An l-2-hydroxyglutarate biosensor based on specific transcriptional regulator LhgR

l-2-Hydroxyglutarate (l-2-HG) plays important roles in diverse physiological processes, such as carbon starvation response, tumorigenesis, and hypoxic adaptation. Despite its importance and intensively studied metabolism, regulation of l-2-HG metabolism remains poorly understood and none of regulator specifically responded to l-2-HG has been identified. Based on bacterial genomic neighborhood analysis of the gene encoding l-2-HG oxidase (LhgO), LhgR, which represses the transcription of lhgO in Pseudomonas putida W619, is identified in this study. LhgR is demonstrated to recognize l-2-HG as its specific effector molecule, and this allosteric transcription factor is then used as a biorecognition element to construct an l-2-HG-sensing FRET sensor. The l-2-HG sensor is able to conveniently monitor the concentrations of l-2-HG in various biological samples. In addition to bacterial l-2-HG generation during carbon starvation, biological function of the l-2-HG dehydrogenase and hypoxia induced l-2-HG accumulation are also revealed by using the l-2-HG sensor in human cells.

L-2-HG is an inhibitor of 2-KG dependent dioxygenases with specific pro-oncogenic capabilities 10,11 . Thus, this oncometabolite is viewed as a biomarker for a variety of cancers and its rapid and sensitive measurement in body fluids is of clinical significance [12][13][14][15] . Importantly, L-2-HG also has endogenous functions in healthy animal cells. For example, this compound was recently identified to aid the proliferation and antitumorigenic abilities of CD8 + T-lymphocytes 16 , to contribute to relieving the cellular reductive stress 3 , and to coordinate glycolytic flux with epigenetic modifications 17 . Considering the various roles of L-2-HG in cell metabolism, the development and optimization of real-time monitoring assays for this metabolite in living cells are required.
Liquid chromatography-tandem mass spectrometry (LC-MS/ MS) 18,19 and gas chromatography-tandem mass spectrometry (GC-MS/MS) 20,21 are often used to assess the extracellular concentrations of L-2-HG. These state-of-the-art methods are timeconsuming, expensive to perform, and require highly skilled personnel. In addition, these destructive methods are also incompatible with real-time monitoring of the fluctuations of L-2-HG concentrations in intact living cells. In this study, we identify and characterize LhgR, an L-2-HG catabolism regulator in Pseudomonas putida W619. Mechanistically, LhgR represses the transcription of LhgO encoding gene lhgO. L-2-HG is a specific effector molecule of LhgR and prevents LhgR binding to the promoter region of lhgO. We then report the development and application of the LhgR-based L-2-HG biosensor via Förster resonance energy transfer (FRET), a technology widely applied to investigate temporal dynamics of various small molecules, such as potassium 22,23 , glycine 24 , and cAMP 25,26 . As-designed sensor quantitatively responds to L-2-HG concentrations in various biological samples with high accuracy and precision. We also use this biosensor to identify the carbon starvation-induced L-2-HG production in bacteria and to demonstrate hypoxia-induced L-2-HG production by LDH and MDH in human cells. Therefore, the LhgR-based biosensor can prove to be a useful tool for real-time measurement of the L-2-HG concentrations in living cells.

Results
LhgR regulates L-2-HG catabolism. In this study, bacteria containing LhgO encoding gene lhgO were selected to study the regulation of L-2-HG metabolism. Homologs of LhgO can be found in 612 different bacterial strains. Similarly organized chromosomal clusters are found in many bacterial genomes, which contain various combinations of genes related to glutarate metabolism (csiD, lhgO, gabT, gabD, and gabP) (Fig. 1a). In Pseudomonas putida KT2440, the glutarate regulon is regulated by allosteric transcription factor CsiR, which is encoded upstream of csiD 27 . The glutarate sensing allosteric transcription factor CsiR and its cognate promoter were cloned into broad host range vectors to create a glutarate biosensor 28 . Interestingly, a different pattern of lhgO gene neighborhood was observed in a few species that do not contain csiD homologs (Fig. 1a). For example, a gene encoding a GntR family protein, lhgR, was found directly upstream of lhgO in P. putida W619. The absence of csiD gene related to glutarate catabolism made us to reason that lhgO of P. putida W619 might be solely involved in L-2-HG metabolism and be L-2-HG inducible.
The lhgO gene in P. putida W619 was cloned into pME6032 vector and transferred into P. putida KT2440 (ΔlhgO). As shown in Fig. 1b, c, the complement of lhgO in P. putida W619 could restore glutarate and L-2-HG utilization abilities of P. putida KT2440 (ΔlhgO), confirming that lhgO encodes a functional L-2-HG catabolic enzyme. To identify the function of LhgR in P. putida W619, the gene segment F2-lhgR-F1-lhgO, which contains the promoter of lhgR (F2), lhgR, the promoter of lhgO (F1), and lhgO, was cloned into pME6032 vector, and the resulting plasmid was transferred into different derivatives of P. putida KT2440 (Fig. 1d). As shown in Fig. 1e, exogenous L-2-HG, but not its mirror-image enantiomer D-2-HG, can induce the expression of lhgO in the gene segment F2-lhgR-F1-lhgO and restore LhgO activity in P. putida KT2440 (ΔcsiRΔlhgO). In addition, the activity of LhgO was also detected in P. putida KT2440 (ΔcsiRΔlhgO) harboring pME6032-F2-lhgR-F1-lhgO when cultured with glutarate as the sole carbon source. However, no activity of LhgO was detected in P. putida KT2440 (ΔcsiRΔc-siDΔlhgO), in which the key gene responsible for L-2-HG production from glutarate was deleted. These results indicated that LhgR represses the expression of LhgO and L-2-HG, but neither D-2-HG nor glutarate can serve as the effector molecule of LhgR.
