Probing coenzyme A homeostasis with semisynthetic biosensors

Abstract

Coenzyme A (CoA) is one of the central cofactors of metabolism, yet a method for measuring its concentration in living cells is missing. Here we introduce the first biosensor for measuring CoA levels in different organelles of mammalian cells. The semisynthetic biosensor is generated through the specific labeling of an engineered GFP-HaloTag fusion protein with a fluorescent ligand. Its readout is based on CoA-dependent changes in Förster resonance energy transfer efficiency between GFP and the fluorescent ligand. Using this biosensor, we probe the role of numerous proteins involved in CoA biosynthesis and transport in mammalian cells. On the basis of these studies, we propose a cellular map of CoA biosynthesis that suggests how pools of cytosolic and mitochondrial CoA are maintained.

Fig. 1. Synthesis and degradation of CoA.

a, Biosynthetic pathway of CoA. b, Hydrolysis of CoA by Nudt8.

Fig. 2. Design of CoA-Snifit.

a, Active site of mtPanK with bound triazole ligand (Protein Data Bank ID: 4BFU). b, Active site of ecPanK with bound CoA (Protein Data Bank ID: 1ESM); only key residues of ecPanK are depicted as green sticks for clarity. c, Chemical structures of TAZ inhibitor and probes used in this study. d, Design principle of CoA-Snifit. CoA (depicted as yellow ball) and the tethered TAZ derivative (TAZ depicted as blue ball) compete for binding to ecPanK, shifting the sensor either to a closed or an open conformation, respectively. FRET efficiency in the closed conformation is higher than that in the open conformation.

Fig. 3. Characterization of CoA-Snifit.

a, Fluorescence emission spectra of CoA-Snifit^G41 in response to different CoA concentrations (12.8 nM to 1 mM) in PBS. b, FRET ratio (F₅₁₀/F_580 nm) of CoA-Snifit^G41 (mean ± s.d., n = 3 independent replicates) in response to different concentrations of analyte. CoA (blue), AcCoA (green), dPCoA (orange), and ATP (red). a.u., arbitrary units. c, FRET ratio (F₅₁₀/F_580 nm) of different CoA-Snifits (mean ± s.d., n = 3 independent replicates) in response to CoA concentrations. d, The FRET ratio (F₅₁₀/F_580 nm) of CoA-Snifit^G41 (mean ± s.d., n = 3 independent replicates) in the presence of 1 mM CoA, ATP, Pan, PPan, PPanSH, Mg²⁺, Ca²⁺, 100 μM ADP, GTP, NAD, NADH, NADP, NADPH, 20 μM propionyl-CoA, butyryl-CoA, hexanoyl-CoA, malonyl-CoA, succinyl-CoA, 3-hydroxy-3-methylglutaryl (HMG-CoA), 1 μM lauroyl-CoA, myristoyl-CoA, and oleoyl-CoA in PBS. e, Representative fluorescence images (n = 3 independent samples) of HEK293 cells stably expressing cytosolic CoA-Snifit^V97T (green) co-stained with Hoechst 33342 (blue). Scale bar, 20 μm. f–h, Representative fluorescence images (n = 3 independent samples) of HEK293 cells stably expressing mitochondrial CoA-Snifit^G41S (green) co-stained with Hoechst 33342 (blue) and MitoTracker Red CMXRos (red). f, Emission of GFP. g, Emission of MitoTracker Red CMXRos. h, Merged image of emission of Hoechst 33342, GFP, and MitoTracker Red CMXRos. Scale bar, 20 μm. i, Representative in-gel fluorescence (n = 2 independent samples) of the cell lysate (n = 3 independent experiments) from HEK293 cells expressing sensor protein of cytosolic CoA-Snifit^V97T and mitochondrial CoA-Snifit^G41S labeled with Halo-SiR, respectively. j, Representative western blot analysis (n = 2 independent experiments) of the cell lysate from unlabeled HEK293 cells expressing CoA-Snifit^V97T and CoA-Snifit^G41S. The housekeeping gene, GAPDH, was used as the loading control. Source data

Fig. 4. Regulation of CoA in HEK293 cells.

