Prion-induced ferroptosis is facilitated by RAC3

Abstract

Prions are infectious agents that initiate transmissible spongiform encephalopathies, causing devastating neuronal destruction in Creutzfeldt-Jakob and Kuru disease. Rapid cell death depends on presence of the endogenous prion protein PrP^C, but its mechanistic contribution to pathogenesis is unclear. Here we investigate the molecular role of PrP^C, reactive oxygen species and lipid metabolism in ferroptosis susceptibility, a regulated cell death process characterized by lipid peroxidation. We discover that elevated expression of the cellular prion PrP^C creates a relaxed oxidative milieu that favors accumulation of unsaturated long-chain phospholipids responsible for ferroptotic death. This condition is sustained by the luminal protein glutathione peroxidase 8, which detoxifies reactive species produced by protein misfolding. Consequently, both PrP^C and infectious Creutzfeldt-Jakob disease (CJD) prions trigger ferroptotic markers and sensitization. This lethality is further enhanced by RAC3, a small GTPase. Depletion of RAC3 is observed solely in pathologically afflicted cortices in CJD patients, revealing a synergistic modulation of lipids and reactive species that drives ferroptosis susceptibility. Together, the results show that PrP^C initially suppresses oxidative stress, attenuates cellular defenses, and establishes a systemic vulnerability to the ferroptotic cascade. These results provide insight into the mechanism underlying regulation of ferroptosis in prion diseases and highlight potential therapeutic targets for diseases involving dysregulated cell death processes.

Fig. 1. PrP^C represses cellular and endoplasmic ROS.

A Cellular free iron content in empty vector control (control OE) and PrP^C OE HT-1080 cells measured by 5 nM calcein or 1 μM FerroOrange staining. Wild-type HT-1080 treated with 500 μM ferric ammonium citrate (FAC) and 100 μM deferoxamine (DFO) compared to DMSO as the positive and negative control, respectively, a typical FACS histogram is depicted. Insets show ferritin heavy chain (FTH, 21kD), ferritin light chain (FTL, 20kD) and PrP^C levels by Western blot with normalization by beta-actin (ACTB, 42kD). B Relative PRNP mRNA expression levels change up to 21 days post-infection with a lentiviral expression vector-containing PrP^C (OE) (encoded by PRNP) and a separate cistron for puromycin resistance, in 1 μg/mL puromycin-containing media. Total RNA was normalized to empty vector control (C). C Cytosolic ROS levels detected by 2,7-Dichlorodihydrofluorescein diacetate (DCFH-DA) using flow cytometry in indicated HT-1080 cells at 7 dpi (days post-infection). A representative flow cytometry histogram of three independent repetitions is depicted (left panel). Boxplot shows (right panel) fluorescence intensity in control and PrP^C OE cells after 0.1 μM RSL3 for 0 h and 2 h, with mean, max and min indicated. D Pearson correlation analysis of GPX8 mRNA expression with PRNP mRNA expression (R = 0.621; two-tailed P < 0.0001; 95% confidence interval 0.587 to 0.651) of 18,575 genes determined for 1393 individual cell lines. GAPDH is included as a reference gene. TPM, transcripts per million. E Subcellular colocalization of PrP^C together with the endoplasmic reticulum-specific marker concanavalin A (10 μg/mL). Pictures shown are representative results of at least three independent repetitions performed with similar outcomes (scale bar = 20 µm). F Kinetic analysis of ER ROS using Hyper-3 ER fluorescent reporter (emission ratio 488/405) upon PrP^C expression (PRNP^tet +dox). Samples were measured by flow cytometry for 1 min to determine baseline intensity before supplementation with DTT (∆ 5 mM) or H₂O₂ (▲ 50 mM). -dox/+dox are samples treated with vehicle or doxycycline, respectively. DTT, dithiothreitol. G Glutathione peroxidase (GPX) activity in PrP^C OE compared to empty vector control showing in bar graph. Insets show PrP^C and GPX8 levels by Western in PrP^C OE and control cells. H Fluorescence intensity of Hyper-3 ER in HT-1080 cells expressing GPX8 compared to empty vector control cells. Average fluorescence intensity values (emission ratio 488/405) were determined by flow cytometry. Insets show GPX8 level by Western. A, C FACS histograms of at least three independent experiments are depicted. Boxplot (C) is shown with whiskers min to max. Relative mRNA expression (B) is shown as mean ± SEM of n = 3 technical replicates. Boxplot data (C) is plotted as whiskers min to max, showing all points of n = 6 each replicates of duplicate repetitions of the experiment with similar results. Significance was determined by two-tailed t-test (C). Activity data (G) is plotted as representative mean ± SEM of n  =  4 biological replicates for independent experiments. Intensity (H) was calculated versus control, shown as mean ± SEM of n = 3 biological replicates. P-values of two-tailed t-test (B, G, H) or ANOVA multiple comparisons with Tukey post-test are shown for comparisons. *P  <  0.05, **P  <  0.01, ***P  <  0.001, ****P  <  0.0001. Source data are provided as a Source Data file.

