An improved pathway for autonomous bioluminescence imaging in eukaryotes

Abstract

The discovery of the bioluminescence pathway in the fungus Neonothopanus nambi enabled engineering of eukaryotes with self-sustained luminescence. However, the brightness of luminescence in heterologous hosts was limited by performance of the native fungal enzymes. Here we report optimized versions of the pathway that enhance bioluminescence by one to two orders of magnitude in plant, fungal and mammalian hosts, and enable longitudinal video-rate imaging.

Fig. 1. Optimization of the fungal bioluminescence pathway in cell culture systems.

a, Biochemical reactions of the fungal bioluminescence pathway are catalyzed by hispidin synthase HispS, hispidin-3-hydroxylase H3H, luciferase Luz and putative caffeoyl pyruvate hydrolase CPH. Optimization of HispS, H3H and Luz-catalyzed steps resulted in two improved versions of the pathway: FBP2 and FBP3. b–d, FBP2 and FBP3 outperform the wild-type (WT) pathway when expressed in mammalian (b), yeast (c) and plant (d) hosts. In experiments in mammalian cells and yeast, each gene was delivered on a separate plasmid; in case of plants, plasmids encoding all genes were used. Experiments in plants and in yeast were performed at room temperature, and in mammalian cells at 37 °C. For mammalian cells, 100 µM caffeic acid was used, and for yeast 100 mM. Comparison in yeast was performed in strains lacking nnCPH. Asterisk indicates samples where luminescence level was close to that of the background; for that reason, fold-change values are not provided. The boxes are the first and the third quartiles, whiskers are the rest of the distribution except outliers, and the orange line is the median. The color of data points in c indicates different yeast strains. The difference between mean values and P values of post-hoc two-sided Conover test (b) or Mann–Whitney U tests (c and d) corrected by the step-down method using Šidák adjustments are indicated below the brackets between the box plots. Kruskal–Wallis H test: H-statistic 33.07, P = 3.6 × 10⁻⁶ (b), H-statistic 61.01, P = 1.8 × 10⁻¹² (c), H-statistic 35.59, P = 9.1 × 10⁻⁸ (d). N = 5–6 biologically independent samples (b); 6–18 biologically independent samples (c); 8–16 plant cell packs (d).

Fig. 2. Stable expression of pathway variants in plants.

a–d, Luminescence of transgenic lines stably expressing different versions of the bioluminescence pathway: average brightness of leaves of 3-week-old Nicotiana benthamiana (N = 14–68 leaves per box plot) (a), but also see Extended Data Fig. 5; average brightness of leaves of 4-week-old Nicotiana tabacum (N = 10–77 leaves per box plot) (b); average brightness of leaves of 3.5-month-old Populus canadensis (N = 10–14 leaves per box plot) (c); average brightness of leaves of 6-week-old Arabidopsis thaliana (N = 20–40 leaves per box plot) (d). The boxes are the first and the third quartiles, whiskers are the rest of the distribution except outliers, and the orange line is the median. The color of data points (if not neon green) indicates different plant lines. The difference between mean values and P values of post-hoc two-sided Mann–Whitney U tests corrected by the step-down method using Šidák adjustments are indicated below the brackets between the box plots. Photos of Arabidopsis thaliana (e), Chrysanthemum morifolium (f), Populus canadensis (g), Petunia hybrida (h and j), Nicotiana tabacum (i) and Nicotiana benthamiana (k) constitutively expressing FBP2, captured on Sony Alpha ILCE-7M3 camera (see Methods for ISO and exposure settings). Imaging was performed at room temperature. WT, wild type.

