A Stereospecific Lactic Acid Exclusion Biosensor for Grass Silage Fed Green Biorefinery Authors: Roderick N. van Roosmalen, Cormac D. Murphy, Joseph B. Sweeney Published: 2026 ABSTRACT ABSTRACTGrass silage fed green biorefineries require high quality grass silage leachates to produce high quality products. Real‐time, cost‐effective methods to monitor key leachate analytes are needed for small‐scale decentralised systems. In this study, Escherichia coli mutants, designated JSP0090 and JSP0094, were created to quantify d‐ and l‐lactic acid concentrations in grass leachate samples using an oxygen probe. The genes encoding either d‐ or l‐lactate dehydrogenase were expressed in an exclusion biosensor strain (JSK0115). This strain required successive gene deletions to prevent interference from sugars, amino acids and organic acids present in grass silage leachates. This strain was incapable of catabolising d‐lactic acid, l‐lactic acid, acetic acid, propionic acid, formic acid, ethanol, glucose, fructose, l‐alanine, VFAs, glycerol, mannitol and succinate. Gene knockouts were achieved using P1 phage lysates and CRISPR Cas 9 methods to target key steps in the catabolism of these compounds, with the exception of succinate. For this metabolite, the di‐carboxylic acid transporters YaaH, DctA, YchM, DcuA and DcuB had to be deleted. The effectiveness of the biosensors for selectively measuring d‐ and l‐lactic acid was assessed in Austrian and Irish grass silage leachate. The concentrations measured were comparable to those obtained using a commercial enzyme kit. Introduction Grass fed green biorefinery (GBR) is an emerging green technology that refines fresh grass or grass silage into a number of higher‐value products (Wild and Kłopotowski 1981; Davies et al. 1999; Kromus et al. 2004; Mandl 2010; Schaffenberger 2013; McEniy and O'Kiely 2014; Xiu and Shahbazi 2015; Farm4More 2019; Santamaria‐Fernandez et al. 2019; Savonen et al. 2020; Jørgensen et al. 2022). The GBR process starts with the mechanical fractionation of grass or grass silage into a leachate and press cake. The press cake can be used as a ruminant feed, fed to an anaerobic digestor or treated to produce additional products such as fibres (Mandl 2010; McEniy and O'Kiely 2014; Stødkilde et al. 2023). The leachate is refined by a number of downstream processes specific to fresh or ensiled grass, into higher value products that include crude protein (Farm4More 2019; Damborg et al. 2020; Møller et al. 2021; Jørgensen et al. 2022; BioRefine 2024), amino acids and lactic acid (Schaffenberger 2013; McEniy and O'Kiely 2014; Schwarz et al. 2018). As high quality grass silage leachates are required to produce high quality GBR products, the presence of key analytes such as lactic acid and amino acids is routinely measured by high‐performance liquid chromatography and gas chromatography (Ecker et al. 2012; Schaffenberger 2013; Schwarz et al. 2018). However, these methods are economically unviable for remote small scale GBRs that typically process approximately 20,000–40,000 t of grass silage per year (Höltinger et al. 2012). Enzymatic assays of key metabolites can also be used, but they require a separate kit per analyte, have a short shelf‐life, require time‐consuming and labour‐intensive sample dilution, kit preparation and calibration processes; meaning that they are not real‐time and would require multiple operators working in shifts to provide continuous analysis of multiple GBR catabolites. There is a need for real‐time, cost‐effective and convenient analytical methods to measure fermentation products for small‐scale decentralised GBRs. One potential method is exclusion biosensors such as the Escherichia coli ‐based acetate and propionate exclusion biosensors developed by Sweeney et al. (2022) that can measure the concentrations of acetate and propionate in anaerobic digestate leachates. We now report the further engineering of this previously reported exclusion strain which uses E. coli as its microbial chassis due to its highly annotated genome and its genetic tractability, to create two new stereospecific lactic acid biosensors for the detection of d‐ and l‐lactic acid in grass silage