Effects of L-2-HG, D-2-HG, glutarate, 2-KG, L-lysine, 5aminovalerate, and succinate on LhgR binding to the lhgO promoter region F1 were also assessed by EMSAs. The release of LhgR from fragments F1 was observed only in the presence of L-2-HG (Fig. 2e). These results suggested that L-2-HG can specifically prevent the binding of LhgR to the promoter of lhgO and induce its expression. LhgR may help P. putida W619 to specifically sense L-2-HG generated by intracellular metabolism or present in habitats and stimulate the catabolism of L-2-HG. In addition, LhgR may self-repress its expression and L-2-HG can also contribute to inducing the expression of lhgR (Supplementary Fig. 1c).
Design and optimization of the L-2-HG-sensing reporter. FRET sensors, which combine a ligand-binding moiety and a pair of donor-acceptor fluorescent pair, allow measurement of ligand concentrations based on the ligand-binding induced changes of FRET efficiency [22][23][24][25][26] . In this study, the L-2-HG-sensing fluorescent reporter (LHGFR) was constructed by fusion of the optimized cyan and yellow fluorescent protein variants, mTFP 30 and Venus 31 , to the N-terminus and C-terminus of LhgR (Supplementary Fig. 2). This first LHGFR was named LHGFR 0N0C , where the subscript indicates the number of amino acids truncated from the N-terminus and C-terminus of LhgR. Subsequently, LHGFR 0N0C was overexpressed in E. coli BL21(DE3) and purified by a Ni-chelating chromatographic column ( Supplementary  Fig. 3). Spectra properties of LHGFR 0N0C reveal the addition of L-2-HG could reduce the emission peak at 492 nm of mTFP and increase the emission peak at 526 nm of Venus ( Supplementary  Fig. 4). Thus, the conformational change of LhgR after the L-2-HG binding may lead to a shortened relative distance and/or favorable orientation of mTFP and Venus, resulting in an increase in FRET (Fig. 3a). In addition, L-2-HG increased the emission ratio of Venus to mTFP in a dose-dependent manner, with a maximum ratio change (ΔR max ) of 11.47 ± 0.38%, an apparent dissociation constant (K d ) of 2.74 ± 0.73 μM, and a Hill slope close to 1 (Fig. 3b).
To increase the magnitude of responses, LHGFR was optimized by truncating N-terminal and C-terminal amino acids of LhgR or by adding a series of artificial linkers between LhgR and various fluorescent proteins 23,32-34 ( Fig. 3c and Supplementary Fig. 5). Truncation of three to seven C-terminal amino acids in LhgR could significantly increase ΔR max of the sensor (Fig. 3c, d and Supplementary Fig. 5). Among the five sensors with increased response magnitude values, LHGFR 0N7C exhibited the highest ΔR max of 60.37 ± 1.30% and K d of 7.22 ± 0.38 μM (Fig. 3d and Supplementary Fig. 6a-e). In addition, LHGFR 0N3C was also a promising sensor with a high ΔR max of 56.13 ± 0.29% and a high K d of 29.33 ± 1.24 μM ( Fig. 3d and Supplementary Fig. 6a).
Characterization of LHGFR in biological samples. Next, we investigated whether LHGFR could be used to quantify L-2-HG concentrations in different biological samples. When L-2-HG with increasing concentrations (0 to 2 mM) were added into serum and urine samples of healthy adults, the response curves were nearly identical with that in assay buffer for both LHGFR 0N3C and LHGFR 0N7C (Fig. 4a-d, Supplementary Fig. 6a and Fig. 6e). Thus, quantitative determination of L-2-HG could be conducted by mixing the target samples with LHGFR and measuring the emission ratios with a conventional fluorescence microplate reader. Based on the response curves established for L-2-HG quantification, both biosensors were used to assay the concentrations of L-2-HG in human serum and urine (Fig. 4e, f).
The results of LHGFR 0N3C and LHGFR 0N7C showed close agreement with the results of LC-MS/MS, the current standard method for clinical assays of L-2-HG (Fig. 4e, f and Supplementary Table 2). In a previous report, L-2-HG was identified to be a metabolic intermediate of glutarate metabolism in P. putida KT2440 7 . LHGFR 0N3C and LHGFR 0N7C also exhibited high accuracy in the quantification of L-2-HG in bacterial culture medium ( Fig. 4g-i). When cultured in the medium containing 20 mM glucose and 5 mM glutarate as carbon sources, the growth of P. putida KT2440 (ΔlhgO) was significantly delayed, which might be due to the possible toxicity of accumulated L-2-HG (Fig. 4j). Nearly identical results of L-2-HG quantification were also obtained by either using LHGFR 0N3C , LHGFR 0N7C , or using LC-MS/MS (Fig. 4k, l and Supplementary Fig. 11). Mutual corroboration between the two biosensors further confirmed their applicability in in vitro L-2-HG quantification of various biological samples. In addition, the response curves of LHGFR 0N3C and LHGFR 0N7C for D-2-HG in serum, urine, and bacterial culture medium were also determined ( Supplementary Fig. 12). The LODs of LHGFR 0N3C and LHGFR 0N7C for D-2-HG were more than 100-fold higher than those for L-2-HG (Supplementary Table 3 and Table 4). LHGFR could detect possible variations of L-2-HG in living bacteria. LHGFR 0N3C and LHGFR 3N7C , a control biosensor that did not respond to L-2-HG in vitro (Fig. 3c), were expressed in E. coli BL21 (DE3). Exogenous L-2-HG was added to the culture system of E. coli BL21(DE3) to achieve final concentrations between 0 and 10 mM and the emission ratio was continuously recorded. As shown in Fig. 5a, exogenous L-2-HG could increase emission ratios of LHGFR 0N3C in a dose-dependent manner. The apparent K d of LHGFR 0N3C expressed in E. coli BL21(DE3) was determined to be 891.72 ± 32.10 μM by fitting emission ratios against exogenous L-2-HG concentrations (Fig. 5b). Maturation time lag and degradation of biosensors may be reasons responsible for higher apparent K d of LHGFR 0N3C when expressed in E. coli 35 . The specificity of LHGFR 0N3C expressed in E. coli BL21(DE3) was also characterized. As shown in Fig. 5c, only exogenous L-2-HG could significantly increase the emission ratio in E. coli BL21(DE3), while glutarate, D-2-HG, and glucose could not.
