Bacterial fermentation of Lemna sp . as a potential substitute of fish meal in shrimp diets

Centro Interdisciplinario de Investigación para el Desarrollo Integral Regional-Instituto Politécnico Nacional, Unidad Sinaloa, Boulevard Juan de Dios Bátiz Paredes 250, Guasave, Sinaloa 81101, Mexico. Centro de Investigación en Biotecnología Aplicada, Ex-Hacienda San Juan Molino, Carretera Estatal TecuexcomacTepetitla Km 1.5, Tlaxcala 90700, Mexico. Centro de Investigaciones Biológicas del Noroeste. Instituto Politécnico Nacional 195, La Paz, Baja California Sur 23096, Mexico.


INTRODUCTION
Worldwide aquaculture has been fast growing in the last decades and has become an important industry in many countries, however fish meal used as the major dietary protein source in compounded feed is currently limited, which has resulted in massive research to identify alternative protein sources (Tacon and Metian, 2008;Olsen and Hasan, 2012).In the last years, most researchers have proposed to use an alternative ingredient for fish flour due to its limited supply and high cost.At present, studies focused on the replacement of this protein with lower cost ingredients from vegetable protein are underway (Amaya et al., 2007;Gamboa-Delgado et al., 2013).
Duckweed (Lemna sp.), as a source of vegetable protein, has a better essential amino acid profile than most vegetable protein sources and, in turn, this profile resembles that of animal protein sources.Duckweed contains from 28 to 43% crude protein, high concentration of trace minerals as potassium and phosphorus, carotenes, and xanthophylls (Chaturvedi et al., 2003).Lemma sp.grows in nutrients-rich waters and has been used as a food supplement for fish (He et al., 2013), and as a source of protein to replace fish flour in diets for carp (Yilmaz et al., 2004).However, the presence of antinutritional factors in plant flours affects negatively their nutritional value (Kumaraguru Vasagam et al., 2007).The fermentation process of plants is a simple and cheap method that might decrease considerably the antinutritional factors and crude fiber content, increasing plant digestibility (Bairagi et al., 2002;Nout, 2009).To this regard, fermentation of duckweed flour (Lemna polyrhiza) by Bacillus sp.decreases fiber and antinutritional factors (Bairagi et al., 2002).Additionally, there are fermentation studies on protein sources with low digestibility, in which microorganisms are used to hydrolyze unusual substrates, such as feathers, to improve their digestibility (Bertsch et al., 2003).
In this study, duckweed was evaluated as a potential substitute of fish meal in shrimp diets (L.vannamei) by improving its nutritional quality through a bacterial fermentation process.

Bacteria
Lactic acid bacteria (LAB) were isolated from the gut and hepatopancreas of healthy juvenile shrimp (L.vannamei and L. stylirostris).Animals were collected from the Navachiste Bay (Guasave, Sinaloa, Mexico).In the laboratory, guts and hepatopancreas were dissected aseptically and homogenized in Eppendorf tubes with 200 μl of sterile saline solution (2.5% NaCl).One hundred microliters of the homogenate were inoculated on MRS agar (BD Difco, USA) supplemented with 2.5% NaCl and 200 mg/L of aniline blue (Sigma-Aldrich, USA) in duplicate.The plates were kept at 30ºC and their replicates at 37ºC for 24 h.Blue colonies were selected and streaked onto MRS plates and incubated as above.The isolates maintained in pure culture were stored at -70ºC in MRS with 15% (v/v) glycerol and 2.5% NaCl.
The bacillus (Ba4) used in this work was originally isolated from sea water of the Navachiste Bay.The Ba4 isolate was characterized to be used as potential probiont for shrimp by Partida-Arangure et al. (2013).
LAB isolates were characterized using Gram stain, cellular morphology, and DNA sequence analysis.In addition, hemolytic activity (HA), hydrophobicity, extracellular enzymatic activity, salinity tolerance, and kinetics of bacterial growth were studied to be used as criteria to select potential bacteria to ferment duckweed flour.

