Repurposing conformational changes in ANL superfamily enzymes to rapidly generate biosensors for organic and amino acids

Organic and amino acids are key metabolites and industrial products whose precise measurement in complex samples is important for diagnosis, monitoring, and bioprocessing. Conventional quantification relies on HPLC, GC, and MS, which are accurate but low-throughput and require specialized equipment. Transcription factor-based biosensors are common but limited by slow transcription/translation kinetics and compatibility issues across organisms. FRET-based biosensors improve response speed but often have narrow dynamic ranges for in vitro screening. cpFP-based sensors provide wider dynamic ranges and spectral simplicity but rapid conversion of ligand-binding domains (LBDs) into functional sensors is challenging and typically limited to non-catalytic LBDs. The authors hypothesize that enzymes with large ligand-induced conformational changes, particularly ANL superfamily ligases with hinge-mediated domain motions, can be repurposed as LBDs by inserting cpEGFP into hinge regions to transduce conformational changes into fluorescence signals, enabling rapid creation of small-molecule biosensors for organic and amino acids.

• Previous genetically encoded biosensors have mainly used non-catalytic LBDs (transcription factors, periplasmic binding proteins, GPCRs), each with limitations in response time, dynamic range, or application scope.
• FRET-based studies (e.g., on NRPS A–PCP di-domains) demonstrated feasibility to monitor conformational dynamics but required chemical labeling and were not genetically encoded, limiting generalizability among low-identity ANL members.
• ANL superfamily enzymes share architecture with large domain motions across hinge A (between N- and C-terminal subunits) and, in NRPSs, hinge B (between C-terminal subunit and PCP), suggesting potential for cpFP insertion to report conformational changes.
• Specificity-conferring codes of NRPS A domains are well-characterized, enabling rational reprogramming of substrate specificity, which can be leveraged to tune biosensor ligand specificity.
• Prior work indicates many NRPS A domains recognize hundreds of acids, offering a large reservoir of potential LBDs.

Design and build of GECFINDERs:

• Identify suitable ANL superfamily members (e.g., Nt4CL2; NRPS A domains such as GrsA-PheA, PpsA-GluA) and locate hinge A and hinge B using crystal structures or homology modeling.
• Construct biosensors by inserting cpEGFP into hinge A or hinge B with two-amino-acid random linkers (XX-cpEGFP-XX) to convert ligand-induced conformational changes into fluorescence changes. Initial attempts with fixed linkers (LE-cpEGFP-TR) guided feasibility.
• Generate mutant libraries: amplify LBD, clone into pET28a(+), design primers with cpEGFP homology arms and degenerate linker codons, assemble with cpEGFP by homologous recombination, DpnI digest, transform into E. coli BAPI.

High-throughput screening and characterization:

• Express libraries in 96-well plates (induction with 0.4 mM IPTG at 16 °C for 48 h), lyse by freeze–thaw, and assay supernatants in buffer with Mg2+ (2.5 mM) and ATP (1 mM).
• Measure fluorescence (e.g., 400/510 and 480/510 nm) before and after adding ligand (typically 1 mM), monitoring for 15 min. Select variants with (F−F0)/F0 ≥ 0.1 or ≤ −0.3, re-screen, sequence, then purify for detailed study.
• Determine dose–response curves (0.75 µM protein, 1 mM ATP, 2.5 mM MgCl2; 100 mM Tris pH 7.5) over 15 min, fit with asymmetric logistic models. Record excitation/emission spectra, pH dependence (6.5–8.5), and quantum yields.

Substrate specificity engineering:

• Apply specificity-conferring code swaps and structure-based redesign to reprogram the A-domain binding pocket (e.g., GrsA-PheA mutations T278M/A301G, I277L/T278L/A301G, S447N, W239S) to switch optimal ligands to Leu or Tyr; adopt literature mutations to generate S-β-phenylalanine specificity. Construct corresponding GECFINDERS and characterize EC50, ΔF/F0, selectivity.

