Genetically encoded fluorescent reporter for polyamines

Abstract

Polyamines are abundant and evolutionarily conserved metabolites that are essential for life. Dietary polyamine supplementation extends life-span and health-span. Dysregulation of polyamine homeostasis is linked to Parkinson's disease and cancer, driving interest in therapeutically targeting this pathway. However, measuring cellular polyamine levels, which vary across cell types and states, remains challenging. We introduce a genetically encoded polyamine reporter for real-time measurement of polyamine concentrations in single living cells. This reporter utilizes the polyamine-responsive ribosomal frameshift motif from the OAZ1 gene. We demonstrate broad applicability of this approach and reveal dynamic changes in polyamine levels in response to genetic and pharmacological perturbations. Using this reporter, we conduct a genome-wide CRISPR screen and uncover an unexpected link between mitochondrial respiration and polyamine import, which are both risk factors for Parkinson's disease. By offering a lens to examine polyamine biology, this reporter may advance our understanding of these ubiquitous metabolites and accelerate therapy development.

Fig. 1. Polyamine sensor design and validation.

a Chemical structures of mammalian polyamines. b Schematic of the polyamine metabolic pathway in mammalian cells. DFMO: difluoromethylornithine, Sard: Sardomozide. c Schematic of polyamine-responsive +1 ribosomal frameshifting during translation of OAZ1 mRNA. d Design for the polyamine sensor. e Immunoblot showing production of the mCherry fusion protein in untreated cells and its reduction following DFMO treatment. f-g Representative fluorescence micrographs (f) and corresponding quantification by flow cytometry (g) under indicated treatments. DFMO, 1 mM (90 h) and Spd: spermidine, 5 μM (18 h). Representative fluorescence micrographs (h) and corresponding quantification by flow cytometry of U-2OS cells (i) upon genetic knockdown of indicated polyamine biosynthesis enzymes. AG (1 mM) was included during spermidine supplementation in (e)–(i). Each data point shown in (g) and (i) represents a single cell. Error bars in (g) and (i) denote the median ± interquartile range. 50 data points are shown. The sample size as follows: (g) control (n = 6095 cells), DFMO (n = 4576 cells) and DFMO/Spd (n = 5874 cells), (i) control (n = 48638 cells), ODC1 KD (n = 47758 cells), SRM KD (n = 42255 cells), ODC1 KD/Spd (n = 46829 cells) and SRM KD/Spd (n = 40990 cells). Flow cytometry quantification and the fluorescent micrographs are representative of 2 independent experiments. Significance values are calculated using two-sided Student’s t test. *** denotes p < 0.0001. Scale bars, 10 µm. Source data is provided as a Source Data file.

Fig. 2. Reporter calibration and reproducibility.

a Schematic of LC/MS protocol for polyamine quantification. Intracellular spermidine concentration measured using LC/MS (b) and F/F₀ (c) in response to various concentrations of DFMO. d Linear regression of the cellular spermidine concentration with F/F₀ (calculated from (c)). e F/F₀ and the spermidine concentration measured using LC/MS upon treatment with ribavirin (100 μM for 72 h; SAT1 activator). The sensor value was calculated using the fit from (d). f Coefficient of variation (CV) was calculated for intra-assay and inter-day variability. CV is calculated as the ratio of standard deviation to the average value. g Intracellular spermidine concentration measured using F/F₀ in response to DFMO and subsequent rescue with spermidine supplementation over time. Cells were treated with 2 mM DFMO for four days followed by simultaneous spermidine supplementation (100 μM) and reporter induction with doxycycline. AG (1 mM) was added with spermidine. Error bars in (b) and (e, left) denote the mean ± standard deviation. Error bars in (c), (e, right) and (g) denote the median ± interquartile range. 50 data points are shown. The sample size as follows: b 0 μM DFMO (n = 5 samples), 50–250 μM DFMO (n = 4 samples) and 400–500 μM DFMO (n = 3 samples), c 0 μM DFMO (n = 43696 cells), 50 μM DFMO (n = 43417 cells), 75 μM DFMO (n = 45531 cells), 100 μM DFMO (n = 49477 cells), 125 μM DFMO (n = 51808 cells), 150 μM DFMO (n = 54552 cells), 175 μM DFMO (n = 57358 cells), 200 μM DFMO (n = 57842 cells), 250 μM DFMO (n = 61712 cells), 300 μM DFMO (n = 64135 cells), 400 μM DFMO (n = 63264 cells) and 500 μM DFMO (n = 63353 cells), e control (n = 5 samples) and ribavirin (n = 3 samples), g control (n = 364 cells), DFMO (n = 358 cells), 2 h DFMO/Spd (n = 557 cells), 4 h DFMO/Spd (n = 248 cells), 6 h DFMO/Spd (n = 233 cells) and 8 h DFMO/Spd (n = 463 cells). Source data are provided as a Source Data file.

