Bioaugmentation of ensiled Caragana korshinskii Kom. with a rapid start-up Pediococcus acidilactici and cellulases: Fermentation proles, bacterial community composition and enzymatic saccharication

The present study investigated the effects of Pediococcus acidilactici and an exogenous brolytic enzyme on fermentation prole, structural carbohydrates degradation, enzymatic saccharication, and dynamics of the bacterial community of Caragana korshinskii silage. Chopped C. korshinskii was either not treated (control) or treated with P. acidilactici (PA) at 100,000 colony-forming units/g fresh weight; Acremonium cellulase (AC) at 0.3 g/kg fresh weight; or a combination of P. acidilactici and Acremonium cellulase (PA+AC). Each treatment was prepared in quadruplicate and ensiled in mini-silo bags for 3, 7, 14, 30, and 60 d, respectively. After 60 d of ensiling, all additives increased lactic acid and acetic acid concentrations (P < 0.001) and decreased propionic acid, non-protein nitrogen (NPN), and ammonia nitrogen (NH 3 -N) concentrations (P < 0.001) compared with the control. Meanwhile, the highest lactic and acetic acids and lowest non-protein nitrogen (NPN) and ammonia nitrogen (NH 3 -N) (P < 0.001) were observed in PA+AC treated silage. In addition, the application of additives decreased the silage pH values during the early-to-mid stage (3 to 30 d) with the PA+AC group having the lowest value. Compared with the control, all treatments increased ferulic acid concentration and degradation of neutral detergent ber (NDF) and acid detergent lignin (ADL) at 60 d, meanwhile, the highest values (P < 0.05) were obtained from PA+AC silage. PA treatment exhibited a lower performance in the degradation of structural carbohydrates but performed the best in glucose yield and cellulose conversion (P < 0.05). The bacterial community in all silages consisted mainly of P. acidilactici over the entire fermentation process, and the highest abundance of 6-phospho-beta-glucosidase was observed in PA treated-silage at the mid-later stage of ensiling. These results indicate that pretreating C. korshinskii improved its silage quality and potential use as a lignocellulosic feedstock for the production of bio-product and biofuel. In WSC WSC silages CP no NH 3 -N of P. acidilactici Acremonium on The addition of P. acidilactici and commercial cellulase (PA-, AC- and PA+AC- treated silages) decreased the pH values at the initial stage of ensiling (3 d), especially in P. acidilactici treatments (PA- and PA+AC- treated silages). Subsequently, the pH values declined continuously and had a signicant difference among the treatments until 30 d, and the lowest pH value was observed in PA+AC treated silage. The pH values of C. korshinskii silages at 60 d were all below 4.95, the addition of AC and PA+AC signicantly decreased the pH values compared with the control and PA groups after 60 d of ensiling. group after 30 and 60 d of ensiling. In addition, the lowest Shannon value was observed in AC group among the four treatments after 30 and 60 d of ensiling. process and reached the lowest at 30 d in PA treated silage. This further proved that P. acidilactici as a fermentation promoter played a role in the ber degradation at the early stage of ensiling. To further demonstrate the prole of lignocellulose degradation, the content of ferulic acid was determined during ensiling. Cellulose is covered by hemicellulose and lignin which limits the degradation of lignocellulose during ensiling (Pérez et al. 2002), reduces the digestibility and utilization of ruminants, and hinders bioenergy production. Ferulic acid was released after the addition of P. acidilactici and Acremonium cellulase primarily through acidolysis and/or enzymatic hydrolysis which break the linkages of the ester bonds. The silage treated with P. acidilactici exhibited a higher concentration of ferulic acid at the initial stage of fermentation, which further explained the reduction of aNDF and ADF contents of the silages. As expected, the highest content of ferulic acid was obtained in PA+AC treated silage after 60 d of ensiling, which is due to the synergistic effects of the additives. WSC amino acids are the basic components of proteins and peptides. In a higher relative abundance of amino acid metabolism observed in the control at 3 d, which suggests that P. acidilactici decreased the action of amino acid metabolism in other treatments mainly due to low pH that inhibits the action of protease. There were no differences among control, PA and PA+AC treated silages in amino acid metabolism at the end of ensiling, however, lower CP content and higher NH 3 -N and NPN concentrations were observed in the control. It could be attributed to higher protein degradation and the accumulation of NH 3 N and NPN contents during the entire process of fermentation, or the different effects of epiphytic and exogenous P. acidilactici.