Mechanistic assays of catalysis post-insertion:

• LC-MS for Nt4CL2 and GECFINDER-4CA to detect 4-coumaroyl-AMP (intermediate) and 4-coumaroyl-CoA (product) with/without CoA.
• HPLC to quantify AMP production in GrsA-PheA and GECFINDER-Phe3 reactions; ESI-TOF-MS to detect PCP thiolation mass shift (+147 Da) to assess the second thioesterification step.

In vitro quantitative assays in complex samples:

• Use GECFINDER-Glu to quantify glutamate in fermentation broths (compare to SBA-40D analyzer). Use GECFINDER-Phe3 to quantify phenylalanine in human serum (compare to HPLC). Set standard curves (e.g., 5 µL serum in 200 µL reaction; read at 460/510 nm after 5 min).

Droplet-based microfluidic HTS with FADS:

• Re-engineer GECFINDER-Phe3 affinity (e.g., F234A) to match higher phenylalanine ranges for screening industrial strains.
• Generate a C. glutamicum library via CRISPR-guided base editing of RBSs for pheAfbr and arofbr; perform droplet co-encapsulation with sensor (12 µM), ATP (2 mM), Mg2+ (5 mM); incubate (∼12–16 h); sort top 0.4% fluorescent droplets at 500 V.
• Validate enrichment via colony PCR; pick and re-screen clones in 96-well plates by GECFINDER-Phe3 assay; reconstruct hits to confirm RBS strength effects; assess correlation between arofbr expression (GFP reporter) and phenylalanine production; measure titers in microplates and shake flasks.

• Rapid conversion of enzymes to cpEGFP-based biosensors: 11 GECFINDERs constructed targeting 4-coumaric acid, ATP, phenylalanine, glutamic acid, proline, isoleucine, histidine, benzoic acid; plus engineered variants for leucine, tyrosine, and S-β-phenylalanine.
• Performance examples (mean ± SD):
  • GECFINDER-4CA (Nt4CL2, hinge A): dynamic range ΔF/F0 = 0.83 ± 0.03 over 0.25–25 µM; ratiometric excitation behavior.
  • GECFINDER-Phe3 (GrsA-PheA-PCP, hinge B): EC50 = 1.91 ± 0.08 µM; ΔF/F0 = 3.39 ± 0.11; high specificity with weak response to Tyr (EC50 224.89 ± 17.87 µM) and Trp.
  • GECFINDER-Glu (PpsA-GluA-PCP): EC50 = 42.92 ± 3.93 µM; ΔF/F0 = 0.95 ± 0.02.
  • GECFINDER-Ile3: EC50 = 186.46 ± 61.05 µM; ΔF/F0 = 0.44 ± 0.01.
  • GECFINDER-His2: EC50 = 13.25 ± 4.26 µM; ΔF/F0 = 0.43 ± 0.02.
  • Engineered specificity:
    • GECFINDER-Leu3: EC50 (Leu) = 200.38 ± 7.76 µM; ΔF/F0 = 2.33 ± 0.05; reduced Phe affinity (EC50 392.24 ± 51.79 µM).
    • GECFINDER-Tyr: EC50 (Tyr) = 89.81 ± 6.71 µM; ΔF/F0 = 0.74 ± 0.02; Phe EC50 = 258.00 ± 37.31 µM.
    • GECFINDER-SβF: EC50 (S-β-Phe) = 7.40 ± 0.45 µM; ΔF/F0 = 2.26 ± 0.02; high stereoselectivity vs R-β-Phe (EC50 54.61 ± 7.66 µM); Phe EC50 increased to 234.25 ± 23.69 µM (123-fold lower affinity than Phe3).
• Mechanism: cpEGFP insertion preserves the first adenylation step (intermediate AMP derivatives detected) but significantly impairs the second thioesterification step (reduced thioester products; no PCP thiolation mass shift), preventing net substrate consumption and allowing accurate measurement with ATP supplementation.
• Specificity: GECFINDERs showed high ligand specificity across 20 amino acids; minimal off-target responses.
• Stability: GECFINDER-Phe3 maintained similar ΔF/F0 and EC50 from pH 6.5–7.5; slight ΔF/F0 decrease at pH 8.5.
• In vitro quantification:
  • GECFINDER-Glu measurements in fermentation broth matched SBA-40D analyzer, enabling 96-sample readouts in ~5 min versus ~100 min sequential SBA analysis.
  • GECFINDER-Phe3 phenylalanine measurements in human serum agreed with HPLC results; assay time ~5 min without derivatization.
• Droplet microfluidics HTS:
  • Proof-of-concept enrichment: sorting top 0.4% droplets yielded 16.7% positives from a 1% starting frequency (32× enrichment).
  • Library screening identified strains exceeding positive control yields by >5-fold; top strain (1-2 F10) reached 2.6 mM phenylalanine in microplate fermentation (24 h) and 1.19 g/L in 72-h shake flask.
  • Positive correlation between arofbr expression level (RBS strength reporter) and phenylalanine production, identifying DAHPS as a rate-limiting step.

