Sustainable Production of Lactic Acid Using a Perennial Ryegrass as Feedstock—A Comparative Study of Fermentation at the Bench- and Reactor-Scale, and Ensiling

Abstract

Perennial ryegrass (Lolium perenne) is an underutilized lignocellulosic biomass that has several benefits such as high availability, renewability, and biomass yield. The grass press-juice obtained from the mechanical pretreatment can be used for the bio-based production of chemicals. Lactic acid is a platform chemical that has attracted consideration due to its broad area of applications. For this reason, the more sustainable production of lactic acid is expected to increase. In this work, lactic acid was produced using complex medium at the bench- and reactor scale, and the results were compared to those obtained using an optimized press-juice medium. Bench-scale fermentations were carried out in a pH-control system and lactic acid production reached approximately 21.84 ± 0.95 g/L in complex medium, and 26.61 ± 1.2 g/L in press-juice medium. In the bioreactor, the production yield was 0.91 ± 0.07 g/g, corresponding to a 1.4-fold increase with respect to the complex medium with fructose. As a comparison to the traditional ensiling process, the ensiling of whole grass fractions of different varieties harvested in summer and autumn was performed. Ensiling showed variations in lactic acid yields, with a yield up to 15.2% dry mass for the late-harvested samples, surpassing typical silage yields of 6–10% dry mass.

1. Introduction

In the past years, fossil resources have been used for energy and material production purposes without considering their limited reserves [1]. Their overuse has caused severe concerns such as global warming and reductions in natural resources. Thus, there is an urgent need to identify alternative resources. Renewable organic materials, such as biomasses, have attracted significant attention for the sustainable production of chemicals, fuels, and materials, offering important economic, environmental, and social benefits [2]. It is estimated that 1 kg of biomass raw material can be transformed into 6 MJ of energy or 0.8 kg of chemicals [2,3]. Lignocellulosic biomass from agriculture, forestry, and urban waste has huge potential due to its high abundance, renewability, and low cost. It has been assessed that more than 200 value-added compounds can be produced from lignocelluloses and some even at the industrial scale through microbial fermentation. In particular, these compounds include platform chemicals such as lactic acid, succinic acid, itaconic acid, glycerol, and sorbitol [3]. Among the others, perennial ryegrasses are considered an important biomass resource due to their high and stable biomass yield, which accounts for approximately 2–20 ton (dry matter)/ha [4].

As mentioned above, lactic acid (LA) is an important platform chemical with a wide range of applications, especially in food, chemical, cosmetic, and pharmaceutical markets. More recently, LA has reached the plastic sector for the production of biodegradable and biocompatible polylactic acid (PLA) polymers [5,6]. Because of the broad application areas, the corresponding market is estimated to reach 1960 kt in 2025 [7,8]. The production of LA using microbial fermentation accounts for more than 90% of the total output with high yields of up to 90–95 wt% [8,9]. The cost of LA depends on its final application and ranges between 1.38 USD/kg for the lowest purity and 1.59 USD/kg for the highest purity [10]. However, feedstock costs account for the 40–70% of total production costs [7,10]. Therefore, the exploitation of underutilized materials such as lignocelluloses can contribute to lowering the final price. Several attempts have been made to use lignocellulosic biomass for LA production. Table 1 summarizes the main findings over the last years.

Traditionally, lactic acid bacteria (LAB) are the microorganism of choice for LA production, as they produce LA as major product using homo- or heterofermentation, and are recognized as safe [11]. Homolactic bacteria such as Lactobacillus, Lactococcus, and Enterococcus produce two molecules of LA per mol of sugar with a theoretical yield of 1 g/g (Figure 1), whereas heterolactic bacteria yield 0.5–0.6 g of LA per g of sugar [12]. Some LAB are defined as facultative heterolactic since they metabolize hexoses and pentoses to produce LA along with by-products (acetic acid, ethanol, and formic acid) [13].

Since homolactic bacteria can reach yields close to the maximum theoretical one, they are more suited for industrial production [11].

Table 1. Microorganism, feedstock, and type of cultivation used for LA production over the last years. SSF indicates Simultaneous Saccharification and Fermentation, and SSCF indicates Simultaneous Saccharification and Co-Fermentation.

Microorganism	Feedstock	Cultivation	Yield [g/g]	Ref.
L. rhamnosus LA-04-01 a	Defatted rice brain hydrolysate	Batch	0.95	[14]
		Continuous	0.98
L. paracasei 7BL a	Wood chips	Fed-Batch	0.96	[15]
	Rice straw		0.97
L. casei G-02 a	Jerusalem artichoke	Fed-Batch (SSF)	0.96	[16]
L. agilis LPB 56 b	Soybean vinasse	Batch	0.85	[17]
L. plantarum NCIMB 8826 a	Delignified hardwood pulp	Batch (SSF)	0.88	[18]
L. paracasei LA104 a	Curcuma longa waste	Batch (SSCF)	0.69	[19]
L. coryniformis ATCC 25600 b			0.65
L. pentosus FL0421 a	Corn stover	Fed-Batch (SSF)	0.66	[20]
L. pentosus DSM20314 a	Wheat bran	Batch	0.73	[21]
L. delbrueckii NBRC 3202 b	Cassava fibrous waste	Batch	0.50	[22]
L. rhamosus ATCC 7469 a	Lignocellulosic mixture	Batch (SSF)	0.97	[23]
L. delbrueckii subsp. bulgarius ATCC 11842 b	Beechwood hydrolysate	Batch (SSF)	0.69	[24]
	Pine hydrolysate		0.40
L. delbrueckii DSM 20074 b	Household bio-waste	Batch	0.65	[25]
L. casei DSM 20011 a	Agro-industrial waste	Batch	0.78	[26]
L. delbrueckii subsp. lactis DSMZ 20072 b	Brewers’ spent grain	Batch	0.89	[27]
L. brevis MTCC 4460 c	Cottonseed cake	Batch (SSCF)	0.22	[28]
	Wheat straw		0.49
	Sugarcane bagasse		0.52
L. buchneri NRRL B-30929 c	Elephant grass liquor	Batch	0.50	[29]
B. coagulans A107 b	Tapioca starch hydrolysate	Continuous	0.80	[7]
B. coagulans AD b	Corn stover hydrolysate	Continuous	0.95	[30]
B. coagulans LA204 b	Corncob	Fed-Batch (SSF)	0.77	[31]
B. coagulans LA1507 b	Sweet sorghum bagasse	Open-Fed-Batch (SSF)	0.44	[30]
E. faecalis RKY1 b	Molasses	Batch	0.95	[32]
E. faecium WH51-1 b	Corn steep water effluent	Batch	0.89	[33]

^a facultative heterofermentative LAB. ^b homofermentative LAB. ^c heterofermentative LAB.