Introduction
Energy security becomes the dominant issue for developing countries, due to increasing global warming and climate change (Ranjan and Moholkar 2012; Alper and Stephanopoulos 2009). Renewable energy resources, such as crop straw, agricultural residues, and energy plants, are widely used in addressing environmental problems and fuel production (Abraham et al. 2020). Lignocellulosic biomass as the most abundant renewable energy resource help to circumvent the competition with cereals in biofuel production (Birgen et al. 2021). Caragana korshinskii Kom., a forage shrub species for vegetation rehabilitation, is widely planted to generally prevent and control deserti cation in the arid and semi-arid regions of China ). More than 4 million tons of C. korshinskii stubble are being generated annually in China after the stems are cut back to give way for new regrowth and make the plant ourish ). As shown in some studies, C. korshinskii exhibited great potential for silage making and as a feedstock for biofuel production (Zhang et al. 2009; Li et al. 2021). However, the utilization e ciency of C. korshinskii is not always satisfactory owing to the high lignin content and the complex ber structure (Xu et al. 2006). In addition, like other highly putrescent materials, C. korshinskii also faces storge and biotransformation challenges, which requires e cient preservation and pretreatment methods for this material to achieve a year-round operation.
Microbial anaerobic fermentation is considered an effective pretreatment method that was applied to preserve and provide high-quality feedstock for animal and biofuel production (Alper and Stephanopoulos 2009;Liu et al. 2019). However, C. korshinskii is di cult to be ensiled due to the higher buffer capacity and crude ber content (Ke et al. 2017;Li et al. 2021). Meanwhile, ingestion, digestion, and absorbption of C. korshinskii are restricted for animals due to the existence of thorns and volatiles like tannin (Jurado et al. 2009; ). Preparing C. korshinskii as silage can preserve its nutrients and soften the stipule thorn, improve its palatability (Zhang et al. 2009), and further enhance the quality of milk and meat when fed to animals (Nolan et al. 2010). Therefore, it is necessary to explore effective methods to break the brous structure of C. korshinskii. Various silage inoculants and exogenous cellulases have been applied to improve fermentation quality by accelerating acidolysis or enzymolysis to decrease the structural complexity of the cell wall Different studies about the application of lactic acid bacteria (LAB) and cellulose enzymes have been explored to degrade the cell wall of herbage, enhance the degradation of ber and improve the ruminal digestibility of silage (Khota et al. 2016; . Pediococcus acidilactici as a homofermentative LAB, can grow rapidly and produce lactic acid at high pH to inhibit spoilage organisms when compared with other bacterial strains, thereby improving fermentation quality and enhancing ber degradability (Porto et al. 2017; Alhaag et al. 2019; Zhang et al. 2020). Hence, the sole application of P. acidilactici can initiate fermentation quickly at the early stage of ensiling (Yang et al. 2019; Bai et al. 2021), while the combination of P. acidilactici and cellulase could be an effective way to improve ber degradability and fermentation quality. We hypothesized that the application of a rapid start-up LAB alone, or in combination with cellulase, before ensiling could either improve fermentation quality or promote ber degradability as well as reduce the loss of nutrients of the ensiled forage. To our best knowledge, however, far less seems to be known about the effects of P. acidilactici and cellulase on the fermentation pro le, the ber degradation, and the bacterial community of C. korshinskii silage. Therefore, the objective of this study was to evaluate the fermentation pro le, structural carbohydrates degradation, and enzymatic sacchari cation of C. Korshinskii silage treated with a rapid start-up Pediococcus acidilactici strain and an exogenous cellulose enzyme.

Materials And Methods
Feedstocks and silage additives Branches (with leaves and pods) of C. korshinskii were manually harvested at the podding stage from Yuzhong County, Gansu Province, China (35°85′N, 104°12′E) on 19 June 2020. P. acidilactici and Acremonium cellulase (AC, Meiji Seika Pharma Co., Ltd, Tokyo, Japan) were applied as silage additives and were supplied as freeze-dried powders. The bacterial strain P. acidilactici was isolated from corn stalk silage and stored in our laboratory. The viable count of P. acidilactici powder was 2.5 × 10 11 colony-forming units (CFU) per gram. Based on the manufacturer's description, Acremonium cellulase activity was more than 1000 U/g, and the patent formula of plant cell wall-degrading enzymes was composed with β-Glucanase, α-Arazyme, α-Galactosidase, β-Galactosidase, and β-Xylanase. Before the onset of the experiment, P. acidilactici and Acremonium cellulase were stored at 4℃.
