Sentinel plants enable quantitative monitoring of bioavailable nitrate in soils and microbial environments Authors: Eugene Li, Chiara Berruto, Tufan M. Oz, Elisa Grillo, Catherine Griffin, Yunqing Wang, Kimberley T. Muchenje, Jolie W. Jones, Gozde S. Demirer Published: 2026 ABSTRACT SignificanceNitrogen availability in soils fluctuates across space and time, yet most measurements rely on laboratory analysis of extracted soil samples. Such measurements provide only snapshots of nitrogen status and do not necessarily reflect the nitrogen that plants experience. Here, we engineer plants that act as living nitrate sensors by converting nitrate perception into a measurable optical signal. Because these sensors operate within intact plants, they report nitrate availability as integrated through plant uptake and physiology rather than through chemical extraction alone. Using this platform, we tracked nitrate levels in agricultural soils and characterized a minimal microbial consortium capable of converting atmospheric nitrogen into plant-available nitrate. This plant-based sensing strategy enables direct monitoring of nitrogen dynamics in soils and microbial environments. Nitrogen availability strongly limits plant productivity and global agricultural output, with nitrate representing the dominant bioavailable nitrogen source in many aerobic soils. Optimizing nitrate availability is therefore central to both crop yield and environmental stewardship. Excess nitrogen fertilization promotes nitrate leaching, eutrophication, and nitrous oxide emissions, whereas insufficient nitrogen reduces crop productivity (1). These competing pressures necessitate precise sensing strategies that quantify bioavailable nitrate with sufficient spatial and temporal resolution to guide nitrogen management decisions. Most soil nitrate measurements rely on extraction-based laboratory assays, such as cadmium reduction colorimetry (2) or ion chromatography (3). These approaches require physical sampling and provide only discrete snapshots of nitrate availability, which varies strongly across space and time (4–6). Field-deployable alternatives can increase sampling density but have their own limitations. Nitrate ion-selective electrodes (7–11) are susceptible to signal drift and interference from competing ions in heterogeneous soils, whereas multispectral and hyperspectral imaging from unmanned aerial vehicles (12–14) infers plant nitrogen status indirectly from canopy reflectance. Collectively, existing methods face trade-offs among analytical accuracy, spatial and temporal resolution, deployment complexity, and biological relevance. Biosensing offers a complementary paradigm in which living systems transduce nitrate exposure into measurable outputs (15). Microbial biosensors have been engineered to use nitrate-responsive promoters to drive optical reporters, enabling detection of nitrate gradients in soil (16). However, as microbes and plant roots differ in their local environments and nitrate uptake dynamics, microbial biosensors reflect nitrate availability to microbes rather than to plants. Deploying engineered microbial reporters also requires maintaining living strains within complex soil communities, which can complicate reproducibility, interpretation, and regulation. Plants integrate soil nitrate through root uptake and internal transport; hence, signals measured in plant tissues provide a plant-relevant readout of soil nitrate dynamics. The transcription factor NLP7 functions as a primary intracellular nitrate sensor that, upon nitrate binding, derepresses nitrate-responsive transcription (17, 18). Liu et al. developed a synthetic nitrate-responsive luciferase reporter containing tandem NLP7-binding sites upstream of a minimal promoter to study nitrate-dependent NLP7 activation in transient assays, and further developed a split mCitrine-NLP7 biosensor to visualize intracellular nitrate dynamics at the single-cell level in transgenic plants (17). Other genetically encoded sensors have also enabled fluorescent readouts of nitrate dynamics in plant systems, including NiTrac (19), a CHL1/NRT1.1-derived reporter of nitrate transport activity, and NitraMeter3.0 (20), a NasR-derived biosensor for imaging cellular nitrate concentration and distribution. These sensors enabled the study of nitrate perception and intracellular dynamics at cellular resolution, and their use has focused primarily on mechanistic interrogation of plant nitrate signaling and transport. Whether nitrate-sensing machinery can be translated into a quantitative, plant-integrated sensing platform suitable for soil and plant–microbe environments and for probing microbial contributions to plant nutrient acquisition remained largely unexplored. Here, we engineered sentinel plants to meet this need. By optimizing promoter and reporter architecture, we built a ratiometric nitrate sensor with high dynamic range, reversibility, and specificity for nitrate, including insensitivity to ammonium, and generated stable transgenic Arabidopsis thaliana lines. We showed that sentinel plants report plant-accessible nitrate across agricultural soils from multiple California field sites and in the crop species, tomato. We further used this platform to characterize a three-member microbial consortium that converts atmospheric nitrogen into plant-available nitrate, thereby increasing tissue nitrate accumulation and plant biomass. Together, these results extend plant nitrate reporters beyond cellular signaling studies toward quantitative monitoring of plant–microbe–soil nitrogen processes. Results Engineering and Optimization of a Genetically Encoded Nitrate Sensor in Arabidopsis. A ratiometric design was used to normalize the nitrate-responsive output to a constitutive reference measured in the same sample. This normalization corrects for variation in transfection and expression efficiency, and tissue input, enabling quantitative comparison of nitrate responses across samples and over time. However, transcriptional interference between adjacent cassettes can compromise quantitative reporter measurements when reporters are encoded on the same plasmid (21, 22). For this, we first optimized the cassette orientation. We used a characterized nitrate-responsive promoter, NRP (23), to drive an enhanced luciferase (ELuc) reporter, together with a constitutive pNOS-driven RedF luciferase normalization cassette (24). The two transcriptional units were assembled in four relative orientations (Fig. 1A): head-to-head (H2H), head-to-tail (H2T), tail-to-head (T2H), and tail-to-tail (T2T), and tested in A. thaliana mesophyll protoplasts exposed to nitrate starvation (0 mM), limitation (1 mM), or sufficiency (10 mM). Cotransfection (CoT) of separate ELuc and RedF plasmids was used as a control to represent minimal transcriptional interference (Fig. 1B). For each orientation, we quantified the basal ELuc/RedF ratio, the induction at 1 and 10 mM, and the agreement with the CoT dose response by ordinary least squares regression against the CoT means (slope, intercept, and R2, Dataset S1). H2H orientation resulted in significantly elevated basal output and low nitrate