agar, 1% casein pep- To imitate the traditional backslopping procedure with ersho, spon- tone, 0.4% beef extract, 0.2% yeast extract, 1% glucose, 0.82% sodium taneously fermented batter was prepared from tef and tap water. The acetate trihydrate, 0.2% ammonium citrate dibasic, 0.02% magnesium fermenting batter was overlaid with water to prevent the formation of sulfate heptahydrate, 0.0025% manganese sulfate, 2% MOPS, 0.2% calci- mold on the surface. After incubation for 48 h at room temperature um chloride, and 0.25% sodium phytate as the sole source of phosphate. (RT), the required amount of fermented batter (‘ersho’) was added to The latter two salts were sterile-filtered and added to the autoclaved the main fermentation. medium at 55 °C. Due to the precipitation of PA, the media was non- transparent and the formation of clearing zones was the indicator 2.8. Determination of pH, PA, iron and zinc for PA hydrolysis. To avoid false-positive results from clearing zones due to acid dissolution, 5 mL cerium sulfate (5%) was used for re- For pH determination during fermentation, an aliquot of batter was precipitation of dissolved but non-degraded PA. diluted with nanopure water (1:1, v/v) and measured immediately using a pH meter (Metrohm unitrode 6.0258.000, Switzerland). 2.7. Tef fermentation and injera preparation Tef flours, freeze-dried batter and injera samples were analyzed in duplicate or triplicate for PA content, as the sum of inositol-6- For the model and full scale tef fermentations, a pre-culture of phosphate and inositol-5-phosphate, according to a modification of phytase positive LAB isolates was grown overnight in liquid MRS me- Makower's method (Makower, 1970). Inorganic phosphate concentra- dium with Tween80 (Biolife, Italy) at 37 °C and shaking at 160 rpm. tion was determined using a colorimetric assay (Van Veldhofen and Optical density (OD600) (BioPhotometer, Eppendorf AG, Switzerland) Mannaerts, 1987), modified for measurement on 96 well-microtitration of the cultures was measured to calculate cell forming units per mL plates. (cfu/mL) based on a standard curve which had been calculated for Flour was mineralized in triplicates with a mixture of nitric acid and each isolate. The respective amount of cells was harvested for 7 min at hydrogen peroxide solution by micro-wave heating in closed Teflon® 14,000 ×g (Biofuge pico, Heraeus, Germany) or 15 min at 12,000 ×g in vessels (MLS ETHOSplus, MLS GmbH, Germany). Iron and zinc concen- a Beckman centrifuge (JA-12, Beckman Coulter GmbH, Germany) for trations in the resulting solutions were measured by graphite furnace greater volumes. The cell pellet was resuspended in peptone-solution (AA240Z, Varian, Australia) and flame (AA240FS, Varian, Australia) – (0.85% NaCl, 0.1% casein peptone) and added to autoclaved tap water atomic absorption spectrometry, respectively. used for the fermentation. All containers and spoons used during fermentation experiments and for sample collection were autoclaved or 2.9. Data analysis disinfected with EtOH (70%). Data analyses were conducted with IBM SPSS Statistics 19 for 2.7.1. Model tef fermentation Windows and Microsoft Office Excel 2010. PA and pH values obtained For model fermentations, 9.6 g sterile tap water, 0.4 mL peptone- from replicated fermentations are given as mean ± standard devia- solution containing the respective starter cultures at 107–108 cfu/g and tion (SD). 6gflour were mixed and incubated in a McCartney bottle (20 mL) for 48 h at 25 °C in accordance with the maximal average annual temperature in central Ethiopia (Conway et al., 2004). In order to distinguish 3. Results PA degradation during fermentation caused by microbial activity from any PA degradation from possible endogenous tef phytases, control 3.1. Enumeration of LAB and other microbes in Ethiopian tef-injera suspensions of tef flour were held under the same conditions without fermentation starter cultures but with the addition of chloramphenicol (200 mg/kg) to prevent bacterial growth and cycloheximide (200 mg/kg) to prevent Ethiopian tef-injera batter samples (n(all) =6,n(LAB) = 13), collected propagation of fungi. The pH was adjusted to 3.6 after 6 h of incubation towards the end of the 37 to 75 hour fermentation process, were with acetic acid. Unless otherwise