Lysoptosis is an evolutionarily conserved cell death pathway moderated by intracellular serpins

Abstract

Lysosomal membrane permeabilization (LMP) and cathepsin release typifies lysosome-dependent cell death (LDCD). However, LMP occurs in most regulated cell death programs suggesting LDCD is not an independent cell death pathway, but is conscripted to facilitate the final cellular demise by other cell death routines. Previously, we demonstrated that Caenorhabditis elegans (C. elegans) null for a cysteine protease inhibitor, srp-6, undergo a specific LDCD pathway characterized by LMP and cathepsin-dependent cytoplasmic proteolysis. We designated this cell death routine, lysoptosis, to distinguish it from other pathways employing LMP. In this study, mouse and human epithelial cells lacking srp-6 homologues, mSerpinb3a and SERPINB3, respectively, demonstrated a lysoptosis phenotype distinct from other cell death pathways. Like in C. elegans, this pathway depended on LMP and released cathepsins, predominantly cathepsin L. These studies suggested that lysoptosis is an evolutionarily-conserved eukaryotic LDCD that predominates in the absence of neutralizing endogenous inhibitors.

Fig. 1. Mouse Serpinb3a protected fetal intestinal epithelial cells (FIECs) from lysoptosis-like death.

a, b mSerpinb3a^+/+ (blue bars with individual data points marked) or mSerpinb3a^−/− (red bars with individual data points marked) FIECs were incubated in 100% (a) or 25% (b) DPBS for the indicated time points. The percent dead (Sytox positive) was calculated as (# Sytox positive nuclei/# of DAPI positive nuclei) × 100. The mean ± SD were from counts covering at least seven fields and compared using a two-tailed t-test (***P < 0.001). A representative experiment is shown. c–l Live-cell, confocal microscopy assessed lysosomal content (red, 10 K TMR-labeled dextran), and plasma membrane permeability (green, SG) in mSerpinb3a^+/+ (c–g) or mSerpinb3a^−/− (h–l). FIECs incubated in 25% (c–f, h–k) or 100% DPBS (g and l). Z series were obtained every 2 min for 1 h, indicated time points are shown (scale bar = 25 µm); inset scale bar = 5 µm). Note marked loss of endolysosomal staining in SG positive mSerpinb3a^−/− versus mSerpinb3a^+/+ FIECs (k vs^. f). FIECs incubated in 100% DPBS did not show SG uptake or loss of TMR-labeled dextran (g, l). Representative transmission electron micrographs (TEM) of mSerpinb3a^+/+ (m–p) or mSerpinb3a^−/− (q–t) FIECs after incubation in 25% DPBS (m, n, o and q, r, s) or 100% DPBS (p and t). Low magnification (scale bar =2 µm) TEM images of mSerpinb3a^+/+ or mSerpinb3a^−/− FIECs (m, p and q, t, respectively) show cell morphology. High magnification (scale bar =100 nm) TEM images (n, o and r, s, respectively) demonstrate cell morphology. n and r were magnified cells from within the hashed box in m and q, respectively. o and s were individual cells from a different field of view. Some mSerpinb3a^+/+ FIECs exhibit hallmarks of apoptotic morphology (n), including chromatin condensation (black arrowheads) and plasma membrane budding (open arrowheads), or necrotic morphology (o), including plasma membrane breaks (black arrows) and organelle degeneration (open arrows). mSerpinb3a^−/− FIECs displayed a necrotic cell death morphology (black and open arrows as above) along with severe vacuolization and cytoplasmic clearing (asterisks) (r, s). Original imaging data files can be found at https://data.mendeley.com/datasets/hk2t9x7d6x/1.

Fig. 2. LMP and lysosomal cysteine peptidases, but not executioner caspases or RIPK1, induced lysoptosis-like death in mSerpinb3a^−/− FIECs.