Table 1. Microorganism, feedstock, and type of cultivation used for LA production over the last years. SSF indicates Simultaneous Saccharification and Fermentation, and SSCF indicates Simultaneous Saccharification and Co-Fermentation.

Microorganism	Feedstock	Cultivation	Yield [g/g]	Ref.
L. rhamnosus LA-04-01 a	Defatted rice brain hydrolysate	Batch	0.95	[14]
		Continuous	0.98
L. paracasei 7BL a	Wood chips	Fed-Batch	0.96	[15]
	Rice straw		0.97
L. casei G-02 a	Jerusalem artichoke	Fed-Batch (SSF)	0.96	[16]
L. agilis LPB 56 b	Soybean vinasse	Batch	0.85	[17]
L. plantarum NCIMB 8826 a	Delignified hardwood pulp	Batch (SSF)	0.88	[18]
L. paracasei LA104 a	Curcuma longa waste	Batch (SSCF)	0.69	[19]
L. coryniformis ATCC 25600 b			0.65
L. pentosus FL0421 a	Corn stover	Fed-Batch (SSF)	0.66	[20]
L. pentosus DSM20314 a	Wheat bran	Batch	0.73	[21]
L. delbrueckii NBRC 3202 b	Cassava fibrous waste	Batch	0.50	[22]
L. rhamosus ATCC 7469 a	Lignocellulosic mixture	Batch (SSF)	0.97	[23]
L. delbrueckii subsp. bulgarius ATCC 11842 b	Beechwood hydrolysate	Batch (SSF)	0.69	[24]
	Pine hydrolysate		0.40
L. delbrueckii DSM 20074 b	Household bio-waste	Batch	0.65	[25]
L. casei DSM 20011 a	Agro-industrial waste	Batch	0.78	[26]
L. delbrueckii subsp. lactis DSMZ 20072 b	Brewers’ spent grain	Batch	0.89	[27]
L. brevis MTCC 4460 c	Cottonseed cake	Batch (SSCF)	0.22	[28]
	Wheat straw		0.49
	Sugarcane bagasse		0.52
L. buchneri NRRL B-30929 c	Elephant grass liquor	Batch	0.50	[29]
B. coagulans A107 b	Tapioca starch hydrolysate	Continuous	0.80	[7]
B. coagulans AD b	Corn stover hydrolysate	Continuous	0.95	[30]
B. coagulans LA204 b	Corncob	Fed-Batch (SSF)	0.77	[31]
B. coagulans LA1507 b	Sweet sorghum bagasse	Open-Fed-Batch (SSF)	0.44	[30]
E. faecalis RKY1 b	Molasses	Batch	0.95	[32]
E. faecium WH51-1 b	Corn steep water effluent	Batch	0.89	[33]

^a facultative heterofermentative LAB. ^b homofermentative LAB. ^c heterofermentative LAB.

Several key parameters influence LA production, including sugar concentration, nutrient availability, pH, temperature, and acid stress [12]. Specifically, LA release can lead to product inhibition due to pH reduction, which in turn affects membrane potential. Therefore, maintaining an optimal pH in the fermentation medium is essential [12]. Regarding the nutrients, LAB have complex nutrients requirement, as they have a limited capacity to synthesize molecules for their growth. For instance, nitrogen must be supplemented in the fermentation media as yeast extract, peptone, or meat extract since LAB are unable to synthesize essential nitrogen-based substances such as amino acids, purines, pyrimidines, and co-factors [34,35]. Both mineral elements and vitamins are crucial in many enzymatic reactions, in membrane transport processes, and in molecules and structural complexes [35,36,37]. However, the addition of those elements increases the overall cost of LA production.

Figure 1. Schematic representation of the LA homo-fermentation pathway from sugar to LA. The red box represents glycolysis, the green box homolactic fermentation. The numbers in the circles are the enzymes involved in that step: (1) hexokinase, (2) phosphoglucose isomerase, (3) phosphofructokinase, (4) aldolase, (5) triose phosphate isomerase, (6) glyceraldehyde 3-phosphate dehydrogenase, (7) phosphoglycerate kinase, (8) phosphoglyceromutase, (9) enolase, (10) pyruvate kinase, (11) lactate dehydrogenase. The figure is adapted from Pessione A. et al. [38] and it was created on BioRender.com.

Figure 1. Schematic representation of the LA homo-fermentation pathway from sugar to LA. The red box represents glycolysis, the green box homolactic fermentation. The numbers in the circles are the enzymes involved in that step: (1) hexokinase, (2) phosphoglucose isomerase, (3) phosphofructokinase, (4) aldolase, (5) triose phosphate isomerase, (6) glyceraldehyde 3-phosphate dehydrogenase, (7) phosphoglycerate kinase, (8) phosphoglyceromutase, (9) enolase, (10) pyruvate kinase, (11) lactate dehydrogenase. The figure is adapted from Pessione A. et al. [38] and it was created on BioRender.com.

The aim of this work was to compare the ability of the homofermentive Lactobaillus delbrueckii subsp. lactis to produce LA in conventional complex medium (De Man—Rogosa—Sharpe (MRS) medium) and in press-juice from a perennial ryegrass (Lolium perenne). Lolium perenne was selected as feedstock since it has several appealing traits including fast growth, low water and fertilizer requirements, and an abundance of above-ground biomass (leaves, stalks, and nodes) which can be harvested and used in biorefineries [39]. The entire production process was designed to enhance sustainability. Generally, lignocellulosic biomasses undergo physical pretreatments prior to any other methods to facilitate enzymatic hydrolysis [40]. In this study, a mechanical pretreatment using a screw press was used as the sole method [41]. Mechanical methods are usually considered more eco-friendly as they avoid the utilization of chemicals or heat [40]. Additionally, in a general biorefinery scheme, the monomeric sugars obtained after saccharification must be recovered. However, the recovery process requires additional energy and power [42]. The use of screw-pressing bypassed the need for an additional recovery step, as the sugars are already contained in the press-juice and can be immediately used in the fermentation process. The remaining solid fraction, also known as press-cake, is generally used as a substrate in enzymatic hydrolysis and in solid-state fermentation [43,44] for animal feed and for biogas production [45].