Fig. 3. Live-cell longitudinal imaging of polyamine dynamics.

a Schematic of sensor design for real-time tracking of polyamines. b Representative fluorescence images of U-2OS cells expressing the degron-based sensor under indicated treatments. DFMO (2 mM), Spd (spermidine, 100 μM) and TMP (trimethoprim, 0.05 μM). AG (1 mM) was included during spermidine supplementation in (b)–(d). U-2OS cells were grown in TMP starting 3 days before DFMO treatment to allow steady-state levels of the reporter expression. Media was replaced every two days with fresh reagents. c Single-cell traces of F/F₀ upon indicated treatments. F is the miRFP670-2 to mCherry fluorescence. F₀ is the average frameshifting efficiency in untreated cells (at the time, t = 0). Arrows indicate the time-point of treatment. d Mean F/F₀ over time from the single-cell traces shown in (c). Shaded bands depict the 95% confidence intervals. The sample size as follows: n = 122 cells in (c) and (d). Fluorescent micrographs are representative of 1 independent experiment. Scale bars, 10 µm. Source data are provided as a Source Data file.

Fig. 4. Single-cell polyamine import assay.

a Schematic of single-cell polyamine import assay. b Representative fluorescence images of cells upon indicated treatments. AMXT-1501 (0.5 μM; polyamine import inhibitor), DFMO (1 mM), and spermidine (5 μM). AG (1 mM) was included during spermidine supplementation in (b) and (c). c Quantification of frameshift efficiency using flow cytometry for treatments in (b). Error bars in (c) denote the median ± interquartile range. 50 data points are shown. Flow cytometry quantification and the fluorescent micrographs are representative of 2 independent experiments. The sample size as follows: (c) control (n = 8089 cells), AMXT (n = 25727 cells), AMXT/DFMO (n = 8439 cells) and AMXT/DFMO/Spd (n = 9262 cells). Significance values are calculated using two-sided Student’s t test. Scale bars, 10 µm.

Fig. 5. Genome-wide CRISPR screen identifies modifiers of polyamine import.

a Schematic of the CRISPR-Cas9 FACS-based genome-wide screen to identify modulators of polyamine import. b -log10 (MaGeCK score) versus median log2 (fold enrichment) for all genes from the spermidine import screen. The horizontal line indicates FDR < 0.05 and the vertical line indicates fold change >1.5. c Graph depicting the fold enrichment for sgRNAs targeting indicated genes. Each data point is for a distinct sgRNA sequence. d Representative images of U-2OS cells corresponding to treatment with DFMO: 1 mM and spermidine: 5 μM. Comparison of polyamine levels in wild-type and upon ATP13A3 knock-out in K562 (e), U-2OS (f), and RPE1 (g) cells under indicated treatments. h Heatmap of mean log2 fold changes for individual sgRNAs targeting genes previously proposed to be involved in polyamine import (n = 2 technical replicates). i Comparison of polyamine levels in U-2OS cells under indicated treatments. Rotenone (1 μM; complex I inhibitor) and antimycin (1 μM; complex III inhibitor) were added 24 h before reporter induction and spermidine addition (5 μM, 18 h). j Total polyamine levels (see enzymatic total polyamine assay in Methods) for the indicated samples (relative abundance). Rotenone (1 μM, 48 h) and spermidine (5 μM, 16 h). 2 mM DFMO (48 h) was used as a positive control. AG (1 mM) was included during spermidine supplementation in (a)–(j). Error bars in (c) denote the mean ± standard deviation (5 independent sgRNAs). Error bars in (e–g, i) denote the median ± interquartile range. 50 data points are shown. Error bars in (j) denote the mean ± standard deviation from 3 independent experiments. Significance values are calculated using two-sided Student’s t test. *** denotes p < 0.0001. Flow cytometry quantification and the fluorescent micrographs are representative of 1 independent experiment. Scale bars, 10 µm. The sample size as follows: e K562 WT (n = 10449 cells), WT/DFMO (n = 30804 cells), WT/DFMO/Spd (n = 9964 cells), ATP13A3 KO (n = 10097 cells), ATP13A3 KO/DFMO (42214 cells), ATP13A3 KO/DFMO/Spd (n = 40848 cells), f U-2OS WT (n = 43369 cells), WT/DFMO (n = 39870 cells), WT/DFMO/Spd (n = 41357 cells), ATP13A3 KO (n = 45404 cells), ATP13A3 KO/DFMO (n = 5741 cells), ATP13A3 KO/DFMO/Spd (n = 5960 cells), (g) RPE-1 WT (n = 9623 cells), WT/DFMO (n = 8437 cells), WT/DFMO/Spd (n = 9437 cells), ATP13A3 KO (n = 9297 cells), ATP13A3 KO/DFMO (n = 9814 cells), ATP13A3 KO/DFMO/Spd (n = 9564 cells) and i control (n = 18636 cells), rotenone (n = 17673 cells), antimycin (n = 18100 cells), rotenone/Spd (n = 17311 cells) and antimycin/Spd (n = 17758 cells). Source data are provided as a Source Data file. Created in BioRender. Sharma, P. (2025) https://BioRender.com/jewt37i.