response. T2H showed a compressed dynamic range, and H2T showed consistently lower output. Only T2T matched CoT across all metrics (Dataset S1). In line with reports that convergent (tail-to-tail) arrangements reduce promoter occlusion and read-through between adjacent cassettes (21, 22), we selected T2T for all subsequent constructs. Optimization of the nitrate sensor in A. thaliana. (A) Schematics of dual luciferase reporter configurations. Head-to-head (H2H), head-to-tail (H2T), tail-to-head (T2H), tail-to-tail (T2T). (B) ELuc/RedF signal ratios for different transcriptional-unit orientations in A. thaliana mesophyll protoplasts at 0, 1, and 10 mM KNO3. Cotransfection (CoT) of separate ELuc and RedF plasmids was used as a reference condition. Data are shown as mean ± SEM; n = 4 technical replicates with 5 × 104 protoplasts per replicate. (C) Luminescence comparison of ELuc/GeNL and RedF/NLuc reporters, expressed from the AtNRP promoter with an Hsp18 terminator (ELuc, GeNL) or the pNOS promoter with a Tnos terminator (RedF, NLuc); within each pair, the constructs are identical except for the reporter open reading frame. Data are shown as mean ± SEM; n = 8 technical replicates with 5 × 104 protoplasts per replicate. Statistical significance was determined using two-tailed unpaired t tests for each pairwise comparison. (D) Schematic of the final NLP7-based nitrate sensor architecture. (E) Minimal promoter comparison using min35S and minDFR in seedlings following transient transformation. Lighter bars indicate 0 mM nitrate, and darker bars indicate 10 mM nitrate. Data are shown as mean ± SEM; n = 4 biological replicates with five pooled seedlings per replicate. Statistical significance was determined by one-way ANOVA with Tukey’s multiple comparisons test. (F) Effect of NLP7 binding site number on sensor output in transiently transformed seedlings under 0 and 10 mM nitrate. Data are shown as mean ± SEM; n = 4 biological replicates with five pooled seedlings per replicate. In all panels, ****P < 0.0001 and outliers were removed using Grubbs’ test, if applicable. Although the NRP promoter was suitable for optimizing cassette orientation, it was not used in the final sensor because it responds to multiple nitrogen sources and lacks specificity for nitrate (23). We therefore developed a biosensor based on the nitrate-responsive transcription factor NLP7 and its cognate binding sites (25). The ELuc/RedF reporter pair used for architectural optimization was replaced with a brighter GeNL/NLuc luciferase pair that provides higher signal output. GeNL generated approximately 10-fold higher signal intensity than ELuc, and NLuc produced approximately 50-fold higher signal intensity than RedF (Fig. 1C). The resulting sensor consisted of tandem NLP7 binding sites positioned upstream of a minimal promoter driving an inducible GeNL reporter, together with a constitutively expressed NLuc normalization cassette arranged in the optimized tail-to-tail configuration (Fig. 1D). To evaluate sensor performance under more physiologically relevant conditions, we transitioned from protoplasts to A. thaliana seedlings grown on MS plates following transient transformation of the sensor circuit (26). Nitrate responses were quantified as the GeNL/NLuc ratio. We first optimized the minimal promoter downstream of the NLP7 binding sites. Replacing the minimal cauliflower mosaic virus 35S promoter (min35S) with a minimal tomato dihydroflavonol 4 reductase promoter (27) (minDFR) increased sensor dynamic range from 15-fold to 53-fold between 0 and 10 mM nitrate, reflecting reduced basal activity under nitrate starvation while preserving strong induction (Fig. 1E). The minDFR promoter was therefore used for further optimization. We next optimized the number of tandem NLP7 binding sites in the synthetic promoter by testing constructs with 2, 4, 6, or 8 binding sites. Sensors containing 2 binding sites did not respond to nitrate, whereas sensors containing 4, 6, or 8 binding sites exhibited 53-fold, 22-fold, and fivefold induction, respectively, between 0 and 10 mM nitrate (Fig. 1F). Although increasing the number of binding sites enabled nitrate-dependent activation, higher copy numbers also increased basal signal under starvation conditions, resulting in reduced fold change. We therefore selected a final tail-to-tail sensor architecture comprising four tandem NLP7 binding sites upstream of the minDFR promoter and a GeNL/NLuc ratiometric readout. To benchmark this design against existing NLP7-based reporters, we compared the optimized sensor with the 4×NRE-min-LUC reporter from Liu et al. (17) in Arabidopsis protoplasts under matched nitrate conditions. Our optimized sensor showed ~2.5-fold higher induction at both 0.25 mM (5.4-fold versus 2.1-fold) and 5 mM nitrate (8.8-fold versus 3.5-fold), indicating an improved dynamic range (SI Appendix, Fig. S1). Characterization of Sensor Behavior in Plants via Transient Transformation. We characterized the performance of the optimized circuit in A. thaliana seedlings following transient transformation. Sensor responsiveness was quantified across nitrate concentrations of 0 to 10 mM, where the sensor exhibited dose-dependent activation that was well described by a saturating Michaelis–Menten relationship (Vmax = 4.20, Km = 0.57 mM, R2 = 0.81; Fig. 2A and Dataset S2). The response was near-linear between 0 and 1 mM and approached saturation above 3 mM nitrate. This is consistent with many nitrate-dependent plant transcriptional and physiological responses (28–30), and indicates that the sensor quantitatively reports nitrate availability across physiologically relevant concentration ranges. Characterization of sensor behavior via transient expression in Arabidopsis. (A) Nitrate dose–response curve, plotted as GeNL/NLuc and fitted to the Michaelis–Menten equation. (B) Sensor responses at 10 mM KCl, NH4Cl, urea, glutamine (Gln), KNO3, and NH4NO3. (C) Sensor responses to phosphate at 1.25 mM, alone or together with 10 mM KNO3. (D) Sensor output at 0 and 10 mM nitrate at pH 5.5, 6.5, and 7.5. (E) Dynamic sensor response during seedling transfer from 10 to 0 mM nitrate plates. In all panels, data are shown as mean ± SEM; n = 4 biological replicates with five pooled seedlings per replicate. Statistical significance in panels B and C was assessed using one-way ANOVA with Tukey’s multiple comparisons test, and in panel D using Šídák’s multiple comparisons test; ****P < 0.0001. To determine sensor specificity, we tested alternative nitrogen sources commonly present in soil. Potassium nitrate robustly induced the sensor, whereas ammonium, urea, and glutamine did not elicit activation, similar to the potassium chloride negative control (Fig. 2B). Ammonium nitrate activated the sensor to a similar extent as potassium nitrate, indicating that ammonium did not inhibit nitrate-dependent reporter output, which was observed in prior sensors with the NRP promoter (23). To assess interference from prevalent soil nutrients, we also evaluated the effect of phosphate. Supplementation with high phosphate (1.25 mM) neither triggered nor attenuated nitrate-induced output (Fig. 2C), indicating