Different juice percentages were tested for online growth measurements to determine the optimal juice concentration for bacterial growth. Then, based on previous findings, the selected press-juice concentration was optimized with glucose, sodium-acetate, and Tween 80 to ensure LA production [27]. The experiments in both complex medium and in press-juice medium were firstly performed at bench-scale using a pH-control system to counteract the pH drops in the cultivations. This system aimed to eliminate the addition of neutralizing agents that make the downstream process difficult. Then, the experiments in both media were scaled up in a bioreactor and various fermentation parameters including sugar uptake and sugar consumption rate, production yield, productivity, and selectivity were compared. Finally, a preliminary study on the ensiling process on different Lolium perenne varieties was carried out.

2. Materials and Methods

2.1. Raw Material

At the Julius Kühn Institute (Braunschweig, Germany), a field experiment was conducted to produce fresh biomass for investigating its use as a raw material in a biorefinery. The study site is located in Braunschweig at an altitude of 80 m above sea level. The region has an annual precipitation of 617 mm and an average temperature of 9.1 °C. The soil is characterized as silty sand with a topsoil depth of 30 cm. Selected varieties of Lolium perenne (

Table 2. Lolium perenne varieties used in this study with information on ploidy and breeder.

Variety	Ploidy	Breeder
Agaska	2×	DLF (Roskilde, Denmark)
Honroso	2×	DSV (Asendorf, Germany)
Arvicola	4×	Feldsaaten Freudenberger (Krefeld, Germany)
Barmigo	4×	Barenbrug (Nijmegen, The Netherlands)
Explosion	4×	DSV (Asendorf, Germany)

) were cultivated in large-scale plots to produce sufficient biomass for subsequent analysis. In order to obtain representative samples for Near Infrared Spectroscopy (NIRS) quality determination, chopped samples of at least 500 g were collected from each plot for dry matter (DM) determination and subsequent grinding. These samples were dried in a drying oven (Memmert GmbH, Schwabach, Germany) at 60 °C for 48 h and milled uniformly with a Brabender cutting mill (Anton Paar TorqueTec GmbH, Duisburg, Germany). Quality parameters including crude protein (CP), water soluble carbohydrates (WSCs), and crude fiber (CF) were analyzed using a Foss NIRSystem 5000 spectrometer (Foss GmbH, Hamburg, Germany) with a wavelength range of 1100 to 2498 nm at a resolution of 2 nm. For the biorefinery application, fresh biomass samples were harvested, vacuum sealed, and immediately cooled for shipment.

These varieties were selected to encompass a range of genetic diversity and breeding backgrounds. The evaluation aimed to assess biomass yield and quality characteristics under the specified conditions, critical for their potential use in biorefinery processes. The implementation of standardized procedures for sample collection, drying, and analysis ensures the reliability and comparability of the results across different varieties.

2.2. Pretreatment and Preparation of Sterile Press-Juice

The pretreatment of the samples and the production of the press-juice can be found elsewhere [41]. Briefly, the raw material was mechanically pretreated using a screw press (Angel Juicer 7500, Luba GmbH, Bad Homburg, Germany). The liquid fraction (press-juice) was centrifuged (4500 rpm, 20 min; Z38K, Hermle Labortechnik GmbH, Wehingen, Germany) to remove the remaining solid particles, and filter sterilized (Stericup^®, 0.2 µm pore size, MerckMillipore, Burlington, MA, USA). The amount of press-juice obtained was approximately 52% of the total mass of the raw material.

The pressed juice (pH 5.3, Microprocessor pH 211, Hanna Instruments Deutschland GmbH, Wertheim, Germany) was analyzed in terms of sugars (approximately 40 g/L), organic nitrogen (approximately 14.87 g/L), and ions (15.75 g/L, including cations and anions). A detailed description of the composition can be found elsewhere [46].

2.3. Microorganism, Preculture, and Complex Medium Preparation

The bacteria Lactobacillus delbrueckii subsp. lactis DSM 20729 was obtained from the German Collection of Microorganisms and Cell Cultures (DSMZ, Braunschweig, Germany).

Precultures and the experiments in complex media were carried out using MRS medium with the following composition: casein peptone 10 g/L, meat extract 10 g/L, yeast extract 5 g/L, Tween-80 1 g/L, K₂HPO₄ 2 g/L, sodium-acetate 5 g/L, (NH₄)₃ citrate 2 g/L, MgSO₄ × 7 H₂O 0.20 g/L, MnSO₄ × H₂O 0.05 g/L, glucose or fructose carbon source approximately 20 g/L. The pH of the medium was adjusted to 6.2 ± 0.1 (Microprocessor pH 211, Hanna Instruments Deutschland GmbH, Wertheim, Germany) using 2.5 M NaOH. In all the experiments, the carbon source (glucose or fructose) was prepared and autoclaved (Systec V-150, Systec GmbH & Co. KG, Linden, Germany) separately to avoid degradation. All the chemicals were purchased from Carl Roth + Co KG (Karslruhe, Germany), except glucose and fructose (Sigma Aldrich, Merck KGaA, Darmstadt, Germany). Precultures were anaerobically grown in 150 mL glass bottles with 100 mL working volumes. The bottles were tightly sealed with rubber septa and incubated at 37 °C, 150 rpm for approximately 24 h (Ecotron, Infors, AG, Bottmingen, Switzerland).