leachates. Materials and Methods Bacterial Strains and Culture Conditions The E. coli strains used in this study were all derived from W3110 and are listed in Table S1. All strains were stored at −80°C in 1 mL aliquots in the appropriate medium containing 20% v/v glycerol [56‐81‐5]. The strain JSK0042 was cultivated in glass test tubes containing 5 mL tryptone soy broth (TSB) with the appropriate antibiotics at 30°C or 37°C and 200 RPM in an orbital shaking incubator. The exclusion strain and its derivatives were cultivated in pyruvate optimal broth (POB), which was based on super optimal broth developed by Hanahan (1983). POB contains double concentrations of NaCl [7647‐14‐5], KCl [7447‐40‐7] and MgCl2 [7786‐30‐3], lacks MgSO4 [7487‐88‐9] and 70 mM pyruvate [113‐24‐6] was added as carbon source. Biosensor strains were cultivated in M9 synthetic medium (Azatian et al. 2019) supplied with the appropriate carbon source. Reference biosensor strains were cultivated using 5 g L−1 d‐xylose [58‐86‐6], which is not expected to be present in grass silage leachates. Strain Construction The plasmids and primers used in this study are listed in Tables S2 and S3. The method described by Sweeney et al. (2015) was used to generate all P1 phage lysates which were used to remove targeted genes by lambda red‐mediated FLP recombinase (Datsenko and Wanner 2000). A consequence of the successive gene deletions that were applied to create the E. coil‐based exclusion strain, was that with each additional gene deletion the number of successful transformants diminished. Eventually, this resulted in strain JSK0095 no longer being transformable with the large (9.3 kb), low copy pCP20 plasmid that was needed to remove its fadK deleted gene's FRT flanked kanamycin cassette. The inability to remove JSK0095's kanamycin cassette meant that subsequent lambda red mediated FLP recombinase gene deletions could no longer be selected for and as such a CRISPR‐Cas9 protocol that was adapted from Reisch and Prather (2015) was applied to create all subsequent gene deletion strains (Table S1). The plasmids pKDsgRNA‐ack (Addgene plasmid # 62654; http://n2t.net/addgene:62654; RRID:Addgene_62654) and pCas9‐CR4 (Addgene plasmid # 62655; http://n2t.net/addgene:62655; RRID:Addgene_62655), were a gift from Kristala Prather (MIT, USA). The protospacer to delete the target gene was found using CRISPOR, http://crispor.tefor.net/crispor.py (Concordet and Haeussler 2018), with E. coli str. K‐12 substr. DH10B as reference genome and 20 bp‐NGG—Sp Cas9 as PAM sequence. The pKDsgRNA backbone was PCR amplified using primers pKDgRNA backbone forward and reverse resulting in a fragment of 4305 base pair (bp). The protospacer was incorporated into the primer, which was used to amplify the guide RNA (gRNA) fragment; the primers also contain a homology arm for Gibson Assembly. The gRNA forward and gRNA reverse primers were used to amplify the gRNA fragment (2663 bp). The backbone and gRNA fragment were assembled into a plasmid using Gibson Assembly. The plasmid pCas9‐CR4 (Table S2) and the plasmid containing the gRNA were transformed into the target strain as described by Reisch and Prather (2015). The construction of the repair fragments deviated from this method because longer repair fragments were needed for the strains used in this study. The fragments were constructed by amplifying 500 bp upstream and downstream of the target gene using primers that introduced overlapping homology arms for Gibson Assembly. The vector pGRD11 (Peist et al. 1997) was linearised by PCR amplification using primers pGDR11 GA lin forward and reverse and all the fragments were assembled into a plasmid. The repair fragment was amplified from this plasmid by PCR. The final step was performed as described by Reisch and Prather (2015) with the only exception that 500 ng mL−1 anhydrotetracycline instead of 100 ng mL−1 was used. Successful gene knockouts were verified by the verification primers listed in Table S3, which amplified short sequences consisting of the junction between the upstream and downstream homology arms that formed when targeted genes had been fully excised. Confirmation of the correct verification sequences for