Besides being a metabolic intermediate of exogenous glutarate catabolism in P. putida KT2440, L-2-HG is also reported as a metabolite produced from endogenous L-lysine during carbon  starvation of E. coli 36,37 . Thus, whether carbon starvation could induce intracellular L-2-HG accumulation of E. coli was investigated. As shown in Fig. 5d, no change in the emission ratio was detected when 20 mM glucose was added to the culture system. However, the emission ratio increased during carbon starvation of E. coli BL21(DE3), suggesting that carbon starvation induced L-2-HG production. The emission ratios also increased after culturing E. coli cells for 6 h with glucose addition (Fig. 5e), which might be due to carbon starvation induced by depletion of exogenous glucose. In addition, the emission ratio increased at the beginning of carbon starvation, reaching a maximum value at 3 h and then decreased to initial levels at 8 h (Fig. 5e). These results confirmed that L-2-HG is a temporary metabolite during carbon starvation and LHGFR 0N3C can be used to monitor the real-time change in intracellular L-2-HG concentrations.
To identify whether carbon starvation-induced endogenous L-2-HG production also results from the glutarate hydroxylase activity of CsiD, gene csiD was disrupted and LHGFR 0N3C was expressed in E. coli MG1655(DE3). Gene lhgO in E. coli MG1655 (DE3) was also disrupted to investigate its role in endogenous L-2-HG catabolism. As expected, the emission ratio of LHGFR 0N3C in E. coli MG1655(DE3) (ΔcsiD) remained unaffected during carbon starvation, whereas disruption of lhgO significantly increased the emission ratio of LHGFR 0N3C in E. coli MG1655(DE3) (ΔlhgO) (Fig. 5f), indicating the roles of CsiD and LhgO in endogenous L-2-HG metabolism during carbon starvation. The performance of LHGFR 0N7C in monitoring L-2-HG fluctuations in living bacteria was also studied and similar results were acquired (Supplementary Fig. 13). As for the control biosensor LHGFR 3N7C , no change of emission ratio could be detected in living bacteria for any of the above-mentioned experiments ( Supplementary Fig. 14).
Monitoring L-2-HG production in human cells by LHGFR. Next, LHGFR 0N3C , LHGFR 0N7C , and LHGFR 3N7C were expressed in the cytosol of HEK293FT cells. The addition of 10 mM L-2-HG affected mTFP and Venus fluorescence intensities, which caused a non-uniform increase in the emission ratio of LHGFR 0N3Cexpressing single cell ( Fig. 6a and Supplementary Movie). The average emission ratio reached a maximum value at 5 min and remained constant during the subsequent confocal imaging period (Fig. 6b). The apparent K d of LHGFR 0N3C expressed in   HEK293FT cells for L-2-HG was determined to be 43.79 ± 3.05 μM (Fig. 6c). Based on the emission ratio of non-permeabilized HEK293FT cells under physiological conditions, the basal L-2-HG concentration in LHGFR 0N3C -expressing cells was 22.95 ± 11.22 μM (Fig. 6c). Only exogenous L-2-HG could significantly increase the emission ratio of LHGFR 0N3C in 10 μM digitoninpermeabilized HEK293FT cells, suggesting the specificity of the biosensor inside living human cells (Fig. 6d). L2HGDH, the only reported enzyme that is able to catabolize L-2-HG in human cells, is localized in mitochondria 13 . The mitochondrial targeting sequence was appended to LHGFR 0N3C to localize the biosensor in mitochondria (Supplementary Fig. 15a). Exogenous L-2-HG also induced an increase in the emission ratio of mitochondrial LHGFR 0N3C (Supplementary Fig. 15b). The emission ratio of LHGFR 0N3C localized in mitochondria was similar to that of in cytosol ( Supplementary Fig. 15c). The uniform distribution of L-2-HG confirmed the presence of a transporter responding for the transport of L-2-HG between cytosol and mitochondria.
HEK293FT cells were then co-transfected with siRNA targeting L2HGDH and LHGFR 0N3C . As shown in Fig. 6e, the transfection of siRNA targeting L2HGDH increased the emission ratio of LHGFR 0N3C , indicating an accumulation of intracellular L-2-HG due to a decrease in L2HGDH levels. Cells with different emission ratios could be easily distinguished in a mixture of HEK293FT cells with or without L2HGDH knockdown ( Supplementary Fig. 16a-b). As expected, overexpression of L2HGDH decreased the emission ratio of LHGFR 0N3C , further supporting the function of L2HGDH in L-2-HG catabolism ( Supplementary Fig. 16c).
The response of LHGFR 0N3C to changes in hypoxia-induced production of L-2-HG in human cells was also studied. The emission ratio of LHGFR 0N3C in HEK293FT cells after 24 h exposure to 2% oxygen was higher than the ratio obtained under normoxic conditions and hypoxia induced a 3.5-fold increase in the concentration of L-2-HG (Fig. 6f). In addition, exogenous cellpermeable dimethyl-2-ketoglutarate significantly increased the emission ratio of LHGFR 0N3C under hypoxic conditions, suggesting that hypoxia-induced L-2-HG might originate from 2-KG (Fig. 6f).
LDHA and MDH2 have been reported to participate in hypoxia-induced L-2-HG production due to their "promiscuous" catalytic activities [1][2][3][4] . In support of this conclusion, siRNAs targeting LDHA and MDH2 were transfected separately or in combination into LHGFR 0N3C -expressing HEK293FT cells. As shown in Fig. 6g, the decrease of LDHA and MDH2 reduced the emission ratio of LHGFR 0N3C under hypoxic conditions, suggesting that these enzymes indeed contribute to the production of L-2-HG from 2-KG. The performance of LHGFR 0N7C in monitoring L-2-HG fluctuations in living HEK293FT cells was also studied and similar results were acquired (Supplementary Fig. 17). In addition, the emission ratio of control cells expressing LHGFR 3N7C remained unaffected by exogenous L-2-HG addition and hypoxia treatment (Supplementary Fig. 18).

