Effects of ferulic acid esterase-producing Lactobacillus fermentum and cellulase additives on the fermentation quality and microbial community of alfalfa silage

Identification of FAE activity in Lactobacillus fermentum 17SD-2

Lactobacillus fermentum 17SD-2 used in this study was isolated from maize silage without any additives, then maintained in De Man, Rogosa, and Sharpe (MRS) broth containing 20% (v/v) glycerol at −80 °C. Before any experiments, the strain was propagated twice and cultured in MRS broth for 24 h at 37 °C, harvested by centrifugation (10,000 rpm, 2 min), and washed twice with phosphate buffered saline (PBS, pH 7.0). The sediment was resuspended in sterile water and 10 µL of the suspension was transformed to MRS plates with glucose omitted but containing 1g/L ethyl ferulate and incubated for 72 h at 37 °C, and Lactobacillus fermentum 5007 (CGMCC 1.3223) was used as negative control because it was previously confirmed to have no FAE activity. After incubation, we observed whether a clear zone formed around the suspension. If a clear zone appeared on the MRS plates, this indicated that the strain L. fermentum 17SD-2 had FAE activity.

Morphological, physiological, and biochemical characteristics of Lactobacillus fermentum 17SD-2

Gram staining and catalase activity of L. fermentum 17SD-2 were determined after 24 h of incubation on MRS agar or broth. To measure the growth curve of Lactobacillus fermentum 17SD-2, the bacterial cells at a density of 1 ×10⁷ CFUs/ml were seeded in MRS broth. The absorbance of Lactobacillus fermentum 17SD-2 culture at 600 nm were regularly monitored with a microplate reader. To evaluate salt tolerance, 2% culture was inoculated into MRS broth containing 3.0% and 6.5% NaCl, and then cultured at 37 °C for 24 h. To evaluate acid tolerance, 2% culture was inoculated into MRS broth with pH 3.5, 4.0, 4.5, 8.0, 8.5, and 9.0, and then cultured at 37 °C for 24 h. To evaluate temperature tolerance, 2% culture was inoculated into MRS broth, and then cultured at different temperatures (including 15, 20, 25, 30, 35, 40, and 45 °C) for 24 h. Then the absorbance of Lactobacillus fermentum 17SD-2 cultures at 600 nm was measured with a microplate reader. The value of OD₆₀₀ greater than 0.3 was considered as positive, expressed with “+”. API 50CH contains a high level of carbohydrates and was used to determine carbon source utilization according to the manufacturer instructions.

Alfalfa harvest and silage preparation

Alfalfa at blooming stage was manually harvested from an experimental field with an area of 5 m ×10 m, located at the Wuqing District (117°10′E, 39°10′N), Tianjin, China on June 7th, 2018. At first, the harvested alfalfa was divided into two groups: fresh (sample1, S1) and wilted (sample 2, S2, wilted in the field for 5 h). Alfalfa with various dry matter (DM) contents (S1: 23.82%; S2: 29.63%) was directly chopped into pieces in length of 2–3 cm using a crop chopper.

A commercial CE (10,000 U/g; Macklin Biochemical Co., Ltd, Shanghai, China) and L. fermentum 17SD-2 were used as additives for silage making. L. fermentum 17SD-2 could grow well at low pH conditions and possess high FAE-activity (clear zone with a diameter of 11.3 mm). The inoculant was added at a level of 10⁶ colony forming unit (CFU) per gram of fresh matter (FM). CE was applied at a ratio of 1 g kg⁻¹ of FM (Chen et al., 2017).

The experimental treatments were designed as follows. Each group (sample 1 and sample 2) included 12 piles (500 g per pile), and they were randomly ensiled with four additives treatments: (i) untreated control (CK); (ii) application of commercial cellulase (CE); (iii) application of L. fermentum 17SD-2 (LF); and (iv) combination of L. fermentum 17SD-2 and commercial CE (LF+CE). Then, chopped material was mixed homogenously with the additives and packed manually into 35 cm × 50 cm polyethylene bags which were tightly vacuumed, and triplicates for per treatment were prepared. A total of 24 bags (2 groups × 4 treatments × 3 replicates) were kept at ambient temperature (21–30 °C). After 60 days of ensiling, three bags per treatment were opened to evaluate their fermentation end products, chemical composition, and microbial communities.