all CRISPR‐Cas9 mediated gene knockouts conducted in this study are presented in Figure S7. For subsequent gene deletions, the same protocol was followed as described by Reisch and Prather (2015). The plasmids used in this study (Table S2) were propagated in and isolated from chemically or electro competent E. coli DH5α, which were cultivated in TSB, supplemented with antibiotics when required. The d‐ and l‐lactic acid biosensor strains were constructed as described by Sweeney et al. (2022) using W3110 as template DNA for the amplification of dld (d‐lactic acid) and lldD (l‐lactic acid). Lactic Acid Analysis d‐ and l‐lactic acid concentrations in the standards and biological samples were measured using an enzymatic analysis kit according to manufacturer protocol (K‐DLATE, Megazyme, Bray, Ireland). Biosensor Setup and Operation The biosensor setup and operation as described in detail by and Sweeney et al. (2022) was adapted and improved for this study. E. coli biosensor strains were grown in M9 medium supplemented with the strain‐specific carbon source until an optical density of ~1.6 (600 nm) was achieved. The cells were harvested, washed and concentrated to 120 mg wet cells mL−1 in phosphate‐buffered saline (PBS). A portion (30 μL) of the cell suspension was immobilised under vacuum onto a cellulose acetate membrane with a pore size of 0.45 μm, which was fixed to the tip of the dissolved oxygen (DO) probe (Vernier, Beaverton, OR, USA). An Arduino Mega 2560 Rev. 3 controlled automated system was employed to automate the addition and replacement of PBS buffer, perform wash steps, agitate the working volume at 450 RPM using a stirrer plate and magnetic flea and automatically inject samples using a nine port Tecan Carvo XL 728329‐D syringe pump. A 3D printed nylon beaker into which the DO probes were immersed was designed to facilitate a biosensor array containing multiple biosensors immersed within the same 20 mL working volume. This meant that one biological sample could be provided to two or more biosensors, immersed together in the same working volume. Raw voltage analogue outputs produced by each biosensor's DO probe within the biosensor array, were logged for 120 s and saved in a single .TXT file at 10–20 min intervals for 20–36 h experimental durations. The data from all sample measurements conducted within the experimental timeframe were extracted into one Excel workbook and processed via a set of algorithms (Supporting Information 3) that first normalised the data and plotted the individual voltage readings over time (Figure S1). A second algorithm calculated the oxygen consumption rate (ocr) from the slope of the linear range of the respective voltage response curve (displayed by trendline in Figure S1). The concentration of d‐ or l‐lactic acid was determined by measuring the oxygen consumption rate in mV (ocr biosensor ) and correcting for background interference and dilution (Supporting Information 2 for comprehensive explanation). Independent sample t‐tests were performed on the average of enzymatic kit and biosensor response calculated concentration of d‐ and l‐lactic acid using IMB SPSS Statistics 29.0.1.0 (IBM, Armonk, New York, USA). There was homogeneity of variances as assessed by Levene's test for all samples except 500 μL l‐lactic acid Austrian leachate biosensor estimate compared to the enzymatic kit analysis (p = 0.06) where a Welch's t‐test was used. All data were normally distributed, as assessed by Shapiro–Wilk's test (p > 0.05). Boxplot analysis showed that there was one outlier in the data of the l‐lactic acid concentration of the Irish leachate when determined with the biosensor using 1000 μL of sample. This outlier was left in the data set when running the statistical test because it was deemed to not influence the outcome of the test. Biological Leachate Austrian grass silage concentrate was produced by Biorefinery Research, Implementation & Development Ltd. (Galway, Ireland) and tbw research GmbH (Austria) as part of the LIFE Farm4More project (LIFE18 CCM/IE/001195). Leachate concentrate was produced using a Bellmer‐Kufferath AKKUPRESS AX—single screw press to dewater a bio‐certified grass clover silage. Suspended