2.4. Cultivation in Different Percentages of Press-Juice

The cultivations in different percentages [10, 25, 30, 50, 60, 75% (v/v) (volume per volume)] of press-juice were carried out in 100 mL serum bottles with 50 mL working volumes, which were tightly sealed with rubber septa. The remaining percentages were filled with MRS medium. Before the inoculation (0.1 OD (optical density)), the bottles were sparged with pure nitrogen. Incubation was carried out at 37 °C and 150 rpm in an incubator shaker (Ecotron, Infors AG, Bottmingen, Switzerland) equipped with a Cell Growth Quantifier (Scientific Bioprocessing, Aquila BioLAB GmbH, Baesweiler, Germany). The system recorded the growth of the bacteria by taking scattered light measurements and the values were directly used to develop growth curves.

2.5. Bench-Scale Fermentation in Schott Bottles

Bench-scale fermentations were carried out in 150 mL Schott bottles with 100 mL working volumes. Before the inoculation (0.1 OD), the bottles were sparged with pure nitrogen. The pH values were constantly measured using pHenomenal^® pH-electrodes (IDP 721, VWR, Darmstadt, Germany) and the Profilux 3.1T aquarium computer with six pH modules (GHL Advanced Technology GmbH & Co. KG, Kaiserslautern, Germany). During the cultivations, the pH value was regulated to 6.8 by automatically adding 2.5 M NaOH as a titrant correction agent (accuracy ± 0.1). A heating and stirring plate (MULTI-HS 6, Carl Roth GmbH & Co. KG, Karlsruhe, Germany) was used for temperature control and for mixing of the Schott bottles. The medium temperature in the bottles was set to 37 °C, and mixing was carried out at 200 rpm using cylindrical magnetic stirrer bars with a diameter of 8 mm and a length of 20 mm (ROTILABO, Carl Roth GmbH & Co. KG, Karlsruhe, Germany). The experiments in complex medium and the experiments in press-juice medium were conducted over 25 and 28 h, respectively.

2.6. Scaled-Up Fermentation in the Bioreactor

Scaled-up fermentations were carried out in a 0.5 L BIO-STAT^® Qplus controlled bioreactor system (Sartorius AG, Göttingen, Germany). In the experiments performed with complex media, the bioreactors were filled with all the medium components and autoclaved (121 °C, 15 min), except for the sugar. Pure nitrogen was sparged overnight to ensure anaerobic conditions. Afterward, the sterilized sugar was added aseptically to the reactors, and nitrogen was sparged for one hour to remove any remaining oxygen. In the experiments with 75% (v/v) press-juice, the bioreactors were sterilized (121 °C, 15 min) before the addition of the press-juice. After sterilization, the press-juice was aseptically filled in, and pure nitrogen was sparged overnight. The optimized nutrient mixture was added afterward, and nitrogen was sparged for one hour. In all the experiments, the temperature was set at 37 °C, the stirring speed at 150 rpm, and the pH at 6.5 (EasyFermBio HB K8 120, Hamiltorn, Hamburg, Germany) by the automatic addition of 2.5 M NaOH with peristaltic pumps. In all the experiments, the reactors were inoculated to 0.1 OD using preculture previously prepared.

2.7. Ensiling of Different Varieties of Lolium perenne

To obtain lactic acid from the ensiling of Lolium perenne fresh grass, the grass was cut into small pieces. Before ensiling, dry matter was determined using a DBS 60 3 moisture analyzer from KERN (Balingen-Frommern, Germany). For each setup, a vacuum bag was filled with 50 g of fresh mass (FM) of the chosen variety, and 10 mL of a silage additive (diluted with deionized water 1:500) was added. This additive, sourced from Lactosan (Kapfenberg, Austria), consisted of a mixture of the strains Lactobacillus plantarum, Lactobacillus rhamnosus, and Pediococcus pentosaceus. After the addition, the mixture was briefly mixed, and the bag was sealed using a VMKH-300 vacuum sealer (Ggmgastro, Ochtrup, Germany). The prepared setups were placed in plastic buckets, purged with nitrogen gas, and sealed. Ensiling was conducted at room temperature for 75 days. To maintain the nitrogen environment inside the buckets, nitrogen was renewed every three days. For sampling, the designated bags were removed from the buckets.

To measure the ensiling products, complete sample-bags were removed from the ensiling and pressed. The press juice obtained was centrifuged at 16,100× g rcf for 15 min using a Centrifuge (5415d, Eppendorf, Hamburg, Germany). Afterward, the samples were diluted to a concentration within the calibration range and filtered through a 0.22 µm pore size polyethersulfone filter.

2.8. Analytical Methods

The analyses of the press-juices were carried out with regard to sugars, protein, amino acids, cations, and anions. Protein concentrations were determined using the Bradford assay (Pierce^® Coomassie Bradford Protein Assay kit, Thermo Scientific, Waltham, MA, USA) and bovine serum albumin as an internal standard (Thermo Fisher Scientific, Waltham, MA, USA). The absorbances of the calibration curve and of the samples were measured at 595 nm (Cary 60 UV-Vis, Agilent Technologies, Santa Clara, CA, USA).

Sugars and lactic acid were analyzed by HPLC (High Performance Liquid Chromatography) [Autosempler AS 6.1L and Azura pump P 6.1L (Kneur GmbH, Berlin, Germany)]. The HPLC was equipped with an Aminex HPX-87H column at 80 °C (Bio-Rad, 300 mm × 7.8 mm, Hercules, CA, USA) and a refractive index detector (RI 101 Shodex, Kawasaki, Japan). The mobile phase was 2.5 mM H₂SO₄ and the flow rate was set at 0.6 mL/min. The instrument control and the data evaluation were carried out with the Clarity Software system (Data Apex, Prague, Czech Republic). Before the analysis, the samples were diluted to a concentration inside the calibration range using external standards, and then filtered through a 0.22 µm pore size nylon filter (KX Syringe Filter Nylon, Cole Parmer GmbH, Wertheim, Germany).

Amino acids were separated using a resolve C18 column (150 mm × 3.9 mm, Waters Corporation, Milford, CT, USA) with a SecurityGuard Cartridge (C18, 4 mm × 3.0 mm ID, Phenomenex, Torrance, CA, USA) set at 30 °C. The detection was performed using an Azura photodiode-array detector DAD 2.1L at 230 nm (Knauer GmbH, Berlin, Germany). The analysis included a precolumn derivatization with ortho-phthaldialdehyde (OPA). The derivatization process is described in the Supplementary Materials. The mobile phase consisted of solvent A (0.025 M sodium-acetate anhydrous and 0.025 M NaH₂PO₄ monohydrate) and solvent B (50% methanol). The pH value of solvent A was adjusted to 7 with 10 M NaOH, then 21 mL of both tetrahydrofuran and methanol were added. The gradient elution program was as follows: from 0 to 50 min, solvent B changed linearly from 0% to 100%; from 50 to 55 min, solvent B was set as isocratic at 100%; from 55 to 60 min, solvent B changed linearly from 100% to 0%; from 60 to 67 min, solvent B was set as isocratic at 0%. The flow rate was 1.0 mL/min. Sample preparation is described in the Supplementary Materials.