Microorganism and fermentation quality analysis

To evaluate microbial counts, 10 g forage (raw materials and silage) was extracted with 90 mL 0.85% sterile physiological saline solution, and serially diluted from 10 ⁻¹ to 10⁻⁶. In total, 100 µL from an appropriate dilution was spread on agar plates. Lactic acid bacteria were enumerated on MRS agar after incubation at 30 °C for 72 h. Molds and yeasts were counted on potato dextrose agar (Nissui) after incubation at 28 °C for 24 h, and yeasts were distinguished from molds or other bacteria by colony appearance and cell morphology observation (Avila et al., 2009). Plates containing a minimum of 30 and a maximum of 300 colony-forming units were enumerated (Reich & Kung, 2010). To determine the fermentation parameters, the sample (25 g) was mixed with 225 mL sterile water and incubated at 4 °C overnight and then filtered through four layers of cheesecloth (Touno et al., 2014). The filtrate was used to measure pH and the concentrations of ammonia-N and organic acid. The ammonia-N content was determined according to phenol-hypochlorite and ninhydrin colorimetric procedures described in a previous study (Broderick & Kang, 1980). Organic acids were analyzed using high-performance liquid chromatography (HPLC), equipped with a UV detector and set as follows: ICSep COREGEL-87H column, eluent five mmol/L H₂SO₄ with a running rate of 0.6 mL/min, temperature of column oven 50 °C, injection volume of 10 µL.

Analysis of chemical composition

To determine the DM content, the samples were dried in a forced-air oven at 65 °C for 48 h and then dried at 103 °C to constant weight. The silage DM losses during drying were calculated according to the formula of Porter et al. (1995). Dried sample was grinded to 1.0 mm particle diameter. The WSC was determined using the anthrone method (Murphy, 1958) and the CP was analyzed according to the standard procedure detailed by the Association of Official Analytical Chemists (AOAC, 1990). The content of NDF and ADF were measured according to the method described in a previous study (Van Soest, Robertson & Lewis, 1991).

Bacterial diversity analysis

Samples (20 g) were mixed with 180 mL of sterile 0.85% NaCl solution with vigorous shaking at 120 r/m for 2 h. The mixture was filtered through four layers cheesecloth and the filtrate was centrifuged at 10,000 r/m for 10 min at 4 °C. The deposit was resuspended in one mL of sterile 0.85% NaCl solution and the microbial pellets were obtained by centrifugation at 12, 000 r/m for 10 min at 4 °C. To extract total DNA in raw material and silage samples (30 samples in total), the DNA Easy Power Soil Kit (Qiagen, Hilden, Germany) was used according to the manufacture’s protocols. The PCR reactions were conducted in a 20 µL mixture (10 ng of template DNA, 2 µL of 2.5 mM dNTPs, 0.8 µL of each primer (5 µM), 0.4 µL of FastPfu Polymerase, and 4 µL of 5 ×FastPfu Buffer, 0.2 µL of BSA, 11.8 µl of ddH₂O). According to Ni et al. (2017), the 16S rDNA V3-V4 regions were amplified using primers 338F (ACTCCTACGGGAGGCAGCAG) and 806R (GGACTACHVGGGTWTCTAAT). Triplicate PCR reaction for each sample was conducted to minimize PCR deviation, and a mixture of the three PCR products was sequenced.