a,b, Fluorescence ratio changes of cytosolic CoA-Snifit^V97T (a) and mitochondrial CoA-Snifit^G41S (b) under the gene knockdown and overexpression conditions. Cells transfected with non-targeting esiRNA for firefly luciferase (siFLUC) were used as the negative control for gene knockdown. Cells, transfected with empty vector, were used as the negative control for gene overexpression. Ratio changes are percentage of normalized FRET values to control, (FRET − FRET_control)/FRET_control. They are presented as mean ± s.e.m., n = 6 FOVs over 4 independent samples with >50 cells per FOV. c, Chemical structures of HoPan and PanK activator PZ-2891. d,e, Regulation of PanKs by HoPan and effect on cytosolic (d) and mitochondrial (e) normalized FRET ratio. Cells transfected with empty vector were used as the negative control. The data are presented as mean ± s.e.m., n = 6 FOVs over 4 independent samples with >50 cells per FOV. f,g, Activation of PanKs by PZ-2891 and effect on cytosolic (f) and mitochondrial (g) normalized FRET ratio. Cells, transfected with empty vector and treated with 0.05% (v/v) DMSO, were used as the negative control. The box plots represent the s.e.m. at the lower and upper box limits and the mean as the middle bar. n = 5 or 6 FOVs over 4 independent samples with >50 cells per FOV. The whiskers extend to ±1.5× the interquartile range beyond the limits of the boxes, respectively. The precise n and P values are listed in the Supplementary Data. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001. NS, not significant (P > 0.05). Two-tailed unpaired t-test. Source data

Fig. 5. Role of transporters in CoA homeostasis.

a,b, Effect of overexpression or knockdown of selected transporters on cytosolic (a) and mitochondrial (b) normalized FRET ratio. Cells transfected with non-targeting esiRNA for firefly luciferase (siFLUC) were used as the negative control for gene knockdown. Cells transfected with empty vector were used as the negative control for gene overexpression. The box plots represent the s.e.m. at the lower and upper box limits and the mean as the middle bar. n = 5 or 6 FOVs over 4 independent samples with >50 cells per FOV. The whiskers extend to ±1.5× the interquartile range beyond the limits of the boxes, respectively. The precise n and P values are listed in the Supplementary Data. ***P ≤ 0.001. Two-tailed unpaired t-test. c, Proposed biosynthetic map of CoA biosynthesis in human cells. Source data

Extended Data Fig. 1. Purity analysis and characterization of ecPanKs and hPanKs.

a. Representative Coomassie staining after protein SDS-PAGE (n = 3 independent experiments). Lane 1: protein marker, lane 2: CoA-Snifit^V97T, lane 3: CoA-Snifit^G41, lane 4: CoA-Snifit^G41S, lane 5: CoA-Snifit^G41N, lane 6: ecPanK^WT, lane 7: ecPanK^F252Y, lane 8: ecPanK^L277F, lane 9: ecPanK^L277Y, lane 10: ecPanK^L277W, lane 11: ecPanK^F252Y,L277W, lane 12: ecPanK^{D127A,F252Y,L277W}. b. Representative in-gel fluorescence scan after SDS-PAGE (n = 3 independent experiments) of Halo-SiR labeled sensor proteins, Lane 1: protein marker, lane 2: CoA-Snifit^V97T, lane 3: CoA-Snifit^G41, lane 4: CoA-Snifit^G41S, lane 5: CoA-Snifit^G41N. c. Representative Coomassie staining (n = 3, independent experiments) of recombinant hPanKs expressed in E. coli. Lane 1: protein marker, lane 2: hPanK1, lane 3: hPanK2, lane 4: hPanK3. d. Fluorescence polarization responses (FP, mean ± s.d., n = 3 independent replicates) of 50 nM TMR-TAZ as a function of ecPanK^WT concentration in PBS. e. Fluorescence polarization responses (mean ± s.d., n = 3 independent replicates) of 50 nM TMR-TAZ and 20 μM ecPanK^WT in the presences of different CoA or AcCoA concentrations. The data (mean ± s.d., n = 3 independent replicates) are fitted to a single-site binding isotherm (dashed line). f-h. Activity assay for hPanK1 (f), hPanK2 (g), and hPanK3 (h) in activity buffer. The absorption (mean ± s.d., n = 3 independent replicates) are recorded at 340 nm at room temperature over 30 min. The titration results showed that all three recombinant hPanK isoforms were active. i. Fluorescence polarization responses (mean ± s.d., n = 3 independent replicates) of 50 nM TMR-TAZ as a function of ecPanK^WT, hPanK1, hPanK2, and hPanK3 concentration in PBS. TMR-TAZ only binds ecPanK^WT but not the three mammalian PanKs. Note that only the catalytic core domains of hPanK1 and hPanK2 were expressed according to the previous report. Source data

Extended Data Fig. 2. In vitro characterization of ecPanK variants and CoA-Snifits.