a-x Representative confocal fluorescence images of mSerpinb3a^+/+ (a-l) or mSerpinb3a^−/− (m-x) FIECs pretreated with diluent (DMSO; a-d; m-p), 2 µm DEVD-CHO (e-h; q-t) or 2 µm E64d (i-l; u-x) and then incubated in 25% DPBS for 1 h (scale bar = 40 µm). FIECs were stained with Hoescht 33342 (H33342; blue), annexin V-FITC (annexin V, green), propidium iodide (PI, red), and Alexafluor⁶⁴⁷ conjugated dextran (dextran⁶⁴⁷, white). y, z Quantification of dead cells (PI-positive; y) or lysosomal area (dextran⁶⁴⁷ staining; z) from different representative experiments (n = 3) over multiple fields (n > 5). i-xii Confocal maximum intensity projections of fluorescently labeled lysosomes from mSerpinb3a^+/+ (i–vi) or mSerpinb3a^−/− (vii–xii) FIECs (scale bar = 25 µm). Cultures were pretreated with DMSO (i-iii; vii-ix) or nec-1 (iv-vi; x-xii) for 1 h prior to incubation in 25% DPBS. Cell numbers, cell death, and lysosomes were quantitated using H33342 (blue), PI (red), and dextran⁶⁴⁷ (white), respectively. Representative images are from the same experiment, which was repeated three times. xii, xiv Quantification of PI positive (xii) or lysosomal (xiv) staining from multiple fields (n ≥ 6). The means ± SD of a representative experiment were compared using a two-way ANOVA with Tukey’s multiple comparisons (**P < 0.01, ***P < 0.001). Original data files can be found at https://data.mendeley.com/datasets/scgbb3s333/1.

Fig. 3. Apoptosis or necroptosis inducers preferentially triggered lysoptosis-like cell death in mSerpinb3a^−/− FIECs.

a–r To assess apoptosis, mSerpinb3a^+/+ or mSerpinb3a^−/− FIECs were treated with 100% PBS (a–f), 1 µm staurosporine (STS) for 16 h (g–l), or 25% PBS for 2 h (m–r). Representative confocal fluorescence maximum intensity projections were of cells stained with Hoechst 33342 (blue), annexin V-FITC (green), and PI (red; scale bars = 25 µm). s–t Quantification of annexin V (s) and PI (t) staining in multiple fields (n ≥ 5) from a representative of three experiments using mSerpinb3a^+/+ (blue) or mSerpinb3a^−/− (red) FIECs treated with 100% PBS (control), 1 µm STS for 16 h or 25% PBS for 2 h. u–x To induce necroptosis, mSerpinb3a^+/+ (u, w), or mSerpinb3a^−/− (v, x) FIECs were pretreated with either DMSO or 5 µm nec-1 and then incubated with 1 µm STS with 10 µm z-VAD-fmk for 16 h. Representative merged confocal fluorescence maximum intensity projections were of cells stained with H33342 (blue) and PI (red, scale bar = 25 µm). Dually labeled nuclei indicated dead cells and were depicted as magenta. y Quantification of PI staining in multiple fields from a representative of three experiments of mSerpinb3a^+/+ or mSerpinb3a^−/− FIECS pretreated with DMSO or 5 µm nec-1 and induced for necroptosis by treatment with 10 µm STS with 1 µm z-VAD-fmk for 16 h. For data in s, t, and y, the means ± SD were compared using a two-tailed t-test (***P < 0.001, **P < 0.01). Original data files can be found https://data.mendeley.com/datasets/4gfrf8xkyf/1.

Fig. 4. Hypotonic stress induces lysoptosis-like death in human tumor cell lines null for SERPINB3.