Discussion
Bacteria have evolved to respond and catabolize a wide range of metabolites. The availability of genomic information from different organisms provided researchers with a new route to predict transcriptional regulators and their physiological functions. In this study, we used a genomic neighborhood analysis approach combined with genetic and biochemical techniques to discover the transcriptional regulator of L-2-HG catabolism. The identified transcriptional regulator, LhgR, was present directly upstream of LhgO in P. putida W619 (Fig. 1a). It specifically binds to the promoter region of lhgO and represses the transcription of lhgO gene. L-2-HG interferes with the DNA-binding activity of LhgR and induces expression of LhgO. LhgR is an example of the bacterial allosteric transcription factor that specially responds to L-2-HG. This finding showed the application of a collection of sequenced genomes in the identification of transcriptional regulators and the approach can be expanded to target other transcriptional regulators in diverse bacteria. L-2-HG is a harbinger of altered metabolism and participates in the pathogenesis of L-2-hydroxyglutaric acidurias and cancer [12][13][14][15]38,39 . Standard methods to measure L-2-HG are based on MS techniques that are time-consuming and require highly skilled workers. LhgR can bind to L-2-HG and then undergo conformational changes, which in turn affects DNA-binding. Thus, a FRET biosensor, LHGFR 0N0C , utilizing the allosteric transcription factor LhgR as an L-2-HG biorecognition element was constructed for a convenient assay of L-2-HG concentrations. As a ratiometric sensor, the emission ratio changes of LHGFR are not affected by the amount of sensor in biological samples or in living cells, and thus allows more accurate measurements. The biosensor was optimized by truncating the N-terminal and Cterminal domains of LhgR or by adding an artificial linker to the N-terminal and C-terminal regions of LhgR. The optimized variants, LHGFR 0N3C and LHGFR 0N7C , increased ΔR max from 11.47 ± 0.38% to 56.13 ± 0.29% and 60.37 ± 1.30%, respectively, with significantly improved sensitivity for L-2-HG detection (Fig. 3b, Supplementary Fig. 6a and Fig. 6e). Besides signal recognition, signal transduction is also an essential aspect of the development of biosensors. Various biosensing systems, including CRISPR-Cas12a-and allosteric transcription factors-mediated small molecule detector (CaT-SMelor) 40 , allosteric transcription factors-based nicked DNA-template-assisted signal transduction (aTF-NAST) 41 , and quantum-dot-allosteric transcription factors-FRET 42 , have been reported. Other biosensors based on L-2-HG responding LhgR and the corresponding transduction mechanisms could also be developed for the detection of L-2-HG.
L-2-HG is a biomarker for L-2-hydroxyglutaric aciduria and a variety of cancers. The rapid, sensitive, and specific measurement of L-2-HG in body fluids is of clinical significance [12][13][14][15]38,43 . The LODs of LHGFR 0N3C and LHGFR 0N7C for L-2-HG in serum and urine were 5.84 μM and 15.74 μM, 1.68 μM and 0.92 μM, Table 3 Table 4). Compared with LHGFR 0N3C , LHGFR 0N7C has a higher affinity with D-2-HG. However, this biosensor is a sensitive variant with low LOD for L-2-HG (Supplementary Table 3). Assessment of L-2-HG by LHGFR 0N7C can be easily conducted after a simple dilution to prevent the possible interference induced by D-2-HG-related diseases. Besides higher sensitivity and specificity, LHGFR 0N3C and LHGFR 0N7C also have superior accuracy and precision over previous MS-based methods for L-2-HG detection ( Fig. 4 and Supplementary Table 2). Being genetically encoded, LHGFR 0N3C and LHGFR 0N7C could be produced in great quantities by recombinant bacteria with low cost and could be applied in future rapid and sensitive clinical diagnosis of L-2-HG-related diseases.

respectively (Supplementary
L-2-HG plays important roles in diverse physiological processes such as hypoxic adaption, immunity, and tumorigenesis, and the establishment of the intracellular detection method of this metabolite is of great research significance 3,12,14,16 . Intracellular L-2-HG concentrations of activated CD8 + T-lymphocytes, renal cancer cells, and various cells under hypoxic conditions were about 25 μM to several hundred micromoles 3,12,16 . Compared with LHGFR 0N7C , LHGFR 0N3C with a higher K d for L-2-HG (29.33 ± 1.24 μM) is a more viable alternative for the detection of intracellular L-2-HG ( Supplementary Fig. 6a and Fig. 6e, Supplementary Table 1). Based on the in vivo response curves of LHGFR 0N3C , the basal L-2-HG concentration in HEK293FT cells under physiological conditions was determined to be 22.95 ± 11.22 μM (Fig. 6c), which is similar to results acquired using MSbased approaches after complicated sample handling and data analysis 2,3 . LHGFR 0N3C also has a low affinity with D-2-HG. The LOD of LHGFR 0N3C for D-2-HG was 872.59 μM and the K d value could not be determined accurately because of the relatively low Fig. 6 Monitoring L-2-HG fluctuations in human cells by LHGFR 0N3C . a Sequential images of mTFP (top), Venus (middle), and Venus/mTFP emission ratio (bottom, pseudocolored) of single HEK293FT cell expressing LHGFR 0N3C . 