a-f. Fluorescence polarization responses of 50 nM TMR-TAZ and 20 μM ecPanK^F252Y (a), 20 μM ecPanK^L277F (b), 20 μM ecPanK^L277Y (c), 20 μM ecPanK^L277W (d), 20 μM ecPanK^F252Y,L277W (e), and 100 μM ecPanK^{D127A,F252Y,L277W} (f) as a function of CoA or AcCoA concentration. The data (mean ± s.d., n = 3 independent replicates) are fitted to a single-site binding isotherm (dashed line). g. Enzymatic assay of ecPanK proteins. Activity of ecPanK^WT and ecPanK^D127A is measured in activity buffer. The absorption was recorded at 340 nm at room temperature over 500 min. The dead mutant ecPanK^D127A showed neglectable activity compared with the wild type ecPanK^WT. h. Emission spectra of CoA-Snifit^G41 upon the titration of 20 μM CoA in the absence or presence of 500 μM di-2-pyridyl thiocarbonate (DPT) in PBS. i.The ratio (GFP/MaP) responses (mean ± s.d., n = 3 independent replicates) of CoA-Snifit^G41 upon the titration of 20 μM CoA in the absence or presence of 500 μM DPT. Source data

Extended Data Fig. 3. Effect of ATP and ADP on the responses of CoA-Snifits.

a-c. the ratio (GFP/MaP) responses (mean ± s.d., n = 3 independent replicates) as a function of different CoA concentrations in the presence of different total ATP and ADP concentrations with [ATP]/[ADP] = 9:1. d-f. the ratio (GFP/MaP) responses (mean ± s.d., n = 3 independent replicates) as a function of different CoA concentrations in the presence of different total ATP and ADP concentrations with ratio of [ATP]/[ADP] = 99:1. g-i. c₅₀ (mean ± s.d., n = 3 independent replicates) as a function of different total ATP and ADP concentrations. Source data

Extended Data Fig. 4. Specificity test of CoA-Snifits.

a. Emission spectra of CoA-Snifit^V97T upon the titration with increasing concentrations of CoA in PBS. b. The ratio (GFP/MaP) responses (mean ± s.d., n = 3 independent replicates) of 100 nM sensor as a function of different CoA (black), AcCoA (red), dPCoA (blue), and ATP (magenta) concentrations in PBS. c. The ratio (GFP/MAP) responses (mean ± s.d., n = 3 independent replicates) of 100 nM sensor in the presence of 1 mM CoA, 100 μM ADP, 100 μM GTP, 1 mM ATP, Pan, PPan, PPanSH, Mg²⁺, Ca²⁺, 100 μM NAD, NADH, NADP, NADPH, 10 μM propionyl-CoA, butyryl-CoA, hexanoyl-CoA, malonyl-CoA, succinoyl-CoA, HMG-CoA, and 1 μM lauroyl-CoA, myristoyl-CoA, oleoyl-CoA in PBS. d. The ratio (GFP/MAP) responses (mean ± s.d., n = 3 independent replicates) of 100 nM sensor in PBS containing 1 mM ATP with 1 mM CoA, 10 μM propionyl-CoA, butyryl-CoA, hexanoyl-CoA, malonyl-CoA, succinoyl-CoA, or HMG-CoA. e. Emission spectra of CoA-Snifit^G41S upon the titration with increasing concentrations of CoA in PBS. f. The ratio (GFP/MaP) responses (mean ± s.d., n = 3 independent replicates) of 100 nM sensor as a function of different CoA (black), AcCoA (red), dPCoA (blue), and ATP (magenta) concentrations in PBS. g. The ratio (GFP/MAP) responses (mean ± s.d., n = 3 independent replicates) of 100 nM sensor in the presence of 5 mM CoA, 100 μM ADP, 100 μM GTP, 1 mM ATP, Pan, PPan, PPanSH, Mg²⁺, Ca²⁺, 100 μM NAD, NADH, NADP, NADPH, 100 μM propionyl-CoA, butyryl-CoA, hexanoyl-CoA, malonyl-CoA, succinoyl-CoA, HMG-CoA, and 10 μM lauroyl-CoA, myristoyl-CoA, oleoyl-CoA in PBS. h. Emission spectra of CoA-Snifit^G41N upon the titration with increasing concentrations of CoA in PBS. i. The ratio (GFP/MaP) responses (mean ± s.d., n = 3 independent replicates) of 100 nM sensor as a function of different CoA (black), AcCoA (red), and dPCoA (blue) concentrations in PBS. j. The ratio (GFP/MAP) responses (mean ± s.d., n = 3 independent replicates) of 100 nM sensor in the presence of 10 mM CoA, 100 μM ADP,100 μM GTP, 1 mM ATP, Pan, PPan, PPanSH, Mg²⁺, Ca²⁺, 100 μM NAD, NADH, NADP, NADPH, 100 μM propionyl-CoA, butyryl-CoA, hexanoyl-CoA, malonyl-CoA, succinyl-CoA, HMG-CoA, and 10 μM lauroyl-CoA, myristoyl-CoA, oleoyl-CoA in PBS. Source data