Bacterial growth kinetics, colony forming unit counts (CFU) and hemolysis assay
The bacillus and LAB were originally isolated and grown at 37 and Flores-Miranda et al. 1517 30ºC, respectively.In this work, bacteria were grown at different temperatures to obtain the temperature range at which both bacteria have good growth.Bacterial growth was determined by reading the absorbance of cultures in a Thermo Spectronic Genesys 2 Spectrophotometer (Thermo Scientific, USA) at 580 nm for 24 h.Ba4 and LAB grew well at 35ºC and this temperature was used for subsequent cultures.We determined bacterial growth kinetics to determine the log growth phase of each LAB isolate.Additionally, count the colony forming units (CFU) and hemolysis assay were carried out according to Leyva-Madrigal et al. (2011).
The isolates with γ-hemolysis activity (lack of hemolysis) were selected for further analysis; β hemolytic isolates were scrapped for their pathogenic potential.

Extracellular enzymatic activity
Extracellular protease and lipase activities were carried out according to León et al. (2000) with some modifications made by Leyva-Madrigal et al. (2011).Additionally, to determine cellulolytic activity, the isolates were inoculated onto Petri plates with carboxymethyl cellulose agar (1%, w/v, Sigma-Aldrich).The plates were incubated at 35°C for 48-72 h.After the incubation time, 10 ml of a Congo red solution (1%, w/v, Sigma-Aldrich) was added to the plates, after 15 min the excess was removed, and then a saline solution (2 M, NaCl) was added, allowing to stand for 15 min to determine isolates showing hydrolysis halos (Teather and Wood, 1982).

Microbial adhesion to solvents (p-xylene)
The microbial adhesion to solvents was measured according to Rosenberg et al. (1980).Additionally, this test was performed on the bacillus (Ba4) isolated by Partida-Arangure et al. (2013).The apolar solvent p-xylene was used because bacterial adhesion to this solvent reflects the hydrophobic or hydrophilic nature of the cell surface.LAB were grown in MRS medium and harvested in the stationary growth phase by centrifugation at 5000 g for 15 min.The biomass was washed twice and resuspended in PBS buffer (137 mM NaCl, 2.7 mM KCl, 2 mM Na 2 HPO 4 , pH 7.4).The absorbance of the cell suspension was measured at 600 nm (A0).One milliliter of cell suspension was added to 3 ml of xylene and then incubated for 10 min at room temperature.A system of two phases was obtained after incubation and then the sample was mixed in a vortex for 2 min.The aqueous phase was removed after 20 min of incubation at room temperature and measured for absorbance at 600 nm (A1).The percentage of bacterial adhesion to the xylene solvent was calculated as (1-A1/A0) × 100.

Salinity tolerance
Salinity tolerance of LAB was obtained at different concentrations of NaCl to be used for fermentation.MRS medium supplemented with 0, 0.5, 1.0, 1.5, 2.0, and 2.5% NaCl was placed in 15-ml Falcon tubes, inoculated with 20 µl of LAB stock, and incubated at 35ºC for 24 h.Absorbance was read in a spectrophotometer Thermo Spectronic Genesys 2 for growth.Sterile medium was used as blank.

DNA extraction
DNA extraction of LAB and Ba4 was achieved according to Moretti et al. (1985).Total DNA was isolated to be used as template in the 16S gene amplification.The samples were subjected to further purification by RNAase treatment and phenolization.

16S ribosomal gene (LAB and Ba4)
Amplification of the 16S ribosomal DNA was performed by single PCR using primers 27f and 1525r (Lane, 1991).Purification of the PCR product was performed with the cleaning kit QIAquick PCR Purification Kit (Invitrogen, USA).Purified PCR products were sent for sequencing to the Biotechnology Institute of UNAM (Mexico).Finally, the sequences obtained were compared with the reported nucleotide sequences in the genomic bank (GeneBank database) using the NCBI BLAST software (www.ncbi.nlm.nih.gov/BLAST/).

Phylogenetic analyses
The phylogenetic analyses were performed with the Molecular Evolutionary Genetics Analysis software (MEGA 5 Beta) (Tamura et al., 2011).Evolutionary relationships between sequences were inferred by using the neighbor-joining method (NJ) (Saitou and Nei, 1987).The robustness of the NJ topology was evaluated by bootstrap test using 1000 replicates.The Thermotoga maritima sequence was used as out group.

Duckweed flour
Wild duckweed (Lemna sp.), identified according to the description of Lot et al. (1999), was collected from a shallow lagoon near a corn-growing area and cultivated in 1000-l plastic tanks with 600 L of freshwater fertilized with 30 mg/L of NPK (nitrogen, phosphorus, and potassium, 1:1:1, w/w).The biomass was harvested every 2 days and sundried for 24 h.Dry plants were passed through a sieve to remove foreign materials and ground in a hammer mill (Thomas Scientific, USA) to a particle size of approximately 450 µm.