solids were removed from the press leachate by sedimentation coarse filtering using a 50 μm textile bag filter. A final concentration of 62% dry matter (DM) was achieved by applying a vacuum evaporation process to produce a storable concentrated product. The concentrated leachate was adjusted to 10% (v/v) (typical grass silage leachate concentration) by the addition of water. The Irish grass silage leachate was provided by John Brennan (UCD) who watered by hand mixing a perennial ryegrass white clover ley with a respective compositional ratio of 60:40 that had been harvested into round bales at UCD Lyons Agricultural Farm to achieve a DM content of 22% and then pressed at 10 mPa for 2 min using a bench top hydraulic press (12‐ton shop floor press, SIP Industrial, UK) that had been modified for grass silage fractionation. Synthetic Leachate (SL) A SL was designed based on the chemical composition of grass silage leachate as described by (Ecker et al. 2012; Schaffenberger 2013; Cerrone et al. 2015; Müller et al. 2016) and contained all the carbohydrates corresponding to the metabolic pathways that had been knocked out of the exclusion strain. The SL, which was adjusted to pH 7, comprised (g L−1) l‐lactic acid [79–33‐4] (10), d‐lactic acid [10326–41‐7] (10), acetate [127‐09‐3] (5), formate [141‐53‐7] (2.5), propionate [137‐40‐6] (5), butyrate [156‐54‐7] (1), valeric acid [109‐52‐4] (1), hexanoic acid [142‐62‐1] (1), ethanol [64‐17‐5] (5), d‐glucose [50‐99‐7] (5), d‐fructose [57‐48‐7] (5), d‐mannitol [69‐65‐8] (2.5) and glycerol [56‐81‐5] (5) and amino acid mixture (20). The concentrations and CAS numbers of the amino acid mixture's individual amino acids are provided in Table S5. A second synthetic leachate (SL2) was developed for d‐ and l‐lactic acid quantification experiments. SL2 which was adjusted to pH 7, contained (g L−1) l‐lactic acid [79‐33‐4] (2.5), d‐lactic acid [10326‐41‐7] (2.5), acetate [127‐09‐3] (2.5), formate [141‐53‐7] (1.25), propionate [137‐40‐6] (1.25), butyrate [156‐54‐7] (0.5), valeric acid [109‐52‐4] (0.5), hexanoic acid [142‐62‐1] (0.5), ethanol [64‐17‐5] (2.5), d‐glucose [50‐99‐7] (2.5), d‐fructose [57‐48‐7] (2.5), d‐mannitol [69‐65‐8] (2.5), glycerol [56‐81‐5] (2.5) and l‐alanine [56‐41‐7] (2.5). Results and Discussion Construction of Exclusion Strain The use of whole cell E. coli biosensors for the measurement of specific metabolites in grass silage leachate samples is hampered by the bacterium's ability to catabolise a range of compounds present in grass silage leachate. To create a metabolite‐specific biosensor that does not experience interference from compounds present in grass silage, key catabolic steps within E. coli need to be removed. The development of the exclusion strains started with strain JSK0042 (Table S1), which already lacks genes for d‐lactic, l‐lactic, acetic, propionic and formic acids and ethanol catabolism. Additional knockout experiments were conducted in this strain to generate an exclusion strain that does not experience interference from compounds present in grass silage leachates. The catabolism of glucose, mannose, fructose, succinate, l‐alanine, volatile fatty acids, glycerol and mannitol needed to be removed, which was based on the known compounds in grass silage leachate (Ecker et al. 2012; Cerrone et al. 2015; Müller et al. 2016). Genes coding for enzymes directly involved in these compounds' catabolism were targeted as described in Table S4, using the iterative engineering approach developed by Sweeney et al. (2022). After each round of mutagenesis, the resulting ‘exclusion’ was confirmed by measuring the oxygen consumption rate (ocr biosensor) using SL and the compound for which the metabolic pathway was deleted. It was inferred that a low or absent stable oxygen consumption rate, here after called biosensor response, confirmed that the exclusion was successful (Figure 1). Development of the exclusion biosensor via gene knockout. The table summarises the biosensor response on the targeted metabolites before and after gene deletion. *Mannitol responses resolved towards 0 after 4–8 h or biosensor operation. For most metabolites, unwanted biosensor responses for specific