10 mM L-2-HG was added at time point zero (min). Elapsed time (in minutes) after the addition of L-2-HG is shown at the top of the images. Scale bar, 10 μm. b Time course of the emission ratio changes inside the region of interest (ROI) depicted from the top-left image of a. c Normalized dose-response curve of LHGFR 0N3C expressed in HEK293FT cells with increasing concentrations (10 nM to 10 mM) of L-2-HG. Cells were permeabilized with 10 μM digitonin. The emission ratio of non-permeabilized HEK293FT cells under physiological conditions is indicated with a black dash line. d Responses of LHGFR 0N3C expressed in HEK293FT cells to exogenously added 1 mM glutarate, L-2-HG, D-2-HG, and glucose. Cells were permeabilized with 10 μM digitonin. All data were normalized to the control (ratio in the absence of any tested compounds). e Identification of the function of L2HGDH in L-2-HG catabolism by LHGFR 0N3C . The emission ratio was measured after co-transfecting siRNA targeting L2HGDH and LHGFR 0N3C for 48 h. ***a: P = 0.0002, **b: P = 0.0058, ***c: P = 0.0002, *d: P = 0.0303. f Detection of hypoxia-induced L-2-HG accumulation by LHGFR 0N3C . Emission ratio changes were recorded after LHGFR 0N3C -expressing HEK293FT cells cultured in normoxia or hypoxia in the absence and presence of 5 mM dimethyl-2-ketoglutarate (DMαKG) for 24 h. The emission ratio was normalized to normoxic conditions without DMαKG. **a: P = 0.0029, **b: P = 0.0030, ***c: P < 0.0001, ***d: P = 0.0002, **e: P = 0.0016, **f: P = 0.0068, *g: P = 0.0126, *h: P = 0.0139. g Identification of the functions of LDHA and MDH2 in L-2-HG anabolism by LHGFR 0N3C . HEK293FT cells were cultured in the presence of 5 mM DMαKG. The emission ratio was normalized to the normoxic conditions treated with negative siRNA. **a: P = 0.0011, *b: P = 0.0400, ***c: P = 0.0002, ***d: P = 0.0004, **e: P = 0.0021, ***f: P = 0.0002. Inconsistent initial emission ratios were detected in HEK293FT cells under different conditions. The normalized emission ratios were thus used to monitor the changes of L-2-HG in different HEK293FT cells. All data shown are means ± s.d. (n = 3, 3, 4, 4, and 4 independent experiments for c, d, e, f, and g). *P < 0.05; **P < 0.01; ***P < 0.001; ns, no significant difference (P ≥ 0.05); one-way ANOVA test with Tukey's Multiple Comparison Test for e and g; two-tailed t-test for f. NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-23723-7 ARTICLE affinity ( Supplementary Fig. 9c and Supplementary Table 1). D-2-HG at a concentration of 240 μM, which is much higher than its basal intracellular concentrations, could barely affect the response of LHGFR 0N3C to L-2-HG (Supplementary Fig. 9e). IDH1-R132H mutation can lead to an extreme intracellular D-2-HG accumulation at millimolar levels 45 . The expression of IDH1-R132H resulted in a 4.9% increase in the emission ratio of LHGFR 0N3C under extreme conditions (Supplementary Fig. 19). Calibrated dose-response curves of LHGFR 0N3C for L-2-HG would be required for monitoring the fluctuations of L-2-HG in cells with IDH mutations.
The potential of LHGFR for real-time monitoring of fluctuations in intracellular L-2-HG concentrations was illustrated by using bacterial cells and HEK293FT cells. It was revealed that carbon starvation also induced temporary intracellular accumulation of L-2-HG in E. coli cells. CsiD and LhgO played indispensable roles in endogenous anabolism and catabolism of L-2-HG, respectively ( Fig. 5f and Supplementary Fig. 13f). In addition, it was identified that the growth of the strain containing lhgO mutation was inhibited when high levels of L-2-HG were present (Fig. 4j-l). Besides being a pathogenic metabolite inducing various cancers and L-2-hydroxyglutaric aciduria in humans [12][13][14][15]38,43 , L-2-HG would also be a toxic metabolite to bacterial cells. L-2-HG catabolizing enzymes, including L2HGDH in humans 9 , dL2HGDH in Drosophila 17 , and LhgO in P. putida 7,8 and E. coli 6,46 , might all exist as detoxification proteins of L-2-HG. The functions of L2HGDH in L-2-HG catabolism and LDHA and MDH2-mediated 2-KG reduction in hypoxia-induced L-2-HG production were also confirmed in HEK293FT cells using LHGFR as an indicator of L-2-HG. LHGFR can be added to the emerging list of metabolite sensors that have been established in mammalian cells, such as probes for ATP 47 , acetylcholine 48 , glycine 24 , and NAD + /NADH 49 . Several genetically encoded fluorescent metabolite sensors, like the NAD + /NADH probe SoNar, have been successfully applied in the screening of anti-tumor agents 49 . L-2-HG has been exploited as a potential therapeutic target in renal cancer 14 or a biomarker for cancer diagnosis and prognostic assessment 15,50 . The L-2-HG biosensors might also be utilized in the diagnosis and screening of anti-tumor agents for L-2-HGrelated cancer.
In summary, a regulatory protein LhgR, which is involved in L-2-HG catabolism and specifically responds to L-2-HG, was identified in P. putida W619. Two LhgR-based L-2-HG biosensors, LHGFR 0N3C and LHGFR 0N7C , with high sensitivity, specificity, and stability, were then constructed. The methods for quantitative estimation of L-2-HG concentrations in various biological samples and living cells by using L-2-HG biosensors were also established. We expect these LhgR-based L-2-HG biosensors to be of practical interest in future research on the metabolism of L-2-HG and the diagnosis and treatment of L-2-HG-related diseases.

Methods
Bacterial strains and culture conditions. The bacterial strains used in this study are listed in Supplementary Data 1. E. coli and its derivatives were cultured in Luria-Bertani (LB) broth at 37°C and 180 rpm. P. putida KT2440 and its derivatives were grown in minimal salt mediums (MSMs) containing different carbon sources at 30°C and 200 rpm. Antibiotics were used at the following concentrations: tetracycline at 30 μg mL −1 ; kanamycin at 50 μg mL −1 ; ampicillin at 100 μg mL −1 ; spectinomycin, at 50 μg mL −1 ; and chloramphenicol at 40 μg mL −1 .
Cloning of F2-lhgR-F1-lhgO and lhgO. All the plasmids and primers used in this study are listed in Supplementary Data 1 and Data 2, respectively. The gene segment F2-lhgR-F1-lhgO of P. putida W619 was synthesized by Tongyong Biosystem Co., Ltd (China). The lhgO gene of P. putida W619 was amplified and cloned into pME6032 plasmid using the restriction sites of EcoRI and KpnI to construct pME6032-lhgO, and the P tac promoter of pME6032 was replaced by the gene segment F2-lhgR-F1-lhgO using the restriction sites of SacI and BamHI to construct pME6032-F2-lhgR-F1-lhgO, then both recombinant plasmids were transferred into different derivatives of P. putida KT2440 by electroporation, respectively.