Water retention capacity
Water retention capacity was modified from Smith et al. (1973).Distilled water was added to a known amount of flour until saturation and then the weight was obtained to determine the percentage of humidity of the sample.

Fermentation of duckweed flour
The optimum conditions for fermentation of duckweed flour were selected based on the results of four experiments.In experiment I and II, the growth of LAB (BALLvHp2) and bacillus (Ba4) in duckweed flour was evaluated, respectively.Flour without bacteria but with molasses was used as control and treatments with different percentages of molasses (3, 5, 10, and 15%) were inoculated with 1 × 10 6 CFU/g of flour.Additionally, the pH was measured every 24 h for 144 h.In experiment III, the percentage of molasses in which microorganisms showed better performance was used.The experiment was designed as follows: 1) BALLvHp2; 2) Ba4, 3) Ba4 + BALLvHp2 inoculated 24 h after Ba4; and 4) Ba4 + BALLvHp2 (inoculated simultaneously).Each treatment had six replicates, one replicate was used every 24 h for the analysis of the microbial concentration and pH.In experiment IV, the fermentation conditions were as follows: duckweed flour added with 10% molasses, 1.5% NaCl, BALLvHp2 and Ba4 (1 × 10 6 CFU/g of flour, inoculated at the same time), and different percentages of moisture (75, 100, and 125%).Each treatment had nine replicates and a replicate was used every 24 h for the analysis of the fermentation process.The fermented samples were diluted in distilled water (1:5) and the pH was measured using an electrode (pH).Microbial growth was determined by counting CFU on MRS and TSA agar plates using the serial dilution method (Shirai et al., 2001).

Chemical composition of ferments
The fermented flour was analyzed at CIBNOR (La Paz, Baja California Sur, Mexico) to determine the content of moisture, protein, crude fat, crude fiber, ash, nitrogen free extract, and gross energy.The amino acid profile was determined at CIAD (Hermosillo, Sonora, Mexico) by liquid chromatography (HPLC) using acid-base digestions (Einarsson et al., 1983).

Determination of tannins and phytic acid
Tannins were determined according to Atanassova and Christova-Bagdassarian (2009), based on the AOAC (1990) and The International Pharmacopoeia.The estimation of phytic acid was according to Wheeler and Ferrel (1971).

Statistical analysis
A factorial analysis of variance (ANOVA) was applied to identify differences among treatments of the fermentation process at 95% interval of confidence (P < 0.05).One-way analysis of variance (ANOVA) was applied to examine the differences in proximate composition of the ferments among treatments (P < 0.05).Where significant ANOVA differences were found, a Tukey's HSD test was used to identify the nature of these differences at P < 0.05.

Bacteria isolation and characterization
Twenty nine presumptive LAB strains were isolated, six isolates from L. stylirostris and 23 isolates from L. vannamei.Six isolates were γ-hemolytic and 23 isolates were β-hemolytic.Five isolates with γ-hemolysis were selected for successive characterization.The selected isolates were Gram positive, catalase negative and coccior cocobacilli-shaped.Selected isolates did not show enzymatic activity (proteases and lipases).However, Ba4 showed cellulolytic activity with a halo of 14.3 ± 0.6 mm (data not shown).Adhesion of selected isolates and Ba4 to the solvent was between 35.52 ± 12.71 and 79.01 ± 5.56%.All isolates showed a log phase between 6 and 12 h.Colony forming units of selected isolates and Ba4 were from 2.315 × 10 9 to 6.020 × 10 9 CFU/ml (Table 1).

Molecular identification of bacteria
The molecular analysis of the isolates BALLvHp5 and BALLvHp2 showed identities of 99.5 and 99.9% with Pediococcus pentosaceus, respectively.The isolate BALLsI2 showed identity of 98.3% with Weisella viridescens.BALLsI6 and BALLsI4 belong to the Weissella genus with identities of 97.9 and 93.4%, respectively.The isolate Ba4 belongs to Bacillus pumilus with a 100% identity (Figure 1).

Water retention of Lemna flour
Results of water retention capacity showed that 1 g of  flour is saturated with 3.97 g of water.This means that the flour has 100% moisture.This result was employed to design the IV fermentation assay with 75, 100, and 125% moisture and the best molasses percentage.