a, b HT3^B3-WT (blue) or HT3^B3-KO (red) cells were incubated with DMSO, 2 µm E64d (a, b), or 10 µm DEVD-CHO (DEVD; b) for 1 h prior to exposure to 100% or 10% DPBS for 4 h, and stained with SG and H33342. Percent dead was calculated as (# Sytox positive nuclei/# of blue nuclei) × 100. c–n Representative confocal images (scale bar = 25 µm) of HT3^B3-WT or HT3^B3-KO cell lines treated with 100% DPBS or 10% DPBS for 4 h and then stained with H33342 (blue; c, f, i, l), Lysotracker™ deep red (white; d, g, j, m), and PI (red; e, h, k, n). o–r To quantify LMP, HT3^B3-WT (o, q), or HT3^B3-KO (p, r) cells were incubated with 3 kDa Cascade blue, 10 kDa Alexa488, 40 kDa TMR, and 70 kDa Texas red conjugated dextrans prior to exposure to 100 or 10% DPBS and imaged using live-cell resonance scanning confocal microscopy (≥10 fields, ≥20 z-planes). The number of lysosomes at each time point (LC_t) were normalized to the initial lysosome count at time zero (LC_t0). s Area under the curve (AUC) for each dextran over time for both HT3^B3-WT (left) or HT3^B3-KO (right) cells treated with 100 or 10% PBS. Statistical significance was determined using a two-way ANOVA with Tukey’s multiple comparisons test. t–w Representative TEM images (scale bars = 500 nm) of HT3^B3-WT or HT3^B3-KO cells treated with 100 or 15% DPBS for 4 h. Vacuolization (closed arrowheads) was noted in the HT3^B3-KO, and to a lesser extent in the HT3^B3-WT cells, treated with 15% DPBS. x Blinded scoring of the percentage of necrotic appearing cells in 2000x magnification TEM images (≥20 fields) of HT3^B3-WT or HT3^B3-KO cells treated with 100 or 15% DPBS for 4 h. Unless otherwise noted, a representative of ≥3 replicates is shown, and the means ± SD were compared using a two-way ANOVA with Tukeys’ multiple comparisons (ns not significant, ***P < 0.001, **P < 0.01). Original data files can be found at https://data.mendeley.com/datasets/gnwb39t76k/1.

Fig. 5. The lysoptosis-like cell death pathway involves promiscuous cytoplasmic proteolytic activity.

a Western blot analysis of lysosomal CTSL (goat polyclonal cathepsin L, Novus, AF952), nucleocytoplasmic SERPINB3 (α-B3; rabbit polyclonal SERPINB3/B4), and cytosolic GAPDH () in supernatants of HT3^B3-WT and HT3^B3-KO cell lines treated with 100 or 10% DPBS for 30 min at 37 °C, 5% CO₂ prior to treatment with DPBS containing 0, 20, or 200 µg/ml digitonin. The relative band intensities (rel. int.) for each lane were calculated (Image Lab, v6.1, Bio-Rad) and are shown under the corresponding lane. Note the detection of CTSL in the 20 µg/ml digitonin supernatants of both HT3^B3-WT and HT3^B3-KO treated with 10% DPBS but not the 100% DPBS controls. Black arrowheads indicate full-length proteins, gray and open arrowheads demark processed active forms of CTSL. b Pan-lysosomal cysteine protease activity (zFR-R110). Activity in supernatants of HT3^B3-WT and HT3^B3-KO cell lines treated with 100 or 10% DPBS for 30 min at 37 °C, 5% CO₂ prior to treatment with DPBS containing 0, 20, or 200 µg/ml digitonin. The activity was normalized to the 200 µg/ml control to account for total lysosomal cysteine protease protein levels. Note, the significantly enhanced cysteine protease activity in HT3^B3-KO cytosolic supernatants (20 µg/ml digitonin) compared to HT3^B3-WT in 10% DPBS cells treatment. c Bulk CTSL-KO SW756^B3-WT or SW756^B3-KO cell lines were generated by CRISPR/cas9 methodology. The percentage of the bulk population containing indels in the CTSL gene was confirmed by NGS sequencing (Table S1) and CTSL protein levels are shown in ref. . Cell lines were then treated with either 100 or 10% DPBS in the presence of SG for 4 h. The % dead was calculated as the number of SG positive cells/the total number of cells as determined by high contrast brightfield imaging × 100. Analyses were compared using a two-way ANOVA with Tukeys’ multiple comparisons (n ≥ 10 replicates; *** P < 0.001). Note, the statistically significant reduction in % dead between SW756^{B3-KO;CTSL-WT} and SW756^{B3-KO;CTSL-KO}. d HT3^B3-WT or HT3^B3-KO cells were incubated with DMSO, 10 µm E64d, or 10 µm cathepsin L selective inhibitor, CAA0225 for 1 h prior to exposure to 100 or 10% DPBS for 14 h, and stained with SG. Percent dead was calculated as ((Sytox positive nuclei/total cells calculated by brightfield microscopy) × 100) (n = 5 replicates; ns not significant, *** P < 0.001). Uncropped immunoblots can be found in Supplementary Fig. 21.