Silage preparation Fresh C. korshinskii was taken to the laboratory and chopped into 2 to 3 cm segments using a manual forage chopper (F80221; Wuyang County Mengba department store, Linyi, China). Subsequently, C. korshinskii was mixed into a pile and randomly separated into 84 sub-samples. Four random fresh sub-samples were collected and frozen at -20°C pending further analysis. The remaining 80 sub-samples (4 treatments × 5 time points × 4 replicates) were randomly subject to the following treatments: 1) distilled water (control); 2) P. acidilactici (PA); 3) Acremonium cellulase (AC); 4) a combination of P. acidilactici and Acremonium cellulase (PA+AC). The application rate of P. acidilactici was 1 × 10 5 CFU/g fresh weight (FW). The application rate of commercial cellulase was 0.3 g/kg FW as described by . To evenly apply the additives to the chopped forages, each additive was diluted in sterile distilled water (10 mL/kg FW). For the control, the same volume of sterile distilled water was applied. After thorough mixing, all the treated samples were ensiled in mini-silo bags (280 mm × 320 mm; Cangzhou Hualiang Packaging Co. Ltd., Hebei, China) and vacuum-sealed with a vacuum sealing machine (DZ400/ZT, Wenzhou Overseas Chinese Packaging machinery factory, Zhejiang, China). The mini-silos were fermented for 3, 7, 14, 30, and 60 d at room temperature of 25 ± 2°C, respectively.

Chemical and ferulic acid analyses
To determine the fermentation parameters, the sample (20 g) from each silo was squeezed with 180 mL of distilled water in a highspeed blender, then ltered through 4 layers of medical gauze. The ltrate was divided into 2 portions. The rst portion was acidized to a pH of around 2.0 by H 2 SO 4 (7.14 M) immediately after measuring the silage pH. The acidized liquid was ltered through a 0.22µm lter for the determination of organic acids (lactic, acetic, propionic, and butyric acids) according to the method of Zhang et al. (2021). Another portion of the ltrate was mixed with trichloroacetic acid (25%, w/v) at a ratio of 1:4 (v/v) and stood for 1 h at room temperature for deposing the true protein. Subsequently, after 15 min of centrifugation at 18,000 × g at 4°C, the supernatant was analyzed for ammonia nitrogen (NH 3 -N) and water-soluble carbohydrates (WSC) following the procedure of Thomas (1977), and nonprotein nitrogen (NPN) was measured as described by Licitra et al. (1996).
To measure the dry matter (DM) content, fresh and silage samples were dried for 72 h at 65°C by using a thermostatic drying oven was dissolved in 5 mL sodium citrate buffer (0.1 M, pH 5.0) which contained Acremonium cellulase (1mg/mL) based on cellulose content. A 200 μL of 2% sodium azide solution was added to avoid microbial contamination. Distilled water was supplemented to bring the total volume to 10 mL before the incubation. Another tube was used as substrate blank with the same dried biomass, the same volume of buffer, antimicrobial agent, and distilled water. An enzyme blank was also prepared in another tube with buffer, azide solution, water, and enzyme solution. All tubes were incubated in a constant temperature shaker for 72 h at 50°C and 160 rpm. At each 12 h interval, sampled and then terminated the reaction at 100°C for 10 min. The glucose and xylose were determined by an Agilent high-performance liquid chromatography 1200 (Agilent Technologies, Inc., Germany; column: Carbomix H-NP10, Sepax Technologies, United States; detector: Refractive Index Detector, Agilent Technologies, Inc., Germany; eluent: 0.6 mL/min, 2.5 mM H 2 SO 4 ; temperature: 55°C) after centrifugation at 10,000 × g for 10 min and ltered through a 0.22-µm lter membrane.