High-throughput sequencing and bacterial diversity analysis

Amplicons were extracted from 2% agarose gels, purified with a Qiagen Gel Extraction Kit (Qiagen, Hilden, Germany) and sequenced at the Shanghai Majorbio Bio-pharm Technology Co., Ltd using paired-end sequencing (2 ×300 bp) with an Illumina MiSeq platform according to the standard protocols. Raw tags were quality-filtered by Trimmomatic (Version 3.29) and merged by FLASH (Version 1.2.7) with the following criteria: (i) the reads were truncated at any site receiving an average quality score <20 over a 50 bp sliding window; (ii) sequences with overlaps longer than 10 bp were merged according to their overlap with mismatches no more than 2 bp; and (iii) sequences of each sample were separated according to barcodes (exactly matching) and primers (allowing 2 nucleotides mismatching) and reads containing ambiguous bases were removed. Operational taxonomic units (OTUs) were clustered with 97% similarity using UPARSE (version 7.1, http://drive5.com/uparse/). OTUs were used to calculate rarefaction and alpha diversity (Mothur (v.1.30.1).

Statistical analysis

Data shown are means ± standard deviation (SD). All microbial counts data were converted to log10 and the results were described on a fresh weight basis. Two-way analysis of variance was used to evaluate the effects of water contents, additives and their interaction on microbial population, pH values, fermentation characteristic, chemical composition and alpha diversity of bacterial community in alfalfa silage. All the statistical analyses were performed using the general linear model procedure of SAS (version 9.0, 2002; SAS Institute, Cary, NC, USA). The means were compared for significance by Duncan’s multiple range method and the significance was declared at P <0.05.

Chemical and microbial composition of raw materials and the characteristics of Lactobacillus fermentum 17SD-2

The chemical composition and microbial populations of the alfalfa crops before ensiling are shown in Table 1. The DM contents of sample 1 and sample 2 were 23.82% and 29.63% of FM, respectively; their CP contents were 22.63 and 25.34% DM, respectively; and NDF and ADF concentrations were 33.3–38.3% and 25.5–29.7% DM, respectively. In this study, the WSC contents were lower than 1.8% of DM in both samples. Additionally, the numbers of epiphytic LAB, yeast, and mold were similar between sample 1 and sample 2. The count of LAB was lower than 10³ CFU/g FM, while yeast and mold counts were around 10⁴ CFU/g FM.

The characteristics of the selected LAB are shown in Table 2. L. fermentum 17SD-2 is Gram-positive, heterofermentation and catalase-negative and was able to grow at a wide range of pH (pH 3.5–9.0), a broad range temperature (15–45 °C), and a salinity of 6.5%. L. fermentum 17SD-2 used the following as carbon sources: ribose, galactose, D-glucose, D-fructose, D-mannose, maltose, lactose, melibiose, sucrose, D-raffinose, gluconate, and 5-keto-gluconate. Furthermore, this strain grew vigorously and possessed high FAE activity (Fig. 1).

Effects of CE, LF, and LF+CE on the pH values and microbial characteristics of alfalfa silage

The final pH values and microbial populations of sample 1 and sample 2 were similar after ensiling (Table 3). Mold was not detected in all silage, and the significance analysis of main factors revealed that all additives (CE, LF, and CE+LF) significantly increased the LAB (P < 0.001) numbers, but decreased the yeast (P < 0.001) numbers. All treatments had lower (P < 0.001) pH values than those of control, particularly the lowest pH (4.6–4.7) in LF+CE treatments, which was much lower than the addition of CE or LF alone.

Effects of CE, LF, and LF+CE on the fermentation quality and chemical composition of alfalfa silage

The fermentation parameters and chemical composition of the alfalfa silage after 60 days of ensiling are shown in Table 4. Lactic acid and acetic acid emerged as the main fermentation end products in all alfalfa silage. All additives remarkably increased the lactic acid content (P < 0.001), in particular, LF+CE had a significant effect on lactic acid and butyric acid content. All treatments had higher lactic acid content and ratio of lactic acid and acetic acid, and LF+CE treatments had the highest lactic acid contents. In this study, all treated silage, especially LF+CE silage, had a lower level of butyric acid content (<1.0% DM), while a large amount of butyric acid (about 2.63% DM) appeared in the control silage of sample 1. In sample 2, no great difference was observed between control and additive treatments (except for LF+CE treatment).

Although all additives decreased ammonia-N (P < 0.001) and increased CP (P < 0.001) contents compared to control, the LF+CE treated silage had the highest contents of CP and lowest ammonia-N. Compared with the addition of CE alone, lower NDF and ADF contents were found in LF+CE treatment. Overall, the silage treated with FAE-producing LAB plus cellulase was better than those of the silage treated with either LAB or cellulase alone.