Fig. 6. Tumor epithelial cell lines null for SERPINB3 undergo lysoptosis-like death when exposed to the apoptosis inducer, staurosporine.

a HT3^B3-WT or HT3^B3-KO cells were pretreated with DMSO, 10 µm DEVD-CHO (DEVD), or E64 for 1 h prior to treatment with DMSO or 10 µm staurosporine (STS) in the presence of NucView caspase-3/7 activity detection reagent for 16 h (% caspase-3/7 positive = (#Nucview positive/# of nuclei) × 100), (Note, this caspase data was performed in conjunction with the Sytox staining shown in q). b Immunoblot analysis of PARP, cleaved PARP (cPARP), caspase-3 cleavage, and actin (α-act) in HT3^B3-WT or HT3^B3-KO cell lines treated with DMSO (-) or 10 µm STS (+). The dashed area is shown with contrast enhancement below (black arrowheads = full length, open arrowheads = cleavage product). c–n Representative confocal images (scale bar = 25 µm) of HT3^B3-WT or HT3^B3-KO cells treated with DMSO or 10 µm STS for 18 h and then stained with H33342 (blue), annexin V (green), and PI (red). o, p Quantification of the percent annexin V and PI-positive cells from n ≥ 10 fields shown in (c–n). q HT3^B3-WT or HT3^B3-KO cells were treated as in (a) and incubated with Sytox™ orange for 16 h (percent dead = (# of Sytox™ positive/# of nuclei) × 100). r–u To quantify lysosomal content, HT3^B3-WT or HT3^B3-KO cells were incubated with fluorescently labeled different molecular weight dextrans as above, prior to exposure to DMSO or 10 µm STS. Live-cell confocal microscopy was used to determine the number of lysosomes at each time point (LC_t) normalized to the initial lysosome count at time zero (LC_t0) over n ≥ 9 fields. v AUC for each dextran over time for the experiment in r–u (means compared using two-way ANOVA with Tukey’s multiple comparisons test). w–z Representative TEM images (scale bars = 500 nm) of HT3^B3-WT or HT3^B3-KO cells treated with DMSO or 10 µm STS for 16 h. Apoptotic morphology indicated with open (membrane budding) and closed arrows (chromatin fragmentation), and necrotic morphology (vacuolization, cytoplasmic clearing) by an asterisk. Unless otherwise noted, a representative of ≥3 replicates is shown, and the means ± SD were compared using a two-tailed t-test (ns not significant, ***P < 0.001). Original imaging data files can be found at https://data.mendeley.com/datasets/6g2typdf63/1. Uncropped immunoblots can be found in Supplementary Fig. 21.

Fig. 7. Initiation of mTP-dependent necrosis by H₂O₂ triggered lysoptosis-like death in HT3^B3-KO cells.

a–n Representative live-cell confocal images of HT3^B3-WT and HT3^B3-KO cells stained with JC-1, then treated with 5 mM H₂O₂. Images were collected using Ex 488 nm with detection windows of Em 498–532 nm (JC-1 monomer, green) and 566–606 nm (JC-1 oligomer, red) every 5 min over 3 h (select time points shown). m Fluorescence intensity of JC-1 oligomer/JC-1 monomer normalized to time 0 (rel. JC-1 oligomer/JC-1 monomer) of the experiment in panels a–n. Time points with no fluorescence in either channel (dead cells) were eliminated from the analysis. n Area under the curve (AUC) for multiple fields for experiment in panels a–n (compared using one-way ANOVA with Tukey multiple comparisons). o Percent dead ((#Sytox™ positive/# of nuclei) × 100) of HT3^B3-WT or HT3^B3-KO cells treated with 5 mM H₂O₂ or DPBS for 4 h. p Percent dead HT3^B3-WT or HT3^B3-KO cells pretreated with DMSO, 10 µm cyclosporin A (CsA), or E64d for 1 h prior to exposure to 5 mM H₂O₂ or DPBS for 4 h. q–t To quantify lysosomal content, HT3^B3-WT or HT3^B3-KO cells were incubated with fluorescently labeled different molecular weight dextrans as above and imaged using live-cell confocal microscopy. The number of lysosomes at each time point (LC_t) were normalized to the lysosome count at time zero (LC_t0). u AUC for each dextran over time for an experiment in q–t (compared by two-way ANOVA with Tukey’s multiple comparisons test). v–y Representative TEM images (scale bars = 2 µm) of HT3^B3-WT or HT3^B3-KO cells treated with DPBS or 5 mM H₂O₂ for 4 h. Insets show normal or abnormal mitochondria (open and dark arrowheads, respectively; scale bars = 500 nm). Yellow arrowheads indicate enlarged vesicle and cytoplasmic clearing. (z-aa) Quantification of TEM images from the experiment in v–y showing the number of necrotic cells per field (z) and abnormal mitochondria per field (aa) (≥5 fields). Unless otherwise noted, a representative of ≥3 replicates is shown, and the means ± SD were compared using a two-tailed t-test (ns not significant, ***P < 0.001, **P < 0.01, *P < 0.05). Original data files can be found at https://data.mendeley.com/datasets/tpp7xnmwxg/1.