Statistical analysis
General linear model procedure of the SPSS 21.0 (Inc., Chicago, IL, United States) was used to analyze the data of fermentation parameters, chemical composition, and ferulic acid content according to 4 × 5 factorial experiment model: where Y ij = response variable; μ = overall mean; E i = effect of the ensiling times (i = 1, 2, 3, 4, 5); T j = effect of additive treatments (j = 1, 2, 3, 4); (E × T) ij = effect of interaction between the ensiling time and additive treatment, and e ij was the residual error. Differences in the pH, structural carbohydrates, ferulic acid, and nonstructural carbohydrate content within the same hydrolysis time were analyzed by one-way ANOVA, and the signi cance was declared at P < 0.05.

Results
Fermentation and chemical characteristics of C. korshinskii before ensiling and ensiled for 60 d The chemical composition of fresh C. korshinskii is shown in Table 1. The fresh C. korshinskii exhibited a DM content of 551 ± 7.57 g/kg FW before ensiling, and the pH value was 6.35 ± 0.01. The concentrations of aNDF, ADF, ADL, hemicellulose, and cellulose were quanti ed before ensiling, which were 335 ± 3.78, 215 ± 3.56, 57.5 ± 1.12, 120 ± 7.34, and 158 ± 2.46 g/kg DM, respectively. In addition, ferulic acid concentration was 1091 ± 1.01 mg/kg DM.   and PA+AC treatments at 7 d, but a marked difference appeared continuously with the advancing period of ensiling. The PA+AC treatment showed the lowest content of ADF from 14 to 60 d of fermentation. The higher content of ADL was obtained after the addition of P. acidilactici combined with Acremonium cellulase throughout the ensiling process except for the 30 d, and the trend was comparable with the group of PA which declined to the lowest at 30 d. With the advancement in the fermentation period, differences among treatments were obtained in hemicellulose and cellulose contents ( Fig. 1E and 1F). At the initial stage of ensiling, PA and AC treated silages had the higher content of hemicellulose, subsequently, AC and PA+AC groups showed higher content of hemicellulose than control and PA treatments from 7 to 14 d. At 30 d, the hemicellulose content was signi cantly different among the four treatments.
The PA-treated silage had the lowest hemicellulose content than other treatments after 30 and 60 d of ensiling. In addition, the trend in cellulose content was similar to that of the ADF.
Concentration of ferulic acid during C. korshinskii ensiling period The effects of additives and ensiling time on ferulic acid are presented in Fig. 2. The gure revealed that the effects are signi cant throughout the ensiling time. At the initial phase of ensiling (3 d), the addition of P. acidilactici with commercial cellulase had a higher ferulic acid concentration (1155 mg/kg DM, P < 0.05) while no signi cant difference appeared among other groups. At 7 and 14 d, the highest concentration of ferulic acid was found in PA treated silage, implying that the P. acidilactici strain contributed to the dramatic increase in the concentration of ferulic acid. Meanwhile, the ferulic acid concentration of AC treated silage declined at 14 d of ensiling, and there was no difference between the control and PA+AC treated silages. An increasing trend was observed in the AC and PA+AC groups in the mid-stage of ensiling, and the highest ferulic acid content (1155 mg/kg DM, P < 0.05) was detected when the mixture of P. acidilactici and commercial cellulase was added compared to other silages. The concentration of ferulic acid had been on a plateau during the initial and mid ensiling phases of control and PA treatments. As the fermentation advances to the end of ensiling, the trends remain similar for the ensiling time at 30 and 60 d except for control silage where ferulic acid concentration declined. The PA+AC treatment had the highest ferulic acid content (1215 mg/kg DM, P < 0.05) when compared with control, PA, and AC groups after 60 d ensilage.