selectivity for nitrate over this major macronutrient. Sensor robustness was further evaluated across a range of pH values, which often fluctuate under field conditions. From pH 5.5 to 7.5, the sensor maintained strong induction at high nitrate, while preserving low basal output under nitrate-free conditions (Fig. 2D), demonstrating the sensor’s potential to detect nitrate across environmentally relevant pH values. Last, to assess whether the sensor responds dynamically and reversibly to changes in nitrate availability, seedlings were first induced with 10 mM nitrate and then transferred to nitrate-free media. Following nitrate removal, sensor output declined toward baseline over 4 d (Fig. 2E), indicating that the sensor reports current nitrate availability rather than retaining a long-term memory of prior exposure. To evaluate whether this sensing architecture functions beyond Arabidopsis, we tested the sensor in the tomato crop (Solanum lycopersicum cv. Micro-Tom). Following transient transformation (Methods), tomato seedlings exhibited clear, dose-dependent increases in GeNL/NLuc signal across the 0, 1, and 10 mM nitrate treatments, with a 4.5-fold increase from 0 to 10 mM (SI Appendix, Fig. S2). Characterization of Sensor Behavior in A. thaliana Stable Sentinel Plants. To enable long-term monitoring of nitrate dynamics, we generated transgenic A. thaliana lines expressing the optimized sensor. Sensor output in this section was measured from all leaves pooled from five seedlings, harvested at a common developmental stage when the first four true leaves were fully expanded (after 3 d of nitrogen starvation followed by 1 d of nitrate treatment). Three independent homozygous lines were analyzed (Dataset S3). To establish the operational range of the stably integrated sensor, we quantified dose–response relationships for each line across 0 to 10 mM KNO3. All lines displayed dose-dependent activation consistent with saturating Michaelis–Menten kinetics, with best-fit parameters of Vmax = 7.27 and Km = 0.12 mM (Line #1), Vmax = 9.29 and Km = 0.15 mM (Line #2), and Vmax = 6.49 and Km = 0.08 mM (Line #3) (Fig. 3A and Dataset S2). Within the low-nitrate range (0 to 0.5 mM), sensor output was well described by linear fits (Fig. 3B and Dataset S2), providing high sensitivity for detecting changes in this range. Notably, all stable lines exhibited higher Vmax and lower apparent Km values than observed in transient assays, indicating stronger maximal reporter output and increased sensitivity to nitrate following genomic integration. Performance of Arabidopsis sentinel plants as nitrate sensors. (A) Nitrate dose–response curves of three independent T2 stable lines grown across a KNO3 gradient (0 to 10 mM); sensor output fits the Michaelis–Menten kinetics. (B) Low-nitrate region (0 to 0.5 mM) for the same three lines; lines are linear fits (parameters in Dataset S2). (C) Heatmap showing mean GeNL/NLuc response for each line under varying nitrogen sources and phosphate conditions. (D) Sensor output in Line #2 under nitrate-free (0 mM) and nitrate-sufficient (10 mM) conditions at pH 5.5, 6.5, and 7.5. Statistical significance was assessed using one-way ANOVA with Šídák’s multiple comparisons test; ****P < 0.0001. (E) Sensor response reversibility in Line #2 during alternating 0 mM and 10 mM KNO3 treatments; dark green shading indicates 10 mM KNO3, and light green shading indicates 0 mM KNO3. (F) Luminescence imaging of Line #2 plants exposed to 0, 0.25, and 5 mM KNO3, and wild-type Arabidopsis as a negative control, all with substrate applied topically. Color scale indicates luminescence intensity (a.u., quantification in SI Appendix, Fig. S5). In panels A–E, data are shown as mean ± SEM; n = 4 biological replicates with five pooled seedlings per replicate. We next examined whether genomic integration of the sensor affected its specificity for different nutrient sources. As observed in transient expression experiments, robust induction occurred exclusively in response to nitrate, whereas urea, ammonium, glutamine, potassium chloride, or phosphate did not elicit responses above the baseline levels (Fig. 3C), demonstrating that nitrate selectivity was maintained in the stable lines. Given its high maximal output and selectivity, we selected Line #2 for subsequent characterization. As with transient assays, sensor performance was maintained over a range of pH values (Fig. 3D), confirming robustness under environmentally relevant pH conditions. For these experiments, seedlings were first subjected to a 3-d nitrogen starvation period before nitrate treatment. To evaluate response dynamics and reversibility, Line #2 seedlings were exposed to alternating periods of nitrate supplementation at 10 mM KNO3 and nitrate withdrawal over 9 d. Sensor output increased rapidly upon nitrate addition, reached saturation, and then declined toward baseline upon nitrate removal, with this pattern reproducibly observed over two successive cycles (Fig. 3E). Conversely, under continuous 10 mM KNO3 for 14 d, output plateaued after the initial induction and remained stable throughout (SI Appendix, Fig. S3). Together, these results show that the readout tracks changes in nitrate levels while remaining steady when nitrate levels are constant. We next asked whether the nitrogen-starvation preconditioning step was required to preserve nitrate-sensing performance. We compared Line #2 seedlings transferred directly or after a 3-d nitrogen starvation period to 0, 0.25, and 5 mM nitrate. In both groups, time was counted from transfer to nitrate plates. Prestarved plants resolved the full nitrate gradient by day 2, whereas directly transferred plants, which began with higher endogenous nitrate levels, achieved separation between 0.25 and 5 mM nitrate by day 4 (SI Appendix, Fig. S4). Thus, prestarvation accelerates baseline alignment but is not required for nitrate sensing, as plants directly transferred showed the same gradient by day 4. To characterize sensor output across tissues and developmental stages, we used Line #2 plants and sampled developing leaves, mature leaves, and roots separately, measuring each tissue at 0.25 mM and 5 mM nitrate at two developmental time points: when plants had 6-to-8 true leaves (Day 7) and 8-to-12 true leaves (Day 14) (SI Appendix, Fig. S6). In every tissue and at both time points, sensor output was significantly higher at 5 mM than at 0.25 mM nitrate. Mature and developing leaves gave similar signals at each nitrate concentration and time point. Roots gave lower signals than leaf tissues, with values near baseline at 0.25 mM and below the leaf signal at 5 mM (SI Appendix, Fig. S6). Finally, to demonstrate the feasibility of intact, direct plant nitrate sensing, we applied the luciferase substrate to the leaves of Line #2 plants with a pipette and avoided homogenization and extraction (Methods). These plants reported nitrate in a dose-dependent manner, distinguishing 0, 0.25, and 5 mM KNO3 (Fig. 3F), with approximately 5.2-fold increases at 0.25 mM and 8.6-fold at 5 mM, and both significantly above the 0 mM KNO3 control and a wild-type Arabidopsis negative control treated with