catabolites were successfully removed by deleting the one or more catabolic conversion or transport steps described in Table S4 except for mannitol, which required 4–8 h of biosensor operation for its knockout strains' biosensor responses to stabilise towards zero. However, it was found that the interference from succinate (4.5 mV s−1) was more challenging to eliminate. The obvious target for the exclusion of this compound was sdhC, which encodes for one of the succinate dehydrogenase subunits responsible for converting succinate to fumarate. However, this enzyme is integral to the tricarboxylic acid cycle, which enables cells to grow on non‐fermentable carbon sources such as acetic acid. The removal of this gene would limit the development of non‐fermentative primary catabolite biosensors including acetic acid. For this reason, an alternative approach to remove succinate catabolism in E. coli while preserving sdhC was sought, which began with attempting to remove its transport into the cell. This strategy was successfully applied to eliminate interference from glucose, mannose and fructose in the strain designated JSK0002 (Figure 1). This method was further applied to create strain JSK0071 by deleting E. coli 's succinate transporters yaaH, ychM and dctA, which are expressed under aerobic growth conditions (Table S4). Unfortunately, this strain still exhibited a high succinate biosensor response (2.5 mV s−1) when grown on 5 g L−1 pyruvate as a sole carbon source (Figure S2). E. coli also possesses two C4 di‐carboxylate transporters DcuB and DcuA whose primarily function as anaerobic fumarate:succinate and l‐aspartate:fumarate antiporters, respectively. These transporters are constitutively expressed under both aerobic and anaerobic conditions (Table S4), hence they were targeted for removal. The gene dcuB removed first to produce the exclusion strain JSKX00125, which did not exhibit a significant reduction in succinate biosensor response (2 mV s−1) compared to JSK0071 (Figure 1, Figure S3A). However, the subsequent removal of dcuA to produce strain JSK0115 resulted in a substantially reduced biosensor response (0.5 mV s−1, Figure 1, Figure S3B). Although, the succinate biosensor response was not completely abolished and likely due to an unknown transport mechanism, this response was judged to be sufficiently low to not cause biosensor interference. Therefore, the strain JSK0115 was selected as the exclusion biosensor strain and was used to develop a d‐ and l‐lactic acid biosensors for grass silage leachate. Constructing d‐ and l‐Lactic Acid Biosensor Strains JSP0090 and JSP0094 The exclusion strain JSK0115 was transformed with two high copy plasmids pGDR11_dld and pGDR11_lldD, which were created by ligating the respective genes into the vector. This resulted in the d‐ and l‐lactic acid biosensor strains designated JSP0090 and JSP0094, respectively. Confirmation of successful expression was via growth experiments on M9 medium supplemented with the appropriate lactic acid stereoisomer as the sole carbon source. To test the biosensor performances, each strain was provided with three different volumes of d‐ or l‐lactic acid standard (2.5 g L−1), SL2 containing both lactic acid enantiomers and SL2 with succinate. The biosensor responses of JSP0090 and JSP0094 when exposed to SL were comparable to those of 500 μL d‐ and l‐lactic acid (Figure S4A,B), which contained similar concentrations of d‐ or l‐lactic acid. It was very encouraging that there were no, or minimal, biosensor responses to synthetic leachate supplemented with succinate or their opposite lactic acid isomers. Although both biosensors exhibited strong responses for their highest d‐ and l‐lactic acid standards (2.5 g L−1), to ensure biosensor responses were in their linear ranges a d‐ or l‐lactic acid standard concentration of 1.25 g L−1 which equated to biosensor responses up to 6 mV s−1 was applied (Figure S4A,B). As d‐ and l‐lactic acid concentrations within grass silage leachates typically far exceed this value, samples were simply diluted to be under the biosensor's upper detection limit, which is a standard practice for enzymatic kit analysis. Furthermore, the data (Figure