Extended Data Fig. 5. Effect of pH on the responses of CoA-Snifits.

a-c. the ratio (GFP/MaP) responses (mean ± s.d., n = 3 independent replicates) of 100 nM CoA-Snifit^V97T (a), CoA-Snifit^G41 (b), and CoA-Snifit^G41S (c) as a function of CoA concentrations at the indicated pH. d-f. The c₅₀ (mean ± s.d., n = 3 independent replicates) of CoA-Snifit^V97T (d), CoA-Snifit^G41 (e), CoA-Snifit^G41S (f) as a function of pH. Source data

Extended Data Fig. 6. Expression and subcellular localization of the sensor proteins.

a. Representative western blot analysis (n = 2 independent experiments) of the lysates of HEK293 cells stably expressing apo-CoA-Snifits. Small impurity bands (< 15%) were observed, possibly due to the degradation of the sensor protein. Lane 1: biotinylated ladder, lane 2: cyto-CoA-Snifit^G41, lane 3: cyto-CoA-Snifit^G41N, lane 4: cyto-CoA-Snifit^V97T, lane 5: mito-CoA-Snifit^G41N, lane 6: mito-CoA-Snifit^G41, lane 7: mito-CoA-Snifit^G41S. b-j. Testing the subcellular localization of unlabeled sensor protein of cytosolic CoA-Snifit^V97T (b-e) and mitochondrial CoA-Snifit^G41S (f-j) in living HEK293 cells. The data are shown as representative images of n = 3 independent samples. b and f. bright field. c and g. emission of Hoechst 33342. d and h. emission of GFP. i. emission of MitoTracker Red CMXRos. e. Merged image of c and d, Scale bar: 20 μm. j. Merged image of g, h, and i, Scale bar: 20 μm. Source data

Extended Data Fig. 7. Labeling specificity of Halo-MaP-TAZ in living cells.

a-e. Evaluating the labeling background of the ligand. The cells were labeled with 1 μM Halo-MaP-TAZ for 12 h and the emission was collected with λ_ex = 535 nm and λ_em = 560 − 620 nm. a. HEK293 cells without labeling. Scale bar: 40 μm. b. HEK293 cells incubated with 1 μM Halo-MaP-TAZ for 12 h. Scale bar: 40 μm. c. HEK293 cells stably expressing cytosolic sensor protein of CoA-Snifit^V97T labeled with 1 μM Halo-MaP-TAZ for 12 h. Scale bar: 40 μm. d. HEK293 cells stably expressing mitochondrial sensor protein of CoA-Snifit^G41S labeled with Halo-MaP-TAZ for 12 h. Scale bar: 40 μm. e. HEK293 cells expressing a nuclear Halo-SNAP-NLS fusion protein were labeled with 1 μM Halo-MaP-TAZ for 12 h and the emission was collected with λ_ex = 535 nm, λ_em = 560 − 620 nm. Scale bar: 40 μm. The average fluorescence intensity ratio between the nuclear and cytosolic signal (F_nuc/F_cyt) is 90, indicating very low background labeling of the probe. The data were shown as representative images of n = 3 independent samples. f. Representative ratio images of the cytosolic sensor protein of CoA-Snifit^V97T labeled with Halo-MaP-TAZ for 0.5, 1, 2, 4, 6, 8, 10, 12 h. Scale bar: 40 μm, ratio bar: 20 − 80. g. The normalized ratio values were calculated from f and shown as mean ± s.d., n = 4 field of views (FOVs) over 4 independent samples with > 50 cells per FOV. The normalized FRET ratio values were calculated for each FOV. h. Representative ratio images of the mitochondrial sensor protein of CoA-Snifit^G41S labeled with Halo-MaP-TAZ for 0.5, 1, 2, 4, 6, 8, 10, 12 h. Scale bar: 40 μm, ratio bar: 20 − 80. i. The normalized ratio values were calculated from h and shown as mean ± s.d., n = 6 field of views (FOVs) over 6 independent samples with > 50 cells per FOV. Source data

Extended Data Fig. 8. Subcellular localization of overexpressed proteins.