Extended Data Fig. 1. Co-expression of HispS homologues with wild-type nnLuz and wild-type nnH3H in plant, yeast, and mammalian hosts.

a. Transient expression in BY-2 plant cell packs (N = 12–16 cell packs per box plot). b. Transient expression in N. benthamiana leaves (N = 10–33 leaves per box plot). c, d. Expression from genome-integrated copy in yeast P. pastoris, luminescence assayed after adding 100 µM (c) and 100 mM (d) caffeic acid, integral signal for 1 hour (N = 6–27 and 3–27 biologically independent samples per box plot in c and d, respectively). e. Transient expression in human cell culture HEK293NT, luminescence assayed after adding 100 µM caffeic acid, integral signal for 23 mins (N = 2–5 biologically independent samples per box plot). The boxes are the first and the third quartiles, whiskers are the rest of the distribution except outliers, the orange line is the median. The colour of data points in case of c. and d. indicates different yeast strains (1–9 strains per box plot). The difference between mean values of nnHispS and mcitHispS and p-values of pairwise post-hoc two-sided Mann-Whitney U-tests (if applicable) corrected by the step-down method using Sidak adjustments are indicated below the brackets between the box plots. Kruskal-Wallis H Test: H-statistic = 155.79, p = 9.7e-28 (a), H-statistic = 158.86, p = 2.3e-28 (b), H-statistic = 182.56, p = 3.0e-33 (c), H-statistic = 24.79, p = 8.3e-04 (d), H-statistic = 155.79, p = 9.7e-28 (e). In these experiments, each gene was delivered on a separate plasmid. Plasmids used in these experiments are listed in Supplementary Table 1.

Extended Data Fig. 2. Effect of phosphopantetheinyl transferase NpgA on luminescence in plant cells.

a. Transient expression in N. benthamiana leaves (N = 9 leaves per box plot). b. Transient expression in P. hybrida flowers (N = 13 and N = 15 leaves for FBP1 + NpgA and FBP1, respectively). c. Transient expression in BY-2 cells (N = 14 plant cell packs per box plot). In these experiments, pX037 plasmid was co-infiltrated with a plasmid encoding NpgA. The boxes are the first and the third quartiles, whiskers are the rest of the distribution except outliers, the orange line is the median. The difference between mean values and p-values of post-hoc two-sided Mann-Whitney U-tests are indicated below the brackets between the box plots, p = 7.9e-04 (a), p = 8.0e-06 (b), p = 7.0e-06 (c). In these experiments, each gene was delivered on a separate plasmid.

Extended Data Fig. 3. Comparison of combinations of nnLuz, nnH3H and HispS variants upon transient expression in BY-2 cells, upon co-expression with NpgA.

In these experiments, each gene was delivered on a separate plasmid. Box and whisker plots (left) are accompanied by colour-coded p-values of post-hoc two-sided Conover’s test corrected by the step-down method using Sidak adjustments (right). NS — non-significant. The boxes are the first and the third quartiles, whiskers are the rest of the distribution except outliers, the orange line is the median. N = 15 plant cell packs for nnHispS + nnH3H WT + nnLuz WT and mcitHispS + nnH3H v2 + nnLuz v4 or 14 plant cell packs for other combinations. Kruskal-Wallis H Test: H-statistic = 99.89, p = 1.1e-18. The difference between several mean values are indicated below the brackets between the box plots.

Extended Data Fig. 4

Insert structure of the plasmids used to create transgenic plant lines.

Extended Data Fig. 5. Average luminescence of leaves of 3-week-old N. benthamiana transformed with various versions of the pathway.

a. Same as Fig. 2a, provided here for easier comparison with (b): results of an experiment, which only had a single FBP3 transgenic line in comparison, for technical reasons. b. Results of an independent experiment, which only had a single FBP2 transgenic line in comparison, for technical reasons. The FBP1 is the plant line NB021 reported in ref. . Photos were captured with ISO 400 and 30 seconds of exposure (see Methods). Box and whisker plots are accompanied by colour-coded p-values of post-hoc two-sided Conover’s test corrected by the step-down method using Sidak adjustments. NS — non-significant. The boxes are the first and the third quartiles, whiskers are the rest of the distribution except outliers, the orange line is the median. The colour of data points indicates different plant lines: for a. NB021 for FBP1 (N = 14 leaves), 3 lines for nnHispS + nnH3H v2 + nnLuz v4 + nnCPH (N = 37 leaves), 4 for FBP2 (N = 68 leaves) and 1 for FBP3 (N = 20 leaves). b. One line for FBP2 (N = 11 leaves) and 4 lines for FBP3 (N = 41 leaves). The difference between mean values is indicated below the brackets between the box plots; in (b) supplied with p-value of post-hoc two-sided Mann-Whitney U-test, p = 7.0e-06. Kruskal-Wallis H Test: H-statistic = 98.97, p = 2.6e-2 (a).