stated, model fermentation experi- assessed for microbial composition. Enterococci and Enterobacteriaceae ments were performed in duplicate. were below the detection limits of 2 lg cfu/g. Spore forming Bacillus species, yeast and mold counts were detected infrequently and were 2.7.2. Full scale tef fermentation thus considered to be present in numbers around the same detection Full scale tef fermentation for injera preparation was performed in limit. Aerobic mesophiles were present on PC agar in all samples in 600 mL batches. For that, 250 g tef flour and 50 g ersho were blended a wide range of 3 to 7 lg after a 24 h incubation and increased up stepwise with 290 g tap water. When defined starter cultures were to 8 lg cfu/g after 72 h due to faint growth of pinpoint colonies added instead of ersho, 275 g of flour and 315 g of liquid, including the representing potential LAB. The highest counts were observed for respective LAB (107–108 cfu/g batter), were mixed. Thereafter 100 g presumptive LAB grown on MRS medium with on average 8 lg cfu/g, tap water was carefully added to cover the surface of the batter. The ranging from 4 to 8 lg cfu/g. walls of the container were cleaned before the container was covered and incubated for 48 h at RT. After 24 and 48 h of fermentation, liquid 3.1.1. Phytase positive LAB present on top of the batter was discarded and replaced by a similar We isolated 76 dominant presumptive LAB from 13 different amount of fresh tap water. Before baking injera from the fermented Ethiopian tef fermentations and screened them in a qualitative assay batter, ‘absit’ was prepared as in the traditional process. For this, the for their ability to degrade PA on mMRS agar medium, containing PA fermented batter was first thoroughly mixed with a spoon, before re- as the phosphorus supply rather than free phosphate. Twenty eight iso- moving and adding a 50 g portion to 150 mL boiling water. The mixture lates exhibited phytase activity in the plate assay. They were further (absit) was stirred, brought back to a boil, cooled to around 60 °C analyzed for their rep-PCR fingerprints in order to differentiate identical and transferred back to the remaining fermented batter. The batter or highly similar isolates and 13 potentially different strains were iden- was incubated for a further 1–2 h at 30 °C until gas production was tified (Fig. 1). Their phylogenetic affiliation was determined according visible. Portions of around 100 g of batter were then baked to injera to 16S rRNA gene sequences (Table 1). M.M. Fischer et al. / International Journal of Food Microbiology 190 (2014) 54–60 57

Fig. 1. Footprints of selected LAB after agarose gel electrophoresis of rep-PCR amplicons: lanes 1–3) Lactobacillus buchneri (MF44, MF58, MF61), lanes 4–6) Lactobacillus casei (MF42, MF50, MF54), lane 7) Lactobacillus brevis (MF67), lane 8) Lactobacillus plantarum (MF79), lane 9) negative control without template, lane 10) Lactobacillus fermentum (MF25), lane 11) Lactobacillus crustorum (MF29), and lanes 12–14) Pediococcus pentosaceus (MF32, MF33, MF35); M) Marker (100 bp & 1 kb).

3.2. Model tef fermentations with LAB isolates identified as PA degrading 3.3. Tef-injera production of full scale fermentations with selected starter cultures The 13 phytase-positive LAB isolates, originating from Ethiopian tef fermentations, were used as starter cultures for these model fermenta- Based on the model fermentation results, LAB isolates of different PA tions. For comparison, a spontaneous and an antibiotic treated fermen- degradation capability, namely L. buchneri MF58, L. brevis MF67 and tation were performed. The pH recorded in fermented tef after 48 h at L. plantarum MF79, were selected as the starter cultures for full scale 25 °C ranged from 3.36 to 3.83 for the inoculated samples and was tef fermentations followed by the baking of injera pancakes. PA degra- 4.08 ± 0.08 in spontaneously fermented tef (Table 2). dation was measured in the fermented batter and compared to tef PA was degraded to a variable extent in all fermentations and also in fermentation with ersho and with ersho plus added phytase (Table 3). the control of non-fermented, antibiotic treated batter where the PA PA degradation was not as high in the full scale fermentation as in the was 