S4A,B) showed that the biosensors could consistently measure lactic acid over 36 h, which is important if the biosensors are to be practically employed to measure analytes in leachates. The Biosensors Accurately Measure Lactic Acid in Grass Silage D‐ and l‐lactic acid biosensor responses (ocrbiosensor) and JSK0115 reference biosensor responses (ocrreference biosensor) for the lactic acid standard and different volumes of two grass silage leachate samples from Ireland and Austria, were collected over a 36 h experimental period but only became stable from 16 h onwards (Figures S5 and S6). For this reason it was decided that lactic acid standards and Austrian and Irish leachate sample data points recorded from the JSK0115 ocrreference biosensor between 20 and 36 h (Figure 2) would be used to correct both JSP0090 (Figure 3A) and JSP0094 (Figure 3B) ocrbiosensor values for the same period. d‐ and l‐lactic acid biosensor response corrections ocrcorrected were required to ensure that background biosensor interference that was captured by the exclusion biosensor's responses ocrreference biosensor (Figure 2) would not result in an over‐ or under‐estimation of d‐ and l‐lactic acid when calculated per biosensor cycle using the equations detailed in Supporting information 2. Each biosensor cycle consisted of a 120‐min time‐period in which all standard volumes (1000, 500 and 200 μL) and 2 leachate sample volumes (1000 and 500 μL) were supplied once to all biosensors. For each cycle, the ocrcorrected biosensor responses for the three standard volumes were used with the enzymatic kit determined d‐ and l‐lactic acid standard concentrations, to create a cycle standard curve (Table S6). The slope and intercept of this standard curve with the respective biological sample dilution factors were used to calculate the d‐ and l‐lactic acid concentration of the biological samples tested in this cycle (Figure S8). The mean and standard deviation of the d‐ and l‐lactic acid estimates for each of the 8 cycles completed over the 20–36‐h time‐period are presented in Table 1. The oxygen consumption rate (ocrreference biosensor) of the JSK0115 reference biosensor, grown on 5 g L−1 d‐xylose, when supplied with 1000 μL 5 g L−1 d‐xylose (purple, dash), 1000, 500 and 200 μL d‐ and l‐lactic acid standard (blue, solid), 1000 and 500 μL Irish leachate (green, dot‐dash) and 1000 and 500 μL Austrian leachate (orange, long dash). The oxygen consumption rate (ocrbiosensor) of (A) the d‐lactic acid biosensor (JSP0090) and (B) l‐lactic acid biosensor (JSP0094), when supplied with 1000, 500 and 200 μL d‐ and l‐lactic acid standard (blue, solid), 1000 and 500 μL Irish leachate (green dot‐dash) and 1000 and 500 μL Austrian leachate (orange, dash). TABLE 1Concentrations of d‐ and l‐lactic acid (g L−1) grass silage samples as measured with an enzymatic kit and individual 1000 and 500 μL corrected biosensor estimates for eight individual cycles.Kit (g L−1)1000 μL Biosensor a (g L−1)500 μL Biosensor a (g L−1) Source of d‐lactic acid Austria6.245 ± 0.1246.024 ± 0.1366.636 ± 0.087Ireland9.125 ± 0.0858.552 ± 0.179.491 ± 0.28Standard1.094 Source of l‐lactic acid Austria6.589 ± 0.046.863 ± 0.1447.416 ± 0.265Ireland10.944 ± 0.5310.704 ± 0.34310.84 ± 0.515Standard1.275 Note: Kit values are the average of three samples; biosensor samples are an average of eight individual 1000 or 500 μL volumes. a Prior to biosensor measurement the leachates were diluted appropriately to bring them within the biosensor detection range. Concentrations of d‐ and l‐lactic acid (g L−1) grass silage samples as measured with an enzymatic kit and individual 1000 and 500 μL corrected biosensor estimates for eight individual cycles. Note: Kit values are the average of three samples; biosensor samples are an average of eight individual 1000 or 500 μL volumes. Prior to biosensor measurement the leachates were diluted appropriately to bring them within the biosensor detection range. An independent t‐test was used to determine if the average concentrations of the Austrian and Irish leachate, determined with the biosensor and enzymatic kit, for the 8 cycles completed within the 20–36‐h experimental time‐period, were