Fig. 2. PrP^C sensitizes cells to ferroptosis via lipid remodeling.

A Cell survival of HT-1080 PrP^C OE (OE) and empty vector control (C) cells treated with cell death inhibitors. Cells were infected for 3 days, selected with puromycin for 24 h, then cultured with inhibitors 10 μM α-tocopherol (aToc) or 2 µM ferrostatin-1 as a ferroptosis inhibitor, 10 μM z-VAD-FMK (zVAD) as an apoptosis inhibitor, and 10 μM necrostatin-1 (Nec-1) as a necroptosis inhibitor, respectively, for 3 days. B (Left) Dose-response curves illustrating sensitivity of PrP^C-expressing HT-1080 cells (PrP^C OE) compared to empty vector control cells (control OE) to IKE treatment over 16 h. (Right) Survival of PrP^C OE cells compared to control against ferroptosis inducer 0.3 μM IKE (16 h). (Lower panels) Sensitivity of GPX8 knockout (KO) cells with (+dox) and without (-dox) PrP^C OE. Inset shows GPX8 protein levels by Western and total protein loading compared to knockout control cells. Viability data are plotted as representative mean ± SD of n  =  3 technical replicates for independent experiments repeated at least three times with similar outcomes. C PrP^C OE effect on lipid peroxidation induced by 0.1 μM RSL3 induction in HT-1080 OE and control cells measured by BODIPY 581/591 C11 (BODIPY-C11). A typical FACS histogram of n = 3 technical replicates of three independent repetitions is depicted. The bar graph shown on the right illustrates the BODIPY-C11 median intensity ± SEM. D Comparative analysis of lipid levels in PrP^C OE HT-1080 cells compared to control cells, highlighting fatty acid acyl chain saturation in normalized lipidomics data. Normalization was performed by glucose peak area for each sample and each feature. XCMS data analysis and tentative annotation (of 617 known lipids) were according to the Bruker UPLC-Q-TOF protocol in positive mode, detecting 10,521 total features. Significant lipids have -log₁₀ P > 1.30. Viability data are representative means of n  =  9 (A) biological or n  =  3 (B) technical replicates for experiments repeated independently at least three times. P-values of two-way ANOVA multiple comparisons with Tukey post-test are shown (A–C). *P  <  0.05, **P  <  0.01, ***P  <  0.001, ****P  <  0.0001. Lipid statistics (D) were determined by P-values (P < 0.05, n = 5, two-tailed Welch’s t-test). n = 5 PrP^C OE and n = 5 control samples. Source data are provided as a Source Data file.

Fig. 3. Native and infectious CJD prions drive ferroptosis sensitivity.