17% lower after 48 h at 25 °C (0.87 g/100 g dm) than at time zero model fermentations (Table 2). For example, in the ersho produced (1.05 g/100 g dm). However when taking PA in the antibiotic treated batter, PA was 0.77 g/100 g after full scale fermentation compared to approach as the 100% control, the spontaneous fermentation decreased 0.61 g/100 g after the model fermentation (Table 2). As in the model the PA content to 71% of the control value, whereas the LAB inoculated fermentations, L. buchneri MF58 had the greatest PA degrading potential fermentations decreased the PA content to 49–89% of that value. reducing the PA to 66% of the content in the ersho fermented batter. Inoculating with the strains of Lactobacillus buchneri and Pediococcus L. brevis MF67 and L. plantarum MF79 were again less effective, whereas pentosaceus resulted in the greatest PA degradation and Lactobacillus the addition of purified A. niger phytase degraded PA almost completely. casei strains in contrast to the lowest. The performance of the starter Baking of the fermented batters into injera pancakes resulted in cultures was monitored by rep-PCR fingerprints. The strains used for a relatively constant loss of 16–17 mg PA/100 g representing 21–33% inoculation were found to be dominant in all tef fermentations over indigenous bacteria. In tef incubated under antibiotic treatment, roughly Table 2 2 lg cfu/g was counted on MRS agar media and 3 lg cfu/g on PC agar Influence of selected lactic acid bacteria (LAB) isolates used as starters for model tef media. No growth was observed on YM-chloramphenicol and KFS agar fermentation on pH and phytic acid (PA) degradation after 48 h at 25 °C. media at the detection limit of 2 lg cfu/g. LAB isolate added pHa PA [g/100 g dm] remainingb,c

Table 1 Uninoculatedd 4.08 ± 0.08 0.62 ± 0.03 ﬁ e Phylogenetic af liation after 16S rRNA gene sequence comparison of the 13 phytase active Uninoculated + AB 3.39 ± 0.01 0.87 ± 0.01 isolates from Ethiopian tef fermentations on MRS medium. L. brevis MF67 n.d.f 0.59 ± 0.00 MF44 3.83 ± 0.03 0.45 ± 0.03 Phylogenetic Isolate Closest Sequence Query NCBI L. buchneri MF58 3.70 ± 0.03 0.43 ± 0.03 group relative identity lengtha accession MF61 n.d. 0.52 ± 0.03 Lactobacillus brevis MF67 L. brevis/ 99%/ 1401/ AY974809.1/ MF42 n.d. 0.64 ± 0.08 L. harbinensis 99% 1401 NR_041263.1 L. casei MF50 3.44 ± 0.01 0.74 ± 0.05 Lactobacillus MF44 L. buchneri 99% 1389 CP002652.1 MF54 n.d. 0.77 ± 0.00 buchneri MF58 L. buchneri 99% 1354 NC_015428.1 L. crustorum MF29 3.36 ± 0.00 0.67 ± 0.00 MF61 L. buchneri 100% 1434 NC_015428.1 L. fermentum MF25 3.52 ± 0.01 0.63 ± 0.00 Lactobacillus casei MF42 L. casei 100% 1442 NC_014334.1 L. plantarum MF79 3.52 ± 0.01 0.62 ± 0.00 MF50 L. casei 99% 1423 HQ534100.1 P. pentosaceus MF32 3.67 ± 0.02 0.47 ± 0.02 MF54 L. casei 100% 1153 NC_014334.1 MF33 3.65 ± 0.00 0.51 ± 0.03 Lactobacillus MF25 L. fermentum 99% 1414 JF757227.1 MF35 3.36 ± 0.01 0.44 ± 0.06 fermentum a Mean from duplicate fermentations started with LAB isolates and four replicates of the Lactobacillus MF29 L. crustorum 99% 1289 AM285451.1 uninoculated tef, respectively ±SD. plantarum MF79 L. plantarum/ 100%/ 1449/ NR_075041.1/ b Mean from fermentations performed at least in duplicates ± SD; mean relative SD L. pentosus 100% 1449 NR_029133.1 from the assay, performed at least in duplicates for each sample, was 4.5%. Pediococcus MF32 P. pentosaceus 99% 1429 NR_042058.1 c Of 1.05 g PA/100 g dry matter (dm) in tef ﬂour. pentosaceus MF33 P. pentosaceus 100% 1404 NC_008525.1 d Spontaneously fermented. MF35 P. pentosaceus 100% 1406 NC_008525.1 e Under antibiotic (AB) treatment; pH adjusted after 6 h to pH 3.6. a 100% coverage for all isolates, besides 99% for MF32. f n.d., not determined. 58 M.M. Fischer et al. / International Journal of Food Microbiology 190 (2014) 54–60

Table 3 similar sorghum-kisra fermentations (Mohammed et al., 1991)havere- Inﬂuence of fermentation starter and additive on phytic acid (PA) degradation and ported the predominance and importance of LAB in the fermentation PA:mineral molar ratios in fermented batter and baked injera. process. Our studies conﬁrmed the predominance of LAB, low numbers Starter for Additive PA [g/100 g dm] PA:mineral molar of Bacillus, yeast and mold species, and the absence of enterococci or a,b c fermentation remaining ratios in baked injera Enterobacteriaceae. The high numbers of LAB are assumed to be respon- In batter In baked injera Iron Zinc sible for the relatively high lactic and acetic acid production and there-