Fermentation assays
In the first fermentation assay, the results showed that the growth of BALLvHp2 was significantly higher in the treatment with 15% molasses during the first 48 h as compared to 10, 5, and 3% molasses; however, the growth of BALLvHp2 decreased after 48 h.Growth with 5 and 10% molasses at 72 h was significantly higher than with 3 and 15% molasses.Bacterial growth in the treatment with 3% molasses was significantly lower than with 5, 10, and 15% molasses at 48 and 72 h (F (18, 56) = 512.90,P = 0.0001).During the fermentation process, the pH decreased significantly from 24 to 144 h with all percentages of molasses as compared with the control with molasses but without bacteria.The pH of treatment with 3% molasses was significantly higher than with 5, 10, and 15% molasses (F = (24, 133) = 201.37,P= 0.0001) (Figure 2).In the second fermentation assay, the results showed that the growth of Ba4 was significantly higher in the treatment with 10% molasses at 48 and 72 h as compared to 5 and 15% molasses.At 96 h, the growth of Ba4 with 5 and 10% molasses was significantly higher than with 15% molasses (F (10, 36) = 158.31,P = 0.0000) (Figure 3).In the third fermentation assay (Figure 4), 10% molasses and bacteria (Ba4 and BALLvHp2) were added to the flour.When bacteria were inoculated alone, higher growth was obtained, but when both bacteria were inoculated, the growth was affected, specially Ba4 (F (25, 72) = 189.71,P = 0.0001).The pH was significantly lower in treatments II, III, and IV as compared to control (F (20, 60) = 262.67,P = 0.0001).
In the fourth fermentation assay, 10% molasses and bacteria (Ba4 and BALLvHp2), inoculated at the same time, were added to the flour with different moisture percentages (75, 100, and 125%).Ba4 showed a similar trend in growth at different percentages of moisture.The growth of BALLvHp2 was significantly lower at 75% moisture as compared with 100 and 125% moisture (F (16, 54) = 39.446,P = 0.0001) (Figure 5).Fermented flour (75, 100, and 125%) showed significantly lower moisture content as compared with the unfermented one (P < 0.05).No significant differences were found in protein content of fermented and unfermented flour (P > 0.05).Fermented flour showed significantly lower ( 50%) ether extract and crude fiber ( 50%) as compared to the unfermented one.Ashes in treatments with 100 and 125% moisture were significantly lower as compared to the unfermented flour (12.3 and 21%, respectively) (P < 0.05).Samples did not show significant differences in the content of nitrogen-free extract (NFE, P > 0.05).Total nitrogen was significantly lower in the fermented flour as compared to the unfermented one (P > 0.05).The energy content of fermented flour was significantly lower than in the unfermented one (P < 0.05).The percentage of tannins in the unfermented flour was 1.18 ± 0.1, whereas in the fermented flour it was 1.11 ± 0.2, 1.04 ± 0.1, and 1.04 ± 0.2% in treatments with 75, 100, and 125% moisture, respectively.No significant differences were found among samples (P > 0.05%).The percentage of phytic acid in the unfermented flour was 2.24 ± 0.21%, whereas in the fermented flour it was 0.91 ± 0.03, 0.73 ± 0.06, and 0.94 ± 0.05% in treatments with 75, 100, and 125% moisture, respectively.Significant differences were found in the three fermented samples when compared with the unfermented sample (P < 0.05).Fermentation reduced phytic acid in 61% (Table 2).