the same substrate (SI Appendix, Fig. S5). Sentinel Plants Report Bioavailable Nitrate Levels in Agricultural Soils. We evaluated whether the nitrate-sensing sentinel plants retain quantitative responsiveness in agricultural soils collected from four different California sites with diverse soil properties (Davis, Riverside, Five Points, and Hanford; Dataset S4). Nitrate concentrations were quantified using commercial soil analysis (NO3− ppm; right y-axis), and sentinel plant output from the same samples was measured as GeNL/NLuc (left y-axis; Fig. 4A). We also included a nitrate-free peat moss sample to define the sensor’s background, which produced a low but nonzero baseline signal. Across field sites, sensor output tracked the analytical nitrate measurements (Spearman ⍴ = 1.0; Dataset S5): Davis soil contained low NO3− and produced weak sensor activation around the same level as the sensor baseline; Riverside and Five Points had intermediate nitrate levels with corresponding sentinel plant signal; and Hanford soil exhibited the highest NO3− together with the strongest sensor response. Thus, despite the complexities of real agricultural soils, the sensor reliably reported relative differences in nitrate availability. Sentinel plant performance in agricultural soils. (A) Sensor output in agricultural soils collected from four California sites (Davis, Riverside, Five Points, and Hanford) and in a nitrate-free peat moss background control, compared with nitrate concentrations quantified by commercial soil analysis. The Left y-axis shows the sentinel plant output as quantified by GeNL/NLuc, and the Right y-axis denotes the ppm-level nitrate concentrations measured analytically. (B) Sensor response to incremental nitrate supplementation in Davis soil brought to final concentrations of 10, 35, 105, and 200 ppm NO3−, alongside a nitrate-free peat moss sample as the 0-ppm background control (leftmost bar). In both panels, data are mean ± SEM; n = 4 biological replicates with each replicate representing an independent soil sample from the same soil batch, in which five seedlings were grown and pooled for luminescence measurements. Statistical analysis was done by one-way ANOVA followed by Tukey’s post hoc test; ns P > 0.05; **P < 0.01; ***P < 0.001. We next tested whether sentinel plants could resolve incremental nitrate additions within a constant soil background. Using low-nitrate Davis soil, we amended the sample to final concentrations of 10, 35, 105, or 200 ppm NO3− equivalents and measured the sensor output (Fig. 4B). Sensor output increased in a dose-dependent manner with nitrate supplementation across 10 to 200 ppm, an agronomically relevant range, indicating that sentinel plants respond quantitatively to increasing nitrate concentrations within agricultural soil matrices. Sentinel Plants for Testing a Synthetic Nitrogen-Fixing Consortium in Liquid Culture. Beyond monitoring nitrate availability in agricultural soils, this genetically encoded sensor can serve as a platform to evaluate synthetic microbial communities for biofertilizer applications. To test whether sentinel plants could monitor microbially mediated nitrogen transformations, we constructed a minimal synthetic microbial consortium to convert atmospheric nitrogen into plant-accessible nitrate via sequential nitrogen fixation and nitrification. Specifically, we assembled a model synthetic microbiome comprising the diazotroph Azotobacter vinelandii (AV), which generates ammonium from atmospheric nitrogen, and the nitrifying bacteria Nitrosomonas europaea and Nitrobacter winogradskyi, which convert ammonium into nitrate. Wild-type Az. vinelandii (DJ strain) tightly regulates nitrogen fixation and does not typically excrete a significant amount of ammonium into the soil. Therefore, we acquired a mutant Az. vinelandii strain FM2 (AV FM2) (31) harboring a deletion in the nifL regulatory gene, resulting in constitutive nitrogenase activity and ammonium excretion. Consistent with previous reports, when grown in liquid culture, the wild-type strain AV DJ yielded negligible extracellular nitrogen, whereas the mutant AV FM2 strain accumulated approximately 10 and 11 mM ammonium after 4 and 7 d, respectively (Fig. 5A), as quantified using an indophenol colorimetric assay (31). Sentinel plants enable evaluation of a nitrogen-fixing consortium. (A) Nitrogen accumulation and transformation in monocultures and cocultures. Ammonium (NH4+, yellow bars) and nitrate (NO3−, pink bars) concentrations were measured over 7 d in wild-type (AV DJ) and the mutant Az. vinelandii FM2 (AV FM2), with or without the addition of nitrifying bacteria (N. europaea and N. winogradskyi) at day 4. (B) Comparison of sensor output (GeNL/NLuc ratio, blue bars, Left y-axis) and chemically measured nitrate concentrations (pink bars, Right y-axis) at day 7 for the culture conditions shown in panel A. (C) Comparison of sensor output (GeNL/NLuc ratio, blue bars, Left y-axis) and chemically measured nitrate concentrations (pink bars, Right y-axis) at day 7 for the culture conditions shown in SI Appendix, Fig. S7. In all panels, data are mean ± SEM; n = 3 biological replicates, where each replicate represents an independent bacterial culture grown in a separate tube. For sentinel plant measurements, five seedlings were grown per replicate and pooled for luminescence analysis. To mimic natural soil nitrification processes, in which ammonium is rapidly converted to nitrate, we supplemented the Az. vinelandii cultures on day 4 with a consortium of nitrifying bacteria (N. europaea and N. winogradskyi). In cultures containing AV FM2, the accumulated ammonium was rapidly oxidized by the nitrifiers between day 4 and 7, resulting in a final accumulation of ~13 mM nitrate by day 7, whereas cocultures containing wild-type AV DJ produced negligible nitrate (Fig. 5A). These ammonium and nitrate levels were quantified using colorimetric assays (31). We next tested whether sentinel plants could detect these microbially driven nitrogen transformations in a hydroponic setup. Sentinel plants were exposed to the cultures for 24 h before luminescence measurement, the earliest time point at which full sensor induction was seen (SI Appendix, Fig. S3). Despite high concentrations of ammonium present in the AV FM2 monocultures, the sensor output remained at basal levels, confirming that the system is not activated by ammonium or other microbial metabolites (Fig. 5B). Robust sensor induction was observed only in the AV FM2-nitrifier cocultures, where the sensor signal matched the 10 mM nitrate positive control, demonstrating high specificity and quantitative response to the final nitrate product even in the presence of active microbial populations (Fig. 5B). Finally, we evaluated whether sentinel plants could be used to screen environmental factors that influence consortium performance, specifically the effect of the carbon source on nitrogen fixation. Because bacterial nitrogen fixation is energetically demanding, the available carbon source strongly influences nitrogenase activity and ammonium production. We therefore grew the AV FM2-nitrifier cocultures using five different