The study addresses the challenge of rapidly creating genetically encoded small-molecule biosensors by exploiting large ligand-induced conformational changes in ANL enzymes. By inserting cpEGFP into flexible hinge regions, the authors converted catalytic domain motions into robust fluorescence signals, overcoming limitations of TF- and FRET-based sensors in speed, dynamic range, and portability. Mechanistic analyses confirmed that only the first adenylation step proceeds while thioesterification is inhibited, ensuring that substrate levels are not significantly perturbed during measurement and enabling accurate quantification with ATP present. The platform demonstrated high specificity and tunable affinity via binding pocket engineering, enabling detection across diverse organic and amino acids and extension to non-natural ligands. Practical relevance was shown in two in vitro applications: rapid, high-throughput quantification in complex samples (matching SBA and HPLC) and compatibility with FADS for screening industrial microbial libraries, where it enabled strong enrichment and discovery of high-producing strains. Given the broad substrate scope of NRPS A domains and related ANL enzymes, the method is poised to scale to hundreds of analytes and be adapted to additional enzymes exhibiting hinge-mediated or allosteric motions.

This work establishes a generalizable strategy to repurpose ANL superfamily enzymes into cpEGFP-based GECFINDER biosensors for organic and amino acids. The authors developed 11 sensors (including engineered variants) with high specificity, tunable affinity, and practical dynamic ranges; elucidated a working mechanism that preserves adenylation while impairing thioesterification; and demonstrated rapid in vitro quantification compatible with POCT and high-throughput FADS-based strain screening. Future directions include: expanding LBD sources across ANL and other enzymes with large domain movements; engineering for reduced ATP dependence and improved signal-to-noise; leveraging structural data and machine learning to design optimal linkers and insertion sites; and translating to intracellular imaging to monitor metabolic dynamics in living cells.

• Dependence on ATP and Mg2+: although intracellular ATP is typically sufficient and ATP consumption is slow due to impaired thioesterification, ATP must be supplied in vitro; ATP effects on baseline fluorescence vary slightly among variants.
• Signal-to-noise varies across targets: some variants (e.g., Ile, His, Pro) show modest ΔF/F0 and may require further engineering for demanding applications.
• Insertion site/linker design rules are not yet predictive; current approach relies on library screening around hinge regions.
• cpEGFP pH sensitivity requires buffered conditions (ΔF/F0 stable ~pH 6.5–7.5; slight reduction at pH 8.5).
• Dynamic range constraints may necessitate affinity tuning (e.g., for high-titer industrial screens).
• Demonstrations are primarily in vitro; intracellular applications remain to be validated and optimized.