Fig. 8. Erastin induces ferroptosis, but minimal LMP, in SW756^B3-WT and SW756^B3-KO cells.

a–d Representative confocal images (scale bar = 25 µm) of SW756^{B3 WT} or SW756^{B3 KO} cells treated with DMSO or 20 µm erastin for 12 h in the presence of Sytox™ Orange (red, open arrowheads), then stained with the ROS indicator, H2-DCFDA (green, white arrowheads) and Mitotracker™ Deep Red (white). e Quantification of the percentage of H2-DCFDA positive cells in SW756^{B3 WT} or SW756^{B3 KO} cells after treatment with DMSO (−) or 20 µm erastin (+) for 12 h (means were compared using a one-way ANOVA with Tukey’s multiple comparisons). f The percent dead 12 h after treatment with erastin ((# Sytox positive nuclei/# blue nuclei) × 100, means compared using a one-way ANOVA with Tukey’s multiple comparisons). g SW756^{B3 WT} or SW756^{B3 KO} cells were incubated for 1 h with DMSO (−), ferrostatin-1 (fer-1), or E64d, prior to addition of DMSO (−) or 20 µm erastin (+) for 12 h (% dead and statistical comparison calculated as in panel f). h–k Representative confocal images (scale bar = 25 µm) of SW756^{B3 WT} or SW756^{B3 KO} cells treated with DMSO or 20 µm erastin for 12 h and stained with Lysotracker™ Red. l Quantification of the lysosomal count per cellular µm² (≥6 fields). m–p SW756^{B3 WT} or SW756^{B3 KO} cells were incubated with fluorescently labeled different molecular weight dextrans as above prior to exposure to either DMSO or 20 µm erastin and imaged using live-cell confocal microscopy. The number of lysosomes were quantified as above. q Area under the curve (AUC) for each dextran over time. (two-way ANOVA with Tukey’s multiple comparisons test). r–u Representative TEM images (scale bars = 2 µm) of SW756^{B3 WT} or SW756^{B3 KO} cells treated with either DMSO or erastin for 12 h. Insets show normal and abnormally appearing mitochondria (open and black arrowheads, respectively; scale bars = 500 nm). v, w Quantification of the number of abnormal mitochondria (v) and necrotic cells (w) in the experiment in panels r–u (≥10 fields). The means ± SD were compared using a two-tailed t-test (ns not significant, ***P < 0.001, **P < 0.01, *P < 0.05). Original data can be found at https://data.mendeley.com/datasets/9m9f4cfhxy/1.

Fig. 9. Tumor epithelial cell lines null for SERPINB3 undergo lysoptosis-like death under conditions that induce RIP1K-dependent necroptosis.