carbon sources. Cultures grown on sugars or sugar alcohols (sucrose, glucose, and mannitol) accumulated high ammonium concentrations at day 4, which were subsequently converted to 6 to 8 mM nitrate by day 7 (SI Appendix, Fig. S7). In contrast, cultures grown on organic acids (malate and succinate) supported significantly less nitrogen fixation, resulting in final nitrate concentrations of ~2 mM, as measured by colorimetric assays. When sentinel plants were exposed to these resulting cultures, sensor output quantitatively tracked the differences in nitrate production generated by the various carbon sources, with strong signals in sugar-derived media and lower signals in organic-acid media, consistent with nitrate concentrations measured by independent chemical assays (Fig. 5C). These results demonstrate that sentinel plants can be used to rapidly screen metabolic outputs of synthetic nitrogen-fixing consortia under varying environmental conditions. Testing the Synthetic Nitrogen-Fixing Consortium in a Model Soil via Sentinel Plants. Ultimately, the practical efficacy of a synthetic nitrogen-fixing consortium depends on its performance in soil rather than liquid culture. To evaluate this, we transitioned to a solid substrate to monitor nitrogen transformations in situ. We selected peat moss as a model soil matrix because of its exceptionally low background levels of inorganic nitrogen (Dataset S4), ensuring that detected nitrate would originate from the introduced microbial consortia. Peat moss pots were inoculated with either engineered AV FM2, wild-type AV DJ, a nondiazotrophic negative control Bacillus subtilis (32), or a heat-killed biomass negative control with AV FM2 (HK AV FM2). To enable nitrification, N. europaea and N. winogradskyi were added to all pots on day 7, and nitrogen dynamics were monitored over a 13-d period. The negative control conditions (heat-killed AV FM2 and B. subtilis) showed negligible nitrogen accumulation throughout the experiment, as measured by both colorimetric assays and sentinel plant outputs (Fig. 6 A and B and SI Appendix, Fig. S8). By day 7, prior to the addition of nitrifiers, both Azotobacter strains had enriched peat moss with ammonium, with the AV FM2 producing significantly more ammonium (~220 ppm) than the wild-type AV DJ (~170 ppm) (SI Appendix, Fig. S8). In situ monitoring of nitrogen transformations via sentinel plants and plant growth promotion. (A) Nitrate accumulation in peat moss inoculated with various bacterial treatments over a 13-d period. Nitrifiers were introduced to all pots at day 7. HK, heat-killed AV FM2 biomass control; BS, B. subtilis nondiazotrophic control; AV DJ, wild-type Az. vinelandii; AV FM2, engineered mutant. (B) In situ sensor performance within the peat moss matrix treated with various bacterial treatments over a 13-d period. Nitrifiers were introduced to all pots at day 7. In panels A and B, n = 3 biological replicates, each an independent inoculated pot containing five sentinel plants. For sentinel measurements in panel B, five seedlings were grown per replicate and pooled for luminescence analysis. (C) Fresh weight of A. thaliana seedlings was assessed 14 d postgermination. n = 10 seedlings, with each seedling representing one replicate. (D) Internal tissue nitrate content (µmol/g FW) of A. thaliana seedlings was assessed 14 d postgermination. Data represent mean ± SEM; n = 3 biological replicates, each consisting of 10 seedlings pooled. Statistical significance was determined by one-way ANOVA with Tukey’s post hoc test; ns P > 0.05; *P < 0.05; **P < 0.01; ****P < 0.0001. Following nitrifier addition, this ammonium pool was progressively converted to nitrate. The higher ammonium production by AV FM2 resulted in significantly greater soil nitrate accumulation, reaching approximately 35 ppm by day 10 and 45 ppm by day 13 (Fig. 6A). In contrast, pots inoculated with AV DJ accumulated substantially less nitrate, plateauing at approximately 10 to 14 ppm during the same period (Fig. 6A). To monitor these nitrogen transformations in situ, sentinel plants were introduced into the inoculated peat moss and exposed for 24 h before luminescence measurement, as above. Sensor output tracked nitrate formation, producing strong signals in AV FM2 pots at days 10 and 13 that corresponded to the elevated soil nitrate concentrations (Fig. 6B). Sensor responses in AV DJ pots were substantially lower, reflecting the reduced nitrate accumulation, whereas HK AV FM2 and B. subtilis control pots remained at baseline levels. To confirm that nitrate production originated from the introduced consortium rather than from the native peat microbiota, we performed 16S rRNA gene amplicon sequencing on peat samples collected at 0, 7, 10, and 13 d from both the AV DJ and FM2 treatments (SI Appendix, Fig. S9). Community profiling confirmed the establishment of Azotobacter in all inoculated pots. Following nitrifier addition, reads corresponding to Nitrosomonas and Nitrobacter increased in abundance, coinciding with nitrate accumulation. Together, these data indicate that the observed ammonium and nitrate pools arise from the activity of the introduced diazotroph-nitrifier consortium and that sentinel plants accurately report these transformations in situ in soil. To evaluate the biofertilizer potential of this three-member synthetic consortium, wild-type A. thaliana seedlings were grown in peat moss pots treated with the consortium, and plant growth was assessed after 14 d. Nitrate produced by the AV FM2-nitrifier consortium supported robust plant growth (SI Appendix, Fig. S10), yielding seedlings with a fresh weight of 15 mg, compared with 1.5 mg in the AV DJ-nitrifier treatment and substantially lower biomass in the heat-killed and B. subtilis control (Fig. 6C). Analysis of seedling tissue confirmed that enhanced growth was associated with increased nitrogen uptake. AV FM2-nitrifier-treated seedlings accumulated significantly higher internal nitrate levels (110 µmol g−1 FW) than AV DJ-nitrifier-treated seedlings (80 µmol g−1 FW), as measured by colorimetric assay (Fig. 6D). Relative to the heat-killed and B. subtilis control, these values correspond to approximately 37-fold and 28-fold increases in internal nitrate content for AV FM2 and AV DJ treatments, respectively. Together, these results demonstrate that the synthetic consortium promotes plant growth in a soil-like environment through a controlled nitrogen-fixation and nitrification pathway that can be accurately and quantitatively monitored by sentinel plants. Discussion This study establishes that plants can function as integrated environmental nitrate sensors that report plant-accessible nutrient availability. We developed a genetically encoded plant biosensing system that enables quantitative monitoring of plant-accessible nitrate across controlled media, agricultural soils, and microbial ecosystems. Because the reporter operates downstream of endogenous nitrate-sensing pathways, the resulting signal reflects nitrate availability experienced by the plant rather than the concentrations measured by soil extraction assays. This distinction is important because plant nutrient