Construction of P. putida KT2440 and E. coli MG1655(DE3) mutants. Genes of P. putida KT2440 were deleted via allele exchange using the pK18 mobsacB system 51 . Briefly, the homologous arms upstream and downstream of the target gene were PCR amplified and fused together by recombinant PCR. The generated fusion fragment was cloned into the suicide plasmid pK18mobsacB. The resulting plasmid was transferred into P. putida KT2440 by electroporation. The single crossover cells and the second crossover cells were sequentially screened from LB plates containing 50 μg mL −1 kanamycin or 10% (wt/vol) sucrose, respectively.
To construct the E. coli MG1655(DE3) (ΔcsiD) mutant strain, the homologous arm upstream of the csiD gene, kanamycin resistance cassette, and the homologous arm downstream of the csiD gene were PCR amplified using the primers csiD-F1/ csiD-R1, csiD-F2/csiD-R2, and csiD-F3/csiD-R3, respectively. The PCR products were fused together by recombinant PCR, and the resulting fusion was transferred into E. coli MG1655(DE3) harboring pTKRed plasmid following isopropyl-β-D-1thiogalactopyranoside (IPTG) induction. The recombinant cells were selected on LB plates containing 50 μg mL −1 kanamycin at 37°C. The pCP20 plasmid was transferred into the selected cells, followed by a second screening on LB plates containing 40 μg mL −1 chloramphenicol at 30°C, then cultured in LB medium at 42°C to eliminate pCP20 plasmid. The lhgO mutant of E. coli MG1655(DE3) was generated by the same process. All mutants were verified by PCR and sequencing.
Enzymatic assay of LhgO. The derivatives of P. putida KT2440 were cultured in 50 mL MSMs with 5 g L −1 different compounds as carbon sources at 30°C and 200 rpm. The cells were harvested at mid-log phase, washed twice and resuspended in phosphate-buffered saline (PBS), then lysed by sonication on ice after the addition of 1 mM phenylmethylsulfonyl fluoride (PMSF). The supernatants obtained were used for further enzyme activity measurements after a centrifugation process (13,000 × g for 10 min at 4°C). Protein concentrations of the supernatants were determined using the Bradford protein assay kit (Sangon, China).
The activity of LhgO was assayed at 30°C by monitoring the reduction of dichlorophenol-indophenol (DCPIP) corresponding to the change of absorbance at 600 nm using a UV/visible spectrophotometer (Ultrospec 2100 pro, Amersham Biosciences, USA). The 800 μL reaction solution contained 0.1 mM L-2-HG, 0.05 mM DCPIP, 0.2 mM phenazine methosulfate (PMS) in PBS and 40 μL crude extracts. One unit of LhgO activity was defined as the amount of enzyme that catalyzed the reduction of 1 μmol of DCPIP per minute.
Expression, purification, and characterization of LhgR. To express and purify the recombinant LhgR, the lhgR gene was PCR amplified using the primer pair lhgR-F/lhgR-R, which contained BamHI and HindIII restriction sites, respectively, and then cloned into the pETDuet-1 plasmid to construct pETDuet-lhgR. The E. coli BL21(DE3) strains harboring pETDuet-lhgR plasmid were grown to an OD 600 of 0.6 in LB medium at 37°C, after which the cells were induced for 12 h with 1 mM IPTG at 16°C. The cells were harvested, washed twice, and resuspended in buffer A (20 mM sodium phosphate and 500 mM sodium chloride, pH 7.4), then lysed by sonication on ice after the addition of 1 mM PMSF and 10% (vol/vol) glycerol. The cell lysate was centrifuged at 13,000 × g for 40 min at 4°C, and the resultant supernatant was loaded onto a HisTrap HP column (5 mL) equilibrated with buffer A. The target protein was eluted with buffer B (20 mM sodium phosphate, 500 mM sodium chloride, and 500 mM imidazole, pH 7.4), analyzed by 12.5% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), and quantified by the Bradford protein assay kit (Sangon, China).
Electrophoretic mobility shift assays. Electrophoretic mobility shift assays (EMSAs) were carried out using the DNA fragment (F1 or F2) and purified LhgR. The DNA fragments were first amplified by primer pairs F1-F/F1-R and F2-F/F2-R, respectively. Then, either fragment at a concentration of 10 nM DNA was incubated with LhgR (0-160 nM) in 20 μL EMSA binding buffer (10 mM Tris-HCl [pH 7.4], 50 mM KCl, 0.5 mM EDTA, 10% [vol/vol] glycerol, and 1 mM dithiothreitol [DTT]). The binding reactions were carried out at 30°C for 30 min. Electrophoresis was performed on 6% native polyacrylamide gels at 4°C and 170 V (constant voltage) for about 45 min, followed by staining with SYBR green I (TaKaRa, China) and photographing. Analysis of the interaction between lhgO promoter region (F1) and LHGFR was performed using the same procedure.
To characterize the effector of LhgR, purified LhgR was first incubated with Llysine, 5-aminovalerate, glutarate, L-2-HG, D-2-HG, 2-KG, or succinate at 30°C for 15 min, followed by incubation with the added DNA fragments at 30°C for 30 min. The mixtures were subsequently subjected to electrophoresis.
DNase I footprinting. DNase I footprinting assays were performed using the 6carboxyfluorescein (FAM) labeled probe and purified LhgR. The DNA fragment F1 was PCR amplified using the primer pair F1-F/F1-R. The PCR products were cloned into the pEASY-Blunt plasmid using pEASY-Blunt Cloning Kit (TransGen, China). The FAM-labeled probes were PCR amplified using the resulting plasmid and the primer pair M13F-FAM/M13R. Then, 350 ng probes were incubated with 2 μg purified LhgR in a total volume of 40 μL for 30 min at 30°C. The DNase I digestion reaction was carried out by adding a total volume of 10 μL solution containing approximately 0.015 units of DNase I (Promega, USA) and 100 nmol CaCl 2 and further incubating for 1 min at 37°C, then stopped by adding a total volume of 140 μL stop solution containing 0.15% (wt/vol) SDS, 200 mM unbuffered sodium acetate, and 30 mM EDTA. The digested DNA fragments were first extracted with phenol-chloroform, then precipitated with ethanol and resuspended in 30 μL MiliQ water. The binding region of LhgR to DNA fragment F2 was analyzed using the same procedure.