Bacterial community composition and functional pro ling in C. korshinskii silage As shown in Table 3, the alpha diversity of C. korshinskii decreased after ensiling. Additives decreased the species richness (Chao 1, Observed species and ACE) compared with the control at the initial-mid phase of ensiling (  The composition of the bacterial community is shown in Fig. 3A. The epiphytic micro ora before ensiling was more complex, primarily comprised Variovorax boronicumulans (5.30%), Methylobacterium goesingense (4.36%), P. acidilactici (2.37), and others (69.48%) at the species level. Regardless of pretreatment, the relative abundance of P. acidilactici dramatically increased and dominated the bacterial community of all the silages after 3 d. In addition to P. acidilactici, there were undesirable bacteria such as Erwinia tasmaniensis in the control after 3 d of ensiling. As the ensiling period advanced from 7 to 60 d, P. acidilactici became the predominant species (> 97%) in the C. korshinskii silage treated with or without additives. The linear discriminant analysis effect size (LEfSe) analysis was used to explore the differences in bacterial communities of the four C. korshinskii silage groups during the ensiling (Fig. 3B). By comparing with the control, the application of additives inhibited the growth of undesirable bacteria during the ensiling periods from 3 to 30 d. The relative abundance of P. acidilactici was signi cantly higher in PA inoculated silage after 3 and 7 d of ensiling, and Lactobacillus paracasei became the signi cantly abundant species in PA inoculated silage from 14 to 60 d of ensiling. In the AC treated silage, Rhizobium soli was higher after 3 d of ensiling, and P. acidilactici and Lactobacillus fermentum were higher after 30 and 60 d of ensiling, respectively. Interestingly, the undesirable bacteria such as Bacillus horikoshii were higher in PA+AC treated silage after 60 d of ensiling. The microbial networks of C. korshinski silage were calculated based on the 16S rRNA gene from the bacteria with a relative abundance greater than 0.001% (Fig. 3C). Simple microbial networks were observed in silages treated with additives, especially in PA inoculated silage. P. acidilactici was positively correlated with B. horikoshii in PA inoculated silage, while negatively correlated with B. horikoshii and other undesirable bacterial species in AC and PA+AC treated silage. The interaction between P. acidilactici and other bacterial species was complicated in the control.
The four KEGG pathways were used to observe functional shifts, which were cellular processes, environmental information processing, genetic information processing, and metabolism, among the bacterial community of the four treatments (Fig. 4A). The relative abundances of cellular community and amino acid metabolism in the control were higher than those in additive treated groups at the initial stage of fermentation (3, 7, and 14 d), while the highest relative abundances were observed in the PA+AC at the 60 d of ensiling.
In addition, the relative abundances of membrane transport, nucleotide metabolism, and carbohydrate metabolism were lower in the control than in the other three groups from 3 to 14 d of ensiling, while PA+AC inoculated silage showed the lowest relative abundances among the four treatments at the end of fermentation point. EC:3.2.1.86 (6-phospho-beta-glucosidase) in enzyme classi cation (EC) database was chosen to explain the ber degradation after fermentation (Fig. 4B). The abundance of 6-phospho-beta-glucosidase was higher in PA inoculated silage than in control and AC treated silage during the entire ensiling period except for 14 d of ensiling. In addition, there were no differences observed in 6-phospho-beta-glucosidase abundance between PA and PA+AC treated silage after 3 and 7 d of ensiling, while the abundance of 6-phospho-beta-glucosidase was higher in PA inoculated silage than in PA+AC treated silage after ensiling for 30 and 60 d.
The effects of additives on enzymatic sacchari cation of C. korshinskii silage The enzymatic sacchari cation results of the 60 d C. korshinskii silages are shown in Fig. 5. The application of P. acidilactici showed the highest glucose yield (P < 0.05) when compared with the other treatments regardless of the hydrolysis times except at 24 h that had no difference compared with the control (Fig. 5A). Meanwhile, the highest cellulose (P < 0.05) conversion was obtained in PA treated silage throughout the incubation period of the enzymatic sacchari cation, and the lowest cellulose conversion was observed in Acremonium cellulase treated silage except that the treatment was not different from the PA+AC group at 36 h (Fig. 5C). For the yield of xylose, the control and PA treated silages were higher than AC and PA+AC groups. control silage exhibited the highest concentration of xylose throughout the incubation period, except at 36 h and no signi cant difference was found between control and PA treatments at 48 h (Fig. 5B). The AC and PA+AC groups had similar xylose yields throughout the hydrolysis time. Although the epiphytic LAB (P. acidilactici) in C. korshinskii before ensiling was the same as the added inoculant, the differences occurred due to the abundance and function. In the present study, the application of P. acidilactici decreased