Ershod –e 0.77 ± 0.04 0.61 ± 0.06 7:1 22:1 by outcompeting non-acid-tolerant species. The fall in pH below 4, Ershod Phytase 0.01 ± 0.04 0.01 ± 0.00 b1 b1 reported from different Ethiopian injera fermentations (Baye et al., L. buchneri MF58 – 0.51 ± 0.03 0.34 ± 0.02 4:1 12:1 2013; Yigzaw et al., 2004) was similarly observed in our model set-up. f L. brevis MF67 – 0.59 ± 0.07 n.d. Certain LAB have previously been reported to be capable of degrading L. plantarum MF79 – 0.66 ± 0.06 0.52 ± 0.02 6:1 18:1 PA (Corsetti and Settanni, 2007) and we have thus screened 76 pre- a Of 1.05 g PA/100 g dry matter (dm) in tef flour. sumptive LAB isolated from Ethiopian tef-injera fermentations for this b Means from triplicate measurements ± SD. property. All thirteen PA degrading isolates have been classified into c With 2.8 mg zinc/100 g dm native in the tef flour and 7.6 mg iron/100 g flour, adapted from USDA National Nutrient Database for Standard Reference (2012). LAB species or groups that have frequently been described to occur in d Spontaneously fermented flour was used as ersho. different European and African sourdoughs used for bread production e –,noadditive. (De Vuyst and Neysens, 2005; Scheirlinck et al., 2007). Furthermore, f n.d., not determined: no injera was baked from that batter. except for the single isolate identified as Lactobacillus crustorum,allthe detected species can be found on the latest qualified presumption of safety (QPS) list, published by EFSA (European Food Safety Authority), of the PA in the respective fermented batters. The PA level in the supporting their safe use as starter culture (EFSA, 2013). injera pancake made from the L. buchneri MF58 fermented tef As the matrix surrounding the substrate has a strong impact on the was 0.34 g/100 g. This represents 57% of the PA content remaining extent of PA degradation by phytases (Bohn et al., 2007; Brejnholt fl in the ersho fermented injera and 32% of the native PA in tef flour et al., 2011) model fermentations with tef our were used to test the (1.05 g/100 g). The fermented batter made with L. brevis MF67 as a starter PA degradation potential of the different LAB species isolated from culture was not baked due to a surface contamination with molds. the Ethiopian fermented tef-injera batter. The highest PA degradation While no formal sensory or textural analysis of the injera pancakes in the model fermentations with selected LAB was seen after incuba- was made, there was a clear difference in visual appearance. Pancakes tion with the L. buchneri and P. pentosaceus strains which showed made from batter fermented with ersho, ersho plus phytase, and with roughly three times higher degradation than observed in the antibiotic L. buchneri MF58 showed the expected three-dimensional structure treated approach, reducing the PA to less than half of the level in with holes or “eyes” formed during the baking of good quality injera native tef. This is in agreement with previous reports from other plant (Yetneberk et al., 2004). The injera from L. buchneri showed larger and sources. Camacho et al. (1991) reported that a strain of L. buchneri fl more evenly distribute holes than the ersho injera (Fig. 2). The injera decreased PA by more than 55% in average, after 12 h of lupine our made from tef batter fermented with L. plantarum MF79, on the other fermentation and several P. pentosaceus strains from sourdough have fi hand, was dense with few holes. been shown to exhibit PA degradation capacity on modi ed MRS agar medium (Raghavendra et al., 2010). Numerous studies have also 3.4. Molar ratios of PA:iron and PA:zinc reported L. plantarum strains to be phytase positive (Lopez et al., 2000; Songré-Ouattara et al., 2008; Sreeramulu et al., 1996; Tang et al., The tef flour used for injera preparation contained 28 mg zinc and 2010) although more detailed studies revealed an extracellular dephos- fi 383 mg iron per kg. PA:mineral molar ratios were calculated according phorylating enzyme rather unspeci c for sodium PA but with a high fi to the determined