Construction and purification of LHGFR. The genes encoding mTFP and Venus were synthesized by Tongyong Biosystem Co., Ltd (China). The mTFP gene and Venus gene were amplified and cloned into pETDuet-1 plasmid using the BamHI and SacI restriction sites, and SalI and NotI restriction sites, respectively. Then either the full-length lhgR gene, its truncated variants, or variants with artificial linkers were inserted between mTFP and Venus by the T5 exonuclease DNA assembly (TEDA) method 52 , respectively. The L-2-HG biosensor LHGFR and its derivatives were expressed and purified using the same procedure. For expression in HEK293FT cells, the codon-optimized LHGFR 0N3C , LHGFR 0N7C , or LHGFR 3N7C sequence was synthesized and cloned into pcDNA3.1 (+) plasmid behind a Kozak sequence, 5′-GCCACC-3′. To construct the plasmid for mitochondrial expression of LHGFR, the gene of mitochondrial targeting sequence (MLSLRQSIRFFKPATRTLCSSRYLL) was PCR amplified using primer pair Mito-F/Mito-R, and primer pair Mito-LHGFR-F/Mito-LHGFR-R was used to amplify the LHGFR fragment. The products were assembled using overlap PCR with Mito-F/Mito-LHGFR-R, and cloned into pcDNA3.1 (+) plasmid behind a Kozak sequence using the NheI and NotI restriction sites.
Characterization of LHGFR in vitro. Purified L-2-HG biosensors and different compounds were diluted by 50 mM Tris-HCl buffer (pH 7.4), mixed together in a black 96-well plate at a volume ratio of 3:1, and the fluorescence intensities were measured using an EnSight microplate reader (PerkinElmer, USA) with excitation at 430 nm, emission at 485 nm (mTFP) and 528 nm (Venus). The dose-response curves were fitted by OriginPro 2016 software (OriginLab) according to the following formula: where R, R min , and R max refer to the emission ratio of Venus to mTFP, ratio in the absence of L-2-HG, and ratio at saturation with L-2-HG, respectively. The [L-2-HG], K d , and p refer to the L-2-HG concentration, apparent dissociation constant, and Hill slope, respectively. Emission spectra were recorded at 430 nm excitation, in steps of 2 nm. Excitation spectra were recorded using emission at 550 nm, excitation from 380 to 535 nm in steps of 2 nm. Kinetics of L-2-HG binding by LHGFR were assessed by using SX-20 stoppedflow fluorimeter (Applied Photophysics, UK). Equal volumes of 1 μM purified LHGFR in 50 mM Tris-HCl buffer (pH 7.4) and L-2-HG-containing buffer (varying concentrations) were mixed, with data detected every four milliseconds at 430 nm excitation. Emission was detected by using a photomultiplier and a 515 nm longpass filter, and the detector voltage was set to 400 V. Apparent rate constants (k app = k on [L-2-HG] + k off ) determined by fitting the Venus fluorescence increase after L-2-HG addition with a single exponential equation were plotted against L-2-HG concentrations ([L-2-HG]). Effects of temperature on LHGFR were detected by analyzing the dose-response curves for L-2-HG at 25, 28, 31, 34, 37, 40, and 45°C, respectively. The reversibility of LHGFR was determined by recording the emission ratios every minute after the addition of 5 μM purified LhgO, the control test without the addition of L-2-HG or purified LhgO was run in parallel. The pH stability of LHGFR was determined using 50 mM Tris-HCl buffer with pH adjusted from 4.0 to 9.0. The background fluorescence without the addition of LHGFR was subtracted.
In order to test the ability of LHGFR for quantitative analysis L-2-HG in different biological samples, purified LHGFR was diluted by 50 mM Tris-HCl (pH 7.4), while varying concentrations of L-2-HG were added into the serum and urine of a healthy adult and bacteria culture medium and filtered through a 0.22 μm filter, respectively. The serum and urine were collected from a healthy adult (the first author of this article). The blood was collected by using promoting coagulating tubes and venous blood collection method, placed for 2 h at room temperature, and serum was prepared by centrifugation for 10 min at 2000 × g and 4°C. The urine collected from the experiment operator was diluted with an equal volume of 50 mM Tris-HCl buffer (pH 7.4). The processed serum and urine were filtered through a 0.22 μm filter and stored at −20°C until L-2-HG addition. The mixtures of purified LHGFR and L-2-HG in different biological samples were then incubated in a black 96-well plate at a volume ratio of 3:1, and the emission ratios were determined using an EnSight microplate reader. The background fluorescence without the addition of LHGFR was subtracted. The formulas for the quantitative analysis of L-2-HG in different biological samples by LHGFR 0N3C and LHGFR 0N7C were as follows: where R refers to the emission ratio of Venus to mTFP determined by LHGFR 0N7C .
Quantification of L-2-HG by HPLC and LC-MS/MS. When L-2-HG was used as a carbon source to cultivate P. putida KT2440, its consumption was analyzed by using high-performance liquid chromatography (HPLC) system (Agilent 1100 series, Agilent Technologies, USA) equipped with an Aminex HPX-87H column (300 × 7.8 mm, Bio-Rad, USA) and a RID detector at 55°C. The mobile phase was 0.1% formic acid at a flow rate of 0.4 mL min −1 .