the relative abundance of Erwinia tasmaniensis and other epiphytic bacteria after 3 d of ensiling, and the P. acidilactici maintained the highest relative abundance throughout the entire fermentation period. This could be attributed to the adaptability as well as the rapid growth and multiplication of P. acidilactici which produces higher lactic acid that swiftly declined the pH to inhibit the growth of spoilage microorganisms throughout the fermentation (Yang et al. 2019; Bai et al. 2021). LEfSe analysis was used to further explore the differences of bacterial community among the control and additive treatments. The application of additives had weakened the growth of epiphytic competitors at the initial-mid stage of ensiling. Subsequently, the competitiveness of some undesirable bacteria such as B. horikoshi resulted in the species increase in the PA+AC treatment at the end of ensiling. In the control treatment, other species subsequently decreased while the epiphytic P. acidilactici abundance increased with the adaptation to the acidic environment at the end of fermentation. Our previous studies showed that a relatively simple bacterial interaction was attributed to a high fermentation quality, which led to a lower alpha diversity Bai et al. 2021). In the current study, the additives simpli ed the bacterial interaction network after ensiling, especially PA treated silage showed the simplest bacterial interaction network structure which could be due to the coaction of epiphytic and exogenous P. acidilactici and result in a better fermentation quality. silages with high relative abundances of nucleotide and carbohydrate metabolisms had a higher relative abundance of P. acidilactici. And the relative abundance of carbohydrate metabolism was higher in AC than in the other treatments, which was in line with the higher WSC content. Generally, amino acids are the basic components of proteins and peptides. In this study, a higher relative abundance of amino acid metabolism was observed in the control at 3 d, which suggests that P. acidilactici decreased the action of amino acid metabolism in other treatments mainly due to low pH that inhibits the action of protease. There were no differences among control, PA and PA+AC treated silages in amino acid metabolism at the end of ensiling, however, lower CP content and higher NH 3  In addition, silages with no additive exhibited a higher xylose yield than AC and PA+AC treatments, it might be because of the highest contents of aNDF and ADF after 60 d of ensiling, and they were easier to hydrolyze into xylose during enzymatic sacchari cation experiment.

Conclusions
After 60 d of ensiling, all additives improved the fermentation quality, as seen in higher lactic and acetic acids contents, and decreased NPN and NH 3 -N concentrations of the silage. All additives also decreased the pH value, with PA+AC treated silage having the lowest pH at the early-to-mid ensiling period. Application of P. acidilactici and brolytic enzyme increased the ferulic acid content and digestibility of aNDF and ADL after 60 d of fermentation, hence, the highest ferulic acid content and lowest aNDF, ADL, and cellulose contents were observed in the PA+AC treated silage. PA treated silage had the highest lignocellulose conversion during enzymatic sacchari cation, and also had the highest abundance of 6-phospho-beta-glucosidase. However, combined P. acidilactici with the Acremonium cellulase exhibited better structural carbohydrates' degradation during the fermentation. Therefore, sole P. acidilactici treatment or in combination with Acremoniuum cellulase is a potential method that can enhance the utilization of C. korshinskii and reduce the waste of the lignocellulose resources, and subsequently serves as a biomass feedstock for biofuel production.   Bacterial community composition, differences and interactional neworks of ensiled C. korshinskii. Treatment: control, without additive; PA, P. acidilactici; AC, Acremonium cellulase; PA+AC, a combination of P. acidilactici and Acremonium cellulase. Arabic number indicating days of ensiling.
A. Comparison of microbiota compositions at species level of fresh and ensiled C. korshinskii as in uenced by additives and ensiling period.
B. Comparison of the communities or species that have signi cant differences among different additive treatments and ensiling time using the LEfSe analysis.
C. Comparison of interaction networks of the C. korshinskii silage microbiota. Node size is scaled based on the overall abundance of each taxon in the microbiota. Edge width is proportional to the strength of association between each metabolite-phylotype pair (as measured by the correlation), red edge indicates positive correlations and green edge indicates negative corrections.

Figure 4
Functional prediction of bacterial changes in C. korshinskii after fermentation using Phylogenetic Investigation of Communities by Reconstruction of Unobserved States (PICRUSt2). Treatment: control, without additive; PA, P. acidilactici; AC, Acremonium cellulase; PA+AC, a combination of P. acidilactici and Acremonium cellulase.
A. Level 2 KEGG orthologue gene of ensiled C. korshinskii as in uenced by additives and ensiling period. Arabic number indicating days of ensiling.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. Graphicabstract.pdf