Cations and anions were analyzed by ion chromatography (IC) (930 Compact IC Flex, Metrohm GmbH & Co. KG, Filderstadt, Germany) with an inline system for dialysis (930 Compact IC Flex, Metrohm, Filderstadt, Germany) and an IC Conductivity Detector (Metrohom, Filderstadt, Germany). Cations were measured with a cation column (Metrosep C6-250/4.0, Metrohm) using 4 mM HNO₃ and 0.7 mM dipicolinic acid as the mobile phase at a flow rate of 0.9 mL/min. Anions were measured with an anion column (Metrosep A Supp5-250/4.0, Metrohom) using 1 mM NaHCO₃ and 3.2 mM Na₂CO₃ as the mobile phase at a flow rate of 0.7 mL/min. The oven temperature in both cases was 35 °C. For each analysis, the samples were diluted to a concentration inside the external calibration range (using 2 mM HNO₃ for cation determination).

Cell growth was monitored by optical density measurements at 600 nm (UV spectrophotometer LAMBA Bio+, Perkin Elmer Corporation, Waltham, MA, USA). The samples collected during the fermentations were centrifuged at 18,407× g rpm for 7 min (Eppendorf, Hamburg, Germany) and the supernatant stored at −20 °C for metabolite analysis.

2.9. Data Processing and Evaluation

The percentage of sugar uptake (S_u) and the sugar consumption rate (r_s) were calculated at the time the cells were growing exponentially and followed Equations (1) and (2), respectively:

$$S_u={\displaystyle \frac{S_i-S_t}{S_i}}\times 100$$

(1)

$$r_s={\displaystyle \frac{S_t-S_i}{t}}$$

(2)

where S_i is the initial sugar(s) concentration(s) and S_t is the sugar(s) concentration(s) at time t.

The selectivity (S) was calculated following Equation (3):

$$S={\displaystyle \frac{{LA}_f}{S_i-S_t}}$$

(3)

The production yield (Y_P/S) of LA was expressed as Equation (4):

$$Y_{P/S}={\displaystyle \frac{{LA}_f-{LA}_i}{S_i-S_f}}$$

(4)

where LA_f indicates the final LA concentration and LA_i is the starting LA concentration.

The volumetric LA productivity (Q_LA) was estimated as (5):

$$Q_{LA}={\displaystyle \frac{C_{LA}}{t}}$$

(5)

C_LA represents the concentration of LA at the end of the fermentation.

The biomass production yield (Y_X/S) followed Equation (6):

$$Y_{X/S}={\displaystyle \frac{X_f-X_i}{S_i-S_f}}$$

(6)

where X_f indicates the final biomass concentration and X_i is the initial biomass concentration.

The volumetric cell productivity (Q_CELL) was expressed as the ratio between the final biomass concentration (C_CELL) and the time t, according to (7):

$$Q_{CELL}={\displaystyle \frac{C_{CELL}}{t}}$$

(7)

3. Results and Discussion

3.1. Influence of Raw Material

The 2021 harvests were analyzed in terms of yield and quality parameters in order to diversify their use in a biorefinery context. In particular, DM, CF, CP, and WSCs were evaluated and the results are reported in

Table 3. Yield and quality parameters of Lolium perenne varieties. DM: dry matter (in tons per hectare); CF: crude fiber; CP: crude protein; WSCs: water soluble carbohydrates; BBCH stage: phenological development stages of plants.

Variety	Cut of the Year	Harvest Date	BBCH Stage	DM (t/ha)	CF (%)	CP (%)	WSCs (%)
Agaska a	2	7 July 2021	49	2.5 ± 0.4	25.2 ± 1.4	11.6 ± 1.1	16.2 ± 0.6
Arvicola a	2	7 July 2021	31	2.5 ± 0.5	21.8 ± 1.4	10.9 ± 0.6	22.2 ± 0.9
Barmigo a	2	7 July 2021	45	2.5 ± 0.4	24.8 ± 0.2	10.5 ± 0.6	18.0 ± 0.3
Explosion b	1	3 June 2022	61	5.2 ± 0.0	25.9 ± 0.3	5.8 ± 0.7	22.6 ± 3.0
Honroso b	1	3 June 2022	59	4.3 ± 0.0	25.3 ± 1.6	7.9 ± 1.3	20.4 ± 0.3

^a Data and standard deviations represent mean values of triplicates. ^b Data and standard deviations represent mean values of duplicates.

.

Agaska exhibited the highest DM yield, followed by Barmigo and Arvicola (Table 3). This suggests that Agaska is the most productive variety for the second cut of the year in terms of biomass yield, which is beneficial for processes requiring large quantities of raw material, such as bioenergy production. However, it is important to note that the first cut of the year typically produces more DM than the second cut. In terms of CF content, Agaska also had the highest amount, indicating a robust cell wall structure. This high CF content can be advantageous for producing materials that require high fiber content. Barmigo and Arvicola had slightly lower CF contents, with Arvicola having the lowest, which might make it more suitable for processes requiring lower fiber content and higher digestibility. Arvicola had the highest CP content, making it potentially more valuable for animal feed applications where protein content is critical. Barmigo and Agaska had lower CP contents, suggesting that while Arvicola is less productive in terms of biomass yield, its higher protein content could offer advantages in feed applications. WSCs are crucial for fermentation processes. Arvicola again led with the highest WSC content, followed by Barmigo and Agaska. The high WSC content in Arvicola highlights its potential suitability for processes like chemical and fuel production, where a high sugar content is desirable. Separately, the comparison between the Explosion and Honroso varieties harvested in 2022 revealed some differences. Both varieties had similar DM yields, with that of Honroso slightly lower than that of Explosion. The CF content was comparable between the two, indicating a robust structure for both varieties. However, Honroso had a higher CP content compared to Explosion, while the WSC content was similar between the two, making both varieties viable options for biorefinery processes. For lactic acid production, both the Explosion and Honroso varieties were particularly suitable due to their high DM yields and adequate carbohydrate contents. The large volume of juice required for LA production was obtained by utilizing both replicates.