A FABP5 changes in PrP^C OE and empty vector control cells by immunostaining and Western with densitometry. Pictures shown are representative results of three independent repetitions performed in triplicate with similar outcomes (scale bar = 20 µm). B Relative expression of ferroptosis markers in PrP^C OE (OE) and control (C) cells detected by qPCR and Western. Lower blots display total protein loading. C Comparison of FABP5 staining intensity in iPS-derived brain organoids infected with normal brain homogenates (NBH) and sCJD prion homogenates (169 days post-infection, dpi). Pictures shown are representative results of at least three independent repetitions performed with similar outcomes of minimum 10 organoids (scale bar = 200 µm (left panels) and 50 µm (right panels)). D Western blotting and densitometry comparing FABP5 protein levels in the brain tissue of mice inoculated with normal brain homogenate (NBH; control) or RML scrapie brain homogenate and sacrificed at 80, 108, and 160 days post inoculation (dpi). E Western blotting and densitometry comparing FABP5 protein levels in the brain tissue of people who died from sporadic CJD with tissue from patients who died of a non-brain-related condition. F Lactate dehydrogenase (LDH) release level of brain organoids treated with increasing RSL3 concentrations and aToc rescue for ferroptosis specificity. G Normalized survival of brain organoids infected with sCJD (MM1) prion homogenates for 90 days compared to NBH controls against an RSL3 treatment at indicated concentrations (scale bar = 200 µm). Cell death was measured by LDH release activity (48 h, relative to DMSO). H Normalized survival of PRNP KO compared to NBH controls treated with RSL3 and measured by LDH release (48 h, relative to DMSO). Western data (A, B) is shown as mean ± SD of n = 3 technical replicates. Relative mRNA expression (B) is shown as mean ± SEM of n = 3 technical replicates. LDH release data and Western data (E) are plotted as representative mean ± SEM of n  =  3 (D, F) or n = 6 (E, G, H) biological replicates for independent experiments. Significance was determined by two-tailed t-test (A, B, F), two-tailed Welch’s t-test (E) and two-way ANOVA comparisons, Tukey post-test (D, G, H). *P  <  0.05, **P  <  0.01, ***P  <  0.001, ****P  <  0.0001. Source data are provided as a Source Data file.

Fig. 4. A CRISPR-activation screen identifies RAC3 in prion-induced ferroptosis.

A Schematic of a whole-genome CRISPR-activation depletion screen. HT-1080 conditionally expressing PrP^C (PRNP^tet) were transduced with a synergistic activation mediator pooled human library and dCas9-VP64 lentiviruses. PrP^C expression was induced for 3 days with doxycycline or vehicle, followed by treatment with 500 nM IKE, resulting in a viability loss of 5–10%. Normalized guide frequencies amplified from viable cells in sensitized PrP^C cells were scored against (-)doxycycline control cells to identify enriched or depleted guides. B Fishtail plot of prion-facilitated ferroptosis by differential CRISPRa guide representation. Effect size is determined by the average PRNP^tet (+)doxycycline / (−)doxycycline ratio. Depleted guides (Effect size <1) are positive regulators of PrP^C-facilitated ferroptosis. C Inverse relationship of RAC3 and PRNP in PrP^C OE or RAC3 OE, respectively, by qPCR and Western blot. Relative mRNA expression is shown as mean ± SD of n = 3 technical replicates. D Gene effect characterized by principal component analysis. CRISPR gene effect (individual gene knockout effect on viability and growth, scaled for whole-genome libraries) in 1095 cell lines is displayed in two dimensions. The distance between two genes in the scatter plot can be interpreted to have similar or inverse effects on cellular processes across individual cell lines, relative to all other cell lines. Significance was determined by two-tailed t-test. *P  <  0.05, **P  <  0.01, ***P  <  0.001, ****P  <  0.0001. Source data are provided as a Source Data file.

Fig. 5. Prion-induced ferroptosis is facilitated by RAC3-driven mesenchymalization.