Effects of CE, LF, and LF+CE on microbial communities after 60 days of ensiling

High-throughput sequencing of 16S rRNA gene amplicons was conducted to systematically describe the bacterial communities in the raw material and silage. Alpha diversity of all samples is shown in Table 5. The coverage of all samples was above 0.99, which indicated that the sampling depth adequately captured most of the bacteria. Compared with sample 1, lower Chao and Shannon indexes were observed in sample 2, which was consistent with the results reported by Wang et al. (2018), indicated that wilting (water content) had an important influence on silage microorganism. Lower observed species than pre-ensiled alfalfa crops and that of control appeared in CE alone or with LF treated silage in sample 2. In contrast, in sample 1, observed species in LF and LF+CE treated silages were higher, especially in LF treatment. Although we could not classify the behind reason clearly, probably related to the high number of the epiphytic detrimental microorganisms.

The dynamic variance of microbial population with different treatments can be demonstrated by principle co-ordinates analysis (PCoA). As shown in Fig. 2A (sample 1), component 1 and component 2 could explain 58.07% and 28.51% of the total variance, respectively; Similarity, in Fig. 2B (sample 1), component 1 and component 2 could explain 50.94% and 41%, respectively. Basically, Raw materials S1 and S2 could be well separated from all silage samples, indicating that ensiling was the main factor affecting anaerobic fermentation. Among silage samples, the distance between CK and CE treatments was near, similar phenomenon was observed between LF and LF+CE treatments, indicated that the addition of L. fermentum 17SD-2 had a significant effect on bacterial community in alfalfa silage. In addition, the variation in microbial community might be one critical factor leading to difference in silage quality.

The bacterial communities at the genus level of all alfalfa silage before and after 60 days of ensiling are shown in Figs. 2C and 2D. Before ensiling, in both raw materials (S1 and S2), Cyanobacteria and Pantoea were the most abundant genera, above 53.86% and 8.33% of the entire community, respectively. However, after fermentation, the high abundance of Cyanobacteria and Pantoea significantly decreased or even disappeared. Enterobacter and Lactobacillus gradually became the most prevalent genera. Although Enterobacter was detected in all silage, their relative abundance decreased from 49.98% to 17.01% with the addition of LF and CE, furthermore, LF alone or with CE treated silage showed much lower Enterobacter abundance than control and CE alone treatment. In addition, a similar phenomenon was observed in sample 2.

Lactobacillus, a desirable genus, was present in all silage and Lactobacillus was highly abundant in LF or LF+CE-treated silage, especially in sample 2, where this genus accounted for 85.36% and 85.50% of the total community, respectively. Many other LAB genera, i.e., Pediococcus, Enterococcus, Weissella, and Lactococcus, were also detected after ensiling. For example, in sample 2 silage, the proportions of Pediococcus and Enterococcus ranged from 10.15% to 1.26% and 11.35% to <0.1%, respectively. Weissella and Lactococcus were mainly detected in control and CE treatment. Additionally, relatively high proportions of Anaerosporobacter (5.96%), Clostridium (4.78%), and Garciella (3.65%) were observed in the control of sample 1.

Relationship between main genera and silage quality

To better understand the relationships between bacterial compositions and silage properties, a Spearman correlation heatmap was constructed at the genus level (Fig. 3). Spearman correlation analysis illustrated that Lactobacillus was positively correlated with LA/AA (r = 0.47), DM (r = 0.69), and CP (r = 0.71) contents, while it was negatively correlated with concentration of ammonia-N (r =  − 0.81) and pH (r =  − 0.47). Enterococcus and Weissella were negatively correlated with LA/AA (r =  − 0.65 and −0.51) and positively with pH (r = 0.64 and 0.51) and ammonia-N (r = 0.75 and 0.75). Enterobacter was positively correlated with the concentration of ammonia-N (r = 0.47), but was negatively correlated with CP (r =  − 0.47) content.