status is shaped not only by bulk soil chemistry but also by root uptake, transport, and local microenvironmental conditions. Sentinel plants, therefore, provide a plant-integrated measurement of nitrate availability that complements analytical measurements of soil chemistry and existing biosensors and electrochemical sensors. The sensing system combines a synthetic NLP7-responsive promoter with a ratiometric luciferase reporter system. Systematic variation of cassette orientation, NLP7 binding-site number, and reporter configuration identified key design considerations for quantitative plant biosensing. Cassette orientation strongly affected signal fidelity: the convergent tail-to-tail configuration most closely matched the interference-free cotransfection reference while retaining nitrate inducibility, consistent with reduced transcriptional interference. This finding agrees with previous studies showing that compositional context influences synthetic gene-circuit performance (22, 33) and establishes cassette orientation as an important design variable in plant synthetic biology. Varying NLP7 binding-site number further showed that promoter composition controls dynamic range and basal activity, with four sites providing the best balance between strong induction and low background. Prior genetically encoded plant nitrate sensors were developed primarily to study plant nitrate signaling and intracellular dynamics at the cellular or tissue scale, ranging from fluorescent single-cell biosensors to synthetic transcriptional reporters used in transient assays. The sentinel plant system described here instead provides a stable, quantitative, whole-plant readout deployable for environmental monitoring across soils and microbial systems. Functional characterization showed that the sensor exhibits a dose-dependent response approximating Michaelis–Menten kinetics, consistent with prior findings of nitrate-responsive transcriptional programs in plants (28). The sensor exhibits strong specificity for nitrate over ammonium, urea, glutamine, and phosphate, thereby avoiding the issues observed in some previous nitrate-responsive promoters, in which ammonium suppressed the response (23). Dynamic experiments further showed that the reporter reversibly tracks changes in nitrate availability and maintains a stable signal under sustained nitrate exposure, so its output reflects current availability rather than a memory of past exposure or time-dependent drift. These properties are important for environmental monitoring, where nutrient levels fluctuate over time and selectivity and reversibility are required for meaningful interpretation. Stable genomic integration preserved the performance across Arabidopsis lines, demonstrating that the sensing architecture functions reliably beyond transient assays. Although absolute signal amplitudes varied among lines, likely due to positional effects associated with transgene insertion, response kinetics and nitrate specificity were consistent. These results indicate that the sensing architecture is robust to genomic context, enabling long-term monitoring of nitrate dynamics in stable plant material. Importantly, transient experiments in tomato demonstrated that the sensing architecture also functions in a crop species. The sensor GeNL/NLuc ratio was lower in tomato than in Arabidopsis, due to higher constitutive NLuc expression from the Nos promoter. Although the dynamic range was lower, the quantitative response indicated that the optimized NLP7-responsive module can function across species. Given that the synthetic promoter uses Arabidopsis NLP7 binding sites, reoptimization for a new host could begin by substituting species-matched binding sites and retuning the minimal promoter. A key step toward practical relevance is demonstrating that plant biosensors remain quantitative in complex soil matrices. Across agricultural soils collected from multiple California field sites, sentinel plant output tracked the relative differences in nitrate levels measured using standard analytical methods. The sensor also resolved incremental nitrate additions within a constant soil background across agronomically relevant concentration ranges. The current sensor exhibits reduced quantitative resolution at low nitrate concentrations (<~10 ppm soil nitrate) due to baseline signal. However, because agricultural soils often exhibit higher nitrate levels (34–37), the sensor remains useful for distinguishing nitrogen-limited from nitrate-replete conditions and for comparing relative differences in plant-accessible nitrate across soils and microbial treatments. These results indicate that plant-based biosensors can operate reliably despite the physical, chemical, and biological heterogeneity inherent to soil environments, demonstrating that engineered plants can potentially report nutrient variation under realistic environmental conditions. An important application of this sensing platform is monitoring microbial nitrogen transformations that generate plant-accessible nitrate. Using sentinel plants as readouts, we characterized a minimal three-member microbial consortium capable of converting atmospheric nitrogen into plant-available nitrate through sequential nitrogen fixation and nitrification. Sentinel plant output quantitatively tracked nitrate generated by this microbial community in both liquid culture and soil-like matrices, demonstrating that plant-based biosensors can report microbial metabolic activity relevant to plant nutrition. The current setup used a low-background peat substrate to isolate consortium-driven nitrogen transformations under controlled conditions. Future studies should evaluate sensor performance and consortium function in agricultural soils to assess robustness under realistic environmental complexity. This consortium represents a potential nitrogen biofertilizer strategy, as its activity increased nitrate availability, tissue nitrate accumulation, and plant biomass. Nitrate concentrations reached 45 ppm by day 13, which falls within the agronomically relevant levels (34–36). At the same time, nitrate is highly mobile in soil, susceptible to leaching, and nitrification processes can contribute to nitrous oxide emissions. Future implementation can incorporate strategies that synchronize nitrate generation with plant demand while minimizing environmental nitrogen loss. This platform also enabled rapid evaluation of environmental factors that influence nitrogen fixation efficiency. Variation in carbon source availability substantially altered ammonium production by Az. vinelandii, thereby influencing downstream nitrate accumulation following nitrification. Carbon sources such as sucrose, glucose, and mannitol supported higher nitrate production than organic acids such as malate and succinate, consistent with previous reports that carbon metabolism strongly regulates diazotrophic nitrogen fixation (38). Sentinel plant output accurately reflected these shifts in nitrate production, illustrating how plant-integrated sensing platforms can accelerate design-build-test cycles for optimizing microbial communities. Extending these capabilities