Overexpression of GFP fusion proteins in HEK293 cells which were co-stained with 1 μg/mL Hoechst 33342 and 100 nM MitoTracker Red CMXRos. The data are shown as representative images of n = 3 independent samples. a. Bright field. b. Emission of Hoechst 33342. c. Emission of GFP. d. Emission of MitoTracker Red CMXRos. e. Merged image of b, c, and d. Scale bar: 20 μm for PanK1, PanK2, PanK3, and CG4241. Scale bar: 10 μm for Nudt8. Source data

Extended Data Fig. 9. Influence of gene overexpression or knockdown of proteins involved in CoA biosynthesis on cytosolic (a and b) and mitochondrial (c and d) FRET ratio.

Cells, transfected with non-targeting esiRNA for firefly luciferase (siFLUC), were used as the negative control for gene knockdown. Cells, transfected with empty vector, were used as the negative control for gene overexpression. The data are presented as mean ± s.e.m., n = 6 field of views (FOVs) over 4 independent samples with > 50 cells per FOV. **P ≤ 0.01, ***P ≤ 0.001, NS p > 0.05, two-tailed unpaired t-test. p values for a (left to right): p < 0.0001 (PANK1), p = 0.0014 (PANK2), p < 0.0001 (PANK3), p = 0.32 (PPCS), p = 0.81 (PPCDC), p = 0.16 (COASY), p = 0.81 (DCAKD), p = 0.69 (ACLY). p values for b (left to right): p < 0.004 (siPANK1), p = 0.40 (siPANK2), p = 0.33 (siPANK3), p = 0.77 (siPPCS), p = 0.63 (siPPCDC), p = 0.12 (siCOASY), p = 0.014 (siDCAKD), p < 0.0001 (siACLY). p values for c (left to right): p < 0.0001 (PANK1), p = 0.0011 (PANK2), p < 0.0001 (PANK3), p = 0.06 (PPCS), p = 0.074 (PPCDC), p = 0.0018 (COASY), p = 0.29 (DCAKD), p = 0.68 (NUDT8). p values for d (left to right): p > 0.05 (siPANK1), p = 0.22 (siPANK2), p = 0.98 (siPANK3), p = 0.06 (siPPCS), p = 0.07 (siPPCDC), p = 0.62 (siCOASY), p = 0.74 (siDCAKD), p < 0.0001 (siNUDT8). Source data

Extended Data Fig. 10. FLIM measurement of CoA in cells.

a-c. Normalized FRET ratios of cytosolic CoA-Snifits in hemolysin permeabilized HEK293 cells. The cells were treated with 1 μg/mL hemolysin in DPBS for 40 min and then incubated with 20 μM CoA for CoA-Snifits^V97T (a), 40 μM CoA for CoA-Snifits^G41 (b), and 100 μM CoA for CoA-Snifits^G41S (c) in DPBS (1 mM Mg²⁺, 1 mM ATP, and 0.2 mg/mL BSA) for 30–45 min. The results showed that the normalized ratio signal gradually increased but did not reach an equilibrium within 30 min, making it difficult to perform calibration titration curves in those permeabilized cells. The data were shown as mean ± s.d., n = 3 field of views (FOVs) over 3 independent samples with > 50 cells per FOV. The normalized FRET ratio values were calculated for each FOV. d. The HEK293 cells stably expressing cytosolic sensor proteins of CoA-Snifits were labeled with Halo-MaP-TAZ for 12 h and were lysed by two flash freeze–thaw cycles in liquid nitrogen. The cell lysate was further cleared by centrifugation at 20,000 g and 4 ^oC for 20 min. The concentration of the sensor was evaluated by measuring the absorbance at 560 nm ((MaP)_560nm = 85000 M⁻¹cm⁻¹). The lysate was diluted in PBS (supplemented with 50 mM HEPES and 0.2 mg/mL BSA) reaching a sensor concentration of 100 nM. The fluorescent titration experiments were performed according to the method described in the main text. Fluorescence intensity ratio (F₅₁₀/F₅₈₀ nm) responses (mean ± s.d., n = 3 independent samples) of the cell lysate of the HEK293 cells containing CoA-Snifit^V97T (black), CoA-Snifit^G41 (red), and CoA-Snifit^G41S (blue) as a function of CoA concentration. e-f. FLIM measurements of CoA-Snifits upon titration with different concentrations of CoA. e. Normalized photon counts of CoA-Snifit^G41 upon the titration with CoA (32 nM to 2.5 mM) in DPBS (supplemented with 25 mM HEPES, 1 mM Mg²⁺, 1 mM ATP, pH 7.4) as a representative example for the CoA-Snifits (n = 3 independent replicates). f. FRET efficiency calculated by fluorescence lifetime responses of CoA-Snifits as a function of CoA concentration (mean ± s.d., n = 3 independent replicates). Source data