a Percent dead ((# Sytox positive nuclei/# of blue nuclei) × 100) of HT3^B3-WT or HT3^B3-KO cells 12 h after incubation with DMSO or 10 ng/µL human TNFα + 0.1 µm BV6 SMAC mimetic + 0.5 µm qVD-OPh + 1 µm cyclohexamide (T + S + Q + C). b Percent dead (calculated as in panel a) of HT3^B3-WT or HT3^B3-KO cells incubated with DMSO, 50 µm nec-1 or 10 µm E64d for 1 h prior to exposure to T + S + Q + C for 8 h. c Immunoblot analysis of RIP kinase 1 (α-RIPK1), phosphorylated RIP kinase 1 (α-pRIPK1), and actin (α-act) in HT29 (control cell line), HT3^B3-WT, and HT3^B3-KO treated with DMSO (−) or T + S + Q + C (+) for 12 h (black arrowheads denote bands for RIPK1 and pRIPK1 based on molecular mass). d–g Representative confocal images (scale bar = 25 µm) of HT3^B3-WT or HT3^B3-KO cells incubated with DMSO or T + S + Q + C and stained with Lysotracker Deep Red (white). h Quantification of the lysosomal count per cellular µm² for the experiment in d–g (≥9 fields, compared using a one-way ANOVA with Tukey’s multiple comparisons). i–l To quantitate lysosomal content, HT3^B3-WT or HT3^B3-KO cells were incubated with 3 kDa Cascade blue, 10 kDa Alexa488, 40 kDa TMR, and 70 kDa Texas red labeled dextrans prior to exposure to either DMSO or T + S + Q + C and imaged using live-cell resonance scanning confocal microscopy (≥20 z-planes, ≥10 fields). The number of lysosomes at each time point (LC_t) were normalized to the lysosome count at time zero (LC_t0). m AUC for each dextran over time for an experiment in i–l (two-way ANOVA with Tukey’s multiple comparisons test). n–q Representative TEM images (scale bars = 2 µm) of HT3^B3-WT or HT3^B3-KO cells treated with DMSO or T + S + Q + C. r Quantification of the number of necrotic cells per field (≥10 fields) for the experiment in n–q. Unless otherwise noted, a representative of ≥3 replicates is shown, and the presented means ± SD were compared using a two-tailed t-test (ns not significant, ***P < 0.001, **P < 0.01, *P < 0.05). Original imaging data can be found at https://data.mendeley.com/datasets/gvgnvkw4ct/1. Uncropped immunoblots can be found in Supplementary Fig. 21.

Fig. 10. Cytosolic LPS induces pyroptosis but not LMP in both HT3^B3-WT and HT3^B3-KO cell lines.

a HT3^B3-WT (blue) or HT3^B3-KO (red) cells were subjected to nucleofection (nuc, +) in the absence (−) or presence (+) of 5 ug/ml LPS-EK (LPS), caspase-1 inhibitor, VX765, or lysosomal cysteine protease inhibitor, E64. An aliquot of cells were also not nucleofected (nuc, −). After 3 h, cells were stained with SG and Hoescht 33342 and imaged. Quantification: (# Sytox positive nuclei/# of blue nuclei) × 100. Means ± SD were compared using a two-tailed t-test (*P < 0.05). b Immunoblot of gasdermin D (GSDMD) and gasdermin E (GSDME) cleavage in THP1 cells treated with α-hemolysin (HlA); positive control. HT3^B3-WT/HT3^B3-KO were nucleofected with 0, 1, or 5 µg of LPS and left to recover for 1 h. Open arrowheads: full-length GSDMD and GSDME based on molecular mass. Black arrowheads: cleaved GSDMD. Dashed box: area contrast enhanced for clarity. Note, the positive control for GSDME cleavage is provided in Fig. S18b. c, d Flow cytometry analysis of HT3^B3-WT/HT3^B3-KO cells treated with a lysosomotropic agent, LLOMe (c) or 1 µg/ml LPS (d). Cells were stained with the lysosomotropic dye, acridine orange (Y-axis), and the cell viability dye, Sytox™ blue (X-axis). Lines indicate the threshold fluorescence levels (gate) to determine each quadrant. Numbers indicate cell percentage in each quadrant. e Schematic representation of each quadrant. Quadrant 1 (Q1) contained live cells positive for lysosomal staining (acridine orange positive, Sytox blue negative). Quadrant 2 (Q2) contained dead cells with positive lysosomal staining (acridine orange positive, Sytox blue positive). Quadrant 3 (Q3) contained dead cells negative for lysosomal staining (acridine orange negative, Sytox blue positive). Quadrant 4 (Q4) contained live cells negative for lysosomal staining (acridine orange negative, Sytox blue negative). Arrows indicate the timing of fluorescence loss for different cell death pathways. If the lysosomal loss occurred prior to plasma membrane permeabilization, then the percentage of cells will increase from Q1 to Q4 to Q3 over time. If the lysosomal loss occurred after plasma membrane loss, then the percentage of cells will increase from Q1 to Q2 to Q3 over time. f, g Graphical representation of the average of three separate experiments indicating the percentage of cells treated with LLOMe (f) or nucleofected LPS (g) in HT3^B3-WT (blue) and HT3^B3-KO (red) cells. Error bars represent means ± SD. Uncropped immunoblots can be found in Supplementary Fig. 21.