PA content and the native zinc content, while the af nity to acetyl phosphate (Zamudio et al., 2001). In our study, tef- amount of iron was corrected due to soil contaminations predicted in fermentation with the single L. plantarum isolate MF79 led to an inter- the Ethiopian flour by using a published iron value of 7.6 mg/100 g for mediate PA reduction, not different from the spontaneous fermentation. the iron content of presumptive non-contaminated tef, taken from the The lowest PA reduction was observed in fermentations started with US National Nutrient Database (U.S. Department of Agriculture and the L. casei isolates. In contrast to the partial PA degradation achieved fi N.D.L.H.P., 2012)(Table 3). According to the PA levels, the PA:molar by application of selected starter cultures, the addition of puri ed ratios were also highest in the traditional injera fermentation. A. niger phytase showed the possibility for complete enzymatic PA turnover during tef fermentation although processing conditions are not 4. Discussion optimal for respective enzyme activity. The contribution of endogenous tef phytases to the PA degradation Earlier microbial investigations of fermented tef-injera batter during fermentation is unclear. Cereals such as wheat and rye are re- (Gashe, 1985; Nigatu and Gashe, 1998; Umeta and Faulks, 1989) and ported to exhibit high phytase activity (Egli et al., 2002) and in wheat

Fig. 2. Texture of baked injera started with ersho (A), started with ersho and added phytase (B), with L. buchneri MF58 as starter culture (C) and with L. plantarum MF79 as starter culture (D). M.M. Fischer et al. / International Journal of Food Microbiology 190 (2014) 54–60 59 and rye sourdough acidification during fermentation is reported to acti- Acknowledgments vate wheat and rye phytases and lead to PA degradation (Leenhardt et al., 2005; Reale et al., 2007). It is possible therefore that the 17% We are deeply grateful for the families who gave us the possibility reduction of PA in the antibiotic treated fermentation was due to to collect samples from their private tef-fermentations. Furthermore tef phytases, however the contribution was minor compared to the PA we thank Susanna Wenk (M. Sc.) for her contribution in the study degradation caused by spontaneous fermentation or by LAB inoculation. and Medicor Foundation Liechtenstein for supporting this project Thus the ability to express enzymes for PA degradation could be espe- financially. cially advantageous for LAB adapted to the tef environment. PA degradation during tef fermentation for injera preparation was comparable to the model fermentations. Additional PA degradation References was observed during baking due to thermal hydrolysis (Mahgoub and Abebe, Y., Bogale, A., Hambidge, K.M., Stoecker, B.J., Bailey, K., Gibson, R.S., 2007. Phytate, Elhag, 1998). But not all fermentation approaches yielded in injera zinc, iron and calcium content of selected raw and prepared foods consumed in rural with typical “eyes”-structure. The different appearance of the injera Sidama, Southern Ethiopia, and implications for bioavailability. J. Food Compos. Anal. 20, 161–168. pancakes might be due to more intense CO2 production by L. buchneri Bae, H.D., Yanke, L.J., Cheng, K.J., Selinger, L.B., 1999. A novel staining method for detecting (Fig. 2C), an obligatory heterofermentative species of LAB, compared phytase activity. J. Microbiol. Methods 39, 17–22. to L. plantarum (Fig. 2D), a facultative heterofermentative species, and Barrangou, R., Lahtinen, S., Ouwehand, F., 2011. Genus Lactobacillus, In: Von Wright, A. fi (Ed.), Lactic Acid Bacteria: Microbiological and Functional Aspects, 4 ed. CRC Press, unde ned mixtures of LAB (Fig. 2A & B) in spontaneously fermented pp. 77–91. dough (Barrangou et al., 2011). Baye, K., Mouquet-Rivier, C., Icard-Verniere, C., Rochette, I., Guyot, J.P., 2013. Influence of The Ethiopian tef flour used in this study was found to have a very flour blend composition on fermentation kinetics and phytate hydrolysis of sourdough used to make injera. Food Chem. 138, 430–436. high iron content, which most probably results, at least in part, from Besrat, A., Admasu, A., M., O, 1980. Critical study of the iron content of tef (Eragrostis tef). soil contamination, introduced during