Extended Data Fig. 6. Imaging of 4-week-old N. tabacum transformed with FBP1 or FBP2.

The FBP1 is the plant line NT001 reported in ref. . The photo (ISO 400, exposure 30 sec) (a) and average luminescence of leaves (b). The boxes are the first and the third quartiles, whiskers are the rest of the distribution except outliers, the orange line is the median. The colour of data points indicates different plant lines (NT001 for FBP1, 5 lines for FBP2); N = 10 and N = 77 leaves for FBP1 and FBP2, respectively. The difference between mean values and p-value of post-hoc two-sided Mann-Whitney U-test are indicated below the brackets between the box plots, p = 3.1e-07.

Extended Data Fig. 7. Comparison of FBP3 pathway to luciferases that require exogenous substrate in BY-2 cells.

(a) Integral luminescence signal collected for 30 minutes. The boxes are the first and the third quartiles, whiskers are the rest of the distribution except outliers, the orange line is the median. N = 5 plant cell packs for NanoLuc and FBP3, and N = 8 for FFLuc. The difference between mean values and p-values of post-hoc two-sided Mann-Whitney U-test corrected by the step-down method using Sidak adjustments are indicated below the brackets between the box plots. (b) - Kinetics. Data shown as mean (solid line) ± SD (area around the solid line). For NanoLuc, 0.75 µL of substrate N113 (N1110, Promega) was used. For FFLuc, we used 100 µM of D-Ln. For FBP3, no substrate was added.

Extended Data Fig. 8. Comparison of FBP3 pathway to luciferases that require exogenous substrate in HEK293NT cells.

(a) Integral luminescence signal collected over 20 minutes. The boxes are the first and the third quartiles, whiskers are the rest of the distribution except outliers, the orange line is the median. N = 5 biologically independent samples per box plot. The difference between mean values and p-values of post-hoc two-sided Mann-Whitney U-test corrected by the step-down method using Sidak adjustments are indicated below the brackets between the box plots. (b) Kinetics. Data shown as mean (solid line) ± SD (area around the solid line). For NanoLuc, 0.5 µL of substrate N113 (N1110, Promega) was used. For FFLuc, we used 100 µM of D-Ln. For nnLuz v4, 100 µM of fungal luciferin was added. For FBP3, 100 µM of caffeic acid.

Extended Data Fig. 9. Autonomous luminescence conferred by bacterial and fungal pathways in plant cells.

(a–c) Luminescence of BY-2 cells infected by different amounts of agrobacteria, expressed as OD₆₀₀. (d) Luminescence of N. benthamiana leaves. In these experiments, each gene was delivered on a separate plasmid. cyto stands for cytoplasmic localisation of iLux enzymes, chlor – for plastid localisation. Box and whisker plots (left) are accompanied by colour-coded p-values of post-hoc two-sided Conover’s test (right) corrected by the step-down method using Sidak adjustments. NS — non-significant. The boxes are the first and the third quartiles, whiskers are the rest of the distribution except outliers, the orange line is the median. N = 4 plant cell packs per box plot (a, b, c) or 12 leaves per box plot (d). The difference between mean values is indicated below the brackets between the box plots. Kruskal-Wallis H Test: H-statistic = 17.60, p = 1.5e-03 (a), H-statistic = 17.39, p = 1.6e-03 (b), H-statistic = 18.29, p = 1.1e-03 (c) and H-statistic = 30.86, p = 3.3e-06 (d).

Extended Data Fig. 10. Comparison of autonomous bacterial bioluminescent system with FBP3 pathway after treatment with 100 µM caffeic acid in transient expression in HEK293NT.

(a) Average signal collected over 20 min. N = 6 biologically independent samples per box plot. (b) Maximum signal. The boxes are the first and the third quartiles, whiskers are the rest of the distribution except outliers, the orange line is the median. The difference between mean values and p-values of post-hoc two-sided Mann-Whitney U-tests are indicated below the brackets between the box plots (c) Kinetics. Data shown as mean (solid line) ± SD (area around the solid line).