To detect L-2-HG concentrations in various biological samples by liquid chromatography-tandem mass spectrometry (LC-MS/MS) system, the samples containing D,L-2-hydroxyglutarate disodium salt (2,3,3-D3) as internal standard (ITSD) were centrifuged at 13,000 × g for 15 min, then filtered through a 0.22 μm filter. The serum samples were mixed with methanol at a volume ratio of 1:3 and vortexed for 2 min to remove protein before centrifugation. Samples were analyzed by using a Thermo ultimate 3000 rapid separation liquid chromatography system (ThermoFisher, USA) coupled with a Bruker impact HD ESI-Q-TOF mass spectrometer (Bruker Daltonics, Germany) in negative ion mode and equipped with a Chirobiotic R column (250 × 4.6 mm, Supelco Analytical, USA). The mobile phase was prepared from (A) 0.1% triethylamine adjusted to pH 4.5 with acetic acid or (B) methanol. The quantification was conducted with an injection volume of 20 μL, a constant 5% gradient of (B) at a flow rate of 0.5 mL min −1 , and a total analysis time of 15 min.
Characterization of LHGFR in living bacteria. E. coli BL21(DE3) strains harboring either pETDuet-LHGFR 0N3C , pETDuet-LHGFR 0N7C , or pETDuet-LHGFR 3N7C were grown to an OD 600 of 0.6 in LB medium at 37°C, after which the cells were induced overnight in the presence of 1 mM IPTG at 16°C. The cultures were collected by centrifugation at 6000 × g for 5 min, washed three times, and resuspended to an OD 600 of 2.5 by carbon starvation medium (MSM containing no carbon source) or glucose medium (MSM containing 20 mM glucose).
To characterize the sensitivity and specificity of LHGFR expressed in E. coli BL21(DE3), 90 μL cell suspensions following 8 h carbon starvation were mixed with 10 μL increasing concentrations of L-2-HG or other compounds, and then added into a black 96-well plate (total 100 μL/per well), the fluorescence intensities were determined using an EnSight microplate reader (PerkinElmer, USA) and the following instrument settings: excitation at 430 nm, emission at 485 nm (mTFP) and 528 nm (Venus), time intervals of 5 min, the temperature at 37°C, and shake at 180 rpm. For carbon starvation experiments, cell suspensions in carbon starvation medium or glucose medium were added into a black 96-well plate (100 μL/per well), then the fluorescence intensities were monitored every ten minutes. In order to analyze functions of CsiD and LhgO in endogenous L-2-HG anabolism and catabolism during carbon starvation, pETDuet-LHGFR 0N3C , pETDuet-LHGFR 0N7C , or pETDuet-LHGFR 3N7C was transferred into E. coli MG1655(DE3) and its variants, and the assays were performed using the same procedure.
For live-cell imaging, HEK293FT cells were plated on a poly-L-lysine pre-coated 35 mm glass-bottom dish and transfected with LHGFR after 24 h. Live-cell imaging was carried out 48 h following transfection by using a Zeiss 880 confocal microscope equipped with an Objective C-Apochromat ×40/1.2 W autocorr M27 lens, a 458 nm argon laser, and a full-spectrum fluorescence detector. The emission of LHGFR expressed in HEK293FT cells was divided into a 463-495 nm channel (mTFP) and a 505-700 channel (Venus). Images were captured using 800 gain, 1024 × 1024 frame size, and 8 bit depth. The fluorescence intensities of each channel were analyzed in ZEN 3.1 software, and raw data were exported to Image-Pro Plus software for ratio image analysis. The Venus/mTFP emission ratio was calculated by dividing pixel-by-pixel a Venus image with a mTFP image. To realtime monitor the emission ratio of LHGFR in single living cells, single-cell with moderately fluorescent was imaged every 30 seconds, and 10 mM L-2-HG was added into the cultures when the ratio of two-channel fluorescence intensities was stable.
Characterization of LHGFR in HEK293FT cells. To characterize the sensitivity and specificity of LHGFR expressed in HEK293FT cells, cells were trypsinized 48 h following transfection and suspended in 1× Hank's balanced salt solution supplemented with 20 mM HEPES. Increasing concentrations of L-2-HG or other compounds including glutarate, D-2-HG, and glucose was mixed with the cell suspensions in a 96-well plate, respectively. Digitonin at a concentration of 10 μM was used to induce cell permeabilization and deplete intracellular L-2-HG for in vivo response curves construction. Then, the fluorescence intensities were determined by a SpectraMax i3 fluorescence plate reader (Molecular Devices, USA) with excitation at 430 nm and emission at 485 nm (mTFP) and 528 nm (Venus). Basal L-2-HG concentration in HEK293FT cells under physiological conditions was determined by substituting the emission ratios of non-permeabilized HEK293FT cells into the calibrated in vivo response curves.
For the detection of hypoxia-induced production of L-2-HG, LHGFR 0N3C or LHGFR 3N7C was expressed in HEK293FT cells and cultured sequentially under normoxic conditions for 24 h and hypoxic conditions for 24 h in the absence or presence of 5 mM dimethyl-2-ketoglutarate (DMαKG). The preparation of cell suspensions and the measurement of emission ratios were performed using the same procedure. The background fluorescence was subtracted at each emission wavelength.
siRNA experiments. The following Silencer Select siRNAs used in this study were purchased from ThermoFisher Scientific (USA): negative control (4390846), L2HGDH-a (s36692), L2HGDH-b (s36693), LDHA (s351), and MDH2 (s8622). To analyze L2HGDH functions in L-2-HG catabolism, siRNA targeting L2HGDH and pcDNA3.1 (+) plasmid encoding either LHGFR 0N3C , LHGFR 0N7C , or LHGFR 3N7C were mixed with Lipofectamine 3000 Transfection Reagent (ThermoFisher, USA) in Opti-MEM Reduced Serum Medium, and the lipoplexes prepared were transfected into HEK293FT cells according to the manufacturer's protocol. The fluorescence intensities were measured by a SpectraMax i3 fluorescence plate reader 48 h following transfection. Similarly, HEK293FT cells were transfected by LHGFR 0N3C or LHGFR 3N7C and siRNAs targeting LDHA and MDH2 separately or in combination. After transfection, cells were cultured sequentially under normoxic conditions for 24 h and hypoxic conditions for 24 h in the presence of 5 mM DMαKG, then the fluorescence intensities were measured. The cells cultured under the normoxic conditions in the presence of 5 mM DMαKG for 48 h were set as control. The background fluorescence was subtracted at each emission wavelength.