the threshing procedure (Besrat Ethiop. Med. J. 18, 45–52. et al., 1980). This soil iron is not expected to be available for absorption Bohn, L., Josefsen, L., Meyer, A.S., Rasmussen, S.K., 2007. Quantitative analysis of phytate globoids isolated from wheat bran and characterization of their sequential dephos- in humans as it does not dissolve in the gastrointestinal tract (Hallberg phorylation by wheat phytase. J. Agric. Food Chem. 55, 7547–7552. and Bjorn-Rasmussen, 1981). We therefore used a published value of Brejnholt, S.M., Dionisio, G., Glitsoe, V., Skov, L.K., Brinch-Pedersen, H., 2011. The degrada- 7.6 mg/100 g for the iron content of presumptive non-contaminated tion of phytate by microbial and wheat phytases is dependent on the phytate matrix – tef, taken from the US National Nutrient Database (U.S. Department of and the phytase origin. J. Sci. Food Agric. 91, 1398 1405. Camacho, L., Sierra, C., Marcus, D., Guzman, E., Campos, R., Vonbaer, D., Trugo, L., 1991. Agriculture and N.D.L.H.P., 2012) and the determined 2.8 mg/100 g for Nutritional quality of lupine (Lupinus albus cv. Multolupa) as affected by lactic acid zinc in order to calculate molar ratios of PA to iron and zinc, respectively, fermentation. Int. J. Food Microbiol. 14, 277–286. as predictor for human absorption (Hurrell, 2004; Lönnerdal, 2002). The Conway, D., Mould, C., Bewket, W., 2004. Over one century of rainfall and temperature observations in Addis Ababa, Ethiopia. Int. J. Climatol. 24, 77–91. PA:iron molar ratio in the baked injera with L. buchneri MF58 as starter Corsetti, A., Settanni, L., 2007. Lactobacilli in sourdough fermentation. Food Res. Int. 40, culture was lower compared to when ersho was used (Table 3) but still 539–558. above the threshold of 1 or preferably 0.4 — above which iron absorp- Dasen, G., Smutny, J., Teuber, M., Meile, L., 1998. Classification and identification of propionibacteria based on ribosomal RNA genes and PCR. Syst. Appl. Microbiol. 21, tion is assumed to be poor (Hurrell, 2004). Considering the achieved de- 251–259. crease of the molar ratio below 6, iron absorption could be improved if DeAngelis,M.,Gallo,G.,Corbo,M.,McSweeney,P.,Faccia,M.,Giovine,M.,Gobbetti,M.,2003. the injera were consumed in meals with small amounts of meat or veg- Phytase activity in sourdough lactic acid bacteria: purification and characterization of a phytase from Lactobacillus sanfranciscensis CB1. Int. J. Food Microbiol. 87, 259–270. etables (Hurrell and Egli, 2010), however this is not the normal practice De Vuyst, L., Neysens, P., 2005. The sourdough microflora: biodiversity and metabolic in Ethiopia as injera is often consumed with pulses, free of ascorbic acid interactions. Trends Food Sci. Technol. 16, 43–56. and also high in PA. In order to improve iron absorption substantially EFSA, 2013. Panel on Biological Hazards (BIOHAZ): scientific opinion on the maintenance of the list of QPS biological agents intentionally added to food and feed (2013 from the tef-injera based diets, PA needs to be further decreased during update). EFSA J. 11, 3449. the tef-injera fermentation or the diet needs to be complemented with Egli, I., Davidsson, L., Juillerat, M.A., Barclay, D., Hurrell, R.F., 2002. The influence of soaking iron absorption enhancing foods. In the case of zinc, using L. buchneri and germination on the phytase activity and phytic acid content of grains and seeds – MF58 as a starter culture for tef-injera production would increase zinc potentially useful for complementary feeding. J. Food Sci. 67, 3484 3488. Egli, I., Davidsson, L., Zeder, C., Walczyk, T., Hurrell, R., 2004. Dephytinization of a comple- availability from low (PA:zinc molar ratios N15) to moderate (PA:zinc mentary food based on wheat and soy increases zinc, but not copper, apparent molar ratios in the range of 5–15) and therefore would be expected to absorption in adults. J. Nutr. 134, 1077–1080. result in a modest but useful increase in zinc absorption (FAO and FAO, WHO, 2004. Vitamin and Mineral Requirements in Human Nutrition: Report of a Joint FAO/WHO Expert Consultation, Bangkok, Thailand, 1998, 2nd ed. WHO, 2004). Gashe, B.A., 1985. Involvement of lactic acid bacteria in the fermentation of TEF (Eragrostis The present study shows for the first time that LAB isolated