3.2. Press-Juice Analysis

LAB have high and complex nutrient requirements to grow and to produce LA. To evaluate the suitability of different press-juices, their nutrient contents including sugars, proteins, amino acids, and ions were analyzed. The results are reported in

Table 4. Analysis of the nutrient contents of different varieties of grass press-juices after mechanical pretreatment (grass samples from 2021). The data presented in the table were obtained from one measurement.

	Arvicola	Agaska	Barmigo
Total sugar [g/L]	23.8	39.5	28.3
Protein [mg/L]	559.4	682.9	641.7
Amino acids [mg/L]	n.a.	3130.5	2560.9
Cations [g/L]	9.3	8.1	8.2
Anions [g/L]	6.0	5.9	4.8

n.a. not analyzed.

and

Table 5. Analysis of the available monosaccharides (glucose and fructose) in grass press-juices (grass samples from 2022).

Sample	Glucose Content [% (m/m)]	Deviation	Fructose Content [% (m/m)]	Deviation
Honroso	2.93%	0.41%	5.32%	1.20%
Explosion	5.09%	0.79%	7.56%	3.00%

.

3.3. Bench-Scale Fermentation in Complex Medium

To assess the ability of L. delbrueckii to grow and to produce LA, bench-scale fermentations in MRS medium were initially performed under pH-controlled conditions using NaOH addition. During LA fermentation, the pH of the broth drops below the LA-pKa value (3.8), which inhibits the metabolic functions of LAB [12,47]. Neutralizing agents such as Ca(OH)₂ or CaCO₃ are typically used to maintain the pH within the optimal range (pH 5–7) [35,47]. However, these agents have some disadvantages including limited LA accumulation and the formation of calcium lactate, which can increase the costs of the downstream processing [47]. Therefore, pH control is critical for achieving successful sustainable LA production.

The fermentations in MRS medium lasted for 25 h (Figure 2). Bacterial growth exhibited a lag-phase of approximately 6 h, during which glucose levels remained constant and no LA was produced. The lag-phase was followed by the exponential growth-phase (6–21 h), during which glucose was rapidly metabolized (87% with respect to the initial glucose concentration) and LA was produced (17.6 ± 0.9 g/L). By the end of the cultivation (25 h), the cell density decreased, the glucose was completely depleted, and LA accumulation reached approximately 22 g/L (Figure 2).

Mussatto et al. demonstrated that pH-controlled experiments yield better results compared to those without pH control. Using MRS medium with an initial glucose concentration of 50 g/L, they achieved approximately 23.5 g/L of LA [48]. The authors noted that glucose consumption was positively affected by the pH. Si et al. investigated different pH values and found that maintaining a pH of 6 during LA production in corn straw hydrolysate led to the highest LA production [49].

3.4. Cultivation in Different Percentages of Press-Juice

To evaluate if L. delbrueckii could grow in press-juice, cultivations in different juice percentages (10, 25, 30, 50, 75% (v/v)) were carried out. Since the experiments aimed to assess bacterial growth, online measurements with CGQ and without pH control were performed.

The results revealed that L. delbrueckii could grow in all the juice percentages, although in different manners (Figure 3). At lower juice concentrations, the latency phase was shorter than in the other cases. This was likely due to the fact more than 50% of the medium at these concentrations consisted of complex media components. Since the preculture was grown in MRS medium, it is possible to state that the bacteria were not significantly perturbated and could adapt rapidly. At higher juice percentages, the lag-phase was extended due to the change of the medium. As depicted in Figure 3, the higher the juice concentration, the longer the lag-phase lasted. This might be attributed to the differences in nutrient availability. Indeed, LAB have elaborated nutrient requirements because of their limited capability to synthesize vitamins and amino acids from inorganic nitrogen sources [35]. Therefore, an optimal fermentation medium should provide all the compounds required to support bacterial growth and metabolic activity. For instance, MRS complex medium includes multiple nitrogen sources (casein peptone, meat extract, and yeast extract) that supply several amino acids, peptides, minerals, and vitamins crucial for LAB growth [27,50]. In contrast, press-juice primarily contains nitrogen sources in the form of amino acids and proteins at a usually lower concentration than the complex medium. Furthermore, proteins in the press-juice can only be used as a nutrient source when proteolysis takes place [51]. According to this, the nutrients in the press-juice are not immediately available for LAB growth and activity.

Despite these challenges, the experiments aimed to establish whether L. delbrueckii could grow in the tested juice percentages to replace as much of the expensive complex media components as possible. Based on these findings, 75% (v/v) was selected as the juice percentage for the further experiments.

3.5. Bench-Scale Fermentation Using the Press-Juice

A concentration of 75% (v/v) of pressed juice was selected for LA production, aiming to replace expensive complex media components such as yeast extract. Previous studies have demonstrated the high nutrient content of juice from grass raw materials [52,53], suggesting that supplementation of only the essential nutrients could be economically advantageous.

Previous studies investigated the optimization of the fermentation medium using brewers’ spent grain liquor. The effect strength of each MRS medium component on LA production was analyzed, and it was found out that sodium-acetate and Tween 80 had the highest impacts [27]. It is reported that sodium-acetate enhances both final cell density and LA concentration, while Tween 80 supplies long-chain unsaturated fatty acids that increase cell membrane permeability [50,54,55].

Based on these results, 75% (v/v) press-juice was supplemented with a mixture containing 25% (v/v) glucose, sodium-acetate, and Tween 80. Preliminary experiments were conducted in Schott bottles equipped with a pH-control system to evaluate the potential of this optimized juice medium for LA production.