statistically the same. There was no statistical difference when comparing the different analytical methods used to determine the l‐lactic acid concentration in the Irish leachate (Figure 4A). However, there was a statistical difference when comparing the different analytical methods used to determine the l‐lactic acid concentration in the Austrian leachate (Figure 4B). Most d‐lactic acid biosensor results for 500 and 1000 μL Austrian and Irish leachates were statistically different from the corresponding enzymatic kit results (Figure 4C,D). (A) Average for eight individual cycles of: l‐lactic acid concentration in Irish leachate, (B) average l‐lactic acid concentration in Austrian leachate, (C) average d‐lactic acid concentration in Irish leachate and (D) average d‐lactic acid concentration in Austrian leachate. Concentrations were determined with 1000 μL biosensor sample Irish leachate (blue), 500 μL biosensor sample Irish leachate (orange), enzymatic kit Irish leachate (red), 1000 μL biosensor sample Austrian leachate (grey), 500 μL biosensor sample Austrian leachate (yellow) and enzymatic kit Austrian leachate (green). Mean and standard deviation are calculated from eight sample with biosensor analysis and three samples from enzymatic kit analysis. p values were determined with a t‐test. ns, not significant. It is evident from the results presented in Table 1, Figure 4 and Figure S8 that the d‐ and l‐lactic acid biosensors JSP0090 and JSP0094 provide a good indication of d‐ and l‐lactic acid when compared to enzymatic kit analysis, however, to make it comparable to chemical analysis, the biosensor system requires further improvements. For example, in this study d‐ and l‐lactic concentrations were calculated per cycle for a single 1000 and 500 μL Irish and Austrian leachate volume using a standard curve slope and intercept that was calculated using three singular standard volumes (1000, 500 and 200 μL) of known d‐ and l‐lactic acid concentration. Further analysis is required to identify d‐ and l‐lactic acid biosensors' linear ranges and to optimise the number of standard concentrations and/or replicates used to create analyte specific standard curves. Improved standard curve slopes to which biosensor's ocrcorrected values would be applied, would result in higher accuracy estimates of analyte concentrations within unknown samples (consample). Conclusion In this study a bacterial exclusion strain was successfully constructed that was unable, or had severely reduced ability, to catabolise d‐lactic acid, l‐lactic acid, acetate, propionate, formate, ethanol, glucose, fructose, l‐alanine, VFAs, glycerol and mannitol. The removal of key aerobic and anaerobic succinate transporter genes substantially reduced the biosensor's response for succinate with the remaining response judged to be small enough to have little or no effect. The exclusion strain designated JSK0115 was used to develop the stereospecific lactic acid biosensor strains JSP0090 (d‐lactic acid) and JSP0094 (l‐lactic acid). The d‐ and l‐lactic acid biosensors were able to indicate the presence of d‐ and l‐lactic acid in grass silage leachate. However, the biosensor's accuracy needs to be improved further to be comparable to analytical methods. This study's exclusion biosensor development approach could further support grass silage fed GBRs if biosensors for additional key performance indicator analytes such as acetate and propionate are created. These biosensors if implemented as an array with this study's d‐ and l‐lactic acid biosensors could provide GBR operators with continuous real‐time analysis of key grass silage quality indicators that would in turn support the biorefining of only high‐quality grass silage leachates into high value products. Author Contributions Joseph B. Sweeney: conceptualization, funding acquisition, supervision, resources, writing – review and editing, visualization, project administration, investigation, writing – original draft. Roderick N. van Roosmalen: conceptualization, investigation, validation, methodology, software, data curation, writing – original draft, writing – review and editing, visualization. Cormac D. Murphy: supervision, resources, writing – review and editing. Funding This work was supported by European Union's Horizon 2020 Research and Innovation Programme AgRefine, 860477; European Climate, Environment and Infrastructure Executive Agency (CINEA) LIFE Programme and Government of Ireland Department of Climate, Energy and the Environment (DCEE), LIFE18CCM/IE/001195. Conflicts of Interest All authors declare the filing of patent application number PCT/EP2023/079126 that includes part of the results reported in this paper. Supporting information Table S1: Escherichia coli strains used in this study. Table S2: Plasmids used in this study. Table S3: A list of primers used in this study. Table S4: Targeted genes that have been deleted to remove E. coli's' ability to catabolise d‐glucose, d‐fructose, l‐alanine, C6–C18 fatty acids, glycerol, d‐mannitol and succinate. Table S5: 20 g L−1 total amino acid composition and CAS number. Table S6: Calculating cycle 1's d‐ and l‐lactic acid standard curve slope (x) and intercept (y) using the 1000, 500 and 200 μL d‐ and l‐lactic acid standard corrected biosensor responses (ocrcorrected) and the corresponding enzymatic kit determined standard concentrations (conin beaker). Figure S1: Example of the mV voltage response curve which is plotted from the 120 s samples collected within the sample period. The slope of the linear range of this voltage response curve is the oxygen consumption rate (ocr). Figure S2: The oxygen consumption rate (ocrbiosensor) of JSK0071, grown on 5 g L−1 pyruvate, when supplied with 1000 μL of 5 g L−1 succinate (orange). The control analyte was 5 g L pyruvate of which 1000 μL was supplied (blue). The replicates from the same experiment are displayed. Figure S3: The oxygen consumption rate (ocrbiosensor) of biosensor strains JSKX0125 (A) and JSK0115 (B), grown on 5 g L−1 d‐xylose, when supplied with 1000 μL of 5 g L−1 succinate (orange, dash) and 1000 μL of synthetic leachate (green, dot‐dash). The control analyte was 5 g L−1 d‐xylose of which 1000 μL was supplied (blue, solid). The replicates from the same experiment are displayed. Figure S4: The corrected oxygen consumption rate (ocrbiosensor) for JSP0090 (A) and JSP0094 (B), grown on 5 g L−1 d‐lactic acid, when supplied with 1000, 500 and 200 μL of 2.5 g L−1 d‐lactic acid (dark blue, solid), 500 μL of synthetic leachate 2 (grey, dot‐dash), 500 μL of synthetic leachate 2 with 2.5 g L−1 succinate (orange, dot‐dot‐dash) and 1000 μL of 2.5 g L−1 l‐lactic acid (green, dash). Figure S5: The full 40 h oxygen consumption rate (ocrreference biosensor) of the 5 g L−1 d‐xylose grown JSK0115 reference biosensor which when subtracted from d‐ and l‐lactic acid biosensors' responses (ocrbiosensor) per cycle from the 20 to 36 h stable experimental period, produced the respective biosensors' corrected responses (ocrcorrected). 1000 μL 5 g L d‐xylose (purple, dash), 1000, 500 and 200 μL d‐ and l‐lactic acid standard (blue, solid), 1000 and 500 μL Irish leachate (green, dot‐dash) and 1000 and 500 μL Austrian leachate (orange, long‐dash). Figure S6: Biosensor performance as measured with oxygen consumption rate (ocrbiosensor) values for 40 h including the 20–36 h stable experimental period of the d‐lactic acid biosensor (JSP0090) and l‐lactic acid biosensor (JSP0094) when supplied with 1000, 500 and 200 μL d‐ and l‐lactic acid standard (blue, solid), 1000 and 500 μL Irish leachate (green dot‐dash) and 1000 and 500 μL Austrian leachate (orange, long‐dash). Figure S7: Agarose gel PCR conformation of the expected verification sequences for CRISPR‐Cas9 mediated gene knockouts using forward and reverse gene knockout verification primers listed in Table S3. Figure S8: d‐lactic acid concentrations and l‐lactic acid concentrations per sample cycle calculated using JSP0090 d‐ and JSP0094 l‐lactic acid biosensor data compared to enzymatic kit determined concentrations. 1000 μL Irish leachate (blue), 500 μL Irish leachate (orange), d‐ or l‐lactic acid concentration of Irish leachate determined with enzymatic kit (green line), 1000 μL Austrian leachate (pink), 500 μL Austrian leachate (grey) and d‐ and l‐lactic acid concentration of Austrian leachate determined with enzymatic kit (red line).