for agricultural deployment will require additional platform engineering. Currently, the sensor relies on a luciferase-based readout that requires exogenous substrate addition, which could be sprayed for in situ detection. However, future iterations should incorporate autonomous bioluminescence pathways (39, 40) or replace luminescence with visible reporters, such as pigment-based reporters [e.g., RUBY (41)] or UV-excitable fluorescent proteins (42), enabling direct visualization and compatibility with camera- or drone-based imaging for in situ large-scale monitoring. Temporal resolution is a further consideration. Because the luminescent output requires reporter transcription, translation, and maturation, the signal develops over an exposure window and reflects nitrate sustained across that window rather than instantaneous changes. This behavior suits multiday soil and microbial dynamics, such as those studied here, but limits the resolution of faster, subday fluctuations. Faster or destabilized reporters could extend the platform toward near-real-time sensing. Field deployment should also account for how diurnal transcriptional dynamics (43) and nitrate interactions with other nutrients (44) influence sensor performance. More broadly, the sensing architecture developed here establishes a framework for engineering plants as environmental biosensors that can monitor diverse soil and rhizosphere processes in agricultural ecosystems. Because the platform relies on modular promoter design and ratiometric normalization, similar approaches could be adapted to detect additional nutrients, metabolites, or environmental cues relevant to plant growth. These sensing modules could also be coupled to regulatory circuits that alter plant behavior in response to nutrient status, extending sentinel plants from passive reporters toward responsive plants with programmable environmental adaptation. Methods Plasmid Construction and Bacterial Strains. Plasmids were designed in SnapGene (SnapGene, GSL Biotech) and assembled by BsaI/SapI-mediated Golden Gate loop assembly; a complete list of standard parts and expression constructs is given in Dataset S6, and all plasmids and sequences are available from Addgene (Dataset S6). Agrobacterium tumefaciens GV3101 (1282PS-12, Intact Genomics) was cultured in LB. Az. vinelandii strains DJ and FM2 (gifts from the Ané Laboratory, University of Wisconsin-Madison, Madison, WI) were cultured in Burk’s medium (45). N. europaea (19718, ATCC) and N. winogradskyi (25391, ATCC) were cultured in ATCC Medium 2265 and DSMZ Medium 756a, respectively. Unless otherwise stated, strains were grown at 30 °C with shaking at 250 rpm. Plant Growth and Transformation. A. thaliana (Col-0) seeds were surface-sterilized with chlorine gas (100 mL bleach plus 3 mL 6 N HCl in a sealed desiccator, ≥3 h), and tomato (Solanum lycopersicum cv. Micro-Tom) seeds in 33% (v/v) bleach with 0.1% Tween-20 for 6 min followed by four to five sterile-water rinses. Seeds were germinated on 1/2-strength MS plates (0.5× Murashige and Skoog Basal Salt Mixture, M524, PhytoTech Labs; 0.05% [w/v] MES; 0.7% Phytagel, P8169, Millipore Sigma; pH 5.7) at 22 °C under a 16/8 h light/dark cycle. For protoplast assays, mesophyll protoplasts were isolated from fully expanded leaves of 4-wk-old A. thaliana (Col-0) and transformed as described by Muchenje et al. (46) Transformed protoplasts were incubated in WI solution with the indicated KNO3 concentrations for 24 h for cassette-orientation and luciferase optimization, or for 3 d for the Liu et al. reporter comparison, after which luminescence was quantified (Sensor Output Measurement and Characterization). Each technical replicate (n) is an independent transfection of 5 × 104 protoplasts, treated and processed independently and assayed on the same day. Each experiment was repeated twice with the same results (n = 4 for Fig. 1B and SI Appendix, Fig. S1; n = 8 for Fig. 1C). For transient assays in seedlings, nitrate-responsive constructs were introduced into 7-d-old A. thaliana seedlings at the two-true-leaf stage by the VAST (Vacuum- and Sonication-Assisted Transient Expression) protocol (26). After 2 d of cocultivation with A. tumefaciens, seedlings were rinsed, transferred to nitrogen-free 1/2 MS plates (M531, PhytoTech Labs) with 10 mM KCl, and starved for 3 d, then moved to nitrogen-free 1/2 MS plates with 100 µM timentin and the indicated treatments (KNO3 concentrations or 10 mM alternative nitrogen sources, with KCl to equalize potassium, or plates buffered to specified pH) for 24 h before quantification. Each biological replicate (n) is a separate pool of five seedlings, treated, harvested, and measured independently on the same day (n = 4 per condition; Figs. 1 E and F and 2). Tomato seedlings were transformed using the same protocol (26), germinated on nitrogen-free 1/2 MS plates until the cotyledons were fully expanded (~9 d), cocultivated for 2 d, and treated with defined concentrations of KNO3 for 24 h (n = 4 pools of five seedlings per condition; SI Appendix, Fig. S2). Stable A. thaliana lines were generated by spraying inflorescences with a suspension of A. tumefaciens GV3101 carrying the sensor constructs, selecting T1 and T2 seeds by the seed-specific FAST-red marker under a fluorescence microscope, and confirming homozygous lines by PCR and electrophoresis. Homozygous lines were germinated on standard 1/2 MS (M524) for 9 d, starved on nitrogen-free 1/2 MS with 10 mM KCl for 3 d, and then treated as above for 24 h before quantification (n = 4 per condition; Fig. 3). For soil assays, agricultural soils from California field sites were sieved (2 mm), mixed 1:1 (w/w) with peat moss to improve aeration, and adjusted to 70% field capacity (determined gravimetrically as 70% of the water retained after gravitational drainage); seedlings were incubated in the indicated soils for 24 h, with each replicate an independent soil sample from the same batch in which five seedlings were grown and pooled (n = 4 per condition; Fig. 4). Sensor Output Measurement and Characterization. For quantification, all leaves pooled from five seedlings (harvested when the two cotyledons and first four true leaves were fully expanded; 3 d of starvation followed by 1 d of nitrate), or 5 × 104 protoplasts, were collected into 2-mL tubes preloaded with 2.8-mm ceramic beads (15-340-160 and 15-340-162, Fisher Scientific). Samples were collected in the afternoon to limit diurnal effects, flash-frozen in liquid nitrogen, bead-beaten (15-340-164, Fisher Scientific) for 40 s, lysed in 100 µL of 1× Cell Lysis Buffer (E1531, Promega), held on ice for 10 min, and centrifuged (5,000× g, 5 min). 