A Survival of HT-1080 cells treated with RAC-inhibitor EHop-016 (2 μM) and ferroptosis inducer (IKE) rescued by 10 μM α-tocopherol (aToc). B Viability bar graph of RAC3 expressing cells (RAC3 OE) compared to empty vector control with ferroptosis inducer 1.25 μM IKE and 10 μM αToc as a ferroptosis inhibitor. A Western indicates increased expression of RAC3 protein. C Cell survival of PrP^C OE, RAC3 OE, and RAC3+PrP^c co-expressing cells was detected in freshly infected cell. Cells were treated subsequently with either 2 μM Ferrostatin-1 or DMSO for 3 days. D A flow cytometry histogram depicts the effect of RAC3 OE on induced ROS level via 0.3 μM RSL3 treatment in HT-1080 cells for 3 h measured by DCFH-DA after the indicated time. E Principal component analysis of all metabolomic and lipidomic mass spectrometry features for RAC3 OE cells and cells treated with IKE, compared to control cells. (PC, principal component). F Bar graph illustrates differential expression of 210 annotated phospholipids and lysophospholipids with varying total numbers of double bonds in RAC3 OE cells compared to control cells with an empty vector. Data are representative means ± SD of n  =  5 technical replicates. G Differential expression of 210 annotated phospholipids and lysophospholipids with varying total numbers of double bonds in IKE-treated cells compared to vehicle-treated controls. Data are representative means ± SD of n  =  5 technical replicates. H Brightfield images of control, RAC3 OE, TGFbeta-1 treated, and RAC3+PrP^c OE co-infected cells indicating various levels of roundness (scale bar = 20 µm). I Aspect ratio depicting the effects of TGFbeta-1 treatment, RAC3 OE cells, and RAC3+PrP^c co-expressing HT-1080 cells (mean shape, length/width ratio). J Relative expression of epithelial and mesenchymal markers in RAC3 OE and TGFbeta-1 treated cells (3 days) compared to empty vector or DMSO-treated control detected by qPCR. Relative mRNA expression is shown as mean ± SEM of n = 3 technical replicates. Viability data (A–C) are representative means ± SEM of n  =  3 biological replicates for experiments repeated independently at least three times. The aspect ratio (I) are representative means of n  =  12 replicates measured by high content microscopy. Significance was determined by two-way ANOVA multiple comparisons with Tukey post-test (A–C, I) or two-tailed t-test (J). *P  <  0.05, **P  <  0.01, ***P  <  0.001, ****P  <  0.0001. Source data are provided as a Source Data file.

Fig. 6. RAC3 expression in RML mice and CJD patient brains.

A Western blotting and densitometry comparing RAC3 protein levels in the brain tissue of mice inoculated with normal brain homogenate (NBH; control) or RML scrapie brain homogenate and sacrificed at 80, 108, and 160 days post inoculation (dpi). Total protein is shown in the lower blot. Data are representative means ± SEM of n  =  3 (NBH) or n  =  4 (RML) biological replicates. B Western blotting and densitometry comparing RAC3 protein levels in the brain tissue of people who died from sporadic CJD with tissue from patients who died of a non-brain-related condition. Total protein is shown in the lower blot. Data are representative means ± SEM of n  =  3 (NBH) or n  =  6 (sCJD) biological replicates. C (Left panels) Immunohistochemical detection of prion protein deposits in the cortex of the superior frontal gyrus of Creutzfeldt-Jakob disease (CJD) case (left column) and control case (upper panels, scale bar = 20 µm). A higher magnification shows a punctate synaptic distribution pattern of prion protein deposits typical for both sporadic MM/MV1 CJD cases, whereas no deposits are detectable in the control case (lower panels; scale bar = 200 µm). (Right panels) Immunohistochemical detection of RAC3 in the superior frontal gyrus in a control case (left column) and a case with Creutzfeldt-Jakob disease (CJD, right column). RAC3 is strongly expressed in the neocortical neuropil of the control case (upper left; scale bar = 2 mm), showing a punctate synaptic distribution pattern at higher magnification without expression in cell bodies (lower left; scale bar = 20 µm). In the CJD case, total immunohistochemical signal for RAC3 is strongly reduced in the overview image (upper right; scale bar = 2 mm) compared to the control case. The synaptic distribution pattern is diffuse and unrecognizable at higher magnification (lower right; scale bar = 20 µm). The baseline characteristics of the 6 enrolled patients (3 neuropathological cases and 3 controls) are listed in Table 1. D A model of prion-induced ferroptosis sensitivity. PrP^C expression lowers ER ROS levels via increased GPX8 activity, leading to a relaxed oxidative environment that favors the accumulation of ferroptosis-sensitive PUFA-PL. Oxidative stress facilitated by RAC3 simultaneously depletes these lipids and triggers mesenchymalization, sensitizing to ferroptosis. PrP^Sc is speculated to contribute to ferroptosis by increased lipid peroxidation, consistent with observations in infected mice, or by cellular sensing of misfolded PrP, resulting in GPX8 upregulation and sensitivity in the presence of RAC3. Significance was determined two-tailed t-test (A, B), P-values = 0.0135 (A) and 0.0003 (B), respectively. Source data are provided as a Source Data file.