The results revealed that both glucose and fructose were consumed during the fermentation, with no discernible preference between the sugars (Figure 4). This finding was in line with previous investigations; Mousavi et al. observed similar sugar utilization patterns in pomegranate juice by different Lactobacilli strains [56]. Costa et al. used a mixture of pear residues and ricotta cheese whey to produce LA, revealing that fructose and glucose were completely metabolized by L. casei within 24 h. The LA reached 43 g/L after 48 h with a yield of 50.6% with regards to total sugars [26]. Volkmar et al. found out that L. delbrueckii can consume glucose and fructose contained in municipal residues, producing approximately 17 g/L of LA [57]. Although the cell growth was not monitored due to the brownish colour of the medium, bacterial growth can be inferred from LA release. During the first 6 h of cultivation, the sugars were not metabolized and no LA was produced. Subsequently, exponential LA production was observed over the next 22 h supported by rapid sugars consumption. By the end of the fermentations, approximately 88% of sugars were consumed and 26.6 ± 1.2 g/L of LA was produced. The results corresponded to a production yield of 0.78 ± 0.03 g_Sugars/g_Lactate. Boakye-Boaten N. et al. produced LA in 90% (v/v) miscanthus press-juice using L. platarum, L. brevis and the combination of both in serum bottles. The results gave 11.7, 10.3, and 10.2 g/L of LA after 48 h of fermentation, respectively [58]. Santamaria-Fernandez M. et al. produced up to 22 g/L of LA from grass clover using L. salivarius [59]. Si et al. used a pH-control system to produce LA from corn straw hydrolysate. The authors obtained 99.8 g/L of LA, which corresponded to a production yield of 0.67 g_LA/g_Sugars [49].

3.6. Comparison between Complex Medium and Press-Juice in the Bioreactor

After the evaluation of LA production in a pH-control system using either MRS medium or press-juice, both the experiments were scaled-up in stirred tank bioreactors. In MRS medium, fermentations were carried out using either glucose or fructose as a carbon source (Figure 5a,b). Fructose was included to provide a further comparative analysis since it is the primary sugar in the press-juice. The experiments with the press-juice were performed as previously described (75% (v/v) press-juice and 25% (v/v) nutrients mixture) (Figure 5c).

The choice of the carbon source significantly influenced the bacterial growth in MRS media. When glucose was used as the carbon source, the lag-phase lasted approximately 2 h (Figure 5a). In contrast, the lag-phase was extended to approximately 4 h when fructose was used as the carbon source (Figure 5b). For the press-juice experiments, OD measurements were not collected due to the brownish color of the medium. However, as for the experiments in Schott bottles, information regarding the growth can be deduced from the LA release. In this case, the lag-phase lasted 8 h, which was longer than in complex media (Figure 5c). During this phase, the cells turned on their metabolic activity, such as the activation of signaling pathways, upregulation of protein assembly, nucleotide metabolism, and other processes necessary for differentiation and duplication. All these activities result in cell division, which is fully activated during the exponential-growth phase [60,61]. The prolonged lag-phase observed in press-juice may stress the bacteria due to suboptimal or insufficient nutrient concentrations. According to Kim J. et al., the nitrogen-to-carbon ratio significantly impacts sugar conversion to LA [35]. Akermann et al. [27] reported that sodium-acetate and Tween 80 were the medium components that most affected the cultivation of L. delbrueckii. Hebert E. et al. studied the nutritional requirements of L. delbrueckii subsp. lactis and they found out that MnSO₄ and FeSO₄ are essential for the growth, since they act as a cofactor in several enzymatic reactions [35,50]. It might be that the formulation of the press-juice medium may not be optimal for bacterial growth. Additionally, the transition from the complex medium used in the preculture to the press-juice could have affected the growth. It is also reported that the microbial lag-phase is an adjustment period before active growth begins, and its duration can vary based on environmental conditions [61].

The exponential-growth phase lasted until 10 h and 18 h when glucose and fructose were used, respectively (Figure 5a,b). Throughout the cultivations, higher cell density was observed with glucose as the carbon source. The same result was obtained by Petrut S. et al., who cultivated L. rhamnosus on different sugars. The authors found out that glucose gave a higher cell density with respect to other sugars such as fructose, sucrose, mannose, and arabinose [62]. Chen H. et al. cultivated L. acidophilus on maltose, glucose, lactose, and whey powder. The authors obtained higher cells concentration (in Colony Forming Units/mL) when the bacteria were grown on glucose and whey powder [63]. During the exponential growth phase, most of the glucose and the fructose were consumed when the complex media were used. In particular, approximately 22.3 ± 3.6 g/L of glucose and approximately 16.5 ± 0.5 g/L of fructose were metabolized, which corresponded to 94.6 ± 4.5% and 77.2 ± 1.4% of sugar utilization (Table 3). The biomass yield was higher when glucose was used as the carbon source (0.17 ± 0.02 g_Biomass/g_Glucose) than with fructose (0.12 ± 0.01 g_Biomass/g_Fructose), indicating that glucose stimulated more cell growth than LA production. When the 75% (v/v) press-juice was used as the fermentation medium, the exponential-phase lasted until 28 h (Figure 5c), likely due to the higher sugar availability. During this phase, the glucose was totally metabolized, and approximately 86.8 ± 0.6% of the fructose was consumed. The corresponding LA concentrations were 14.1 ± 0.8 g/L in the experiment in MRS with glucose, 18.3 ± 2.2 g/L in the case of MRS with fructose, and 25.6 ± 3.4 g/L in pressed juice. Extending the fermentation time in complex media led to LA degradation.

The production yield and the productivity are crucial parameters in fermentation processes, and they are summarized in

Table 6. Comparison of the fermentation parameters between the bioreactor fermentations in MRS medium (with glucose or fructose) and the bioreactor fermentations in optimized 75% (v/v) press-juice. S_i is the initial sugar(s) concentration; S_u and r_S were evaluated during the exponential-growth phase.