DISCUSSION
Fermentation process of foods contributes to enhance flavor and texture, and to improve shelf-life and digestibility (Nout, 2009).In this work, 29 presumptive LAB were isolated from wild juveniles of L. stylirostris and L. vannamei.After their characterization, five of these isolates (presumptive LAB) showed potential as fermentative bacteria, including medium and high adhesion capacity, gamma hemolysis, cellulolytic activity (only the bacillus Ba4), and high growth.Lactic acid bacteria are widely used as starter cultures for the production of fermented foods (Ammor and Mayo, 2007).
In addition to their role in fermentation, bacteria are an important source of high quality protein with up to 70% (Aas et al., 2006).The growth of bacteria during the fermentation process was better with 10% molasses and the pH decreased significantly from 24 to 144 h with all percentages of molasses.On the other hand, the growth of bacteria in flour with different percentages of moisture was very similar, differing at specific points of the fermentation time.
In this work, fermentation of Lemna flour resulted in a decrease in pH and in the levels of moisture, ether extract (lipids), crude fiber (insoluble carbohydrates), ashes, total nitrogen, and energy.On the other hand, protein, total free amino acid, and nitrogen-free extract (soluble carbohydrates, such as starch and sugar) were similar between unfermented and fermented flours.
The protein content of fermented Lemna sp. was similar to results obtained by Cruz et al. (2011) in Lemna minor (24.4%) and Spirodela polyrrhiza (24.1%).On the other hand, the crude protein found in unfermented Lemna sp. was higher than that found by Cruz et al. (2011) (15.7%) and Yilmaz et al. (2004) (18.4%) in unfermented flour of L. minor; however, it was lower when compared with unfermented L. minor studied by Oludayo-Olaniyi and Omoniyi-Oladunjoye (2012) (42.2%), and with S. polyrrhiza (29.05%) (Anderson et al., 2011).Although there were not significant differences between fermented and unfermented flour on crude protein content, the observed decrement in fermented flour might be due to possible metabolic utilization by P. pentosaceus and B. pumilus, which hydrolyze protein from plants to equilibrate the C:N ratio, since molasses has low protein content (4%) and higher (50%) carbohydrate content (Cleasby, 1963).Similarly, Oyarekua (2011) observed a decrement of protein content during fermentation of pigeon pea (Cajanus cajan) without molasses or other carbon sources.On the other hand, total free aminoacids were similar between unfermented and fermented flour.Duckweed has a better essential amino acid profile than most vegetable protein sources (Chaturvedy et al., 2003).Moreover, the amino acid composition is similar to that of fish meal, although, it is deficient in lysine (Aas et al., 2006).
In this work, crude fiber (insoluble carbohydrates) of Lemna sp.decreased from 6.6% in unfermented flour to 3.4% in fermented flour.Fermentation reduces crude fiber content due to the cellulolytic activity of B. pulmilus.Similar results were observed by Saha et al. (2011) when fermenting Eichhornia crassipes with B. subtilis, Bacillus megaterium, and Lactobacillus acidophilus.According to Akiyama (1993), the excess of crude fiber affects the palatability and digestibility of diets in shrimp.Therefore, the reduction of crude fiber in fermented flour should be beneficial for shrimp nutrition.
The ash content of fermented flour of Lemna sp. was lower than in the unfermented one, but high if we consider the tolerance in diets for L. vannamei.Akiyama (1993) observed that the excess of ashes in diets for shrimp can affect their palatability and digestibility.However, it is important to consider that according to results obtained of protein in fermented flour, we might substitute only 22% of fish protein, therefore the excess of ashes will be diluted.
Regarding the lipid content of fermented flour of Lemna sp., the present study demonstrates that the fermentation process decreases lipids.However, lipids are low when compared with shrimp tolerance.Nguyen et al. (2012) found that a diet with high lipid content (21.1%) affects survival and weight of L. vannamei.In the same way, Akiyama et al. (1992) reported that diets above 10% lipids affect survival and growth of Penaeus monodon, with the optimum content between 6.0 and 7.5%.Adebowale and Maliki (2011) found that the fermentation process of C. cajan decreased lipids, possibly due to the activities of lipolytic enzymes that hydrolyze lipids.The fermented flour showed lower energy content than the unfermented flour.According to Adebowale and Maliki (2011), the results obtained could be attributed to the decrease in lipids and NFE.
Fermentation of Lemna flour improves its nutritional value and suitability as an alternative protein source in formulated diets for white shrimp L. vannamei.However, further research is needed about the effect in shrimp of diets with fermented Lemna sp., especially on growth performance and gene expression of trypsin and chymotrypsin.

Figure 1 .
Figure1.Neighbor-joining tree derived from partial 16S rDNA gene sequences of the five LAB isolates (BALLvHp2, BALLvHp5, BALLsI2, BALLsI4, and BALLsI6, blue marker), a bacillus isolate (Ba4, red marker), and 19 reference strains (sequence accession numbers are given after the species name).Thermotoga maritima was chosen as out group.The scale bar shows nucleotide substitution rate per site.Bootstrap probabilities as determined for 1000 replicates are given as percentage.

Figure 3 .
Figure 3. Growth kinetics of isolate Ba4 in Lemna flour supplemented with molasses at different percentages (5, 10, and 15%).Growth of bacteria (CFU/g flour) at different fermentation times.Error bars = mean ± SD.Different letters denote significant differences at P < 0.05.

Table 1 .
Characterization of LAB isolates from L. stylirostris and L. vannamei.
p-Xylene adhesion value is the mean ± standard deviation from three replicates.
bResults are shown as the mean ± standard deviation.Different letters denote significant differences.