from tef) an Ethiopian fermented food. J. Food Sci. 50, 800–801. fermented Ethiopian tef-injera batter degrade PA during fermentation Gevers, D., Huys, G., Swings, J., 2001. Applicability of rep-PCR fingerprinting for identification of Lactobacillus species. FEMS Microbiol. Lett. 205, 31–36. within the food matrix. The power of our study was limited by the use Goldenberger, D., Perschil, I., Ritzler, M., Altwegg, M., 1995. A simple universal DNA of non-sterile flour, but as LAB starter cultures dominated the lactic acid extraction procedure using SDS and proteinase-K is compatible with direct PCR fermentation, this constant microbial background was not expected to amplification. PCR Methods Appl. 4, 368–370. fl fi Goldenberger, D., Kunzli, A., Vogt, P., Zbinden, R., Altwegg, M., 1997. Molecular diagnosis greatly in uence the overall ndings. Another potentially unfavorable as- of bacterial endocarditis by broad-range PCR amplification and direct sequencing. pect was that fermentations were performed in Zurich not in Ethiopia but J. Clin. Microbiol. 35, 2733–2739. since we used Ethiopian tef flour, the outcome of our ersho mediated tef- Greisen, K., Loeffelholz, M., Purohit, A., Leong, D., 1994. PCR primers and probes for the 16S ribosomal-RNA gene of most species of pathogenic bacteria, including bacteria injera was estimated to be closely related to traditional Ethiopian injera found in cerebrospinal-fluid. J. Clin. Microbiol. 32, 335–351. production. The model fermentation set-up was established to evaluate Hallberg, L., Bjorn-Rasmussen, E., 1981. Measurement of iron-absorption from meals the ability of LAB to degrade PA directly in the tef-matrix, although con- contaminated with iron. Am. J. Clin. Nutr. 34, 2808–2815. tribution of unspecificacidphosphatases,whetherextra-orintracellular Hellström, A.M., Almgren, A., Carlsson, N.G., Svanberg, U., Andlid, T.A., 2012. Degradation of phytate by Pichia kudriavzevii TY13 and Hanseniaspora guilliermondii TY14 in enzymes, on PA turnover could not be distinguished by this model. Tanzanian togwa. Int. J. Food Microbiol. 153, 73–77. In conclusion, we have shown that certain strains belonging to dif- Hurrell, R.F., 2002. Fortification: overcoming technical and practical barriers. J. Nutr. 132, – ferent LAB species, isolated from tef-injera fermentation, partially de- 806s 812s. Hurrell, R.F., 2004. Phytic acid degradation as a means of improving iron absorption. Int. J. grade PA. Injera pancakes prepared with L. buchneri MF58 fermented Vitam. Nutr. Res. 74, 445–452. tef contained substantially less PA than those prepared by traditional Hurrell, R.F., Egli, I.M., 2010. Iron bioavailability and dietary reference values. Am. J. Clin. fermentation and could help to improve zinc absorption. PA levels, Nutr. 91, 1461s–1467s. Kastner, S., 2008. The Inoculum of the Cassava Product Attiéké: Microbiological Diversity, however, are still relatively high and further PA degradation is probably Impact on Product Quality and Safety, and Development of a Controlled Fermentation needed for increased iron absorption. Process. Eidgenössische Technische Hochschule Zürich, Switzerland. 60 M.M. Fischer et al. / International Journal of Food Microbiology 190 (2014) 54–60

Konietzny, U., Greiner, R., 2002. Molecular and catalytic properties of phytate-degrading Schlemmer, U., Frolich, W., Prieto, R.M., Grases, F., 2009. Phytate in foods and significance enzymes (phytases). Int. J. Food Sci. Technol. 37, 791–812. for humans: food sources, intake, processing, bioavailability, protective role and Leenhardt, F., Levrat-Verny, M.A., Chanliaud, E., Remesy, C., 2005. Moderate decrease of analysis. Mol. Nutr. Food Res. 53, S330–S375. pH by sourdough fermentation is sufficient to reduce phytate content of whole Songré-Ouattara, L.T., Mouquet-Rivier, C., Icard-Vernière, C., Humblot, C., Diawara, B., wheat flour through endogenous phytase activity. J. Agric. Food Chem. 53, 98–102. Guyot, J.P., 2008. Enzyme activities of lactic acid bacteria from a pearl millet Lönnerdal, B., 2002. Phytic acid — trace element (Zn, Cu, Mn) interactions. Int. J. Food Sci. fermented gruel (ben-saalga) of functional interest in nutrition. Int. J. Food Microbiol. Technol. 37, 749–758. 