15 µL of clarified extract plus 30 µL of luciferase reagent was read at 25 °C on a Tecan Spark (100-ms integration). GeNL/NLuc was measured with the Nano-Glo Luciferase Assay System (N1110, Promega; GeNL 520 to 590 nm, NLuc 430 to 485 nm) and the ELuc/RedF pair with the Luciferase Assay System (E1500, Promega; ELuc 490 to 545 nm, RedF 610 to 700 nm); signals were resolved by spectral deconvolution (Promega ChromaLuc Calculator, TM062). To benchmark the sensor against an existing NLP7-based reporter, the 4×NRE-min-LUC reporter (17) (gift from the Sheen laboratory, Massachusetts General Hospital and Harvard Medical School, Boston, MA) was assayed in parallel in A. thaliana protoplasts at 0, 0.25, or 5 mM KNO3 (3-d incubation). It was read using the Dual-Luciferase Reporter Assay System (E1910, Promega) and our sensor (Promega N1110), with fold induction calculated relative to 0 mM (n = 4; SI Appendix, Fig. S1). To test temporal behavior, starved Line #2 seedlings were held on continuous 10 mM KNO3 and read over 14 d (n = 4; SI Appendix, Fig. S3); and to test whether starvation is required, Line #2 seedlings were moved to 0, 0.25, or 5 mM KNO3 either directly from standard 1/2 MS or after a 3-d starvation, with time 0 defined as transfer and output read at the indicated time points through day 6 for directly transferred seedlings and through day 4 for starved seedlings (n = 4 per condition; SI Appendix, Fig. S4). For tissue and developmental profiling, starved Line #2 seedlings exposed to 0.25 or 5 mM KNO3 were sampled at two developmental time points: when plants had 6-to-8 true leaves (Day 7) and 8-to-12 true leaves (Day 14); each plant’s leaves were divided into a younger and an older half, roots were collected separately, and each fraction (pooled from five seedlings) was processed as above (n = 4; SI Appendix, Fig. S6). For whole-plant imaging, roots were removed from starved Line #2 seedlings that had been treated for 24 h with 0, 0.25, or 5 mM KNO3, and the seedlings were arranged adaxial side up on a transparent plate over a black imaging stage. 100 µL of Nano-Glo Live Cell substrate (N2011, Promega), diluted 1:50 in LCS Dilution Buffer with 0.01% Triton X-100, was pipetted onto the leaves without homogenization or vacuum infiltration. After 3 min, seedlings were imaged on an IVIS system (8 s exposure, binning factor 2, f/1.2 aperture) and total luminescence per seedling quantified as integrated density in ImageJ/Fiji (n = 5 per condition; Fig. 3F and SI Appendix, Fig. S5). Synthetic Consortium Cultivation. Nitrate-forming medium (NFM) was a base medium (153 mg M407, PhytoTech Labs, plus 10 g sucrose to 500 mL, autoclaved) supplemented at 1:99 with a filter-sterilized 100× phosphate buffer (1.10 g KH2PO4, 5.77 g K2HPO4 3H2O, and 7.5 g KHCO3 to 50 mL). For liquid coculture, Az. vinelandii (DJ or FM2) was grown in NFM (Starting OD600 = 0.01, 4 d), after which pelleted N. europaea and N. winogradskyi were resuspended directly into the culture in equal volume (no added medium), and cocultures were held at 30 °C, 250 rpm. Each replicate is an independent culture in a separate tube, assayed on the same day, with each experiment repeated twice (n = 3 per condition; Fig. 5 and SI Appendix, Fig. S7); for the sentinel-plant readout, five seedlings were grown per replicate and pooled. For peat-moss cocultivation, 25 g of peat moss in a Magenta box (30930007, Magenta) was adjusted to ~pH 7.5 with KOH and inoculated with overnight Az. vinelandii in NFM to ~1 × 108 cells per gram at 70% field capacity, mixed, sealed with Micropore tape (19571, Qiagen), and incubated at 22 °C (16/8 h), with moisture maintained using NFM. On day 7, pelleted N. europaea and N. winogradskyi (55 mL of each culture) were added. Ammonium and nitrate were measured on days 0, 7, 10, and 13, and peat was sampled for 16S sequencing at the same time points. Each replicate is an independently inoculated pot, with each experiment repeated twice (n = 3 per condition; Fig. 6 A and B and SI Appendix, Fig. S8). Plant fresh weight was measured on n = 10 seedlings per treatment (Fig. 6C), and internal tissue nitrate on n = 3 pools of 10 seedlings (Fig. 6D). Quantification of Nitrogen Species. Ammonium was quantified by a Berthelot-reaction assay adapted from Mus et al. (31) (50 µL of sample or NH4Cl standard [0 to 5 mM] reacted with 100 µL phenol-nitroprusside [1% phenol, 0.005% sodium nitroprusside], then 100 µL alkaline hypochlorite [1% sodium hypochlorite, 0.5% NaOH], 37 °C for 30 min, read at 620 nm). Nitrate and nitrite were quantified by a Griess assay adapted from Hachiya et al. (47) (50 µL of sample or standard [0 to 320 µM] mixed with 50 µL sulfanilamide–VCl3 [0.25% VCl3, 1% sulfanilamide in 1 M HCl] then 50 µL N-(1-naphthyl)ethylenediamine [0.02%], 6 to 8 h at room temperature). Concentrations were interpolated from standard curves using at least two technical and three biological replicates, unless otherwise noted. Nitrogen is reported in mM (mmol N L−1) for liquid culture and ppm (mg N kg−1 dry soil) for soil, as nitrogen equivalents (NH4+–N and NO3-–N). 16S rRNA Gene Amplicon Sequencing and Microbial Community Analysis. Microbial community composition was assessed using full-length 16S rRNA gene amplicon sequencing. DNA was extracted from peat moss samples using the DNeasy PowerSoil Pro Kit (47014, Qiagen) according to the manufacturer’s instructions. Extracted DNA was used as a template to amplify the full-length 16S rRNA gene (~1.5 kb) using the Oxford Nanopore Microbial Amplicon Barcoding Kit 24 (SQK-MAB114.24) following the manufacturer’s protocol (Oxford Nanopore Technologies). Briefly, approximately 10 ng of genomic DNA per sample was used as input for PCR amplification with kit-provided primers targeting the bacterial 16S rRNA gene. Amplicons were barcoded during PCR to enable multiplexing of samples. Barcoded PCR products were pooled and purified using magnetic bead cleanup prior to adapter attachment. Libraries were prepared according to the Oxford Nanopore microbial amplicon sequencing workflow and loaded onto R10.4.1 flow cells (FLO-MIN114) for sequencing on a Nanopore platform. The analysis pipeline was adapted from Aja-Macaya et al. (48) In summary, reads were first base-called using the Dorado basecaller super accurate (SUP) model, then demultiplexed with Dorado. The resulting reads were quality-controlled using chopper, filtered for a minimum quality score of 20, a minimum length of 1,200, and a maximum length of 1,900, and trimmed by 20 nt at the front and back of the sequence. Reads were then aligned and identified with Emu using the default database (48). Statistical Analysis and Reproducibility. Cassette orientations were compared to the cotransfection reference by ordinary least squares regression of each orientation’s mean dose response against the corresponding CoT means, with slope, intercept, and R2 reported in Dataset S1. Low-nitrate linear fits (0 to 0.5 mM) and Michaelis–Menten fits are reported in Dataset S2. Agreement between sentinel-plant output and analytically measured soil nitrate was assessed by Spearman rank correlation (Dataset S5). Outliers were identified by Grubbs’ test where indicated. Group comparisons used two-tailed unpaired t tests, one-way ANOVA with Tukey’s post hoc test, or one-way ANOVA with Šídák’s multiple-comparisons test, as stated in each figure legend. Significance is denoted ns P > 0.05, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Analyses were performed in GraphPad Prism. Supplementary Material Appendix 01 (PDF) Dataset S01 (XLSX) Dataset S02 (XLSX) Dataset S03 (XLSX) Dataset S04 (XLSX) Dataset S05 (XLSX) Dataset S06 (XLSX)