	MRS Medium with Glucose 1	MRS Medium with Fructose 1	75% (v/v) Press-Juice 2
Si [g/L]	23.5 ± 2.9	21.3 ± 0.3	33.3 ± 0.84
Sugar(s) remaining [g/L]	0 ± 0.00	0.21 ± 00	2.9 ± 0.00
Su [%]	94.6 ± 4.5	77.2 ± 1.43	102.7 ± 3.5
rS [g/Lh]	−2.23 ± 0.36	−0.91 ± 0.03	−1.03 ± 0.04
CLA [g/L]	13.47 ± 0.57	21.45 ± 1.45	27.81 ± 2.69
S [g/g]	0.59 ± 0.11	1.02 ± 0.08	0.84 ± 0.01
YP/S [g/g]	0.63 ± 0.02	1.02 ± 0.08	0.91 ± 0.07
QLA [g/L h]	0.48 ± 0.02	0.69 ± 0.00	0.87 ± 0.10
Fermentation efficiency [%]	51.3 ± 1.9	101.7 ± 8.4	91.4 ± 6.7
YX/S [g/g]	0.17 ± 0.02	0.12 ± 0.01	n.a
QCELLS [g/L h]	0.16 ± 0.00	0.09 ± 0.01	n.a

¹ Parameters calculated at 29 h. ² Parameters calculated at 32 h. n.a. not applicable.

. The highest production yield was achieved in MRS with fructose as the carbon source, reaching the maximum theoretical yield of 1 g/g [12]. However, when 75% (v/v) press-juice was used, the production yield was slightly lower (0.91 ± 0.07 g_LA/g_Sugars). This result is comparable to those of Chen C. et al., who obtained a yield of 0.93 g_LA/g_Sugars with L. plantarum 23 using microalgae hydrolysate [64]. Similarly, LA produced from orange peel wastes using L. delbrueckii subsp. delbrueckii CECT286 reached a production yield of 0.95 g_LA/g_Sugars [65]. Thus, the results of this experiments are consistent with those reported in the literature.

In terms of productivity (Q_LA), the highest value was obtained when 75% (v/v) press-juice was used (0.87 ± 0.10 g/Lh), outperforming other studies. For instance, Pontes et al. obtained an LA productivity of 0.81 g/Lh using forest and marginal lands lignocellulosic biomass and L. rhamnosus ATCC 7469 [23,66].

3.7. Lactic Acid Yields from Ensiling Whole Grass Fractions of Different Varieties of Lolium perenne

As a preliminary study to assess the influence of different varieties on ensiling, a classical whole grass ensiling process was conducted for the first project crops in the summer and autumn of 2020. LA yields after ensiling the whole grass fraction of Lolium perenne varieties harvested in 2020 are shown in Figure 6a. The pH values dropped rapidly within the first 15 days, stabilizing around pH 4, which is indicative of a successful fermentation process. The maximum LA yield was reached after 2–3 weeks of ensiling. Samples harvested in October showed substantially higher LA yields of up to 15.2% of the grass dry mass after ensiling compared to those harvested in May and June. Among the October-harvested samples, Agaska had the highest LA yield, while Honroso had the lowest with 12.8% DM. For the samples harvested in May and June, Explosion had the highest LA yield at 11.8% DM, while Arvicola had the lowest at 8.6% DM. Among the different Lolium perenne varieties, Agaska showed the least impact, with a 24% decrease in lactic acid yield when harvested in May/June compared to October, while Barmigo exhibited the greatest reduction, with a 41.7% decrease.

High-quality silages exhibit LA yields of 6–10% based on dry matter. This range was achieved across all varieties and experiments [67]. Notably, the grasses harvested in October even exceeded 10%. The yields obtained in this study are consistent with values from the literature, with the 15.2% yield of Agaska harvested in October surpassing them. This superior yield positions Agaska as a highly efficient variety for LA production through ensiling. Johnson et al. reported yields of up to approximately 6.5% DM in 16 days with silages in vacuum bags without the addition of a silage additive [68]. Haag et al. achieved yields of 7% DM for untreated silages and silages with a biological silage additive, and up to 13% DM for silages with both biological and chemical additives [69]. Kung et al. reported yields of around 14% DM for silages with a biological additive [67]. Fazzino et al. reached 54.5 g/kg of total solids of LA after 8 weeks of ensiling using orange peel waste [70].

The WSC values for the 2020 crops could not be assessed as precisely as in the following years. Given that the WSC contents are assumed to be relatively similar across the varieties, the differences in LA yields are likely due to the inherent genetic variability among the Lolium perenne varieties, impacting their fermentation efficiency. The WSC content may include oligosaccharides that cannot be converted into LA by microorganisms. Factors such as microbial efficiency, nutrient availability, and specific conditions during ensiling also significantly influence LA yields.

The average proportions of the by-product acetic acid during the ensiling are shown in Figure 6b. In high-quality silages, the acetic acid content should be between 1% and 3% DM [67]. The upper limit for acetic acid content was maintained across all varieties. The highest acetic acid content was found in the variety Agaska (October) with 2.0% DM, while the lowest was in the variety Agaska May/June at 0.2% DM. Low acetic acid content is desirable in silage because it increases aerobic stability. However, since this work primarily focuses on lactic acid production, falling below the lower limit is not considered negative.

4. Conclusions

In this study, it was demonstrated that press-juice from perennial ryegrass can effectively replace or partially substitute expensive media components that are generally used in fermentation processes. The experiments performed in Schott bottles confirmed that the pH-control system is a fundamental parameter to control in LA production. This strategy not only eliminates the addition of neutralizing agents which complicate the downstream process, but it also reduces the total costs. The fermentation in bioreactors proved that the optimized press-juice medium allowed the production of more than 27 g/L of LA, which corresponded to a production yield of 0.91 ± 0.07 g_LA/g_Sugars. Overall, the use of press-juice as a fermentation medium not only allowed us to approach the maximum theoretical yield but also provided high productivity, making it a promising and economically viable substitute to traditional complex media for LA production.

The whole grass fraction of Lolium perenne also showed significant potential for lactic acid production. The Agaska variety, in particular, achieved a lactic acid yield of 15.2% dry mass (DM) when harvested in October, significantly surpassing typical silage yields of 6–10% DM. This study highlights the significant potential of certain Lolium perenne varieties, particularly Agaska and Explosion, for high-efficiency lactic acid production through ensiling. Barmigo and Arvicola also show reasonable efficiency, while Honroso appears less suitable for maximizing lactic acid yields. These findings suggest that selecting appropriate grass varieties is critical for optimizing lactic acid yields of whole grass ensiling. Additionally, the impact of harvest time on ensiling outcomes cannot be overlooked, as autumn harvests tend to produce better lactic acid yields.