128, 395–400. Lopez, H.W., Ouvry, A., Bervas, E., Guy, C., Messager, A., Demigne, C., Remesy, C., 2000. Sreeramulu, G., Srinivasa, D.S., Nand, K., Joseph, R., 1996. Lactobacillus amylovorus as a Strains of lactic acid bacteria isolated from sour doughs degrade phytic acid and phytase producer in submerged culture. Lett. Appl. Microbiol. 23, 385–388. improve calcium and magnesium solubility from whole wheat flour. J. Agric. Food Stewart, R., Getachew, A., 1962. Investigations of the nature of injera. Econ. Bot. 16, Chem. 48, 2281–2285. 127–130. Mahgoub, S.E.O., Elhag, S.A., 1998. Effect of milling, soaking, malting, heat treatment and Tang, A.L., Wilcox, G., Walker, K.Z., Shah, N.P., Ashton, J.F., Stojanovska, L., 2010. Phytase ac- fermentation on phytate level of four Sudanese sorghum cultivars. Food Chem. 61, tivity from Lactobacillus spp. in calcium-fortified soymilk. J. Food Sci. 75, M373–M376. 77–80. Troesch, B., Egli, I., Zeder, C., Hurrell, R.F., de Pee, S., Zimmermann, M.B., 2009. Optimiza- Makower, R., 1970. Extraction and determination of phytic acid in beans (Phaseolus tion of a phytase-containing micronutrient powder with low amounts of highly vulgaris). Cereal Chem. 47, 288–295. bioavailable iron for in-home fortification of complementary foods. Am. J. Clin. Minervini, F., De Angelis, M., Di Cagno, R., Gobbetti, M., 2014. Ecological parameters Nutr. 89, 539–544. influencing microbial diversity and stability of traditional sourdough. Int. J. Food U.S. Department of Agriculture, N.D.L.H.P., 2012. USDA National Nutrient Database for Microbiol. 171, 136–146. Standard Reference. Release 25 [Online]. Available at http://www.ars.usda.gov/ba/ Mohammed, S.I., Steenson, L.R., Kirleis, A.W., 1991. Isolation and characterization of mi- bhnrc/ndl (Accessed: June 2013). croorganisms associated with the traditional sorghum fermentation for production Umeta, M., Faulks, R.M., 1989. Lactic acid and volatile (C2–C6) fatty acid production in the of Sudanese kisra. Appl. Environ. Microbiol. 57, 2529–2533. fermentation and baking of tef (Eragrostis tef). J. Cereal Sci. 9, 91–95. Nielsen, D.S., Teniola, O.D., Ban-Koffi, L., Owusu, M., Andersson, T.S., Holzapfel, W.H., 2007. Umeta, M., West, C.E., Fufa, H., 2005. Content of zinc, iron, calcium and their absorption The microbiology of Ghanaian cocoa fermentations analysed using culture-dependent inhibitors in foods commonly consumed in Ethiopia. J. Food Compos. Anal. 18, and culture-independent methods. Int. J. Food Microbiol. 114, 168–186. 803–817. Nigatu, A., Gashe, B.A., 1998. Effect of heat treatment on the antimicrobial properties of tef Urga, K., Narasimha, H.V., 1998. Phytate:zinc and phytate × calcium:zinc molar ratios in dough, injera, kocho and aradisame and the fate of selected pathogens. World J. selected diets of Ethiopians. Bull. Chem. Soc. Ethiop. 12, 1–7. Microbiol. Biotechnol. 14, 63–69. Van Veldhofen, P., Mannaerts, G., 1987. Inorganic and organic phosphate measurements Raghavendra, P., Rao, T.S.C., Halami, P.M., 2010. Evaluation of beneficial attributes for in the nanomolar range. Anal. Biochem. 161, 45–48. phytate-degrading Pediococcus pentosaceus CFR R123. Benefic. Microbes 1, 259–264. Yetneberk, S., de Kock, H.L., Rooney, L.W., Taylor, J.R.N., 2004. Effects of sorghum cultivar Ramakrishnan, U., 2002. Prevalence of micronutrient malnutrition worldwide. Nutr. Rev. on injera quality. Cereal Chem. 81, 314–321. 60, S46–S52. Yigzaw, Y., Gorton, L., Solomon, T., Akalu, G., 2004. Fermentation of seeds of teff (Eragrostis Reale, A., Konietzny, U., Coppola, R., Sorrentino, E., Greiner, R., 2007. The importance teff), grass-pea (Lathyrus sativus), and their mixtures: aspects of nutrition and food of lactic acid bacteria for phytate degradation during cereal dough fermentation. safety. J. Agric. Food Chem. 52, 1163–1169. J. Agric. Food Chem. 55, 2993–2997. Zamudio, M., Gonzalez, A., Medina, J.A., 2001. Lactobacillus plantarum phytase activity is Scheirlinck, I., van der Meulen, R., van Schoor, A., Huys, G., Vandamme, P., de Vuyst, L., due to non-specific acid phosphatase. Lett. Appl. Microbiol. 32, 181–184. Vancanneyt, M., 2007. Lactobacillus crustorum sp nov., isolated from two traditional Zegeye, A., 1997. Acceptability of injera with stewed chicken. Food Qual. Prefer. 8, Belgian wheat sourdoughs. Int. J. Syst. Evol. Microbiol. 57, 1461–1467. 293–295.